code stringlengths 31 1.05M | apis list | extract_api stringlengths 97 1.91M |
|---|---|---|
import tensorflow as tf
from scipy.stats import rankdata
import numpy as np
import os
import time
import datetime
from argparse import ArgumentParser, ArgumentDefaultsHelpFormatter
from builddata_softplus import *
from capsuleNet import CapsE
# Parameters
# ==================================================
parser = ArgumentParser("CapsE", formatter_class=ArgumentDefaultsHelpFormatter, conflict_handler='resolve')
parser.add_argument("--data", default="./data/", help="Data sources.")
parser.add_argument("--run_folder", default="./", help="Data sources.")
parser.add_argument("--name", default="WN18RR", help="Name of the dataset.")
parser.add_argument("--embedding_dim", default=100, type=int,
help="Dimensionality of character embedding (default: 128)")
parser.add_argument("--filter_size", default=1, type=int, help="Comma-separated filter sizes (default: '3,4,5')")
parser.add_argument("--num_filters", default=400, type=int, help="Number of filters per filter size (default: 128)")
parser.add_argument("--learning_rate", default=0.00001, type=float, help="Learning rate")
parser.add_argument("--batch_size", default=128, type=int, help="Batch Size")
parser.add_argument("--neg_ratio", default=1.0, help="Number of negative triples generated by positive (default: 1.0)")
parser.add_argument("--useInitialization", default=True, type=bool, help="Using the pretrained embeddings")
parser.add_argument("--num_epochs", default=51, type=int, help="Number of training epochs")
parser.add_argument("--savedEpochs", default=10, type=int, help="")
parser.add_argument("--allow_soft_placement", default=True, type=bool, help="Allow device soft device placement")
parser.add_argument("--log_device_placement", default=False, type=bool, help="Log placement of ops on devices")
parser.add_argument("--model_name", default='wn18rr_400_4', help="")
parser.add_argument("--useConstantInit", action='store_true')
parser.add_argument('--iter_routing', default=1, type=int, help='number of iterations in routing algorithm')
parser.add_argument('--num_outputs_secondCaps', default=1, type=int, help='')
parser.add_argument('--vec_len_secondCaps', default=10, type=int, help='')
parser.add_argument("--model_index", default='30')
parser.add_argument("--num_splits", default=8, type=int)
parser.add_argument("--testIdx", default=1, type=int, help="From 0 to 7")
parser.add_argument("--decode", action='store_false')
args = parser.parse_args()
print(args)
# Load data
print("Loading data...")
train, valid, test, words_indexes, indexes_words, \
headTailSelector, entity2id, id2entity, relation2id, id2relation = build_data(path=args.data, name=args.name)
data_size = len(train)
train_batch = Batch_Loader(train, words_indexes, indexes_words, headTailSelector, \
entity2id, id2entity, relation2id, id2relation, batch_size=args.batch_size,
neg_ratio=args.neg_ratio)
entity_array = np.array(list(train_batch.indexes_ents.keys()))
initialization = []
if args.useInitialization == True:
print("Using pre-trained initialization.")
initialization = np.empty([len(words_indexes), args.embedding_dim]).astype(np.float32)
initEnt, initRel = init_norm_Vector(args.data + args.name + '/relation2vec' + str(args.embedding_dim) + '.init',
args.data + args.name + '/entity2vec' + str(args.embedding_dim) + '.init',
args.embedding_dim)
for _word in words_indexes:
if _word in relation2id:
index = relation2id[_word]
_ind = words_indexes[_word]
initialization[_ind] = initRel[index]
elif _word in entity2id:
index = entity2id[_word]
_ind = words_indexes[_word]
initialization[_ind] = initEnt[index]
else:
print('*****************Error********************!')
break
initialization = np.array(initialization, dtype=np.float32)
assert len(words_indexes) % (len(entity2id) + len(relation2id)) == 0
print("Loading data... finished!")
x_valid = np.array(list(valid.keys())).astype(np.int32)
y_valid = np.array(list(valid.values())).astype(np.float32)
len_valid = len(x_valid)
batch_valid = int(len_valid / (args.num_splits - 1))
x_test = np.array(list(test.keys())).astype(np.int32)
y_test = np.array(list(test.values())).astype(np.float32)
len_test = len(x_test)
batch_test = int(len_test / (args.num_splits - 1))
# uncomment when tuning hyper-parameters on the validation set
# x_test = x_valid
# y_test = y_valid
# len_test = len_valid
# batch_test = batch_valid
##########################################
if args.decode == False:
lstModelNames = list(args.model_name.split(","))
for _model_name in lstModelNames:
out_dir = os.path.abspath(os.path.join(args.run_folder, "runs_CapsE", _model_name))
print("Evaluating {}\n".format(out_dir))
checkpoint_dir = os.path.abspath(os.path.join(out_dir, "checkpoints"))
checkpoint_prefix = os.path.join(checkpoint_dir, "model")
lstModelIndexes = list(args.model_index.split(","))
for _model_index in lstModelIndexes:
_file = checkpoint_prefix + "-" + _model_index
lstHT = []
for _index in range(args.num_splits):
with open(_file + '.eval.' + str(_index) + '.txt') as f:
for _line in f:
if _line.strip() != '':
lstHT.append(list(map(float, _line.strip().split())))
lstHT = np.array(lstHT)
print(_file, 'mr, mrr, hits@1, hits@10 --> ', np.sum(lstHT, axis=0) / (2 * len_test))
print('------------------------------------')
else:
with tf.Graph().as_default():
tf.set_random_seed(1234)
session_conf = tf.ConfigProto(allow_soft_placement=args.allow_soft_placement,
log_device_placement=args.log_device_placement)
session_conf.gpu_options.allow_growth = True
sess = tf.Session(config=session_conf)
with sess.as_default():
global_step = tf.Variable(0, name="global_step", trainable=False)
capse = CapsE(sequence_length=x_valid.shape[1],
initialization=initialization,
embedding_size=args.embedding_dim,
filter_size=args.filter_size,
num_filters=args.num_filters,
vocab_size=len(words_indexes),
iter_routing=args.iter_routing,
batch_size=2 * args.batch_size,
num_outputs_secondCaps=args.num_outputs_secondCaps,
vec_len_secondCaps=args.vec_len_secondCaps,
useConstantInit=args.useConstantInit
)
# Output directory for models and summaries
lstModelNames = list(args.model_name.split(","))
for _model_name in lstModelNames:
out_dir = os.path.abspath(os.path.join(args.run_folder, "runs_CapsE", _model_name))
print("Evaluating {}\n".format(out_dir))
# Checkpoint directory. Tensorflow assumes this directory already exists so we need to create it
checkpoint_dir = os.path.abspath(os.path.join(out_dir, "checkpoints"))
checkpoint_prefix = os.path.join(checkpoint_dir, "model")
lstModelIndexes = list(args.model_index.split(","))
for _model_index in lstModelIndexes:
_file = checkpoint_prefix + "-" + _model_index
capse.saver.restore(sess, _file)
print("Loaded model", _file)
# Predict function to predict scores for test data
def predict(x_batch, y_batch, writer=None):
feed_dict = {
capse.input_x: x_batch,
capse.input_y: y_batch
}
scores = sess.run([capse.predictions], feed_dict)
return scores
def test_prediction(x_batch, y_batch, head_or_tail='head'):
hits10 = 0.0
mrr = 0.0
mr = 0.0
hits1 = 0.0
for i in range(len(x_batch)):
new_x_batch = np.tile(x_batch[i], (len(entity2id), 1))
new_y_batch = np.tile(y_batch[i], (len(entity2id), 1))
if head_or_tail == 'head':
new_x_batch[:, 0] = entity_array
else: # 'tail'
new_x_batch[:, 2] = entity_array
lstIdx = []
for tmpIdxTriple in range(len(new_x_batch)):
tmpTriple = (new_x_batch[tmpIdxTriple][0], new_x_batch[tmpIdxTriple][1],
new_x_batch[tmpIdxTriple][2])
if (tmpTriple in train) or (tmpTriple in valid) or (
tmpTriple in test): # also remove the valid test triple
lstIdx.append(tmpIdxTriple)
new_x_batch = np.delete(new_x_batch, lstIdx, axis=0)
new_y_batch = np.delete(new_y_batch, lstIdx, axis=0)
# thus, insert the valid test triple again, to the beginning of the array
new_x_batch = np.insert(new_x_batch, 0, x_batch[i], axis=0) # thus, the index of the valid test triple is equal to 0
new_y_batch = np.insert(new_y_batch, 0, y_batch[i], axis=0)
# for running with a batch size
while len(new_x_batch) % ((int(args.neg_ratio) + 1) * args.batch_size) != 0:
new_x_batch = np.append(new_x_batch, [x_batch[i]], axis=0)
new_y_batch = np.append(new_y_batch, [y_batch[i]], axis=0)
results = []
listIndexes = range(0, len(new_x_batch), (int(args.neg_ratio) + 1) * args.batch_size)
for tmpIndex in range(len(listIndexes) - 1):
results = np.append(results, predict(
new_x_batch[listIndexes[tmpIndex]:listIndexes[tmpIndex + 1]],
new_y_batch[listIndexes[tmpIndex]:listIndexes[tmpIndex + 1]]))
results = np.append(results, predict(new_x_batch[listIndexes[-1]:], new_y_batch[listIndexes[-1]:]))
results = np.reshape(results, -1)
results_with_id = rankdata(results, method='ordinal')
_filter = results_with_id[0]
mr += _filter
mrr += 1.0 / _filter
if _filter <= 10:
hits10 += 1
if _filter == 1:
hits1 += 1
return np.array([mr, mrr, hits1, hits10])
if args.testIdx < (args.num_splits - 1):
head_results = test_prediction(
x_test[batch_test * args.testIdx: batch_test * (args.testIdx + 1)],
y_test[batch_test * args.testIdx: batch_test * (args.testIdx + 1)],
head_or_tail='head')
tail_results = test_prediction(
x_test[batch_test * args.testIdx: batch_test * (args.testIdx + 1)],
y_test[batch_test * args.testIdx: batch_test * (args.testIdx + 1)],
head_or_tail='tail')
else:
head_results = test_prediction(x_test[batch_test * args.testIdx: len_test],
y_test[batch_test * args.testIdx: len_test],
head_or_tail='head')
tail_results = test_prediction(x_test[batch_test * args.testIdx: len_test],
y_test[batch_test * args.testIdx: len_test],
head_or_tail='tail')
wri = open(_file + '.eval.' + str(args.testIdx) + '.txt', 'w')
for _val in head_results:
wri.write(str(_val) + ' ')
wri.write('\n')
for _val in tail_results:
wri.write(str(_val) + ' ')
wri.write('\n')
wri.close()
| [
"numpy.sum",
"argparse.ArgumentParser",
"tensorflow.Session",
"scipy.stats.rankdata",
"tensorflow.set_random_seed",
"numpy.insert",
"tensorflow.ConfigProto",
"numpy.append",
"tensorflow.Variable",
"numpy.array",
"numpy.reshape",
"tensorflow.Graph",
"os.path.join",
"numpy.delete"
] | [((319, 421), 'argparse.ArgumentParser', 'ArgumentParser', (['"""CapsE"""'], {'formatter_class': 'ArgumentDefaultsHelpFormatter', 'conflict_handler': '"""resolve"""'}), "('CapsE', formatter_class=ArgumentDefaultsHelpFormatter,\n conflict_handler='resolve')\n", (333, 421), False, 'from argparse import ArgumentParser, ArgumentDefaultsHelpFormatter\n'), ((3735, 3777), 'numpy.array', 'np.array', (['initialization'], {'dtype': 'np.float32'}), '(initialization, dtype=np.float32)\n', (3743, 3777), True, 'import numpy as np\n'), ((4797, 4834), 'os.path.join', 'os.path.join', (['checkpoint_dir', '"""model"""'], {}), "(checkpoint_dir, 'model')\n", (4809, 4834), False, 'import os\n'), ((5412, 5436), 'tensorflow.set_random_seed', 'tf.set_random_seed', (['(1234)'], {}), '(1234)\n', (5430, 5436), True, 'import tensorflow as tf\n'), ((5454, 5568), 'tensorflow.ConfigProto', 'tf.ConfigProto', ([], {'allow_soft_placement': 'args.allow_soft_placement', 'log_device_placement': 'args.log_device_placement'}), '(allow_soft_placement=args.allow_soft_placement,\n log_device_placement=args.log_device_placement)\n', (5468, 5568), True, 'import tensorflow as tf\n'), ((5632, 5663), 'tensorflow.Session', 'tf.Session', ([], {'config': 'session_conf'}), '(config=session_conf)\n', (5642, 5663), True, 'import tensorflow as tf\n'), ((4601, 4657), 'os.path.join', 'os.path.join', (['args.run_folder', '"""runs_CapsE"""', '_model_name'], {}), "(args.run_folder, 'runs_CapsE', _model_name)\n", (4613, 4657), False, 'import os\n'), ((4737, 4773), 'os.path.join', 'os.path.join', (['out_dir', '"""checkpoints"""'], {}), "(out_dir, 'checkpoints')\n", (4749, 4773), False, 'import os\n'), ((5218, 5233), 'numpy.array', 'np.array', (['lstHT'], {}), '(lstHT)\n', (5226, 5233), True, 'import numpy as np\n'), ((5707, 5758), 'tensorflow.Variable', 'tf.Variable', (['(0)'], {'name': '"""global_step"""', 'trainable': '(False)'}), "(0, name='global_step', trainable=False)\n", (5718, 5758), True, 'import tensorflow as tf\n'), ((5385, 5395), 'tensorflow.Graph', 'tf.Graph', ([], {}), '()\n', (5393, 5395), True, 'import tensorflow as tf\n'), ((6728, 6765), 'os.path.join', 'os.path.join', (['checkpoint_dir', '"""model"""'], {}), "(checkpoint_dir, 'model')\n", (6740, 6765), False, 'import os\n'), ((5283, 5304), 'numpy.sum', 'np.sum', (['lstHT'], {'axis': '(0)'}), '(lstHT, axis=0)\n', (5289, 5304), True, 'import numpy as np\n'), ((6424, 6480), 'os.path.join', 'os.path.join', (['args.run_folder', '"""runs_CapsE"""', '_model_name'], {}), "(args.run_folder, 'runs_CapsE', _model_name)\n", (6436, 6480), False, 'import os\n'), ((6666, 6702), 'os.path.join', 'os.path.join', (['out_dir', '"""checkpoints"""'], {}), "(out_dir, 'checkpoints')\n", (6678, 6702), False, 'import os\n'), ((9464, 9498), 'numpy.array', 'np.array', (['[mr, mrr, hits1, hits10]'], {}), '([mr, mrr, hits1, hits10])\n', (9472, 9498), True, 'import numpy as np\n'), ((8079, 8117), 'numpy.delete', 'np.delete', (['new_x_batch', 'lstIdx'], {'axis': '(0)'}), '(new_x_batch, lstIdx, axis=0)\n', (8088, 8117), True, 'import numpy as np\n'), ((8139, 8177), 'numpy.delete', 'np.delete', (['new_y_batch', 'lstIdx'], {'axis': '(0)'}), '(new_y_batch, lstIdx, axis=0)\n', (8148, 8177), True, 'import numpy as np\n'), ((8281, 8326), 'numpy.insert', 'np.insert', (['new_x_batch', '(0)', 'x_batch[i]'], {'axis': '(0)'}), '(new_x_batch, 0, x_batch[i], axis=0)\n', (8290, 8326), True, 'import numpy as np\n'), ((8406, 8451), 'numpy.insert', 'np.insert', (['new_y_batch', '(0)', 'y_batch[i]'], {'axis': '(0)'}), '(new_y_batch, 0, y_batch[i], axis=0)\n', (8415, 8451), True, 'import numpy as np\n'), ((9190, 9213), 'numpy.reshape', 'np.reshape', (['results', '(-1)'], {}), '(results, -1)\n', (9200, 9213), True, 'import numpy as np\n'), ((9239, 9274), 'scipy.stats.rankdata', 'rankdata', (['results'], {'method': '"""ordinal"""'}), "(results, method='ordinal')\n", (9247, 9274), False, 'from scipy.stats import rankdata\n'), ((8598, 8642), 'numpy.append', 'np.append', (['new_x_batch', '[x_batch[i]]'], {'axis': '(0)'}), '(new_x_batch, [x_batch[i]], axis=0)\n', (8607, 8642), True, 'import numpy as np\n'), ((8665, 8709), 'numpy.append', 'np.append', (['new_y_batch', '[y_batch[i]]'], {'axis': '(0)'}), '(new_y_batch, [y_batch[i]], axis=0)\n', (8674, 8709), True, 'import numpy as np\n')] |
#%%
import matplotlib.pyplot as plt
from numpy import sin, pi
from cmath import phase
import os
def stringToComplex(s):
s = s.replace('\n', '')
s = s.replace('j', '')
s = s.replace(' ', '')
real, imag = tuple( s.split('+') )
real, imag = float(real), float(imag)
return complex(real, imag)
def f(x):
y = [(1/n) * sin(2*pi*100*n*x) for n in range(1, 10, 2)]
return sum(y)
#%%
n = 2**12
data = [ f(t/n) for t in range(n) ]
plt.figure(figsize = (12, 5))
plt.plot(data)
plt.grid()
plt.savefig('images/input.png', dpi=300)
plt.show()
python_dir = os.path.dirname(__file__)
input_path = os.path.join(python_dir, "data/input.txt")
output_path = os.path.join(python_dir, "data/output.txt")
buildandrun_path = os.path.join(python_dir, "BuildAndRun.bat")
with open(input_path, "w+") as file:
file.write(str(n) + '\n')
for x in data:
file.write(str(x) + '\n')
#%%
os.system(buildandrun_path)
#%%
with open(output_path, "r") as file:
data.clear()
for x in file.readlines():
data.append( stringToComplex(x) )
fig, (ax1, ax2) = plt.subplots(2, 1, figsize = (12, 8))
ax1.plot([abs(x) for x in data])
ax1.set_title('Valor absoluto')
ax1.grid()
ax2.plot([phase(x) for x in data])
ax2.set_title('Fase')
ax2.grid()
plt.tight_layout()
plt.savefig('images/output.png', dpi=300)
plt.show()
# %%
| [
"matplotlib.pyplot.tight_layout",
"matplotlib.pyplot.show",
"os.path.join",
"matplotlib.pyplot.plot",
"cmath.phase",
"os.path.dirname",
"os.system",
"matplotlib.pyplot.subplots",
"matplotlib.pyplot.figure",
"numpy.sin",
"matplotlib.pyplot.grid",
"matplotlib.pyplot.savefig"
] | [((457, 484), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(12, 5)'}), '(figsize=(12, 5))\n', (467, 484), True, 'import matplotlib.pyplot as plt\n'), ((487, 501), 'matplotlib.pyplot.plot', 'plt.plot', (['data'], {}), '(data)\n', (495, 501), True, 'import matplotlib.pyplot as plt\n'), ((502, 512), 'matplotlib.pyplot.grid', 'plt.grid', ([], {}), '()\n', (510, 512), True, 'import matplotlib.pyplot as plt\n'), ((513, 553), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""images/input.png"""'], {'dpi': '(300)'}), "('images/input.png', dpi=300)\n", (524, 553), True, 'import matplotlib.pyplot as plt\n'), ((554, 564), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (562, 564), True, 'import matplotlib.pyplot as plt\n'), ((579, 604), 'os.path.dirname', 'os.path.dirname', (['__file__'], {}), '(__file__)\n', (594, 604), False, 'import os\n'), ((618, 660), 'os.path.join', 'os.path.join', (['python_dir', '"""data/input.txt"""'], {}), "(python_dir, 'data/input.txt')\n", (630, 660), False, 'import os\n'), ((675, 718), 'os.path.join', 'os.path.join', (['python_dir', '"""data/output.txt"""'], {}), "(python_dir, 'data/output.txt')\n", (687, 718), False, 'import os\n'), ((738, 781), 'os.path.join', 'os.path.join', (['python_dir', '"""BuildAndRun.bat"""'], {}), "(python_dir, 'BuildAndRun.bat')\n", (750, 781), False, 'import os\n'), ((908, 935), 'os.system', 'os.system', (['buildandrun_path'], {}), '(buildandrun_path)\n', (917, 935), False, 'import os\n'), ((1099, 1134), 'matplotlib.pyplot.subplots', 'plt.subplots', (['(2)', '(1)'], {'figsize': '(12, 8)'}), '(2, 1, figsize=(12, 8))\n', (1111, 1134), True, 'import matplotlib.pyplot as plt\n'), ((1310, 1328), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (1326, 1328), True, 'import matplotlib.pyplot as plt\n'), ((1333, 1374), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""images/output.png"""'], {'dpi': '(300)'}), "('images/output.png', dpi=300)\n", (1344, 1374), True, 'import matplotlib.pyplot as plt\n'), ((1379, 1389), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (1387, 1389), True, 'import matplotlib.pyplot as plt\n'), ((343, 368), 'numpy.sin', 'sin', (['(2 * pi * 100 * n * x)'], {}), '(2 * pi * 100 * n * x)\n', (346, 368), False, 'from numpy import sin, pi\n'), ((1239, 1247), 'cmath.phase', 'phase', (['x'], {}), '(x)\n', (1244, 1247), False, 'from cmath import phase\n')] |
import numpy
from matchms import Spectrum
from matplotlib import pyplot as plt
def test_spectrum_plot_with_histogram_unspecified():
mz = numpy.array([10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110], dtype="float")
intensities = numpy.array([1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9], dtype="float")
spectrum = Spectrum(mz=mz, intensities=intensities)
fig = spectrum.plot()
assert fig is not None
assert hasattr(fig, "axes")
assert isinstance(fig.axes, list)
assert len(fig.axes) == 1
assert isinstance(fig.axes[0], plt.Axes)
assert hasattr(fig.axes[0], "lines")
assert isinstance(fig.axes[0].lines, list)
assert len(fig.axes[0].lines) == 11
assert isinstance(fig.axes[0].lines[0], plt.Line2D)
assert hasattr(fig.axes[0].lines[0], "_x")
def test_spectrum_plot_with_histogram_false():
mz = numpy.array([10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110], dtype="float")
intensities = numpy.array([1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9], dtype="float")
spectrum = Spectrum(mz=mz, intensities=intensities)
fig = spectrum.plot(with_histogram=False)
assert fig is not None
assert hasattr(fig, "axes")
assert isinstance(fig.axes, list)
assert len(fig.axes) == 1
assert isinstance(fig.axes[0], plt.Axes)
assert hasattr(fig.axes[0], "lines")
assert isinstance(fig.axes[0].lines, list)
assert len(fig.axes[0].lines) == 11
assert isinstance(fig.axes[0].lines[0], plt.Line2D)
assert hasattr(fig.axes[0].lines[0], "_x")
def test_spectrum_plot_with_histogram_true():
mz = numpy.array([10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110], dtype="float")
intensities = numpy.array([1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9], dtype="float")
spectrum = Spectrum(mz=mz, intensities=intensities)
fig = spectrum.plot(with_histogram=True)
assert fig is not None
assert hasattr(fig, "axes")
assert isinstance(fig.axes, list)
assert len(fig.axes) == 2
assert isinstance(fig.axes[0], plt.Axes)
assert hasattr(fig.axes[0], "lines")
assert isinstance(fig.axes[0].lines, list)
assert len(fig.axes[0].lines) == 11
assert isinstance(fig.axes[0].lines[0], plt.Line2D)
assert hasattr(fig.axes[0].lines[0], "_x")
def test_spectrum_plot_with_histogram_true_and_intensity_limit():
mz = numpy.array([10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110], dtype="float")
intensities = numpy.array([1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9], dtype="float")
spectrum = Spectrum(mz=mz, intensities=intensities)
fig = spectrum.plot(with_histogram=True, intensity_to=10.0)
assert fig is not None
assert hasattr(fig, "axes")
assert isinstance(fig.axes, list)
assert len(fig.axes) == 2
assert isinstance(fig.axes[0], plt.Axes)
assert hasattr(fig.axes[0], "lines")
assert isinstance(fig.axes[0].lines, list)
assert len(fig.axes[0].lines) == 11
assert isinstance(fig.axes[0].lines[0], plt.Line2D)
assert hasattr(fig.axes[0].lines[0], "_x")
def test_spectrum_plot_with_histogram_true_and_expfit_true_and_intensity_limit():
mz = numpy.array([10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110], dtype="float")
intensities = numpy.array([1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9], dtype="float")
spectrum = Spectrum(mz=mz, intensities=intensities)
fig = spectrum.plot(with_histogram=True, with_expfit=True, intensity_to=10.0)
assert fig is not None
assert hasattr(fig, "axes")
assert isinstance(fig.axes, list)
assert len(fig.axes) == 2
assert isinstance(fig.axes[0], plt.Axes)
assert hasattr(fig.axes[0], "lines")
assert isinstance(fig.axes[0].lines, list)
assert len(fig.axes[0].lines) == 11
assert isinstance(fig.axes[0].lines[0], plt.Line2D)
assert hasattr(fig.axes[0].lines[0], "_x")
| [
"matchms.Spectrum",
"numpy.array"
] | [((144, 218), 'numpy.array', 'numpy.array', (['[10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110]'], {'dtype': '"""float"""'}), "([10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110], dtype='float')\n", (155, 218), False, 'import numpy\n'), ((237, 298), 'numpy.array', 'numpy.array', (['[1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9]'], {'dtype': '"""float"""'}), "([1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9], dtype='float')\n", (248, 298), False, 'import numpy\n'), ((314, 354), 'matchms.Spectrum', 'Spectrum', ([], {'mz': 'mz', 'intensities': 'intensities'}), '(mz=mz, intensities=intensities)\n', (322, 354), False, 'from matchms import Spectrum\n'), ((845, 919), 'numpy.array', 'numpy.array', (['[10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110]'], {'dtype': '"""float"""'}), "([10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110], dtype='float')\n", (856, 919), False, 'import numpy\n'), ((938, 999), 'numpy.array', 'numpy.array', (['[1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9]'], {'dtype': '"""float"""'}), "([1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9], dtype='float')\n", (949, 999), False, 'import numpy\n'), ((1015, 1055), 'matchms.Spectrum', 'Spectrum', ([], {'mz': 'mz', 'intensities': 'intensities'}), '(mz=mz, intensities=intensities)\n', (1023, 1055), False, 'from matchms import Spectrum\n'), ((1565, 1639), 'numpy.array', 'numpy.array', (['[10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110]'], {'dtype': '"""float"""'}), "([10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110], dtype='float')\n", (1576, 1639), False, 'import numpy\n'), ((1658, 1719), 'numpy.array', 'numpy.array', (['[1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9]'], {'dtype': '"""float"""'}), "([1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9], dtype='float')\n", (1669, 1719), False, 'import numpy\n'), ((1735, 1775), 'matchms.Spectrum', 'Spectrum', ([], {'mz': 'mz', 'intensities': 'intensities'}), '(mz=mz, intensities=intensities)\n', (1743, 1775), False, 'from matchms import Spectrum\n'), ((2304, 2378), 'numpy.array', 'numpy.array', (['[10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110]'], {'dtype': '"""float"""'}), "([10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110], dtype='float')\n", (2315, 2378), False, 'import numpy\n'), ((2397, 2458), 'numpy.array', 'numpy.array', (['[1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9]'], {'dtype': '"""float"""'}), "([1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9], dtype='float')\n", (2408, 2458), False, 'import numpy\n'), ((2474, 2514), 'matchms.Spectrum', 'Spectrum', ([], {'mz': 'mz', 'intensities': 'intensities'}), '(mz=mz, intensities=intensities)\n', (2482, 2514), False, 'from matchms import Spectrum\n'), ((3078, 3152), 'numpy.array', 'numpy.array', (['[10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110]'], {'dtype': '"""float"""'}), "([10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110], dtype='float')\n", (3089, 3152), False, 'import numpy\n'), ((3171, 3232), 'numpy.array', 'numpy.array', (['[1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9]'], {'dtype': '"""float"""'}), "([1, 1, 5, 5, 5, 5, 7, 7, 7, 9, 9], dtype='float')\n", (3182, 3232), False, 'import numpy\n'), ((3248, 3288), 'matchms.Spectrum', 'Spectrum', ([], {'mz': 'mz', 'intensities': 'intensities'}), '(mz=mz, intensities=intensities)\n', (3256, 3288), False, 'from matchms import Spectrum\n')] |
# third party
import numpy as np
import pytest
# syft absolute
from syft import deserialize
from syft import serialize
from syft.core.adp.entity import Entity
from syft.core.tensor.tensor import Tensor
gonzalo = Entity(name="Gonzalo")
@pytest.fixture(scope="function")
def x() -> Tensor:
x = Tensor(np.array([[1, 2, 3], [4, 5, 6]]))
x = x.private(min_val=-1, max_val=7, entity=gonzalo)
return x
@pytest.fixture(scope="function")
def y() -> Tensor:
y = Tensor(np.array([[-1, -2, -3], [-4, -5, -6]]))
y = y.private(min_val=-7, max_val=1, entity=gonzalo)
return y
#
# ######################### ADD ############################
#
# MADHAVA: this needs fixing
@pytest.mark.xfail
def test_add(x: Tensor) -> None:
z = x + x
assert isinstance(z, Tensor), "Add: Result is not a Tensor"
assert (
z.child.min_vals == 2 * x.child.min_vals
).all(), "(Add, Minval) Result is not correct"
assert (
z.child.max_vals == 2 * x.child.max_vals
).all(), "(Add, Maxval) Result is not correct"
# MADHAVA: this needs fixing
@pytest.mark.xfail
def test_single_entity_phi_tensor_serde(x: Tensor) -> None:
blob = serialize(x.child)
x2 = deserialize(blob)
assert (x.child.min_vals == x2.min_vals).all()
assert (x.child.max_vals == x2.max_vals).all()
# def test_add(x,y):
# z = x+y
# assert isinstance(z, Tensor), "Add: Result is not a Tensor"
# assert z.child.min_vals == x.child.min_vals + y.child.min_vals, "(Add, Minval) Result is not correct"
# assert z.child.max_vals == x.child.max_vals + y.child.max_vals, "(Add, Maxval) Result is not correct"
#
# ######################### SUB ############################
#
# def test_sub(x):
# z=x-x
# assert isinstance(z, Tensor), "Sub: Result is not a Tensor"
# assert z.child.min_vals == 0 * x.child.min_vals, "(Sub, Minval) Result is not correct"
# assert z.child.max_vals == 0 * x.child.max_vals, "(Sub, Maxval) Result is not correct"
#
# def test_sub(x,y):
# z=x-y
# assert isinstance(z, Tensor), "Sub: Result is not a Tensor"
# assert z.child.min_vals == x.child.min_vals - y.child.min_vals, "(Sub, Minval) Result is not correct"
# assert z.child.max_vals == x.child.max_vals - y.child.max_vals, "(Sub, Maxval) Result is not correct"
#
# ######################### MUL ############################
#
# def test_mul(x):
# z = x*x
# assert isinstance(z, Tensor), "Mul: Result is not a Tensor"
# assert z.child.min_vals == x.child.min_vals ** 2, "(Mul, Minval) Result is not correct"
# assert z.child.max_vals == x.child.max_vals ** 2, "(Mul, Maxval) Result is not correct"
#
# def test_mul(x,y):
# z = x*y
# assert isinstance(z, Tensor), "Mul: Result is not a Tensor"
# assert z.child.min_vals == x.child.min_vals ** 2, "(Mul, Minval) Result is not correct"
# assert z.child.max_vals == x.child.max_vals ** 2, "(Mul, Maxval) Result is not correct"
| [
"syft.deserialize",
"syft.serialize",
"pytest.fixture",
"syft.core.adp.entity.Entity",
"numpy.array"
] | [((214, 236), 'syft.core.adp.entity.Entity', 'Entity', ([], {'name': '"""Gonzalo"""'}), "(name='Gonzalo')\n", (220, 236), False, 'from syft.core.adp.entity import Entity\n'), ((240, 272), 'pytest.fixture', 'pytest.fixture', ([], {'scope': '"""function"""'}), "(scope='function')\n", (254, 272), False, 'import pytest\n'), ((414, 446), 'pytest.fixture', 'pytest.fixture', ([], {'scope': '"""function"""'}), "(scope='function')\n", (428, 446), False, 'import pytest\n'), ((1165, 1183), 'syft.serialize', 'serialize', (['x.child'], {}), '(x.child)\n', (1174, 1183), False, 'from syft import serialize\n'), ((1193, 1210), 'syft.deserialize', 'deserialize', (['blob'], {}), '(blob)\n', (1204, 1210), False, 'from syft import deserialize\n'), ((307, 339), 'numpy.array', 'np.array', (['[[1, 2, 3], [4, 5, 6]]'], {}), '([[1, 2, 3], [4, 5, 6]])\n', (315, 339), True, 'import numpy as np\n'), ((481, 519), 'numpy.array', 'np.array', (['[[-1, -2, -3], [-4, -5, -6]]'], {}), '([[-1, -2, -3], [-4, -5, -6]])\n', (489, 519), True, 'import numpy as np\n')] |
# -*- coding: utf-8 -*-
'''
Multiclass Budget Support Vector Machine under POM6
'''
__author__ = "<NAME>"
__date__ = "Apr. 2021"
import numpy as np
from MMLL.models.Common_to_all_POMs import Common_to_all_POMs
from transitions import State
from transitions.extensions import GraphMachine
from sklearn.metrics import roc_curve, auc
from pympler import asizeof #asizeof.asizeof(my_object)
import pickle
import dill
import time
class Model():
"""
Multiclass Budget Support Vector Machine.
"""
def __init__(self):
self.C = None
self.w_dict = {}
self.sigma = None
self.Csvm = None
self.classes = None
self.is_trained = False
self.supported_formats = ['pkl']
t = time.time()
seed = int((t - int(t)) * 10000)
np.random.seed(seed=seed)
def predict(self, X):
"""
Predicts outputs given the inputs
Parameters
----------
X: ndarray
Matrix with the input values
Returns
-------
prediction_values: ndarray
"""
NP = X.shape[0]
NC = self.C.shape[0]
XC2 = -2 * np.dot(X, self.C.T)
XC2 += np.sum(np.multiply(X, X), axis=1).reshape((NP, 1))
XC2 += np.sum(np.multiply(self.C, self.C), axis=1).reshape((1, NC))
# Gauss
KXC = np.exp(-XC2 / 2.0 / (self.sigma ** 2))
#1 ./ ( 1 + ((x).^2 / (2 * sigma ^2 )));
#KXC = 1 / (1 + (XC2 / 2.0 / (self.sigma ** 2) ) )
# bias
KXC = np.hstack( (np.ones((NP, 1)), KXC))
preds_dict = {}
NCLA = len(self.classes)
O = []
for cla in self.classes:
o = np.dot(KXC, self.w_dict[cla]).ravel()
preds_dict.update({cla: o})
O.append(o)
O = np.array(O)
winners = list(np.argmax(O, axis=0))
o = np.array([self.classes[pos] for pos in winners]).ravel()
return o
def predict_soft(self, X):
"""
Predicts outputs given the inputs
Parameters
----------
X: ndarray
Matrix with the input values
Returns
-------
prediction_values: ndarray
"""
NP = X.shape[0]
NC = self.C.shape[0]
XC2 = -2 * np.dot(X, self.C.T)
XC2 += np.sum(np.multiply(X, X), axis=1).reshape((NP, 1))
XC2 += np.sum(np.multiply(self.C, self.C), axis=1).reshape((1, NC))
# Gauss
KXC = np.exp(-XC2 / 2.0 / (self.sigma ** 2))
#1 ./ ( 1 + ((x).^2 / (2 * sigma ^2 )));
#KXC = 1 / (1 + (XC2 / 2.0 / (self.sigma ** 2) ) )
# bias
KXC = np.hstack( (np.ones((NP, 1)), KXC))
preds_dict = {}
NCLA = len(self.classes)
O = []
for cla in self.classes:
o = np.dot(KXC, self.w_dict[cla]).ravel()
preds_dict.update({cla: o})
O.append(o)
O = np.array(O)
winners = list(np.argmax(O, axis=0))
o = np.array([self.classes[pos] for pos in winners]).ravel()
return preds_dict
def save(self, filename=None):
"""
Saves the trained model to file. The valid file extensions are:
- "pkl": saves the model as a Python3 pickle file
Parameters
----------
filename: string
path+filename
"""
if filename is None:
print('=' * 80)
print('Model Save Error: A valid filename must be provided, otherwise nothing is saved. The valid file extensions are:')
print('\t - "pkl": saves the model as a Python3 pickle file')
print('\t - "onnx": saves the model using Open Neural Network Exchange format (ONNX)')
print('\t - "pmml": saves the model using Predictive Model Markup Language (PMML)')
print('=' * 80)
else:
# Checking filename extension
extension = filename.split('.')[-1]
if extension not in self.supported_formats:
print('=' * 80)
print('Model Save Error: Unsupported format. The valid file extensions are:')
print('\t - "pkl": saves the model as a Python3 pickle file')
print('=' * 80)
else:
if not self.is_trained:
print('=' * 80)
print('Model Save Error: model not trained yet, nothing to save.')
print('=' * 80)
else:
try:
if extension == 'pkl':
with open(filename, 'wb') as f:
pickle.dump(self, f)
print('=' * 80)
print('Model saved at %s in pickle format.' %filename)
print('=' * 80)
except:
print('=' * 80)
print('Model Save Error: model cannot be saved at %s, please check the provided path/filename.' %filename)
print('=' * 80)
raise
class MBSVM_Master(Common_to_all_POMs):
"""
This class implements the Multiclass Budget Support Vector Machine, run at Master node. It inherits from Common_to_all_POMs.
"""
def __init__(self, master_address, workers_addresses, model_type, comms, logger, verbose=True, **kwargs):
"""
Create a :class:`BSVM_Master` instance.
Parameters
----------
master_address: string
address of the master node
workers_addresses: list of strings
list of the addresses of the workers
comms: comms object instance
object providing communications
logger: class:`logging.Logger`
logging object instance
verbose: boolean
indicates if messages are print or not on screen
kwargs: Keyword arguments.
"""
super().__init__()
self.pom = 6
self.model_type = model_type
self.name = self.model_type + '_Master' # Name
self.master_address = master_address
self.workers_addresses = workers_addresses
self.epsilon = 0.00000001 # to avoid log(0)
self.landa = 0.5
try:
kwargs.update(kwargs['model_parameters'])
del kwargs['model_parameters']
except Exception as err:
pass
self.process_kwargs(kwargs)
# Convert workers_addresses -> '0', '1', + send_to dict
self.broadcast_addresses = workers_addresses
self.Nworkers = len(workers_addresses) # Nworkers
self.workers_addresses = list(range(self.Nworkers))
self.workers_addresses = [str(x) for x in self.workers_addresses]
self.send_to = {}
self.receive_from = {}
for k in range(self.Nworkers):
self.send_to.update({str(k): workers_addresses[k]})
self.receive_from.update({workers_addresses[k]: str(k)})
self.logger = logger # logger
self.comms = comms # comms lib
self.verbose = verbose # print on screen when true
self.NI = None
self.state_dict = {} # dictionary storing the execution state
for k in range(0, self.Nworkers):
self.state_dict.update({self.workers_addresses[k]: ''})
# we extract the model_parameters as extra kwargs, to be all jointly processed
try:
kwargs.update(kwargs['model_parameters'])
del kwargs['model_parameters']
except Exception as err:
pass
self.process_kwargs(kwargs)
self.create_FSM_master()
self.FSMmaster.master_address = master_address
self.message_counter = 100 # used to number the messages
self.cryptonode_address = None
self.KTK_dict = {}
self.KTy_dict = {}
#self.NC = self.C.shape[0]
self.NI = self.C.shape[1]
self.newNI_dict = {}
self.model = Model()
self.model.C = self.C
self.model.sigma = np.sqrt(self.NI) * self.fsigma
self.model.Csvm = self.Csvm
self.Kacum_dict = {}
self.grady_dict = {}
self.s0_dict = {}
self.s1_dict = {}
self.grads_dict = {}
self.Ztr_dict = {}
self.NPtr_dict = {}
self.eps = 0.0000001
try:
if self.target_data_description['NT'] == 1:
if self.target_data_description['output_type'][0]['type'] == 'cat':
self.classes = self.target_data_description['output_type'][0]['values']
else:
self.display('Target values must be categorical (string)')
sys.exit()
else:
self.display('The case with more than one target is not covered yet.')
sys.exit()
except Exception as err:
self.display('The target_data_description is not well defined, please check.', str(err))
raise
t = time.time()
seed = int((t - int(t)) * 10000)
np.random.seed(seed=seed)
def create_FSM_master(self):
"""
Creates a Finite State Machine to be run at the Master Node
Parameters
----------
None
"""
self.display(self.name + ': creating FSM')
states_master = [
State(name='waiting_order', on_enter=['while_waiting_order']),
State(name='update_tr_data', on_enter=['while_update_tr_data']),
State(name='getting_KTK', on_enter=['while_getting_KTK']),
State(name='selecting_C', on_enter=['while_selecting_C']),
State(name='sending_C', on_enter=['while_sending_C']),
State(name='computing_XTw', on_enter=['while_computing_XTw']),
State(name='computing_oi', on_enter=['while_computing_oi']),
State(name='updating_w', on_enter=['while_updating_w'])
]
transitions_master = [
['go_update_tr_data', 'waiting_order', 'update_tr_data'],
['go_waiting_order', 'update_tr_data', 'waiting_order'],
['go_selecting_C', 'waiting_order', 'selecting_C'],
['go_waiting_order', 'selecting_C', 'waiting_order'],
['go_sending_C', 'waiting_order', 'sending_C'],
['go_waiting_order', 'sending_C', 'waiting_order'],
['go_computing_oi', 'waiting_order', 'computing_oi'],
['go_waiting_order', 'computing_oi', 'waiting_order'],
['go_getting_KTK', 'waiting_order', 'getting_KTK'],
['go_waiting_order', 'getting_KTK', 'waiting_order'],
['go_computing_XTw', 'waiting_order', 'computing_XTw'],
['go_waiting_order', 'computing_XTw', 'waiting_order'],
['go_updating_w', 'waiting_order', 'updating_w'],
['go_waiting_order', 'updating_w', 'waiting_order']
]
class FSM_master(object):
self.name = 'FSM_master'
def while_waiting_order(self, MLmodel):
MLmodel.display(MLmodel.name + ': WAITING for instructions...')
return
def while_update_tr_data(self, MLmodel):
try:
action = 'update_tr_data'
data = {}
packet = {'action': action, 'to': 'MLmodel', 'data': data, 'sender': MLmodel.master_address}
message_id = MLmodel.master_address+'_'+str(MLmodel.message_counter)
packet.update({'message_id': message_id})
MLmodel.message_counter += 1
size_bytes = asizeof.asizeof(dill.dumps(packet))
MLmodel.display('COMMS_MASTER_BROADCAST %s, id = %s, bytes=%s' % (action, message_id, str(size_bytes)), verbose=False)
MLmodel.comms.broadcast(packet)
MLmodel.display(MLmodel.name + ': broadcasted update_tr_data to all Workers')
except Exception as err:
raise
'''
message = "ERROR: %s %s" % (str(err), str(type(err)))
MLmodel.display('\n ' + '='*50 + '\n' + message + '\n ' + '='*50 + '\n' )
MLmodel.display('ERROR AT while_update_tr_data')
import code
code.interact(local=locals())
'''
return
def while_selecting_C(self, MLmodel):
try:
action = 'selecting_C'
data = {'C': MLmodel.model.C, 'sigma': MLmodel.model.sigma}
packet = {'action': action, 'to': 'MLmodel', 'data': data, 'sender': MLmodel.master_address}
message_id = MLmodel.master_address+'_'+str(MLmodel.message_counter)
packet.update({'message_id': message_id})
MLmodel.message_counter += 1
size_bytes = asizeof.asizeof(dill.dumps(packet))
MLmodel.display('COMMS_MASTER_BROADCAST %s, id = %s, bytes=%s' % (action, message_id, str(size_bytes)), verbose=False)
if MLmodel.selected_workers is None:
MLmodel.comms.broadcast(packet)
MLmodel.display(MLmodel.name + ': broadcasted C to all Workers')
else:
recipients = [MLmodel.send_to[w] for w in MLmodel.selected_workers]
MLmodel.comms.broadcast(packet, recipients)
MLmodel.display(MLmodel.name + ': broadcasted C to Workers: %s' % str([MLmodel.receive_from[w] for w in MLmodel.selected_workers]))
except Exception as err:
raise
'''
print('ERROR AT while_selecting_C')
import code
code.interact(local=locals())
'''
return
def while_sending_C(self, MLmodel):
try:
action = 'sending_C'
data = {'C': MLmodel.model.C, 'sigma': MLmodel.model.sigma, 'Csvm': MLmodel.model.Csvm, 'classes': MLmodel.classes}
packet = {'action': action, 'to': 'MLmodel', 'data': data, 'sender': MLmodel.master_address}
message_id = MLmodel.master_address+'_'+str(MLmodel.message_counter)
packet.update({'message_id': message_id})
MLmodel.message_counter += 1
size_bytes = asizeof.asizeof(dill.dumps(packet))
MLmodel.display('COMMS_MASTER_BROADCAST %s, id = %s, bytes=%s' % (action, message_id, str(size_bytes)), verbose=False)
if MLmodel.selected_workers is None:
MLmodel.comms.broadcast(packet)
MLmodel.display(MLmodel.name + ': broadcasted C to all Workers')
else:
recipients = [MLmodel.send_to[w] for w in MLmodel.selected_workers]
MLmodel.comms.broadcast(packet, recipients)
MLmodel.display(MLmodel.name + ': broadcasted C to Workers: %s' % str([MLmodel.receive_from[w] for w in MLmodel.selected_workers]))
except Exception as err:
raise
'''
print('ERROR AT while_sending_C')
import code
code.interact(local=locals())
'''
return
def while_computing_XTw(self, MLmodel):
try:
#action = 'computing_XTw'
MLmodel.ACxaxb_dict = {}
xaxbP_dict = {}
for cla in MLmodel.classes:
MLmodel.display('PROC_MASTER_START', verbose=False)
MLmodel.x = MLmodel.model.w_dict[cla].T
NItrain = MLmodel.x.shape[1]
K = int(NItrain / 2)
# Guardar
tmp_dict = {}
tmp_dict.update({'A': np.random.uniform(-10, 10, K).reshape((1, K))})
tmp_dict.update({'C': np.random.uniform(-10, 10, K).reshape((1, K))})
tmp_dict.update({'xa': MLmodel.x[:, 0:K]})
tmp_dict.update({'xb': MLmodel.x[:, K:]})
MLmodel.ACxaxb_dict.update({cla: tmp_dict})
# Enviar
#xa_ = MLmodel.xa + MLmodel.A
#xb_ = MLmodel.xb + MLmodel.C
#P = MLmodel.A + MLmodel.C # warning, check the sum is nonzero (low prob...)
tmp_dict = {}
tmp_dict.update({'xa_': MLmodel.ACxaxb_dict[cla]['xa'] + MLmodel.ACxaxb_dict[cla]['A']})
tmp_dict.update({'xb_': MLmodel.ACxaxb_dict[cla]['xb'] + MLmodel.ACxaxb_dict[cla]['C']})
tmp_dict.update({'P': MLmodel.ACxaxb_dict[cla]['A'] + MLmodel.ACxaxb_dict[cla]['C']})
xaxbP_dict.update({cla: tmp_dict})
MLmodel.display('PROC_MASTER_END', verbose=False)
# broadcasts xaxbP_dict
action = 'sending_xaxbP'
data = {'xaxbP_dict': xaxbP_dict, 'classes': MLmodel.classes}
del xaxbP_dict
packet = {'action': action, 'to': 'MLmodel', 'data': data, 'sender': MLmodel.master_address}
del data
message_id = MLmodel.master_address+'_'+str(MLmodel.message_counter)
packet.update({'message_id': message_id})
MLmodel.message_counter += 1
size_bytes = asizeof.asizeof(dill.dumps(packet))
MLmodel.display('COMMS_MASTER_BROADCAST %s, id = %s, bytes=%s' % (action, message_id, str(size_bytes)), verbose=False)
if MLmodel.selected_workers is None:
MLmodel.comms.broadcast(packet)
MLmodel.display(MLmodel.name + ': computing_XTw with all Workers')
else:
recipients = [MLmodel.send_to[w] for w in MLmodel.selected_workers]
MLmodel.comms.broadcast(packet, recipients)
MLmodel.display(MLmodel.name + ': computing_XTw with Workers: %s' % str(MLmodel.selected_workers))
except Exception as err:
raise
'''
message = "ERROR: %s %s" % (str(err), str(type(err)))
MLmodel.display('\n ' + '='*50 + '\n' + message + '\n ' + '='*50 + '\n' )
MLmodel.display('ERROR AT while_computing_XTw')
import code
code.interact(local=locals())
'''
return
def while_computing_oi(self, MLmodel):
# MLmodel.s0_dict, MLmodel.s1_dict
try:
MLmodel.o_dict = {}
for addr in MLmodel.workers_addresses:
MLmodel.display('PROC_MASTER_START', verbose=False)
# We need to compute these and send them to every worker:
Rzz_dict = {}
rzt_dict = {}
for cla in MLmodel.classes:
#MLmodel.display('PROC_MASTER_START', verbose=False)
U0 = MLmodel.s0_dict[addr]['ya_0_dict'][cla] * (MLmodel.ACxaxb_dict[cla]['xa'] + 2 * MLmodel.ACxaxb_dict[cla]['A']) + MLmodel.s0_dict[addr]['yb_0_dict'][cla] * (MLmodel.ACxaxb_dict[cla]['xb'] + 2 * MLmodel.ACxaxb_dict[cla]['C']) + MLmodel.s0_dict[addr]['Q0_dict'][cla] * (MLmodel.ACxaxb_dict[cla]['A'] + 2 * MLmodel.ACxaxb_dict[cla]['C'])
u0 = np.sum(U0, axis=1)
del U0
s0 = u0 + MLmodel.s0_dict[addr]['v0_dict'][cla]
del u0
NPtr0 = s0.shape[0]
y0 = -np.ones(NPtr0)
e0 = y0 - s0
del s0
# Weighting values a
a0 = np.ones(NPtr0)
ey0 = e0 * y0
which0 = ey0 >= MLmodel.eps
a0[which0] = 2 * MLmodel.Csvm / ey0[which0]
which0 = ey0 < MLmodel.eps
a0[which0] = 2 * MLmodel.Csvm / MLmodel.eps
which0 = ey0 < 0
a0[which0] = 0
a0 = a0.reshape((-1, 1))
del ey0, e0, which0
Rzz0 = MLmodel.Ztr_dict[addr]['Ztr0_dict'][cla] * a0
rzt0 = np.dot(Rzz0.T, y0)
Rzz0 = np.dot(Rzz0.T, MLmodel.Ztr_dict[addr]['Ztr0_dict'][cla])
del a0, y0
#U1 = MLmodel.s1_dict[addr]['ya_1'] * (MLmodel.xa + 2 * MLmodel.A) + MLmodel.s1_dict[addr]['yb_1'] * (MLmodel.xb + 2 * MLmodel.C) + MLmodel.s1_dict[addr]['Q1'] * (MLmodel.A + 2 * MLmodel.C)
U1 = MLmodel.s1_dict[addr]['ya_1_dict'][cla] * (MLmodel.ACxaxb_dict[cla]['xa'] + 2 * MLmodel.ACxaxb_dict[cla]['A']) + MLmodel.s1_dict[addr]['yb_1_dict'][cla] * (MLmodel.ACxaxb_dict[cla]['xb'] + 2 * MLmodel.ACxaxb_dict[cla]['C']) + MLmodel.s1_dict[addr]['Q1_dict'][cla] * (MLmodel.ACxaxb_dict[cla]['A'] + 2 * MLmodel.ACxaxb_dict[cla]['C'])
u1 = np.sum(U1, axis=1)
del U1
s1 = u1 + MLmodel.s1_dict[addr]['v1_dict'][cla]
del u1
NPtr1 = s1.shape[0]
y1 = np.ones(NPtr1)
e1 = y1 - s1
del s1
# Weighting values a
a1 = np.ones(NPtr1)
ey1 = e1 * y1
which1 = ey1 >= MLmodel.eps
a1[which1] = 2 * MLmodel.Csvm / ey1[which1]
which1 = ey1 < MLmodel.eps
a1[which1] = 2 * MLmodel.Csvm / MLmodel.eps
which1 = ey1 < 0
a1[which1] = 0
a1 = a1.reshape((-1, 1))
del ey1, e1, which1
Rzz1 = MLmodel.Ztr_dict[addr]['Ztr1_dict'][cla] * a1
rzt1 = np.dot(Rzz1.T, y1)
Rzz1 = np.dot(Rzz1.T, MLmodel.Ztr_dict[addr]['Ztr1_dict'][cla])
del a1, y1
Rzz_dict.update({cla: Rzz0 + Rzz1})
rzt_dict.update({cla: rzt0 + rzt1})
# Needed ?
#MLmodel.NPtr_dict.update({addr: NPtr0 + NPtr1})
MLmodel.display('PROC_MASTER_END', verbose=False)
action = 'sending_Rzz_rzt'
data = {'Rzz_dict': Rzz_dict, 'rzt_dict': rzt_dict}
#del Rzz0, Rzz1, rzt0, rzt1
packet = {'action': action, 'to': 'MLmodel', 'data': data, 'sender': MLmodel.master_address}
del data
message_id = MLmodel.master_address+'_'+str(MLmodel.message_counter)
packet.update({'message_id': message_id})
MLmodel.message_counter += 1
size_bytes = asizeof.asizeof(dill.dumps(packet))
MLmodel.display('COMMS_MASTER_SEND %s to %s, id = %s, bytes=%s' % (action, addr, message_id, str(size_bytes)), verbose=False)
MLmodel.comms.send(packet, MLmodel.send_to[addr])
#del packet, size_bytes, message_id
del packet
MLmodel.display(MLmodel.name + ' %s: sent sending_Rzz to %s' % (str(MLmodel.master_address), str(addr)))
#del MLmodel.xa, MLmodel.xb, MLmodel.A, MLmodel.C
except:
raise
'''
print('ERROR AT while_computing_oi')
import code
code.interact(local=locals())
'''
return
def while_getting_KTK(self, MLmodel):
try:
action = 'compute_KTK'
data = {'w': MLmodel.model.w}
packet = {'action': action, 'to': 'MLmodel', 'data': data, 'sender': MLmodel.master_address}
message_id = MLmodel.master_address+'_'+str(MLmodel.message_counter)
packet.update({'message_id': message_id})
MLmodel.message_counter += 1
size_bytes = asizeof.asizeof(dill.dumps(packet))
MLmodel.display('COMMS_MASTER_BROADCAST %s, id = %s, bytes=%s' % (action, message_id, str(size_bytes)), verbose=False)
if MLmodel.selected_workers is None:
MLmodel.comms.broadcast(packet)
MLmodel.display(MLmodel.name + ': broadcasted compute_KTK to all workers')
else:
recipients = [MLmodel.send_to[w] for w in MLmodel.selected_workers]
MLmodel.comms.broadcast(packet, recipients)
MLmodel.display(MLmodel.name + ': broadcasted compute_KTK to Workers: %s' % str([MLmodel.receive_from[w] for w in MLmodel.selected_workers]))
except Exception as err:
raise
'''
print('ERROR AT while_getting_KTK')
import code
code.interact(local=locals())
'''
return
def while_updating_w(self, MLmodel):
MLmodel.display('PROC_MASTER_START', verbose=False)
MLmodel.w_old_dict = dict(MLmodel.model.w_dict)
try:
MLmodel.w_new_dict = {}
for cla in MLmodel.classes:
MLmodel.KTK_accum = np.zeros((MLmodel.NItrain, MLmodel.NItrain))
MLmodel.KTy_accum = np.zeros((MLmodel.NItrain, 1))
for waddr in MLmodel.workers_addresses:
MLmodel.KTK_accum += MLmodel.KTK_dict[waddr][cla]
MLmodel.KTy_accum += MLmodel.KTy_dict[waddr][cla].reshape((-1, 1))
#MLmodel.model.w_dict[cla] = np.dot(np.linalg.inv(MLmodel.KTK_accum + MLmodel.Kcc), MLmodel.KTy_accum)
MLmodel.w_new_dict[cla] = np.dot(np.linalg.inv(MLmodel.KTK_accum + MLmodel.Kcc), MLmodel.KTy_accum)
MLmodel.display('PROC_MASTER_END', verbose=False)
except Exception as err:
raise
'''
print('ERROR AT while_updating_w')
import code
code.interact(local=locals())
'''
return
def while_Exit(self, MLmodel):
#print('while_Exit')
return
self.FSMmaster = FSM_master()
self.grafmachine_master = GraphMachine(model=self.FSMmaster,
states=states_master,
transitions=transitions_master,
initial='waiting_order',
show_auto_transitions=False, # default value is False
title="Finite State Machine modelling the behaviour of the master",
show_conditions=False)
return
def reset(self, NI):
"""
Create some empty variables needed by the Master Node
Parameters
----------
NI: integer
Number of input features
"""
self.NI = NI
self.model.w = np.random.normal(0, 0.001, (self.NI + 1, 1)) # weights in plaintext, first value is bias
self.w_old = np.random.normal(0, 1.0, (self.NI + 1, 1))
self.XTDaX_accum = np.zeros((self.NI + 1, self.NI + 1)) # Cov. matrix in plaintext
self.XTDast_accum = np.zeros((self.NI + 1, 1)) # Cov. matrix in plaintext
self.preds_dict = {} # dictionary storing the prediction errors
self.XTX_dict = {}
self.XTy_dict = {}
self.display(self.name + ': Resetting local data')
def train_Master(self):
"""
This is the main training loop, it runs the following actions until
the stop condition is met:
- Update the execution state
- Process the received packets
- Perform actions according to the state
Parameters
----------
None
"""
self.display(self.name + ': Starting training')
self.display('MASTER_INIT', verbose=False)
self.FSMmaster.go_update_tr_data(self)
self.run_Master()
self.display('PROC_MASTER_START', verbose=False)
# Checking the new NI values
newNIs = list(set(list(self.newNI_dict.values())))
if len(newNIs) > 1:
message = 'ERROR: the training data has different number of features...'
self.display(message)
self.display(list(self.newNI_dict.values()))
raise Exception(message)
else:
self.reset(newNIs[0])
## Adding bias to validation data, if any
#if self.Xval_b is not None:
# self.Xval_b = self.add_bias(self.Xval_b).astype(float)
# self.yval = self.yval.astype(float)
'''
self.FSMmaster.go_selecting_C(self)
self.run_Master()
# Selecting centroids with largest projection
Ncandidates = self.C.shape[0]
Kacum_total = np.zeros(Ncandidates)
for addr in self.workers_addresses:
Kacum_total += self.Kacum_dict[addr].ravel()
index = np.argsort(-Kacum_total)
self.C = self.C[index[0: self.NC], :]
'''
self.model.C = self.C
self.NC = self.C.shape[0]
# computing Kcc, only once
X = self.model.C
XC2 = -2 * np.dot(X, self.model.C.T)
XC2 += np.sum(np.multiply(X, X), axis=1).reshape((self.NC, 1))
XC2 += np.sum(np.multiply(self.model.C, self.model.C), axis=1).reshape((1, self.NC))
KCC = np.exp(-XC2 / 2.0 / (self.model.sigma ** 2))
self.Kcc = np.zeros((self.NC + 1, self.NC + 1))
self.Kcc[1:, 1:] = KCC
self.Kcc[0, 0] = 0.00001
self.Bob_data_s = False
self.Bob_data_grad = False
self.NI = self.NC + 1
# Checking dimensions
if int(self.NI / 2) != self.NI / 2: # add one value
self.w_orig_size = self.NI
self.NItrain = self.NI + 1
# Adding row and column to Kcc
self.Kcc = np.hstack((self.Kcc, np.zeros((self.NI, 1))))
self.Kcc = np.vstack((self.Kcc, np.zeros((1, self.NI + 1))))
self.Kcc[self.NI, self.NI] = 1.0
else:
self.w_orig_size = self.NI
self.NItrain = self.NI
# Computing and storing KXC_val
if self.Xval is not None:
XC2 = -2 * np.dot(self.Xval, self.C.T)
XC2 += np.sum(np.multiply(self.Xval, self.Xval), axis=1).reshape((-1, 1))
XC2 += np.sum(np.multiply(self.C, self.C), axis=1).reshape((1, self.NC))
# Gauss
KXC_val = np.exp(-XC2 / 2.0 / (self.model.sigma ** 2))
self.KXC_val = np.hstack( (np.ones((self.Xval.shape[0], 1)), KXC_val)) # NP_val x NC + 1
#self.yval.astype(float).reshape((-1, 1))
self.stop_training = False
self.kiter = 0
self.display('PROC_MASTER_END', verbose=False)
self.FSMmaster.go_sending_C(self)
self.run_Master()
self.ceval_acum = 100
# Checking dimensions
if int(self.NI / 2) != self.NI / 2: # add one value
self.w_orig_size = self.NI
self.NItrain = self.NI + 1
else:
self.w_orig_size = self.NI
self.NItrain = self.NI
self.model.w_dict = {}
self.w_old_dict = {}
self.model.classes = self.classes
for cla in self.classes:
self.model.w_dict.update({cla: np.random.normal(0, 0.001, (self.NItrain, 1))})
self.w_old_dict.update({cla: np.random.normal(0, 0.001, (self.NItrain, 1))})
self.ACC_val = 0
self.ACC_val_old = 0
while not self.stop_training:
self.display('MASTER_ITER_START', verbose=False)
self.FSMmaster.go_computing_XTw(self)
self.run_Master()
# We receive self.s_dict, self.Ztr_dict (once)
self.FSMmaster.go_computing_oi(self)
self.run_Master()
# Updating w
self.FSMmaster.go_updating_w(self)
self.FSMmaster.go_waiting_order(self)
self.display('PROC_MASTER_START', verbose=False)
self.kiter += 1
# Stop if Maxiter is reached
if self.kiter == self.Nmaxiter:
self.stop_training = True
if self.Xval is None: # A validation set is not provided
for cla in self.classes:
self.model.w_dict[cla] = (1 - self.landa) * self.w_old_dict[cla] + self.landa * self.w_new_dict[cla]
inc_w = 0
for cla in self.classes:
inc_w += np.linalg.norm(self.model.w_dict[cla] - self.w_old_dict[cla]) / np.linalg.norm(self.w_old_dict[cla])
message = 'Maxiter = %d, iter = %d, inc_w = %f' % (self.Nmaxiter, self.kiter, inc_w)
#self.display(message)
print(message)
if inc_w <= self.conv_stop:
self.stop_training = True
else:
self.ceval_acum_old = self.ceval_acum
NIval = self.KXC_val.shape[1]
O = []
for cla in self.classes:
#w_ = self.model.w_dict[cla][0: NIval]
w_ = ((1 - self.landa) * self.w_old_dict[cla] + self.landa * self.w_new_dict[cla])[0: NIval]
o_val = np.dot(self.KXC_val, w_).ravel()
O.append(o_val)
O = np.array(O)
winners = list(np.argmax(O, axis=0))
preds_val = np.array([self.classes[pos] for pos in winners]).ravel()
ACC_val = np.mean(preds_val.ravel() == self.yval)
if ACC_val > self.ACC_val_old:
# retain the new
for cla in self.classes:
self.model.w_dict[cla] = (1 - self.landa) * self.w_old_dict[cla] + self.landa * self.w_new_dict[cla]
self.ACC_val_old = ACC_val
message = 'Maxiter = %d, iter = %d, ACC val = %f' % (self.Nmaxiter, self.kiter, ACC_val)
print(message)
else: # restore the previous one and stop
self.model.w_dict = dict(self.w_old_dict)
self.stop_training = True
'''
self.ceval_acum = 0
for cla in self.classes:
yval = np.array(self.yval == cla).astype(float).reshape((-1, 1))
w_ = self.model.w_dict[cla][0: NIval]
w_old_ = self.w_old_dict[cla][0: NIval]
CE_val = []
landas = np.arange(0, 1.0, 0.001)
Xw = np.dot(self.KXC_val, w_)
Xw_old = np.dot(self.KXC_val, w_old_)
for landa in landas:
w_tmp = landa * w_ + (1 - landa) * w_old_
o_tmp = landa * Xw + (1 - landa) * Xw_old
ce_val = np.mean(self.cross_entropy(self.sigm(o_tmp), yval, self.epsilon))
CE_val.append(ce_val)
min_pos = np.argmin(CE_val)
CEval_opt = CE_val[min_pos]
indices = np.array(range(len(landas)))[CE_val == CEval_opt]
min_pos = indices[0] # first
landa_opt = landas[min_pos]
self.ceval_acum += CEval_opt
self.model.w_dict[cla] = (1.0 - landa_opt) * self.w_old_dict[cla] + landa_opt * self.model.w_dict[cla]
message = 'Class = %s, landa_opt = %f' % (cla, landa_opt)
self.display(message)
self.ceval_acum = self.ceval_acum / len(self.classes)
if self.ceval_acum < self.ceval_acum_old:
message = 'Maxiter = %d, iter = %d, CE val = %f' % (self.Nmaxiter, self.kiter, self.ceval_acum)
print(message)
self.w_old_dict = dict(self.model.w_dict)
else:
self.stop_training = True
# We retain the last weight values
self.model.w_dict = dict(self.w_old_dict)
'''
self.display('PROC_MASTER_END', verbose=False)
self.display('MASTER_ITER_END', verbose=False)
self.display(self.name + ': Training is done')
self.model.niter = self.kiter
self.model.is_trained = True
# reduciendo a dimensiรณn original
for cla in self.classes:
self.model.w_dict[cla] = self.model.w_dict[cla][0:self.w_orig_size, :]
self.display('MASTER_FINISH', verbose=False)
def Update_State_Master(self):
"""
We update control the flow given some conditions and parameters
Parameters
----------
None
"""
if self.chekAllStates('ACK_update_tr_data'):
self.FSMmaster.go_waiting_order(self)
if self.chekAllStates('ACK_projecting_C'):
self.FSMmaster.go_waiting_order(self)
if self.chekAllStates('ACK_storing_C'):
self.FSMmaster.go_waiting_order(self)
if self.chekAllStates('ACK_sending_s'):
if not self.Bob_data_s:
self.Bob_data_s = True
self.FSMmaster.go_waiting_order(self)
if self.chekAllStates('ACK_sending_KTK'):
self.FSMmaster.go_waiting_order(self)
def ProcessReceivedPacket_Master(self, packet, sender):
"""
Process the received packet at Master and take some actions, possibly changing the state
Parameters
----------
packet: packet object
packet received (usually a dict with various content)
sender: string
id of the sender
"""
if packet is not None:
try:
#sender = packet['sender']
sender = self.receive_from[packet['sender']]
if packet['action'][0:3] == 'ACK':
self.display(self.name + ': received ACK from %s: %s' % (str(sender), packet['action']))
self.state_dict[sender] = packet['action']
try:
self.display('COMMS_MASTER_RECEIVED %s from %s, id=%s' % (packet['action'], sender, str(packet['message_id'])), verbose=False)
except:
self.display('MASTER MISSING message_id in %s from %s' % (packet['action'], sender), verbose=False)
pass
if packet['action'] == 'ACK_update_tr_data':
self.newNI_dict.update({sender: packet['data']['newNI']})
if packet['action'] == 'ACK_projecting_C':
self.Kacum_dict.update({sender: packet['data']['Kacum']})
if packet['action'] == 'ACK_sending_s':
if not self.Bob_data_s:
self.s0_dict.update({sender: {'ya_0_dict': packet['data']['ya_0_dict'], 'yb_0_dict': packet['data']['yb_0_dict'], 'Q0_dict': packet['data']['Q0_dict'], 'v0_dict': packet['data']['v0_dict']}})
self.s1_dict.update({sender: {'ya_1_dict': packet['data']['ya_1_dict'], 'yb_1_dict': packet['data']['yb_1_dict'], 'Q1_dict': packet['data']['Q1_dict'], 'v1_dict': packet['data']['v1_dict']}})
self.Ztr_dict.update({sender: {'Ztr0_dict': packet['data']['Ztr0_dict'], 'Ztr1_dict': packet['data']['Ztr1_dict']}})
else:
self.s0_dict[sender]['v0_dict'] = packet['data']['v0_dict']
self.s1_dict[sender]['v1_dict'] = packet['data']['v1_dict']
if packet['action'] == 'ACK_sending_KTK':
self.KTK_dict.update({sender: packet['data']['KTK_dict']})
self.KTy_dict.update({sender: packet['data']['KTy_dict']})
except Exception as err:
raise
'''
print('ERROR AT ProcessReceivedPacket_Master')
import code
code.interact(local=locals())
'''
return
#===============================================================
# Worker
#===============================================================
class MBSVM_Worker(Common_to_all_POMs):
'''
Class implementing Multiclass Budget Support Vector Machine, run at Worker
'''
def __init__(self, master_address, worker_address, model_type, comms, logger, verbose=True, Xtr_b=None, ytr=None):
"""
Create a :class:`BSVM_Worker` instance.
Parameters
----------
master_address: string
address of the master node
worker_address: string
id of this worker
model_type: string
type of ML model
comms: comms object instance
object providing communications
logger: class:`logging.Logger`
logging object instance
verbose: boolean
indicates if messages are print or not on screen
Xtr_b: ndarray
2-D numpy array containing the input training patterns
ytr: ndarray
1-D numpy array containing the target training values
"""
self.pom = 6
self.master_address = master_address
self.worker_address = worker_address # The id of this Worker
#self.workers_addresses = workers_addresses # The id of this Worker
self.model_type = model_type
self.comms = comms # The comms library
self.logger = logger # logger
self.name = model_type + '_Worker' # Name
self.verbose = verbose # print on screen when true
self.Xtr_b = Xtr_b
self.ytr = ytr
self.NPtr = len(ytr)
self.w = None
self.create_FSM_worker()
self.message_id = 0 # used to number the messages
self.eps = 0.0000001
self.Bob_data_s = False
self.Bob_data_grad = False
self.message_counter = 100
t = time.time()
seed = int((t - int(t)) * 10000)
np.random.seed(seed=seed)
def create_FSM_worker(self):
"""
Creates a Finite State Machine to be run at the Worker Node
Parameters
----------
None
"""
self.name = 'FSM_worker'
self.display(self.name + ' %s: creating FSM' % (str(self.worker_address)))
class FSM_worker(object):
name = 'FSM_worker'
def while_waiting_order(self, MLmodel):
MLmodel.display(MLmodel.name + ' %s: WAITING for instructions...' % (str(MLmodel.worker_address)))
return
def while_setting_tr_data(self, MLmodel, packet):
try:
NPtr, newNI = MLmodel.Xtr_b.shape
#MLmodel.Xtr_b = MLmodel.add_bias(MLmodel.Xtr_b).astype(float)
#MLmodel.ytr = MLmodel.ytr.astype(float)
action = 'ACK_update_tr_data'
data = {'newNI': newNI}
packet = {'action': action, 'data': data, 'sender': MLmodel.worker_address}
message_id = 'worker_' + MLmodel.worker_address + '_' + str(MLmodel.message_counter)
packet.update({'message_id': message_id})
MLmodel.message_counter += 1
size_bytes = asizeof.asizeof(dill.dumps(packet))
MLmodel.display('COMMS_WORKER_SEND %s to %s, id = %s, bytes=%s' % (action, MLmodel.master_address, message_id, str(size_bytes)), verbose=False)
MLmodel.comms.send(packet, MLmodel.master_address)
MLmodel.display(MLmodel.name + ' %s: sent ACK_update_tr_data' % (str(MLmodel.worker_address)))
except Exception as err:
raise
'''
message = "ERROR: %s %s" % (str(err), str(type(err)))
MLmodel.display('\n ' + '='*50 + '\n' + message + '\n ' + '='*50 + '\n' )
import code
code.interact(local=locals())
#MLmodel.display('ERROR AT while_computing_XTDaX')
'''
def while_projecting_C(self, MLmodel, packet):
# We project X over C and return accumulated
try:
MLmodel.display('PROC_WORKER_START', verbose=False)
MLmodel.C = packet['data']['C']
NC = MLmodel.C.shape[0]
MLmodel.sigma = packet['data']['sigma']
NI = MLmodel.Xtr_b.shape[1]
NP = MLmodel.Xtr_b.shape[0]
#MLmodel.sigma = np.sqrt(NI) * MLmodel.fsigma
X = MLmodel.Xtr_b
XC2 = -2 * np.dot(X, MLmodel.C.T)
XC2 += np.sum(np.multiply(X, X), axis=1).reshape((NP, 1))
XC2 += np.sum(np.multiply(MLmodel.C, MLmodel.C), axis=1).reshape((1, NC))
# Gauss
KXC = np.exp(-XC2 / 2.0 / (MLmodel.sigma ** 2))
#Kacum contains the number of closest patterns
winners = list(np.argmax(KXC, axis=1))
Kacum = np.zeros((NC, 1))
for kc in range(NC):
Kacum[kc] = winners.count(kc)
#Kacum = np.sum(KXC, axis = 0)
MLmodel.display('PROC_WORKER_END', verbose=False)
action = 'ACK_projecting_C'
data = {'Kacum': Kacum}
packet = {'action': action, 'data': data, 'sender': MLmodel.worker_address}
message_id = 'worker_' + MLmodel.worker_address + '_' + str(MLmodel.message_counter)
packet.update({'message_id': message_id})
MLmodel.message_counter += 1
size_bytes = asizeof.asizeof(dill.dumps(packet))
MLmodel.display('COMMS_WORKER_SEND %s to %s, id = %s, bytes=%s' % (action, MLmodel.master_address, message_id, str(size_bytes)), verbose=False)
MLmodel.comms.send(packet, MLmodel.master_address)
MLmodel.display(MLmodel.name + ' %s: sent ACK_projecting_C' % (str(MLmodel.worker_address)))
except Exception as err:
raise
'''
print('ERROR AT while_projecting_C')
import code
code.interact(local=locals())
'''
return
def while_storing_C(self, MLmodel, packet):
# We store C and compute KXC
try:
MLmodel.display('PROC_WORKER_START', verbose=False)
MLmodel.C = packet['data']['C']
MLmodel.Csvm = packet['data']['Csvm']
MLmodel.classes = packet['data']['classes']
NC = MLmodel.C.shape[0]
MLmodel.sigma = packet['data']['sigma']
NI = MLmodel.Xtr_b.shape[1]
NP = MLmodel.Xtr_b.shape[0]
#MLmodel.sigma = np.sqrt(NI) * MLmodel.fsigma
X = MLmodel.Xtr_b
XC2 = -2 * np.dot(X, MLmodel.C.T)
XC2 += np.sum(np.multiply(X, X), axis=1).reshape((NP, 1))
XC2 += np.sum(np.multiply(MLmodel.C, MLmodel.C), axis=1).reshape((1, NC))
# Gauss
KXC = np.exp(-XC2 / 2.0 / (MLmodel.sigma ** 2))
# Poly
#KXC = 1 / (1 + (XC2 / 2.0 / (MLmodel.sigma ** 2) ) )
MLmodel.KXC = np.hstack( (np.ones((NP, 1)), KXC)) # NP x NC + 1
print('KXC min max', np.min(MLmodel.KXC), np.max(MLmodel.KXC))
# Checking NI
NI = MLmodel.KXC.shape[1]
NPtr = MLmodel.KXC.shape[0]
if NI/2 != int(NI/2):
MLmodel.KXC = np.hstack((MLmodel.KXC, np.random.normal(0, 0.01, (MLmodel.NPtr, 1))))
MLmodel.NPtr_train = MLmodel.KXC.shape[0]
MLmodel.NI_train = MLmodel.KXC.shape[1]
# RMD
MLmodel.Cmat_dict = {}
MLmodel.Dmat_dict = {}
MLmodel.Ztr0_dict = {}
MLmodel.Ztr1_dict = {}
for cla in MLmodel.classes:
MLmodel.Cmat_dict.update({cla: np.random.normal(0, 1, (MLmodel.NI_train, MLmodel.NI_train))})
MLmodel.Dmat_dict.update({cla: np.linalg.inv(MLmodel.Cmat_dict[cla])})
ytr = np.array(MLmodel.ytr == cla).astype(float).reshape((-1, 1))
which0 = (ytr == 0).ravel()
MLmodel.Ztr0_dict.update({cla: np.dot(MLmodel.KXC[which0, :], MLmodel.Cmat_dict[cla])})
which1 = (ytr == 1).ravel()
MLmodel.Ztr1_dict.update({cla: np.dot(MLmodel.KXC[which1, :], MLmodel.Cmat_dict[cla])})
MLmodel.display('PROC_WORKER_END', verbose=False)
action = 'ACK_storing_C'
data = {}
packet = {'action': action, 'data': data, 'sender': MLmodel.worker_address}
message_id = 'worker_' + MLmodel.worker_address + '_' + str(MLmodel.message_counter)
packet.update({'message_id': message_id})
MLmodel.message_counter += 1
size_bytes = asizeof.asizeof(dill.dumps(packet))
MLmodel.display('COMMS_WORKER_SEND %s to %s, id = %s, bytes=%s' % (action, MLmodel.master_address, message_id, str(size_bytes)), verbose=False)
MLmodel.comms.send(packet, MLmodel.master_address)
MLmodel.display(MLmodel.name + ' %s: sent ACK_storing_C' % (str(MLmodel.worker_address)))
except Exception as err:
raise
'''
print('ERROR AT while_storing_C')
import code
code.interact(local=locals())
'''
def while_computing_s(self, MLmodel, packet):
try:
MLmodel.display('PROC_WORKER_START', verbose=False)
MLmodel.classes = packet['data']['classes']
if not MLmodel.Bob_data_s:
NI_train = MLmodel.KXC.shape[1]
NPtr_train = MLmodel.Xtr_b.shape[0]
# MLmodel.Cmat_dict
# MLmodel.Dmat_dict
# MLmodel.Ztr0_dict
# MLmodel.Ztr1_dict
K = int(NI_train / 2)
MLmodel.yas0_dict = {}
MLmodel.yas1_dict = {}
MLmodel.ybs0_dict = {}
MLmodel.ybs1_dict = {}
MLmodel.Bs0_dict = {}
MLmodel.Ds0_dict = {}
MLmodel.Bs1_dict = {}
MLmodel.Ds1_dict = {}
# Send once
MLmodel.Qs0_dict = {}
MLmodel.Qs1_dict = {}
MLmodel.ya_s0_dict = {}
MLmodel.yb_s0_dict = {}
MLmodel.ya_s1_dict = {}
MLmodel.yb_s1_dict = {}
for cla in MLmodel.classes:
ytr = np.array(MLmodel.ytr == cla).astype(float).reshape((-1, 1))
which0 = (ytr == 0).ravel()
NPtr0 = np.sum(which0)
which1 = (ytr == 1).ravel()
NPtr1 = np.sum(which1)
aux = MLmodel.KXC[:, 0:K]
MLmodel.yas0_dict.update({cla: aux[which0, :]})
MLmodel.yas1_dict.update({cla: aux[which1, :]})
aux = MLmodel.KXC[:, K:]
MLmodel.ybs0_dict.update({cla: aux[which0, :]})
MLmodel.ybs1_dict.update({cla: aux[which1, :]})
MLmodel.Bs0_dict.update({cla: np.random.uniform(-10, 10, (NPtr0, K))})
MLmodel.Ds0_dict.update({cla: np.random.uniform(-10, 10, (NPtr0, K))})
MLmodel.Bs1_dict.update({cla: np.random.uniform(-10, 10, (NPtr1, K))})
MLmodel.Ds1_dict.update({cla: np.random.uniform(-10, 10, (NPtr1, K))})
# Send once
#MLmodel.Qs = MLmodel.Bs - MLmodel.Ds # warning, check the sum is nonzero (low prob...)
MLmodel.Qs0_dict.update({cla: MLmodel.Bs0_dict[cla] - MLmodel.Ds0_dict[cla]})
MLmodel.Qs1_dict.update({cla: MLmodel.Bs1_dict[cla] - MLmodel.Ds1_dict[cla]})
#MLmodel.ya_s = MLmodel.Bs - MLmodel.yas
MLmodel.ya_s0_dict.update({cla: MLmodel.Bs0_dict[cla] - MLmodel.yas0_dict[cla]})
MLmodel.ya_s1_dict.update({cla: MLmodel.Bs1_dict[cla] - MLmodel.yas1_dict[cla]})
#MLmodel.yb_s = MLmodel.Ds - MLmodel.ybs
MLmodel.yb_s0_dict.update({cla: MLmodel.Ds0_dict[cla] - MLmodel.ybs0_dict[cla]})
MLmodel.yb_s1_dict.update({cla: MLmodel.Ds1_dict[cla] - MLmodel.ybs1_dict[cla]})
v0_dict = {}
v1_dict = {}
for cla in MLmodel.classes:
xa_ = packet['data']['xaxbP_dict'][cla]['xa_']
xb_ = packet['data']['xaxbP_dict'][cla]['xb_']
P = packet['data']['xaxbP_dict'][cla]['P']
V0 = xa_ * (2 * MLmodel.yas0_dict[cla] - MLmodel.Bs0_dict[cla]) + xb_ * (2 * MLmodel.ybs0_dict[cla] - MLmodel.Ds0_dict[cla]) + P * (MLmodel.Ds0_dict[cla] - 2 * MLmodel.Bs0_dict[cla])
v0 = np.sum(V0, axis=1)
v0_dict.update({cla: v0})
V1 = xa_ * (2 * MLmodel.yas1_dict[cla] - MLmodel.Bs1_dict[cla]) + xb_ * (2 * MLmodel.ybs1_dict[cla] - MLmodel.Ds1_dict[cla]) + P * (MLmodel.Ds1_dict[cla] - 2 * MLmodel.Bs1_dict[cla])
v1 = np.sum(V1, axis=1)
v1_dict.update({cla: v1})
MLmodel.display('PROC_WORKER_END', verbose=False)
# send to Master ya_, yb_, Q, v
action = 'ACK_sending_s'
#message_id = 'worker_' + MLmodel.worker_address + '_' + str(MLmodel.message_counter)
if not MLmodel.Bob_data_s:
data = {'ya_0_dict': MLmodel.ya_s0_dict, 'yb_0_dict': MLmodel.yb_s0_dict, 'Q0_dict': MLmodel.Qs0_dict, 'v0_dict': v0_dict, 'Ztr0_dict': MLmodel.Ztr0_dict, 'Ztr1_dict': MLmodel.Ztr1_dict}
data.update({'ya_1_dict': MLmodel.ya_s1_dict, 'yb_1_dict': MLmodel.yb_s1_dict, 'Q1_dict': MLmodel.Qs1_dict, 'v1_dict': v1_dict})
MLmodel.Bob_data_s = True
else:
data = {'v0_dict': v0_dict, 'v1_dict': v1_dict}
#del v0, v1
packet = {'action': action, 'data': data, 'sender': MLmodel.worker_address}
del data
message_id = 'worker_' + MLmodel.worker_address + '_' + str(MLmodel.message_counter)
packet.update({'message_id': message_id})
MLmodel.message_counter += 1
size_bytes = asizeof.asizeof(dill.dumps(packet))
MLmodel.display('COMMS_WORKER_SEND %s to %s, id = %s, bytes=%s' % (action, MLmodel.master_address, message_id, str(size_bytes)), verbose=False)
MLmodel.comms.send(packet, MLmodel.master_address)
del packet#, size_bytes
MLmodel.display(MLmodel.name + ' %s: sent ACK_sending_s' % (str(MLmodel.worker_address)))
except:
raise
'''
print('ERROR AT while_computing_s')
import code
code.interact(local=locals())
'''
return
def while_computing_KTK(self, MLmodel, packet):
try:
MLmodel.display('PROC_WORKER_START', verbose=False)
KTK_dict = {}
KTy_dict = {}
for cla in MLmodel.classes:
KTK = np.dot(MLmodel.Dmat_dict[cla].T, packet['data']['Rzz_dict'][cla])
KTK = np.dot(KTK, MLmodel.Dmat_dict[cla])
KTK_dict.update({cla: KTK})
KTy = np.dot(MLmodel.Dmat_dict[cla].T, packet['data']['rzt_dict'][cla])
KTy_dict.update({cla: KTy})
MLmodel.display('PROC_WORKER_END', verbose=False)
action = 'ACK_sending_KTK'
data = {'KTK_dict': KTK_dict, 'KTy_dict': KTy_dict}
#del KTK, KTy
packet = {'action': action, 'data': data, 'sender': MLmodel.worker_address}
message_id = 'worker_' + MLmodel.worker_address + '_' + str(MLmodel.message_counter)
packet.update({'message_id': message_id})
MLmodel.message_counter += 1
size_bytes = asizeof.asizeof(dill.dumps(packet))
MLmodel.display('COMMS_WORKER_SEND %s to %s, id = %s, bytes=%s' % (action, MLmodel.master_address, message_id, str(size_bytes)), verbose=False)
#MLmodel.comms.send(MLmodel.master_address, packet)
MLmodel.comms.send(packet, MLmodel.master_address)
MLmodel.display(MLmodel.name + ' %s: sent ACK_sending_KTK' % (str(MLmodel.worker_address)))
except Exception as err:
raise
'''
print('ERROR AT while_computing_KTK')
import code
code.interact(local=locals())
pass
'''
return
states_worker = [
State(name='waiting_order', on_enter=['while_waiting_order']),
State(name='setting_tr_data', on_enter=['while_setting_tr_data']),
State(name='projecting_C', on_enter=['while_projecting_C']),
State(name='storing_C', on_enter=['while_storing_C']),
State(name='computing_s', on_enter=['while_computing_s']),
State(name='computing_KTK', on_enter=['while_computing_KTK']),
State(name='computing_KXC', on_enter=['while_computing_KXC']),
State(name='Exit', on_enter=['while_Exit'])
]
transitions_worker = [
['go_setting_tr_data', 'waiting_order', 'setting_tr_data'],
['done_setting_tr_data', 'setting_tr_data', 'waiting_order'],
['go_projecting_C', 'waiting_order', 'projecting_C'],
['done_projecting_C', 'projecting_C', 'waiting_order'],
['go_storing_C', 'waiting_order', 'storing_C'],
['done_storing_C', 'storing_C', 'waiting_order'],
['go_computing_s', 'waiting_order', 'computing_s'],
['done_computing_s', 'computing_s', 'waiting_order'],
['go_computing_KXC', 'waiting_order', 'computing_KXC'],
['done_computing_KXC', 'computing_KXC', 'waiting_order'],
['go_computing_KTK', 'waiting_order', 'computing_KTK'],
['done_computing_KTK', 'computing_KTK', 'waiting_order'],
['go_exit', 'waiting_order', 'Exit']
]
self.FSMworker = FSM_worker()
self.grafmachine_worker = GraphMachine(model=self.FSMworker,
states=states_worker,
transitions=transitions_worker,
initial='waiting_order',
show_auto_transitions=False, # default value is False
title="Finite State Machine modelling the behaviour of worker No. %s" % str(self.worker_address),
show_conditions=False)
return
def ProcessReceivedPacket_Worker(self, packet, sender):
"""
Take an action after receiving a packet
Parameters
----------
packet: packet object
packet received (usually a dict with various content)
sender: string
id of the sender
"""
self.terminate = False
try:
self.display('COMMS_WORKER_RECEIVED %s from %s, id=%s' % (packet['action'], sender, str(packet['message_id'])), verbose=False)
except:
self.display('WORKER MISSING message_id in %s from %s' % (packet['action'], sender), verbose=False)
pass
if packet is not None:
try:
# Exit the process
if packet['action'] == 'STOP':
self.display(self.name + ' %s: terminated by Master' % (str(self.worker_address)))
self.display('EXIT_WORKER')
self.terminate = True
if packet['action'] == 'update_tr_data':
# We update the training data
self.FSMworker.go_setting_tr_data(self, packet)
self.FSMworker.done_setting_tr_data(self)
if packet['action'] == 'compute_KTK':
self.FSMworker.go_computing_KTK(self, packet)
self.FSMworker.done_computing_KTK(self)
if packet['action'] == 'selecting_C':
#self.C = packet['data']['C']
self.FSMworker.go_projecting_C(self, packet)
self.FSMworker.done_projecting_C(self)
if packet['action'] == 'sending_C':
#self.C = packet['data']['C']
self.FSMworker.go_storing_C(self, packet)
self.FSMworker.done_storing_C(self)
if packet['action'] == 'sending_xaxbP':
self.FSMworker.go_computing_s(self, packet)
self.FSMworker.done_computing_s(self)
if packet['action'] == 'sending_Rzz_rzt':
self.FSMworker.go_computing_KTK(self, packet)
self.FSMworker.done_computing_KTK(self)
except Exception as err:
raise
'''
print('ERROR AT CheckNewPacket_worker')
import code
code.interact(local=locals())
'''
return self.terminate
| [
"pickle.dump",
"numpy.random.seed",
"numpy.sum",
"numpy.argmax",
"numpy.ones",
"numpy.linalg.norm",
"numpy.exp",
"numpy.random.normal",
"transitions.extensions.GraphMachine",
"numpy.multiply",
"numpy.max",
"transitions.State",
"dill.dumps",
"numpy.min",
"numpy.linalg.inv",
"numpy.dot",... | [((770, 781), 'time.time', 'time.time', ([], {}), '()\n', (779, 781), False, 'import time\n'), ((833, 858), 'numpy.random.seed', 'np.random.seed', ([], {'seed': 'seed'}), '(seed=seed)\n', (847, 858), True, 'import numpy as np\n'), ((1410, 1446), 'numpy.exp', 'np.exp', (['(-XC2 / 2.0 / self.sigma ** 2)'], {}), '(-XC2 / 2.0 / self.sigma ** 2)\n', (1416, 1446), True, 'import numpy as np\n'), ((1876, 1887), 'numpy.array', 'np.array', (['O'], {}), '(O)\n', (1884, 1887), True, 'import numpy as np\n'), ((2579, 2615), 'numpy.exp', 'np.exp', (['(-XC2 / 2.0 / self.sigma ** 2)'], {}), '(-XC2 / 2.0 / self.sigma ** 2)\n', (2585, 2615), True, 'import numpy as np\n'), ((3045, 3056), 'numpy.array', 'np.array', (['O'], {}), '(O)\n', (3053, 3056), True, 'import numpy as np\n'), ((9574, 9585), 'time.time', 'time.time', ([], {}), '()\n', (9583, 9585), False, 'import time\n'), ((9637, 9662), 'numpy.random.seed', 'np.random.seed', ([], {'seed': 'seed'}), '(seed=seed)\n', (9651, 9662), True, 'import numpy as np\n'), ((28916, 29168), 'transitions.extensions.GraphMachine', 'GraphMachine', ([], {'model': 'self.FSMmaster', 'states': 'states_master', 'transitions': 'transitions_master', 'initial': '"""waiting_order"""', 'show_auto_transitions': '(False)', 'title': '"""Finite State Machine modelling the behaviour of the master"""', 'show_conditions': '(False)'}), "(model=self.FSMmaster, states=states_master, transitions=\n transitions_master, initial='waiting_order', show_auto_transitions=\n False, title=\n 'Finite State Machine modelling the behaviour of the master',\n show_conditions=False)\n", (28928, 29168), False, 'from transitions.extensions import GraphMachine\n'), ((29534, 29578), 'numpy.random.normal', 'np.random.normal', (['(0)', '(0.001)', '(self.NI + 1, 1)'], {}), '(0, 0.001, (self.NI + 1, 1))\n', (29550, 29578), True, 'import numpy as np\n'), ((29650, 29692), 'numpy.random.normal', 'np.random.normal', (['(0)', '(1.0)', '(self.NI + 1, 1)'], {}), '(0, 1.0, (self.NI + 1, 1))\n', (29666, 29692), True, 'import numpy as np\n'), ((29721, 29757), 'numpy.zeros', 'np.zeros', (['(self.NI + 1, self.NI + 1)'], {}), '((self.NI + 1, self.NI + 1))\n', (29729, 29757), True, 'import numpy as np\n'), ((29817, 29843), 'numpy.zeros', 'np.zeros', (['(self.NI + 1, 1)'], {}), '((self.NI + 1, 1))\n', (29825, 29843), True, 'import numpy as np\n'), ((32150, 32192), 'numpy.exp', 'np.exp', (['(-XC2 / 2.0 / self.model.sigma ** 2)'], {}), '(-XC2 / 2.0 / self.model.sigma ** 2)\n', (32156, 32192), True, 'import numpy as np\n'), ((32216, 32252), 'numpy.zeros', 'np.zeros', (['(self.NC + 1, self.NC + 1)'], {}), '((self.NC + 1, self.NC + 1))\n', (32224, 32252), True, 'import numpy as np\n'), ((45284, 45295), 'time.time', 'time.time', ([], {}), '()\n', (45293, 45295), False, 'import time\n'), ((45347, 45372), 'numpy.random.seed', 'np.random.seed', ([], {'seed': 'seed'}), '(seed=seed)\n', (45361, 45372), True, 'import numpy as np\n'), ((1210, 1229), 'numpy.dot', 'np.dot', (['X', 'self.C.T'], {}), '(X, self.C.T)\n', (1216, 1229), True, 'import numpy as np\n'), ((1912, 1932), 'numpy.argmax', 'np.argmax', (['O'], {'axis': '(0)'}), '(O, axis=0)\n', (1921, 1932), True, 'import numpy as np\n'), ((2379, 2398), 'numpy.dot', 'np.dot', (['X', 'self.C.T'], {}), '(X, self.C.T)\n', (2385, 2398), True, 'import numpy as np\n'), ((3081, 3101), 'numpy.argmax', 'np.argmax', (['O'], {'axis': '(0)'}), '(O, axis=0)\n', (3090, 3101), True, 'import numpy as np\n'), ((8585, 8601), 'numpy.sqrt', 'np.sqrt', (['self.NI'], {}), '(self.NI)\n', (8592, 8601), True, 'import numpy as np\n'), ((9946, 10007), 'transitions.State', 'State', ([], {'name': '"""waiting_order"""', 'on_enter': "['while_waiting_order']"}), "(name='waiting_order', on_enter=['while_waiting_order'])\n", (9951, 10007), False, 'from transitions import State\n'), ((10022, 10085), 'transitions.State', 'State', ([], {'name': '"""update_tr_data"""', 'on_enter': "['while_update_tr_data']"}), "(name='update_tr_data', on_enter=['while_update_tr_data'])\n", (10027, 10085), False, 'from transitions import State\n'), ((10100, 10157), 'transitions.State', 'State', ([], {'name': '"""getting_KTK"""', 'on_enter': "['while_getting_KTK']"}), "(name='getting_KTK', on_enter=['while_getting_KTK'])\n", (10105, 10157), False, 'from transitions import State\n'), ((10172, 10229), 'transitions.State', 'State', ([], {'name': '"""selecting_C"""', 'on_enter': "['while_selecting_C']"}), "(name='selecting_C', on_enter=['while_selecting_C'])\n", (10177, 10229), False, 'from transitions import State\n'), ((10244, 10297), 'transitions.State', 'State', ([], {'name': '"""sending_C"""', 'on_enter': "['while_sending_C']"}), "(name='sending_C', on_enter=['while_sending_C'])\n", (10249, 10297), False, 'from transitions import State\n'), ((10314, 10375), 'transitions.State', 'State', ([], {'name': '"""computing_XTw"""', 'on_enter': "['while_computing_XTw']"}), "(name='computing_XTw', on_enter=['while_computing_XTw'])\n", (10319, 10375), False, 'from transitions import State\n'), ((10390, 10449), 'transitions.State', 'State', ([], {'name': '"""computing_oi"""', 'on_enter': "['while_computing_oi']"}), "(name='computing_oi', on_enter=['while_computing_oi'])\n", (10395, 10449), False, 'from transitions import State\n'), ((10466, 10521), 'transitions.State', 'State', ([], {'name': '"""updating_w"""', 'on_enter': "['while_updating_w']"}), "(name='updating_w', on_enter=['while_updating_w'])\n", (10471, 10521), False, 'from transitions import State\n'), ((31943, 31968), 'numpy.dot', 'np.dot', (['X', 'self.model.C.T'], {}), '(X, self.model.C.T)\n', (31949, 31968), True, 'import numpy as np\n'), ((33270, 33312), 'numpy.exp', 'np.exp', (['(-XC2 / 2.0 / self.model.sigma ** 2)'], {}), '(-XC2 / 2.0 / self.model.sigma ** 2)\n', (33276, 33312), True, 'import numpy as np\n'), ((62395, 62456), 'transitions.State', 'State', ([], {'name': '"""waiting_order"""', 'on_enter': "['while_waiting_order']"}), "(name='waiting_order', on_enter=['while_waiting_order'])\n", (62400, 62456), False, 'from transitions import State\n'), ((62471, 62536), 'transitions.State', 'State', ([], {'name': '"""setting_tr_data"""', 'on_enter': "['while_setting_tr_data']"}), "(name='setting_tr_data', on_enter=['while_setting_tr_data'])\n", (62476, 62536), False, 'from transitions import State\n'), ((62551, 62610), 'transitions.State', 'State', ([], {'name': '"""projecting_C"""', 'on_enter': "['while_projecting_C']"}), "(name='projecting_C', on_enter=['while_projecting_C'])\n", (62556, 62610), False, 'from transitions import State\n'), ((62625, 62678), 'transitions.State', 'State', ([], {'name': '"""storing_C"""', 'on_enter': "['while_storing_C']"}), "(name='storing_C', on_enter=['while_storing_C'])\n", (62630, 62678), False, 'from transitions import State\n'), ((62695, 62752), 'transitions.State', 'State', ([], {'name': '"""computing_s"""', 'on_enter': "['while_computing_s']"}), "(name='computing_s', on_enter=['while_computing_s'])\n", (62700, 62752), False, 'from transitions import State\n'), ((62769, 62830), 'transitions.State', 'State', ([], {'name': '"""computing_KTK"""', 'on_enter': "['while_computing_KTK']"}), "(name='computing_KTK', on_enter=['while_computing_KTK'])\n", (62774, 62830), False, 'from transitions import State\n'), ((62845, 62906), 'transitions.State', 'State', ([], {'name': '"""computing_KXC"""', 'on_enter': "['while_computing_KXC']"}), "(name='computing_KXC', on_enter=['while_computing_KXC'])\n", (62850, 62906), False, 'from transitions import State\n'), ((62921, 62964), 'transitions.State', 'State', ([], {'name': '"""Exit"""', 'on_enter': "['while_Exit']"}), "(name='Exit', on_enter=['while_Exit'])\n", (62926, 62964), False, 'from transitions import State\n'), ((1605, 1621), 'numpy.ones', 'np.ones', (['(NP, 1)'], {}), '((NP, 1))\n', (1612, 1621), True, 'import numpy as np\n'), ((1947, 1995), 'numpy.array', 'np.array', (['[self.classes[pos] for pos in winners]'], {}), '([self.classes[pos] for pos in winners])\n', (1955, 1995), True, 'import numpy as np\n'), ((2774, 2790), 'numpy.ones', 'np.ones', (['(NP, 1)'], {}), '((NP, 1))\n', (2781, 2790), True, 'import numpy as np\n'), ((3116, 3164), 'numpy.array', 'np.array', (['[self.classes[pos] for pos in winners]'], {}), '([self.classes[pos] for pos in winners])\n', (3124, 3164), True, 'import numpy as np\n'), ((33025, 33052), 'numpy.dot', 'np.dot', (['self.Xval', 'self.C.T'], {}), '(self.Xval, self.C.T)\n', (33031, 33052), True, 'import numpy as np\n'), ((36259, 36270), 'numpy.array', 'np.array', (['O'], {}), '(O)\n', (36267, 36270), True, 'import numpy as np\n'), ((1253, 1270), 'numpy.multiply', 'np.multiply', (['X', 'X'], {}), '(X, X)\n', (1264, 1270), True, 'import numpy as np\n'), ((1320, 1347), 'numpy.multiply', 'np.multiply', (['self.C', 'self.C'], {}), '(self.C, self.C)\n', (1331, 1347), True, 'import numpy as np\n'), ((1757, 1786), 'numpy.dot', 'np.dot', (['KXC', 'self.w_dict[cla]'], {}), '(KXC, self.w_dict[cla])\n', (1763, 1786), True, 'import numpy as np\n'), ((2422, 2439), 'numpy.multiply', 'np.multiply', (['X', 'X'], {}), '(X, X)\n', (2433, 2439), True, 'import numpy as np\n'), ((2489, 2516), 'numpy.multiply', 'np.multiply', (['self.C', 'self.C'], {}), '(self.C, self.C)\n', (2500, 2516), True, 'import numpy as np\n'), ((2926, 2955), 'numpy.dot', 'np.dot', (['KXC', 'self.w_dict[cla]'], {}), '(KXC, self.w_dict[cla])\n', (2932, 2955), True, 'import numpy as np\n'), ((31992, 32009), 'numpy.multiply', 'np.multiply', (['X', 'X'], {}), '(X, X)\n', (32003, 32009), True, 'import numpy as np\n'), ((32064, 32103), 'numpy.multiply', 'np.multiply', (['self.model.C', 'self.model.C'], {}), '(self.model.C, self.model.C)\n', (32075, 32103), True, 'import numpy as np\n'), ((32685, 32707), 'numpy.zeros', 'np.zeros', (['(self.NI, 1)'], {}), '((self.NI, 1))\n', (32693, 32707), True, 'import numpy as np\n'), ((32755, 32781), 'numpy.zeros', 'np.zeros', (['(1, self.NI + 1)'], {}), '((1, self.NI + 1))\n', (32763, 32781), True, 'import numpy as np\n'), ((33356, 33388), 'numpy.ones', 'np.ones', (['(self.Xval.shape[0], 1)'], {}), '((self.Xval.shape[0], 1))\n', (33363, 33388), True, 'import numpy as np\n'), ((34152, 34197), 'numpy.random.normal', 'np.random.normal', (['(0)', '(0.001)', '(self.NItrain, 1)'], {}), '(0, 0.001, (self.NItrain, 1))\n', (34168, 34197), True, 'import numpy as np\n'), ((34242, 34287), 'numpy.random.normal', 'np.random.normal', (['(0)', '(0.001)', '(self.NItrain, 1)'], {}), '(0, 0.001, (self.NItrain, 1))\n', (34258, 34287), True, 'import numpy as np\n'), ((36303, 36323), 'numpy.argmax', 'np.argmax', (['O'], {'axis': '(0)'}), '(O, axis=0)\n', (36312, 36323), True, 'import numpy as np\n'), ((48412, 48451), 'numpy.exp', 'np.exp', (['(-XC2 / 2.0 / MLmodel.sigma ** 2)'], {}), '(-XC2 / 2.0 / MLmodel.sigma ** 2)\n', (48418, 48451), True, 'import numpy as np\n'), ((48635, 48652), 'numpy.zeros', 'np.zeros', (['(NC, 1)'], {}), '((NC, 1))\n', (48643, 48652), True, 'import numpy as np\n'), ((51024, 51063), 'numpy.exp', 'np.exp', (['(-XC2 / 2.0 / MLmodel.sigma ** 2)'], {}), '(-XC2 / 2.0 / MLmodel.sigma ** 2)\n', (51030, 51063), True, 'import numpy as np\n'), ((12290, 12308), 'dill.dumps', 'dill.dumps', (['packet'], {}), '(packet)\n', (12300, 12308), False, 'import dill\n'), ((13660, 13678), 'dill.dumps', 'dill.dumps', (['packet'], {}), '(packet)\n', (13670, 13678), False, 'import dill\n'), ((15310, 15328), 'dill.dumps', 'dill.dumps', (['packet'], {}), '(packet)\n', (15320, 15328), False, 'import dill\n'), ((18683, 18701), 'dill.dumps', 'dill.dumps', (['packet'], {}), '(packet)\n', (18693, 18701), False, 'import dill\n'), ((26299, 26317), 'dill.dumps', 'dill.dumps', (['packet'], {}), '(packet)\n', (26309, 26317), False, 'import dill\n'), ((27720, 27764), 'numpy.zeros', 'np.zeros', (['(MLmodel.NItrain, MLmodel.NItrain)'], {}), '((MLmodel.NItrain, MLmodel.NItrain))\n', (27728, 27764), True, 'import numpy as np\n'), ((27810, 27840), 'numpy.zeros', 'np.zeros', (['(MLmodel.NItrain, 1)'], {}), '((MLmodel.NItrain, 1))\n', (27818, 27840), True, 'import numpy as np\n'), ((33080, 33113), 'numpy.multiply', 'np.multiply', (['self.Xval', 'self.Xval'], {}), '(self.Xval, self.Xval)\n', (33091, 33113), True, 'import numpy as np\n'), ((33167, 33194), 'numpy.multiply', 'np.multiply', (['self.C', 'self.C'], {}), '(self.C, self.C)\n', (33178, 33194), True, 'import numpy as np\n'), ((35370, 35431), 'numpy.linalg.norm', 'np.linalg.norm', (['(self.model.w_dict[cla] - self.w_old_dict[cla])'], {}), '(self.model.w_dict[cla] - self.w_old_dict[cla])\n', (35384, 35431), True, 'import numpy as np\n'), ((35434, 35470), 'numpy.linalg.norm', 'np.linalg.norm', (['self.w_old_dict[cla]'], {}), '(self.w_old_dict[cla])\n', (35448, 35470), True, 'import numpy as np\n'), ((36354, 36402), 'numpy.array', 'np.array', (['[self.classes[pos] for pos in winners]'], {}), '([self.classes[pos] for pos in winners])\n', (36362, 36402), True, 'import numpy as np\n'), ((46721, 46739), 'dill.dumps', 'dill.dumps', (['packet'], {}), '(packet)\n', (46731, 46739), False, 'import dill\n'), ((48159, 48181), 'numpy.dot', 'np.dot', (['X', 'MLmodel.C.T'], {}), '(X, MLmodel.C.T)\n', (48165, 48181), True, 'import numpy as np\n'), ((48582, 48604), 'numpy.argmax', 'np.argmax', (['KXC'], {'axis': '(1)'}), '(KXC, axis=1)\n', (48591, 48604), True, 'import numpy as np\n'), ((49359, 49377), 'dill.dumps', 'dill.dumps', (['packet'], {}), '(packet)\n', (49369, 49377), False, 'import dill\n'), ((50771, 50793), 'numpy.dot', 'np.dot', (['X', 'MLmodel.C.T'], {}), '(X, MLmodel.C.T)\n', (50777, 50793), True, 'import numpy as np\n'), ((51300, 51319), 'numpy.min', 'np.min', (['MLmodel.KXC'], {}), '(MLmodel.KXC)\n', (51306, 51319), True, 'import numpy as np\n'), ((51321, 51340), 'numpy.max', 'np.max', (['MLmodel.KXC'], {}), '(MLmodel.KXC)\n', (51327, 51340), True, 'import numpy as np\n'), ((53190, 53208), 'dill.dumps', 'dill.dumps', (['packet'], {}), '(packet)\n', (53200, 53208), False, 'import dill\n'), ((57952, 57970), 'numpy.sum', 'np.sum', (['V0'], {'axis': '(1)'}), '(V0, axis=1)\n', (57958, 57970), True, 'import numpy as np\n'), ((58260, 58278), 'numpy.sum', 'np.sum', (['V1'], {'axis': '(1)'}), '(V1, axis=1)\n', (58266, 58278), True, 'import numpy as np\n'), ((59611, 59629), 'dill.dumps', 'dill.dumps', (['packet'], {}), '(packet)\n', (59621, 59629), False, 'import dill\n'), ((60608, 60673), 'numpy.dot', 'np.dot', (['MLmodel.Dmat_dict[cla].T', "packet['data']['Rzz_dict'][cla]"], {}), "(MLmodel.Dmat_dict[cla].T, packet['data']['Rzz_dict'][cla])\n", (60614, 60673), True, 'import numpy as np\n'), ((60707, 60742), 'numpy.dot', 'np.dot', (['KTK', 'MLmodel.Dmat_dict[cla]'], {}), '(KTK, MLmodel.Dmat_dict[cla])\n', (60713, 60742), True, 'import numpy as np\n'), ((60828, 60893), 'numpy.dot', 'np.dot', (['MLmodel.Dmat_dict[cla].T', "packet['data']['rzt_dict'][cla]"], {}), "(MLmodel.Dmat_dict[cla].T, packet['data']['rzt_dict'][cla])\n", (60834, 60893), True, 'import numpy as np\n'), ((61585, 61603), 'dill.dumps', 'dill.dumps', (['packet'], {}), '(packet)\n', (61595, 61603), False, 'import dill\n'), ((20901, 20919), 'numpy.sum', 'np.sum', (['U0'], {'axis': '(1)'}), '(U0, axis=1)\n', (20907, 20919), True, 'import numpy as np\n'), ((21336, 21350), 'numpy.ones', 'np.ones', (['NPtr0'], {}), '(NPtr0)\n', (21343, 21350), True, 'import numpy as np\n'), ((21966, 21984), 'numpy.dot', 'np.dot', (['Rzz0.T', 'y0'], {}), '(Rzz0.T, y0)\n', (21972, 21984), True, 'import numpy as np\n'), ((22021, 22077), 'numpy.dot', 'np.dot', (['Rzz0.T', "MLmodel.Ztr_dict[addr]['Ztr0_dict'][cla]"], {}), "(Rzz0.T, MLmodel.Ztr_dict[addr]['Ztr0_dict'][cla])\n", (22027, 22077), True, 'import numpy as np\n'), ((22743, 22761), 'numpy.sum', 'np.sum', (['U1'], {'axis': '(1)'}), '(U1, axis=1)\n', (22749, 22761), True, 'import numpy as np\n'), ((22998, 23012), 'numpy.ones', 'np.ones', (['NPtr1'], {}), '(NPtr1)\n', (23005, 23012), True, 'import numpy as np\n'), ((23177, 23191), 'numpy.ones', 'np.ones', (['NPtr1'], {}), '(NPtr1)\n', (23184, 23191), True, 'import numpy as np\n'), ((23807, 23825), 'numpy.dot', 'np.dot', (['Rzz1.T', 'y1'], {}), '(Rzz1.T, y1)\n', (23813, 23825), True, 'import numpy as np\n'), ((23862, 23918), 'numpy.dot', 'np.dot', (['Rzz1.T', "MLmodel.Ztr_dict[addr]['Ztr1_dict'][cla]"], {}), "(Rzz1.T, MLmodel.Ztr_dict[addr]['Ztr1_dict'][cla])\n", (23868, 23918), True, 'import numpy as np\n'), ((24920, 24938), 'dill.dumps', 'dill.dumps', (['packet'], {}), '(packet)\n', (24930, 24938), False, 'import dill\n'), ((28277, 28323), 'numpy.linalg.inv', 'np.linalg.inv', (['(MLmodel.KTK_accum + MLmodel.Kcc)'], {}), '(MLmodel.KTK_accum + MLmodel.Kcc)\n', (28290, 28323), True, 'import numpy as np\n'), ((36166, 36190), 'numpy.dot', 'np.dot', (['self.KXC_val', 'w_'], {}), '(self.KXC_val, w_)\n', (36172, 36190), True, 'import numpy as np\n'), ((51220, 51236), 'numpy.ones', 'np.ones', (['(NP, 1)'], {}), '((NP, 1))\n', (51227, 51236), True, 'import numpy as np\n'), ((55417, 55431), 'numpy.sum', 'np.sum', (['which0'], {}), '(which0)\n', (55423, 55431), True, 'import numpy as np\n'), ((55526, 55540), 'numpy.sum', 'np.sum', (['which1'], {}), '(which1)\n', (55532, 55540), True, 'import numpy as np\n'), ((4908, 4928), 'pickle.dump', 'pickle.dump', (['self', 'f'], {}), '(self, f)\n', (4919, 4928), False, 'import pickle\n'), ((21157, 21171), 'numpy.ones', 'np.ones', (['NPtr0'], {}), '(NPtr0)\n', (21164, 21171), True, 'import numpy as np\n'), ((48217, 48234), 'numpy.multiply', 'np.multiply', (['X', 'X'], {}), '(X, X)\n', (48228, 48234), True, 'import numpy as np\n'), ((48296, 48329), 'numpy.multiply', 'np.multiply', (['MLmodel.C', 'MLmodel.C'], {}), '(MLmodel.C, MLmodel.C)\n', (48307, 48329), True, 'import numpy as np\n'), ((50829, 50846), 'numpy.multiply', 'np.multiply', (['X', 'X'], {}), '(X, X)\n', (50840, 50846), True, 'import numpy as np\n'), ((50908, 50941), 'numpy.multiply', 'np.multiply', (['MLmodel.C', 'MLmodel.C'], {}), '(MLmodel.C, MLmodel.C)\n', (50919, 50941), True, 'import numpy as np\n'), ((51583, 51627), 'numpy.random.normal', 'np.random.normal', (['(0)', '(0.01)', '(MLmodel.NPtr, 1)'], {}), '(0, 0.01, (MLmodel.NPtr, 1))\n', (51599, 51627), True, 'import numpy as np\n'), ((52068, 52128), 'numpy.random.normal', 'np.random.normal', (['(0)', '(1)', '(MLmodel.NI_train, MLmodel.NI_train)'], {}), '(0, 1, (MLmodel.NI_train, MLmodel.NI_train))\n', (52084, 52128), True, 'import numpy as np\n'), ((52187, 52224), 'numpy.linalg.inv', 'np.linalg.inv', (['MLmodel.Cmat_dict[cla]'], {}), '(MLmodel.Cmat_dict[cla])\n', (52200, 52224), True, 'import numpy as np\n'), ((52427, 52481), 'numpy.dot', 'np.dot', (['MLmodel.KXC[which0, :]', 'MLmodel.Cmat_dict[cla]'], {}), '(MLmodel.KXC[which0, :], MLmodel.Cmat_dict[cla])\n', (52433, 52481), True, 'import numpy as np\n'), ((52593, 52647), 'numpy.dot', 'np.dot', (['MLmodel.KXC[which1, :]', 'MLmodel.Cmat_dict[cla]'], {}), '(MLmodel.KXC[which1, :], MLmodel.Cmat_dict[cla])\n', (52599, 52647), True, 'import numpy as np\n'), ((56019, 56057), 'numpy.random.uniform', 'np.random.uniform', (['(-10)', '(10)', '(NPtr0, K)'], {}), '(-10, 10, (NPtr0, K))\n', (56036, 56057), True, 'import numpy as np\n'), ((56120, 56158), 'numpy.random.uniform', 'np.random.uniform', (['(-10)', '(10)', '(NPtr0, K)'], {}), '(-10, 10, (NPtr0, K))\n', (56137, 56158), True, 'import numpy as np\n'), ((56221, 56259), 'numpy.random.uniform', 'np.random.uniform', (['(-10)', '(10)', '(NPtr1, K)'], {}), '(-10, 10, (NPtr1, K))\n', (56238, 56259), True, 'import numpy as np\n'), ((56322, 56360), 'numpy.random.uniform', 'np.random.uniform', (['(-10)', '(10)', '(NPtr1, K)'], {}), '(-10, 10, (NPtr1, K))\n', (56339, 56360), True, 'import numpy as np\n'), ((16938, 16967), 'numpy.random.uniform', 'np.random.uniform', (['(-10)', '(10)', 'K'], {}), '(-10, 10, K)\n', (16955, 16967), True, 'import numpy as np\n'), ((17033, 17062), 'numpy.random.uniform', 'np.random.uniform', (['(-10)', '(10)', 'K'], {}), '(-10, 10, K)\n', (17050, 17062), True, 'import numpy as np\n'), ((52258, 52286), 'numpy.array', 'np.array', (['(MLmodel.ytr == cla)'], {}), '(MLmodel.ytr == cla)\n', (52266, 52286), True, 'import numpy as np\n'), ((55263, 55291), 'numpy.array', 'np.array', (['(MLmodel.ytr == cla)'], {}), '(MLmodel.ytr == cla)\n', (55271, 55291), True, 'import numpy as np\n')] |
import pytest
import os
import pandas as pd
import riptable as rt
from enum import IntEnum
from numpy.testing import assert_array_equal
from riptable import *
from riptable import save_sds, load_sds
from riptable import FastArray, Categorical, CatZero
from riptable.rt_categorical import Categories
from riptable.rt_enum import (
INVALID_DICT,
)
from riptable.rt_enum import (
DisplayLength,
DisplayJustification,
DisplayColumnColors,
)
from riptable.rt_enum import CategoryMode, TypeRegister
from riptable.rt_numpy import isnan, isnotnan, arange, ones
from riptable.tests.test_utils import (
get_categorical_data_factory_method,
get_all_categorical_data,
)
from riptable.rt_sds import SDSMakeDirsOn
from riptable.tests.utils import LikertDecision
# change to true since we write into /tests directory
SDSMakeDirsOn()
three_unicode = np.array(["AAPL\u2080", "AMZN\u2082", "IBM\u2081"])
three_bytes = FastArray([b'a', b'b', b'c'])
three_ints = FastArray([1, 2, 3])
compare_func_names = ['__ne__', '__eq__', '__ge__', '__gt__', '__le__', '__lt__']
int_success = [
np.array([True, False, True]), # ne
np.array([False, True, False]), # eq
np.array([False, True, True]), # ge
np.array([False, False, True]), # gt
np.array([True, True, False]), # le
np.array([True, False, False]), # lt
]
same_success = [
np.array([False, False, False]), # ne
np.array([True, True, True]), # eq
np.array([True, True, True]), # ge
np.array([False, False, False]), # gt
np.array([True, True, True]), # le
np.array([False, False, False]), # lt
]
diff_success = [
np.array([True, False, True]), # ne
np.array([False, True, False]), # eq
np.array([False, True, False]), # ge
np.array([False, False, False]), # gt
np.array([True, True, True]), # le
np.array([True, False, True]), # lt
]
ShowCompareInfo = False
list_bytes = [b'b', b'b', b'a', b'd', b'c']
list_unicode = ['b', 'b', 'a', 'd', 'c']
list_true_unicode = [u'b\u2082', u'b\u2082', u'a\u2082', u'd\u2082', u'c\u2082']
decision_dict = dict(zip(LikertDecision.__members__.keys(), [int(v) for v in LikertDecision.__members__.values()],))
def array_equal(arr1, arr2):
subr = arr1 - arr2
sumr = sum(subr == 0)
result = sumr == len(arr1)
if not result:
print("array comparison failed", arr1, arr2)
return result
class TestCategorical:
def _notimpl(self):
pytest.skip("This test needs to be implemented.")
def test_constructor(self):
# from pandas categorical
# from single parameter
# from two parameters
# ndarray
# python list
self._notimpl()
def test_ctor_list(self):
c_bytes = Categorical(list_bytes)
assert c_bytes.dtype == np.int8, f"Dtype {c_bytes.dtype} was not correct for construction from small list."
assert len(c_bytes) == 5, f"Length of underlying index array was incorrect for construction from bytes."
unique_bytes = np.unique(list_bytes)
assert np.all(
c_bytes._categories_wrap._list == unique_bytes
), f"Categories did not generate a unique list of categories from input bytes list."
c_unicode = Categorical(list_unicode)
assert c_unicode.dtype == np.int8, f"Dtype {c_unicode.dtype} was not correct for construction from small list."
assert len(c_unicode) == 5, f"Length of underlying index array was incorrect for construction from unicode."
assert (
len(c_unicode._categories_wrap) == 4
), f"Length of unique categories was incorrect for construction from unicode."
assert (
c_unicode._categories_wrap._list[0] == b'a'
), f"Unique categories were not sorted for construction from unicode."
assert c_unicode._categories_wrap._list.dtype.char == 'S', f"Unicode strings were not flipped to byte strings."
c_true_unicode = Categorical(list_true_unicode)
assert (
c_true_unicode.dtype == np.int8
), f"Dtype {c_true_unicode.dtype} was not correct for construction from small list."
assert (
len(c_true_unicode) == 5
), f"Length of underlying index array was incorrect for construction from true unicode."
assert (
len(c_true_unicode._categories_wrap) == 4
), f"Length of unique categories was incorrect for construction from true unicode."
assert (
c_true_unicode._categories_wrap._list[0] == u'a\u2082'
), f"Unique categories were not sorted for construction from true unicode."
assert (
c_true_unicode._categories_wrap._list.dtype.char == 'U'
), f"Unicode strings were not flipped to byte strings."
def test_ctor_nparray(self):
c_bytes = Categorical(np.array(list_bytes))
assert c_bytes.dtype == np.int8, f"Dtype {c_bytes.dtype} was not correct for construction from small list."
assert len(c_bytes) == 5, f"Length of underlying index array was incorrect for construction from bytes."
unique_bytes = np.unique(list_bytes)
assert np.all(
c_bytes._categories_wrap._list == unique_bytes
), f"Categories did not generate a unique list of categories from input bytes list."
c_unicode = Categorical(np.array(list_unicode))
assert c_unicode.dtype == np.int8, f"Dtype {c_unicode.dtype} was not correct for construction from small list."
assert len(c_unicode) == 5, f"Length of underlying index array was incorrect for construction from unicode."
assert (
len(c_unicode._categories_wrap._list) == 4
), f"Length of unique categories was incorrect for construction from unicode."
assert (
c_unicode._categories_wrap._list[0] == b'a'
), f"Unique categories were not sorted for construction from unicode."
assert c_unicode._categories_wrap._list.dtype.char == 'S', f"Unicode strings were not flipped to byte strings."
c_true_unicode = Categorical(np.array(list_true_unicode))
assert (
c_true_unicode.dtype == np.int8
), f"Dtype {c_true_unicode.dtype} was not correct for construction from small list."
assert (
len(c_true_unicode) == 5
), f"Length of underlying index array was incorrect for construction from true unicode."
assert (
len(c_true_unicode._categories_wrap._list) == 4
), f"Length of unique categories was incorrect for construction from true unicode."
assert (
c_true_unicode._categories_wrap._list[0] == u'a\u2082'
), f"Unique categories were not sorted for construction from true unicode."
assert (
c_true_unicode._categories_wrap._list.dtype.char == 'U'
), f"Unicode strings were not flipped to byte strings."
def test_ctor_values_and_cats(self):
v_bytes = [b'IBM', b'AAPL', b'AMZN', b'IBM', b'hello']
v_str = ['IBM', 'AAPL', 'AMZN', 'IBM', 'hello']
v_true = [
u'IBM\u2082',
u'AAPL\u2082',
u'AMZN\u2082',
u'IBM\u2082',
u'hello\u2082',
]
c_bytes = [b'AAPL', b'AMZN', b'IBM']
c_str = ['AAPL', 'AMZN', 'IBM']
c_true = [u'AAPL\u2082', u'AMZN\u2082', u'IBM\u2082']
v_correct = [2, 0, 1, 2, 3]
c_correct = [b'AAPL', b'AMZN', b'IBM', b'inv']
valid_v = [
v_bytes,
v_str,
np.array(v_bytes),
np.array(v_str),
FastArray(v_bytes),
FastArray(v_str),
]
valid_c = [
c_bytes,
c_str,
np.array(c_bytes),
np.array(c_str),
FastArray(c_bytes),
FastArray(c_str),
]
for v in valid_v:
vdt = None
if hasattr(v, 'dtype'):
vdt = v.dtype
else:
vdt = type(v)
for c in valid_c:
cdt = None
if hasattr(c, 'dtype'):
cdt = c.dtype
else:
cdt = type(c)
# error if no invalid provided
with pytest.raises(ValueError):
cat = Categorical(v, c)
# accept invalid and correctly assign
# cat = Categorical(v, c, invalid_category=b'inv')
# self.assertEqual(cat._categories.dtype.char, 'S', msg=f"Categorical from v: {vdt} and c: {cdt} did not flip categories to bytestring")
# v_is_correct = bool(np.all(v_correct == cat.view(FastArray)))
# self.assertTrue(v_is_correct, msg=f"Did not create the correct underlying index array from v: {vdt} and c: {cdt}")
# c_is_correct = bool(np.all(c_correct == cat._categories))
# self.assertTrue(c_is_correct, msg=f"Did not create the correct categories from v: {vdt} and c: {cdt}")
# v = v_true
# vdt = "TRUE unicode"
# for c in valid_c:
# if hasattr(c,'dtype'):
# cdt = c.dtype
# else:
# cdt = type(c)
# cat = Categorical(v,c)
# ---------------------------------------------------------------------------
def test_ctor_bad_index(self):
idx_list = [1, 2, 3, 4, 5]
str_list = ['a', 'b']
with pytest.raises(ValueError):
c = Categorical(idx_list, str_list)
# ---------------------------------------------------------------------------
def test_ctor_non_unique(self):
'''
riptable categoricals, like pandas categoricals, do not allow a non-unique list of categories when an index array is provided.
'''
idx_list = [0, 1]
str_list = ['b', 'b', 'a']
c = Categorical(idx_list, str_list)
# ---------------------------------------------------------------------------
def test_ctor_enum(self):
codes = [1, 44, 44, 133, 75]
c = Categorical(codes, LikertDecision)
# ---------------------------------------------------------------------------
def test_compare_enum_int(self):
compare_func_names = [
'__ne__',
'__eq__',
'__ge__',
'__gt__',
'__le__',
'__lt__',
]
codes = [1, 44, 44, 133, 75]
valid_idx = 44
bad_idx = 43
valid_idx_correct = [
FastArray([True, False, False, True, True]),
FastArray([False, True, True, False, False]),
FastArray([False, True, True, True, True]),
FastArray([False, False, False, True, True]),
FastArray([True, True, True, False, False]),
FastArray([True, False, False, False, False]),
]
bad_idx_correct = [
FastArray([True, True, True, True, True]),
FastArray([False, False, False, False, False]),
FastArray([False, True, True, True, True]),
FastArray([False, True, True, True, True]),
FastArray([True, False, False, False, False]),
FastArray([True, False, False, False, False]),
]
for d in (LikertDecision, decision_dict):
c = Categorical(codes, d)
# test valid integer code
for name, correct in zip(compare_func_names, valid_idx_correct):
func = c.__getattribute__(name)
result = func(valid_idx)
was_correct = bool(np.all(correct == result))
assert (
was_correct
), f"Categorical enum comparison failed with good integer index on {name} operation. {c.view(FastArray)} code: {valid_idx}"
# test invalid integer code
for name, correct in zip(compare_func_names, bad_idx_correct):
func = c.__getattribute__(name)
result = func(bad_idx)
was_correct = bool(np.all(correct == result))
assert was_correct, f"Categorical enum comparison failed with good integer index on {name} operation"
# ---------------------------------------------------------------------------
def test_compare_enum_str(self):
compare_func_names = [
'__ne__',
'__eq__',
'__ge__',
'__gt__',
'__le__',
'__lt__',
]
codes = [1, 44, 44, 133, 75]
valid_idx = 'StronglyAgree'
bad_idx = 'x'
valid_idx_correct = [
FastArray([True, False, False, True, True]),
FastArray([False, True, True, False, False]),
FastArray([False, True, True, True, True]),
FastArray([False, False, False, True, True]),
FastArray([True, True, True, False, False]),
FastArray([True, False, False, False, False]),
]
for d in (LikertDecision, decision_dict):
c = Categorical(codes, d)
# test valid category string
for name, correct in zip(compare_func_names, valid_idx_correct):
func = c.__getattribute__(name)
result = func(valid_idx)
was_correct = bool(np.all(correct == result))
assert was_correct, f"Categorical enum comparison failed with good category string on {name} operation"
# test invalid category string
for name in compare_func_names:
func = c.__getattribute__(name)
with pytest.raises(ValueError):
result = func(bad_idx)
def test_map(self):
c = Categorical(['b', 'b', 'c', 'a', 'd'], ordered=False)
mapping = {'a': 'AA', 'b': 'BB', 'c': 'CC', 'd': 'DD'}
result = c.map(mapping)
correct = FastArray([b'BB', b'BB', b'CC', b'AA', b'DD'])
assert bool(np.all(result == correct))
c = Categorical(['b', 'b', 'c', 'a', 'd'], ordered=False, base_index=0)
result = c.map(mapping)
assert bool(np.all(result == correct))
c = Categorical(['b', 'b', 'c', 'a', 'd'], ordered=False)
mapping = {'a': 'AA', 'b': 'BB', 'c': 'CC'}
result = c.map(mapping, invalid='INVALID')
correct = FastArray([b'BB', b'BB', b'CC', b'AA', b'INVALID'])
assert bool(np.all(result == correct))
c = Categorical(['b', 'b', 'c', 'a', 'd'], ordered=False, base_index=0)
result = c.map(mapping, invalid='INVALID')
assert bool(np.all(result == correct))
c = Categorical(['b', 'b', 'c', 'a', 'd'], ordered=False)
mapping = {'a': 1.0, 'b': 2.0, 'c': 3.0}
result = c.map(mapping, invalid=666)
correct = FastArray([2.0, 2.0, 3.0, 1.0, 666.0])
assert bool(np.all(result == correct))
c = Categorical(['b', 'b', 'c', 'a', 'd'], ordered=False, base_index=0)
result = c.map(mapping, invalid=666)
assert bool(np.all(result == correct))
c = Categorical(['b', 'b', 'c', 'a', 'd'], ordered=False)
result = c.map(mapping)
assert np.isnan(result[4])
c = Categorical(['b', 'b', 'c', 'a', 'd'], ordered=False, base_index=0)
result = c.map(mapping)
assert np.isnan(result[4])
c = Categorical(['b', 'b', 'c', 'a', 'd'], ordered=False)
mapping = FastArray(['w', 'x', 'y', 'z'])
result = c.map(mapping)
correct = FastArray([b'w', b'w', b'x', b'y', b'z'])
assert bool(np.all(result == correct))
c = Categorical(['b', 'b', 'c', 'a', 'd'], ordered=False, base_index=0)
result = c.map(mapping)
assert bool(np.all(result == correct))
c = Categorical([2, 2, 3, 1, 4, 0], ['a', 'b', 'c', 'd'])
mapping = {'a': 1.0, 'b': 2.0, 'c': 3.0}
result = c.map(mapping, invalid=666)
correct = FastArray([2.0, 2.0, 3.0, 1.0, 666.0, 666.0])
assert bool(np.all(result == correct))
# ---------------------------------------------------------------------------
def test_from_category(self):
c = Categorical(['a', 'a', 'b', 'c', 'a'])
bin = c.from_category('a')
assert bin == 1
c = Categorical(['a', 'a', 'b', 'c', 'a'], base_index=0)
bin = c.from_category(b'a')
assert bin == 0
with pytest.raises(ValueError):
bin = c.from_category('z')
c = Categorical(np.arange(5, 10))
bin = c.from_category(5)
assert bin == 1
with pytest.raises(ValueError):
bin = c.from_category(100)
c = Categorical([FastArray(['a', 'b', 'c']), np.arange(3)])
bin = c.from_category(('c', 2))
assert bin == 3
# ---------------------------------------------------------------------------
def test_getitem_enum_int(self):
codes = [1, 44, 44, 133, 75]
correct_strings = [
'StronglyDisagree',
'StronglyAgree',
'StronglyAgree',
'Agree',
'Disagree',
]
c = Categorical(codes, LikertDecision)
# getitem good init
for idx in range(5):
assert correct_strings[idx] == c[idx], f"Failed to return correct string for valid index in categorical."
# getitem bad init
with pytest.raises(IndexError):
result = c[5]
# ---------------------------------------------------------------------------
def test_getitem_enum_int_list(self):
codes = [1, 44, 44, 133, 75]
correct_strings = [
'StronglyDisagree',
'StronglyAgree',
'StronglyAgree',
'Agree',
'Disagree',
]
c = Categorical(codes, LikertDecision)
result = c[[1, 4]]
assert isinstance(
result, Categorical
), f"Failed to return Categorical when indexing by integer list. Returned {type(result)} instead."
assert result[0] == 'StronglyAgree'
assert result[1] == 'Disagree'
result = c[np.array([1, 4])]
assert isinstance(
result, Categorical
), f"Failed to return Categorical when indexing by integer list. Returned {type(result)} instead."
assert result[0] == 'StronglyAgree'
assert result[1] == 'Disagree'
result = c[FastArray([1, 4])]
assert isinstance(
result, Categorical
), f"Failed to return Categorical when indexing by integer list. Returned {type(result)} instead."
assert result[0] == 'StronglyAgree'
assert result[1] == 'Disagree'
def test_getitem_enum(self):
self._notimpl()
def test_setitem_enum(self):
self._notimpl()
# -------------------------------------------- MATLAB ----------------------------------
def test_ctor_matlab(self):
idx_list = [1.0, 2.0, 3.0, 4.0, 5.0]
str_list = ['a', 'b', 'c', 'd', 'e']
with pytest.raises(TypeError):
c = Categorical(idx_list, str_list)
c = Categorical(idx_list, str_list, from_matlab=True)
assert c[0] == 'a'
assert c.dtype == np.dtype(np.int8)
# def test_ctor_matlab_non_unique(self):
# idx_list = [1.0, 2.0, 3.0, 4.0, 5.0]
# str_list = ['a','b','c','d','d']
# with self.assertRaises(ValueError, msg=f"Failed to raise error when MATLab categories were not unique."):
# c = Categorical(idx_list, str_list, from_matlab=True)
# ------------------------------- PANDAS CATEGORICAL ----------------------------------
def test_ctor_pandas_cat(self):
idx_list = [0, 1, 2, 3, 4]
str_list = ['a', 'b', 'c', 'd', 'e']
pd_c = pd.Categorical.from_codes(idx_list, str_list)
pd_c = Categorical(pd_c)
rt_c = Categorical(idx_list, str_list)
cats_match = bool(np.all(pd_c.category_array == rt_c.category_array))
assert cats_match, f"Failed to create matching categories from pandas categorical"
# idx_match = bool(np.all(pd_c.view(np.ndarray)+1 == rt_c.view(np.ndarray)))
# self.assertTrue(idx_match, msg=f"Failed to create matching unerlying array from pandas categorical")
# convert pandas invalid bytes
pd_c = pd.Categorical.from_codes([-1, 0, 1, 2], ['a', 'b', 'c'])
pd_c = Categorical(pd_c)
cat_list = pd_c.category_array
assert len(cat_list) == 3
no_negative = bool(np.all(pd_c.view(FastArray) >= 0))
assert no_negative
# convert pandas invalid unicode
pd_c = pd.Categorical.from_codes([-1, 0, 1, 2], [u'\u2082', u'\u2083', u'\u2084'])
pd_c = Categorical(pd_c)
cat_list = pd_c.category_array
assert len(cat_list) == 3
no_negative = bool(np.all(pd_c.view(FastArray) >= 0))
assert no_negative
# --------------------------------RIPTABLE CATEGORICAL ----------------------------------------
# def test_ctor_rt_cat(self):
# c_unicode = Categorical(list_unicode)
# c = c_unicode.copy(forceunicode=True)
# self.assertEqual(c._categories_wrap._list.dtype.char, 'U', msg=f"Failed to force unicode on categorical copy.")
# ------------------------------------CUSTOM CATEGORIES ----------------------------------
def test_ctor_list_unique(self):
unique_str = ['a', 'b', 'c', 'd', 'e', 'f']
str_list = ['a', 'b', 'c', 'd', 'e']
c = Categorical(str_list, unique_str)
cats_match = bool(np.all(c._categories_wrap._list == unique_str))
assert cats_match, f"Failed to create matching categories from unique category input."
# ------------------------------------INTEGER ARRAY ----------------------------------
def test_ctor_integer_array(self):
lis = [1, 4, 9, 16, 25]
c = Categorical(lis)
for v1, v2 in zip(c, lis):
assert v1 == v2
# ------------------------------------GARBAGE ----------------------------------
def test_ctor_garbage(self):
with pytest.raises(TypeError):
c = Categorical(1, 2)
# ------------------------------------TEST FORCE DTYPE ----------------------------------
def test_init_with_dtype(self):
int_types = [np.int8, np.int16, np.int32, np.int64]
float_types = [np.float32, np.float64]
uint_types = [np.uint8, np.uint16, np.uint32, np.uint64]
arr = ['a', 'b', 'c', 'd', 'e']
for dt in int_types:
c = Categorical(arr, dtype=dt)
assert c.dtype == dt, f"Failed to force the correct dtype {dt} for categorical."
for dt in float_types + uint_types:
with pytest.raises(TypeError):
c = Categorical(arr, dtype=dt)
# ------------------------------------TEST CONVERT VALUE-------------------------------------
def test_possibly_convert_value(self):
'''
TODO: fix for new Categories class
'''
self._notimpl()
def test_categories_bad_init(self):
tup = ('a', 'b', 'c')
with pytest.raises(TypeError):
cat = Categories(tup)
def test_categories_len(self):
cats_from_list = Categorical(['a', 'b', 'c'], ordered=True, base_index=1, filter=None)._categories_wrap
assert len(cats_from_list) == 3
cats_from_enum = Categorical(FastArray([144]), LikertDecision)._categories_wrap
assert len(cats_from_enum) == 144
def test_get_categories(self):
c_list = [
'StronglyAgree',
'Agree',
'Disagree',
'StronglyDisagree',
'NeitherAgreeNorDisagree',
]
cats_from_list = Categories(c_list, unicode=True)
cats_from_enum = Categories(LikertDecision)
get_cats_match = bool(np.all(cats_from_list.get_categories() == cats_from_enum.get_categories()))
assert get_cats_match
def test_possibly_add_categories(self):
self._notimpl()
# uniquify and sort
# raise exception for adding cats to intenum, etc.
def test_categories_preserves_subtype(self):
# Test the Categorical.categories() method preserves the array type for the category data.
# This is important because we want the array(s) returned by this method to have the same type
# as the internal data (i.e. what's returned by Categorical.category_array or Categorical.category_dict).
# Single-key Categorical
dates = rt.Date(
[
'2019-03-15',
'2019-04-18',
'2019-05-17',
'2019-06-21',
'2019-07-19',
'2019-08-16',
'2019-09-20',
'2019-10-18',
'2019-11-15',
'2019-12-20',
]
)
dates.name = 'dates'
dates_cat = rt.Cat(dates)
cats = dates_cat.categories()
assert type(dates) == type(cats)
# Multi-key Categorical
datestrs = rt.FA(
[
'2019-03-15',
'2019-04-18',
'2019-05-17',
'2019-06-21',
'2019-07-19',
'2019-08-16',
'2019-09-20',
'2019-10-18',
'2019-11-15',
'2019-12-20',
]
)
datestrs.name = 'datestrs'
mcat = rt.Cat([dates, datestrs])
mcats = mcat.categories()
assert type(mcats['key_0']) == type(dates)
assert type(mcats['key_1']) == type(datestrs)
# Empty single-key Categorical
dates = rt.Date([])
dates_cat = rt.Cat(dates)
cats = dates_cat.categories()
assert type(dates) == type(cats)
def test_make_unique(self):
# SJK: changed this test on 8/21/2018 - count now comes from the grouping object, not Categories.make unique
values = FastArray(['a', 'b', 'c', 'c', 'd', 'a', 'b'])
# c = Categories([],base_index=1)
# index, cat_len, filter = c.make_unique(values)
cat = Categorical(values, ordered=True, base_index=1, filter=None)
index = cat._fa
c = cat._categories_wrap
assert len(index) == 7
assert max(index) == 4
assert c._mode == CategoryMode.StringArray
assert c._list.dtype.char == 'S'
assert c.isbytes
univals = values.astype('U')
cat = Categorical(univals, ordered=True, base_index=1, filter=None, unicode=True)
index = cat._fa
c = cat._categories_wrap
assert len(index) == 7
assert max(index) == 4
assert c._mode == CategoryMode.StringArray
assert c._list.dtype.char == 'U'
assert c.isunicode
@pytest.mark.xfail(
reason='20200416 This test was previously overridden by a later test in the file with the same name. Need to revisit and get back in a working state.'
)
def test_force_base_index(self):
filter = FastArray([True, True, False, False, True])
c = Categorical(['a', 'a', 'b', 'c', 'a'])
assert c.base_index == 1, 'Did not default base index to 1'
assert c._fa[0] == 1, 'Did not default base index to 1'
c = Categorical(['a', 'a', 'b', 'c', 'a'], base_index=0)
assert c.base_index == 0, 'Did not force base index to 0'
assert c._fa[0] == 0, 'Did not force base index to 0'
c = Categorical(['a', 'a', 'b', 'c', 'a'], filter=filter)
assert len(c.category_array) == 1
assert c._fa[2] == 0, 'Did not default base index to 1'
c = Categorical(['a', 'a', 'b', 'c', 'a'], base_index=0, filter=filter)
assert len(c.category_array) == 1
assert c._fa[2] == INVALID_DICT[c.dtype.num], 'Did not force base index to 0'
with pytest.raises(ValueError):
c = Categorical(['a', 'a', 'b', 'c', 'a'], base_index=99, filter=filter)
c = Categorical(['a', 'a', 'b', 'c', 'a'], ['a', 'b', 'c'])
assert c.base_index == 1, 'Did not default base index to 1'
assert c._fa[0] == 1, 'Did not default base index to 1'
c = Categorical(['a', 'a', 'b', 'c', 'a'], ['a', 'b', 'c'], base_index=0)
assert c.base_index == 0, 'Did not force base index to 0'
assert c._fa[0] == 0, 'Did not force base index to 0'
with pytest.raises(NotImplementedError):
c = Categorical(['a', 'a', 'b', 'c', 'a'], ['a', 'b', 'c'], base_index=0, filter=filter)
with pytest.raises(ValueError):
c = Categorical([1.0, 2.0, 3.0], ['a', 'b', 'c'], from_matlab=True, base_index=0)
pdc = pd.Categorical(['a', 'a', 'b', 'c', 'a'])
with pytest.raises(ValueError):
c = Categorical(pdc, base_index=0)
def test_is_in_unique_strings(self):
values = ['a', 'b', 'c', 'c', 'd', 'a', 'b']
good_cats = ['a', 'b', 'c', 'd']
incomplete_cats = ['a', 'b', 'c']
bad_cats = ['a', 'a', 'b']
invalid = 'invalid'
###--------REMOVED from_provided_categories, rewrite these tests to go through main constructor
# valid bytes
c = Categorical(values, good_cats, ordered=True, base_index=1, unicode=False, filter=None)
cats = c._categories_wrap
assert len(c) == 7
assert max(c._fa) == 4
assert cats._mode == CategoryMode.StringArray
assert cats._list.dtype.char == 'S'
assert cats.isbytes
# valid unicode
c = Categorical(values, good_cats, ordered=True, base_index=1, unicode=True, filter=None)
cats = c._categories_wrap
assert len(c) == 7
assert max(c._fa) == 4
assert cats._mode == CategoryMode.StringArray
assert cats._list.dtype.char == 'U'
assert cats.isunicode
# non-unique categories
# 4/12/2019 - no longer checks for uniqueness
# with self.assertRaises(ValueError):
# c = Categories.from_provided_categories(values, bad_cats, ordered=True, base_index=1, unicode=False, filter=None)
# not all values found in categories
with pytest.raises(ValueError):
c = Categorical(values, incomplete_cats, ordered=True, base_index=1, unicode=False, filter=None,)
# insert invalid True
# 5/16/2019 invalid must appear in provided uniques
with pytest.raises(ValueError):
c = Categorical(
values, incomplete_cats, ordered=True, base_index=1, unicode=True, filter=None, invalid=invalid,
)
cats = c._categories_wrap
assert len(c) == 7
assert max(c._fa) == 3
assert cats._mode == CategoryMode.StringArray
assert cats._list.dtype.char == 'U'
assert cats.isunicode
def test_getitem_enum_str(self):
codes = [1, 44, 44, 133, 75]
correct = [True, False, False, False, False]
valid_str = 'StronglyDisagree'
invalid_str = 'q'
c = Categorical(codes, LikertDecision)
# with self.assertRaises(IndexError):
mask = c[valid_str]
is_correct = bool(np.all(mask == correct))
assert is_correct
with pytest.raises(ValueError):
mask = c[invalid_str]
assert sum(mask) == 0
def test_match_str_to_category(self):
single_byte = b'a'
single_unicode = 'a'
single_true_unicode = u'\u2082'
byte_values = [b'a', b'b', b'c', b'c', b'd', b'a', b'b']
values = FastArray(['a', 'b', 'c', 'c', 'd', 'a', 'b'])
true_unicode = [u'\u2082', u'\u2083', u'\u2082']
# 4/25/2019 - changed these tests to construct a Categorical, rather than
# a Categories object directly. Categorical will always make a Categories object.
# (held in _categories_wrap)
c = Categorical(values, ordered=True, base_index=1, filter=None)
matching_char = c._categories_wrap.match_str_to_category(single_unicode)
assert isinstance(matching_char, bytes)
with pytest.raises(TypeError):
matching = c._categories_wrap.match_str_to_category(single_true_unicode)
univals = np.array(['a', 'b', 'c', 'c', 'd', 'a', 'b'])
c = Categorical(univals, ordered=True, base_index=1, filter=None, unicode=True)
matching_char = c._categories_wrap.match_str_to_category(single_byte)
assert isinstance(matching_char, str)
c = Categorical(values, ordered=True, base_index=1, filter=None)
matching = c._categories_wrap.match_str_to_category(values)
assert matching.dtype.char == 'S'
with pytest.raises(TypeError):
matching = c._categories_wrap.match_str_to_category(true_unicode)
c = Categorical(univals, ordered=True, base_index=1, filter=None, unicode=True)
matching = c._categories_wrap.match_str_to_category(values)
assert matching.dtype.char == 'U'
# Categories object being removed
# Disabling these tests - methods will move into Categorical
# 4/24/2019
# def test_get_category_index(self):
# values = FastArray(['a', 'b', 'c', 'c', 'd', 'a', 'b', 'g'])
# _, c, _, _ = Categories.from_array(values, ordered=True, base_index=1, filter=None)
# # when found, will return exact index
# str_idx = c.get_category_index('b')
# self.assertEqual(str_idx, 2)
# # when ordered, will return floating point for LTE GTE
# str_idx = c.get_category_index('e')
# self.assertEqual(str_idx, 4.5)
# # when unordered, will return invalid index (length of string array)
# c._sorted = False
# str_idx = c.get_category_index('e')
# self.assertEqual(str_idx, 6)
# def test_get_category_match_index(self):
# values = FastArray(['a', 'b', 'c', 'c', 'd', 'a', 'b', 'g'])
# _, c, _, _ = Categories.from_array(values, ordered=False, base_index=1, filter=None)
# string_matches = c.get_category_match_index(['a','b'])
# self.assertEqual(string_matches, [1,2])
# c._mode = CategoryMode.IntEnum
# with self.assertRaises(NotImplementedError):
# string_matches = c.get_category_match_index(['a','b'])
def test_possibly_invalid(self):
values = ['a', 'b', 'c', 'c', 'd', 'a', 'b', 'g']
c = Categorical(values, base_index=1)
out_of_range = -50
sentinel = INVALID_DICT[c.dtype.num]
c.view(FastArray)[0] = out_of_range
# c.view(FastArray)[1] = sentinel
# **changed invalid, all will display as bad code if changed underneath and not in range
assert c[0] == "!<-50>"
# self.assertEqual(c[1], "!<inv>")
def test_categories_getitem_str_list(self):
codes = [1, 44, 44, 133, 75]
correct = FastArray([False, True, True, False, True])
c = Categorical(codes, LikertDecision)
mask = c[['StronglyAgree', 'Disagree']]
is_correct = bool(np.all(mask == correct))
assert is_correct
mask = c[[b'StronglyAgree', b'Disagree']]
is_correct = bool(np.all(mask == correct))
assert is_correct
def test_categories_print_repr(self):
self._notimpl()
def test_enum_dict_warning(self):
class DupeEnum(IntEnum):
code_a = 1
code_b = 1
code_c = 1
code_d = 2
with pytest.warns(UserWarning):
c = Categorical([1, 2], DupeEnum)
# ------------------------- TEST MERGE -------------------------------------------
# def test_merge(self):
# from riptable.rt_categorical import categorical_merge
# c_bytes = Categorical(['b','b','b','a','b','b'], ['a','b'])
# c_unicode = Categorical(["AAPL\u2080","AMZN\u2082"])
# result = categorical_merge([c_bytes, c_unicode])
# # self.assertTrue(result[0]._categories_wrap._list is result[1]._categories_wrap._list, msg=f"Categorical merge did not assign the same dictionary to both arrays.")
# self.assertEqual(result[0]._categories_wrap._list.dtype.char, 'U', msg=f"{result[0]._categories_wrap._list.dtype.char} was not 'U'. dictionary was not flipped to unicode.")
# for item in c_bytes._categories_wrap._list:
# self.assertTrue(item.decode() in result[0]._categories_wrap._list, msg=f"{item} did not appear in final categories")
# for item in c_unicode._categories_wrap._list:
# self.assertTrue(item in result[0]._categories_wrap._list, msg=f"{item} did not appear in final categories")
# c1 = Categorical([1, 1, 3, 2, 2], [1, 2, 3, 4, 5], from_matlab=True)
# c2 = Categorical([2, 2, 4, 4, 3], [1, 2, 3, 4, 5], from_matlab=True)
# [cm1, cm2] = categorical_merge([c1, c2])
# self.assertTrue((cm1 == [1, 1, 3, 2, 2]).all())
# self.assertTrue((cm2 == [2, 2, 4, 4, 3]).all())
# ------------------------- TEST HSTACK -------------------------------------------
def test_hstack(self):
c1 = Categorical(['a', 'a', 'c', 'b', 'b'])
c2 = Categorical(['b', 'b', 'd', 'd', 'c'])
cm = Categorical.hstack([c1, c2])
assert (cm.as_string_array == ['a', 'a', 'c', 'b', 'b', 'b', 'b', 'd', 'd', 'c']).all()
c1 = Categorical([1, 1, 3, 2, 2], [1, 2, 3, 4, 5], from_matlab=True)
c2 = Categorical([2, 2, 4, 4, 3], [1, 2, 3, 4, 5], from_matlab=True)
cm = Categorical.hstack([c1, c2])
assert (cm == [1, 1, 3, 2, 2, 2, 2, 4, 4, 3]).all()
def test_hstack_fails_for_different_mode_cats(self):
# Create a dictionary-mode Categorical (from ISO3166 data).
# The dictionary is created manually below instead of using e.g.
# {k: int(v) for (k, v) in ISOCountryCode.__members__.items()}
# so the dictionary we give to Categorical does not have the insert ordering
# imply an ordering of the keys/values.
country_code_dict = {
'IRL': 372, 'USA': 840, 'AUS': 36, 'HKG': 344, 'JPN': 392,
'MEX': 484, 'KHM': 116, 'THA': 764, 'JAM': 388, 'ARM': 51
}
# The values for the Categorical's backing array.
# This includes some value(s) not in the dictionary and not all values in the dictionary are used here.
country_num_codes = [36, 36, 344, 840, 840, 372, 840, 372, 840, 124, 840, 124, 36, 484]
cat1 = rt.Categorical(country_num_codes, country_code_dict)
assert cat1.category_mode == CategoryMode.Dictionary
# Create a single-key, string-mode Categorical.
cat2 = rt.Categorical(['AUS', 'AUS', 'HKG', 'USA', 'USA', 'IRL', 'USA', 'IRL', 'USA', 'KHM', 'IRL', 'AUS', 'MEX'])
assert cat2.category_mode != CategoryMode.Dictionary
# Try to hstack the two Categoricals. This should fail due to the CategoryMode values being different.
with pytest.raises((ValueError, TypeError)):
rt.hstack([cat1, cat2])
def test_align(self):
c1 = Categorical(['a', 'b', 'c'])
c2 = Categorical(['d', 'e', 'f'])
c3 = Categorical(['c', 'f', 'z'])
cm = Categorical.align([c1, c2, c3])
assert (cm[0].as_string_array == ['a', 'b', 'c']).all()
assert (cm[1].as_string_array == ['d', 'e', 'f']).all()
assert (cm[2].as_string_array == ['c', 'f', 'z']).all()
assert (cm[0].categories() == FastArray([b'Filtered', b'a', b'b', b'c', b'd', b'e', b'f', b'z'])).all()
assert (cm[0].categories() == cm[1].categories()).all()
assert (cm[0].categories() == cm[2].categories()).all()
c1 = Categorical([1, 1, 3, 2, 2], [1, 2, 3, 4, 5], from_matlab=True)
c2 = Categorical([2, 2, 4, 4, 3], [1, 2, 3, 4, 5], from_matlab=True)
cm = Categorical.align([c1, c2])
assert (cm[0] == [1, 1, 3, 2, 2]).all()
assert (cm[1] == [2, 2, 4, 4, 3]).all()
# Multikey with nested Categorical
c1 = Categorical([Categorical(['a']), FastArray([1])])
c2 = Categorical([Categorical(['b']), FastArray([2])])
cm = Categorical.align([c1, c2])
assert cm[0][0] == ('a', 1)
assert cm[1][0] == ('b', 2)
assert cm[0].category_dict == cm[1].category_dict
def test_categorical_merge_dict(self):
from riptable.rt_categorical import categorical_merge_dict
d1 = {'a': 1, 'b': 2, 'c': 3, 'd': 4, 'e': 5}
d2 = {'a': 1, 'e': 5, 'b': 2, 'f': 6}
c1 = Categorical([3, 3, 4, 3, 1, 2, 5], d1)
c2 = Categorical([1, 1, 5, 2, 2, 1, 5], d2)
combined = categorical_merge_dict([c1, c2], return_type=dict)
for i in range(1, 6):
assert i in combined.values()
def test_getitem_empty(self):
c = Categorical([0, 1, 2], ['a', 'b', 'c'])
empty_list = c[[]]
assert isinstance(empty_list, Categorical)
dict_matches = bool(np.all(empty_list.categories() == c.categories()))
assert dict_matches
with pytest.raises(IndexError):
empty_np = c[np.array([])]
assert isinstance(empty_np, Categorical)
dict_matches = bool(np.all(empty_np.categories() == c.categories()))
assert dict_matches
def test_iter_groups(self):
correct_keys = FastArray(['a', 'b', 'c', 'd', 'e'])
correct_idx = [[8], [3], [5, 6], [2, 9], [0, 1, 4, 7]]
str_arr = FastArray(['e', 'e', 'd', 'b', 'e', 'c', 'c', 'e', 'a', 'd'])
c = Categorical(str_arr)
for i, tup in enumerate(c.iter_groups()):
assert tup[0] == correct_keys[i]
assert bool(np.all(tup[1] == correct_idx[i]))
def test_enum_dict_multi(self):
self._notimpl()
# not implemented
def test_enum_init_errors(self):
with pytest.raises(TypeError):
c = Categorical(['a', 'b', 'c'], LikertDecision)
def test_custom_invalid_category(self):
# 5/16/2019 invalid must appear in provided uniques
c = Categorical(
['a', 'b', 'c', 'my_invalid'], ['a', 'b', 'c', 'my_invalid'], invalid='my_invalid', base_index=1,
)
assert c[3] == 'my_invalid'
assert c.isnan()[3]
assert len(c.category_array) == 4
@pytest.mark.xfail(reason="After invalid_set, the custom invalid value is not displayed.")
def test_invalid_set(self):
c = Categorical(
['a', 'b', 'c', 'my_invalid'], ['a', 'b', 'c', 'my_invalid'], invalid='my_invalid', base_index=1,
)
# set a new string to be displayed for invalid items and validate
custom_invalid = "custom_invalid"
c.invalid_set(custom_invalid)
assert c[3] == custom_invalid
assert c.isnan()[3]
assert len(c.category_array) == 4
def test_lock_unlock(self):
self._notimpl()
# halfway implemented
def test_set_item(self):
self._notimpl()
# when index needs to be fixed after categories are added
# setitem with integer / invalid integer
# setitem with string / invalid category
def test_return_empty_cat(self):
self._notimpl()
# this code still needs to get written
def test_getitem_np_str(self):
c = Categorical(['a', 'a', 'b', 'a', 'c', 'c', 'b'])
correct = FastArray([True, True, True, True, False, False, True])
with pytest.raises(IndexError):
result = c[np.array(['a', 'b'])]
# self.assertTrue(array_equal(result, correct), msg=f"incorrect getitem result when indexing by numpy array of strings")
with pytest.raises(IndexError):
result = c[np.array(['a', 'b']).astype('S')]
# self.assertTrue(array_equal(result, correct), msg=f"incorrect getitem result when indexing by numpy array of strings")
def test_getitem_slice(self):
c = Categorical(['a', 'a', 'b', 'a', 'c', 'c', 'b'])
result = c[:3]
assert isinstance(result, Categorical)
match_fa = bool(np.all(result.view(FastArray) == [1, 1, 2]))
assert match_fa
assert len(result) == 3
assert len(result._categories_wrap) == 3
def test_categorical_compare_check(self):
self._notimpl()
# Categories have different modes
# categories are both enum
# compare cat to empty list
# non-categorical input
# convert all to unicode if one is unicode
# this keyword wasn't used anywhere, removed from copy()
# def test_copy_invalid(self):
# c = Categorical(['a','a','b','a','c','c','b'])
# invalid_copy = c.copy(fill_invalid=True)
# all_invalid = bool(np.all(invalid_copy.view(FastArray)==-128))
# self.assertTrue(all_invalid)
# for idx, item in enumerate(c.categories()):
# self.assertEqual(item, invalid_copy.categories()[idx])
# self.assertFalse(c.categories() is invalid_copy.categories())
def test_fill_invalid(self):
values = list('aabaccb')
c = Categorical(values, base_index=1)
c.fill_invalid(inplace=True)
assert_array_equal(FastArray([c.filtered_name] * len(values)), c.expand_array)
assert_array_equal(FastArray([0] * len(values)), c._fa)
expected = FastArray(sorted(set(values))).astype('|S1')
assert_array_equal(expected, c.category_array)
assert_array_equal(expected, c.category_dict[next(iter(c.category_dict))]) # values of first key
def test_force_unicode(self):
c = Categorical(['a', 'a', 'b', 'a', 'c', 'c', 'b'], unicode=True)
result_dtype = c.categories().dtype.char
assert result_dtype == 'U', f"Failed to force unicode when constructing categorical from list of string values"
def test_categories_shallow_copy(self):
codes = [10, 10, 20, 10, 30, 20, 10]
d = {10: 'a', 20: 'b', 30: 'c'}
c = Categorical(codes, d)
original_cats = c._categories_wrap
new_cats = original_cats.copy(deep=False)
assert (
original_cats._str_to_int_dict is new_cats._str_to_int_dict
), f"Categories did not use same str_to_int dictionary after shallow copy."
assert (
original_cats._int_to_str_dict is new_cats._int_to_str_dict
), f"Categories did not use same int_to_str dictionary after shallow copy."
# 5/16/2019 invalid category must be in user provided
# def test_two_lists_invalid(self):
# c = Categorical(['a','a','b','a','c','c','b'],np.array(['a','b']), invalid='inv', base_index=1)
# self.assertEqual(c[4],FILTERED_LONG_NAME)
@pytest.mark.xfail(
reason='20200416 This test was previously overridden by a later test in the file with the same name. Need to revisit and get back in a working state.'
)
def test_getitem_enum_list(self):
c = Categorical([44, 133, 133, 75, 144, 1], LikertDecision)
with pytest.raises(IndexError):
result = c[[b'NeitherAgreeNorDisagree']]
correct = FastArray([False, False, False, False, True, False])
# self.assertTrue(array_equal(result, correct))
result = c[[4]]
assert result[0] == 'NeitherAgreeNorDisagree'
def test_non_unique(self):
with pytest.raises(ValueError):
c = Categorical(['a', 'a', 'b', 'a', 'c', 'c', 'b'], ['a', 'a', 'b'])
def test_match_to_category(self):
c = Categorical(['a', 'a', 'b', 'a', 'c', 'c', 'b'])
result = c._categories_wrap.match_str_to_category('a')
assert b'a' == result
with pytest.raises(TypeError):
result = c._categories_wrap.match_str_to_category([1, 2, 3])
with pytest.raises(TypeError):
result = c._categories_wrap.match_str_to_category({1, 2, 3})
c1 = Categorical(['abc', 'def', 'abc', 'abc'], np.array(['abc', 'def']), unicode=True)
result = c1._categories_wrap.match_str_to_category([b'a'])
assert result.dtype.char == 'U'
# ------------------------------------TEST SET ITEM------------------------------------------
def test_set_item_str_index(self):
c = Categorical(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])
correct = [2, 2, 2, 2, 2, 2]
c['a'] = 'b'
is_correct = bool(np.all(c.view(FastArray) == correct))
assert is_correct, f"Category was not correctly changed with set item on a string."
with pytest.raises(ValueError):
c['b'] = 'c'
def test_set_item_int_index(self):
c = Categorical(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])
correct = [1, 2, 2, 1, 2, 2]
c[0] = 'a'
is_correct = bool(np.all(c.view(FastArray) == correct))
assert is_correct, f"Category was not correctly changed with set item on an int."
with pytest.raises(ValueError):
c[0] = 'c'
# ------------------------------------TEST CALCULATE DTYPE ----------------------------------
def test_get_dtype_from_len(self):
'''
Categorical will select different types
'''
dtype_sizes = {
np.int8: 1,
np.int16: 101,
np.int32: 50001,
} # , np.int64:2000000001 }
for dt, sz in dtype_sizes.items():
LENGTH = 6
NO_CODES = sz
alphabet = list('abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789')
np_alphabet = np.array(alphabet, dtype="|U1")
np_codes = np.random.choice(np_alphabet, [NO_CODES, LENGTH])
codes = ["".join(np_codes[i]) for i in range(len(np_codes))]
c = Categorical(["".join(np_codes[i]) for i in range(len(np_codes))])
# only perform the test if there are enough uniques
if len(c._categories_wrap._list) >= sz:
assert c.dtype == dt, f"Categorical did not set dtype to {dt} for array of size {sz}."
# -------SINGLE INTEGER
def test_getitem_int(self):
'''
Single integer index should return the corresponding category in unicode format.
'''
c = Categorical(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])
assert c[0] == 'b', f"Get item with integer did not return the correct category."
assert isinstance(c[0], str), f"Get item with integer did not return as unicode."
assert c[3] == 'a', f"Get item with integer did not return the correct category."
assert isinstance(c[3], str), f"Get item with integer did not return as unicode."
with pytest.raises(IndexError):
d = c[10]
# --------INTEGER MASK
def test_getitem_int_mask(self):
py_mask = [0, 3]
c = Categorical(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])
for mask in [py_mask, np.array(py_mask)]:
d = c[mask]
assert isinstance(
d, Categorical
), f"Get item with integer mask did not return a categorical. Returned {type(d).__name__} instead."
assert len(d) == len(
mask
), f"Get item with integer mask did not return categorical of {len(mask)}. returned {len(d)} instead."
has_same_cats = bool(np.all(d._categories_wrap._list == c._categories_wrap._list))
assert (
has_same_cats
), f"Failed to copy the same categories to new categorical after getitem with integer mask."
d = c[[0, 10]]
assert d._fa[1] == 0, f"Failed to put invalid for out of range index."
# -------BOOLEAN MASK
def test_getitem_bool_mask(self):
py_mask = [True, True, True, False, True, True]
c = Categorical(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])
for mask in [py_mask, np.array(py_mask)]:
d = c[mask]
assert not (
b'a' in d.as_string_array
), f"b'a' does not get trimmed out of categorical with getitem from boolean array."
assert 5 == len(
d
), f"Length {len(d)} did not match 5 in categorical getitem with a boolean array of the same size."
has_same_cats = bool(np.all(d._categories_wrap._list == c._categories_wrap._list))
assert (
has_same_cats
), f"Failed to copy the same categories to new categorical after getitem with integer mask."
# -------SINGLE STRING
def test_getitem_single_string(self):
b_result = [True, True, True, False, True, True]
c = Categorical(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])
idx = b'c'
# with self.assertRaises(IndexError):
d = c[idx]
has_true = bool(np.any(d))
assert not has_true, f"Failed to return an array of all false for getitem with {idx}"
assert isinstance(d, FastArray), f"Get item input {idx} did not return FastArray"
assert d.dtype.char == '?', f"Get item input {idx} did not return FastArray"
idx = idx.decode()
# with self.assertRaises(IndexError):
d = c[idx]
has_true = bool(np.any(d))
assert not has_true, f"Failed to return an array of all false for getitem with {idx}"
assert isinstance(d, FastArray), f"Get item input {idx} did not return FastArray"
assert d.dtype.char == '?', f"Get item input {idx} did not return FastArray"
idx = b'b'
# with self.assertRaises(IndexError):
d = c[idx]
is_correct = bool(np.all(d == b_result))
assert is_correct, f"Did not return the correct array for getitem with {idx}"
assert isinstance(d, FastArray), f"Get item input {idx} did not return FastArray"
assert d.dtype.char == '?', f"Get item input {idx} did not return FastArray"
idx = idx.decode()
# with self.assertRaises(IndexError):
d = c[idx]
is_correct = bool(np.all(d == b_result))
assert is_correct, f"Did not return the correct array for getitem with {idx}"
assert isinstance(d, FastArray), f"Get item input {idx} did not return FastArray"
assert d.dtype.char == '?', f"Get item input {idx} did not return FastArray"
# ------MULTIPLE STRINGS
def test_getitem_multiple_strings(self):
c = Categorical(['b', 'b', 'b', 'a', 'b', 'b'])
inputs = {
(b'b',): [True, True, True, False, True, True], # single in (list)
(b'c',): [False, False, False, False, False, False], # single not in (list)
(b'a', b'b'): [True, True, True, True, True, True], # both in (list)
(b'c', b'd'): [False, False, False, False, False, False,], # both not in (list)
(b'b', b'c'): [True, True, True, False, True, True], # mixed (list)
}
for idx, correct in inputs.items():
idx = list(idx)
d = c[idx]
is_correct = bool(np.all(d == correct))
assert is_correct, f"Indexing categorical {c} by {idx} did not return the correct result."
assert d.dtype.char == '?', f"Get item input {idx} did not return FastArray"
idx = [b.decode() for b in idx]
d = c[idx]
is_correct = bool(np.all(d == correct))
assert is_correct, f"Indexing categorical {c} by {idx} did not return the correct result."
assert d.dtype.char == '?', f"Get item input {idx} did not return FastArray"
# ------NUMERIC GETITEM
def test_getitem_numeric_categories(self):
# before it was fixed, a bug was returning a string of the numeric category
nums = np.array([1, 1, 2, 3, 4, 5, 1, 1, 1])
c = Categorical(nums)
assert c[0] == 1
assert isinstance(c[0], (int, np.integer))
nums = nums.astype(np.float32)
c = Categorical(nums)
assert c[0] == 1.0
assert isinstance(c[0], (float, np.floating)), f"Expected float, got {type(c[0])}"
# ------------------------- TEST COMPARE CHECK -------------------------------------------
def test_compare_check(self):
'''
Test comparison between two 'equal' categoricals with different underlying arrays.
'''
compare_ops = {
'__ne__': [False, False, False, False, False, False],
'__eq__': [True, True, True, True, True, True],
'__ge__': [True, True, True, True, True, True],
'__gt__': [False, False, False, False, False, False],
'__le__': [True, True, True, True, True, True],
'__lt__': [False, False, False, False, False, False],
}
c = Categorical(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b', 'c'])
d = Categorical(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])
for name, correct in compare_ops.items():
func = c.__getattribute__(name)
result = func(d)
is_correct = bool(np.all(result == correct))
assert is_correct, f"Compare operation betweeen two equal categoricals did not return the correct result."
def test_compare_return_type(self):
'''
Test comparison operations with single strings to make sure FastArray of boolean is returned.
'''
c = Categorical(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])
scalars = ['a', 'c']
compare_ops = ['__ne__', '__eq__', '__ge__', '__gt__', '__le__', '__lt__']
for s in scalars:
for op in compare_ops:
func = c.__getattribute__(op)
result = func(s)
assert isinstance(result, FastArray), f"comparison {op} with input {s} did not return FastArray"
assert result.dtype.char == '?', f"comparison {op} with input {s} did not return boolean"
def test_compare_different_modes(self):
c1 = Categorical(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])
c2 = Categorical([0, 1], {0: 'a', 1: 'b'})
with pytest.raises(TypeError):
c1 == c2
def test_compare_conflicting_dicts(self):
c1 = Categorical([0, 1], {0: 'a', 1: 'b'})
c2 = Categorical([0, 1], {1: 'a', 0: 'b'})
with pytest.raises(ValueError):
c1 == c2
def test_compare_safe_dicts(self):
c1 = Categorical([0, 1], {0: 'a', 1: 'b'})
c2 = Categorical([2, 1], {2: 'c', 1: 'b'})
correct = FastArray([False, True])
result = c1 == c2
match = bool(np.all(correct == result))
assert match
def test_isnan(self):
c = Categorical([1, 1, 3, 2, 2], ['a', 'b', 'c'], base_index=1, invalid='a')
is_correct = [True, True, False, False, False]
is_not_correct = [False, False, True, True, True]
assert bool(np.all(is_correct == isnan(c)))
assert bool(np.all(is_correct == c.isnan()))
assert bool(np.all(is_not_correct == isnotnan(c)))
assert bool(np.all(is_not_correct == c.isnotnan()))
# ------------------------------------------------------
def test_get_categories(self):
# string list
c = Categorical(['a', 'b', 'c', 'd', 'e'])
catsarray = c.category_array
assert isinstance(catsarray, np.ndarray)
catsdict = c.category_dict
assert isinstance(catsdict, dict)
assert len(catsdict) == 1
with pytest.raises(TypeError):
catscodes = c.category_codes
with pytest.raises(TypeError):
catsmapping = c.category_mapping
# numeric list
c = Categorical(np.array([1, 2, 3, 4, 5]))
catsarray = c.category_array
assert isinstance(catsarray, np.ndarray)
catsdict = c.category_dict
assert isinstance(catsdict, dict)
assert len(catsdict) == 1
with pytest.raises(TypeError):
catscodes = c.category_codes
with pytest.raises(TypeError):
catsmapping = c.category_mapping
# dict/enum
c = Categorical([1, 2, 3, 4], {1: 'a', 2: 'b', 3: 'c', 4: 'd'})
catsarray = c.category_array
assert isinstance(catsarray, np.ndarray)
catsdict = c.category_dict
assert isinstance(catsdict, dict)
assert len(catsdict) == 1
catscodes = c.category_codes
assert isinstance(catscodes, np.ndarray)
catsmapping = c.category_mapping
assert isinstance(catsmapping, dict)
# multikey
c = Categorical([np.arange(5), np.random.rand(5)])
with pytest.raises(TypeError):
catsarray = c.category_array
catsdict = c.category_dict
assert isinstance(catsdict, dict)
assert len(catsdict), 2
with pytest.raises(TypeError):
catscodes = c.category_codes
with pytest.raises(TypeError):
catsmapping = c.category_mapping
# ------------------------------------------------------
def test_force_base_index2(self):
c = Categorical(['a', 'a', 'b', 'c', 'a'])
assert c.base_index == 1
assert c._fa[0] == 1
c = Categorical(['a', 'a', 'b', 'c', 'a'], base_index=0)
assert c.base_index == 0
assert c._fa[0] == 0
codes = np.array([0, 0, 1, 2, 0])
cats = np.array(['a', 'b', 'c'])
# c = Categorical(codes, cats)
# self.assertEqual(c.base_index, 0)
# self.assertEqual(c._fa[0], 0)
codes += 1
c = Categorical(codes, cats, base_index=1)
assert c.base_index == 1
assert c._fa[0] == 1
codes = codes.astype(np.float32)
c = Categorical(codes, cats, from_matlab=True)
assert c.base_index == 1
assert c._fa[0] == 1
with pytest.raises(ValueError):
c = Categorical(codes, cats, from_matlab=True, base_index=0)
c = Categorical(np.array(['a', 'a', 'b', 'c', 'a']), np.array(['a', 'b', 'c']))
assert c.base_index == 1
assert c._fa[0] == 1
c = Categorical(np.array(['a', 'a', 'b', 'c', 'a']), np.array(['a', 'b', 'c']), base_index=0)
assert c.base_index == 0
assert c._fa[0] == 0
# ------------------------------------------------------
def test_ordered(self):
c = Categorical(['c', 'c', 'a', 'b', 'c'])
cats = c.category_array
assert cats[0] == b'a'
c = Categorical(['c', 'c', 'a', 'b', 'c'], ordered=False)
cats = c.category_array
assert cats[0] == b'c'
c = Categorical(['c', 'c', 'a', 'b', 'c'], ['c', 'a', 'b'])
cats = c.category_array
assert cats[0] == b'c'
assert not c.ordered
c = Categorical(['c', 'c', 'a', 'b', 'c'], ['a', 'b', 'c'])
assert c.ordered
## removed this test - side-effect of search sorted with unsorted array (not categorical related)
## false claim that categories are ordered in keyword
# c = Categorical(['c','c','a','c','c'], ['c','a','b'], ordered=True)
# self.assertTrue(bool(np.all(c!='c')))
# self.assertTrue(bool(np.all(c!=b'c')))
c = Categorical(['c', 'c', 'a', 'b', 'c'], ['c', 'a', 'b'], ordered=False)
cats = c.category_array
assert cats[0] == b'c'
assert not c.ordered
codes = FastArray([0, 0, 1, 2, 0])
cats = FastArray(['c', 'b', 'a'], unicode=True)
c = Categorical(codes, cats)
assert c.category_array[0] == 'c'
assert not c.ordered
# with self.assertWarns(UserWarning):
# c = Categorical(codes, cats, ordered=True)
# self.assertEqual(c.category_array[0], b'c')
# self.assertFalse(c.ordered)
# ------------------------------------------------------
def test_keywords_not_allowed(self):
# filter + base index 0
f = np.array([True, False, True])
with pytest.raises(ValueError):
c = Categorical(['a', 'b', 'c'], filter=f, base_index=0)
# ------------------------------------------------------
def test_display_properties(self):
'''
Categoricals take over their display properties to appear like strings (not the underlying integer array)
(see Utils.rt_display_properties)
'''
c = Categorical(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])
item_format, convert_func = c.display_query_properties()
assert item_format.length == DisplayLength.Long, f"Incorrect length for item format."
assert item_format.justification == DisplayJustification.Left
assert item_format.invalid == None
assert item_format.can_have_spaces == True
assert item_format.decoration == None
assert item_format.color == DisplayColumnColors.Default
assert convert_func.__name__ == 'display_convert_func'
# this could change, right now the convert function just does a str over the item
assert convert_func(1, item_format) == '1', f"Incorrect convert function was returned."
# ------------------------------------------------------
# -----MISC. COVER TESTS--------------------------------
def test_non_array_dict_categories_ctor(self):
with pytest.raises(TypeError):
c = Categories(['garbage', 'list'])
def test_too_many_args_categories_ctor(self):
with pytest.raises(ValueError):
c = Categories(FastArray([1]), FastArray([2]), FastArray([3]))
def test_filter_and_invalid(self):
c = Categorical(
['a', 'a', 'b', 'c', 'c'], ['c'], invalid='a', filter=FastArray([True, True, False, True, True]),
)
c.filtered_set_name('a')
assert bool(np.all(c._fa == [0, 0, 0, 1, 1]))
for i in range(3):
assert c[i] == 'a'
for i in range(3, 5):
assert c[i] == 'c'
def test_zero_base_with_invalid(self):
with pytest.raises(ValueError):
c = Categorical(['a', 'b', 'c'], ['b', 'c'], base_index=0)
# removed this property from Categories 04/24/2019
# def test_multikey_labels(self):
# c = Categorical([FastArray(['a','b','c']), FastArray([1,2,3])])
# labels = c._categories_wrap.multikey_labels
# self.assertTrue(isinstance(labels[0], tuple))
# self.assertEqual(labels[0][0],'a')
def test_ncols_non_multikey(self):
c = Categorical(['a', 'b', 'c'])
assert c._categories_wrap.ncols == 1
# now checks for single / multikey / enum, not CategoryMode
# def test_len_undefined_mode(self):
# c = Categorical(['a','b','c'])
# c._categories_wrap._mode = CategoryMode.Default
# self.assertEqual(len(c._categories_wrap),0)
def test_categories_copy_shallow(self):
c = Categorical(['a', 'b', 'c'])
copycat = c._categories_wrap.copy(deep=False)
assert isinstance(copycat, Categories)
def test_categories_copy_deep(self):
c = Categorical([1, 2, 3], {1: 'a', 2: 'b', 3: 'c'})
copycat = c._categories_wrap.copy(deep=False)
assert isinstance(copycat, Categories)
# impossible path, unless mode is forced like below. disabling 4/24/2019
# c._categories_wrap._mode = CategoryMode.Default
# with self.assertRaises(NotImplementedError):
# c = c._categories_wrap.copy()
def test_wrap_get_categories(self):
c = Categorical(['a', 'b', 'c'])
arr = c._categories_wrap.get_categories()
assert isinstance(arr, FastArray)
c = Categorical([FastArray(['a', 'b', 'c']), FastArray([1, 2, 3])])
d = c._categories_wrap.get_categories()
assert isinstance(d, dict)
def test_get_string_mode_nums(self):
c = Categorical(np.arange(5))
assert not c._categories_wrap.isbytes
assert not c._categories_wrap.isunicode
def test_pop_single_arr(self):
c = Categorical([np.array(['a', 'b', 'c'])])
d = Categorical(np.array(['a', 'b', 'c']))
assert bool(np.all(c == d))
c = Categorical({'test': np.array(['a', 'b', 'c'])})
d = Categorical(np.array(['a', 'b', 'c']))
assert bool(np.all(c == d))
def test_from_cat_as_array(self):
c = Categorical(FastArray([1, 2, 3]), _from_categorical=np.array(['a', 'b', 'c']))
assert isinstance(c.category_array, FastArray)
assert c.base_index == 1
def test_from_pandas_object(self):
pdc = pd.Categorical(['a', 'b', 'c'])
c = Categorical(pdc, unicode=True)
assert c.category_array.dtype.char == 'U'
c = Categorical(pdc, unicode=False)
assert c.category_array.dtype.char == 'S'
pdc = pd.Categorical(three_unicode)
c = Categorical(pdc)
assert c.category_array.dtype.char == 'U'
def test_empty_init(self):
with pytest.raises(ValueError):
c = Categorical({})
with pytest.raises(ValueError):
c = Categorical([])
def test_multi_with_cats(self):
with pytest.raises(NotImplementedError):
c = Categorical(
[FastArray(['a', 'b', 'c', 'a']), FastArray([1, 2, 3, 1])],
[FastArray(['a', 'b', 'c']), FastArray([1, 2, 3])],
)
# 5/9/2019 removed this warning to reduce constructor paths
# def test_unicode_warn(self):
# with self.assertWarns(UserWarning):
# c = Categorical([1,2,3],{1:'a',2:'b',3:'c'}, unicode=False)
def test_map_non_integer(self):
with pytest.raises(TypeError):
c = Categorical([1.0, 2.0, 3.0], {1: 'a', 2: 'b', 3: 'c'})
def test_category_multi_arrays(self):
with pytest.raises(TypeError):
c = Categorical([1, 2, 3], [np.arange(5), np.arange(5)])
def test_getitem_enum_list2(self):
c = Categorical([1, 1, 2, 3, 1], {'a': 1, 'b': 2, 'c': 3})
d = c[[1, 2, 3]]
assert d[0] == 'a'
def test_tuple_compare_error(self):
c = Categorical([FastArray(['a', 'b', 'c', 'a']), FastArray([1, 2, 3, 1])])
with pytest.raises(ValueError):
_ = c == ('a', 'b', 'c')
def test_filter_out_bytes_from_unicode(self):
c = Categorical(['a', 'a', 'b', 'c', 'a'], unicode=True, invalid=b'a')
assert bool(np.all(c._fa == [1, 1, 2, 3, 1]))
assert c.category_array.dtype.char == 'U'
assert 'a' in c.category_array
def test_bytes_compare_multikey(self):
c = Categorical([np.array(['a', 'b', 'c', 'a']), FastArray([1, 2, 3, 1])], unicode=True)
cols = c.category_dict
bytescol = list(cols.values())[0]
assert bytescol.dtype.char == 'U'
result = c == (b'a', 1)
assert bool(np.all(FastArray([True, False, False, True]) == result))
def test_cat_zero_wronge_base(self):
with pytest.raises(ValueError):
c = CatZero(['a', 'a', 'b', 'c', 'a'], base_index=1)
def test_preserve_name(self):
ds = TypeRegister.Dataset({'strcol': np.random.choice(['a', 'b', 'c'], 10), 'numcol': arange(10)})
c = Categorical(ds.strcol)
assert c.get_name() == 'strcol'
c = Categorical([ds.strcol, ds.numcol])
ds2 = c.sum(arange(10))
labels = ds2.label_get_names()
assert labels[0] == 'strcol'
assert labels[1] == 'numcol'
ds = TypeRegister.Dataset({'mycodes': np.random.randint(1, 4, 10)})
c = Categorical(ds.mycodes, {'a': 1, 'b': 2, 'c': 3})
assert c.get_name() == 'mycodes'
codes = np.random.randint(1, 4, 10)
cats = FastArray(['a', 'b', 'c'])
cats.set_name('test')
c = Categorical(codes, cats)
assert c.get_name(), 'test'
def test_subarray_name(self):
c = Categorical(['a', 'b'])
c1 = c[[0]]
assert c1.get_name() == c.get_name()
# Make sure there is no "quantum effect" that printing the array changes it's name.
_ = str(c1)
assert c1.get_name() == c.get_name()
def test_construct_from_categorical(self):
c = Categorical(['a', 'a', 'b', 'c', 'a'])
d = Categorical(c)
assert isinstance(d.category_array, np.ndarray)
assert isinstance(d.expand_array, np.ndarray)
d2 = Categorical([c])
assert isinstance(d2.category_array, np.ndarray)
assert isinstance(d2.expand_array, np.ndarray)
def test_total_size(self):
c = Categorical(['a', 'a', 'b', 'c', 'a'])
assert c._total_size == 8
c = Categorical([arange(5, dtype=np.int32), arange(5, dtype=np.int32)])
assert c._total_size == 45
c = Categorical([arange(5, dtype=np.int64), arange(5, dtype=np.int64)])
assert c._total_size == 85
# removed while modifying groupby calculation behavior
# def test_hold_dataset(self):
# ds = TypeRegister.Dataset({'strcol':np.random.choice(['a','b','c'],30), 'numcol':arange(30)})
# c = ds.cat('strcol')
# self.assertTrue(isinstance(c._dataset, TypeRegister.Dataset))
# result = c.sum()
# self.assertTrue(isinstance(result, TypeRegister.Dataset))
# self.assertEqual(result._nrows, 3)
def test_expand_dict(self):
og_strings = FastArray(['a', 'a', 'b', 'c', 'a'])
og_nums = arange(5)
c = Categorical([og_strings, og_nums])
d = c.expand_dict
assert isinstance(d, dict)
assert len(d) == 2
dictlist = list(d.values())
assert bool(np.all(dictlist[0] == og_strings))
assert bool(np.all(dictlist[1] == og_nums))
c = Categorical([1, 2, 3], {'a': 1, 'b': 2, 'c': 3})
d = c.expand_dict
assert isinstance(d, dict)
assert len(d) == 1
dictlist = list(d.values())
assert bool(np.all(dictlist[0] == arange(1, 4)))
c = Categorical(np.random.randint(0, 10, 100_100))
with pytest.warns(UserWarning):
d = c.expand_dict
def test_expand_array(self):
c = Categorical([1, 2, 3], {'a': 1, 'b': 2, 'c': 3})
arr = c.expand_array
assert bool(np.all(arr == arange(1, 4)))
c = Categorical([FastArray(['a', 'b', 'c', 'a']), FastArray([1, 2, 3, 1])])
# expand array now works on multikey categoricals, returns a tuple of expanded arrays SJK: 4/29/2019
multi_expand = c.expand_array
assert isinstance(multi_expand, tuple)
assert len(multi_expand) == 2
assert bool(np.all(FastArray(['a', 'b', 'c', 'a']) == multi_expand[0]))
assert bool(np.all(FastArray([1, 2, 3, 1]) == multi_expand[1]))
c._fa[:] = 0
multi_expand = c.expand_array
assert bool(np.all(isnan(multi_expand[1])))
assert bool(np.all(multi_expand[0] == b'Filtered'))
def test_true_false_spacer(self):
c = Categorical(['a', 'b', 'c'])
t_true = c._tf_spacer(['test', True])
assert t_true == 'testTrue '
t_false = c._tf_spacer(['test', False])
assert t_false == 'testFalse'
def test_mapping_hstack(self):
c1 = Categorical([1, 1, 1, 1, 2, 3], {'a': 1, 'b': 2, 'c': 3})
c2 = Categorical([1, 1, 1, 1, 3, 4], {'a': 1, 'c': 3, 'd': 4})
stacked = Categorical.hstack([c1, c2])
assert stacked.unique_count == 4
assert stacked.from_category('b') == 2
assert stacked.from_category('d') == 4
assert len(stacked) == 12
c1 = Categorical([1, 1, 1, 1, 2, 3], {'a': 1, 'b': 2, 'd': 3})
c2 = Categorical([1, 1, 1, 1, 3, 4], {'a': 1, 'c': 3, 'd': 4})
# removed, hstack now relies on unique codes only SJK: 3/5/2019
# with self.assertRaises(TypeError):
# c3 = Categorical.hstack([c1, c2])
def test_matlab_nan(self):
dts = [np.int8, np.int16, np.int32, np.int64]
matlab_float_idx = FastArray([1.0, 0.0, np.nan])
matlab_cats = ['a', 'b']
for dt in dts:
c = Categorical(matlab_float_idx, matlab_cats, dtype=dt, from_matlab=True)
assert bool(np.all(c._fa == [1, 0, 0])), f'failed to flip nan to zero for dtype {dt}'
assert np.dtype(dt) == c.dtype
def test_from_provided_with_filter(self):
# not found and filter
c = Categorical(
['a', 'a', 'b', 'c', 'd'],
['a', 'b', 'c'],
filter=FastArray([False, False, True, True, False]),
invalid='INVALID',
)
c.filtered_set_name('INVALID')
correct = FastArray([b'INVALID', b'INVALID', b'b', b'c', b'INVALID'])
assert bool(np.all(c.expand_array == correct))
# filter only (uses default invalid)
c = Categorical(['a', 'a', 'b', 'c'], ['a', 'b', 'c'], filter=FastArray([False, False, True, True]),)
f = c.filtered_name
correct = FastArray([f, f, b'b', b'c'])
assert bool(np.all(c.expand_array == correct))
# even though filtered out, categories still untouched
correct = FastArray([b'a', b'b', b'c'])
assert bool(np.all(c.category_array == correct))
# filtering not allowed for base index 0
with pytest.raises(ValueError):
c = Categorical(
['a', 'a', 'b', 'c'], ['a', 'b', 'c'], filter=FastArray([False, False, True, True]), base_index=0,
)
def test_numeric_invalid(self):
# 5/16/2019 invalid category must be in provided uniques
c = Categorical([1.0, 1.0, 2.0], [1.0, 2.0], invalid=2.0)
assert c._fa[2] == 2
num = c.sum(arange(1, 4)).col_0[0]
assert num == 3
def test_get_groupings(self):
g, f, n = (
FastArray([2, 3, 0, 4, 1]),
FastArray([0, 0, 2, 4]),
FastArray([0, 2, 2, 1]),
)
c = Categorical(['b', 'c', 'a', 'a', 'b'], base_index=0)
gg = c.get_groupings()
group = gg['iGroup']
first = gg['iFirstGroup']
ncount = gg['nCountGroup']
assert bool(np.all(g == group))
assert bool(np.all(f == first))
assert bool(np.all(n == ncount))
c = Categorical(['b', 'c', 'a', 'a', 'b'], base_index=1)
gg = c.get_groupings()
group = gg['iGroup']
first = gg['iFirstGroup']
ncount = gg['nCountGroup']
assert bool(np.all(g == group))
assert bool(np.all(f == first))
assert bool(np.all(n == ncount))
def test_repr(self):
# just make sure no error for coverage
c = Categorical(['a', 'b', 'c'])
r = c.__repr__()
assert r, f"Representation should not be empty for Categorical '{c}'."
assert isinstance(r, str)
def test_copy_deep(self):
c = Categorical(['a', 'b', 'c'])
d = c.copy(deep=True)
d[0] = 'b'
assert c[0] == 'a'
assert c._fa[0] == 1
assert d[0] == 'b'
assert d._fa[0] == 2
def test_copy_new_filter(self):
a = Categorical('A B A B A B'.split())
b = Categorical('B A B A B A'.split())
c = a.copy()
f = c == 'A'
c[f] = b[f]
assert c[0] == 'B'
assert c[1] == 'B'
assert a[0] == 'A'
assert a[1] == 'B'
assert b[0] == 'B'
assert b[1] == 'A'
def test_setitem_tuple(self):
c = Categorical([arange(5), arange(5)])
c[0] = (1, 1)
assert c._fa[0] == 2
def test_nunique(self):
codes = np.random.randint(0, 3, 1000)
d = {0: 'All', 1: 'ManualAndQuasi', 2: 'Manual'}
c = Categorical(codes, d)
n = c.nunique()
assert n == 3
assert len(c.unique()) == 3
codes = np.ones(1000, dtype=np.int32)
c = Categorical(codes, d)
n = c.nunique()
assert n == 1
assert len(c.unique()) == 1
codes = arange(5)
c = Categorical(codes, d)
n = c.nunique()
assert n == 5
assert len(c.unique()) == 5
c = Categorical(['a', 'a', 'b', 'c', 'd'], ['a', 'b', 'c', 'd'])
n = c.nunique()
assert n == 4
assert len(c.unique()) == 4
c = Categorical(['a', 'a', 'b', 'c', 'd'], ['a', 'b', 'c', 'd'], base_index=0)
n = c.nunique()
assert n == 4
assert len(c.unique()) == 4
c = Categorical(['a', 'a', 'b', 'c', 'd'])
c[2] = 0
n = c.nunique()
assert n == 3
assert len(c.unique()) == 3
assert c.unique_count == 4
c = Categorical([arange(3), np.array(['a', 'b', 'c'])])
c[0] = 0
n = c.nunique()
assert n == 2
assert c.unique_count == 3
# The following assertion is moved to it's own unit pytest along with an xfail.
# found below and named test_multikey_categorical_unique.
# assert len(c.unique()) == 2
def test_unique(self):
l = list('xyyz')
c, c_sub = rt.Cat(l), rt.Cat(l[:3])
assert_array_equal(c.unique(), c.category_array, 'mismatch between unique categories and category array')
assert_array_equal(c.unique(), c.category_array.unique(), 'mismatch between unique categories and expanded category array')
assert c.nunique() == 3, 'mismatch in number of unique categories'
assert_array_equal(c[:3].unique(), c_sub.category_array, 'mismatch between unique categories and category array with sliced categorical')
assert_array_equal(c[:3].unique(), c_sub.category_array.unique(), 'mismatch between unique categories and expanded category array with sliced categorical')
assert c[:3].nunique() == 2, 'mismatch in number of unique categories with sliced categorical'
def test_scalar_unique(self):
idx = ones(100)
cats = 700_000.0
c = Categorical(idx, cats, from_matlab=True)
assert isinstance(c, Categorical)
assert c.unique_count == 1
def test_stack_multikey(self):
# TODO pytest parameterize the strings
strs = FA(np.random.choice(['aaaaa', 'b', 'ccc'], 23))
flts = np.random.choice([7.14, 6.66, 5.03], 23)
c1 = Categorical([strs, flts])
c1_str = Categorical(strs)
c1_flt = Categorical(flts)
strs2 = FA(np.random.choice(['b', 'aaaaa'], 17))
flts2 = np.random.choice([5.03, 7.14], 17)
c2 = Categorical([strs2, flts2])
c2_str = Categorical(strs2)
c2_flt = Categorical(flts2)
fa_str = hstack([strs, strs2])
fa_flt = hstack([flts, flts2])
# TODO add assertions for multikey Categoricals
c_str = Categorical(fa_str)
c_flt = Categorical(fa_flt)
# TODO move these into SDS save / load tests
paths = [r'riptable/tests/temp/ds1.sds', r'riptable/tests/temp/ds2.sds']
ds1 = Dataset(
{
'mkcat': c1,
'strcat': c1_str,
'fltcat': c1_flt,
'strfa': strs,
'fltfa': flts,
}
)
ds2 = Dataset(
{
'mkcat': c2,
'strcat': c2_str,
'fltcat': c2_flt,
'strfa': strs2,
'fltfa': flts2,
}
)
ds1.save(paths[0])
ds2.save(paths[1])
# normal dataset hstack
hstack_ds = hstack([ds1, ds2])
assert isinstance(hstack_ds, Dataset)
# dataset hstack from load
stack_load_ds = load_sds(paths, stack=True)
assert isinstance(stack_load_ds, PDataset)
# multikey cat hstack
hstack_mkcats = hstack([c1, c2])
assert isinstance(hstack_mkcats, Categorical)
# normal array hstack
hstack_strs = hstack([strs, strs2])
hstack_flts = hstack([flts, flts2])
# single cat hstack
hstack_cstrs = hstack([c1_str, c2_str])
assert isinstance(hstack_cstrs, Categorical)
hstack_cflts = hstack([c1_flt, c2_flt])
assert isinstance(hstack_cflts, Categorical)
assert bool(np.all(hstack_strs == hstack_cstrs.expand_array))
assert bool(np.all(hstack_flts == hstack_cflts.expand_array))
mktup = [*hstack_mkcats.category_dict.values()]
assert bool(np.all(hstack_mkcats._expand_array(mktup[0]) == fa_str))
assert bool(np.all(hstack_mkcats._expand_array(mktup[1]) == fa_flt))
mktup2 = [*stack_load_ds.mkcat.category_dict.values()]
assert bool(np.all(stack_load_ds.mkcat._expand_array(mktup2[0]) == fa_str))
assert bool(np.all(stack_load_ds.mkcat._expand_array(mktup2[1]) == fa_flt))
mktup3 = [*hstack_ds.mkcat.category_dict.values()]
assert bool(np.all(hstack_ds.mkcat._expand_array(mktup3[0]) == fa_str))
assert bool(np.all(hstack_ds.mkcat._expand_array(mktup3[1]) == fa_flt))
for p in paths:
os.remove(p)
# TO TEST:
# regular python Enum
# apply / apply_dataset, etc.
# def test_sort_copy(self):
# c = Categorical(np.random.choice(['a','b','c'], 15))
# d = c.sort_copy()
# c = Categorical([np.random.choice(['a','b','c'], 15), np.random.randint(0,3,15)])
# d = c.sort_copy()
# ----------------------------------------------------------
# def test_str_repr(self):
# '''
# SJK: We're still in the early stages of deciding how to print out or summarize a categorical in the workspace.
# Comment it out if repr or str changes, and I will fix up.
# '''
# # no break
# input = ['b', 'b', 'b', 'a', 'b', 'b']
# str_string = ', '.join(input)
# repr_string = "Categorical(["+str_string+"])"
# c = Categorical(input)
# self.assertEqual(str(c),str_string, msg=f"__str__ did not produce the correct string {str_string} for categorical. got {str_string} instead")
# self.assertEqual(c.__repr__(),repr_string, msg=f"__repr__ did not produce the correct string {str_string} for categorical. got {str_string} instead")
# # add break
# slice_size = 5
# input = ['b', 'b', 'b', 'a', 'b', 'b', 'b', 'b', 'b', 'a', 'b', 'b', 'c', 'c']
# str_string = ', '.join(input[:slice_size]+['...']+input[-slice_size:])
# repr_string = "Categorical(["+str_string+"])"
# c = Categorical(input)
# self.assertEqual(str(c),str_string, msg=f"__str__ did not produce the correct string {str_string} for categorical. got {str_string} instead")
# self.assertEqual(c.__repr__(),repr_string, msg=f"__repr__ did not produce the correct string {str_string} for categorical. got {str_string} instead")
def test_as_string_array(self):
# SJK 10/4/2018 - as string array now returns bytes OR unicode (whatever type the string based categorical is holding)
f = np.array([b'b', b'b', b'b', b'a', b'b', b'b'])
c = Categorical(f)
is_equal = bool(np.all(c.as_string_array == f))
assert isinstance(c.as_string_array, FastArray), f"Categorical did not return a fastarray in as_string_array"
assert (
is_equal
), f"Categorical returned an incorrect string array {c.as_string_array} view of itself. Expected {f}"
def test_indexing_numeric(self):
c = Cat([1.1, 2.2, 3.3])
result = c['2.2']
assert np.all(result == [False, True, False])
def test_fill_forward(self):
fa = FA([1., np.nan, 1.])
c = Cat([1,1,1])
c.fill_forward(fa, inplace=True)
assert np.all(fa == [1., 1., 1.])
# TODO pytest parameterize `compare_func_names`
def test_all_compare_tests(self):
# with scalar
# cat(unicode)
i = 2
c1 = Categorical(three_ints)
if ShowCompareInfo:
print("Categorical:", c1)
if ShowCompareInfo:
print("Compare unicode to int scalar: 2")
self.compare_cat_test(c1, compare_func_names, int_success, i)
# cat(unicode) / unicode, unicode list
i = "AMZN\u2082"
c3 = Categorical(three_unicode)
if ShowCompareInfo:
print("Categorical:", c3)
if ShowCompareInfo:
print("Compare unicode cat to unicode string")
self.compare_cat_test(c3, compare_func_names, int_success, i)
if ShowCompareInfo:
print("Compare to list of unicode string")
self.compare_cat_test(c3, compare_func_names, int_success, [i])
if ShowCompareInfo:
print("Compare to a numpy array of unicode string")
self.compare_cat_test(c3, compare_func_names, int_success, np.array([i]))
# cat(bytes) / bytes, bytes list
i = b'b'
c4 = Categorical(three_bytes)
if ShowCompareInfo:
print("Categorical:", c4)
if ShowCompareInfo:
print("Compare bytes cat to bytestring")
self.compare_cat_test(c4, compare_func_names, int_success, i)
if ShowCompareInfo:
print("Compare to bytestring in list")
self.compare_cat_test(c4, compare_func_names, int_success, [i])
if ShowCompareInfo:
print("Compare to bytestring in numpy array")
self.compare_cat_test(c4, compare_func_names, int_success, np.array([i]))
# cat(bytes) / unicode, unicode list
i = "b"
c5 = Categorical(three_bytes)
if ShowCompareInfo:
print("Categorical:", c5)
if ShowCompareInfo:
print("Compare bytes cat to unicode string")
self.compare_cat_test(c5, compare_func_names, int_success, i)
if ShowCompareInfo:
print("Compare to unicode string in list")
self.compare_cat_test(c5, compare_func_names, int_success, [i])
if ShowCompareInfo:
print("Compare to unicode string in numpy array")
self.compare_cat_test(c5, compare_func_names, int_success, np.array([i]))
# equal categoricals (same dictionary)
# cat(bytes) / cat(bytes)
if ShowCompareInfo:
print("Compare two equal categoricals:")
if ShowCompareInfo:
print("Both from byte lists:")
c1 = Categorical(three_bytes)
c2 = Categorical(three_bytes)
if ShowCompareInfo:
print("cat1:", c1)
if ShowCompareInfo:
print("cat2:", c2)
self.compare_cat_test(c1, compare_func_names, same_success, c2)
# cat(unicode) / cat(unicode)
if ShowCompareInfo:
print("Both from unicode lists:")
c1 = Categorical(three_unicode)
c2 = Categorical(three_unicode)
if ShowCompareInfo:
print("cat1:", c1)
if ShowCompareInfo:
print("cat2:", c2)
self.compare_cat_test(c1, compare_func_names, same_success, c2)
# cat(unicode) / cat(bytes)
if ShowCompareInfo:
print("unicode/bytes list")
c1 = Categorical(["a", "b", "c"])
c2 = Categorical(three_bytes)
if ShowCompareInfo:
print("cat1:", c1)
if ShowCompareInfo:
print("cat2:", c2)
self.compare_cat_test(c1, compare_func_names, same_success, c2)
# unequal categoricals (same dictionary)
# cat(bytes) / cat(bytes)
if ShowCompareInfo:
print("Compare two unequal categoricals (same dict):")
if ShowCompareInfo:
print("both bytes")
c1 = Categorical([0, 1, 0], three_bytes)
c2 = Categorical([2, 1, 2], three_bytes)
if ShowCompareInfo:
print("cat1:", c1)
if ShowCompareInfo:
print("cat2:", c2)
self.compare_cat_test(c1, compare_func_names, diff_success, c2)
# cat(unicode) / cat(unicode)
if ShowCompareInfo:
print("both unicode")
c1 = Categorical([0, 1, 0], three_unicode)
c2 = Categorical([2, 1, 2], three_unicode)
if ShowCompareInfo:
print("cat1:", c1)
if ShowCompareInfo:
print("cat2:", c2)
self.compare_cat_test(c1, compare_func_names, diff_success, c2)
## cat(bytes) / int list (matching)
# if ShowCompareInfo: print("Compare categorical to matching int list")
# if ShowCompareInfo: print("bytes")
# i = [1,2,3]
# c1 = Categorical(three_bytes)
# self.compare_cat_test(c1,compare_func_names,same_success,i)
## cat(unicode) / int list (matching)
# if ShowCompareInfo: print("unicode")
# c1 = Categorical(three_unicode)
# self.compare_cat_test(c1,compare_func_names,same_success,i)
## cat(bytes) / int list (non-matching)
# if ShowCompareInfo: print("Compare categorical to non-matching int list")
# if ShowCompareInfo: print("bytes")
# i = [3,2,1]
# c1 = Categorical(three_bytes)
# self.compare_cat_test(c1,compare_func_names,int_success,i)
## cat(unicode) / int list(non-matching)
# if ShowCompareInfo: print("unicode")
# c1 = Categorical(three_unicode)
# self.compare_cat_test(c1,compare_func_names,int_success,i)
# def cat_slicing(self):
# three_unicode =FA(["AAPL\u2080","AMZN\u2082","IBM\u2081"])
# three_bytes = FA([b'a',b'b',b'c'])
# num_rows=8
# idx_size=15
# get_item_dicts = {
# "single_slices" : {
# ":2" : slice(None,2,None),
# "-2:": slice(-2,None,None),
# "2:5": slice(2,5,None),
# "5:" : slice(5,None,None),
# ":" : slice(None,None,None)
# },
# "bool_arrays" : {
# "python_bool" : [True, False, True, False, False, True, True, True, False, True, False, False, True, False, True],
# "numpy_bool" : np.array([True, False, True, False, False, True, True, True, False, True, False, False, True, False, True])
# },
# "int_indices" : { "int_idx_size"+str(idx_size) : np.random.randint(low=0,high=num_rows,size=idx_size) for idx_size in range(1,num_rows) }
# }
# failures = 0
# idx_list = np.random.randint(low=0,high=8,size=15)
# s_list = np.array([b'adam',b'bob',b'charlie',b'david',b'edward',b'frank',b'greg',b'harold'])
# c = Categorical(idx_list, s_list)
# for key, test_dict in get_item_dicts.items():
# print("\n\n"+key)
# for call_str, val in test_dict.items():
# success = s_list[idx_list[val]]
# if np.all(c[val].as_string_array == success):
# message = "success"
# else:
# message = "failure"
# failures += 1
# print(call_str, message)
# print("Tests complete with",failures,"errors")
# return c
@pytest.mark.xfail(
reason="RIP-215 - lead to inconsistent Categorical state; please add hypothesis tests when resolved."
)
def test_category_add(self):
cat = Categorical(list("bbcdebc"))
e = "a"
cat.category_add(e)
assert e in cat, "expect the added category to be added to the Categorical"
assert e in cat._categories, "expect the added category to be added to the Categorical._categories"
assert e in cat.category_array, "expect the added category to be added to the Categorical.category_array"
assert e in cat.category_dict, "expect the added category to be added to the Categorical.category_dict"
@pytest.mark.xfail(
reason="RIP-215 - lead to inconsistent Categorical state; please add hypothesis tests when resolved."
)
def test_category_remove(self):
cat = Categorical(list("bbcdebc"))
e = cat[0]
cat.category_remove(e)
assert e not in cat, "expect the removed category to be removed from the Categorical"
assert e not in cat._categories, "expect the removed category to be removed from the Categorical._categories"
assert (
e not in cat.category_array
), "expect the removed category to be removed from the Categorical.category_array"
assert (
e not in cat.category_dict
), "expect the removed category to be removed from the Categorical.category_dict"
# TODO move this to testing utils
def compare_cat_test(self, cat, compare_func_names, success_bools, i):
for fname, success in zip(compare_func_names, success_bools):
func = getattr(cat, fname)
result = func(i)
assert np.all(result == success), f'fail on {fname} {cat} {i}'
if ShowCompareInfo:
if np.all(result == success):
message = "succeeded"
else:
message = "failed"
print(fname, message)
def test_duplicated(self):
result = Cat([2, 3, 2], list('qwery')).duplicated()
assert np.all(result == FA([False, False, True]))
def test_cat_copy(self):
# add deep copy for enum, single, multi
x = arange(6, dtype=uint16) // 2
c = Cat(x, {0: 'Run', 1: 'Stop', 2: 'Start'}, dtype=uint16)
c[1] = 'Start'
a = c.copy()
d = a[:5]
a[1] = 'Run'
b = a[:5]
assert a._fa[1] == 0
assert b._fa[1] == 0
assert c._fa[1] == 2
assert d._fa[1] == 0
def test_assinglekey(self):
c = Cat([1, 2, 1, 2, 1, 2], {'Sunny': 1, 'Thunderstorms': 2})
# insert bad value
c._fa[3] = 17
c1 = c.as_singlekey(ordered=False)
c2 = c.as_singlekey(ordered=True)
assert np.all(c1.expand_array == c2.expand_array)
c = Cat([-1, -2, -1, -2, -1, -2], {'Sunny': -1, 'Thunderstorms': -2})
c._fa[3] = 17
c3 = c.as_singlekey(ordered=False)
c2 = c.as_singlekey(ordered=True)
assert np.all(c1.expand_array == c2.expand_array)
assert np.all(c3.expand_array == c2.expand_array)
# Cannot use the pytest.mark.parameterize decorator within classes that inherit from unittest.TestCase.
# Will need to migrate for unittest to pytest and fold the following categorical tests into Categorical_Test.
@pytest.mark.parametrize(
"categoricals",
[
# Categorical constructed from python list data
pytest.param(
[
Categorical(data)
for data in get_categorical_data_factory_method([CategoryMode.StringArray, CategoryMode.NumericArray])
],
id="cat_with_list_values",
),
# Categorical constructed from numpy array
pytest.param(
[
Categorical(np.array(data))
for data in get_categorical_data_factory_method([CategoryMode.StringArray, CategoryMode.NumericArray])
],
id="cat_with_np_array_values",
),
# Categorical constructred from riptable fast array
pytest.param(
[
Categorical(rt.FastArray(data))
for data in get_categorical_data_factory_method([CategoryMode.StringArray, CategoryMode.NumericArray])
],
id="cat_with_rt_fastarray_values",
),
# failed test cases
pytest.param(
[Categorical(data) for data in get_categorical_data_factory_method(CategoryMode.MultiKey)],
marks=[
pytest.mark.xfail(
reason="RIP-410 - Bug for MultiKey Categoricals: AttributeError: 'Categorical' object has no attribute 'ismultikey_labels'"
)
],
id="cat_with_tuple_values",
),
],
)
def test_one_hot_encode(categoricals):
for categorical in categoricals:
col_names, encoded_arrays = categorical.one_hot_encode()
category_array = categorical.category_array.astype('U')
# Test 1.1 The col_names are the same as the category array.
assert not set(category_array).symmetric_difference(set(col_names)), (
f"The column names should be the same as the names in the category array",
f"category array {category_array}\ncolumn names {col_names}",
)
# Test 1.2 The encoded_arrays dtypes are consistent with one another.
encoded_arrays_dtypes = set([fa.dtype for fa in encoded_arrays])
assert (
len(encoded_arrays_dtypes) == 1
), f"Encoded array dtypes should be consistent, got {encoded_arrays_dtypes}"
# todo for each category, assert the mask of the categorical is in the encoded_arrays
@pytest.mark.parametrize(
"categoricals",
[
# Categorical constructed from python list data
pytest.param(
[Categorical(data) for data in get_categorical_data_factory_method([CategoryMode.StringArray])],
id="cat_with_list_values",
),
# Categorical constructed from numpy array
pytest.param(
[Categorical(np.array(data)) for data in get_categorical_data_factory_method([CategoryMode.StringArray])],
id="cat_with_np_array_values",
),
# Categorical constructred from riptable fast array
pytest.param(
[
Categorical(rt.FastArray(data))
for data in get_categorical_data_factory_method([CategoryMode.StringArray])
],
id="cat_with_rt_fastarray_values",
),
],
)
def test_shift_cat(categoricals):
# todo Handle numeric invalid types for categoricals with values other than strings.
filtered_name = rt.rt_enum.FILTERED_LONG_NAME.encode("utf-8")
for categorical in categoricals:
cat_len = len(categorical)
for i in range(-cat_len + 1, cat_len): # exhaustive shift of all Categorical values.
# shift the categorical i-places
shift_cat = categorical.shift_cat(i)
# The category array should remain unchanged.
assert_array_equal(shift_cat.category_array, categorical.category_array)
# The underlying FastArray should have the items shifted to the i-th position.
if i > 0: # shift forwards case
assert_array_equal(
shift_cat._fa[i:], categorical._fa[:-i], f"FastArray items should be shifted by {i} postions.",
)
# The Categorical should have the values shifted to the i-th position.
cat_values, shift_cat_values = (
categorical.expand_array,
shift_cat.expand_array,
)
assert_array_equal(
shift_cat_values[i:], cat_values[:-i], f"Categorical values should be shifted by {i} positions.",
)
# The underlying FastArray should have the first i-items to be the invalid value.
# The Categorical values should have the first i-items be the filtered or invalid name.
# Need to handle other invalid values and other Categorical base indexing.
assert_array_equal(
shift_cat_values[:i],
np.full(i, filtered_name),
f"Shifted Categorical values up to {i}-th position should be '{filtered_name}'.",
)
assert_array_equal(
shift_cat._fa[:i],
np.zeros(i),
f"Shifted Categorical underlying FastArray items up to {i}-th position should be the invalid value 0.",
)
elif i < 0: # shifted backwards case
i = abs(i) # slicing arithmetic based on positional value of i
assert_array_equal(
shift_cat._fa[: cat_len - i],
categorical._fa[i:],
f"FastArray items should be shifted by -{i} postions.",
)
cat_values, shift_cat_values = (
categorical.expand_array,
shift_cat.expand_array,
)
assert_array_equal(
shift_cat_values[: cat_len - i],
cat_values[i:],
f"Categorical values should be shifted by -{i} positions.",
)
assert_array_equal(
shift_cat_values[-i:],
np.full(i, filtered_name),
f"Shifted Categorical values up to -{i}-th position should be '{filtered_name}'.",
)
assert_array_equal(
shift_cat._fa[-i:],
np.zeros(i),
f"Shifted Categorical underlying FastArray items up to -{i}-th position should be the invalid value 0.",
)
elif i == 0: # zero-th shift case
# test for equality
assert_array_equal(shift_cat.category_array, categorical.category_array)
assert_array_equal(shift_cat._fa, categorical._fa)
cat_values, shift_cat_values = (
categorical.expand_array,
shift_cat.expand_array,
)
assert_array_equal(shift_cat_values, cat_values)
# shift overflow for backward and forward case up to two values
for i in list(range(-cat_len - 2, -cat_len)) + list(range(cat_len, cat_len + 2)):
shift_cat = categorical.shift_cat(i)
assert_array_equal(shift_cat.category_array, categorical.category_array)
# Investigate possible bug with expanding Categorical values. E.g.:
# given:
# Categorical([a, a, a, a, a, a, a, a, a, a]) Length: 10
# FastArray([1, 1, 1, 1, 1, 1, 1, 1, 1, 1], dtype=int8) Base Index: 1
# FastArray([b'a'], dtype='|S1') Unique count: 1
# shifted categorical
# Categorical([Filtered, Filtered, Filtered, Filtered, Filtered, Filtered, Filtered, Filtered, Filtered, Filtered]) Length: 10
# FastArray([0, 0, 0, 0, 0, 0, 0, 0, 0, 0], dtype=int8) Base Index: 1
# FastArray([b'a'], dtype='|S1') Unique count: 1
# got
# E x: FastArray([b'Filtered', b'Filtered', b'Filtered', b'Filtered',
# E b'Filtered', b'Filtered', b'Filtered', b'Filtered',
# E b'Filtered', b'a'], dtype='|S8')
# E y: array([b'Filtered', b'Filtered', b'Filtered', b'Filtered', b'Filtered',
# E b'Filtered', b'Filtered', b'Filtered', b'Filtered', b'Filtered'],
# E dtype='|S8')
# Expected all values to be b'Filtered', but saw b'a'.
# todo assert_array_equal(shift_cat_values, np.full(cat_len, filtered_name), f"Overflow shifted Categorical values. All values are expected to be invalid '{filtered_name}'.")
assert_array_equal(
shift_cat._fa,
np.zeros(cat_len),
f"Overflow shifted Categorical underlying FastArray items. All values are expected to be invalid value 0.",
)
@pytest.mark.parametrize(
# TODO - add base 0 and base 1 indexing w/ expectations
"categoricals",
[
# Categorical constructed from python list data
pytest.param(
[Categorical(data) for data in get_categorical_data_factory_method([CategoryMode.StringArray])],
id="cat_with_list_values",
),
# Categorical constructed from numpy array
pytest.param(
[Categorical(np.array(data)) for data in get_categorical_data_factory_method([CategoryMode.StringArray])],
id="cat_with_np_array_values",
),
# Categorical constructred from riptable fast array
pytest.param(
[
Categorical(rt.FastArray(data))
for data in get_categorical_data_factory_method([CategoryMode.StringArray])
],
id="cat_with_rt_fastarray_values",
),
],
)
@pytest.mark.parametrize("misc", [None, "INVALID"]) # TODO - add numeric values
@pytest.mark.parametrize("inplace", [False, True])
def test_shrink(categoricals, misc, inplace):
for categorical in categoricals:
cat = categorical.copy(deep=True) # deep copy so test data remains unchanged with inplace shrinks
# Test 1 Shrink with empty values.
# Shrink to empty categories.
shrink_cat = cat.shrink([], misc=misc, inplace=inplace)
# Type is preserved after shrinking.
assert isinstance(shrink_cat, Categorical), "shrink_cat should be a Categorical."
if misc is None:
# For base index 1 Categorical, the underlying FastArray should be all zeros.
assert_array_equal(shrink_cat._fa, np.zeros(len(cat)))
# The Categorical categories should be empty.
expected_category_array = np.empty(0)
assert_array_equal(
shrink_cat.category_array, expected_category_array, f"Category dictionary values should be empty.",
)
for arr in shrink_cat.category_dict.values():
assert_array_equal(
arr, expected_category_array, f"Category dictionary values should be empty.",
)
# TODO expanding shrink categorical does not return original types invalid value; instead it returns nans
# N.B, when shrinking, the category array type changes to float64
# E x: FastArray([nan])
# E y: array([b'Filtered'], dtype='|S8')
# assert_array_equal(shrink_cat.expand_array, np.full(len(cat), filtered_name), f"Given empty values, shrink categorical values should all be invalid '{filtered_name}'.")
else: # single categories being the specified misc
# TODO - consider any constraints to assert on for the dtype?
# The invalid value based on the dtype: e.g., for U32 its -2147483646
# assert_array_equal(shrink_cat._fa, InvalidValuesForDtype)
# assert_array_equal(shrink_cat.expand_array, InvalidValuesForDtypeExpanded)
# The categories should only contain the misc value.
expected_category_array = np.array(misc)
assert_array_equal(
shrink_cat.category_array,
expected_category_array,
f"Category array should only contain the '{misc}' category.",
)
for arr in shrink_cat.category_dict.values():
assert_array_equal(
arr,
expected_category_array,
f"Category dictionary values should only contain the '{misc}' category.",
)
# Test 2 Shrink with same categories
cat = categorical.copy(deep=True)
# Shrink to all the same categories.
shrink_cat = cat.shrink(cat.category_array, misc=misc, inplace=inplace)
# Type is preserved after shrinking.
assert isinstance(shrink_cat, Categorical), "shrink_cat should be a Categorical."
if misc is None: # TODO handle the misc not None case
shrink_cat_values, cat_values = shrink_cat.expand_array, cat.expand_array
assert_array_equal(shrink_cat_values, cat_values)
assert_array_equal(shrink_cat._fa, cat._fa)
assert_array_equal(shrink_cat.category_array, cat.category_array)
for arr, expected_arr in zip(shrink_cat.category_dict.values(), cat.category_dict.values()):
assert_array_equal(arr, expected_arr)
# TODO Test 3 Shrink with subset of categories
cat = categorical.copy(deep=True)
# Shrink to all the same categories.
n = int(len(cat) / 2)
shrink_cat = cat.shrink(cat.category_array[:n], misc=misc, inplace=inplace)
# Type is preserved after shrinking.
assert isinstance(shrink_cat, Categorical), "shrink_cat should be a Categorical."
@pytest.mark.parametrize(
"categoricals",
[
# TODO - test categorical construction using numpy and riptable arrays as a separate test
# Categorical constructed from python list data
pytest.param([Categorical(data) for data in get_categorical_data_factory_method()], id="cat_with_list_values",),
# Categorical constructed from numpy array
pytest.param(
[
Categorical(np.array(data))
for data in get_categorical_data_factory_method([CategoryMode.StringArray, CategoryMode.NumericArray])
],
id="cat_with_np_array_values",
),
# Categorical constructred from riptable fast array
pytest.param(
[
Categorical(rt.FastArray(data))
for data in get_categorical_data_factory_method([CategoryMode.StringArray, CategoryMode.NumericArray])
],
id="cat_with_rt_fastarray_values",
),
],
)
def test_sds(categoricals, tmpdir):
dir = tmpdir.mkdir("test_categorical_sds")
for i, cat in enumerate(categoricals):
name = "categorical_" + str(i)
p = str(dir.join(name))
save_sds(p, cat)
cat2 = load_sds(p)
# Test 1 Saved and loaded categoricals should be the same.
# TODO vary the meta version optional parameter when calling Categorical._load_from_sds_meta_data
assert isinstance(cat2, Categorical)
assert_array_equal(cat2._fa, cat._fa)
if not cat.ismultikey: # MultiKey Categorical's do not support category_array operation
assert_array_equal(cat2.category_array, cat.category_array)
for actual, expected in zip(cat2.category_dict.values(), cat.category_dict.values()):
assert_array_equal(actual, expected)
cat2_values, cat_values = cat2.expand_array, cat.expand_array
assert_array_equal(cat2_values, cat_values)
# Test 2 As and from meta data Categoricals should be the same.
cat3 = Categorical._from_meta_data(*cat._as_meta_data(name=name))
# Saved and loaded categoricals should be the same.
assert isinstance(cat3, Categorical)
assert_array_equal(cat3._fa, cat._fa)
if not cat.ismultikey: # MultiKey Categorical's do not support category_array operation
assert_array_equal(cat3.category_array, cat.category_array)
for actual, expected in zip(cat3.category_dict.values(), cat.category_dict.values()):
assert_array_equal(actual, expected)
cat3_values, cat_values = cat3.expand_array, cat.expand_array
assert_array_equal(cat3_values, cat_values)
@pytest.mark.parametrize(
"categoricals",
[
# TODO handle CategoryMode IntEnum and Default
[
Categorical(data)
for data in get_categorical_data_factory_method([CategoryMode.StringArray, CategoryMode.NumericArray])
]
+ [
Categorical(data, base_index=0)
for data in get_categorical_data_factory_method([CategoryMode.StringArray, CategoryMode.NumericArray])
]
],
)
def test_from_bin(categoricals):
for cat in categoricals:
cat_arr_len = len(cat.category_array)
# Test 1 All bin values are in the category array.
if cat.base_index == 0:
for i in range(cat_arr_len):
assert cat.from_bin(i) in cat.category_array
elif cat.base_index == 1:
for i in range(1, cat_arr_len + 1):
assert cat.from_bin(i) in cat.category_array
else:
raise ValueError(f"Unhandled Categorical base index {cat.base_index}")
# Test 2 Handling of invalid input types: base_index and bin.
# The bin is not an integer.
with pytest.raises(TypeError):
cat.from_bin(str(i))
cat.from_bin(float(i))
# Bin value out of range.
with pytest.raises(ValueError):
cat.from_bin(-1)
if cat.base_index == 0:
cat.from_bin(cat_arr_len)
elif cat.base_index == 1:
cat.from_bin(0)
cat.from_bin(cat_arr_len + 1)
else:
raise ValueError(f"Unhandled Categorical base index {cat.base_index}")
# The base index is None.
cat.grouping._base_index = None
with pytest.raises(TypeError):
cat.from_bin(1)
@pytest.mark.parametrize("cat", get_all_categorical_data())
def test_argsort(cat):
assert_array_equal(
cat.argsort(),
np.argsort(cat._fa),
"Categorical argsort should be equivalent to the argsort of the underlying FastArray",
)
@pytest.mark.parametrize(
"cats",
[
pytest.param(
[
Categorical(data)
for data in get_categorical_data_factory_method([CategoryMode.StringArray, CategoryMode.NumericArray])
]
),
pytest.param(
[Categorical(data) for data in get_categorical_data_factory_method(CategoryMode.MultiKey)],
marks=[
pytest.mark.xfail(reason="NotImplementedError: Add categories not supported for MultiKey Categoricals")
],
),
],
) # TODO parameterize across base index 0 and 1
def test_auto_add(cats):
for cat in cats:
alpha, beta = "alpha", "beta"
first_index, last_index = 0, len(cat) - 1
# Test 1 auto_add_on will allow addition of a category if the Categorical is unlocked,
# otherwise an error is raised.
cat.auto_add_on()
cat.unlock() # Categorical is unlocked
# Test 1.1 When unlocked and attempting to add a category, the categories should be added.
# set the first and last categories
cat[first_index] = cat[last_index] = alpha
# auto_add_on and unlock should not allow setting beyond the first and last index of categories
with pytest.raises(IndexError): # index out of bounds
cat[first_index - 1] = alpha
cat[last_index + 1] = alpha
# category is added at specified index
first_category = cat.category_array[cat._fa[first_index] - 1]
# TODO normalize the category_array value, which is sometimes a numpy str_ or bytes_ to an ascii and compare
# assert cat.category_array[cat._fa[first_index]-1] == alpha
# assert at.category_array[cat._fa[last_index]-1] == alpha
# added category is in category array and dictionary
assert alpha in cat.category_array
for categories in cat.category_dict.values():
assert alpha in categories
# Test 1.2 When locked and attempting to add a category, an error is raised and the categories should not be added.
cat.lock() # Categorical is locked
with pytest.raises(IndexError): # cannot add a category since index is locked
cat[first_index] = beta
assert beta not in cat.category_array
for categories in cat.category_dict.values():
assert beta not in categories
# Test 2 auto_add_off will prevent category assignment of non-existing categories and raise an error
cat.auto_add_off()
# Test 2.1 Unlocked case
cat.unlock() # Categorical is unlocked
with pytest.raises(ValueError): # cannot automatically add categories while auto_add_categories is False
cat[first_index] = beta
# Test 2.2 Locked case
cat.lock()
with pytest.raises(IndexError): # cannot add a category since index is locked
cat[first_index] = beta
@pytest.mark.xfail(reason="rt_numpy.unique() needs to handles multikey categoricals")
def test_multikey_categorical_unique():
c = Categorical([arange(3), FA(list('abc'))])
assert len(c.unique()) == c.nunique()
@pytest.mark.parametrize("values", [list_bytes, list_unicode, list_true_unicode])
def test_categorical_convert(values):
categories = list(set(values))
# pd_c is a pandas Categorical with a missing category.
# pandas Categorical will designate the values with a missing category by -1.
pd_c = pd.Categorical(values, categories=categories[:-1])
# The output of categorical_convert, when applied to a pandas Categorical, can be used to
# construct a riptable Categorical. We test that this handles missing categories correctly.
rt_values, rt_categories = rt.categorical_convert(pd_c)
cat = rt.Categorical(rt_values, categories=rt_categories)
# The invalid category should not be in the Categorical.
missing_category = categories[-1]
assert missing_category not in cat
assert missing_category not in cat._categories
assert missing_category not in cat.category_array
assert missing_category not in cat.category_dict[next(iter(cat.category_dict))] # values of first key
# All other category values should be in the Categorical.
for e in categories[:-1]:
# assert e in cat # uncomment when test_categorical_convert_xfail is fixed
assert e in cat._categories
assert e in cat.category_array
assert e in cat.category_dict[next(iter(cat.category_dict))] # values of first key
@pytest.mark.xfail(reason="RIP-396 - category not in Categorical, but is in Categorical.category_array")
@pytest.mark.parametrize("values", [list_bytes,])
def test_categorical_convert_xfail(values):
categories = list(set(values))
# pd_c is a pandas Categorical with a missing category.
# pandas Categorical will designate the values with a missing category by -1.
pd_c = pd.Categorical(values, categories=categories[:-1])
rt_values, rt_categories = rt.categorical_convert(pd_c)
cat = rt.Categorical(rt_values, categories=rt_categories)
# All other category values should be in the Categorical.
for e in categories[:-1]:
assert e in cat
def test_build_dicts_enum():
str_to_int, int_to_str = Categories.build_dicts_enum(LikertDecision)
codes = list(str_to_int.values()) * 2
c = Categorical(codes, categories=LikertDecision)
c2 = Categorical(codes, categories=str_to_int)
c3 = Categorical(codes, categories=int_to_str)
# c is the our oracle Categorical.
# Categoricals constructed from any of the dictionaries built by build_dicts_enum
# should construct the same Categorical as c.
assert_array_equal(c, c2)
assert_array_equal(c, c3)
@pytest.mark.parametrize("values", [list("abcdef"), [b"a", b"b", b"c", b"d", b"e", b"f"]])
def test_build_dicts_python(values):
# int
d = {k: v for k, v in enumerate(values)}
str_to_int, int_to_str = Categories.build_dicts_python(d)
codes = list(d.keys()) * 2
c = Categorical(codes, categories=d)
c2 = Categorical(codes, categories=str_to_int)
c3 = Categorical(codes, categories=int_to_str)
# c is the our oracle Categorical.
# Categoricals constructed from any of the dictionaries built by build_dicts_python
# should construct the same Categorical as c.
assert_array_equal(c, c2)
assert_array_equal(c, c3)
@pytest.mark.parametrize(
"a,b,a_in_b,b_in_a",
[
pytest.param(Cat(list('abc')), Cat(list('a')), FA([True, False, False]), FA([True]), id='single_key_overlap'),
pytest.param(
Cat([FA(list('abc')), FA([1,2,3])]),
Cat([FA(list('a')), FA([1])]),
FA([True, False, False]),
FA([True]),
id='single_multikey_overlap'
),
pytest.param(
Cat([FA(list('abc')), FA([1,2,3])]),
Cat([FA(list('ab')), FA([1,2])]),
FA([True, True, False]),
FA([True, True]),
id='two_multikey_overlap'
),
pytest.param(
Cat([FA(list('abcde')), FA([1,2,3,4,5])]),
Cat([FA(list('dc')), FA([4,5])]),
FA([False, False, False, True, False]),
FA([True, False]),
id='single_multikey_overlap2'
),
pytest.param(
Cat([FA(list('abcde')), FA([1,2,3,4,5])]),
Cat([FA(list('aba')), FA([1,2,1])]),
FA([True, True, False, False, False]),
FA([True, True, True]),
id='repeated_key_multikey_overlap'
),
pytest.param(
Cat([FA(list('abcdeab')), FA([1,2,3,4,5,1,6])]),
Cat([FA(list('aba')), FA([1, 2, 1])]),
FA([True, True, False, False, False, True, False]),
FA([True, True, True]),
id='repeated_key_multikey_overlap2'
),
]
)
def test_multikey_categorical_isin(a, b, a_in_b, b_in_a):
assert_array_equal(a_in_b, a.isin(b))
assert_array_equal(b_in_a, b.isin(a))
# TODO this is a good candidate for a hypothesis test once the CategoricalStrategy is able to generate MultiKey Categoricals
f_msg = 'expected to be consistent with cat1.as_singlekey().isin(cat2.as_singlekey()) operation.'
assert_array_equal(a.as_singlekey().isin(b.as_singlekey()), a.isin(b), f_msg)
assert_array_equal(b.as_singlekey().isin(a.as_singlekey()), b.isin(a), f_msg)
_make_unique_test_cases = pytest.mark.parametrize('cat, expected', [
(rt.Cat([1, 1, 2, 2], ['a', 'a']), rt.Cat([1, 1, 1, 1], ['a'])),
(rt.Cat([2, 2, 2, 2], ['a', 'a']), rt.Cat([1, 1, 1, 1], ['a'])),
(rt.Cat([1, 2, 3, 3], ['a', 'a', 'b']), rt.Cat([1, 1, 2, 2], ['a', 'b'])),
(rt.Cat([0, 0, 1, 1], ['a', 'a'], base_index=0), rt.Cat([0, 0, 0, 0], ['a'], base_index=0)),
(rt.Cat([1, 1, 1, 1], ['a', 'a'], base_index=0), rt.Cat([0, 0, 0, 0], ['a'], base_index=0)),
(rt.Cat([0, 0, 1, 1], ['a', 'b'], base_index=0), rt.Cat([0, 0, 1, 1], ['a', 'b'], base_index=0)),
(rt.Cat([1, 1, 2, 2, 3], [99, 99, 101], ), rt.Cat([1, 1, 1, 1, 2], [99, 101])),
(rt.Cat([0, 0, 1, 1], [99, 99], base_index=0), rt.Cat([0, 0, 0, 0], [99], base_index=0)),
(rt.Cat([0, 0, 1, 1], [99, 101], base_index=0), rt.Cat([0, 0, 1, 1], [99, 101], base_index=0)),
(rt.Cat([0, 0, 1, 1, 2, 2], ['a', 'a'], ), rt.Cat([0, 0, 1, 1, 1, 1], ['a'], )),
(rt.Cat([0, 0, 1, 1, 2, 2, 3, 3], ['a', 'a', 'b'], ), rt.Cat([0, 0, 1, 1, 1, 1, 2, 2], ['a', 'b'], )),
])
@_make_unique_test_cases
def test_category_make_unique_not_inplace(cat, expected):
res = cat.category_make_unique()
assert (res == expected).all()
@pytest.mark.parametrize('base_index', [0, 1])
def test_category_make_unique_multikey(base_index):
c1 = Categorical(np.arange(10) % 2, ['a', 'a'], base_index=base_index)
c2 = Categorical(np.arange(10) % 3, ['a', 'b', 'c'], base_index=base_index)
cat = Categorical([c1, c2], base_index=base_index)
res = cat.category_make_unique()
assert list(cat) == list(res)
| [
"riptable.rt_enum.FILTERED_LONG_NAME.encode",
"os.remove",
"riptable.tests.utils.LikertDecision.__members__.values",
"riptable.FastArray",
"riptable.hstack",
"riptable.rt_categorical.Categories",
"pytest.mark.parametrize",
"riptable.tests.test_utils.get_all_categorical_data",
"riptable.rt_numpy.isno... | [((861, 876), 'riptable.rt_sds.SDSMakeDirsOn', 'SDSMakeDirsOn', ([], {}), '()\n', (874, 876), False, 'from riptable.rt_sds import SDSMakeDirsOn\n'), ((963, 992), 'riptable.FastArray', 'FastArray', (["[b'a', b'b', b'c']"], {}), "([b'a', b'b', b'c'])\n", (972, 992), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((1007, 1027), 'riptable.FastArray', 'FastArray', (['[1, 2, 3]'], {}), '([1, 2, 3])\n', (1016, 1027), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((110068, 110118), 'pytest.mark.parametrize', 'pytest.mark.parametrize', (['"""misc"""', "[None, 'INVALID']"], {}), "('misc', [None, 'INVALID'])\n", (110091, 110118), False, 'import pytest\n'), ((110150, 110199), 'pytest.mark.parametrize', 'pytest.mark.parametrize', (['"""inplace"""', '[False, True]'], {}), "('inplace', [False, True])\n", (110173, 110199), False, 'import pytest\n'), ((121970, 122059), 'pytest.mark.xfail', 'pytest.mark.xfail', ([], {'reason': '"""rt_numpy.unique() needs to handles multikey categoricals"""'}), "(reason=\n 'rt_numpy.unique() needs to handles multikey categoricals')\n", (121987, 122059), False, 'import pytest\n'), ((122196, 122281), 'pytest.mark.parametrize', 'pytest.mark.parametrize', (['"""values"""', '[list_bytes, list_unicode, list_true_unicode]'], {}), "('values', [list_bytes, list_unicode, list_true_unicode]\n )\n", (122219, 122281), False, 'import pytest\n'), ((123592, 123705), 'pytest.mark.xfail', 'pytest.mark.xfail', ([], {'reason': '"""RIP-396 - category not in Categorical, but is in Categorical.category_array"""'}), "(reason=\n 'RIP-396 - category not in Categorical, but is in Categorical.category_array'\n )\n", (123609, 123705), False, 'import pytest\n'), ((123698, 123745), 'pytest.mark.parametrize', 'pytest.mark.parametrize', (['"""values"""', '[list_bytes]'], {}), "('values', [list_bytes])\n", (123721, 123745), False, 'import pytest\n'), ((128816, 128861), 'pytest.mark.parametrize', 'pytest.mark.parametrize', (['"""base_index"""', '[0, 1]'], {}), "('base_index', [0, 1])\n", (128839, 128861), False, 'import pytest\n'), ((27402, 27581), 'pytest.mark.xfail', 'pytest.mark.xfail', ([], {'reason': '"""20200416 This test was previously overridden by a later test in the file with the same name. Need to revisit and get back in a working state."""'}), "(reason=\n '20200416 This test was previously overridden by a later test in the file with the same name. Need to revisit and get back in a working state.'\n )\n", (27419, 27581), False, 'import pytest\n'), ((43126, 43220), 'pytest.mark.xfail', 'pytest.mark.xfail', ([], {'reason': '"""After invalid_set, the custom invalid value is not displayed."""'}), "(reason=\n 'After invalid_set, the custom invalid value is not displayed.')\n", (43143, 43220), False, 'import pytest\n'), ((47569, 47748), 'pytest.mark.xfail', 'pytest.mark.xfail', ([], {'reason': '"""20200416 This test was previously overridden by a later test in the file with the same name. Need to revisit and get back in a working state."""'}), "(reason=\n '20200416 This test was previously overridden by a later test in the file with the same name. Need to revisit and get back in a working state.'\n )\n", (47586, 47748), False, 'import pytest\n'), ((96527, 96657), 'pytest.mark.xfail', 'pytest.mark.xfail', ([], {'reason': '"""RIP-215 - lead to inconsistent Categorical state; please add hypothesis tests when resolved."""'}), "(reason=\n 'RIP-215 - lead to inconsistent Categorical state; please add hypothesis tests when resolved.'\n )\n", (96544, 96657), False, 'import pytest\n'), ((97218, 97348), 'pytest.mark.xfail', 'pytest.mark.xfail', ([], {'reason': '"""RIP-215 - lead to inconsistent Categorical state; please add hypothesis tests when resolved."""'}), "(reason=\n 'RIP-215 - lead to inconsistent Categorical state; please add hypothesis tests when resolved.'\n )\n", (97235, 97348), False, 'import pytest\n'), ((103431, 103476), 'riptable.rt_enum.FILTERED_LONG_NAME.encode', 'rt.rt_enum.FILTERED_LONG_NAME.encode', (['"""utf-8"""'], {}), "('utf-8')\n", (103467, 103476), True, 'import riptable as rt\n'), ((118721, 118747), 'riptable.tests.test_utils.get_all_categorical_data', 'get_all_categorical_data', ([], {}), '()\n', (118745, 118747), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((122508, 122558), 'pandas.Categorical', 'pd.Categorical', (['values'], {'categories': 'categories[:-1]'}), '(values, categories=categories[:-1])\n', (122522, 122558), True, 'import pandas as pd\n'), ((122785, 122813), 'riptable.categorical_convert', 'rt.categorical_convert', (['pd_c'], {}), '(pd_c)\n', (122807, 122813), True, 'import riptable as rt\n'), ((122825, 122876), 'riptable.Categorical', 'rt.Categorical', (['rt_values'], {'categories': 'rt_categories'}), '(rt_values, categories=rt_categories)\n', (122839, 122876), True, 'import riptable as rt\n'), ((123984, 124034), 'pandas.Categorical', 'pd.Categorical', (['values'], {'categories': 'categories[:-1]'}), '(values, categories=categories[:-1])\n', (123998, 124034), True, 'import pandas as pd\n'), ((124069, 124097), 'riptable.categorical_convert', 'rt.categorical_convert', (['pd_c'], {}), '(pd_c)\n', (124091, 124097), True, 'import riptable as rt\n'), ((124109, 124160), 'riptable.Categorical', 'rt.Categorical', (['rt_values'], {'categories': 'rt_categories'}), '(rt_values, categories=rt_categories)\n', (124123, 124160), True, 'import riptable as rt\n'), ((124346, 124389), 'riptable.rt_categorical.Categories.build_dicts_enum', 'Categories.build_dicts_enum', (['LikertDecision'], {}), '(LikertDecision)\n', (124373, 124389), False, 'from riptable.rt_categorical import Categories\n'), ((124444, 124489), 'riptable.Categorical', 'Categorical', (['codes'], {'categories': 'LikertDecision'}), '(codes, categories=LikertDecision)\n', (124455, 124489), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((124500, 124541), 'riptable.Categorical', 'Categorical', (['codes'], {'categories': 'str_to_int'}), '(codes, categories=str_to_int)\n', (124511, 124541), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((124552, 124593), 'riptable.Categorical', 'Categorical', (['codes'], {'categories': 'int_to_str'}), '(codes, categories=int_to_str)\n', (124563, 124593), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((124779, 124804), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['c', 'c2'], {}), '(c, c2)\n', (124797, 124804), False, 'from numpy.testing import assert_array_equal\n'), ((124810, 124835), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['c', 'c3'], {}), '(c, c3)\n', (124828, 124835), False, 'from numpy.testing import assert_array_equal\n'), ((125057, 125089), 'riptable.rt_categorical.Categories.build_dicts_python', 'Categories.build_dicts_python', (['d'], {}), '(d)\n', (125086, 125089), False, 'from riptable.rt_categorical import Categories\n'), ((125131, 125163), 'riptable.Categorical', 'Categorical', (['codes'], {'categories': 'd'}), '(codes, categories=d)\n', (125142, 125163), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((125174, 125215), 'riptable.Categorical', 'Categorical', (['codes'], {'categories': 'str_to_int'}), '(codes, categories=str_to_int)\n', (125185, 125215), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((125226, 125267), 'riptable.Categorical', 'Categorical', (['codes'], {'categories': 'int_to_str'}), '(codes, categories=int_to_str)\n', (125237, 125267), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((125455, 125480), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['c', 'c2'], {}), '(c, c2)\n', (125473, 125480), False, 'from numpy.testing import assert_array_equal\n'), ((125486, 125511), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['c', 'c3'], {}), '(c, c3)\n', (125504, 125511), False, 'from numpy.testing import assert_array_equal\n'), ((129083, 129127), 'riptable.Categorical', 'Categorical', (['[c1, c2]'], {'base_index': 'base_index'}), '([c1, c2], base_index=base_index)\n', (129094, 129127), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((2166, 2199), 'riptable.tests.utils.LikertDecision.__members__.keys', 'LikertDecision.__members__.keys', ([], {}), '()\n', (2197, 2199), False, 'from riptable.tests.utils import LikertDecision\n'), ((2530, 2579), 'pytest.skip', 'pytest.skip', (['"""This test needs to be implemented."""'], {}), "('This test needs to be implemented.')\n", (2541, 2579), False, 'import pytest\n'), ((2839, 2862), 'riptable.Categorical', 'Categorical', (['list_bytes'], {}), '(list_bytes)\n', (2850, 2862), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((3341, 3366), 'riptable.Categorical', 'Categorical', (['list_unicode'], {}), '(list_unicode)\n', (3352, 3366), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((4066, 4096), 'riptable.Categorical', 'Categorical', (['list_true_unicode'], {}), '(list_true_unicode)\n', (4077, 4096), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((10120, 10151), 'riptable.Categorical', 'Categorical', (['idx_list', 'str_list'], {}), '(idx_list, str_list)\n', (10131, 10151), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((10319, 10353), 'riptable.Categorical', 'Categorical', (['codes', 'LikertDecision'], {}), '(codes, LikertDecision)\n', (10330, 10353), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((14036, 14089), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'c', 'a', 'd']"], {'ordered': '(False)'}), "(['b', 'b', 'c', 'a', 'd'], ordered=False)\n", (14047, 14089), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((14206, 14252), 'riptable.FastArray', 'FastArray', (["[b'BB', b'BB', b'CC', b'AA', b'DD']"], {}), "([b'BB', b'BB', b'CC', b'AA', b'DD'])\n", (14215, 14252), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((14314, 14381), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'c', 'a', 'd']"], {'ordered': '(False)', 'base_index': '(0)'}), "(['b', 'b', 'c', 'a', 'd'], ordered=False, base_index=0)\n", (14325, 14381), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((14478, 14531), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'c', 'a', 'd']"], {'ordered': '(False)'}), "(['b', 'b', 'c', 'a', 'd'], ordered=False)\n", (14489, 14531), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((14656, 14707), 'riptable.FastArray', 'FastArray', (["[b'BB', b'BB', b'CC', b'AA', b'INVALID']"], {}), "([b'BB', b'BB', b'CC', b'AA', b'INVALID'])\n", (14665, 14707), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((14769, 14836), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'c', 'a', 'd']"], {'ordered': '(False)', 'base_index': '(0)'}), "(['b', 'b', 'c', 'a', 'd'], ordered=False, base_index=0)\n", (14780, 14836), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((14952, 15005), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'c', 'a', 'd']"], {'ordered': '(False)'}), "(['b', 'b', 'c', 'a', 'd'], ordered=False)\n", (14963, 15005), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((15121, 15159), 'riptable.FastArray', 'FastArray', (['[2.0, 2.0, 3.0, 1.0, 666.0]'], {}), '([2.0, 2.0, 3.0, 1.0, 666.0])\n', (15130, 15159), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((15221, 15288), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'c', 'a', 'd']"], {'ordered': '(False)', 'base_index': '(0)'}), "(['b', 'b', 'c', 'a', 'd'], ordered=False, base_index=0)\n", (15232, 15288), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((15398, 15451), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'c', 'a', 'd']"], {'ordered': '(False)'}), "(['b', 'b', 'c', 'a', 'd'], ordered=False)\n", (15409, 15451), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((15534, 15601), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'c', 'a', 'd']"], {'ordered': '(False)', 'base_index': '(0)'}), "(['b', 'b', 'c', 'a', 'd'], ordered=False, base_index=0)\n", (15545, 15601), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((15686, 15739), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'c', 'a', 'd']"], {'ordered': '(False)'}), "(['b', 'b', 'c', 'a', 'd'], ordered=False)\n", (15697, 15739), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((15759, 15790), 'riptable.FastArray', 'FastArray', (["['w', 'x', 'y', 'z']"], {}), "(['w', 'x', 'y', 'z'])\n", (15768, 15790), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((15843, 15884), 'riptable.FastArray', 'FastArray', (["[b'w', b'w', b'x', b'y', b'z']"], {}), "([b'w', b'w', b'x', b'y', b'z'])\n", (15852, 15884), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((15946, 16013), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'c', 'a', 'd']"], {'ordered': '(False)', 'base_index': '(0)'}), "(['b', 'b', 'c', 'a', 'd'], ordered=False, base_index=0)\n", (15957, 16013), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((16110, 16163), 'riptable.Categorical', 'Categorical', (['[2, 2, 3, 1, 4, 0]', "['a', 'b', 'c', 'd']"], {}), "([2, 2, 3, 1, 4, 0], ['a', 'b', 'c', 'd'])\n", (16121, 16163), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((16279, 16324), 'riptable.FastArray', 'FastArray', (['[2.0, 2.0, 3.0, 1.0, 666.0, 666.0]'], {}), '([2.0, 2.0, 3.0, 1.0, 666.0, 666.0])\n', (16288, 16324), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((16506, 16544), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {}), "(['a', 'a', 'b', 'c', 'a'])\n", (16517, 16544), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((16621, 16673), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {'base_index': '(0)'}), "(['a', 'a', 'b', 'c', 'a'], base_index=0)\n", (16632, 16673), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((17497, 17531), 'riptable.Categorical', 'Categorical', (['codes', 'LikertDecision'], {}), '(codes, LikertDecision)\n', (17508, 17531), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((18169, 18203), 'riptable.Categorical', 'Categorical', (['codes', 'LikertDecision'], {}), '(codes, LikertDecision)\n', (18180, 18203), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((19524, 19573), 'riptable.Categorical', 'Categorical', (['idx_list', 'str_list'], {'from_matlab': '(True)'}), '(idx_list, str_list, from_matlab=True)\n', (19535, 19573), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((20199, 20244), 'pandas.Categorical.from_codes', 'pd.Categorical.from_codes', (['idx_list', 'str_list'], {}), '(idx_list, str_list)\n', (20224, 20244), True, 'import pandas as pd\n'), ((20261, 20278), 'riptable.Categorical', 'Categorical', (['pd_c'], {}), '(pd_c)\n', (20272, 20278), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((20295, 20326), 'riptable.Categorical', 'Categorical', (['idx_list', 'str_list'], {}), '(idx_list, str_list)\n', (20306, 20326), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((20758, 20815), 'pandas.Categorical.from_codes', 'pd.Categorical.from_codes', (['[-1, 0, 1, 2]', "['a', 'b', 'c']"], {}), "([-1, 0, 1, 2], ['a', 'b', 'c'])\n", (20783, 20815), True, 'import pandas as pd\n'), ((20832, 20849), 'riptable.Categorical', 'Categorical', (['pd_c'], {}), '(pd_c)\n', (20843, 20849), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((21076, 21136), 'pandas.Categorical.from_codes', 'pd.Categorical.from_codes', (['[-1, 0, 1, 2]', "[u'โ', u'โ', u'โ']"], {}), "([-1, 0, 1, 2], [u'โ', u'โ', u'โ'])\n", (21101, 21136), True, 'import pandas as pd\n'), ((21168, 21185), 'riptable.Categorical', 'Categorical', (['pd_c'], {}), '(pd_c)\n', (21179, 21185), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((21956, 21989), 'riptable.Categorical', 'Categorical', (['str_list', 'unique_str'], {}), '(str_list, unique_str)\n', (21967, 21989), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((22341, 22357), 'riptable.Categorical', 'Categorical', (['lis'], {}), '(lis)\n', (22352, 22357), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((24237, 24269), 'riptable.rt_categorical.Categories', 'Categories', (['c_list'], {'unicode': '(True)'}), '(c_list, unicode=True)\n', (24247, 24269), False, 'from riptable.rt_categorical import Categories\n'), ((24296, 24322), 'riptable.rt_categorical.Categories', 'Categories', (['LikertDecision'], {}), '(LikertDecision)\n', (24306, 24322), False, 'from riptable.rt_categorical import Categories\n'), ((25048, 25205), 'riptable.Date', 'rt.Date', (["['2019-03-15', '2019-04-18', '2019-05-17', '2019-06-21', '2019-07-19',\n '2019-08-16', '2019-09-20', '2019-10-18', '2019-11-15', '2019-12-20']"], {}), "(['2019-03-15', '2019-04-18', '2019-05-17', '2019-06-21',\n '2019-07-19', '2019-08-16', '2019-09-20', '2019-10-18', '2019-11-15',\n '2019-12-20'])\n", (25055, 25205), True, 'import riptable as rt\n'), ((25459, 25472), 'riptable.Cat', 'rt.Cat', (['dates'], {}), '(dates)\n', (25465, 25472), True, 'import riptable as rt\n'), ((25609, 25760), 'riptable.FA', 'rt.FA', (["['2019-03-15', '2019-04-18', '2019-05-17', '2019-06-21', '2019-07-19',\n '2019-08-16', '2019-09-20', '2019-10-18', '2019-11-15', '2019-12-20']"], {}), "(['2019-03-15', '2019-04-18', '2019-05-17', '2019-06-21', '2019-07-19',\n '2019-08-16', '2019-09-20', '2019-10-18', '2019-11-15', '2019-12-20'])\n", (25614, 25760), True, 'import riptable as rt\n'), ((26019, 26044), 'riptable.Cat', 'rt.Cat', (['[dates, datestrs]'], {}), '([dates, datestrs])\n', (26025, 26044), True, 'import riptable as rt\n'), ((26246, 26257), 'riptable.Date', 'rt.Date', (['[]'], {}), '([])\n', (26253, 26257), True, 'import riptable as rt\n'), ((26279, 26292), 'riptable.Cat', 'rt.Cat', (['dates'], {}), '(dates)\n', (26285, 26292), True, 'import riptable as rt\n'), ((26545, 26591), 'riptable.FastArray', 'FastArray', (["['a', 'b', 'c', 'c', 'd', 'a', 'b']"], {}), "(['a', 'b', 'c', 'c', 'd', 'a', 'b'])\n", (26554, 26591), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((26710, 26770), 'riptable.Categorical', 'Categorical', (['values'], {'ordered': '(True)', 'base_index': '(1)', 'filter': 'None'}), '(values, ordered=True, base_index=1, filter=None)\n', (26721, 26770), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((27071, 27146), 'riptable.Categorical', 'Categorical', (['univals'], {'ordered': '(True)', 'base_index': '(1)', 'filter': 'None', 'unicode': '(True)'}), '(univals, ordered=True, base_index=1, filter=None, unicode=True)\n', (27082, 27146), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((27644, 27687), 'riptable.FastArray', 'FastArray', (['[True, True, False, False, True]'], {}), '([True, True, False, False, True])\n', (27653, 27687), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((27703, 27741), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {}), "(['a', 'a', 'b', 'c', 'a'])\n", (27714, 27741), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((27891, 27943), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {'base_index': '(0)'}), "(['a', 'a', 'b', 'c', 'a'], base_index=0)\n", (27902, 27943), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((28089, 28142), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {'filter': 'filter'}), "(['a', 'a', 'b', 'c', 'a'], filter=filter)\n", (28100, 28142), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((28266, 28333), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {'base_index': '(0)', 'filter': 'filter'}), "(['a', 'a', 'b', 'c', 'a'], base_index=0, filter=filter)\n", (28277, 28333), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((28608, 28663), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']", "['a', 'b', 'c']"], {}), "(['a', 'a', 'b', 'c', 'a'], ['a', 'b', 'c'])\n", (28619, 28663), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((28813, 28882), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']", "['a', 'b', 'c']"], {'base_index': '(0)'}), "(['a', 'a', 'b', 'c', 'a'], ['a', 'b', 'c'], base_index=0)\n", (28824, 28882), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((29322, 29363), 'pandas.Categorical', 'pd.Categorical', (["['a', 'a', 'b', 'c', 'a']"], {}), "(['a', 'a', 'b', 'c', 'a'])\n", (29336, 29363), True, 'import pandas as pd\n'), ((29846, 29936), 'riptable.Categorical', 'Categorical', (['values', 'good_cats'], {'ordered': '(True)', 'base_index': '(1)', 'unicode': '(False)', 'filter': 'None'}), '(values, good_cats, ordered=True, base_index=1, unicode=False,\n filter=None)\n', (29857, 29936), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((30197, 30286), 'riptable.Categorical', 'Categorical', (['values', 'good_cats'], {'ordered': '(True)', 'base_index': '(1)', 'unicode': '(True)', 'filter': 'None'}), '(values, good_cats, ordered=True, base_index=1, unicode=True,\n filter=None)\n', (30208, 30286), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((31730, 31764), 'riptable.Categorical', 'Categorical', (['codes', 'LikertDecision'], {}), '(codes, LikertDecision)\n', (31741, 31764), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((32265, 32311), 'riptable.FastArray', 'FastArray', (["['a', 'b', 'c', 'c', 'd', 'a', 'b']"], {}), "(['a', 'b', 'c', 'c', 'd', 'a', 'b'])\n", (32274, 32311), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((32597, 32657), 'riptable.Categorical', 'Categorical', (['values'], {'ordered': '(True)', 'base_index': '(1)', 'filter': 'None'}), '(values, ordered=True, base_index=1, filter=None)\n', (32608, 32657), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((32995, 33070), 'riptable.Categorical', 'Categorical', (['univals'], {'ordered': '(True)', 'base_index': '(1)', 'filter': 'None', 'unicode': '(True)'}), '(univals, ordered=True, base_index=1, filter=None, unicode=True)\n', (33006, 33070), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((33212, 33272), 'riptable.Categorical', 'Categorical', (['values'], {'ordered': '(True)', 'base_index': '(1)', 'filter': 'None'}), '(values, ordered=True, base_index=1, filter=None)\n', (33223, 33272), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((33519, 33594), 'riptable.Categorical', 'Categorical', (['univals'], {'ordered': '(True)', 'base_index': '(1)', 'filter': 'None', 'unicode': '(True)'}), '(univals, ordered=True, base_index=1, filter=None, unicode=True)\n', (33530, 33594), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((35133, 35166), 'riptable.Categorical', 'Categorical', (['values'], {'base_index': '(1)'}), '(values, base_index=1)\n', (35144, 35166), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((35614, 35657), 'riptable.FastArray', 'FastArray', (['[False, True, True, False, True]'], {}), '([False, True, True, False, True])\n', (35623, 35657), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((35671, 35705), 'riptable.Categorical', 'Categorical', (['codes', 'LikertDecision'], {}), '(codes, LikertDecision)\n', (35682, 35705), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((37853, 37891), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'c', 'b', 'b']"], {}), "(['a', 'a', 'c', 'b', 'b'])\n", (37864, 37891), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((37906, 37944), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'd', 'd', 'c']"], {}), "(['b', 'b', 'd', 'd', 'c'])\n", (37917, 37944), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((37959, 37987), 'riptable.Categorical.hstack', 'Categorical.hstack', (['[c1, c2]'], {}), '([c1, c2])\n', (37977, 37987), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((38099, 38162), 'riptable.Categorical', 'Categorical', (['[1, 1, 3, 2, 2]', '[1, 2, 3, 4, 5]'], {'from_matlab': '(True)'}), '([1, 1, 3, 2, 2], [1, 2, 3, 4, 5], from_matlab=True)\n', (38110, 38162), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((38177, 38240), 'riptable.Categorical', 'Categorical', (['[2, 2, 4, 4, 3]', '[1, 2, 3, 4, 5]'], {'from_matlab': '(True)'}), '([2, 2, 4, 4, 3], [1, 2, 3, 4, 5], from_matlab=True)\n', (38188, 38240), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((38255, 38283), 'riptable.Categorical.hstack', 'Categorical.hstack', (['[c1, c2]'], {}), '([c1, c2])\n', (38273, 38283), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((39231, 39283), 'riptable.Categorical', 'rt.Categorical', (['country_num_codes', 'country_code_dict'], {}), '(country_num_codes, country_code_dict)\n', (39245, 39283), True, 'import riptable as rt\n'), ((39421, 39532), 'riptable.Categorical', 'rt.Categorical', (["['AUS', 'AUS', 'HKG', 'USA', 'USA', 'IRL', 'USA', 'IRL', 'USA', 'KHM',\n 'IRL', 'AUS', 'MEX']"], {}), "(['AUS', 'AUS', 'HKG', 'USA', 'USA', 'IRL', 'USA', 'IRL',\n 'USA', 'KHM', 'IRL', 'AUS', 'MEX'])\n", (39435, 39532), True, 'import riptable as rt\n'), ((39839, 39867), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (39850, 39867), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((39882, 39910), 'riptable.Categorical', 'Categorical', (["['d', 'e', 'f']"], {}), "(['d', 'e', 'f'])\n", (39893, 39910), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((39925, 39953), 'riptable.Categorical', 'Categorical', (["['c', 'f', 'z']"], {}), "(['c', 'f', 'z'])\n", (39936, 39953), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((39968, 39999), 'riptable.Categorical.align', 'Categorical.align', (['[c1, c2, c3]'], {}), '([c1, c2, c3])\n', (39985, 39999), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((40452, 40515), 'riptable.Categorical', 'Categorical', (['[1, 1, 3, 2, 2]', '[1, 2, 3, 4, 5]'], {'from_matlab': '(True)'}), '([1, 1, 3, 2, 2], [1, 2, 3, 4, 5], from_matlab=True)\n', (40463, 40515), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((40530, 40593), 'riptable.Categorical', 'Categorical', (['[2, 2, 4, 4, 3]', '[1, 2, 3, 4, 5]'], {'from_matlab': '(True)'}), '([2, 2, 4, 4, 3], [1, 2, 3, 4, 5], from_matlab=True)\n', (40541, 40593), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((40608, 40635), 'riptable.Categorical.align', 'Categorical.align', (['[c1, c2]'], {}), '([c1, c2])\n', (40625, 40635), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((40922, 40949), 'riptable.Categorical.align', 'Categorical.align', (['[c1, c2]'], {}), '([c1, c2])\n', (40939, 40949), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((41317, 41355), 'riptable.Categorical', 'Categorical', (['[3, 3, 4, 3, 1, 2, 5]', 'd1'], {}), '([3, 3, 4, 3, 1, 2, 5], d1)\n', (41328, 41355), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((41370, 41408), 'riptable.Categorical', 'Categorical', (['[1, 1, 5, 2, 2, 1, 5]', 'd2'], {}), '([1, 1, 5, 2, 2, 1, 5], d2)\n', (41381, 41408), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((41431, 41481), 'riptable.rt_categorical.categorical_merge_dict', 'categorical_merge_dict', (['[c1, c2]'], {'return_type': 'dict'}), '([c1, c2], return_type=dict)\n', (41453, 41481), False, 'from riptable.rt_categorical import categorical_merge_dict\n'), ((41606, 41645), 'riptable.Categorical', 'Categorical', (['[0, 1, 2]', "['a', 'b', 'c']"], {}), "([0, 1, 2], ['a', 'b', 'c'])\n", (41617, 41645), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((42148, 42184), 'riptable.FastArray', 'FastArray', (["['a', 'b', 'c', 'd', 'e']"], {}), "(['a', 'b', 'c', 'd', 'e'])\n", (42157, 42184), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((42268, 42329), 'riptable.FastArray', 'FastArray', (["['e', 'e', 'd', 'b', 'e', 'c', 'c', 'e', 'a', 'd']"], {}), "(['e', 'e', 'd', 'b', 'e', 'c', 'c', 'e', 'a', 'd'])\n", (42277, 42329), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((42343, 42363), 'riptable.Categorical', 'Categorical', (['str_arr'], {}), '(str_arr)\n', (42354, 42363), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((42874, 42987), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c', 'my_invalid']", "['a', 'b', 'c', 'my_invalid']"], {'invalid': '"""my_invalid"""', 'base_index': '(1)'}), "(['a', 'b', 'c', 'my_invalid'], ['a', 'b', 'c', 'my_invalid'],\n invalid='my_invalid', base_index=1)\n", (42885, 42987), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((43262, 43375), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c', 'my_invalid']", "['a', 'b', 'c', 'my_invalid']"], {'invalid': '"""my_invalid"""', 'base_index': '(1)'}), "(['a', 'b', 'c', 'my_invalid'], ['a', 'b', 'c', 'my_invalid'],\n invalid='my_invalid', base_index=1)\n", (43273, 43375), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((44146, 44194), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'a', 'c', 'c', 'b']"], {}), "(['a', 'a', 'b', 'a', 'c', 'c', 'b'])\n", (44157, 44194), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((44214, 44269), 'riptable.FastArray', 'FastArray', (['[True, True, True, True, False, False, True]'], {}), '([True, True, True, True, False, False, True])\n', (44223, 44269), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((44766, 44814), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'a', 'c', 'c', 'b']"], {}), "(['a', 'a', 'b', 'a', 'c', 'c', 'b'])\n", (44777, 44814), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((45944, 45977), 'riptable.Categorical', 'Categorical', (['values'], {'base_index': '(1)'}), '(values, base_index=1)\n', (45955, 45977), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((46247, 46293), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['expected', 'c.category_array'], {}), '(expected, c.category_array)\n', (46265, 46293), False, 'from numpy.testing import assert_array_equal\n'), ((46451, 46513), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'a', 'c', 'c', 'b']"], {'unicode': '(True)'}), "(['a', 'a', 'b', 'a', 'c', 'c', 'b'], unicode=True)\n", (46462, 46513), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((46832, 46853), 'riptable.Categorical', 'Categorical', (['codes', 'd'], {}), '(codes, d)\n', (46843, 46853), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((47807, 47862), 'riptable.Categorical', 'Categorical', (['[44, 133, 133, 75, 144, 1]', 'LikertDecision'], {}), '([44, 133, 133, 75, 144, 1], LikertDecision)\n', (47818, 47862), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((47977, 48029), 'riptable.FastArray', 'FastArray', (['[False, False, False, False, True, False]'], {}), '([False, False, False, False, True, False])\n', (47986, 48029), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((48381, 48429), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'a', 'c', 'c', 'b']"], {}), "(['a', 'a', 'b', 'a', 'c', 'c', 'b'])\n", (48392, 48429), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((49118, 49173), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']", "['a', 'b']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])\n", (49129, 49173), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((49514, 49569), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']", "['a', 'b']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])\n", (49525, 49569), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((51108, 51163), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']", "['a', 'b']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])\n", (51119, 51163), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((51699, 51754), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']", "['a', 'b']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])\n", (51710, 51754), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((52688, 52743), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']", "['a', 'b']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])\n", (52699, 52743), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((53549, 53604), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']", "['a', 'b']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])\n", (53560, 53604), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((55306, 55349), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'])\n", (55317, 55349), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((56707, 56724), 'riptable.Categorical', 'Categorical', (['nums'], {}), '(nums)\n', (56718, 56724), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((56856, 56873), 'riptable.Categorical', 'Categorical', (['nums'], {}), '(nums)\n', (56867, 56873), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((57678, 57738), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']", "['a', 'b', 'c']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b', 'c'])\n", (57689, 57738), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((57752, 57807), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']", "['a', 'b']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])\n", (57763, 57807), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((58297, 58352), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']", "['a', 'b']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])\n", (58308, 58352), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((58893, 58948), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']", "['a', 'b']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])\n", (58904, 58948), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((58963, 59004), 'riptable.Categorical', 'Categorical', (['[0, 1]', "{(0): 'a', (1): 'b'}"], {}), "([0, 1], {(0): 'a', (1): 'b'})\n", (58974, 59004), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((59126, 59167), 'riptable.Categorical', 'Categorical', (['[0, 1]', "{(0): 'a', (1): 'b'}"], {}), "([0, 1], {(0): 'a', (1): 'b'})\n", (59137, 59167), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((59178, 59219), 'riptable.Categorical', 'Categorical', (['[0, 1]', "{(1): 'a', (0): 'b'}"], {}), "([0, 1], {(1): 'a', (0): 'b'})\n", (59189, 59219), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((59335, 59376), 'riptable.Categorical', 'Categorical', (['[0, 1]', "{(0): 'a', (1): 'b'}"], {}), "([0, 1], {(0): 'a', (1): 'b'})\n", (59346, 59376), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((59387, 59428), 'riptable.Categorical', 'Categorical', (['[2, 1]', "{(2): 'c', (1): 'b'}"], {}), "([2, 1], {(2): 'c', (1): 'b'})\n", (59398, 59428), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((59444, 59468), 'riptable.FastArray', 'FastArray', (['[False, True]'], {}), '([False, True])\n', (59453, 59468), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((59609, 59681), 'riptable.Categorical', 'Categorical', (['[1, 1, 3, 2, 2]', "['a', 'b', 'c']"], {'base_index': '(1)', 'invalid': '"""a"""'}), "([1, 1, 3, 2, 2], ['a', 'b', 'c'], base_index=1, invalid='a')\n", (59620, 59681), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((60161, 60199), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c', 'd', 'e']"], {}), "(['a', 'b', 'c', 'd', 'e'])\n", (60172, 60199), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((61054, 61121), 'riptable.Categorical', 'Categorical', (['[1, 2, 3, 4]', "{(1): 'a', (2): 'b', (3): 'c', (4): 'd'}"], {}), "([1, 2, 3, 4], {(1): 'a', (2): 'b', (3): 'c', (4): 'd'})\n", (61065, 61121), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((62052, 62090), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {}), "(['a', 'a', 'b', 'c', 'a'])\n", (62063, 62090), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((62170, 62222), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {'base_index': '(0)'}), "(['a', 'a', 'b', 'c', 'a'], base_index=0)\n", (62181, 62222), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((62535, 62573), 'riptable.Categorical', 'Categorical', (['codes', 'cats'], {'base_index': '(1)'}), '(codes, cats, base_index=1)\n', (62546, 62573), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((62695, 62737), 'riptable.Categorical', 'Categorical', (['codes', 'cats'], {'from_matlab': '(True)'}), '(codes, cats, from_matlab=True)\n', (62706, 62737), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((63349, 63387), 'riptable.Categorical', 'Categorical', (["['c', 'c', 'a', 'b', 'c']"], {}), "(['c', 'c', 'a', 'b', 'c'])\n", (63360, 63387), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((63468, 63521), 'riptable.Categorical', 'Categorical', (["['c', 'c', 'a', 'b', 'c']"], {'ordered': '(False)'}), "(['c', 'c', 'a', 'b', 'c'], ordered=False)\n", (63479, 63521), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((63602, 63657), 'riptable.Categorical', 'Categorical', (["['c', 'c', 'a', 'b', 'c']", "['c', 'a', 'b']"], {}), "(['c', 'c', 'a', 'b', 'c'], ['c', 'a', 'b'])\n", (63613, 63657), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((63768, 63823), 'riptable.Categorical', 'Categorical', (["['c', 'c', 'a', 'b', 'c']", "['a', 'b', 'c']"], {}), "(['c', 'c', 'a', 'b', 'c'], ['a', 'b', 'c'])\n", (63779, 63823), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((64215, 64285), 'riptable.Categorical', 'Categorical', (["['c', 'c', 'a', 'b', 'c']", "['c', 'a', 'b']"], {'ordered': '(False)'}), "(['c', 'c', 'a', 'b', 'c'], ['c', 'a', 'b'], ordered=False)\n", (64226, 64285), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((64400, 64426), 'riptable.FastArray', 'FastArray', (['[0, 0, 1, 2, 0]'], {}), '([0, 0, 1, 2, 0])\n', (64409, 64426), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((64443, 64483), 'riptable.FastArray', 'FastArray', (["['c', 'b', 'a']"], {'unicode': '(True)'}), "(['c', 'b', 'a'], unicode=True)\n", (64452, 64483), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((64497, 64521), 'riptable.Categorical', 'Categorical', (['codes', 'cats'], {}), '(codes, cats)\n', (64508, 64521), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((65397, 65452), 'riptable.Categorical', 'Categorical', (["['b', 'b', 'b', 'a', 'b', 'b']", "['a', 'b']"], {}), "(['b', 'b', 'b', 'a', 'b', 'b'], ['a', 'b'])\n", (65408, 65452), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((67528, 67556), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (67539, 67556), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((67925, 67953), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (67936, 67953), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((68114, 68168), 'riptable.Categorical', 'Categorical', (['[1, 2, 3]', "{(1): 'a', (2): 'b', (3): 'c'}"], {}), "([1, 2, 3], {(1): 'a', (2): 'b', (3): 'c'})\n", (68125, 68168), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((68565, 68593), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (68576, 68593), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((69645, 69676), 'pandas.Categorical', 'pd.Categorical', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (69659, 69676), True, 'import pandas as pd\n'), ((69690, 69720), 'riptable.Categorical', 'Categorical', (['pdc'], {'unicode': '(True)'}), '(pdc, unicode=True)\n', (69701, 69720), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((69787, 69818), 'riptable.Categorical', 'Categorical', (['pdc'], {'unicode': '(False)'}), '(pdc, unicode=False)\n', (69798, 69818), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((69887, 69916), 'pandas.Categorical', 'pd.Categorical', (['three_unicode'], {}), '(three_unicode)\n', (69901, 69916), True, 'import pandas as pd\n'), ((69930, 69946), 'riptable.Categorical', 'Categorical', (['pdc'], {}), '(pdc)\n', (69941, 69946), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((71046, 71100), 'riptable.Categorical', 'Categorical', (['[1, 1, 2, 3, 1]', "{'a': 1, 'b': 2, 'c': 3}"], {}), "([1, 1, 2, 3, 1], {'a': 1, 'b': 2, 'c': 3})\n", (71057, 71100), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((71428, 71494), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {'unicode': '(True)', 'invalid': "b'a'"}), "(['a', 'a', 'b', 'c', 'a'], unicode=True, invalid=b'a')\n", (71439, 71494), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((72323, 72345), 'riptable.Categorical', 'Categorical', (['ds.strcol'], {}), '(ds.strcol)\n', (72334, 72345), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((72402, 72437), 'riptable.Categorical', 'Categorical', (['[ds.strcol, ds.numcol]'], {}), '([ds.strcol, ds.numcol])\n', (72413, 72437), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((72679, 72728), 'riptable.Categorical', 'Categorical', (['ds.mycodes', "{'a': 1, 'b': 2, 'c': 3}"], {}), "(ds.mycodes, {'a': 1, 'b': 2, 'c': 3})\n", (72690, 72728), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((72834, 72860), 'riptable.FastArray', 'FastArray', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (72843, 72860), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((72905, 72929), 'riptable.Categorical', 'Categorical', (['codes', 'cats'], {}), '(codes, cats)\n', (72916, 72929), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((73017, 73040), 'riptable.Categorical', 'Categorical', (["['a', 'b']"], {}), "(['a', 'b'])\n", (73028, 73040), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((73331, 73369), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {}), "(['a', 'a', 'b', 'c', 'a'])\n", (73342, 73369), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((73383, 73397), 'riptable.Categorical', 'Categorical', (['c'], {}), '(c)\n', (73394, 73397), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((73526, 73542), 'riptable.Categorical', 'Categorical', (['[c]'], {}), '([c])\n', (73537, 73542), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((73704, 73742), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {}), "(['a', 'a', 'b', 'c', 'a'])\n", (73715, 73742), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((74518, 74554), 'riptable.FastArray', 'FastArray', (["['a', 'a', 'b', 'c', 'a']"], {}), "(['a', 'a', 'b', 'c', 'a'])\n", (74527, 74554), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((74574, 74583), 'riptable.rt_numpy.arange', 'arange', (['(5)'], {}), '(5)\n', (74580, 74583), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((74599, 74633), 'riptable.Categorical', 'Categorical', (['[og_strings, og_nums]'], {}), '([og_strings, og_nums])\n', (74610, 74633), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((74886, 74934), 'riptable.Categorical', 'Categorical', (['[1, 2, 3]', "{'a': 1, 'b': 2, 'c': 3}"], {}), "([1, 2, 3], {'a': 1, 'b': 2, 'c': 3})\n", (74897, 74934), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((75304, 75352), 'riptable.Categorical', 'Categorical', (['[1, 2, 3]', "{'a': 1, 'b': 2, 'c': 3}"], {}), "([1, 2, 3], {'a': 1, 'b': 2, 'c': 3})\n", (75315, 75352), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((76143, 76171), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (76154, 76171), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((76399, 76456), 'riptable.Categorical', 'Categorical', (['[1, 1, 1, 1, 2, 3]', "{'a': 1, 'b': 2, 'c': 3}"], {}), "([1, 1, 1, 1, 2, 3], {'a': 1, 'b': 2, 'c': 3})\n", (76410, 76456), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((76471, 76528), 'riptable.Categorical', 'Categorical', (['[1, 1, 1, 1, 3, 4]', "{'a': 1, 'c': 3, 'd': 4}"], {}), "([1, 1, 1, 1, 3, 4], {'a': 1, 'c': 3, 'd': 4})\n", (76482, 76528), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((76548, 76576), 'riptable.Categorical.hstack', 'Categorical.hstack', (['[c1, c2]'], {}), '([c1, c2])\n', (76566, 76576), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((76766, 76823), 'riptable.Categorical', 'Categorical', (['[1, 1, 1, 1, 2, 3]', "{'a': 1, 'b': 2, 'd': 3}"], {}), "([1, 1, 1, 1, 2, 3], {'a': 1, 'b': 2, 'd': 3})\n", (76777, 76823), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((76838, 76895), 'riptable.Categorical', 'Categorical', (['[1, 1, 1, 1, 3, 4]', "{'a': 1, 'c': 3, 'd': 4}"], {}), "([1, 1, 1, 1, 3, 4], {'a': 1, 'c': 3, 'd': 4})\n", (76849, 76895), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((77182, 77211), 'riptable.FastArray', 'FastArray', (['[1.0, 0.0, np.nan]'], {}), '([1.0, 0.0, np.nan])\n', (77191, 77211), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((77848, 77907), 'riptable.FastArray', 'FastArray', (["[b'INVALID', b'INVALID', b'b', b'c', b'INVALID']"], {}), "([b'INVALID', b'INVALID', b'b', b'c', b'INVALID'])\n", (77857, 77907), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((78171, 78200), 'riptable.FastArray', 'FastArray', (["[f, f, b'b', b'c']"], {}), "([f, f, b'b', b'c'])\n", (78180, 78200), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((78340, 78369), 'riptable.FastArray', 'FastArray', (["[b'a', b'b', b'c']"], {}), "([b'a', b'b', b'c'])\n", (78349, 78369), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((78800, 78853), 'riptable.Categorical', 'Categorical', (['[1.0, 1.0, 2.0]', '[1.0, 2.0]'], {'invalid': '(2.0)'}), '([1.0, 1.0, 2.0], [1.0, 2.0], invalid=2.0)\n', (78811, 78853), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((79154, 79206), 'riptable.Categorical', 'Categorical', (["['b', 'c', 'a', 'a', 'b']"], {'base_index': '(0)'}), "(['b', 'c', 'a', 'a', 'b'], base_index=0)\n", (79165, 79206), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((79481, 79533), 'riptable.Categorical', 'Categorical', (["['b', 'c', 'a', 'a', 'b']"], {'base_index': '(1)'}), "(['b', 'c', 'a', 'a', 'b'], base_index=1)\n", (79492, 79533), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((79880, 79908), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (79891, 79908), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((80096, 80124), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (80107, 80124), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((80950, 80971), 'riptable.Categorical', 'Categorical', (['codes', 'd'], {}), '(codes, d)\n', (80961, 80971), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((81119, 81140), 'riptable.Categorical', 'Categorical', (['codes', 'd'], {}), '(codes, d)\n', (81130, 81140), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((81245, 81254), 'riptable.rt_numpy.arange', 'arange', (['(5)'], {}), '(5)\n', (81251, 81254), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((81268, 81289), 'riptable.Categorical', 'Categorical', (['codes', 'd'], {}), '(codes, d)\n', (81279, 81289), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((81390, 81450), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'd']", "['a', 'b', 'c', 'd']"], {}), "(['a', 'a', 'b', 'c', 'd'], ['a', 'b', 'c', 'd'])\n", (81401, 81450), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((81551, 81625), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'd']", "['a', 'b', 'c', 'd']"], {'base_index': '(0)'}), "(['a', 'a', 'b', 'c', 'd'], ['a', 'b', 'c', 'd'], base_index=0)\n", (81562, 81625), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((81726, 81764), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'd']"], {}), "(['a', 'a', 'b', 'c', 'd'])\n", (81737, 81764), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((83169, 83178), 'riptable.rt_numpy.ones', 'ones', (['(100)'], {}), '(100)\n', (83173, 83178), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((83218, 83258), 'riptable.Categorical', 'Categorical', (['idx', 'cats'], {'from_matlab': '(True)'}), '(idx, cats, from_matlab=True)\n', (83229, 83258), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((83559, 83584), 'riptable.Categorical', 'Categorical', (['[strs, flts]'], {}), '([strs, flts])\n', (83570, 83584), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((83603, 83620), 'riptable.Categorical', 'Categorical', (['strs'], {}), '(strs)\n', (83614, 83620), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((83639, 83656), 'riptable.Categorical', 'Categorical', (['flts'], {}), '(flts)\n', (83650, 83656), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((83783, 83810), 'riptable.Categorical', 'Categorical', (['[strs2, flts2]'], {}), '([strs2, flts2])\n', (83794, 83810), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((83829, 83847), 'riptable.Categorical', 'Categorical', (['strs2'], {}), '(strs2)\n', (83840, 83847), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((83866, 83884), 'riptable.Categorical', 'Categorical', (['flts2'], {}), '(flts2)\n', (83877, 83884), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((84043, 84062), 'riptable.Categorical', 'Categorical', (['fa_str'], {}), '(fa_str)\n', (84054, 84062), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((84080, 84099), 'riptable.Categorical', 'Categorical', (['fa_flt'], {}), '(fa_flt)\n', (84091, 84099), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((84939, 84966), 'riptable.load_sds', 'load_sds', (['paths'], {'stack': '(True)'}), '(paths, stack=True)\n', (84947, 84966), False, 'from riptable import save_sds, load_sds\n'), ((88397, 88411), 'riptable.Categorical', 'Categorical', (['f'], {}), '(f)\n', (88408, 88411), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((89247, 89270), 'riptable.Categorical', 'Categorical', (['three_ints'], {}), '(three_ints)\n', (89258, 89270), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((89584, 89610), 'riptable.Categorical', 'Categorical', (['three_unicode'], {}), '(three_unicode)\n', (89595, 89610), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((90250, 90274), 'riptable.Categorical', 'Categorical', (['three_bytes'], {}), '(three_bytes)\n', (90261, 90274), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((90901, 90925), 'riptable.Categorical', 'Categorical', (['three_bytes'], {}), '(three_bytes)\n', (90912, 90925), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((91740, 91764), 'riptable.Categorical', 'Categorical', (['three_bytes'], {}), '(three_bytes)\n', (91751, 91764), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((91779, 91803), 'riptable.Categorical', 'Categorical', (['three_bytes'], {}), '(three_bytes)\n', (91790, 91803), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((92130, 92156), 'riptable.Categorical', 'Categorical', (['three_unicode'], {}), '(three_unicode)\n', (92141, 92156), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((92171, 92197), 'riptable.Categorical', 'Categorical', (['three_unicode'], {}), '(three_unicode)\n', (92182, 92197), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((92516, 92544), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (92527, 92544), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((92559, 92583), 'riptable.Categorical', 'Categorical', (['three_bytes'], {}), '(three_bytes)\n', (92570, 92583), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((93039, 93074), 'riptable.Categorical', 'Categorical', (['[0, 1, 0]', 'three_bytes'], {}), '([0, 1, 0], three_bytes)\n', (93050, 93074), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((93089, 93124), 'riptable.Categorical', 'Categorical', (['[2, 1, 2]', 'three_bytes'], {}), '([2, 1, 2], three_bytes)\n', (93100, 93124), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((93439, 93476), 'riptable.Categorical', 'Categorical', (['[0, 1, 0]', 'three_unicode'], {}), '([0, 1, 0], three_unicode)\n', (93450, 93476), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((93491, 93528), 'riptable.Categorical', 'Categorical', (['[2, 1, 2]', 'three_unicode'], {}), '([2, 1, 2], three_unicode)\n', (93502, 93528), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((115370, 115386), 'riptable.save_sds', 'save_sds', (['p', 'cat'], {}), '(p, cat)\n', (115378, 115386), False, 'from riptable import save_sds, load_sds\n'), ((115403, 115414), 'riptable.load_sds', 'load_sds', (['p'], {}), '(p)\n', (115411, 115414), False, 'from riptable import save_sds, load_sds\n'), ((115647, 115684), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['cat2._fa', 'cat._fa'], {}), '(cat2._fa, cat._fa)\n', (115665, 115684), False, 'from numpy.testing import assert_array_equal\n'), ((116081, 116124), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['cat2_values', 'cat_values'], {}), '(cat2_values, cat_values)\n', (116099, 116124), False, 'from numpy.testing import assert_array_equal\n'), ((116391, 116428), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['cat3._fa', 'cat._fa'], {}), '(cat3._fa, cat._fa)\n', (116409, 116428), False, 'from numpy.testing import assert_array_equal\n'), ((116825, 116868), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['cat3_values', 'cat_values'], {}), '(cat3_values, cat_values)\n', (116843, 116868), False, 'from numpy.testing import assert_array_equal\n'), ((7774, 7792), 'riptable.FastArray', 'FastArray', (['v_bytes'], {}), '(v_bytes)\n', (7783, 7792), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((7807, 7823), 'riptable.FastArray', 'FastArray', (['v_str'], {}), '(v_str)\n', (7816, 7823), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((7974, 7992), 'riptable.FastArray', 'FastArray', (['c_bytes'], {}), '(c_bytes)\n', (7983, 7992), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((8007, 8023), 'riptable.FastArray', 'FastArray', (['c_str'], {}), '(c_str)\n', (8016, 8023), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((9684, 9709), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (9697, 9709), False, 'import pytest\n'), ((9728, 9759), 'riptable.Categorical', 'Categorical', (['idx_list', 'str_list'], {}), '(idx_list, str_list)\n', (9739, 9759), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((10786, 10829), 'riptable.FastArray', 'FastArray', (['[True, False, False, True, True]'], {}), '([True, False, False, True, True])\n', (10795, 10829), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((10844, 10888), 'riptable.FastArray', 'FastArray', (['[False, True, True, False, False]'], {}), '([False, True, True, False, False])\n', (10853, 10888), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((10903, 10945), 'riptable.FastArray', 'FastArray', (['[False, True, True, True, True]'], {}), '([False, True, True, True, True])\n', (10912, 10945), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((10960, 11004), 'riptable.FastArray', 'FastArray', (['[False, False, False, True, True]'], {}), '([False, False, False, True, True])\n', (10969, 11004), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((11019, 11062), 'riptable.FastArray', 'FastArray', (['[True, True, True, False, False]'], {}), '([True, True, True, False, False])\n', (11028, 11062), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((11077, 11122), 'riptable.FastArray', 'FastArray', (['[True, False, False, False, False]'], {}), '([True, False, False, False, False])\n', (11086, 11122), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((11177, 11218), 'riptable.FastArray', 'FastArray', (['[True, True, True, True, True]'], {}), '([True, True, True, True, True])\n', (11186, 11218), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((11233, 11279), 'riptable.FastArray', 'FastArray', (['[False, False, False, False, False]'], {}), '([False, False, False, False, False])\n', (11242, 11279), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((11294, 11336), 'riptable.FastArray', 'FastArray', (['[False, True, True, True, True]'], {}), '([False, True, True, True, True])\n', (11303, 11336), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((11351, 11393), 'riptable.FastArray', 'FastArray', (['[False, True, True, True, True]'], {}), '([False, True, True, True, True])\n', (11360, 11393), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((11408, 11453), 'riptable.FastArray', 'FastArray', (['[True, False, False, False, False]'], {}), '([True, False, False, False, False])\n', (11417, 11453), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((11468, 11513), 'riptable.FastArray', 'FastArray', (['[True, False, False, False, False]'], {}), '([True, False, False, False, False])\n', (11477, 11513), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((11594, 11615), 'riptable.Categorical', 'Categorical', (['codes', 'd'], {}), '(codes, d)\n', (11605, 11615), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((12925, 12968), 'riptable.FastArray', 'FastArray', (['[True, False, False, True, True]'], {}), '([True, False, False, True, True])\n', (12934, 12968), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((12983, 13027), 'riptable.FastArray', 'FastArray', (['[False, True, True, False, False]'], {}), '([False, True, True, False, False])\n', (12992, 13027), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((13042, 13084), 'riptable.FastArray', 'FastArray', (['[False, True, True, True, True]'], {}), '([False, True, True, True, True])\n', (13051, 13084), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((13099, 13143), 'riptable.FastArray', 'FastArray', (['[False, False, False, True, True]'], {}), '([False, False, False, True, True])\n', (13108, 13143), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((13158, 13201), 'riptable.FastArray', 'FastArray', (['[True, True, True, False, False]'], {}), '([True, True, True, False, False])\n', (13167, 13201), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((13216, 13261), 'riptable.FastArray', 'FastArray', (['[True, False, False, False, False]'], {}), '([True, False, False, False, False])\n', (13225, 13261), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((13342, 13363), 'riptable.Categorical', 'Categorical', (['codes', 'd'], {}), '(codes, d)\n', (13353, 13363), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((16752, 16777), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (16765, 16777), False, 'import pytest\n'), ((16939, 16964), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (16952, 16964), False, 'import pytest\n'), ((17756, 17781), 'pytest.raises', 'pytest.raises', (['IndexError'], {}), '(IndexError)\n', (17769, 17781), False, 'import pytest\n'), ((18804, 18821), 'riptable.FastArray', 'FastArray', (['[1, 4]'], {}), '([1, 4])\n', (18813, 18821), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((19434, 19458), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (19447, 19458), False, 'import pytest\n'), ((19477, 19508), 'riptable.Categorical', 'Categorical', (['idx_list', 'str_list'], {}), '(idx_list, str_list)\n', (19488, 19508), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((22559, 22583), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (22572, 22583), False, 'import pytest\n'), ((22602, 22619), 'riptable.Categorical', 'Categorical', (['(1)', '(2)'], {}), '(1, 2)\n', (22613, 22619), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((23017, 23043), 'riptable.Categorical', 'Categorical', (['arr'], {'dtype': 'dt'}), '(arr, dtype=dt)\n', (23028, 23043), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((23605, 23629), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (23618, 23629), False, 'import pytest\n'), ((23650, 23665), 'riptable.rt_categorical.Categories', 'Categories', (['tup'], {}), '(tup)\n', (23660, 23665), False, 'from riptable.rt_categorical import Categories\n'), ((23730, 23799), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']"], {'ordered': '(True)', 'base_index': '(1)', 'filter': 'None'}), "(['a', 'b', 'c'], ordered=True, base_index=1, filter=None)\n", (23741, 23799), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((28480, 28505), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (28493, 28505), False, 'import pytest\n'), ((28524, 28592), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']"], {'base_index': '(99)', 'filter': 'filter'}), "(['a', 'a', 'b', 'c', 'a'], base_index=99, filter=filter)\n", (28535, 28592), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((29029, 29063), 'pytest.raises', 'pytest.raises', (['NotImplementedError'], {}), '(NotImplementedError)\n', (29042, 29063), False, 'import pytest\n'), ((29082, 29170), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'c', 'a']", "['a', 'b', 'c']"], {'base_index': '(0)', 'filter': 'filter'}), "(['a', 'a', 'b', 'c', 'a'], ['a', 'b', 'c'], base_index=0,\n filter=filter)\n", (29093, 29170), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((29183, 29208), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (29196, 29208), False, 'import pytest\n'), ((29227, 29304), 'riptable.Categorical', 'Categorical', (['[1.0, 2.0, 3.0]', "['a', 'b', 'c']"], {'from_matlab': '(True)', 'base_index': '(0)'}), "([1.0, 2.0, 3.0], ['a', 'b', 'c'], from_matlab=True, base_index=0)\n", (29238, 29304), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((29378, 29403), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (29391, 29403), False, 'import pytest\n'), ((29422, 29452), 'riptable.Categorical', 'Categorical', (['pdc'], {'base_index': '(0)'}), '(pdc, base_index=0)\n', (29433, 29452), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((30836, 30861), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (30849, 30861), False, 'import pytest\n'), ((30880, 30977), 'riptable.Categorical', 'Categorical', (['values', 'incomplete_cats'], {'ordered': '(True)', 'base_index': '(1)', 'unicode': '(False)', 'filter': 'None'}), '(values, incomplete_cats, ordered=True, base_index=1, unicode=\n False, filter=None)\n', (30891, 30977), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((31082, 31107), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (31095, 31107), False, 'import pytest\n'), ((31126, 31239), 'riptable.Categorical', 'Categorical', (['values', 'incomplete_cats'], {'ordered': '(True)', 'base_index': '(1)', 'unicode': '(True)', 'filter': 'None', 'invalid': 'invalid'}), '(values, incomplete_cats, ordered=True, base_index=1, unicode=\n True, filter=None, invalid=invalid)\n', (31137, 31239), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((31938, 31963), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (31951, 31963), False, 'import pytest\n'), ((32803, 32827), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (32816, 32827), False, 'import pytest\n'), ((33399, 33423), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (33412, 33423), False, 'import pytest\n'), ((36225, 36250), 'pytest.warns', 'pytest.warns', (['UserWarning'], {}), '(UserWarning)\n', (36237, 36250), False, 'import pytest\n'), ((36269, 36298), 'riptable.Categorical', 'Categorical', (['[1, 2]', 'DupeEnum'], {}), '([1, 2], DupeEnum)\n', (36280, 36298), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((39719, 39757), 'pytest.raises', 'pytest.raises', (['(ValueError, TypeError)'], {}), '((ValueError, TypeError))\n', (39732, 39757), False, 'import pytest\n'), ((39772, 39795), 'riptable.hstack', 'rt.hstack', (['[cat1, cat2]'], {}), '([cat1, cat2])\n', (39781, 39795), True, 'import riptable as rt\n'), ((41853, 41878), 'pytest.raises', 'pytest.raises', (['IndexError'], {}), '(IndexError)\n', (41866, 41878), False, 'import pytest\n'), ((42665, 42689), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (42678, 42689), False, 'import pytest\n'), ((42708, 42752), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']", 'LikertDecision'], {}), "(['a', 'b', 'c'], LikertDecision)\n", (42719, 42752), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((44284, 44309), 'pytest.raises', 'pytest.raises', (['IndexError'], {}), '(IndexError)\n', (44297, 44309), False, 'import pytest\n'), ((44501, 44526), 'pytest.raises', 'pytest.raises', (['IndexError'], {}), '(IndexError)\n', (44514, 44526), False, 'import pytest\n'), ((47877, 47902), 'pytest.raises', 'pytest.raises', (['IndexError'], {}), '(IndexError)\n', (47890, 47902), False, 'import pytest\n'), ((48217, 48242), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (48230, 48242), False, 'import pytest\n'), ((48261, 48326), 'riptable.Categorical', 'Categorical', (["['a', 'a', 'b', 'a', 'c', 'c', 'b']", "['a', 'a', 'b']"], {}), "(['a', 'a', 'b', 'a', 'c', 'c', 'b'], ['a', 'a', 'b'])\n", (48272, 48326), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((48541, 48565), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (48554, 48565), False, 'import pytest\n'), ((48657, 48681), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (48670, 48681), False, 'import pytest\n'), ((49406, 49431), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (49419, 49431), False, 'import pytest\n'), ((49798, 49823), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (49811, 49823), False, 'import pytest\n'), ((51542, 51567), 'pytest.raises', 'pytest.raises', (['IndexError'], {}), '(IndexError)\n', (51555, 51567), False, 'import pytest\n'), ((59015, 59039), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (59028, 59039), False, 'import pytest\n'), ((59230, 59255), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (59243, 59255), False, 'import pytest\n'), ((60416, 60440), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (60429, 60440), False, 'import pytest\n'), ((60498, 60522), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (60511, 60522), False, 'import pytest\n'), ((60864, 60888), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (60877, 60888), False, 'import pytest\n'), ((60946, 60970), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (60959, 60970), False, 'import pytest\n'), ((61588, 61612), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (61601, 61612), False, 'import pytest\n'), ((61782, 61806), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (61795, 61806), False, 'import pytest\n'), ((61864, 61888), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (61877, 61888), False, 'import pytest\n'), ((62818, 62843), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (62831, 62843), False, 'import pytest\n'), ((62862, 62918), 'riptable.Categorical', 'Categorical', (['codes', 'cats'], {'from_matlab': '(True)', 'base_index': '(0)'}), '(codes, cats, from_matlab=True, base_index=0)\n', (62873, 62918), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((64999, 65024), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (65012, 65024), False, 'import pytest\n'), ((65043, 65095), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']"], {'filter': 'f', 'base_index': '(0)'}), "(['a', 'b', 'c'], filter=f, base_index=0)\n", (65054, 65095), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((66341, 66365), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (66354, 66365), False, 'import pytest\n'), ((66384, 66415), 'riptable.rt_categorical.Categories', 'Categories', (["['garbage', 'list']"], {}), "(['garbage', 'list'])\n", (66394, 66415), False, 'from riptable.rt_categorical import Categories\n'), ((66483, 66508), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (66496, 66508), False, 'import pytest\n'), ((67048, 67073), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (67061, 67073), False, 'import pytest\n'), ((67092, 67146), 'riptable.Categorical', 'Categorical', (["['a', 'b', 'c']", "['b', 'c']"], {'base_index': '(0)'}), "(['a', 'b', 'c'], ['b', 'c'], base_index=0)\n", (67103, 67146), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((69431, 69451), 'riptable.FastArray', 'FastArray', (['[1, 2, 3]'], {}), '([1, 2, 3])\n', (69440, 69451), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((70046, 70071), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (70059, 70071), False, 'import pytest\n'), ((70090, 70105), 'riptable.Categorical', 'Categorical', (['{}'], {}), '({})\n', (70101, 70105), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((70122, 70147), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (70135, 70147), False, 'import pytest\n'), ((70166, 70181), 'riptable.Categorical', 'Categorical', (['[]'], {}), '([])\n', (70177, 70181), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((70235, 70269), 'pytest.raises', 'pytest.raises', (['NotImplementedError'], {}), '(NotImplementedError)\n', (70248, 70269), False, 'import pytest\n'), ((70738, 70762), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (70751, 70762), False, 'import pytest\n'), ((70781, 70841), 'riptable.Categorical', 'Categorical', (['[1.0, 2.0, 3.0]', "{(1): 'a', (2): 'b', (3): 'c'}"], {}), "([1.0, 2.0, 3.0], {(1): 'a', (2): 'b', (3): 'c'})\n", (70792, 70841), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((70895, 70919), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (70908, 70919), False, 'import pytest\n'), ((71297, 71322), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (71310, 71322), False, 'import pytest\n'), ((72072, 72097), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (72085, 72097), False, 'import pytest\n'), ((72116, 72164), 'riptable.CatZero', 'CatZero', (["['a', 'a', 'b', 'c', 'a']"], {'base_index': '(1)'}), "(['a', 'a', 'b', 'c', 'a'], base_index=1)\n", (72123, 72164), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((72459, 72469), 'riptable.rt_numpy.arange', 'arange', (['(10)'], {}), '(10)\n', (72465, 72469), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((75197, 75222), 'pytest.warns', 'pytest.warns', (['UserWarning'], {}), '(UserWarning)\n', (75209, 75222), False, 'import pytest\n'), ((77289, 77359), 'riptable.Categorical', 'Categorical', (['matlab_float_idx', 'matlab_cats'], {'dtype': 'dt', 'from_matlab': '(True)'}), '(matlab_float_idx, matlab_cats, dtype=dt, from_matlab=True)\n', (77300, 77359), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((78494, 78519), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (78507, 78519), False, 'import pytest\n'), ((79024, 79050), 'riptable.FastArray', 'FastArray', (['[2, 3, 0, 4, 1]'], {}), '([2, 3, 0, 4, 1])\n', (79033, 79050), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((79065, 79088), 'riptable.FastArray', 'FastArray', (['[0, 0, 2, 4]'], {}), '([0, 0, 2, 4])\n', (79074, 79088), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((79103, 79126), 'riptable.FastArray', 'FastArray', (['[0, 2, 2, 1]'], {}), '([0, 2, 2, 1])\n', (79112, 79126), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((82348, 82357), 'riptable.Cat', 'rt.Cat', (['l'], {}), '(l)\n', (82354, 82357), True, 'import riptable as rt\n'), ((82359, 82372), 'riptable.Cat', 'rt.Cat', (['l[:3]'], {}), '(l[:3])\n', (82365, 82372), True, 'import riptable as rt\n'), ((86368, 86380), 'os.remove', 'os.remove', (['p'], {}), '(p)\n', (86377, 86380), False, 'import os\n'), ((98809, 98832), 'riptable.rt_numpy.arange', 'arange', (['(6)'], {'dtype': 'uint16'}), '(6, dtype=uint16)\n', (98815, 98832), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((103816, 103888), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shift_cat.category_array', 'categorical.category_array'], {}), '(shift_cat.category_array, categorical.category_array)\n', (103834, 103888), False, 'from numpy.testing import assert_array_equal\n'), ((107386, 107458), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shift_cat.category_array', 'categorical.category_array'], {}), '(shift_cat.category_array, categorical.category_array)\n', (107404, 107458), False, 'from numpy.testing import assert_array_equal\n'), ((110994, 111116), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shrink_cat.category_array', 'expected_category_array', 'f"""Category dictionary values should be empty."""'], {}), "(shrink_cat.category_array, expected_category_array,\n f'Category dictionary values should be empty.')\n", (111012, 111116), False, 'from numpy.testing import assert_array_equal\n'), ((112382, 112518), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shrink_cat.category_array', 'expected_category_array', 'f"""Category array should only contain the \'{misc}\' category."""'], {}), '(shrink_cat.category_array, expected_category_array,\n f"Category array should only contain the \'{misc}\' category.")\n', (112400, 112518), False, 'from numpy.testing import assert_array_equal\n'), ((113389, 113438), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shrink_cat_values', 'cat_values'], {}), '(shrink_cat_values, cat_values)\n', (113407, 113438), False, 'from numpy.testing import assert_array_equal\n'), ((113452, 113495), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shrink_cat._fa', 'cat._fa'], {}), '(shrink_cat._fa, cat._fa)\n', (113470, 113495), False, 'from numpy.testing import assert_array_equal\n'), ((113509, 113574), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shrink_cat.category_array', 'cat.category_array'], {}), '(shrink_cat.category_array, cat.category_array)\n', (113527, 113574), False, 'from numpy.testing import assert_array_equal\n'), ((115796, 115855), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['cat2.category_array', 'cat.category_array'], {}), '(cat2.category_array, cat.category_array)\n', (115814, 115855), False, 'from numpy.testing import assert_array_equal\n'), ((115964, 116000), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['actual', 'expected'], {}), '(actual, expected)\n', (115982, 116000), False, 'from numpy.testing import assert_array_equal\n'), ((116540, 116599), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['cat3.category_array', 'cat.category_array'], {}), '(cat3.category_array, cat.category_array)\n', (116558, 116599), False, 'from numpy.testing import assert_array_equal\n'), ((116708, 116744), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['actual', 'expected'], {}), '(actual, expected)\n', (116726, 116744), False, 'from numpy.testing import assert_array_equal\n'), ((118027, 118051), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (118040, 118051), False, 'import pytest\n'), ((118174, 118199), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (118187, 118199), False, 'import pytest\n'), ((118629, 118653), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (118642, 118653), False, 'import pytest\n'), ((120253, 120278), 'pytest.raises', 'pytest.raises', (['IndexError'], {}), '(IndexError)\n', (120266, 120278), False, 'import pytest\n'), ((121152, 121177), 'pytest.raises', 'pytest.raises', (['IndexError'], {}), '(IndexError)\n', (121165, 121177), False, 'import pytest\n'), ((121647, 121672), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (121660, 121672), False, 'import pytest\n'), ((121853, 121878), 'pytest.raises', 'pytest.raises', (['IndexError'], {}), '(IndexError)\n', (121866, 121878), False, 'import pytest\n'), ((122118, 122127), 'riptable.rt_numpy.arange', 'arange', (['(3)'], {}), '(3)\n', (122124, 122127), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((127651, 127683), 'riptable.Cat', 'rt.Cat', (['[1, 1, 2, 2]', "['a', 'a']"], {}), "([1, 1, 2, 2], ['a', 'a'])\n", (127657, 127683), True, 'import riptable as rt\n'), ((127685, 127712), 'riptable.Cat', 'rt.Cat', (['[1, 1, 1, 1]', "['a']"], {}), "([1, 1, 1, 1], ['a'])\n", (127691, 127712), True, 'import riptable as rt\n'), ((127721, 127753), 'riptable.Cat', 'rt.Cat', (['[2, 2, 2, 2]', "['a', 'a']"], {}), "([2, 2, 2, 2], ['a', 'a'])\n", (127727, 127753), True, 'import riptable as rt\n'), ((127755, 127782), 'riptable.Cat', 'rt.Cat', (['[1, 1, 1, 1]', "['a']"], {}), "([1, 1, 1, 1], ['a'])\n", (127761, 127782), True, 'import riptable as rt\n'), ((127791, 127828), 'riptable.Cat', 'rt.Cat', (['[1, 2, 3, 3]', "['a', 'a', 'b']"], {}), "([1, 2, 3, 3], ['a', 'a', 'b'])\n", (127797, 127828), True, 'import riptable as rt\n'), ((127830, 127862), 'riptable.Cat', 'rt.Cat', (['[1, 1, 2, 2]', "['a', 'b']"], {}), "([1, 1, 2, 2], ['a', 'b'])\n", (127836, 127862), True, 'import riptable as rt\n'), ((127871, 127917), 'riptable.Cat', 'rt.Cat', (['[0, 0, 1, 1]', "['a', 'a']"], {'base_index': '(0)'}), "([0, 0, 1, 1], ['a', 'a'], base_index=0)\n", (127877, 127917), True, 'import riptable as rt\n'), ((127919, 127960), 'riptable.Cat', 'rt.Cat', (['[0, 0, 0, 0]', "['a']"], {'base_index': '(0)'}), "([0, 0, 0, 0], ['a'], base_index=0)\n", (127925, 127960), True, 'import riptable as rt\n'), ((127969, 128015), 'riptable.Cat', 'rt.Cat', (['[1, 1, 1, 1]', "['a', 'a']"], {'base_index': '(0)'}), "([1, 1, 1, 1], ['a', 'a'], base_index=0)\n", (127975, 128015), True, 'import riptable as rt\n'), ((128017, 128058), 'riptable.Cat', 'rt.Cat', (['[0, 0, 0, 0]', "['a']"], {'base_index': '(0)'}), "([0, 0, 0, 0], ['a'], base_index=0)\n", (128023, 128058), True, 'import riptable as rt\n'), ((128067, 128113), 'riptable.Cat', 'rt.Cat', (['[0, 0, 1, 1]', "['a', 'b']"], {'base_index': '(0)'}), "([0, 0, 1, 1], ['a', 'b'], base_index=0)\n", (128073, 128113), True, 'import riptable as rt\n'), ((128115, 128161), 'riptable.Cat', 'rt.Cat', (['[0, 0, 1, 1]', "['a', 'b']"], {'base_index': '(0)'}), "([0, 0, 1, 1], ['a', 'b'], base_index=0)\n", (128121, 128161), True, 'import riptable as rt\n'), ((128172, 128210), 'riptable.Cat', 'rt.Cat', (['[1, 1, 2, 2, 3]', '[99, 99, 101]'], {}), '([1, 1, 2, 2, 3], [99, 99, 101])\n', (128178, 128210), True, 'import riptable as rt\n'), ((128214, 128248), 'riptable.Cat', 'rt.Cat', (['[1, 1, 1, 1, 2]', '[99, 101]'], {}), '([1, 1, 1, 1, 2], [99, 101])\n', (128220, 128248), True, 'import riptable as rt\n'), ((128257, 128301), 'riptable.Cat', 'rt.Cat', (['[0, 0, 1, 1]', '[99, 99]'], {'base_index': '(0)'}), '([0, 0, 1, 1], [99, 99], base_index=0)\n', (128263, 128301), True, 'import riptable as rt\n'), ((128303, 128343), 'riptable.Cat', 'rt.Cat', (['[0, 0, 0, 0]', '[99]'], {'base_index': '(0)'}), '([0, 0, 0, 0], [99], base_index=0)\n', (128309, 128343), True, 'import riptable as rt\n'), ((128352, 128397), 'riptable.Cat', 'rt.Cat', (['[0, 0, 1, 1]', '[99, 101]'], {'base_index': '(0)'}), '([0, 0, 1, 1], [99, 101], base_index=0)\n', (128358, 128397), True, 'import riptable as rt\n'), ((128399, 128444), 'riptable.Cat', 'rt.Cat', (['[0, 0, 1, 1]', '[99, 101]'], {'base_index': '(0)'}), '([0, 0, 1, 1], [99, 101], base_index=0)\n', (128405, 128444), True, 'import riptable as rt\n'), ((128455, 128493), 'riptable.Cat', 'rt.Cat', (['[0, 0, 1, 1, 2, 2]', "['a', 'a']"], {}), "([0, 0, 1, 1, 2, 2], ['a', 'a'])\n", (128461, 128493), True, 'import riptable as rt\n'), ((128497, 128530), 'riptable.Cat', 'rt.Cat', (['[0, 0, 1, 1, 1, 1]', "['a']"], {}), "([0, 0, 1, 1, 1, 1], ['a'])\n", (128503, 128530), True, 'import riptable as rt\n'), ((128541, 128590), 'riptable.Cat', 'rt.Cat', (['[0, 0, 1, 1, 2, 2, 3, 3]', "['a', 'a', 'b']"], {}), "([0, 0, 1, 1, 2, 2, 3, 3], ['a', 'a', 'b'])\n", (128547, 128590), True, 'import riptable as rt\n'), ((128594, 128638), 'riptable.Cat', 'rt.Cat', (['[0, 0, 1, 1, 1, 1, 2, 2]', "['a', 'b']"], {}), "([0, 0, 1, 1, 1, 1, 2, 2], ['a', 'b'])\n", (128600, 128638), True, 'import riptable as rt\n'), ((2218, 2253), 'riptable.tests.utils.LikertDecision.__members__.values', 'LikertDecision.__members__.values', ([], {}), '()\n', (2251, 2253), False, 'from riptable.tests.utils import LikertDecision\n'), ((17034, 17060), 'riptable.FastArray', 'FastArray', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (17043, 17060), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((23203, 23227), 'pytest.raises', 'pytest.raises', (['TypeError'], {}), '(TypeError)\n', (23216, 23227), False, 'import pytest\n'), ((23250, 23276), 'riptable.Categorical', 'Categorical', (['arr'], {'dtype': 'dt'}), '(arr, dtype=dt)\n', (23261, 23276), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((23898, 23914), 'riptable.FastArray', 'FastArray', (['[144]'], {}), '([144])\n', (23907, 23914), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((40807, 40825), 'riptable.Categorical', 'Categorical', (["['a']"], {}), "(['a'])\n", (40818, 40825), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((40827, 40841), 'riptable.FastArray', 'FastArray', (['[1]'], {}), '([1])\n', (40836, 40841), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((40871, 40889), 'riptable.Categorical', 'Categorical', (["['b']"], {}), "(['b'])\n", (40882, 40889), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((40891, 40905), 'riptable.FastArray', 'FastArray', (['[2]'], {}), '([2])\n', (40900, 40905), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((66538, 66552), 'riptable.FastArray', 'FastArray', (['[1]'], {}), '([1])\n', (66547, 66552), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((66554, 66568), 'riptable.FastArray', 'FastArray', (['[2]'], {}), '([2])\n', (66563, 66568), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((66570, 66584), 'riptable.FastArray', 'FastArray', (['[3]'], {}), '([3])\n', (66579, 66584), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((66721, 66763), 'riptable.FastArray', 'FastArray', (['[True, True, False, True, True]'], {}), '([True, True, False, True, True])\n', (66730, 66763), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((68716, 68742), 'riptable.FastArray', 'FastArray', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (68725, 68742), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((68744, 68764), 'riptable.FastArray', 'FastArray', (['[1, 2, 3]'], {}), '([1, 2, 3])\n', (68753, 68764), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((71224, 71255), 'riptable.FastArray', 'FastArray', (["['a', 'b', 'c', 'a']"], {}), "(['a', 'b', 'c', 'a'])\n", (71233, 71255), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((71257, 71280), 'riptable.FastArray', 'FastArray', (['[1, 2, 3, 1]'], {}), '([1, 2, 3, 1])\n', (71266, 71280), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((71745, 71768), 'riptable.FastArray', 'FastArray', (['[1, 2, 3, 1]'], {}), '([1, 2, 3, 1])\n', (71754, 71768), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((72297, 72307), 'riptable.rt_numpy.arange', 'arange', (['(10)'], {}), '(10)\n', (72303, 72307), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((73806, 73831), 'riptable.rt_numpy.arange', 'arange', (['(5)'], {'dtype': 'np.int32'}), '(5, dtype=np.int32)\n', (73812, 73831), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((73833, 73858), 'riptable.rt_numpy.arange', 'arange', (['(5)'], {'dtype': 'np.int32'}), '(5, dtype=np.int32)\n', (73839, 73858), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((73925, 73950), 'riptable.rt_numpy.arange', 'arange', (['(5)'], {'dtype': 'np.int64'}), '(5, dtype=np.int64)\n', (73931, 73950), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((73952, 73977), 'riptable.rt_numpy.arange', 'arange', (['(5)'], {'dtype': 'np.int64'}), '(5, dtype=np.int64)\n', (73958, 73977), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((75461, 75492), 'riptable.FastArray', 'FastArray', (["['a', 'b', 'c', 'a']"], {}), "(['a', 'b', 'c', 'a'])\n", (75470, 75492), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((75494, 75517), 'riptable.FastArray', 'FastArray', (['[1, 2, 3, 1]'], {}), '([1, 2, 3, 1])\n', (75503, 75517), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((76003, 76025), 'riptable.rt_numpy.isnan', 'isnan', (['multi_expand[1]'], {}), '(multi_expand[1])\n', (76008, 76025), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((77700, 77744), 'riptable.FastArray', 'FastArray', (['[False, False, True, True, False]'], {}), '([False, False, True, True, False])\n', (77709, 77744), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((78083, 78120), 'riptable.FastArray', 'FastArray', (['[False, False, True, True]'], {}), '([False, False, True, True])\n', (78092, 78120), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((80725, 80734), 'riptable.rt_numpy.arange', 'arange', (['(5)'], {}), '(5)\n', (80731, 80734), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((80736, 80745), 'riptable.rt_numpy.arange', 'arange', (['(5)'], {}), '(5)\n', (80742, 80745), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((81932, 81941), 'riptable.rt_numpy.arange', 'arange', (['(3)'], {}), '(3)\n', (81938, 81941), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((100134, 100151), 'riptable.Categorical', 'Categorical', (['data'], {}), '(data)\n', (100145, 100151), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((101077, 101094), 'riptable.Categorical', 'Categorical', (['data'], {}), '(data)\n', (101088, 101094), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((104046, 104164), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shift_cat._fa[i:]', 'categorical._fa[:-i]', 'f"""FastArray items should be shifted by {i} postions."""'], {}), "(shift_cat._fa[i:], categorical._fa[:-i],\n f'FastArray items should be shifted by {i} postions.')\n", (104064, 104164), False, 'from numpy.testing import assert_array_equal\n'), ((104470, 104590), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shift_cat_values[i:]', 'cat_values[:-i]', 'f"""Categorical values should be shifted by {i} positions."""'], {}), "(shift_cat_values[i:], cat_values[:-i],\n f'Categorical values should be shifted by {i} positions.')\n", (104488, 104590), False, 'from numpy.testing import assert_array_equal\n'), ((102558, 102575), 'riptable.Categorical', 'Categorical', (['data'], {}), '(data)\n', (102569, 102575), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((111222, 111322), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['arr', 'expected_category_array', 'f"""Category dictionary values should be empty."""'], {}), "(arr, expected_category_array,\n f'Category dictionary values should be empty.')\n", (111240, 111322), False, 'from numpy.testing import assert_array_equal\n'), ((112658, 112784), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['arr', 'expected_category_array', 'f"""Category dictionary values should only contain the \'{misc}\' category."""'], {}), '(arr, expected_category_array,\n f"Category dictionary values should only contain the \'{misc}\' category.")\n', (112676, 112784), False, 'from numpy.testing import assert_array_equal\n'), ((113698, 113735), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['arr', 'expected_arr'], {}), '(arr, expected_arr)\n', (113716, 113735), False, 'from numpy.testing import assert_array_equal\n'), ((109339, 109356), 'riptable.Categorical', 'Categorical', (['data'], {}), '(data)\n', (109350, 109356), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((114378, 114395), 'riptable.Categorical', 'Categorical', (['data'], {}), '(data)\n', (114389, 114395), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((117008, 117025), 'riptable.Categorical', 'Categorical', (['data'], {}), '(data)\n', (117019, 117025), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((117179, 117210), 'riptable.Categorical', 'Categorical', (['data'], {'base_index': '(0)'}), '(data, base_index=0)\n', (117190, 117210), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((119061, 119078), 'riptable.Categorical', 'Categorical', (['data'], {}), '(data)\n', (119072, 119078), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((119263, 119280), 'riptable.Categorical', 'Categorical', (['data'], {}), '(data)\n', (119274, 119280), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((8470, 8495), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (8483, 8495), False, 'import pytest\n'), ((8524, 8541), 'riptable.Categorical', 'Categorical', (['v', 'c'], {}), '(v, c)\n', (8535, 8541), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((13923, 13948), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (13936, 13948), False, 'import pytest\n'), ((40234, 40300), 'riptable.FastArray', 'FastArray', (["[b'Filtered', b'a', b'b', b'c', b'd', b'e', b'f', b'z']"], {}), "([b'Filtered', b'a', b'b', b'c', b'd', b'e', b'f', b'z'])\n", (40243, 40300), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((59839, 59847), 'riptable.rt_numpy.isnan', 'isnan', (['c'], {}), '(c)\n', (59844, 59847), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((59950, 59961), 'riptable.rt_numpy.isnotnan', 'isnotnan', (['c'], {}), '(c)\n', (59958, 59961), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((70319, 70350), 'riptable.FastArray', 'FastArray', (["['a', 'b', 'c', 'a']"], {}), "(['a', 'b', 'c', 'a'])\n", (70328, 70350), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((70352, 70375), 'riptable.FastArray', 'FastArray', (['[1, 2, 3, 1]'], {}), '([1, 2, 3, 1])\n', (70361, 70375), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((70396, 70422), 'riptable.FastArray', 'FastArray', (["['a', 'b', 'c']"], {}), "(['a', 'b', 'c'])\n", (70405, 70422), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((70424, 70444), 'riptable.FastArray', 'FastArray', (['[1, 2, 3]'], {}), '([1, 2, 3])\n', (70433, 70444), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((71964, 72001), 'riptable.FastArray', 'FastArray', (['[True, False, False, True]'], {}), '([True, False, False, True])\n', (71973, 72001), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((75106, 75118), 'riptable.rt_numpy.arange', 'arange', (['(1)', '(4)'], {}), '(1, 4)\n', (75112, 75118), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((75418, 75430), 'riptable.rt_numpy.arange', 'arange', (['(1)', '(4)'], {}), '(1, 4)\n', (75424, 75430), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((75786, 75817), 'riptable.FastArray', 'FastArray', (["['a', 'b', 'c', 'a']"], {}), "(['a', 'b', 'c', 'a'])\n", (75795, 75817), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((75867, 75890), 'riptable.FastArray', 'FastArray', (['[1, 2, 3, 1]'], {}), '([1, 2, 3, 1])\n', (75876, 75890), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((78614, 78651), 'riptable.FastArray', 'FastArray', (['[False, False, True, True]'], {}), '([False, False, True, True])\n', (78623, 78651), False, 'from riptable import FastArray, Categorical, CatZero\n'), ((78905, 78917), 'riptable.rt_numpy.arange', 'arange', (['(1)', '(4)'], {}), '(1, 4)\n', (78911, 78917), False, 'from riptable.rt_numpy import isnan, isnotnan, arange, ones\n'), ((100181, 100276), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray, CategoryMode.NumericArray]'], {}), '([CategoryMode.StringArray, CategoryMode\n .NumericArray])\n', (100216, 100276), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((100504, 100599), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray, CategoryMode.NumericArray]'], {}), '([CategoryMode.StringArray, CategoryMode\n .NumericArray])\n', (100539, 100599), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((100795, 100813), 'riptable.FastArray', 'rt.FastArray', (['data'], {}), '(data)\n', (100807, 100813), True, 'import riptable as rt\n'), ((100844, 100939), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray, CategoryMode.NumericArray]'], {}), '([CategoryMode.StringArray, CategoryMode\n .NumericArray])\n', (100879, 100939), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((101107, 101165), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['CategoryMode.MultiKey'], {}), '(CategoryMode.MultiKey)\n', (101142, 101165), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((101206, 101358), 'pytest.mark.xfail', 'pytest.mark.xfail', ([], {'reason': '"""RIP-410 - Bug for MultiKey Categoricals: AttributeError: \'Categorical\' object has no attribute \'ismultikey_labels\'"""'}), '(reason=\n "RIP-410 - Bug for MultiKey Categoricals: AttributeError: \'Categorical\' object has no attribute \'ismultikey_labels\'"\n )\n', (101223, 101358), False, 'import pytest\n'), ((105580, 105708), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shift_cat._fa[:cat_len - i]', 'categorical._fa[i:]', 'f"""FastArray items should be shifted by -{i} postions."""'], {}), "(shift_cat._fa[:cat_len - i], categorical._fa[i:],\n f'FastArray items should be shifted by -{i} postions.')\n", (105598, 105708), False, 'from numpy.testing import assert_array_equal\n'), ((105967, 106097), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shift_cat_values[:cat_len - i]', 'cat_values[i:]', 'f"""Categorical values should be shifted by -{i} positions."""'], {}), "(shift_cat_values[:cat_len - i], cat_values[i:],\n f'Categorical values should be shifted by -{i} positions.')\n", (105985, 106097), False, 'from numpy.testing import assert_array_equal\n'), ((102588, 102651), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray]'], {}), '([CategoryMode.StringArray])\n', (102623, 102651), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((102835, 102898), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray]'], {}), '([CategoryMode.StringArray])\n', (102870, 102898), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((103085, 103103), 'riptable.FastArray', 'rt.FastArray', (['data'], {}), '(data)\n', (103097, 103103), True, 'import riptable as rt\n'), ((103134, 103197), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray]'], {}), '([CategoryMode.StringArray])\n', (103169, 103197), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((109369, 109432), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray]'], {}), '([CategoryMode.StringArray])\n', (109404, 109432), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((109616, 109679), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray]'], {}), '([CategoryMode.StringArray])\n', (109651, 109679), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((109866, 109884), 'riptable.FastArray', 'rt.FastArray', (['data'], {}), '(data)\n', (109878, 109884), True, 'import riptable as rt\n'), ((109915, 109978), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray]'], {}), '([CategoryMode.StringArray])\n', (109950, 109978), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((114408, 114445), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', ([], {}), '()\n', (114443, 114445), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((114641, 114736), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray, CategoryMode.NumericArray]'], {}), '([CategoryMode.StringArray, CategoryMode\n .NumericArray])\n', (114676, 114736), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((114932, 114950), 'riptable.FastArray', 'rt.FastArray', (['data'], {}), '(data)\n', (114944, 114950), True, 'import riptable as rt\n'), ((114981, 115076), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray, CategoryMode.NumericArray]'], {}), '([CategoryMode.StringArray, CategoryMode\n .NumericArray])\n', (115016, 115076), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((117051, 117146), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray, CategoryMode.NumericArray]'], {}), '([CategoryMode.StringArray, CategoryMode\n .NumericArray])\n', (117086, 117146), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((117236, 117331), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray, CategoryMode.NumericArray]'], {}), '([CategoryMode.StringArray, CategoryMode\n .NumericArray])\n', (117271, 117331), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((119108, 119203), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['[CategoryMode.StringArray, CategoryMode.NumericArray]'], {}), '([CategoryMode.StringArray, CategoryMode\n .NumericArray])\n', (119143, 119203), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((119293, 119351), 'riptable.tests.test_utils.get_categorical_data_factory_method', 'get_categorical_data_factory_method', (['CategoryMode.MultiKey'], {}), '(CategoryMode.MultiKey)\n', (119328, 119351), False, 'from riptable.tests.test_utils import get_categorical_data_factory_method, get_all_categorical_data\n'), ((119392, 119505), 'pytest.mark.xfail', 'pytest.mark.xfail', ([], {'reason': '"""NotImplementedError: Add categories not supported for MultiKey Categoricals"""'}), "(reason=\n 'NotImplementedError: Add categories not supported for MultiKey Categoricals'\n )\n", (119409, 119505), False, 'import pytest\n'), ((106789, 106861), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shift_cat.category_array', 'categorical.category_array'], {}), '(shift_cat.category_array, categorical.category_array)\n', (106807, 106861), False, 'from numpy.testing import assert_array_equal\n'), ((106879, 106929), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shift_cat._fa', 'categorical._fa'], {}), '(shift_cat._fa, categorical._fa)\n', (106897, 106929), False, 'from numpy.testing import assert_array_equal\n'), ((107108, 107156), 'numpy.testing.assert_array_equal', 'assert_array_equal', (['shift_cat_values', 'cat_values'], {}), '(shift_cat_values, cat_values)\n', (107126, 107156), False, 'from numpy.testing import assert_array_equal\n')] |
import numpy as np
from scipy import io, sparse, linalg
# run this from elegant scipy chapter
chem = np.load('chem-network.npy')
gap = np.load('gap-network.npy')
neuron_types = np.load('neuron-types.npy')
neuron_ids = np.load('neurons.npy')
A = chem + gap
n = A.shape[0]
c = (A + A.T) / 2
d = sparse.diags([np.sum(c, axis=0)], [0])
d = d.toarray()
L = np.array(d - c)
b = np.sum(c * np.sign(A - A.T), axis=1)
z = np.linalg.pinv(L) @ b
# IPython log file
dinv2 = np.copy(d)
diag = (np.arange(n), np.arange(n))
dinv2[diag] = dinv[diag] ** (-.5)
q = dinv2 @ L @ dinv2
eigvals, vec = linalg.eig(q)
x = dinv2 @ vec[:, 1]
x.shape
from matplotlib import pyplot as plt
from matplotlib import colors
ii, jj = np.nonzero(c)
plt.scatter(x, z, c=neuron_types, cmap=colors.ListedColormap(((1, 0, 0), (0, 1, 0), (0, 0, 1))), zorder=1)
for src, dst in zip(ii, jj):
plt.plot(x[[src, dst]], z[[src, dst]], c=(0.85, 0.85, 0.85), lw=0.2, alpha=0.5, zorder=0)
for x0, z0, neuron_id in zip(x, z, neuron_ids):
plt.text(x0, z0, ' ' + neuron_id,
horizontalalignment='left', verticalalignment='center',
fontsize=4, zorder=2)
| [
"numpy.load",
"numpy.sum",
"numpy.copy",
"matplotlib.pyplot.plot",
"scipy.linalg.eig",
"numpy.nonzero",
"matplotlib.pyplot.text",
"numpy.array",
"numpy.arange",
"numpy.sign",
"numpy.linalg.pinv",
"matplotlib.colors.ListedColormap"
] | [((102, 129), 'numpy.load', 'np.load', (['"""chem-network.npy"""'], {}), "('chem-network.npy')\n", (109, 129), True, 'import numpy as np\n'), ((136, 162), 'numpy.load', 'np.load', (['"""gap-network.npy"""'], {}), "('gap-network.npy')\n", (143, 162), True, 'import numpy as np\n'), ((178, 205), 'numpy.load', 'np.load', (['"""neuron-types.npy"""'], {}), "('neuron-types.npy')\n", (185, 205), True, 'import numpy as np\n'), ((219, 241), 'numpy.load', 'np.load', (['"""neurons.npy"""'], {}), "('neurons.npy')\n", (226, 241), True, 'import numpy as np\n'), ((353, 368), 'numpy.array', 'np.array', (['(d - c)'], {}), '(d - c)\n', (361, 368), True, 'import numpy as np\n'), ((463, 473), 'numpy.copy', 'np.copy', (['d'], {}), '(d)\n', (470, 473), True, 'import numpy as np\n'), ((581, 594), 'scipy.linalg.eig', 'linalg.eig', (['q'], {}), '(q)\n', (591, 594), False, 'from scipy import io, sparse, linalg\n'), ((701, 714), 'numpy.nonzero', 'np.nonzero', (['c'], {}), '(c)\n', (711, 714), True, 'import numpy as np\n'), ((414, 431), 'numpy.linalg.pinv', 'np.linalg.pinv', (['L'], {}), '(L)\n', (428, 431), True, 'import numpy as np\n'), ((482, 494), 'numpy.arange', 'np.arange', (['n'], {}), '(n)\n', (491, 494), True, 'import numpy as np\n'), ((496, 508), 'numpy.arange', 'np.arange', (['n'], {}), '(n)\n', (505, 508), True, 'import numpy as np\n'), ((855, 949), 'matplotlib.pyplot.plot', 'plt.plot', (['x[[src, dst]]', 'z[[src, dst]]'], {'c': '(0.85, 0.85, 0.85)', 'lw': '(0.2)', 'alpha': '(0.5)', 'zorder': '(0)'}), '(x[[src, dst]], z[[src, dst]], c=(0.85, 0.85, 0.85), lw=0.2, alpha=\n 0.5, zorder=0)\n', (863, 949), True, 'from matplotlib import pyplot as plt\n'), ((997, 1113), 'matplotlib.pyplot.text', 'plt.text', (['x0', 'z0', "(' ' + neuron_id)"], {'horizontalalignment': '"""left"""', 'verticalalignment': '"""center"""', 'fontsize': '(4)', 'zorder': '(2)'}), "(x0, z0, ' ' + neuron_id, horizontalalignment='left',\n verticalalignment='center', fontsize=4, zorder=2)\n", (1005, 1113), True, 'from matplotlib import pyplot as plt\n'), ((308, 325), 'numpy.sum', 'np.sum', (['c'], {'axis': '(0)'}), '(c, axis=0)\n', (314, 325), True, 'import numpy as np\n'), ((384, 400), 'numpy.sign', 'np.sign', (['(A - A.T)'], {}), '(A - A.T)\n', (391, 400), True, 'import numpy as np\n'), ((754, 810), 'matplotlib.colors.ListedColormap', 'colors.ListedColormap', (['((1, 0, 0), (0, 1, 0), (0, 0, 1))'], {}), '(((1, 0, 0), (0, 1, 0), (0, 0, 1)))\n', (775, 810), False, 'from matplotlib import colors\n')] |
'''
Define the operations used to denoise the image.
'''
import cv2
import numpy as np
def denoise(frame, useMorphOps = True, useGaussianBlur = True):
if useMorphOps:
kernel = np.ones((5,5),np.uint8)
frame = cv2.morphologyEx(frame, cv2.MORPH_OPEN, kernel)
frame = cv2.morphologyEx(frame, cv2.MORPH_CLOSE, kernel)
if useGaussianBlur:
frame = cv2.GaussianBlur(frame, (5, 5), 0)
return frame
| [
"cv2.morphologyEx",
"numpy.ones",
"cv2.GaussianBlur"
] | [((189, 214), 'numpy.ones', 'np.ones', (['(5, 5)', 'np.uint8'], {}), '((5, 5), np.uint8)\n', (196, 214), True, 'import numpy as np\n'), ((229, 276), 'cv2.morphologyEx', 'cv2.morphologyEx', (['frame', 'cv2.MORPH_OPEN', 'kernel'], {}), '(frame, cv2.MORPH_OPEN, kernel)\n', (245, 276), False, 'import cv2\n'), ((293, 341), 'cv2.morphologyEx', 'cv2.morphologyEx', (['frame', 'cv2.MORPH_CLOSE', 'kernel'], {}), '(frame, cv2.MORPH_CLOSE, kernel)\n', (309, 341), False, 'import cv2\n'), ((382, 416), 'cv2.GaussianBlur', 'cv2.GaussianBlur', (['frame', '(5, 5)', '(0)'], {}), '(frame, (5, 5), 0)\n', (398, 416), False, 'import cv2\n')] |
""" ###################### EDGE ####################"""
import numpy as np
from collections import defaultdict
class vertex:
def __init__(self, type, node, id):
"""
:param node:
"""
self.id = id
self.Type = type
self.Cells = node
self.Trains = []
self.TrainsDir = [] # 0 = A->B, 1 = B->A
self.Links = []
self.TrainsTraversal = defaultdict(list)
self.is_signal_on = False
self.is_starting_edge = False
self.is_safe = True
self.occupancy = 0
self.extended_capacity = len(node)
self.capacity = len(node)
def __str__(self):
"""
:return:
"""
return 'Type : ' + str(self.Type) \
+ '; Trains: ' + str(self.Trains) \
+ '; Safety Status: ' + str(self.is_safe)
def other_end(self, first):
return self.Cells[0] if self.Cells[-1] == first else self.Cells[-1]
def setCosts(self):
"""
:return:
"""
#self.setCollision()
#self.CostCollisionLockTotal = 0
#self.CostTransitionTimeTotal = 0
#self.CostDeadLockTotal = 0
#self.CostTotal = 0
#self.CostPerTrain = []
#self.DeadlockCostPerTrain = []
#if self.signal_time > 1:
# self.signal_time -= 1
#elif self.signal_time == 1:
# self.signal_time -= 1
# self.signal_deadlocks = []
#start_times = [item[0] for item in self.TrainsTime]
agent_dirs = np.unique(self.TrainsDir)
if self.is_starting_edge:
self.is_safe = True
else:
self.is_safe = True if len(agent_dirs) <= 1 else False
#self.is_safe = True if len(np.unique(self.TrainsDir)) <= 1 else False
def setExtendedCapacity(self):
"""
:return:
"""
if self.is_safe:
pending_to_explore = []
explored = []
explored.append(self.id)
for vertex in self.Links:
if vertex[1].is_safe:
pending_to_explore.append(vertex[1])
capacity = 0
while len(pending_to_explore):
vertex = pending_to_explore.pop()
explored.append(vertex.id)
capacity += len(vertex.Cells)
for next_vertex in vertex.Links:
if next_vertex[1].is_safe and next_vertex[1].id not in explored:
pending_to_explore.append(next_vertex[1])
self.extended_capacity = capacity
if self.extended_capacity > 2*self.capacity:
self.extended_capacity = 2*self.capacity
class Global_Graph:
def __init__(self):
"""
"""
self.vertices = {}
self.num_vertices = 0
#self.Deadlocks = []
#self.LastUpdated = 0
#self.CostTotalEnv = 0
def __str__(self):
"""
:return:
"""
return 'Cost: ' + str(self.CostTotalEnv) + ' Deadlocks: ' + str(self.Deadlocks)
def setCosts(self):
"""
:return:
"""
#cost = 0
for vertex in self.vertices:
if len(self.vertices[vertex].Trains):
self.vertices[vertex].setCosts()
#cost += self.vertices[vertex].CostTotal
#for vertex in self.vertices:
# if len(self.vertices[vertex].Trains):
# self.vertices[vertex].setExtendedCapacity()
#self.CostTotalEnv = cost
def add_edge_vertex(self, type, cells):
"""
:param node:
:return:
"""
if str(cells[0])[1:-1]+","+str(cells[-1])[1:-1] not in self.vertices\
and str(cells[-1])[1:-1]+","+str(cells[0])[1:-1] not in self.vertices:
new_edge_vertex = vertex(type, cells, str(cells[0])[1:-1]+","+str(cells[-1])[1:-1])
self.vertices[str(cells[0])[1:-1]+","+str(cells[-1])[1:-1]] = new_edge_vertex
self.num_vertices += 1
return new_edge_vertex
elif str(cells[0])[1:-1]+","+str(cells[-1])[1:-1] in self.vertices:
return self.vertices[str(cells[0])[1:-1]+","+str(cells[-1])[1:-1]]
elif str(cells[-1])[1:-1]+","+str(cells[0])[1:-1] in self.vertices:
return self.vertices[str(cells[-1])[1:-1]+","+str(cells[0])[1:-1]]
def add_signal_vertex(self, type, node):
"""
:param node:
:return:
"""
if str(node)[1:-1] not in self.vertices:
new_vertex = vertex(type, [node], str(node)[1:-1])
self.vertices[str(node)[1:-1]] = new_vertex
self.num_vertices = self.num_vertices + 1
return new_vertex
return self.vertices[str(node)[1:-1]]
if __name__ == "__main__":
# create a graph of 4 nodes
#
# if a node is added - only node list is updated
# call graph insert method
# if an edge is added - possibly two nodes will be added
g = Global_Graph()
g.add_vertex('a')
g.add_edge('a','b')
g.add_edge('a','c')
g.add_edge('b','c')
g.add_edge('b','d')
g.add_edge('c','d')
source_vert = g.vert_dict['a']
for edge in g.vert_dict['a'].edges:
if edge.end == 'c':
edge.cost_triples.append([1,2,3])
#print("found")
#edge_temp = g.edge_dict['ab']
#edge_temp.cost_triples.append([1,2,3])
#g.add_vertex('b')
#g.add_vertex('c')
#g.add_vertex('d')
#g.add_vertex('e')
#print("done") | [
"collections.defaultdict",
"numpy.unique"
] | [((417, 434), 'collections.defaultdict', 'defaultdict', (['list'], {}), '(list)\n', (428, 434), False, 'from collections import defaultdict\n'), ((1552, 1577), 'numpy.unique', 'np.unique', (['self.TrainsDir'], {}), '(self.TrainsDir)\n', (1561, 1577), True, 'import numpy as np\n')] |
import numpy as np
from matplotlib import pyplot as plt, gridspec as gridspec
import seaborn as sns
import matplotlib as mpl
import matplotlib.cm as cm
from rl_agents.utils import remap, constrain
class DQNGraphics(object):
"""
Graphical visualization of the DQNAgent state-action values.
"""
RED = (255, 0, 0)
BLACK = (0, 0, 0)
MIN_ATTENTION = 0.01
@classmethod
def display(cls, agent, surface, sim_surface=None, display_text=True):
"""
Display the action-values for the current state
:param agent: the DQNAgent to be displayed
:param surface: the pygame surface on which the agent is displayed
:param sim_surface: the pygame surface on which the env is rendered
:param display_text: whether to display the action values as text
"""
import pygame
action_values = agent.get_state_action_values(agent.previous_state)
action_distribution = agent.action_distribution(agent.previous_state)
cell_size = (surface.get_width() // len(action_values), surface.get_height())
pygame.draw.rect(surface, cls.BLACK, (0, 0, surface.get_width(), surface.get_height()), 0)
# Display node value
for action, value in enumerate(action_values):
cmap = cm.jet_r
norm = mpl.colors.Normalize(vmin=0, vmax=1/(1-agent.config["gamma"]))
color = cmap(norm(value), bytes=True)
pygame.draw.rect(surface, color, (cell_size[0]*action, 0, cell_size[0], cell_size[1]), 0)
if display_text:
font = pygame.font.Font(None, 15)
text = "v={:.2f} / p={:.2f}".format(value, action_distribution[action])
text = font.render(text,
1, (10, 10, 10), (255, 255, 255))
surface.blit(text, (cell_size[0]*action, 0))
if sim_surface and hasattr(agent.value_net, "get_attention_matrix"):
cls.display_vehicles_attention(agent, sim_surface)
@classmethod
def display_vehicles_attention(cls, agent, sim_surface):
import pygame
try:
state = agent.previous_state
if (not hasattr(cls, "state")) or (cls.state != state).any():
cls.v_attention = cls.compute_vehicles_attention(agent, state)
cls.state = state
for head in range(list(cls.v_attention.values())[0].shape[0]):
attention_surface = pygame.Surface(sim_surface.get_size(), pygame.SRCALPHA)
for vehicle, attention in cls.v_attention.items():
if attention[head] < cls.MIN_ATTENTION:
continue
width = attention[head] * 5
desat = remap(attention[head], (0, 0.5), (0.7, 1), clip=True)
colors = sns.color_palette("dark", desat=desat)
color = np.array(colors[(2*head) % (len(colors) - 1)]) * 255
color = (*color, remap(attention[head], (0, 0.5), (100, 200), clip=True))
if vehicle is agent.env.vehicle:
pygame.draw.circle(attention_surface, color,
sim_surface.vec2pix(agent.env.vehicle.position),
max(sim_surface.pix(width / 2), 1))
else:
pygame.draw.line(attention_surface, color,
sim_surface.vec2pix(agent.env.vehicle.position),
sim_surface.vec2pix(vehicle.position),
max(sim_surface.pix(width), 1))
sim_surface.blit(attention_surface, (0, 0))
except ValueError as e:
print("Unable to display vehicles attention", e)
@classmethod
def compute_vehicles_attention(cls, agent, state):
import torch
state_t = torch.tensor([state], dtype=torch.float).to(agent.device)
attention = agent.value_net.get_attention_matrix(state_t).squeeze(0).squeeze(1).detach().cpu().numpy()
ego, others, mask = agent.value_net.split_input(state_t)
mask = mask.squeeze()
v_attention = {}
for v_index in range(state.shape[0]):
if mask[v_index]:
continue
v_position = {}
for feature in ["x", "y"]:
v_feature = state[v_index, agent.env.observation_type.features.index(feature)]
v_feature = remap(v_feature, [-1, 1], agent.env.observation_type.features_range[feature])
v_position[feature] = v_feature
v_position = np.array([v_position["x"], v_position["y"]])
if not agent.env.observation_type.absolute and v_index > 0:
v_position += agent.env.unwrapped.vehicle.position
vehicle = min(agent.env.road.vehicles, key=lambda v: np.linalg.norm(v.position - v_position))
v_attention[vehicle] = attention[:, v_index]
return v_attention
class ValueFunctionViewer(object):
def __init__(self, agent, state_sampler):
self.agent = agent
self.state_sampler = state_sampler
self.values_history = np.array([])
self.figure = None
self.axes = []
def display(self):
if not self.state_sampler:
return
if not self.figure:
plt.ion()
self.figure = plt.figure('Value function')
gs = gridspec.GridSpec(2, 2)
self.axes.append(plt.subplot(gs[0, :]))
self.axes.append(plt.subplot(gs[1, 0]))
self.axes.append(plt.subplot(gs[1, 1]))
xx, _, _ = self.state_sampler.states_mesh()
cax1 = self.axes[1].imshow(xx)
cax2 = self.axes[2].imshow(xx)
self.figure.colorbar(cax1, ax=self.axes[1])
self.figure.colorbar(cax2, ax=self.axes[2])
self.plot_values()
self.plot_value_map()
def plot_value_map(self):
xx, yy, states = self.state_sampler.states_mesh()
values, actions = self.agent.get_batch_state_values(states)
values, actions = np.reshape(values, np.shape(xx)), np.reshape(actions, np.shape(xx))
self.axes[1].clear()
self.axes[2].clear()
self.axes[1].imshow(values)
self.axes[2].imshow(actions)
plt.pause(0.001)
plt.draw()
def plot_values(self):
states = self.state_sampler.states_list()
values, _ = self.agent.get_batch_state_values(states)
self.values_history = np.vstack((self.values_history, values)) if self.values_history.size else values
self.axes[0].clear()
self.axes[0].set_xlabel('Episode')
self.axes[0].set_ylabel('Value')
self.axes[0].plot(self.values_history)
plt.pause(0.001)
plt.draw()
| [
"matplotlib.pyplot.subplot",
"torch.tensor",
"matplotlib.colors.Normalize",
"pygame.draw.rect",
"matplotlib.pyplot.draw",
"matplotlib.pyplot.ion",
"matplotlib.pyplot.figure",
"rl_agents.utils.remap",
"numpy.array",
"numpy.shape",
"seaborn.color_palette",
"pygame.font.Font",
"numpy.linalg.nor... | [((5249, 5261), 'numpy.array', 'np.array', (['[]'], {}), '([])\n', (5257, 5261), True, 'import numpy as np\n'), ((6396, 6412), 'matplotlib.pyplot.pause', 'plt.pause', (['(0.001)'], {}), '(0.001)\n', (6405, 6412), True, 'from matplotlib import pyplot as plt, gridspec as gridspec\n'), ((6421, 6431), 'matplotlib.pyplot.draw', 'plt.draw', ([], {}), '()\n', (6429, 6431), True, 'from matplotlib import pyplot as plt, gridspec as gridspec\n'), ((6852, 6868), 'matplotlib.pyplot.pause', 'plt.pause', (['(0.001)'], {}), '(0.001)\n', (6861, 6868), True, 'from matplotlib import pyplot as plt, gridspec as gridspec\n'), ((6877, 6887), 'matplotlib.pyplot.draw', 'plt.draw', ([], {}), '()\n', (6885, 6887), True, 'from matplotlib import pyplot as plt, gridspec as gridspec\n'), ((1329, 1395), 'matplotlib.colors.Normalize', 'mpl.colors.Normalize', ([], {'vmin': '(0)', 'vmax': "(1 / (1 - agent.config['gamma']))"}), "(vmin=0, vmax=1 / (1 - agent.config['gamma']))\n", (1349, 1395), True, 'import matplotlib as mpl\n'), ((1454, 1549), 'pygame.draw.rect', 'pygame.draw.rect', (['surface', 'color', '(cell_size[0] * action, 0, cell_size[0], cell_size[1])', '(0)'], {}), '(surface, color, (cell_size[0] * action, 0, cell_size[0],\n cell_size[1]), 0)\n', (1470, 1549), False, 'import pygame\n'), ((4692, 4736), 'numpy.array', 'np.array', (["[v_position['x'], v_position['y']]"], {}), "([v_position['x'], v_position['y']])\n", (4700, 4736), True, 'import numpy as np\n'), ((5430, 5439), 'matplotlib.pyplot.ion', 'plt.ion', ([], {}), '()\n', (5437, 5439), True, 'from matplotlib import pyplot as plt, gridspec as gridspec\n'), ((5466, 5494), 'matplotlib.pyplot.figure', 'plt.figure', (['"""Value function"""'], {}), "('Value function')\n", (5476, 5494), True, 'from matplotlib import pyplot as plt, gridspec as gridspec\n'), ((5512, 5535), 'matplotlib.gridspec.GridSpec', 'gridspec.GridSpec', (['(2)', '(2)'], {}), '(2, 2)\n', (5529, 5535), True, 'from matplotlib import pyplot as plt, gridspec as gridspec\n'), ((6602, 6642), 'numpy.vstack', 'np.vstack', (['(self.values_history, values)'], {}), '((self.values_history, values))\n', (6611, 6642), True, 'import numpy as np\n'), ((1597, 1623), 'pygame.font.Font', 'pygame.font.Font', (['None', '(15)'], {}), '(None, 15)\n', (1613, 1623), False, 'import pygame\n'), ((3961, 4001), 'torch.tensor', 'torch.tensor', (['[state]'], {'dtype': 'torch.float'}), '([state], dtype=torch.float)\n', (3973, 4001), False, 'import torch\n'), ((4541, 4618), 'rl_agents.utils.remap', 'remap', (['v_feature', '[-1, 1]', 'agent.env.observation_type.features_range[feature]'], {}), '(v_feature, [-1, 1], agent.env.observation_type.features_range[feature])\n', (4546, 4618), False, 'from rl_agents.utils import remap, constrain\n'), ((5565, 5586), 'matplotlib.pyplot.subplot', 'plt.subplot', (['gs[0, :]'], {}), '(gs[0, :])\n', (5576, 5586), True, 'from matplotlib import pyplot as plt, gridspec as gridspec\n'), ((5617, 5638), 'matplotlib.pyplot.subplot', 'plt.subplot', (['gs[1, 0]'], {}), '(gs[1, 0])\n', (5628, 5638), True, 'from matplotlib import pyplot as plt, gridspec as gridspec\n'), ((5669, 5690), 'matplotlib.pyplot.subplot', 'plt.subplot', (['gs[1, 1]'], {}), '(gs[1, 1])\n', (5680, 5690), True, 'from matplotlib import pyplot as plt, gridspec as gridspec\n'), ((6207, 6219), 'numpy.shape', 'np.shape', (['xx'], {}), '(xx)\n', (6215, 6219), True, 'import numpy as np\n'), ((6242, 6254), 'numpy.shape', 'np.shape', (['xx'], {}), '(xx)\n', (6250, 6254), True, 'import numpy as np\n'), ((2770, 2823), 'rl_agents.utils.remap', 'remap', (['attention[head]', '(0, 0.5)', '(0.7, 1)'], {'clip': '(True)'}), '(attention[head], (0, 0.5), (0.7, 1), clip=True)\n', (2775, 2823), False, 'from rl_agents.utils import remap, constrain\n'), ((2853, 2891), 'seaborn.color_palette', 'sns.color_palette', (['"""dark"""'], {'desat': 'desat'}), "('dark', desat=desat)\n", (2870, 2891), True, 'import seaborn as sns\n'), ((3010, 3065), 'rl_agents.utils.remap', 'remap', (['attention[head]', '(0, 0.5)', '(100, 200)'], {'clip': '(True)'}), '(attention[head], (0, 0.5), (100, 200), clip=True)\n', (3015, 3065), False, 'from rl_agents.utils import remap, constrain\n'), ((4941, 4980), 'numpy.linalg.norm', 'np.linalg.norm', (['(v.position - v_position)'], {}), '(v.position - v_position)\n', (4955, 4980), True, 'import numpy as np\n')] |
import datetime as DT
import numpy as NP
import matplotlib.pyplot as PLT
import matplotlib.colors as PLTC
import scipy.constants as FCNST
from astropy.io import fits
from astropy.io import ascii
from astropy.table import Table
import progressbar as PGB
import antenna_array as AA
import geometry as GEOM
import my_DSP_modules as DSP
import sim_observe as SIM
import ipdb as PDB
itr = 4
# Antenna initialization
lat = -26.701 # Latitude of MWA in degrees
f0 = 150e6 # Center frequency
antenna_file = '/data3/t_nithyanandan/project_MWA/MWA_128T_antenna_locations_MNRAS_2012_Beardsley_et_al.txt'
ant_info = NP.loadtxt(antenna_file, skiprows=6, comments='#', usecols=(0,1,2,3))
ant_info[:,1] -= NP.mean(ant_info[:,1])
ant_info[:,2] -= NP.mean(ant_info[:,2])
ant_info[:,3] -= NP.mean(ant_info[:,3])
max_antenna_radius = 75.0 # in meters
# max_antenna_radius = 75.0 # in meters
# core_ind = NP.logical_and((NP.abs(ant_info[:,1]) < max_antenna_radius), (NP.abs(ant_info[:,2]) < max_antenna_radius))
core_ind = NP.logical_and((NP.abs(ant_info[:,1]) < max_antenna_radius), (NP.abs(ant_info[:,2]) < max_antenna_radius))
ant_info = ant_info[core_ind,:]
# ant_info = ant_info[:30,:]
n_antennas = ant_info.shape[0]
nx = 4 # dipoles along x
ny = 4 # dipoles along y
dx = 1.1 # dipole spacing along x
dy = 1.1 # dipole spacing along y
nchan = 16
f_center = f0
channel_width = 40e3
bandwidth = nchan * channel_width
dt = 1/bandwidth
# ant_locs = NP.asarray([[0.0, 0.0, 0.0],[100.0, 0.0, 0.0],[50.0, 400.0, 0.0]])
# src_flux = [1.0]
# skypos = NP.asarray([0.0, 0.0]).reshape(-1,2)
# src_flux = [1.0, 1.0]
# skypos = NP.asarray([[0.0, 0.0], [0.1, 0.0]])
src_seed = 50
NP.random.seed(src_seed)
# n_src = NP.random.poisson(lam=5)
n_src = 10
lmrad = NP.random.uniform(low=0.0, high=0.5, size=n_src).reshape(-1,1)
lmang = NP.random.uniform(low=0.0, high=2*NP.pi, size=n_src).reshape(-1,1)
skypos = NP.hstack((lmrad * NP.cos(lmang), lmrad * NP.sin(lmang)))
src_flux = NP.ones(n_src)
# n_src = 4
# src_flux = NP.ones(n_src)
# skypos = 0.25*NP.hstack((NP.cos(2.0*NP.pi*NP.arange(n_src).reshape(-1,1)/n_src),
# NP.sin(2.0*NP.pi*NP.arange(n_src).reshape(-1,1)/n_src)))
# src_flux = [1.0, 1.0, 1.0, 1.0]
# skypos = NP.asarray([[0.25, 0.0], [0.0, -0.25], [-0.25, 0.0], [0.0, 0.25]])
# skypos = NP.asarray([[0.0, 0.0], [0.2, 0.0], [0.0, 0.4], [0.0, -0.5]])
nvect = NP.sqrt(1.0-NP.sum(skypos**2, axis=1)).reshape(-1,1)
skypos = NP.hstack((skypos,nvect))
ants = []
aar = AA.AntennaArray()
for i in xrange(n_antennas):
ant = AA.Antenna('{0:0d}'.format(int(ant_info[i,0])), lat, ant_info[i,1:], f0, nsamples=nchan/2)
ant.f = ant.f0 + DSP.spectax(nchan, dt, shift=True)
ants += [ant]
aar = aar + ant
iar = AA.InterferometerArray(antenna_array=aar)
iar.grid()
antpos_info = aar.antenna_positions(sort=True)
Ef_runs = None
count = 0
for i in xrange(itr):
E_timeseries_dict = SIM.stochastic_E_timeseries(f_center, nchan/2, 2*channel_width,
flux_ref=src_flux, skypos=skypos,
antpos=antpos_info['positions'],
tshift=False)
timestamp = str(DT.datetime.now())
antenna_level_update_info = {}
antenna_level_update_info['antenna_array'] = {}
antenna_level_update_info['antenna_array']['timestamp'] = timestamp
antenna_level_update_info['antennas'] = []
for label in iar.antenna_array.antennas:
adict = {}
adict['label'] = label
adict['action'] = 'modify'
adict['timestamp'] = timestamp
ind = antpos_info['labels'].index(label)
adict['t'] = E_timeseries_dict['t']
adict['Et'] = {}
adict['flags'] = {}
for pol in ['P1', 'P2']:
adict['flags'][pol] = False
adict['Et'][pol] = E_timeseries_dict['Et'][:,ind]
# adict['Et_P1'] = E_timeseries_dict['Et'][:,ind]
# adict['Et_P2'] = E_timeseries_dict['Et'][:,ind]
# adict['flag_P1'] = False
# adict['flag_P2'] = False
antenna_level_update_info['antennas'] += [adict]
interferometer_level_update_info = {}
interferometer_level_update_info['interferometers'] = []
for label in iar.interferometers:
idict = {}
idict['label'] = label
idict['action'] = 'modify'
idict['gridfunc_freq'] = 'scale'
idict['gridmethod'] = 'NN'
idict['distNN'] = 0.5 * FCNST.c / f0
idict['tol'] = 1.0e-6
idict['maxmatch'] = 1
idict['wtsinfo'] = {}
for pol in ['P11', 'P12', 'P21', 'P22']:
# idict['wtsinfo'][pol] = [{'orientation':0.0, 'lookup':'/data3/t_nithyanandan/project_MOFF/simulated/MWA/data/lookup/V_illumination_lookup_zenith.txt'}]
idict['wtsinfo'][pol] = [{'orientation':0.0, 'lookup':'/data3/t_nithyanandan/project_MOFF/simulated/LWA/data/lookup/E_illumination_isotropic_radiators_lookup_zenith.txt'}]
interferometer_level_update_info['interferometers'] += [idict]
iar.update(antenna_level_updates=antenna_level_update_info, interferometer_level_updates=interferometer_level_update_info, do_correlate='FX', parallel=True, verbose=True)
iar.grid_convolve(pol='P11', method='NN', distNN=0.5*FCNST.c/f0, tol=1.0e-6, maxmatch=1, identical_interferometers=True, gridfunc_freq='scale', mapping='weighted', wts_change=False, parallel=True, pp_method='queue')
imgobj = AA.NewImage(interferometer_array=iar, pol='P11')
imgobj.imagr(weighting='natural', pol='P11')
if i == 0:
avg_img = imgobj.img['P11']
else:
avg_img += imgobj.img['P11']
avg_img /= itr
fig = PLT.figure()
ax = fig.add_subplot(111)
imgplot = ax.imshow(NP.mean(avg_img, axis=2), aspect='equal', origin='lower', extent=(imgobj.gridl.min(), imgobj.gridl.max(), imgobj.gridm.min(), imgobj.gridm.max()))
posplot, = ax.plot(skypos[:,0], skypos[:,1], 'o', mfc='none', mec='black', mew=1, ms=8)
ax.set_xlim(imgobj.gridl.min(), imgobj.gridl.max())
ax.set_ylim(imgobj.gridm.min(), imgobj.gridm.max())
PLT.savefig('/data3/t_nithyanandan/project_MOFF/simulated/MWA/figures/FX_image_random_source_positions_{0:0d}_iterations.png'.format(itr), bbox_inches=0)
fig = PLT.figure()
ax = fig.add_subplot(111)
imgplot = ax.imshow(NP.mean(imgobj.beam['P11'], axis=2), aspect='equal', origin='lower', extent=(imgobj.gridl.min(), imgobj.gridl.max(), imgobj.gridm.min(), imgobj.gridm.max()))
ax.set_xlim(imgobj.gridl.min(), imgobj.gridl.max())
ax.set_ylim(imgobj.gridm.min(), imgobj.gridm.max())
PLT.savefig('/data3/t_nithyanandan/project_MOFF/simulated/MWA/figures/FX_psf_square_illumination.png'.format(itr), bbox_inches=0)
| [
"antenna_array.InterferometerArray",
"numpy.random.uniform",
"numpy.random.seed",
"numpy.abs",
"numpy.sum",
"sim_observe.stochastic_E_timeseries",
"numpy.ones",
"datetime.datetime.now",
"numpy.hstack",
"matplotlib.pyplot.figure",
"numpy.mean",
"antenna_array.NewImage",
"numpy.loadtxt",
"nu... | [((608, 680), 'numpy.loadtxt', 'NP.loadtxt', (['antenna_file'], {'skiprows': '(6)', 'comments': '"""#"""', 'usecols': '(0, 1, 2, 3)'}), "(antenna_file, skiprows=6, comments='#', usecols=(0, 1, 2, 3))\n", (618, 680), True, 'import numpy as NP\n'), ((696, 719), 'numpy.mean', 'NP.mean', (['ant_info[:, 1]'], {}), '(ant_info[:, 1])\n', (703, 719), True, 'import numpy as NP\n'), ((736, 759), 'numpy.mean', 'NP.mean', (['ant_info[:, 2]'], {}), '(ant_info[:, 2])\n', (743, 759), True, 'import numpy as NP\n'), ((776, 799), 'numpy.mean', 'NP.mean', (['ant_info[:, 3]'], {}), '(ant_info[:, 3])\n', (783, 799), True, 'import numpy as NP\n'), ((1664, 1688), 'numpy.random.seed', 'NP.random.seed', (['src_seed'], {}), '(src_seed)\n', (1678, 1688), True, 'import numpy as NP\n'), ((1959, 1973), 'numpy.ones', 'NP.ones', (['n_src'], {}), '(n_src)\n', (1966, 1973), True, 'import numpy as NP\n'), ((2439, 2465), 'numpy.hstack', 'NP.hstack', (['(skypos, nvect)'], {}), '((skypos, nvect))\n', (2448, 2465), True, 'import numpy as NP\n'), ((2482, 2499), 'antenna_array.AntennaArray', 'AA.AntennaArray', ([], {}), '()\n', (2497, 2499), True, 'import antenna_array as AA\n'), ((2731, 2772), 'antenna_array.InterferometerArray', 'AA.InterferometerArray', ([], {'antenna_array': 'aar'}), '(antenna_array=aar)\n', (2753, 2772), True, 'import antenna_array as AA\n'), ((5702, 5714), 'matplotlib.pyplot.figure', 'PLT.figure', ([], {}), '()\n', (5712, 5714), True, 'import matplotlib.pyplot as PLT\n'), ((6261, 6273), 'matplotlib.pyplot.figure', 'PLT.figure', ([], {}), '()\n', (6271, 6273), True, 'import matplotlib.pyplot as PLT\n'), ((2905, 3061), 'sim_observe.stochastic_E_timeseries', 'SIM.stochastic_E_timeseries', (['f_center', '(nchan / 2)', '(2 * channel_width)'], {'flux_ref': 'src_flux', 'skypos': 'skypos', 'antpos': "antpos_info['positions']", 'tshift': '(False)'}), "(f_center, nchan / 2, 2 * channel_width,\n flux_ref=src_flux, skypos=skypos, antpos=antpos_info['positions'],\n tshift=False)\n", (2932, 3061), True, 'import sim_observe as SIM\n'), ((5482, 5530), 'antenna_array.NewImage', 'AA.NewImage', ([], {'interferometer_array': 'iar', 'pol': '"""P11"""'}), "(interferometer_array=iar, pol='P11')\n", (5493, 5530), True, 'import antenna_array as AA\n'), ((5761, 5785), 'numpy.mean', 'NP.mean', (['avg_img'], {'axis': '(2)'}), '(avg_img, axis=2)\n', (5768, 5785), True, 'import numpy as NP\n'), ((6320, 6355), 'numpy.mean', 'NP.mean', (["imgobj.beam['P11']"], {'axis': '(2)'}), "(imgobj.beam['P11'], axis=2)\n", (6327, 6355), True, 'import numpy as NP\n'), ((1026, 1048), 'numpy.abs', 'NP.abs', (['ant_info[:, 1]'], {}), '(ant_info[:, 1])\n', (1032, 1048), True, 'import numpy as NP\n'), ((1072, 1094), 'numpy.abs', 'NP.abs', (['ant_info[:, 2]'], {}), '(ant_info[:, 2])\n', (1078, 1094), True, 'import numpy as NP\n'), ((1743, 1791), 'numpy.random.uniform', 'NP.random.uniform', ([], {'low': '(0.0)', 'high': '(0.5)', 'size': 'n_src'}), '(low=0.0, high=0.5, size=n_src)\n', (1760, 1791), True, 'import numpy as NP\n'), ((1814, 1868), 'numpy.random.uniform', 'NP.random.uniform', ([], {'low': '(0.0)', 'high': '(2 * NP.pi)', 'size': 'n_src'}), '(low=0.0, high=2 * NP.pi, size=n_src)\n', (1831, 1868), True, 'import numpy as NP\n'), ((2651, 2685), 'my_DSP_modules.spectax', 'DSP.spectax', (['nchan', 'dt'], {'shift': '(True)'}), '(nchan, dt, shift=True)\n', (2662, 2685), True, 'import my_DSP_modules as DSP\n'), ((3227, 3244), 'datetime.datetime.now', 'DT.datetime.now', ([], {}), '()\n', (3242, 3244), True, 'import datetime as DT\n'), ((1909, 1922), 'numpy.cos', 'NP.cos', (['lmang'], {}), '(lmang)\n', (1915, 1922), True, 'import numpy as NP\n'), ((1932, 1945), 'numpy.sin', 'NP.sin', (['lmang'], {}), '(lmang)\n', (1938, 1945), True, 'import numpy as NP\n'), ((2389, 2416), 'numpy.sum', 'NP.sum', (['(skypos ** 2)'], {'axis': '(1)'}), '(skypos ** 2, axis=1)\n', (2395, 2416), True, 'import numpy as NP\n')] |
from torch.autograd import Variable
from net_gan_mnist import *
import torch
import torch.nn as nn
import numpy as np
from init import *
class MNISTGanTrainer(object):
def __init__(self, batch_size=64, latent_dims=100):
super(MNISTGanTrainer, self).__init__()
self.dis = Dis28x28()
self.gen = Gen28x28(latent_dims)
self.dis_opt = torch.optim.Adam(self.dis.parameters(), lr=0.0002, betas=(0.5, 0.999), weight_decay=0.0005)
self.gen_opt = torch.optim.Adam(self.gen.parameters(), lr=0.0002, betas=(0.5, 0.999), weight_decay=0.0005)
self.true_labels = Variable(torch.LongTensor(np.ones(batch_size, dtype=np.int)))
self.fake_labels = Variable(torch.LongTensor(np.zeros(batch_size, dtype=np.int)))
self.dis.apply(xavier_weights_init)
self.gen.apply(xavier_weights_init)
def cuda(self):
self.dis.cuda()
self.gen.cuda()
self.true_labels = self.true_labels.cuda()
self.fake_labels = self.fake_labels.cuda()
def dis_update(self, images, noise):
self.dis.zero_grad()
true_outputs = self.dis(images)
true_loss = nn.functional.cross_entropy(true_outputs, self.true_labels)
_, true_predicts = torch.max(true_outputs.data, 1)
true_acc = (true_predicts == 1).sum()/(1.0*true_predicts.size(0))
fake_images = self.gen(noise)
fake_outputs = self.dis(fake_images)
fake_loss = nn.functional.cross_entropy(fake_outputs, self.fake_labels)
_, fake_predicts = torch.max(fake_outputs.data, 1)
fake_acc = (fake_predicts == 0).sum() / (1.0 * fake_predicts.size(0))
d_loss = true_loss + fake_loss
d_loss.backward()
self.dis_opt.step()
return 0.5 * (true_acc + fake_acc)
def gen_update(self, noise):
self.gen.zero_grad()
fake_images = self.gen(noise)
fake_outputs = self.dis(fake_images)
fake_loss = nn.functional.cross_entropy(fake_outputs, self.true_labels)
fake_loss.backward()
self.gen_opt.step()
return fake_images
| [
"numpy.zeros",
"numpy.ones",
"torch.max",
"torch.nn.functional.cross_entropy"
] | [((1144, 1203), 'torch.nn.functional.cross_entropy', 'nn.functional.cross_entropy', (['true_outputs', 'self.true_labels'], {}), '(true_outputs, self.true_labels)\n', (1171, 1203), True, 'import torch.nn as nn\n'), ((1231, 1262), 'torch.max', 'torch.max', (['true_outputs.data', '(1)'], {}), '(true_outputs.data, 1)\n', (1240, 1262), False, 'import torch\n'), ((1440, 1499), 'torch.nn.functional.cross_entropy', 'nn.functional.cross_entropy', (['fake_outputs', 'self.fake_labels'], {}), '(fake_outputs, self.fake_labels)\n', (1467, 1499), True, 'import torch.nn as nn\n'), ((1527, 1558), 'torch.max', 'torch.max', (['fake_outputs.data', '(1)'], {}), '(fake_outputs.data, 1)\n', (1536, 1558), False, 'import torch\n'), ((1939, 1998), 'torch.nn.functional.cross_entropy', 'nn.functional.cross_entropy', (['fake_outputs', 'self.true_labels'], {}), '(fake_outputs, self.true_labels)\n', (1966, 1998), True, 'import torch.nn as nn\n'), ((628, 661), 'numpy.ones', 'np.ones', (['batch_size'], {'dtype': 'np.int'}), '(batch_size, dtype=np.int)\n', (635, 661), True, 'import numpy as np\n'), ((717, 751), 'numpy.zeros', 'np.zeros', (['batch_size'], {'dtype': 'np.int'}), '(batch_size, dtype=np.int)\n', (725, 751), True, 'import numpy as np\n')] |
from controller import *
import random
from keras.optimizers import Adam
from keras.models import Sequential
from keras.layers.core import Dense, Dropout
from keras.utils.np_utils import to_categorical
import numpy as np
class CellItemType(enum.Enum):
WALL = -1
EMPTY = 0
BODY = 1
HEAD = 2
FRUIT = 4
def __int__(self):
return self.value
class AIController(Controller):
player = None
game = None
neural_network = None
learning_rate = 0.0005
first_layer = 150
second_layer = 150
third_layer = 150
train_flag = True
def init(self, player, game):
self.player = player
self.game = game
self.reward = 0
self.score = 0
self.last_state = self.get_snake_vision()
self.last_decision = None
if not self.neural_network:
self.create_network()
def get_input_size(self):
return len(self.last_state)
def scan(self, board, start_pos, itemType, direction):
i = 1
while True:
x = start_pos.x + i * self.player.step * direction[0]
y = start_pos.y + i * self.player.step * direction[1]
if x < self.get_min_x() or x >= self.get_max_x() or y < self.get_min_y() or y >= self.get_max_y():
if itemType == CellItemType.WALL:
return 1 / start_pos.distance(Position(x, y))
break
curr_idx = self.coordinates_to_board_index(x, y)
if board[curr_idx] == int(itemType):
return 1 / start_pos.distance(Position(x, y))
i += 1
#print(" i = ", i)
return 1
def get_snake_vision(self):
board = self.board_state_to_list()
directions = [(0, -1), (1, -1), (1, 0), (1, 1), (0, 1), (-1, 1), (-1, 0), (-1, -1)] # up, up-right, right...
directions_for_move = None
if self.player.last_move == Move.UP:
directions_for_move = directions[-3:] + directions[:4]
elif self.player.last_move == Move.RIGHT:
directions_for_move = directions[-1:] + directions[:6]
elif self.player.last_move == Move.DOWN:
directions_for_move = directions[1:]
elif self.player.last_move == Move.LEFT:
directions_for_move = directions[3:] + directions[:2]
vision = []
for cell in (CellItemType.WALL, CellItemType.FRUIT, CellItemType.BODY):
for direction in directions_for_move:
vision.append(self.scan(board, self.player.positions[0], cell, direction))
return np.asarray(vision)
def make_move(self):
self.reward -= 0
self.last_state = self.get_snake_vision()
prediction = self.neural_network.predict(self.last_state.reshape((1, self.get_input_size())))
#print(" ---------------------------------------- ")
#print("Snake vision = ", self.last_state)
#print("Prediction = ", prediction)
#print("------------------------------------------\n\n")
# Predictions will be a [[0.3333328 0.3332601 0.33340713]] type np array
self.last_decision = to_categorical(np.argmax(prediction[0]), num_classes=3)
print("Current Reward = ", self.reward)
if self.last_decision[0]: # left
self.player.turn_left()
elif self.last_decision[1]: # forward
pass
elif self.last_decision[2]: # right
self.player.turn_right()
def set_reward(self):
#self.reward = 0
if self.player.get_score() > self.score:
self.score = self.player.get_score()
self.reward += 500
elif self.game.is_end():
self.reward += -500
# else:
# self.reward = - self.player.positions[0].distance(self.game.fruit.position) / self.player.step
def update_state(self):
if self.train_flag:
self.set_reward()
#print(self.reward)
target_f = self.neural_network.predict(self.last_state.reshape((1, self.get_input_size())))
target_f[0][np.argmax(self.last_decision)] = self.reward
self.neural_network.fit(self.last_state.reshape((1, self.get_input_size())), target_f, epochs=1, verbose=0)
def get_board_width(self):
return (self.game.board_rect.right - self.game.board_rect.left) / self.player.step
def get_board_height(self):
return (self.game.board_rect.bottom - self.game.board_rect.top) / self.player.step
def get_min_x(self):
return self.game.board_rect.left
def get_min_y(self):
return self.game.board_rect.top
def get_max_x(self):
return self.game.board_rect.right
def get_max_y(self):
return self.game.board_rect.bottom
def coordinates_to_board_index(self, x, y):
tmp_x = (x - self.get_min_x()) / self.player.step
tmp_y = (y - self.get_min_y()) / self.player.step
width = self.get_board_width()
return int(tmp_y * width + tmp_x)
def board_state_to_list(self):
board = []
for row in range(self.game.board_rect.top, self.game.board_rect.bottom, self.player.step):
for col in range(self.game.board_rect.left, self.game.board_rect.right, self.player.step):
board.append(CellItemType.EMPTY.value)
board[self.coordinates_to_board_index(self.game.fruit.position.x, self.game.fruit.position.y)] = CellItemType.FRUIT.value
for pos in self.player.positions:
board[self.coordinates_to_board_index(pos.x, pos.y)] = CellItemType.BODY.value
snake_head = self.player.positions[0]
board[self.coordinates_to_board_index(snake_head.x, snake_head.y)] = CellItemType.HEAD.value
return np.asarray(board)
def create_network(self):
self.neural_network = Sequential()
self.neural_network.add(Dense(self.first_layer, activation='relu', input_dim=self.get_input_size()))
self.neural_network.add(Dense(self.second_layer, activation='relu'))
self.neural_network.add(Dense(self.third_layer, activation='relu'))
self.neural_network.add(Dense(3, activation='softmax'))
opt = Adam(self.learning_rate)
self.neural_network.compile(loss='mse', optimizer=opt)
| [
"keras.layers.core.Dense",
"numpy.argmax",
"numpy.asarray",
"keras.optimizers.Adam",
"keras.models.Sequential"
] | [((2644, 2662), 'numpy.asarray', 'np.asarray', (['vision'], {}), '(vision)\n', (2654, 2662), True, 'import numpy as np\n'), ((5964, 5981), 'numpy.asarray', 'np.asarray', (['board'], {}), '(board)\n', (5974, 5981), True, 'import numpy as np\n'), ((6052, 6064), 'keras.models.Sequential', 'Sequential', ([], {}), '()\n', (6062, 6064), False, 'from keras.models import Sequential\n'), ((6414, 6438), 'keras.optimizers.Adam', 'Adam', (['self.learning_rate'], {}), '(self.learning_rate)\n', (6418, 6438), False, 'from keras.optimizers import Adam\n'), ((3228, 3252), 'numpy.argmax', 'np.argmax', (['prediction[0]'], {}), '(prediction[0])\n', (3237, 3252), True, 'import numpy as np\n'), ((6206, 6249), 'keras.layers.core.Dense', 'Dense', (['self.second_layer'], {'activation': '"""relu"""'}), "(self.second_layer, activation='relu')\n", (6211, 6249), False, 'from keras.layers.core import Dense, Dropout\n'), ((6283, 6325), 'keras.layers.core.Dense', 'Dense', (['self.third_layer'], {'activation': '"""relu"""'}), "(self.third_layer, activation='relu')\n", (6288, 6325), False, 'from keras.layers.core import Dense, Dropout\n'), ((6359, 6389), 'keras.layers.core.Dense', 'Dense', (['(3)'], {'activation': '"""softmax"""'}), "(3, activation='softmax')\n", (6364, 6389), False, 'from keras.layers.core import Dense, Dropout\n'), ((4224, 4253), 'numpy.argmax', 'np.argmax', (['self.last_decision'], {}), '(self.last_decision)\n', (4233, 4253), True, 'import numpy as np\n')] |
#!/usr/bin/env python
# -*- coding: utf-8 -*-
#%%
from numpy import *
import numpy as np
import torch
import aTEAM.nn.functional as aF
#%%
a = np.arange(10)
a = a[:,None]+a[None,:]
b = torch.from_numpy(a)
print(np.roll(a, shift=[0,1],axis=[1,0])-aF.roll(b, shift=[0,1],axis=[1,0]).data.numpy())
print(np.roll(a, shift=[2,1])-aF.roll(b, shift=[2,1]).data.numpy())
print(np.roll(a, shift=[2,-1], axis=[0,1])-aF.roll(b, shift=[2,-1], axis=[0,1]).data.numpy())
#%%
| [
"aTEAM.nn.functional.roll",
"numpy.roll",
"numpy.arange",
"torch.from_numpy"
] | [((143, 156), 'numpy.arange', 'np.arange', (['(10)'], {}), '(10)\n', (152, 156), True, 'import numpy as np\n'), ((185, 204), 'torch.from_numpy', 'torch.from_numpy', (['a'], {}), '(a)\n', (201, 204), False, 'import torch\n'), ((212, 249), 'numpy.roll', 'np.roll', (['a'], {'shift': '[0, 1]', 'axis': '[1, 0]'}), '(a, shift=[0, 1], axis=[1, 0])\n', (219, 249), True, 'import numpy as np\n'), ((302, 326), 'numpy.roll', 'np.roll', (['a'], {'shift': '[2, 1]'}), '(a, shift=[2, 1])\n', (309, 326), True, 'import numpy as np\n'), ((370, 408), 'numpy.roll', 'np.roll', (['a'], {'shift': '[2, -1]', 'axis': '[0, 1]'}), '(a, shift=[2, -1], axis=[0, 1])\n', (377, 408), True, 'import numpy as np\n'), ((247, 284), 'aTEAM.nn.functional.roll', 'aF.roll', (['b'], {'shift': '[0, 1]', 'axis': '[1, 0]'}), '(b, shift=[0, 1], axis=[1, 0])\n', (254, 284), True, 'import aTEAM.nn.functional as aF\n'), ((326, 350), 'aTEAM.nn.functional.roll', 'aF.roll', (['b'], {'shift': '[2, 1]'}), '(b, shift=[2, 1])\n', (333, 350), True, 'import aTEAM.nn.functional as aF\n'), ((407, 445), 'aTEAM.nn.functional.roll', 'aF.roll', (['b'], {'shift': '[2, -1]', 'axis': '[0, 1]'}), '(b, shift=[2, -1], axis=[0, 1])\n', (414, 445), True, 'import aTEAM.nn.functional as aF\n')] |
"""Most test exploit the special case where simulate_moments just returns parameters."""
import itertools
import warnings
import numpy as np
import pandas as pd
import pytest
from estimagic.estimation.estimate_msm import estimate_msm
from estimagic.shared.check_option_dicts import check_numdiff_options
from estimagic.shared.check_option_dicts import check_optimization_options
from numpy.testing import assert_array_almost_equal as aaae
def _sim_pd(params):
return params["value"]
def _sim_np(params):
return params["value"].to_numpy()
def _sim_dict_pd(params):
return {"simulated_moments": params["value"], "other": "bla"}
def _sim_dict_np(params):
return {"simulated_moments": params["value"].to_numpy(), "other": "bla"}
cov_np = np.diag([1, 2, 3.0])
cov_pd = pd.DataFrame(cov_np)
funcs = [_sim_pd, _sim_np, _sim_dict_pd, _sim_dict_np]
covs = [cov_np, cov_pd]
test_cases = list(itertools.product(funcs, covs))
@pytest.mark.parametrize("simulate_moments, moments_cov", test_cases)
def test_estimate_msm(simulate_moments, moments_cov):
start_params = pd.DataFrame()
start_params["value"] = [3, 2, 1]
expected_params = pd.DataFrame()
expected_params["value"] = np.zeros(3)
# abuse simulate_moments to get empirical moments in correct format
empirical_moments = simulate_moments(expected_params)
if isinstance(empirical_moments, dict):
empirical_moments = empirical_moments["simulated_moments"]
optimize_options = {"algorithm": "scipy_lbfgsb"}
# catching warnings is necessary because the very special case with diagonal
# weighting and diagonal jacobian leads to singular matrices while calculating
# sensitivity to removal of moments.
with warnings.catch_warnings():
warnings.filterwarnings("ignore", message="Standard matrix inversion failed")
calculated = estimate_msm(
simulate_moments=simulate_moments,
empirical_moments=empirical_moments,
moments_cov=moments_cov,
params=start_params,
optimize_options=optimize_options,
)
calculated_params = calculated["optimize_res"]["solution_params"][["value"]]
# check that minimization works
aaae(calculated_params["value"].to_numpy(), expected_params["value"].to_numpy())
# check that cov works
calculated_cov = calculated["cov"]
if isinstance(calculated_cov, pd.DataFrame):
calculated_cov = calculated_cov.to_numpy()
# this works only in the very special case with diagonal moments cov and
# jac = identity matrix
expected_cov = np.diag([1, 2, 3])
aaae(calculated_cov, expected_cov)
def test_check_and_process_numdiff_options_with_invalid_entries():
with pytest.raises(ValueError):
check_numdiff_options({"func": lambda x: x}, "estimate_msm")
def test_check_and_process_optimize_options_with_invalid_entries():
with pytest.raises(ValueError):
check_optimization_options({"criterion": lambda x: x}, "estimate_msm")
| [
"pandas.DataFrame",
"estimagic.shared.check_option_dicts.check_numdiff_options",
"warnings.filterwarnings",
"estimagic.estimation.estimate_msm.estimate_msm",
"numpy.zeros",
"pytest.raises",
"estimagic.shared.check_option_dicts.check_optimization_options",
"warnings.catch_warnings",
"pytest.mark.para... | [((761, 781), 'numpy.diag', 'np.diag', (['[1, 2, 3.0]'], {}), '([1, 2, 3.0])\n', (768, 781), True, 'import numpy as np\n'), ((791, 811), 'pandas.DataFrame', 'pd.DataFrame', (['cov_np'], {}), '(cov_np)\n', (803, 811), True, 'import pandas as pd\n'), ((947, 1015), 'pytest.mark.parametrize', 'pytest.mark.parametrize', (['"""simulate_moments, moments_cov"""', 'test_cases'], {}), "('simulate_moments, moments_cov', test_cases)\n", (970, 1015), False, 'import pytest\n'), ((912, 942), 'itertools.product', 'itertools.product', (['funcs', 'covs'], {}), '(funcs, covs)\n', (929, 942), False, 'import itertools\n'), ((1089, 1103), 'pandas.DataFrame', 'pd.DataFrame', ([], {}), '()\n', (1101, 1103), True, 'import pandas as pd\n'), ((1165, 1179), 'pandas.DataFrame', 'pd.DataFrame', ([], {}), '()\n', (1177, 1179), True, 'import pandas as pd\n'), ((1211, 1222), 'numpy.zeros', 'np.zeros', (['(3)'], {}), '(3)\n', (1219, 1222), True, 'import numpy as np\n'), ((2600, 2618), 'numpy.diag', 'np.diag', (['[1, 2, 3]'], {}), '([1, 2, 3])\n', (2607, 2618), True, 'import numpy as np\n'), ((2623, 2657), 'numpy.testing.assert_array_almost_equal', 'aaae', (['calculated_cov', 'expected_cov'], {}), '(calculated_cov, expected_cov)\n', (2627, 2657), True, 'from numpy.testing import assert_array_almost_equal as aaae\n'), ((1734, 1759), 'warnings.catch_warnings', 'warnings.catch_warnings', ([], {}), '()\n', (1757, 1759), False, 'import warnings\n'), ((1769, 1846), 'warnings.filterwarnings', 'warnings.filterwarnings', (['"""ignore"""'], {'message': '"""Standard matrix inversion failed"""'}), "('ignore', message='Standard matrix inversion failed')\n", (1792, 1846), False, 'import warnings\n'), ((1868, 2042), 'estimagic.estimation.estimate_msm.estimate_msm', 'estimate_msm', ([], {'simulate_moments': 'simulate_moments', 'empirical_moments': 'empirical_moments', 'moments_cov': 'moments_cov', 'params': 'start_params', 'optimize_options': 'optimize_options'}), '(simulate_moments=simulate_moments, empirical_moments=\n empirical_moments, moments_cov=moments_cov, params=start_params,\n optimize_options=optimize_options)\n', (1880, 2042), False, 'from estimagic.estimation.estimate_msm import estimate_msm\n'), ((2736, 2761), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (2749, 2761), False, 'import pytest\n'), ((2771, 2831), 'estimagic.shared.check_option_dicts.check_numdiff_options', 'check_numdiff_options', (["{'func': lambda x: x}", '"""estimate_msm"""'], {}), "({'func': lambda x: x}, 'estimate_msm')\n", (2792, 2831), False, 'from estimagic.shared.check_option_dicts import check_numdiff_options\n'), ((2911, 2936), 'pytest.raises', 'pytest.raises', (['ValueError'], {}), '(ValueError)\n', (2924, 2936), False, 'import pytest\n'), ((2946, 3016), 'estimagic.shared.check_option_dicts.check_optimization_options', 'check_optimization_options', (["{'criterion': lambda x: x}", '"""estimate_msm"""'], {}), "({'criterion': lambda x: x}, 'estimate_msm')\n", (2972, 3016), False, 'from estimagic.shared.check_option_dicts import check_optimization_options\n')] |
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Sun May 9 11:59:51 2021
@author: Daniel
"""
import numpy as np
from get_sudoku import get_sudoku_
from copy import deepcopy
def look_row(row):
#creates local set for current square
local_set_row = {i for i in range(1,10)}
#iterates through elements of row of current square
for element in row:
#eleminates number from local set if present in row
#this will reduce the possibilities of numbers to be put on current square
if element in local_set_row:
local_set_row.remove(element)
return local_set_row
def look_column(column):
#creates local set for current square
local_set_column = {i for i in range(1,10)}
#iterates through elements of column of current square
for element in column:
#eliminates number from local set if present in row
#this will reduce the possibilities of numbers to be put on current square
if element in local_set_column:
local_set_column.remove(element)
return local_set_column
def get_square(sudoku, row, column):
#finds square where current cell is located
if row in range(0,3):
if column in range(0,3):
search_square = sudoku[0:3,0:3]
if column in range(3,6):
search_square = sudoku[0:3,3:6]
if column in range(6,9):
search_square = sudoku[0:3,6:9]
if row in range(3,6):
if column in range(0,3):
search_square = sudoku[3:6,0:3]
if column in range(3,6):
search_square = sudoku[3:6,3:6]
if column in range(6,9):
search_square = sudoku[3:6,6:9]
if row in range(6,9):
if column in range(0,3):
search_square = sudoku[6:9,0:3]
if column in range(3,6):
search_square = sudoku[6:9,3:6]
if column in range(6,9):
search_square = sudoku[6:9,6:9]
return search_square
def look_square(sudoku, row, column):
#creates local set for current square
local_set_square = {i for i in range(1,10)}
#finds square where current cell is located
search_square = get_square(sudoku, row, column)
#iterates through elements of column of current square
for row in search_square:
for column in row:
#eleminates number from local set if present in row
#this will reduce the possibilities of numbers to be put on current square
if column in local_set_square:
local_set_square.remove(column)
return local_set_square
def intercept_locals(local_set_row, local_set_column, local_set_square):
#intercepts sets of possible numbers to put on current square according to
#evaluation of row, column and big square
local_set = local_set_row.intersection(local_set_column)
local_set = local_set.intersection(local_set_square)
return local_set
def get_candidates(sudoku):
candidates = sudoku.tolist()
#starts searching by row
for i in range(0,9):
#iterates through column of said row
for j in range(0,9):
#only interested in squares that don't already have a number assigned
if sudoku[i,j] == 0:
#look row
set_row = look_row(sudoku[i])
#look column
set_column = look_column(sudoku[:,j])
#look big square
set_square = look_square(sudoku, i, j)
#intercept all
local_set = intercept_locals(set_row, set_column, set_square)
#
candidates[i][j] = list(local_set)
else:
candidates[i][j] = [sudoku[i,j]]
return candidates
def filter_candidates(sudoku):
test_sudoku = sudoku.copy()
candidates = get_candidates(sudoku)
filtered_candidates = deepcopy(candidates)
for i in range(9):
for j in range(9):
# Check for empty cells
if sudoku[i][j] == 0:
for candidate in candidates[i][j]:
# Use test candidate
test_sudoku[i][j] = candidate
# Remove candidate if it produces an invalid grid
if not check_solution(fill_singles(test_sudoku)):
filtered_candidates[i][j].remove(candidate)
# Revert changes
test_sudoku[i][j] = 0
return filtered_candidates
def fill_singles(sudoku, candidates=None):
sudoku = sudoku.copy()
if not candidates:
candidates = get_candidates(sudoku)
singles = True
while singles:
singles = False
for i in range(0,9):
for j in range(0,9):
if type(candidates[i][j]) == list:
if len(candidates[i][j]) == 1 and sudoku[i, j] == 0:
sudoku[i, j] = candidates[i][j][0]
candidates = merge(get_candidates(sudoku), candidates)
singles = True
return sudoku
def merge(candidates_1, candidates_2):
candidates_min = []
for i in range(9):
row = []
for j in range(9):
if len(candidates_1[i][j]) < len(candidates_2[i][j]):
row.append(candidates_1[i][j][:])
else:
row.append(candidates_2[i][j][:])
candidates_min.append(row)
return candidates_min
def make_guess(sudoku, candidates=None):
print("Making guesses, still on it :(")
min_len = 9
sudoku = sudoku.copy()
if not candidates:
candidates = get_candidates(sudoku)
for i in range(9):
for j in range(9):
#checks if current cell in memory is set to unique value or is still undecided
if type(candidates[i][j]) == list:
#finds list with lowest lentgh to maximize hypothesis of getting substitution right
if len(candidates[i][j]) < min_len:
min_len = len(candidates[i][j])
for i in range(9):
for j in range(9):
if type(candidates[i][j]) == list:
if len(candidates[i][j]) == min_len:
for guess in candidates[i][j]:
sudoku[i][j] = guess
solution = filtered_solve(sudoku)
if solution is not None:
return solution
# Discarding incorrect guess
sudoku[i][j] = 0
def check_solution(sudoku):
'''checks whether solution is valid (particularly useful after making guess)'''
candidates = get_candidates(sudoku)
for i in range(0, 9):
for j in range(0, 9):
#checks if in candidates list there are lists with no values (=cells with no candidates)
if type(candidates[i][j]) == list:
if len(candidates[i][j]) == 0:
return False
return True
def is_solution(sudoku):
if np.all(np.sum(sudoku, axis=1) == 45) and \
np.all(np.sum(sudoku, axis=0) == 45):
if sum(map(sum, sudoku[0:3, 0:3])) == 45 and sum(map(sum, sudoku[0:3, 3:6])) == 45 and sum(map(sum, sudoku[0:3, 6:9])) == 45 and \
sum(map(sum, sudoku[3:6, 0:3])) == 45 and sum(map(sum, sudoku[3:6, 3:6])) == 45 and sum(map(sum, sudoku[3:6, 6:9])) == 45 and \
sum(map(sum, sudoku[6:9, 0:3])) == 45 and sum(map(sum, sudoku[6:9, 3:6])) == 45 and sum(map(sum, sudoku[6:9, 6:9])) == 45:
return True
return False
def filtered_solve(sudoku):
candidates = filter_candidates(sudoku)
sudoku = fill_singles(sudoku, candidates)
if is_solution(sudoku):
return sudoku
if not check_solution(sudoku):
return None
return make_guess(sudoku, candidates)
def solve(sudoku):
sudoku = fill_singles(sudoku)
if is_solution(sudoku):
print(sudoku)
return sudoku
if not check_solution(sudoku):
return None
return make_guess(sudoku)
solve(get_sudoku_())
| [
"copy.deepcopy",
"numpy.sum",
"get_sudoku.get_sudoku_"
] | [((4135, 4155), 'copy.deepcopy', 'deepcopy', (['candidates'], {}), '(candidates)\n', (4143, 4155), False, 'from copy import deepcopy\n'), ((8732, 8745), 'get_sudoku.get_sudoku_', 'get_sudoku_', ([], {}), '()\n', (8743, 8745), False, 'from get_sudoku import get_sudoku_\n'), ((7630, 7652), 'numpy.sum', 'np.sum', (['sudoku'], {'axis': '(1)'}), '(sudoku, axis=1)\n', (7636, 7652), True, 'import numpy as np\n'), ((7680, 7702), 'numpy.sum', 'np.sum', (['sudoku'], {'axis': '(0)'}), '(sudoku, axis=0)\n', (7686, 7702), True, 'import numpy as np\n')] |
import math
import numpy as np
import torch
from sklearn.metrics import average_precision_score, roc_auc_score
def choose_target(model,memory_s, memory_g, src_mem):
u = model.memory_merge(memory_s[1], memory_g[1]) #[num_nodes,mem_d]
u_norm = torch.norm(u, dim=1) #[num_nodes, 1]
u_normalized = u/u_norm.view(-1, 1) #[num_nodes,mem_d]
src_mem_norm = torch.norm(src_mem, dim=1) #[batch_size, 1]
src_mem_normalized = src_mem / src_mem_norm.view(-1, 1) #[batch_size, mem_d]
cos_similarity = torch.matmul(src_mem_normalized, u_normalized.t()) #[batch_size, num_nodes]
cos_similarity, idx = torch.sort(cos_similarity, descending=True)
return cos_similarity, idx
def recall(des_node, idx, top_k):
bs = idx.shape[0]
idx = idx[:, :top_k] #[bs,top_k]
recall = np.array([a in idx[i] for i, a in enumerate(des_node)])#[bs,1]
recall = recall.sum() / recall.size
return recall
def MRR(des_node, idx):
bs = idx.shape[0]
mrr = np.array([float(np.where(idx[i].cpu() == a)[0] + 1) for i, a in enumerate(des_node)])#[bs,1]
mrr = (1 / mrr).mean()
return mrr
def eval_edge_prediction(model, negative_edge_sampler, data, n_neighbors, batch_size=200):
# Ensures the random sampler uses a seed for evaluation (i.e. we sample always the same
# negatives for validation / test set)
assert negative_edge_sampler.seed is not None
negative_edge_sampler.reset_random_state()
val_mrr, val_recall_20, val_recall_50 = [], [], []
with torch.no_grad():
model = model.eval()
# While usually the test batch size is as big as it fits in memory, here we keep it the same
# size as the training batch size, since it allows the memory to be updated more frequently,
# and later test batches to access information from interactions in previous test batches
# through the memory
TEST_BATCH_SIZE = batch_size
num_test_instance = len(data.sources)
num_test_batch = math.ceil(num_test_instance / TEST_BATCH_SIZE)
for k in range(num_test_batch):
s_idx = k * TEST_BATCH_SIZE
e_idx = min(num_test_instance, s_idx + TEST_BATCH_SIZE)
sources_batch = data.sources[s_idx:e_idx]
destinations_batch = data.destinations[s_idx:e_idx]
timestamps_batch = data.timestamps[s_idx:e_idx]
edge_idxs_batch = data.edge_idxs[s_idx: e_idx]
size = len(sources_batch)
_, negative_samples = negative_edge_sampler.sample(size)
src_mem, des_mem = model(sources_batch, destinations_batch,
negative_samples, timestamps_batch,
edge_idxs_batch, test=True)
src_cos_sim, src_idx = choose_target(model, model.memory_s.memory, model.memory_g.memory, src_mem)
des_cos_sim, des_idx = choose_target(model, model.memory_s.memory, model.memory_g.memory, des_mem)
recall_20 = (recall(destinations_batch, src_idx, 20) + recall(sources_batch, des_idx, 20)) / 2
recall_50 = (recall(destinations_batch, src_idx, 50) + recall(sources_batch, des_idx, 50)) / 2
mrr = (MRR(destinations_batch, src_idx) + MRR(sources_batch, des_idx)) / 2
true_label = np.concatenate([np.ones(size), np.zeros(size)])
val_mrr.append(mrr)
val_recall_20.append(recall_20)
val_recall_50.append(recall_50)
return np.mean(val_mrr), np.mean(val_recall_20), np.mean(val_recall_50)
| [
"math.ceil",
"torch.norm",
"numpy.zeros",
"numpy.ones",
"numpy.mean",
"torch.no_grad",
"torch.sort"
] | [((248, 268), 'torch.norm', 'torch.norm', (['u'], {'dim': '(1)'}), '(u, dim=1)\n', (258, 268), False, 'import torch\n'), ((361, 387), 'torch.norm', 'torch.norm', (['src_mem'], {'dim': '(1)'}), '(src_mem, dim=1)\n', (371, 387), False, 'import torch\n'), ((605, 648), 'torch.sort', 'torch.sort', (['cos_similarity'], {'descending': '(True)'}), '(cos_similarity, descending=True)\n', (615, 648), False, 'import torch\n'), ((1458, 1473), 'torch.no_grad', 'torch.no_grad', ([], {}), '()\n', (1471, 1473), False, 'import torch\n'), ((1909, 1955), 'math.ceil', 'math.ceil', (['(num_test_instance / TEST_BATCH_SIZE)'], {}), '(num_test_instance / TEST_BATCH_SIZE)\n', (1918, 1955), False, 'import math\n'), ((3323, 3339), 'numpy.mean', 'np.mean', (['val_mrr'], {}), '(val_mrr)\n', (3330, 3339), True, 'import numpy as np\n'), ((3341, 3363), 'numpy.mean', 'np.mean', (['val_recall_20'], {}), '(val_recall_20)\n', (3348, 3363), True, 'import numpy as np\n'), ((3365, 3387), 'numpy.mean', 'np.mean', (['val_recall_50'], {}), '(val_recall_50)\n', (3372, 3387), True, 'import numpy as np\n'), ((3178, 3191), 'numpy.ones', 'np.ones', (['size'], {}), '(size)\n', (3185, 3191), True, 'import numpy as np\n'), ((3193, 3207), 'numpy.zeros', 'np.zeros', (['size'], {}), '(size)\n', (3201, 3207), True, 'import numpy as np\n')] |
__author__ = 'Zander'
from pygame import gfxdraw
from Vector2 import Vector2
from Vector4 import Vector4
from Vector3 import Vector3
from Matrix4 import Matrix4
import math, pygame
import numpy as np
class Renderer:
def __init__(self, screen, width, height, scale=1):
self.width = width
self.height = height
self.screen = screen
self.matrix = Matrix4()
self.scale = scale
self.wireframe = False
self.pixels = []
def clear(self):
self.screen.fill((0,0,0))
self.pixels = []
def drawScreen(self):
self.pixels = sorted(self.pixels, key=lambda x: x[3])
for p in self.pixels:
gfxdraw.pixel(self.screen, p[0], p[1], p[2])
def putPixel(self, vector, color):
if not vector.x < 0 and not vector.x > self.width:
if not vector.y < 0 and not vector.y > self.height:
self.pixels += [[int(vector.x), int(vector.y), color, vector.z]]
def drawLine(self, point0, point1, color):
dist = (point0 - point1).length
if dist < 2:
return
middlePoint = point0 + (point1 - point0)/Vector2(2, 2)
self.putPixel(middlePoint, color)
self.drawLine(point0, middlePoint, color)
self.drawLine(middlePoint, point1, color)
def drawScanLine(self, pointA, pointB, y, sz, ez, sShade, eShade, color):
if pointA > pointB:
temp = pointA
pointA = pointB
pointB = temp
z_slope = (ez - sz)/(pointB - pointA)
shadingGradient = (eShade - sShade)/(pointB - pointA)
for x in range(int(pointA), int(pointB)):
if x > self.width:
return
if y > self.height:
return
color *= sShade
color = self.clamp(color, 0, 255)
self.putPixel(Vector3(x, y, sz), (color, color, color))
sz += z_slope
sShade += shadingGradient
def clamp (self, value, min, max):
if value < min:
return min
elif value > max:
return max
else:
return value
"""a.y <= b.y <= c.y"""
def drawTriangle(self, pointA, pointB, pointC, shades, color):
#Uses rasterization to draw triangle
#slopes dx/dy
drawUpper = True
#handles weird exceptions
if int(pointA.y - pointC.y) != 0:
slopeAC = (pointA.x - pointC.x)/(pointA.y - pointC.y)
slopeACZ = (pointA.z - pointC.z)/(pointA.y - pointC.y)
#new beta test
shadingGradientAC = (shades[0] - shades[2])/(pointA.y - pointC.y)
else:
return
if int(pointA.y - pointB.y) != 0:
slopeAB = (pointA.x - pointB.x)/(pointA.y - pointB.y)
slopeABZ = (pointA.z - pointB.z)/(pointA.y - pointB.y)
#new beta code
shadingGradientAB = (shades[0] - shades[1])/(pointA.y - pointB.y)
else:
drawUpper = False
self.drawScanLine(pointA.x, pointB.x, pointA.y, pointA.z, pointB.z, shades[0], shades[0], color)
slopeBC = (pointB.x - pointC.x)/(pointB.y - pointC.y)
slopeBCZ = (pointB.z - pointC.z)/(pointB.y - pointC.y)
#new beta
shadingGradientBC = (shades[1] - shades[2])/(pointB.y - pointC.y)
sx, ex = pointA.x, pointA.x
sz, ez = pointA.z, pointA.z
#new beta test
sShade, eShade = shades[0], shades[0]
if drawUpper:
for y in range(int(pointA.y), int(pointB.y)):
self.drawScanLine(sx, ex, y, sz, ez, sShade, eShade, color)
sx += slopeAC
ex += slopeAB
sz += slopeACZ
ez += slopeABZ
sShade += shadingGradientAC
eShade += shadingGradientAB
else:
ex = pointB.x
ez = pointB.z
eShade = shades[1]
for y in range(int(pointB.y), int(pointC.y)):
self.drawScanLine(sx, ex, y, sz, ez, sShade, eShade, color)
sx += slopeAC
ex += slopeBC
sz += slopeACZ
ez += slopeBCZ
sShade += shadingGradientAC
eShade += shadingGradientBC
def drawBline(self, point0, point1, color):
x0 = int(point0.x)
y0 = int(point0.y)
x1 = int(point1.x)
y1 = int(point1.y)
dx = abs(x1 - x0)
dy = abs(y1 - y0)
if (x0 < x1):
sx = 1
else:
sx = -1
if y0 < y1:
sy = 1
else:
sy = -1
err = dx - dy
while x0 != x1 or y0 != y1:
self.putPixel(Vector3(x0, y0, 0), color)
e2 = 2 * err
if e2 > -dy:
err -= dy
x0 += sx
elif e2 < dx:
err += dx
y0 += sy
def project(self, vector, aspectRatio, zNear, zFar, fov_degree):
if (aspectRatio > 1):
sX = 1/aspectRatio
else:
sX = 1
if (aspectRatio > 1):
sY = 1
else:
sY = aspectRatio
fov = 1/math.tan(math.radians(fov_degree/2))
scaleX = fov * sX
scaleY = fov * sY
projectionMatrix = self.matrix.get_projection_matrix(zNear, zFar, scaleX, scaleY)
projectedVector = Vector4(vector.x, vector.y, vector.z, 1).dot_product_with_matrix(projectionMatrix)
return Vector3(projectedVector[0] + self.width/2, projectedVector[1]+ self.height/2, projectedVector[2])
def render(self, camera, lights, meshes):
polygons=[]
for mesh in meshes:
rotationMatrix = self.matrix.get_rotation_matrix(mesh.rotation.x, mesh.rotation.y, mesh.rotation.z)
translationMatrix = self.matrix.get_translation_matrix(mesh.position - camera.position)
worldMatrix = np.dot(self.matrix.get_scaling_matrix(Vector3(self.scale,self.scale,self.scale)), np.dot(rotationMatrix, translationMatrix))
for face in mesh.faces:
# print face[0], face[1], face[2]
vertexA = mesh.vertices[face[0][0]]
vertexB = mesh.vertices[face[0][1]]
vertexC = mesh.vertices[face[0][2]]
vertexA = vertexA.dot_product_with_matrix(worldMatrix)
vertexB = vertexB.dot_product_with_matrix(worldMatrix)
vertexC = vertexC.dot_product_with_matrix(worldMatrix)
pixelA = self.project(vertexA, 800/600, 1, 1000, 70)
pixelB = self.project(vertexB, 800/600, 1, 1000, 70)
pixelC = self.project(vertexC, 800/600, 1, 1000, 70)
if not self.wireframe:
#still in devoplment.... BETA!!!!
norm1 = mesh.normals[face[1][0]]
norm2 = mesh.normals[face[1][1]]
norm3 = mesh.normals[face[1][2]]
norm1 = norm1.dot_product_with_matrix(worldMatrix)
norm2 = norm2.dot_product_with_matrix(worldMatrix)
norm3 = norm3.dot_product_with_matrix(worldMatrix)
#sorts pixels
pixels = [pixelA, pixelB, pixelC]
pixels = sorted(pixels, key=lambda x: x.y)
color = 255
#gouraud shading
# vertices = [[vertexA, pixelA.y], [vertexB, pixelB.y], [vertexC, pixelC.y]]
# vertices = sorted(vertices, key=lambda x: x[1])
#
# diffuseIntensities = []
# for light in lights:
# lightingDistanceA = light.position - vertices[0][0]
# lightingDistanceB = light.position - vertices[1][0]
# lightingDistanceC = light.position - vertices[2][0]
#
# diffuseIntensities += [light.getIntensity(vertices[0][0]) * norm1.normalize().dot_product(lightingDistanceA.normalize())]
# diffuseIntensities += [light.getIntensity(vertices[1][0]) * norm1.normalize().dot_product(lightingDistanceB.normalize())]
# diffuseIntensities += [light.getIntensity(vertices[2][0]) * norm1.normalize().dot_product(lightingDistanceC.normalize())]
#
# self.drawTriangle(pixels[0], pixels[1], pixels[2], diffuseIntensities, color)
#flat shading
nFace = (norm1 + norm2 + norm3)/Vector3(3,3,3)
center = (vertexA + vertexB + vertexC)/Vector3(3,3,3)
diffuseIntensity = 0
for light in lights:
lightingDistance = light.position - center
diffuseIntensity += light.getIntensity(center) * \
nFace.normalize().dot_product(lightingDistance.normalize())
color *= diffuseIntensity
color += 50
color = self.clamp(color, 0, 255)
polygons += [(self.screen, (int(color), int(color), int(color)),
[[pixelA.x, pixelA.y], [pixelB.x, pixelB.y], [pixelC.x, pixelC.y]], vertexA.z)]
else:
self.drawBline(pixelA, pixelB, (255,255,255))
self.drawBline(pixelB, pixelC, (255,255,255))
self.drawBline(pixelC, pixelA, (255,255,255))
if not self.wireframe:
polygons = sorted(polygons, key=lambda x: x[3])
for polygon in polygons:
pygame.draw.polygon(polygon[0], polygon[1], polygon[2])
self.drawScreen() | [
"Vector2.Vector2",
"pygame.gfxdraw.pixel",
"math.radians",
"Vector3.Vector3",
"Matrix4.Matrix4",
"numpy.dot",
"pygame.draw.polygon",
"Vector4.Vector4"
] | [((383, 392), 'Matrix4.Matrix4', 'Matrix4', ([], {}), '()\n', (390, 392), False, 'from Matrix4 import Matrix4\n'), ((5489, 5596), 'Vector3.Vector3', 'Vector3', (['(projectedVector[0] + self.width / 2)', '(projectedVector[1] + self.height / 2)', 'projectedVector[2]'], {}), '(projectedVector[0] + self.width / 2, projectedVector[1] + self.\n height / 2, projectedVector[2])\n', (5496, 5596), False, 'from Vector3 import Vector3\n'), ((688, 732), 'pygame.gfxdraw.pixel', 'gfxdraw.pixel', (['self.screen', 'p[0]', 'p[1]', 'p[2]'], {}), '(self.screen, p[0], p[1], p[2])\n', (701, 732), False, 'from pygame import gfxdraw\n'), ((1154, 1167), 'Vector2.Vector2', 'Vector2', (['(2)', '(2)'], {}), '(2, 2)\n', (1161, 1167), False, 'from Vector2 import Vector2\n'), ((1871, 1888), 'Vector3.Vector3', 'Vector3', (['x', 'y', 'sz'], {}), '(x, y, sz)\n', (1878, 1888), False, 'from Vector3 import Vector3\n'), ((4705, 4723), 'Vector3.Vector3', 'Vector3', (['x0', 'y0', '(0)'], {}), '(x0, y0, 0)\n', (4712, 4723), False, 'from Vector3 import Vector3\n'), ((5193, 5221), 'math.radians', 'math.radians', (['(fov_degree / 2)'], {}), '(fov_degree / 2)\n', (5205, 5221), False, 'import math, pygame\n'), ((5390, 5430), 'Vector4.Vector4', 'Vector4', (['vector.x', 'vector.y', 'vector.z', '(1)'], {}), '(vector.x, vector.y, vector.z, 1)\n', (5397, 5430), False, 'from Vector4 import Vector4\n'), ((6002, 6043), 'numpy.dot', 'np.dot', (['rotationMatrix', 'translationMatrix'], {}), '(rotationMatrix, translationMatrix)\n', (6008, 6043), True, 'import numpy as np\n'), ((9708, 9763), 'pygame.draw.polygon', 'pygame.draw.polygon', (['polygon[0]', 'polygon[1]', 'polygon[2]'], {}), '(polygon[0], polygon[1], polygon[2])\n', (9727, 9763), False, 'import math, pygame\n'), ((5958, 6001), 'Vector3.Vector3', 'Vector3', (['self.scale', 'self.scale', 'self.scale'], {}), '(self.scale, self.scale, self.scale)\n', (5965, 6001), False, 'from Vector3 import Vector3\n'), ((8586, 8602), 'Vector3.Vector3', 'Vector3', (['(3)', '(3)', '(3)'], {}), '(3, 3, 3)\n', (8593, 8602), False, 'from Vector3 import Vector3\n'), ((8660, 8676), 'Vector3.Vector3', 'Vector3', (['(3)', '(3)', '(3)'], {}), '(3, 3, 3)\n', (8667, 8676), False, 'from Vector3 import Vector3\n')] |
import astropy.units as u
from numpy.linalg import norm
from .izzo import lambert as lambert_izzo
class Maneuver:
r"""Class to represent a Maneuver.
Each ``Maneuver`` consists on a list of impulses :math:`\Delta v_i`
(changes in velocity) each one applied at a certain instant :math:`t_i`.
You can access them directly indexing the ``Maneuver`` object itself.
>>> man = Maneuver((0 * u.s, [1, 0, 0] * u.km / u.s),
... (10 * u.s, [1, 0, 0] * u.km / u.s))
>>> man[0]
(<Quantity 0. s>, <Quantity [1., 0., 0.] km / s>)
>>> man.impulses[1]
(<Quantity 10. s>, <Quantity [1., 0., 0.] km / s>)
"""
def __init__(self, *args):
r"""Constructor.
Parameters
----------
impulses : list
List of pairs (delta_time, delta_velocity)
"""
self.impulses = args
# HACK: Change API or validation code
_dts, _dvs = zip(*args)
self._dts, self._dvs = self._initialize(
[(_dt * u.one).value for _dt in _dts] * (_dts[0] * u.one).unit,
[(_dv * u.one).value for _dv in _dvs] * (_dvs[0] * u.one).unit,
)
try:
if not all(len(dv) == 3 for dv in self._dvs):
raise TypeError
except TypeError:
raise ValueError("Delta-V must be three dimensions vectors")
def __repr__(self):
return f"Number of impulses: {len(self.impulses)}, Total cost: {self.get_total_cost():.6f}"
@u.quantity_input(dts=u.s, dvs=u.m / u.s)
def _initialize(self, dts, dvs):
return dts, dvs
# def __getitem__(self, key):
# return self.impulses[key]
@classmethod
def lambert(cls, orbit_i, orbit_f, method=lambert_izzo, short=True, **kwargs):
"""Computes Lambert maneuver between two different points.
Parameters
----------
orbit_i: ~poliastro.twobody.Orbit
Initial orbit
orbit_f: ~poliastro.twobody.Orbit
Final orbit
method: function
Method for solving Lambert's problem
short: keyword, boolean
Selects between short and long solution
"""
# Get initial algorithm conditions
k = orbit_i.attractor.k
r_i = orbit_i.r
r_f = orbit_f.r
# Time of flight is solved by subtracting both orbit epochs
tof = orbit_f.epoch - orbit_i.epoch
# Compute all possible solutions to the Lambert transfer
sols = list(method(k, r_i, r_f, tof, **kwargs))
# Return short or long solution
if short:
dv_a, dv_b = sols[0]
else:
dv_a, dv_b = sols[-1]
return cls(
(0 * u.s, (dv_a - orbit_i.v).decompose()),
(tof.to(u.s), (orbit_f.v - dv_b).decompose()),
)
def get_total_time(self):
"""Returns total time of the maneuver.
"""
total_time = sum(self._dts, 0 * u.s)
return total_time
def get_total_cost(self):
"""Returns total cost of the maneuver.
"""
dvs = [norm(dv) for dv in self._dvs]
return sum(dvs, 0 * u.km / u.s)
| [
"numpy.linalg.norm",
"astropy.units.quantity_input"
] | [((1476, 1516), 'astropy.units.quantity_input', 'u.quantity_input', ([], {'dts': 'u.s', 'dvs': '(u.m / u.s)'}), '(dts=u.s, dvs=u.m / u.s)\n', (1492, 1516), True, 'import astropy.units as u\n'), ((3069, 3077), 'numpy.linalg.norm', 'norm', (['dv'], {}), '(dv)\n', (3073, 3077), False, 'from numpy.linalg import norm\n')] |
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import numpy as np
import importlib
from cosyai.dataset.base import _BaseDataset
from cosyai.util import check_config_none
class RandSet(_BaseDataset):
def __init__(self, conf):
super().__init__(conf)
check_config_none(conf, ["input_dim", "output_dim", "dataset_size"])
self._create(conf.task, conf.input_dim, conf.output_dim,
conf.dataset_size, conf.split_weight)
self._transform(conf.backend)
def _create(self,
task,
input_dim,
output_dim,
num,
split_weight,):
split_weight = split_weight or [7, 1, 2]
x = np.random.rand(num, 1, input_dim,)
if task == "classification":
y = np.random.randint(0, 2, (num, output_dim))
elif task == "regression":
y = np.random.rand(num, output_dim)
else:
raise NotImplementedError()
indices = np.cumsum(
np.asarray(split_weight) * num / np.sum(split_weight), dtype=int)
self.train_set = x[:indices[0]], y[:indices[0]]
self.eval_set = x[indices[0]:indices[1]], y[indices[0]:indices[1]]
self.test_set = x[indices[1]:], y[indices[1]:]
def _transform(self, backend):
module = importlib.import_module('cosyai.backend.' + backend +
'.util')
self.train_set = module.data_transformer(*self.train_set)
self.eval_set = module.data_transformer(*self.eval_set)
self.test_set = module.data_transformer(*self.test_set)
| [
"numpy.sum",
"importlib.import_module",
"numpy.asarray",
"cosyai.util.check_config_none",
"numpy.random.randint",
"numpy.random.rand"
] | [((270, 338), 'cosyai.util.check_config_none', 'check_config_none', (['conf', "['input_dim', 'output_dim', 'dataset_size']"], {}), "(conf, ['input_dim', 'output_dim', 'dataset_size'])\n", (287, 338), False, 'from cosyai.util import check_config_none\n'), ((717, 750), 'numpy.random.rand', 'np.random.rand', (['num', '(1)', 'input_dim'], {}), '(num, 1, input_dim)\n', (731, 750), True, 'import numpy as np\n'), ((1334, 1396), 'importlib.import_module', 'importlib.import_module', (["('cosyai.backend.' + backend + '.util')"], {}), "('cosyai.backend.' + backend + '.util')\n", (1357, 1396), False, 'import importlib\n'), ((806, 848), 'numpy.random.randint', 'np.random.randint', (['(0)', '(2)', '(num, output_dim)'], {}), '(0, 2, (num, output_dim))\n', (823, 848), True, 'import numpy as np\n'), ((900, 931), 'numpy.random.rand', 'np.random.rand', (['num', 'output_dim'], {}), '(num, output_dim)\n', (914, 931), True, 'import numpy as np\n'), ((1061, 1081), 'numpy.sum', 'np.sum', (['split_weight'], {}), '(split_weight)\n', (1067, 1081), True, 'import numpy as np\n'), ((1028, 1052), 'numpy.asarray', 'np.asarray', (['split_weight'], {}), '(split_weight)\n', (1038, 1052), True, 'import numpy as np\n')] |
'''
Code from
https://github.com/matheusgadelha/MRTNet/blob/master/models/AutoEncoder.py
https://github.com/matheusgadelha/MRTNet/blob/master/models/MRTDecoder.py
revised by <NAME>
'''
import torch
import torch.nn as nn
import numpy as np
import math
from torch.nn import Sequential, Linear, ModuleList
tree_arch = {}
tree_arch[4] = [4, 8, 8, 8]
tree_arch[6] = [2, 4, 4, 4, 4, 4]
tree_arch[8] = [2, 2, 2, 2, 2, 4, 4, 4]
tree_arch[11] = [2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2]
class SRTDecoder(nn.Module):
def __init__(self, z_dim, nlevels, feat_dims, num_output_points):
super(SRTDecoder, self).__init__()
self.z_dim = z_dim
self.nlevels = nlevels
self.feat_dims = feat_dims
self.num_output_points = num_output_points
self.tarch, self.num_nodes = self.get_arch()
self.base_size = int(self.tarch[0])
self.fc1 = Linear(self.z_dim, self.base_size * self.feat_dims[0])
upconv_list = []
for level in range(1, self.nlevels):
upconv_list.append(self.upconv(level))
self.upconv_list = ModuleList(upconv_list)
self.final_conv = nn.Sequential()
self.final_conv.add_module('final_conv1',
nn.ConvTranspose1d(self.feat_dims[-1], 32, kernel_size=1, stride=1, padding=0))
self.final_conv.add_module('relu_final',
nn.ReLU(inplace=True))
self.final_conv.add_module('final_conv2',
nn.ConvTranspose1d(32, 3, kernel_size=1, stride=1, padding=0))
self.final_conv.add_module('tanh_final',
nn.Tanh())
def get_arch(self):
logmult = int(math.log2(self.num_output_points / 2048))
tarch = tree_arch[self.nlevels]
if self.num_output_points == 16384:
while logmult > 0:
last_min_pos = np.where(tarch == np.min(tarch))[0][-1]
tarch[last_min_pos] *= 2
logmult -= 1
# number of node for each level
num_nodes = []
for i, up_ratio in enumerate(tarch):
if i == 0:
num_nodes.append(up_ratio)
else:
last_num_node = num_nodes[-1]
num_nodes.append(up_ratio * last_num_node)
return tarch, num_nodes
def upconv(self, level):
in_channels = self.feat_dims[level-1]
out_channels = self.feat_dims[level]
up_ratio = self.tarch[level]
return Sequential(
nn.ConvTranspose1d(in_channels, out_channels, kernel_size=up_ratio, stride=up_ratio, padding=0),
nn.LeakyReLU(0.2, inplace=True)
)
def forward(self, z):
batch_size = z.shape[0]
node = self.fc1(z).view(batch_size, -1, self.base_size)
for upconv in self.upconv_list:
node = upconv(node)
out = self.final_conv(node)
out = torch.transpose(out, 1, 2).contiguous()
return out | [
"torch.nn.ReLU",
"torch.nn.Sequential",
"torch.nn.ModuleList",
"torch.nn.Tanh",
"torch.nn.ConvTranspose1d",
"numpy.min",
"torch.nn.Linear",
"torch.nn.LeakyReLU",
"math.log2",
"torch.transpose"
] | [((876, 930), 'torch.nn.Linear', 'Linear', (['self.z_dim', '(self.base_size * self.feat_dims[0])'], {}), '(self.z_dim, self.base_size * self.feat_dims[0])\n', (882, 930), False, 'from torch.nn import Sequential, Linear, ModuleList\n'), ((1088, 1111), 'torch.nn.ModuleList', 'ModuleList', (['upconv_list'], {}), '(upconv_list)\n', (1098, 1111), False, 'from torch.nn import Sequential, Linear, ModuleList\n'), ((1139, 1154), 'torch.nn.Sequential', 'nn.Sequential', ([], {}), '()\n', (1152, 1154), True, 'import torch.nn as nn\n'), ((1221, 1299), 'torch.nn.ConvTranspose1d', 'nn.ConvTranspose1d', (['self.feat_dims[-1]', '(32)'], {'kernel_size': '(1)', 'stride': '(1)', 'padding': '(0)'}), '(self.feat_dims[-1], 32, kernel_size=1, stride=1, padding=0)\n', (1239, 1299), True, 'import torch.nn as nn\n'), ((1366, 1387), 'torch.nn.ReLU', 'nn.ReLU', ([], {'inplace': '(True)'}), '(inplace=True)\n', (1373, 1387), True, 'import torch.nn as nn\n'), ((1455, 1516), 'torch.nn.ConvTranspose1d', 'nn.ConvTranspose1d', (['(32)', '(3)'], {'kernel_size': '(1)', 'stride': '(1)', 'padding': '(0)'}), '(32, 3, kernel_size=1, stride=1, padding=0)\n', (1473, 1516), True, 'import torch.nn as nn\n'), ((1583, 1592), 'torch.nn.Tanh', 'nn.Tanh', ([], {}), '()\n', (1590, 1592), True, 'import torch.nn as nn\n'), ((1645, 1685), 'math.log2', 'math.log2', (['(self.num_output_points / 2048)'], {}), '(self.num_output_points / 2048)\n', (1654, 1685), False, 'import math\n'), ((2485, 2585), 'torch.nn.ConvTranspose1d', 'nn.ConvTranspose1d', (['in_channels', 'out_channels'], {'kernel_size': 'up_ratio', 'stride': 'up_ratio', 'padding': '(0)'}), '(in_channels, out_channels, kernel_size=up_ratio, stride=\n up_ratio, padding=0)\n', (2503, 2585), True, 'import torch.nn as nn\n'), ((2609, 2640), 'torch.nn.LeakyReLU', 'nn.LeakyReLU', (['(0.2)'], {'inplace': '(True)'}), '(0.2, inplace=True)\n', (2621, 2640), True, 'import torch.nn as nn\n'), ((2915, 2941), 'torch.transpose', 'torch.transpose', (['out', '(1)', '(2)'], {}), '(out, 1, 2)\n', (2930, 2941), False, 'import torch\n'), ((1851, 1864), 'numpy.min', 'np.min', (['tarch'], {}), '(tarch)\n', (1857, 1864), True, 'import numpy as np\n')] |
# Copyright 2020 <NAME>
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
"""
A few routines to generate random data
"""
from typing import Optional, Tuple
import numpy as np
from scipy.stats import cauchy, expon, gamma
from .utils import iscomplex, tensor_H
def crandn(*shape, dtype=np.complex128):
""" wrapper for numpy.random.randn that can produce complex numbers """
out = np.zeros(shape, dtype=dtype)
if iscomplex(out):
out = np.random.randn(*shape) + 1j * np.random.randn(*shape)
return out.astype(dtype)
else:
out = np.random.randn(*shape)
return out.astype(dtype)
def rand_psd(*shape, inner=None, dtype=np.complex):
""" Random PSD matrices """
if inner is None:
shape = shape + (shape[-1],)
else:
shape = shape + (inner,)
X = crandn(*shape, dtype=dtype)
V = X @ tensor_H(X) / shape[-1]
return 0.5 * (V + tensor_H(V))
def circular_symmetric(
loc=0, scale=1.0, dim=1, size=None, distrib="laplace", dtype=np.complex128
):
if size is None:
size = [1]
elif not isinstance(size, tuple):
size = (size,)
# generate first normal vectors
out = crandn(*((dim,) + size), dtype=dtype)
out /= np.linalg.norm(out, axis=0)
# generate the norms according to exponential distribution
if distrib == "laplace":
# the marginal of the norm of symmetric circularl Laplace distributed
# vectors is a gamma random variable with shape=dim and scale=1
if dtype in [np.complex128, np.complex64]:
a = 2 * dim
else:
a = dim
norms = gamma.rvs(a, size=size)
elif distrib == "gauss":
raise NotImplementedError
norms = cauchy.rvs(loc=loc, scale=scale, size=size)
else:
raise NotImplementedError()
out *= norms[None, ...]
return out
def rand_mixture(
n_freq: int,
n_sources: int,
n_frames: int,
distrib: Optional[str] = "laplace",
scale: Optional[float] = 1.0,
dtype: Optional[np.dtype] = np.complex128,
) -> Tuple[np.array, np.array, np.array]:
A = crandn(n_freq, n_sources, n_sources, dtype=dtype)
groundtruths = circular_symmetric(
dim=n_freq,
size=(n_sources, n_frames),
distrib=distrib,
scale=scale,
dtype=dtype,
)
mixtures = A @ groundtruths
return mixtures, groundtruths, A
| [
"scipy.stats.gamma.rvs",
"numpy.random.randn",
"numpy.zeros",
"scipy.stats.cauchy.rvs",
"numpy.linalg.norm"
] | [((1397, 1425), 'numpy.zeros', 'np.zeros', (['shape'], {'dtype': 'dtype'}), '(shape, dtype=dtype)\n', (1405, 1425), True, 'import numpy as np\n'), ((2233, 2260), 'numpy.linalg.norm', 'np.linalg.norm', (['out'], {'axis': '(0)'}), '(out, axis=0)\n', (2247, 2260), True, 'import numpy as np\n'), ((1575, 1598), 'numpy.random.randn', 'np.random.randn', (['*shape'], {}), '(*shape)\n', (1590, 1598), True, 'import numpy as np\n'), ((2629, 2652), 'scipy.stats.gamma.rvs', 'gamma.rvs', (['a'], {'size': 'size'}), '(a, size=size)\n', (2638, 2652), False, 'from scipy.stats import cauchy, expon, gamma\n'), ((1463, 1486), 'numpy.random.randn', 'np.random.randn', (['*shape'], {}), '(*shape)\n', (1478, 1486), True, 'import numpy as np\n'), ((2732, 2775), 'scipy.stats.cauchy.rvs', 'cauchy.rvs', ([], {'loc': 'loc', 'scale': 'scale', 'size': 'size'}), '(loc=loc, scale=scale, size=size)\n', (2742, 2775), False, 'from scipy.stats import cauchy, expon, gamma\n'), ((1494, 1517), 'numpy.random.randn', 'np.random.randn', (['*shape'], {}), '(*shape)\n', (1509, 1517), True, 'import numpy as np\n')] |
"""
Build JobStats (returned to the client after job completion) - based mostly on the DataFrame of collected metrics from
the invoker and all workers.
"""
# Copyright 2021 The Funnel Rocket Maintainers
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import logging
import sys
from typing import Optional, Union, List, Dict
import pandas
import numpy as np
from pandas import DataFrame
from frocket.common.config import config
from frocket.common.dataset import DatasetPartsInfo, PartNamingMethod
from frocket.common.tasks.base import JobStats, JobDatasetStats, JobInvokerStats, TimingStats, JobWorkerStats
from frocket.invoker.metrics_frame import MetricsFrame, METRIC_NAME_COLUMN, METRIC_VALUE_COLUMN, METRIC_SOURCE_COLUMN
from frocket.common.metrics import MetricName, ComponentLabel, SUCCESS_LABEL, MetricLabelEnum, \
WorkerStartupLabel, LoadFromLabel
logger = logging.getLogger(__name__)
TASK_COMPLETION_GRANULARITY_SECONDS = 0.25 # Data series of task success over time is measured in this resolution
TIMING_PERCENTILES = [float(pct) for pct in config.get('stats.timing.percentiles').split(',')]
MIN_METRICS_FOR_PERCENTILES = 20 # Below this sample count, don't return percentiles
MIN_METRICS_FOR_99_PERCENTILE = 100 # Below this count, don't return 99th percentile
# List of keys to pull from Pandas' describe()
TIMING_DESCRIBE_KEYS = ['min', 'mean', 'max'] + [f"{int(pct*100)}%" for pct in TIMING_PERCENTILES]
def build_stats(frame: MetricsFrame, parts_info: DatasetPartsInfo = None) -> JobStats:
df = frame.dataframe
if df is None: # In job failure cases
return JobStats()
if parts_info:
ds_stats = JobDatasetStats(total_size=parts_info.total_size, parts=parts_info.total_parts)
else:
ds_stats = None
# Invoker stats
all_task_rows_df = _filter_by_label(df, ComponentLabel.WORKER)
successful_task_rows_df = _filter_by_success(all_task_rows_df)
total_tasks = _count_tasks(all_task_rows_df)
failed_tasks = total_tasks - _count_tasks(successful_task_rows_df)
invoker_stats = JobInvokerStats(
enqueue_time=_sum_value(df, MetricName.ASYNC_ENQUEUE_SECONDS, single_value=True),
poll_time=_sum_value(df, MetricName.ASYNC_POLL_SECONDS, single_value=True),
total_tasks=total_tasks,
failed_tasks=failed_tasks,
task_success_over_time=_task_success_over_time(successful_task_rows_df)
# TODO backlog add: lost_task_retries as counted by the invoker; support sync. invokers?
)
# Worker stats
worker_stats = JobWorkerStats(
cold_tasks=_count_tasks(_filter_by_label(successful_task_rows_df, WorkerStartupLabel.COLD)),
warm_tasks=_count_tasks(_filter_by_label(successful_task_rows_df, WorkerStartupLabel.WARM)),
scanned_rows=_sum_value(successful_task_rows_df, MetricName.SCANNED_ROWS, as_int=True),
scanned_groups=_sum_value(successful_task_rows_df, MetricName.SCANNED_GROUPS, as_int=True),
cache=_cache_performance(successful_task_rows_df),
invoke_latency=_timing_stats(successful_task_rows_df, MetricName.INVOKE_TO_RUN_SECONDS),
load_time=_timing_stats(successful_task_rows_df, MetricName.TASK_TOTAL_LOAD_SECONDS),
total_time=_timing_stats(successful_task_rows_df, MetricName.TASK_TOTAL_RUN_SECONDS)
# TODO backlog add: loaded_column_types - mapping of column type to count, which affects load time
)
job_stats = JobStats(
total_time=_sum_value(df, MetricName.INVOKER_TOTAL_SECONDS, single_value=True),
cost=_total_cost(df),
dataset=ds_stats,
invoker=invoker_stats,
worker=worker_stats)
return job_stats
def _task_success_over_time(task_rows_df: DataFrame) -> Dict[float, int]:
"""Return a sparse series of data points - for each time slot (e.g. 0.25 secs) since the job started, return how
many tasks completed successfully in that slot. Non-cumulative, does not include zeros."""
task_duration_rows = _filter_by_metrics(
task_rows_df, metrics=[MetricName.INVOKE_TO_RUN_SECONDS, MetricName.TASK_TOTAL_RUN_SECONDS])
task_durations = task_duration_rows.groupby(METRIC_SOURCE_COLUMN)[METRIC_VALUE_COLUMN].sum()
quantized_task_durations = \
np.ceil(task_durations / TASK_COMPLETION_GRANULARITY_SECONDS) * TASK_COMPLETION_GRANULARITY_SECONDS
return quantized_task_durations.value_counts().sort_index().to_dict()
def _cache_performance(task_rows_df: DataFrame) -> Dict[str, int]:
return {
# Note the 'source' is always the case for locally-loaded files, in which case caching is N/A.
'source': _count_tasks(_filter_by_label(task_rows_df, LoadFromLabel.SOURCE)),
'diskCache': _count_tasks(_filter_by_label(task_rows_df, LoadFromLabel.DISK_CACHE))
}
def _sum_value(df: DataFrame, metric: MetricName,
single_value: bool = False,
as_int: bool = False) -> Union[float, int, None]:
df = _filter_by_metrics(df, metric)
if single_value:
assert len(df) <= 1
if df.empty:
return None
else:
values_sum = df[METRIC_VALUE_COLUMN].sum()
return int(values_sum) if as_int else float(values_sum)
def _count(df: DataFrame, metric: MetricName) -> int:
return _filter_by_metrics(df, metric)[METRIC_VALUE_COLUMN].count()
def _timing_stats(task_rows_df: DataFrame, metric: MetricName) -> TimingStats:
values_df = _filter_by_metrics(task_rows_df, metric)[METRIC_VALUE_COLUMN]
if len(values_df) < MIN_METRICS_FOR_PERCENTILES:
percentiles = [0.5]
else:
percentiles = TIMING_PERCENTILES
if len(values_df) < MIN_METRICS_FOR_99_PERCENTILE:
percentiles = [pct for pct in percentiles if pct < 0.99]
raw_stats = values_df.describe(percentiles=percentiles).to_dict()
return {k: v for k, v in raw_stats.items()
if k in TIMING_DESCRIBE_KEYS and not np.isnan(v)}
def _filter_by_metrics(df: DataFrame, metrics: Union[MetricName, List[MetricName]]) -> DataFrame:
if type(metrics) is MetricName:
return df[df[METRIC_NAME_COLUMN] == metrics.name]
else:
return df[df[METRIC_NAME_COLUMN].isin([m.name for m in metrics])]
def _filter_by_label(df: DataFrame, label: MetricLabelEnum) -> DataFrame:
return df[df[label.label_name] == label.label_value.lower()]
def _filter_by_success(df: DataFrame, value: bool = True) -> DataFrame:
return df[df[SUCCESS_LABEL] == str(value)]
def _count_tasks(task_rows_df: DataFrame) -> int:
"""Each task attempt (e.g. task index 117, attempt 2) has a unique name in the source column, which ofc appears in
multiple rows. This count the unique count of task attempt IDs in the given DF."""
return task_rows_df[METRIC_SOURCE_COLUMN].nunique()
def _total_cost(df: DataFrame) -> Optional[float]:
cost_reports_df = _filter_by_metrics(df, MetricName.COST_DOLLARS)
num_reports = len(cost_reports_df)
if num_reports == 0:
logger.debug(f"Total cost: no metrics found")
return None
else:
total_cost = float(cost_reports_df[METRIC_VALUE_COLUMN].sum())
logger.debug(f"Total cost: ${total_cost:.6f} (sum of {num_reports} metric reports)")
return total_cost
# Stand-alone testing
if __name__ == "__main__":
config.init_logging(force_level=logging.DEBUG, force_console_output=True)
filename = config.get('metrics.export.lastrun', None)
if not filename:
sys.exit('No lastrun file defined')
df = pandas.read_parquet(filename)
dummy_frame = MetricsFrame([])
dummy_frame._df = df
dummy_parts_info = DatasetPartsInfo(naming_method=PartNamingMethod.LIST, total_parts=4, total_size=1024)
build_stats(dummy_frame, dummy_parts_info)
| [
"frocket.invoker.metrics_frame.MetricsFrame",
"frocket.common.config.config.init_logging",
"numpy.ceil",
"frocket.common.config.config.get",
"frocket.common.tasks.base.JobDatasetStats",
"sys.exit",
"numpy.isnan",
"pandas.read_parquet",
"frocket.common.tasks.base.JobStats",
"frocket.common.dataset.... | [((1379, 1406), 'logging.getLogger', 'logging.getLogger', (['__name__'], {}), '(__name__)\n', (1396, 1406), False, 'import logging\n'), ((7793, 7866), 'frocket.common.config.config.init_logging', 'config.init_logging', ([], {'force_level': 'logging.DEBUG', 'force_console_output': '(True)'}), '(force_level=logging.DEBUG, force_console_output=True)\n', (7812, 7866), False, 'from frocket.common.config import config\n'), ((7882, 7924), 'frocket.common.config.config.get', 'config.get', (['"""metrics.export.lastrun"""', 'None'], {}), "('metrics.export.lastrun', None)\n", (7892, 7924), False, 'from frocket.common.config import config\n'), ((8000, 8029), 'pandas.read_parquet', 'pandas.read_parquet', (['filename'], {}), '(filename)\n', (8019, 8029), False, 'import pandas\n'), ((8048, 8064), 'frocket.invoker.metrics_frame.MetricsFrame', 'MetricsFrame', (['[]'], {}), '([])\n', (8060, 8064), False, 'from frocket.invoker.metrics_frame import MetricsFrame, METRIC_NAME_COLUMN, METRIC_VALUE_COLUMN, METRIC_SOURCE_COLUMN\n'), ((8113, 8202), 'frocket.common.dataset.DatasetPartsInfo', 'DatasetPartsInfo', ([], {'naming_method': 'PartNamingMethod.LIST', 'total_parts': '(4)', 'total_size': '(1024)'}), '(naming_method=PartNamingMethod.LIST, total_parts=4,\n total_size=1024)\n', (8129, 8202), False, 'from frocket.common.dataset import DatasetPartsInfo, PartNamingMethod\n'), ((2108, 2118), 'frocket.common.tasks.base.JobStats', 'JobStats', ([], {}), '()\n', (2116, 2118), False, 'from frocket.common.tasks.base import JobStats, JobDatasetStats, JobInvokerStats, TimingStats, JobWorkerStats\n'), ((2158, 2237), 'frocket.common.tasks.base.JobDatasetStats', 'JobDatasetStats', ([], {'total_size': 'parts_info.total_size', 'parts': 'parts_info.total_parts'}), '(total_size=parts_info.total_size, parts=parts_info.total_parts)\n', (2173, 2237), False, 'from frocket.common.tasks.base import JobStats, JobDatasetStats, JobInvokerStats, TimingStats, JobWorkerStats\n'), ((4743, 4804), 'numpy.ceil', 'np.ceil', (['(task_durations / TASK_COMPLETION_GRANULARITY_SECONDS)'], {}), '(task_durations / TASK_COMPLETION_GRANULARITY_SECONDS)\n', (4750, 4804), True, 'import numpy as np\n'), ((7954, 7989), 'sys.exit', 'sys.exit', (['"""No lastrun file defined"""'], {}), "('No lastrun file defined')\n", (7962, 7989), False, 'import sys\n'), ((1567, 1605), 'frocket.common.config.config.get', 'config.get', (['"""stats.timing.percentiles"""'], {}), "('stats.timing.percentiles')\n", (1577, 1605), False, 'from frocket.common.config import config\n'), ((6410, 6421), 'numpy.isnan', 'np.isnan', (['v'], {}), '(v)\n', (6418, 6421), True, 'import numpy as np\n')] |
import warnings
from io import BytesIO
from tempfile import NamedTemporaryFile
import onnx
import torch.nn
from torch.nn import Module
import torch.nn.functional as F
from torchvision import models
from numpy.testing import assert_almost_equal
import numpy as np
import tensorflow as tf
from onnx2keras import onnx2keras, compatible_data_format, OnnxConstant, OnnxTensor, InterleavedImageBatch, \
ensure_data_format, OptimizationMissingWarning
def make_onnx_model(net, indata, opset_version=None):
fd = BytesIO()
torch.onnx.export(net, indata, fd, opset_version=opset_version)
# with open("/tmp/t.onnx", "wb") as debug: torch.onnx.export(net, indata, debug, opset_version=opset_version)
fd.seek(0)
return onnx.load(fd)
def convert_and_compare_output(net, indata, precition=5, image_out=True, savable=True, missing_optimizations=False, opset_version=None):
try:
return _convert_and_compare_output(net, indata, precition, image_out, savable, missing_optimizations, opset_version)
except AssertionError:
return _convert_and_compare_output(net, indata, precition, image_out, savable, missing_optimizations, opset_version)
def _convert_and_compare_output(net, indata, precition=5, image_out=True, savable=True, missing_optimizations=False, opset_version=None):
torch_indata = torch.tensor(indata)
y1 = net(torch_indata).detach().numpy()
onnx_model = make_onnx_model(net, torch.zeros_like(torch_indata), opset_version)
with warnings.catch_warnings(record=True) as warns:
warnings.simplefilter("always")
kernas_net = onnx2keras(onnx_model)
warns = [w for w in warns if w.category is OptimizationMissingWarning]
if not missing_optimizations:
assert len(warns) == 0
if savable:
with NamedTemporaryFile() as f:
f.close()
kernas_net.save(f.name)
y2 = kernas_net.predict(indata.transpose(0, 2, 3, 1))
if image_out:
y2 = y2.transpose(0, 3, 1, 2)
assert_almost_equal(y1, y2, precition)
return kernas_net
class GlobalAvgPool(Module):
def forward(self, x):
return x.mean([2, 3])
class TestUtils:
def test_compatible_data_format(self):
assert compatible_data_format(OnnxConstant, OnnxConstant)
assert compatible_data_format(OnnxTensor, OnnxTensor)
assert compatible_data_format(OnnxConstant, OnnxTensor)
assert compatible_data_format(OnnxTensor, OnnxConstant)
assert compatible_data_format(InterleavedImageBatch, InterleavedImageBatch)
assert not compatible_data_format(OnnxTensor, InterleavedImageBatch)
assert not compatible_data_format(OnnxConstant, InterleavedImageBatch)
assert not compatible_data_format(InterleavedImageBatch, OnnxTensor)
assert not compatible_data_format(InterleavedImageBatch, OnnxConstant)
class TestOnnx:
def test_conv(self):
net = torch.nn.Sequential(torch.nn.Conv2d(3, 16, 7), torch.nn.ReLU())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_conv_no_bias(self):
net = torch.nn.Sequential(torch.nn.Conv2d(3, 16, 7, bias=False), torch.nn.ReLU())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_conv_padding(self):
net = torch.nn.Sequential(torch.nn.Conv2d(1, 16, 3, padding=1), torch.nn.ReLU())
x = np.random.rand(1, 1, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_prelu(self):
net = torch.nn.Sequential(torch.nn.Conv2d(3, 16, 7), torch.nn.PReLU())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_prelu_per_channel(self):
act = torch.nn.PReLU(num_parameters=16)
with torch.no_grad():
act.weight[:] = torch.tensor(range(16))
net = torch.nn.Sequential(torch.nn.Conv2d(3, 16, 7), act)
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x, 5)
def test_maxpool(self):
net = torch.nn.Sequential(torch.nn.MaxPool2d(2))
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_maxpool_resnet(self):
net = torch.nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
x = np.random.rand(1, 192, 272, 64).astype(np.float32)
convert_and_compare_output(net, x)
def test_concat(self):
for axis in range(1,4):
class Dbl(torch.nn.Module):
def forward(self, x):
return torch.cat((x, x), axis)
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(Dbl(), x)
def test_conv_transpose(self):
net = torch.nn.Sequential(torch.nn.ConvTranspose2d(3, 16, 5, 2), torch.nn.ReLU())
x = np.random.rand(1, 3, 112, 112).astype(np.float32)
convert_and_compare_output(net, x)
def test_conv_transpose_padding(self):
net = torch.nn.Sequential(torch.nn.ConvTranspose2d(3, 16, 4, 2, padding=1), torch.nn.ReLU())
x = np.random.rand(1, 3, 112, 112).astype(np.float32)
convert_and_compare_output(net, x)
def test_conv_different_padding(self):
net = torch.nn.Sequential(torch.nn.Conv2d(3, 64, kernel_size=7, stride=1, padding=(3, 4)))
x = np.random.rand(1, 3, 384, 544).astype(np.float32)
convert_and_compare_output(net, x)
def test_conv_transpose_no_bias(self):
net = torch.nn.Sequential(torch.nn.ConvTranspose2d(3, 16, 5, 2, bias=False), torch.nn.ReLU())
x = np.random.rand(1, 3, 112, 112).astype(np.float32)
convert_and_compare_output(net, x)
def test_conv_transpose_grouped_no_bias(self):
net = torch.nn.Sequential(torch.nn.ConvTranspose2d(16, 16, 5, 2, groups=2, bias=False), torch.nn.ReLU())
x = np.random.rand(1, 16, 112, 112).astype(np.float32)
convert_and_compare_output(net, x)
def test_conv_transpose_grouped_bias(self):
net = torch.nn.Sequential(torch.nn.ConvTranspose2d(16, 16, 5, 2, groups=2), torch.nn.ReLU())
x = np.random.rand(1, 16, 112, 112).astype(np.float32)
convert_and_compare_output(net, x)
def test_conv_transpose_grouped_fully(self):
net = torch.nn.Sequential(torch.nn.ConvTranspose2d(16, 16, 5, 2, groups=16), torch.nn.ReLU())
x = np.random.rand(1, 16, 112, 112).astype(np.float32)
convert_and_compare_output(net, x)
def test_conv_transpose_output_padding(self):
net = torch.nn.Sequential(torch.nn.ConvTranspose2d(16, 16, 3, 2, output_padding=1), torch.nn.ReLU())
x = np.random.rand(1, 16, 112, 112).astype(np.float32)
convert_and_compare_output(net, x)
def test_conv_stride2_padding_strange(self):
net = torch.nn.Sequential(torch.nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3))
x = np.random.rand(1, 3, 384, 544).astype(np.float32)
convert_and_compare_output(net, x)
def test_conv_stride2_padding_simple_odd(self):
net = torch.nn.Sequential(torch.nn.Conv2d(3, 64, kernel_size=3, stride=2, padding=1))
x = np.random.rand(1, 3, 223, 223).astype(np.float32)
kernas_net = convert_and_compare_output(net, x)
assert [l.__class__.__name__ for l in kernas_net.layers] == ['InputLayer', 'Conv2D']
def test_conv_stride2_padding_simple_even(self):
net = torch.nn.Sequential(torch.nn.Conv2d(3, 64, kernel_size=3, stride=2, padding=1))
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
kernas_net = convert_and_compare_output(net, x)
# assert [l.__class__.__name__ for l in kernas_net.layers] == ['InputLayer', 'Conv2D']
def test_batchnorm(self):
bn = torch.nn.BatchNorm2d(3)
bn.running_mean.uniform_()
bn.running_var.uniform_()
net = torch.nn.Sequential(bn, torch.nn.ReLU())
net.eval()
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_batchnorm1d(self):
bn = torch.nn.BatchNorm1d(3)
bn.running_mean.uniform_()
bn.running_var.uniform_()
net = torch.nn.Sequential(GlobalAvgPool(), bn, torch.nn.ReLU())
net.eval()
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
def test_clamp(self):
class Clamp(Module):
def forward(self, x):
return torch.clamp(x, 0.3, 0.7)
net = torch.nn.Sequential(torch.nn.ReLU(), Clamp(), torch.nn.ReLU())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x, savable=False)
def test_relu6(self):
class Clamp(Module):
def forward(self, x):
return torch.clamp(x, 0, 6)
net = torch.nn.Sequential(torch.nn.ReLU(), Clamp(), torch.nn.ReLU())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_leaky_relu(self):
net = torch.nn.Sequential(torch.nn.Conv2d(3, 3, 3), torch.nn.LeakyReLU(), torch.nn.Conv2d(3, 3, 3))
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_depthwise(self):
net = torch.nn.Sequential(torch.nn.Conv2d(3, 3, 7, groups=3), torch.nn.ReLU())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_groupwise(self):
net = torch.nn.Conv2d(8, 8, 7, groups=4)
x = np.random.rand(1, 8, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_depthwise_no_bias(self):
net = torch.nn.Sequential(torch.nn.Conv2d(3, 3, 7, groups=3, bias=False), torch.nn.ReLU())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_add(self):
class AddTst(Module):
def __init__(self):
Module.__init__(self)
self.conv1 = torch.nn.Conv2d(3, 3, 7)
self.conv2 = torch.nn.Conv2d(3, 3, 7)
def forward(self, x):
return self.conv1(x).relu_() + self.conv2(x).relu_()
net = torch.nn.Sequential(AddTst(), torch.nn.ReLU())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_global_avrage_pooling(self):
net = torch.nn.Sequential(GlobalAvgPool(), torch.nn.ReLU())
x = np.random.rand(1, 3, 16, 16).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
def test_dropout(self):
net = torch.nn.Sequential(GlobalAvgPool(), torch.nn.Dropout(), torch.nn.ReLU())
net.eval()
x = np.random.rand(1, 3, 16, 16).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
def test_linear(self):
net = torch.nn.Sequential(GlobalAvgPool(), torch.nn.Linear(3, 8), torch.nn.ReLU())
net.eval()
x = np.random.rand(5, 3, 16, 16).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
def test_linear_no_bias(self):
net = torch.nn.Sequential(GlobalAvgPool(), torch.nn.Linear(3, 8, bias=False), torch.nn.ReLU())
net.eval()
x = np.random.rand(5, 3, 16, 16).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
def test_mobilenet_v2(self):
net = models.mobilenet_v2()
net.eval()
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
def test_avg_pool_pad(self):
class PadTst(Module):
def forward(self, x):
return F.avg_pool2d(x, kernel_size=3, stride=1, padding=1)
net = torch.nn.Sequential(PadTst(), torch.nn.ReLU())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_avg_pool_pad_asym(self):
class PadTst(Module):
def forward(self, x):
return F.avg_pool2d(x, kernel_size=(3, 6), stride=(1, 2), padding=(1, 2))
net = torch.nn.Sequential(PadTst(), torch.nn.ReLU())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_gloabl_avg_pool(self):
class AvgTst(Module):
def forward(self, x):
return F.adaptive_avg_pool2d(x, (1, 1))
net = torch.nn.Sequential(AvgTst(), torch.nn.ReLU())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_flatten(self):
class Tst(Module):
def forward(self, x):
return torch.flatten(x, 1)
net = torch.nn.Sequential(Tst(), torch.nn.ReLU())
x = np.random.rand(1, 3, 1, 1).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
def test_vector_pad(self):
class VectorPad2D(Module):
def forward(self, x):
tt = [torch.nn.functional.pad(x[:, i:i + 1], [1,1,1,1], 'constant', [1,2,3][i])
for i in range(x.shape[1])]
return torch.cat(tt, 1)
net = torch.nn.Sequential(VectorPad2D(), torch.nn.ReLU())
x = np.random.rand(2, 3, 5, 5).astype(np.float32)
convert_and_compare_output(net, x)
def test_vector_pad_addhack(self):
class VectorPad2D(Module):
def forward(self, x):
c = torch.tensor([1,2,3]).reshape(1, 3, 1, 1)
return torch.nn.functional.pad(x - c, [1,1,1,1]) + c
net = torch.nn.Sequential(VectorPad2D(), torch.nn.ReLU())
x = np.random.rand(1, 3, 5, 5).astype(np.float32)
convert_and_compare_output(net, x)
def test_vector_pad_addhack_asym(self):
class VectorPad2D(Module):
def forward(self, x):
c = torch.tensor([1,2,3]).reshape(1, 3, 1, 1)
return torch.nn.functional.pad(x - c, [1,0,1,0]) + c
net = torch.nn.Sequential(VectorPad2D(), torch.nn.ReLU())
x = np.random.rand(1, 3, 5, 5).astype(np.float32)
convert_and_compare_output(net, x)
def test_sigmoid(self):
net = torch.nn.Sequential(torch.nn.Conv2d(3, 16, 7), torch.nn.Sigmoid())
x = np.random.rand(1, 3, 224, 224).astype(np.float32)
convert_and_compare_output(net, x)
def test_upsample_nearest(self):
net = torch.nn.Sequential(torch.nn.UpsamplingNearest2d(scale_factor=2), torch.nn.ReLU())
x = np.random.rand(1, 3, 32, 32).astype(np.float32)
convert_and_compare_output(net, x)
def test_upsample_nearest_v11(self):
net = torch.nn.Sequential(torch.nn.UpsamplingNearest2d(scale_factor=2), torch.nn.ReLU())
x = np.random.rand(1, 3, 32, 32).astype(np.float32)
convert_and_compare_output(net, x, opset_version=11)
def test_upsample_bilinear(self):
net = torch.nn.Sequential(torch.nn.UpsamplingBilinear2d(scale_factor=2), torch.nn.ReLU())
x = np.random.rand(1, 3, 32, 32).astype(np.float32)
convert_and_compare_output(net, x, opset_version=11)
def test_interpolate_nearest(self):
class Net(Module):
def forward(self, x):
return F.interpolate(x, scale_factor=2, mode="nearest")
net = torch.nn.Sequential(Net(), torch.nn.ReLU())
x = np.random.rand(1, 3, 32, 32).astype(np.float32)
convert_and_compare_output(net, x)
def test_interpolate_bilinear(self):
class Net(Module):
def forward(self, x):
return F.interpolate(x, scale_factor=2, mode="bilinear", align_corners=True)
net = torch.nn.Sequential(Net(), torch.nn.ReLU())
x = np.random.rand(1, 3, 32, 32).astype(np.float32)
convert_and_compare_output(net, x, opset_version=11)
def test_eq_mul(self):
class EqProd(Module):
def forward(self, x):
maxmap = F.max_pool2d(x, 3, 1, 1, 1, False, False)
return x * (maxmap == x)
net = torch.nn.Sequential(EqProd(), torch.nn.ReLU())
x = np.random.rand(1, 3, 5, 5).astype(np.float32)
convert_and_compare_output(net, x)
def test_adaptive_avgpool_reshape(self):
class Net(Module):
def forward(self, x):
return F.adaptive_avg_pool2d(x, 1).reshape(x.shape[0], -1)
net = torch.nn.Sequential(Net(), torch.nn.ReLU())
x = np.random.rand(1, 3, 16, 16).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
x = np.random.rand(4, 3, 16, 16).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
def test_bmm(self):
class Net(Module):
def forward(self, x):
x = x.reshape(1, 16, 16)
return torch.bmm(x, x)
net = torch.nn.Sequential(Net(), torch.nn.ReLU())
x = np.random.rand(1, 1, 16, 16).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
def test_matmul(self):
class Net(Module):
def forward(self, x):
x = x.reshape(1, 1, 16, 16)
return torch.matmul(x, x)
net = torch.nn.Sequential(Net(), torch.nn.ReLU())
x = np.random.rand(1, 1, 16, 16).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
def test_unsupported_optimasation(self):
class Reshape(Module):
def forward(self, x):
return x.reshape(4, 4, 16, 16)
net = torch.nn.Sequential(GlobalAvgPool(), torch.nn.Linear(3, 4 * 16 * 16), Reshape(),
torch.nn.Conv2d(4, 3, 3), torch.nn.ReLU())
net.eval()
x = np.random.rand(4, 3, 16, 16).astype(np.float32)
convert_and_compare_output(net, x, missing_optimizations=True)
def test_sqrt(self):
class Sq(Module):
def forward(self, x):
return torch.sqrt(x)
net = torch.nn.Sequential(Sq(), torch.nn.ReLU())
x = np.random.rand(4, 3, 16, 16).astype(np.float32)
is_tf1 = tuple(map(int, tf.__version__.split('.'))) < (2, 0, 0)
convert_and_compare_output(net, x, savable=(not is_tf1))
def test_abs(self):
class Abs(Module):
def forward(self, x):
return torch.abs(x)
net = torch.nn.Sequential(Abs(), torch.nn.ReLU())
x = np.random.rand(4, 3, 16, 16).astype(np.float32)
convert_and_compare_output(net, x)
def test_neg(self):
class Neg(Module):
def forward(self, x):
return -x
net = torch.nn.Sequential(Neg(), torch.nn.ReLU())
x = np.random.rand(4, 3, 16, 16).astype(np.float32)
convert_and_compare_output(net, x)
def test_center_crop(self):
class CenterCrop8x8(Module):
def forward(self, x):
n, c, h, w = x.shape
dx = (w - 8) // 2
dy = (h - 8) // 2
crop = x[:, :, dy:dy+8, dx:dx+8]
return crop
net = torch.nn.Sequential(CenterCrop8x8(), torch.nn.ReLU())
x = np.random.rand(4, 3, 16, 32).astype(np.float32)
convert_and_compare_output(net, x, opset_version=11)
def test_mul(self):
class Mul(Module):
def forward(self, x):
return x * x
net = torch.nn.Sequential(Mul(), torch.nn.ReLU())
x = np.random.rand(4, 3, 16, 32).astype(np.float32)
convert_and_compare_output(net, x, opset_version=11)
def test_mul_const(self):
class Mul(Module):
def forward(self, x):
return (2 * x) * (x * 2)
net = torch.nn.Sequential(Mul(), torch.nn.ReLU())
x = np.random.rand(4, 3, 16, 32).astype(np.float32)
convert_and_compare_output(net, x, opset_version=11)
def test_concat_for_OnnxTensor(self):
batch = 4
class Net(Module):
def __init__(self):
super().__init__()
self.conv1 = torch.nn.Conv2d(3, 4, 5)
self.conv2 = torch.nn.Conv2d(3, 4, 5)
self.ClassHead = torch.nn.ModuleList([torch.nn.Conv2d(4, 16, 3), torch.nn.Conv2d(4, 16, 3)])
def forward(self, x):
features = [self.conv1(x), self.conv2(x)]
classifications = torch.cat([self.ClassHead[i](feature).reshape(batch, -1, 2) for i, feature in enumerate(features)], dim=1)
return classifications
x = np.random.rand(batch, 3, 16, 32).astype(np.float32)
convert_and_compare_output(Net(), x, missing_optimizations=True, image_out=False)
def test_transpose_to_onnx_testor(self):
class Net(Module):
def forward(self, x):
return x.permute(0,2,3,1)
x = np.random.rand(2, 3, 16, 32).astype(np.float32)
convert_and_compare_output(Net(), x, image_out=False)
def test_concat_for_OnnxTensor_optimized(self):
batch = 4
class Net(Module):
def __init__(self):
super().__init__()
self.conv1 = torch.nn.Conv2d(3, 4, 5)
self.conv2 = torch.nn.Conv2d(3, 4, 5)
self.ClassHead = torch.nn.ModuleList([torch.nn.Conv2d(4, 16, 3), torch.nn.Conv2d(4, 16, 3)])
def forward(self, x):
features = [self.conv1(x), self.conv2(x)]
classifications = torch.cat([self.ClassHead[i](feature).permute(0,2,3,1).reshape(batch, -1, 2) for i, feature in enumerate(features)], dim=1)
return classifications
x = np.random.rand(batch, 3, 16, 32).astype(np.float32)
convert_and_compare_output(Net(), x, image_out=False)
def test_gather(self):
class Net(Module):
def forward(self, x):
return x.permute(0,2,3,1).select(2,1)
net = torch.nn.Sequential(Net(), torch.nn.ReLU())
x = np.random.rand(2, 3, 16, 32).astype(np.float32)
convert_and_compare_output(net, x, image_out=False)
# def test_inception_v3(self):
# net = models.Inception3(aux_logits=False)
# net.eval()
# x = np.random.rand(1, 3, 299, 299).astype(np.float32)
# convert_and_compare_output(net, x, image_out=False)
| [
"tempfile.NamedTemporaryFile",
"io.BytesIO",
"warnings.simplefilter",
"numpy.testing.assert_almost_equal",
"onnx2keras.onnx2keras",
"torch.nn.functional.avg_pool2d",
"tensorflow.__version__.split",
"torch.nn.functional.adaptive_avg_pool2d",
"warnings.catch_warnings",
"torch.nn.Module.__init__",
... | [((515, 524), 'io.BytesIO', 'BytesIO', ([], {}), '()\n', (522, 524), False, 'from io import BytesIO\n'), ((733, 746), 'onnx.load', 'onnx.load', (['fd'], {}), '(fd)\n', (742, 746), False, 'import onnx\n'), ((2004, 2042), 'numpy.testing.assert_almost_equal', 'assert_almost_equal', (['y1', 'y2', 'precition'], {}), '(y1, y2, precition)\n', (2023, 2042), False, 'from numpy.testing import assert_almost_equal\n'), ((1489, 1525), 'warnings.catch_warnings', 'warnings.catch_warnings', ([], {'record': '(True)'}), '(record=True)\n', (1512, 1525), False, 'import warnings\n'), ((1544, 1575), 'warnings.simplefilter', 'warnings.simplefilter', (['"""always"""'], {}), "('always')\n", (1565, 1575), False, 'import warnings\n'), ((1597, 1619), 'onnx2keras.onnx2keras', 'onnx2keras', (['onnx_model'], {}), '(onnx_model)\n', (1607, 1619), False, 'from onnx2keras import onnx2keras, compatible_data_format, OnnxConstant, OnnxTensor, InterleavedImageBatch, ensure_data_format, OptimizationMissingWarning\n'), ((2228, 2278), 'onnx2keras.compatible_data_format', 'compatible_data_format', (['OnnxConstant', 'OnnxConstant'], {}), '(OnnxConstant, OnnxConstant)\n', (2250, 2278), False, 'from onnx2keras import onnx2keras, compatible_data_format, OnnxConstant, OnnxTensor, InterleavedImageBatch, ensure_data_format, OptimizationMissingWarning\n'), ((2294, 2340), 'onnx2keras.compatible_data_format', 'compatible_data_format', (['OnnxTensor', 'OnnxTensor'], {}), '(OnnxTensor, OnnxTensor)\n', (2316, 2340), False, 'from onnx2keras import onnx2keras, compatible_data_format, OnnxConstant, OnnxTensor, InterleavedImageBatch, ensure_data_format, OptimizationMissingWarning\n'), ((2356, 2404), 'onnx2keras.compatible_data_format', 'compatible_data_format', (['OnnxConstant', 'OnnxTensor'], {}), '(OnnxConstant, OnnxTensor)\n', (2378, 2404), False, 'from onnx2keras import onnx2keras, compatible_data_format, OnnxConstant, OnnxTensor, InterleavedImageBatch, ensure_data_format, OptimizationMissingWarning\n'), ((2420, 2468), 'onnx2keras.compatible_data_format', 'compatible_data_format', (['OnnxTensor', 'OnnxConstant'], {}), '(OnnxTensor, OnnxConstant)\n', (2442, 2468), False, 'from onnx2keras import onnx2keras, compatible_data_format, OnnxConstant, OnnxTensor, InterleavedImageBatch, ensure_data_format, OptimizationMissingWarning\n'), ((2484, 2552), 'onnx2keras.compatible_data_format', 'compatible_data_format', (['InterleavedImageBatch', 'InterleavedImageBatch'], {}), '(InterleavedImageBatch, InterleavedImageBatch)\n', (2506, 2552), False, 'from onnx2keras import onnx2keras, compatible_data_format, OnnxConstant, OnnxTensor, InterleavedImageBatch, ensure_data_format, OptimizationMissingWarning\n'), ((11595, 11616), 'torchvision.models.mobilenet_v2', 'models.mobilenet_v2', ([], {}), '()\n', (11614, 11616), False, 'from torchvision import models\n'), ((1801, 1821), 'tempfile.NamedTemporaryFile', 'NamedTemporaryFile', ([], {}), '()\n', (1819, 1821), False, 'from tempfile import NamedTemporaryFile\n'), ((2572, 2629), 'onnx2keras.compatible_data_format', 'compatible_data_format', (['OnnxTensor', 'InterleavedImageBatch'], {}), '(OnnxTensor, InterleavedImageBatch)\n', (2594, 2629), False, 'from onnx2keras import onnx2keras, compatible_data_format, OnnxConstant, OnnxTensor, InterleavedImageBatch, ensure_data_format, OptimizationMissingWarning\n'), ((2649, 2708), 'onnx2keras.compatible_data_format', 'compatible_data_format', (['OnnxConstant', 'InterleavedImageBatch'], {}), '(OnnxConstant, InterleavedImageBatch)\n', (2671, 2708), False, 'from onnx2keras import onnx2keras, compatible_data_format, OnnxConstant, OnnxTensor, InterleavedImageBatch, ensure_data_format, OptimizationMissingWarning\n'), ((2728, 2785), 'onnx2keras.compatible_data_format', 'compatible_data_format', (['InterleavedImageBatch', 'OnnxTensor'], {}), '(InterleavedImageBatch, OnnxTensor)\n', (2750, 2785), False, 'from onnx2keras import onnx2keras, compatible_data_format, OnnxConstant, OnnxTensor, InterleavedImageBatch, ensure_data_format, OptimizationMissingWarning\n'), ((2805, 2864), 'onnx2keras.compatible_data_format', 'compatible_data_format', (['InterleavedImageBatch', 'OnnxConstant'], {}), '(InterleavedImageBatch, OnnxConstant)\n', (2827, 2864), False, 'from onnx2keras import onnx2keras, compatible_data_format, OnnxConstant, OnnxTensor, InterleavedImageBatch, ensure_data_format, OptimizationMissingWarning\n'), ((2998, 3028), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (3012, 3028), True, 'import numpy as np\n'), ((3227, 3257), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (3241, 3257), True, 'import numpy as np\n'), ((3455, 3485), 'numpy.random.rand', 'np.random.rand', (['(1)', '(1)', '(224)', '(224)'], {}), '(1, 1, 224, 224)\n', (3469, 3485), True, 'import numpy as np\n'), ((3666, 3696), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (3680, 3696), True, 'import numpy as np\n'), ((4006, 4036), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (4020, 4036), True, 'import numpy as np\n'), ((4200, 4230), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (4214, 4230), True, 'import numpy as np\n'), ((4410, 4441), 'numpy.random.rand', 'np.random.rand', (['(1)', '(192)', '(272)', '(64)'], {}), '(1, 192, 272, 64)\n', (4424, 4441), True, 'import numpy as np\n'), ((4946, 4976), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(112)', '(112)'], {}), '(1, 3, 112, 112)\n', (4960, 4976), True, 'import numpy as np\n'), ((5196, 5226), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(112)', '(112)'], {}), '(1, 3, 112, 112)\n', (5210, 5226), True, 'import numpy as np\n'), ((5444, 5474), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(384)', '(544)'], {}), '(1, 3, 384, 544)\n', (5458, 5474), True, 'import numpy as np\n'), ((5695, 5725), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(112)', '(112)'], {}), '(1, 3, 112, 112)\n', (5709, 5725), True, 'import numpy as np\n'), ((5965, 5996), 'numpy.random.rand', 'np.random.rand', (['(1)', '(16)', '(112)', '(112)'], {}), '(1, 16, 112, 112)\n', (5979, 5996), True, 'import numpy as np\n'), ((6221, 6252), 'numpy.random.rand', 'np.random.rand', (['(1)', '(16)', '(112)', '(112)'], {}), '(1, 16, 112, 112)\n', (6235, 6252), True, 'import numpy as np\n'), ((6479, 6510), 'numpy.random.rand', 'np.random.rand', (['(1)', '(16)', '(112)', '(112)'], {}), '(1, 16, 112, 112)\n', (6493, 6510), True, 'import numpy as np\n'), ((6745, 6776), 'numpy.random.rand', 'np.random.rand', (['(1)', '(16)', '(112)', '(112)'], {}), '(1, 16, 112, 112)\n', (6759, 6776), True, 'import numpy as np\n'), ((6995, 7025), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(384)', '(544)'], {}), '(1, 3, 384, 544)\n', (7009, 7025), True, 'import numpy as np\n'), ((7247, 7277), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(223)', '(223)'], {}), '(1, 3, 223, 223)\n', (7261, 7277), True, 'import numpy as np\n'), ((7606, 7636), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (7620, 7636), True, 'import numpy as np\n'), ((8030, 8060), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (8044, 8060), True, 'import numpy as np\n'), ((8365, 8395), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (8379, 8395), True, 'import numpy as np\n'), ((8702, 8732), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (8716, 8732), True, 'import numpy as np\n'), ((9033, 9063), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (9047, 9063), True, 'import numpy as np\n'), ((9278, 9308), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (9292, 9308), True, 'import numpy as np\n'), ((9501, 9531), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (9515, 9531), True, 'import numpy as np\n'), ((9686, 9716), 'numpy.random.rand', 'np.random.rand', (['(1)', '(8)', '(224)', '(224)'], {}), '(1, 8, 224, 224)\n', (9700, 9716), True, 'import numpy as np\n'), ((9929, 9959), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (9943, 9959), True, 'import numpy as np\n'), ((10125, 10146), 'torch.nn.Module.__init__', 'Module.__init__', (['self'], {}), '(self)\n', (10140, 10146), False, 'from torch.nn import Module\n'), ((10431, 10461), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (10445, 10461), True, 'import numpy as np\n'), ((10647, 10675), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(16)', '(16)'], {}), '(1, 3, 16, 16)\n', (10661, 10675), True, 'import numpy as np\n'), ((10903, 10931), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(16)', '(16)'], {}), '(1, 3, 16, 16)\n', (10917, 10931), True, 'import numpy as np\n'), ((11161, 11189), 'numpy.random.rand', 'np.random.rand', (['(5)', '(3)', '(16)', '(16)'], {}), '(5, 3, 16, 16)\n', (11175, 11189), True, 'import numpy as np\n'), ((11439, 11467), 'numpy.random.rand', 'np.random.rand', (['(5)', '(3)', '(16)', '(16)'], {}), '(5, 3, 16, 16)\n', (11453, 11467), True, 'import numpy as np\n'), ((11648, 11678), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (11662, 11678), True, 'import numpy as np\n'), ((11879, 11930), 'torch.nn.functional.avg_pool2d', 'F.avg_pool2d', (['x'], {'kernel_size': '(3)', 'stride': '(1)', 'padding': '(1)'}), '(x, kernel_size=3, stride=1, padding=1)\n', (11891, 11930), True, 'import torch.nn.functional as F\n'), ((12004, 12034), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (12018, 12034), True, 'import numpy as np\n'), ((12223, 12289), 'torch.nn.functional.avg_pool2d', 'F.avg_pool2d', (['x'], {'kernel_size': '(3, 6)', 'stride': '(1, 2)', 'padding': '(1, 2)'}), '(x, kernel_size=(3, 6), stride=(1, 2), padding=(1, 2))\n', (12235, 12289), True, 'import torch.nn.functional as F\n'), ((12363, 12393), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (12377, 12393), True, 'import numpy as np\n'), ((12580, 12612), 'torch.nn.functional.adaptive_avg_pool2d', 'F.adaptive_avg_pool2d', (['x', '(1, 1)'], {}), '(x, (1, 1))\n', (12601, 12612), True, 'import torch.nn.functional as F\n'), ((12686, 12716), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (12700, 12716), True, 'import numpy as np\n'), ((12982, 13008), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(1)', '(1)'], {}), '(1, 3, 1, 1)\n', (12996, 13008), True, 'import numpy as np\n'), ((13453, 13479), 'numpy.random.rand', 'np.random.rand', (['(2)', '(3)', '(5)', '(5)'], {}), '(2, 3, 5, 5)\n', (13467, 13479), True, 'import numpy as np\n'), ((13860, 13886), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(5)', '(5)'], {}), '(1, 3, 5, 5)\n', (13874, 13886), True, 'import numpy as np\n'), ((14272, 14298), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(5)', '(5)'], {}), '(1, 3, 5, 5)\n', (14286, 14298), True, 'import numpy as np\n'), ((14483, 14513), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (14497, 14513), True, 'import numpy as np\n'), ((14723, 14751), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(32)', '(32)'], {}), '(1, 3, 32, 32)\n', (14737, 14751), True, 'import numpy as np\n'), ((14965, 14993), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(32)', '(32)'], {}), '(1, 3, 32, 32)\n', (14979, 14993), True, 'import numpy as np\n'), ((15223, 15251), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(32)', '(32)'], {}), '(1, 3, 32, 32)\n', (15237, 15251), True, 'import numpy as np\n'), ((15457, 15505), 'torch.nn.functional.interpolate', 'F.interpolate', (['x'], {'scale_factor': '(2)', 'mode': '"""nearest"""'}), "(x, scale_factor=2, mode='nearest')\n", (15470, 15505), True, 'import torch.nn.functional as F\n'), ((15576, 15604), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(32)', '(32)'], {}), '(1, 3, 32, 32)\n', (15590, 15604), True, 'import numpy as np\n'), ((15793, 15862), 'torch.nn.functional.interpolate', 'F.interpolate', (['x'], {'scale_factor': '(2)', 'mode': '"""bilinear"""', 'align_corners': '(True)'}), "(x, scale_factor=2, mode='bilinear', align_corners=True)\n", (15806, 15862), True, 'import torch.nn.functional as F\n'), ((15933, 15961), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(32)', '(32)'], {}), '(1, 3, 32, 32)\n', (15947, 15961), True, 'import numpy as np\n'), ((16159, 16200), 'torch.nn.functional.max_pool2d', 'F.max_pool2d', (['x', '(3)', '(1)', '(1)', '(1)', '(False)', '(False)'], {}), '(x, 3, 1, 1, 1, False, False)\n', (16171, 16200), True, 'import torch.nn.functional as F\n'), ((16315, 16341), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(5)', '(5)'], {}), '(1, 3, 5, 5)\n', (16329, 16341), True, 'import numpy as np\n'), ((16656, 16684), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(16)', '(16)'], {}), '(1, 3, 16, 16)\n', (16670, 16684), True, 'import numpy as np\n'), ((16776, 16804), 'numpy.random.rand', 'np.random.rand', (['(4)', '(3)', '(16)', '(16)'], {}), '(4, 3, 16, 16)\n', (16790, 16804), True, 'import numpy as np\n'), ((17120, 17148), 'numpy.random.rand', 'np.random.rand', (['(1)', '(1)', '(16)', '(16)'], {}), '(1, 1, 16, 16)\n', (17134, 17148), True, 'import numpy as np\n'), ((17473, 17501), 'numpy.random.rand', 'np.random.rand', (['(1)', '(1)', '(16)', '(16)'], {}), '(1, 1, 16, 16)\n', (17487, 17501), True, 'import numpy as np\n'), ((17942, 17970), 'numpy.random.rand', 'np.random.rand', (['(4)', '(3)', '(16)', '(16)'], {}), '(4, 3, 16, 16)\n', (17956, 17970), True, 'import numpy as np\n'), ((18253, 18281), 'numpy.random.rand', 'np.random.rand', (['(4)', '(3)', '(16)', '(16)'], {}), '(4, 3, 16, 16)\n', (18267, 18281), True, 'import numpy as np\n'), ((18630, 18658), 'numpy.random.rand', 'np.random.rand', (['(4)', '(3)', '(16)', '(16)'], {}), '(4, 3, 16, 16)\n', (18644, 18658), True, 'import numpy as np\n'), ((18903, 18931), 'numpy.random.rand', 'np.random.rand', (['(4)', '(3)', '(16)', '(16)'], {}), '(4, 3, 16, 16)\n', (18917, 18931), True, 'import numpy as np\n'), ((19360, 19388), 'numpy.random.rand', 'np.random.rand', (['(4)', '(3)', '(16)', '(32)'], {}), '(4, 3, 16, 32)\n', (19374, 19388), True, 'import numpy as np\n'), ((19654, 19682), 'numpy.random.rand', 'np.random.rand', (['(4)', '(3)', '(16)', '(32)'], {}), '(4, 3, 16, 32)\n', (19668, 19682), True, 'import numpy as np\n'), ((19966, 19994), 'numpy.random.rand', 'np.random.rand', (['(4)', '(3)', '(16)', '(32)'], {}), '(4, 3, 16, 32)\n', (19980, 19994), True, 'import numpy as np\n'), ((20731, 20763), 'numpy.random.rand', 'np.random.rand', (['batch', '(3)', '(16)', '(32)'], {}), '(batch, 3, 16, 32)\n', (20745, 20763), True, 'import numpy as np\n'), ((21034, 21062), 'numpy.random.rand', 'np.random.rand', (['(2)', '(3)', '(16)', '(32)'], {}), '(2, 3, 16, 32)\n', (21048, 21062), True, 'import numpy as np\n'), ((21827, 21859), 'numpy.random.rand', 'np.random.rand', (['batch', '(3)', '(16)', '(32)'], {}), '(batch, 3, 16, 32)\n', (21841, 21859), True, 'import numpy as np\n'), ((22154, 22182), 'numpy.random.rand', 'np.random.rand', (['(2)', '(3)', '(16)', '(32)'], {}), '(2, 3, 16, 32)\n', (22168, 22182), True, 'import numpy as np\n'), ((4709, 4739), 'numpy.random.rand', 'np.random.rand', (['(1)', '(3)', '(224)', '(224)'], {}), '(1, 3, 224, 224)\n', (4723, 4739), True, 'import numpy as np\n'), ((18333, 18358), 'tensorflow.__version__.split', 'tf.__version__.split', (['"""."""'], {}), "('.')\n", (18353, 18358), True, 'import tensorflow as tf\n'), ((16534, 16561), 'torch.nn.functional.adaptive_avg_pool2d', 'F.adaptive_avg_pool2d', (['x', '(1)'], {}), '(x, 1)\n', (16555, 16561), True, 'import torch.nn.functional as F\n')] |
#!/usr/bin/env python
# -*- coding: utf-8 -*-
#
# nnutil2 - Tensorflow utilities for training neural networks
# Copyright (c) 2019, <NAME> <<EMAIL>>
#
# This file is part of 'nnutil2'.
#
# This file may be modified and distributed under the terms of the 3-clause BSD
# license. See the LICENSE file for details.
import tensorflow as tf
import numpy as np
def is_tensor(x):
return isinstance(x, (tf.Tensor, tf.SparseTensor, tf.RaggedTensor))
def as_tensor(structure):
if tf.nest.is_nested(structure):
return tf.nest.map_structure(as_tensor, structure)
elif isinstance(structure, tf.Tensor):
return structure
elif isinstance(structure, tf.SparseTensor):
return tf.sparse.to_dense(structure)
elif isinstance(structure, tf.RaggedTensor):
return tf.sparse.to_dense(structure)
elif isinstance(structure, np.ndarray):
return tf.constant(structure, shape=structure.shape, dtype=tf.dtype.as_dtype(structure.dtype))
elif isinstance(structure, (int, np.integer, np.signedinteger, float, np.floating, str, bytes)):
return tf.constant(structure, shape=(), dtype=tf.dtype.as_dtype(type(structure)))
else:
raise Exception("Cannot handle nested structure of type: {}".format(type(structure)))
def as_numpy(structure):
if tf.nest.is_nested(structure):
return tf.nest.map_structure(as_numpy, structure)
elif is_tensor(structure):
return structure.numpy()
elif isinstance(structure, np.ndarray):
return structure
elif isinstance(structure, (int, np.integer, np.signedinteger, float, np.floating, str, bytes)):
dtype = tf.dtype.as_dtype(type(structure)).as_numpy_dtype()
return np.array(structure, shape=(), dtype=dtype)
else:
raise Exception("Cannot handle nested structure of type: {}".format(type(structure)))
| [
"tensorflow.dtype.as_dtype",
"tensorflow.sparse.to_dense",
"tensorflow.nest.is_nested",
"tensorflow.nest.map_structure",
"numpy.array"
] | [((483, 511), 'tensorflow.nest.is_nested', 'tf.nest.is_nested', (['structure'], {}), '(structure)\n', (500, 511), True, 'import tensorflow as tf\n'), ((1310, 1338), 'tensorflow.nest.is_nested', 'tf.nest.is_nested', (['structure'], {}), '(structure)\n', (1327, 1338), True, 'import tensorflow as tf\n'), ((528, 571), 'tensorflow.nest.map_structure', 'tf.nest.map_structure', (['as_tensor', 'structure'], {}), '(as_tensor, structure)\n', (549, 571), True, 'import tensorflow as tf\n'), ((1355, 1397), 'tensorflow.nest.map_structure', 'tf.nest.map_structure', (['as_numpy', 'structure'], {}), '(as_numpy, structure)\n', (1376, 1397), True, 'import tensorflow as tf\n'), ((706, 735), 'tensorflow.sparse.to_dense', 'tf.sparse.to_dense', (['structure'], {}), '(structure)\n', (724, 735), True, 'import tensorflow as tf\n'), ((801, 830), 'tensorflow.sparse.to_dense', 'tf.sparse.to_dense', (['structure'], {}), '(structure)\n', (819, 830), True, 'import tensorflow as tf\n'), ((1718, 1760), 'numpy.array', 'np.array', (['structure'], {'shape': '()', 'dtype': 'dtype'}), '(structure, shape=(), dtype=dtype)\n', (1726, 1760), True, 'import numpy as np\n'), ((943, 977), 'tensorflow.dtype.as_dtype', 'tf.dtype.as_dtype', (['structure.dtype'], {}), '(structure.dtype)\n', (960, 977), True, 'import tensorflow as tf\n')] |
#!/usr/bin/env python3.7
#
# Copyright (c) University of Luxembourg 2021.
# Created by <NAME>, <EMAIL>, SnT, 2021.
#
import argparse
import numpy
from utilities import print_new_test, is_int, cosine, euclidean, searchStringInFile
parser = argparse.ArgumentParser()
parser.add_argument('--name', type=str)
parser.add_argument('--cov_a', type=str)
parser.add_argument('--cov_b', type=str)
parser.add_argument('--result', type=str)
parser.add_argument('--line', type=int)
parser.add_argument('--operator', type=str)
args = parser.parse_args()
cov_a = args.cov_a
cov_b = args.cov_b
name = args.name
result = args.result
lineNumber = int(args.line) - 1
operator = args.operator
def getCoverageAsList(test):
global name
coverage_line_splitted = searchStringInFile(test, name).split(':')
if len(coverage_line_splitted) == 1:
raise ValueError("coverage not found")
coverage = coverage_line_splitted[1]
coverage_frequencies = coverage.split(',')
return coverage_frequencies
def get_distance(testA, testB):
try:
coverageA_frequencies = getCoverageAsList(testA)
coverageB_frequencies = getCoverageAsList(testB)
except ValueError as err:
print(err.args)
return -1
covAList = []
for i in coverageA_frequencies:
if is_int(i):
covAList.append(int(i))
else:
covAList.append(int(0))
covBList = []
for i in coverageB_frequencies:
if is_int(i):
covBList.append(int(i))
else:
covBList.append(int(0))
# SDL, LOD corrections
global lineNumber
if operator == "SDL":
if len(covAList) == len(covBList):
del covAList[lineNumber]
del covBList[lineNumber]
else:
while len(covAList) != len(covBList):
if len(covAList) > len(covBList):
del covAList[lineNumber]
else:
covAList.insert(len(covAList), int(0))
del covAList[lineNumber]
del covBList[lineNumber]
elif operator == 'LOD':
while len(covAList) != len(covBList):
if len(covAList) > len(covBList):
del covAList[lineNumber]
else:
covAList.insert(len(covAList), int(0))
print(covAList)
print(covBList)
A = numpy.array(covAList)
B = numpy.array(covBList)
distance = cosine(A, B)
print(distance)
return distance
distance = get_distance(cov_a, cov_b)
print_new_test(result, distance)
| [
"utilities.searchStringInFile",
"argparse.ArgumentParser",
"utilities.cosine",
"utilities.print_new_test",
"numpy.array",
"utilities.is_int"
] | [((243, 268), 'argparse.ArgumentParser', 'argparse.ArgumentParser', ([], {}), '()\n', (266, 268), False, 'import argparse\n'), ((2739, 2771), 'utilities.print_new_test', 'print_new_test', (['result', 'distance'], {}), '(result, distance)\n', (2753, 2771), False, 'from utilities import print_new_test, is_int, cosine, euclidean, searchStringInFile\n'), ((2578, 2599), 'numpy.array', 'numpy.array', (['covAList'], {}), '(covAList)\n', (2589, 2599), False, 'import numpy\n'), ((2608, 2629), 'numpy.array', 'numpy.array', (['covBList'], {}), '(covBList)\n', (2619, 2629), False, 'import numpy\n'), ((2646, 2658), 'utilities.cosine', 'cosine', (['A', 'B'], {}), '(A, B)\n', (2652, 2658), False, 'from utilities import print_new_test, is_int, cosine, euclidean, searchStringInFile\n'), ((1514, 1523), 'utilities.is_int', 'is_int', (['i'], {}), '(i)\n', (1520, 1523), False, 'from utilities import print_new_test, is_int, cosine, euclidean, searchStringInFile\n'), ((1677, 1686), 'utilities.is_int', 'is_int', (['i'], {}), '(i)\n', (1683, 1686), False, 'from utilities import print_new_test, is_int, cosine, euclidean, searchStringInFile\n'), ((860, 890), 'utilities.searchStringInFile', 'searchStringInFile', (['test', 'name'], {}), '(test, name)\n', (878, 890), False, 'from utilities import print_new_test, is_int, cosine, euclidean, searchStringInFile\n')] |
import numpy
def permute_node(node, permutation_index, axis=-1):
"""Permute index of `node` array
Args:
node (numpy.ndarray): the array whose `axis` to be permuted.
permutation_index (numpy.ndarray): 1d numpy array whose size should be
same as permutation axis of `node`.
axis (int): permutation axis.
Returns (numpy.ndarray): permutated `node` array.
"""
if node.shape[axis] != len(permutation_index):
raise ValueError(
'node.shape[{}] = {} and len(permutation_index) = {} do not match!'
.format(axis, node.shape[axis], len(permutation_index)))
out_node = numpy.take(node, permutation_index, axis=axis).copy()
return out_node
def permute_adj(adj, permutation_index, axis=None):
"""Permute index of adjacency matrix array
Args:
adj (numpy.ndarray): the array whose `axis` to be permuted.
It is considered as adjacency matrix.
permutation_index (numpy.ndarray): 1d numpy array whose size should be
same as permutation axis of `node`.
axis (list or tuple or None): list of 2d int, indicates the permutation
axis. When None is passed (default), it uses -1 and -2 as `axis`,
it means that last 2 axis are considered to be permuted.
Returns (numpy.ndarray): permutated `adj` array.
"""
if axis is not None:
if not isinstance(axis, (list, tuple)):
raise TypeError('axis must be list or tuple, got {}'
.format(type(axis)))
if len(axis) != 2:
raise ValueError('axis length must 2, got {}'.format(len(axis)))
else:
axis = [-1, -2] # default value is to use last 2 axis
num_node = len(permutation_index)
for ax in axis:
if adj.shape[ax] != len(permutation_index):
raise ValueError(
'adj.shape[{}] = {} and len(permutation_index) = {} do not '
'match!'.format(axis, adj.shape[axis], len(permutation_index)))
out_adj = numpy.zeros_like(adj)
ndim = adj.ndim
for i in range(num_node):
for j in range(num_node):
in_indices = [slice(None)] * ndim
out_indices = [slice(None)] * ndim
in_indices[axis[0]] = i
in_indices[axis[1]] = j
out_indices[axis[0]] = permutation_index[i]
out_indices[axis[1]] = permutation_index[j]
out_adj[tuple(in_indices)] = adj[tuple(out_indices)]
return out_adj
| [
"numpy.zeros_like",
"numpy.take"
] | [((2052, 2073), 'numpy.zeros_like', 'numpy.zeros_like', (['adj'], {}), '(adj)\n', (2068, 2073), False, 'import numpy\n'), ((654, 700), 'numpy.take', 'numpy.take', (['node', 'permutation_index'], {'axis': 'axis'}), '(node, permutation_index, axis=axis)\n', (664, 700), False, 'import numpy\n')] |
import cv2
import numpy as np
from random import randint
from scipy.optimize import least_squares
from math import sqrt, atan2
def get_8_points(len_features):
"""
Function to get 8 indices of random points
Implements the 8-point algorithm
:param len_features: total no. of features retrieved from feature extractor
:return: a list containing indices of 8 random feature points
"""
eight_points = []
# Iterate until 8 points are not stored in the list
while len(eight_points) != 8:
# Get a index of a random point
index = randint(0, len_features - 1)
# Add only distinct points to the list
if index not in eight_points:
eight_points.append(index)
return eight_points
class MotionEstimator:
"""
A class to estimate 3D motion of a camera
"""
def __init__(self, focal_lengths, principal_pts):
# Define K-matrix using camera parameters
self.k_mat = np.array([[focal_lengths[0], 0, principal_pts[0]],
[0, focal_lengths[1], principal_pts[1]],
[0, 0, 1]])
# Define no. of iterations for RANSAC
self.iterations = 60
# Define threshold for outlier rejection in RANSAC
self.epsilon = 0.01
# Create SIFT detector object
self.sift = cv2.xfeatures2d.SIFT_create()
# Define parameters for Flann-based matcher
index_params = dict(algorithm=0, trees=5)
search_params = dict(checks=50)
# Create object of Flann-based matcher
self.matcher = cv2.FlannBasedMatcher(index_params, search_params)
# Define original homogeneous matrix for the camera pose
# Camera is considered to be at origin
self.original_h = np.identity(4)
self.h_mat_last_row = np.array([0, 0, 0, 1])
self.u = np.zeros((3, 1))
self.v = np.identity(3)
def extract_features(self, curr_img, next_img):
"""
Method to extract matching key features from 2 images
:param curr_img: current image frame from the dataset
:param next_img: next image frame from the dataset
:return: a tuple of lists containing matching key features from both images
"""
# Get key-points and descriptors for both frames
key_points_curr, descriptor_curr = self.sift.detectAndCompute(curr_img, None)
key_points_next, descriptor_next = self.sift.detectAndCompute(next_img, None)
# Get matches between the current and next frame
matches = self.matcher.knnMatch(descriptor_curr, descriptor_next, k=2)
# Define empty list to store features of current and next frame
features_curr, features_next = [], []
# Employ ratio test
for _, (m, n) in enumerate(matches):
if m.distance < 0.5 * n.distance:
features_curr.append(key_points_curr[m.queryIdx].pt)
features_next.append(key_points_next[m.trainIdx].pt)
return features_curr, features_next
def ransac_with_8_point(self, features_curr, features_next):
"""
Method to ransac on features of current and next image frame using 8-point algorithm
:param features_curr: a list of feature points in the current image frame
:param features_next: a list of feature points in the next image frame
:return: a tuple containing the 3x3 fundamental matrix, inliers in the current and next image frame
"""
count_inliers = 0
final_fundamental_mat = np.zeros((3, 3))
# Define list to store inliers from current and next image frame
inliers_curr, inliers_next = [], []
for _ in range(self.iterations):
count = 0
good_features_curr, good_features_next = [], []
temp_features_curr, temp_features_next = [], []
# Get 8 random points and extract features at those points
for pt in get_8_points(len(features_curr)):
good_features_curr.append([features_curr[pt][0], features_curr[pt][1]])
good_features_next.append([features_next[pt][0], features_next[pt][1]])
# Calculate fundamental matrix using these 8 features
fundamental_mat = self.calc_fundamental_matrix(good_features_curr, good_features_next)
# Employ outlier rejection
for i in range(len(features_curr)):
if self.check_fundamental_matrix(features_curr[i], features_next[i], fundamental_mat):
count += 1
temp_features_curr.append(features_curr[i])
temp_features_next.append(features_next[i])
if count > count_inliers:
count_inliers = count
final_fundamental_mat = fundamental_mat
inliers_curr, inliers_next = temp_features_curr, temp_features_next
return final_fundamental_mat, inliers_curr, inliers_next
@staticmethod
def calc_fundamental_matrix(features_curr, features_next):
"""
Method to estimate the fundamental matrix using 8 points
:param features_curr: a list of 8 feature points in the current image frame
:param features_next: a list of 8 feature points in the next image frame
:return: a 3x3 fundamental matrix
"""
a_mat = np.empty((8, 9))
# Iterate over all features
for i in range(len(features_curr)):
# Get positions of features in current and next frame
x_curr, y_curr = features_curr[i][0], features_curr[i][1]
x_next, y_next = features_next[i][0], features_next[i][1]
# Fill ith column of A matrix
a_mat[i] = np.array([x_next * x_curr, x_next * y_curr, x_next,
y_next * x_curr, y_next * y_curr, y_next,
x_curr, y_curr, 1])
# Get SVD of A matrix
_, _, v = np.linalg.svd(a_mat, full_matrices=True)
# Get SVD of last column of V matrix
u, s, v_new = np.linalg.svd(v[-1].reshape(3, 3))
# Restrain fundamental matrix to a rank of 2
s_new = np.array([[s[0], 0, 0],
[0, s[1], 0],
[0, 0, 0]])
f_mat = u @ s_new @ v_new
return f_mat
def check_fundamental_matrix(self, feature_curr, feature_next, fundamental_mat):
"""
Method to calculate transpose(x2).F.x1 and check if it satisfies the desired threshold
:param feature_curr: a tuple of feature in current image frame
:param feature_next: a tuple of corresponding feature in next image frame
:param fundamental_mat: a 3x3 array of fundamental matrix
:return: true if estimate is less than threshold
"""
# Get transpose of current features
fc_new = np.array([feature_curr[0], feature_curr[1], 1]).T
fn_new = np.array([feature_next[0], feature_next[1], 1])
# Estimate transpose(x2).F.x1
est = abs(np.squeeze(np.matmul(np.matmul(fn_new, fundamental_mat), fc_new)))
return est < self.epsilon
def calc_essential_matrix(self, fundamental_mat):
"""
Method to calculate the essential matrix
:param fundamental_mat: a numpy 3x3 array of fundamental matrix
:return: a numpy 3x3 array of essential matrix
"""
temp = np.matmul(np.matmul(self.k_mat.T, fundamental_mat), self.k_mat)
u, _, v = np.linalg.svd(temp, full_matrices=True)
sigma = np.array([[1, 0, 0], [0, 1, 0], [0, 0, 0]])
essential_mat = np.matmul(np.matmul(u, sigma), v)
return essential_mat
@staticmethod
def estimate_camera_pose(essential_mat):
"""
Estimate the various possible poses of the camera
:param essential_mat: a numpy 3x3 array of essential matrix
:return: a tuple containing 4 camera centers and 4 rotation matrices
"""
# Define empty lists to store camera poses
camera_centers, rotation_matrices = [], []
u, _, v = np.linalg.svd(essential_mat)
w_mat = np.array([[0, -1, 0], [1, 0, 0], [0, 0, 1]])
# Get the 4 pose configurations of the camera
for i in range(4):
# Evaluate center of camera for camera pose
cam_center = u[:, 2]
# Check if additive inverse needs to be taken
if i % 2 == 1:
cam_center = -cam_center
# Evaluate rotation matrix for camera pose
# Check whether transpose of W matrix needs to be taken
if i < 2:
rotation_mat = u @ w_mat @ v
else:
rotation_mat = u @ w_mat.T @ v
# Check for negative determinant condition
if np.linalg.det(rotation_mat) < 0:
cam_center, rotation_mat = -cam_center, -rotation_mat
camera_centers.append(cam_center.reshape((3, 1)))
rotation_matrices.append(rotation_mat)
return camera_centers, rotation_matrices
def disambiguate_camera_pose(self, camera_centers, rotation_matrices, inliers_curr, inliers_next):
"""
Method to find the unique camera pose using the 4 possible poses
:param camera_centers: a list of camera centers
:param rotation_matrices: a list of rotation matrices
:param inliers_curr: a list of inliers in the current image frame
:param inliers_next: a list of inliers in the next image frame
:return: a tuple containing the pose of the camera
"""
check = 0
rotation_final, cam_center_final, x_final = None, None, None
# Iterate over all the rotation matrices
for i in range(len(rotation_matrices)):
euler_angles = self.get_euler_angles(rotation_matrices[i])
if -50 < euler_angles[0] < 50 and -50 < euler_angles[2] < 50:
count = 0
x_inlier = None
cam_pose_new = np.hstack((rotation_matrices[i], camera_centers[i]))
for j in range(len(inliers_curr)):
temp_x = self.get_triangulation_point(cam_pose_new, inliers_curr[j], inliers_next[j])
r_mat_row = rotation_matrices[i][2, :].reshape((1, 3))
if np.squeeze(r_mat_row @ (temp_x - camera_centers[i])):
count += 1
x_inlier = temp_x
if check < count:
check = count
rotation_final = rotation_matrices[i]
cam_center_final = camera_centers[i]
x_final = x_inlier
if cam_center_final[2] > 0:
cam_center_final = -cam_center_final
return cam_center_final, rotation_final, x_final
@ staticmethod
def get_euler_angles(rotation_mat):
"""
Method to get Euler angles using a 3x3 rotation matrix
:param rotation_mat: a 3x3 numpy array of rotation matrix
:return: a tuple of Euler angles in x,y, and z directions respectively
"""
psi = sqrt((rotation_mat[0][0] ** 2) + (rotation_mat[1][0] ** 2))
if not psi < 1e-6:
x = atan2(rotation_mat[2][1], rotation_mat[2][2])
y = atan2(-rotation_mat[2][1], psi)
z = atan2(rotation_mat[1][0], rotation_mat[0][0])
else:
x = atan2(-rotation_mat[1][2], rotation_mat[1][1])
y = atan2(-rotation_mat[2][0], psi)
z = 0
return (x * 180 / np.pi), (y * 180 / np.pi), (z * 180 / np.pi)
def get_triangulation_point(self, camera_pose, inlier_curr, inlier_next):
"""
Method to employ triangular check for cheirality condition
Method to triangulate inliers
:param camera_pose: new camera pose
:param inlier_curr: an inlier in the current image frame
:param inlier_next: an inlier in the next image frame
:return:
"""
x_old = np.array([[0, -1, inlier_curr[1]],
[1, 0, -inlier_curr[0]],
[inlier_curr[1], inlier_curr[0], 0]])
x_new = np.array([[0, -1, inlier_next[1]],
[1, 0, -inlier_next[0]],
[inlier_next[1], inlier_next[0], 0]])
a_old = x_old @ self.original_h[0:3, :]
a_new = x_new @ camera_pose
a_mat = np.vstack((a_old, a_new))
_, _, v = np.linalg.svd(a_mat)
x_final = (v[-1] / v[-1][3]).reshape((4, 1))
return x_final[0:3].reshape((3, 1))
def get_homogeneous_matrix(self, rotation_mat, translation_mat):
"""
Method to get the homogeneous matrix
:param rotation_mat: a 3x3 numpy array of rotation matrix
:param translation_mat: a 3x1 translation matrix
:return: a 4x4 numpy array
"""
# Homogenoeous matrix is of the form: H = [r t
# 0 1]
# where r is a 3x3 rotational matrix and t is a 3x1 translation matrix
# Generate a 3x4 matrix of the form [r t]
z = np.column_stack((rotation_mat, translation_mat))
# Append the constant last row in the 3x4 matrix and return it
return np.vstack((z, self.h_mat_last_row))
@staticmethod
def get_drift(drift, custom_pose, opencv_pose):
"""
Evalute drift between 2 trajectories
:param drift: drift from the previous frame
:param custom_pose: pose of the camera using our pipeline
:param opencv_pose: pose of the camera using opencv's in-built functions
:return: drift of the current frame
"""
drift += sqrt(((custom_pose[0] - opencv_pose[1]) ** 2) + ((custom_pose[0] - opencv_pose[1]) ** 2))
return drift
def get_triangulation_error(self, x_linear, camera_pose):
"""
Get the error in linear triangulation
:param x_linear: X returned from linear triangulation
:param camera_pose: a tuple containing rotation and translation matrices
:return: error between the current and next frame
"""
p_old = np.matmul(self.k_mat, self.original_h[0:3, :])
p_new = np.matmul(self.k_mat, camera_pose)
x_linear = np.reshape(x_linear, (np.shape(x_linear)[0], 1))
x_0 = np.insert(x_linear, 3, 1)
x_pt = np.reshape(x_0, (4, 1))
error_1 = (self.u[0] - (np.dot(p_old[0], x_pt) / np.dot(p_old[2], x_pt))) ** 2 + (
self.u[1] - (np.dot(p_old[1], x_pt) / np.dot(p_new[2], x_pt))) ** 2
error_2 = (self.v[0] - (np.dot(p_new[0], x_pt) / np.dot(p_new[2], x_pt))) ** 2 + (
self.v[1] - (np.dot(p_new[1], x_pt) / np.dot(p_old[2], x_pt))) ** 2
error_total = error_1 + error_2
return error_total
def nonlinear_triangulation(self, x_linear, camera_pose, inlier_curr, inlier_next):
"""
Method to employ non-linear triangulation
:param x_linear: X returned from linear triangulation
:param camera_pose: a tuple containing rotation and translation matrices
:param inlier_curr: inliers from the current image frame
:param inlier_next: inliers from the next image frame
:return:
"""
x_init = np.reshape(x_linear, (np.shape(x_linear)[0]))
error = self.get_triangulation_error(x_linear, camera_pose)
refined_pts = least_squares(error, x_init, args=(self, x_linear, camera_pose, inlier_curr, inlier_next))
return refined_pts
| [
"math.atan2",
"numpy.empty",
"numpy.shape",
"numpy.linalg.svd",
"random.randint",
"numpy.identity",
"numpy.insert",
"scipy.optimize.least_squares",
"numpy.reshape",
"numpy.linalg.det",
"math.sqrt",
"cv2.FlannBasedMatcher",
"numpy.hstack",
"numpy.squeeze",
"numpy.dot",
"numpy.vstack",
... | [((575, 603), 'random.randint', 'randint', (['(0)', '(len_features - 1)'], {}), '(0, len_features - 1)\n', (582, 603), False, 'from random import randint\n'), ((964, 1071), 'numpy.array', 'np.array', (['[[focal_lengths[0], 0, principal_pts[0]], [0, focal_lengths[1],\n principal_pts[1]], [0, 0, 1]]'], {}), '([[focal_lengths[0], 0, principal_pts[0]], [0, focal_lengths[1],\n principal_pts[1]], [0, 0, 1]])\n', (972, 1071), True, 'import numpy as np\n'), ((1350, 1379), 'cv2.xfeatures2d.SIFT_create', 'cv2.xfeatures2d.SIFT_create', ([], {}), '()\n', (1377, 1379), False, 'import cv2\n'), ((1592, 1642), 'cv2.FlannBasedMatcher', 'cv2.FlannBasedMatcher', (['index_params', 'search_params'], {}), '(index_params, search_params)\n', (1613, 1642), False, 'import cv2\n'), ((1781, 1795), 'numpy.identity', 'np.identity', (['(4)'], {}), '(4)\n', (1792, 1795), True, 'import numpy as np\n'), ((1826, 1848), 'numpy.array', 'np.array', (['[0, 0, 0, 1]'], {}), '([0, 0, 0, 1])\n', (1834, 1848), True, 'import numpy as np\n'), ((1866, 1882), 'numpy.zeros', 'np.zeros', (['(3, 1)'], {}), '((3, 1))\n', (1874, 1882), True, 'import numpy as np\n'), ((1900, 1914), 'numpy.identity', 'np.identity', (['(3)'], {}), '(3)\n', (1911, 1914), True, 'import numpy as np\n'), ((3553, 3569), 'numpy.zeros', 'np.zeros', (['(3, 3)'], {}), '((3, 3))\n', (3561, 3569), True, 'import numpy as np\n'), ((5362, 5378), 'numpy.empty', 'np.empty', (['(8, 9)'], {}), '((8, 9))\n', (5370, 5378), True, 'import numpy as np\n'), ((5958, 5998), 'numpy.linalg.svd', 'np.linalg.svd', (['a_mat'], {'full_matrices': '(True)'}), '(a_mat, full_matrices=True)\n', (5971, 5998), True, 'import numpy as np\n'), ((6170, 6219), 'numpy.array', 'np.array', (['[[s[0], 0, 0], [0, s[1], 0], [0, 0, 0]]'], {}), '([[s[0], 0, 0], [0, s[1], 0], [0, 0, 0]])\n', (6178, 6219), True, 'import numpy as np\n'), ((6936, 6983), 'numpy.array', 'np.array', (['[feature_next[0], feature_next[1], 1]'], {}), '([feature_next[0], feature_next[1], 1])\n', (6944, 6983), True, 'import numpy as np\n'), ((7493, 7532), 'numpy.linalg.svd', 'np.linalg.svd', (['temp'], {'full_matrices': '(True)'}), '(temp, full_matrices=True)\n', (7506, 7532), True, 'import numpy as np\n'), ((7549, 7592), 'numpy.array', 'np.array', (['[[1, 0, 0], [0, 1, 0], [0, 0, 0]]'], {}), '([[1, 0, 0], [0, 1, 0], [0, 0, 0]])\n', (7557, 7592), True, 'import numpy as np\n'), ((8091, 8119), 'numpy.linalg.svd', 'np.linalg.svd', (['essential_mat'], {}), '(essential_mat)\n', (8104, 8119), True, 'import numpy as np\n'), ((8136, 8180), 'numpy.array', 'np.array', (['[[0, -1, 0], [1, 0, 0], [0, 0, 1]]'], {}), '([[0, -1, 0], [1, 0, 0], [0, 0, 1]])\n', (8144, 8180), True, 'import numpy as np\n'), ((11117, 11172), 'math.sqrt', 'sqrt', (['(rotation_mat[0][0] ** 2 + rotation_mat[1][0] ** 2)'], {}), '(rotation_mat[0][0] ** 2 + rotation_mat[1][0] ** 2)\n', (11121, 11172), False, 'from math import sqrt, atan2\n'), ((12002, 12103), 'numpy.array', 'np.array', (['[[0, -1, inlier_curr[1]], [1, 0, -inlier_curr[0]], [inlier_curr[1],\n inlier_curr[0], 0]]'], {}), '([[0, -1, inlier_curr[1]], [1, 0, -inlier_curr[0]], [inlier_curr[1],\n inlier_curr[0], 0]])\n', (12010, 12103), True, 'import numpy as np\n'), ((12168, 12269), 'numpy.array', 'np.array', (['[[0, -1, inlier_next[1]], [1, 0, -inlier_next[0]], [inlier_next[1],\n inlier_next[0], 0]]'], {}), '([[0, -1, inlier_next[1]], [1, 0, -inlier_next[0]], [inlier_next[1],\n inlier_next[0], 0]])\n', (12176, 12269), True, 'import numpy as np\n'), ((12418, 12443), 'numpy.vstack', 'np.vstack', (['(a_old, a_new)'], {}), '((a_old, a_new))\n', (12427, 12443), True, 'import numpy as np\n'), ((12462, 12482), 'numpy.linalg.svd', 'np.linalg.svd', (['a_mat'], {}), '(a_mat)\n', (12475, 12482), True, 'import numpy as np\n'), ((13129, 13177), 'numpy.column_stack', 'np.column_stack', (['(rotation_mat, translation_mat)'], {}), '((rotation_mat, translation_mat))\n', (13144, 13177), True, 'import numpy as np\n'), ((13264, 13299), 'numpy.vstack', 'np.vstack', (['(z, self.h_mat_last_row)'], {}), '((z, self.h_mat_last_row))\n', (13273, 13299), True, 'import numpy as np\n'), ((13700, 13790), 'math.sqrt', 'sqrt', (['((custom_pose[0] - opencv_pose[1]) ** 2 + (custom_pose[0] - opencv_pose[1]) **\n 2)'], {}), '((custom_pose[0] - opencv_pose[1]) ** 2 + (custom_pose[0] - opencv_pose\n [1]) ** 2)\n', (13704, 13790), False, 'from math import sqrt, atan2\n'), ((14161, 14207), 'numpy.matmul', 'np.matmul', (['self.k_mat', 'self.original_h[0:3, :]'], {}), '(self.k_mat, self.original_h[0:3, :])\n', (14170, 14207), True, 'import numpy as np\n'), ((14224, 14258), 'numpy.matmul', 'np.matmul', (['self.k_mat', 'camera_pose'], {}), '(self.k_mat, camera_pose)\n', (14233, 14258), True, 'import numpy as np\n'), ((14341, 14366), 'numpy.insert', 'np.insert', (['x_linear', '(3)', '(1)'], {}), '(x_linear, 3, 1)\n', (14350, 14366), True, 'import numpy as np\n'), ((14382, 14405), 'numpy.reshape', 'np.reshape', (['x_0', '(4, 1)'], {}), '(x_0, (4, 1))\n', (14392, 14405), True, 'import numpy as np\n'), ((15434, 15528), 'scipy.optimize.least_squares', 'least_squares', (['error', 'x_init'], {'args': '(self, x_linear, camera_pose, inlier_curr, inlier_next)'}), '(error, x_init, args=(self, x_linear, camera_pose, inlier_curr,\n inlier_next))\n', (15447, 15528), False, 'from scipy.optimize import least_squares\n'), ((5730, 5847), 'numpy.array', 'np.array', (['[x_next * x_curr, x_next * y_curr, x_next, y_next * x_curr, y_next * y_curr,\n y_next, x_curr, y_curr, 1]'], {}), '([x_next * x_curr, x_next * y_curr, x_next, y_next * x_curr, y_next *\n y_curr, y_next, x_curr, y_curr, 1])\n', (5738, 5847), True, 'import numpy as np\n'), ((6869, 6916), 'numpy.array', 'np.array', (['[feature_curr[0], feature_curr[1], 1]'], {}), '([feature_curr[0], feature_curr[1], 1])\n', (6877, 6916), True, 'import numpy as np\n'), ((7421, 7461), 'numpy.matmul', 'np.matmul', (['self.k_mat.T', 'fundamental_mat'], {}), '(self.k_mat.T, fundamental_mat)\n', (7430, 7461), True, 'import numpy as np\n'), ((7627, 7646), 'numpy.matmul', 'np.matmul', (['u', 'sigma'], {}), '(u, sigma)\n', (7636, 7646), True, 'import numpy as np\n'), ((11220, 11265), 'math.atan2', 'atan2', (['rotation_mat[2][1]', 'rotation_mat[2][2]'], {}), '(rotation_mat[2][1], rotation_mat[2][2])\n', (11225, 11265), False, 'from math import sqrt, atan2\n'), ((11282, 11313), 'math.atan2', 'atan2', (['(-rotation_mat[2][1])', 'psi'], {}), '(-rotation_mat[2][1], psi)\n', (11287, 11313), False, 'from math import sqrt, atan2\n'), ((11330, 11375), 'math.atan2', 'atan2', (['rotation_mat[1][0]', 'rotation_mat[0][0]'], {}), '(rotation_mat[1][0], rotation_mat[0][0])\n', (11335, 11375), False, 'from math import sqrt, atan2\n'), ((11406, 11452), 'math.atan2', 'atan2', (['(-rotation_mat[1][2])', 'rotation_mat[1][1]'], {}), '(-rotation_mat[1][2], rotation_mat[1][1])\n', (11411, 11452), False, 'from math import sqrt, atan2\n'), ((11469, 11500), 'math.atan2', 'atan2', (['(-rotation_mat[2][0])', 'psi'], {}), '(-rotation_mat[2][0], psi)\n', (11474, 11500), False, 'from math import sqrt, atan2\n'), ((8802, 8829), 'numpy.linalg.det', 'np.linalg.det', (['rotation_mat'], {}), '(rotation_mat)\n', (8815, 8829), True, 'import numpy as np\n'), ((10008, 10060), 'numpy.hstack', 'np.hstack', (['(rotation_matrices[i], camera_centers[i])'], {}), '((rotation_matrices[i], camera_centers[i]))\n', (10017, 10060), True, 'import numpy as np\n'), ((15320, 15338), 'numpy.shape', 'np.shape', (['x_linear'], {}), '(x_linear)\n', (15328, 15338), True, 'import numpy as np\n'), ((7061, 7095), 'numpy.matmul', 'np.matmul', (['fn_new', 'fundamental_mat'], {}), '(fn_new, fundamental_mat)\n', (7070, 7095), True, 'import numpy as np\n'), ((10316, 10368), 'numpy.squeeze', 'np.squeeze', (['(r_mat_row @ (temp_x - camera_centers[i]))'], {}), '(r_mat_row @ (temp_x - camera_centers[i]))\n', (10326, 10368), True, 'import numpy as np\n'), ((14300, 14318), 'numpy.shape', 'np.shape', (['x_linear'], {}), '(x_linear)\n', (14308, 14318), True, 'import numpy as np\n'), ((14438, 14460), 'numpy.dot', 'np.dot', (['p_old[0]', 'x_pt'], {}), '(p_old[0], x_pt)\n', (14444, 14460), True, 'import numpy as np\n'), ((14463, 14485), 'numpy.dot', 'np.dot', (['p_old[2]', 'x_pt'], {}), '(p_old[2], x_pt)\n', (14469, 14485), True, 'import numpy as np\n'), ((14530, 14552), 'numpy.dot', 'np.dot', (['p_old[1]', 'x_pt'], {}), '(p_old[1], x_pt)\n', (14536, 14552), True, 'import numpy as np\n'), ((14555, 14577), 'numpy.dot', 'np.dot', (['p_new[2]', 'x_pt'], {}), '(p_new[2], x_pt)\n', (14561, 14577), True, 'import numpy as np\n'), ((14617, 14639), 'numpy.dot', 'np.dot', (['p_new[0]', 'x_pt'], {}), '(p_new[0], x_pt)\n', (14623, 14639), True, 'import numpy as np\n'), ((14642, 14664), 'numpy.dot', 'np.dot', (['p_new[2]', 'x_pt'], {}), '(p_new[2], x_pt)\n', (14648, 14664), True, 'import numpy as np\n'), ((14709, 14731), 'numpy.dot', 'np.dot', (['p_new[1]', 'x_pt'], {}), '(p_new[1], x_pt)\n', (14715, 14731), True, 'import numpy as np\n'), ((14734, 14756), 'numpy.dot', 'np.dot', (['p_old[2]', 'x_pt'], {}), '(p_old[2], x_pt)\n', (14740, 14756), True, 'import numpy as np\n')] |
import time
import configparser
import numpy as np
import re
class PowerSupplyCalib(object):
def __init__(self, ps, vstart, vend, vstep, scan_speed=1.5):
self.vstart = vstart
self.vend = vend
self.vstep = vstep
self.scan_vals = np.arange(self.vstart, self.vend+(0.5*self.vstep), self.vstep)
self.ps = ps
self.scan_speed = scan_speed
self._scans = {}
self._volt_key = 'VoltCalib'
self._curr_key = 'CurrCalib'
self._key_regex = re.compile('address\.([0-9]{3})')
self._min_volt_addr = 59
self._max_volt_addr = 84
self._start_volt_val = 8
self._volt_per_addr = 8
self._min_curr_addr = 110
self._max_curr_addr = 142
self._start_curr_val = 0
self._curr_per_addr = 32
self._calib_types = {
self._volt_key: range(self._min_volt_addr, self._max_volt_addr, 1),
self._curr_key: range(self._min_curr_addr, self._max_curr_addr, 1),
}
self.prntstr = '{0:.1f}: {1:.2f}'
def scan(self, name=None, residual=False):
scan_result = []
if self.ps.voltage() < 0.01:
self.ps.on()
for val in self.scan_vals:
self.ps.voltage(val)
time.sleep(self.scan_speed)
if residual:
setp, _ = self.ps.setpoint()
result = self.ps.voltage() - setp
else:
result = self.ps.voltage()
scan_result.append(result)
if name is None:
return scan_result
else:
self._scans[name] = scan_result
def dump(self, filename):
config = configparser.ConfigParser()
for k, v in self._calib_types.items():
config.add_section(k)
for addr in v:
config[k]['address.{:0>3d}'.format(addr)] = self.address(addr)
with open(filename, 'w') as configfile:
config.write(configfile)
def load(self, filename):
config = configparser.ConfigParser()
config.read(filename)
for calib_type in self._calib_types.keys():
for addr_key, value in config[calib_type].items():
match = self._key_regex.match(addr_key)
if match:
self.address(int(match.group(1)), value)
@property
def volt_addr(self):
return self._calib_types[self._volt_key]
@property
def curr_addr(self):
return self._calib_types[self._curr_key]
def set_scan(self, vstart, vend, vstep, scan_speed=None):
self._scans = {}
self.vstart = vstart
self.vend = vend
self.vstep = vstep
self.scan_vals = np.arange(self.vstart, self.vend+(0.5*self.vstep), self.vstep)
if scan_speed is not None:
self.scan_speed = scan_speed
def set_scan_address(self, address):
voltages = self.get_voltages(address)
self.set_scan(voltages[0], voltages[-1], 0.1)
def get_scan(self, name, show=False):
if show and name in self._scans:
print('\n'.join(self.prntstr.format(setp, res) for setp, res in zip(self.scan_vals, self._scans[name])))
else:
return self._scans.get(name)
def residual(self, name1, name2, show=False):
avals = self.get_scan(name1)
bvals = self.get_scan(name2)
residuals = [a-b for a, b in zip(avals, bvals)]
if show:
print('\n'.join(self.prntstr.format(setp, res) for setp, res in zip(self.scan_vals, residuals)))
else:
return residuals
def address(self, addr, val=None):
if val is None:
return self.ps.cmd('GEEP', '{:0>3d}'.format(addr))
else:
self.ps.cmd('SEEP', '{addr:0>3d}{val}'.format(addr=addr, val=val))
def get_volt_address(self, voltage):
return (int(voltage*10) - self._start_volt_val) // self._volt_per_addr + self._min_volt_addr
def get_curr_address(self, current):
return (int(current*100) - self._start_curr_val) // self._curr_per_addr + self._min_curr_addr
def get_voltages(self, address):
voltint = self._volt_per_addr * (address - self._min_volt_addr) + self._start_volt_val
return [volt/10.0 for volt in range(voltint, voltint+self._volt_per_addr, 1)]
def get_currents(self, address):
currint = self._curr_per_addr * (address - self._min_curr_addr) + self._start_curr_val
return [curr/100.0 for curr in range(currint, currint+self._curr_per_addr, 1)]
| [
"time.sleep",
"configparser.ConfigParser",
"numpy.arange",
"re.compile"
] | [((268, 332), 'numpy.arange', 'np.arange', (['self.vstart', '(self.vend + 0.5 * self.vstep)', 'self.vstep'], {}), '(self.vstart, self.vend + 0.5 * self.vstep, self.vstep)\n', (277, 332), True, 'import numpy as np\n'), ((514, 548), 're.compile', 're.compile', (['"""address\\\\.([0-9]{3})"""'], {}), "('address\\\\.([0-9]{3})')\n", (524, 548), False, 'import re\n'), ((1680, 1707), 'configparser.ConfigParser', 'configparser.ConfigParser', ([], {}), '()\n', (1705, 1707), False, 'import configparser\n'), ((2028, 2055), 'configparser.ConfigParser', 'configparser.ConfigParser', ([], {}), '()\n', (2053, 2055), False, 'import configparser\n'), ((2716, 2780), 'numpy.arange', 'np.arange', (['self.vstart', '(self.vend + 0.5 * self.vstep)', 'self.vstep'], {}), '(self.vstart, self.vend + 0.5 * self.vstep, self.vstep)\n', (2725, 2780), True, 'import numpy as np\n'), ((1270, 1297), 'time.sleep', 'time.sleep', (['self.scan_speed'], {}), '(self.scan_speed)\n', (1280, 1297), False, 'import time\n')] |
import numpy as np
import pickle
import contrib_to_behavior
import matplotlib
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
from sklearn import svm
matplotlib.rcParams['pdf.fonttype'] = 42
matplotlib.rcParams['ps.fonttype'] = 42
plt.rcParams["font.family"] = "arial"
class neural_analysis:
def __init__(self, model_filename, ABBA=False, old_format = False):
x = pickle.load(open(model_filename, 'rb'))
self.ABBA = ABBA
# reshape STP depression
self.syn_x = np.stack(x['syn_x'],axis=2)
self.syn_x = np.stack(self.syn_x,axis=1)
if self.syn_x.shape[0] == 0:
self.syn_x = None
else:
num_neurons, trial_length, num_blocks, trials_per_block = self.syn_x.shape
self.syn_x = np.reshape(self.syn_x,(num_neurons,trial_length,num_blocks*trials_per_block))
# reshape STP facilitation
self.syn_u = np.stack(x['syn_u'],axis=2)
self.syn_u = np.stack(self.syn_u,axis=1)
if self.syn_u.shape[0] == 0:
self.syn_u = None
else:
num_neurons, trial_length, num_blocks, trials_per_block = self.syn_u.shape
self.syn_u = np.reshape(self.syn_u,(num_neurons,trial_length,num_blocks*trials_per_block))
# reshape RNN outputs
self.rnn_outputs = np.stack(x['hidden_state'],axis=2)
self.rnn_outputs = np.stack(self.rnn_outputs,axis=1)
num_neurons, trial_length, num_blocks, trials_per_block = self.rnn_outputs.shape
self.rnn_outputs = np.reshape(self.rnn_outputs,(num_neurons,trial_length,num_blocks*trials_per_block))
# reshape desired outputs
self.desired_outputs = x['desired_output']
if old_format:
self.desired_outputs = np.transpose(self.desired_outputs,(2,0,1))
# reshape train_mask
self.train_mask = x['train_mask']
self.train_mask = np.transpose(self.train_mask,(0,1))
# reshape RNN inputs
self.rnn_inputs = x['rnn_input']
self.rnn_inputs = np.transpose(self.rnn_inputs,(2,0,1))
# reshape model outputs
self.model_outputs = np.stack(x['model_outputs'],axis=2)
self.model_outputs = np.stack(self.model_outputs,axis=1)
num_classes = self.model_outputs.shape[0]
self.model_outputs = np.reshape(self.model_outputs,(num_classes,trial_length,num_blocks*trials_per_block))
"""
rnn_inputs, desired_outputs, rnn_outputs, model_outputs
should be of shape neurons X time X trials
print(self.rnn_inputs.shape, self.desired_outputs.shape,self.rnn_outputs.shape,self.model_outputs.shape, self.train_mask.shape)
"""
# reshape trial_conds
self.sample_dir = x['sample_dir']
self.test_dir = x['test_dir']
self.match = x['match']
self.rule = x['rule']
self.catch = x['catch']
self.probe = x['probe']
# for the ABBA trials
if self.ABBA:
self.num_test_stim = x['num_test_stim']
self.repeat_test_stim = x['repeat_test_stim']
self.ABBA_delay = x['params']['ABBA_delay']
# other info
#self.EI_list = x['params']['EI_list']
self.num_rules = len(x['params']['possible_rules'])
self.possible_rules = x['params']['possible_rules']
self.num_motion_dirs = x['params']['num_motion_dirs']
self.U = x['U']
self.W_rnn = x['w_rnn']
self.b_rnn = x['b_rnn']
self.W_in = x['w_in']
self.EI_list = x['params']['EI_list']
self.dead_time = x['params']['dead_time']
self.fix_time = x['params']['fix_time']
self.delta_t = x['params']['dt']
if self.ABBA:
self.max_num_tests = x['params']['max_num_tests']
self.ABBA_accuracy_match, self.ABBA_accuracy_non_match = self.performance_ABBA()
else:
pass
#accuracy = self.performance()
#print(accuracy)
def calc_native_tuning(self):
rule = 0
sample_rng = range(8+20,8+20+20)
#sample_rng = range(88,108)
num_dirs = self.num_motion_dirs
num_input_neurons, trial_length, num_trials = self.rnn_inputs.shape
mean_input_resp = np.zeros((num_input_neurons, num_dirs))
num_rnn_neurons = self.rnn_outputs.shape[0]
native_tuning = np.zeros((num_rnn_neurons, num_dirs))
for d in range(num_dirs):
ind = np.where((self.rule == self.possible_rules[rule])*(self.sample_dir==d))
#ind = np.where((self.rule == self.possible_rules[rule])*(self.test_dir==d))
s = np.mean(self.rnn_inputs[:,:,ind[0]],axis=2)
mean_input_resp[:,d] = np.mean(s[:,sample_rng],axis=1)
native_tuning = np.dot(self.W_in, mean_input_resp)
return native_tuning
def motion_tuning(self):
num_neurons, trial_length, num_trials = self.rnn_outputs.shape
sample_pd = np.zeros((num_neurons, trial_length))
sample_pev = np.zeros((num_neurons, trial_length))
sample_amp = np.zeros((num_neurons, trial_length))
test_pd = np.zeros((num_neurons, 2, trial_length))
test_pev = np.zeros((num_neurons, 2, trial_length))
test_amp = np.zeros((num_neurons, 2, trial_length))
sample_dir = np.ones((num_trials, 3))
sample_dir[:,1] = np.cos(2*np.pi*self.sample_dir/self.num_motion_dirs)
sample_dir[:,2] = np.sin(2*np.pi*self.sample_dir/self.num_motion_dirs)
test_dir = np.ones((num_trials, 3))
test_dir[:,1] = np.cos(2*np.pi*self.test_dir/self.num_motion_dirs)
test_dir[:,2] = np.sin(2*np.pi*self.test_dir/self.num_motion_dirs)
for n in range(num_neurons):
for t in range(trial_length):
h = np.linalg.lstsq(sample_dir, self.rnn_outputs[n,t,:])
pred_err = self.rnn_outputs[n,t,:] - np.dot(h[0], sample_dir.T)
mse = np.mean(pred_err**2)
response_var = np.var(self.rnn_outputs[n,t,:])
sample_pev[n,t] = 1 - (mse)/(response_var+1e-9)
sample_pd[n,t] = np.arctan2(h[0][2],h[0][1])
sample_amp[n,t] = np.sqrt(h[0][0]**2+h[0][1]**2)
for m in range(2):
ind = np.where(self.match==m)[0]
h = np.linalg.lstsq(test_dir[ind], self.rnn_outputs[n,t,ind])
pred_err = self.rnn_outputs[n,t,ind] - np.dot(h[0], test_dir[ind].T)
mse = np.mean(pred_err**2)
response_var = np.var(self.rnn_outputs[n,t,ind])
test_pev[n,m,t] = 1 - (mse)/(response_var+1e-9)
test_pd[n,m,t] = np.arctan2(h[0][2],h[0][1])
test_amp[n,m,t] = np.sqrt(h[0][0]**2+h[0][1]**2)
return sample_pd, sample_pev, sample_amp, test_pd, test_pev, test_amp
def recreate_effective_weight_matrix(self, EI = False):
rule = 0
num_neurons, trial_length, num_trials = self.syn_u.shape
W = np.zeros((num_neurons,num_neurons,self.num_motion_dirs, trial_length))
mean_efficacy = np.zeros((num_neurons,self.num_motion_dirs, trial_length))
for d in range(self.num_motion_dirs):
ind = np.where((self.rule == self.possible_rules[rule])*(self.sample_dir==d)*(self.match==1))[0]
mean_efficacy[:,d,:] = np.mean(self.syn_u[:,:,ind]*self.syn_x[:,:,ind],axis=2)
if EI:
ei_diag = np.diag(self.EI_list)
W_rnn = np.dot(np.maximum(0,self.W_rnn), ei_diag)
else:
W_rnn = self.W_rnn
for n1 in range(num_neurons):
for n2 in range(num_neurons):
for d in range(self.num_motion_dirs):
W[n1,n2,d,:] = mean_efficacy[n2,d,:]*W_rnn[n1,n2]
return W
def recreate_output_current(self, EI = False):
rule = 0
num_neurons, trial_length, num_trials = self.syn_u.shape
out_current = np.zeros((num_neurons,self.num_motion_dirs, self.num_motion_dirs, trial_length))
out_current = np.zeros((num_neurons,self.num_motion_dirs, self.num_motion_dirs, trial_length))
for s in range(self.num_motion_dirs):
for t in range(self.num_motion_dirs):
ind = np.where((self.rule == self.possible_rules[rule])*(self.sample_dir==s)*(self.test_dir==t))[0]
out_current[:,s,t,:] = np.mean(self.syn_u[:,:,ind]*self.syn_x[:,:,ind]*self.rnn_outputs[:,:,ind],axis=2)
"""
if EI:
ei_diag = np.diag(self.EI_list)
W_rnn = np.dot(np.maximum(0,self.W_rnn), ei_diag)
else:
W_rnn = self.W_rnn
for n1 in range(num_neurons):
for n2 in range(num_neurons):
for s in range(self.num_motion_dirs):
for t in range(self.num_motion_dirs):
out_current[n1,n2,s,t,:] = post_syn[n2,s,t,:]*W_rnn[n1,n2]
"""
return out_current
def performance(self):
n = 18 # number of time steps to measure during test, this will be the basis of performance
time_correct = np.zeros((self.num_rules, self.num_motion_dirs, 2))
count = np.zeros((self.num_rules, self.num_motion_dirs, 2))
for i in range(len(self.sample_dir)):
if self.catch[i]==0:
s = np.int_(self.sample_dir[i])
m = np.int_(self.match[i])
r = np.int_(np.where(self.rule[i]==self.possible_rules)[0])
count[r,s,m] +=1
if m==1:
score=np.mean((self.model_outputs[2,-n:,i]>self.model_outputs[1,-n:,i])*(self.model_outputs[2,-n:,i]>self.model_outputs[0,-n:,i]))
else:
score=np.mean((self.model_outputs[1,-n:,i]>self.model_outputs[2,-n:,i])*(self.model_outputs[1,-n:,i]>self.model_outputs[0,-n:,i]))
time_correct[r,s,m] += score
return time_correct/count
def performance_ABBA(self):
ABBA_delay = self.ABBA_delay//self.delta_t
eof = (self.dead_time+self.fix_time)//self.delta_t
eos = eof + ABBA_delay
# performance is measured with and without a repeated distractor
time_correct_match = np.zeros((self.max_num_tests))
time_correct_non_match = np.zeros((self.max_num_tests))
time_match = np.zeros((self.max_num_tests))
time_non_match = np.zeros((self.max_num_tests))
for i in range(len(self.sample_dir)):
for j in range(self.num_test_stim[i]):
# will discard the first time point of each test stim
test_rng = range(1+eos+(2*j+1)*ABBA_delay, eos+(2*j+2)*ABBA_delay)
matching_stim = self.match[i]==1 and j==self.num_test_stim[i]-1
if matching_stim:
time_match[j] += ABBA_delay-1 # -1 because we're discarding the first time point of each test stim
time_correct_match[j] += np.sum((self.model_outputs[2,test_rng,i]>self.model_outputs[1,test_rng,i])*(self.model_outputs[2,test_rng,i]>self.model_outputs[0,test_rng,i]))
else:
time_non_match[j] += ABBA_delay-1
time_correct_non_match[j] += np.sum((self.model_outputs[1,test_rng,i]>self.model_outputs[2,test_rng,i])*(self.model_outputs[1,test_rng,i]>self.model_outputs[0,test_rng,i]))
auccracy_match = time_correct_match/time_match
auccracy_non_match = time_correct_non_match/time_non_match
print('Accuracy')
print(time_correct_non_match, time_non_match)
print(time_correct_match, time_match)
return auccracy_match, auccracy_non_match
def show_results(self):
print(self.results)
def plot_example_neurons(self, example_numbers):
mean_resp = calc_mean_responses(self)
1/0
f = plt.figure(figsize=(12,8))
ax = f.add_subplot(1, 3, 1)
ax.imshow(trial_info['sample_direction'],interpolation='none',aspect='auto')
ax = f.add_subplot(1, 3, 2)
ax.imshow(trial_info['test_direction'],interpolation='none',aspect='auto')
ax = f.add_subplot(1, 3, 3)
ax.imshow(trial_info['match'],interpolation='none',aspect='auto')
plt.show()
1/0
def calculate_svms(self, num_reps = 3, DMC = [False], decode_test = False):
lin_clf = svm.SVC(C=1,kernel='linear',decision_function_shape='ovr', shrinking=False, tol=1e-4)
num_neurons, trial_length, num_trials = self.rnn_outputs.shape
spike_decoding = np.zeros((trial_length,self.num_rules,num_reps))
synapse_decoding = np.zeros((trial_length,self.num_rules,num_reps))
spike_decoding_test = np.zeros((trial_length,self.num_rules,num_reps))
synapse_decoding_test = np.zeros((trial_length,self.num_rules,num_reps))
N = self.num_motion_dirs
sample_cat = np.floor(self.sample_dir/(self.num_motion_dirs/2)*np.ones_like(self.sample_dir))
if self.ABBA:
test_dir = self.test_dir[:,0]
else:
test_dir = self.test_dir
test_cat = np.floor(test_dir/(self.num_motion_dirs/2)*np.ones_like(test_dir))
for r in range(self.num_rules):
if self.ABBA:
ind = np.where((self.num_test_stim>=4))[0]
else:
ind = np.where((self.rule==self.possible_rules[r]))[0]
for t in range(trial_length):
if DMC[r]:
spike_decoding[t,r,:] = self.calc_svm_equal_trials(lin_clf,self.rnn_outputs[:,t,ind].T, sample_cat[ind],num_reps,2)
if decode_test:
spike_decoding_test[t,r,:] = self.calc_svm_equal_trials(lin_clf,self.rnn_outputs[:,t,ind].T, test_cat[ind],num_reps,2)
else:
spike_decoding[t,r,:] = self.calc_svm_equal_trials(lin_clf,self.rnn_outputs[:,t,ind].T, self.sample_dir[ind],num_reps,N)
if decode_test:
spike_decoding_test[t,r,:] = self.calc_svm_equal_trials(lin_clf,self.rnn_outputs[:,t,ind].T, test_dir[ind],num_reps,N)
if self.syn_x is not None:
effective_current = self.syn_x[:,t,ind].T*self.syn_u[:,t,ind].T
if DMC[r]:
synapse_decoding[t,r,:] = self.calc_svm_equal_trials(lin_clf,effective_current, sample_cat[ind],num_reps,2)
if decode_test:
synapse_decoding_test[t,r,:] = self.calc_svm_equal_trials(lin_clf,effective_current, test_cat[ind],num_reps,2)
else:
synapse_decoding[t,r,:] = self.calc_svm_equal_trials(lin_clf,effective_current, self.sample_dir[ind],num_reps,N)
if decode_test:
synapse_decoding_test[t,r,:] = self.calc_svm_equal_trials(lin_clf,effective_current, test_dir[ind],num_reps,N)
return spike_decoding, synapse_decoding, spike_decoding_test, synapse_decoding_test
def calculate_autocorr(self, time_start, time_end):
num_neurons, trial_length, num_trials = self.rnn_outputs.shape
num_lags = time_end-time_start
spike_autocorr = np.zeros((num_neurons, num_lags))
syn_x_autocorr = np.zeros((num_neurons, num_lags))
syn_adapt_autocorr = np.zeros((num_neurons, num_lags))
for n in range(num_neurons):
count = np.zeros((num_lags))
for i in range(time_start, time_end):
for j in range(time_start, time_end):
lag = np.abs(i-j)
for s in range(4):
ind = np.where(self.sample_dir==s)
ind = np.where(self.match==1)
ind = ind[0]
count[lag] += 1
r1 = np.corrcoef(self.rnn_outputs[n,i,ind], self.rnn_outputs[n,j,ind])
spike_autocorr[n, lag] += r1[0,1]
if self.syn_x is not None:
r1 = np.corrcoef(self.syn_x[n,i,ind], self.syn_x[n,j,ind])
syn_x_autocorr[n, lag] += r1[0,1]
if self.sa is not None:
r1 = np.corrcoef(self.sa[n,i,ind], self.sa[n,j,ind])
syn_adapt_autocorr[n, lag] += r1[0,1]
spike_autocorr[n,:] /= count
syn_x_autocorr[n,:] /= count
syn_adapt_autocorr[n,:] /= count
return spike_autocorr,syn_x_autocorr,syn_adapt_autocorr
def calc_mean_responses(self):
num_rules = self.num_rules
num_dirs = self.num_motion_dirs
num_neurons, trial_length, num_trials = self.rnn_outputs.shape
num_classes = self.model_outputs.shape[0]
mean_resp = np.zeros((num_neurons, num_rules, num_dirs, trial_length))
mean_out_match = np.zeros((num_classes, num_rules, trial_length))
mean_out_non_match = np.zeros((num_classes, num_rules, trial_length))
for n in range(num_neurons):
for r in range(num_rules):
for d in range(num_dirs):
if self.ABBA:
ind = np.where((self.num_test_stim>=4)*(self.sample_dir==d))[0]
else:
ind = np.where((self.rule == self.possible_rules[r])*(self.sample_dir==d))[0]
mean_resp[n,r,d,:] = np.mean(self.rnn_outputs[n,:,ind],axis=0)
for n in range(num_classes):
for r in range(num_rules):
ind_match = np.where((self.rule == self.possible_rules[r])*(self.match==1)*(self.catch==0))
ind_non_match = np.where((self.rule == self.possible_rules[r])*(self.match==0)*(self.catch==0))
mean_out_match[n,r,:] = np.mean(self.model_outputs[n,:,ind_match[0]],axis=0)
mean_out_non_match[n,r,:] = np.mean(self.model_outputs[n,:,ind_non_match[0]],axis=0)
return mean_resp, mean_out_match, mean_out_non_match
def decoding_accuracy_postle(self, num_reps = 10):
lin_clf = svm.SVC(C=1,kernel='linear',decision_function_shape='ovr', shrinking=False, tol=1e-5)
num_neurons, trial_length, num_trials = self.rnn_outputs.shape
sample_pev = np.zeros((num_neurons, 2,2,2,2,trial_length))
sample_stp_pev = np.zeros((num_neurons, 2,2,2,2,trial_length))
sample_decoding = np.zeros((2,2,2,2,trial_length,num_reps))
sample_stp_decoding = np.zeros((2,2,2,2,trial_length,num_reps))
model_output = np.zeros((2,2,3,trial_length))
# r1 and r2 refer to the first and second rule (attention) cue
# m refers to the modality
# p refers to the presence or absence of a probe
for m1 in range(2):
for m2 in range(2):
ind = np.where((self.match[:,0] == m1)*(self.match[:,1] == m2)*(self.probe[:,1]==0))[0]
model_output[m1,m2,:,:] = np.mean(self.model_outputs[:,:,ind],axis=2)
for r1 in range(2):
for r2 in range(2):
for p in range(2):
ind = np.where((self.rule[:,0] == r1)*(self.rule[:,1] == r2)*(self.probe[:,1]==p))[0]
#ind = np.where((self.rule[:,0] == r1)*(self.rule[:,1] == r2)*(self.probe[:,1]>=0))[0]
for m in range(2):
for t in range(trial_length):
for n in range(num_neurons):
sample_pev[n,r1,r2,p,m,t] = self.calc_pev(self.rnn_outputs[n,t,ind], self.sample_dir[ind,m])
sample_decoding[r1,r2,p,m,t,:] = self.calc_svm_equal_trials(lin_clf,self.rnn_outputs[:,t,ind].T, self.sample_dir[ind,m],num_reps, self.num_motion_dirs)
if self.syn_x is not None:
for n in range(num_neurons):
effective_current = self.syn_x[n,t,ind]*self.syn_u[n,t,ind]
sample_stp_pev[n,r1,r2,p,m,t] = self.calc_pev(effective_current, self.sample_dir[ind,m])
effective_current = self.syn_x[:,t,ind]*self.syn_u[:,t,ind]
sample_stp_decoding[r1,r2,p,m,t,:] = self.calc_svm_equal_trials(lin_clf,effective_current.T, self.sample_dir[ind,m],num_reps, self.num_motion_dirs)
return sample_pev, sample_stp_pev, sample_decoding, sample_stp_decoding, model_output
@staticmethod
def calc_svm_equal_trials(lin_clf, y, conds, num_reps, num_conds):
# normalize values between 0 and 1
for i in range(y.shape[1]):
m1 = y[:,i].min()
m2 = y[:,i].max()
y[:,i] -= m1
if m2>m1:
y[:,i] /=(m2-m1)
"""
Want to ensure that all conditions have the same number of trials
Will find the min number of trials per conditions, and remove trials above the min number
"""
num_trials = np.zeros((num_conds))
for i in range(num_conds):
num_trials[i] = np.sum(conds==i)
min_num_trials = int(np.min(num_trials))
conds_equal = np.zeros((min_num_trials*num_conds))
y_equal = np.zeros((min_num_trials*num_conds, y.shape[1]))
for i in range(num_conds):
ind = np.where(conds==i)[0]
ind = ind[:min_num_trials]
conds_equal[i*min_num_trials:(i+1)*min_num_trials] = i
y_equal[i*min_num_trials:(i+1)*min_num_trials, :] = y[ind,:]
train_pct = 0.75
score = np.zeros((num_reps))
for r in range(num_reps):
q = np.random.permutation(len(conds_equal))
i = np.int_(np.round(len(conds_equal)*train_pct))
train_ind = q[:i]
test_ind = q[i:]
lin_clf.fit(y_equal[train_ind,:], conds_equal[train_ind])
#dec = lin_clf.decision_function(y[test_ind,:])
dec = lin_clf.predict(y_equal[test_ind,:])
for i in range(len(test_ind)):
if conds_equal[test_ind[i]]==dec[i]:
score[r] += 1/len(test_ind)
return score
@staticmethod
def calc_svm(lin_clf, y, conds, num_reps):
num_conds = len(np.unique(conds))
y = np.squeeze(y).T
# normalize values between 0 and 1
for i in range(y.shape[1]):
m1 = y[:,i].min()
m2 = y[:,i].max()
y[:,i] -= m1
if m2>m1:
y[:,i] /=(m2-m1)
train_pct = 0.75
score = np.zeros((num_reps))
for r in range(num_reps):
q = np.random.permutation(len(conds))
i = np.int_(np.round(len(conds)*train_pct))
train_ind = q[:i]
test_ind = q[i:]
lin_clf.fit(y[train_ind,:], conds[train_ind])
dec = lin_clf.decision_function(y[test_ind,:])
if num_conds>2:
dec = np.argmax(dec, 1)
else:
dec = np.int_(np.sign(dec)*0.5+0.5)
for i in range(len(test_ind)):
if conds[test_ind[i]]==dec[i]:
score[r] += 1/len(test_ind)
return score
def calculate_pevs(self):
num_neurons, trial_length, num_trials = self.rnn_outputs.shape
sample_pev = np.zeros((num_neurons, self.num_rules,trial_length))
test_pev = np.zeros((num_neurons, self.num_rules,trial_length))
rule_pev = np.zeros((num_neurons,trial_length))
match_pev = np.zeros((num_neurons, self.num_rules,trial_length))
sample_stp_pev = np.zeros((num_neurons, self.num_rules,trial_length))
sample_cat_pev = np.zeros((num_neurons, self.num_rules,trial_length))
sample_cat_stp_pev = np.zeros((num_neurons, self.num_rules,trial_length))
test_stp_pev = np.zeros((num_neurons, self.num_rules,trial_length))
for r in range(self.num_rules):
if self.ABBA:
ind = np.where((self.num_test_stim>=4))[0]
else:
ind = np.where((self.rule == self.possible_rules[r]))[0]
ind_test = np.where((self.rule == self.possible_rules[r])*(self.match == 0))[0]
for n in range(num_neurons):
for t in range(trial_length):
sample_pev[n,r,t] = self.calc_pev(self.rnn_outputs[n,t,ind], self.sample_dir[ind])
sample_cat_pev[n,r,t] = self.calc_pev(self.rnn_outputs[n,t,ind], np.floor(self.sample_dir[ind]/(self.num_motion_dirs/2)))
if not self.ABBA:
test_pev[n,r,t] = self.calc_pev(self.rnn_outputs[n,t,ind_test], self.test_dir[ind_test])
rule_pev[n,t] = self.calc_pev(self.rnn_outputs[n,t,:], self.rule)
match_pev[n,r,t] = self.calc_pev(self.rnn_outputs[n,t,ind], self.match[ind])
if self.syn_x is not None:
effective_current = self.syn_x[n,t,ind]*self.syn_u[n,t,ind]
sample_stp_pev[n,r,t] = self.calc_pev(effective_current, self.sample_dir[ind])
if not self.ABBA:
test_stp_pev[n,r,t] = self.calc_pev(effective_current, self.test_dir[ind_test])
sample_cat_stp_pev[n,r,t] = self.calc_pev(effective_current, np.floor(self.sample_dir[ind]/(self.num_motion_dirs/2)))
return sample_pev, test_pev, rule_pev, match_pev, sample_stp_pev, sample_cat_pev, sample_cat_stp_pev, test_stp_pev
@staticmethod
def calc_pev(x, conds):
unique_conds = np.unique(conds)
m = len(unique_conds)
lx = len(x)
xr = x - np.mean(x)
xm = np.zeros((1,m))
countx = np.zeros((1,m))
for (j,i) in enumerate(unique_conds):
ind = np.where(conds==i)
countx[0,j] = len(ind[0])
xm[0,j] = np.mean(xr[ind[0]])
gm = np.mean(xr)
df1 = np.sum(countx>0)-1
df2 = lx - df1 - 1
xc = xm - gm
ix = np.where(countx==0)
xc[ix] = 0
RSS = np.dot(countx, np.transpose(xc**2))
#TSS = (xr - gm)**2
TSS = np.dot(np.transpose(xr - gm),xr - gm)
#print(TSS.shape)
SSE = TSS - RSS
if df2 > 0:
mse = SSE/df2
else:
mse = np.NaN
F = (RSS/df1)/mse
"""
Table = np.zeros((3,5))
Table[:,0] = [RSS,SSE,TSS]
Table[:,1] = [df1,df2,df1+df2]
Table[:,2] = [RSS/df1,mse,999];
Table[:,3] = [F,999,999]
"""
SS_groups = RSS;
SS_total = TSS;
df_groups = df1;
MS_error = mse;
pev = (SS_groups-df_groups*MS_error)/(SS_total+MS_error)
if np.isnan(pev):
pev = 0
return pev
def plot_all_figures(self, rule,dt=25, STP=False, DMC = [False], f=None, start_sp=0, num_rows=3, tight=False, two_rules = False, decode_test = False):
font = {'family' : 'normal',
'weight' : 'bold',
'size' : 12}
mean_resp, mean_out_match, mean_out_non_match = self.calc_mean_responses()
spike_decode, synapse_decode, spike_decode_test, synapse_decode_test = self.calculate_svms(DMC=DMC,decode_test=decode_test)
sample_pev, test_pev, rule_pev, _, sample_stp_pev, sample_cat_pev, sample_cat_stp_pev, test_stp_pev = self.calculate_pevs()
if DMC[0]:
sample_pev = sample_cat_pev
sample_stp_pev = sample_cat_stp_pev
chance_level = 1/2
else:
chance_level = 1/8
if two_rules:
num_cols = 4
else:
num_cols = 3
# find good example neuron
mean_pev = np.mean(sample_pev[:, rule, 30:],axis=1)
ind = np.argsort(mean_pev)
example_neuron = ind[-1]
trial_length_steps = sample_pev.shape[2]
trial_length = np.int_(trial_length_steps*dt)
t = np.arange(0,trial_length,dt)
t -= 900 # assuming 400 ms dead time, 500 ms fixation
if self.ABBA:
t0,t1,t2,t3 = np.where(t==-500), np.where(t==0),np.where(t==500),np.where(t==1500)
else:
t0,t1,t2,t3 = np.where(t==-500), np.where(t==0),np.where(t==500),np.where(t==1500)
if f is None:
f = plt.figure(figsize=(8,2*num_rows))
ax = f.add_subplot(num_rows, num_cols, start_sp+1)
if self.ABBA:
# plot accuracy bar plot instead
x=np.array([0,1,2,3])
ax.bar(x+0.1, self.ABBA_accuracy_match,width=0.2,color='r',align='center')
ax.bar(x-0.1, self.ABBA_accuracy_non_match,width=0.2,color='b',align='center')
ax.set_title('Accuracy')
ax.set_ylabel('Fraction correct')
ax.set_xlabel('Num. of distractors')
else:
ax.hold(True)
if two_rules:
ax.plot(t, mean_out_match[0,0,:] ,'k',linewidth=2,label='Fixation')
ax.plot(t, mean_out_match[1,0,:] ,'m',linewidth=2,label='Non-match')
ax.plot(t, mean_out_match[2,0,:] ,'g',linewidth=2,label='Match')
ax.plot(t, mean_out_match[0,1,:] ,'k--',linewidth=2,label='Fixation')
ax.plot(t, mean_out_match[1,1,:] ,'m--',linewidth=2,label='Non-match')
ax.plot(t, mean_out_match[2,1,:] ,'g--',linewidth=2,label='Match')
else:
ax.plot(t, mean_out_match[0,rule,:] ,'k',linewidth=2,label='Fixation')
ax.plot(t, mean_out_match[1,rule,:] ,'m',linewidth=2,label='Non-match')
ax.plot(t, mean_out_match[2,rule,:] ,'g',linewidth=2,label='Match')
#plt.legend(loc=3)
self.add_subplot_fixings(ax)
ax.set_title('Network output - match trials')
if self.ABBA:
pass
else:
ax = f.add_subplot(num_rows, num_cols, start_sp+2)
ax.hold(True)
if two_rules:
ax.plot(t, mean_out_non_match[0,0,:] ,'k',linewidth=2,label='Fixation')
ax.plot(t, mean_out_non_match[1,0,:] ,'m',linewidth=2,label='Non-match')
ax.plot(t, mean_out_non_match[2,0,:] ,'g',linewidth=2,label='Match')
ax.plot(t, mean_out_non_match[0,1,:] ,'k--',linewidth=2,label='Fixation')
ax.plot(t, mean_out_non_match[1,1,:] ,'m--',linewidth=2,label='Non-match')
ax.plot(t, mean_out_non_match[2,1,:] ,'g--',linewidth=2,label='Match')
else:
ax.plot(t, mean_out_non_match[0,rule,:] ,'k',linewidth=2)
ax.plot(t, mean_out_non_match[1,rule,:] ,'m',linewidth=2)
ax.plot(t, mean_out_non_match[2,rule,:] ,'g',linewidth=2)
self.add_subplot_fixings(ax)
ax.set_title('Network output - non-match trials')
ax = f.add_subplot(num_rows, num_cols, start_sp+3)
ax.hold(True)
# if plotting the result of the delayed rule task, show rule PEV instead of example neuron
if two_rules:
max_val = np.max(rule_pev)
ax.plot(t,np.mean(rule_pev, axis=0), linewidth=2)
self.add_subplot_fixings(ax,chance_level=0,ylim=0.2)
ax.set_title('Rule selectivity')
ax.set_ylabel('Normalized PEV')
else:
"""
max_val = np.max(mean_resp[example_neuron,rule,:,:])
print(max_val)
for i in range(8):
ax.plot(t,mean_resp[example_neuron,rule,i,:],color=[1-i/7,0,i/7], linewidth=1)
self.add_subplot_fixings(ax,chance_level=0,ylim=max_val*1.05)
ax.set_title('Example neuron')
ax.set_ylabel('Activity (a.u.)')
# plot the mean population response from those neurons whose synapses are informative of sample
"""
#syn_pev = np.mean(sample_stp_pev[:,0,t2[0]:t3[0]], axis=1)
#ind_syn = np.where(syn_pev > 0.1)[0]
#print('Informative synapses ', ind_syn)
s = np.mean(mean_resp[:,rule,:,:],axis=0)
max_val = np.max(s)
for i in range(8):
ax.plot(t,s[i,:],color=[1-i/7,0,i/7], linewidth=1)
self.add_subplot_fixings(ax,chance_level=0,ylim=0.5)
ax.set_title('Mean response from synpases informative neurons')
ax.set_ylabel('Activity (a.u.)')
ax.set_ylim([0, 0.5])
if two_rules:
ax = f.add_subplot(num_rows, num_cols, start_sp+5)
im = ax.imshow(sample_pev[:,0,:],aspect='auto',interpolation=None)
f.colorbar(im,orientation='vertical')
ax.spines['right'].set_visible(False)
ax.set_ylabel('Neuron number')
ax.set_xlabel('Time relative to sample onset (ms)')
ax.spines['top'].set_visible(False)
ax.set_xticks([t0[0], t1[0], t2[0], t3[0]])
ax.set_xticklabels([-500,0,500,1500])
ax.set_title('Neuronal sample \nselectvity - DMS task')
ax = f.add_subplot(num_rows, num_cols, start_sp+6)
im = ax.imshow(sample_pev[:,1,:],aspect='auto',interpolation=None)
f.colorbar(im,orientation='vertical')
ax.spines['right'].set_visible(False)
ax.set_ylabel('Neuron number')
ax.set_xlabel('Time relative to sample onset (ms)')
ax.spines['top'].set_visible(False)
ax.set_xticks([t0[0], t1[0], t2[0], t3[0]])
ax.set_xticklabels([-500,0,500,1500])
ax.set_title('Neuronal sample \nselectvity - DMrS task')
else:
ax = f.add_subplot(num_rows, 3, start_sp+4)
im = ax.imshow(sample_pev[:,rule,:],aspect='auto',interpolation=None)
f.colorbar(im,orientation='vertical')
ax.spines['right'].set_visible(False)
ax.set_ylabel('Neuron number')
ax.set_xlabel('Time relative to sample onset (ms)')
ax.spines['top'].set_visible(False)
if DMC:
ax.set_title('Neuronal sample \ncategory selectvity')
else:
ax.set_title('Neuronal sample selectvity')
if self.ABBA:
ax.set_xticks([t0[0], t1[0], t2[0], t3[0]])
ax.set_xticklabels([-500,0,500,1500])
else:
ax.set_xticks([t0[0], t1[0], t2[0], t3[0]])
ax.set_xticklabels([-500,0,500,1500])
if two_rules:
ax = f.add_subplot(num_rows, num_cols, start_sp+7)
plt.hold(True)
u = np.mean(sample_pev[:,0,:],axis=0)
se = np.std(sample_pev[:,0,:],axis=0)/np.sqrt(sample_pev.shape[0])
ax.plot(t,u,'g')
sample_max1 = np.max(u)
ax.fill_between(t,u-se,u+se,facecolor=(0,1,0,0.5))
u = np.mean(sample_pev[:,1,:],axis=0)
se = np.std(sample_pev[:,1,:],axis=0)/np.sqrt(sample_pev.shape[0])
ax.plot(t,u,'m')
sample_max = np.max(u)
sample_max = np.max([sample_max, sample_max1])
ax.fill_between(t,u-se,u+se,facecolor=(1,0,1,0.5))
else:
ax = f.add_subplot(num_rows, num_cols, start_sp+5)
u = np.mean(sample_pev[:,rule,:],axis=0)
se = np.std(sample_pev[:,rule,:],axis=0)/np.sqrt(sample_pev.shape[0])
ax.plot(t,u,'k')
sample_max = np.max(u)
ax.fill_between(t,u-se,u+se,facecolor=(0,0,0,0.5))
self.add_subplot_fixings(ax,chance_level=0,ylim=sample_max*2)
if DMC:
ax.set_title('Neuronal sample \ncategory selectivity')
else:
ax.set_title('Neuronal sample selectivity')
ax.set_ylabel('Normalized PEV')
if two_rules:
ax = f.add_subplot(num_rows, num_cols, start_sp+8)
u = np.mean(spike_decode[:,0,:],axis=1)
se = np.std(spike_decode[:,0,:],axis=1)
ax.plot(t,u,'g')
ax.fill_between(t,u-se,u+se,facecolor=(0,1,0,0.5))
u = np.mean(spike_decode[:,1,:],axis=1)
se = np.std(spike_decode[:,1,:],axis=1)
ax.plot(t,u,'m')
ax.fill_between(t,u-se,u+se,facecolor=(1,0,1,0.5))
self.add_subplot_fixings(ax, chance_level=chance_level)
else:
ax = f.add_subplot(num_rows, num_cols, start_sp+6)
u = np.mean(spike_decode[:,rule,:],axis=1)
se = np.std(spike_decode[:,rule,:],axis=1)
ax.plot(t,u,'k')
ax.fill_between(t,u-se,u+se,facecolor=(0,0,0,0.5))
u = np.mean(spike_decode_test[:,rule,:],axis=1)
se = np.std(spike_decode_test[:,rule,:],axis=1)
ax.plot(t,u,'c')
ax.fill_between(t,u-se,u+se,facecolor=(0,1,1,0.5))
self.add_subplot_fixings(ax, chance_level=chance_level)
if DMC:
ax.set_title('Neuronal sample \ncategory decoding')
else:
ax.set_title('Neuronal sample decoding')
ax.set_ylabel('Decoding accuracy')
# add short term plasticity plots
if STP:
if two_rules:
ax = f.add_subplot(num_rows, num_cols, start_sp+9)
im = ax.imshow(sample_stp_pev[:,0,:],aspect='auto',interpolation=None)
f.colorbar(im,orientation='vertical')
ax.spines['right'].set_visible(False)
ax.set_ylabel('Neuron number')
ax.set_xlabel('Time relative to sample onset (ms)')
ax.spines['top'].set_visible(False)
ax.set_xticks([t0[0], t1[0], t2[0], t3[0]])
ax.set_xticklabels([-500,0,500,1500])
ax.set_title('Synaptic sample \nselectvity - DMS task')
ax = f.add_subplot(num_rows, num_cols, start_sp+10)
im = ax.imshow(sample_stp_pev[:,1,:],aspect='auto',interpolation=None)
f.colorbar(im,orientation='vertical')
ax.spines['right'].set_visible(False)
ax.set_ylabel('Neuron number')
ax.set_xlabel('Time relative to sample onset (ms)')
ax.spines['top'].set_visible(False)
ax.set_xticks([t0[0], t1[0], t2[0], t3[0]])
ax.set_xticklabels([-500,0,500,1500])
ax.set_title('Synaptic sample \nselectvity - DMrS task')
else:
ax = f.add_subplot(num_rows, 3, start_sp+7)
im = ax.imshow(sample_stp_pev[:,rule,:],aspect='auto',interpolation=None)
f.colorbar(im,orientation='vertical')
ax.spines['right'].set_visible(False)
ax.set_ylabel('Neuron number')
ax.set_xlabel('Time relative to sample onset (ms)')
ax.spines['top'].set_visible(False)
if DMC:
ax.set_title('Synaptic sample \ncategory selectvity')
else:
ax.set_title('Synaptic sample selectvity')
if self.ABBA:
ax.set_xticks([t0[0], t1[0], t2[0], t3[0]])
ax.set_xticklabels([-500,0,500,1500])
else:
ax.set_xticks([t0[0], t1[0], t2[0], t3[0]])
ax.set_xticklabels([-500,0,500,1500])
if two_rules:
ax = f.add_subplot(num_rows, num_cols, start_sp+11)
plt.hold(True)
u = np.mean(sample_stp_pev[:,0,:],axis=0)
se = np.std(sample_stp_pev[:,0,:],axis=0)/np.sqrt(sample_pev.shape[0])
ax.plot(t,u,'g')
ax.fill_between(t,u-se,u+se,facecolor=(0,1,0,0.5))
u = np.mean(sample_stp_pev[:,1,:],axis=0)
se = np.std(sample_stp_pev[:,1,:],axis=0)/np.sqrt(sample_pev.shape[0])
ax.plot(t,u,'m')
ax.fill_between(t,u-se,u+se,facecolor=(1,0,1,0.5))
ax.set_title('Synaptic sample selectivity')
else:
ax = f.add_subplot(num_rows, num_cols, start_sp+8)
u = np.mean(sample_stp_pev[:,rule,:],axis=0)
se = np.std(sample_stp_pev[:,rule,:],axis=0)/np.sqrt(sample_pev.shape[0])
ax.plot(t,u,'k')
ax.fill_between(t,u-se,u+se,facecolor=(0,0,0,0.5))
if DMC:
ax.set_title('Synaptic sample \ncategory selectivity')
else:
ax.set_title('Synaptic sample selectivity')
self.add_subplot_fixings(ax,chance_level=0,ylim=sample_max*2)
ax.set_ylabel('Normalized PEV')
if two_rules:
ax = f.add_subplot(num_rows, num_cols, start_sp+12)
u = np.mean(synapse_decode[:,0,:],axis=1)
se = np.std(synapse_decode[:,0,:],axis=1)
ax.plot(t,u,'g')
ax.fill_between(t,u-se,u+se,facecolor=(0,1,0,0.5))
u = np.mean(synapse_decode[:,1,:],axis=1)
se = np.std(synapse_decode[:,1,:],axis=1)
ax.plot(t,u,'m')
ax.fill_between(t,u-se,u+se,facecolor=(0.5,0,0.5))
ax.set_title('Synaptic sample decoding')
else:
ax = f.add_subplot(num_rows, num_cols, start_sp+9)
u = np.mean(synapse_decode[:,rule,:],axis=1)
se = np.std(synapse_decode[:,rule,:],axis=1)
ax.plot(t,u,'k')
ax.fill_between(t,u-se,u+se,facecolor=(0,0,0,0.5))
u = np.mean(synapse_decode_test[:,rule,:],axis=1)
se = np.std(synapse_decode_test[:,rule,:],axis=1)
ax.plot(t,u,'c')
ax.fill_between(t,u-se,u+se,facecolor=(0,1,1,0.5))
if DMC:
ax.set_title('Synaptic sample \ncategory decoding')
else:
ax.set_title('Synaptic sample decoding')
self.add_subplot_fixings(ax, chance_level=chance_level)
ax.set_ylabel('Decoding accuracy')
if tight:
plt.tight_layout()
plt.savefig('DMS summary.pdf', format='pdf')
plt.show()
def plot_postle_figure(self,dt=20, STP=False, tight=False):
# declare that we're analyzing a postle task
self.postle = True
sample_pev, sample_stp_pev, sample_decoding, sample_stp_decoding, model_output = self.decoding_accuracy_postle(num_reps=10)
t = np.arange(0,220*dt,dt)
t -= 900 # assuming 400 ms dead time, 500 ms fixation
t0,t1,t2,t3,t4,t5,t6 = np.where(t==-500), np.where(t==0), np.where(t==500), np.where(t==1000), np.where(t==1500), np.where(t==2000), np.where(t==2500)
f = plt.figure(figsize=(6,6))
for i in range(2):
for j in range(2):
ax = f.add_subplot(3, 2, 2*i+j+1)
u = np.mean(sample_decoding[i,j,0,0,:,:],axis=1)
se = np.std(sample_decoding[i,j,0,0,:,:],axis=1)
ax.fill_between(t,u-se,u+se,facecolor=(0,1,0,0.5))
ax.plot(t,u,'g')
u = np.mean(sample_decoding[i,j,0,1,:,:],axis=1)
se = np.std(sample_decoding[i,j,0,1,:,:],axis=1)
ax.fill_between(t,u-se,u+se,facecolor=(1,0.6,0,0.5))
ax.plot(t,u,color=[1,0.6,0])
if STP:
u = np.mean(sample_stp_decoding[i,j,0,0,:,:],axis=1)
se = np.std(sample_stp_decoding[i,j,0,0,:,:],axis=1)
ax.fill_between(t,u-se,u+se,facecolor=(1,0,1,0.5))
ax.plot(t,u,'m')
u = np.mean(sample_stp_decoding[i,j,0,1,:,:],axis=1)
se = np.std(sample_stp_decoding[i,j,0,1,:,:],axis=1)
ax.fill_between(t,u-se,u+se,facecolor=(0,1,1,0.5))
ax.plot(t,u,'c')
self.add_subplot_fixings(ax, chance_level=1/8, ylim=1.1)
ax.set_ylabel('Decoding accuracy')
ax = f.add_subplot(3, 2, 5)
u = np.mean(sample_decoding[0,0,1,1,:,:],axis=1)
se = np.std(sample_decoding[0,0,1,1,:,:],axis=1)
#u = np.mean(np.mean(sample_decoding[0,:,1,1,:,:],axis=0),axis=1)
#se = np.std(np.mean(sample_decoding[0,:,1,1,:,:],axis=0),axis=1)
ax.fill_between(t,u-se,u+se,facecolor=(0,0,0,0.5))
ax.plot(t,u,'k')
u = np.mean(sample_decoding[0,0,0,1,:,:],axis=1)
se = np.std(sample_decoding[0,0,0,1,:,:],axis=1)
#u = np.mean(np.mean(sample_decoding[0,:,0,1,:,:],axis=0),axis=1)
#se = np.std(np.mean(sample_decoding[0,:,0,1,:,:],axis=0),axis=1)
ax.fill_between(t,u-se,u+se,facecolor=(1,0.6,0,0.5))
ax.plot(t,u,color=[1,0.6,0])
self.add_subplot_fixings(ax, chance_level=1/8, ylim=1.1)
ax.plot([2400,2400],[-2, 99],'y--')
ax.set_ylabel('Decoding accuracy')
ax = f.add_subplot(3, 2, 6)
u = np.mean(sample_decoding[1,1,1,0,:,:],axis=1)
se = np.std(sample_decoding[1,1,1,0,:,:],axis=1)
#u = np.mean(np.mean(sample_decoding[1,:,1,0,:,:],axis=0),axis=1)
#se = np.std(np.mean(sample_decoding[1,:,1,0,:,:],axis=0),axis=1)
ax.fill_between(t,u-se,u+se,facecolor=(0,0,0,0.5))
ax.plot(t,u,'k')
u = np.mean(sample_decoding[1,1,0,0,:,:],axis=1)
se = np.std(sample_decoding[1,1,0,0,:,:],axis=1)
#u = np.mean(np.mean(sample_decoding[1,:,0,0,:,:],axis=0),axis=1)
#se = np.std(np.mean(sample_decoding[1,:,0,0,:,:],axis=0),axis=1)
ax.fill_between(t,u-se,u+se,facecolor=(0,1,0,0.5))
ax.plot(t,u,'g')
self.add_subplot_fixings(ax, chance_level=1/8, ylim=1.1)
ax.plot([2400,2400],[-2, 99],'y--')
ax.set_ylabel('Decoding accuracy')
plt.tight_layout()
plt.savefig('postle summary.pdf', format='pdf')
plt.show()
return sample_decoding, sample_stp_decoding
def plot_ABBA_figures(self,dt=25, STP=False, tight=False):
mean_resp, mean_out_match, mean_out_non_match = self.calc_mean_responses()
spike_decode, synapse_decode, spike_decode_test, synapse_decode_test = self.calculate_svms(DMC=[False],decode_test=True)
#sample_pev, test_pev, rule_pev, _, sample_stp_pev, sample_cat_pev, sample_cat_stp_pev, test_stp_pev = self.calculate_pevs()
chance_level = 1/2
trial_length_steps = self.rnn_outputs.shape[1]
trial_length = np.int_(trial_length_steps*dt)
t = np.arange(0,trial_length,dt)
t -= 900 # assuming 400 ms dead time, 500 ms fixation
t0,t1,t2,t3 = np.where(t==-500), np.where(t==0),np.where(t==500),np.where(t==1500)
f = plt.figure(figsize=(6,4))
ax = f.add_subplot(2, 2, 1)
# plot accuracy bars
x=np.array([0,1,2,3])
ax.bar(x+0.1, self.ABBA_accuracy_match,width=0.2,color='r',align='center')
ax.bar(x-0.1, self.ABBA_accuracy_non_match,width=0.2,color='b',align='center')
ax.set_title('Accuracy')
ax.set_ylabel('Fraction correct')
ax.set_xlabel('Num. of distractors')
ax.spines['right'].set_visible(False)
ax.spines['top'].set_visible(False)
ax = f.add_subplot(2, 2, 2)
ax.hold(True)
s = np.mean(mean_resp[:,0,:,:],axis=0)
max_val = np.max(s)
for i in range(8):
ax.plot(t,s[i,:],color=[1-i/7,0,i/7], linewidth=1)
self.add_subplot_fixings(ax,chance_level=0,ylim=0.5)
ax.set_title('Mean response from synpases informative neurons')
ax.set_ylabel('Activity (a.u.)')
ax.set_ylim([0, 0.5])
ax = f.add_subplot(2, 2, 3)
u = np.mean(spike_decode[:,0,:],axis=1)
se = np.std(spike_decode[:,0,:],axis=1)
ax.plot(t,u,'b')
ax.fill_between(t,u-se,u+se,facecolor=(0,0,1,0.5))
u = np.mean(spike_decode_test[:,0,:],axis=1)
se = np.std(spike_decode_test[:,0,:],axis=1)
ax.plot(t,u,'r')
ax.fill_between(t,u-se,u+se,facecolor=(1,0,0,0.5))
self.add_subplot_fixings(ax, chance_level=1/8, ylim=1.1)
ax.set_ylabel('Decoding accuracy')
ax = f.add_subplot(2, 2, 4)
u = np.mean(spike_decode[:,0,:],axis=1)
se = np.std(spike_decode[:,0,:],axis=1)
ax.plot(t,u,'b')
ax.fill_between(t,u-se,u+se,facecolor=(0,0,1,0.5))
u = np.mean(synapse_decode_test[:,0,:],axis=1)
se = np.std(synapse_decode_test[:,0,:],axis=1)
ax.plot(t,u,'r')
ax.fill_between(t,u-se,u+se,facecolor=(1,0,0,0.5))
self.add_subplot_fixings(ax, chance_level=1/8, ylim=1.1)
ax.set_ylabel('Decoding accuracy')
if tight:
plt.tight_layout()
plt.savefig('ABBA summary.pdf', format='pdf')
plt.show()
def add_subplot_fixings(self, ax, chance_level = 0, ylim = 1.1, delayed_rule=False):
ax.plot([0,0],[-2, 99],'k--')
if self.ABBA:
ax.plot([500,500],[-2, 99],'k--')
ax.plot([1000,1000],[-2, 99],'k--')
ax.plot([1500,1500],[-2, 99],'k--')
ax.plot([2000,2000],[-2, 99],'k--')
ax.set_xlim([-500,2500])
ax.set_xticks([-500,0,500,1000,1500,2000,2500])
elif self.postle:
ax.plot([500,500],[-2, 99],'k--')
ax.plot([1000,1000],[-2, 99],'k--')
ax.plot([1500,1500],[-2, 99],'k--')
ax.plot([2000,2000],[-2, 99],'k--')
ax.plot([2500,2500],[-2, 99],'k--')
ax.plot([3000,3000],[-2, 99],'k--')
ax.set_xlim([-500,3500])
ax.set_xticks([-500,0,500,1000,1500,2000,2500,3000])
else:
ax.plot([500,500],[-2, 99],'k--')
ax.plot([1500,1500],[-2, 99],'k--')
ax.set_xlim([-500,2000])
ax.set_xticks([-500,0,500,1500])
if delayed_rule:
ax.set_xticks([-500,0,500,1000,1500])
ax.plot([1000,1000],[-2, 99],'k--')
ax.plot([-700,3600],[chance_level, chance_level],'k--')
ax.set_ylim([-0.1, ylim])
ax.spines['right'].set_visible(False)
ax.spines['top'].set_visible(False)
ax.xaxis.set_ticks_position('bottom')
ax.yaxis.set_ticks_position('left')
#ax.set_title('Tuning similarity between proximal and distal neurons')
ax.set_ylabel('Response')
ax.set_xlabel('Time relative to sample onset (ms)')
def compare_two_tasks(fn1, fn2, DMC=False, ABBA_flag=False,rule = 0, dt=25):
# enter the two filenames, fn1 and fn2
font = {'family' : 'normal',
'weight' : 'bold',
'size' : 12}
na1 = neural_analysis(fn1, ABBA = ABBA_flag)
na2 = neural_analysis(fn2, ABBA = ABBA_flag)
mean_resp1, _, _ = na1.calc_mean_responses()
svm_results1 = na1.calculate_svms(ABBA = ABBA_flag)
sample_pev1, _, rule_pev1, _, sample_stp_pev1, _ , sample_cat_pev1, sample_cat_stp_pev1 = na1.calculate_pevs(ABBA = ABBA_flag)
mean_resp2, _, _ = na2.calc_mean_responses()
svm_results2 = na2.calculate_svms()
sample_pev2, _, rule_pev2, _, sample_stp_pev2, _ ,sample_cat_pev2, sample_cat_stp_pev2 = na2.calculate_pevs()
if DMC:
sample_pev1 = sample_cat_pev1
sample_pev2 = sample_cat_pev2
sample_stp_pev1 = sample_cat_stp_pev1
sample_stp_pev2 = sample_cat_stp_pev2
svm_results1['sample_full'] = svm_results1['sample_full_cat']
svm_results2['sample_full'] = svm_results2['sample_full_cat']
svm_results1['sample_full_stp'] = svm_results1['sample_full_cat_stp']
svm_results2['sample_full_stp'] = svm_results2['sample_full_cat_stp']
if na1.num_rules>1 and False:
# not sure if I want this. If there are more than one task rules, this part will average
# across all rules
sample_pev1[:,0,:] = np.mean(sample_cat_pev1,axis=1)
sample_pev2[:,0,:] = np.mean(sample_cat_pev2,axis=1)
sample_stp_pev1[:,0,:] = np.mean(sample_cat_stp_pev1,axis=1)
sample_stp_pev2[:,0,:] = np.mean(sample_cat_stp_pev2,axis=1)
svm_results1['sample_full'][:,0,:] = np.mean(svm_results1['sample_full'],axis=1)
svm_results2['sample_full'][:,0,:] = np.mean(svm_results2['sample_full'],axis=1)
svm_results1['sample_full_stp'][:,0,:] = np.mean(svm_results1['sample_full_stp'],axis=1)
svm_results2['sample_full_stp'][:,0,:] = np.mean(svm_results2['sample_full_stp'],axis=1)
rule = 0
f = plt.figure(figsize=(8,4))
t = np.arange(0,2700,dt)
t -= 900
t0,t1,t2,t3 = np.where(t==-500), np.where(t==0),np.where(t==500),np.where(t==1500)
if ABBA_flag:
t = np.arange(0,200+500+1500+300+300,dt)
t = np.arange(0,200+500+250+2000,dt)
t -= 700
t0,t1,t2,t3,t4,t5,t6 = np.where(t==-500), np.where(t==0),np.where(t==500),np.where(t==1000),np.where(t==1500),np.where(t==2000), np.where(t==2500)
ax = f.add_subplot(2, 3, 1)
ax.hold(True)
u1 = np.mean(np.mean(mean_resp1[:,rule,:,:],axis=1),axis=0)
u2 = np.mean(np.mean(mean_resp2[:,rule,:,:],axis=1),axis=0)
se1 = np.std(np.mean(mean_resp1[:,rule,:,:],axis=1),axis=0)/np.sqrt(mean_resp1.shape[0])
se2 = np.std(np.mean(mean_resp2[:,rule,:,:],axis=1),axis=0)/np.sqrt(mean_resp1.shape[0])
ax.fill_between(t,u1-se1,u1+se1,facecolor=(0,1,0))
ax.fill_between(t,u2-se2,u2+se2,facecolor=(1,0,1))
ax.plot(t,u1,'g',label='without STP',color=(0,0.5,0),linewidth=2)
ax.plot(t,u2,'m',label='with STP',color=(0.5,0,0.5),linewidth=2)
na1.add_subplot_fixings(ax, chance_level=0, ylim=6, ABBA_task=ABBA_flag)
green_patch = mpatches.Patch(color='green', label='without STP')
magenta_patch = mpatches.Patch(color='magenta', label='with STP')
plt.legend(loc=0, handles=[green_patch,magenta_patch],prop={'size':6})
ax.set_title('Mean population response')
ax.set_ylabel('Mean response')
ax = f.add_subplot(2, 3, 2)
ax.hold(True)
u1 = np.mean(sample_pev1[:,rule,:],axis=0)
u2 = np.mean(sample_pev2[:,rule,:],axis=0)
u3 = np.mean(rule_pev2,axis=0)
se1 = np.std(sample_pev1[:,rule,:],axis=0)/np.sqrt(sample_pev1.shape[0])
se2 = np.std(sample_pev2[:,rule,:],axis=0)/np.sqrt(sample_pev1.shape[0])
ax.fill_between(t,u1-se1,u1+se1,facecolor=(0,1,0))
ax.fill_between(t,u2-se2,u2+se2,facecolor=(1,0,1))
ax.plot(t,u1,'g',label='without STP',color=(0,0.5,0),linewidth=2)
ax.plot(t,u2,'m',label='with STP',color=(0.5,0,0.5),linewidth=2)
ax.plot(t,u3,label='rule with STP',color=(0,0,0),linewidth=2)
na1.add_subplot_fixings(ax, chance_level=0, ylim=0.35,ABBA_task=ABBA_flag)
#green_patch = mpatches.Patch(color='green', label='without STP')
#magenta_patch = mpatches.Patch(color='magenta', label='with STP')
#plt.legend(loc=0, handles=[green_patch,magenta_patch])
ax.set_title('Neuron sample selectivity')
ax.set_ylabel('Normalized PEV')
ax = f.add_subplot(2, 3, 3)
ax.hold(True)
u1 = np.mean(svm_results1['sample_full'][:,rule,:],axis=1)
u2 = np.mean(svm_results2['sample_full'][:,rule,:],axis=1)
se1 = np.std(svm_results1['sample_full'][:,rule,:],axis=1)
se2 = np.std(svm_results2['sample_full'][:,rule,:],axis=1)
se1[np.isnan(se1)] = 0
se2[np.isnan(se2)] = 0
ax.fill_between(t,u1-se1,u1+se1,facecolor=(0,1,0))
ax.fill_between(t,u2-se2,u2+se2,facecolor=(1,0,1))
ax.plot(t,u1,'g',label='without STP',color=(0,0.5,0),linewidth=2)
ax.plot(t,u2,'m',label='with STP',color=(0.5,0,0.5),linewidth=2)
na1.add_subplot_fixings(ax, chance_level=1/8, ylim=1.1,ABBA_task=ABBA_flag)
#green_patch = mpatches.Patch(color='green', label='without STP')
#magenta_patch = mpatches.Patch(color='magenta', label='with STP')
#plt.legend(loc=0, handles=[green_patch,magenta_patch])
ax.set_title('Neuron sample decoding accuracy')
ax.set_ylabel('Decoding accuracy')
ax = f.add_subplot(2, 3, 4)
im = ax.imshow(sample_stp_pev2[:,rule,:],aspect='auto',interpolation=None)
if not ABBA_flag:
ax.set_xticks([t0[0], t1[0], t2[0], t3[0]])
ax.set_xticklabels([-500,0,500,1500])
else:
ax.set_xticks([t0[0], t1[0], t2[0], t3[0], t4[0], t5[0], t6[0]])
ax.set_xticklabels([-500,0,500,1000,1500,2000,2000,2500])
f.colorbar(im,orientation='vertical')
ax.spines['right'].set_visible(False)
ax.spines['top'].set_visible(False)
ax.set_ylabel('Neuron number')
ax.set_xlabel('Time relative to sample onset (ms)')
ax.set_title('Synaptic sample selectivity')
ax = f.add_subplot(2, 3, 5)
ax.hold(True)
u2 = np.mean(sample_stp_pev2[:,rule,:],axis=0)
se2 = np.std(sample_stp_pev2[:,rule,:],axis=0)/np.sqrt(sample_pev1.shape[0])
ax.fill_between(t,u2-se2,u2+se2,facecolor=(1,0,1))
ax.plot(t,u2,'m',label='with STP',color=(0.5,0,0.5),linewidth=2)
na1.add_subplot_fixings(ax, chance_level=0, ylim=0.3,ABBA_task=ABBA_flag)
#green_patch = mpatches.Patch(color='green', label='without STP')
#magenta_patch = mpatches.Patch(color='magenta', label='with STP')
#plt.legend(loc=0, handles=[green_patch,magenta_patch])
ax.set_title('Synaptic sample selectivity')
ax.set_ylabel('Normalized PEV')
ax = f.add_subplot(2, 3, 6)
ax.hold(True)
u2 = np.mean(svm_results2['sample_full_stp'][:,rule,:],axis=1)
se2 = np.std(svm_results2['sample_full_stp'][:,rule,:],axis=1)
se2[np.isnan(se2)] = 0
ax.fill_between(t,u2-se2,u2+se2,facecolor=(1,0,1))
ax.plot(t,u2,'m',label='with STP',color=(0.5,0,0.5),linewidth=2)
na1.add_subplot_fixings(ax, chance_level=1/8, ylim=1.1,ABBA_task=ABBA_flag)
#green_patch = mpatches.Patch(color='green', label='without STP')
#magenta_patch = mpatches.Patch(color='magenta', label='with STP')
#plt.legend(loc=0, handles=[green_patch,magenta_patch])
ax.set_title('Synaptic sample decoding accuray')
ax.set_ylabel('Decoding accuracy')
plt.tight_layout()
plt.savefig('DMS comparison.pdf', format='pdf')
plt.show()
def compare_two_tasks_two_rules(fn1, fn2, DMC=False, ABBA=False, dt=25):
# enter the two filenames, fn1 and fn2
font = {'family' : 'normal',
'weight' : 'bold',
'size' : 12}
na1 = neural_analysis(fn1)
na2 = neural_analysis(fn2)
mean_resp1, _, _ = na1.calc_mean_responses()
svm_results1 = na1.calculate_svms()
sample_pev1, _, rule_pev1, _, sample_stp_pev1, _ , sample_cat_pev1, sample_cat_stp_pev1 = na1.calculate_pevs()
mean_resp2, _, _ = na2.calc_mean_responses()
svm_results2 = na2.calculate_svms()
sample_pev2, _, rule_pev2, _, sample_stp_pev2, _ ,sample_cat_pev2, sample_cat_stp_pev2 = na2.calculate_pevs()
if DMC:
sample_pev1 = sample_cat_pev1
sample_pev2 = sample_cat_pev2
sample_stp_pev1 = sample_cat_stp_pev1
sample_stp_pev2 = sample_cat_stp_pev2
svm_results1['sample_full'] = svm_results1['sample_full_cat']
svm_results2['sample_full'] = svm_results2['sample_full_cat']
svm_results1['sample_full_stp'] = svm_results1['sample_full_cat_stp']
svm_results2['sample_full_stp'] = svm_results2['sample_full_cat_stp']
f = plt.figure(figsize=(8,9))
t = np.arange(0,2700,dt)
t -= 900
t0,t1,t2,t3 = np.where(t==-500), np.where(t==0),np.where(t==500),np.where(t==1500)
if ABBA:
t = np.arange(0,200+500+1500+300+300,dt)
t -= 700
t0,t1,t2,t3,t4,t5,t6,t7 = np.where(t==-500), np.where(t==0),np.where(t==300),np.where(t==600),np.where(t==900),np.where(t==1200), np.where(t==1500), np.where(t==1800)
ax = f.add_subplot(5, 2, 1)
ax.hold(True)
u1 = np.mean(np.mean(np.mean(mean_resp1[:,:,:,:],axis=1),axis=1),axis=0)
u2 = np.mean(np.mean(np.mean(mean_resp2[:,:,:,:],axis=1),axis=1),axis=0)
se1 = np.std(np.mean(np.mean(mean_resp1[:,:,:,:],axis=1),axis=1),axis=0)/np.sqrt(mean_resp1.shape[0])
se2 = np.std(np.mean(np.mean(mean_resp2[:,:,:,:],axis=1),axis=1),axis=0)/np.sqrt(mean_resp1.shape[0])
ax.fill_between(t,u1-se1,u1+se1,facecolor=(0,1,0))
ax.fill_between(t,u2-se2,u2+se2,facecolor=(1,0,1))
ax.plot(t,u1,'g',label='without STP',color=(0,0.5,0),linewidth=2)
ax.plot(t,u2,'m',label='with STP',color=(0.5,0,0.5),linewidth=2)
na1.add_subplot_fixings(ax, chance_level=0, ylim=4, ABBA_task=ABBA,delayed_rule=True)
green_patch = mpatches.Patch(color='green', label='without STP')
magenta_patch = mpatches.Patch(color='magenta', label='with STP')
plt.legend(loc=0, handles=[green_patch,magenta_patch],prop={'size':6})
ax.set_title('Mean population response')
ax.set_ylabel('Mean response')
ax = f.add_subplot(5, 2, 2)
ax.hold(True)
u1 = np.mean(rule_pev1,axis=0)
u2 = np.mean(rule_pev2,axis=0)
se1 = np.std(rule_pev1,axis=0)/np.sqrt(sample_pev1.shape[0])
se2 = np.std(rule_pev2,axis=0)/np.sqrt(sample_pev1.shape[0])
ax.fill_between(t,u1-se1,u1+se1,facecolor=(0,1,0))
ax.fill_between(t,u2-se2,u2+se2,facecolor=(1,0,1))
ax.plot(t,u1,'g',label='without STP',color=(0,0.5,0),linewidth=2)
ax.plot(t,u2,'m',label='with STP',color=(0.5,0,0.5),linewidth=2)
na1.add_subplot_fixings(ax, chance_level=0, ylim=0.3,ABBA_task=ABBA,delayed_rule=True)
ax.set_title('Neuron rule selectivity')
ax.set_ylabel('Normalized PEV')
for rule in range(2):
ax = f.add_subplot(5, 2, 3+2*rule)
ax.hold(True)
u1 = np.mean(sample_pev1[:,rule,:],axis=0)
u2 = np.mean(sample_pev2[:,rule,:],axis=0)
se1 = np.std(sample_pev1[:,rule,:],axis=0)/np.sqrt(sample_pev1.shape[0])
se2 = np.std(sample_pev2[:,rule,:],axis=0)/np.sqrt(sample_pev1.shape[0])
ax.fill_between(t,u1-se1,u1+se1,facecolor=(0,1,0))
ax.fill_between(t,u2-se2,u2+se2,facecolor=(1,0,1))
ax.plot(t,u1,'g',label='without STP',color=(0,0.5,0),linewidth=2)
ax.plot(t,u2,'m',label='with STP',color=(0.5,0,0.5),linewidth=2)
na1.add_subplot_fixings(ax, chance_level=0, ylim=0.7,ABBA_task=ABBA,delayed_rule=True)
ax.set_title('Neuron sample selectivity')
ax.set_ylabel('Normalized PEV')
ax = f.add_subplot(5, 2, 4+2*rule)
ax.hold(True)
u1 = np.mean(svm_results1['sample_full'][:,rule,:],axis=1)
u2 = np.mean(svm_results2['sample_full'][:,rule,:],axis=1)
se1 = np.std(svm_results1['sample_full'][:,rule,:],axis=1)
se2 = np.std(svm_results2['sample_full'][:,rule,:],axis=1)
se1[np.isnan(se1)] = 0
se2[np.isnan(se2)] = 0
ax.fill_between(t,u1-se1,u1+se1,facecolor=(0,1,0))
ax.fill_between(t,u2-se2,u2+se2,facecolor=(1,0,1))
ax.plot(t,u1,'g',label='without STP',color=(0,0.5,0),linewidth=2)
ax.plot(t,u2,'m',label='with STP',color=(0.5,0,0.5),linewidth=2)
na1.add_subplot_fixings(ax, chance_level=1/8, ylim=1.1,ABBA_task=ABBA,delayed_rule=True)
ax.set_title('Neuron sample decoding accuracy')
ax.set_ylabel('Decoding accuracy')
ax = f.add_subplot(5, 2, 7+2*rule)
ax.hold(True)
u1 = np.mean(sample_stp_pev1[:,rule,:],axis=0)
u2 = np.mean(sample_stp_pev2[:,rule,:],axis=0)
se1 = np.std(sample_stp_pev1[:,rule,:],axis=0)/np.sqrt(sample_pev1.shape[0])
se2 = np.std(sample_stp_pev2[:,rule,:],axis=0)/np.sqrt(sample_pev1.shape[0])
ax.fill_between(t,u1-se1,u1+se1,facecolor=(0,1,0))
ax.fill_between(t,u2-se2,u2+se2,facecolor=(1,0,1))
ax.plot(t,u1,'g',label='without STP',color=(0,0.5,0),linewidth=2)
ax.plot(t,u2,'m',label='with STP',color=(0.5,0,0.5),linewidth=2)
na1.add_subplot_fixings(ax, chance_level=0, ylim=0.7,ABBA_task=ABBA,delayed_rule=True)
ax.set_title('Synaptic sample selectivity')
ax.set_ylabel('Normalized PEV')
ax = f.add_subplot(5, 2, 8+2*rule)
ax.hold(True)
u1 = np.mean(svm_results1['sample_full_stp'][:,rule,:],axis=1)
u2 = np.mean(svm_results2['sample_full_stp'][:,rule,:],axis=1)
se1 = np.std(svm_results1['sample_full_stp'][:,rule,:],axis=1)
se2 = np.std(svm_results2['sample_full_stp'][:,rule,:],axis=1)
se1[np.isnan(se1)] = 0
se2[np.isnan(se2)] = 0
ax.fill_between(t,u1-se1,u1+se1,facecolor=(0,1,0))
ax.fill_between(t,u2-se2,u2+se2,facecolor=(1,0,1))
ax.plot(t,u1,'g',label='without STP',color=(0,0.5,0),linewidth=2)
ax.plot(t,u2,'m',label='with STP',color=(0.5,0,0.5),linewidth=2)
na1.add_subplot_fixings(ax, chance_level=1/8, ylim=1.1,ABBA_task=ABBA,delayed_rule=True)
ax.set_title('Synaptic sample decoding accuracy')
ax.set_ylabel('Decoding accuracy')
plt.tight_layout()
plt.savefig('Two rules comparison.pdf', format='pdf')
plt.show()
def plt_dual_figures(fn1, fn2, ABBA=False, DMC=False, two_rules=False):
# assume fn1 has no STP, and fn2 does
if two_rules:
fig_handle = plt.figure(figsize=(10,10))
sp = 8
else:
fig_handle = plt.figure(figsize=(8,10))
sp = 6
na = neural_analysis(fn1, ABBA=ABBA)
na.plot_all_figures(rule=0, STP=False, ABBA=ABBA, DMC=DMC, two_rules=two_rules,f=fig_handle, start_sp=0, num_rows=5, tight=False)
na = neural_analysis(fn2, ABBA=ABBA)
na.plot_all_figures(rule=0, STP=True, ABBA=ABBA, DMC=DMC, two_rules=two_rules,f=fig_handle, start_sp=sp, num_rows=5, tight=True)
def plot_summary_decoding_figure():
fn1 = 'C:/Users/nicol_000/Projects/RNN STP Model/saved_model_files/DMS.pkl'
fn2 = 'C:/Users/nicol_000/Projects/RNN STP Model/saved_model_files/DMS_std_stf.pkl'
fn3 = 'C:/Users/nicol_000/Projects/RNN STP Model/saved_model_files/DMC.pkl'
fn4 = 'C:/Users/nicol_000/Projects/RNN STP Model/saved_model_files/DMC_std_stf.pkl'
fn5 = 'C:/Users/nicol_000/Projects/RNN STP Model/saved_model_files/DMS_rotation.pkl'
fn6 = 'C:/Users/nicol_000/Projects/RNN STP Model/saved_model_files/DMS_rotate_std_stf_v3.pkl'
fn7 = 'C:/Users/nicol_000/Projects/RNN STP Model/saved_model_files/DMS_and_rotate_v3.pkl'
fn8 = 'C:/Users/nicol_000/Projects/RNN STP Model/saved_model_files/DMS_and_rotate_std_stf_v3.pkl'
fn9 = 'C:/Users/nicol_000/Projects/RNN STP Model/saved_model_files/ABBA.pkl'
fn10 = 'C:/Users/nicol_000/Projects/RNN STP Model/saved_model_files/ABBA_std_stf_v2.pkl'
fig_handle = plt.figure(figsize=(6,10))
num_rows = 5
plot_decoding_pairs(fn1, fn2, fig_handle, num_rows=num_rows, start_sp=0, DMC=False, ABBA=False, two_rules=False)
plot_decoding_pairs(fn3, fn4, fig_handle, num_rows=num_rows, start_sp=2, DMC=True, ABBA=False, two_rules=False)
plot_decoding_pairs(fn5, fn6, fig_handle, num_rows=num_rows, start_sp=4, DMC=False, ABBA=False, two_rules=False)
plot_decoding_pairs(fn7, fn8, fig_handle, num_rows=num_rows, start_sp=6, DMC=False, ABBA=False, two_rules=True)
plot_decoding_pairs(fn9, fn10, fig_handle, num_rows=num_rows, start_sp=8, DMC=False, ABBA=True, two_rules=False)
plt.tight_layout()
plt.savefig('Summary.pdf', format='pdf')
plt.show()
def plot_decoding_pairs(fn1, fn2, f, num_rows, start_sp, DMC=False, ABBA=False, two_rules=False):
dt = 25
na = neural_analysis(fn1, ABBA=False)
svm_results1 = na.calculate_svms()
na = neural_analysis(fn2, ABBA=False)
svm_results2 = na.calculate_svms()
trial_length_steps = svm_results1['sample_full'].shape[0]
trial_length = np.int_(trial_length_steps*dt)
t = np.arange(0,trial_length,dt)
t -= 900 # assuming 400 ms dead time, 500 ms fixation
if DMC:
svm_results1['sample_full'] = svm_results1['sample_full_cat']
svm_results2['sample_full'] = svm_results2['sample_full_cat']
svm_results1['sample_full_stp'] = svm_results1['sample_full_cat_stp']
svm_results2['sample_full_stp'] = svm_results2['sample_full_cat_stp']
if two_rules:
svm_results1['sample_full'] = np.mean(svm_results1['sample_full'],axis=1)
svm_results2['sample_full'] = np.mean(svm_results2['sample_full'],axis=1)
svm_results1['sample_full_stp'] = np.mean(svm_results1['sample_full_stp'],axis=1)
svm_results2['sample_full_stp'] = np.mean(svm_results2['sample_full_stp'],axis=1)
else:
svm_results1['sample_full'] = np.squeeze(svm_results1['sample_full'][:,0,:])
svm_results2['sample_full'] = np.squeeze(svm_results2['sample_full'][:,0,:])
svm_results1['sample_full_stp'] = np.squeeze(svm_results1['sample_full_stp'][:,0,:])
svm_results2['sample_full_stp'] = np.squeeze(svm_results2['sample_full_stp'][:,0,:])
print(svm_results1['sample_full'].shape)
ax = f.add_subplot(num_rows, 2, start_sp+1)
u = np.mean(svm_results1['sample_full'],axis=1)
se = np.std(svm_results1['sample_full'],axis=1)
print(u.shape, se.shape, t.shape)
ax.plot(t,u,'g')
ax.fill_between(t,u-se,u+se,facecolor=(0,0.5,0))
u = np.mean(svm_results2['sample_full'],axis=1)
se = np.std(svm_results2['sample_full'],axis=1)
ax.plot(t,u,'m')
ax.fill_between(t,u-se,u+se,facecolor=(0.5,0,0.5))
if DMC:
ax.set_title('Neuronal sample \category decoding')
cl = 1/2
else:
ax.set_title('Neuronal sample decoding')
cl = 1/8
na.add_subplot_fixings(ax, chance_level=cl)
ax.set_ylabel('Decoding accuracy')
ax = f.add_subplot(num_rows, 2, start_sp+2)
u = np.mean(svm_results2['sample_full_stp'],axis=1)
se = np.std(svm_results2['sample_full_stp'],axis=1)
ax.plot(t,u,'m')
ax.fill_between(t,u-se,u+se,facecolor=(0.5,0,0.5))
if DMC:
ax.set_title('Synaptic sample \category decoding')
cl = 1/2
else:
ax.set_title('Synaptic sample decoding')
cl = 1/8
na.add_subplot_fixings(ax, chance_level=cl)
def plot_summary_results(old_format = False):
dt = 20
t = np.arange(0,2900,dt)
t -= 900
t0,t1,t2,t3 = np.where(t==-500), np.where(t==0),np.where(t==500),np.where(t==1500)
num_svm_reps = 2
trial_length = (400+500+500+1000+500)//dt
N = 11
data_dir = 'D:/Masse/RNN STP/saved_models/'
fn = ['DMS_', 'DMS_stp_', 'DMC_stp_', 'DMrS_stp_']
titles = ['DMS no STP', 'DMS', 'DMC', 'DMrS']
spike_decoding = np.zeros((4, N, trial_length, num_svm_reps))
synapse_decoding = np.zeros((4, N, trial_length, num_svm_reps))
spike_decoding_test = np.zeros((4, N, trial_length, num_svm_reps))
synapse_decoding_test = np.zeros((4, N, trial_length, num_svm_reps))
perf = np.zeros((4, N))
perf_shuffled_hidden = np.zeros((4, N))
perf_shuffled_stp = np.zeros((4, N))
"""
Calculate the spiking and synaptic sample decoding accuracy across all networks
Calculate the behavioral performance
"""
for i in range(N):
print('Group ', i)
for j in range(1,4):
if j == 2:
DMC = [True]
else:
DMC = [False]
f = data_dir + fn[j] + str(i) + '.pkl'
na = neural_analysis(f, ABBA=False, old_format = old_format)
perf[j,i] = get_perf(na.desired_outputs, na.model_outputs, na.train_mask, na.rule)
spike_decode, synapse_decode, spike_decode_test, synapse_decode_test = na.calculate_svms(num_reps = num_svm_reps, DMC = DMC)
spike_decoding[j,i,:,:] = spike_decode[:,0,:]
synapse_decoding[j,i,:,:] = synapse_decode[:,0,:]
spike_decoding_test[j,i,:,:] = spike_decode_test[:,0,:]
synapse_decoding_test[j,i,:,:] = synapse_decode_test[:,0,:]
a = contrib_to_behavior.Analysis(f,old_format = old_format)
perf[j,i], perf_shuffled_hidden[j,i], perf_shuffled_stp[j,i] = a.simulate_network()
"""
Calculate the mean decoding accuracy for the last 500 ms of the delay
"""
d = range(1900//dt,2400//dt)
delay_accuracy = np.mean(np.mean(spike_decoding[:,:,d,:],axis=3),axis=2)
ind_example = [0]
for j in range(1,4):
ind_good_perf = np.where(perf[j,:] > 0.9)[0]
ind_sort = np.argsort(delay_accuracy[j,ind_good_perf])[0]
ind_example.append(ind_good_perf[ind_sort])
"""
Plot decoding accuracy from example models
Only consider models with performance accuracy above 99.0%
Will use the model with the lowest spike decoding value during the last 500 ms of the delay
"""
print(ind_example)
f = plt.figure(figsize=(6,4))
for j in range(1,4):
if j == 2:
chance_level = 1/2
else:
chance_level = 1/8
ax = f.add_subplot(2, 2, j+1)
u = np.mean(spike_decoding[j,ind_example[j],:,:],axis=1)
se = np.std(spike_decoding[j,ind_example[j],:,:],axis=1)
ax.plot(t,u,'g')
ax.fill_between(t,u-se,u+se,facecolor=(0,1,0,0.5))
u = np.mean(synapse_decoding[j,ind_example[j],:,:],axis=1)
se = np.std(synapse_decoding[j,ind_example[j],:,:],axis=1)
ax.plot(t,u,'m')
ax.fill_between(t,u-se,u+se,facecolor=(1,0,1,0.5))
na.add_subplot_fixings(ax, chance_level=chance_level)
ax.set_title(titles[j])
ax.set_ylabel('Decoding accuracy')
ax.set_ylim([0, 1])
plt.tight_layout()
plt.savefig('Example models.pdf', format='pdf')
plt.show()
"""
Plot mean decoding accuracy across all models
Only use models with performance accuracy above 85%
"""
print(ind_example)
f = plt.figure(figsize=(6,4))
for j in range(1,4):
if j == 2:
chance_level = 1/2
else:
chance_level = 1/8
ind_good_models = np.where(perf[j,:] > 0.85)[0]
ax = f.add_subplot(2, 2, j+1)
u = np.mean(np.mean(spike_decoding[j,ind_good_models,:,:],axis=2),axis=0)
se = np.std(np.mean(spike_decoding[j,ind_good_models,:,:],axis=2),axis=0)/np.sqrt(len(ind_good_models))
ax.plot(t,u,'g')
ax.fill_between(t,u-se,u+se,facecolor=(0,1,0,0.5))
u = np.mean(np.mean(synapse_decoding[j,ind_good_models,:,:],axis=2),axis=0)
se = np.std(np.mean(synapse_decoding[j,ind_good_models,:,:],axis=2),axis=0)/np.sqrt(len(ind_good_models))
ax.plot(t,u,'m')
ax.fill_between(t,u-se,u+se,facecolor=(1,0,1,0.5))
na.add_subplot_fixings(ax, chance_level=chance_level)
ax.set_title(titles[j])
ax.set_ylabel('Decoding accuracy')
ax.set_ylim([0, 1])
plt.tight_layout()
plt.savefig('Average models.pdf', format='pdf')
plt.show()
"""
Plot decoding accuracy across all models using heatmaps
Only use models with performance accuracy above 97.5%
"""
print(ind_example)
f = plt.figure(figsize=(6,4))
for j in range(1,4):
if j == 2:
chance_level = 1/2
else:
chance_level = 1/8
ind_good_models = np.where(perf[j,:] > 0.975)[0]
ax = f.add_subplot(2, 2, j+1)
u = np.mean(synapse_decoding[j,ind_good_models,:,:],axis=2)
im = ax.imshow(u,aspect='auto',interpolation=None)
f.colorbar(im,orientation='vertical')
ax.spines['right'].set_visible(False)
ax.set_ylabel('Model number')
ax.set_xlabel('Time relative to sample onset (ms)')
ax.spines['top'].set_visible(False)
ax.set_title(titles[j])
ax.set_xticks([t0[0], t1[0], t2[0], t3[0]])
ax.set_xticklabels([-500,0,500,1500])
plt.tight_layout()
plt.savefig('All models.pdf', format='pdf')
plt.show()
print(ind_example)
return spike_decoding, synapse_decoding, spike_decoding_test, synapse_decoding_test, perf, perf_shuffled_hidden, perf_shuffled_stp
def plot_variable_delay_results():
"""
Plot a model that was trained on a variable delay
"""
data_dir = 'C:/Users/Freedmanlab/Documents/Masse/STP/saved_models/'
dt = 25
num_svm_reps = 5
t = np.arange(0,2900,dt)
t -= 900
fn = 'DMS_EI_std_stf_var_delay_1_iter1000.pkl'
f = data_dir + fn
na = neural_analysis(f, ABBA=False)
perf = get_perf(na.desired_outputs, na.model_outputs, na.train_mask)
print('Model accuracy = ', perf)
spike_decode, synapse_decode = na.calculate_svms(num_reps = num_svm_reps, DMC = False)
f = plt.figure(figsize=(3,2))
chance_level = 1/8
ax = f.add_subplot(1, 1, 1)
u = np.mean(spike_decode[:,0,:],axis=1)
se = np.std(spike_decode[:,0,:],axis=1)
ax.plot(t,u,'g')
ax.fill_between(t,u-se,u+se,facecolor=(0,1,0,0.5))
u = np.mean(synapse_decode[:,0,:],axis=1)
se = np.std(synapse_decode[:,0,:],axis=1)
ax.plot(t,u,'m')
ax.fill_between(t,u-se,u+se,facecolor=(1,0,1,0.5))
na.add_subplot_fixings(ax, chance_level=chance_level)
ax.set_ylabel('Decoding accuracy')
ax.set_ylim([0, 1])
plt.tight_layout()
plt.savefig('Var delay model.pdf', format='pdf')
plt.show()
def plot_multiple_delay_results():
dt = 20
num_svm_reps = 5
N = 8
data_dir = 'D:/Masse/RNN STP/saved_models/'
delay = [1000,1500,2000]
num_delays = len(delay)
mean_decoding = np.zeros((num_delays, N))
std_decoding = np.zeros((num_delays, N))
perf = np.zeros((num_delays, N))
for i in range(N):
print('Group ', i)
for j in range(num_delays):
if j==0:
f = data_dir + 'DMS_stp_' + str(i) + '.pkl'
else:
f = data_dir + 'DMS_stp_delay_' + str(delay[j]) + '_' + str(i) + '.pkl'
try:
na = neural_analysis(f, ABBA=False, old_format = False)
spike_decode, synapse_decode,_,_ = na.calculate_svms(num_reps = num_svm_reps, DMC = [False])
perf[j,i] = get_perf(na.desired_outputs, na.model_outputs, na.train_mask, na.rule)
except:
na = neural_analysis(f, ABBA=False, old_format = True)
spike_decode, synapse_decode,_,_ = na.calculate_svms(num_reps = num_svm_reps, DMC = [False])
perf[j,i] = get_perf(na.desired_outputs, na.model_outputs, na.train_mask, na.rule)
# look at last 100 ms of delay epoch
# variable delay
delay_end = (400+500+500+delay[j])//dt
delay_start = (400+500+500+delay[j]-100)//dt
# variable tau
#delay_end = (400+500+500+1000)//dt
#delay_start = (400+500+500+900)//dt
mean_decoding[j,i] = np.mean(spike_decode[delay_start:delay_end,0,:])
std_decoding[j,i] = np.std(np.mean(spike_decode[delay_start:delay_end,0,:],axis=0))
print(i,j,perf[j,i],mean_decoding[j,i],std_decoding[j,i])
f = plt.figure(figsize=(3,2))
chance_level = 1/8
ax = f.add_subplot(1, 1, 1)
for i, d in enumerate(delay):
# only use models with over 90% accuracy
ind_good_model = np.where(perf[i,:]>0.90)[0]
ax.plot([d]*len(ind_good_model),mean_decoding[i,ind_good_model],'k.')
ax.plot([0,3000],[chance_level,chance_level],'k--')
ax.set_ylim([0, 1])
ax.set_xlim([400, 2100])
ax.set_xticks(delay)
ax.set_xticklabels(delay)
return mean_decoding, std_decoding, perf
def plot_summary_results_v2(old_format = False):
dt = 20
t = np.arange(0,2900,dt)
t -= 900
t0,t1,t2,t3 = np.where(t==-500), np.where(t==0),np.where(t==500),np.where(t==1500)
num_svm_reps = 2
trial_length = (400+500+500+1000+500)//dt
N = 20
data_dir = 'D:/Masse/RNN STP/saved_models/'
fn = ['DMS_stp_', 'DMC_stp_', 'DMrS_stp_', 'DMS_DMrS_stp_']
titles = ['DMS', 'DMC', 'DMrS', 'DMS_DMrS']
num_tasks = len(fn)
"""
the DMS_DMrS will produce two decoding/accuracy scores, one for each task
thus, will show num_tasks+1 set of values
"""
spike_decoding = np.zeros((num_tasks+1, N, trial_length, num_svm_reps))
synapse_decoding = np.zeros((num_tasks+1, N, trial_length, num_svm_reps))
spike_decoding_test = np.zeros((num_tasks+1, N, trial_length, num_svm_reps))
synapse_decoding_test = np.zeros((num_tasks+1, N, trial_length, num_svm_reps))
perf = np.zeros((num_tasks+1, N))
perf_shuffled_hidden = np.zeros((num_tasks+1, N))
perf_shuffled_stp = np.zeros((num_tasks+1, N))
"""
Calculate the spiking and synaptic sample decoding accuracy across all networks
Calculate the behavioral performance
"""
for i in range(N):
print('Group ', i)
for j in range(num_tasks):
if fn[j] == 'DMC_stp_':
DMC = [True]
elif fn[j] == 'DMS_DMrS_stp_':
DMC = [False, False]
else:
DMC = [False]
f = data_dir + fn[j] + str(i) + '.pkl'
try:
na = neural_analysis(f, ABBA=False, old_format = old_format)
except:
na = neural_analysis(f, ABBA=False, old_format = not old_format)
#perf_temp = get_perf(na.desired_outputs, na.model_outputs, na.train_mask, na.rule)
spike_decode, synapse_decode, spike_decode_test, synapse_decode_test = na.calculate_svms(num_reps = num_svm_reps, DMC = DMC)
try:
a = contrib_to_behavior.Analysis(f,old_format = old_format)
perf_temp, perf_shuffled_hidden_temp, perf_shuffled_stp_temp = a.simulate_network()
except:
a = contrib_to_behavior.Analysis(f,old_format = not old_format)
perf_temp, perf_shuffled_hidden_temp, perf_shuffled_stp_temp = a.simulate_network()
if j<3:
print(perf_temp)
perf[j,i] = perf_temp
perf_shuffled_hidden[j,i] = perf_shuffled_hidden_temp
perf_shuffled_stp[j,i] = perf_shuffled_stp_temp
spike_decoding[j,i,:,:] = spike_decode[:,0,:]
synapse_decoding[j,i,:,:] = synapse_decode[:,0,:]
spike_decoding_test[j,i,:,:] = spike_decode_test[:,0,:]
synapse_decoding_test[j,i,:,:] = synapse_decode_test[:,0,:]
else:
perf[j:,i] = perf_temp
perf_shuffled_hidden[j:,i] = perf_shuffled_hidden_temp
perf_shuffled_stp[j:,i] = perf_shuffled_stp_temp
spike_decoding[j:,i,:,:] = np.transpose(spike_decode[:,:,:],(1,0,2))
synapse_decoding[j:,i,:,:] = np.transpose(synapse_decode[:,:,:],(1,0,2))
spike_decoding_test[j:,i,:,:] = np.transpose(spike_decode_test[:,:,:],(1,0,2))
synapse_decoding_test[j:,i,:,:] = np.transpose(synapse_decode_test[:,:,:],(1,0,2))
print(spike_decoding.shape)
"""
Calculate the mean decoding accuracy for the last 500 ms of the delay
"""
dt=20
d = range(1900//dt,2400//dt)
delay_accuracy = np.mean(np.mean(spike_decoding[:,:,d,:],axis=3),axis=2)
fn = ['DMS_stp_', 'DMC_stp_', 'DMrS_stp_', 'DMS_DMrS_stp_']
titles = ['DMS', 'DMC', 'DMrS', 'DMS + DMrS']
# combine the DMS and DMrS trials for the DMS_DMrS task
delay_accuracy[3,:] = np.mean(delay_accuracy[3:,:],axis=0)
perf_combined = perf[:num_tasks,:]
perf_combined[num_tasks-1,:] = np.mean(perf[num_tasks:,:],axis=0)
# will find 2 examples for each task
ind_example = np.zeros((num_tasks, 3),dtype=np.int8)
for j in range(num_tasks):
ind_good_perf = np.where(perf_combined[j,:] > 0.9)[0]
ind_sort = np.argsort(delay_accuracy[j,ind_good_perf])
#ind_example[j,0] = ind_good_perf[ind_sort][-1]
ind_example[j,0]= ind_good_perf[ind_sort][len(ind_sort)//2]
ind_example[j,1]= ind_good_perf[ind_sort][0]
f = plt.figure(figsize=(6,8.5))
for j in range(num_tasks):
if fn[j] == 'DMC_stp_':
chance_level = 1/2
else:
chance_level = 1/8
for i in range(2):
ax = f.add_subplot(num_tasks+1, 2, j*2+i+1)
u = np.mean(spike_decoding[j,ind_example[j,i],:,:],axis=1)
se = np.std(spike_decoding[j,ind_example[j,i],:,:],axis=1)
ax.plot(t,u,'g')
ax.fill_between(t,u-se,u+se,facecolor=(0,1,0,0.5))
u = np.mean(synapse_decoding[j,ind_example[j,i],:,:],axis=1)
se = np.std(synapse_decoding[j,ind_example[j,i],:,:],axis=1)
ax.plot(t,u,'m')
ax.fill_between(t,u-se,u+se,facecolor=(1,0,1,0.5))
na.add_subplot_fixings(ax, chance_level=chance_level)
if j == 3:
# DMS_DMrS task
u = np.mean(spike_decoding[j+1,ind_example[j,i],:,:],axis=1)
se = np.std(spike_decoding[j+1,ind_example[j,i],:,:],axis=1)
ax.plot(t,u,'b')
ax.fill_between(t,u-se,u+se,facecolor=(0,0,1,0.5))
u = np.mean(synapse_decoding[j+1,ind_example[j,i],:,:],axis=1)
se = np.std(synapse_decoding[j+1,ind_example[j,i],:,:],axis=1)
ax.plot(t,u,'r')
ax.fill_between(t,u-se,u+se,facecolor=(1,0,0,0.5))
ax.set_xticks([-500,0,500,1000,1500])
ax.plot([1000,1000],[-2, 99],'k--')
ax.set_yticks([0,0.5,1])
ax.set_title(titles[j])
ax.set_ylabel('Decoding accuracy')
ax.set_ylim([0, 1])
plt.tight_layout()
plt.savefig('Summary1.pdf', format='pdf')
plt.show()
col=['b','r','g','c','k']
marker = ['o','v','^','s','D']
"""
Normalize delay decoding
"""
for j in range(num_tasks+1):
if j == 1:
delay_accuracy[j,:] = (delay_accuracy[j,:]-0.5)*2
else:
delay_accuracy[j,:] = (delay_accuracy[j,:]-1/8)*8/7
f = plt.figure(figsize=(6.5,3))
ax = f.add_subplot(1, 3, 1)
for j in range(num_tasks+1):
ind_good_models = np.where(perf[j,:] > 0.9)[0]
#ax.plot(delay_accuracy[j,ind_good_models], perf_shuffled_hidden[j,ind_good_models]
# -perf[j,ind_good_models],marker[j], color=col[j], markersize=3)
ax.plot(delay_accuracy[j,ind_good_models], perf_shuffled_hidden[j,ind_good_models]
-perf[j,ind_good_models],marker[j], color=col[j], markersize=3)
ax.set_xlim(-0.1,1.02)
ax.set_aspect(1.12/0.5)
ax.spines['right'].set_visible(False)
ax.spines['top'].set_visible(False)
ax.xaxis.set_ticks_position('bottom')
ax.yaxis.set_ticks_position('left')
ax.set_yticks([-0.5,-0.25,0])
ax.set_xticks([0,0.5,1])
ax.set_ylabel('Delta acc. shuffled spike rate')
ax.set_xlabel('Normalized delay decoding acc.')
ax = f.add_subplot(1, 3, 2)
for j in range(num_tasks+1):
ind_good_models = np.where(perf[j,:] > 0.9)[0]
#ax.plot(delay_accuracy[j,ind_good_models], perf_shuffled_hidden[j,ind_good_models]
# -perf[j,ind_good_models],marker[j], color=col[j], markersize=3)
ax.plot(delay_accuracy[j,ind_good_models], perf_shuffled_stp[j,ind_good_models]
-perf[j,ind_good_models],marker[j], color=col[j], markersize=3)
ax.set_xlim(-0.1,1.02)
ax.spines['right'].set_visible(False)
ax.spines['top'].set_visible(False)
ax.xaxis.set_ticks_position('bottom')
ax.yaxis.set_ticks_position('left')
ax.set_yticks([-0.5,-0.25,0])
ax.set_xticks([0,0.5,1])
ax.set_aspect(1.12/0.5)
ax.set_ylabel('Delta acc. shuffled STP')
ax.set_xlabel('Normalized delay decoding acc.')
ax = f.add_subplot(1, 3, 3)
for j in range(num_tasks+1):
ind_good_models = np.where(perf[j,:] > 0.9)[0]
ax.plot(perf_shuffled_stp[j,ind_good_models]-perf[j,ind_good_models], perf_shuffled_hidden[j,ind_good_models]
-perf[j,ind_good_models],marker[j], color=col[j], markersize=3)
ax.set_aspect(1)
ax.spines['right'].set_visible(False)
ax.spines['top'].set_visible(False)
ax.xaxis.set_ticks_position('bottom')
ax.yaxis.set_ticks_position('left')
ax.set_yticks([-0.5,-0.25,0])
ax.set_xticks([-0.5,-0.25,0])
ax.set_ylabel('Delta acc. shuffled spike rate')
ax.set_xlabel('Delta acc. shuffled STP')
plt.tight_layout()
plt.savefig('Summary2.pdf', format='pdf')
plt.show()
return spike_decoding, synapse_decoding, spike_decoding_test, synapse_decoding_test, perf, perf_shuffled_hidden, perf_shuffled_stp, ind_example
def get_perf(y, y_hat, mask, rule):
"""
only examine time points when test stimulus is on
in another words, when y[0,:,:] is not 0
"""
print('Neural analysis: get_perf')
print(y.shape, y_hat.shape, mask.shape)
mask *= np.logical_or(y[1,:,:]>0,y[2,:,:]>0)
#mask *= y[0,:,:]==0
y = np.argmax(y, axis = 0)
y_hat = np.argmax(y_hat, axis = 0)
return np.sum(np.float32(y == y_hat)*np.squeeze(mask))/np.sum(mask)
| [
"numpy.sum",
"numpy.arctan2",
"numpy.maximum",
"numpy.argmax",
"numpy.abs",
"numpy.floor",
"numpy.ones",
"numpy.isnan",
"numpy.argsort",
"matplotlib.pyplot.figure",
"numpy.mean",
"numpy.arange",
"numpy.sin",
"sklearn.svm.SVC",
"matplotlib.patches.Patch",
"numpy.diag",
"matplotlib.pyp... | [((54515, 54541), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(8, 4)'}), '(figsize=(8, 4))\n', (54525, 54541), True, 'import matplotlib.pyplot as plt\n'), ((54549, 54571), 'numpy.arange', 'np.arange', (['(0)', '(2700)', 'dt'], {}), '(0, 2700, dt)\n', (54558, 54571), True, 'import numpy as np\n'), ((55684, 55734), 'matplotlib.patches.Patch', 'mpatches.Patch', ([], {'color': '"""green"""', 'label': '"""without STP"""'}), "(color='green', label='without STP')\n", (55698, 55734), True, 'import matplotlib.patches as mpatches\n'), ((55755, 55804), 'matplotlib.patches.Patch', 'mpatches.Patch', ([], {'color': '"""magenta"""', 'label': '"""with STP"""'}), "(color='magenta', label='with STP')\n", (55769, 55804), True, 'import matplotlib.patches as mpatches\n'), ((55809, 55882), 'matplotlib.pyplot.legend', 'plt.legend', ([], {'loc': '(0)', 'handles': '[green_patch, magenta_patch]', 'prop': "{'size': 6}"}), "(loc=0, handles=[green_patch, magenta_patch], prop={'size': 6})\n", (55819, 55882), True, 'import matplotlib.pyplot as plt\n'), ((56024, 56064), 'numpy.mean', 'np.mean', (['sample_pev1[:, rule, :]'], {'axis': '(0)'}), '(sample_pev1[:, rule, :], axis=0)\n', (56031, 56064), True, 'import numpy as np\n'), ((56071, 56111), 'numpy.mean', 'np.mean', (['sample_pev2[:, rule, :]'], {'axis': '(0)'}), '(sample_pev2[:, rule, :], axis=0)\n', (56078, 56111), True, 'import numpy as np\n'), ((56118, 56144), 'numpy.mean', 'np.mean', (['rule_pev2'], {'axis': '(0)'}), '(rule_pev2, axis=0)\n', (56125, 56144), True, 'import numpy as np\n'), ((57039, 57095), 'numpy.mean', 'np.mean', (["svm_results1['sample_full'][:, rule, :]"], {'axis': '(1)'}), "(svm_results1['sample_full'][:, rule, :], axis=1)\n", (57046, 57095), True, 'import numpy as np\n'), ((57102, 57158), 'numpy.mean', 'np.mean', (["svm_results2['sample_full'][:, rule, :]"], {'axis': '(1)'}), "(svm_results2['sample_full'][:, rule, :], axis=1)\n", (57109, 57158), True, 'import numpy as np\n'), ((57166, 57221), 'numpy.std', 'np.std', (["svm_results1['sample_full'][:, rule, :]"], {'axis': '(1)'}), "(svm_results1['sample_full'][:, rule, :], axis=1)\n", (57172, 57221), True, 'import numpy as np\n'), ((57229, 57284), 'numpy.std', 'np.std', (["svm_results2['sample_full'][:, rule, :]"], {'axis': '(1)'}), "(svm_results2['sample_full'][:, rule, :], axis=1)\n", (57235, 57284), True, 'import numpy as np\n'), ((58674, 58718), 'numpy.mean', 'np.mean', (['sample_stp_pev2[:, rule, :]'], {'axis': '(0)'}), '(sample_stp_pev2[:, rule, :], axis=0)\n', (58681, 58718), True, 'import numpy as np\n'), ((59348, 59408), 'numpy.mean', 'np.mean', (["svm_results2['sample_full_stp'][:, rule, :]"], {'axis': '(1)'}), "(svm_results2['sample_full_stp'][:, rule, :], axis=1)\n", (59355, 59408), True, 'import numpy as np\n'), ((59416, 59475), 'numpy.std', 'np.std', (["svm_results2['sample_full_stp'][:, rule, :]"], {'axis': '(1)'}), "(svm_results2['sample_full_stp'][:, rule, :], axis=1)\n", (59422, 59475), True, 'import numpy as np\n'), ((60006, 60024), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (60022, 60024), True, 'import matplotlib.pyplot as plt\n'), ((60029, 60076), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""DMS comparison.pdf"""'], {'format': '"""pdf"""'}), "('DMS comparison.pdf', format='pdf')\n", (60040, 60076), True, 'import matplotlib.pyplot as plt\n'), ((60081, 60091), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (60089, 60091), True, 'import matplotlib.pyplot as plt\n'), ((61305, 61331), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(8, 9)'}), '(figsize=(8, 9))\n', (61315, 61331), True, 'import matplotlib.pyplot as plt\n'), ((61339, 61361), 'numpy.arange', 'np.arange', (['(0)', '(2700)', 'dt'], {}), '(0, 2700, dt)\n', (61348, 61361), True, 'import numpy as np\n'), ((62504, 62554), 'matplotlib.patches.Patch', 'mpatches.Patch', ([], {'color': '"""green"""', 'label': '"""without STP"""'}), "(color='green', label='without STP')\n", (62518, 62554), True, 'import matplotlib.patches as mpatches\n'), ((62575, 62624), 'matplotlib.patches.Patch', 'mpatches.Patch', ([], {'color': '"""magenta"""', 'label': '"""with STP"""'}), "(color='magenta', label='with STP')\n", (62589, 62624), True, 'import matplotlib.patches as mpatches\n'), ((62629, 62702), 'matplotlib.pyplot.legend', 'plt.legend', ([], {'loc': '(0)', 'handles': '[green_patch, magenta_patch]', 'prop': "{'size': 6}"}), "(loc=0, handles=[green_patch, magenta_patch], prop={'size': 6})\n", (62639, 62702), True, 'import matplotlib.pyplot as plt\n'), ((62845, 62871), 'numpy.mean', 'np.mean', (['rule_pev1'], {'axis': '(0)'}), '(rule_pev1, axis=0)\n', (62852, 62871), True, 'import numpy as np\n'), ((62880, 62906), 'numpy.mean', 'np.mean', (['rule_pev2'], {'axis': '(0)'}), '(rule_pev2, axis=0)\n', (62887, 62906), True, 'import numpy as np\n'), ((66833, 66851), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (66849, 66851), True, 'import matplotlib.pyplot as plt\n'), ((66856, 66909), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""Two rules comparison.pdf"""'], {'format': '"""pdf"""'}), "('Two rules comparison.pdf', format='pdf')\n", (66867, 66909), True, 'import matplotlib.pyplot as plt\n'), ((66914, 66924), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (66922, 66924), True, 'import matplotlib.pyplot as plt\n'), ((68516, 68543), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(6, 10)'}), '(figsize=(6, 10))\n', (68526, 68543), True, 'import matplotlib.pyplot as plt\n'), ((69150, 69168), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (69166, 69168), True, 'import matplotlib.pyplot as plt\n'), ((69173, 69213), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""Summary.pdf"""'], {'format': '"""pdf"""'}), "('Summary.pdf', format='pdf')\n", (69184, 69213), True, 'import matplotlib.pyplot as plt\n'), ((69218, 69228), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (69226, 69228), True, 'import matplotlib.pyplot as plt\n'), ((69592, 69624), 'numpy.int_', 'np.int_', (['(trial_length_steps * dt)'], {}), '(trial_length_steps * dt)\n', (69599, 69624), True, 'import numpy as np\n'), ((69640, 69670), 'numpy.arange', 'np.arange', (['(0)', 'trial_length', 'dt'], {}), '(0, trial_length, dt)\n', (69649, 69670), True, 'import numpy as np\n'), ((70883, 70927), 'numpy.mean', 'np.mean', (["svm_results1['sample_full']"], {'axis': '(1)'}), "(svm_results1['sample_full'], axis=1)\n", (70890, 70927), True, 'import numpy as np\n'), ((70936, 70979), 'numpy.std', 'np.std', (["svm_results1['sample_full']"], {'axis': '(1)'}), "(svm_results1['sample_full'], axis=1)\n", (70942, 70979), True, 'import numpy as np\n'), ((71099, 71143), 'numpy.mean', 'np.mean', (["svm_results2['sample_full']"], {'axis': '(1)'}), "(svm_results2['sample_full'], axis=1)\n", (71106, 71143), True, 'import numpy as np\n'), ((71152, 71195), 'numpy.std', 'np.std', (["svm_results2['sample_full']"], {'axis': '(1)'}), "(svm_results2['sample_full'], axis=1)\n", (71158, 71195), True, 'import numpy as np\n'), ((71583, 71631), 'numpy.mean', 'np.mean', (["svm_results2['sample_full_stp']"], {'axis': '(1)'}), "(svm_results2['sample_full_stp'], axis=1)\n", (71590, 71631), True, 'import numpy as np\n'), ((71640, 71687), 'numpy.std', 'np.std', (["svm_results2['sample_full_stp']"], {'axis': '(1)'}), "(svm_results2['sample_full_stp'], axis=1)\n", (71646, 71687), True, 'import numpy as np\n'), ((72056, 72078), 'numpy.arange', 'np.arange', (['(0)', '(2900)', 'dt'], {}), '(0, 2900, dt)\n', (72065, 72078), True, 'import numpy as np\n'), ((72434, 72478), 'numpy.zeros', 'np.zeros', (['(4, N, trial_length, num_svm_reps)'], {}), '((4, N, trial_length, num_svm_reps))\n', (72442, 72478), True, 'import numpy as np\n'), ((72502, 72546), 'numpy.zeros', 'np.zeros', (['(4, N, trial_length, num_svm_reps)'], {}), '((4, N, trial_length, num_svm_reps))\n', (72510, 72546), True, 'import numpy as np\n'), ((72573, 72617), 'numpy.zeros', 'np.zeros', (['(4, N, trial_length, num_svm_reps)'], {}), '((4, N, trial_length, num_svm_reps))\n', (72581, 72617), True, 'import numpy as np\n'), ((72646, 72690), 'numpy.zeros', 'np.zeros', (['(4, N, trial_length, num_svm_reps)'], {}), '((4, N, trial_length, num_svm_reps))\n', (72654, 72690), True, 'import numpy as np\n'), ((72702, 72718), 'numpy.zeros', 'np.zeros', (['(4, N)'], {}), '((4, N))\n', (72710, 72718), True, 'import numpy as np\n'), ((72746, 72762), 'numpy.zeros', 'np.zeros', (['(4, N)'], {}), '((4, N))\n', (72754, 72762), True, 'import numpy as np\n'), ((72787, 72803), 'numpy.zeros', 'np.zeros', (['(4, N)'], {}), '((4, N))\n', (72795, 72803), True, 'import numpy as np\n'), ((74620, 74646), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(6, 4)'}), '(figsize=(6, 4))\n', (74630, 74646), True, 'import matplotlib.pyplot as plt\n'), ((75405, 75423), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (75421, 75423), True, 'import matplotlib.pyplot as plt\n'), ((75428, 75475), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""Example models.pdf"""'], {'format': '"""pdf"""'}), "('Example models.pdf', format='pdf')\n", (75439, 75475), True, 'import matplotlib.pyplot as plt\n'), ((75484, 75494), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (75492, 75494), True, 'import matplotlib.pyplot as plt\n'), ((75664, 75690), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(6, 4)'}), '(figsize=(6, 4))\n', (75674, 75690), True, 'import matplotlib.pyplot as plt\n'), ((76633, 76651), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (76649, 76651), True, 'import matplotlib.pyplot as plt\n'), ((76656, 76703), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""Average models.pdf"""'], {'format': '"""pdf"""'}), "('Average models.pdf', format='pdf')\n", (76667, 76703), True, 'import matplotlib.pyplot as plt\n'), ((76709, 76719), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (76717, 76719), True, 'import matplotlib.pyplot as plt\n'), ((76899, 76925), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(6, 4)'}), '(figsize=(6, 4))\n', (76909, 76925), True, 'import matplotlib.pyplot as plt\n'), ((77645, 77663), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (77661, 77663), True, 'import matplotlib.pyplot as plt\n'), ((77668, 77711), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""All models.pdf"""'], {'format': '"""pdf"""'}), "('All models.pdf', format='pdf')\n", (77679, 77711), True, 'import matplotlib.pyplot as plt\n'), ((77717, 77727), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (77725, 77727), True, 'import matplotlib.pyplot as plt\n'), ((78125, 78147), 'numpy.arange', 'np.arange', (['(0)', '(2900)', 'dt'], {}), '(0, 2900, dt)\n', (78134, 78147), True, 'import numpy as np\n'), ((78487, 78513), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(3, 2)'}), '(figsize=(3, 2))\n', (78497, 78513), True, 'import matplotlib.pyplot as plt\n'), ((78576, 78614), 'numpy.mean', 'np.mean', (['spike_decode[:, 0, :]'], {'axis': '(1)'}), '(spike_decode[:, 0, :], axis=1)\n', (78583, 78614), True, 'import numpy as np\n'), ((78621, 78658), 'numpy.std', 'np.std', (['spike_decode[:, 0, :]'], {'axis': '(1)'}), '(spike_decode[:, 0, :], axis=1)\n', (78627, 78658), True, 'import numpy as np\n'), ((78740, 78780), 'numpy.mean', 'np.mean', (['synapse_decode[:, 0, :]'], {'axis': '(1)'}), '(synapse_decode[:, 0, :], axis=1)\n', (78747, 78780), True, 'import numpy as np\n'), ((78787, 78826), 'numpy.std', 'np.std', (['synapse_decode[:, 0, :]'], {'axis': '(1)'}), '(synapse_decode[:, 0, :], axis=1)\n', (78793, 78826), True, 'import numpy as np\n'), ((79025, 79043), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (79041, 79043), True, 'import matplotlib.pyplot as plt\n'), ((79048, 79096), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""Var delay model.pdf"""'], {'format': '"""pdf"""'}), "('Var delay model.pdf', format='pdf')\n", (79059, 79096), True, 'import matplotlib.pyplot as plt\n'), ((79102, 79112), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (79110, 79112), True, 'import matplotlib.pyplot as plt\n'), ((79319, 79344), 'numpy.zeros', 'np.zeros', (['(num_delays, N)'], {}), '((num_delays, N))\n', (79327, 79344), True, 'import numpy as np\n'), ((79364, 79389), 'numpy.zeros', 'np.zeros', (['(num_delays, N)'], {}), '((num_delays, N))\n', (79372, 79389), True, 'import numpy as np\n'), ((79401, 79426), 'numpy.zeros', 'np.zeros', (['(num_delays, N)'], {}), '((num_delays, N))\n', (79409, 79426), True, 'import numpy as np\n'), ((80988, 81014), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(3, 2)'}), '(figsize=(3, 2))\n', (80998, 81014), True, 'import matplotlib.pyplot as plt\n'), ((81593, 81615), 'numpy.arange', 'np.arange', (['(0)', '(2900)', 'dt'], {}), '(0, 2900, dt)\n', (81602, 81615), True, 'import numpy as np\n'), ((82142, 82198), 'numpy.zeros', 'np.zeros', (['(num_tasks + 1, N, trial_length, num_svm_reps)'], {}), '((num_tasks + 1, N, trial_length, num_svm_reps))\n', (82150, 82198), True, 'import numpy as np\n'), ((82220, 82276), 'numpy.zeros', 'np.zeros', (['(num_tasks + 1, N, trial_length, num_svm_reps)'], {}), '((num_tasks + 1, N, trial_length, num_svm_reps))\n', (82228, 82276), True, 'import numpy as np\n'), ((82301, 82357), 'numpy.zeros', 'np.zeros', (['(num_tasks + 1, N, trial_length, num_svm_reps)'], {}), '((num_tasks + 1, N, trial_length, num_svm_reps))\n', (82309, 82357), True, 'import numpy as np\n'), ((82384, 82440), 'numpy.zeros', 'np.zeros', (['(num_tasks + 1, N, trial_length, num_svm_reps)'], {}), '((num_tasks + 1, N, trial_length, num_svm_reps))\n', (82392, 82440), True, 'import numpy as np\n'), ((82450, 82478), 'numpy.zeros', 'np.zeros', (['(num_tasks + 1, N)'], {}), '((num_tasks + 1, N))\n', (82458, 82478), True, 'import numpy as np\n'), ((82504, 82532), 'numpy.zeros', 'np.zeros', (['(num_tasks + 1, N)'], {}), '((num_tasks + 1, N))\n', (82512, 82532), True, 'import numpy as np\n'), ((82555, 82583), 'numpy.zeros', 'np.zeros', (['(num_tasks + 1, N)'], {}), '((num_tasks + 1, N))\n', (82563, 82583), True, 'import numpy as np\n'), ((85424, 85462), 'numpy.mean', 'np.mean', (['delay_accuracy[3:, :]'], {'axis': '(0)'}), '(delay_accuracy[3:, :], axis=0)\n', (85431, 85462), True, 'import numpy as np\n'), ((85535, 85571), 'numpy.mean', 'np.mean', (['perf[num_tasks:, :]'], {'axis': '(0)'}), '(perf[num_tasks:, :], axis=0)\n', (85542, 85571), True, 'import numpy as np\n'), ((85634, 85673), 'numpy.zeros', 'np.zeros', (['(num_tasks, 3)'], {'dtype': 'np.int8'}), '((num_tasks, 3), dtype=np.int8)\n', (85642, 85673), True, 'import numpy as np\n'), ((86023, 86051), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(6, 8.5)'}), '(figsize=(6, 8.5))\n', (86033, 86051), True, 'import matplotlib.pyplot as plt\n'), ((87607, 87625), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (87623, 87625), True, 'import matplotlib.pyplot as plt\n'), ((87630, 87671), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""Summary1.pdf"""'], {'format': '"""pdf"""'}), "('Summary1.pdf', format='pdf')\n", (87641, 87671), True, 'import matplotlib.pyplot as plt\n'), ((87680, 87690), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (87688, 87690), True, 'import matplotlib.pyplot as plt\n'), ((88012, 88040), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(6.5, 3)'}), '(figsize=(6.5, 3))\n', (88022, 88040), True, 'import matplotlib.pyplot as plt\n'), ((90430, 90448), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (90446, 90448), True, 'import matplotlib.pyplot as plt\n'), ((90453, 90494), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""Summary2.pdf"""'], {'format': '"""pdf"""'}), "('Summary2.pdf', format='pdf')\n", (90464, 90494), True, 'import matplotlib.pyplot as plt\n'), ((90503, 90513), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (90511, 90513), True, 'import matplotlib.pyplot as plt\n'), ((90916, 90961), 'numpy.logical_or', 'np.logical_or', (['(y[1, :, :] > 0)', '(y[2, :, :] > 0)'], {}), '(y[1, :, :] > 0, y[2, :, :] > 0)\n', (90929, 90961), True, 'import numpy as np\n'), ((90987, 91007), 'numpy.argmax', 'np.argmax', (['y'], {'axis': '(0)'}), '(y, axis=0)\n', (90996, 91007), True, 'import numpy as np\n'), ((91022, 91046), 'numpy.argmax', 'np.argmax', (['y_hat'], {'axis': '(0)'}), '(y_hat, axis=0)\n', (91031, 91046), True, 'import numpy as np\n'), ((542, 570), 'numpy.stack', 'np.stack', (["x['syn_x']"], {'axis': '(2)'}), "(x['syn_x'], axis=2)\n", (550, 570), True, 'import numpy as np\n'), ((591, 619), 'numpy.stack', 'np.stack', (['self.syn_x'], {'axis': '(1)'}), '(self.syn_x, axis=1)\n', (599, 619), True, 'import numpy as np\n'), ((955, 983), 'numpy.stack', 'np.stack', (["x['syn_u']"], {'axis': '(2)'}), "(x['syn_u'], axis=2)\n", (963, 983), True, 'import numpy as np\n'), ((1004, 1032), 'numpy.stack', 'np.stack', (['self.syn_u'], {'axis': '(1)'}), '(self.syn_u, axis=1)\n', (1012, 1032), True, 'import numpy as np\n'), ((1378, 1413), 'numpy.stack', 'np.stack', (["x['hidden_state']"], {'axis': '(2)'}), "(x['hidden_state'], axis=2)\n", (1386, 1413), True, 'import numpy as np\n'), ((1440, 1474), 'numpy.stack', 'np.stack', (['self.rnn_outputs'], {'axis': '(1)'}), '(self.rnn_outputs, axis=1)\n', (1448, 1474), True, 'import numpy as np\n'), ((1590, 1682), 'numpy.reshape', 'np.reshape', (['self.rnn_outputs', '(num_neurons, trial_length, num_blocks * trials_per_block)'], {}), '(self.rnn_outputs, (num_neurons, trial_length, num_blocks *\n trials_per_block))\n', (1600, 1682), True, 'import numpy as np\n'), ((1975, 2012), 'numpy.transpose', 'np.transpose', (['self.train_mask', '(0, 1)'], {}), '(self.train_mask, (0, 1))\n', (1987, 2012), True, 'import numpy as np\n'), ((2117, 2157), 'numpy.transpose', 'np.transpose', (['self.rnn_inputs', '(2, 0, 1)'], {}), '(self.rnn_inputs, (2, 0, 1))\n', (2129, 2157), True, 'import numpy as np\n'), ((2226, 2262), 'numpy.stack', 'np.stack', (["x['model_outputs']"], {'axis': '(2)'}), "(x['model_outputs'], axis=2)\n", (2234, 2262), True, 'import numpy as np\n'), ((2291, 2327), 'numpy.stack', 'np.stack', (['self.model_outputs'], {'axis': '(1)'}), '(self.model_outputs, axis=1)\n', (2299, 2327), True, 'import numpy as np\n'), ((2406, 2500), 'numpy.reshape', 'np.reshape', (['self.model_outputs', '(num_classes, trial_length, num_blocks * trials_per_block)'], {}), '(self.model_outputs, (num_classes, trial_length, num_blocks *\n trials_per_block))\n', (2416, 2500), True, 'import numpy as np\n'), ((4386, 4425), 'numpy.zeros', 'np.zeros', (['(num_input_neurons, num_dirs)'], {}), '((num_input_neurons, num_dirs))\n', (4394, 4425), True, 'import numpy as np\n'), ((4502, 4539), 'numpy.zeros', 'np.zeros', (['(num_rnn_neurons, num_dirs)'], {}), '((num_rnn_neurons, num_dirs))\n', (4510, 4539), True, 'import numpy as np\n'), ((4906, 4940), 'numpy.dot', 'np.dot', (['self.W_in', 'mean_input_resp'], {}), '(self.W_in, mean_input_resp)\n', (4912, 4940), True, 'import numpy as np\n'), ((5117, 5154), 'numpy.zeros', 'np.zeros', (['(num_neurons, trial_length)'], {}), '((num_neurons, trial_length))\n', (5125, 5154), True, 'import numpy as np\n'), ((5176, 5213), 'numpy.zeros', 'np.zeros', (['(num_neurons, trial_length)'], {}), '((num_neurons, trial_length))\n', (5184, 5213), True, 'import numpy as np\n'), ((5235, 5272), 'numpy.zeros', 'np.zeros', (['(num_neurons, trial_length)'], {}), '((num_neurons, trial_length))\n', (5243, 5272), True, 'import numpy as np\n'), ((5291, 5331), 'numpy.zeros', 'np.zeros', (['(num_neurons, 2, trial_length)'], {}), '((num_neurons, 2, trial_length))\n', (5299, 5331), True, 'import numpy as np\n'), ((5351, 5391), 'numpy.zeros', 'np.zeros', (['(num_neurons, 2, trial_length)'], {}), '((num_neurons, 2, trial_length))\n', (5359, 5391), True, 'import numpy as np\n'), ((5411, 5451), 'numpy.zeros', 'np.zeros', (['(num_neurons, 2, trial_length)'], {}), '((num_neurons, 2, trial_length))\n', (5419, 5451), True, 'import numpy as np\n'), ((5482, 5506), 'numpy.ones', 'np.ones', (['(num_trials, 3)'], {}), '((num_trials, 3))\n', (5489, 5506), True, 'import numpy as np\n'), ((5533, 5591), 'numpy.cos', 'np.cos', (['(2 * np.pi * self.sample_dir / self.num_motion_dirs)'], {}), '(2 * np.pi * self.sample_dir / self.num_motion_dirs)\n', (5539, 5591), True, 'import numpy as np\n'), ((5614, 5672), 'numpy.sin', 'np.sin', (['(2 * np.pi * self.sample_dir / self.num_motion_dirs)'], {}), '(2 * np.pi * self.sample_dir / self.num_motion_dirs)\n', (5620, 5672), True, 'import numpy as np\n'), ((5696, 5720), 'numpy.ones', 'np.ones', (['(num_trials, 3)'], {}), '((num_trials, 3))\n', (5703, 5720), True, 'import numpy as np\n'), ((5745, 5801), 'numpy.cos', 'np.cos', (['(2 * np.pi * self.test_dir / self.num_motion_dirs)'], {}), '(2 * np.pi * self.test_dir / self.num_motion_dirs)\n', (5751, 5801), True, 'import numpy as np\n'), ((5822, 5878), 'numpy.sin', 'np.sin', (['(2 * np.pi * self.test_dir / self.num_motion_dirs)'], {}), '(2 * np.pi * self.test_dir / self.num_motion_dirs)\n', (5828, 5878), True, 'import numpy as np\n'), ((7277, 7349), 'numpy.zeros', 'np.zeros', (['(num_neurons, num_neurons, self.num_motion_dirs, trial_length)'], {}), '((num_neurons, num_neurons, self.num_motion_dirs, trial_length))\n', (7285, 7349), True, 'import numpy as np\n'), ((7381, 7440), 'numpy.zeros', 'np.zeros', (['(num_neurons, self.num_motion_dirs, trial_length)'], {}), '((num_neurons, self.num_motion_dirs, trial_length))\n', (7389, 7440), True, 'import numpy as np\n'), ((8311, 8396), 'numpy.zeros', 'np.zeros', (['(num_neurons, self.num_motion_dirs, self.num_motion_dirs, trial_length)'], {}), '((num_neurons, self.num_motion_dirs, self.num_motion_dirs,\n trial_length))\n', (8319, 8396), True, 'import numpy as np\n'), ((8423, 8508), 'numpy.zeros', 'np.zeros', (['(num_neurons, self.num_motion_dirs, self.num_motion_dirs, trial_length)'], {}), '((num_neurons, self.num_motion_dirs, self.num_motion_dirs,\n trial_length))\n', (8431, 8508), True, 'import numpy as np\n'), ((9609, 9660), 'numpy.zeros', 'np.zeros', (['(self.num_rules, self.num_motion_dirs, 2)'], {}), '((self.num_rules, self.num_motion_dirs, 2))\n', (9617, 9660), True, 'import numpy as np\n'), ((9677, 9728), 'numpy.zeros', 'np.zeros', (['(self.num_rules, self.num_motion_dirs, 2)'], {}), '((self.num_rules, self.num_motion_dirs, 2))\n', (9685, 9728), True, 'import numpy as np\n'), ((10758, 10786), 'numpy.zeros', 'np.zeros', (['self.max_num_tests'], {}), '(self.max_num_tests)\n', (10766, 10786), True, 'import numpy as np\n'), ((10824, 10852), 'numpy.zeros', 'np.zeros', (['self.max_num_tests'], {}), '(self.max_num_tests)\n', (10832, 10852), True, 'import numpy as np\n'), ((10876, 10904), 'numpy.zeros', 'np.zeros', (['self.max_num_tests'], {}), '(self.max_num_tests)\n', (10884, 10904), True, 'import numpy as np\n'), ((10934, 10962), 'numpy.zeros', 'np.zeros', (['self.max_num_tests'], {}), '(self.max_num_tests)\n', (10942, 10962), True, 'import numpy as np\n'), ((12520, 12547), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(12, 8)'}), '(figsize=(12, 8))\n', (12530, 12547), True, 'import matplotlib.pyplot as plt\n'), ((12905, 12915), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (12913, 12915), True, 'import matplotlib.pyplot as plt\n'), ((13046, 13140), 'sklearn.svm.SVC', 'svm.SVC', ([], {'C': '(1)', 'kernel': '"""linear"""', 'decision_function_shape': '"""ovr"""', 'shrinking': '(False)', 'tol': '(0.0001)'}), "(C=1, kernel='linear', decision_function_shape='ovr', shrinking=\n False, tol=0.0001)\n", (13053, 13140), False, 'from sklearn import svm\n'), ((13228, 13278), 'numpy.zeros', 'np.zeros', (['(trial_length, self.num_rules, num_reps)'], {}), '((trial_length, self.num_rules, num_reps))\n', (13236, 13278), True, 'import numpy as np\n'), ((13304, 13354), 'numpy.zeros', 'np.zeros', (['(trial_length, self.num_rules, num_reps)'], {}), '((trial_length, self.num_rules, num_reps))\n', (13312, 13354), True, 'import numpy as np\n'), ((13383, 13433), 'numpy.zeros', 'np.zeros', (['(trial_length, self.num_rules, num_reps)'], {}), '((trial_length, self.num_rules, num_reps))\n', (13391, 13433), True, 'import numpy as np\n'), ((13464, 13514), 'numpy.zeros', 'np.zeros', (['(trial_length, self.num_rules, num_reps)'], {}), '((trial_length, self.num_rules, num_reps))\n', (13472, 13514), True, 'import numpy as np\n'), ((15946, 15979), 'numpy.zeros', 'np.zeros', (['(num_neurons, num_lags)'], {}), '((num_neurons, num_lags))\n', (15954, 15979), True, 'import numpy as np\n'), ((16005, 16038), 'numpy.zeros', 'np.zeros', (['(num_neurons, num_lags)'], {}), '((num_neurons, num_lags))\n', (16013, 16038), True, 'import numpy as np\n'), ((16068, 16101), 'numpy.zeros', 'np.zeros', (['(num_neurons, num_lags)'], {}), '((num_neurons, num_lags))\n', (16076, 16101), True, 'import numpy as np\n'), ((17711, 17769), 'numpy.zeros', 'np.zeros', (['(num_neurons, num_rules, num_dirs, trial_length)'], {}), '((num_neurons, num_rules, num_dirs, trial_length))\n', (17719, 17769), True, 'import numpy as np\n'), ((17795, 17843), 'numpy.zeros', 'np.zeros', (['(num_classes, num_rules, trial_length)'], {}), '((num_classes, num_rules, trial_length))\n', (17803, 17843), True, 'import numpy as np\n'), ((17873, 17921), 'numpy.zeros', 'np.zeros', (['(num_classes, num_rules, trial_length)'], {}), '((num_classes, num_rules, trial_length))\n', (17881, 17921), True, 'import numpy as np\n'), ((19073, 19166), 'sklearn.svm.SVC', 'svm.SVC', ([], {'C': '(1)', 'kernel': '"""linear"""', 'decision_function_shape': '"""ovr"""', 'shrinking': '(False)', 'tol': '(1e-05)'}), "(C=1, kernel='linear', decision_function_shape='ovr', shrinking=\n False, tol=1e-05)\n", (19080, 19166), False, 'from sklearn import svm\n'), ((19260, 19309), 'numpy.zeros', 'np.zeros', (['(num_neurons, 2, 2, 2, 2, trial_length)'], {}), '((num_neurons, 2, 2, 2, 2, trial_length))\n', (19268, 19309), True, 'import numpy as np\n'), ((19331, 19380), 'numpy.zeros', 'np.zeros', (['(num_neurons, 2, 2, 2, 2, trial_length)'], {}), '((num_neurons, 2, 2, 2, 2, trial_length))\n', (19339, 19380), True, 'import numpy as np\n'), ((19412, 19458), 'numpy.zeros', 'np.zeros', (['(2, 2, 2, 2, trial_length, num_reps)'], {}), '((2, 2, 2, 2, trial_length, num_reps))\n', (19420, 19458), True, 'import numpy as np\n'), ((19484, 19530), 'numpy.zeros', 'np.zeros', (['(2, 2, 2, 2, trial_length, num_reps)'], {}), '((2, 2, 2, 2, trial_length, num_reps))\n', (19492, 19530), True, 'import numpy as np\n'), ((19558, 19591), 'numpy.zeros', 'np.zeros', (['(2, 2, 3, trial_length)'], {}), '((2, 2, 3, trial_length))\n', (19566, 19591), True, 'import numpy as np\n'), ((22082, 22101), 'numpy.zeros', 'np.zeros', (['num_conds'], {}), '(num_conds)\n', (22090, 22101), True, 'import numpy as np\n'), ((22255, 22291), 'numpy.zeros', 'np.zeros', (['(min_num_trials * num_conds)'], {}), '(min_num_trials * num_conds)\n', (22263, 22291), True, 'import numpy as np\n'), ((22310, 22360), 'numpy.zeros', 'np.zeros', (['(min_num_trials * num_conds, y.shape[1])'], {}), '((min_num_trials * num_conds, y.shape[1]))\n', (22318, 22360), True, 'import numpy as np\n'), ((22655, 22673), 'numpy.zeros', 'np.zeros', (['num_reps'], {}), '(num_reps)\n', (22663, 22673), True, 'import numpy as np\n'), ((23691, 23709), 'numpy.zeros', 'np.zeros', (['num_reps'], {}), '(num_reps)\n', (23699, 23709), True, 'import numpy as np\n'), ((24526, 24579), 'numpy.zeros', 'np.zeros', (['(num_neurons, self.num_rules, trial_length)'], {}), '((num_neurons, self.num_rules, trial_length))\n', (24534, 24579), True, 'import numpy as np\n'), ((24598, 24651), 'numpy.zeros', 'np.zeros', (['(num_neurons, self.num_rules, trial_length)'], {}), '((num_neurons, self.num_rules, trial_length))\n', (24606, 24651), True, 'import numpy as np\n'), ((24670, 24707), 'numpy.zeros', 'np.zeros', (['(num_neurons, trial_length)'], {}), '((num_neurons, trial_length))\n', (24678, 24707), True, 'import numpy as np\n'), ((24727, 24780), 'numpy.zeros', 'np.zeros', (['(num_neurons, self.num_rules, trial_length)'], {}), '((num_neurons, self.num_rules, trial_length))\n', (24735, 24780), True, 'import numpy as np\n'), ((24805, 24858), 'numpy.zeros', 'np.zeros', (['(num_neurons, self.num_rules, trial_length)'], {}), '((num_neurons, self.num_rules, trial_length))\n', (24813, 24858), True, 'import numpy as np\n'), ((24883, 24936), 'numpy.zeros', 'np.zeros', (['(num_neurons, self.num_rules, trial_length)'], {}), '((num_neurons, self.num_rules, trial_length))\n', (24891, 24936), True, 'import numpy as np\n'), ((24965, 25018), 'numpy.zeros', 'np.zeros', (['(num_neurons, self.num_rules, trial_length)'], {}), '((num_neurons, self.num_rules, trial_length))\n', (24973, 25018), True, 'import numpy as np\n'), ((25041, 25094), 'numpy.zeros', 'np.zeros', (['(num_neurons, self.num_rules, trial_length)'], {}), '((num_neurons, self.num_rules, trial_length))\n', (25049, 25094), True, 'import numpy as np\n'), ((26889, 26905), 'numpy.unique', 'np.unique', (['conds'], {}), '(conds)\n', (26898, 26905), True, 'import numpy as np\n'), ((26997, 27013), 'numpy.zeros', 'np.zeros', (['(1, m)'], {}), '((1, m))\n', (27005, 27013), True, 'import numpy as np\n'), ((27030, 27046), 'numpy.zeros', 'np.zeros', (['(1, m)'], {}), '((1, m))\n', (27038, 27046), True, 'import numpy as np\n'), ((27231, 27242), 'numpy.mean', 'np.mean', (['xr'], {}), '(xr)\n', (27238, 27242), True, 'import numpy as np\n'), ((27337, 27358), 'numpy.where', 'np.where', (['(countx == 0)'], {}), '(countx == 0)\n', (27345, 27358), True, 'import numpy as np\n'), ((28058, 28071), 'numpy.isnan', 'np.isnan', (['pev'], {}), '(pev)\n', (28066, 28071), True, 'import numpy as np\n'), ((29117, 29158), 'numpy.mean', 'np.mean', (['sample_pev[:, rule, 30:]'], {'axis': '(1)'}), '(sample_pev[:, rule, 30:], axis=1)\n', (29124, 29158), True, 'import numpy as np\n'), ((29172, 29192), 'numpy.argsort', 'np.argsort', (['mean_pev'], {}), '(mean_pev)\n', (29182, 29192), True, 'import numpy as np\n'), ((29307, 29339), 'numpy.int_', 'np.int_', (['(trial_length_steps * dt)'], {}), '(trial_length_steps * dt)\n', (29314, 29339), True, 'import numpy as np\n'), ((29359, 29389), 'numpy.arange', 'np.arange', (['(0)', 'trial_length', 'dt'], {}), '(0, trial_length, dt)\n', (29368, 29389), True, 'import numpy as np\n'), ((44196, 44222), 'numpy.arange', 'np.arange', (['(0)', '(220 * dt)', 'dt'], {}), '(0, 220 * dt, dt)\n', (44205, 44222), True, 'import numpy as np\n'), ((44453, 44479), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(6, 6)'}), '(figsize=(6, 6))\n', (44463, 44479), True, 'import matplotlib.pyplot as plt\n'), ((45855, 45905), 'numpy.mean', 'np.mean', (['sample_decoding[0, 0, 1, 1, :, :]'], {'axis': '(1)'}), '(sample_decoding[0, 0, 1, 1, :, :], axis=1)\n', (45862, 45905), True, 'import numpy as np\n'), ((45913, 45962), 'numpy.std', 'np.std', (['sample_decoding[0, 0, 1, 1, :, :]'], {'axis': '(1)'}), '(sample_decoding[0, 0, 1, 1, :, :], axis=1)\n', (45919, 45962), True, 'import numpy as np\n'), ((46210, 46260), 'numpy.mean', 'np.mean', (['sample_decoding[0, 0, 0, 1, :, :]'], {'axis': '(1)'}), '(sample_decoding[0, 0, 0, 1, :, :], axis=1)\n', (46217, 46260), True, 'import numpy as np\n'), ((46268, 46317), 'numpy.std', 'np.std', (['sample_decoding[0, 0, 0, 1, :, :]'], {'axis': '(1)'}), '(sample_decoding[0, 0, 0, 1, :, :], axis=1)\n', (46274, 46317), True, 'import numpy as np\n'), ((46768, 46818), 'numpy.mean', 'np.mean', (['sample_decoding[1, 1, 1, 0, :, :]'], {'axis': '(1)'}), '(sample_decoding[1, 1, 1, 0, :, :], axis=1)\n', (46775, 46818), True, 'import numpy as np\n'), ((46826, 46875), 'numpy.std', 'np.std', (['sample_decoding[1, 1, 1, 0, :, :]'], {'axis': '(1)'}), '(sample_decoding[1, 1, 1, 0, :, :], axis=1)\n', (46832, 46875), True, 'import numpy as np\n'), ((47123, 47173), 'numpy.mean', 'np.mean', (['sample_decoding[1, 1, 0, 0, :, :]'], {'axis': '(1)'}), '(sample_decoding[1, 1, 0, 0, :, :], axis=1)\n', (47130, 47173), True, 'import numpy as np\n'), ((47181, 47230), 'numpy.std', 'np.std', (['sample_decoding[1, 1, 0, 0, :, :]'], {'axis': '(1)'}), '(sample_decoding[1, 1, 0, 0, :, :], axis=1)\n', (47187, 47230), True, 'import numpy as np\n'), ((47630, 47648), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (47646, 47648), True, 'import matplotlib.pyplot as plt\n'), ((47657, 47704), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""postle summary.pdf"""'], {'format': '"""pdf"""'}), "('postle summary.pdf', format='pdf')\n", (47668, 47704), True, 'import matplotlib.pyplot as plt\n'), ((47717, 47727), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (47725, 47727), True, 'import matplotlib.pyplot as plt\n'), ((48365, 48397), 'numpy.int_', 'np.int_', (['(trial_length_steps * dt)'], {}), '(trial_length_steps * dt)\n', (48372, 48397), True, 'import numpy as np\n'), ((48417, 48447), 'numpy.arange', 'np.arange', (['(0)', 'trial_length', 'dt'], {}), '(0, trial_length, dt)\n', (48426, 48447), True, 'import numpy as np\n'), ((48613, 48639), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(6, 4)'}), '(figsize=(6, 4))\n', (48623, 48639), True, 'import matplotlib.pyplot as plt\n'), ((48723, 48745), 'numpy.array', 'np.array', (['[0, 1, 2, 3]'], {}), '([0, 1, 2, 3])\n', (48731, 48745), True, 'import numpy as np\n'), ((49205, 49243), 'numpy.mean', 'np.mean', (['mean_resp[:, 0, :, :]'], {'axis': '(0)'}), '(mean_resp[:, 0, :, :], axis=0)\n', (49212, 49243), True, 'import numpy as np\n'), ((49258, 49267), 'numpy.max', 'np.max', (['s'], {}), '(s)\n', (49264, 49267), True, 'import numpy as np\n'), ((49619, 49657), 'numpy.mean', 'np.mean', (['spike_decode[:, 0, :]'], {'axis': '(1)'}), '(spike_decode[:, 0, :], axis=1)\n', (49626, 49657), True, 'import numpy as np\n'), ((49668, 49705), 'numpy.std', 'np.std', (['spike_decode[:, 0, :]'], {'axis': '(1)'}), '(spike_decode[:, 0, :], axis=1)\n', (49674, 49705), True, 'import numpy as np\n'), ((49799, 49842), 'numpy.mean', 'np.mean', (['spike_decode_test[:, 0, :]'], {'axis': '(1)'}), '(spike_decode_test[:, 0, :], axis=1)\n', (49806, 49842), True, 'import numpy as np\n'), ((49853, 49895), 'numpy.std', 'np.std', (['spike_decode_test[:, 0, :]'], {'axis': '(1)'}), '(spike_decode_test[:, 0, :], axis=1)\n', (49859, 49895), True, 'import numpy as np\n'), ((50142, 50180), 'numpy.mean', 'np.mean', (['spike_decode[:, 0, :]'], {'axis': '(1)'}), '(spike_decode[:, 0, :], axis=1)\n', (50149, 50180), True, 'import numpy as np\n'), ((50191, 50228), 'numpy.std', 'np.std', (['spike_decode[:, 0, :]'], {'axis': '(1)'}), '(spike_decode[:, 0, :], axis=1)\n', (50197, 50228), True, 'import numpy as np\n'), ((50322, 50367), 'numpy.mean', 'np.mean', (['synapse_decode_test[:, 0, :]'], {'axis': '(1)'}), '(synapse_decode_test[:, 0, :], axis=1)\n', (50329, 50367), True, 'import numpy as np\n'), ((50378, 50422), 'numpy.std', 'np.std', (['synapse_decode_test[:, 0, :]'], {'axis': '(1)'}), '(synapse_decode_test[:, 0, :], axis=1)\n', (50384, 50422), True, 'import numpy as np\n'), ((50738, 50748), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (50746, 50748), True, 'import matplotlib.pyplot as plt\n'), ((53873, 53905), 'numpy.mean', 'np.mean', (['sample_cat_pev1'], {'axis': '(1)'}), '(sample_cat_pev1, axis=1)\n', (53880, 53905), True, 'import numpy as np\n'), ((53934, 53966), 'numpy.mean', 'np.mean', (['sample_cat_pev2'], {'axis': '(1)'}), '(sample_cat_pev2, axis=1)\n', (53941, 53966), True, 'import numpy as np\n'), ((53999, 54035), 'numpy.mean', 'np.mean', (['sample_cat_stp_pev1'], {'axis': '(1)'}), '(sample_cat_stp_pev1, axis=1)\n', (54006, 54035), True, 'import numpy as np\n'), ((54068, 54104), 'numpy.mean', 'np.mean', (['sample_cat_stp_pev2'], {'axis': '(1)'}), '(sample_cat_stp_pev2, axis=1)\n', (54075, 54104), True, 'import numpy as np\n'), ((54149, 54193), 'numpy.mean', 'np.mean', (["svm_results1['sample_full']"], {'axis': '(1)'}), "(svm_results1['sample_full'], axis=1)\n", (54156, 54193), True, 'import numpy as np\n'), ((54238, 54282), 'numpy.mean', 'np.mean', (["svm_results2['sample_full']"], {'axis': '(1)'}), "(svm_results2['sample_full'], axis=1)\n", (54245, 54282), True, 'import numpy as np\n'), ((54331, 54379), 'numpy.mean', 'np.mean', (["svm_results1['sample_full_stp']"], {'axis': '(1)'}), "(svm_results1['sample_full_stp'], axis=1)\n", (54338, 54379), True, 'import numpy as np\n'), ((54428, 54476), 'numpy.mean', 'np.mean', (["svm_results2['sample_full_stp']"], {'axis': '(1)'}), "(svm_results2['sample_full_stp'], axis=1)\n", (54435, 54476), True, 'import numpy as np\n'), ((54606, 54625), 'numpy.where', 'np.where', (['(t == -500)'], {}), '(t == -500)\n', (54614, 54625), True, 'import numpy as np\n'), ((54625, 54641), 'numpy.where', 'np.where', (['(t == 0)'], {}), '(t == 0)\n', (54633, 54641), True, 'import numpy as np\n'), ((54640, 54658), 'numpy.where', 'np.where', (['(t == 500)'], {}), '(t == 500)\n', (54648, 54658), True, 'import numpy as np\n'), ((54657, 54676), 'numpy.where', 'np.where', (['(t == 1500)'], {}), '(t == 1500)\n', (54665, 54676), True, 'import numpy as np\n'), ((54705, 54751), 'numpy.arange', 'np.arange', (['(0)', '(200 + 500 + 1500 + 300 + 300)', 'dt'], {}), '(0, 200 + 500 + 1500 + 300 + 300, dt)\n', (54714, 54751), True, 'import numpy as np\n'), ((54754, 54794), 'numpy.arange', 'np.arange', (['(0)', '(200 + 500 + 250 + 2000)', 'dt'], {}), '(0, 200 + 500 + 250 + 2000, dt)\n', (54763, 54794), True, 'import numpy as np\n'), ((55043, 55085), 'numpy.mean', 'np.mean', (['mean_resp1[:, rule, :, :]'], {'axis': '(1)'}), '(mean_resp1[:, rule, :, :], axis=1)\n', (55050, 55085), True, 'import numpy as np\n'), ((55107, 55149), 'numpy.mean', 'np.mean', (['mean_resp2[:, rule, :, :]'], {'axis': '(1)'}), '(mean_resp2[:, rule, :, :], axis=1)\n', (55114, 55149), True, 'import numpy as np\n'), ((55218, 55246), 'numpy.sqrt', 'np.sqrt', (['mean_resp1.shape[0]'], {}), '(mean_resp1.shape[0])\n', (55225, 55246), True, 'import numpy as np\n'), ((55311, 55339), 'numpy.sqrt', 'np.sqrt', (['mean_resp1.shape[0]'], {}), '(mean_resp1.shape[0])\n', (55318, 55339), True, 'import numpy as np\n'), ((56154, 56193), 'numpy.std', 'np.std', (['sample_pev1[:, rule, :]'], {'axis': '(0)'}), '(sample_pev1[:, rule, :], axis=0)\n', (56160, 56193), True, 'import numpy as np\n'), ((56191, 56220), 'numpy.sqrt', 'np.sqrt', (['sample_pev1.shape[0]'], {}), '(sample_pev1.shape[0])\n', (56198, 56220), True, 'import numpy as np\n'), ((56231, 56270), 'numpy.std', 'np.std', (['sample_pev2[:, rule, :]'], {'axis': '(0)'}), '(sample_pev2[:, rule, :], axis=0)\n', (56237, 56270), True, 'import numpy as np\n'), ((56268, 56297), 'numpy.sqrt', 'np.sqrt', (['sample_pev1.shape[0]'], {}), '(sample_pev1.shape[0])\n', (56275, 56297), True, 'import numpy as np\n'), ((57290, 57303), 'numpy.isnan', 'np.isnan', (['se1'], {}), '(se1)\n', (57298, 57303), True, 'import numpy as np\n'), ((57317, 57330), 'numpy.isnan', 'np.isnan', (['se2'], {}), '(se2)\n', (57325, 57330), True, 'import numpy as np\n'), ((58726, 58769), 'numpy.std', 'np.std', (['sample_stp_pev2[:, rule, :]'], {'axis': '(0)'}), '(sample_stp_pev2[:, rule, :], axis=0)\n', (58732, 58769), True, 'import numpy as np\n'), ((58767, 58796), 'numpy.sqrt', 'np.sqrt', (['sample_pev1.shape[0]'], {}), '(sample_pev1.shape[0])\n', (58774, 58796), True, 'import numpy as np\n'), ((59481, 59494), 'numpy.isnan', 'np.isnan', (['se2'], {}), '(se2)\n', (59489, 59494), True, 'import numpy as np\n'), ((61391, 61410), 'numpy.where', 'np.where', (['(t == -500)'], {}), '(t == -500)\n', (61399, 61410), True, 'import numpy as np\n'), ((61410, 61426), 'numpy.where', 'np.where', (['(t == 0)'], {}), '(t == 0)\n', (61418, 61426), True, 'import numpy as np\n'), ((61425, 61443), 'numpy.where', 'np.where', (['(t == 500)'], {}), '(t == 500)\n', (61433, 61443), True, 'import numpy as np\n'), ((61442, 61461), 'numpy.where', 'np.where', (['(t == 1500)'], {}), '(t == 1500)\n', (61450, 61461), True, 'import numpy as np\n'), ((61485, 61531), 'numpy.arange', 'np.arange', (['(0)', '(200 + 500 + 1500 + 300 + 300)', 'dt'], {}), '(0, 200 + 500 + 1500 + 300 + 300, dt)\n', (61494, 61531), True, 'import numpy as np\n'), ((62012, 62040), 'numpy.sqrt', 'np.sqrt', (['mean_resp1.shape[0]'], {}), '(mean_resp1.shape[0])\n', (62019, 62040), True, 'import numpy as np\n'), ((62118, 62146), 'numpy.sqrt', 'np.sqrt', (['mean_resp1.shape[0]'], {}), '(mean_resp1.shape[0])\n', (62125, 62146), True, 'import numpy as np\n'), ((62916, 62941), 'numpy.std', 'np.std', (['rule_pev1'], {'axis': '(0)'}), '(rule_pev1, axis=0)\n', (62922, 62941), True, 'import numpy as np\n'), ((62941, 62970), 'numpy.sqrt', 'np.sqrt', (['sample_pev1.shape[0]'], {}), '(sample_pev1.shape[0])\n', (62948, 62970), True, 'import numpy as np\n'), ((62981, 63006), 'numpy.std', 'np.std', (['rule_pev2'], {'axis': '(0)'}), '(rule_pev2, axis=0)\n', (62987, 63006), True, 'import numpy as np\n'), ((63006, 63035), 'numpy.sqrt', 'np.sqrt', (['sample_pev1.shape[0]'], {}), '(sample_pev1.shape[0])\n', (63013, 63035), True, 'import numpy as np\n'), ((63565, 63605), 'numpy.mean', 'np.mean', (['sample_pev1[:, rule, :]'], {'axis': '(0)'}), '(sample_pev1[:, rule, :], axis=0)\n', (63572, 63605), True, 'import numpy as np\n'), ((63616, 63656), 'numpy.mean', 'np.mean', (['sample_pev2[:, rule, :]'], {'axis': '(0)'}), '(sample_pev2[:, rule, :], axis=0)\n', (63623, 63656), True, 'import numpy as np\n'), ((64353, 64409), 'numpy.mean', 'np.mean', (["svm_results1['sample_full'][:, rule, :]"], {'axis': '(1)'}), "(svm_results1['sample_full'][:, rule, :], axis=1)\n", (64360, 64409), True, 'import numpy as np\n'), ((64420, 64476), 'numpy.mean', 'np.mean', (["svm_results2['sample_full'][:, rule, :]"], {'axis': '(1)'}), "(svm_results2['sample_full'][:, rule, :], axis=1)\n", (64427, 64476), True, 'import numpy as np\n'), ((64488, 64543), 'numpy.std', 'np.std', (["svm_results1['sample_full'][:, rule, :]"], {'axis': '(1)'}), "(svm_results1['sample_full'][:, rule, :], axis=1)\n", (64494, 64543), True, 'import numpy as np\n'), ((64555, 64610), 'numpy.std', 'np.std', (["svm_results2['sample_full'][:, rule, :]"], {'axis': '(1)'}), "(svm_results2['sample_full'][:, rule, :], axis=1)\n", (64561, 64610), True, 'import numpy as np\n'), ((65218, 65262), 'numpy.mean', 'np.mean', (['sample_stp_pev1[:, rule, :]'], {'axis': '(0)'}), '(sample_stp_pev1[:, rule, :], axis=0)\n', (65225, 65262), True, 'import numpy as np\n'), ((65273, 65317), 'numpy.mean', 'np.mean', (['sample_stp_pev2[:, rule, :]'], {'axis': '(0)'}), '(sample_stp_pev2[:, rule, :], axis=0)\n', (65280, 65317), True, 'import numpy as np\n'), ((66024, 66084), 'numpy.mean', 'np.mean', (["svm_results1['sample_full_stp'][:, rule, :]"], {'axis': '(1)'}), "(svm_results1['sample_full_stp'][:, rule, :], axis=1)\n", (66031, 66084), True, 'import numpy as np\n'), ((66095, 66155), 'numpy.mean', 'np.mean', (["svm_results2['sample_full_stp'][:, rule, :]"], {'axis': '(1)'}), "(svm_results2['sample_full_stp'][:, rule, :], axis=1)\n", (66102, 66155), True, 'import numpy as np\n'), ((66167, 66226), 'numpy.std', 'np.std', (["svm_results1['sample_full_stp'][:, rule, :]"], {'axis': '(1)'}), "(svm_results1['sample_full_stp'][:, rule, :], axis=1)\n", (66173, 66226), True, 'import numpy as np\n'), ((66238, 66297), 'numpy.std', 'np.std', (["svm_results2['sample_full_stp'][:, rule, :]"], {'axis': '(1)'}), "(svm_results2['sample_full_stp'][:, rule, :], axis=1)\n", (66244, 66297), True, 'import numpy as np\n'), ((67093, 67121), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(10, 10)'}), '(figsize=(10, 10))\n', (67103, 67121), True, 'import matplotlib.pyplot as plt\n'), ((67167, 67194), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(8, 10)'}), '(figsize=(8, 10))\n', (67177, 67194), True, 'import matplotlib.pyplot as plt\n'), ((70096, 70140), 'numpy.mean', 'np.mean', (["svm_results1['sample_full']"], {'axis': '(1)'}), "(svm_results1['sample_full'], axis=1)\n", (70103, 70140), True, 'import numpy as np\n'), ((70178, 70222), 'numpy.mean', 'np.mean', (["svm_results2['sample_full']"], {'axis': '(1)'}), "(svm_results2['sample_full'], axis=1)\n", (70185, 70222), True, 'import numpy as np\n'), ((70264, 70312), 'numpy.mean', 'np.mean', (["svm_results1['sample_full_stp']"], {'axis': '(1)'}), "(svm_results1['sample_full_stp'], axis=1)\n", (70271, 70312), True, 'import numpy as np\n'), ((70354, 70402), 'numpy.mean', 'np.mean', (["svm_results2['sample_full_stp']"], {'axis': '(1)'}), "(svm_results2['sample_full_stp'], axis=1)\n", (70361, 70402), True, 'import numpy as np\n'), ((70450, 70498), 'numpy.squeeze', 'np.squeeze', (["svm_results1['sample_full'][:, 0, :]"], {}), "(svm_results1['sample_full'][:, 0, :])\n", (70460, 70498), True, 'import numpy as np\n'), ((70535, 70583), 'numpy.squeeze', 'np.squeeze', (["svm_results2['sample_full'][:, 0, :]"], {}), "(svm_results2['sample_full'][:, 0, :])\n", (70545, 70583), True, 'import numpy as np\n'), ((70624, 70676), 'numpy.squeeze', 'np.squeeze', (["svm_results1['sample_full_stp'][:, 0, :]"], {}), "(svm_results1['sample_full_stp'][:, 0, :])\n", (70634, 70676), True, 'import numpy as np\n'), ((70717, 70769), 'numpy.squeeze', 'np.squeeze', (["svm_results2['sample_full_stp'][:, 0, :]"], {}), "(svm_results2['sample_full_stp'][:, 0, :])\n", (70727, 70769), True, 'import numpy as np\n'), ((72108, 72127), 'numpy.where', 'np.where', (['(t == -500)'], {}), '(t == -500)\n', (72116, 72127), True, 'import numpy as np\n'), ((72127, 72143), 'numpy.where', 'np.where', (['(t == 0)'], {}), '(t == 0)\n', (72135, 72143), True, 'import numpy as np\n'), ((72142, 72160), 'numpy.where', 'np.where', (['(t == 500)'], {}), '(t == 500)\n', (72150, 72160), True, 'import numpy as np\n'), ((72159, 72178), 'numpy.where', 'np.where', (['(t == 1500)'], {}), '(t == 1500)\n', (72167, 72178), True, 'import numpy as np\n'), ((74085, 74128), 'numpy.mean', 'np.mean', (['spike_decoding[:, :, d, :]'], {'axis': '(3)'}), '(spike_decoding[:, :, d, :], axis=3)\n', (74092, 74128), True, 'import numpy as np\n'), ((74816, 74872), 'numpy.mean', 'np.mean', (['spike_decoding[j, ind_example[j], :, :]'], {'axis': '(1)'}), '(spike_decoding[j, ind_example[j], :, :], axis=1)\n', (74823, 74872), True, 'import numpy as np\n'), ((74882, 74937), 'numpy.std', 'np.std', (['spike_decoding[j, ind_example[j], :, :]'], {'axis': '(1)'}), '(spike_decoding[j, ind_example[j], :, :], axis=1)\n', (74888, 74937), True, 'import numpy as np\n'), ((75030, 75088), 'numpy.mean', 'np.mean', (['synapse_decoding[j, ind_example[j], :, :]'], {'axis': '(1)'}), '(synapse_decoding[j, ind_example[j], :, :], axis=1)\n', (75037, 75088), True, 'import numpy as np\n'), ((75098, 75155), 'numpy.std', 'np.std', (['synapse_decoding[j, ind_example[j], :, :]'], {'axis': '(1)'}), '(synapse_decoding[j, ind_example[j], :, :], axis=1)\n', (75104, 75155), True, 'import numpy as np\n'), ((77152, 77211), 'numpy.mean', 'np.mean', (['synapse_decoding[j, ind_good_models, :, :]'], {'axis': '(2)'}), '(synapse_decoding[j, ind_good_models, :, :], axis=2)\n', (77159, 77211), True, 'import numpy as np\n'), ((81645, 81664), 'numpy.where', 'np.where', (['(t == -500)'], {}), '(t == -500)\n', (81653, 81664), True, 'import numpy as np\n'), ((81664, 81680), 'numpy.where', 'np.where', (['(t == 0)'], {}), '(t == 0)\n', (81672, 81680), True, 'import numpy as np\n'), ((81679, 81697), 'numpy.where', 'np.where', (['(t == 500)'], {}), '(t == 500)\n', (81687, 81697), True, 'import numpy as np\n'), ((81696, 81715), 'numpy.where', 'np.where', (['(t == 1500)'], {}), '(t == 1500)\n', (81704, 81715), True, 'import numpy as np\n'), ((85174, 85217), 'numpy.mean', 'np.mean', (['spike_decoding[:, :, d, :]'], {'axis': '(3)'}), '(spike_decoding[:, :, d, :], axis=3)\n', (85181, 85217), True, 'import numpy as np\n'), ((85785, 85829), 'numpy.argsort', 'np.argsort', (['delay_accuracy[j, ind_good_perf]'], {}), '(delay_accuracy[j, ind_good_perf])\n', (85795, 85829), True, 'import numpy as np\n'), ((91109, 91121), 'numpy.sum', 'np.sum', (['mask'], {}), '(mask)\n', (91115, 91121), True, 'import numpy as np\n'), ((812, 898), 'numpy.reshape', 'np.reshape', (['self.syn_x', '(num_neurons, trial_length, num_blocks * trials_per_block)'], {}), '(self.syn_x, (num_neurons, trial_length, num_blocks *\n trials_per_block))\n', (822, 898), True, 'import numpy as np\n'), ((1225, 1311), 'numpy.reshape', 'np.reshape', (['self.syn_u', '(num_neurons, trial_length, num_blocks * trials_per_block)'], {}), '(self.syn_u, (num_neurons, trial_length, num_blocks *\n trials_per_block))\n', (1235, 1311), True, 'import numpy as np\n'), ((1826, 1871), 'numpy.transpose', 'np.transpose', (['self.desired_outputs', '(2, 0, 1)'], {}), '(self.desired_outputs, (2, 0, 1))\n', (1838, 1871), True, 'import numpy as np\n'), ((4593, 4668), 'numpy.where', 'np.where', (['((self.rule == self.possible_rules[rule]) * (self.sample_dir == d))'], {}), '((self.rule == self.possible_rules[rule]) * (self.sample_dir == d))\n', (4601, 4668), True, 'import numpy as np\n'), ((4770, 4816), 'numpy.mean', 'np.mean', (['self.rnn_inputs[:, :, ind[0]]'], {'axis': '(2)'}), '(self.rnn_inputs[:, :, ind[0]], axis=2)\n', (4777, 4816), True, 'import numpy as np\n'), ((4849, 4882), 'numpy.mean', 'np.mean', (['s[:, sample_rng]'], {'axis': '(1)'}), '(s[:, sample_rng], axis=1)\n', (4856, 4882), True, 'import numpy as np\n'), ((7639, 7701), 'numpy.mean', 'np.mean', (['(self.syn_u[:, :, ind] * self.syn_x[:, :, ind])'], {'axis': '(2)'}), '(self.syn_u[:, :, ind] * self.syn_x[:, :, ind], axis=2)\n', (7646, 7701), True, 'import numpy as np\n'), ((7745, 7766), 'numpy.diag', 'np.diag', (['self.EI_list'], {}), '(self.EI_list)\n', (7752, 7766), True, 'import numpy as np\n'), ((16160, 16178), 'numpy.zeros', 'np.zeros', (['num_lags'], {}), '(num_lags)\n', (16168, 16178), True, 'import numpy as np\n'), ((22167, 22185), 'numpy.sum', 'np.sum', (['(conds == i)'], {}), '(conds == i)\n', (22173, 22185), True, 'import numpy as np\n'), ((22213, 22231), 'numpy.min', 'np.min', (['num_trials'], {}), '(num_trials)\n', (22219, 22231), True, 'import numpy as np\n'), ((23385, 23401), 'numpy.unique', 'np.unique', (['conds'], {}), '(conds)\n', (23394, 23401), True, 'import numpy as np\n'), ((23415, 23428), 'numpy.squeeze', 'np.squeeze', (['y'], {}), '(y)\n', (23425, 23428), True, 'import numpy as np\n'), ((26973, 26983), 'numpy.mean', 'np.mean', (['x'], {}), '(x)\n', (26980, 26983), True, 'import numpy as np\n'), ((27119, 27139), 'numpy.where', 'np.where', (['(conds == i)'], {}), '(conds == i)\n', (27127, 27139), True, 'import numpy as np\n'), ((27198, 27217), 'numpy.mean', 'np.mean', (['xr[ind[0]]'], {}), '(xr[ind[0]])\n', (27205, 27217), True, 'import numpy as np\n'), ((27257, 27275), 'numpy.sum', 'np.sum', (['(countx > 0)'], {}), '(countx > 0)\n', (27263, 27275), True, 'import numpy as np\n'), ((27405, 27426), 'numpy.transpose', 'np.transpose', (['(xc ** 2)'], {}), '(xc ** 2)\n', (27417, 27426), True, 'import numpy as np\n'), ((27475, 27496), 'numpy.transpose', 'np.transpose', (['(xr - gm)'], {}), '(xr - gm)\n', (27487, 27496), True, 'import numpy as np\n'), ((29727, 29764), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(8, 2 * num_rows)'}), '(figsize=(8, 2 * num_rows))\n', (29737, 29764), True, 'import matplotlib.pyplot as plt\n'), ((29924, 29946), 'numpy.array', 'np.array', (['[0, 1, 2, 3]'], {}), '([0, 1, 2, 3])\n', (29932, 29946), True, 'import numpy as np\n'), ((32550, 32566), 'numpy.max', 'np.max', (['rule_pev'], {}), '(rule_pev)\n', (32556, 32566), True, 'import numpy as np\n'), ((33534, 33575), 'numpy.mean', 'np.mean', (['mean_resp[:, rule, :, :]'], {'axis': '(0)'}), '(mean_resp[:, rule, :, :], axis=0)\n', (33541, 33575), True, 'import numpy as np\n'), ((33594, 33603), 'numpy.max', 'np.max', (['s'], {}), '(s)\n', (33600, 33603), True, 'import numpy as np\n'), ((36069, 36083), 'matplotlib.pyplot.hold', 'plt.hold', (['(True)'], {}), '(True)\n', (36077, 36083), True, 'import matplotlib.pyplot as plt\n'), ((36100, 36136), 'numpy.mean', 'np.mean', (['sample_pev[:, 0, :]'], {'axis': '(0)'}), '(sample_pev[:, 0, :], axis=0)\n', (36107, 36136), True, 'import numpy as np\n'), ((36268, 36277), 'numpy.max', 'np.max', (['u'], {}), '(u)\n', (36274, 36277), True, 'import numpy as np\n'), ((36357, 36393), 'numpy.mean', 'np.mean', (['sample_pev[:, 1, :]'], {'axis': '(0)'}), '(sample_pev[:, 1, :], axis=0)\n', (36364, 36393), True, 'import numpy as np\n'), ((36524, 36533), 'numpy.max', 'np.max', (['u'], {}), '(u)\n', (36530, 36533), True, 'import numpy as np\n'), ((36559, 36592), 'numpy.max', 'np.max', (['[sample_max, sample_max1]'], {}), '([sample_max, sample_max1])\n', (36565, 36592), True, 'import numpy as np\n'), ((36749, 36788), 'numpy.mean', 'np.mean', (['sample_pev[:, rule, :]'], {'axis': '(0)'}), '(sample_pev[:, rule, :], axis=0)\n', (36756, 36788), True, 'import numpy as np\n'), ((36922, 36931), 'numpy.max', 'np.max', (['u'], {}), '(u)\n', (36928, 36931), True, 'import numpy as np\n'), ((37360, 37398), 'numpy.mean', 'np.mean', (['spike_decode[:, 0, :]'], {'axis': '(1)'}), '(spike_decode[:, 0, :], axis=1)\n', (37367, 37398), True, 'import numpy as np\n'), ((37413, 37450), 'numpy.std', 'np.std', (['spike_decode[:, 0, :]'], {'axis': '(1)'}), '(spike_decode[:, 0, :], axis=1)\n', (37419, 37450), True, 'import numpy as np\n'), ((37556, 37594), 'numpy.mean', 'np.mean', (['spike_decode[:, 1, :]'], {'axis': '(1)'}), '(spike_decode[:, 1, :], axis=1)\n', (37563, 37594), True, 'import numpy as np\n'), ((37609, 37646), 'numpy.std', 'np.std', (['spike_decode[:, 1, :]'], {'axis': '(1)'}), '(spike_decode[:, 1, :], axis=1)\n', (37615, 37646), True, 'import numpy as np\n'), ((37910, 37951), 'numpy.mean', 'np.mean', (['spike_decode[:, rule, :]'], {'axis': '(1)'}), '(spike_decode[:, rule, :], axis=1)\n', (37917, 37951), True, 'import numpy as np\n'), ((37966, 38006), 'numpy.std', 'np.std', (['spike_decode[:, rule, :]'], {'axis': '(1)'}), '(spike_decode[:, rule, :], axis=1)\n', (37972, 38006), True, 'import numpy as np\n'), ((38112, 38158), 'numpy.mean', 'np.mean', (['spike_decode_test[:, rule, :]'], {'axis': '(1)'}), '(spike_decode_test[:, rule, :], axis=1)\n', (38119, 38158), True, 'import numpy as np\n'), ((38173, 38218), 'numpy.std', 'np.std', (['spike_decode_test[:, rule, :]'], {'axis': '(1)'}), '(spike_decode_test[:, rule, :], axis=1)\n', (38179, 38218), True, 'import numpy as np\n'), ((43851, 43861), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (43859, 43861), True, 'import matplotlib.pyplot as plt\n'), ((44312, 44331), 'numpy.where', 'np.where', (['(t == -500)'], {}), '(t == -500)\n', (44320, 44331), True, 'import numpy as np\n'), ((44331, 44347), 'numpy.where', 'np.where', (['(t == 0)'], {}), '(t == 0)\n', (44339, 44347), True, 'import numpy as np\n'), ((44347, 44365), 'numpy.where', 'np.where', (['(t == 500)'], {}), '(t == 500)\n', (44355, 44365), True, 'import numpy as np\n'), ((44365, 44384), 'numpy.where', 'np.where', (['(t == 1000)'], {}), '(t == 1000)\n', (44373, 44384), True, 'import numpy as np\n'), ((44384, 44403), 'numpy.where', 'np.where', (['(t == 1500)'], {}), '(t == 1500)\n', (44392, 44403), True, 'import numpy as np\n'), ((44403, 44422), 'numpy.where', 'np.where', (['(t == 2000)'], {}), '(t == 2000)\n', (44411, 44422), True, 'import numpy as np\n'), ((44422, 44441), 'numpy.where', 'np.where', (['(t == 2500)'], {}), '(t == 2500)\n', (44430, 44441), True, 'import numpy as np\n'), ((48531, 48550), 'numpy.where', 'np.where', (['(t == -500)'], {}), '(t == -500)\n', (48539, 48550), True, 'import numpy as np\n'), ((48550, 48566), 'numpy.where', 'np.where', (['(t == 0)'], {}), '(t == 0)\n', (48558, 48566), True, 'import numpy as np\n'), ((48565, 48583), 'numpy.where', 'np.where', (['(t == 500)'], {}), '(t == 500)\n', (48573, 48583), True, 'import numpy as np\n'), ((48582, 48601), 'numpy.where', 'np.where', (['(t == 1500)'], {}), '(t == 1500)\n', (48590, 48601), True, 'import numpy as np\n'), ((50653, 50671), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (50669, 50671), True, 'import matplotlib.pyplot as plt\n'), ((50684, 50729), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""ABBA summary.pdf"""'], {'format': '"""pdf"""'}), "('ABBA summary.pdf', format='pdf')\n", (50695, 50729), True, 'import matplotlib.pyplot as plt\n'), ((54835, 54854), 'numpy.where', 'np.where', (['(t == -500)'], {}), '(t == -500)\n', (54843, 54854), True, 'import numpy as np\n'), ((54854, 54870), 'numpy.where', 'np.where', (['(t == 0)'], {}), '(t == 0)\n', (54862, 54870), True, 'import numpy as np\n'), ((54869, 54887), 'numpy.where', 'np.where', (['(t == 500)'], {}), '(t == 500)\n', (54877, 54887), True, 'import numpy as np\n'), ((54886, 54905), 'numpy.where', 'np.where', (['(t == 1000)'], {}), '(t == 1000)\n', (54894, 54905), True, 'import numpy as np\n'), ((54904, 54923), 'numpy.where', 'np.where', (['(t == 1500)'], {}), '(t == 1500)\n', (54912, 54923), True, 'import numpy as np\n'), ((54922, 54941), 'numpy.where', 'np.where', (['(t == 2000)'], {}), '(t == 2000)\n', (54930, 54941), True, 'import numpy as np\n'), ((54941, 54960), 'numpy.where', 'np.where', (['(t == 2500)'], {}), '(t == 2500)\n', (54949, 54960), True, 'import numpy as np\n'), ((55171, 55213), 'numpy.mean', 'np.mean', (['mean_resp1[:, rule, :, :]'], {'axis': '(1)'}), '(mean_resp1[:, rule, :, :], axis=1)\n', (55178, 55213), True, 'import numpy as np\n'), ((55264, 55306), 'numpy.mean', 'np.mean', (['mean_resp2[:, rule, :, :]'], {'axis': '(1)'}), '(mean_resp2[:, rule, :, :], axis=1)\n', (55271, 55306), True, 'import numpy as np\n'), ((61573, 61592), 'numpy.where', 'np.where', (['(t == -500)'], {}), '(t == -500)\n', (61581, 61592), True, 'import numpy as np\n'), ((61592, 61608), 'numpy.where', 'np.where', (['(t == 0)'], {}), '(t == 0)\n', (61600, 61608), True, 'import numpy as np\n'), ((61607, 61625), 'numpy.where', 'np.where', (['(t == 300)'], {}), '(t == 300)\n', (61615, 61625), True, 'import numpy as np\n'), ((61624, 61642), 'numpy.where', 'np.where', (['(t == 600)'], {}), '(t == 600)\n', (61632, 61642), True, 'import numpy as np\n'), ((61641, 61659), 'numpy.where', 'np.where', (['(t == 900)'], {}), '(t == 900)\n', (61649, 61659), True, 'import numpy as np\n'), ((61658, 61677), 'numpy.where', 'np.where', (['(t == 1200)'], {}), '(t == 1200)\n', (61666, 61677), True, 'import numpy as np\n'), ((61677, 61696), 'numpy.where', 'np.where', (['(t == 1500)'], {}), '(t == 1500)\n', (61685, 61696), True, 'import numpy as np\n'), ((61696, 61715), 'numpy.where', 'np.where', (['(t == 1800)'], {}), '(t == 1800)\n', (61704, 61715), True, 'import numpy as np\n'), ((61806, 61845), 'numpy.mean', 'np.mean', (['mean_resp1[:, :, :, :]'], {'axis': '(1)'}), '(mean_resp1[:, :, :, :], axis=1)\n', (61813, 61845), True, 'import numpy as np\n'), ((61883, 61922), 'numpy.mean', 'np.mean', (['mean_resp2[:, :, :, :]'], {'axis': '(1)'}), '(mean_resp2[:, :, :, :], axis=1)\n', (61890, 61922), True, 'import numpy as np\n'), ((63668, 63707), 'numpy.std', 'np.std', (['sample_pev1[:, rule, :]'], {'axis': '(0)'}), '(sample_pev1[:, rule, :], axis=0)\n', (63674, 63707), True, 'import numpy as np\n'), ((63705, 63734), 'numpy.sqrt', 'np.sqrt', (['sample_pev1.shape[0]'], {}), '(sample_pev1.shape[0])\n', (63712, 63734), True, 'import numpy as np\n'), ((63749, 63788), 'numpy.std', 'np.std', (['sample_pev2[:, rule, :]'], {'axis': '(0)'}), '(sample_pev2[:, rule, :], axis=0)\n', (63755, 63788), True, 'import numpy as np\n'), ((63786, 63815), 'numpy.sqrt', 'np.sqrt', (['sample_pev1.shape[0]'], {}), '(sample_pev1.shape[0])\n', (63793, 63815), True, 'import numpy as np\n'), ((64620, 64633), 'numpy.isnan', 'np.isnan', (['se1'], {}), '(se1)\n', (64628, 64633), True, 'import numpy as np\n'), ((64651, 64664), 'numpy.isnan', 'np.isnan', (['se2'], {}), '(se2)\n', (64659, 64664), True, 'import numpy as np\n'), ((65329, 65372), 'numpy.std', 'np.std', (['sample_stp_pev1[:, rule, :]'], {'axis': '(0)'}), '(sample_stp_pev1[:, rule, :], axis=0)\n', (65335, 65372), True, 'import numpy as np\n'), ((65370, 65399), 'numpy.sqrt', 'np.sqrt', (['sample_pev1.shape[0]'], {}), '(sample_pev1.shape[0])\n', (65377, 65399), True, 'import numpy as np\n'), ((65414, 65457), 'numpy.std', 'np.std', (['sample_stp_pev2[:, rule, :]'], {'axis': '(0)'}), '(sample_stp_pev2[:, rule, :], axis=0)\n', (65420, 65457), True, 'import numpy as np\n'), ((65455, 65484), 'numpy.sqrt', 'np.sqrt', (['sample_pev1.shape[0]'], {}), '(sample_pev1.shape[0])\n', (65462, 65484), True, 'import numpy as np\n'), ((66307, 66320), 'numpy.isnan', 'np.isnan', (['se1'], {}), '(se1)\n', (66315, 66320), True, 'import numpy as np\n'), ((66338, 66351), 'numpy.isnan', 'np.isnan', (['se2'], {}), '(se2)\n', (66346, 66351), True, 'import numpy as np\n'), ((73761, 73815), 'contrib_to_behavior.Analysis', 'contrib_to_behavior.Analysis', (['f'], {'old_format': 'old_format'}), '(f, old_format=old_format)\n', (73789, 73815), False, 'import contrib_to_behavior\n'), ((74204, 74230), 'numpy.where', 'np.where', (['(perf[j, :] > 0.9)'], {}), '(perf[j, :] > 0.9)\n', (74212, 74230), True, 'import numpy as np\n'), ((74252, 74296), 'numpy.argsort', 'np.argsort', (['delay_accuracy[j, ind_good_perf]'], {}), '(delay_accuracy[j, ind_good_perf])\n', (74262, 74296), True, 'import numpy as np\n'), ((75836, 75863), 'numpy.where', 'np.where', (['(perf[j, :] > 0.85)'], {}), '(perf[j, :] > 0.85)\n', (75844, 75863), True, 'import numpy as np\n'), ((75924, 75981), 'numpy.mean', 'np.mean', (['spike_decoding[j, ind_good_models, :, :]'], {'axis': '(2)'}), '(spike_decoding[j, ind_good_models, :, :], axis=2)\n', (75931, 75981), True, 'import numpy as np\n'), ((76202, 76261), 'numpy.mean', 'np.mean', (['synapse_decoding[j, ind_good_models, :, :]'], {'axis': '(2)'}), '(synapse_decoding[j, ind_good_models, :, :], axis=2)\n', (76209, 76261), True, 'import numpy as np\n'), ((77071, 77099), 'numpy.where', 'np.where', (['(perf[j, :] > 0.975)'], {}), '(perf[j, :] > 0.975)\n', (77079, 77099), True, 'import numpy as np\n'), ((80739, 80789), 'numpy.mean', 'np.mean', (['spike_decode[delay_start:delay_end, 0, :]'], {}), '(spike_decode[delay_start:delay_end, 0, :])\n', (80746, 80789), True, 'import numpy as np\n'), ((81182, 81208), 'numpy.where', 'np.where', (['(perf[i, :] > 0.9)'], {}), '(perf[i, :] > 0.9)\n', (81190, 81208), True, 'import numpy as np\n'), ((85728, 85763), 'numpy.where', 'np.where', (['(perf_combined[j, :] > 0.9)'], {}), '(perf_combined[j, :] > 0.9)\n', (85736, 85763), True, 'import numpy as np\n'), ((86289, 86348), 'numpy.mean', 'np.mean', (['spike_decoding[j, ind_example[j, i], :, :]'], {'axis': '(1)'}), '(spike_decoding[j, ind_example[j, i], :, :], axis=1)\n', (86296, 86348), True, 'import numpy as np\n'), ((86361, 86419), 'numpy.std', 'np.std', (['spike_decoding[j, ind_example[j, i], :, :]'], {'axis': '(1)'}), '(spike_decoding[j, ind_example[j, i], :, :], axis=1)\n', (86367, 86419), True, 'import numpy as np\n'), ((86523, 86584), 'numpy.mean', 'np.mean', (['synapse_decoding[j, ind_example[j, i], :, :]'], {'axis': '(1)'}), '(synapse_decoding[j, ind_example[j, i], :, :], axis=1)\n', (86530, 86584), True, 'import numpy as np\n'), ((86597, 86657), 'numpy.std', 'np.std', (['synapse_decoding[j, ind_example[j, i], :, :]'], {'axis': '(1)'}), '(synapse_decoding[j, ind_example[j, i], :, :], axis=1)\n', (86603, 86657), True, 'import numpy as np\n'), ((86883, 86946), 'numpy.mean', 'np.mean', (['spike_decoding[j + 1, ind_example[j, i], :, :]'], {'axis': '(1)'}), '(spike_decoding[j + 1, ind_example[j, i], :, :], axis=1)\n', (86890, 86946), True, 'import numpy as np\n'), ((86957, 87019), 'numpy.std', 'np.std', (['spike_decoding[j + 1, ind_example[j, i], :, :]'], {'axis': '(1)'}), '(spike_decoding[j + 1, ind_example[j, i], :, :], axis=1)\n', (86963, 87019), True, 'import numpy as np\n'), ((87121, 87186), 'numpy.mean', 'np.mean', (['synapse_decoding[j + 1, ind_example[j, i], :, :]'], {'axis': '(1)'}), '(synapse_decoding[j + 1, ind_example[j, i], :, :], axis=1)\n', (87128, 87186), True, 'import numpy as np\n'), ((87197, 87261), 'numpy.std', 'np.std', (['synapse_decoding[j + 1, ind_example[j, i], :, :]'], {'axis': '(1)'}), '(synapse_decoding[j + 1, ind_example[j, i], :, :], axis=1)\n', (87203, 87261), True, 'import numpy as np\n'), ((88131, 88157), 'numpy.where', 'np.where', (['(perf[j, :] > 0.9)'], {}), '(perf[j, :] > 0.9)\n', (88139, 88157), True, 'import numpy as np\n'), ((88993, 89019), 'numpy.where', 'np.where', (['(perf[j, :] > 0.9)'], {}), '(perf[j, :] > 0.9)\n', (89001, 89019), True, 'import numpy as np\n'), ((89846, 89872), 'numpy.where', 'np.where', (['(perf[j, :] > 0.9)'], {}), '(perf[j, :] > 0.9)\n', (89854, 89872), True, 'import numpy as np\n'), ((5974, 6028), 'numpy.linalg.lstsq', 'np.linalg.lstsq', (['sample_dir', 'self.rnn_outputs[n, t, :]'], {}), '(sample_dir, self.rnn_outputs[n, t, :])\n', (5989, 6028), True, 'import numpy as np\n'), ((6129, 6151), 'numpy.mean', 'np.mean', (['(pred_err ** 2)'], {}), '(pred_err ** 2)\n', (6136, 6151), True, 'import numpy as np\n'), ((6181, 6214), 'numpy.var', 'np.var', (['self.rnn_outputs[n, t, :]'], {}), '(self.rnn_outputs[n, t, :])\n', (6187, 6214), True, 'import numpy as np\n'), ((6310, 6338), 'numpy.arctan2', 'np.arctan2', (['h[0][2]', 'h[0][1]'], {}), '(h[0][2], h[0][1])\n', (6320, 6338), True, 'import numpy as np\n'), ((6372, 6408), 'numpy.sqrt', 'np.sqrt', (['(h[0][0] ** 2 + h[0][1] ** 2)'], {}), '(h[0][0] ** 2 + h[0][1] ** 2)\n', (6379, 6408), True, 'import numpy as np\n'), ((7513, 7612), 'numpy.where', 'np.where', (['((self.rule == self.possible_rules[rule]) * (self.sample_dir == d) * (self.\n match == 1))'], {}), '((self.rule == self.possible_rules[rule]) * (self.sample_dir == d) *\n (self.match == 1))\n', (7521, 7612), True, 'import numpy as np\n'), ((7794, 7819), 'numpy.maximum', 'np.maximum', (['(0)', 'self.W_rnn'], {}), '(0, self.W_rnn)\n', (7804, 7819), True, 'import numpy as np\n'), ((8764, 8860), 'numpy.mean', 'np.mean', (['(self.syn_u[:, :, ind] * self.syn_x[:, :, ind] * self.rnn_outputs[:, :, ind])'], {'axis': '(2)'}), '(self.syn_u[:, :, ind] * self.syn_x[:, :, ind] * self.rnn_outputs[:,\n :, ind], axis=2)\n', (8771, 8860), True, 'import numpy as np\n'), ((9828, 9855), 'numpy.int_', 'np.int_', (['self.sample_dir[i]'], {}), '(self.sample_dir[i])\n', (9835, 9855), True, 'import numpy as np\n'), ((9876, 9898), 'numpy.int_', 'np.int_', (['self.match[i]'], {}), '(self.match[i])\n', (9883, 9898), True, 'import numpy as np\n'), ((13626, 13655), 'numpy.ones_like', 'np.ones_like', (['self.sample_dir'], {}), '(self.sample_dir)\n', (13638, 13655), True, 'import numpy as np\n'), ((13834, 13856), 'numpy.ones_like', 'np.ones_like', (['test_dir'], {}), '(test_dir)\n', (13846, 13856), True, 'import numpy as np\n'), ((18517, 18609), 'numpy.where', 'np.where', (['((self.rule == self.possible_rules[r]) * (self.match == 1) * (self.catch == 0))'], {}), '((self.rule == self.possible_rules[r]) * (self.match == 1) * (self.\n catch == 0))\n', (18525, 18609), True, 'import numpy as np\n'), ((18629, 18721), 'numpy.where', 'np.where', (['((self.rule == self.possible_rules[r]) * (self.match == 0) * (self.catch == 0))'], {}), '((self.rule == self.possible_rules[r]) * (self.match == 0) * (self.\n catch == 0))\n', (18637, 18721), True, 'import numpy as np\n'), ((18749, 18804), 'numpy.mean', 'np.mean', (['self.model_outputs[n, :, ind_match[0]]'], {'axis': '(0)'}), '(self.model_outputs[n, :, ind_match[0]], axis=0)\n', (18756, 18804), True, 'import numpy as np\n'), ((18846, 18905), 'numpy.mean', 'np.mean', (['self.model_outputs[n, :, ind_non_match[0]]'], {'axis': '(0)'}), '(self.model_outputs[n, :, ind_non_match[0]], axis=0)\n', (18853, 18905), True, 'import numpy as np\n'), ((19985, 20031), 'numpy.mean', 'np.mean', (['self.model_outputs[:, :, ind]'], {'axis': '(2)'}), '(self.model_outputs[:, :, ind], axis=2)\n', (19992, 20031), True, 'import numpy as np\n'), ((22412, 22432), 'numpy.where', 'np.where', (['(conds == i)'], {}), '(conds == i)\n', (22420, 22432), True, 'import numpy as np\n'), ((24092, 24109), 'numpy.argmax', 'np.argmax', (['dec', '(1)'], {}), '(dec, 1)\n', (24101, 24109), True, 'import numpy as np\n'), ((29507, 29526), 'numpy.where', 'np.where', (['(t == -500)'], {}), '(t == -500)\n', (29515, 29526), True, 'import numpy as np\n'), ((29526, 29542), 'numpy.where', 'np.where', (['(t == 0)'], {}), '(t == 0)\n', (29534, 29542), True, 'import numpy as np\n'), ((29541, 29559), 'numpy.where', 'np.where', (['(t == 500)'], {}), '(t == 500)\n', (29549, 29559), True, 'import numpy as np\n'), ((29558, 29577), 'numpy.where', 'np.where', (['(t == 1500)'], {}), '(t == 1500)\n', (29566, 29577), True, 'import numpy as np\n'), ((29616, 29635), 'numpy.where', 'np.where', (['(t == -500)'], {}), '(t == -500)\n', (29624, 29635), True, 'import numpy as np\n'), ((29635, 29651), 'numpy.where', 'np.where', (['(t == 0)'], {}), '(t == 0)\n', (29643, 29651), True, 'import numpy as np\n'), ((29650, 29668), 'numpy.where', 'np.where', (['(t == 500)'], {}), '(t == 500)\n', (29658, 29668), True, 'import numpy as np\n'), ((29667, 29686), 'numpy.where', 'np.where', (['(t == 1500)'], {}), '(t == 1500)\n', (29675, 29686), True, 'import numpy as np\n'), ((32589, 32614), 'numpy.mean', 'np.mean', (['rule_pev'], {'axis': '(0)'}), '(rule_pev, axis=0)\n', (32596, 32614), True, 'import numpy as np\n'), ((36151, 36186), 'numpy.std', 'np.std', (['sample_pev[:, 0, :]'], {'axis': '(0)'}), '(sample_pev[:, 0, :], axis=0)\n', (36157, 36186), True, 'import numpy as np\n'), ((36184, 36212), 'numpy.sqrt', 'np.sqrt', (['sample_pev.shape[0]'], {}), '(sample_pev.shape[0])\n', (36191, 36212), True, 'import numpy as np\n'), ((36408, 36443), 'numpy.std', 'np.std', (['sample_pev[:, 1, :]'], {'axis': '(0)'}), '(sample_pev[:, 1, :], axis=0)\n', (36414, 36443), True, 'import numpy as np\n'), ((36441, 36469), 'numpy.sqrt', 'np.sqrt', (['sample_pev.shape[0]'], {}), '(sample_pev.shape[0])\n', (36448, 36469), True, 'import numpy as np\n'), ((36803, 36841), 'numpy.std', 'np.std', (['sample_pev[:, rule, :]'], {'axis': '(0)'}), '(sample_pev[:, rule, :], axis=0)\n', (36809, 36841), True, 'import numpy as np\n'), ((36839, 36867), 'numpy.sqrt', 'np.sqrt', (['sample_pev.shape[0]'], {}), '(sample_pev.shape[0])\n', (36846, 36867), True, 'import numpy as np\n'), ((41002, 41016), 'matplotlib.pyplot.hold', 'plt.hold', (['(True)'], {}), '(True)\n', (41010, 41016), True, 'import matplotlib.pyplot as plt\n'), ((41037, 41077), 'numpy.mean', 'np.mean', (['sample_stp_pev[:, 0, :]'], {'axis': '(0)'}), '(sample_stp_pev[:, 0, :], axis=0)\n', (41044, 41077), True, 'import numpy as np\n'), ((41282, 41322), 'numpy.mean', 'np.mean', (['sample_stp_pev[:, 1, :]'], {'axis': '(0)'}), '(sample_stp_pev[:, 1, :], axis=0)\n', (41289, 41322), True, 'import numpy as np\n'), ((41698, 41741), 'numpy.mean', 'np.mean', (['sample_stp_pev[:, rule, :]'], {'axis': '(0)'}), '(sample_stp_pev[:, rule, :], axis=0)\n', (41705, 41741), True, 'import numpy as np\n'), ((42372, 42412), 'numpy.mean', 'np.mean', (['synapse_decode[:, 0, :]'], {'axis': '(1)'}), '(synapse_decode[:, 0, :], axis=1)\n', (42379, 42412), True, 'import numpy as np\n'), ((42431, 42470), 'numpy.std', 'np.std', (['synapse_decode[:, 0, :]'], {'axis': '(1)'}), '(synapse_decode[:, 0, :], axis=1)\n', (42437, 42470), True, 'import numpy as np\n'), ((42588, 42628), 'numpy.mean', 'np.mean', (['synapse_decode[:, 1, :]'], {'axis': '(1)'}), '(synapse_decode[:, 1, :], axis=1)\n', (42595, 42628), True, 'import numpy as np\n'), ((42647, 42686), 'numpy.std', 'np.std', (['synapse_decode[:, 1, :]'], {'axis': '(1)'}), '(synapse_decode[:, 1, :], axis=1)\n', (42653, 42686), True, 'import numpy as np\n'), ((42963, 43006), 'numpy.mean', 'np.mean', (['synapse_decode[:, rule, :]'], {'axis': '(1)'}), '(synapse_decode[:, rule, :], axis=1)\n', (42970, 43006), True, 'import numpy as np\n'), ((43025, 43067), 'numpy.std', 'np.std', (['synapse_decode[:, rule, :]'], {'axis': '(1)'}), '(synapse_decode[:, rule, :], axis=1)\n', (43031, 43067), True, 'import numpy as np\n'), ((43185, 43233), 'numpy.mean', 'np.mean', (['synapse_decode_test[:, rule, :]'], {'axis': '(1)'}), '(synapse_decode_test[:, rule, :], axis=1)\n', (43192, 43233), True, 'import numpy as np\n'), ((43252, 43299), 'numpy.std', 'np.std', (['synapse_decode_test[:, rule, :]'], {'axis': '(1)'}), '(synapse_decode_test[:, rule, :], axis=1)\n', (43258, 43299), True, 'import numpy as np\n'), ((43759, 43777), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (43775, 43777), True, 'import matplotlib.pyplot as plt\n'), ((43794, 43838), 'matplotlib.pyplot.savefig', 'plt.savefig', (['"""DMS summary.pdf"""'], {'format': '"""pdf"""'}), "('DMS summary.pdf', format='pdf')\n", (43805, 43838), True, 'import matplotlib.pyplot as plt\n'), ((44616, 44666), 'numpy.mean', 'np.mean', (['sample_decoding[i, j, 0, 0, :, :]'], {'axis': '(1)'}), '(sample_decoding[i, j, 0, 0, :, :], axis=1)\n', (44623, 44666), True, 'import numpy as np\n'), ((44682, 44731), 'numpy.std', 'np.std', (['sample_decoding[i, j, 0, 0, :, :]'], {'axis': '(1)'}), '(sample_decoding[i, j, 0, 0, :, :], axis=1)\n', (44688, 44731), True, 'import numpy as np\n'), ((44846, 44896), 'numpy.mean', 'np.mean', (['sample_decoding[i, j, 0, 1, :, :]'], {'axis': '(1)'}), '(sample_decoding[i, j, 0, 1, :, :], axis=1)\n', (44853, 44896), True, 'import numpy as np\n'), ((44912, 44961), 'numpy.std', 'np.std', (['sample_decoding[i, j, 0, 1, :, :]'], {'axis': '(1)'}), '(sample_decoding[i, j, 0, 1, :, :], axis=1)\n', (44918, 44961), True, 'import numpy as np\n'), ((61960, 61999), 'numpy.mean', 'np.mean', (['mean_resp1[:, :, :, :]'], {'axis': '(1)'}), '(mean_resp1[:, :, :, :], axis=1)\n', (61967, 61999), True, 'import numpy as np\n'), ((62066, 62105), 'numpy.mean', 'np.mean', (['mean_resp2[:, :, :, :]'], {'axis': '(1)'}), '(mean_resp2[:, :, :, :], axis=1)\n', (62073, 62105), True, 'import numpy as np\n'), ((76006, 76063), 'numpy.mean', 'np.mean', (['spike_decoding[j, ind_good_models, :, :]'], {'axis': '(2)'}), '(spike_decoding[j, ind_good_models, :, :], axis=2)\n', (76013, 76063), True, 'import numpy as np\n'), ((76286, 76345), 'numpy.mean', 'np.mean', (['synapse_decoding[j, ind_good_models, :, :]'], {'axis': '(2)'}), '(synapse_decoding[j, ind_good_models, :, :], axis=2)\n', (76293, 76345), True, 'import numpy as np\n'), ((80827, 80885), 'numpy.mean', 'np.mean', (['spike_decode[delay_start:delay_end, 0, :]'], {'axis': '(0)'}), '(spike_decode[delay_start:delay_end, 0, :], axis=0)\n', (80834, 80885), True, 'import numpy as np\n'), ((83535, 83589), 'contrib_to_behavior.Analysis', 'contrib_to_behavior.Analysis', (['f'], {'old_format': 'old_format'}), '(f, old_format=old_format)\n', (83563, 83589), False, 'import contrib_to_behavior\n'), ((84628, 84674), 'numpy.transpose', 'np.transpose', (['spike_decode[:, :, :]', '(1, 0, 2)'], {}), '(spike_decode[:, :, :], (1, 0, 2))\n', (84640, 84674), True, 'import numpy as np\n'), ((84715, 84763), 'numpy.transpose', 'np.transpose', (['synapse_decode[:, :, :]', '(1, 0, 2)'], {}), '(synapse_decode[:, :, :], (1, 0, 2))\n', (84727, 84763), True, 'import numpy as np\n'), ((84807, 84858), 'numpy.transpose', 'np.transpose', (['spike_decode_test[:, :, :]', '(1, 0, 2)'], {}), '(spike_decode_test[:, :, :], (1, 0, 2))\n', (84819, 84858), True, 'import numpy as np\n'), ((84904, 84957), 'numpy.transpose', 'np.transpose', (['synapse_decode_test[:, :, :]', '(1, 0, 2)'], {}), '(synapse_decode_test[:, :, :], (1, 0, 2))\n', (84916, 84957), True, 'import numpy as np\n'), ((91068, 91090), 'numpy.float32', 'np.float32', (['(y == y_hat)'], {}), '(y == y_hat)\n', (91078, 91090), True, 'import numpy as np\n'), ((91091, 91107), 'numpy.squeeze', 'np.squeeze', (['mask'], {}), '(mask)\n', (91101, 91107), True, 'import numpy as np\n'), ((6080, 6106), 'numpy.dot', 'np.dot', (['h[0]', 'sample_dir.T'], {}), '(h[0], sample_dir.T)\n', (6086, 6106), True, 'import numpy as np\n'), ((6549, 6608), 'numpy.linalg.lstsq', 'np.linalg.lstsq', (['test_dir[ind]', 'self.rnn_outputs[n, t, ind]'], {}), '(test_dir[ind], self.rnn_outputs[n, t, ind])\n', (6564, 6608), True, 'import numpy as np\n'), ((6722, 6744), 'numpy.mean', 'np.mean', (['(pred_err ** 2)'], {}), '(pred_err ** 2)\n', (6729, 6744), True, 'import numpy as np\n'), ((6778, 6813), 'numpy.var', 'np.var', (['self.rnn_outputs[n, t, ind]'], {}), '(self.rnn_outputs[n, t, ind])\n', (6784, 6813), True, 'import numpy as np\n'), ((6917, 6945), 'numpy.arctan2', 'np.arctan2', (['h[0][2]', 'h[0][1]'], {}), '(h[0][2], h[0][1])\n', (6927, 6945), True, 'import numpy as np\n'), ((6983, 7019), 'numpy.sqrt', 'np.sqrt', (['(h[0][0] ** 2 + h[0][1] ** 2)'], {}), '(h[0][0] ** 2 + h[0][1] ** 2)\n', (6990, 7019), True, 'import numpy as np\n'), ((8631, 8733), 'numpy.where', 'np.where', (['((self.rule == self.possible_rules[rule]) * (self.sample_dir == s) * (self.\n test_dir == t))'], {}), '((self.rule == self.possible_rules[rule]) * (self.sample_dir == s) *\n (self.test_dir == t))\n', (8639, 8733), True, 'import numpy as np\n'), ((10059, 10202), 'numpy.mean', 'np.mean', (['((self.model_outputs[2, -n:, i] > self.model_outputs[1, -n:, i]) * (self.\n model_outputs[2, -n:, i] > self.model_outputs[0, -n:, i]))'], {}), '((self.model_outputs[2, -n:, i] > self.model_outputs[1, -n:, i]) * (\n self.model_outputs[2, -n:, i] > self.model_outputs[0, -n:, i]))\n', (10066, 10202), True, 'import numpy as np\n'), ((10232, 10375), 'numpy.mean', 'np.mean', (['((self.model_outputs[1, -n:, i] > self.model_outputs[2, -n:, i]) * (self.\n model_outputs[1, -n:, i] > self.model_outputs[0, -n:, i]))'], {}), '((self.model_outputs[1, -n:, i] > self.model_outputs[2, -n:, i]) * (\n self.model_outputs[1, -n:, i] > self.model_outputs[0, -n:, i]))\n', (10239, 10375), True, 'import numpy as np\n'), ((11522, 11687), 'numpy.sum', 'np.sum', (['((self.model_outputs[2, test_rng, i] > self.model_outputs[1, test_rng, i]) *\n (self.model_outputs[2, test_rng, i] > self.model_outputs[0, test_rng, i]))'], {}), '((self.model_outputs[2, test_rng, i] > self.model_outputs[1, test_rng,\n i]) * (self.model_outputs[2, test_rng, i] > self.model_outputs[0,\n test_rng, i]))\n', (11528, 11687), True, 'import numpy as np\n'), ((11793, 11958), 'numpy.sum', 'np.sum', (['((self.model_outputs[1, test_rng, i] > self.model_outputs[2, test_rng, i]) *\n (self.model_outputs[1, test_rng, i] > self.model_outputs[0, test_rng, i]))'], {}), '((self.model_outputs[1, test_rng, i] > self.model_outputs[2, test_rng,\n i]) * (self.model_outputs[1, test_rng, i] > self.model_outputs[0,\n test_rng, i]))\n', (11799, 11958), True, 'import numpy as np\n'), ((13955, 13988), 'numpy.where', 'np.where', (['(self.num_test_stim >= 4)'], {}), '(self.num_test_stim >= 4)\n', (13963, 13988), True, 'import numpy as np\n'), ((14032, 14077), 'numpy.where', 'np.where', (['(self.rule == self.possible_rules[r])'], {}), '(self.rule == self.possible_rules[r])\n', (14040, 14077), True, 'import numpy as np\n'), ((16311, 16324), 'numpy.abs', 'np.abs', (['(i - j)'], {}), '(i - j)\n', (16317, 16324), True, 'import numpy as np\n'), ((18340, 18384), 'numpy.mean', 'np.mean', (['self.rnn_outputs[n, :, ind]'], {'axis': '(0)'}), '(self.rnn_outputs[n, :, ind], axis=0)\n', (18347, 18384), True, 'import numpy as np\n'), ((19861, 19953), 'numpy.where', 'np.where', (['((self.match[:, 0] == m1) * (self.match[:, 1] == m2) * (self.probe[:, 1] == 0))'], {}), '((self.match[:, 0] == m1) * (self.match[:, 1] == m2) * (self.probe[\n :, 1] == 0))\n', (19869, 19953), True, 'import numpy as np\n'), ((25191, 25224), 'numpy.where', 'np.where', (['(self.num_test_stim >= 4)'], {}), '(self.num_test_stim >= 4)\n', (25199, 25224), True, 'import numpy as np\n'), ((25268, 25313), 'numpy.where', 'np.where', (['(self.rule == self.possible_rules[r])'], {}), '(self.rule == self.possible_rules[r])\n', (25276, 25313), True, 'import numpy as np\n'), ((25346, 25413), 'numpy.where', 'np.where', (['((self.rule == self.possible_rules[r]) * (self.match == 0))'], {}), '((self.rule == self.possible_rules[r]) * (self.match == 0))\n', (25354, 25413), True, 'import numpy as np\n'), ((41096, 41135), 'numpy.std', 'np.std', (['sample_stp_pev[:, 0, :]'], {'axis': '(0)'}), '(sample_stp_pev[:, 0, :], axis=0)\n', (41102, 41135), True, 'import numpy as np\n'), ((41133, 41161), 'numpy.sqrt', 'np.sqrt', (['sample_pev.shape[0]'], {}), '(sample_pev.shape[0])\n', (41140, 41161), True, 'import numpy as np\n'), ((41341, 41380), 'numpy.std', 'np.std', (['sample_stp_pev[:, 1, :]'], {'axis': '(0)'}), '(sample_stp_pev[:, 1, :], axis=0)\n', (41347, 41380), True, 'import numpy as np\n'), ((41378, 41406), 'numpy.sqrt', 'np.sqrt', (['sample_pev.shape[0]'], {}), '(sample_pev.shape[0])\n', (41385, 41406), True, 'import numpy as np\n'), ((41760, 41802), 'numpy.std', 'np.std', (['sample_stp_pev[:, rule, :]'], {'axis': '(0)'}), '(sample_stp_pev[:, rule, :], axis=0)\n', (41766, 41802), True, 'import numpy as np\n'), ((41800, 41828), 'numpy.sqrt', 'np.sqrt', (['sample_pev.shape[0]'], {}), '(sample_pev.shape[0])\n', (41807, 41828), True, 'import numpy as np\n'), ((45135, 45189), 'numpy.mean', 'np.mean', (['sample_stp_decoding[i, j, 0, 0, :, :]'], {'axis': '(1)'}), '(sample_stp_decoding[i, j, 0, 0, :, :], axis=1)\n', (45142, 45189), True, 'import numpy as np\n'), ((45209, 45262), 'numpy.std', 'np.std', (['sample_stp_decoding[i, j, 0, 0, :, :]'], {'axis': '(1)'}), '(sample_stp_decoding[i, j, 0, 0, :, :], axis=1)\n', (45215, 45262), True, 'import numpy as np\n'), ((45410, 45464), 'numpy.mean', 'np.mean', (['sample_stp_decoding[i, j, 0, 1, :, :]'], {'axis': '(1)'}), '(sample_stp_decoding[i, j, 0, 1, :, :], axis=1)\n', (45417, 45464), True, 'import numpy as np\n'), ((45484, 45537), 'numpy.std', 'np.std', (['sample_stp_decoding[i, j, 0, 1, :, :]'], {'axis': '(1)'}), '(sample_stp_decoding[i, j, 0, 1, :, :], axis=1)\n', (45490, 45537), True, 'import numpy as np\n'), ((83731, 83789), 'contrib_to_behavior.Analysis', 'contrib_to_behavior.Analysis', (['f'], {'old_format': '(not old_format)'}), '(f, old_format=not old_format)\n', (83759, 83789), False, 'import contrib_to_behavior\n'), ((6481, 6506), 'numpy.where', 'np.where', (['(self.match == m)'], {}), '(self.match == m)\n', (6489, 6506), True, 'import numpy as np\n'), ((6666, 6695), 'numpy.dot', 'np.dot', (['h[0]', 'test_dir[ind].T'], {}), '(h[0], test_dir[ind].T)\n', (6672, 6695), True, 'import numpy as np\n'), ((9927, 9972), 'numpy.where', 'np.where', (['(self.rule[i] == self.possible_rules)'], {}), '(self.rule[i] == self.possible_rules)\n', (9935, 9972), True, 'import numpy as np\n'), ((16413, 16443), 'numpy.where', 'np.where', (['(self.sample_dir == s)'], {}), '(self.sample_dir == s)\n', (16421, 16443), True, 'import numpy as np\n'), ((16472, 16497), 'numpy.where', 'np.where', (['(self.match == 1)'], {}), '(self.match == 1)\n', (16480, 16497), True, 'import numpy as np\n'), ((16627, 16696), 'numpy.corrcoef', 'np.corrcoef', (['self.rnn_outputs[n, i, ind]', 'self.rnn_outputs[n, j, ind]'], {}), '(self.rnn_outputs[n, i, ind], self.rnn_outputs[n, j, ind])\n', (16638, 16696), True, 'import numpy as np\n'), ((20159, 20248), 'numpy.where', 'np.where', (['((self.rule[:, 0] == r1) * (self.rule[:, 1] == r2) * (self.probe[:, 1] == p))'], {}), '((self.rule[:, 0] == r1) * (self.rule[:, 1] == r2) * (self.probe[:,\n 1] == p))\n', (20167, 20248), True, 'import numpy as np\n'), ((25707, 25766), 'numpy.floor', 'np.floor', (['(self.sample_dir[ind] / (self.num_motion_dirs / 2))'], {}), '(self.sample_dir[ind] / (self.num_motion_dirs / 2))\n', (25715, 25766), True, 'import numpy as np\n'), ((16864, 16921), 'numpy.corrcoef', 'np.corrcoef', (['self.syn_x[n, i, ind]', 'self.syn_x[n, j, ind]'], {}), '(self.syn_x[n, i, ind], self.syn_x[n, j, ind])\n', (16875, 16921), True, 'import numpy as np\n'), ((17094, 17145), 'numpy.corrcoef', 'np.corrcoef', (['self.sa[n, i, ind]', 'self.sa[n, j, ind]'], {}), '(self.sa[n, i, ind], self.sa[n, j, ind])\n', (17105, 17145), True, 'import numpy as np\n'), ((18113, 18173), 'numpy.where', 'np.where', (['((self.num_test_stim >= 4) * (self.sample_dir == d))'], {}), '((self.num_test_stim >= 4) * (self.sample_dir == d))\n', (18121, 18173), True, 'import numpy as np\n'), ((18227, 18299), 'numpy.where', 'np.where', (['((self.rule == self.possible_rules[r]) * (self.sample_dir == d))'], {}), '((self.rule == self.possible_rules[r]) * (self.sample_dir == d))\n', (18235, 18299), True, 'import numpy as np\n'), ((24158, 24170), 'numpy.sign', 'np.sign', (['dec'], {}), '(dec)\n', (24165, 24170), True, 'import numpy as np\n'), ((26613, 26672), 'numpy.floor', 'np.floor', (['(self.sample_dir[ind] / (self.num_motion_dirs / 2))'], {}), '(self.sample_dir[ind] / (self.num_motion_dirs / 2))\n', (26621, 26672), True, 'import numpy as np\n')] |
import unittest
import numpy as np
from trip_kinematics.Utility import Rotation as R
class TestStates(unittest.TestCase):
"""Correct results were generated using scipy.spatial.transform.Rotation.
"""
def test_from_euler_to_quat(self):
from_euler_cases = [
([1, 2, 3],
[0.4359528440735657, -0.7182870182434115, 0.3106224510657039, 0.44443511344300074]),
([1, 0, 0], [0.8775825618903728, 0.479425538604203, 0.0, 0.0]),
([0, 1, 0], [0.8775825618903728, 0.0, 0.479425538604203, 0.0]),
([0, 0, 1], [0.8775825618903728, 0.0, 0.0, 0.479425538604203]),
([1.9259795237086745, -1.3166224746837234, -1.6487569080546618],
[0.6754185988029391, 0.1844402562591348, -0.7139352011035228, 0.00938279741930681]),
([1.7518638481261277, -1.4762648402511551, 2.783050177056892],
[-0.4241491835815902, 0.5252649613496548, 0.482281714511987, 0.5582101201991901]),
([-1.5637740430088356, 0.9967772877753394, -1.281072618629502],
[0.7010098084209453, -0.2935149227036473, 0.6418274036266723, -0.1024295982699634]),
([-2.018987322821037, -2.477643741973166, -1.7329756283622468],
[-0.4975785988500945, -0.562138163026927, -0.1155880687282515, -0.6504272611159234]),
([-1.547364165870453, 3.1155873968096914, 2.5081666410682404],
[-0.6610669854921286, -0.6825367059399726, 0.2141370877143812, 0.22644953831298628]),
([0.9232800471655072, 2.9636239629991614, 1.3121379453450235],
[0.3336790888032567, -0.5126269396587794, 0.7307896657891122, -0.3030154299822701]),
([-3.0969110304794603, 1.6540648344396605, -1.1444726410810802],
[0.4111294068459875, -0.5601567984771901, 0.38036785991321215, 0.6103419230982696]),
]
for euler_angles, quat in from_euler_cases:
assert np.allclose(R.from_euler('xyz', euler_angles, degrees=False).as_quat(), quat)
euler_angles_deg = np.array(euler_angles) * (180 / np.pi)
assert np.allclose(R.from_euler('xyz', euler_angles_deg, degrees=True).as_quat(), quat)
def test_from_matrix_to_quat(self):
test_cases = [
(np.array([[0.41198224566568303, -0.8337376517741568, -0.3676304629248995],
[-0.058726644927620864, -0.4269176212762076, 0.902381585483331],
[-0.9092974268256819, -0.35017548837401474, -0.2248450953661529]]),
[-0.43595284407356566, 0.7182870182434113, -0.31062245106570385, -0.4444351134430007]),
(np.array([[1., 0., 0.],
[0., 0.54030231, -0.84147098],
[0., 0.84147098, 0.54030231]]),
[0.8775825618903726, 0.47942553860420295, 0.0, 0.0]),
(np.array([[0.54030231, 0., 0.84147098],
[0., 1., 0.],
[-0.84147098, 0., 0.54030231]]),
[0.8775825618903726, 0.0, 0.47942553860420295, 0.0]),
(np.array([[0.54030231, -0.84147098, 0.],
[0.84147098, 0.54030231, 0.],
[0., 0., 1.]]),
[0.8775825618903726, 0.0, 0.0, 0.47942553860420295]),
(np.array([[-0.01958302, -0.27603141, -0.9609491],
[-0.25068215, 0.93178751, -0.26254618],
[0.96787136, 0.23575134, -0.08744336]]),
[-0.6754185988029392, -0.18444025625913482, 0.7139352011035228, -0.00938279741930680]),
(np.array([[-0.08838838, 0.98018011, 0.17729764],
[0.03312264, -0.17500364, 0.98401048],
[0.99553523, 0.09284766, -0.01699786]]),
[-0.4241491835815902, 0.5252649613496548, 0.4822817145119869, 0.5582101201991901]),
(np.array([[0.15513152, -0.23316354, 0.95998384],
[-0.52038015, 0.80671434, 0.28002943],
[-0.83972538, -0.54299793, 0.00381315]]),
[0.7010098084209453, -0.2935149227036473, 0.6418274036266723, -0.10242959826996342]),
(np.array([[0.12716755, -0.51732444, 0.84628827],
[0.7772303, -0.47810987, -0.40905258],
[0.61623167, 0.7097791, 0.34128017]]),
[0.49757859885009453, 0.5621381630269271, 0.11558806872825153, 0.6504272611159234]),
(np.array([[0.80573183, 0.00708378, -0.59223816],
[-0.59170947, -0.0342715, -0.80542248],
[-0.02600233, 0.99938745, -0.02342209]]),
[0.6610669854921286, 0.6825367059399726, -0.2141370877143812, -0.22644953831298634]),
(np.array([[-0.25174377, -0.54702511, 0.7983662],
[-0.95146476, 0.29079054, -0.10077531],
[-0.17703071, -0.78498687, -0.59367983]]),
[0.3336790888032567, -0.5126269396587794, 0.7307896657891123, -0.30301542998227005]),
(np.array([[-0.03439394, -0.92799031, -0.37101353],
[0.07572774, -0.3725858, 0.92490277],
[-0.99653518, 0.00371504, 0.0830893]]),
[0.4111294068459875, -0.5601567984771901, 0.38036785991321215, 0.6103419230982696])
]
for matrix, quat in test_cases:
assert np.allclose(R.from_matrix(matrix).as_quat(), quat)
def main():
test_cases = [
(np.array([[0.41198224566568303, -0.8337376517741568, -0.3676304629248995],
[-0.058726644927620864, -0.4269176212762076, 0.902381585483331],
[-0.9092974268256819, -0.35017548837401474, -0.2248450953661529]]),
[-0.43595284407356566, 0.7182870182434113, -0.31062245106570385, -0.4444351134430007]),
(np.array([[1., 0., 0.],
[0., 0.54030231, -0.84147098],
[0., 0.84147098, 0.54030231]]),
[0.8775825618903726, 0.47942553860420295, 0.0, 0.0]),
(np.array([[0.54030231, 0., 0.84147098],
[0., 1., 0.],
[-0.84147098, 0., 0.54030231]]),
[0.8775825618903726, 0.0, 0.47942553860420295, 0.0]),
(np.array([[0.54030231, -0.84147098, 0.],
[0.84147098, 0.54030231, 0.],
[0., 0., 1.]]),
[0.8775825618903726, 0.0, 0.0, 0.47942553860420295]),
(np.array([[-0.01958302, -0.27603141, -0.9609491],
[-0.25068215, 0.93178751, -0.26254618],
[0.96787136, 0.23575134, -0.08744336]]),
[-0.6754185988029392, -0.18444025625913482, 0.7139352011035228, -0.009382797419306801]),
(np.array([[-0.08838838, 0.98018011, 0.17729764],
[0.03312264, -0.17500364, 0.98401048],
[0.99553523, 0.09284766, -0.01699786]]),
[-0.4241491835815902, 0.5252649613496548, 0.4822817145119869, 0.5582101201991901]),
(np.array([[0.15513152, -0.23316354, 0.95998384],
[-0.52038015, 0.80671434, 0.28002943],
[-0.83972538, -0.54299793, 0.00381315]]),
[0.7010098084209453, -0.2935149227036473, 0.6418274036266723, -0.10242959826996342]),
(np.array([[0.12716755, -0.51732444, 0.84628827],
[0.7772303, -0.47810987, -0.40905258],
[0.61623167, 0.7097791, 0.34128017]]),
[0.49757859885009453, 0.5621381630269271, 0.11558806872825153, 0.6504272611159234]),
(np.array([[0.80573183, 0.00708378, -0.59223816],
[-0.59170947, -0.0342715, -0.80542248],
[-0.02600233, 0.99938745, -0.02342209]]),
[0.6610669854921286, 0.6825367059399726, -0.2141370877143812, -0.22644953831298634]),
(np.array([[-0.25174377, -0.54702511, 0.7983662],
[-0.95146476, 0.29079054, -0.10077531],
[-0.17703071, -0.78498687, -0.59367983]]),
[0.3336790888032567, -0.5126269396587794, 0.7307896657891123, -0.30301542998227005]),
(np.array([[-0.03439394, -0.92799031, -0.37101353],
[0.07572774, -0.3725858, 0.92490277],
[-0.99653518, 0.00371504, 0.0830893]]),
[0.4111294068459875, -0.5601567984771901, 0.38036785991321215, 0.6103419230982696])
]
for matrix, quat in test_cases:
assert np.allclose(R.from_matrix(matrix).as_quat(), quat)
if __name__ == '__main__':
main()
| [
"trip_kinematics.Utility.Rotation.from_matrix",
"numpy.array",
"trip_kinematics.Utility.Rotation.from_euler"
] | [((5515, 5730), 'numpy.array', 'np.array', (['[[0.41198224566568303, -0.8337376517741568, -0.3676304629248995], [-\n 0.058726644927620864, -0.4269176212762076, 0.902381585483331], [-\n 0.9092974268256819, -0.35017548837401474, -0.2248450953661529]]'], {}), '([[0.41198224566568303, -0.8337376517741568, -0.3676304629248995],\n [-0.058726644927620864, -0.4269176212762076, 0.902381585483331], [-\n 0.9092974268256819, -0.35017548837401474, -0.2248450953661529]])\n', (5523, 5730), True, 'import numpy as np\n'), ((5871, 5965), 'numpy.array', 'np.array', (['[[1.0, 0.0, 0.0], [0.0, 0.54030231, -0.84147098], [0.0, 0.84147098, 0.54030231]\n ]'], {}), '([[1.0, 0.0, 0.0], [0.0, 0.54030231, -0.84147098], [0.0, 0.84147098,\n 0.54030231]])\n', (5879, 5965), True, 'import numpy as np\n'), ((6072, 6166), 'numpy.array', 'np.array', (['[[0.54030231, 0.0, 0.84147098], [0.0, 1.0, 0.0], [-0.84147098, 0.0, 0.54030231]\n ]'], {}), '([[0.54030231, 0.0, 0.84147098], [0.0, 1.0, 0.0], [-0.84147098, 0.0,\n 0.54030231]])\n', (6080, 6166), True, 'import numpy as np\n'), ((6273, 6368), 'numpy.array', 'np.array', (['[[0.54030231, -0.84147098, 0.0], [0.84147098, 0.54030231, 0.0], [0.0, 0.0, 1.0]\n ]'], {}), '([[0.54030231, -0.84147098, 0.0], [0.84147098, 0.54030231, 0.0], [\n 0.0, 0.0, 1.0]])\n', (6281, 6368), True, 'import numpy as np\n'), ((6474, 6607), 'numpy.array', 'np.array', (['[[-0.01958302, -0.27603141, -0.9609491], [-0.25068215, 0.93178751, -\n 0.26254618], [0.96787136, 0.23575134, -0.08744336]]'], {}), '([[-0.01958302, -0.27603141, -0.9609491], [-0.25068215, 0.93178751,\n -0.26254618], [0.96787136, 0.23575134, -0.08744336]])\n', (6482, 6607), True, 'import numpy as np\n'), ((6754, 6886), 'numpy.array', 'np.array', (['[[-0.08838838, 0.98018011, 0.17729764], [0.03312264, -0.17500364, \n 0.98401048], [0.99553523, 0.09284766, -0.01699786]]'], {}), '([[-0.08838838, 0.98018011, 0.17729764], [0.03312264, -0.17500364, \n 0.98401048], [0.99553523, 0.09284766, -0.01699786]])\n', (6762, 6886), True, 'import numpy as np\n'), ((7027, 7160), 'numpy.array', 'np.array', (['[[0.15513152, -0.23316354, 0.95998384], [-0.52038015, 0.80671434, \n 0.28002943], [-0.83972538, -0.54299793, 0.00381315]]'], {}), '([[0.15513152, -0.23316354, 0.95998384], [-0.52038015, 0.80671434, \n 0.28002943], [-0.83972538, -0.54299793, 0.00381315]])\n', (7035, 7160), True, 'import numpy as np\n'), ((7303, 7433), 'numpy.array', 'np.array', (['[[0.12716755, -0.51732444, 0.84628827], [0.7772303, -0.47810987, -\n 0.40905258], [0.61623167, 0.7097791, 0.34128017]]'], {}), '([[0.12716755, -0.51732444, 0.84628827], [0.7772303, -0.47810987, -\n 0.40905258], [0.61623167, 0.7097791, 0.34128017]])\n', (7311, 7433), True, 'import numpy as np\n'), ((7575, 7709), 'numpy.array', 'np.array', (['[[0.80573183, 0.00708378, -0.59223816], [-0.59170947, -0.0342715, -\n 0.80542248], [-0.02600233, 0.99938745, -0.02342209]]'], {}), '([[0.80573183, 0.00708378, -0.59223816], [-0.59170947, -0.0342715, \n -0.80542248], [-0.02600233, 0.99938745, -0.02342209]])\n', (7583, 7709), True, 'import numpy as np\n'), ((7852, 7987), 'numpy.array', 'np.array', (['[[-0.25174377, -0.54702511, 0.7983662], [-0.95146476, 0.29079054, -\n 0.10077531], [-0.17703071, -0.78498687, -0.59367983]]'], {}), '([[-0.25174377, -0.54702511, 0.7983662], [-0.95146476, 0.29079054, \n -0.10077531], [-0.17703071, -0.78498687, -0.59367983]])\n', (7860, 7987), True, 'import numpy as np\n'), ((8130, 8261), 'numpy.array', 'np.array', (['[[-0.03439394, -0.92799031, -0.37101353], [0.07572774, -0.3725858, \n 0.92490277], [-0.99653518, 0.00371504, 0.0830893]]'], {}), '([[-0.03439394, -0.92799031, -0.37101353], [0.07572774, -0.3725858,\n 0.92490277], [-0.99653518, 0.00371504, 0.0830893]])\n', (8138, 8261), True, 'import numpy as np\n'), ((2073, 2095), 'numpy.array', 'np.array', (['euler_angles'], {}), '(euler_angles)\n', (2081, 2095), True, 'import numpy as np\n'), ((2294, 2509), 'numpy.array', 'np.array', (['[[0.41198224566568303, -0.8337376517741568, -0.3676304629248995], [-\n 0.058726644927620864, -0.4269176212762076, 0.902381585483331], [-\n 0.9092974268256819, -0.35017548837401474, -0.2248450953661529]]'], {}), '([[0.41198224566568303, -0.8337376517741568, -0.3676304629248995],\n [-0.058726644927620864, -0.4269176212762076, 0.902381585483331], [-\n 0.9092974268256819, -0.35017548837401474, -0.2248450953661529]])\n', (2302, 2509), True, 'import numpy as np\n'), ((2666, 2760), 'numpy.array', 'np.array', (['[[1.0, 0.0, 0.0], [0.0, 0.54030231, -0.84147098], [0.0, 0.84147098, 0.54030231]\n ]'], {}), '([[1.0, 0.0, 0.0], [0.0, 0.54030231, -0.84147098], [0.0, 0.84147098,\n 0.54030231]])\n', (2674, 2760), True, 'import numpy as np\n'), ((2883, 2977), 'numpy.array', 'np.array', (['[[0.54030231, 0.0, 0.84147098], [0.0, 1.0, 0.0], [-0.84147098, 0.0, 0.54030231]\n ]'], {}), '([[0.54030231, 0.0, 0.84147098], [0.0, 1.0, 0.0], [-0.84147098, 0.0,\n 0.54030231]])\n', (2891, 2977), True, 'import numpy as np\n'), ((3100, 3195), 'numpy.array', 'np.array', (['[[0.54030231, -0.84147098, 0.0], [0.84147098, 0.54030231, 0.0], [0.0, 0.0, 1.0]\n ]'], {}), '([[0.54030231, -0.84147098, 0.0], [0.84147098, 0.54030231, 0.0], [\n 0.0, 0.0, 1.0]])\n', (3108, 3195), True, 'import numpy as np\n'), ((3317, 3450), 'numpy.array', 'np.array', (['[[-0.01958302, -0.27603141, -0.9609491], [-0.25068215, 0.93178751, -\n 0.26254618], [0.96787136, 0.23575134, -0.08744336]]'], {}), '([[-0.01958302, -0.27603141, -0.9609491], [-0.25068215, 0.93178751,\n -0.26254618], [0.96787136, 0.23575134, -0.08744336]])\n', (3325, 3450), True, 'import numpy as np\n'), ((3612, 3744), 'numpy.array', 'np.array', (['[[-0.08838838, 0.98018011, 0.17729764], [0.03312264, -0.17500364, \n 0.98401048], [0.99553523, 0.09284766, -0.01699786]]'], {}), '([[-0.08838838, 0.98018011, 0.17729764], [0.03312264, -0.17500364, \n 0.98401048], [0.99553523, 0.09284766, -0.01699786]])\n', (3620, 3744), True, 'import numpy as np\n'), ((3901, 4034), 'numpy.array', 'np.array', (['[[0.15513152, -0.23316354, 0.95998384], [-0.52038015, 0.80671434, \n 0.28002943], [-0.83972538, -0.54299793, 0.00381315]]'], {}), '([[0.15513152, -0.23316354, 0.95998384], [-0.52038015, 0.80671434, \n 0.28002943], [-0.83972538, -0.54299793, 0.00381315]])\n', (3909, 4034), True, 'import numpy as np\n'), ((4193, 4323), 'numpy.array', 'np.array', (['[[0.12716755, -0.51732444, 0.84628827], [0.7772303, -0.47810987, -\n 0.40905258], [0.61623167, 0.7097791, 0.34128017]]'], {}), '([[0.12716755, -0.51732444, 0.84628827], [0.7772303, -0.47810987, -\n 0.40905258], [0.61623167, 0.7097791, 0.34128017]])\n', (4201, 4323), True, 'import numpy as np\n'), ((4481, 4615), 'numpy.array', 'np.array', (['[[0.80573183, 0.00708378, -0.59223816], [-0.59170947, -0.0342715, -\n 0.80542248], [-0.02600233, 0.99938745, -0.02342209]]'], {}), '([[0.80573183, 0.00708378, -0.59223816], [-0.59170947, -0.0342715, \n -0.80542248], [-0.02600233, 0.99938745, -0.02342209]])\n', (4489, 4615), True, 'import numpy as np\n'), ((4774, 4909), 'numpy.array', 'np.array', (['[[-0.25174377, -0.54702511, 0.7983662], [-0.95146476, 0.29079054, -\n 0.10077531], [-0.17703071, -0.78498687, -0.59367983]]'], {}), '([[-0.25174377, -0.54702511, 0.7983662], [-0.95146476, 0.29079054, \n -0.10077531], [-0.17703071, -0.78498687, -0.59367983]])\n', (4782, 4909), True, 'import numpy as np\n'), ((5068, 5199), 'numpy.array', 'np.array', (['[[-0.03439394, -0.92799031, -0.37101353], [0.07572774, -0.3725858, \n 0.92490277], [-0.99653518, 0.00371504, 0.0830893]]'], {}), '([[-0.03439394, -0.92799031, -0.37101353], [0.07572774, -0.3725858,\n 0.92490277], [-0.99653518, 0.00371504, 0.0830893]])\n', (5076, 5199), True, 'import numpy as np\n'), ((8467, 8488), 'trip_kinematics.Utility.Rotation.from_matrix', 'R.from_matrix', (['matrix'], {}), '(matrix)\n', (8480, 8488), True, 'from trip_kinematics.Utility import Rotation as R\n'), ((1975, 2023), 'trip_kinematics.Utility.Rotation.from_euler', 'R.from_euler', (['"""xyz"""', 'euler_angles'], {'degrees': '(False)'}), "('xyz', euler_angles, degrees=False)\n", (1987, 2023), True, 'from trip_kinematics.Utility import Rotation as R\n'), ((2144, 2195), 'trip_kinematics.Utility.Rotation.from_euler', 'R.from_euler', (['"""xyz"""', 'euler_angles_deg'], {'degrees': '(True)'}), "('xyz', euler_angles_deg, degrees=True)\n", (2156, 2195), True, 'from trip_kinematics.Utility import Rotation as R\n'), ((5429, 5450), 'trip_kinematics.Utility.Rotation.from_matrix', 'R.from_matrix', (['matrix'], {}), '(matrix)\n', (5442, 5450), True, 'from trip_kinematics.Utility import Rotation as R\n')] |
import os
os.environ['CUDA_VISIBLE_DEVICES'] = '1'
import cv2
import glob
import numpy as np
from PIL import Image
from core.utils import load_image, deprocess_image, preprocess_image
from core.networks import unet_spp_large_swish_generator_model
from core.dcp import estimate_transmission
img_size = 512
def preprocess_image(cv_img):
cv_img = cv2.resize(cv_img, (img_size,img_size))
img = np.array(cv_img)
img = (img - 127.5) / 127.5
return img
def load_image(path):
img = Image.open(path)
return img
def deprocess_image(img):
img = img * 127.5 + 127.5
return img.astype('uint8')
def get_file_name(path):
basename = os.path.basename(path)
onlyname = os.path.splitext(basename)[0]
return onlyname
def preprocess_cv2_image(cv_img):
cv_img = cv2.cvtColor(cv_img, cv2.COLOR_BGR2RGB)
cv_img = cv2.resize(cv_img, (img_size, img_size))
img = np.array(cv_img)
img = (img - 127.5) / 127.5
return img
def preprocess_depth_img(cv_img):
cv_img = cv2.resize(cv_img, (img_size, img_size))
img = np.array(cv_img)
img = np.reshape(img, (img_size, img_size, 1))
img = 2*(img - 0.5)
return img
g = unet_spp_large_swish_generator_model()
weight_path = "./weights/generator_185_26.h5"
g.load_weights(weight_path)
g.summary()
output_dir = "outputs/generator_185_26"
if not os.path.exists(output_dir):
os.makedirs(output_dir)
if __name__ == "__main__":
img_src = glob.glob("path/to/test/image/folder/*.jpg")
cnt=0
for img_path in img_src:
img_name = get_file_name(img_path)
ori_image = cv2.imread(img_path)
h, w, _ = ori_image.shape
# ori_image_resized = cv2.resize(ori_image, (img_size,img_size))
# cv2.imwrite(f"{img_name}_resized.jpg", ori_image_resized)
t = estimate_transmission(ori_image)
t = preprocess_depth_img(t)
ori_image = preprocess_cv2_image(ori_image)
x_test = np.concatenate((ori_image, t), axis=2)
x_test = np.reshape(x_test, (1,img_size,img_size,4))
generated_images = g.predict(x=x_test)
de_test = deprocess_image(generated_images)
de_test = np.reshape(de_test, (img_size,img_size,3))
# pred_image_resized = cv2.cvtColor(de_test, cv2.COLOR_BGR2RGB)
# cv2.imwrite(f"{img_name}_resized_pred.jpg", pred_image_resized)
de_test = cv2.resize(de_test, (w, h))
rgb_de_test = cv2.cvtColor(de_test, cv2.COLOR_BGR2RGB)
cv2.imwrite(f"{output_dir}/{img_name}.jpg", rgb_de_test)
cnt+=1
print(cnt, len(img_src))
# if cnt==10: break
print("Done!")
| [
"core.utils.deprocess_image",
"os.makedirs",
"core.dcp.estimate_transmission",
"os.path.basename",
"cv2.cvtColor",
"numpy.concatenate",
"cv2.imwrite",
"os.path.exists",
"PIL.Image.open",
"cv2.imread",
"numpy.array",
"numpy.reshape",
"os.path.splitext",
"core.networks.unet_spp_large_swish_g... | [((1240, 1278), 'core.networks.unet_spp_large_swish_generator_model', 'unet_spp_large_swish_generator_model', ([], {}), '()\n', (1276, 1278), False, 'from core.networks import unet_spp_large_swish_generator_model\n'), ((371, 411), 'cv2.resize', 'cv2.resize', (['cv_img', '(img_size, img_size)'], {}), '(cv_img, (img_size, img_size))\n', (381, 411), False, 'import cv2\n'), ((422, 438), 'numpy.array', 'np.array', (['cv_img'], {}), '(cv_img)\n', (430, 438), True, 'import numpy as np\n'), ((526, 542), 'PIL.Image.open', 'Image.open', (['path'], {}), '(path)\n', (536, 542), False, 'from PIL import Image\n'), ((699, 721), 'os.path.basename', 'os.path.basename', (['path'], {}), '(path)\n', (715, 721), False, 'import os\n'), ((842, 881), 'cv2.cvtColor', 'cv2.cvtColor', (['cv_img', 'cv2.COLOR_BGR2RGB'], {}), '(cv_img, cv2.COLOR_BGR2RGB)\n', (854, 881), False, 'import cv2\n'), ((896, 936), 'cv2.resize', 'cv2.resize', (['cv_img', '(img_size, img_size)'], {}), '(cv_img, (img_size, img_size))\n', (906, 936), False, 'import cv2\n'), ((948, 964), 'numpy.array', 'np.array', (['cv_img'], {}), '(cv_img)\n', (956, 964), True, 'import numpy as np\n'), ((1067, 1107), 'cv2.resize', 'cv2.resize', (['cv_img', '(img_size, img_size)'], {}), '(cv_img, (img_size, img_size))\n', (1077, 1107), False, 'import cv2\n'), ((1119, 1135), 'numpy.array', 'np.array', (['cv_img'], {}), '(cv_img)\n', (1127, 1135), True, 'import numpy as np\n'), ((1147, 1187), 'numpy.reshape', 'np.reshape', (['img', '(img_size, img_size, 1)'], {}), '(img, (img_size, img_size, 1))\n', (1157, 1187), True, 'import numpy as np\n'), ((1421, 1447), 'os.path.exists', 'os.path.exists', (['output_dir'], {}), '(output_dir)\n', (1435, 1447), False, 'import os\n'), ((1454, 1477), 'os.makedirs', 'os.makedirs', (['output_dir'], {}), '(output_dir)\n', (1465, 1477), False, 'import os\n'), ((1525, 1569), 'glob.glob', 'glob.glob', (['"""path/to/test/image/folder/*.jpg"""'], {}), "('path/to/test/image/folder/*.jpg')\n", (1534, 1569), False, 'import glob\n'), ((738, 764), 'os.path.splitext', 'os.path.splitext', (['basename'], {}), '(basename)\n', (754, 764), False, 'import os\n'), ((1680, 1700), 'cv2.imread', 'cv2.imread', (['img_path'], {}), '(img_path)\n', (1690, 1700), False, 'import cv2\n'), ((1896, 1928), 'core.dcp.estimate_transmission', 'estimate_transmission', (['ori_image'], {}), '(ori_image)\n', (1917, 1928), False, 'from core.dcp import estimate_transmission\n'), ((2041, 2079), 'numpy.concatenate', 'np.concatenate', (['(ori_image, t)'], {'axis': '(2)'}), '((ori_image, t), axis=2)\n', (2055, 2079), True, 'import numpy as np\n'), ((2100, 2146), 'numpy.reshape', 'np.reshape', (['x_test', '(1, img_size, img_size, 4)'], {}), '(x_test, (1, img_size, img_size, 4))\n', (2110, 2146), True, 'import numpy as np\n'), ((2213, 2246), 'core.utils.deprocess_image', 'deprocess_image', (['generated_images'], {}), '(generated_images)\n', (2228, 2246), False, 'from core.utils import load_image, deprocess_image, preprocess_image\n'), ((2266, 2310), 'numpy.reshape', 'np.reshape', (['de_test', '(img_size, img_size, 3)'], {}), '(de_test, (img_size, img_size, 3))\n', (2276, 2310), True, 'import numpy as np\n'), ((2480, 2507), 'cv2.resize', 'cv2.resize', (['de_test', '(w, h)'], {}), '(de_test, (w, h))\n', (2490, 2507), False, 'import cv2\n'), ((2533, 2573), 'cv2.cvtColor', 'cv2.cvtColor', (['de_test', 'cv2.COLOR_BGR2RGB'], {}), '(de_test, cv2.COLOR_BGR2RGB)\n', (2545, 2573), False, 'import cv2\n'), ((2583, 2639), 'cv2.imwrite', 'cv2.imwrite', (['f"""{output_dir}/{img_name}.jpg"""', 'rgb_de_test'], {}), "(f'{output_dir}/{img_name}.jpg', rgb_de_test)\n", (2594, 2639), False, 'import cv2\n')] |
from OT.PSD import OT_PSD
from basic.select import select_file
from basic.filter import MA
from matplotlib import rcParams
rcParams["font.family"] = "sans-serif"
rcParams["font.sans-serif"] = ["Arial"]
rcParams.update({'font.size': 18})
import pandas as pd
import numpy as np
import random
import math
import matplotlib.pyplot as plt
# plt.rc('text', usetex=True)
# plt.rc('font', **{'family' : 'sans-serif'})
# plt.rc('font', **{'sans-serif' : 'Arial'})
# plt.rc('legend', fontsize=18)
## generate #n uniform RV range from 0 to max
def get_uRV(max, n):
RVs = np.array([int(math.floor(random.uniform(0, max))) for i in range(n)])
return RVs
def get_boot_select(signals):
n = len(signals)
signals_boot = []
index = get_uRV(n, n)
for i in index:
signals_boot += [list(signals)[i]]
return signals_boot
def get_excel_data(data):
n_traces = int(data.shape[1] / 4)
signals = []
dt = []
t_end = [] # ending time
Fs = []
for i in range(n_traces):
## select and remove nan data
signal = data[:, 1 + 4 * i]
signals += [signal[~np.isnan(signal)]]
dt += [data[1, 0 + 4 * i] - data[0, 0 + 4 * i]]
t_end += [dt[-1] * len(signals[-1])]
Fs += [1 / dt[-1]]
return signals, Fs
def connect_traces(signals):
signal_connect = []
L_cum = 0
## for connecting all traces
for fs,signal in zip(Fs,signals):
signal_connect = np.append(signal_connect, signal+L_cum)
L_cum += np.mean(signal[-20:])
return signal_connect
### import data
# path = select_file()
path = r'C:\Users\pine\Desktop\Data\time trace\m51 all traces\m51_2.0uM_All.xlsx'
## x:1.8,
df = pd.read_excel(path)
data = np.array(df.dropna(axis='columns', how='all'))
signals, Fs = get_excel_data(data)
freq_all_c = []
psd_all_c = []
t_AFC_all = []
AFC_all = []
n_boot = 2
for i in range(n_boot):
Fs_spatial = 5
F_resolution = 0.002
signals_boot = get_boot_select(signals)
signal_connect = connect_traces(signals_boot)
PSD_connect = OT_PSD(signal_connect, fs=np.mean(Fs), Fs_spatial=Fs_spatial, F_resolution=F_resolution, bintype='set_width')
t_AFC_conn, AFC_conn = PSD_connect.t_ACF, PSD_connect.ACF
freq_conn, psd_conn = PSD_connect.get_PSD()
freq_all_c = np.append(freq_all_c, freq_conn)
psd_all_c = np.append(psd_all_c, psd_conn)
t_AFC_all += [t_AFC_conn[:200]]
AFC_all += [AFC_conn[:200]]
t_AFC = np.mean(np.array(t_AFC_all), axis=0)
AFC = np.mean(np.array(AFC_all), axis=0)
# fig, ax = plt.subplots(figsize=(10,8))
# ax.plot(freq_conn, psd_conn, '.')
# ax.set_xlim(0, Fs_spatial/2)
# ax.set_ylim(0, psd[np.argsort(psd)[-2]]*2)
# ax.set_xlabel('spatial frequency (1/count)')
# ax.set_ylabel('PSD')
fig, ax = plt.subplots(figsize=(10,8))
ax.plot(t_AFC, AFC, '-.')
ax.set_xlim(0, 40)
# ax.set_xlim(0, 100)
# ax.set_ylim(0.000, 0.1)
ax.set_xlabel('Distance (count)')
ax.set_ylabel('Autocorrelation')
fig, ax = plt.subplots(figsize=(10,8))
f = np.sort(freq_all_c)
p = psd_all_c[np.argsort(freq_all_c)]
ax.plot(MA(f,15,mode='silding'), MA(p,15,mode='silding'), '-')
# ax.plot(f, p, '-')
ax.set_xlim(0, 0.5)
# ax.set_xlim(0, 100)
ax.set_ylim(0.000, 5e7)
ax.set_xlabel('Spatial frequency (1/count)')
ax.set_ylabel('Power spectral density(a.u.)')
ax.annotate(r'$\frac{1}{7.5}$', xy=(1/7.5, 3e7), xytext=(1/7.5, 5e7),
arrowprops=dict(facecolor='black', shrink=0.05)
)
plt.figure()
plt.plot(signal_connect) | [
"matplotlib.pyplot.plot",
"random.uniform",
"matplotlib.rcParams.update",
"numpy.isnan",
"pandas.read_excel",
"numpy.sort",
"matplotlib.pyplot.figure",
"numpy.append",
"numpy.array",
"numpy.argsort",
"basic.filter.MA",
"numpy.mean",
"matplotlib.pyplot.subplots"
] | [((202, 236), 'matplotlib.rcParams.update', 'rcParams.update', (["{'font.size': 18}"], {}), "({'font.size': 18})\n", (217, 236), False, 'from matplotlib import rcParams\n'), ((1686, 1705), 'pandas.read_excel', 'pd.read_excel', (['path'], {}), '(path)\n', (1699, 1705), True, 'import pandas as pd\n'), ((2757, 2786), 'matplotlib.pyplot.subplots', 'plt.subplots', ([], {'figsize': '(10, 8)'}), '(figsize=(10, 8))\n', (2769, 2786), True, 'import matplotlib.pyplot as plt\n'), ((2957, 2986), 'matplotlib.pyplot.subplots', 'plt.subplots', ([], {'figsize': '(10, 8)'}), '(figsize=(10, 8))\n', (2969, 2986), True, 'import matplotlib.pyplot as plt\n'), ((2990, 3009), 'numpy.sort', 'np.sort', (['freq_all_c'], {}), '(freq_all_c)\n', (2997, 3009), True, 'import numpy as np\n'), ((3436, 3448), 'matplotlib.pyplot.figure', 'plt.figure', ([], {}), '()\n', (3446, 3448), True, 'import matplotlib.pyplot as plt\n'), ((3449, 3473), 'matplotlib.pyplot.plot', 'plt.plot', (['signal_connect'], {}), '(signal_connect)\n', (3457, 3473), True, 'import matplotlib.pyplot as plt\n'), ((2287, 2319), 'numpy.append', 'np.append', (['freq_all_c', 'freq_conn'], {}), '(freq_all_c, freq_conn)\n', (2296, 2319), True, 'import numpy as np\n'), ((2336, 2366), 'numpy.append', 'np.append', (['psd_all_c', 'psd_conn'], {}), '(psd_all_c, psd_conn)\n', (2345, 2366), True, 'import numpy as np\n'), ((2452, 2471), 'numpy.array', 'np.array', (['t_AFC_all'], {}), '(t_AFC_all)\n', (2460, 2471), True, 'import numpy as np\n'), ((2495, 2512), 'numpy.array', 'np.array', (['AFC_all'], {}), '(AFC_all)\n', (2503, 2512), True, 'import numpy as np\n'), ((3024, 3046), 'numpy.argsort', 'np.argsort', (['freq_all_c'], {}), '(freq_all_c)\n', (3034, 3046), True, 'import numpy as np\n'), ((3056, 3081), 'basic.filter.MA', 'MA', (['f', '(15)'], {'mode': '"""silding"""'}), "(f, 15, mode='silding')\n", (3058, 3081), False, 'from basic.filter import MA\n'), ((3081, 3106), 'basic.filter.MA', 'MA', (['p', '(15)'], {'mode': '"""silding"""'}), "(p, 15, mode='silding')\n", (3083, 3106), False, 'from basic.filter import MA\n'), ((1443, 1484), 'numpy.append', 'np.append', (['signal_connect', '(signal + L_cum)'], {}), '(signal_connect, signal + L_cum)\n', (1452, 1484), True, 'import numpy as np\n'), ((1500, 1521), 'numpy.mean', 'np.mean', (['signal[-20:]'], {}), '(signal[-20:])\n', (1507, 1521), True, 'import numpy as np\n'), ((2075, 2086), 'numpy.mean', 'np.mean', (['Fs'], {}), '(Fs)\n', (2082, 2086), True, 'import numpy as np\n'), ((592, 614), 'random.uniform', 'random.uniform', (['(0)', 'max'], {}), '(0, max)\n', (606, 614), False, 'import random\n'), ((1108, 1124), 'numpy.isnan', 'np.isnan', (['signal'], {}), '(signal)\n', (1116, 1124), True, 'import numpy as np\n')] |
import copy
import h5py
import math
import numpy as np
import os
import torch
from torch.utils.data import Dataset
import sys
BASE_DIR = os.path.dirname(os.path.abspath(__file__))
ROOR_DIR = os.path.dirname(BASE_DIR)
sys.path.append(ROOR_DIR)
from utils import random_select_points, shift_point_cloud, jitter_point_cloud, \
generate_random_rotation_matrix, generate_random_tranlation_vector, \
transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc
half1 = ['airplane', 'bathtub', 'bed', 'bench', 'bookshelf', 'bottle', 'bowl',
'car', 'chair', 'cone', 'cup', 'curtain', 'desk', 'door', 'dresser',
'flower_pot', 'glass_box', 'guitar', 'keyboard', 'lamp']
half1_symmetric = ['bottle', 'bowl', 'cone', 'cup', 'flower_pot', 'lamp']
half2 = ['laptop', 'mantel', 'monitor', 'night_stand', 'person', 'piano',
'plant', 'radio', 'range_hood', 'sink', 'sofa', 'stairs', 'stool',
'table', 'tent', 'toilet', 'tv_stand', 'vase', 'wardrobe', 'xbox']
half2_symmetric = ['tent', 'vase']
class ModelNet40(Dataset):
def __init__(self, root, split, npts, p_keep, noise, unseen, ao=False,
normal=False):
super(ModelNet40, self).__init__()
self.single = False # for specific-class visualization
assert split in ['train', 'val', 'test']
self.split = split
self.npts = npts
self.p_keep = p_keep
self.noise = noise
self.unseen = unseen
self.ao = ao # Asymmetric Objects
self.normal = normal
self.half = half1 if split in 'train' else half2
self.symmetric = half1_symmetric + half2_symmetric
self.label2cat, self.cat2label = self.label2category(
os.path.join(root, 'shape_names.txt'))
self.half_labels = [self.cat2label[cat] for cat in self.half]
self.symmetric_labels = [self.cat2label[cat] for cat in self.symmetric]
files = [os.path.join(root, 'ply_data_train{}.h5'.format(i))
for i in range(5)]
if split == 'test':
files = [os.path.join(root, 'ply_data_test{}.h5'.format(i))
for i in range(2)]
self.data, self.labels = self.decode_h5(files)
print(f'split: {self.split}, unique_ids: {len(np.unique(self.labels))}')
if self.split == 'train':
self.Rs = [generate_random_rotation_matrix() for _ in range(len(self.data))]
self.ts = [generate_random_tranlation_vector() for _ in range(len(self.data))]
def label2category(self, file):
with open(file, 'r') as f:
label2cat = [category.strip() for category in f.readlines()]
cat2label = {label2cat[i]: i for i in range(len(label2cat))}
return label2cat, cat2label
def decode_h5(self, files):
points, normal, label = [], [], []
for file in files:
f = h5py.File(file, 'r')
cur_points = f['data'][:].astype(np.float32)
cur_normal = f['normal'][:].astype(np.float32)
cur_label = f['label'][:].flatten().astype(np.int32)
if self.unseen:
idx = np.isin(cur_label, self.half_labels)
cur_points = cur_points[idx]
cur_normal = cur_normal[idx]
cur_label = cur_label[idx]
if self.ao and self.split in ['val', 'test']:
idx = ~np.isin(cur_label, self.symmetric_labels)
cur_points = cur_points[idx]
cur_normal = cur_normal[idx]
cur_label = cur_label[idx]
if self.single:
idx = np.isin(cur_label, [8])
cur_points = cur_points[idx]
cur_normal = cur_normal[idx]
cur_label = cur_label[idx]
points.append(cur_points)
normal.append(cur_normal)
label.append(cur_label)
points = np.concatenate(points, axis=0)
normal = np.concatenate(normal, axis=0)
data = np.concatenate([points, normal], axis=-1).astype(np.float32)
label = np.concatenate(label, axis=0)
return data, label
def compose(self, item, p_keep):
tgt_cloud = self.data[item, ...]
if self.split != 'train':
np.random.seed(item)
R, t = generate_random_rotation_matrix(), generate_random_tranlation_vector()
else:
tgt_cloud = flip_pc(tgt_cloud)
R, t = generate_random_rotation_matrix(), generate_random_tranlation_vector()
src_cloud = random_crop(copy.deepcopy(tgt_cloud), p_keep=p_keep[0])
src_size = math.ceil(self.npts * p_keep[0])
tgt_size = self.npts
if len(p_keep) > 1:
tgt_cloud = random_crop(copy.deepcopy(tgt_cloud),
p_keep=p_keep[1])
tgt_size = math.ceil(self.npts * p_keep[1])
src_cloud_points = transform(src_cloud[:, :3], R, t)
src_cloud_normal = transform(src_cloud[:, 3:], R)
src_cloud = np.concatenate([src_cloud_points, src_cloud_normal],
axis=-1)
src_cloud = random_select_points(src_cloud, m=src_size)
tgt_cloud = random_select_points(tgt_cloud, m=tgt_size)
if self.split == 'train' or self.noise:
src_cloud[:, :3] = jitter_point_cloud(src_cloud[:, :3])
tgt_cloud[:, :3] = jitter_point_cloud(tgt_cloud[:, :3])
tgt_cloud, src_cloud = shuffle_pc(tgt_cloud), shuffle_pc(
src_cloud)
return src_cloud, tgt_cloud, R, t
def __getitem__(self, item):
src_cloud, tgt_cloud, R, t = self.compose(item=item,
p_keep=self.p_keep)
if not self.normal:
tgt_cloud, src_cloud = tgt_cloud[:, :3], src_cloud[:, :3]
return tgt_cloud, src_cloud, R, t
def __len__(self):
return len(self.data)
| [
"sys.path.append",
"numpy.isin",
"os.path.abspath",
"h5py.File",
"numpy.random.seed",
"copy.deepcopy",
"math.ceil",
"os.path.dirname",
"utils.generate_random_rotation_matrix",
"utils.generate_random_tranlation_vector",
"numpy.unique",
"utils.jitter_point_cloud",
"utils.shuffle_pc",
"utils.... | [((193, 218), 'os.path.dirname', 'os.path.dirname', (['BASE_DIR'], {}), '(BASE_DIR)\n', (208, 218), False, 'import os\n'), ((219, 244), 'sys.path.append', 'sys.path.append', (['ROOR_DIR'], {}), '(ROOR_DIR)\n', (234, 244), False, 'import sys\n'), ((155, 180), 'os.path.abspath', 'os.path.abspath', (['__file__'], {}), '(__file__)\n', (170, 180), False, 'import os\n'), ((3897, 3927), 'numpy.concatenate', 'np.concatenate', (['points'], {'axis': '(0)'}), '(points, axis=0)\n', (3911, 3927), True, 'import numpy as np\n'), ((3945, 3975), 'numpy.concatenate', 'np.concatenate', (['normal'], {'axis': '(0)'}), '(normal, axis=0)\n', (3959, 3975), True, 'import numpy as np\n'), ((4068, 4097), 'numpy.concatenate', 'np.concatenate', (['label'], {'axis': '(0)'}), '(label, axis=0)\n', (4082, 4097), True, 'import numpy as np\n'), ((4604, 4636), 'math.ceil', 'math.ceil', (['(self.npts * p_keep[0])'], {}), '(self.npts * p_keep[0])\n', (4613, 4636), False, 'import math\n'), ((4894, 4927), 'utils.transform', 'transform', (['src_cloud[:, :3]', 'R', 't'], {}), '(src_cloud[:, :3], R, t)\n', (4903, 4927), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((4955, 4985), 'utils.transform', 'transform', (['src_cloud[:, 3:]', 'R'], {}), '(src_cloud[:, 3:], R)\n', (4964, 4985), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((5006, 5067), 'numpy.concatenate', 'np.concatenate', (['[src_cloud_points, src_cloud_normal]'], {'axis': '(-1)'}), '([src_cloud_points, src_cloud_normal], axis=-1)\n', (5020, 5067), True, 'import numpy as np\n'), ((5123, 5166), 'utils.random_select_points', 'random_select_points', (['src_cloud'], {'m': 'src_size'}), '(src_cloud, m=src_size)\n', (5143, 5166), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((5187, 5230), 'utils.random_select_points', 'random_select_points', (['tgt_cloud'], {'m': 'tgt_size'}), '(tgt_cloud, m=tgt_size)\n', (5207, 5230), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((1725, 1762), 'os.path.join', 'os.path.join', (['root', '"""shape_names.txt"""'], {}), "(root, 'shape_names.txt')\n", (1737, 1762), False, 'import os\n'), ((2883, 2903), 'h5py.File', 'h5py.File', (['file', '"""r"""'], {}), "(file, 'r')\n", (2892, 2903), False, 'import h5py\n'), ((4250, 4270), 'numpy.random.seed', 'np.random.seed', (['item'], {}), '(item)\n', (4264, 4270), True, 'import numpy as np\n'), ((4399, 4417), 'utils.flip_pc', 'flip_pc', (['tgt_cloud'], {}), '(tgt_cloud)\n', (4406, 4417), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((4541, 4565), 'copy.deepcopy', 'copy.deepcopy', (['tgt_cloud'], {}), '(tgt_cloud)\n', (4554, 4565), False, 'import copy\n'), ((4833, 4865), 'math.ceil', 'math.ceil', (['(self.npts * p_keep[1])'], {}), '(self.npts * p_keep[1])\n', (4842, 4865), False, 'import math\n'), ((5311, 5347), 'utils.jitter_point_cloud', 'jitter_point_cloud', (['src_cloud[:, :3]'], {}), '(src_cloud[:, :3])\n', (5329, 5347), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((5379, 5415), 'utils.jitter_point_cloud', 'jitter_point_cloud', (['tgt_cloud[:, :3]'], {}), '(tgt_cloud[:, :3])\n', (5397, 5415), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((5447, 5468), 'utils.shuffle_pc', 'shuffle_pc', (['tgt_cloud'], {}), '(tgt_cloud)\n', (5457, 5468), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((5470, 5491), 'utils.shuffle_pc', 'shuffle_pc', (['src_cloud'], {}), '(src_cloud)\n', (5480, 5491), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((2353, 2386), 'utils.generate_random_rotation_matrix', 'generate_random_rotation_matrix', ([], {}), '()\n', (2384, 2386), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((2442, 2477), 'utils.generate_random_tranlation_vector', 'generate_random_tranlation_vector', ([], {}), '()\n', (2475, 2477), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((3135, 3171), 'numpy.isin', 'np.isin', (['cur_label', 'self.half_labels'], {}), '(cur_label, self.half_labels)\n', (3142, 3171), True, 'import numpy as np\n'), ((3611, 3634), 'numpy.isin', 'np.isin', (['cur_label', '[8]'], {}), '(cur_label, [8])\n', (3618, 3634), True, 'import numpy as np\n'), ((3991, 4032), 'numpy.concatenate', 'np.concatenate', (['[points, normal]'], {'axis': '(-1)'}), '([points, normal], axis=-1)\n', (4005, 4032), True, 'import numpy as np\n'), ((4290, 4323), 'utils.generate_random_rotation_matrix', 'generate_random_rotation_matrix', ([], {}), '()\n', (4321, 4323), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((4325, 4360), 'utils.generate_random_tranlation_vector', 'generate_random_tranlation_vector', ([], {}), '()\n', (4358, 4360), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((4437, 4470), 'utils.generate_random_rotation_matrix', 'generate_random_rotation_matrix', ([], {}), '()\n', (4468, 4470), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((4472, 4507), 'utils.generate_random_tranlation_vector', 'generate_random_tranlation_vector', ([], {}), '()\n', (4505, 4507), False, 'from utils import random_select_points, shift_point_cloud, jitter_point_cloud, generate_random_rotation_matrix, generate_random_tranlation_vector, transform, random_crop, shuffle_pc, random_scale_point_cloud, flip_pc\n'), ((4730, 4754), 'copy.deepcopy', 'copy.deepcopy', (['tgt_cloud'], {}), '(tgt_cloud)\n', (4743, 4754), False, 'import copy\n'), ((3386, 3427), 'numpy.isin', 'np.isin', (['cur_label', 'self.symmetric_labels'], {}), '(cur_label, self.symmetric_labels)\n', (3393, 3427), True, 'import numpy as np\n'), ((2268, 2290), 'numpy.unique', 'np.unique', (['self.labels'], {}), '(self.labels)\n', (2277, 2290), True, 'import numpy as np\n')] |
from astropy.table import Table
from astropy.io import fits
from GPSTiming.interpolation import interpolate_boardtimes
from numpy import diff
def test_time_differences_greater_zero(path: str):
fits_file = fits.open(path)
table = Table(fits_file[1].data)
fits_file.close()
table = interpolate_boardtimes(table)
assert diff(table["InterpolatedUnixTime"]).all() > 0
def main():
path = '/net/big-tank/POOL/projects/fact/gps_timestamp_data/2014/01/01/20140101_073_v1.1.1_gps_timestamp_data_timestamps.fits'
test_time_differences_greater_zero(path)
if __name__ == "__main__":
main()
| [
"numpy.diff",
"astropy.table.Table",
"astropy.io.fits.open",
"GPSTiming.interpolation.interpolate_boardtimes"
] | [((211, 226), 'astropy.io.fits.open', 'fits.open', (['path'], {}), '(path)\n', (220, 226), False, 'from astropy.io import fits\n'), ((239, 263), 'astropy.table.Table', 'Table', (['fits_file[1].data'], {}), '(fits_file[1].data)\n', (244, 263), False, 'from astropy.table import Table\n'), ((298, 327), 'GPSTiming.interpolation.interpolate_boardtimes', 'interpolate_boardtimes', (['table'], {}), '(table)\n', (320, 327), False, 'from GPSTiming.interpolation import interpolate_boardtimes\n'), ((339, 374), 'numpy.diff', 'diff', (["table['InterpolatedUnixTime']"], {}), "(table['InterpolatedUnixTime'])\n", (343, 374), False, 'from numpy import diff\n')] |
import matplotlib.pyplot as plt
import matplotlib.cm as cm
import numpy as np
class TransitionPlot:
N_COL = 8
OBJ_NAMES = ['BG', 'SQ', 'SC', 'BQ', 'BC', 'P1', 'P2', 'P3', 'P4']
def __init__(self, num_obj_slots):
assert num_obj_slots in (4,8,9)
self.COLORS = [cm.rainbow(x) for x in np.linspace(0, 1, num_obj_slots)]
self.OBJ_ROW = np.ceil(num_obj_slots / 4).astype(np.int32)
self.N_ROW = 4 + self.OBJ_ROW + 8
self.FIGURE_SIZE = (self.N_COL*2, self.N_ROW*2)
print(num_obj_slots)
plt.subplots(figsize=self.FIGURE_SIZE)
self.obs_axs = []
self.act_axs = []
self.latent_axs = []
self.obj_axs = []
self.next_obj_axs = []
for i in [0, 3]:
self.obs_axs.append(
plt.subplot2grid(
shape=(self.N_ROW, self.N_COL), loc=(0, i*2), colspan=2, rowspan=2
)
)
for i in [1, 2]:
self.act_axs.append(
plt.subplot2grid(
shape=(self.N_ROW, self.N_COL), loc=(0, i*2), colspan=2, rowspan=2
)
)
for i in range(self.OBJ_ROW):
for j in range(4):
self.obj_axs.append(
plt.subplot2grid(shape=(self.N_ROW, self.N_COL), loc=(2 + i, 0 + j), colspan=1, rowspan=1)
)
self.next_obj_axs.append(
plt.subplot2grid(shape=(self.N_ROW, self.N_COL), loc=(2 + i, 4 + j), colspan=1, rowspan=1)
)
self.latent_axs.append(
plt.subplot2grid(
shape=(self.N_ROW, self.N_COL), loc=(2 + self.OBJ_ROW, 0), colspan=4, rowspan=4
)
)
self.latent_axs.append(
plt.subplot2grid(
shape=(self.N_ROW, self.N_COL), loc=(2 + self.OBJ_ROW, 4), colspan=4, rowspan=4
)
)
self.latent_axs.append(
plt.subplot2grid(
shape=(self.N_ROW, self.N_COL), loc=(2 + self.OBJ_ROW + 4, 4), colspan=4, rowspan=4
)
)
plt.tight_layout()
def reset(self):
for ax in self.obs_axs + self.act_axs + self.latent_axs + self.obj_axs + self.next_obj_axs:
ax.cla()
for ax in self.obj_axs + self.next_obj_axs + self.act_axs + self.obs_axs:
ax.axis('off')
for i in range(3):
self.latent_axs[i].set_xlim(-5, 5)
self.latent_axs[i].set_ylim(-5, 5)
self.latent_axs[0].set_title("Pre State Latent", fontsize=6)
self.latent_axs[1].set_title("Next State Latent", fontsize=6)
self.latent_axs[2].set_title("Pre State Latent +\n Transition", fontsize=6)
self.obs_axs[0].set_title("Pre State Latent", fontsize=6)
self.obs_axs[1].set_title("Next State Latent", fontsize=6)
self.act_axs[0].set_title("Action Moving Object", fontsize=6)
self.act_axs[1].set_title("Action Target Object", fontsize=6)
def plt_observations(self, obs, next_obs):
np_obs = np.transpose(obs[0].cpu().numpy(), (1,2,0))
np_next_obs = np.transpose(next_obs[0].cpu().numpy(), (1,2,0))
# print(np_obs.shape, np_next_obs.shape)
self.obs_axs[0].imshow(np_obs)
self.obs_axs[1].imshow(np_next_obs)
def plt_action(self, action):
np_mov_obj = np.transpose(action[0][0].cpu().numpy(), (1,2,0))
np_tar_obj = np.transpose(action[0][1].cpu().numpy(), (1,2,0))
# print(np_obs.shape, np_next_obs.shape)
self.act_axs[0].imshow(np_mov_obj)
self.act_axs[1].imshow(np_tar_obj)
def plt_objects(self, objs, next_objs):
for i in range(objs.size()[1]):
np_obj = np.transpose(objs[0][i].cpu().numpy(), (1,2,0))
np_next_obj = np.transpose(next_objs[0][i].cpu().numpy(), (1,2,0))
self.obj_axs[i].imshow(np_obj)
self.next_obj_axs[i].imshow(np_next_obj)
def plt_latent(self, state, next_state, pred_state):
legend = [[], [], []]
for i in range(state.size()[1]):
np_state = state[0][i].cpu().numpy()
np_next_state = next_state[0][i].cpu().numpy()
np_pred_state = pred_state[0][i].cpu().numpy()
# if i == 0:
# print(np_state, np_next_state, np_pred_state)
# print("-*40")
self.latent_axs[0].scatter(np_state[0], np_state[1], color=self.COLORS[i], marker='x', s=10)
self.latent_axs[1].scatter(np_next_state[0], np_next_state[1], color=self.COLORS[i], marker='x', s=10)
self.latent_axs[2].scatter(np_pred_state[0], np_pred_state[1], color=self.COLORS[i], marker='x', s=10)
#
for j, st in enumerate([np_state, np_next_state, np_pred_state]):
legend[j].append("{}-({:.2f},{:.2f})".format(self.OBJ_NAMES[i], st[0], st[1]))
for ax, lgd in zip(self.latent_axs, legend):
ax.legend(lgd, prop={'size': 6}, loc=2, ncol=2)
def show(self, interval=0.5):
plt.pause(interval)
def close(self):
plt.close()
| [
"numpy.ceil",
"matplotlib.pyplot.close",
"matplotlib.pyplot.subplot2grid",
"matplotlib.pyplot.subplots",
"matplotlib.cm.rainbow",
"matplotlib.pyplot.pause",
"numpy.linspace",
"matplotlib.pyplot.tight_layout"
] | [((549, 587), 'matplotlib.pyplot.subplots', 'plt.subplots', ([], {'figsize': 'self.FIGURE_SIZE'}), '(figsize=self.FIGURE_SIZE)\n', (561, 587), True, 'import matplotlib.pyplot as plt\n'), ((2120, 2138), 'matplotlib.pyplot.tight_layout', 'plt.tight_layout', ([], {}), '()\n', (2136, 2138), True, 'import matplotlib.pyplot as plt\n'), ((5057, 5076), 'matplotlib.pyplot.pause', 'plt.pause', (['interval'], {}), '(interval)\n', (5066, 5076), True, 'import matplotlib.pyplot as plt\n'), ((5107, 5118), 'matplotlib.pyplot.close', 'plt.close', ([], {}), '()\n', (5116, 5118), True, 'import matplotlib.pyplot as plt\n'), ((290, 303), 'matplotlib.cm.rainbow', 'cm.rainbow', (['x'], {}), '(x)\n', (300, 303), True, 'import matplotlib.cm as cm\n'), ((1603, 1704), 'matplotlib.pyplot.subplot2grid', 'plt.subplot2grid', ([], {'shape': '(self.N_ROW, self.N_COL)', 'loc': '(2 + self.OBJ_ROW, 0)', 'colspan': '(4)', 'rowspan': '(4)'}), '(shape=(self.N_ROW, self.N_COL), loc=(2 + self.OBJ_ROW, 0),\n colspan=4, rowspan=4)\n', (1619, 1704), True, 'import matplotlib.pyplot as plt\n'), ((1786, 1887), 'matplotlib.pyplot.subplot2grid', 'plt.subplot2grid', ([], {'shape': '(self.N_ROW, self.N_COL)', 'loc': '(2 + self.OBJ_ROW, 4)', 'colspan': '(4)', 'rowspan': '(4)'}), '(shape=(self.N_ROW, self.N_COL), loc=(2 + self.OBJ_ROW, 4),\n colspan=4, rowspan=4)\n', (1802, 1887), True, 'import matplotlib.pyplot as plt\n'), ((1969, 2074), 'matplotlib.pyplot.subplot2grid', 'plt.subplot2grid', ([], {'shape': '(self.N_ROW, self.N_COL)', 'loc': '(2 + self.OBJ_ROW + 4, 4)', 'colspan': '(4)', 'rowspan': '(4)'}), '(shape=(self.N_ROW, self.N_COL), loc=(2 + self.OBJ_ROW + 4,\n 4), colspan=4, rowspan=4)\n', (1985, 2074), True, 'import matplotlib.pyplot as plt\n'), ((313, 345), 'numpy.linspace', 'np.linspace', (['(0)', '(1)', 'num_obj_slots'], {}), '(0, 1, num_obj_slots)\n', (324, 345), True, 'import numpy as np\n'), ((370, 396), 'numpy.ceil', 'np.ceil', (['(num_obj_slots / 4)'], {}), '(num_obj_slots / 4)\n', (377, 396), True, 'import numpy as np\n'), ((801, 891), 'matplotlib.pyplot.subplot2grid', 'plt.subplot2grid', ([], {'shape': '(self.N_ROW, self.N_COL)', 'loc': '(0, i * 2)', 'colspan': '(2)', 'rowspan': '(2)'}), '(shape=(self.N_ROW, self.N_COL), loc=(0, i * 2), colspan=2,\n rowspan=2)\n', (817, 891), True, 'import matplotlib.pyplot as plt\n'), ((1013, 1103), 'matplotlib.pyplot.subplot2grid', 'plt.subplot2grid', ([], {'shape': '(self.N_ROW, self.N_COL)', 'loc': '(0, i * 2)', 'colspan': '(2)', 'rowspan': '(2)'}), '(shape=(self.N_ROW, self.N_COL), loc=(0, i * 2), colspan=2,\n rowspan=2)\n', (1029, 1103), True, 'import matplotlib.pyplot as plt\n'), ((1277, 1371), 'matplotlib.pyplot.subplot2grid', 'plt.subplot2grid', ([], {'shape': '(self.N_ROW, self.N_COL)', 'loc': '(2 + i, 0 + j)', 'colspan': '(1)', 'rowspan': '(1)'}), '(shape=(self.N_ROW, self.N_COL), loc=(2 + i, 0 + j),\n colspan=1, rowspan=1)\n', (1293, 1371), True, 'import matplotlib.pyplot as plt\n'), ((1448, 1542), 'matplotlib.pyplot.subplot2grid', 'plt.subplot2grid', ([], {'shape': '(self.N_ROW, self.N_COL)', 'loc': '(2 + i, 4 + j)', 'colspan': '(1)', 'rowspan': '(1)'}), '(shape=(self.N_ROW, self.N_COL), loc=(2 + i, 4 + j),\n colspan=1, rowspan=1)\n', (1464, 1542), True, 'import matplotlib.pyplot as plt\n')] |
import time
import numpy as np
import mxnet as mx
import matplotlib.pyplot as plt
from mxnet import nd
from mxnet import autograd
from mxnet import gluon
from mxnet.gluon import nn
def gpu_exists():
try:
mx.nd.zeros((1, ), ctx=mx.gpu(0))
except:
return False
return True
data_ctx = mx.cpu()
if gpu_exists():
print("Using GPU for model context.")
model_ctx = mx.gpu(0)
else:
print("Using CPU for model context.")
model_ctx = mx.cpu(0)
mx.random.seed(1)
# %%
# Load MNIST
mnist_data = mx.test_utils.get_mnist()
n_samples = 10
def show_samples(n_samples, mnist_data):
idx_list = np.random.choice(len(mnist_data["train_data"]), n_samples)
fig, axs = plt.subplots(1, n_samples)
for i, j in enumerate(idx_list):
axs[i].imshow(mnist_data["train_data"][j][0], cmap="Greys")
axs[i].get_xaxis().set_ticks([])
axs[i].get_yaxis().set_ticks([])
plt.show()
train_data = np.reshape(mnist_data["train_data"], (-1, 28*28))
test_data = np.reshape(mnist_data["test_data"], (-1, 28*28))
class VAE(gluon.HybridBlock):
def __init__(self, n_hidden=400, n_latent=2, n_layers=1, n_output=768, batch_size=100):
# TODO Continue implementation
pass
| [
"matplotlib.pyplot.show",
"mxnet.random.seed",
"mxnet.test_utils.get_mnist",
"numpy.reshape",
"mxnet.cpu",
"mxnet.gpu",
"matplotlib.pyplot.subplots"
] | [((314, 322), 'mxnet.cpu', 'mx.cpu', ([], {}), '()\n', (320, 322), True, 'import mxnet as mx\n'), ((484, 501), 'mxnet.random.seed', 'mx.random.seed', (['(1)'], {}), '(1)\n', (498, 501), True, 'import mxnet as mx\n'), ((535, 560), 'mxnet.test_utils.get_mnist', 'mx.test_utils.get_mnist', ([], {}), '()\n', (558, 560), True, 'import mxnet as mx\n'), ((953, 1004), 'numpy.reshape', 'np.reshape', (["mnist_data['train_data']", '(-1, 28 * 28)'], {}), "(mnist_data['train_data'], (-1, 28 * 28))\n", (963, 1004), True, 'import numpy as np\n'), ((1015, 1065), 'numpy.reshape', 'np.reshape', (["mnist_data['test_data']", '(-1, 28 * 28)'], {}), "(mnist_data['test_data'], (-1, 28 * 28))\n", (1025, 1065), True, 'import numpy as np\n'), ((399, 408), 'mxnet.gpu', 'mx.gpu', (['(0)'], {}), '(0)\n', (405, 408), True, 'import mxnet as mx\n'), ((473, 482), 'mxnet.cpu', 'mx.cpu', (['(0)'], {}), '(0)\n', (479, 482), True, 'import mxnet as mx\n'), ((709, 735), 'matplotlib.pyplot.subplots', 'plt.subplots', (['(1)', 'n_samples'], {}), '(1, n_samples)\n', (721, 735), True, 'import matplotlib.pyplot as plt\n'), ((927, 937), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (935, 937), True, 'import matplotlib.pyplot as plt\n'), ((241, 250), 'mxnet.gpu', 'mx.gpu', (['(0)'], {}), '(0)\n', (247, 250), True, 'import mxnet as mx\n')] |
#!/usr/bin/env python3
import os
import sys
import importlib
import h5py
import random
import numpy as np
from argparse import ArgumentParser
def main():
args = parse_args()
if not importlib.util.find_spec('chimeranet'):
print('ChimeraNet is not installed, import from source.')
sys.path.append(os.path.join(os.path.split(__file__)[0], '..'))
from keras.callbacks import Callback, CSVLogger
from chimeranet.models import probe_model_shape, load_model, ChimeraPPModel
class NameDataset(Callback):
def __init__(self, name, val_name=None):
self.name = name
self.val_name = val_name
def on_epoch_end(self, epoch, logs):
logs['dataset'] = self.name
logs['val_dataset'] = self.val_name
with h5py.File(args.train_data, 'r') as f:
_, T, F, C = f['y/embedding'].shape
# build/load model
if args.input_model is not None:
T_, F_, C_, D_ = probe_model_shape(args.input_model)
assert T == T_ and F == F_ and C == C_,\
'Incompatible dataset with the model'
model = load_model(args.input_model)
else:
cm = ChimeraPPModel(T, F, C, args.embedding_dims)
model = cm.build_model()
model.compile(
'rmsprop',
loss={
'embedding': cm.loss_deepclustering(),
'mask': cm.loss_mask()
},
loss_weights={
'embedding': 0.9,
'mask': 0.1
}
)
# train
train_generator = generate_data(
args.train_data, args.batch_size, shuffle=True)
train_steps = get_dataset_size(
args.train_data) // args.batch_size
if args.validation_data:
validation_generator = generate_data(
args.validation_data, args.batch_size)
validation_steps = get_dataset_size(
args.validation_data) // args.batch_size
else:
validation_generator = None
validation_steps = None
model.fit_generator(
train_generator,
steps_per_epoch=train_steps,
validation_data=validation_generator,
validation_steps=validation_steps,
initial_epoch=args.initial_epoch,
epochs=args.stop_epoch,
callbacks=[
NameDataset(args.train_data, args.validation_data),
CSVLogger(args.log, append=args.initial_epoch > 0),
],
)
model.save(args.output_model)
def get_dataset_size(filename):
with h5py.File(filename, 'r') as f:
sample_size = f['x'].shape[0]
return sample_size
def generate_data(filename, batch_size, shuffle=False):
while True:
for x, y in generate_data_one(filename, batch_size, shuffle):
yield x, y
def generate_data_one(filename, batch_size, shuffle=False):
with h5py.File(filename, 'r') as f:
sample_size = get_dataset_size(filename)
sample_idx = list(range(sample_size))
if shuffle:
random.shuffle(sample_idx)
sample_idxs = [
sorted(sample_idx[batch_i*batch_size:(batch_i+1)*batch_size])
for batch_i in range(sample_size // batch_size)
]
for sample_idx in sample_idxs:
x = f['x'][sample_idx]
y = dict(
(k, f['y/{}'.format(k)][sample_idx])
for k in ('mask', 'embedding')
)
if shuffle:
idx = np.arange(batch_size)
np.random.shuffle(idx)
x = x[idx]
y = dict((k, v[idx]) for k, v in y.items())
yield x, y
def parse_args():
parser = ArgumentParser()
parser.add_argument(
'-i', '--train-data', type=str, required=True,
metavar='PATH', help='Train dataset path'
)
parser.add_argument(
'-o', '--output-model', type=str, required=True,
metavar='PATH', help='Output model path'
)
parser.add_argument(
'-m', '--input-model', type=str,
metavar='PATH',
help='Input model path (train from this model)'
)
parser.add_argument(
'-d', '--embedding-dims', type=int, default=20,
metavar='D',
help='Dimension of embedding, ignored -m is given (default=20)'
)
parser.add_argument(
'-b', '--batch-size', type=int, default=32,
metavar='B',
help='Batch size of train/validation'
)
parser.add_argument(
'--validation-data', type=str,
metavar='PATH', help='Validation dtaset path'
)
parser.add_argument(
'--log', type=str,
metavar='PATH', help='Log path'
)
parser.add_argument(
'--stop-epoch', type=int, default=None,
metavar='N',
help='Train stops on this epoch (default=initial_epoch+1)'
)
parser.add_argument(
'--initial-epoch', type=int, default=0,
metavar='N',
help='Train starts on this epoch (default=0)'
)
args = parser.parse_args()
if args.stop_epoch is None:
args.stop_epoch = args.initial_epoch + 1
return args
if __name__ == '__main__':
main()
| [
"chimeranet.models.ChimeraPPModel",
"chimeranet.models.probe_model_shape",
"h5py.File",
"chimeranet.models.load_model",
"argparse.ArgumentParser",
"importlib.util.find_spec",
"random.shuffle",
"numpy.arange",
"keras.callbacks.CSVLogger",
"os.path.split",
"numpy.random.shuffle"
] | [((3651, 3667), 'argparse.ArgumentParser', 'ArgumentParser', ([], {}), '()\n', (3665, 3667), False, 'from argparse import ArgumentParser\n'), ((192, 230), 'importlib.util.find_spec', 'importlib.util.find_spec', (['"""chimeranet"""'], {}), "('chimeranet')\n", (216, 230), False, 'import importlib\n'), ((794, 825), 'h5py.File', 'h5py.File', (['args.train_data', '"""r"""'], {}), "(args.train_data, 'r')\n", (803, 825), False, 'import h5py\n'), ((962, 997), 'chimeranet.models.probe_model_shape', 'probe_model_shape', (['args.input_model'], {}), '(args.input_model)\n', (979, 997), False, 'from chimeranet.models import probe_model_shape, load_model, ChimeraPPModel\n'), ((1113, 1141), 'chimeranet.models.load_model', 'load_model', (['args.input_model'], {}), '(args.input_model)\n', (1123, 1141), False, 'from chimeranet.models import probe_model_shape, load_model, ChimeraPPModel\n'), ((1165, 1209), 'chimeranet.models.ChimeraPPModel', 'ChimeraPPModel', (['T', 'F', 'C', 'args.embedding_dims'], {}), '(T, F, C, args.embedding_dims)\n', (1179, 1209), False, 'from chimeranet.models import probe_model_shape, load_model, ChimeraPPModel\n'), ((2511, 2535), 'h5py.File', 'h5py.File', (['filename', '"""r"""'], {}), "(filename, 'r')\n", (2520, 2535), False, 'import h5py\n'), ((2839, 2863), 'h5py.File', 'h5py.File', (['filename', '"""r"""'], {}), "(filename, 'r')\n", (2848, 2863), False, 'import h5py\n'), ((2997, 3023), 'random.shuffle', 'random.shuffle', (['sample_idx'], {}), '(sample_idx)\n', (3011, 3023), False, 'import random\n'), ((2366, 2416), 'keras.callbacks.CSVLogger', 'CSVLogger', (['args.log'], {'append': '(args.initial_epoch > 0)'}), '(args.log, append=args.initial_epoch > 0)\n', (2375, 2416), False, 'from keras.callbacks import Callback, CSVLogger\n'), ((3448, 3469), 'numpy.arange', 'np.arange', (['batch_size'], {}), '(batch_size)\n', (3457, 3469), True, 'import numpy as np\n'), ((3486, 3508), 'numpy.random.shuffle', 'np.random.shuffle', (['idx'], {}), '(idx)\n', (3503, 3508), True, 'import numpy as np\n'), ((335, 358), 'os.path.split', 'os.path.split', (['__file__'], {}), '(__file__)\n', (348, 358), False, 'import os\n')] |
import numpy as np
import time
from tscc.optimization.optimizescale import optimizescale
# add constraints according to the a,b,c coefficients of quadratic function
def computescale(a, b, c, iters, stateRobo, limitsRobo,
weightSlack, guessScale, deltaT):
nObst = len(a)
for i in range(iters):
listLeft = []
listRight = []
for j in range(nObst):
# concave case that leads to linearization
if (a[j] <= 0 and c[j] <= 0 and b[j] >= 0):
tempLeft = (a[j] + b[j]/(2*guessScale))
tempRight = - (c[j] + (b[j]/2.0)*guessScale)
listLeft.append(tempLeft)
listRight.append(tempRight)
# convex case with a as coefficient of s^2
elif (a[j] >= 0):
root1 = (-b[j] + np.sqrt(b[j]**2 - 4*a[j]*c[j]))/(2*a[j])
root2 = (-b[j] - np.sqrt(b[j]**2 - 4*a[j]*c[j]))/(2*a[j])
scaleMax = max(root1, root2)
scaleMin = min(root1, root2)
if (scaleMin >= 0):
listLeft.append(-1.0)
listRight.append(-1*scaleMin**2)
listLeft.append(1.0)
listRight.append(scaleMax**2)
else:
if (scaleMax < 0):
raise ValueError('Invalid constraint')
else:
listLeft.append(1.0)
listRight.append(scaleMax**2)
# convex case with c as coefficient of (1/s)^2
elif (a[j] <= 0 and c[j] >= 0):
root1 = (-b[j] + np.sqrt(b[j]**2 - 4*a[j]*c[j]))/(2*a[j])
root2 = (-b[j] - np.sqrt(b[j]**2 - 4*a[j]*c[j]))/(2*a[j])
scaleMax = max(root1, root2)
scaleMin = min(root1, root2)
if(scaleMin > 0):
listLeft.append(-1.0)
listRight.append(-1.0/scaleMin**2)
listLeft.append(1.0)
listRight.append(1.0/scaleMax**2)
else:
if (scaleMax <= 0):
raise ValueError('Invalid constraint')
else:
listLeft.append(1.0)
listRight.append(1/scaleMax**2)
# satisfied for any scale greater than zero, so no constriants
else:
pass
guessScale = optimizescale(listLeft, listRight, stateRobo,
limitsRobo, weightSlack, deltaT)
return guessScale
| [
"tscc.optimization.optimizescale.optimizescale",
"numpy.sqrt"
] | [((2468, 2546), 'tscc.optimization.optimizescale.optimizescale', 'optimizescale', (['listLeft', 'listRight', 'stateRobo', 'limitsRobo', 'weightSlack', 'deltaT'], {}), '(listLeft, listRight, stateRobo, limitsRobo, weightSlack, deltaT)\n', (2481, 2546), False, 'from tscc.optimization.optimizescale import optimizescale\n'), ((833, 869), 'numpy.sqrt', 'np.sqrt', (['(b[j] ** 2 - 4 * a[j] * c[j])'], {}), '(b[j] ** 2 - 4 * a[j] * c[j])\n', (840, 869), True, 'import numpy as np\n'), ((907, 943), 'numpy.sqrt', 'np.sqrt', (['(b[j] ** 2 - 4 * a[j] * c[j])'], {}), '(b[j] ** 2 - 4 * a[j] * c[j])\n', (914, 943), True, 'import numpy as np\n'), ((1647, 1683), 'numpy.sqrt', 'np.sqrt', (['(b[j] ** 2 - 4 * a[j] * c[j])'], {}), '(b[j] ** 2 - 4 * a[j] * c[j])\n', (1654, 1683), True, 'import numpy as np\n'), ((1721, 1757), 'numpy.sqrt', 'np.sqrt', (['(b[j] ** 2 - 4 * a[j] * c[j])'], {}), '(b[j] ** 2 - 4 * a[j] * c[j])\n', (1728, 1757), True, 'import numpy as np\n')] |
#!/usr/bin/env python
"""plots.py: plots utility functions."""
__author__ = "<NAME>."
__copyright__ = "Copyright 2020, SuperDARN@VT"
__credits__ = []
__license__ = "MIT"
__version__ = "1.0"
__maintainer__ = "<NAME>."
__email__ = "<EMAIL>"
__status__ = "Research"
import os
import sys
sys.path.extend(["code/", "code/rt/", "code/sd/"])
import matplotlib as mpl
import matplotlib.pyplot as plt
from matplotlib.collections import LineCollection
import matplotlib.dates as mdates
import plot_utils
from matplotlib.colors import LogNorm
import numpy
import pydarn
import datetime as dt
import pickle
import numpy as np
class TracePlots(object):
"""
Plot ray-traced using .pickle files stored under data/{date}/{rad}/ folder
"""
def __init__(self, ev, rad, bm, f, dstart, dend):
self.ev = ev
self.rad = rad
self.freq = f
self.bm = bm
self.folder = "data/{dn}/{rad}/".format(dn=ev.strftime("%Y-%m-%d"), rad=rad)
file = self.folder + "%s_%s_%d.pickle"%(dstart.strftime("%H%M"), dend.strftime("%H%M"), f)
with open(file, "rb") as f: self.rdict = pickle.load(f)
self.dstart, self.dend = dstart, dend
return
def plot(self, ix_date, jx_azimuth, fname):
rdict = self.rdict
fname = self.folder + fname
fig = plt.figure(figsize=(14, 8))
ax, aax = plot_utils.curved_earth_axes()
elevbeg = list(rdict[ix_date][jx_azimuth].keys())[0]
edenstht = rdict[ix_date][jx_azimuth][elevbeg]["edensTHT"]
edensArr = rdict[ix_date][jx_azimuth][elevbeg]["edensARR"]
rays = rdict[ix_date][jx_azimuth][elevbeg]["ray_parameters"]
ionos = rdict[ix_date][jx_azimuth][elevbeg]["ionospheric_scatter"]
X, Y = numpy.meshgrid(edenstht, ax.Re + numpy.linspace(60,560,500))
im = aax.pcolormesh(X, Y, edensArr, cmap="cividis", norm=mpl.colors.LogNorm(vmax=1e12, vmin=1e8))
cbax = plot_utils.add_cbar(im, ax)
_ = cbax.set_ylabel(r"N$_{el}$ [$m^{-3}$]", fontsize=14)
ax.set_title(ix_date.strftime("%Y-%m-%d %H:%M") + " (%s, Beam-%d, %d MHz)"%(self.rad.upper(), self.bm, self.freq), fontsize=14)
ax.set_ylabel(r"Alt. [km]", size=16)
ax.set_xlabel(r"Ground range [km]", size=16)
for _el in rdict[ix_date][jx_azimuth].keys():
rays = rdict[ix_date][jx_azimuth][_el]["ray_parameters"]
aax.plot(rays["th"], numpy.array(rays["r"])*1e-3, c="#9658B1",
zorder=8, linewidth=1.)
range_markers = [0] + list(numpy.arange(180, 5000, 225))
x, y = [], []
for _el in rdict[ix_date][jx_azimuth].keys():
rays = rdict[ix_date][jx_azimuth][_el]["ray_parameters"]
grans = numpy.array(rays["gran"])*1e-3
th = numpy.array(rays["th"])
r = numpy.array(rays["r"])
for rm in range_markers:
inds = (grans >= rm)
if inds.any():
x.append( th[inds][0] )
y.append( r[inds][0]*1e-3 )
aax.scatter(x, y, color="w",s=1)
scatter_color = "k"
for _el in rdict[ix_date][jx_azimuth].keys():
ionos = rdict[ix_date][jx_azimuth][_el]["ionospheric_scatter"]
if ionos["nstep"] <= 0: continue
t = ionos["theta"]
r = ionos["altitude"]*1e-3
spts = numpy.array([t, r]).T.reshape(-1, 1, 2)
h = ionos["h"]*1e-3
rel = numpy.radians( ionos["ranelev"] )
r = numpy.sqrt( r**2 + h**2 + 2*r*h*numpy.sin( rel ) )
t = t + numpy.arcsin( h/r * numpy.cos( rel ) )
epts = numpy.array([t, r]).T.reshape(-1, 1, 2)
segments = numpy.concatenate([spts, epts], axis=1)
lcol = LineCollection( segments, zorder=10,linewidths=5. )
_ = lcol.set_color(scatter_color)
aax.add_collection( lcol )
for _el in rdict[ix_date][jx_azimuth].keys():
gscat = rdict[ix_date][jx_azimuth][_el]["ground_scatter"]
if gscat is not None: aax.scatter(gscat["theta"], ax.Re*numpy.ones(gscat["theta"].shape),
color=scatter_color, zorder=20)
ax.grid()
fig.savefig(fname)
img = plt.imread(fname)
img = img[220:620, 120:1380, :]
fig = plt.figure(figsize=(8, 3), dpi=180)
ax = fig.add_subplot(111)
ax.imshow(img)
ax.set_xticks([])
ax.set_yticks([])
fig.patch.set_visible(False)
ax.axis("off")
fig.savefig(fname, bbox_inches="tight")
plt.close()
return
def plot_rays(rad, ev, fs, dstart, dend, ix_date=5):
hdw = pydarn.read_hdw_file(rad)
for f in fs:
u = dstart
while u < dend:
fold = "data/{dn}/{rad}/".format(dn=ev.strftime("%Y-%m-%d"), rad=rad) + "traces_%02d/"%(f)
if not os.path.exists(fold): os.system("mkdir " + fold)
for bm in range(1):
fname = "traces_%02d/%s_%d_%d.png"%(f, u.strftime("%H%M"), bm, f)
boresite = hdw.boresight
offset = hdw.beams/2. - 0.5
beam_azim = round(boresite + (bm - offset)*hdw.beam_separation,2)
tp = TracePlots(ev, rad, bm, f, u, u)
tp.plot(u, beam_azim, fname)
u = u + dt.timedelta(minutes=ix_date)
return
def plot_2D_electron_density(H, T, edens, f0, lat, lon, p_time, fname, location):
p_time = p_time.replace(minute=int(p_time.minute/5)*5) if numpy.mod(p_time.minute, 5) != 0 else p_time
fig = plt.figure(figsize=(5, 3), dpi=180)
ax = fig.add_subplot(111)
ax.xaxis.set_major_formatter(mdates.DateFormatter(r"$%H^{%M}$"))
im = ax.contourf(T, H, edens.T/1e12, cmap="plasma", vmin=0, vmax=0.5)
pos = ax.get_position()
cpos = [pos.x1 + 0.025, pos.y0 + 0.0125,
0.015, pos.height * 0.8] # this list defines (left, bottom, width, height
cax = fig.add_axes(cpos)
boundaries = np.linspace(0, 0.5, 5)
cb2 = mpl.colorbar.ColorbarBase(cax, cmap=plt.cm.plasma, boundaries=boundaries, spacing="proportional", orientation="vertical")
cb2.set_label(r"$N_e, \times 10^{12}m^{-3}$")
maxx = numpy.argmax(edens, axis=1)
h = [H[i] for i in maxx]
ax.set_ylim(60, 460)
ax.set_ylabel("Height, km")
ax.set_xlim(T[0], T[-1])
ax.set_xlabel("Time, UT")
t_prior = p_time - dt.timedelta(minutes=30)
ax.axvline(t_prior, color="b", lw=0.8, ls="--")
t_post = p_time + dt.timedelta(minutes=30)
ax.axvline(t_post, color="r", lw=0.8, ls="--")
ax.axvline(p_time, color="k", lw=0.8, ls="--")
ax.text(0.01, 1.05, "Date: "+T[0].strftime("%Y-%m-%d"), va="center", ha="left", transform=ax.transAxes)
ax.text(0.99, 1.05, r"$T^{O_{max}}$: "+p_time.strftime("%H:%M UT"), va="center", ha="right", transform=ax.transAxes)
ax.text(1.02, 0.99, r"Location: %s"%(location), va="top", ha="center", rotation=90,
transform=ax.transAxes, fontdict={"size":6})
#fig.savefig(fname.format(d=2), bbox_inches="tight")
t_id = T.index(p_time)
tpri_id = T.index(t_prior)
tpos_id = T.index(t_post)
fig, axes = plt.subplots(figsize=(3, 3), dpi=150)
ax = axes
ax.set_ylabel("Height, km")
ax.set_xlabel(r"$N_e$, $m^{-3}$")
ax.semilogx(edens[t_id, :], H, color="k", lw=1., ls="-", label=r"$N_e[T^{O_{max}}]$")
ax.semilogx(edens[tpri_id, :], H, color="b", lw=1., ls="-", label=r"$N_e[T^{O_{max}}-30min]$")
ax.semilogx(edens[tpos_id, :], H, color="r", lw=1., ls="-", label=r"$N_e[T^{O_{max}}+30min]$")
ax.axhline(H[np.argmax(edens[t_id, :])], color="k", lw=0.6, ls="--")
ax.axhline(H[np.argmax(edens[tpri_id, :])], color="b", lw=0.6, ls="--")
ax.axhline(H[np.argmax(edens[tpos_id, :])], color="r", lw=0.6, ls="--")
ax.legend(loc=2)
ax.set_ylim(50,400)
ax.set_xlim(1e8,1e12)
ax.text(0.01, 1.05, "Date: "+T[0].strftime("%Y-%m-%d"), va="center", ha="left", transform=ax.transAxes)
ax.text(0.99, 1.05, r"$T^{O_{max}}$: "+p_time.strftime("%H:%M UT"), va="center", ha="right", transform=ax.transAxes)
ax.text(1.05, 0.99, "Location: "+location, va="top", ha="right", rotation=90, transform=ax.transAxes)
#fig.savefig(fname.format(d=1), bbox_inches="tight")
fig = plt.figure(figsize=(5, 3), dpi=180)
ax = fig.add_subplot(111)
ax.xaxis.set_major_formatter(mdates.DateFormatter(r"$%H^{%M}$"))
hF1, hF2 = H.tolist().index(150), H.tolist().index(240)
ax.semilogy(T, edens[:, hF1], "bo", ms=.8, ls="None", alpha=0.5, label=r"$N_e(hF_1=150km)$")
ax.semilogy(T, edens[:, hF2], "ro", ms=.8, ls="None", alpha=0.5, label=r"$N_e(hF_2=240km)$")
ax.axvline(T[np.argmin(edens[:, hF1])], ls="--", lw=0.8, color="b")
ax.axvline(T[np.argmin(edens[:, hF2])], ls="--", lw=0.8, color="r")
ax.legend(loc=4)
ax.set_xlim(T[0], T[-1])
ax.set_xlabel("Time, UT")
ax.set_ylabel(r"$N_e$, $m^{-3}$")
#fig.savefig(fname.format(d=0), bbox_inches="tight")
return
def plot_params(H, T, params, p_time, fname, location, pname, vlim, dlim):
keys = {"opdt":[r"$\frac{dO^+}{dt}, s^{-1}$", r"\frac{dO^+}{dt}"],
"dfield":[r"$D_{\vec{E}\times\vec{B}}, s^{-1}$", r"D_{\vec{E}\times\vec{B}}"],
"amb_diff":[r"$D_{\alpha}, s^{-1}$", r"D_{\alpha}"],
"dwind":[r"$D_{\vec{w}}, s^{-1}$", r"D_{\vec{w}}"],
"WI":[r"$W_I, ms^{-1}$", r"W_I"],
"U":[r"$U, ms^{-1}$", r"U"],
"V":[r"$V, ms^{-1}$", r"V"],
}
p_time = p_time.replace(minute=int(p_time.minute/5)*5) if numpy.mod(p_time.minute, 5) != 0 else p_time
fig = plt.figure(figsize=(5, 3), dpi=180)
ax = fig.add_subplot(111)
ax.xaxis.set_major_formatter(mdates.DateFormatter(r"$%H^{%M}$"))
im = ax.contourf(T, H, params.T, cmap="plasma", vmax=vlim[1], vmin=vlim[0])
pos = ax.get_position()
cpos = [pos.x1 + 0.025, pos.y0 + 0.0125,
0.015, pos.height * 0.8] # this list defines (left, bottom, width, height
cax = fig.add_axes(cpos)
boundaries = np.linspace(vlim[0], vlim[1], int((vlim[1]-vlim[0])/dlim))
cb2 = mpl.colorbar.ColorbarBase(cax, cmap=plt.cm.plasma, boundaries=boundaries, spacing="proportional", orientation="vertical")
cb2.set_label(keys[pname][0])
ax.set_ylim(60, 460)
ax.set_ylabel("Height, km")
ax.set_xlim(T[0], T[-1])
ax.set_xlabel("Time, UT")
t_prior = p_time - dt.timedelta(minutes=30)
ax.axvline(t_prior, color="b", lw=0.8, ls="--")
t_post = p_time + dt.timedelta(minutes=30)
ax.axvline(t_post, color="r", lw=0.8, ls="--")
ax.axvline(p_time, color="k", lw=0.8, ls="--")
ax.text(0.01, 1.05, "Date: "+T[0].strftime("%Y-%m-%d"), va="center", ha="left", transform=ax.transAxes)
ax.text(0.99, 1.05, r"$T^{O_{max}}$: "+p_time.strftime("%H:%M UT"), va="center", ha="right", transform=ax.transAxes)
ax.text(1.02, 0.99, r"Location: %s"%(location), va="top", ha="center", rotation=90,
transform=ax.transAxes, fontdict={"size":6})
#fig.savefig(fname.format(d=2), bbox_inches="tight")
t_id = T.index(p_time)
tpri_id = T.index(t_prior)
tpos_id = T.index(t_post)
fig, axes = plt.subplots(figsize=(3, 3), dpi=150)
ax = axes
ax.set_ylabel("Height, km")
ax.set_xlabel(keys[pname][0])
ax.plot(params[t_id, :], H, color="k", lw=1., ls="-", label=r"$%s[T^{O_{max}}]$"%keys[pname][1])
ax.plot(params[tpri_id, :], H, color="b", lw=1., ls="-", label=r"$%s[T^{O_{max}}-30min]$"%keys[pname][1])
ax.plot(params[tpos_id, :], H, color="r", lw=1., ls="-", label=r"$%s[T^{O_{max}}+30min]$"%keys[pname][1])
ax.axhline(H[np.argmax(params[t_id, :])], color="k", lw=0.6, ls="--")
ax.axhline(H[np.argmax(params[tpri_id, :])], color="b", lw=0.6, ls="--")
ax.axhline(H[np.argmax(params[tpos_id, :])], color="r", lw=0.6, ls="--")
ax.legend(loc=0)
ax.set_ylim(50,400)
ax.set_xlim(vlim)
ax.text(0.01, 1.05, "Date: "+T[0].strftime("%Y-%m-%d"), va="center", ha="left", transform=ax.transAxes)
ax.text(0.99, 1.05, r"$T^{O_{max}}$: "+p_time.strftime("%H:%M UT"), va="center", ha="right", transform=ax.transAxes)
ax.text(1.05, 0.99, "Location: "+location, va="top", ha="right", rotation=90, transform=ax.transAxes)
#fig.savefig(fname.format(d=1), bbox_inches="tight")
fig = plt.figure(figsize=(5, 3), dpi=180)
ax = fig.add_subplot(111)
ax.xaxis.set_major_formatter(mdates.DateFormatter(r"$%H^{%M}$"))
hF1, hF2 = H.tolist().index(150), H.tolist().index(240)
ax.plot(T, params[:, hF1], "bo", ms=.8, ls="None", alpha=0.5, label=r"$%s(hF_1=150km)$"%keys[pname][1])
ax.plot(T, params[:, hF2], "ro", ms=.8, ls="None", alpha=0.5, label=r"$%s(hF_2=240km)$"%keys[pname][1])
ax.axvline(T[np.argmin(params[:, hF1])], ls="--", lw=0.8, color="b")
ax.axvline(T[np.argmin(params[:, hF2])], ls="--", lw=0.8, color="r")
ax.legend(loc=0)
ax.set_xlim(T[0], T[-1])
ax.set_ylim(vlim)
ax.set_xlabel("Time, UT")
ax.set_ylabel(keys[pname][0])
#fig.savefig(fname.format(d=0), bbox_inches="tight")
return
def create_ts_label_plots(T, params, H, ylabel, yscale=None, ylim=None, xlabel=r"Time ($\tau$), UT"):
fig = plt.figure(figsize=(5, 3), dpi=180)
ax = fig.add_subplot(111)
ax.xaxis.set_major_formatter(mdates.DateFormatter(r"$%H^{%M}$"))
hF1, hF2 = H.tolist().index(150), H.tolist().index(240)
for lab, shape, param in zip([r"$\mathcal{E}$", r"$\mathcal{D}$"], ["o", "D"], params):
ax.plot(T, param[:,hF1], "r"+shape, ms=.8, ls="None", alpha=0.5, label=r"$hF_1=150km$ [%s]"%lab)
ax.plot(T, param[:,hF2], "b"+shape, ms=.8, ls="None", alpha=0.5, label=r"$hF_2=240km$ [%s]"%lab)
ax.legend(loc=4)
ax.set_xlabel(xlabel)
ax.set_ylabel(ylabel)
if yscale: ax.set_yscale(yscale)
if ylim: ax.set_ylim(ylim)
ax.set_xlim(T[0], T[-1])
return fig, ax
def normalize_data(data):
return (data - np.min(data)) / (np.max(data) - np.min(data))
def create_ts_plot(H, T, params, label, vlim=[0,1], xlim=[dt.datetime(2017,8,21,16), dt.datetime(2017,8,21,19,30)],
h1=150, h2=220, mult=None, units=r"m^{-3}s^{-%d}"):
fig = plt.figure(figsize=(5, 6), dpi=180)
ax = fig.add_subplot(211)
ax.xaxis.set_major_formatter(mdates.DateFormatter(r"$%H^{%M}$"))
hF1, hF2 = H.tolist().index(h1), H.tolist().index(h2)
u = params[:, hF1] if mult==1. else params[:, hF1] * mult[:, hF1]
ax.plot(T, u, "bo", ms=.8, ls="None", alpha=0.5, label=r"$%s(hF_1=%d km)$"%(label, h1))
ax.axvline(T[np.argmin(u)], ls="--", lw=0.8, color="b")
u = params[:, hF2] if mult==1. else params[:, hF2] * mult[:, hF2]
ax.plot(T, u, "ro", ms=.8, ls="None", alpha=0.5, label=r"$%s(hF_2=%d km)$"%(label, h2))
ax.axvline(T[np.argmin(params[:, hF2])], ls="--", lw=0.8, color="r")
ax.legend(loc=2)
ax.set_xlim(xlim)
#ax.set_ylim(vlim)
ax.set_ylabel(r"$%s$, $%s$"%(label, units%1))
ax = fig.add_subplot(212)
ax.xaxis.set_major_formatter(mdates.DateFormatter(r"$%H^{%M}$"))
hF1, hF2 = H.tolist().index(h1), H.tolist().index(h2)
u = params[:, hF1] if mult==1. else params[:, hF1] * mult[:, hF1]
u = np.diff(u, prepend=u[0])
ax.plot(T, u, "bo", ms=.8, ls="None", alpha=0.5, label=r"$%s(hF_1=%d km)$"%(label, h1))
#ax.axvline(T[np.argmin(u)], ls="--", lw=0.8, color="b")
u = params[:, hF2] if mult==1. else params[:, hF2] * mult[:, hF2]
u = np.diff(u, prepend=u[0])
ax.plot(T, u, "ro", ms=.8, ls="None", alpha=0.5, label=r"$%s(hF_2=%d km)$"%(label, h2))
ax.axhline(0, color="k", ls="--", lw=0.8)
#ax.axvline(T[np.argmin(u)], ls="--", lw=0.8, color="r")
ax.legend(loc=2)
ax.set_xlim(xlim)
ax.set_xlabel("Time, UT")
ax.set_ylabel(r"$\frac{\partial}{\partial t}$ $%s$, $%s$"%(label, units%2))
return
def create_rate_ts_plot(H, T, params, label, vlim=[0,1], xlim=[dt.datetime(2017,8,21,16), dt.datetime(2017,8,21,19,30)],
h1=150, h2=220, mult=None, units=r"s^{-2}"):
fig = plt.figure(figsize=(5, 3), dpi=180)
ax = fig.add_subplot(111)
ax.xaxis.set_major_formatter(mdates.DateFormatter(r"$%H^{%M}$"))
hF1, hF2 = H.tolist().index(h1), H.tolist().index(h2)
#params = normalize_data(params)
#params *= mult
u = params[:, hF1] if mult==1. else params[:, hF1] * mult[:, hF1]
u = np.diff(u, prepend=u[0])
ax.plot(T, u, "bo", ms=.8, ls="None", alpha=0.5, label=r"$%s(hF_1=%d km)$"%(label, h1))
ax.axvline(T[np.argmin(u)], ls="--", lw=0.8, color="b")
u = params[:, hF2] if mult==1. else params[:, hF2] * mult[:, hF2]
u = np.diff(u, prepend=u[0])
ax.plot(T, u, "ro", ms=.8, ls="None", alpha=0.5, label=r"$%s(hF_2=%d km)$"%(label, h2))
ax.axvline(T[np.argmin(u)], ls="--", lw=0.8, color="r")
ax.legend(loc=0)
ax.set_xlim(xlim)
#ax.set_ylim(vlim)
ax.set_xlabel("Time, UT")
ax.set_ylabel(r"$%s$, $%s$"%(label, units))
return
if __name__ == "__main__":
plot_rays("cvw", ev=dt.datetime(2017,8,21), fs=[12, 14, 16, 18], dstart=dt.datetime(2017,8,21,14,10),
dend=dt.datetime(2017,8,21,20,10), ix_date=5)
| [
"numpy.argmax",
"numpy.ones",
"numpy.argmin",
"matplotlib.pyplot.figure",
"pickle.load",
"matplotlib.colors.LogNorm",
"numpy.arange",
"numpy.sin",
"matplotlib.pyplot.imread",
"matplotlib.pyplot.close",
"sys.path.extend",
"os.path.exists",
"matplotlib.dates.DateFormatter",
"numpy.max",
"d... | [((287, 337), 'sys.path.extend', 'sys.path.extend', (["['code/', 'code/rt/', 'code/sd/']"], {}), "(['code/', 'code/rt/', 'code/sd/'])\n", (302, 337), False, 'import sys\n'), ((4759, 4784), 'pydarn.read_hdw_file', 'pydarn.read_hdw_file', (['rad'], {}), '(rad)\n', (4779, 4784), False, 'import pydarn\n'), ((5661, 5696), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(5, 3)', 'dpi': '(180)'}), '(figsize=(5, 3), dpi=180)\n', (5671, 5696), True, 'import matplotlib.pyplot as plt\n'), ((6090, 6112), 'numpy.linspace', 'np.linspace', (['(0)', '(0.5)', '(5)'], {}), '(0, 0.5, 5)\n', (6101, 6112), True, 'import numpy as np\n'), ((6123, 6248), 'matplotlib.colorbar.ColorbarBase', 'mpl.colorbar.ColorbarBase', (['cax'], {'cmap': 'plt.cm.plasma', 'boundaries': 'boundaries', 'spacing': '"""proportional"""', 'orientation': '"""vertical"""'}), "(cax, cmap=plt.cm.plasma, boundaries=boundaries,\n spacing='proportional', orientation='vertical')\n", (6148, 6248), True, 'import matplotlib as mpl\n'), ((6306, 6333), 'numpy.argmax', 'numpy.argmax', (['edens'], {'axis': '(1)'}), '(edens, axis=1)\n', (6318, 6333), False, 'import numpy\n'), ((7269, 7306), 'matplotlib.pyplot.subplots', 'plt.subplots', ([], {'figsize': '(3, 3)', 'dpi': '(150)'}), '(figsize=(3, 3), dpi=150)\n', (7281, 7306), True, 'import matplotlib.pyplot as plt\n'), ((8382, 8417), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(5, 3)', 'dpi': '(180)'}), '(figsize=(5, 3), dpi=180)\n', (8392, 8417), True, 'import matplotlib.pyplot as plt\n'), ((9728, 9763), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(5, 3)', 'dpi': '(180)'}), '(figsize=(5, 3), dpi=180)\n', (9738, 9763), True, 'import matplotlib.pyplot as plt\n'), ((10232, 10357), 'matplotlib.colorbar.ColorbarBase', 'mpl.colorbar.ColorbarBase', (['cax'], {'cmap': 'plt.cm.plasma', 'boundaries': 'boundaries', 'spacing': '"""proportional"""', 'orientation': '"""vertical"""'}), "(cax, cmap=plt.cm.plasma, boundaries=boundaries,\n spacing='proportional', orientation='vertical')\n", (10257, 10357), True, 'import matplotlib as mpl\n'), ((11294, 11331), 'matplotlib.pyplot.subplots', 'plt.subplots', ([], {'figsize': '(3, 3)', 'dpi': '(150)'}), '(figsize=(3, 3), dpi=150)\n', (11306, 11331), True, 'import matplotlib.pyplot as plt\n'), ((12435, 12470), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(5, 3)', 'dpi': '(180)'}), '(figsize=(5, 3), dpi=180)\n', (12445, 12470), True, 'import matplotlib.pyplot as plt\n'), ((13309, 13344), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(5, 3)', 'dpi': '(180)'}), '(figsize=(5, 3), dpi=180)\n', (13319, 13344), True, 'import matplotlib.pyplot as plt\n'), ((14289, 14324), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(5, 6)', 'dpi': '(180)'}), '(figsize=(5, 6), dpi=180)\n', (14299, 14324), True, 'import matplotlib.pyplot as plt\n'), ((15298, 15322), 'numpy.diff', 'np.diff', (['u'], {'prepend': 'u[0]'}), '(u, prepend=u[0])\n', (15305, 15322), True, 'import numpy as np\n'), ((15555, 15579), 'numpy.diff', 'np.diff', (['u'], {'prepend': 'u[0]'}), '(u, prepend=u[0])\n', (15562, 15579), True, 'import numpy as np\n'), ((16144, 16179), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(5, 3)', 'dpi': '(180)'}), '(figsize=(5, 3), dpi=180)\n', (16154, 16179), True, 'import matplotlib.pyplot as plt\n'), ((16473, 16497), 'numpy.diff', 'np.diff', (['u'], {'prepend': 'u[0]'}), '(u, prepend=u[0])\n', (16480, 16497), True, 'import numpy as np\n'), ((16729, 16753), 'numpy.diff', 'np.diff', (['u'], {'prepend': 'u[0]'}), '(u, prepend=u[0])\n', (16736, 16753), True, 'import numpy as np\n'), ((1325, 1352), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(14, 8)'}), '(figsize=(14, 8))\n', (1335, 1352), True, 'import matplotlib.pyplot as plt\n'), ((1371, 1401), 'plot_utils.curved_earth_axes', 'plot_utils.curved_earth_axes', ([], {}), '()\n', (1399, 1401), False, 'import plot_utils\n'), ((1957, 1984), 'plot_utils.add_cbar', 'plot_utils.add_cbar', (['im', 'ax'], {}), '(im, ax)\n', (1976, 1984), False, 'import plot_utils\n'), ((4331, 4348), 'matplotlib.pyplot.imread', 'plt.imread', (['fname'], {}), '(fname)\n', (4341, 4348), True, 'import matplotlib.pyplot as plt\n'), ((4403, 4438), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '(8, 3)', 'dpi': '(180)'}), '(figsize=(8, 3), dpi=180)\n', (4413, 4438), True, 'import matplotlib.pyplot as plt\n'), ((4664, 4675), 'matplotlib.pyplot.close', 'plt.close', ([], {}), '()\n', (4673, 4675), True, 'import matplotlib.pyplot as plt\n'), ((5760, 5793), 'matplotlib.dates.DateFormatter', 'mdates.DateFormatter', (['"""$%H^{%M}$"""'], {}), "('$%H^{%M}$')\n", (5780, 5793), True, 'import matplotlib.dates as mdates\n'), ((6502, 6526), 'datetime.timedelta', 'dt.timedelta', ([], {'minutes': '(30)'}), '(minutes=30)\n', (6514, 6526), True, 'import datetime as dt\n'), ((6601, 6625), 'datetime.timedelta', 'dt.timedelta', ([], {'minutes': '(30)'}), '(minutes=30)\n', (6613, 6625), True, 'import datetime as dt\n'), ((8481, 8514), 'matplotlib.dates.DateFormatter', 'mdates.DateFormatter', (['"""$%H^{%M}$"""'], {}), "('$%H^{%M}$')\n", (8501, 8514), True, 'import matplotlib.dates as mdates\n'), ((9827, 9860), 'matplotlib.dates.DateFormatter', 'mdates.DateFormatter', (['"""$%H^{%M}$"""'], {}), "('$%H^{%M}$')\n", (9847, 9860), True, 'import matplotlib.dates as mdates\n'), ((10527, 10551), 'datetime.timedelta', 'dt.timedelta', ([], {'minutes': '(30)'}), '(minutes=30)\n', (10539, 10551), True, 'import datetime as dt\n'), ((10626, 10650), 'datetime.timedelta', 'dt.timedelta', ([], {'minutes': '(30)'}), '(minutes=30)\n', (10638, 10650), True, 'import datetime as dt\n'), ((12534, 12567), 'matplotlib.dates.DateFormatter', 'mdates.DateFormatter', (['"""$%H^{%M}$"""'], {}), "('$%H^{%M}$')\n", (12554, 12567), True, 'import matplotlib.dates as mdates\n'), ((13408, 13441), 'matplotlib.dates.DateFormatter', 'mdates.DateFormatter', (['"""$%H^{%M}$"""'], {}), "('$%H^{%M}$')\n", (13428, 13441), True, 'import matplotlib.dates as mdates\n'), ((14150, 14178), 'datetime.datetime', 'dt.datetime', (['(2017)', '(8)', '(21)', '(16)'], {}), '(2017, 8, 21, 16)\n', (14161, 14178), True, 'import datetime as dt\n'), ((14177, 14209), 'datetime.datetime', 'dt.datetime', (['(2017)', '(8)', '(21)', '(19)', '(30)'], {}), '(2017, 8, 21, 19, 30)\n', (14188, 14209), True, 'import datetime as dt\n'), ((14388, 14421), 'matplotlib.dates.DateFormatter', 'mdates.DateFormatter', (['"""$%H^{%M}$"""'], {}), "('$%H^{%M}$')\n", (14408, 14421), True, 'import matplotlib.dates as mdates\n'), ((15125, 15158), 'matplotlib.dates.DateFormatter', 'mdates.DateFormatter', (['"""$%H^{%M}$"""'], {}), "('$%H^{%M}$')\n", (15145, 15158), True, 'import matplotlib.dates as mdates\n'), ((16007, 16035), 'datetime.datetime', 'dt.datetime', (['(2017)', '(8)', '(21)', '(16)'], {}), '(2017, 8, 21, 16)\n', (16018, 16035), True, 'import datetime as dt\n'), ((16034, 16066), 'datetime.datetime', 'dt.datetime', (['(2017)', '(8)', '(21)', '(19)', '(30)'], {}), '(2017, 8, 21, 19, 30)\n', (16045, 16066), True, 'import datetime as dt\n'), ((16243, 16276), 'matplotlib.dates.DateFormatter', 'mdates.DateFormatter', (['"""$%H^{%M}$"""'], {}), "('$%H^{%M}$')\n", (16263, 16276), True, 'import matplotlib.dates as mdates\n'), ((1119, 1133), 'pickle.load', 'pickle.load', (['f'], {}), '(f)\n', (1130, 1133), False, 'import pickle\n'), ((2843, 2866), 'numpy.array', 'numpy.array', (["rays['th']"], {}), "(rays['th'])\n", (2854, 2866), False, 'import numpy\n'), ((2883, 2905), 'numpy.array', 'numpy.array', (["rays['r']"], {}), "(rays['r'])\n", (2894, 2905), False, 'import numpy\n'), ((3529, 3560), 'numpy.radians', 'numpy.radians', (["ionos['ranelev']"], {}), "(ionos['ranelev'])\n", (3542, 3560), False, 'import numpy\n'), ((3771, 3810), 'numpy.concatenate', 'numpy.concatenate', (['[spts, epts]'], {'axis': '(1)'}), '([spts, epts], axis=1)\n', (3788, 3810), False, 'import numpy\n'), ((3830, 3881), 'matplotlib.collections.LineCollection', 'LineCollection', (['segments'], {'zorder': '(10)', 'linewidths': '(5.0)'}), '(segments, zorder=10, linewidths=5.0)\n', (3844, 3881), False, 'from matplotlib.collections import LineCollection\n'), ((5606, 5633), 'numpy.mod', 'numpy.mod', (['p_time.minute', '(5)'], {}), '(p_time.minute, 5)\n', (5615, 5633), False, 'import numpy\n'), ((7696, 7721), 'numpy.argmax', 'np.argmax', (['edens[t_id, :]'], {}), '(edens[t_id, :])\n', (7705, 7721), True, 'import numpy as np\n'), ((7769, 7797), 'numpy.argmax', 'np.argmax', (['edens[tpri_id, :]'], {}), '(edens[tpri_id, :])\n', (7778, 7797), True, 'import numpy as np\n'), ((7845, 7873), 'numpy.argmax', 'np.argmax', (['edens[tpos_id, :]'], {}), '(edens[tpos_id, :])\n', (7854, 7873), True, 'import numpy as np\n'), ((8788, 8812), 'numpy.argmin', 'np.argmin', (['edens[:, hF1]'], {}), '(edens[:, hF1])\n', (8797, 8812), True, 'import numpy as np\n'), ((8860, 8884), 'numpy.argmin', 'np.argmin', (['edens[:, hF2]'], {}), '(edens[:, hF2])\n', (8869, 8884), True, 'import numpy as np\n'), ((9673, 9700), 'numpy.mod', 'numpy.mod', (['p_time.minute', '(5)'], {}), '(p_time.minute, 5)\n', (9682, 9700), False, 'import numpy\n'), ((11750, 11776), 'numpy.argmax', 'np.argmax', (['params[t_id, :]'], {}), '(params[t_id, :])\n', (11759, 11776), True, 'import numpy as np\n'), ((11824, 11853), 'numpy.argmax', 'np.argmax', (['params[tpri_id, :]'], {}), '(params[tpri_id, :])\n', (11833, 11853), True, 'import numpy as np\n'), ((11901, 11930), 'numpy.argmax', 'np.argmax', (['params[tpos_id, :]'], {}), '(params[tpos_id, :])\n', (11910, 11930), True, 'import numpy as np\n'), ((12863, 12888), 'numpy.argmin', 'np.argmin', (['params[:, hF1]'], {}), '(params[:, hF1])\n', (12872, 12888), True, 'import numpy as np\n'), ((12936, 12961), 'numpy.argmin', 'np.argmin', (['params[:, hF2]'], {}), '(params[:, hF2])\n', (12945, 12961), True, 'import numpy as np\n'), ((14045, 14057), 'numpy.min', 'np.min', (['data'], {}), '(data)\n', (14051, 14057), True, 'import numpy as np\n'), ((14062, 14074), 'numpy.max', 'np.max', (['data'], {}), '(data)\n', (14068, 14074), True, 'import numpy as np\n'), ((14077, 14089), 'numpy.min', 'np.min', (['data'], {}), '(data)\n', (14083, 14089), True, 'import numpy as np\n'), ((14662, 14674), 'numpy.argmin', 'np.argmin', (['u'], {}), '(u)\n', (14671, 14674), True, 'import numpy as np\n'), ((14885, 14910), 'numpy.argmin', 'np.argmin', (['params[:, hF2]'], {}), '(params[:, hF2])\n', (14894, 14910), True, 'import numpy as np\n'), ((16607, 16619), 'numpy.argmin', 'np.argmin', (['u'], {}), '(u)\n', (16616, 16619), True, 'import numpy as np\n'), ((16863, 16875), 'numpy.argmin', 'np.argmin', (['u'], {}), '(u)\n', (16872, 16875), True, 'import numpy as np\n'), ((17113, 17137), 'datetime.datetime', 'dt.datetime', (['(2017)', '(8)', '(21)'], {}), '(2017, 8, 21)\n', (17124, 17137), True, 'import datetime as dt\n'), ((17165, 17197), 'datetime.datetime', 'dt.datetime', (['(2017)', '(8)', '(21)', '(14)', '(10)'], {}), '(2017, 8, 21, 14, 10)\n', (17176, 17197), True, 'import datetime as dt\n'), ((17214, 17246), 'datetime.datetime', 'dt.datetime', (['(2017)', '(8)', '(21)', '(20)', '(10)'], {}), '(2017, 8, 21, 20, 10)\n', (17225, 17246), True, 'import datetime as dt\n'), ((1807, 1835), 'numpy.linspace', 'numpy.linspace', (['(60)', '(560)', '(500)'], {}), '(60, 560, 500)\n', (1821, 1835), False, 'import numpy\n'), ((1901, 1959), 'matplotlib.colors.LogNorm', 'mpl.colors.LogNorm', ([], {'vmax': '(1000000000000.0)', 'vmin': '(100000000.0)'}), '(vmax=1000000000000.0, vmin=100000000.0)\n', (1919, 1959), True, 'import matplotlib as mpl\n'), ((2600, 2628), 'numpy.arange', 'numpy.arange', (['(180)', '(5000)', '(225)'], {}), '(180, 5000, 225)\n', (2612, 2628), False, 'import numpy\n'), ((2795, 2820), 'numpy.array', 'numpy.array', (["rays['gran']"], {}), "(rays['gran'])\n", (2806, 2820), False, 'import numpy\n'), ((4967, 4987), 'os.path.exists', 'os.path.exists', (['fold'], {}), '(fold)\n', (4981, 4987), False, 'import os\n'), ((4989, 5015), 'os.system', 'os.system', (["('mkdir ' + fold)"], {}), "('mkdir ' + fold)\n", (4998, 5015), False, 'import os\n'), ((2449, 2471), 'numpy.array', 'numpy.array', (["rays['r']"], {}), "(rays['r'])\n", (2460, 2471), False, 'import numpy\n'), ((5420, 5449), 'datetime.timedelta', 'dt.timedelta', ([], {'minutes': 'ix_date'}), '(minutes=ix_date)\n', (5432, 5449), True, 'import datetime as dt\n'), ((3439, 3458), 'numpy.array', 'numpy.array', (['[t, r]'], {}), '([t, r])\n', (3450, 3458), False, 'import numpy\n'), ((3611, 3625), 'numpy.sin', 'numpy.sin', (['rel'], {}), '(rel)\n', (3620, 3625), False, 'import numpy\n'), ((3670, 3684), 'numpy.cos', 'numpy.cos', (['rel'], {}), '(rel)\n', (3679, 3684), False, 'import numpy\n'), ((3708, 3727), 'numpy.array', 'numpy.array', (['[t, r]'], {}), '([t, r])\n', (3719, 3727), False, 'import numpy\n'), ((4159, 4191), 'numpy.ones', 'numpy.ones', (["gscat['theta'].shape"], {}), "(gscat['theta'].shape)\n", (4169, 4191), False, 'import numpy\n')] |
#!/usr/bin/python3
# Tested with Python 3.8.6
#------------------------------------------------------------------------------
# runEmceeAfterglow.py
#------------------------------------------------------------------------------
# Authors: <NAME>, <NAME>
# Oregon State University
#------------------------------------------------------------------------------
"""
Use emcee module to generate parameter fits based on DL afterglow code.
Imported files
--------------
init_params.py
plots.py
multibandDataMooley.py
exceptionHandler.py
<afterglowModel.py>
Functions
--------------
cleanTempFolder(None)
Remove files from /temp folder.
Used by: main()
createParamFiles(*args)
Store emcee parameter values in .dat file.
Used by: main()
runAfterglow(*args)
Call afterglow script to calculate lightcurves.
Used by: logLikelihood()
logPrior(*agrs)
Create parameter lables using math text.
Used by: logProbability()
logLikelihood(*args)
Used by: logProbability()
logProbability(*args)
Used by: main()
main(None)
Run emcee package, save and plot results.
"""
# Standard Python library imports
import numpy as np
from math import log10
import time
import shutil # for cleaning temp folder
import os
os.environ["OMP_NUM_THREADS"] = "1"
import sys
print("Python version {}".format(sys.version))
import multiprocessing
from multiprocessing import Pool
# Emcee imports
import emcee
print("emcee version", emcee.__version__)
import tqdm # for progress bar
# Companion scripts
from cleanDataGW170817 import time_obs, flux_obs, flux_uncert
from init_params import params_list
# import <yourAfterglowModel> as run_ag
from exceptionHandler import exception_handler
def cleanTempFolder():
"""Remove files from temp folder."""
folder = "./temp/"
for filename in os.listdir(folder):
file_path = folder + filename
if os.path.isfile(file_path) or os.path.islink(file_path):
os.unlink(file_path)
elif os.path.isdir(file_path):
shutil.rmtree(file_path)
def createParamFiles(params_list):
"""
Determine whether a duplicate file exists
using pid and timestamp as identifiers.
If file exists, create a new one with a different timestamp.
If not, create a temp file containing emcee parameter samples.
Parameters
----------
params_list: list
List of emcee parameter values.
Returns
----------
params_datafile: .dat file
Contains emcee parameter values for use by runAfterglow.
"""
folder = "./temp/"
filename = "params-pid-{}-time-{}.dat".format(
str(os.getpid()), str(time.time()))
# Ensure filenames are unique
if os.path.isfile(folder + filename):
new_filename = 'params-pid-{}-time-{}.dat'.format(
str(os.getpid()), str(time.time()))
params_datafile = folder + new_filename
else:
params_datafile = folder + filename
dataout = []
for i, item in enumerate(params_list):
dataout.append(item[0])
np.savetxt(params_datafile, dataout, fmt='%s')
return params_datafile
def runAfterglow(eps_E, eps_B, p_e,
n_ISM, E_j, E_c, theta_j,
theta_c, theta_obs, Gamma_0):
"""
Convert emcee parameter values from log space to linear space.
Call afterglow script runAfterglowDL.py to calculate lightcurves.
Parameters
----------
theta = {eps_E,...,Gamma_0}
Returns
----------
lightcurve: float
"""
params = zip((eps_E, eps_B, p_e,
n_ISM, E_j, E_c, theta_j,
theta_c, theta_obs, Gamma_0))
params_list = (list(params))
params_datafile = createParamFiles(params_list)
lightcurve = run_ag.main(params_datafile=params_datafile)
return lightcurve
def logPrior(theta):
"""
Define flat ("uninformative") prior distributions for
a set of parameters.
Parameters
----------
theta: set of parameters
Returns
----------
0.0 if sample drawn within the bounds, -infinity otherwise.
"""
eps_E, eps_B, p_e, \
n_ISM, E_j, E_c, theta_j, \
theta_c, theta_obs, Gamma_0 = theta
# NOTE: eps_E, eps_B, n_ISM, E_j, E_c, and Gamma_0 are flat priors
# in log space
if (-4 < eps_E < -0.3 and
-4 < eps_B < -0.3 and
2 < p_e < 2.5 and
-4 < n_ISM < -0.3 and
-4 < E_j < 50 and
-4 < E_c < 49 and
0 < theta_j < 10 and
theta_j + 0.6 < theta_c < 20 and
0 < theta_obs < 90 and
-4 < Gamma_0 < 2.7):
return 0.0
return -np.inf
def logLikelihood(theta, x, y, yerr):
"""
Define log-likelihood function assuming
a Gaussian distribution.
Parameters
----------
theta: set of parameters
y: array, float
Observed flux
yerr: array, float
Observed flux uncertainty
Returns
----------
-0.5 * np.sum(((y-model)/yerr)**2): float
Likelihood function
"""
eps_E, eps_B, p_e, \
n_ISM, E_j, E_c, theta_j, \
theta_c, theta_obs, Gamma_0 = theta
lightcurve = runAfterglow(eps_E, eps_B, p_e,
n_ISM, E_j, E_c, theta_j,
theta_c, theta_obs, Gamma_0)
model = lightcurve
return -0.5 * np.sum(((y-model)/yerr)**2)
def logProbability(theta, x, y, yerr):
"""Define full log-probabilty function."""
if not np.isfinite(logPrior(theta)):
return -np.inf
return logPrior(theta) + logLikelihood(theta, x, y, yerr)
def emceeSampler(params_list):
""""
Run emcee sampler and check for convergence every n steps.
Parameters
----------
params_list: list, float
NOTE: This is a global variable,
imported from init_params.py (see imports list, line 63).
Returns
----------
None
"""
def _prepEmcee(params_list, Gaussian_ball=False):
"""
Iniitalize walkers around initial guess.
If 'Gaussian_ball' is set to True, initialize walkers in a small
Gaussian ball around the inital guess.
"""
num_params = len(params_list)
print("# of parameters emcee is fitting: {}".format(num_params))
print("Initial parameter guesses:{}".format(params_list))
params_list = np.reshape(params_list, (1, num_params))
if Gaussian_ball == True:
pos = params_list + 1e-4 * np.random.randn(n_walkers, num_params)
else:
pos = params_list * np.random.randn(n_walkers, num_params)
print(pos)
print("Initial walkers set.")
nwalkers, ndim = pos.shape
return nwalkers, ndim, pos
def _createBackendFile():
"""Generate a .h5 backend file to save and monitor progress."""
print(os.getcwd())
folder = "./backend"
datestamp = time.strftime("%Y%m%d-%H%M")
filename = "backend-file-{}.h5".format(datestamp)
backend = emcee.backends.HDFBackend(folder+filename)
return backend
def _saveResults(backend, samples):
"""Rename backend file to match when the emcee run completed."""
datestamp = time.strftime("%Y%m%d-%H%M")
backend_folder = './backend/'
filename = "backend-file-{}.h5".format(datestamp)
os.rename(backend.filename,
backend_folder + filename)
def _runEmcee(backend, nwalkers, ndim, pos):
"""
Set up a pool process to run emcee in parallel.
Run emcee sampler and check for convergence very n steps,
where n is user-defined.
"""
backend.reset(nwalkers, ndim)
index = 0
autocorr = np.empty(max_iter)
old_tau = np.inf
# Set up parallel processing
with Pool(processes = n_processes) as pool:
sampler = emcee.EnsembleSampler(nwalkers,
ndim,
logProbability,
args = (x,y,yerr),
backend=backend,
pool=pool)
# Run emcee
for sample in sampler.sample(
pos, iterations=max_iter, progress=True):
#print("log_prob = {} ".format(sampler.get_log_prob()))
#print("tau = {}".format(sampler.get_autocorr_time()))
#print("acceptance fraction = {} ".format(sampler.acceptance_fraction))
# Check for convergence very "check_iter" steps
if sampler.iteration % check_iter:
continue
tau = sampler.get_autocorr_time(tol=0)
autocorr[index] = np.mean(tau)
index += 1
converged = np.all(tau * 100 < sampler.iteration)
converged &= np.all(np.abs(old_tau - tau) / tau < 0.01)
if converged:
break
old_tau = tau
# Get samples
samples = sampler.chain[:, :, :].reshape((-1,ndim))
print(samples.shape, samples)
return samples
backend = _createBackendFile()
nwalkers, ndim, pos = _prepEmcee(params_list)
samples = _runEmcee(backend, nwalkers, ndim, pos)
_saveResults(backend, samples)
print("Emcee run complete. Access backend file to plot.")
@exception_handler
def main():
"""Clean temp folder and run emcee sampler."""
cleanTempFolder()
emceeSampler(params_list)
if __name__ == "__main__":
# Global variables from imports
x = time_obs
y = flux_obs
yerr = flux_uncert
num_params = len(params_list)
params = np.reshape(params_list, (1, num_params))
# User-defined global variables
n_walkers = 20
n_processes = 1
max_iter = 1
# Check for convergence every n iterations
# NOTE: max_iter must be divisible by n
check_iter = 1
# Run script
main()
| [
"numpy.sum",
"os.unlink",
"numpy.abs",
"numpy.empty",
"time.strftime",
"os.path.isfile",
"os.path.islink",
"numpy.mean",
"shutil.rmtree",
"numpy.random.randn",
"emcee.backends.HDFBackend",
"numpy.savetxt",
"numpy.reshape",
"os.rename",
"multiprocessing.Pool",
"os.listdir",
"numpy.all... | [((1831, 1849), 'os.listdir', 'os.listdir', (['folder'], {}), '(folder)\n', (1841, 1849), False, 'import os\n'), ((2724, 2757), 'os.path.isfile', 'os.path.isfile', (['(folder + filename)'], {}), '(folder + filename)\n', (2738, 2757), False, 'import os\n'), ((3065, 3111), 'numpy.savetxt', 'np.savetxt', (['params_datafile', 'dataout'], {'fmt': '"""%s"""'}), "(params_datafile, dataout, fmt='%s')\n", (3075, 3111), True, 'import numpy as np\n'), ((9819, 9859), 'numpy.reshape', 'np.reshape', (['params_list', '(1, num_params)'], {}), '(params_list, (1, num_params))\n', (9829, 9859), True, 'import numpy as np\n'), ((5353, 5386), 'numpy.sum', 'np.sum', (['(((y - model) / yerr) ** 2)'], {}), '(((y - model) / yerr) ** 2)\n', (5359, 5386), True, 'import numpy as np\n'), ((6375, 6415), 'numpy.reshape', 'np.reshape', (['params_list', '(1, num_params)'], {}), '(params_list, (1, num_params))\n', (6385, 6415), True, 'import numpy as np\n'), ((6924, 6952), 'time.strftime', 'time.strftime', (['"""%Y%m%d-%H%M"""'], {}), "('%Y%m%d-%H%M')\n", (6937, 6952), False, 'import time\n'), ((7029, 7073), 'emcee.backends.HDFBackend', 'emcee.backends.HDFBackend', (['(folder + filename)'], {}), '(folder + filename)\n', (7054, 7073), False, 'import emcee\n'), ((7230, 7258), 'time.strftime', 'time.strftime', (['"""%Y%m%d-%H%M"""'], {}), "('%Y%m%d-%H%M')\n", (7243, 7258), False, 'import time\n'), ((7363, 7417), 'os.rename', 'os.rename', (['backend.filename', '(backend_folder + filename)'], {}), '(backend.filename, backend_folder + filename)\n', (7372, 7417), False, 'import os\n'), ((7755, 7773), 'numpy.empty', 'np.empty', (['max_iter'], {}), '(max_iter)\n', (7763, 7773), True, 'import numpy as np\n'), ((1900, 1925), 'os.path.isfile', 'os.path.isfile', (['file_path'], {}), '(file_path)\n', (1914, 1925), False, 'import os\n'), ((1929, 1954), 'os.path.islink', 'os.path.islink', (['file_path'], {}), '(file_path)\n', (1943, 1954), False, 'import os\n'), ((1968, 1988), 'os.unlink', 'os.unlink', (['file_path'], {}), '(file_path)\n', (1977, 1988), False, 'import os\n'), ((2002, 2026), 'os.path.isdir', 'os.path.isdir', (['file_path'], {}), '(file_path)\n', (2015, 2026), False, 'import os\n'), ((2650, 2661), 'os.getpid', 'os.getpid', ([], {}), '()\n', (2659, 2661), False, 'import os\n'), ((2668, 2679), 'time.time', 'time.time', ([], {}), '()\n', (2677, 2679), False, 'import time\n'), ((6862, 6873), 'os.getcwd', 'os.getcwd', ([], {}), '()\n', (6871, 6873), False, 'import os\n'), ((7860, 7887), 'multiprocessing.Pool', 'Pool', ([], {'processes': 'n_processes'}), '(processes=n_processes)\n', (7864, 7887), False, 'from multiprocessing import Pool\n'), ((7921, 8025), 'emcee.EnsembleSampler', 'emcee.EnsembleSampler', (['nwalkers', 'ndim', 'logProbability'], {'args': '(x, y, yerr)', 'backend': 'backend', 'pool': 'pool'}), '(nwalkers, ndim, logProbability, args=(x, y, yerr),\n backend=backend, pool=pool)\n', (7942, 8025), False, 'import emcee\n'), ((2040, 2064), 'shutil.rmtree', 'shutil.rmtree', (['file_path'], {}), '(file_path)\n', (2053, 2064), False, 'import shutil\n'), ((2834, 2845), 'os.getpid', 'os.getpid', ([], {}), '()\n', (2843, 2845), False, 'import os\n'), ((2852, 2863), 'time.time', 'time.time', ([], {}), '()\n', (2861, 2863), False, 'import time\n'), ((6574, 6612), 'numpy.random.randn', 'np.random.randn', (['n_walkers', 'num_params'], {}), '(n_walkers, num_params)\n', (6589, 6612), True, 'import numpy as np\n'), ((8841, 8853), 'numpy.mean', 'np.mean', (['tau'], {}), '(tau)\n', (8848, 8853), True, 'import numpy as np\n'), ((8909, 8946), 'numpy.all', 'np.all', (['(tau * 100 < sampler.iteration)'], {}), '(tau * 100 < sampler.iteration)\n', (8915, 8946), True, 'import numpy as np\n'), ((6489, 6527), 'numpy.random.randn', 'np.random.randn', (['n_walkers', 'num_params'], {}), '(n_walkers, num_params)\n', (6504, 6527), True, 'import numpy as np\n'), ((8983, 9004), 'numpy.abs', 'np.abs', (['(old_tau - tau)'], {}), '(old_tau - tau)\n', (8989, 9004), True, 'import numpy as np\n')] |
#!/usr/bin/python
"""
"""
# ----
import json
import tempfile
import itertools
import subprocess
import matplotlib.pyplot as plt
import scipy.stats as stats
from matplotlib.font_manager import FontProperties
# ----
import numpy as np
import pandas as pd
import statsmodels.api as sm
## ----------------------------------------------------------------
## Commit time related plots
## ----------------------------------------------------------------
class CommitData(object):
"Prepare commit data into ready to plot form"
def __init__(self, logs):
# Clean data
dfs = {k : d.commit for k,d in logs.items() if not d.commit.empty}
# Get summary stats
pts = pd.concat([d[['H','at']] for _,d in dfs.items()])
t0 = np.min(pts['at']).tz_localize(None)
ts = (pts['at'].dt.tz_localize(None) - t0).astype('timedelta64') / 1e9
hs = pts['H']
# Calculate relative time
for k, df in dfs.items() :
df['dt'] = (df['at'].dt.tz_localize(None) - t0).astype('timedelta64') / 1e9
self.fit = logs.fitTvsH
self.TPS = logs.TPS
# Store precalculated data
self.tMin = np.min(ts)
self.tMax = np.max(ts)
self.hMin = np.min(hs)
self.hMax = np.max(hs)
self.dfs = dfs
def plot_points(self, ax, reltime=False):
"Simply plot points"
for k,df in self.dfs.items() :
xs = df['dt' if reltime else 'at']
ys = df['H']
ax.plot(xs, ys, '+', label=k)
self.add_title_h("Height vs time")
p = self.fit.params
hs = np.asarray([ np.min(df['H']), np.max(df['H'])])
ax.plot( hs * p[1] + p[0], hs, '-', color='gray', lw=0.5)
plt.xlabel("time")
plt.ylabel("H")
def plot_residuals_HvsT(self, ax):
for k,df in self.dfs.items() :
p = self.fit.params
xs = df['dt']
ys = df['H'] - (xs / p[1] - p[0]/p[1])
ax.plot(xs, ys, '+', label=k)
self.add_title_h("Height residuals")
plt.xlabel("time")
plt.ylabel("ฮH")
def plot_residuals_TvsH(self, ax):
for k,df in self.dfs.items() :
p = self.fit.params
xs = df['H']
ys = df['dt'] - (xs * p[1] + p[0])
ax.plot(xs, ys, '+', label=k)
self.add_title_h("Time residuals")
plt.xlabel("H")
plt.ylabel("ฮt")
def plot_ntx(self,ax):
for k in self.dfs:
df = self.dfs[k]
break
tot = np.sum(df['Ntx'])
avg = np.average(df['Ntx'])
plt.title("Block size (ฮผ=%.2f, tot=%i)" % (avg,tot))
plt.xlabel("Height")
plt.ylabel("N of transactions")
plt.axhline(y=0, color='k')
plt.axhline(y=avg, color='k')
plt.plot(df['H'], df['Ntx'],'+')
def plot_ntx_distr(self,ax):
for k in self.dfs:
df = self.dfs[k]
break
ntx = df['Ntx']
mu = np.average(ntx)
sig = np.std(ntx)
n1 = np.min(ntx)
n2 = np.max(ntx)
#
plt.title("Block size (ฮผ=%.2f)" % (mu))
plt.xlabel("N of transactions")
plt.ylabel("prob. density")
#
plt.hist(ntx, n2-n1+1, density=True)
x = np.linspace(n1, n2, 200)
plt.plot(x, stats.norm.pdf(x, mu, sig), color='r')
plt.axvline(mu, color='r')
def plot_n_signatures(self,ax):
for k in self.dfs:
df = self.dfs[k]
break
df = df[df['H']>1]
avg = np.average(df['nsign'])
plt.title("N signatures for block (avg = %.2f)" % avg)
plt.plot(df['H'] - 1, df['nsign'],'+')
def add_title_h(self,s ):
plt.title("%s (%.03f s/block, %.f tps)" %
(s,
float(self.fit.params[1]),
self.TPS
))
# ----------------------------------------------------------------
# Plotting routines
# ----------------------------------------------------------------
class SimplePlot(object):
"Simple plotter"
def __init__(self):
self.fig = plt.figure()
self.ax = plt.subplot(111)
plt.grid()
def __enter__(self):
return self.ax
def __exit__(self, type, value, trace):
pass
class LegendPlot(object):
"Plotter with legend"
def __init__(self):
fig,ax = figure_with_legend()
self.fig = fig
self.ax = ax
plt.grid()
def __enter__(self):
return self.ax
def __exit__(self, type, value, trace):
add_legend(self.ax)
def figure_with_legend():
fig = plt.figure(figsize=[9, 4.8])
ax = plt.subplot(111)
box = ax.get_position()
ax.set_position([box.x0, box.y0, box.width * 0.8, box.height])
return fig,ax
def add_legend(ax) :
fontP = FontProperties()
fontP.set_size('small')
ax.legend(loc='center left', bbox_to_anchor=(1, 0.5), prop=fontP)
# ----------------------------------------------------------------
# Plotting routines
# ----------------------------------------------------------------
def plot_round(logs):
"""
Plot round growth
"""
dfs = {k : v.round for k,v in logs.items() if not v.round.empty}
#
fig,ax = figure_with_legend()
plt.title("Round vs T")
plt.grid()
plt.xlabel("Time")
plt.ylabel("Round")
for k,v in dfs.items():
plt.plot(v['at'], v['R'], lw=0.5, marker='x', markersize=2, label=k)
add_legend(ax)
return fig
def plot_round_distr(logs):
"Plot distribution of round numbers per figure"
dfs = {k : v.roundDistr for k,v in logs.items() if not v.round.empty}
n = len(dfs)
fig,ax = figure_with_legend()
for i,(nm,rs) in enumerate(dfs.items()):
xs = np.asarray(range(len(rs)))
plt.bar(xs + i / n, rs, width=1/n, align='edge', label=nm)
print("%12s: <R> = %.2f" % (nm, np.average(xs, weights=rs)))
add_legend(ax)
return fig
def plot_mempool_size(dfs):
"Plot mempool size over time"
fig,ax = figure_with_legend()
plt.grid()
plt.title("Mempool size")
colors = plt.rcParams['axes.prop_cycle'].by_key()['color']
def pickColor(i):
return colors[i % len(colors)]
for i,k in enumerate(dfs) :
df = dfs[k].mempool
dB = df[df['msg'] == "Mempool before filtering"]
dA = df[df['msg'] == "Mempool after filtering"]
plt.plot(dB['at'], dB['size'], '+', color=pickColor(i), ms=3, label=k+' before')
plt.plot(dA['at'], dA['size'], 'x', color=pickColor(i), ms=3, label=k+' after')
add_legend(ax)
return fig
def plot_mempool_added(dfs):
"Plot N of tx added to mempool over time"
fig = plt.figure()
plt.grid()
plt.title("Number of transaction added to mempool")
colors = plt.rcParams['axes.prop_cycle'].by_key()['color']
for i,df in enumerate(dfs) :
df = dfs[df].mempool
d = df[df['msg'] == "Mempool after filtering"]
plt.plot(d['at'], d['added'], '+', color=colors[i], ms=3)
plt.plot(d['at'], d['discarded'], 'v', color=colors[i], ms=3)
plt.plot(d['at'], d['filtered'], 'x', color=colors[i], ms=3)
return fig
def plot_gossip(logs, key):
"""
Plot statistics about gossip"
"""
fig,ax = figure_with_legend()
plt.grid()
plt.title("Gossip statistics for: "+key)
for i,(k,d) in enumerate(logs.items()):
tx = d[key]
tx = tx - tx.values[0]
ax.plot(d['at'], tx, '+', label=k, markersize=1.5)
add_legend(ax)
return fig
def plot_gossip_rxtx_ratio(dfs, key):
"Plot statistics about gossip"
dfs = [d.gossip() for d in dfs]
fig = plt.figure()
plt.grid()
plt.title("Rx/Tx ratio for: "+key)
for d in dfs:
plt.plot(d['at'], d['Rx' + key]/d['Tx'+key], '+')
return fig
| [
"matplotlib.pyplot.title",
"numpy.sum",
"matplotlib.pyplot.bar",
"matplotlib.pyplot.figure",
"matplotlib.pyplot.axvline",
"matplotlib.font_manager.FontProperties",
"numpy.std",
"numpy.max",
"numpy.linspace",
"matplotlib.pyplot.axhline",
"numpy.average",
"numpy.min",
"matplotlib.pyplot.ylabel... | [((4652, 4680), 'matplotlib.pyplot.figure', 'plt.figure', ([], {'figsize': '[9, 4.8]'}), '(figsize=[9, 4.8])\n', (4662, 4680), True, 'import matplotlib.pyplot as plt\n'), ((4691, 4707), 'matplotlib.pyplot.subplot', 'plt.subplot', (['(111)'], {}), '(111)\n', (4702, 4707), True, 'import matplotlib.pyplot as plt\n'), ((4855, 4871), 'matplotlib.font_manager.FontProperties', 'FontProperties', ([], {}), '()\n', (4869, 4871), False, 'from matplotlib.font_manager import FontProperties\n'), ((5300, 5323), 'matplotlib.pyplot.title', 'plt.title', (['"""Round vs T"""'], {}), "('Round vs T')\n", (5309, 5323), True, 'import matplotlib.pyplot as plt\n'), ((5328, 5338), 'matplotlib.pyplot.grid', 'plt.grid', ([], {}), '()\n', (5336, 5338), True, 'import matplotlib.pyplot as plt\n'), ((5343, 5361), 'matplotlib.pyplot.xlabel', 'plt.xlabel', (['"""Time"""'], {}), "('Time')\n", (5353, 5361), True, 'import matplotlib.pyplot as plt\n'), ((5366, 5385), 'matplotlib.pyplot.ylabel', 'plt.ylabel', (['"""Round"""'], {}), "('Round')\n", (5376, 5385), True, 'import matplotlib.pyplot as plt\n'), ((6089, 6099), 'matplotlib.pyplot.grid', 'plt.grid', ([], {}), '()\n', (6097, 6099), True, 'import matplotlib.pyplot as plt\n'), ((6104, 6129), 'matplotlib.pyplot.title', 'plt.title', (['"""Mempool size"""'], {}), "('Mempool size')\n", (6113, 6129), True, 'import matplotlib.pyplot as plt\n'), ((6725, 6737), 'matplotlib.pyplot.figure', 'plt.figure', ([], {}), '()\n', (6735, 6737), True, 'import matplotlib.pyplot as plt\n'), ((6742, 6752), 'matplotlib.pyplot.grid', 'plt.grid', ([], {}), '()\n', (6750, 6752), True, 'import matplotlib.pyplot as plt\n'), ((6757, 6808), 'matplotlib.pyplot.title', 'plt.title', (['"""Number of transaction added to mempool"""'], {}), "('Number of transaction added to mempool')\n", (6766, 6808), True, 'import matplotlib.pyplot as plt\n'), ((7327, 7337), 'matplotlib.pyplot.grid', 'plt.grid', ([], {}), '()\n', (7335, 7337), True, 'import matplotlib.pyplot as plt\n'), ((7342, 7384), 'matplotlib.pyplot.title', 'plt.title', (["('Gossip statistics for: ' + key)"], {}), "('Gossip statistics for: ' + key)\n", (7351, 7384), True, 'import matplotlib.pyplot as plt\n'), ((7697, 7709), 'matplotlib.pyplot.figure', 'plt.figure', ([], {}), '()\n', (7707, 7709), True, 'import matplotlib.pyplot as plt\n'), ((7714, 7724), 'matplotlib.pyplot.grid', 'plt.grid', ([], {}), '()\n', (7722, 7724), True, 'import matplotlib.pyplot as plt\n'), ((7729, 7765), 'matplotlib.pyplot.title', 'plt.title', (["('Rx/Tx ratio for: ' + key)"], {}), "('Rx/Tx ratio for: ' + key)\n", (7738, 7765), True, 'import matplotlib.pyplot as plt\n'), ((1178, 1188), 'numpy.min', 'np.min', (['ts'], {}), '(ts)\n', (1184, 1188), True, 'import numpy as np\n'), ((1209, 1219), 'numpy.max', 'np.max', (['ts'], {}), '(ts)\n', (1215, 1219), True, 'import numpy as np\n'), ((1240, 1250), 'numpy.min', 'np.min', (['hs'], {}), '(hs)\n', (1246, 1250), True, 'import numpy as np\n'), ((1271, 1281), 'numpy.max', 'np.max', (['hs'], {}), '(hs)\n', (1277, 1281), True, 'import numpy as np\n'), ((1742, 1760), 'matplotlib.pyplot.xlabel', 'plt.xlabel', (['"""time"""'], {}), "('time')\n", (1752, 1760), True, 'import matplotlib.pyplot as plt\n'), ((1769, 1784), 'matplotlib.pyplot.ylabel', 'plt.ylabel', (['"""H"""'], {}), "('H')\n", (1779, 1784), True, 'import matplotlib.pyplot as plt\n'), ((2069, 2087), 'matplotlib.pyplot.xlabel', 'plt.xlabel', (['"""time"""'], {}), "('time')\n", (2079, 2087), True, 'import matplotlib.pyplot as plt\n'), ((2096, 2112), 'matplotlib.pyplot.ylabel', 'plt.ylabel', (['"""ฮH"""'], {}), "('ฮH')\n", (2106, 2112), True, 'import matplotlib.pyplot as plt\n'), ((2390, 2405), 'matplotlib.pyplot.xlabel', 'plt.xlabel', (['"""H"""'], {}), "('H')\n", (2400, 2405), True, 'import matplotlib.pyplot as plt\n'), ((2414, 2430), 'matplotlib.pyplot.ylabel', 'plt.ylabel', (['"""ฮt"""'], {}), "('ฮt')\n", (2424, 2430), True, 'import matplotlib.pyplot as plt\n'), ((2547, 2564), 'numpy.sum', 'np.sum', (["df['Ntx']"], {}), "(df['Ntx'])\n", (2553, 2564), True, 'import numpy as np\n'), ((2579, 2600), 'numpy.average', 'np.average', (["df['Ntx']"], {}), "(df['Ntx'])\n", (2589, 2600), True, 'import numpy as np\n'), ((2609, 2662), 'matplotlib.pyplot.title', 'plt.title', (["('Block size (ฮผ=%.2f, tot=%i)' % (avg, tot))"], {}), "('Block size (ฮผ=%.2f, tot=%i)' % (avg, tot))\n", (2618, 2662), True, 'import matplotlib.pyplot as plt\n'), ((2670, 2690), 'matplotlib.pyplot.xlabel', 'plt.xlabel', (['"""Height"""'], {}), "('Height')\n", (2680, 2690), True, 'import matplotlib.pyplot as plt\n'), ((2699, 2730), 'matplotlib.pyplot.ylabel', 'plt.ylabel', (['"""N of transactions"""'], {}), "('N of transactions')\n", (2709, 2730), True, 'import matplotlib.pyplot as plt\n'), ((2739, 2766), 'matplotlib.pyplot.axhline', 'plt.axhline', ([], {'y': '(0)', 'color': '"""k"""'}), "(y=0, color='k')\n", (2750, 2766), True, 'import matplotlib.pyplot as plt\n'), ((2777, 2806), 'matplotlib.pyplot.axhline', 'plt.axhline', ([], {'y': 'avg', 'color': '"""k"""'}), "(y=avg, color='k')\n", (2788, 2806), True, 'import matplotlib.pyplot as plt\n'), ((2815, 2848), 'matplotlib.pyplot.plot', 'plt.plot', (["df['H']", "df['Ntx']", '"""+"""'], {}), "(df['H'], df['Ntx'], '+')\n", (2823, 2848), True, 'import matplotlib.pyplot as plt\n'), ((2994, 3009), 'numpy.average', 'np.average', (['ntx'], {}), '(ntx)\n', (3004, 3009), True, 'import numpy as np\n'), ((3024, 3035), 'numpy.std', 'np.std', (['ntx'], {}), '(ntx)\n', (3030, 3035), True, 'import numpy as np\n'), ((3050, 3061), 'numpy.min', 'np.min', (['ntx'], {}), '(ntx)\n', (3056, 3061), True, 'import numpy as np\n'), ((3076, 3087), 'numpy.max', 'np.max', (['ntx'], {}), '(ntx)\n', (3082, 3087), True, 'import numpy as np\n'), ((3106, 3143), 'matplotlib.pyplot.title', 'plt.title', (["('Block size (ฮผ=%.2f)' % mu)"], {}), "('Block size (ฮผ=%.2f)' % mu)\n", (3115, 3143), True, 'import matplotlib.pyplot as plt\n'), ((3154, 3185), 'matplotlib.pyplot.xlabel', 'plt.xlabel', (['"""N of transactions"""'], {}), "('N of transactions')\n", (3164, 3185), True, 'import matplotlib.pyplot as plt\n'), ((3194, 3221), 'matplotlib.pyplot.ylabel', 'plt.ylabel', (['"""prob. density"""'], {}), "('prob. density')\n", (3204, 3221), True, 'import matplotlib.pyplot as plt\n'), ((3240, 3280), 'matplotlib.pyplot.hist', 'plt.hist', (['ntx', '(n2 - n1 + 1)'], {'density': '(True)'}), '(ntx, n2 - n1 + 1, density=True)\n', (3248, 3280), True, 'import matplotlib.pyplot as plt\n'), ((3291, 3315), 'numpy.linspace', 'np.linspace', (['n1', 'n2', '(200)'], {}), '(n1, n2, 200)\n', (3302, 3315), True, 'import numpy as np\n'), ((3383, 3409), 'matplotlib.pyplot.axvline', 'plt.axvline', (['mu'], {'color': '"""r"""'}), "(mu, color='r')\n", (3394, 3409), True, 'import matplotlib.pyplot as plt\n'), ((3562, 3585), 'numpy.average', 'np.average', (["df['nsign']"], {}), "(df['nsign'])\n", (3572, 3585), True, 'import numpy as np\n'), ((3594, 3648), 'matplotlib.pyplot.title', 'plt.title', (["('N signatures for block (avg = %.2f)' % avg)"], {}), "('N signatures for block (avg = %.2f)' % avg)\n", (3603, 3648), True, 'import matplotlib.pyplot as plt\n'), ((3657, 3696), 'matplotlib.pyplot.plot', 'plt.plot', (["(df['H'] - 1)", "df['nsign']", '"""+"""'], {}), "(df['H'] - 1, df['nsign'], '+')\n", (3665, 3696), True, 'import matplotlib.pyplot as plt\n'), ((4140, 4152), 'matplotlib.pyplot.figure', 'plt.figure', ([], {}), '()\n', (4150, 4152), True, 'import matplotlib.pyplot as plt\n'), ((4172, 4188), 'matplotlib.pyplot.subplot', 'plt.subplot', (['(111)'], {}), '(111)\n', (4183, 4188), True, 'import matplotlib.pyplot as plt\n'), ((4197, 4207), 'matplotlib.pyplot.grid', 'plt.grid', ([], {}), '()\n', (4205, 4207), True, 'import matplotlib.pyplot as plt\n'), ((4482, 4492), 'matplotlib.pyplot.grid', 'plt.grid', ([], {}), '()\n', (4490, 4492), True, 'import matplotlib.pyplot as plt\n'), ((5422, 5490), 'matplotlib.pyplot.plot', 'plt.plot', (["v['at']", "v['R']"], {'lw': '(0.5)', 'marker': '"""x"""', 'markersize': '(2)', 'label': 'k'}), "(v['at'], v['R'], lw=0.5, marker='x', markersize=2, label=k)\n", (5430, 5490), True, 'import matplotlib.pyplot as plt\n'), ((5826, 5886), 'matplotlib.pyplot.bar', 'plt.bar', (['(xs + i / n)', 'rs'], {'width': '(1 / n)', 'align': '"""edge"""', 'label': 'nm'}), "(xs + i / n, rs, width=1 / n, align='edge', label=nm)\n", (5833, 5886), True, 'import matplotlib.pyplot as plt\n'), ((6998, 7055), 'matplotlib.pyplot.plot', 'plt.plot', (["d['at']", "d['added']", '"""+"""'], {'color': 'colors[i]', 'ms': '(3)'}), "(d['at'], d['added'], '+', color=colors[i], ms=3)\n", (7006, 7055), True, 'import matplotlib.pyplot as plt\n'), ((7064, 7125), 'matplotlib.pyplot.plot', 'plt.plot', (["d['at']", "d['discarded']", '"""v"""'], {'color': 'colors[i]', 'ms': '(3)'}), "(d['at'], d['discarded'], 'v', color=colors[i], ms=3)\n", (7072, 7125), True, 'import matplotlib.pyplot as plt\n'), ((7134, 7194), 'matplotlib.pyplot.plot', 'plt.plot', (["d['at']", "d['filtered']", '"""x"""'], {'color': 'colors[i]', 'ms': '(3)'}), "(d['at'], d['filtered'], 'x', color=colors[i], ms=3)\n", (7142, 7194), True, 'import matplotlib.pyplot as plt\n'), ((7790, 7843), 'matplotlib.pyplot.plot', 'plt.plot', (["d['at']", "(d['Rx' + key] / d['Tx' + key])", '"""+"""'], {}), "(d['at'], d['Rx' + key] / d['Tx' + key], '+')\n", (7798, 7843), True, 'import matplotlib.pyplot as plt\n'), ((3336, 3362), 'scipy.stats.norm.pdf', 'stats.norm.pdf', (['x', 'mu', 'sig'], {}), '(x, mu, sig)\n', (3350, 3362), True, 'import scipy.stats as stats\n'), ((767, 784), 'numpy.min', 'np.min', (["pts['at']"], {}), "(pts['at'])\n", (773, 784), True, 'import numpy as np\n'), ((1633, 1648), 'numpy.min', 'np.min', (["df['H']"], {}), "(df['H'])\n", (1639, 1648), True, 'import numpy as np\n'), ((1650, 1665), 'numpy.max', 'np.max', (["df['H']"], {}), "(df['H'])\n", (1656, 1665), True, 'import numpy as np\n'), ((5925, 5951), 'numpy.average', 'np.average', (['xs'], {'weights': 'rs'}), '(xs, weights=rs)\n', (5935, 5951), True, 'import numpy as np\n')] |
"""
Regression Using Decision Tree, Random Tree, Bootstrap Aggregating,
and Boosting.
Copyright (c) 2020 <NAME>
"""
import numpy as np
class RTLearner:
def __init__(self, leaf=1, tol=1.0e-6):
"""
leaf Lowest number of leaves
tol Tolerance to group close-valued leaves
"""
self.leaf = leaf
self.tol = tol
def buildTree(self, X, Y):
"""
Builds the decision-tree table.
column 0 = feature used for the split (-1 indicates a leaf)
column 1 = split value
column 2 = relative position left branch (0 indicates no left branch)
column 3 = relative position right branch (0 indicates no right branch)
"""
# Return the mean if equal or less than the lowest number of leaves
if (X.shape[0] <= self.leaf):
return np.array([-1, Y.mean(), 0, 0])
# Return the mean if all remaining leaves have close values
Ym = Y.mean()
deltaY = np.absolute(Y-Ym)
if (all(deltaY <= self.tol)):
return np.array([-1, Ym, 0, 0])
# Keep splitting
else:
# Randomly pick a feature and split
idx = np.random.randint(0, X.shape[1])
i = np.random.randint(0, X.shape[0], size=2)
split_value = (X[i[0], idx] + X[i[1], idx]) / 2.0
# Build the left branch dataset
X_left = X[X[:, idx] <= split_value]
Y_left = Y[X[:, idx] <= split_value]
# Return the mean if there is no split (because all data end up in
# the left branch).
if (X_left.shape[0] == X.shape[0]):
return np.array([-1, Y_left.mean(), 0, 0])
# Keep splitting
else:
# Build the right branch dataset
X_right = X[X[:, idx] > split_value]
Y_right = Y[X[:, idx] > split_value]
# Search the two new branches
left_branch = self.buildTree(X_left, Y_left)
right_branch = self.buildTree(X_right, Y_right)
# Return the sub-tree table
k = divmod(len(left_branch), 4)[0]
root = np.array([idx, split_value, 1, k+1])
return np.concatenate((root, left_branch, right_branch))
def createModel(self, X, Y):
"""
Wrapper for building the decision-tree table.
"""
# Build the tree-table as 1-dim array
a = self.buildTree(X, Y)
# Reshape it as an (n_row, 4) matrix
n_row = divmod(len(a), 4)[0]
self.treeTable = a.reshape(n_row, 4)
def evalData(self, X):
"""
Evaluates a dataset of features with the created decision-tree table.
column 0 = feature used for the split (-1 indicates a leaf)
column 1 = split value
column 2 = relative position left branch (0 indicates no left branch)
column 3 = relative position right branch (0 indicates no right branch)
"""
n = X.shape[0] # Number of data to evaluate
pred_Y = np.empty(n) # Allocate prediction array
# Loop over the dataset
for i in range(n):
# Start from the root node of the decision-tree table
row = 0
feature = int(round(self.treeTable[0, 0]))
# Move along the decision-tree table until a leaf is found
while (feature != -1):
# Next node is on the left branch
if (X[i, feature] <= self.treeTable[row, 1]):
delta = int(round(self.treeTable[row, 2]))
# Next node is on the right branch
else:
delta = int(round(self.treeTable[row, 3]))
# Get the feature of the next node
row += delta
feature = int(round(self.treeTable[row, 0]))
# Set the leaf value as predicted value
pred_Y[i] = self.treeTable[row, 1]
return pred_Y
| [
"numpy.absolute",
"numpy.empty",
"numpy.random.randint",
"numpy.array",
"numpy.concatenate"
] | [((1001, 1020), 'numpy.absolute', 'np.absolute', (['(Y - Ym)'], {}), '(Y - Ym)\n', (1012, 1020), True, 'import numpy as np\n'), ((3122, 3133), 'numpy.empty', 'np.empty', (['n'], {}), '(n)\n', (3130, 3133), True, 'import numpy as np\n'), ((1076, 1100), 'numpy.array', 'np.array', (['[-1, Ym, 0, 0]'], {}), '([-1, Ym, 0, 0])\n', (1084, 1100), True, 'import numpy as np\n'), ((1207, 1239), 'numpy.random.randint', 'np.random.randint', (['(0)', 'X.shape[1]'], {}), '(0, X.shape[1])\n', (1224, 1239), True, 'import numpy as np\n'), ((1256, 1296), 'numpy.random.randint', 'np.random.randint', (['(0)', 'X.shape[0]'], {'size': '(2)'}), '(0, X.shape[0], size=2)\n', (1273, 1296), True, 'import numpy as np\n'), ((2215, 2253), 'numpy.array', 'np.array', (['[idx, split_value, 1, k + 1]'], {}), '([idx, split_value, 1, k + 1])\n', (2223, 2253), True, 'import numpy as np\n'), ((2275, 2324), 'numpy.concatenate', 'np.concatenate', (['(root, left_branch, right_branch)'], {}), '((root, left_branch, right_branch))\n', (2289, 2324), True, 'import numpy as np\n')] |
# -*- coding: utf-8 -*-
""" Module summary description.
More detailed description.
"""
import numpy as np
import networkx as nx
from math import sqrt as msqrt
from numba import njit
from shapely.errors import TopologicalError
from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, \
Point, MultiLineString, JOIN_STYLE
from shapely.ops import cascaded_union, linemerge, unary_union, transform
from gistools.coordinates import r_tree_idx
from gistools.graph import part_graph
from gistools.utils.check.type import is_iterable, type_assert
def add_points_to_line(line, threshold):
""" Add point coordinates to line geometry
:param line:
:param threshold:
:return:
"""
return linemerge(cut_(line, threshold))
def aggregate_partitions(polygons, weights, nparts, division,
weight_attr, split, recursive, **metis_options):
""" Aggregate polygons into partitions
:param polygons: polygons to aggregate
:param weights: polygons' corresponding weight
:param nparts: number of partitions
:param division: list of final relative weights of each partition
:param weight_attr:
:param split:
:param recursive:
:param metis_options:
:return:
"""
if "contig" not in metis_options.keys():
metis_options["contig"] = False
graph = polygon_collection_to_graph(polygons, weights, split,
metis_options["contig"], weight_attr)
tpweights = [(d,) for d in division]
partition = part_graph(graph, nparts, weight_attr, tpweights, recursive, **metis_options)
# Return unions of polygons belonging to each part (no multi-polygons)
return explode([no_artifact_unary_union([polygons[n] for n in part]) for part in partition])
def area_partition_polygon(polygon, unit_area, disaggregation_factor, precision,
recursive, split, **metis_options):
""" Partition polygon into a subset of polygons of equal area
:param polygon: polygon intended to be partitioned
:param unit_area: area of a sub-polygon
:param disaggregation_factor: factor use to discretize polygons before aggregation
:param recursive: k-way or recursive method for partitioning
:param precision: metric precision for sub-polygon area
:param split: function used to split polygon into smaller unit blocks
:param metis_options: specific METIS options (see METIS manual)
:return:
"""
nparts = int(polygon.area/unit_area)
if nparts <= 1 and (polygon.area - unit_area) < unit_area/disaggregation_factor:
return [polygon]
# Split polygon into sub-elements
split_poly = split_polygon(polygon, split, unit_area/disaggregation_factor, get_explode=True)
division = [unit_area/polygon.area] * nparts
if polygon.area % unit_area != 0: # and (polygon.area - nparts * unit_area) >= unit_area/disaggregation_factor:
division += [(polygon.area - nparts * unit_area)/polygon.area]
nparts += 1
area = [int(poly.area / precision) for poly in split_poly]
return aggregate_partitions(split_poly, area, nparts, division, "area",
split, recursive, **metis_options)
def centroid(point_collection):
""" Retrieve centroid of multiple points
:param point_collection:
:return:
"""
x_centroid = np.mean([pt.x for pt in point_collection])
y_centroid = np.mean([pt.y for pt in point_collection])
return Point([x_centroid, y_centroid])
def connect_lines_to_point(line_collection, point):
""" Connect a set of lines to some point
:param line_collection:
:param point:
:return:
"""
new_line_collection = []
for line in line_collection:
if Point(line.coords[0]).distance(point) < Point(line.coords[-1]).distance(point):
new_line_collection.append(LineString(point.coords[:] + line.coords[:]))
else:
new_line_collection.append(LineString(line.coords[:] + point.coords[:]))
return new_line_collection
def cut(line, threshold, count=0):
""" Cut a line in segments
Cut a line in segments whose length
is below a threshold value. This method
is more randomless regarding the final
size of the line segments. See 'cut_'
function for more accuracy
:param line:
:param threshold:
:param count:
:return:
"""
result = []
if threshold < 0 or threshold >= line.length or count == 250:
return [line]
# Recursively cut line in 2 at midpoint
p = line.interpolate(0.5, normalized=True)
split_line = cut_at_point(line, p)
for sub_line in split_line:
result.extend(cut(sub_line, threshold, count + 1))
return result
def cut_(line, threshold):
""" Cut a line in segments (method 2)
This method cuts a line in as many segments as necessary,
depending on the given threshold. For instance, a line
of 105m will be cut into 10 pieces of 10m + 1 piece of 5m
if threshold=10
:param line: LineString
:param threshold: minimum sub line piece size
:return:
"""
if threshold < 0 or threshold >= line.length:
return [line]
result = []
while "It remains line to cut":
split_line = cut_at_distance(line, threshold/line.length, normalized=True)
result.append(split_line[0])
if split_line[1].length > threshold:
line = split_line[1]
else:
result.append(split_line[1])
break
return result
def cut_at_distance(line, distance, normalized=False):
""" Cut line at given distance from starting point
:param line:
:param distance:
:param normalized:
:return:
"""
if normalized:
length = 1
else:
length = line.length
if distance <= 0.0 or distance >= length:
return [line]
coords = list(line.coords)
for i, p in enumerate(coords):
pd = line.project(Point(p), normalized=normalized)
if pd == distance:
return [LineString(coords[:i+1]), LineString(coords[i:])]
elif pd > distance:
cp = line.interpolate(distance, normalized=normalized)
try:
return [LineString(coords[:i] + [(cp.x, cp.y)]), LineString([(cp.x, cp.y)] + coords[i:])]
except ValueError:
return [LineString(coords[:i] + [(cp.x, cp.y, cp.z)]), LineString([(cp.x, cp.y, cp.z)] + coords[i:])]
def cut_at_point(line, point):
""" Cut line at point
Cut line at point, which can be within
or without the geometry
:param line:
:param point:
:return:
"""
d = line.project(point)
return cut_at_distance(line, d)
def cut_at_points(line, points):
""" Cut line at multiple points
:param line:
:param points:
:return:
"""
cut_line = []
distance = [line.project(point) for point in points]
sorted_points = [point for _, point in sorted(zip(distance, points))]
for idx, point in enumerate(sorted_points):
cut_line.extend(cut_at_point(line, point))
if idx < len(sorted_points) - 1:
line = cut_line.pop()
return cut_line
def dissolve(geometry_collection):
""" Recursively join contiguous geometries in collection
:param geometry_collection:
:return:
"""
if not is_iterable(geometry_collection):
raise TypeError("Input must be a collection but is '{}'".format(type(geometry_collection)))
while "There is still geometries to aggregate":
joint = []
idx = r_tree_idx(geometry_collection)
geom_idx = []
increment = 0
while len(geom_idx) < len(geometry_collection):
if increment not in geom_idx:
geom = geometry_collection[increment]
union_idx, union = intersecting_features(geom, geometry_collection, idx)
if len(union) > 0:
joint.append(cascaded_union(union))
for ix in union_idx:
idx.delete(ix, geometry_collection[ix].bounds)
geom_idx.extend(union_idx)
increment += 1
if len(joint) < len(geometry_collection):
geometry_collection = joint
else:
break
return joint
def explode(geometry_collection):
""" Convert multi-part geometry collection into single-part
:param geometry_collection: valid geometry collection
:return:
"""
single = []
if not is_iterable(geometry_collection):
geometry_collection = [geometry_collection]
for geom in geometry_collection:
try:
single.extend(geom)
except TypeError:
single.append(geom)
return single
def fishnet(polygon, threshold):
""" Intersect polygon with a regular grid or "fishnet"
:param polygon:
:param threshold:
:return:
"""
return polygon_to_mesh(polygon, threshold, mesh)
def hexana(polygon, threshold):
""" Split a polygon using a honeycomb grid
:param polygon: original polygon to split
:param threshold: unit hexagon surface
:return: list of polygons
"""
return polygon_to_mesh(polygon, threshold, honeycomb)
# Thanks to https://gist.github.com/urschrei/17cf0be92ca90a244a91
@njit()
def honeycomb_nb(startx, starty, endx, endy, radius):
"""
Calculate a grid of hexagon coordinates of the given radius
given lower-left and upper-right coordinates
Returns a list of lists containing 6 tuples of x, y point coordinates
These can be used to construct valid regular hexagonal polygons
- update 04/23/2019:
* can give either radius or area of unit hexagon
* return a list of shapely Polygon
You will probably want to use projected coordinates for this
"""
# calculate side length given radius
sl = (2 * radius) * np.tan(np.pi / 6)
# calculate radius for a given side-length
# (a * (math.cos(math.pi / 6) / math.sin(math.pi / 6)) / 2)
# see http://www.calculatorsoup.com/calculators/geometry-plane/polygon.php
# calculate coordinates of the hexagon points
# sin(30)
p = sl * 0.5
b = sl * np.cos(np.radians(30))
w = b * 2
h = 2 * sl
# offset start and end coordinates by hex widths and heights to guarantee coverage
startx = startx - w
starty = starty - h
endx = endx + w
endy = endy + h
origx = startx
# offsets for moving along and up rows
xoffset = b
yoffset = 3 * p
row = 1
while starty < endy:
if row % 2 == 0:
startx = origx + xoffset
else:
startx = origx
while startx < endx:
p1x = startx
p1y = starty + p
p2x = startx
p2y = starty + (3 * p)
p3x = startx + b
p3y = starty + h
p4x = startx + w
p4y = starty + (3 * p)
p5x = startx + w
p5y = starty + p
p6x = startx + b
p6y = starty
poly = [
(p1x, p1y),
(p2x, p2y),
(p3x, p3y),
(p4x, p4y),
(p5x, p5y),
(p6x, p6y),
(p1x, p1y)]
yield poly
startx += w
starty += yoffset
row += 1
def honeycomb(startx, starty, endx, endy, radius=None, area=None):
"""
Parameters
----------
startx
starty
endx
endy
radius
area
Returns
-------
"""
if not radius:
radius = msqrt(area / (2*msqrt(3)))
return (Polygon(poly) for poly in honeycomb_nb(startx, starty, endx, endy, radius))
def intersecting_features(geometry, geometry_collection, r_tree=None):
""" Return list of geometries intersecting with given geometry
:param geometry:
:param geometry_collection:
:param r_tree: rtree index corresponding to geometry collection
:return:
"""
is_intersecting = intersects(geometry, geometry_collection, r_tree)
return [i for i in range(len(geometry_collection)) if is_intersecting[i]], \
[geom for i, geom in enumerate(geometry_collection) if is_intersecting[i]]
def intersects(geometry, geometry_collection, r_tree=None):
""" Return if geometry intersects with geometries of collection
Use this function with large geometry collections
:param geometry:
:param geometry_collection:
:param r_tree:
:return: list of boolean of length = length(geometry_collection)
"""
# Use Rtree to speed up !
if r_tree is None:
r_tree = r_tree_idx(geometry_collection)
list_of_intersecting_features = list(r_tree.intersection(geometry.bounds))
return [False if f not in list_of_intersecting_features else geometry.intersects(geometry_collection[f]) for f in
range(len(geometry_collection))]
def is_in_collection(geometry, geometry_collection, r_tree):
""" Test if geometry is present in collection (using shapely 'equals' method)
:param geometry:
:param geometry_collection:
:param r_tree:
:return:
"""
_, list_of_intersecting_features = intersecting_features(geometry, geometry_collection, r_tree)
for geom in list_of_intersecting_features:
if geometry.equals(geom):
return True
return False
def is_line_connected_to(line, geometry_collection):
""" Is line connected to one of the geometries in collection ?
:param line:
:param geometry_collection:
:return:
"""
return [other.intersects(Point(line.coords[0])) for other in geometry_collection], [other.intersects(Point(
line.coords[-1])) for other in geometry_collection]
def katana(polygon, threshold, count=0):
""" Split a polygon
See https://snorfalorpagus.net/blog/2016/03/13/splitting-large-polygons-for-faster-intersections/
Copyright (c) 2016, <NAME>
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the
following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following
disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the
following disclaimer in the documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED
WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE
GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
:param polygon: Shapely polygon
:param threshold:
:param count:
:return:
"""
if count == 0:
if not polygon.is_valid:
polygon = polygon.buffer(0, 0)
result = []
width = polygon.bounds[2] - polygon.bounds[0]
height = polygon.bounds[3] - polygon.bounds[1]
if width * height <= threshold or count == 250:
return [polygon]
if height >= width:
a = box(polygon.bounds[0], polygon.bounds[1], polygon.bounds[2], polygon.bounds[1] + height/2)
b = box(polygon.bounds[0], polygon.bounds[1] + height/2, polygon.bounds[2], polygon.bounds[3])
else:
a = box(polygon.bounds[0], polygon.bounds[1], polygon.bounds[0] + width/2, polygon.bounds[3])
b = box(polygon.bounds[0] + width/2, polygon.bounds[1], polygon.bounds[2], polygon.bounds[3])
for sword in (a, b,):
split_poly = polygon.intersection(sword)
if not isinstance(split_poly, GeometryCollection):
split_poly = [split_poly]
for sub_poly in split_poly:
if isinstance(sub_poly, (Polygon, MultiPolygon)):
result.extend(katana(sub_poly, threshold, count+1))
return result
def katana_centroid(polygon, threshold, count=0):
""" Split a polygon in equal areas
Thanks to https://snorfalorpagus.net/blog/2016/03/13/splitting-large-polygons-for-faster-intersections/ and
<NAME> in http://community-gispython-org-community-projects.955323.n3.nabble.com/Community-Spliting-a
-polygon- into-two-polygons-with-the-same-area-td4024026.html#a4024033, we merge here both approaches to split a
polygon into a number of sub-polygons of almost equal areas.
:param polygon: Shapely polygon
:param threshold:
:param count:
:return:
"""
if count == 0:
if not polygon.is_valid:
polygon = polygon.buffer(0, 0)
result = []
width = polygon.bounds[2] - polygon.bounds[0]
height = polygon.bounds[3] - polygon.bounds[1]
if width * height <= threshold or count == 250:
return [polygon]
if height >= width:
a = box(polygon.bounds[0], polygon.bounds[1], polygon.bounds[2], polygon.centroid.y)
b = box(polygon.bounds[0], polygon.centroid.y, polygon.bounds[2], polygon.bounds[3])
else:
a = box(polygon.bounds[0], polygon.bounds[1], polygon.centroid.x, polygon.bounds[3])
b = box(polygon.centroid.x, polygon.bounds[1], polygon.bounds[2], polygon.bounds[3])
for sword in (a, b,):
split_poly = polygon.intersection(sword)
if not isinstance(split_poly, GeometryCollection):
split_poly = [split_poly]
for sub_poly in split_poly:
if isinstance(sub_poly, (Polygon, MultiPolygon)):
result.extend(katana_centroid(sub_poly, threshold, count+1))
return result
def length_of_segments(line):
""" Retrieve segment length in line
:param line:
:return:
"""
return np.diff([line.project(Point(p)) for p in line.coords])
def mask(polygon_collection, mask_collection, fast_intersection_surface):
""" Geometry mask
:param polygon_collection:
:param mask_collection:
:param fast_intersection_surface:
:return:
"""
# Retrieve base layer and mask geometry, split it for faster intersection
# and explode it (to be sure there is no multi-parts)
geometry = split_polygon_collection(polygon_collection, fast_intersection_surface, get_explode=True)
mask_geometry = split_polygon_collection(mask_collection, fast_intersection_surface, get_explode=True)
# Use Rtree to speed up !
idx = r_tree_idx(mask_geometry)
# 0. Initialization
result = []
for geom in geometry:
list_of_intersecting_mask = list(idx.intersection(geom.bounds))
within = [geom.within(mask_geometry[n]) for n in list_of_intersecting_mask]
if not any(within):
is_intersecting = [geom.intersects(mask_geometry[n]) for n in list_of_intersecting_mask]
if any(is_intersecting):
difference = geom.difference(cascaded_union([mask_geometry[n] for n in list_of_intersecting_mask]))
if not difference.is_empty:
result.append(difference)
else:
result.append(geom)
# Multi to single + dissolve coincident polygons
result = explode(result)
result = [no_artifact_unary_union(poly) for poly in dissolve(result)]
return result
def merge(line_collection):
""" Merge connected lines
:param line_collection:
:return:
"""
# Merge MultiLinestring objects returned by the "join" function
merged_line = [linemerge(line) if isinstance(line, MultiLineString) else line for line in dissolve(line_collection)]
# Keep only single parts
return explode(merged_line)
def mesh(startx, starty, endx, endy, side=None, area=None):
""" Compute a mesh grid
:param startx:
:param starty:
:param endx:
:param endy:
:param side:
:param area:
:return:
"""
if not side:
side = msqrt(area)
startx = startx - side/2
starty = starty - side/2
endx = endx + side/2
endy = endy + side/2
origx = startx
polygons = []
while starty < endy:
startx = origx
while startx < endx:
poly = [
(startx, starty),
(startx, starty + side),
(startx + side, starty + side),
(startx + side, starty)]
polygons.append(Polygon(poly))
startx += side
starty += side
return polygons
def nearest_feature(geometry, geometry_collection, r_tree=None):
""" Return nearest feature from geometry collection to given geometry
If some of the geometries intersect, the nearest feature is the one whose centroid is the closest to the centroid
of the given geometry (but distance remains 0)
:param geometry:
:param geometry_collection:
:param r_tree: rtree index corresponding to geometry collection
:return: nearest feature index and corresponding distance
"""
# Use Rtree to speed up !
if r_tree is None:
r_tree = r_tree_idx(geometry_collection)
# Look if some geometries intersect
list_of_intersecting_features, _ = intersecting_features(geometry, geometry_collection, r_tree)
if list_of_intersecting_features:
distance = [geometry.centroid.distance(geometry_collection[n].centroid) for n in list_of_intersecting_features]
return list_of_intersecting_features[np.argmin(distance)], 0
else:
list_of_nearest_features = list(r_tree.nearest(geometry.bounds, 1))
distance = [geometry.distance(geometry_collection[n]) for n in list_of_nearest_features]
return list_of_nearest_features[np.argmin(distance)], np.min(distance)
def no_artifact_unary_union(geoms, eps=0.00001):
""" Make unary union that does not return artifacts
Thanks to https://gis.stackexchange.com/questions/277334/shapely-polygon-union-results-in-strange-artifacts-of
-tiny-non-overlapping-area
:param geoms: list of geoms to aggregate
:param eps: buffering precision
:return:
"""
return unary_union(geoms).buffer(eps, 1, join_style=JOIN_STYLE.mitre).buffer(-eps, 1, join_style=JOIN_STYLE.mitre)
def overlapping_features(geometry, geometry_collection, r_tree=None):
""" Return list of geometries overlapping with given geometry
Overlapping geometry is either overlapping in the shapely way,
or within or containing the other geometry
:param geometry:
:param geometry_collection:
:param r_tree:
:return:
"""
idx, list_of_intersecting_features = intersecting_features(geometry, geometry_collection, r_tree)
_overlaps = [[i, geom] for i, geom in zip(idx, list_of_intersecting_features) if geom.overlaps(geometry) or
geom.within(geometry) or geom.contains(geometry)]
return [overlap[0] for overlap in _overlaps], [overlap[1] for overlap in _overlaps]
def overlaps(geometry, geometry_collection, r_tree=None):
""" Return if geometry overlaps with geometries of collection
Overlapping is regarded as any area shared by two geometries
:param geometry:
:param geometry_collection:
:param r_tree:
:return:
"""
is_intersecting = intersects(geometry, geometry_collection, r_tree)
return [False if not is_intersecting[i] else geom.overlaps(geometry) or geom.within(geometry) or geom.contains(
geometry) for i, geom in enumerate(geometry_collection)]
def polygon_to_mesh(polygon, threshold, method):
"""
:param polygon:
:param threshold:
:param method: {'hexana', 'fishnet'}
:return:
"""
grid = method(*polygon.bounds, area=threshold)
split = []
for unit in grid:
if unit.within(polygon):
split.append(unit)
elif unit.overlaps(polygon):
split.append(unit.intersection(polygon))
return explode(split)
def polygon_collection_to_graph(polygon_collection, weights, split, is_contiguous, weight_attr="weight"):
""" Convert collection of polygons to networkx graph
Conversion of a polygon collection into a graph allows
later graph partitioning
:param polygon_collection:
:param weights: weight of each polygon in collection
:param split: split function
:param is_contiguous: True or False (metis options)
:param weight_attr: name of weight attribute
:return:
"""
if not is_iterable(polygon_collection):
raise TypeError("Input must be a collection but is '{}'".format(type(polygon_collection)))
if 'katana' in split.__name__:
is_katana = True
else:
is_katana = False
r_tree = r_tree_idx(polygon_collection)
graph = nx.Graph()
for n, polygon in enumerate(polygon_collection):
list_of_intersecting_features, _ = intersecting_features(polygon, polygon_collection, r_tree)
list_of_intersecting_features.remove(n)
if list_of_intersecting_features or not is_contiguous:
if is_katana:
graph.add_edges_from([(n, feature) for feature in list_of_intersecting_features
if not isinstance(polygon.intersection(polygon_collection[feature]), Point)])
else:
graph.add_edges_from([(n, feature) for feature in list_of_intersecting_features])
graph.add_node(n, **{weight_attr: weights[n]})
return graph
def radius_of_curvature(line, method="osculating"):
""" Compute curvature radius of LineString
:param line:
:param method: method for computing radius of curvature {'circumscribe', 'osculating'}
:return:
"""
def norm(xx, yy):
return np.sqrt(xx ** 2 + yy ** 2)
def tangent_vector(xi, yi):
return (xi[2::] - xi[:-2]) / norm(xi[2::] - xi[:-2], yi[2::] - yi[:-2]), \
(yi[2::] - yi[:-2]) / norm(xi[2::] - xi[:-2], yi[2::] - yi[:-2])
if method == "osculating":
if len(line.coords) >= 3:
x = np.array(line.coords.xy[0])
y = np.array(line.coords.xy[1])
xi1 = np.concatenate((x[1::], [x[-1]]))
yi1 = np.concatenate((y[1::], [y[-1]]))
xi_1 = np.concatenate(([x[0]], x[:-1]))
yi_1 = np.concatenate(([y[0]], y[:-1]))
tangent_vector_xi1, tangent_vector_yi1 = tangent_vector(xi1, yi1)
tangent_vector_xi_1, tangent_vector_yi_1 = tangent_vector(xi_1, yi_1)
coefficient_of_curvature = \
norm(tangent_vector_xi1 - tangent_vector_xi_1, tangent_vector_yi1 - tangent_vector_yi_1) /\
norm(x[2::] - x[:-2], y[2::] - y[:-2])
coefficient_of_curvature[coefficient_of_curvature == 0] = 1e-6
rad_of_curvature = 1 / coefficient_of_curvature
else:
return np.array([10000])
elif method == "circumscribe":
segment_length = length_of_segments(line)
a, b = segment_length[:-1:], segment_length[1::]
c = []
if len(line.coords) > 3:
length_2_by_2_start = length_of_segments(LineString(line.coords[::2]))
length_2_by_2_end = length_of_segments(LineString(line.coords[1::2]))
for n in range(len(length_2_by_2_end)):
c.extend([length_2_by_2_start[n], length_2_by_2_end[n]])
if len(length_2_by_2_start) > len(length_2_by_2_end):
c.append(length_2_by_2_start[-1])
elif len(line.coords) == 3:
c = LineString(line.coords[::2]).length
elif len(line.coords) < 3:
return np.array([10000])
heron = (a + b + c) * (b + c - a) * (c + a - b) * (a + b - c)
heron[heron < 0] = 0
divider = np.sqrt(heron)
divider[divider == 0] = 0.1
rad_of_curvature = a * b * c / divider
else:
rad_of_curvature = []
# Return values and add replicate to beginning of array (as result of curvature computation returns an array with
# length = length(line.coords) - 2): return array with length = length(line.coords) - 1
return np.concatenate(([rad_of_curvature[0]], rad_of_curvature))
def shape_factor(polygon, convex_hull):
""" Compute shape factor of given polygon
Compute shape factor (here, circularity) of
a given polygon using either convex hull or not
:param polygon:
:param convex_hull: should convex hull be used for computing shape ? (bool)
:return:
"""
if convex_hull:
return 4 * np.pi * polygon.convex_hull.area / (polygon.convex_hull.length ** 2)
else:
return 4 * np.pi * polygon.area / (polygon.length ** 2)
@type_assert(polygon1=(Polygon, MultiPolygon), polygon2=(Polygon, MultiPolygon), normalized=bool)
def shared_area(polygon1, polygon2, normalized=False):
""" Get area shared by 2 polygons
:param polygon1:
:param polygon2:
:param normalized:
:return:
"""
if not polygon1.intersects(polygon2):
return 0
else:
new_poly = polygon1.intersection(polygon2)
if normalized:
return new_poly.area / polygon1.area
else:
return new_poly.area
@type_assert(polygon=(Polygon, MultiPolygon), normalized=bool)
def shared_area_among_collection(polygon: Polygon, polygon_collection, normalized: bool = False, r_tree=None):
""" Get area shared by a polygon with polygons from a collection
:param polygon:
:param polygon_collection:
:param normalized:
:param r_tree:
:return:
"""
if not is_iterable(polygon_collection):
raise TypeError("Input 2 must be a collection but is '{}'".format(type(polygon_collection)))
poly_intersects = intersects(polygon, polygon_collection, r_tree)
return [shared_area(polygon, poly, normalized) if poly_intersects[n] else 0 for n, poly in enumerate(
polygon_collection)]
def split_collection(geometry_collection, threshold, method, get_explode):
""" Split geometry collection
:param geometry_collection:
:param threshold:
:param method:
:param get_explode:
:return:
"""
if not is_iterable(geometry_collection):
raise TypeError("Geometry must be a collection")
new_collection = []
for geom in geometry_collection:
try:
new_collection.extend(method(geom, threshold))
except TopologicalError:
new_collection.append(geom)
if get_explode:
new_collection = explode(new_collection)
# Return new collection
return new_collection
def split_line_collection(line_collection, threshold, method="cut", get_explode=False):
"""
:param line_collection:
:param threshold:
:param method:
:param get_explode:
:return:
"""
split_method = {'cut': cut}
return split_collection(line_collection, threshold, split_method[method], get_explode)
def split_polygon(polygon, method, threshold, get_explode):
""" Split polygon with respect to method
Split polygon and return exploded (no multi part) if necessary
:param polygon:
:param method:
:param threshold:
:param get_explode: (boolean) return exploded collection
:return:
"""
sp_poly = method(polygon, threshold)
if get_explode:
return explode(sp_poly)
else:
return sp_poly
def split_polygon_collection(polygon_collection, threshold, method="katana", get_explode=False):
""" Split a collection of polygons
:param polygon_collection: collection of shapely polygons
:param threshold: threshold surface under which no more splitting must be achieved
:param method: method used for splitting
:param get_explode:
:return: new polygon collection with only Polygon geometries (no MultiPolygon geometries)
"""
split_method = {'katana': katana, 'katana_centroid': katana_centroid}
return split_collection(polygon_collection, threshold, split_method[method], get_explode)
def to_2d(geometry):
""" Convert 3D geometry to 2D
Credit to @feenster and @hunt3ri from
https://github.com/hotosm/tasking-manager/blob/master/server/services/grid/grid_service.py
:param geometry:
:return:
"""
def _to_2d(x, y, z):
return tuple(filter(None, [x, y]))
return transform(_to_2d, geometry)
| [
"shapely.ops.unary_union",
"numba.njit",
"shapely.ops.transform",
"numpy.argmin",
"gistools.coordinates.r_tree_idx",
"numpy.mean",
"shapely.geometry.box",
"shapely.geometry.Point",
"shapely.geometry.Polygon",
"shapely.geometry.LineString",
"numpy.tan",
"numpy.radians",
"shapely.ops.cascaded_... | [((9332, 9338), 'numba.njit', 'njit', ([], {}), '()\n', (9336, 9338), False, 'from numba import njit\n'), ((29009, 29109), 'gistools.utils.check.type.type_assert', 'type_assert', ([], {'polygon1': '(Polygon, MultiPolygon)', 'polygon2': '(Polygon, MultiPolygon)', 'normalized': 'bool'}), '(polygon1=(Polygon, MultiPolygon), polygon2=(Polygon,\n MultiPolygon), normalized=bool)\n', (29020, 29109), False, 'from gistools.utils.check.type import is_iterable, type_assert\n'), ((29528, 29589), 'gistools.utils.check.type.type_assert', 'type_assert', ([], {'polygon': '(Polygon, MultiPolygon)', 'normalized': 'bool'}), '(polygon=(Polygon, MultiPolygon), normalized=bool)\n', (29539, 29589), False, 'from gistools.utils.check.type import is_iterable, type_assert\n'), ((1559, 1636), 'gistools.graph.part_graph', 'part_graph', (['graph', 'nparts', 'weight_attr', 'tpweights', 'recursive'], {}), '(graph, nparts, weight_attr, tpweights, recursive, **metis_options)\n', (1569, 1636), False, 'from gistools.graph import part_graph\n'), ((3399, 3441), 'numpy.mean', 'np.mean', (['[pt.x for pt in point_collection]'], {}), '([pt.x for pt in point_collection])\n', (3406, 3441), True, 'import numpy as np\n'), ((3459, 3501), 'numpy.mean', 'np.mean', (['[pt.y for pt in point_collection]'], {}), '([pt.y for pt in point_collection])\n', (3466, 3501), True, 'import numpy as np\n'), ((3514, 3545), 'shapely.geometry.Point', 'Point', (['[x_centroid, y_centroid]'], {}), '([x_centroid, y_centroid])\n', (3519, 3545), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((18905, 18930), 'gistools.coordinates.r_tree_idx', 'r_tree_idx', (['mask_geometry'], {}), '(mask_geometry)\n', (18915, 18930), False, 'from gistools.coordinates import r_tree_idx\n'), ((25055, 25085), 'gistools.coordinates.r_tree_idx', 'r_tree_idx', (['polygon_collection'], {}), '(polygon_collection)\n', (25065, 25085), False, 'from gistools.coordinates import r_tree_idx\n'), ((25098, 25108), 'networkx.Graph', 'nx.Graph', ([], {}), '()\n', (25106, 25108), True, 'import networkx as nx\n'), ((28455, 28512), 'numpy.concatenate', 'np.concatenate', (['([rad_of_curvature[0]], rad_of_curvature)'], {}), '(([rad_of_curvature[0]], rad_of_curvature))\n', (28469, 28512), True, 'import numpy as np\n'), ((32632, 32659), 'shapely.ops.transform', 'transform', (['_to_2d', 'geometry'], {}), '(_to_2d, geometry)\n', (32641, 32659), False, 'from shapely.ops import cascaded_union, linemerge, unary_union, transform\n'), ((7382, 7414), 'gistools.utils.check.type.is_iterable', 'is_iterable', (['geometry_collection'], {}), '(geometry_collection)\n', (7393, 7414), False, 'from gistools.utils.check.type import is_iterable, type_assert\n'), ((7603, 7634), 'gistools.coordinates.r_tree_idx', 'r_tree_idx', (['geometry_collection'], {}), '(geometry_collection)\n', (7613, 7634), False, 'from gistools.coordinates import r_tree_idx\n'), ((8539, 8571), 'gistools.utils.check.type.is_iterable', 'is_iterable', (['geometry_collection'], {}), '(geometry_collection)\n', (8550, 8571), False, 'from gistools.utils.check.type import is_iterable, type_assert\n'), ((9922, 9939), 'numpy.tan', 'np.tan', (['(np.pi / 6)'], {}), '(np.pi / 6)\n', (9928, 9939), True, 'import numpy as np\n'), ((11657, 11670), 'shapely.geometry.Polygon', 'Polygon', (['poly'], {}), '(poly)\n', (11664, 11670), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((12659, 12690), 'gistools.coordinates.r_tree_idx', 'r_tree_idx', (['geometry_collection'], {}), '(geometry_collection)\n', (12669, 12690), False, 'from gistools.coordinates import r_tree_idx\n'), ((15703, 15800), 'shapely.geometry.box', 'box', (['polygon.bounds[0]', 'polygon.bounds[1]', 'polygon.bounds[2]', '(polygon.bounds[1] + height / 2)'], {}), '(polygon.bounds[0], polygon.bounds[1], polygon.bounds[2], polygon.bounds\n [1] + height / 2)\n', (15706, 15800), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((15806, 15902), 'shapely.geometry.box', 'box', (['polygon.bounds[0]', '(polygon.bounds[1] + height / 2)', 'polygon.bounds[2]', 'polygon.bounds[3]'], {}), '(polygon.bounds[0], polygon.bounds[1] + height / 2, polygon.bounds[2],\n polygon.bounds[3])\n', (15809, 15902), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((15919, 16014), 'shapely.geometry.box', 'box', (['polygon.bounds[0]', 'polygon.bounds[1]', '(polygon.bounds[0] + width / 2)', 'polygon.bounds[3]'], {}), '(polygon.bounds[0], polygon.bounds[1], polygon.bounds[0] + width / 2,\n polygon.bounds[3])\n', (15922, 16014), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((16021, 16116), 'shapely.geometry.box', 'box', (['(polygon.bounds[0] + width / 2)', 'polygon.bounds[1]', 'polygon.bounds[2]', 'polygon.bounds[3]'], {}), '(polygon.bounds[0] + width / 2, polygon.bounds[1], polygon.bounds[2],\n polygon.bounds[3])\n', (16024, 16116), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((17384, 17469), 'shapely.geometry.box', 'box', (['polygon.bounds[0]', 'polygon.bounds[1]', 'polygon.bounds[2]', 'polygon.centroid.y'], {}), '(polygon.bounds[0], polygon.bounds[1], polygon.bounds[2], polygon.centroid.y\n )\n', (17387, 17469), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((17477, 17562), 'shapely.geometry.box', 'box', (['polygon.bounds[0]', 'polygon.centroid.y', 'polygon.bounds[2]', 'polygon.bounds[3]'], {}), '(polygon.bounds[0], polygon.centroid.y, polygon.bounds[2], polygon.bounds[3]\n )\n', (17480, 17562), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((17580, 17665), 'shapely.geometry.box', 'box', (['polygon.bounds[0]', 'polygon.bounds[1]', 'polygon.centroid.x', 'polygon.bounds[3]'], {}), '(polygon.bounds[0], polygon.bounds[1], polygon.centroid.x, polygon.bounds[3]\n )\n', (17583, 17665), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((17673, 17758), 'shapely.geometry.box', 'box', (['polygon.centroid.x', 'polygon.bounds[1]', 'polygon.bounds[2]', 'polygon.bounds[3]'], {}), '(polygon.centroid.x, polygon.bounds[1], polygon.bounds[2], polygon.bounds[3]\n )\n', (17676, 17758), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((20368, 20379), 'math.sqrt', 'msqrt', (['area'], {}), '(area)\n', (20373, 20379), True, 'from math import sqrt as msqrt\n'), ((21475, 21506), 'gistools.coordinates.r_tree_idx', 'r_tree_idx', (['geometry_collection'], {}), '(geometry_collection)\n', (21485, 21506), False, 'from gistools.coordinates import r_tree_idx\n'), ((24812, 24843), 'gistools.utils.check.type.is_iterable', 'is_iterable', (['polygon_collection'], {}), '(polygon_collection)\n', (24823, 24843), False, 'from gistools.utils.check.type import is_iterable, type_assert\n'), ((26075, 26101), 'numpy.sqrt', 'np.sqrt', (['(xx ** 2 + yy ** 2)'], {}), '(xx ** 2 + yy ** 2)\n', (26082, 26101), True, 'import numpy as np\n'), ((29896, 29927), 'gistools.utils.check.type.is_iterable', 'is_iterable', (['polygon_collection'], {}), '(polygon_collection)\n', (29907, 29927), False, 'from gistools.utils.check.type import is_iterable, type_assert\n'), ((30482, 30514), 'gistools.utils.check.type.is_iterable', 'is_iterable', (['geometry_collection'], {}), '(geometry_collection)\n', (30493, 30514), False, 'from gistools.utils.check.type import is_iterable, type_assert\n'), ((5997, 6005), 'shapely.geometry.Point', 'Point', (['p'], {}), '(p)\n', (6002, 6005), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((10232, 10246), 'numpy.radians', 'np.radians', (['(30)'], {}), '(30)\n', (10242, 10246), True, 'import numpy as np\n'), ((19954, 19969), 'shapely.ops.linemerge', 'linemerge', (['line'], {}), '(line)\n', (19963, 19969), False, 'from shapely.ops import cascaded_union, linemerge, unary_union, transform\n'), ((22121, 22137), 'numpy.min', 'np.min', (['distance'], {}), '(distance)\n', (22127, 22137), True, 'import numpy as np\n'), ((26381, 26408), 'numpy.array', 'np.array', (['line.coords.xy[0]'], {}), '(line.coords.xy[0])\n', (26389, 26408), True, 'import numpy as np\n'), ((26425, 26452), 'numpy.array', 'np.array', (['line.coords.xy[1]'], {}), '(line.coords.xy[1])\n', (26433, 26452), True, 'import numpy as np\n'), ((26471, 26503), 'numpy.concatenate', 'np.concatenate', (['(x[1:], [x[-1]])'], {}), '((x[1:], [x[-1]]))\n', (26485, 26503), True, 'import numpy as np\n'), ((26523, 26555), 'numpy.concatenate', 'np.concatenate', (['(y[1:], [y[-1]])'], {}), '((y[1:], [y[-1]]))\n', (26537, 26555), True, 'import numpy as np\n'), ((26576, 26608), 'numpy.concatenate', 'np.concatenate', (['([x[0]], x[:-1])'], {}), '(([x[0]], x[:-1]))\n', (26590, 26608), True, 'import numpy as np\n'), ((26628, 26660), 'numpy.concatenate', 'np.concatenate', (['([y[0]], y[:-1])'], {}), '(([y[0]], y[:-1]))\n', (26642, 26660), True, 'import numpy as np\n'), ((27195, 27212), 'numpy.array', 'np.array', (['[10000]'], {}), '([10000])\n', (27203, 27212), True, 'import numpy as np\n'), ((28093, 28107), 'numpy.sqrt', 'np.sqrt', (['heron'], {}), '(heron)\n', (28100, 28107), True, 'import numpy as np\n'), ((3905, 3949), 'shapely.geometry.LineString', 'LineString', (['(point.coords[:] + line.coords[:])'], {}), '(point.coords[:] + line.coords[:])\n', (3915, 3949), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((4004, 4048), 'shapely.geometry.LineString', 'LineString', (['(line.coords[:] + point.coords[:])'], {}), '(line.coords[:] + point.coords[:])\n', (4014, 4048), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((6077, 6103), 'shapely.geometry.LineString', 'LineString', (['coords[:i + 1]'], {}), '(coords[:i + 1])\n', (6087, 6103), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((6103, 6125), 'shapely.geometry.LineString', 'LineString', (['coords[i:]'], {}), '(coords[i:])\n', (6113, 6125), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((13620, 13641), 'shapely.geometry.Point', 'Point', (['line.coords[0]'], {}), '(line.coords[0])\n', (13625, 13641), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((13696, 13718), 'shapely.geometry.Point', 'Point', (['line.coords[-1]'], {}), '(line.coords[-1])\n', (13701, 13718), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((18265, 18273), 'shapely.geometry.Point', 'Point', (['p'], {}), '(p)\n', (18270, 18273), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((20817, 20830), 'shapely.geometry.Polygon', 'Polygon', (['poly'], {}), '(poly)\n', (20824, 20830), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((21852, 21871), 'numpy.argmin', 'np.argmin', (['distance'], {}), '(distance)\n', (21861, 21871), True, 'import numpy as np\n'), ((22099, 22118), 'numpy.argmin', 'np.argmin', (['distance'], {}), '(distance)\n', (22108, 22118), True, 'import numpy as np\n'), ((3786, 3807), 'shapely.geometry.Point', 'Point', (['line.coords[0]'], {}), '(line.coords[0])\n', (3791, 3807), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((3826, 3848), 'shapely.geometry.Point', 'Point', (['line.coords[-1]'], {}), '(line.coords[-1])\n', (3831, 3848), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((11633, 11641), 'math.sqrt', 'msqrt', (['(3)'], {}), '(3)\n', (11638, 11641), True, 'from math import sqrt as msqrt\n'), ((19366, 19435), 'shapely.ops.cascaded_union', 'cascaded_union', (['[mask_geometry[n] for n in list_of_intersecting_mask]'], {}), '([mask_geometry[n] for n in list_of_intersecting_mask])\n', (19380, 19435), False, 'from shapely.ops import cascaded_union, linemerge, unary_union, transform\n'), ((22505, 22523), 'shapely.ops.unary_union', 'unary_union', (['geoms'], {}), '(geoms)\n', (22516, 22523), False, 'from shapely.ops import cascaded_union, linemerge, unary_union, transform\n'), ((27458, 27486), 'shapely.geometry.LineString', 'LineString', (['line.coords[::2]'], {}), '(line.coords[::2])\n', (27468, 27486), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((27539, 27568), 'shapely.geometry.LineString', 'LineString', (['line.coords[1::2]'], {}), '(line.coords[1::2])\n', (27549, 27568), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((6263, 6302), 'shapely.geometry.LineString', 'LineString', (['(coords[:i] + [(cp.x, cp.y)])'], {}), '(coords[:i] + [(cp.x, cp.y)])\n', (6273, 6302), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((6304, 6343), 'shapely.geometry.LineString', 'LineString', (['([(cp.x, cp.y)] + coords[i:])'], {}), '([(cp.x, cp.y)] + coords[i:])\n', (6314, 6343), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((7991, 8012), 'shapely.ops.cascaded_union', 'cascaded_union', (['union'], {}), '(union)\n', (8005, 8012), False, 'from shapely.ops import cascaded_union, linemerge, unary_union, transform\n'), ((27866, 27894), 'shapely.geometry.LineString', 'LineString', (['line.coords[::2]'], {}), '(line.coords[::2])\n', (27876, 27894), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((27957, 27974), 'numpy.array', 'np.array', (['[10000]'], {}), '([10000])\n', (27965, 27974), True, 'import numpy as np\n'), ((6400, 6445), 'shapely.geometry.LineString', 'LineString', (['(coords[:i] + [(cp.x, cp.y, cp.z)])'], {}), '(coords[:i] + [(cp.x, cp.y, cp.z)])\n', (6410, 6445), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n'), ((6447, 6492), 'shapely.geometry.LineString', 'LineString', (['([(cp.x, cp.y, cp.z)] + coords[i:])'], {}), '([(cp.x, cp.y, cp.z)] + coords[i:])\n', (6457, 6492), False, 'from shapely.geometry import MultiPolygon, GeometryCollection, Polygon, box, LineString, Point, MultiLineString, JOIN_STYLE\n')] |
#! /usr/bin/env python
# -*- coding: utf-8 -*-
import os
import random
import logging
import datetime
import numpy as np
# padle
import paddle
logger = logging.getLogger(__name__)
def get_logger(log_file=None):# {{{
"""Set logger and return it.
If the log_file is not None, log will be written into log_file.
Else, log will be shown in the screen.
Args:
log_file (str): If log_file is not None, log will be written
into the log_file.
Return:
~Logger
* **logger**: An Logger object with customed config.
"""
# Basic config
logging.basicConfig(
format='%(asctime)s - %(levelname)s - %(message)s',
datefmt='%m/%d/%Y %H:%M:%S',
level=logging.INFO,
)
logger = logging.getLogger(__name__)
# Add filehandler
if log_file is not None:
file_handler = logging.FileHandler(log_file, mode='w')
file_handler.setLevel(logging.INFO)
file_handler.setFormatter(
logging.Formatter('%(asctime)s - %(levelname)s - %(message)s')
)
logger.addHandler(file_handler)
return logger# }}}
def set_seed_logger(args, cfg):# {{{
"""Experiments preparation, e.g., fix random seed, prepare checkpoint dir
and set logger.
Args:
args (parser.Argument): An parser.Argument object.
cfg (yacs.config): An yacs.config.CfgNode object.
Return:
~(Logger, str):
* **logger**: An Logger object with customed config.
* **save_dir**: Checkpoint dir to save models.
"""
seed = cfg['MISC']['SEED']
# Set random seed
paddle.seed(seed)
random.seed(seed)
np.random.seed(seed)
os.environ['PYTHONHASHSEED'] = str(seed)
# Prepare save dir
if cfg['OUTPUT']['SAVE_NAME']:
prefix = cfg['OUTPUT']['SAVE_NAME'] + '_'
else:
prefix = ''
exp_name = prefix + datetime.datetime.now().strftime('%yY_%mM_%dD_%HH')
save_dir = os.path.join(cfg['OUTPUT']['CHECKPOINT_DIR'], exp_name)
if not os.path.exists(save_dir):
os.makedirs(save_dir, exist_ok=True)
# Build logger
log_file = os.path.join(save_dir, 'log.txt')
logger = get_logger(log_file)
return logger, save_dir# }}}
def dump_cfg(cfg, cfg_file):# {{{
"""Dump config of each experiment into file for backup.
Args:
cfg (yacs.config): An yacs.config.CfgNode object.
cfg_file (str): Dump config to this file.
"""
logger.info('Dump configs into {}'.format(cfg_file))
logger.info('Using configs: ')
logger.info(cfg)
with open(cfg_file, 'w') as f:
f.write(cfg.dump())# }}}
def compute_ranks(dataset, results):# {{{
sims = np.array([results[i] for i in range(len(dataset))])
labels = np.array([dataset.get_label(i) for i in range(len(dataset))])
if dataset.has_caption_indices:
num_captions_per_image = dataset.num_captions_per_image
else:
num_captions_per_image = len(dataset.image_keys) * dataset.num_captions_per_image
sims = sims.reshape([-1, num_captions_per_image])
labels = labels.reshape([-1, num_captions_per_image])
# Compute i2t ranks
i2t_ranks = []
for sim, label in zip(sims, labels):
inds = np.argsort(sim)[::-1]
rank = num_captions_per_image
for r, ind in enumerate(inds):
if label[ind] == 1:
rank = r
break
i2t_ranks.append(rank)
# Compute t2i ranks
t2i_ranks = []
if not dataset.has_caption_indices:
sims = np.swapaxes(sims, 0, 1)
labels = np.swapaxes(labels, 0, 1)
for sim, label in zip(sims, labels):
inds = np.argsort(sim)[::-1]
rank = num_captions_per_image
for r, ind in enumerate(inds):
if label[ind] == 1:
rank = r
break
t2i_ranks.append(rank)
return i2t_ranks, t2i_ranks# }}}
def get_retrieval_results(dataset, results):
sims = np.array([results[i] for i in range(len(dataset))])
images = np.array([dataset.get_result(i)[0] for i in range(len(dataset))])
captions = np.array([dataset.get_result(i)[1] for i in range(len(dataset))])
if dataset.has_caption_indices:
num_captions_per_image = dataset.num_captions_per_image
else:
num_captions_per_image = len(dataset.image_keys) * dataset.num_captions_per_image
sims = sims.reshape([-1, num_captions_per_image]) # num_image x num_captions
images = images.reshape([-1, num_captions_per_image]) # num_images x num_captions
captions = captions.reshape([-1, num_captions_per_image]) # num_images x num_captions
# Get i2t results
i2t_results = {}
for i, (sim, cap) in enumerate(zip(sims, captions)):
inds = np.argsort(sim)[::-1]
tmp_results = []
for ind in inds:
tmp_results.append(cap[ind])
i2t_results[images[i][0]] = tmp_results[:10]
# Get t2i results
t2i_results = {}
if not dataset.has_caption_indices:
sims = np.swapaxes(sims, 0, 1) # num_captions x num_images
images = np.swapaxes(images, 0, 1) # num_captions x num_images
for t, (sim, image) in enumerate(zip(sims, images)):
inds = np.argsort(sim)[::-1]
tmp_results = []
for ind in inds:
tmp_results.append(image[ind])
t2i_results[captions[0][t]] = tmp_results
return i2t_results, t2i_results
| [
"numpy.random.seed",
"logging.FileHandler",
"logging.basicConfig",
"os.makedirs",
"os.path.exists",
"datetime.datetime.now",
"logging.Formatter",
"numpy.argsort",
"paddle.seed",
"random.seed",
"numpy.swapaxes",
"os.path.join",
"logging.getLogger"
] | [((155, 182), 'logging.getLogger', 'logging.getLogger', (['__name__'], {}), '(__name__)\n', (172, 182), False, 'import logging\n'), ((600, 724), 'logging.basicConfig', 'logging.basicConfig', ([], {'format': '"""%(asctime)s - %(levelname)s - %(message)s"""', 'datefmt': '"""%m/%d/%Y %H:%M:%S"""', 'level': 'logging.INFO'}), "(format='%(asctime)s - %(levelname)s - %(message)s',\n datefmt='%m/%d/%Y %H:%M:%S', level=logging.INFO)\n", (619, 724), False, 'import logging\n'), ((765, 792), 'logging.getLogger', 'logging.getLogger', (['__name__'], {}), '(__name__)\n', (782, 792), False, 'import logging\n'), ((1620, 1637), 'paddle.seed', 'paddle.seed', (['seed'], {}), '(seed)\n', (1631, 1637), False, 'import paddle\n'), ((1642, 1659), 'random.seed', 'random.seed', (['seed'], {}), '(seed)\n', (1653, 1659), False, 'import random\n'), ((1664, 1684), 'numpy.random.seed', 'np.random.seed', (['seed'], {}), '(seed)\n', (1678, 1684), True, 'import numpy as np\n'), ((1960, 2015), 'os.path.join', 'os.path.join', (["cfg['OUTPUT']['CHECKPOINT_DIR']", 'exp_name'], {}), "(cfg['OUTPUT']['CHECKPOINT_DIR'], exp_name)\n", (1972, 2015), False, 'import os\n'), ((2133, 2166), 'os.path.join', 'os.path.join', (['save_dir', '"""log.txt"""'], {}), "(save_dir, 'log.txt')\n", (2145, 2166), False, 'import os\n'), ((868, 907), 'logging.FileHandler', 'logging.FileHandler', (['log_file'], {'mode': '"""w"""'}), "(log_file, mode='w')\n", (887, 907), False, 'import logging\n'), ((2027, 2051), 'os.path.exists', 'os.path.exists', (['save_dir'], {}), '(save_dir)\n', (2041, 2051), False, 'import os\n'), ((2061, 2097), 'os.makedirs', 'os.makedirs', (['save_dir'], {'exist_ok': '(True)'}), '(save_dir, exist_ok=True)\n', (2072, 2097), False, 'import os\n'), ((3541, 3564), 'numpy.swapaxes', 'np.swapaxes', (['sims', '(0)', '(1)'], {}), '(sims, 0, 1)\n', (3552, 3564), True, 'import numpy as np\n'), ((3582, 3607), 'numpy.swapaxes', 'np.swapaxes', (['labels', '(0)', '(1)'], {}), '(labels, 0, 1)\n', (3593, 3607), True, 'import numpy as np\n'), ((5053, 5076), 'numpy.swapaxes', 'np.swapaxes', (['sims', '(0)', '(1)'], {}), '(sims, 0, 1)\n', (5064, 5076), True, 'import numpy as np\n'), ((5123, 5148), 'numpy.swapaxes', 'np.swapaxes', (['images', '(0)', '(1)'], {}), '(images, 0, 1)\n', (5134, 5148), True, 'import numpy as np\n'), ((999, 1061), 'logging.Formatter', 'logging.Formatter', (['"""%(asctime)s - %(levelname)s - %(message)s"""'], {}), "('%(asctime)s - %(levelname)s - %(message)s')\n", (1016, 1061), False, 'import logging\n'), ((3233, 3248), 'numpy.argsort', 'np.argsort', (['sim'], {}), '(sim)\n', (3243, 3248), True, 'import numpy as np\n'), ((4788, 4803), 'numpy.argsort', 'np.argsort', (['sim'], {}), '(sim)\n', (4798, 4803), True, 'import numpy as np\n'), ((1893, 1916), 'datetime.datetime.now', 'datetime.datetime.now', ([], {}), '()\n', (1914, 1916), False, 'import datetime\n'), ((3672, 3687), 'numpy.argsort', 'np.argsort', (['sim'], {}), '(sim)\n', (3682, 3687), True, 'import numpy as np\n'), ((5258, 5273), 'numpy.argsort', 'np.argsort', (['sim'], {}), '(sim)\n', (5268, 5273), True, 'import numpy as np\n')] |
import numpy as np
from cloudvolume import CloudVolume
from cloudvolume.lib import Bbox
from cloudvolume.storage import Storage
from chunkflow.chunk.validate import validate_by_template_matching
from tinybrain import downsample_with_averaging
from chunkflow.chunk import Chunk
from .base import OperatorBase
class CutoutOperator(OperatorBase):
def __init__(self,
volume_path: str,
mip: int = 0,
expand_margin_size=(0, 0, 0),
fill_missing: bool = False,
validate_mip: int = None,
blackout_sections: bool = None,
dry_run: bool = False,
name: str = 'cutout',
verbose: bool = True,
use_https: bool = True):
super().__init__(name=name, verbose=verbose)
self.volume_path = volume_path
self.mip = mip
self.expand_margin_size = expand_margin_size
self.fill_missing = fill_missing
self.validate_mip = validate_mip
self.blackout_sections = blackout_sections
self.dry_run = dry_run
self.use_https = use_https
if blackout_sections:
with Storage(volume_path) as stor:
self.blackout_section_ids = stor.get_json(
'blackout_section_ids.json')['section_ids']
def __call__(self, output_bbox):
#gevent.monkey.patch_all(thread=False)
vol = CloudVolume(self.volume_path,
bounded=False,
fill_missing=self.fill_missing,
progress=self.verbose,
mip=self.mip,
cache=False,
use_https=self.use_https,
green_threads=True)
chunk_slices = tuple(
slice(s.start - m, s.stop + m)
for s, m in zip(output_bbox.to_slices(), self.expand_margin_size))
if self.dry_run:
input_bbox = Bbox.from_slices(chunk_slices)
return Chunk.from_bbox(input_bbox)
if self.verbose:
print('cutout {} from {}'.format(chunk_slices[::-1],
self.volume_path))
# always reverse the indexes since cloudvolume use x,y,z indexing
chunk = vol[chunk_slices[::-1]]
# the cutout is fortran ordered, so need to transpose and make it C order
chunk = chunk.transpose()
# we can delay this transpose later
# actually we do not need to make it contiguous
# chunk = np.ascontiguousarray(chunk)
# if the channel number is 1, squeeze it as 3d array
# this should not be neccessary
# TODO: remove this step and use 4D array all over this package.
# always use 4D array will simplify some operations
global_offset = tuple(s.start for s in chunk_slices)
if chunk.shape[0] == 1:
chunk = np.squeeze(chunk, axis=0)
else:
global_offset = (chunk.shape[0], ) + global_offset
chunk = Chunk(chunk, global_offset=global_offset)
if self.blackout_sections:
chunk = self._blackout_sections(chunk)
if self.validate_mip:
self._validate_chunk(chunk, vol)
return chunk
def _blackout_sections(self, chunk):
"""
make some sections black.
this was normally used for the section with bad alignment.
The ConvNet was supposed to handle them better with black image.
TODO: make this function as a separate operator
"""
# current code only works with 3d image
assert chunk.ndim == 3, "current code assumes that the chunk is 3D image."
for z in self.blackout_section_ids:
z0 = z - chunk.global_offset[0]
if z0 >= 0 and z0 < chunk.shape[0]:
chunk[z0, :, :] = 0
return chunk
def _validate_chunk(self, chunk, vol):
"""
check that all the input voxels was downloaded without black region
We have found some black regions in previous inference run,
so hopefully this will solve the problem.
"""
if chunk.ndim == 4 and chunk.shape[0] > 1:
chunk = chunk[0, :, :, :]
validate_vol = CloudVolume(self.volume_path,
bounded=False,
fill_missing=self.fill_missing,
progress=self.verbose,
mip=self.validate_mip,
cache=False,
green_threads=True)
chunk_mip = self.mip
if self.verbose:
print('validate chunk in mip {}'.format(self.validate_mip))
assert self.validate_mip >= chunk_mip
# only use the region corresponds to higher mip level
# clamp the surrounding regions in XY plane
# this assumes that the input dataset was downsampled starting from the
# beginning offset in the info file
global_offset = chunk.global_offset
# factor3 follows xyz order in CloudVolume
factor3 = np.array([
2**(self.validate_mip - chunk_mip), 2
**(self.validate_mip - chunk_mip), 1
],
dtype=np.int32)
clamped_offset = tuple(go + f - (go - vo) % f for go, vo, f in zip(
global_offset[::-1], vol.voxel_offset, factor3))
clamped_stop = tuple(
go + s - (go + s - vo) % f
for go, s, vo, f in zip(global_offset[::-1], chunk.shape[::-1],
vol.voxel_offset, factor3))
clamped_slices = tuple(
slice(o, s) for o, s in zip(clamped_offset, clamped_stop))
clamped_bbox = Bbox.from_slices(clamped_slices)
clamped_input = chunk.cutout(clamped_slices[::-1])
# transform to xyz order
clamped_input = np.transpose(clamped_input)
# get the corresponding bounding box for validation
validate_bbox = vol.bbox_to_mip(clamped_bbox,
mip=chunk_mip,
to_mip=self.validate_mip)
#validate_bbox = clamped_bbox // factor3
# downsample the input using avaraging
# keep the z as it is since the mip only applies to xy plane
# recursivly downsample the input
# if we do it directly, the downsampled input will not be the same with the recursive one
# because of the rounding error of integer division
for _ in range(self.validate_mip - chunk_mip):
clamped_input = downsample_with_averaging(clamped_input, (2, 2, 1))
# validation by template matching
assert validate_by_template_matching(clamped_input)
validate_input = validate_vol[validate_bbox.to_slices()]
if validate_input.shape[3] == 1:
validate_input = np.squeeze(validate_input, axis=3)
# use the validate input to check the downloaded input
assert np.alltrue(validate_input == clamped_input)
| [
"chunkflow.chunk.validate.validate_by_template_matching",
"numpy.transpose",
"cloudvolume.lib.Bbox.from_slices",
"cloudvolume.CloudVolume",
"numpy.array",
"numpy.alltrue",
"numpy.squeeze",
"cloudvolume.storage.Storage",
"tinybrain.downsample_with_averaging",
"chunkflow.chunk.Chunk",
"chunkflow.c... | [((1446, 1627), 'cloudvolume.CloudVolume', 'CloudVolume', (['self.volume_path'], {'bounded': '(False)', 'fill_missing': 'self.fill_missing', 'progress': 'self.verbose', 'mip': 'self.mip', 'cache': '(False)', 'use_https': 'self.use_https', 'green_threads': '(True)'}), '(self.volume_path, bounded=False, fill_missing=self.fill_missing,\n progress=self.verbose, mip=self.mip, cache=False, use_https=self.\n use_https, green_threads=True)\n', (1457, 1627), False, 'from cloudvolume import CloudVolume\n'), ((3106, 3147), 'chunkflow.chunk.Chunk', 'Chunk', (['chunk'], {'global_offset': 'global_offset'}), '(chunk, global_offset=global_offset)\n', (3111, 3147), False, 'from chunkflow.chunk import Chunk\n'), ((4348, 4511), 'cloudvolume.CloudVolume', 'CloudVolume', (['self.volume_path'], {'bounded': '(False)', 'fill_missing': 'self.fill_missing', 'progress': 'self.verbose', 'mip': 'self.validate_mip', 'cache': '(False)', 'green_threads': '(True)'}), '(self.volume_path, bounded=False, fill_missing=self.fill_missing,\n progress=self.verbose, mip=self.validate_mip, cache=False,\n green_threads=True)\n', (4359, 4511), False, 'from cloudvolume import CloudVolume\n'), ((5240, 5349), 'numpy.array', 'np.array', (['[2 ** (self.validate_mip - chunk_mip), 2 ** (self.validate_mip - chunk_mip), 1]'], {'dtype': 'np.int32'}), '([2 ** (self.validate_mip - chunk_mip), 2 ** (self.validate_mip -\n chunk_mip), 1], dtype=np.int32)\n', (5248, 5349), True, 'import numpy as np\n'), ((5876, 5908), 'cloudvolume.lib.Bbox.from_slices', 'Bbox.from_slices', (['clamped_slices'], {}), '(clamped_slices)\n', (5892, 5908), False, 'from cloudvolume.lib import Bbox\n'), ((6025, 6052), 'numpy.transpose', 'np.transpose', (['clamped_input'], {}), '(clamped_input)\n', (6037, 6052), True, 'import numpy as np\n'), ((6857, 6901), 'chunkflow.chunk.validate.validate_by_template_matching', 'validate_by_template_matching', (['clamped_input'], {}), '(clamped_input)\n', (6886, 6901), False, 'from chunkflow.chunk.validate import validate_by_template_matching\n'), ((7152, 7195), 'numpy.alltrue', 'np.alltrue', (['(validate_input == clamped_input)'], {}), '(validate_input == clamped_input)\n', (7162, 7195), True, 'import numpy as np\n'), ((2020, 2050), 'cloudvolume.lib.Bbox.from_slices', 'Bbox.from_slices', (['chunk_slices'], {}), '(chunk_slices)\n', (2036, 2050), False, 'from cloudvolume.lib import Bbox\n'), ((2070, 2097), 'chunkflow.chunk.Chunk.from_bbox', 'Chunk.from_bbox', (['input_bbox'], {}), '(input_bbox)\n', (2085, 2097), False, 'from chunkflow.chunk import Chunk\n'), ((2978, 3003), 'numpy.squeeze', 'np.squeeze', (['chunk'], {'axis': '(0)'}), '(chunk, axis=0)\n', (2988, 3003), True, 'import numpy as np\n'), ((6747, 6798), 'tinybrain.downsample_with_averaging', 'downsample_with_averaging', (['clamped_input', '(2, 2, 1)'], {}), '(clamped_input, (2, 2, 1))\n', (6772, 6798), False, 'from tinybrain import downsample_with_averaging\n'), ((7038, 7072), 'numpy.squeeze', 'np.squeeze', (['validate_input'], {'axis': '(3)'}), '(validate_input, axis=3)\n', (7048, 7072), True, 'import numpy as np\n'), ((1194, 1214), 'cloudvolume.storage.Storage', 'Storage', (['volume_path'], {}), '(volume_path)\n', (1201, 1214), False, 'from cloudvolume.storage import Storage\n')] |
#!/usr/bin/python3
# -*- coding: utf-8 -*-
import logging
import matplotlib
import multiprocessing as mp
import numpy as np
import os
import sys
# Fix problem: no $DISPLAY environment variable
matplotlib.use('Agg')
from argparse import ArgumentParser
from datetime import datetime as dt
from pprint import pprint
from config import cfg
from core.train import train_net
from core.test import test_net
def get_args_from_command_line():
parser = ArgumentParser(description='Parser of Runner of Pix2Vox')
parser.add_argument('--gpu',
dest='gpu_id',
help='GPU device id to use [cuda0]',
default=cfg.CONST.DEVICE,
type=str)
parser.add_argument('--rand', dest='randomize', help='Randomize (do not use a fixed seed)', action='store_true')
parser.add_argument('--test', dest='test', help='Test neural networks', action='store_true')
parser.add_argument('--batch-size',
dest='batch_size',
help='name of the net',
default=cfg.CONST.BATCH_SIZE,
type=int)
parser.add_argument('--epoch', dest='epoch', help='number of epoches', default=cfg.TRAIN.NUM_EPOCHES, type=int)
parser.add_argument('--weights', dest='weights', help='Initialize network from the weights file', default=None)
parser.add_argument('--out', dest='out_path', help='Set output path', default=cfg.DIR.OUT_PATH)
args = parser.parse_args()
return args
def main():
# Get args from command line
args = get_args_from_command_line()
if args.gpu_id is not None:
cfg.CONST.DEVICE = args.gpu_id
if not args.randomize:
np.random.seed(cfg.CONST.RNG_SEED)
if args.batch_size is not None:
cfg.CONST.BATCH_SIZE = args.batch_size
if args.epoch is not None:
cfg.TRAIN.NUM_EPOCHES = args.epoch
if args.out_path is not None:
cfg.DIR.OUT_PATH = args.out_path
if args.weights is not None:
cfg.CONST.WEIGHTS = args.weights
if not args.test:
cfg.TRAIN.RESUME_TRAIN = True
# Print config
print('Use config:')
pprint(cfg)
# Set GPU to use
if type(cfg.CONST.DEVICE) == str:
os.environ["CUDA_VISIBLE_DEVICES"] = cfg.CONST.DEVICE
# Start train/test process
if not args.test:
train_net(cfg)
else:
# print('WEIGHTS' in cfg.CONST)
# print(os.path.exists(cfg.CONST.WEIGHTS))
if 'WEIGHTS' in cfg.CONST and os.path.exists(cfg.CONST.WEIGHTS):
test_net(cfg)
else:
print('[FATAL] %s Please specify the file path of checkpoint.' % (dt.now()))
sys.exit(2)
if __name__ == '__main__':
# Check python version
if sys.version_info < (3, 0):
raise Exception("Please Check python version")
# Setup logger
mp.log_to_stderr()
logger = mp.get_logger()
logger.setLevel(logging.INFO)
main()
| [
"numpy.random.seed",
"argparse.ArgumentParser",
"os.path.exists",
"core.test.test_net",
"multiprocessing.log_to_stderr",
"matplotlib.use",
"pprint.pprint",
"core.train.train_net",
"datetime.datetime.now",
"sys.exit",
"multiprocessing.get_logger"
] | [((194, 215), 'matplotlib.use', 'matplotlib.use', (['"""Agg"""'], {}), "('Agg')\n", (208, 215), False, 'import matplotlib\n'), ((452, 509), 'argparse.ArgumentParser', 'ArgumentParser', ([], {'description': '"""Parser of Runner of Pix2Vox"""'}), "(description='Parser of Runner of Pix2Vox')\n", (466, 509), False, 'from argparse import ArgumentParser\n'), ((2191, 2202), 'pprint.pprint', 'pprint', (['cfg'], {}), '(cfg)\n', (2197, 2202), False, 'from pprint import pprint\n'), ((2898, 2916), 'multiprocessing.log_to_stderr', 'mp.log_to_stderr', ([], {}), '()\n', (2914, 2916), True, 'import multiprocessing as mp\n'), ((2930, 2945), 'multiprocessing.get_logger', 'mp.get_logger', ([], {}), '()\n', (2943, 2945), True, 'import multiprocessing as mp\n'), ((1733, 1767), 'numpy.random.seed', 'np.random.seed', (['cfg.CONST.RNG_SEED'], {}), '(cfg.CONST.RNG_SEED)\n', (1747, 1767), True, 'import numpy as np\n'), ((2387, 2401), 'core.train.train_net', 'train_net', (['cfg'], {}), '(cfg)\n', (2396, 2401), False, 'from core.train import train_net\n'), ((2541, 2574), 'os.path.exists', 'os.path.exists', (['cfg.CONST.WEIGHTS'], {}), '(cfg.CONST.WEIGHTS)\n', (2555, 2574), False, 'import os\n'), ((2588, 2601), 'core.test.test_net', 'test_net', (['cfg'], {}), '(cfg)\n', (2596, 2601), False, 'from core.test import test_net\n'), ((2717, 2728), 'sys.exit', 'sys.exit', (['(2)'], {}), '(2)\n', (2725, 2728), False, 'import sys\n'), ((2694, 2702), 'datetime.datetime.now', 'dt.now', ([], {}), '()\n', (2700, 2702), True, 'from datetime import datetime as dt\n')] |
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import sys
import os
import os.path as osp
import time
import cv2
import numpy as np
import yaml
from six import text_type as _text_type
from paddlelite.lite import *
class Predictor:
def __init__(self, model_nb, model_yaml, thread_num, shape):
if not osp.exists(model_nb):
print("model nb file is not exists in {}".format(model_xml))
self.model_nb = model_nb
self.shape = shape
config = MobileConfig()
config.set_model_from_file(model_nb)
config.set_threads(thread_num)
if not osp.exists(model_yaml):
print("model yaml file is not exists in {}".format(model_yaml))
with open(model_yaml) as f:
self.info = yaml.load(f.read(), Loader=yaml.Loader)
self.model_type = self.info['_Attributes']['model_type']
self.model_name = self.info['Model']
self.num_classes = self.info['_Attributes']['num_classes']
self.labels = self.info['_Attributes']['labels']
if self.info['Model'] == 'MaskRCNN':
if self.info['_init_params']['with_fpn']:
self.mask_head_resolution = 28
else:
self.mask_head_resolution = 14
transforms_mode = self.info.get('TransformsMode', 'RGB')
if transforms_mode == 'RGB':
to_rgb = True
else:
to_rgb = False
self.transforms = self.build_transforms(self.info['Transforms'],
to_rgb)
self.predictor = create_paddle_predictor(config)
self.total_time = 0
self.count_num = 0
def build_transforms(self, transforms_info, to_rgb=True):
if self.model_type == "classifier":
import transforms.cls_transforms as transforms
elif self.model_type == "detector":
import transforms.det_transforms as transforms
elif self.model_type == "segmenter":
import transforms.seg_transforms as transforms
op_list = list()
for op_info in transforms_info:
op_name = list(op_info.keys())[0]
op_attr = op_info[op_name]
if not hasattr(transforms, op_name):
raise Exception(
"There's no operator named '{}' in transforms of {}".
format(op_name, self.model_type))
op_list.append(getattr(transforms, op_name)(**op_attr))
eval_transforms = transforms.Compose(op_list)
if hasattr(eval_transforms, 'to_rgb'):
eval_transforms.to_rgb = to_rgb
self.arrange_transforms(eval_transforms)
return eval_transforms
def arrange_transforms(self, eval_transforms):
if self.model_type == 'classifier':
import transforms.cls_transforms as transforms
arrange_transform = transforms.ArrangeClassifier
elif self.model_type == 'segmenter':
import transforms.seg_transforms as transforms
arrange_transform = transforms.ArrangeSegmenter
elif self.model_type == 'detector':
import transforms.det_transforms as transforms
arrange_name = 'Arrange{}'.format(self.model_name)
arrange_transform = getattr(transforms, arrange_name)
else:
raise Exception("Unrecognized model type: {}".format(
self.model_type))
if type(eval_transforms.transforms[-1]).__name__.startswith('Arrange'):
eval_transforms.transforms[-1] = arrange_transform(mode='test')
else:
eval_transforms.transforms.append(arrange_transform(mode='test'))
def raw_predict(self, preprocessed_input):
self.count_num += 1
input_tensor = self.predictor.get_input(0)
input_tensor.resize(self.shape)
input_tensor.set_float_data(preprocessed_input['image'])
if self.model_name == "YOLOv3":
input_size_tensor = self.predictor.get_input(1)
input_size_tensor.resize([1, 2])
input_size_tensor.set_float_data(preprocessed_input['im_size'])
#Start inference
start_time = time.time()
self.predictor.run()
time_use = time.time() - start_time
if (self.count_num >= 20):
self.total_time += time_use
if (self.count_num >= 120):
print("avgtime:", self.total_time * 10)
#Processing output blob
print("Processing output blob")
return
def preprocess(self, image):
res = dict()
if self.model_type == "classifier":
im, = self.transforms(image)
im = np.expand_dims(im, axis=0).copy()
im = im.flatten()
res['image'] = im
elif self.model_type == "detector":
if self.model_name == "YOLOv3":
im, im_shape = self.transforms(image)
im = np.expand_dims(im, axis=0).copy()
im_shape = np.expand_dims(im_shape, axis=0).copy()
res['image'] = im
res['im_size'] = im_shape
if self.model_name.count('RCNN') > 0:
im, im_resize_info, im_shape = self.transforms(image)
im = np.expand_dims(im, axis=0).copy()
im_resize_info = np.expand_dims(im_resize_info, axis=0).copy()
im_shape = np.expand_dims(im_shape, axis=0).copy()
res['image'] = im
res['im_info'] = im_resize_info
res['im_shape'] = im_shape
elif self.model_type == "segmenter":
im, im_info = self.transforms(image)
im = np.expand_dims(im, axis=0).copy()
#np.savetxt('./input_data.txt',im.flatten())
res['image'] = im
res['im_info'] = im_info
return res
def classifier_postprocess(self, topk=1):
output_tensor = self.predictor.get_output(0)
output_data = output_tensor.float_data()
true_topk = min(self.num_classes, topk)
pred_label = np.argsort(-np.array(output_data))[:true_topk]
result = [{
'category_id': l,
'category': self.labels[l],
'score': output_data[l],
} for l in pred_label]
print(result)
return result
def segmenter_postprocess(self, preprocessed_inputs):
out_label_tensor = self.predictor.get_output(0)
out_label = out_label_tensor.float_data()
label_shape = tuple(out_label_tensor.shape())
label_map = np.array(out_label).astype('uint8')
label_map = label_map.reshap(label_shape)
label_map = np.squeeze(label_map)
out_score_tensor = self.predictor.get_output(1)
out_score = out_score_tensor.float_data()
score_shape = tuple(out_score_tensor.shape())
score_map = np.array(out_score)
score_map = score_map.reshap(score_shape)
score_map = np.transpose(score_map, (1, 2, 0))
im_info = preprocessed_inputs['im_info']
for info in im_info[::-1]:
if info[0] == 'resize':
w, h = info[1][1], info[1][0]
label_map = cv2.resize(label_map, (w, h), cv2.INTER_NEAREST)
score_map = cv2.resize(score_map, (w, h), cv2.INTER_LINEAR)
elif info[0] == 'padding':
w, h = info[1][1], info[1][0]
label_map = label_map[0:h, 0:w]
score_map = score_map[0:h, 0:w, :]
else:
raise Exception("Unexpected info '{}' in im_info".format(info[
0]))
return {'label_map': label_map, 'score_map': score_map}
def detector_postprocess(self, preprocessed_inputs):
out_tensor = self.predictor.get_output(0)
out_data = out_tensor.float_data()
out_shape = tuple(out_tensor.shape())
out_data = np.array(out_data)
outputs = label_data.reshap(out_shape)
result = []
for out in outputs:
result.append(out.tolist())
return result
def predict(self, image, topk=1, threshold=0.5):
preprocessed_input = self.preprocess(image)
self.raw_predict(preprocessed_input)
if self.model_type == "classifier":
results = self.classifier_postprocess(topk)
elif self.model_type == "detector":
results = self.detector_postprocess(preprocessed_input)
elif self.model_type == "segmenter":
pass
results = self.segmenter_postprocess(preprocessed_input)
| [
"numpy.transpose",
"os.path.exists",
"numpy.expand_dims",
"time.time",
"numpy.array",
"transforms.det_transforms.Compose",
"numpy.squeeze",
"cv2.resize"
] | [((3044, 3071), 'transforms.det_transforms.Compose', 'transforms.Compose', (['op_list'], {}), '(op_list)\n', (3062, 3071), True, 'import transforms.det_transforms as transforms\n'), ((4717, 4728), 'time.time', 'time.time', ([], {}), '()\n', (4726, 4728), False, 'import time\n'), ((7190, 7211), 'numpy.squeeze', 'np.squeeze', (['label_map'], {}), '(label_map)\n', (7200, 7211), True, 'import numpy as np\n'), ((7393, 7412), 'numpy.array', 'np.array', (['out_score'], {}), '(out_score)\n', (7401, 7412), True, 'import numpy as np\n'), ((7483, 7517), 'numpy.transpose', 'np.transpose', (['score_map', '(1, 2, 0)'], {}), '(score_map, (1, 2, 0))\n', (7495, 7517), True, 'import numpy as np\n'), ((8424, 8442), 'numpy.array', 'np.array', (['out_data'], {}), '(out_data)\n', (8432, 8442), True, 'import numpy as np\n'), ((876, 896), 'os.path.exists', 'osp.exists', (['model_nb'], {}), '(model_nb)\n', (886, 896), True, 'import os.path as osp\n'), ((1162, 1184), 'os.path.exists', 'osp.exists', (['model_yaml'], {}), '(model_yaml)\n', (1172, 1184), True, 'import os.path as osp\n'), ((4777, 4788), 'time.time', 'time.time', ([], {}), '()\n', (4786, 4788), False, 'import time\n'), ((7084, 7103), 'numpy.array', 'np.array', (['out_label'], {}), '(out_label)\n', (7092, 7103), True, 'import numpy as np\n'), ((7713, 7761), 'cv2.resize', 'cv2.resize', (['label_map', '(w, h)', 'cv2.INTER_NEAREST'], {}), '(label_map, (w, h), cv2.INTER_NEAREST)\n', (7723, 7761), False, 'import cv2\n'), ((7790, 7837), 'cv2.resize', 'cv2.resize', (['score_map', '(w, h)', 'cv2.INTER_LINEAR'], {}), '(score_map, (w, h), cv2.INTER_LINEAR)\n', (7800, 7837), False, 'import cv2\n'), ((5210, 5236), 'numpy.expand_dims', 'np.expand_dims', (['im'], {'axis': '(0)'}), '(im, axis=0)\n', (5224, 5236), True, 'import numpy as np\n'), ((6608, 6629), 'numpy.array', 'np.array', (['output_data'], {}), '(output_data)\n', (6616, 6629), True, 'import numpy as np\n'), ((5467, 5493), 'numpy.expand_dims', 'np.expand_dims', (['im'], {'axis': '(0)'}), '(im, axis=0)\n', (5481, 5493), True, 'import numpy as np\n'), ((5528, 5560), 'numpy.expand_dims', 'np.expand_dims', (['im_shape'], {'axis': '(0)'}), '(im_shape, axis=0)\n', (5542, 5560), True, 'import numpy as np\n'), ((5785, 5811), 'numpy.expand_dims', 'np.expand_dims', (['im'], {'axis': '(0)'}), '(im, axis=0)\n', (5799, 5811), True, 'import numpy as np\n'), ((5852, 5890), 'numpy.expand_dims', 'np.expand_dims', (['im_resize_info'], {'axis': '(0)'}), '(im_resize_info, axis=0)\n', (5866, 5890), True, 'import numpy as np\n'), ((5925, 5957), 'numpy.expand_dims', 'np.expand_dims', (['im_shape'], {'axis': '(0)'}), '(im_shape, axis=0)\n', (5939, 5957), True, 'import numpy as np\n'), ((6201, 6227), 'numpy.expand_dims', 'np.expand_dims', (['im'], {'axis': '(0)'}), '(im, axis=0)\n', (6215, 6227), True, 'import numpy as np\n')] |
"""FFX.py v1.3 (Sept 16, 2011)
This module implements the Fast Function Extraction (FFX) algorithm.
Reference: <NAME>, FFX: Fast, Scalable, Deterministic Symbolic
Regression Technology, Genetic Programming Theory and Practice IX, Edited by R.
Riolo, <NAME>, and <NAME>, Springer, 2011. http://www.trent.st/ffx
HOW TO USE THIS MODULE:
Easiest to use by calling runffx.py. Its code has example usage patterns.
The main routines are:
models = MultiFFXModelFactory().build(train_X, train_y, test_X, test_y, varnames)
yhat = model.simulate(X)
print model
Can expand / restrict the set of functions via the user-changeable constants (right below licence).
FFX Software Licence Agreement (like BSD, but adapted for non-commercial gain only)
Copyright (c) 2011, Solido Design Automation Inc. Authored by <NAME>.
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Usage does not involve commercial gain.
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the documentation
and/or other materials provided with the distribution.
* Neither the name of the associated institutions nor the names of its
contributors may be used to endorse or promote products derived from this
software without specific prior written permission.
For permissions beyond the scope of this license, please contact <NAME> (<EMAIL>).
THIS SOFTWARE IS PROVIDED BY THE DEVELOPERS ''AS IS'' AND ANY EXPRESS OR
IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
EVENT SHALL THE DEVELOPERS OR THEIR INSTITUTIONS BE LIABLE FOR ANY DIRECT,
INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;LOSS OF USE,
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Patent pending.
"""
from __future__ import print_function
import itertools
import math
import signal
import time
import sys
from functools import wraps
# 3rd party dependencies
import numpy
import scipy
from sklearn.linear_model import ElasticNet
# user-changeable constants
CONSIDER_INTER = True # consider interactions?
CONSIDER_DENOM = True # consider denominator?
CONSIDER_EXPON = True # consider exponents?
CONSIDER_NONLIN = True # consider abs() and log()?
CONSIDER_THRESH = True # consider hinge functions?
if sys.version_info >= (3, 0):
# hacky python3 compatibility
# so as not to rely on six
xrange = range
itertools.izip = zip
# Make dependency on pandas optional.
try:
import pandas
except ImportError:
pandas = None
INF = float('Inf')
# maximum time (s) for regularization update during pathwise learn.
MAX_TIME_REGULARIZE_UPDATE = 5
# GTH = Greater-Than Hinge function, LTH = Less-Than Hinge function
OP_ABS, OP_MAX0, OP_MIN0, OP_LOG10, OP_GTH, OP_LTH = 1, 2, 3, 4, 5, 6
def _approachStr(approach):
assert len(approach) == 5
assert set(approach).issubset([0, 1])
return 'inter%d denom%d expon%d nonlin%d thresh%d' % \
(approach[0], approach[1], approach[2], approach[3], approach[4])
#=========================================================================
# strategy
class FFXBuildStrategy(object):
"""All parameter settings. Put magic numbers here."""
def __init__(self, approach):
"""
@arguments
approach -- 5-d list of [use_inter, use_denom, use_expon, use_nonlin, use_thresh]
"""
assert len(approach) == 5
assert set(approach).issubset([0, 1])
self.approach = approach
self.num_alphas = 1000
# final round will stop if either of these is hit
self.final_target_train_nmse = 0.01 # 0.01 = 1%
self.final_max_num_bases = 250
# aggressive pruning (note: lasso has l1_ratio=1.0, ridge regression
# has l1_ratio=0.0)
self._l1_ratio = 0.95
# eps -- Length of the path. eps=1e-3 means that alpha_min / alpha_max
# = 1e-3.
self._eps = 1e-70
# will use all if 'nonlin1', else []
self.all_nonlin_ops = [OP_ABS, OP_LOG10]
# will use all if 'thresh1', else []
self.all_threshold_ops = [OP_GTH, OP_LTH]
self.num_thrs_per_var = 5
# will use all if 'expon1', else [1.0]
self.all_expr_exponents = [-1.0, -0.5, +0.5, +1.0]
def includeInteractions(self):
return bool(self.approach[0])
def includeDenominator(self):
return bool(self.approach[1])
def exprExponents(self):
if self.approach[2]:
return self.all_expr_exponents
else:
return [1.0]
def nonlinOps(self):
if self.approach[3]:
return self.all_nonlin_ops
else:
return []
def thresholdOps(self):
if self.approach[4]:
return self.all_threshold_ops
else:
return []
def eps(self):
return self._eps
def l1_ratio(self):
return self._l1_ratio
def numAlphas(self):
return self.num_alphas
#=========================================================================
#models / bases
class FFXModel:
def __init__(self, coefs_n, bases_n, coefs_d, bases_d, varnames=None):
"""
@arguments
coefs_n -- 1d array of float -- coefficients for numerator.
bases_n -- list of *Base -- bases for numerator
coefs_d -- 1d array of float -- coefficients for denominator
bases_d -- list of *Base -- bases for denominator
varnames -- list of string
"""
# preconditions
# offset + numer_bases == numer_coefs
assert 1 + len(bases_n) == len(coefs_n)
assert len(bases_d) == len(coefs_d) # denom_bases == denom_coefs
# make sure that the coefs line up with their 'pretty' versions
coefs_n = numpy.array([float(coefStr(coef)) for coef in coefs_n])
coefs_d = numpy.array([float(coefStr(coef)) for coef in coefs_d])
# reorder numerator bases from highest-to-lowest influence
# -but keep offset 0th of course
offset = coefs_n[0]
coefs_n2 = coefs_n[1:]
I = numpy.argsort(numpy.abs(coefs_n2))[::-1]
coefs_n = [offset] + [coefs_n2[i] for i in I]
bases_n = [bases_n[i] for i in I]
# reorder denominator bases from highest-to-lowest influence
I = numpy.argsort(numpy.abs(coefs_d))[::-1]
coefs_d = [coefs_d[i] for i in I]
bases_d = [bases_d[i] for i in I]
# store values
self.varnames = varnames
self.coefs_n = coefs_n
self.bases_n = bases_n
self.coefs_d = coefs_d
self.bases_d = bases_d
def numBases(self):
"""Return total number of bases"""
return len(self.bases_n) + len(self.bases_d)
def simulate(self, X):
"""
@arguments
X -- 2d array of [sample_i][var_i] : float
@return
y -- 1d array of [sample_i] : float
"""
N = X.shape[0]
# numerator
y = numpy.zeros(N, dtype=float)
y += self.coefs_n[0]
for (coef, base) in itertools.izip(self.coefs_n[1:], self.bases_n):
y += coef * base.simulate(X)
# denominator
if self.bases_d:
denom_y = numpy.zeros(N, dtype=float)
denom_y += 1.0
for (coef, base) in itertools.izip(self.coefs_d, self.bases_d):
denom_y += coef * base.simulate(X)
y /= denom_y
return y
def __str__(self):
return self.str2()
def str2(self, maxlen=100000):
include_denom = bool(self.bases_d)
s = ''
# numerator
if include_denom and len(self.coefs_n) > 1:
s += '('
numer_s = ['%s' % coefStr(self.coefs_n[0])]
for (coef, base) in itertools.izip(self.coefs_n[1:], self.bases_n):
numer_s += ['%s*%s' % (coefStr(coef), base)]
s += ' + '.join(numer_s)
if include_denom and len(self.coefs_n) > 1:
s += ')'
# denominator
if self.bases_d:
s += ' / ('
denom_s = ['1.0']
for (coef, base) in itertools.izip(self.coefs_d, self.bases_d):
denom_s += ['%s*%s' % (coefStr(coef), base)]
s += ' + '.join(denom_s)
s += ')'
# change xi to actual variable names
for var_i in xrange(len(self.varnames) - 1, -1, -1):
s = s.replace('x%d' % var_i, self.varnames[var_i])
s = s.replace('+ -', '- ')
# truncate long strings
if len(s) > maxlen:
s = s[:maxlen] + '...'
return s
def complexity(self):
# Define complexity as the number of nodes needed in the
# corresponding GP tree.
# We have a leading constant, then for each base we have a
# coefficient, a multiply, and a plus, plus the complexity of
# the base itself.
num_complexity = 1 + sum(3 + b.complexity() for b in self.bases_n)
if self.bases_d:
denom_complexity = 1 + sum(3 + b.complexity()
for b in self.bases_d)
# add 1 for the division
return num_complexity + 1 + denom_complexity
else:
return num_complexity
class SimpleBase:
"""e.g. x4^2"""
def __init__(self, var, exponent):
self.var = var
self.exponent = exponent
def simulate(self, X):
"""
@arguments
X -- 2d array of [sample_i][var_i] : float
@return
y -- 1d array of [sample_i] : float
"""
return X[:, self.var] ** self.exponent
def __str__(self):
if self.exponent == 1:
return 'x%d' % self.var
else:
return 'x%d^%g' % (self.var, self.exponent)
def complexity(self):
if self.exponent == 1:
return 1
else:
return 3
class OperatorBase:
"""e.g. log(x4^2)"""
def __init__(self, simple_base, nonlin_op, thr=INF):
"""
@arguments
simple_base -- SimpleBase
nonlin_op -- one of OPS
thr -- None or float -- depends on nonlin_op
"""
self.simple_base = simple_base
self.nonlin_op = nonlin_op
self.thr = thr
def simulate(self, X):
"""
@arguments
X -- 2d array of [sample_i][var_i] : float
@return
y -- 1d array of [sample_i] : float
"""
op = self.nonlin_op
ok = True
y_lin = self.simple_base.simulate(X)
if op == OP_ABS:
ya = numpy.abs(y_lin)
elif op == OP_MAX0:
ya = numpy.clip(y_lin, 0.0, INF)
elif op == OP_MIN0:
ya = numpy.clip(y_lin, -INF, 0.0)
elif op == OP_LOG10:
# safeguard against: log() on values <= 0.0
mn, mx = min(y_lin), max(y_lin)
if mn <= 0.0 or scipy.isnan(mn) or mx == INF or scipy.isnan(mx):
ok = False
else:
ya = numpy.log10(y_lin)
elif op == OP_GTH:
ya = numpy.clip(self.thr - y_lin, 0.0, INF)
elif op == OP_LTH:
ya = numpy.clip(y_lin - self.thr, 0.0, INF)
else:
raise 'Unknown op %d' % op
if ok: # could always do ** exp, but faster ways if exp is 0,1
y = ya
else:
y = INF * numpy.ones(X.shape[0], dtype=float)
return y
def __str__(self):
op = self.nonlin_op
simple_s = str(self.simple_base)
if op == OP_ABS:
return 'abs(%s)' % simple_s
elif op == OP_MAX0:
return 'max(0, %s)' % simple_s
elif op == OP_MIN0:
return 'min(0, %s)' % simple_s
elif op == OP_LOG10:
return 'log10(%s)' % simple_s
elif op == OP_GTH:
return 'max(0,%s-%s)' % (coefStr(self.thr), simple_s)
elif op == OP_LTH:
return ('max(0,%s-%s)' % (simple_s, coefStr(self.thr))).replace('--', '+')
else:
raise 'Unknown op %d' % op
def complexity(self):
"""Return an integer measure of model complexity. It's intended to
measure the number of nodes in the GP tree corresponding to
the model. We assume the GP language includes: +, -, *, /,
MAX0, MIN0, LOG10 but not GTH, LTH. Thus, MAX0(x) returns the
value max(0, x) but contributes only 1 + complexity(x) to the
complexity count. GTH(thr, x) returns the value max(0, thr-x)
but because it would be implemented in GP as MAX0(thr-x) it contributes
3 + complexity(x) to the count."""
op = self.nonlin_op
if op == OP_ABS: return 1 + self.simple_base.complexity()
elif op == OP_MAX0: return 1 + self.simple_base.complexity()
elif op == OP_MIN0: return 1 + self.simple_base.complexity()
elif op == OP_LOG10: return 1 + self.simple_base.complexity()
elif op == OP_GTH: return 3 + self.simple_base.complexity()
elif op == OP_LTH: return 3 + self.simple_base.complexity()
else: raise 'Unknown op %d' % op
class ProductBase:
"""e.g. x2^2 * log(x1^3)"""
def __init__(self, base1, base2):
self.base1 = base1
self.base2 = base2
def simulate(self, X):
"""
@arguments
X -- 2d array of [sample_i][var_i] : float
@return
y -- 1d array of [sample_i] : float
"""
yhat1 = self.base1.simulate(X)
yhat2 = self.base2.simulate(X)
return yhat1 * yhat2
def __str__(self):
return '%s * %s' % (self.base1, self.base2)
def complexity(self):
return 1 + self.base1.complexity() + self.base2.complexity()
class ConstantModel:
"""e.g. 3.2"""
def __init__(self, constant, numvars):
"""
@description
Constructor.
@arguments
constant -- float -- constant value returned by this model
numvars -- int -- number of input variables to this model
"""
self.constant = float(constant)
self.numvars = numvars
def numBases(self):
"""Return total number of bases"""
return 0
def simulate(self, X):
"""
@arguments
X -- 2d array of [sample_i][var_i] : float
@return
y -- 1d array of [sample_i] : float
"""
N = X.shape[0]
if scipy.isnan(self.constant): # corner case
yhat = numpy.array([INF] * N)
else: # typical case
yhat = numpy.ones(N, dtype=float) * self.constant
return yhat
def __str__(self):
return self.str2()
def str2(self, dummy_arg=None):
return coefStr(self.constant)
def complexity(self):
return 1
#==============================================================================
# Model factories
class MultiFFXModelFactory:
def build(self, train_X, train_y, test_X, test_y, varnames=None, verbose=False):
"""
@description
Builds FFX models at many different settings, then merges the results
into a single Pareto Optimal Set.
@argument
train_X -- 2d array of [sample_i][var_i] : float -- training inputs
train_y -- 1d array of [sample_i] : float -- training outputs
test_X -- 2d array -- testing inputs
test_y -- 1d array -- testing outputs
varnames -- list of string -- variable names
@return
models -- list of FFXModel -- Pareto-optimal set of models
"""
if pandas is not None and isinstance(train_X, pandas.DataFrame):
varnames = train_X.columns
train_X = train_X.as_matrix()
test_X = test_X.as_matrix()
if isinstance(train_X, numpy.ndarray) and varnames is None:
raise Exception('varnames required for numpy.ndarray')
if verbose:
print('Build(): begin. {2} variables, {1} training samples, {0} test samples'.format(
test_X.shape[0], *train_X.shape))
models = []
min_y = min(min(train_y), min(test_y))
max_y = max(max(train_y), max(test_y))
# build all combinations of approaches, except for (a) features we don't consider
# and (b) too many features at once
approaches = []
if verbose:
print("Learning Approaches Considered:")
print("=========================================")
print("Inter Denom Expon Nonlin Threshold")
print("=========================================")
if CONSIDER_INTER:
inters = [1] # inter=0 is a subset of inter=1
else:
inters = [0]
for inter in inters:
for denom in [0, 1]:
if denom == 1 and not CONSIDER_DENOM:
continue
for expon in [0, 1]:
if expon == 1 and not CONSIDER_EXPON:
continue
if expon == 1 and inter == 1:
continue # never need both exponent and inter
for nonlin in [0, 1]:
if nonlin == 1 and not CONSIDER_NONLIN:
continue
for thresh in [0, 1]:
if thresh == 1 and not CONSIDER_THRESH:
continue
approach = [inter, denom, expon, nonlin, thresh]
if sum(approach) >= 4:
continue # not too many features at once
approaches.append(approach)
if verbose:
# " ", _approachStr(approach)
print(
'\t'.join(['Yes' if a else 'No' for a in approach]))
for (i, approach) in enumerate(approaches):
if verbose:
print('-' * 200)
print('Build with approach %d/%d (%s): begin' %
(i + 1, len(approaches), _approachStr(approach)))
ss = FFXBuildStrategy(approach)
next_models = FFXModelFactory().build(train_X, train_y, ss, varnames, verbose)
# set test_nmse on each model
for model in next_models:
test_yhat = model.simulate(test_X)
model.test_nmse = nmse(test_yhat, test_y, min_y, max_y)
# pareto filter
if verbose:
print(' STEP 3: Nondominated filter')
next_models = self._nondominatedModels(next_models)
models += next_models
if verbose:
print('Build with approach %d/%d (%s): done. %d model(s).' %
(i + 1, len(approaches), _approachStr(approach), len(next_models)))
print('Models:')
for model in next_models:
print("num_bases=%d, test_nmse=%.6f, model=%s" %
(model.numBases(), model.test_nmse, model.str2(500)))
# final pareto filter
models = self._nondominatedModels(models)
# log nondominated models
if verbose:
print('=' * 200)
print('%d nondominated models (wrt test error & num. bases):' %
len(models))
for (i, model) in enumerate(models):
print("Nondom model %d/%d: num_bases=%d, test_nmse=%.6f, model=%s" %
(i + 1, len(models), model.numBases(), model.test_nmse, model.str2(500)))
return models
def _FFXapproach(self, inter, denom, expon, nonlin, thresh):
return 'FFX inter%d denom%d expon%d nonlin%d thresh%d' % \
(inter, denom, expon, nonlin, thresh)
def _nondominatedModels(self, models):
test_nmses = [model.test_nmse for model in models]
num_bases = [model.numBases() for model in models]
I = nondominatedIndices2d(test_nmses, num_bases)
models = [models[i] for i in I]
I = numpy.argsort([model.numBases() for model in models])
models = [models[i] for i in I]
return models
class FFXModelFactory:
def build(self, X, y, ss, varnames=None, verbose=False):
"""
@description
Build FFX models at the settings of input solution strategy 'ss'.
@argument
X -- 2d array of [var_i][sample_i] : float -- training inputs
y -- 1d array of [sample_i] : float -- training outputs
varnames -- list of string -- variable names
ss -- FFXSolutionStrategy
@return
models -- list of FFXModel -- Pareto-optimal set of models
"""
if pandas is not None and isinstance(X, pandas.DataFrame):
varnames = X.columns
X = X.as_matrix()
if isinstance(X, numpy.ndarray) and varnames is None:
raise Exception('varnames required for numpy.ndarray')
if X.ndim == 1:
self.nrow, self.ncol = len(X), 1
elif X.ndim == 2:
self.nrow, self.ncol = X.shape
else:
raise Exception('X is wrong dimensionality.')
y = numpy.asarray(y)
if self.nrow != len(y):
raise Exception('X sample count and y sample count do not match')
# if y has shape (N, 1) then we reshape to just (N,)
if len(y.shape) > 1:
assert y.shape[1] == 1
y = numpy.reshape(y, (y.shape[0],))
if self.ncol == 0:
print(' Corner case: no input vars, so return a ConstantModel')
return [ConstantModel(y.mean(), 0)]
# Main work...
# build up each combination of all {var_i} x {op_j}, except for
# when a combination is unsuccessful
if verbose:
print(' STEP 1A: Build order-1 bases: begin')
order1_bases = []
for var_i in range(self.ncol):
for exponent in ss.exprExponents():
if exponent == 0.0:
continue
#'lin' version of base
simple_base = SimpleBase(var_i, exponent)
lin_yhat = simple_base.simulate(X)
# checking exponent is a speedup
if exponent in [1.0, 2.0] or not yIsPoor(lin_yhat):
order1_bases.append(simple_base)
# add e.g. OP_ABS, OP_MAX0, OP_MIN0, OP_LOG10
for nonlin_op in ss.nonlinOps():
# ignore cases where op has no effect
if nonlin_op == OP_ABS and exponent in [-2, +2]:
continue
if nonlin_op == OP_MAX0 and min(lin_yhat) >= 0:
continue
if nonlin_op == OP_MIN0 and max(lin_yhat) <= 0:
continue
nonsimple_base = OperatorBase(
simple_base, nonlin_op, None)
nonsimple_base.var = var_i # easy access when considering interactions
nonlin_yhat = nonsimple_base.simulate(X)
if numpy.all(nonlin_yhat == lin_yhat):
continue # op has no effect
if not yIsPoor(nonlin_yhat):
order1_bases.append(nonsimple_base)
# add e.g. OP_GTH, OP_LTH
if exponent == 1.0 and ss.thresholdOps():
minx, maxx = min(X[:, var_i]), max(X[:, var_i])
rangex = maxx - minx
stepx = 0.8 * rangex / float(ss.num_thrs_per_var + 1)
thrs = numpy.arange(
minx + 0.2 * rangex, maxx - 0.2 * rangex + 0.1 * rangex, stepx)
for threshold_op in ss.thresholdOps():
for thr in thrs:
nonsimple_base = OperatorBase(
simple_base, threshold_op, thr)
# easy access when considering interactions
nonsimple_base.var = var_i
order1_bases.append(nonsimple_base)
if verbose:
print(' STEP 1A: Build order-1 bases: done. Have %d order-1 bases.' %
len(order1_bases))
var1_models = None
if ss.includeInteractions():
# find base-1 influences
if verbose:
print(' STEP 1B: Find order-1 base infls: begin')
max_num_bases = len(order1_bases) # no limit
target_train_nmse = 0.01
models = self._basesToModels(
ss, varnames, order1_bases, X, y, max_num_bases, target_train_nmse, verbose)
if models is None: # fit failed.
model = ConstantModel(y[0], 0)
return [model]
var1_models = models
# use most-explaining model (which also has the max num bases)
model = models[-1]
order1_bases = model.bases_n + model.bases_d
if len(order1_bases) == 0: # the most-explaining model is a constant model
model = ConstantModel(y[0], 0)
return [model]
# order bases by influence
order1_infls = numpy.abs(
list(model.coefs_n[1:]) + list(model.coefs_d)) # influences
I = numpy.argsort(-1 * order1_infls)
order1_bases = [order1_bases[i] for i in I]
if verbose:
print(' STEP 1B: Find order-1 base infls: done')
# don't let inter coeffs swamp linear ones; but handle more when n
# small
n_order1_bases = len(order1_bases)
max_n_order2_bases = 3 * math.sqrt(n_order1_bases) # default
max_n_order2_bases = max(max_n_order2_bases, 10) # lower limit
max_n_order2_bases = max(
max_n_order2_bases, 2 * n_order1_bases) # ""
if ss.includeDenominator():
max_n_order2_bases = min(
max_n_order2_bases, 4000) # upper limit
else:
max_n_order2_bases = min(max_n_order2_bases, 8000) # ""
# build up order2 bases
if verbose:
print(' STEP 1C: Build order-2 bases: begin')
# -always have all xi*xi terms
order2_bases = []
order2_basetups = set() # set of (basei_id, basej_id) tuples
for i, basei in enumerate(order1_bases):
if basei.__class__ != SimpleBase:
continue # just xi
if basei.exponent != 1.0:
continue # just exponent==1
combined_base = SimpleBase(basei.var, 2)
order2_bases.append(combined_base)
tup = (id(basei), id(basei))
order2_basetups.add(tup)
# -then add other terms
for max_base_i in xrange(len(order1_bases)):
for i in xrange(max_base_i):
basei = order1_bases[i]
for j in xrange(max_base_i):
if j >= i:
continue # disallow mirror image
basej = order1_bases[j]
tup = (id(basei), id(basej))
if tup in order2_basetups:
continue # no duplicate pairs
combined_base = ProductBase(basei, basej)
order2_bases.append(combined_base)
order2_basetups.add(tup)
if len(order2_bases) >= max_n_order2_bases:
break # for j
if len(order2_bases) >= max_n_order2_bases:
break # for i
if len(order2_bases) >= max_n_order2_bases:
break # for max_base_i
if verbose:
print(' STEP 1C: Build order-2 bases: done. Have %d order-2'
' bases.' % len(order2_bases))
bases = order1_bases + order2_bases
else:
bases = order1_bases
# all bases. Stop based on target nmse, not number of bases
if verbose:
print(' STEP 2: Regress on all %d bases: begin.' % len(bases))
var2_models = self._basesToModels(
ss, varnames, bases, X, y, ss.final_max_num_bases, ss.final_target_train_nmse, verbose)
if verbose:
print(' STEP 2: Regress on all %d bases: done.' % len(bases))
# combine models having 1-var with models having 2-vars
if var1_models is None and var2_models is None:
models = []
elif var1_models is None and var2_models is not None:
models = var2_models
elif var1_models is not None and var2_models is None:
models = var1_models
else: # var1_models is not None and var2_models is not None
models = var1_models + var2_models
# add constant; done
models = [ConstantModel(numpy.mean(y), X.shape[0])] + models
return models
def _basesToModels(self, ss, varnames, bases, X, y, max_num_bases, target_train_nmse, verbose):
# compute regress_X
if ss.includeDenominator():
regress_X = numpy.zeros((self.nrow, len(bases) * 2), dtype=float)
else:
regress_X = numpy.zeros((self.nrow, len(bases)), dtype=float)
for i, base in enumerate(bases):
base_y = base.simulate(X)
regress_X[:, i] = base_y # numerators
if ss.includeDenominator():
regress_X[:, len(bases) + i] = -1.0 * \
base_y * y # denominators
# compute models.
models = self._pathwiseLearn(ss, varnames, bases, X, regress_X, y,
max_num_bases, target_train_nmse, verbose)
return models
def _pathwiseLearn(self, ss, varnames, bases, X_orig, X_orig_regress, y_orig,
max_num_bases, target_nmse, verbose=False, **fit_params):
"""Adapted from enet_path() in sklearn.linear_model.
http://scikit-learn.sourceforge.net/modules/linear_model.html
Compute Elastic-Net path with coordinate descent.
Returns list of model (or None if failure)."""
if verbose:
print(' Pathwise learn: begin. max_num_bases=%d' % max_num_bases)
max_iter = 1000 # default 1000. magic number.
# Condition X and y:
# -"unbias" = rescale so that (mean=0, stddev=1) -- subtract each row's
# mean, then divide by stddev
# -X transposed
# -X as fortran array
(X_unbiased, y_unbiased, X_avgs, X_stds, y_avg,
y_std) = self._unbiasedXy(X_orig_regress, y_orig)
# make data contiguous in memory
X_unbiased = numpy.asfortranarray(X_unbiased)
n_samples = X_unbiased.shape[0]
vals = numpy.dot(X_unbiased.T, y_unbiased)
vals = [val for val in vals if not scipy.isnan(val)]
if vals:
alpha_max = numpy.abs(max(vals) / (n_samples * ss.l1_ratio()))
else:
alpha_max = 1.0 # backup: pick a value from the air
# alphas = lotsa alphas at beginning, and usual rate for rest
st, fin = numpy.log10(alpha_max * ss.eps()), numpy.log10(alpha_max)
alphas1 = numpy.logspace(
st, fin, num=ss.numAlphas() * 10)[::-1][:ss.numAlphas() // 4]
alphas2 = numpy.logspace(st, fin, num=ss.numAlphas())
alphas = sorted(set(alphas1).union(alphas2), reverse=True)
if 'precompute' not in fit_params or fit_params['precompute'] is True:
fit_params['precompute'] = numpy.dot(X_unbiased.T, X_unbiased)
# if not 'Xy' in fit_params or fit_params['Xy'] is None:
# fit_params['Xy'] = numpy.dot(X_unbiased.T, y_unbiased)
models = [] # build this up
nmses = [] # for detecting stagnation
cur_unbiased_coefs = None # init values for coefs
start_time = time.time()
for (alpha_i, alpha) in enumerate(alphas):
# compute (unbiased) coefficients. Recall that mean=0 so no
# intercept needed
clf = ElasticNetWithTimeout(alpha=alpha, l1_ratio=ss.l1_ratio(), fit_intercept=False,
max_iter=max_iter, **fit_params)
try:
clf.fit(X_unbiased, y_unbiased)
except TimeoutError:
print(' Regularized update failed. Returning None')
return None # failure
except ValueError:
print(' Regularized update failed with ValueError.')
print(' X_unbiased:')
print(X_unbiased)
print(' y_unbiased:')
print(y_unbiased)
sys.exit(1)
cur_unbiased_coefs = clf.coef_.copy()
if cur_unbiased_coefs.shape == tuple():
# This happens when we have only one variable because
# ElasticNet calls numpy.squeeze(), which reduces a
# single element array to a 0-d array. That would
# crash us below in list(cur_unbiased_coefs). We just
# undo the squeeze.
cur_unbiased_coefs = cur_unbiased_coefs.reshape((1,))
# compute model; update models
# -"rebias" means convert from (mean=0, stddev=1) to original (mean, stddev)
coefs = self._rebiasCoefs(
[0.0] + list(cur_unbiased_coefs), X_stds, X_avgs, y_std, y_avg)
(coefs_n, bases_n, coefs_d, bases_d) = self._allocateToNumerDenom(
ss, bases, coefs)
model = FFXModel(coefs_n, bases_n, coefs_d, bases_d, varnames)
models.append(model)
# update nmses
nmse_unbiased = nmse(numpy.dot(cur_unbiased_coefs, X_unbiased.T), y_unbiased,
min(y_unbiased), max(y_unbiased))
nmses.append(nmse_unbiased)
# log
num_bases = len(numpy.nonzero(cur_unbiased_coefs)[0])
if verbose and ((alpha_i == 0) or (alpha_i + 1) % 50 == 0):
print(' alpha %d/%d (%3e): num_bases=%d, nmse=%.6f, time %.2f s' %
(alpha_i + 1, len(alphas), alpha, num_bases,
nmse_unbiased, time.time() - start_time))
# maybe stop
if scipy.isinf(nmses[-1]):
if verbose:
print(' Pathwise learn: Early stop because nmse is inf')
return None
if nmse_unbiased < target_nmse:
if verbose:
print(' Pathwise learn: Early stop because nmse < target')
return models
if num_bases > max_num_bases:
if verbose:
print(' Pathwise learn: Early stop because num bases > %d' % max_num_bases)
return models
if len(nmses) > 15 and round(nmses[-1], 4) == round(nmses[-15], 4):
if verbose:
print(' Pathwise learn: Early stop because nmses stagnated')
return models
if verbose:
print(' Pathwise learn: done')
return models
def _allocateToNumerDenom(self, ss, bases, coefs):
"""Prune out nonzero coefficients/bases. Allocate to numerator vs. denominator."""
if ss.includeDenominator():
# offset + numer_bases + denom_bases
assert 1 + len(bases) + len(bases) == len(coefs)
n_bases = len(bases)
coefs_n = [coefs[0]] + \
[coef for coef in coefs[1:n_bases + 1] if coef != 0]
bases_n = [base for (base, coef) in itertools.izip(
bases, coefs[1:n_bases + 1]) if coef != 0]
coefs_d = [coef for coef in coefs[n_bases + 1:] if coef != 0]
bases_d = [base for (base, coef) in itertools.izip(
bases, coefs[n_bases + 1:]) if coef != 0]
else:
# offset + numer_bases + denom_bases
assert 1 + len(bases) == len(coefs)
coefs_n = [coefs[0]] + [coef for coef in coefs[1:] if coef != 0]
bases_n = [base for (base, coef) in itertools.izip(
bases, coefs[1:]) if coef != 0]
coefs_d = []
bases_d = []
return (coefs_n, bases_n, coefs_d, bases_d)
def _unbiasedXy(self, Xin, yin):
"""Make all input rows of X, and y, to have mean=0 stddev=1 """
# unbiased X
X_avgs, X_stds = Xin.mean(0), Xin.std(0)
# if any stddev was 0, overwrite with 1 to prevent divide by
# zero. Because we then return the overwritten value,
#_rebiasCoefs will end up with the right rebiased values. Same
# for y below.
numpy.copyto(X_stds, 1.0, where=~(X_stds > 0.0))
X_unbiased = (Xin - X_avgs) / X_stds
# unbiased y
y_avg, y_std = yin.mean(0), yin.std(0)
# if stddev was 0, overwrite with 1
if not y_std > 0.0:
y_std = 1.0
y_unbiased = (yin - y_avg) / y_std
assert numpy.all(numpy.isfinite(X_unbiased))
assert numpy.all(numpy.isfinite(y_unbiased))
return (X_unbiased, y_unbiased, X_avgs, X_stds, y_avg, y_std)
def _rebiasCoefs(self, unbiased_coefs, X_stds, X_avgs, y_std, y_avg):
"""Given the coefficients that were learned using unbiased training data, rebias the
coefficients so that they are at the scale of the real training X and y."""
# preconditions
assert unbiased_coefs is not None
assert len(unbiased_coefs) == (len(X_stds) + 1) == (len(X_avgs) + 1), \
(len(unbiased_coefs), (len(X_stds) + 1), (len(X_avgs) + 1))
# main work
n = len(X_stds)
coefs = numpy.zeros(n + 1, dtype=float)
coefs[0] = unbiased_coefs[0] * y_std + y_avg
for j in range(1, n + 1):
coefs[j] = unbiased_coefs[j] * y_std / X_stds[j - 1]
coefs[0] -= coefs[j] * X_avgs[j - 1]
return coefs
#=========================================================================
# Revise linear_model.coordinate_descent.ElasticNet.fit() to handle when it hangs
# http://www.saltycrane.com/blog/2010/04/using-python-timeout-decorator-uploading-s3/
class TimeoutError(Exception):
def __init__(self, value="Timed Out"):
self.value = value
def __str__(self):
return repr(self.value)
def timeout(seconds_before_timeout):
def decorate(f):
# Just do without the timeout on Windows: see
# https://github.com/natekupp/ffx/issues/17
if not hasattr(signal, "SIGALRM"):
return f
def handler(signum, frame):
raise TimeoutError()
@wraps(f)
def new_f(*args, **kwargs):
old = signal.signal(signal.SIGALRM, handler)
signal.alarm(seconds_before_timeout)
try:
result = f(*args, **kwargs)
finally:
signal.signal(signal.SIGALRM, old)
signal.alarm(0)
return result
return new_f
return decorate
class ElasticNetWithTimeout(ElasticNet):
# if this freezes, then exit with a TimeoutError
@timeout(MAX_TIME_REGULARIZE_UPDATE)
def fit(self, *args, **kwargs):
return ElasticNet.fit(self, *args, **kwargs)
#=========================================================================
# utility classes / functions
def nondominatedIndices2d(cost0s, cost1s):
"""
@description
Find indices of individuals that are on the nondominated 2-d tradeoff.
@arguments
cost0s -- 1d array of float [model_i] -- want to minimize this. E.g. complexity.
cost1s -- 1d array of float [model_i] -- want to minimize this too. E.g. nmse.
@return
nondomI -- list of int -- nondominated indices; each is in range [0, #inds - 1]
ALWAYS returns at least one entry if there is valid data
"""
cost0s, cost1s = numpy.asarray(cost0s), numpy.asarray(cost1s)
n_points = len(cost0s)
assert n_points == len(cost1s)
if n_points == 0: # corner case
return []
# indices of cost0s sorted for ascending order
I = numpy.argsort(cost0s)
#'cur' == best at this cost0s
best_cost = [cost0s[I[0]], cost1s[I[0]]]
best_cost_index = I[0]
nondom_locs = []
for i in xrange(n_points):
loc = I[i] # traverse cost0s in ascending order
if cost0s[loc] == best_cost[0]:
if cost1s[loc] < best_cost[1]:
best_cost_index = loc
best_cost = [cost0s[loc], cost1s[loc]]
else: # cost0s[loc] > best_cost[0] because
# loc indexes cost0s in ascending order
if not nondom_locs:
# initial value
nondom_locs = [best_cost_index]
elif best_cost[1] < cost1s[nondom_locs[-1]]:
# if the current cost is lower than the last item
# on the non-dominated list, add it's index to
# the non-dominated list
nondom_locs.append(best_cost_index)
# set up "last tested value"
best_cost_index = loc
best_cost = [cost0s[loc], cost1s[loc]]
if not nondom_locs:
# if none are non-dominated, return the last one
nondom_locs = [loc]
elif best_cost[1] < cost1s[nondom_locs[-1]]:
# if the current cost is lower than the last item
# on the non-dominated list, add it's index to
# the non-dominated list
nondom_locs.append(best_cost_index)
# return the non-dominated in sorted order
nondomI = sorted(nondom_locs)
return nondomI
def nmse(yhat, y, min_y, max_y):
"""
@description
Calculates the normalized mean-squared error.
@arguments
yhat -- 1d array or list of floats -- estimated values of y
y -- 1d array or list of floats -- true values
min_y, max_y -- float, float -- roughly the min and max; they
do not have to be the perfect values of min and max, because
they're just here to scale the output into a roughly [0,1] range
@return
nmse -- float -- normalized mean-squared error
"""
# base case: no entries
if len(yhat) == 0:
return 0.0
# base case: both yhat and y are constant, and same values
if (max_y == min_y) and (max(yhat) == min(yhat) == max(y) == min(y)):
return 0.0
# main case
assert max_y > min_y, 'max_y=%g was not > min_y=%g' % (max_y, min_y)
yhat_a, y_a = numpy.asarray(yhat), numpy.asarray(y)
y_range = float(max_y - min_y)
try:
result = math.sqrt(numpy.mean(((yhat_a - y_a) / y_range) ** 2))
if scipy.isnan(result):
return INF
return result
except:
return INF
def yIsPoor(y):
"""Returns True if y is not usable"""
return max(scipy.isinf(y)) or max(scipy.isnan(y))
def coefStr(x):
"""Gracefully print a number to 3 significant digits. See _testCoefStr in
unit tests"""
if x == 0.0:
s = '0'
elif numpy.abs(x) < 1e-4: s = ('%.2e' % x).replace('e-0', 'e-')
elif numpy.abs(x) < 1e-3: s = '%.6f' % x
elif numpy.abs(x) < 1e-2: s = '%.5f' % x
elif numpy.abs(x) < 1e-1: s = '%.4f' % x
elif numpy.abs(x) < 1e0: s = '%.3f' % x
elif numpy.abs(x) < 1e1: s = '%.2f' % x
elif numpy.abs(x) < 1e2: s = '%.1f' % x
elif numpy.abs(x) < 1e4: s = '%.0f' % x
else: s = ('%.2e' % x).replace('e+0', 'e')
return s
def basesStr(bases):
"""Pretty print list of bases"""
return ', '.join([str(base) for base in bases])
def rail(x, minx, maxx):
return max(minx, max(maxx, x))
| [
"numpy.abs",
"numpy.ones",
"numpy.clip",
"numpy.argsort",
"numpy.mean",
"numpy.arange",
"numpy.isfinite",
"numpy.reshape",
"signal.alarm",
"numpy.log10",
"scipy.isinf",
"math.sqrt",
"numpy.asarray",
"numpy.asfortranarray",
"scipy.isnan",
"functools.wraps",
"numpy.dot",
"signal.sign... | [((41297, 41318), 'numpy.argsort', 'numpy.argsort', (['cost0s'], {}), '(cost0s)\n', (41310, 41318), False, 'import numpy\n'), ((7638, 7665), 'numpy.zeros', 'numpy.zeros', (['N'], {'dtype': 'float'}), '(N, dtype=float)\n', (7649, 7665), False, 'import numpy\n'), ((7723, 7769), 'itertools.izip', 'itertools.izip', (['self.coefs_n[1:]', 'self.bases_n'], {}), '(self.coefs_n[1:], self.bases_n)\n', (7737, 7769), False, 'import itertools\n'), ((8426, 8472), 'itertools.izip', 'itertools.izip', (['self.coefs_n[1:]', 'self.bases_n'], {}), '(self.coefs_n[1:], self.bases_n)\n', (8440, 8472), False, 'import itertools\n'), ((15129, 15155), 'scipy.isnan', 'scipy.isnan', (['self.constant'], {}), '(self.constant)\n', (15140, 15155), False, 'import scipy\n'), ((21956, 21972), 'numpy.asarray', 'numpy.asarray', (['y'], {}), '(y)\n', (21969, 21972), False, 'import numpy\n'), ((31793, 31825), 'numpy.asfortranarray', 'numpy.asfortranarray', (['X_unbiased'], {}), '(X_unbiased)\n', (31813, 31825), False, 'import numpy\n'), ((31882, 31917), 'numpy.dot', 'numpy.dot', (['X_unbiased.T', 'y_unbiased'], {}), '(X_unbiased.T, y_unbiased)\n', (31891, 31917), False, 'import numpy\n'), ((32995, 33006), 'time.time', 'time.time', ([], {}), '()\n', (33004, 33006), False, 'import time\n'), ((37843, 37891), 'numpy.copyto', 'numpy.copyto', (['X_stds', '(1.0)'], {'where': '(~(X_stds > 0.0))'}), '(X_stds, 1.0, where=~(X_stds > 0.0))\n', (37855, 37891), False, 'import numpy\n'), ((38854, 38885), 'numpy.zeros', 'numpy.zeros', (['(n + 1)'], {'dtype': 'float'}), '(n + 1, dtype=float)\n', (38865, 38885), False, 'import numpy\n'), ((39822, 39830), 'functools.wraps', 'wraps', (['f'], {}), '(f)\n', (39827, 39830), False, 'from functools import wraps\n'), ((40390, 40427), 'sklearn.linear_model.ElasticNet.fit', 'ElasticNet.fit', (['self', '*args'], {}), '(self, *args, **kwargs)\n', (40404, 40427), False, 'from sklearn.linear_model import ElasticNet\n'), ((41074, 41095), 'numpy.asarray', 'numpy.asarray', (['cost0s'], {}), '(cost0s)\n', (41087, 41095), False, 'import numpy\n'), ((41097, 41118), 'numpy.asarray', 'numpy.asarray', (['cost1s'], {}), '(cost1s)\n', (41110, 41118), False, 'import numpy\n'), ((43668, 43687), 'numpy.asarray', 'numpy.asarray', (['yhat'], {}), '(yhat)\n', (43681, 43687), False, 'import numpy\n'), ((43689, 43705), 'numpy.asarray', 'numpy.asarray', (['y'], {}), '(y)\n', (43702, 43705), False, 'import numpy\n'), ((43833, 43852), 'scipy.isnan', 'scipy.isnan', (['result'], {}), '(result)\n', (43844, 43852), False, 'import scipy\n'), ((7882, 7909), 'numpy.zeros', 'numpy.zeros', (['N'], {'dtype': 'float'}), '(N, dtype=float)\n', (7893, 7909), False, 'import numpy\n'), ((7969, 8011), 'itertools.izip', 'itertools.izip', (['self.coefs_d', 'self.bases_d'], {}), '(self.coefs_d, self.bases_d)\n', (7983, 8011), False, 'import itertools\n'), ((8771, 8813), 'itertools.izip', 'itertools.izip', (['self.coefs_d', 'self.bases_d'], {}), '(self.coefs_d, self.bases_d)\n', (8785, 8813), False, 'import itertools\n'), ((11239, 11255), 'numpy.abs', 'numpy.abs', (['y_lin'], {}), '(y_lin)\n', (11248, 11255), False, 'import numpy\n'), ((15191, 15213), 'numpy.array', 'numpy.array', (['([INF] * N)'], {}), '([INF] * N)\n', (15202, 15213), False, 'import numpy\n'), ((22225, 22256), 'numpy.reshape', 'numpy.reshape', (['y', '(y.shape[0],)'], {}), '(y, (y.shape[0],))\n', (22238, 22256), False, 'import numpy\n'), ((26257, 26289), 'numpy.argsort', 'numpy.argsort', (['(-1 * order1_infls)'], {}), '(-1 * order1_infls)\n', (26270, 26289), False, 'import numpy\n'), ((32274, 32296), 'numpy.log10', 'numpy.log10', (['alpha_max'], {}), '(alpha_max)\n', (32285, 32296), False, 'import numpy\n'), ((32653, 32688), 'numpy.dot', 'numpy.dot', (['X_unbiased.T', 'X_unbiased'], {}), '(X_unbiased.T, X_unbiased)\n', (32662, 32688), False, 'import numpy\n'), ((35421, 35443), 'scipy.isinf', 'scipy.isinf', (['nmses[-1]'], {}), '(nmses[-1])\n', (35432, 35443), False, 'import scipy\n'), ((38171, 38197), 'numpy.isfinite', 'numpy.isfinite', (['X_unbiased'], {}), '(X_unbiased)\n', (38185, 38197), False, 'import numpy\n'), ((38224, 38250), 'numpy.isfinite', 'numpy.isfinite', (['y_unbiased'], {}), '(y_unbiased)\n', (38238, 38250), False, 'import numpy\n'), ((39885, 39923), 'signal.signal', 'signal.signal', (['signal.SIGALRM', 'handler'], {}), '(signal.SIGALRM, handler)\n', (39898, 39923), False, 'import signal\n'), ((39936, 39972), 'signal.alarm', 'signal.alarm', (['seconds_before_timeout'], {}), '(seconds_before_timeout)\n', (39948, 39972), False, 'import signal\n'), ((40118, 40133), 'signal.alarm', 'signal.alarm', (['(0)'], {}), '(0)\n', (40130, 40133), False, 'import signal\n'), ((43777, 43820), 'numpy.mean', 'numpy.mean', (['(((yhat_a - y_a) / y_range) ** 2)'], {}), '(((yhat_a - y_a) / y_range) ** 2)\n', (43787, 43820), False, 'import numpy\n'), ((44005, 44019), 'scipy.isinf', 'scipy.isinf', (['y'], {}), '(y)\n', (44016, 44019), False, 'import scipy\n'), ((44028, 44042), 'scipy.isnan', 'scipy.isnan', (['y'], {}), '(y)\n', (44039, 44042), False, 'import scipy\n'), ((44201, 44213), 'numpy.abs', 'numpy.abs', (['x'], {}), '(x)\n', (44210, 44213), False, 'import numpy\n'), ((6765, 6784), 'numpy.abs', 'numpy.abs', (['coefs_n2'], {}), '(coefs_n2)\n', (6774, 6784), False, 'import numpy\n'), ((6984, 7002), 'numpy.abs', 'numpy.abs', (['coefs_d'], {}), '(coefs_d)\n', (6993, 7002), False, 'import numpy\n'), ((11301, 11328), 'numpy.clip', 'numpy.clip', (['y_lin', '(0.0)', 'INF'], {}), '(y_lin, 0.0, INF)\n', (11311, 11328), False, 'import numpy\n'), ((12041, 12076), 'numpy.ones', 'numpy.ones', (['X.shape[0]'], {'dtype': 'float'}), '(X.shape[0], dtype=float)\n', (12051, 12076), False, 'import numpy\n'), ((15263, 15289), 'numpy.ones', 'numpy.ones', (['N'], {'dtype': 'float'}), '(N, dtype=float)\n', (15273, 15289), False, 'import numpy\n'), ((26620, 26645), 'math.sqrt', 'math.sqrt', (['n_order1_bases'], {}), '(n_order1_bases)\n', (26629, 26645), False, 'import math\n'), ((34839, 34882), 'numpy.dot', 'numpy.dot', (['cur_unbiased_coefs', 'X_unbiased.T'], {}), '(cur_unbiased_coefs, X_unbiased.T)\n', (34848, 34882), False, 'import numpy\n'), ((40071, 40105), 'signal.signal', 'signal.signal', (['signal.SIGALRM', 'old'], {}), '(signal.SIGALRM, old)\n', (40084, 40105), False, 'import signal\n'), ((44269, 44281), 'numpy.abs', 'numpy.abs', (['x'], {}), '(x)\n', (44278, 44281), False, 'import numpy\n'), ((11374, 11402), 'numpy.clip', 'numpy.clip', (['y_lin', '(-INF)', '(0.0)'], {}), '(y_lin, -INF, 0.0)\n', (11384, 11402), False, 'import numpy\n'), ((29971, 29984), 'numpy.mean', 'numpy.mean', (['y'], {}), '(y)\n', (29981, 29984), False, 'import numpy\n'), ((31961, 31977), 'scipy.isnan', 'scipy.isnan', (['val'], {}), '(val)\n', (31972, 31977), False, 'import scipy\n'), ((33809, 33820), 'sys.exit', 'sys.exit', (['(1)'], {}), '(1)\n', (33817, 33820), False, 'import sys\n'), ((35050, 35083), 'numpy.nonzero', 'numpy.nonzero', (['cur_unbiased_coefs'], {}), '(cur_unbiased_coefs)\n', (35063, 35083), False, 'import numpy\n'), ((36755, 36798), 'itertools.izip', 'itertools.izip', (['bases', 'coefs[1:n_bases + 1]'], {}), '(bases, coefs[1:n_bases + 1])\n', (36769, 36798), False, 'import itertools\n'), ((36952, 36994), 'itertools.izip', 'itertools.izip', (['bases', 'coefs[n_bases + 1:]'], {}), '(bases, coefs[n_bases + 1:])\n', (36966, 36994), False, 'import itertools\n'), ((37263, 37295), 'itertools.izip', 'itertools.izip', (['bases', 'coefs[1:]'], {}), '(bases, coefs[1:])\n', (37277, 37295), False, 'import itertools\n'), ((44314, 44326), 'numpy.abs', 'numpy.abs', (['x'], {}), '(x)\n', (44323, 44326), False, 'import numpy\n'), ((23941, 23975), 'numpy.all', 'numpy.all', (['(nonlin_yhat == lin_yhat)'], {}), '(nonlin_yhat == lin_yhat)\n', (23950, 23975), False, 'import numpy\n'), ((24486, 24562), 'numpy.arange', 'numpy.arange', (['(minx + 0.2 * rangex)', '(maxx - 0.2 * rangex + 0.1 * rangex)', 'stepx'], {}), '(minx + 0.2 * rangex, maxx - 0.2 * rangex + 0.1 * rangex, stepx)\n', (24498, 24562), False, 'import numpy\n'), ((44359, 44371), 'numpy.abs', 'numpy.abs', (['x'], {}), '(x)\n', (44368, 44371), False, 'import numpy\n'), ((11560, 11575), 'scipy.isnan', 'scipy.isnan', (['mn'], {}), '(mn)\n', (11571, 11575), False, 'import scipy\n'), ((11592, 11607), 'scipy.isnan', 'scipy.isnan', (['mx'], {}), '(mx)\n', (11603, 11607), False, 'import scipy\n'), ((11675, 11693), 'numpy.log10', 'numpy.log10', (['y_lin'], {}), '(y_lin)\n', (11686, 11693), False, 'import numpy\n'), ((11738, 11776), 'numpy.clip', 'numpy.clip', (['(self.thr - y_lin)', '(0.0)', 'INF'], {}), '(self.thr - y_lin, 0.0, INF)\n', (11748, 11776), False, 'import numpy\n'), ((44404, 44416), 'numpy.abs', 'numpy.abs', (['x'], {}), '(x)\n', (44413, 44416), False, 'import numpy\n'), ((11821, 11859), 'numpy.clip', 'numpy.clip', (['(y_lin - self.thr)', '(0.0)', 'INF'], {}), '(y_lin - self.thr, 0.0, INF)\n', (11831, 11859), False, 'import numpy\n'), ((35353, 35364), 'time.time', 'time.time', ([], {}), '()\n', (35362, 35364), False, 'import time\n'), ((44449, 44461), 'numpy.abs', 'numpy.abs', (['x'], {}), '(x)\n', (44458, 44461), False, 'import numpy\n'), ((44494, 44506), 'numpy.abs', 'numpy.abs', (['x'], {}), '(x)\n', (44503, 44506), False, 'import numpy\n'), ((44539, 44551), 'numpy.abs', 'numpy.abs', (['x'], {}), '(x)\n', (44548, 44551), False, 'import numpy\n')] |
"""
DataContainer class for linking directories containing different sorts of data.
This is meant to make plotting and analysis easier.
TO DO
-----
- request random subsets.
- make sure input directories are iterable
- add features to existing files.
"""
__date__ = "July-November 2019"
import h5py
try:
from numba.errors import NumbaPerformanceWarning
except NameError:
pass
import numpy as np
import os
from scipy.io import wavfile
from sklearn.decomposition import PCA
from time import strptime, mktime
import torch
import umap
import warnings
from ava.models.vae import VAE
from ava.models.vae_dataset import get_syllable_partition, \
get_syllable_data_loaders, get_hdf5s_from_dir
AUDIO_FIELDS = ['audio']
FILENAME_FIELDS = ['sap_time']
SEGMENT_FIELDS = ['segments', 'segment_audio']
PROJECTION_FIELDS = ['latent_means', 'latent_mean_pca', 'latent_mean_umap']
SPEC_FIELDS = ['specs', 'onsets', 'offsets', 'audio_filenames']
MUPET_FIELDS = ['syllable_number', 'syllable_start_time', 'syllable_end_time',
'inter-syllable_interval', 'syllable_duration', 'starting_frequency',
'final_frequency', 'minimum_frequency', 'maximum_frequency',
'mean_frequency', 'frequency_bandwidth', 'total_syllable_energy',
'peak_syllable_amplitude', 'cluster']
DEEPSQUEAK_FIELDS = ['id', 'label', 'accepted', 'score', 'begin_time',
'end_time', 'call_length', 'principal_frequency', 'low_freq', 'high_freq',
'delta_freq', 'frequency_standard_deviation', 'slope', 'sinuosity',
'mean_power', 'tonality']
SAP_FIELDS = ['syllable_duration_sap', 'syllable_start', 'mean_amplitude',
'mean_pitch', 'mean_FM', 'mean_AM2', 'mean_entropy', 'mean_pitch_goodness',
'mean_mean_freq', 'pitch_variance', 'FM_variance', 'entropy_variance',
'pitch_goodness_variance', 'mean_freq_variance', 'AM_variance']
ALL_FIELDS = AUDIO_FIELDS + FILENAME_FIELDS + SEGMENT_FIELDS + \
PROJECTION_FIELDS + SPEC_FIELDS + MUPET_FIELDS + DEEPSQUEAK_FIELDS + \
SAP_FIELDS
"""All fields that can be requested by a DataContainer object."""
MUPET_ONSET_COL = MUPET_FIELDS.index('syllable_start_time')
DEEPSQUEAK_ONSET_COL = DEEPSQUEAK_FIELDS.index('begin_time')
SAP_ONSET_COL = SAP_FIELDS.index('syllable_start')
PRETTY_NAMES = {
'audio': 'Audio',
'segments': 'Segments',
'segment_audio': 'Segment Audio',
'latent_means': 'Latent Means',
'latent_mean_pca': 'Latent Mean PCA Projection',
'latent_mean_umap': 'Latent Mean UMAP Projection',
'specs': 'Spectrograms',
'onsets': 'Onsets (s)',
'offsets': 'Offsets (s)',
'aduio_filenames': 'Filenames',
'syllable_number': 'Syllable Number',
'syllable_start_time': 'Onsets (s)',
'syllable_duration': 'Duration (ms)',
'starting_frequency': 'Starting Freq. (kHz)',
'final_frequency': 'Final Freq. (kHz)',
'minimum_frequency': 'Min Freq. (kHz)',
'maximum_frequency': 'Max Freq. (kHz)',
'mean_frequency': 'Mean Freq. (kHz)',
'frequency_bandwidth': 'Freq. Bandwidth (kHz)',
'total_syllable_energy': 'Total Energy (dB)',
'peak_syllable_amplitude': 'Peak Amplitude (dB)',
'cluster': 'Cluster',
'id': 'Syllabler Number',
'label': 'Label',
'accepted': 'Accepted',
'score': 'DeepSqueak Detection Score',
'begin_time': 'Onsets (s)',
'end_time': 'Offsets (s)',
'call_length': 'Duration (ms)',
'principal_frequency': 'Principal Freq. (kHz)',
'low_freq': 'Minimum Freq. (kHz)',
'high_freq': 'Max Freq. (kHz)',
'delta_freq': 'Freq. Bandwidth (kHz)',
'frequency_standard_deviation': 'Freq Std. Dev. (kHz)',
'slope': 'Freq. Mod. (kHz/s)',
'sinuosity': 'Sinuosity',
'mean_power': 'Power (dB/Hz)',
'tonality': 'Tonality',
'syllable_duration_sap': 'Duration (s)',
'syllable_start': 'Onset (s)',
'mean_amplitude': 'Amplitude',
'mean_pitch': 'Pitch',
'mean_FM': 'Freq. Mod.',
'mean_AM2': 'Amp. Mod.',
'mean_entropy': 'Entropy',
'mean_pitch_goodness': 'Goodness of Pitch',
'mean_mean_freq': 'Mean Frequency',
'pitch_variance': 'Pitch Variance',
'FM_variance': 'Freq. Mod. Var.',
'entropy_variance': 'Entropy Var.',
'pitch_goodness_variance': 'Goodness of Pitch Var.',
'mean_freq_variance': 'Freq. Var.',
'AM_variance': 'Amp. Mod. Var.',
}
PRETTY_NAMES_NO_UNITS = {}
for k in PRETTY_NAMES:
PRETTY_NAMES_NO_UNITS[k] = ' '.join(PRETTY_NAMES[k].split('(')[0].split(' '))
class DataContainer():
"""
Link directories containing different data sources for easy plotting.
The idea here is for plotting and analysis tools to accept a DataContainer,
from which they can request different types of data. Those requests can then
be handled here in a central location, which can cut down on redundant code
and processing steps.
Attributes
----------
audio_dirs : {list of str, None}, optional
Directories containing audio. Defaults to None.
segment_dirs : {list of str, None}, optional
Directories containing segmenting decisions.
spec_dirs : list of {str, None}, optional
Directories containing hdf5 files of spectrograms. These should be
files output by ava.preprocessing.preprocessing. Defaults to None.
model_filename : {str, None}, optional
The VAE checkpoint to load. Written by models.vae.save_state.
Defaults to None.
projection_dirs : list of {str, None}, optional
Directory containing different projections. This is where things
like latent means, their projections, and handcrafted features
found in feature_dirs are saved. Defaults to None.
plots_dir : str, optional
Directory to save plots. Defaults to '' (current working directory).
feature_dirs : list of {str, None}, optional
Directory containing text files with different syllable features.
For exmaple, this could contain exported MUPET, DeepSqueak or SAP
syllable tables. Defaults to None.
template_dir : {str, None}, optional
Directory continaing audio files of song templates. Defaults to
None.
Methods
-------
request(field)
Request some type of data.
Notes
-----
Supported directory structure:
::
โโโ animal_1
โ โโโ audio (raw audio)
โ โ โโโ foo.wav
โ โ โโโ bar.wav
โ โ โโโ baz.wav
โ โโโ features (output of MUPET, DeepSqueak, SAP, ...)
โ โ โโโ foo.csv
โ โ โโโ bar.csv
โ โ โโโ baz.csv
โ โโโ spectrograms (used to train models, written by
โ โ โโโ syllables_000.hdf5 preprocessing.preprocess.process_sylls)
โ โ โโโ syllables_001.hdf5
โ โโโ projections (latent means, UMAP, PCA, tSNE
โ โโโ syllables_000.hdf5 projections, copies of features in
โ โโโ syllables_001.hdf5 experiment_1/features. These are
โ written by a DataContainer object.)
โโโ animal_2
โ โโโ audio
โ โ โโโ 1.wav
โ โ โโโ 2.wav
โ โโโ features
โ โ โโโ 1.csv
โ โ โโโ 2.csv
โ โโโ spectrograms
โ โ โโโ syllables_000.hdf5
โ โ โโโ syllables_001.hdf5
โ โโโ projections
โ โโโ syllables_000.hdf5
โ โโโ syllables_001.hdf5
.
.
.
There should be a 1-to-1 correspondence between, for example, the syllables
in `animal_1/audio/baz.wav` and the features described in
`animal_1/features/baz.csv`. Analogously, the fifth entry in
`animal_2/spectrograms/syllables_000.hdf5` should describe the same syllable
as the fifth entry in `animal_2/projections/syllables_000.hdf5`. There is no
strict relationship, however, between individual files in `animal_1/audio`
and `animal_1/spectrograms`. The hdf5 files in the spectrograms and
projections directories should contain a subset of the syllables in the
audio and features directories.
Then a DataContainer object can be initialized as:
>>> from ava.data.data_container import DataContainer
>>> audio_dirs = ['animal_1/audio', 'animal_2/audio']
>>> spec_dirs = ['animal_1/spectrograms', 'animal_2/spectrograms']
>>> model_filename = 'checkpoint.tar'
>>> dc = DataContainer(audio_dirs=audio_dirs, spec_dirs=spec_dirs, \
model_filename=model_filename)
>>> latent_means = dc.request('latent_means')
It's fine to leave some of the initialization parameters unspecified. If the
DataContainer object is asked to do something it can't, it will hopefully
complain politely. Or at least informatively.
"""
def __init__(self, audio_dirs=None, segment_dirs=None, spec_dirs=None, \
feature_dirs=None, projection_dirs=None, plots_dir='', \
model_filename=None, template_dir=None, verbose=True):
self.audio_dirs = audio_dirs
self.segment_dirs = segment_dirs
self.spec_dirs = spec_dirs
self.feature_dirs = feature_dirs
self.projection_dirs = projection_dirs
self.plots_dir = plots_dir
self.model_filename = model_filename
self.template_dir = template_dir
self.verbose = verbose
self.sylls_per_file = None # syllables in each hdf5 file in spec_dirs
self.fields = self._check_for_fields()
if self.plots_dir not in [None, ''] and not os.path.exists(self.plots_dir):
os.makedirs(self.plots_dir)
def request(self, field):
"""
Request some type of data.
Parameters
----------
field : str
The type of data being requested. Should come from ...
Raises
------
`NotImplementedError`
when `field` is not recognized.
Note
----
Besides `__init__` and `clear_projections`, this should be the only
external-facing method.
"""
if field not in ALL_FIELDS:
print(str(field) + " is not a valid field!")
raise NotImplementedError
# If it's not here, make it and return it.
if field not in self.fields:
if self.verbose:
print("Making field:", field)
data = self._make_field(field)
# Otherwise, read it and return it.
else:
if self.verbose:
print("Reading field:", field)
data = self._read_field(field)
if self.verbose:
print("\tDone with:", field)
return data
def clear_projections(self):
"""Remove all projections."""
for proj_dir in self.projection_dirs:
if not os.path.exists(proj_dir):
continue
fns = [os.path.join(proj_dir, i) for i in os.listdir(proj_dir)]
fns = [i for i in fns if len(i) > 5 and i[-5:] == '.hdf5']
for fn in fns:
os.remove(fn)
self.fields = self._check_for_fields()
def clean_projections(self):
"""
Remove all projections.
Note
----
* This will be deprecated in 0.3.0. Use ``clear_projections`` instead.
"""
warnings.warn(
"clean_projections will be deprecated in v0.3.0. " + \
"Use clear_projections instead.",
UserWarning
)
self.clear_projections()
def _make_field(self, field):
"""Make a field."""
if field == 'latent_means':
data = self._make_latent_means()
elif field == 'latent_mean_pca':
data = self._make_latent_mean_pca_projection()
elif field == 'latent_mean_umap':
data = self._make_latent_mean_umap_projection()
elif field in MUPET_FIELDS:
data = self._make_feature_field(field, kind='mupet')
elif field in DEEPSQUEAK_FIELDS:
data = self._make_feature_field(field, kind='deepsqueak')
elif field in SAP_FIELDS:
data = self._make_feature_field(field, kind='sap')
elif field in FILENAME_FIELDS:
data = self._read_filename_field(field)
elif field == 'specs':
raise NotImplementedError
else:
raise NotImplementedError
# Add this field to the collection of fields that have been computed.
self.fields[field] = 1
if self.verbose:
print("Making field:", field)
return data
def _read_field(self, field):
"""
Read a field from memory.
Parameters
----------
field : str
Field name to read from file. See ``ALL_FIELDS`` for possible
fields.
"""
if field in AUDIO_FIELDS:
raise NotImplementedError
elif field == 'segments':
return self._read_segments()
elif field == 'segment_audio':
return self._read_segment_audio()
elif field in PROJECTION_FIELDS:
load_dirs = self.projection_dirs
elif field in SPEC_FIELDS:
load_dirs = self.spec_dirs
elif field in MUPET_FIELDS:
load_dirs = self.projection_dirs
elif field in DEEPSQUEAK_FIELDS:
load_dirs = self.projection_dirs
elif field in SAP_FIELDS:
load_dirs = self.projection_dirs
else:
raise Exception("Can\'t read field: "+field+"\n This should have \
been caught in self.request!")
to_return = []
for i in range(len(self.spec_dirs)):
spec_dir, load_dir = self.spec_dirs[i], load_dirs[i]
hdf5s = get_hdf5s_from_dir(spec_dir)
for j, hdf5 in enumerate(hdf5s):
filename = os.path.join(load_dir, os.path.split(hdf5)[-1])
with h5py.File(filename, 'r') as f:
assert field in f, "Can\'t find field \'"+field+"\' in"+ \
" file \'"+filename+"\'!"
if field == 'audio_filenames':
data = np.array([k.decode('UTF-8') for k in f[field]])
to_return.append(data)
else:
to_return.append(np.array(f[field]))
return np.concatenate(to_return)
def _read_segment_audio(self):
"""
Read all the segmented audio and return it.
result[audio_dir][audio_filename] = [audio_1, audio_2, ..., audio_n]
"""
self._check_for_dirs(['audio_dirs'], 'audio')
segments = self.request('segments')
result = {}
for audio_dir in self.audio_dirs:
dir_result = {}
audio_fns = [i for i in os.listdir(audio_dir) if _is_wav_file(i) \
and i in segments[audio_dir]]
for audio_fn in audio_fns:
fs, audio = wavfile.read(os.path.join(audio_dir, audio_fn))
fn_result = []
for seg in segments[audio_dir][audio_fn]:
i1 = int(round(seg[0]*fs))
i2 = int(round(seg[1]*fs))
fn_result.append(audio[i1:i2])
dir_result[audio_fn] = fn_result
result[audio_dir] = dir_result
return result
def _read_segments(self):
"""
Return all the segmenting decisions.
Return a dictionary mapping audio directories to audio filenames to
numpy arrays of shape [num_segments,2] containing onset and offset
times.
TO DO: add support for other delimiters, file extstensions, etc.
Returns
-------
segments : dict
Maps audio directories to audio filenames to numpy arrays.
"""
self._check_for_dirs(['audio_dirs', 'segment_dirs'], 'segments')
result = {}
for audio_dir, seg_dir in zip(self.audio_dirs, self.segment_dirs):
dir_result = {}
seg_fns = [os.path.join(seg_dir, i) for i in os.listdir(seg_dir) \
if _is_seg_file(i)]
audio_fns = [os.path.split(i)[1][:-4]+'.wav' for i in seg_fns]
for audio_fn, seg_fn in zip(audio_fns, seg_fns):
segs = _read_columns(seg_fn, delimiter='\t', unpack=False, \
skiprows=0)
if len(segs) > 0:
dir_result[audio_fn] = segs
result[audio_dir] = dir_result
return result
def _make_latent_means(self):
"""
Write latent means for the syllables in self.spec_dirs.
Returns
-------
latent_means : numpy.ndarray
Latent means of shape (max_num_syllables, z_dim)
Note
----
* Duplicated code with ``_write_projection``?
"""
self._check_for_dirs(['projection_dirs', 'spec_dirs', 'model_filename'],\
'latent_means')
# First, see how many syllables are in each file.
temp = get_hdf5s_from_dir(self.spec_dirs[0])
assert len(temp) > 0, "Found no specs in" + self.spec_dirs[0]
hdf5_file = temp[0]
with h5py.File(hdf5_file, 'r') as f:
self.sylls_per_file = len(f['specs'])
spf = self.sylls_per_file
# Load the model, making sure to get z_dim correct.
map_loc = 'cuda' if torch.cuda.is_available() else 'cpu'
z_dim = torch.load(self.model_filename, map_location=map_loc)['z_dim']
model = VAE(z_dim=z_dim)
model.load_state(self.model_filename)
# For each directory...
all_latent = []
for i in range(len(self.spec_dirs)):
spec_dir, proj_dir = self.spec_dirs[i], self.projection_dirs[i]
# Make the projection directory if it doesn't exist.
if proj_dir != '' and not os.path.exists(proj_dir):
os.makedirs(proj_dir)
# Make a DataLoader for the syllables.
partition = get_syllable_partition([spec_dir], 1, shuffle=False)
try:
loader = get_syllable_data_loaders(partition, \
shuffle=(False,False))['train']
# Get the latent means from the model.
latent_means = model.get_latent(loader)
all_latent.append(latent_means)
# Write them to the corresponding projection directory.
hdf5s = get_hdf5s_from_dir(spec_dir)
assert len(latent_means) // len(hdf5s) == spf
for j in range(len(hdf5s)):
filename = os.path.join(proj_dir, os.path.split(hdf5s[j])[-1])
data = latent_means[j*spf:(j+1)*spf]
with h5py.File(filename, 'a') as f:
f.create_dataset('latent_means', data=data)
except AssertionError: # No specs in this directory
pass
return np.concatenate(all_latent)
def _read_filename_field(self, field):
if field == 'sap_time':
data = self._make_sap_time()
else:
raise NotImplementedError
return data
def _make_sap_time(self):
"""Return time in seconds, following SAP conventions."""
onsets = self.request('syllable_start')
fns = self.request('audio_filenames')
result = np.zeros(lemn(onsets))
for i, onset, fn in zip(range(len(onsets)), onsets, fns):
# December 29, 1899, 7pm is the SAP anchor time.
anchor = mktime(strptime("1899 12 29 19", "%Y %m %d %H"))
temp = os.path.split(fn)[-1].split('_')[1].split('.')
day = float(temp[0])
millisecond = float(temp[1])
time = anchor + 24*60*60*day + 1e-3*millisecond
result[i] = time + onset
return result
def _make_latent_mean_umap_projection(self):
"""Project latent means to two dimensions with UMAP."""
# Get latent means.
latent_means = self.request('latent_means')
# UMAP them.
transform = umap.UMAP(n_components=2, n_neighbors=20, min_dist=0.1, \
metric='euclidean', random_state=42)
if self.verbose:
print("Running UMAP... (n="+str(len(latent_means))+")")
# https://github.com/lmcinnes/umap/issues/252
with warnings.catch_warnings():
try:
warnings.filterwarnings("ignore", \
category=NumbaPerformanceWarning)
except NameError:
pass
embedding = transform.fit_transform(latent_means)
if self.verbose:
print("Done.")
# Write to files.
self._write_projection("latent_mean_umap", embedding)
return embedding
def _make_latent_mean_pca_projection(self):
"""Project latent means to two dimensions with PCA."""
# Get latent means.
latent_means = self.request('latent_means')
# UMAP them.
transform = PCA(n_components=2, copy=False, random_state=42)
if self.verbose:
print("Running PCA...")
embedding = transform.fit_transform(latent_means)
if self.verbose:
print("Done.")
# Write to files.
self._write_projection("latent_mean_pca", embedding)
return embedding
def _make_feature_field(self, field, kind):
"""
Read a feature from a text file and put it in an hdf5 file.
Read from self.feature_dirs and write to self.projection_dirs. This
could be a bit tricky because we need to match up the syllables in the
text file with the ones in the hdf5 file.
Parameters
----------
field : str
Name of data being requested. See ``ALL_FIELDS`` for a complete
list.
kind : str, 'mupet' or 'deepsqueak'
Is this a MUPET or a DeepSqueak field?
Returns
-------
data : numpy.ndarray
Requested data.
"""
self._check_for_dirs( \
['spec_dirs', 'feature_dirs', 'projection_dirs'], field)
# FInd which column the field is stored in.
if kind == 'mupet':
file_fields = MUPET_FIELDS
onset_col = MUPET_ONSET_COL
elif kind == 'deepsqueak':
file_fields = DEEPSQUEAK_FIELDS
onset_col = DEEPSQUEAK_ONSET_COL
elif kind == 'sap':
file_fields = SAP_FIELDS
onset_col = SAP_ONSET_COL
else:
assert NotImplementedError
field_col = file_fields.index(field)
to_return = []
# Run through each directory.
for i in range(len(self.spec_dirs)):
spec_dir = self.spec_dirs[i]
feature_dir = self.feature_dirs[i]
proj_dir = self.projection_dirs[i]
hdf5s = get_hdf5s_from_dir(spec_dir)
current_fn, k = None, None
for hdf5 in hdf5s:
# Get the filenames and onsets from self.spec_dirs.
with h5py.File(hdf5, 'r') as f:
audio_filenames = np.array(f['audio_filenames'])
spec_onsets = np.array(f['onsets'])
# if kind == 'sap': # SAP writes onsets in milliseconds.
# spec_onsets /= 1e3
feature_arr = np.zeros(len(spec_onsets))
# Loop through each syllable.
for j in range(len(spec_onsets)):
audio_fn, spec_onset = audio_filenames[j], spec_onsets[j]
audio_fn = audio_fn.decode('UTF-8')
# Update the feature file, if needed.
if audio_fn != current_fn:
current_fn = audio_fn
feature_fn = os.path.split(audio_fn)[-1][:-4]
if kind == 'deepsqueak': # DeepSqueak appends '_Stats'
feature_fn += '_Stats' # when exporting features.
feature_fn += '.csv'
feature_fn = os.path.join(feature_dir, feature_fn)
# Read the onsets and features.
feature_onsets, features = \
_read_columns(feature_fn, [onset_col, field_col])
if kind == 'sap': # SAP writes onsets in milliseconds.
feature_onsets /= 1e3
k = 0
# Look for the corresponding onset in the feature file.
while spec_onset > feature_onsets[k] + 0.01:
k += 1
assert k < len(feature_onsets)
if abs(spec_onset - feature_onsets[k]) > 0.01:
print("Mismatch between spec_dirs and feature_dirs!")
print("hdf5 file:", hdf5)
print("\tindex:", j)
print("audio filename:", audio_fn)
print("feature filename:", feature_fn)
print("Didn't find spec_onset", spec_onset)
print("in feature onsets of min:", \
np.min(feature_onsets), "max:", \
np.max(feature_onsets))
print("field:", field)
print("kind:", kind)
quit()
# And add it to the feature array.
feature_arr[j] = features[k]
# Write the fields to self.projection_dirs.
write_fn = os.path.join(proj_dir, os.path.split(hdf5)[-1])
with h5py.File(write_fn, 'a') as f:
f.create_dataset(field, data=feature_arr)
to_return.append(feature_arr)
self.fields[field] = 1
return np.concatenate(to_return)
def _write_projection(self, key, data):
"""Write the given projection to self.projection_dirs."""
sylls_per_file = self.sylls_per_file
# For each directory...
k = 0
for i in range(len(self.projection_dirs)):
spec_dir, proj_dir = self.spec_dirs[i], self.projection_dirs[i]
hdf5s = get_hdf5s_from_dir(spec_dir)
for j in range(len(hdf5s)):
filename = os.path.join(proj_dir, os.path.split(hdf5s[j])[-1])
to_write = data[k:k+sylls_per_file]
with h5py.File(filename, 'a') as f:
f.create_dataset(key, data=to_write)
k += sylls_per_file
def _check_for_fields(self):
"""Check to see which fields are saved."""
fields = {}
# If self.spec_dirs is registered, assume everything is there.
if self.spec_dirs is not None:
for field in SPEC_FIELDS:
fields[field] = 1
# Same for self.audio_dirs.
if self.audio_dirs is not None:
fields['audio'] = 1
# Same for self.segment_dirs.
if self.segment_dirs is not None:
fields['segments'] = 1
fields['segment_audio'] = 1
# If self.projection_dirs is registered, see what we have.
# If it's in one file, assume it's in all of them.
if self.projection_dirs is not None:
if os.path.exists(self.projection_dirs[0]):
hdf5s = get_hdf5s_from_dir(self.projection_dirs[0])
if len(hdf5s) > 0:
hdf5 = hdf5s[0]
if os.path.exists(hdf5):
with h5py.File(hdf5, 'r') as f:
for key in f.keys():
if key in ALL_FIELDS:
fields[key] = 1
self.sylls_per_file = len(f[key])
return fields
def _check_for_dirs(self, dir_names, field):
"""Check that the given directories exist."""
for dir_name in dir_names:
if dir_name == 'audio_dirs':
temp = self.audio_dirs
elif dir_name == 'segment_dirs':
temp = self.segment_dirs
elif dir_name == 'spec_dirs':
temp = self.spec_dirs
elif dir_name == 'feature_dirs':
temp = self.feature_dirs
elif dir_name == 'projection_dirs':
temp = self.projection_dirs
elif dir_name == 'model_filename':
temp = self.model_filename
else:
raise NotImplementedError
assert temp is not None, dir_name + " must be specified before " + \
field + " is made!"
def _read_columns(filename, columns=(0,1), delimiter=',', skiprows=1, \
unpack=True):
"""
A wrapper around numpy.loadtxt to handle empty files.
TO DO: Add categorical variables.
"""
data = np.loadtxt(filename, delimiter=delimiter, usecols=columns, \
skiprows=skiprows).reshape(-1,len(columns))
if unpack:
return tuple(data[:,i] for i in range(data.shape[1]))
return data
def _is_seg_file(filename):
"""Is this a segmenting file?"""
return len(filename) > 4 and filename[-4:] == '.txt'
def _is_wav_file(filename):
"""Is this a wav file?"""
return len(filename) > 4 and filename[-4:] == '.wav'
if __name__ == '__main__':
pass
###
| [
"os.remove",
"ava.models.vae.VAE",
"os.path.join",
"ava.models.vae_dataset.get_syllable_partition",
"torch.load",
"os.path.exists",
"numpy.max",
"warnings.catch_warnings",
"numpy.loadtxt",
"h5py.File",
"ava.models.vae_dataset.get_hdf5s_from_dir",
"umap.UMAP",
"numpy.min",
"torch.cuda.is_av... | [((10192, 10309), 'warnings.warn', 'warnings.warn', (["('clean_projections will be deprecated in v0.3.0. ' +\n 'Use clear_projections instead.')", 'UserWarning'], {}), "('clean_projections will be deprecated in v0.3.0. ' +\n 'Use clear_projections instead.', UserWarning)\n", (10205, 10309), False, 'import warnings\n'), ((12632, 12657), 'numpy.concatenate', 'np.concatenate', (['to_return'], {}), '(to_return)\n', (12646, 12657), True, 'import numpy as np\n'), ((14819, 14856), 'ava.models.vae_dataset.get_hdf5s_from_dir', 'get_hdf5s_from_dir', (['self.spec_dirs[0]'], {}), '(self.spec_dirs[0])\n', (14837, 14856), False, 'from ava.models.vae_dataset import get_syllable_partition, get_syllable_data_loaders, get_hdf5s_from_dir\n'), ((15247, 15263), 'ava.models.vae.VAE', 'VAE', ([], {'z_dim': 'z_dim'}), '(z_dim=z_dim)\n', (15250, 15263), False, 'from ava.models.vae import VAE\n'), ((16378, 16404), 'numpy.concatenate', 'np.concatenate', (['all_latent'], {}), '(all_latent)\n', (16392, 16404), True, 'import numpy as np\n'), ((17344, 17440), 'umap.UMAP', 'umap.UMAP', ([], {'n_components': '(2)', 'n_neighbors': '(20)', 'min_dist': '(0.1)', 'metric': '"""euclidean"""', 'random_state': '(42)'}), "(n_components=2, n_neighbors=20, min_dist=0.1, metric='euclidean',\n random_state=42)\n", (17353, 17440), False, 'import umap\n'), ((18106, 18154), 'sklearn.decomposition.PCA', 'PCA', ([], {'n_components': '(2)', 'copy': '(False)', 'random_state': '(42)'}), '(n_components=2, copy=False, random_state=42)\n', (18109, 18154), False, 'from sklearn.decomposition import PCA\n'), ((21794, 21819), 'numpy.concatenate', 'np.concatenate', (['to_return'], {}), '(to_return)\n', (21808, 21819), True, 'import numpy as np\n'), ((8816, 8843), 'os.makedirs', 'os.makedirs', (['self.plots_dir'], {}), '(self.plots_dir)\n', (8827, 8843), False, 'import os\n'), ((12179, 12207), 'ava.models.vae_dataset.get_hdf5s_from_dir', 'get_hdf5s_from_dir', (['spec_dir'], {}), '(spec_dir)\n', (12197, 12207), False, 'from ava.models.vae_dataset import get_syllable_partition, get_syllable_data_loaders, get_hdf5s_from_dir\n'), ((14950, 14975), 'h5py.File', 'h5py.File', (['hdf5_file', '"""r"""'], {}), "(hdf5_file, 'r')\n", (14959, 14975), False, 'import h5py\n'), ((15127, 15152), 'torch.cuda.is_available', 'torch.cuda.is_available', ([], {}), '()\n', (15150, 15152), False, 'import torch\n'), ((15174, 15227), 'torch.load', 'torch.load', (['self.model_filename'], {'map_location': 'map_loc'}), '(self.model_filename, map_location=map_loc)\n', (15184, 15227), False, 'import torch\n'), ((15648, 15700), 'ava.models.vae_dataset.get_syllable_partition', 'get_syllable_partition', (['[spec_dir]', '(1)'], {'shuffle': '(False)'}), '([spec_dir], 1, shuffle=False)\n', (15670, 15700), False, 'from ava.models.vae_dataset import get_syllable_partition, get_syllable_data_loaders, get_hdf5s_from_dir\n'), ((17575, 17600), 'warnings.catch_warnings', 'warnings.catch_warnings', ([], {}), '()\n', (17598, 17600), False, 'import warnings\n'), ((19631, 19659), 'ava.models.vae_dataset.get_hdf5s_from_dir', 'get_hdf5s_from_dir', (['spec_dir'], {}), '(spec_dir)\n', (19649, 19659), False, 'from ava.models.vae_dataset import get_syllable_partition, get_syllable_data_loaders, get_hdf5s_from_dir\n'), ((22119, 22147), 'ava.models.vae_dataset.get_hdf5s_from_dir', 'get_hdf5s_from_dir', (['spec_dir'], {}), '(spec_dir)\n', (22137, 22147), False, 'from ava.models.vae_dataset import get_syllable_partition, get_syllable_data_loaders, get_hdf5s_from_dir\n'), ((23003, 23042), 'os.path.exists', 'os.path.exists', (['self.projection_dirs[0]'], {}), '(self.projection_dirs[0])\n', (23017, 23042), False, 'import os\n'), ((24201, 24278), 'numpy.loadtxt', 'np.loadtxt', (['filename'], {'delimiter': 'delimiter', 'usecols': 'columns', 'skiprows': 'skiprows'}), '(filename, delimiter=delimiter, usecols=columns, skiprows=skiprows)\n', (24211, 24278), True, 'import numpy as np\n'), ((8781, 8811), 'os.path.exists', 'os.path.exists', (['self.plots_dir'], {}), '(self.plots_dir)\n', (8795, 8811), False, 'import os\n'), ((9787, 9811), 'os.path.exists', 'os.path.exists', (['proj_dir'], {}), '(proj_dir)\n', (9801, 9811), False, 'import os\n'), ((9836, 9861), 'os.path.join', 'os.path.join', (['proj_dir', 'i'], {}), '(proj_dir, i)\n', (9848, 9861), False, 'import os\n'), ((9977, 9990), 'os.remove', 'os.remove', (['fn'], {}), '(fn)\n', (9986, 9990), False, 'import os\n'), ((14008, 14032), 'os.path.join', 'os.path.join', (['seg_dir', 'i'], {}), '(seg_dir, i)\n', (14020, 14032), False, 'import os\n'), ((15569, 15590), 'os.makedirs', 'os.makedirs', (['proj_dir'], {}), '(proj_dir)\n', (15580, 15590), False, 'import os\n'), ((15993, 16021), 'ava.models.vae_dataset.get_hdf5s_from_dir', 'get_hdf5s_from_dir', (['spec_dir'], {}), '(spec_dir)\n', (16011, 16021), False, 'from ava.models.vae_dataset import get_syllable_partition, get_syllable_data_loaders, get_hdf5s_from_dir\n'), ((16891, 16931), 'time.strptime', 'strptime', (['"""1899 12 29 19"""', '"""%Y %m %d %H"""'], {}), "('1899 12 29 19', '%Y %m %d %H')\n", (16899, 16931), False, 'from time import strptime, mktime\n'), ((17614, 17681), 'warnings.filterwarnings', 'warnings.filterwarnings', (['"""ignore"""'], {'category': 'NumbaPerformanceWarning'}), "('ignore', category=NumbaPerformanceWarning)\n", (17637, 17681), False, 'import warnings\n'), ((23056, 23099), 'ava.models.vae_dataset.get_hdf5s_from_dir', 'get_hdf5s_from_dir', (['self.projection_dirs[0]'], {}), '(self.projection_dirs[0])\n', (23074, 23099), False, 'from ava.models.vae_dataset import get_syllable_partition, get_syllable_data_loaders, get_hdf5s_from_dir\n'), ((9871, 9891), 'os.listdir', 'os.listdir', (['proj_dir'], {}), '(proj_dir)\n', (9881, 9891), False, 'import os\n'), ((12316, 12340), 'h5py.File', 'h5py.File', (['filename', '"""r"""'], {}), "(filename, 'r')\n", (12325, 12340), False, 'import h5py\n'), ((13004, 13025), 'os.listdir', 'os.listdir', (['audio_dir'], {}), '(audio_dir)\n', (13014, 13025), False, 'import os\n'), ((13140, 13173), 'os.path.join', 'os.path.join', (['audio_dir', 'audio_fn'], {}), '(audio_dir, audio_fn)\n', (13152, 13173), False, 'import os\n'), ((14042, 14061), 'os.listdir', 'os.listdir', (['seg_dir'], {}), '(seg_dir)\n', (14052, 14061), False, 'import os\n'), ((15539, 15563), 'os.path.exists', 'os.path.exists', (['proj_dir'], {}), '(proj_dir)\n', (15553, 15563), False, 'import os\n'), ((15722, 15782), 'ava.models.vae_dataset.get_syllable_data_loaders', 'get_syllable_data_loaders', (['partition'], {'shuffle': '(False, False)'}), '(partition, shuffle=(False, False))\n', (15747, 15782), False, 'from ava.models.vae_dataset import get_syllable_partition, get_syllable_data_loaders, get_hdf5s_from_dir\n'), ((19777, 19797), 'h5py.File', 'h5py.File', (['hdf5', '"""r"""'], {}), "(hdf5, 'r')\n", (19786, 19797), False, 'import h5py\n'), ((19827, 19857), 'numpy.array', 'np.array', (["f['audio_filenames']"], {}), "(f['audio_filenames'])\n", (19835, 19857), True, 'import numpy as np\n'), ((19877, 19898), 'numpy.array', 'np.array', (["f['onsets']"], {}), "(f['onsets'])\n", (19885, 19898), True, 'import numpy as np\n'), ((21648, 21672), 'h5py.File', 'h5py.File', (['write_fn', '"""a"""'], {}), "(write_fn, 'a')\n", (21657, 21672), False, 'import h5py\n'), ((22295, 22319), 'h5py.File', 'h5py.File', (['filename', '"""a"""'], {}), "(filename, 'a')\n", (22304, 22319), False, 'import h5py\n'), ((23152, 23172), 'os.path.exists', 'os.path.exists', (['hdf5'], {}), '(hdf5)\n', (23166, 23172), False, 'import os\n'), ((12282, 12301), 'os.path.split', 'os.path.split', (['hdf5'], {}), '(hdf5)\n', (12295, 12301), False, 'import os\n'), ((16224, 16248), 'h5py.File', 'h5py.File', (['filename', '"""a"""'], {}), "(filename, 'a')\n", (16233, 16248), False, 'import h5py\n'), ((20530, 20567), 'os.path.join', 'os.path.join', (['feature_dir', 'feature_fn'], {}), '(feature_dir, feature_fn)\n', (20542, 20567), False, 'import os\n'), ((21614, 21633), 'os.path.split', 'os.path.split', (['hdf5'], {}), '(hdf5)\n', (21627, 21633), False, 'import os\n'), ((22217, 22240), 'os.path.split', 'os.path.split', (['hdf5s[j]'], {}), '(hdf5s[j])\n', (22230, 22240), False, 'import os\n'), ((12603, 12621), 'numpy.array', 'np.array', (['f[field]'], {}), '(f[field])\n', (12611, 12621), True, 'import numpy as np\n'), ((14104, 14120), 'os.path.split', 'os.path.split', (['i'], {}), '(i)\n', (14117, 14120), False, 'import os\n'), ((16143, 16166), 'os.path.split', 'os.path.split', (['hdf5s[j]'], {}), '(hdf5s[j])\n', (16156, 16166), False, 'import os\n'), ((21319, 21341), 'numpy.min', 'np.min', (['feature_onsets'], {}), '(feature_onsets)\n', (21325, 21341), True, 'import numpy as np\n'), ((21361, 21383), 'numpy.max', 'np.max', (['feature_onsets'], {}), '(feature_onsets)\n', (21367, 21383), True, 'import numpy as np\n'), ((23185, 23205), 'h5py.File', 'h5py.File', (['hdf5', '"""r"""'], {}), "(hdf5, 'r')\n", (23194, 23205), False, 'import h5py\n'), ((20331, 20354), 'os.path.split', 'os.path.split', (['audio_fn'], {}), '(audio_fn)\n', (20344, 20354), False, 'import os\n'), ((16943, 16960), 'os.path.split', 'os.path.split', (['fn'], {}), '(fn)\n', (16956, 16960), False, 'import os\n')] |
import os
import sys
import logging as log
import numpy as np
import cv2
from openvino.inference_engine import IENetwork, IECore
class ModelDetection:
'''
Class for the Face Detection Model.
'''
def __init__(self, model_name, device='CPU', extensions=None, threshold=0.5):
self.threshold = threshold
self.model_name = model_name
self.device = device
self.extensions = extensions
def load_model(self):
## Get model_bin and model_xml
model_bin = self.model_name + ".bin"
model_xml = self.model_name + ".xml"
plugin = IECore()
network = IENetwork(model=model_xml, weights=model_bin)
## Add extension if any
if self.extensions and "CPU" in self.device: # Add a CPU extension, if applicable
plugin.add_extension(self.extensions, self.device)
## (Additional) Check unsupported layer
supported_layers = plugin.query_network(network=network, device_name=self.device)
unsupported_layers = [l for l in network.layers.keys() if l not in supported_layers]
if len(unsupported_layers) > 2:
print("Unsupported layers found: {}".format(unsupported_layers))
print("Check whether extensions are available to add to IECore.")
exit(1)
## Load network
self.exec_network = plugin.load_network(network, self.device)
self.input_blob = next(iter(network.inputs))
self.output_blob = next(iter(network.outputs))
self.n, self.c, self.h, self.w = network.inputs[self.input_blob].shape
self.plugin = plugin
self.network = network
def predict(self, image):
frame_shape = image.shape
image = self.preprocess_input(image)
self.exec_network.requests[0].infer({self.input_blob: image})
outputs = self.exec_network.requests[0].outputs[self.output_blob]
self.outputs = outputs
boxes, scores = self.preprocess_output(outputs, frame_shape)
return boxes, scores
def check_model(self):
raise NotImplementedError
def preprocess_input(self, image):
img = cv2.dnn.blobFromImage(image, size=(self.w, self.h))
return img
def preprocess_output(self, outputs, frame_shape):
img_h, img_w, _ = frame_shape
boxes = []
scores = []
res = outputs
people = res[0][:, np.where((res[0][0][:, 2] > self.threshold))]
for person in people[0][0]:
box = person[3:7] * np.array([img_w, img_h, img_w, img_h])
box = int(box[0]), int(box[1]), int(box[2]), int(box[3])
boxes.append(box)
scores.append(person[2])
return boxes, scores
| [
"openvino.inference_engine.IENetwork",
"openvino.inference_engine.IECore",
"cv2.dnn.blobFromImage",
"numpy.where",
"numpy.array"
] | [((601, 609), 'openvino.inference_engine.IECore', 'IECore', ([], {}), '()\n', (607, 609), False, 'from openvino.inference_engine import IENetwork, IECore\n'), ((628, 673), 'openvino.inference_engine.IENetwork', 'IENetwork', ([], {'model': 'model_xml', 'weights': 'model_bin'}), '(model=model_xml, weights=model_bin)\n', (637, 673), False, 'from openvino.inference_engine import IENetwork, IECore\n'), ((2218, 2269), 'cv2.dnn.blobFromImage', 'cv2.dnn.blobFromImage', (['image'], {'size': '(self.w, self.h)'}), '(image, size=(self.w, self.h))\n', (2239, 2269), False, 'import cv2\n'), ((2471, 2513), 'numpy.where', 'np.where', (['(res[0][0][:, 2] > self.threshold)'], {}), '(res[0][0][:, 2] > self.threshold)\n', (2479, 2513), True, 'import numpy as np\n'), ((2585, 2623), 'numpy.array', 'np.array', (['[img_w, img_h, img_w, img_h]'], {}), '([img_w, img_h, img_w, img_h])\n', (2593, 2623), True, 'import numpy as np\n')] |
# Copyright 2017 Google Inc. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Utilities for running predictions.
Includes (from the Cloud ML SDK):
- _predict_lib
Important changes:
- Remove interfaces for TensorFlowModel (they don't change behavior).
- Set from_client(skip_preprocessing=True) and remove the pre-processing code.
"""
import __builtin__
import base64
import collections
from contextlib import contextmanager
import importlib
import inspect
import json
import logging
import os
import pickle
import pydoc # used for importing python classes from their FQN
import StringIO
import timeit
from _interfaces import Model
from _interfaces import PredictionClient
from enum import Enum
import numpy as np
from tensorflow.python.client import session as tf_session
from tensorflow.python.framework import dtypes
from tensorflow.python.saved_model import loader
from tensorflow.python.saved_model import signature_constants
from tensorflow.python.saved_model import tag_constants
# --------------------------
# prediction.prediction_lib
# --------------------------
class UserClassType(Enum):
model_class = "model_class"
processor_class = "processor_class"
ENGINE = "Prediction-Engine"
ENGINE_RUN_TIME = "Prediction-Engine-Run-Time"
FRAMEWORK = "Framework"
SCIKIT_LEARN_FRAMEWORK_NAME = "scikit_learn"
XGBOOST_FRAMEWORK_NAME = "xgboost"
TENSORFLOW_FRAMEWORK_NAME = "tensorflow"
PREPROCESS_TIME = "Prediction-Preprocess-Time"
POSTPROCESS_TIME = "Prediction-Postprocess-Time"
# Keys for the name of the methods that the user provided `Processor`
# class should implement.
PREPROCESS_KEY = "preprocess"
POSTPROCESS_KEY = "postprocess"
FROM_MODEL_KEY = "from_model_path"
# Additional TF keyword arguments
INPUTS_KEY = "inputs"
OUTPUTS_KEY = "outputs"
SIGNATURE_KEY = "signature_name"
# Stats
COLUMNARIZE_TIME = "Prediction-Columnarize-Time"
UNALIAS_TIME = "Prediction-Unalias-Time"
ENCODE_TIME = "Prediction-Encode-Time"
SESSION_RUN_TIME = "Prediction-Session-Run-Time"
ALIAS_TIME = "Prediction-Alias-Time"
ROWIFY_TIME = "Prediction-Rowify-Time"
# TODO(b/67586901): Consider removing INPUT_PROCESSING_TIME during cleanup.
SESSION_RUN_ENGINE_NAME = "TF_SESSION_RUN"
# Scikit-learn and XGBoost related constants
MODEL_FILE_NAME_JOBLIB = "model.joblib"
MODEL_FILE_NAME_PICKLE = "model.pkl"
MODEL_FILE_NAME_BST = "model.bst"
PredictionErrorType = collections.namedtuple(
"PredictionErrorType", ("message", "code"))
class PredictionError(Exception):
"""Customer exception for known prediction exception."""
# The error code for prediction.
FAILED_TO_LOAD_MODEL = PredictionErrorType(
message="Failed to load model", code=0)
INVALID_INPUTS = PredictionErrorType("Invalid inputs", code=1)
FAILED_TO_RUN_MODEL = PredictionErrorType(
message="Failed to run the provided model", code=2)
INVALID_OUTPUTS = PredictionErrorType(
message="There was a problem processing the outputs", code=3)
INVALID_USER_CODE = PredictionErrorType(
message="There was a problem processing the user code", code=4)
# When adding new exception, please update the ERROR_MESSAGE_ list as well as
# unittest.
def __init__(self, error_code, error_detail, *args):
super(PredictionError, self).__init__(error_code, error_detail, *args)
@property
def error_code(self):
return self.args[0].code
@property
def error_message(self):
return self.args[0].message
@property
def error_detail(self):
return self.args[1]
def __str__(self):
return ("%s: %s (Error code: %d)" % (self.error_message,
self.error_detail, self.error_code))
MICRO = 1000000
MILLI = 1000
class Timer(object):
"""Context manager for timing code blocks.
The object is intended to be used solely as a context manager and not
as a general purpose object.
The timer starts when __enter__ is invoked on the context manager
and stopped when __exit__ is invoked. After __exit__ is called,
the duration properties report the amount of time between
__enter__ and __exit__ and thus do not change. However, if any of the
duration properties are called between the call to __enter__ and __exit__,
then they will return the "live" value of the timer.
If the same Timer object is re-used in multiple with statements, the values
reported will reflect the latest call. Do not use the same Timer object in
nested with blocks with the same Timer context manager.
Example usage:
with Timer() as timer:
foo()
print(timer.duration_secs)
"""
def __init__(self, timer_fn=None):
self.start = None
self.end = None
self._get_time = timer_fn or timeit.default_timer
def __enter__(self):
self.end = None
self.start = self._get_time()
return self
def __exit__(self, exc_type, value, traceback):
self.end = self._get_time()
return False
@property
def seconds(self):
now = self._get_time()
return (self.end or now) - (self.start or now)
@property
def microseconds(self):
return int(MICRO * self.seconds)
@property
def milliseconds(self):
return int(MILLI * self.seconds)
class Stats(dict):
"""An object for tracking stats.
This class is dict-like, so stats are accessed/stored like so:
stats = Stats()
stats["count"] = 1
stats["foo"] = "bar"
This class also facilitates collecting timing information via the
context manager obtained using the "time" method. Reported timings
are in microseconds.
Example usage:
with stats.time("foo_time"):
foo()
print(stats["foo_time"])
"""
@contextmanager
def time(self, name, timer_fn=None):
with Timer(timer_fn) as timer:
yield timer
self[name] = timer.microseconds
def columnarize(instances):
"""Columnarize inputs.
Each line in the input is a dictionary of input names to the value
for that input (a single instance). For each input "column", this method
appends each of the input values to a list. The result is a dict mapping
input names to a batch of input data. This can be directly used as the
feed dict during prediction.
For example,
instances = [{"a": [1.0, 2.0], "b": "a"},
{"a": [3.0, 4.0], "b": "c"},
{"a": [5.0, 6.0], "b": "e"},]
batch = prediction_server_lib.columnarize(instances)
assert batch == {"a": [[1.0, 2.0], [3.0, 4.0], [5.0, 6.0]],
"b": ["a", "c", "e"]}
Arguments:
instances: (list of dict) where the dictionaries map input names
to the values for those inputs.
Returns:
A dictionary mapping input names to values, as described above.
"""
columns = collections.defaultdict(list)
for instance in instances:
for k, v in instance.iteritems():
columns[k].append(v)
return columns
def rowify(columns):
"""Converts columnar input to row data.
Consider the following code:
columns = {"prediction": np.array([1, # 1st instance
0, # 2nd
1]), # 3rd
"scores": np.array([[0.1, 0.9], # 1st instance
[0.7, 0.3], # 2nd
[0.4, 0.6]])} # 3rd
Then rowify will return the equivalent of:
[{"prediction": 1, "scores": [0.1, 0.9]},
{"prediction": 0, "scores": [0.7, 0.3]},
{"prediction": 1, "scores": [0.4, 0.6]}]
(each row is yielded; no list is actually created).
Arguments:
columns: (dict) mapping names to numpy arrays, where the arrays
contain a batch of data.
Raises:
PredictionError: if the outer dimension of each input isn't identical
for each of element.
Yields:
A map with a single instance, as described above. Note: instances
is not a numpy array.
"""
sizes_set = {e.shape[0] for e in columns.itervalues()}
# All the elements in the length array should be identical. Otherwise,
# raise an exception.
if len(sizes_set) != 1:
sizes_dict = {name: e.shape[0] for name, e in columns.iteritems()}
raise PredictionError(
PredictionError.INVALID_OUTPUTS,
"Bad output from running tensorflow session: outputs had differing "
"sizes in the batch (outer) dimension. See the outputs and their "
"size: %s. Check your model for bugs that effect the size of the "
"outputs." % sizes_dict)
# Pick an arbitrary value in the map to get it's size.
num_instances = len(next(columns.itervalues()))
for row in xrange(num_instances):
yield {name: output[row, ...].tolist()
for name, output in columns.iteritems()}
def canonicalize_single_tensor_input(instances, tensor_name):
"""Canonicalize single input tensor instances into list of dicts.
Instances that are single input tensors may or may not be provided with their
tensor name. The following are both valid instances:
1) instances = [{"x": "a"}, {"x": "b"}, {"x": "c"}]
2) instances = ["a", "b", "c"]
This function canonicalizes the input instances to be of type 1).
Arguments:
instances: single input tensor instances as supplied by the user to the
predict method.
tensor_name: the expected name of the single input tensor.
Raises:
PredictionError: if the wrong tensor name is supplied to instances.
Returns:
A list of dicts. Where each dict is a single instance, mapping the
tensor_name to the value (as supplied by the original instances).
"""
# Input is a single string tensor, the tensor name might or might not
# be given.
# There are 3 cases (assuming the tensor name is "t", tensor = "abc"):
# 1) {"t": "abc"}
# 2) "abc"
# 3) {"y": ...} --> wrong tensor name is given.
def parse_single_tensor(x, tensor_name):
if not isinstance(x, dict):
# case (2)
return {tensor_name: x}
elif len(x) == 1 and tensor_name == x.keys()[0]:
# case (1)
return x
else:
raise PredictionError(PredictionError.INVALID_INPUTS,
"Expected tensor name: %s, got tensor name: %s." %
(tensor_name, x.keys()))
if not isinstance(instances, list):
instances = [instances]
instances = [parse_single_tensor(x, tensor_name) for x in instances]
return instances
class BaseModel(Model):
"""The base definition of an internal Model interface.
"""
def __init__(self, client):
"""Constructs a BaseModel.
Args:
client: An instance of PredictionClient for performing prediction.
"""
self._client = client
self._user_processor = None
def preprocess(self, instances, stats=None, **kwargs):
"""Runs the preprocessing function on the instances.
Args:
instances: list of instances as provided to the predict() method.
stats: Stats object for recording timing information.
**kwargs: Additional keyword arguments for preprocessing.
Returns:
A new list of preprocessed instances. Each instance is as described
in the predict() method.
"""
pass
def postprocess(self, predicted_output, original_input=None, stats=None,
**kwargs):
"""Runs the postprocessing function on the instances.
Args:
predicted_output: list of instances returned by the predict() method on
preprocessed instances.
original_input: List of instances, before any pre-processing was applied.
stats: Stats object for recording timing information.
**kwargs: Additional keyword arguments for postprocessing.
Returns:
A new list of postprocessed instances.
"""
pass
def predict(self, instances, stats=None, **kwargs):
"""Runs preprocessing, predict, and postprocessing on the input."""
stats = stats or Stats()
self._validate_kwargs(kwargs)
with stats.time(PREPROCESS_TIME):
preprocessed = self.preprocess(instances, stats=stats, **kwargs)
with stats.time(ENGINE_RUN_TIME):
predicted_outputs = self._client.predict(
preprocessed, stats=stats, **kwargs)
with stats.time(POSTPROCESS_TIME):
postprocessed = self.postprocess(
predicted_outputs, original_input=instances, stats=stats, **kwargs)
return instances, postprocessed
def _validate_kwargs(self, kwargs):
"""Validates and sets defaults for extra predict keyword arguments.
Modifies the keyword args dictionary in-place. Keyword args will be included
into pre/post-processing and the client predict method.
Can raise Exception to error out of request on bad keyword args.
If no additional args are required, pass.
Args:
kwargs: Dictionary (str->str) of keyword arguments to check.
"""
pass
# TODO(b/34686738): when we no longer load the model to get the signature
# consider making this a named constructor on SessionClient.
def load_model(
model_path,
tags=(tag_constants.SERVING,),
config=None):
"""Loads the model at the specified path.
Args:
model_path: the path to either session_bundle or SavedModel
tags: the tags that determines the model to load.
config: tf.ConfigProto containing session configuration options.
Returns:
A pair of (Session, map<string, SignatureDef>) objects.
Raises:
PredictionError: if the model could not be loaded.
"""
if loader.maybe_saved_model_directory(model_path):
try:
logging.info("Importing tensorflow.contrib in load_model")
import tensorflow.contrib # pylint: disable=redefined-outer-name, unused-variable, g-import-not-at-top
session = tf_session.Session(target="", graph=None, config=config)
meta_graph = loader.load(session, tags=list(tags), export_dir=model_path)
except Exception as e: # pylint: disable=broad-except
raise PredictionError(PredictionError.FAILED_TO_LOAD_MODEL,
"Failed to load the model due to bad model data."
" tags: %s\n%s" % (list(tags), str(e)))
else:
raise PredictionError(PredictionError.FAILED_TO_LOAD_MODEL,
"Cloud ML only supports TF 1.0 or above and models "
"saved in SavedModel format.")
if session is None:
raise PredictionError(PredictionError.FAILED_TO_LOAD_MODEL,
"Failed to create session when loading the model")
if not meta_graph.signature_def:
raise PredictionError(PredictionError.FAILED_TO_LOAD_MODEL,
"MetaGraph must have at least one signature_def.")
# Remove invalid signatures from the signature map.
invalid_signatures = []
for signature_name in meta_graph.signature_def:
try:
signature = meta_graph.signature_def[signature_name]
_update_dtypes(session.graph, signature.inputs)
_update_dtypes(session.graph, signature.outputs)
except ValueError as e:
logging.warn("Error updating signature %s: %s", signature_name, str(e))
invalid_signatures.append(signature_name)
for signature_name in invalid_signatures:
del meta_graph.signature_def[signature_name]
return session, meta_graph.signature_def
def _update_dtypes(graph, interface):
"""Adds dtype to TensorInfos in interface if necessary.
If already present, validates TensorInfo matches values in the graph.
TensorInfo is updated in place.
Args:
graph: the TensorFlow graph; used to lookup datatypes of tensors.
interface: map from alias to TensorInfo object.
Raises:
ValueError: if the data type in the TensorInfo does not match the type
found in graph.
"""
for alias, info in interface.iteritems():
# Postpone conversion to enum for better error messages.
dtype = graph.get_tensor_by_name(info.name).dtype
if not info.dtype:
info.dtype = dtype.as_datatype_enum
elif info.dtype != dtype.as_datatype_enum:
raise ValueError("Specified data types do not match for alias %s. "
"Graph has %d while TensorInfo reports %d." %
(alias, dtype, info.dtype))
# (TODO:b/68775232): Move this to a Tensorflow specific library.
class TensorFlowClient(PredictionClient):
"""A client for Prediction that uses Session.run."""
def __init__(self, signature_map, *args, **kwargs):
self._signature_map = signature_map
super(TensorFlowClient, self).__init__(*args, **kwargs)
@property
def signature_map(self):
return self._signature_map
def get_signature(self, signature_name=None):
"""Gets tensorflow signature for the given signature_name.
Args:
signature_name: string The signature name to use to choose the signature
from the signature map.
Returns:
a pair of signature_name and signature. The first element is the
signature name in string that is actually used. The second one is the
signature.
Raises:
PredictionError: when the signature is not found with the given signature
name or when there are more than one signatures in the signature map.
"""
# The way to find signature is:
# 1) if signature_name is specified, try to find it in the signature_map. If
# not found, raise an exception.
# 2) if signature_name is not specified, check if signature_map only
# contains one entry. If so, return the only signature.
# 3) Otherwise, use the default signature_name and do 1).
if not signature_name and len(self.signature_map) == 1:
return self.signature_map.keys()[0], self.signature_map.values()[0]
key = (signature_name or
signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY)
if key in self.signature_map:
return key, self.signature_map[key]
else:
raise PredictionError(
PredictionError.INVALID_INPUTS,
"No signature found for signature key %s." % signature_name)
# (TODO:b/68775232): Move this to a Tensorflow specific library.
class SessionClient(TensorFlowClient):
"""A client for Prediction that uses Session.run."""
def __init__(self, session, signature_map):
self._session = session
super(SessionClient, self).__init__(signature_map)
def predict(self, inputs, stats=None,
signature_name=None, **unused_kwargs):
"""Produces predictions for the given inputs.
Args:
inputs: a dict mapping input names to values
stats: Stats object for recording timing information.
signature_name: name of SignatureDef to use in this prediction
**unused_kwargs: placeholder, pre/postprocess may have additional args
Returns:
A dict mapping output names to output values, similar to the input
dict.
"""
stats = stats or Stats()
stats[ENGINE] = "SessionRun"
stats[FRAMEWORK] = TENSORFLOW_FRAMEWORK_NAME
with stats.time(UNALIAS_TIME):
_, signature = self.get_signature(signature_name)
fetches = [output.name for output in signature.outputs.values()]
try:
unaliased = {signature.inputs[key].name: val
for key, val in inputs.iteritems()}
except Exception as e:
raise PredictionError(PredictionError.INVALID_INPUTS,
"Input mismatch: " + str(e))
with stats.time(SESSION_RUN_TIME):
try:
# TODO(b/33849399): measure the actual session.run() time, even in the
# case of ModelServer.
outputs = self._session.run(fetches=fetches, feed_dict=unaliased)
except Exception as e:
logging.error("Exception during running the graph: " + str(e))
raise PredictionError(PredictionError.FAILED_TO_RUN_MODEL,
"Exception during running the graph: " + str(e))
with stats.time(ALIAS_TIME):
return dict(zip(signature.outputs.iterkeys(), outputs))
def load_model_class(client, model_path):
"""Loads in the user specified custom Model class.
Args:
client: An instance of ModelServerClient for performing prediction.
model_path: the path to either session_bundle or SavedModel
Returns:
An instance of a Model.
Returns None if the user didn't specify the name of the custom
python class to load in the create_version_request.
Raises:
PredictionError: for any of the following:
(1) the user provided python model class cannot be found
(2) if the loaded class does not implement the Model interface.
"""
model_class = load_custom_class(UserClassType.model_class)
if not model_class:
return None
model_instance = model_class.from_client(client, model_path)
_validate_model_class(model_instance)
return model_instance
def load_custom_class(class_type):
"""Loads in the user specified custom class.
Args:
class_type: An instance of UserClassType specifying what type of class to
load.
Returns:
An instance of a class specified by the user in the `create_version_request`
or None if no such class was specified.
Raises:
PredictionError: if the user provided python class cannot be found.
"""
create_version_json = os.environ.get("create_version_request")
if not create_version_json:
return None
create_version_request = json.loads(create_version_json)
if not create_version_request:
return None
version = create_version_request.get("version")
if not version:
return None
class_name = version.get(class_type.name)
if not class_name:
return None
custom_class = pydoc.locate(class_name)
# TODO(b/37749453): right place to generate errors?
if not custom_class:
package_uris = [str(s) for s in version.get("package_uris")]
raise PredictionError(PredictionError.INVALID_USER_CODE,
"%s cannot be found. Please make sure "
"(1) %s is the fully qualified function "
"name, and (2) %s uses the correct package "
"name as provided by the package_uris: %s" %
(class_name, class_type.name, class_type.name,
package_uris))
return custom_class
def _validate_model_class(user_class):
"""Validates a user provided instance of a Model implementation.
Args:
user_class: An instance of a Model implementation.
Raises:
PredictionError: for any of the following:
(1) the user model class does not have the correct method signatures for
the predict method
"""
user_class_name = type(user_class).__name__
# Can't use isinstance() because the user doesn't have access to our Model
# class. We can only inspect the user_class to check if it conforms to the
# Model interface.
if not hasattr(user_class, "predict"):
raise PredictionError(PredictionError.INVALID_USER_CODE,
"The provided model class, %s, is missing the "
"required predict method." % user_class_name)
# Check the predict method has the correct number of arguments
user_signature = inspect.getargspec(user_class.predict)[0]
model_signature = inspect.getargspec(Model.predict)[0]
user_predict_num_args = len(user_signature)
predict_num_args = len(model_signature)
if predict_num_args is not user_predict_num_args:
raise PredictionError(PredictionError.INVALID_USER_CODE,
"The provided model class, %s, has a predict method "
"with an invalid signature. Expected signature: %s "
"User signature: %s" %
(user_class_name, model_signature, user_signature))
# TODO(user): Make this generic so it can load any Processor class, not just
# from the create_version_request.
def _new_processor_class(model_path=None):
user_processor_cls = load_custom_class(UserClassType.processor_class)
if user_processor_cls:
user_preprocess_fn = getattr(user_processor_cls, PREPROCESS_KEY, None)
user_postprocess_fn = getattr(user_processor_cls, POSTPROCESS_KEY, None)
user_from_model_path_fn = getattr(user_processor_cls, FROM_MODEL_KEY, None)
_validate_fn_signature(user_preprocess_fn,
["self", "instances"],
PREPROCESS_KEY, user_processor_cls.__name__)
_validate_fn_signature(user_postprocess_fn,
["self", "instances"],
POSTPROCESS_KEY, user_processor_cls.__name__)
_validate_fn_signature(user_from_model_path_fn, ["cls", "model_path"],
FROM_MODEL_KEY, user_processor_cls.__name__)
if user_from_model_path_fn:
return user_from_model_path_fn(model_path) # pylint: disable=not-callable
# Call the constructor if no `from_model_path` method provided.
return user_processor_cls()
def _validate_fn_signature(fn, required_arg_names, expected_fn_name, cls_name):
if not fn:
return
if not callable(fn):
raise PredictionError(
PredictionError.INVALID_USER_CODE,
"The provided %s function in the Processor class "
"%s is not callable." % (expected_fn_name, cls_name))
for arg in required_arg_names:
if arg not in inspect.getargspec(fn).args:
raise PredictionError(
PredictionError.INVALID_USER_CODE,
"The provided %s function in the Processor class "
"has an invalid signature. It should take %s as arguments but "
"takes %s" %
(fn.__name__, required_arg_names, inspect.getargspec(fn).args))
# (TODO:b/68775232): Move this to a Tensorflow specific library.
class TensorFlowModel(BaseModel):
"""The default implementation of the Model interface that uses TensorFlow.
This implementation optionally performs preprocessing and postprocessing
using the provided functions. These functions accept a single instance
as input and produce a corresponding output to send to the prediction
client.
"""
def __init__(self, client):
"""Constructs a TensorFlowModel.
Args:
client: An instance of ModelServerClient or SessionClient.
"""
super(TensorFlowModel, self).__init__(client)
self._preprocess_fn = None
self._postprocess_fn = None
processor_cls = _new_processor_class()
if processor_cls:
self._preprocess_fn = getattr(processor_cls, PREPROCESS_KEY, None)
self._postprocess_fn = getattr(processor_cls, POSTPROCESS_KEY, None)
def _get_columns(self, instances, stats, signature):
"""Columnarize the instances, appending input_name, if necessary.
Instances are the same instances passed to the predict() method. Since
models with a single input can accept the raw input without the name,
we create a dict here with that name.
This list of instances is then converted into a column-oriented format:
The result is a dictionary mapping input name to a list of values for just
that input (one entry per row in the original instances list).
Args:
instances: the list of instances as provided to the predict() method.
stats: Stats object for recording timing information.
signature: SignatureDef for the current request.
Returns:
A dictionary mapping input names to their values.
Raises:
PredictionError: if an error occurs during prediction.
"""
with stats.time(COLUMNARIZE_TIME):
columns = columnarize(instances)
for k, v in columns.iteritems():
if k not in signature.inputs.keys():
raise PredictionError(
PredictionError.INVALID_INPUTS,
"Unexpected tensor name: %s" % k)
# Detect whether or not the user omits an input in one or more inputs.
# TODO(b/34686738): perform this check in columnarize?
if isinstance(v, list) and len(v) != len(instances):
raise PredictionError(
PredictionError.INVALID_INPUTS,
"Input %s was missing in at least one input instance." % k)
return columns
# TODO(b/34686738): can this be removed?
def is_single_input(self, signature):
"""Returns True if the graph only has one input tensor."""
return len(signature.inputs) == 1
# TODO(b/34686738): can this be removed?
def is_single_string_input(self, signature):
"""Returns True if the graph only has one string input tensor."""
if self.is_single_input(signature):
dtype = signature.inputs.values()[0].dtype
return dtype == dtypes.string.as_datatype_enum
return False
def get_signature(self, signature_name=None):
return self._client.get_signature(signature_name)
def preprocess(self, instances, stats=None, signature_name=None, **kwargs):
_, signature = self.get_signature(signature_name)
preprocessed = self._canonicalize_input(instances, signature)
if self._preprocess_fn:
try:
preprocessed = self._preprocess_fn(preprocessed, **kwargs)
except Exception as e:
logging.error("Exception during preprocessing: " + str(e))
raise PredictionError(PredictionError.INVALID_INPUTS,
"Exception during preprocessing: " + str(e))
return self._get_columns(preprocessed, stats, signature)
def _canonicalize_input(self, instances, signature):
"""Preprocess single-input instances to be dicts if they aren't already."""
# The instances should be already (b64-) decoded here.
if not self.is_single_input(signature):
return instances
tensor_name = signature.inputs.keys()[0]
return canonicalize_single_tensor_input(instances, tensor_name)
def postprocess(self, predicted_output, original_input=None, stats=None,
signature_name=None, **kwargs):
"""Performs the necessary transformations on the prediction results.
The transformations include rowifying the predicted results, and also
making sure that each input/output is a dict mapping input/output alias to
the value for that input/output.
Args:
predicted_output: list of instances returned by the predict() method on
preprocessed instances.
original_input: List of instances, before any pre-processing was applied.
stats: Stats object for recording timing information.
signature_name: the signature name to find out the signature.
**kwargs: Additional keyword arguments for postprocessing
Returns:
A list which is a dict mapping output alias to the output.
"""
_, signature = self.get_signature(signature_name)
with stats.time(ROWIFY_TIME):
# When returned element only contains one result (batch size == 1),
# tensorflow's session.run() will return a scalar directly instead of a
# a list. So we need to listify that scalar.
# TODO(b/34686738): verify this behavior is correct.
def listify(value):
if not hasattr(value, "shape"):
return np.asarray([value], dtype=np.object)
elif not value.shape:
# TODO(b/34686738): pretty sure this is a bug that only exists because
# samples like iris have a bug where they use tf.squeeze which removes
# the batch dimension. The samples should be fixed.
return np.expand_dims(value, axis=0)
else:
return value
postprocessed_outputs = {
alias: listify(val)
for alias, val in predicted_output.iteritems()
}
postprocessed_outputs = rowify(postprocessed_outputs)
postprocessed_outputs = list(postprocessed_outputs)
if self._postprocess_fn:
try:
postprocessed_outputs = self._postprocess_fn(postprocessed_outputs,
**kwargs)
except Exception as e:
logging.error("Exception during postprocessing: %s", e)
raise PredictionError(PredictionError.INVALID_INPUTS,
"Exception during postprocessing: " + str(e))
with stats.time(ENCODE_TIME):
try:
postprocessed_outputs = encode_base64(
postprocessed_outputs, signature.outputs)
except PredictionError as e:
logging.error("Encode base64 failed: %s", e)
raise PredictionError(PredictionError.INVALID_OUTPUTS,
"Prediction failed during encoding instances: {0}"
.format(e.error_detail))
except ValueError as e:
logging.error("Encode base64 failed: %s", e)
raise PredictionError(PredictionError.INVALID_OUTPUTS,
"Prediction failed during encoding instances: {0}"
.format(e))
except Exception as e: # pylint: disable=broad-except
logging.error("Encode base64 failed: %s", e)
raise PredictionError(PredictionError.INVALID_OUTPUTS,
"Prediction failed during encoding instances")
return postprocessed_outputs
@classmethod
def from_client(cls, client, unused_model_path, **unused_kwargs):
"""Creates a TensorFlowModel from a SessionClient and model data files."""
return cls(client)
@property
def signature_map(self):
return self._client.signature_map
# This class is specific to Scikit-learn, and should be moved to a separate
# module. However due to gcloud's complicated copying mechanism we need to keep
# things in one file for now.
class SklearnClient(PredictionClient):
"""A loaded scikit-learn model to be used for prediction."""
def __init__(self, predictor):
self._predictor = predictor
def predict(self, inputs, stats=None, **kwargs):
stats = stats or Stats()
stats[FRAMEWORK] = SCIKIT_LEARN_FRAMEWORK_NAME
stats[ENGINE] = SCIKIT_LEARN_FRAMEWORK_NAME
try:
return self._predictor.predict(inputs, **kwargs)
except Exception as e:
logging.exception("Exception while predicting with sklearn model.")
raise PredictionError(PredictionError.FAILED_TO_RUN_MODEL,
"Exception during sklearn prediction: " + str(e))
# (TODO:b/68775232) This class is specific to Xgboost, and should be moved to a
# separate module. However due to gcloud's complicated copying mechanism we need
# to keep things in one file for now.
class XgboostClient(PredictionClient):
"""A loaded xgboost model to be used for prediction."""
def __init__(self, booster):
self._booster = booster
def predict(self, inputs, stats=None, **kwargs):
stats = stats or Stats()
stats[FRAMEWORK] = XGBOOST_FRAMEWORK_NAME
stats[ENGINE] = XGBOOST_FRAMEWORK_NAME
# TODO(b/64574886): Move this to the top once b/64574886 is resolved.
# Before then, it would work in production since we install xgboost in
# the Dockerfile, but the problem is the unit test that will fail to build
# and run since xgboost can not be added as a dependency to this target.
import xgboost as xgb # pylint: disable=g-import-not-at-top
try:
inputs_dmatrix = xgb.DMatrix(inputs)
except Exception as e:
logging.exception("Could not initialize DMatrix from inputs: ")
raise PredictionError(
PredictionError.FAILED_TO_RUN_MODEL,
"Could not initialize DMatrix from inputs: " + str(e))
try:
return self._booster.predict(inputs_dmatrix, **kwargs)
except Exception as e:
logging.exception("Exception during predicting with xgboost model: ")
raise PredictionError(PredictionError.FAILED_TO_RUN_MODEL,
"Exception during xgboost prediction: " + str(e))
# (TODO:b/68775232) Move this to a separate Scikit-learn specific library.
class SklearnModel(BaseModel):
"""The implementation of Scikit-learn Model.
"""
def __init__(self, client):
super(SklearnModel, self).__init__(client)
self._user_processor = _new_processor_class()
if self._user_processor and hasattr(self._user_processor, PREPROCESS_KEY):
self._preprocess = self._user_processor.preprocess
else:
self._preprocess = self._null_processor
if self._user_processor and hasattr(self._user_processor, POSTPROCESS_KEY):
self._postprocess = self._user_processor.postprocess
else:
self._postprocess = self._null_processor
def predict(self, instances, stats=None, **kwargs):
"""Override the predict method to remove TF-specific args from kwargs."""
kwargs.pop(SIGNATURE_KEY, None)
return super(SklearnModel, self).predict(instances, stats, **kwargs)
def preprocess(self, instances, stats=None, **kwargs):
# TODO(b/67383676) Consider changing this to a more generic type.
return self._preprocess(np.array(instances), **kwargs)
def postprocess(self, predicted_outputs, original_input=None, stats=None,
**kwargs):
# TODO(b/67383676) Consider changing this to a more generic type.
post_processed = self._postprocess(predicted_outputs, **kwargs)
if isinstance(post_processed, np.ndarray):
return post_processed.tolist()
if isinstance(post_processed, list):
return post_processed
raise PredictionError(
PredictionError.INVALID_OUTPUTS,
"Bad output type returned after running %s"
"The post-processing function should return either "
"a numpy ndarray or a list."
% self._postprocess.__name__)
def _null_processor(self, instances, **unused_kwargs):
return instances
# (TODO:b/68775232): Move this to a XGboost specific library.
class XGBoostModel(SklearnModel):
"""The implementation of XGboost Model.
"""
def __init__(self, client):
super(XGBoostModel, self).__init__(client)
def create_sklearn_client(model_path, unused_tags):
"""Returns a prediction client for the corresponding sklearn model."""
logging.info("Loading the scikit-learn model file from %s", model_path)
sklearn_predictor = _load_joblib_or_pickle_model(model_path)
if not sklearn_predictor:
error_msg = "Could not find either {} or {} in {}".format(
MODEL_FILE_NAME_JOBLIB, MODEL_FILE_NAME_PICKLE, model_path)
logging.critical(error_msg)
raise PredictionError(PredictionError.FAILED_TO_LOAD_MODEL, error_msg)
return SklearnClient(sklearn_predictor)
def create_sklearn_model(model_path, unused_flags):
"""Returns a sklearn model from the given model_path."""
return SklearnModel(create_sklearn_client(model_path, None))
def create_xgboost_client(model_path, unused_tags):
"""Returns a prediction client for the corresponding xgboost model."""
logging.info("Loading the xgboost model from %s", model_path)
# TODO(b/64574886): Move this to the top once b/64574886 is resolved. Before
# then, it would work in production since we install xgboost in the
# Dockerfile, but the problem is the unit test that will fail to build and run
# since xgboost can not be added as a dependency to this target.
import xgboost as xgb # pylint: disable=g-import-not-at-top
try:
booster = _load_joblib_or_pickle_model(model_path) or xgb.Booster(
model_file=os.path.join(model_path, MODEL_FILE_NAME_BST))
except xgb.core.XGBoostError as e:
error_msg = "Could not load the model: {}. {}.".format(
os.path.join(model_path, MODEL_FILE_NAME_BST), str(e))
logging.critical(error_msg)
raise PredictionError(PredictionError.FAILED_TO_LOAD_MODEL, error_msg)
return XgboostClient(booster)
def create_xgboost_model(model_path, unused_flags):
"""Returns a xgboost model from the given model_path."""
return XGBoostModel(create_xgboost_client(model_path, None))
# (TODO:b/68775232): Move this to a Tensorflow specific library.
def create_tf_session_client(model_dir, tags):
return SessionClient(*load_model(model_dir, tags))
# (TODO:b/68775232): Move this to a separate utils library.
_PICKLE_MODULE_WHITELIST = [
"sklearn", "copy_reg", "xgboost", "numpy", "scipy", "pandas"
]
_PICKLE_CLASS_WHITELIST = {
"__builtin__": (__builtin__, [
"basestring",
"bool",
"buffer",
"bytearray",
"bytes",
"complex",
"dict",
"enumerate",
"float",
"frozenset",
"int",
"list",
"long",
"reversed",
"set",
"slice",
"str",
"tuple",
"unicode",
"xrange",
"object",
],),
}
class _RestrictedUnpickler(pickle.Unpickler):
"""Restricted Unpickler implementation.
Prevents execution of code from pickled data by allowing only importing
whitelisted modules.
"""
def find_class(self, module_name, name):
if module_name.split(".")[0] in _PICKLE_MODULE_WHITELIST:
module = importlib.import_module(module_name)
return getattr(module, name)
(module, safe_names) = _PICKLE_CLASS_WHITELIST.get(module_name, (None, []))
if name in safe_names:
return getattr(module, name)
# Forbid everything else.
raise pickle.UnpicklingError("Importing global module: %s.%s is forbidden" %
(module_name, name))
@classmethod
def load_string(class_, pickle_string):
return class_(StringIO.StringIO(pickle_string)).load()
def _load_joblib_or_pickle_model(model_path):
"""Loads either a .joblib or .pkl file.
Loads one of MODEL_FILE_NAME_JOBLIB or MODEL_FILE_NAME_PICKLE files if they
exist.
Arguments:
model_path: The path to the directory that contains the model file.
Raises:
PredictionError: If there is a problem while loading the file.
Returns:
A loaded scikit-learn predictor object or None if neither
MODEL_FILE_NAME_JOBLIB nor MODEL_FILE_NAME_PICKLE files are found.
"""
try:
# If we put this at the top, we need to add a dependency to sklearn
# anywhere that prediction_lib is called.
from sklearn.externals import joblib # pylint: disable=g-import-not-at-top
except Exception as e:
error_msg = "Could not import sklearn module."
logging.critical(error_msg)
raise PredictionError(PredictionError.FAILED_TO_LOAD_MODEL, error_msg)
try:
if os.path.exists(os.path.join(model_path, MODEL_FILE_NAME_JOBLIB)):
model_file_name = os.path.join(model_path, MODEL_FILE_NAME_JOBLIB)
logging.info("Loading model %s using joblib.", model_file_name)
return joblib.load(os.path.join(model_path, MODEL_FILE_NAME_JOBLIB))
elif os.path.exists(os.path.join(model_path, MODEL_FILE_NAME_PICKLE)):
model_file_name = os.path.join(model_path, MODEL_FILE_NAME_PICKLE)
logging.info("Loading model %s using pickle.", model_file_name)
with open(os.path.join(model_path, MODEL_FILE_NAME_PICKLE), "rb") as f:
return _RestrictedUnpickler.load_string(f.read())
except Exception as e:
error_msg = "Could not load the model: {}. {}.".format(
model_file_name, str(e))
logging.critical(error_msg)
raise PredictionError(PredictionError.FAILED_TO_LOAD_MODEL, error_msg)
return None
_FRAMEWORK_TO_MODEL_MAP = {
TENSORFLOW_FRAMEWORK_NAME: (TensorFlowModel, create_tf_session_client),
SCIKIT_LEARN_FRAMEWORK_NAME: (SklearnModel, create_sklearn_client),
XGBOOST_FRAMEWORK_NAME: (XGBoostModel, create_xgboost_client)
}
def create_model(client,
model_path,
framework=TENSORFLOW_FRAMEWORK_NAME,
**unused_kwargs):
"""Creates and returns the appropriate model.
Creates and returns a Model if no user specified model is
provided. Otherwise, the user specified model is imported, created, and
returned.
Args:
client: An instance of PredictionClient for performing prediction.
model_path: The path to the exported model (e.g. session_bundle or
SavedModel)
framework: The framework used to train the model.
Returns:
An instance of the appropriate model class.
"""
if framework is TENSORFLOW_FRAMEWORK_NAME:
logging.info("Importing tensorflow.contrib in create_model")
import tensorflow.contrib # pylint: disable=redefined-outer-name, unused-variable, g-import-not-at-top
model_cls = _FRAMEWORK_TO_MODEL_MAP[framework][0]
return (load_model_class(client, model_path) or
model_cls(client))
def create_client(framework, model_path, tags):
framework = framework or TENSORFLOW_FRAMEWORK_NAME
create_client_fn = _FRAMEWORK_TO_MODEL_MAP[framework][1]
return create_client_fn(model_path, tags)
def decode_base64(data):
if isinstance(data, list):
return [decode_base64(val) for val in data]
elif isinstance(data, dict):
if data.viewkeys() == {"b64"}:
return base64.b64decode(data["b64"])
else:
return {k: decode_base64(v) for k, v in data.iteritems()}
else:
return data
def encode_base64(instances, outputs_map):
"""Encodes binary data in a JSON-friendly way."""
if not isinstance(instances, list):
raise ValueError("only lists allowed in output; got %s" %
(type(instances),))
if not instances:
return instances
first_value = instances[0]
if not isinstance(first_value, dict):
if len(outputs_map) != 1:
return ValueError("The first instance was a string, but there are "
"more than one output tensor, so dict expected.")
# Only string tensors whose name ends in _bytes needs encoding.
tensor_name, tensor_info = outputs_map.items()[0]
tensor_type = tensor_info.dtype
if tensor_type == dtypes.string and tensor_name.endswith("_bytes"):
instances = _encode_str_tensor(instances)
return instances
encoded_data = []
for instance in instances:
encoded_instance = {}
for tensor_name, tensor_info in outputs_map.iteritems():
tensor_type = tensor_info.dtype
tensor_data = instance[tensor_name]
if tensor_type == dtypes.string and tensor_name.endswith("_bytes"):
tensor_data = _encode_str_tensor(tensor_data)
encoded_instance[tensor_name] = tensor_data
encoded_data.append(encoded_instance)
return encoded_data
def _encode_str_tensor(data):
if isinstance(data, list):
return [_encode_str_tensor(val) for val in data]
return {"b64": base64.b64encode(data)}
def local_predict(
model_dir=None,
tags=(tag_constants.SERVING,),
signature_name=None,
instances=None,
framework=TENSORFLOW_FRAMEWORK_NAME):
"""Run a prediction locally."""
instances = decode_base64(instances)
client = create_client(framework, model_dir, tags)
model = create_model(client, model_dir, framework)
_, predictions = model.predict(instances, signature_name=signature_name)
return {"predictions": list(predictions)}
| [
"tensorflow.python.saved_model.loader.maybe_saved_model_directory",
"base64.b64decode",
"collections.defaultdict",
"logging.critical",
"os.path.join",
"StringIO.StringIO",
"logging.error",
"json.loads",
"tensorflow.python.client.session.Session",
"importlib.import_module",
"pydoc.locate",
"num... | [((2882, 2948), 'collections.namedtuple', 'collections.namedtuple', (['"""PredictionErrorType"""', "('message', 'code')"], {}), "('PredictionErrorType', ('message', 'code'))\n", (2904, 2948), False, 'import collections\n'), ((7173, 7202), 'collections.defaultdict', 'collections.defaultdict', (['list'], {}), '(list)\n', (7196, 7202), False, 'import collections\n'), ((13862, 13908), 'tensorflow.python.saved_model.loader.maybe_saved_model_directory', 'loader.maybe_saved_model_directory', (['model_path'], {}), '(model_path)\n', (13896, 13908), False, 'from tensorflow.python.saved_model import loader\n'), ((21572, 21612), 'os.environ.get', 'os.environ.get', (['"""create_version_request"""'], {}), "('create_version_request')\n", (21586, 21612), False, 'import os\n'), ((21686, 21717), 'json.loads', 'json.loads', (['create_version_json'], {}), '(create_version_json)\n', (21696, 21717), False, 'import json\n'), ((21949, 21973), 'pydoc.locate', 'pydoc.locate', (['class_name'], {}), '(class_name)\n', (21961, 21973), False, 'import pydoc\n'), ((38126, 38197), 'logging.info', 'logging.info', (['"""Loading the scikit-learn model file from %s"""', 'model_path'], {}), "('Loading the scikit-learn model file from %s', model_path)\n", (38138, 38197), False, 'import logging\n'), ((38875, 38936), 'logging.info', 'logging.info', (['"""Loading the xgboost model from %s"""', 'model_path'], {}), "('Loading the xgboost model from %s', model_path)\n", (38887, 38936), False, 'import logging\n'), ((23483, 23521), 'inspect.getargspec', 'inspect.getargspec', (['user_class.predict'], {}), '(user_class.predict)\n', (23501, 23521), False, 'import inspect\n'), ((23545, 23578), 'inspect.getargspec', 'inspect.getargspec', (['Model.predict'], {}), '(Model.predict)\n', (23563, 23578), False, 'import inspect\n'), ((38424, 38451), 'logging.critical', 'logging.critical', (['error_msg'], {}), '(error_msg)\n', (38440, 38451), False, 'import logging\n'), ((41258, 41354), 'pickle.UnpicklingError', 'pickle.UnpicklingError', (["('Importing global module: %s.%s is forbidden' % (module_name, name))"], {}), "('Importing global module: %s.%s is forbidden' % (\n module_name, name))\n", (41280, 41354), False, 'import pickle\n'), ((44197, 44257), 'logging.info', 'logging.info', (['"""Importing tensorflow.contrib in create_model"""'], {}), "('Importing tensorflow.contrib in create_model')\n", (44209, 44257), False, 'import logging\n'), ((46430, 46452), 'base64.b64encode', 'base64.b64encode', (['data'], {}), '(data)\n', (46446, 46452), False, 'import base64\n'), ((13925, 13983), 'logging.info', 'logging.info', (['"""Importing tensorflow.contrib in load_model"""'], {}), "('Importing tensorflow.contrib in load_model')\n", (13937, 13983), False, 'import logging\n'), ((14110, 14166), 'tensorflow.python.client.session.Session', 'tf_session.Session', ([], {'target': '""""""', 'graph': 'None', 'config': 'config'}), "(target='', graph=None, config=config)\n", (14128, 14166), True, 'from tensorflow.python.client import session as tf_session\n'), ((35361, 35380), 'xgboost.DMatrix', 'xgb.DMatrix', (['inputs'], {}), '(inputs)\n', (35372, 35380), True, 'import xgboost as xgb\n'), ((37010, 37029), 'numpy.array', 'np.array', (['instances'], {}), '(instances)\n', (37018, 37029), True, 'import numpy as np\n'), ((39605, 39632), 'logging.critical', 'logging.critical', (['error_msg'], {}), '(error_msg)\n', (39621, 39632), False, 'import logging\n'), ((41003, 41039), 'importlib.import_module', 'importlib.import_module', (['module_name'], {}), '(module_name)\n', (41026, 41039), False, 'import importlib\n'), ((42278, 42305), 'logging.critical', 'logging.critical', (['error_msg'], {}), '(error_msg)\n', (42294, 42305), False, 'import logging\n'), ((42410, 42458), 'os.path.join', 'os.path.join', (['model_path', 'MODEL_FILE_NAME_JOBLIB'], {}), '(model_path, MODEL_FILE_NAME_JOBLIB)\n', (42422, 42458), False, 'import os\n'), ((42485, 42533), 'os.path.join', 'os.path.join', (['model_path', 'MODEL_FILE_NAME_JOBLIB'], {}), '(model_path, MODEL_FILE_NAME_JOBLIB)\n', (42497, 42533), False, 'import os\n'), ((42540, 42603), 'logging.info', 'logging.info', (['"""Loading model %s using joblib."""', 'model_file_name'], {}), "('Loading model %s using joblib.', model_file_name)\n", (42552, 42603), False, 'import logging\n'), ((43156, 43183), 'logging.critical', 'logging.critical', (['error_msg'], {}), '(error_msg)\n', (43172, 43183), False, 'import logging\n'), ((25627, 25649), 'inspect.getargspec', 'inspect.getargspec', (['fn'], {}), '(fn)\n', (25645, 25649), False, 'import inspect\n'), ((34220, 34287), 'logging.exception', 'logging.exception', (['"""Exception while predicting with sklearn model."""'], {}), "('Exception while predicting with sklearn model.')\n", (34237, 34287), False, 'import logging\n'), ((35414, 35477), 'logging.exception', 'logging.exception', (['"""Could not initialize DMatrix from inputs: """'], {}), "('Could not initialize DMatrix from inputs: ')\n", (35431, 35477), False, 'import logging\n'), ((35722, 35791), 'logging.exception', 'logging.exception', (['"""Exception during predicting with xgboost model: """'], {}), "('Exception during predicting with xgboost model: ')\n", (35739, 35791), False, 'import logging\n'), ((39546, 39591), 'os.path.join', 'os.path.join', (['model_path', 'MODEL_FILE_NAME_BST'], {}), '(model_path, MODEL_FILE_NAME_BST)\n', (39558, 39591), False, 'import os\n'), ((42629, 42677), 'os.path.join', 'os.path.join', (['model_path', 'MODEL_FILE_NAME_JOBLIB'], {}), '(model_path, MODEL_FILE_NAME_JOBLIB)\n', (42641, 42677), False, 'import os\n'), ((42704, 42752), 'os.path.join', 'os.path.join', (['model_path', 'MODEL_FILE_NAME_PICKLE'], {}), '(model_path, MODEL_FILE_NAME_PICKLE)\n', (42716, 42752), False, 'import os\n'), ((42779, 42827), 'os.path.join', 'os.path.join', (['model_path', 'MODEL_FILE_NAME_PICKLE'], {}), '(model_path, MODEL_FILE_NAME_PICKLE)\n', (42791, 42827), False, 'import os\n'), ((42834, 42897), 'logging.info', 'logging.info', (['"""Loading model %s using pickle."""', 'model_file_name'], {}), "('Loading model %s using pickle.', model_file_name)\n", (42846, 42897), False, 'import logging\n'), ((44886, 44915), 'base64.b64decode', 'base64.b64decode', (["data['b64']"], {}), "(data['b64'])\n", (44902, 44915), False, 'import base64\n'), ((31300, 31336), 'numpy.asarray', 'np.asarray', (['[value]'], {'dtype': 'np.object'}), '([value], dtype=np.object)\n', (31310, 31336), True, 'import numpy as np\n'), ((32136, 32191), 'logging.error', 'logging.error', (['"""Exception during postprocessing: %s"""', 'e'], {}), "('Exception during postprocessing: %s', e)\n", (32149, 32191), False, 'import logging\n'), ((32520, 32564), 'logging.error', 'logging.error', (['"""Encode base64 failed: %s"""', 'e'], {}), "('Encode base64 failed: %s', e)\n", (32533, 32564), False, 'import logging\n'), ((32802, 32846), 'logging.error', 'logging.error', (['"""Encode base64 failed: %s"""', 'e'], {}), "('Encode base64 failed: %s', e)\n", (32815, 32846), False, 'import logging\n'), ((33102, 33146), 'logging.error', 'logging.error', (['"""Encode base64 failed: %s"""', 'e'], {}), "('Encode base64 failed: %s', e)\n", (33115, 33146), False, 'import logging\n'), ((39394, 39439), 'os.path.join', 'os.path.join', (['model_path', 'MODEL_FILE_NAME_BST'], {}), '(model_path, MODEL_FILE_NAME_BST)\n', (39406, 39439), False, 'import os\n'), ((41459, 41491), 'StringIO.StringIO', 'StringIO.StringIO', (['pickle_string'], {}), '(pickle_string)\n', (41476, 41491), False, 'import StringIO\n'), ((31608, 31637), 'numpy.expand_dims', 'np.expand_dims', (['value'], {'axis': '(0)'}), '(value, axis=0)\n', (31622, 31637), True, 'import numpy as np\n'), ((42914, 42962), 'os.path.join', 'os.path.join', (['model_path', 'MODEL_FILE_NAME_PICKLE'], {}), '(model_path, MODEL_FILE_NAME_PICKLE)\n', (42926, 42962), False, 'import os\n'), ((25932, 25954), 'inspect.getargspec', 'inspect.getargspec', (['fn'], {}), '(fn)\n', (25950, 25954), False, 'import inspect\n')] |
#!/usr/bin/env python
# -*- coding: utf-8 -*-
#
# Project: Fast Azimuthal integration
# https://github.com/silx-kit/pyFAI
#
# Copyright (C) 2017-2020 European Synchrotron Radiation Facility, Grenoble, France
#
# Principal author: <NAME> (<EMAIL>)
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
# .
# The above copyright notice and this permission notice shall be included in
# all copies or substantial portions of the Software.
# .
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
# THE SOFTWARE.
"""
Description of detectors which are not flat.
Mainly cylindrical curved imaging-plates for now.
"""
__author__ = "<NAME>"
__contact__ = "<EMAIL>"
__license__ = "MIT"
__copyright__ = "European Synchrotron Radiation Facility, Grenoble, France"
__date__ = "25/06/2020"
__status__ = "production"
import numpy
import logging
logger = logging.getLogger(__name__)
import json
from collections import OrderedDict
from ._common import Detector
from pyFAI.utils import mathutil
try:
from ..ext import bilinear
except ImportError:
logger.debug("Backtrace", exc_info=True)
bilinear = None
class CylindricalDetector(Detector):
"Abstract base class for all cylindrical detecors"
MANUFACTURER = None
IS_FLAT = False
force_pixel = True
def __init__(self, pixel1=24.893e-6, pixel2=24.893e-6, radius=0.29989):
Detector.__init__(self, pixel1, pixel2)
self.radius = radius
self._pixel_corners = None
def get_config(self):
"""Return the configuration with arguments to the constructor
:return: dict with param for serialization
"""
return OrderedDict((("pixel1", self._pixel1),
("pixel2", self._pixel2),
("radius", self.radius)))
def set_config(self, config):
"""Sets the configuration of the detector.
The configuration is either a python dictionary or a JSON string or a
file containing this JSON configuration
keys in that dictionary are: pixel1, pixel2, radius
:param config: string or JSON-serialized dict
:return: self
"""
if not isinstance(config, dict):
try:
config = json.loads(config)
except Exception as err: # IGNORE:W0703:
logger.error("Unable to parse config %s with JSON: %s, %s",
config, err)
raise err
pixel1 = config.get("pixel1")
if pixel1:
self.set_pixel1(pixel1)
pixel2 = config.get("pixel2")
if pixel2:
self.set_pixel1(pixel2)
radius = config.get("radius")
if radius:
self.radius = radius
self._pixel_corners = None
return self
def _get_compact_pixel_corners(self):
"The core function which calculates the pixel corner coordinates"
raise NotImplementedError("This is an abtract class")
def get_pixel_corners(self, correct_binning=False, use_cython=True):
"""
Calculate the position of the corner of the pixels
This should be overwritten by class representing non-contiguous detector (Xpad, ...)
:param correct_binning: If True, check that the produced array have the right shape regarding binning
:param use_cython: set to False for testing
:return: 4D array containing:
pixel index (slow dimension)
pixel index (fast dimension)
corner index (A, B, C or D), triangles or hexagons can be handled the same way
vertex position (z,y,x)
"""
if self._pixel_corners is None:
with self._sem:
if self._pixel_corners is None:
p1, p2, p3 = self._get_compact_pixel_corners()
if bilinear and use_cython:
d1 = mathutil.expand2d(p1, self.shape[1] + 1, False)
d2 = mathutil.expand2d(p2, self.shape[0] + 1, True)
d3 = mathutil.expand2d(p3, self.shape[0] + 1, True)
corners = bilinear.convert_corner_2D_to_4D(3, d1, d2, d3)
else:
p1.shape = -1, 1
p1.strides = p1.strides[0], 0
p2.shape = 1, -1
p2.strides = 0, p2.strides[1]
p3.shape = 1, -1
p3.strides = 0, p3.strides[1]
corners = numpy.zeros((self.shape[0], self.shape[1], 4, 3), dtype=numpy.float32)
corners[:, :, 0, 0] = p3[:, :-1]
corners[:, :, 0, 1] = p1[:-1, :]
corners[:, :, 0, 2] = p2[:, :-1]
corners[:, :, 1, 0] = p3[:, :-1]
corners[:, :, 1, 1] = p1[1:, :]
corners[:, :, 1, 2] = p2[:, :-1]
corners[:, :, 2, 1] = p1[1:, :]
corners[:, :, 2, 2] = p2[:, 1:]
corners[:, :, 2, 0] = p3[:, 1:]
corners[:, :, 3, 0] = p3[:, 1:]
corners[:, :, 3, 1] = p1[:-1, :]
corners[:, :, 3, 2] = p2[:, 1:]
self._pixel_corners = corners
if correct_binning and self._pixel_corners.shape[:2] != self.shape:
return self._rebin_pixel_corners()
else:
return self._pixel_corners
def calc_cartesian_positions(self, d1=None, d2=None, center=True, use_cython=True):
"""
Calculate the position of each pixel center in cartesian coordinate
and in meter of a couple of coordinates.
The half pixel offset is taken into account here !!!
Adapted to Nexus detector definition
:param d1: the Y pixel positions (slow dimension)
:type d1: ndarray (1D or 2D)
:param d2: the X pixel positions (fast dimension)
:type d2: ndarray (1D or 2D)
:param center: retrieve the coordinate of the center of the pixel
:param use_cython: set to False to test Python implementeation
:return: position in meter of the center of each pixels.
:rtype: ndarray
d1 and d2 must have the same shape, returned array will have
the same shape.
"""
if (d1 is None) or d2 is None:
d1 = mathutil.expand2d(numpy.arange(self.shape[0]).astype(numpy.float32), self.shape[1], False)
d2 = mathutil.expand2d(numpy.arange(self.shape[1]).astype(numpy.float32), self.shape[0], True)
corners = self.get_pixel_corners()
if center:
# avoid += It modifies in place and segfaults
d1 = d1 + 0.5
d2 = d2 + 0.5
if bilinear and use_cython:
p1, p2, p3 = bilinear.calc_cartesian_positions(d1.ravel(), d2.ravel(), corners, is_flat=False)
p1.shape = d1.shape
p2.shape = d2.shape
p3.shape = d2.shape
else:
i1 = d1.astype(int).clip(0, corners.shape[0] - 1)
i2 = d2.astype(int).clip(0, corners.shape[1] - 1)
delta1 = d1 - i1
delta2 = d2 - i2
pixels = corners[i1, i2]
if pixels.ndim == 3:
A0 = pixels[:, 0, 0]
A1 = pixels[:, 0, 1]
A2 = pixels[:, 0, 2]
B0 = pixels[:, 1, 0]
B1 = pixels[:, 1, 1]
B2 = pixels[:, 1, 2]
C0 = pixels[:, 2, 0]
C1 = pixels[:, 2, 1]
C2 = pixels[:, 2, 2]
D0 = pixels[:, 3, 0]
D1 = pixels[:, 3, 1]
D2 = pixels[:, 3, 2]
else:
A0 = pixels[:, :, 0, 0]
A1 = pixels[:, :, 0, 1]
A2 = pixels[:, :, 0, 2]
B0 = pixels[:, :, 1, 0]
B1 = pixels[:, :, 1, 1]
B2 = pixels[:, :, 1, 2]
C0 = pixels[:, :, 2, 0]
C1 = pixels[:, :, 2, 1]
C2 = pixels[:, :, 2, 2]
D0 = pixels[:, :, 3, 0]
D1 = pixels[:, :, 3, 1]
D2 = pixels[:, :, 3, 2]
# points A and D are on the same dim1 (Y), they differ in dim2 (X)
# points B and C are on the same dim1 (Y), they differ in dim2 (X)
# points A and B are on the same dim2 (X), they differ in dim1 (Y)
# points C and D are on the same dim2 (X), they differ in dim1 (
p1 = A1 * (1.0 - delta1) * (1.0 - delta2) \
+ B1 * delta1 * (1.0 - delta2) \
+ C1 * delta1 * delta2 \
+ D1 * (1.0 - delta1) * delta2
p2 = A2 * (1.0 - delta1) * (1.0 - delta2) \
+ B2 * delta1 * (1.0 - delta2) \
+ C2 * delta1 * delta2 \
+ D2 * (1.0 - delta1) * delta2
p3 = A0 * (1.0 - delta1) * (1.0 - delta2) \
+ B0 * delta1 * (1.0 - delta2) \
+ C0 * delta1 * delta2 \
+ D0 * (1.0 - delta1) * delta2
# To ensure numerical consitency with cython procedure.
p1 = p1.astype(numpy.float32)
p2 = p2.astype(numpy.float32)
p3 = p3.astype(numpy.float32)
return p1, p2, p3
class Aarhus(CylindricalDetector):
"""
Cylindrical detector made of a bent imaging-plate.
Developped at the Danish university of Aarhus
r = 1.2m or 0.3m
Credits:
Private communication;
<NAME>,
Center for Materials Crystallography & Dept. of Chemistry and iNANO,
Aarhus University
The image has to be laid-out horizontally
Nota: the detector is bend towards the sample, hence reducing the sample-detector distance.
This is why z<0 (or p3<0)
"""
MANUFACTURER = "Aarhus University"
MAX_SHAPE = (1000, 16000)
def __init__(self, pixel1=24.893e-6, pixel2=24.893e-6, radius=0.29989):
CylindricalDetector.__init__(self, pixel1, pixel2, radius)
def _get_compact_pixel_corners(self):
"The core function which calculates the pixel corner coordinates"
p1 = (numpy.arange(self.shape[0] + 1.0) * self._pixel1).astype(numpy.float32)
t2 = numpy.arange(self.shape[1] + 1.0) * (self._pixel2 / self.radius)
p2 = (self.radius * numpy.sin(t2)).astype(numpy.float32)
p3 = (self.radius * (numpy.cos(t2) - 1.0)).astype(numpy.float32)
return p1, p2, p3
class Rapid(CylindricalDetector):
"""
Cylindrical detector: Rigaku R-axis RAPID II
Unlike the Aarhus detector, the detectors is bent the other direction.
It covers 210ยฐ
r = 127.26mm
pixel size 100ยตm but can be binned 2x2
Credits:
Private communication;
Dr. <NAME>
Department of Condensed Matter Physics
Institute of Physics
P.J. ล afรกrik University, Koลกice, Slovakia
The image has to be laid-out horizontally
Nota: the detector is bend towards the sample, hence reducing the sample-detector distance.
This is why z<0 (or p3<0)
"""
MANUFACTURER = "Rigaku"
aliases = ["RapidII"]
MAX_SHAPE = (2560, 4700)
def __init__(self, pixel1=0.1e-3, pixel2=0.1e-3, radius=0.12726):
CylindricalDetector.__init__(self, pixel1, pixel2, radius)
def _get_compact_pixel_corners(self):
"The core function which calculates the pixel corner coordinates"
p1 = (numpy.arange(self.shape[0] + 1.0) * self._pixel1).astype(numpy.float32)
t2 = numpy.arange(self.shape[1] + 1.0) * (self._pixel2 / self.radius)
p2 = (self.radius * numpy.sin(t2)).astype(numpy.float32)
p3 = (self.radius * (numpy.cos(t2) - 1.0)).astype(numpy.float32)
return p1, p2, p3
| [
"json.loads",
"pyFAI.utils.mathutil.expand2d",
"numpy.zeros",
"numpy.sin",
"numpy.arange",
"numpy.cos",
"collections.OrderedDict",
"logging.getLogger"
] | [((1690, 1717), 'logging.getLogger', 'logging.getLogger', (['__name__'], {}), '(__name__)\n', (1707, 1717), False, 'import logging\n'), ((2486, 2580), 'collections.OrderedDict', 'OrderedDict', (["(('pixel1', self._pixel1), ('pixel2', self._pixel2), ('radius', self.radius))"], {}), "((('pixel1', self._pixel1), ('pixel2', self._pixel2), ('radius',\n self.radius)))\n", (2497, 2580), False, 'from collections import OrderedDict\n'), ((11170, 11203), 'numpy.arange', 'numpy.arange', (['(self.shape[1] + 1.0)'], {}), '(self.shape[1] + 1.0)\n', (11182, 11203), False, 'import numpy\n'), ((12436, 12469), 'numpy.arange', 'numpy.arange', (['(self.shape[1] + 1.0)'], {}), '(self.shape[1] + 1.0)\n', (12448, 12469), False, 'import numpy\n'), ((3080, 3098), 'json.loads', 'json.loads', (['config'], {}), '(config)\n', (3090, 3098), False, 'import json\n'), ((11085, 11118), 'numpy.arange', 'numpy.arange', (['(self.shape[0] + 1.0)'], {}), '(self.shape[0] + 1.0)\n', (11097, 11118), False, 'import numpy\n'), ((11263, 11276), 'numpy.sin', 'numpy.sin', (['t2'], {}), '(t2)\n', (11272, 11276), False, 'import numpy\n'), ((12351, 12384), 'numpy.arange', 'numpy.arange', (['(self.shape[0] + 1.0)'], {}), '(self.shape[0] + 1.0)\n', (12363, 12384), False, 'import numpy\n'), ((12529, 12542), 'numpy.sin', 'numpy.sin', (['t2'], {}), '(t2)\n', (12538, 12542), False, 'import numpy\n'), ((4768, 4815), 'pyFAI.utils.mathutil.expand2d', 'mathutil.expand2d', (['p1', '(self.shape[1] + 1)', '(False)'], {}), '(p1, self.shape[1] + 1, False)\n', (4785, 4815), False, 'from pyFAI.utils import mathutil\n'), ((4845, 4891), 'pyFAI.utils.mathutil.expand2d', 'mathutil.expand2d', (['p2', '(self.shape[0] + 1)', '(True)'], {}), '(p2, self.shape[0] + 1, True)\n', (4862, 4891), False, 'from pyFAI.utils import mathutil\n'), ((4921, 4967), 'pyFAI.utils.mathutil.expand2d', 'mathutil.expand2d', (['p3', '(self.shape[0] + 1)', '(True)'], {}), '(p3, self.shape[0] + 1, True)\n', (4938, 4967), False, 'from pyFAI.utils import mathutil\n'), ((5395, 5465), 'numpy.zeros', 'numpy.zeros', (['(self.shape[0], self.shape[1], 4, 3)'], {'dtype': 'numpy.float32'}), '((self.shape[0], self.shape[1], 4, 3), dtype=numpy.float32)\n', (5406, 5465), False, 'import numpy\n'), ((7309, 7336), 'numpy.arange', 'numpy.arange', (['self.shape[0]'], {}), '(self.shape[0])\n', (7321, 7336), False, 'import numpy\n'), ((7417, 7444), 'numpy.arange', 'numpy.arange', (['self.shape[1]'], {}), '(self.shape[1])\n', (7429, 7444), False, 'import numpy\n'), ((11329, 11342), 'numpy.cos', 'numpy.cos', (['t2'], {}), '(t2)\n', (11338, 11342), False, 'import numpy\n'), ((12595, 12608), 'numpy.cos', 'numpy.cos', (['t2'], {}), '(t2)\n', (12604, 12608), False, 'import numpy\n')] |
import h5py
import math
import os
import numpy as np
import torch
import torch.optim as optim
import torch.nn as nn
from net import classifier
from torchlight import torchlight
def weights_init(m):
classname = m.__class__.__name__
if classname.find('Conv1d') != -1:
m.weight.data.normal_(0.0, 0.02)
if m.bias is not None:
m.bias.data.fill_(0)
elif classname.find('Conv2d') != -1:
m.weight.data.normal_(0.0, 0.02)
if m.bias is not None:
m.bias.data.fill_(0)
elif classname.find('BatchNorm') != -1:
m.weight.data.normal_(1.0, 0.02)
m.bias.data.fill_(0)
def find_all_substr(a_str, sub):
start = 0
while True:
start = a_str.find(sub, start)
if start == -1:
return
yield start
start += len(sub) # use start += 1 to find overlapping matches
def get_best_epoch_and_accuracy(path_to_model_files):
all_models = os.listdir(path_to_model_files)
acc_list = np.zeros(len(all_models))
for i, model in enumerate(all_models):
acc = str.split(model, '_')
if len(acc) > 1:
acc_list[i] = float(acc[1][3:])
best_model = all_models[np.argmax(acc_list)]
all_us = list(find_all_substr(best_model, '_'))
return int(best_model[5:all_us[0]]), float(best_model[all_us[0]+4:all_us[1]])
class Processor(object):
"""
Processor for gait generation
"""
def __init__(self, args, data_loader, C, num_classes, graph_dict, device='cuda:0'):
self.args = args
self.data_loader = data_loader
self.num_classes = num_classes
self.result = dict()
self.iter_info = dict()
self.epoch_info = dict()
self.meta_info = dict(epoch=0, iter=0)
self.device = device
self.io = torchlight.IO(
self.args.work_dir,
save_log=self.args.save_log,
print_log=self.args.print_log)
# model
if not os.path.isdir(self.args.work_dir):
os.mkdir(self.args.work_dir)
self.model = classifier.Classifier(C, num_classes, graph_dict)
self.model.cuda('cuda:0')
self.model.apply(weights_init)
self.loss = nn.CrossEntropyLoss()
self.best_loss = math.inf
self.step_epochs = [math.ceil(float(self.args.num_epoch * x)) for x in self.args.step]
self.best_epoch = None
self.best_accuracy = np.zeros((1, np.max(self.args.topk)))
self.accuracy_updated = False
# optimizer
if self.args.optimizer == 'SGD':
self.optimizer = optim.SGD(
self.model.parameters(),
lr=self.args.base_lr,
momentum=0.9,
nesterov=self.args.nesterov,
weight_decay=self.args.weight_decay)
elif self.args.optimizer == 'Adam':
self.optimizer = optim.Adam(
self.model.parameters(),
lr=self.args.base_lr,
weight_decay=self.args.weight_decay)
else:
raise ValueError()
self.lr = self.args.base_lr
def adjust_lr(self):
# if self.args.optimizer == 'SGD' and\
if self.meta_info['epoch'] in self.step_epochs:
lr = self.args.base_lr * (
0.1 ** np.sum(self.meta_info['epoch'] >= np.array(self.step_epochs)))
for param_group in self.optimizer.param_groups:
param_group['lr'] = lr
self.lr = lr
def show_epoch_info(self):
for k, v in self.epoch_info.items():
self.io.print_log('\t{}: {}'.format(k, v))
if self.args.pavi_log:
self.io.log('train', self.meta_info['iter'], self.epoch_info)
def show_iter_info(self):
if self.meta_info['iter'] % self.args.log_interval == 0:
info = '\tIter {} Done.'.format(self.meta_info['iter'])
for k, v in self.iter_info.items():
if isinstance(v, float):
info = info + ' | {}: {:.4f}'.format(k, v)
else:
info = info + ' | {}: {}'.format(k, v)
self.io.print_log(info)
if self.args.pavi_log:
self.io.log('train', self.meta_info['iter'], self.iter_info)
def show_topk(self, k):
rank = self.result.argsort()
hit_top_k = [l in rank[i, -k:] for i, l in enumerate(self.label)]
accuracy = 100. * sum(hit_top_k) * 1.0 / len(hit_top_k)
if accuracy > self.best_accuracy[0, k-1]:
self.best_accuracy[0, k-1] = accuracy
self.accuracy_updated = True
else:
self.accuracy_updated = False
print_epoch = self.best_epoch if self.best_epoch is not None else 0
self.io.print_log('\tTop{}: {:.2f}%. Best so far: {:.2f}% (epoch: {:d}).'.
format(k, accuracy, self.best_accuracy[0, k-1], print_epoch))
def per_train(self):
self.model.train()
self.adjust_lr()
loader = self.data_loader['train']
loss_value = []
for data, label in loader:
# get data
data = data.float().to(self.device)
label = label.long().to(self.device)
# forward
output, _ = self.model(data)
loss = self.loss(output, label)
# backward
self.optimizer.zero_grad()
loss.backward()
self.optimizer.step()
# statistics
self.iter_info['loss'] = loss.data.item()
self.iter_info['lr'] = '{:.6f}'.format(self.lr)
loss_value.append(self.iter_info['loss'])
self.show_iter_info()
self.meta_info['iter'] += 1
self.epoch_info['mean_loss'] = np.mean(loss_value)
self.show_epoch_info()
self.io.print_timer()
# for k in self.args.topk:
# self.calculate_topk(k, show=False)
# if self.accuracy_updated:
# self.model.extract_feature()
def per_test(self, evaluation=True):
self.model.eval()
loader = self.data_loader['test']
loss_value = []
result_frag = []
label_frag = []
for data, label in loader:
# get data
data = data.float().to(self.device)
label = label.long().to(self.device)
# inference
with torch.no_grad():
output, _ = self.model(data)
result_frag.append(output.data.cpu().numpy())
# get loss
if evaluation:
loss = self.loss(output, label)
loss_value.append(loss.item())
label_frag.append(label.data.cpu().numpy())
self.result = np.concatenate(result_frag)
if evaluation:
self.label = np.concatenate(label_frag)
self.epoch_info['mean_loss'] = np.mean(loss_value)
self.show_epoch_info()
# show top-k accuracy
for k in self.args.topk:
self.show_topk(k)
def train(self):
for epoch in range(self.args.start_epoch, self.args.num_epoch):
self.meta_info['epoch'] = epoch
# training
self.io.print_log('Training epoch: {}'.format(epoch))
self.per_train()
self.io.print_log('Done.')
# evaluation
if (epoch % self.args.eval_interval == 0) or (
epoch + 1 == self.args.num_epoch):
self.io.print_log('Eval epoch: {}'.format(epoch))
self.per_test()
self.io.print_log('Done.')
# save model and weights
if self.accuracy_updated:
torch.save(self.model.state_dict(),
os.path.join(self.args.work_dir,
'epoch{}_acc{:.2f}_model.pth.tar'.format(epoch, self.best_accuracy.item())))
if self.epoch_info['mean_loss'] < self.best_loss:
self.best_loss = self.epoch_info['mean_loss']
self.best_epoch = epoch
def test(self):
# the path of weights must be appointed
if self.args.weights is None:
raise ValueError('Please appoint --weights.')
self.io.print_log('Model: {}.'.format(self.args.model))
self.io.print_log('Weights: {}.'.format(self.args.weights))
# evaluation
self.io.print_log('Evaluation Start:')
self.per_test()
self.io.print_log('Done.\n')
# save the output of model
if self.args.save_result:
result_dict = dict(
zip(self.data_loader['test'].dataset.sample_name,
self.result))
self.io.save_pkl(result_dict, 'test_result.pkl')
def save_best_feature(self, ftype_real, ftype_synth, data, joints, coords):
if self.best_epoch is None:
self.best_epoch, best_accuracy = get_best_epoch_and_accuracy(self.args.work_dir)
else:
best_accuracy = self.best_accuracy.item()
filename = os.path.join(self.args.work_dir,
'epoch{}_acc{:.2f}_model.pth.tar'.format(self.best_epoch, best_accuracy))
self.model.load_state_dict(torch.load(filename))
features = np.empty((0, 64))
fr = h5py.File('../data/features'+ftype_real+'.h5', 'r')
fl = h5py.File('../data/features'+ftype_synth+'.h5', 'r')
frkeys = fr.keys()
flkeys = fl.keys()
df_save = h5py.File('../data/deepFeatures'+ftype_real+'+'+ftype_synth+'.h5', 'w')
for i, (each_data, each_key) in enumerate(zip(data[:len(frkeys)], frkeys)):
# get data
each_data = np.reshape(each_data, (1, each_data.shape[0], joints, coords, 1))
each_data = np.moveaxis(each_data, [1, 2, 3], [2, 3, 1])
each_data = torch.from_numpy(each_data).float().to(self.device)
# get feature
with torch.no_grad():
_, feature = self.model(each_data)
fname = [each_key][0]+'_real'
df_save.create_dataset(fname, data=feature)
features = np.append(features, np.array(feature).reshape((1, feature.shape[0])), axis=0)
for i, (each_data, each_key) in enumerate(zip(data[len(frkeys):], flkeys)):
# get data
each_data = np.reshape(each_data, (1, each_data.shape[0], joints, coords, 1))
each_data = np.moveaxis(each_data, [1, 2, 3], [2, 3, 1])
each_data = torch.from_numpy(each_data).float().to(self.device)
# get feature
with torch.no_grad():
_, feature = self.model(each_data)
fname = [each_key][0]+'_synth'
df_save.create_dataset(fname, data=feature)
features = np.append(features, np.array(feature).reshape((1, feature.shape[0])), axis=0)
df_save.close()
return features
| [
"os.mkdir",
"h5py.File",
"numpy.moveaxis",
"numpy.argmax",
"os.path.isdir",
"numpy.empty",
"torch.load",
"torch.nn.CrossEntropyLoss",
"net.classifier.Classifier",
"numpy.max",
"numpy.mean",
"numpy.array",
"numpy.reshape",
"torchlight.torchlight.IO",
"torch.no_grad",
"os.listdir",
"nu... | [((953, 984), 'os.listdir', 'os.listdir', (['path_to_model_files'], {}), '(path_to_model_files)\n', (963, 984), False, 'import os\n'), ((1202, 1221), 'numpy.argmax', 'np.argmax', (['acc_list'], {}), '(acc_list)\n', (1211, 1221), True, 'import numpy as np\n'), ((1819, 1917), 'torchlight.torchlight.IO', 'torchlight.IO', (['self.args.work_dir'], {'save_log': 'self.args.save_log', 'print_log': 'self.args.print_log'}), '(self.args.work_dir, save_log=self.args.save_log, print_log=\n self.args.print_log)\n', (1832, 1917), False, 'from torchlight import torchlight\n'), ((2079, 2128), 'net.classifier.Classifier', 'classifier.Classifier', (['C', 'num_classes', 'graph_dict'], {}), '(C, num_classes, graph_dict)\n', (2100, 2128), False, 'from net import classifier\n'), ((2222, 2243), 'torch.nn.CrossEntropyLoss', 'nn.CrossEntropyLoss', ([], {}), '()\n', (2241, 2243), True, 'import torch.nn as nn\n'), ((5777, 5796), 'numpy.mean', 'np.mean', (['loss_value'], {}), '(loss_value)\n', (5784, 5796), True, 'import numpy as np\n'), ((6753, 6780), 'numpy.concatenate', 'np.concatenate', (['result_frag'], {}), '(result_frag)\n', (6767, 6780), True, 'import numpy as np\n'), ((9324, 9341), 'numpy.empty', 'np.empty', (['(0, 64)'], {}), '((0, 64))\n', (9332, 9341), True, 'import numpy as np\n'), ((9355, 9410), 'h5py.File', 'h5py.File', (["('../data/features' + ftype_real + '.h5')", '"""r"""'], {}), "('../data/features' + ftype_real + '.h5', 'r')\n", (9364, 9410), False, 'import h5py\n'), ((9420, 9476), 'h5py.File', 'h5py.File', (["('../data/features' + ftype_synth + '.h5')", '"""r"""'], {}), "('../data/features' + ftype_synth + '.h5', 'r')\n", (9429, 9476), False, 'import h5py\n'), ((9545, 9624), 'h5py.File', 'h5py.File', (["('../data/deepFeatures' + ftype_real + '+' + ftype_synth + '.h5')", '"""w"""'], {}), "('../data/deepFeatures' + ftype_real + '+' + ftype_synth + '.h5', 'w')\n", (9554, 9624), False, 'import h5py\n'), ((1982, 2015), 'os.path.isdir', 'os.path.isdir', (['self.args.work_dir'], {}), '(self.args.work_dir)\n', (1995, 2015), False, 'import os\n'), ((2029, 2057), 'os.mkdir', 'os.mkdir', (['self.args.work_dir'], {}), '(self.args.work_dir)\n', (2037, 2057), False, 'import os\n'), ((6829, 6855), 'numpy.concatenate', 'np.concatenate', (['label_frag'], {}), '(label_frag)\n', (6843, 6855), True, 'import numpy as np\n'), ((6899, 6918), 'numpy.mean', 'np.mean', (['loss_value'], {}), '(loss_value)\n', (6906, 6918), True, 'import numpy as np\n'), ((9283, 9303), 'torch.load', 'torch.load', (['filename'], {}), '(filename)\n', (9293, 9303), False, 'import torch\n'), ((9749, 9814), 'numpy.reshape', 'np.reshape', (['each_data', '(1, each_data.shape[0], joints, coords, 1)'], {}), '(each_data, (1, each_data.shape[0], joints, coords, 1))\n', (9759, 9814), True, 'import numpy as np\n'), ((9839, 9883), 'numpy.moveaxis', 'np.moveaxis', (['each_data', '[1, 2, 3]', '[2, 3, 1]'], {}), '(each_data, [1, 2, 3], [2, 3, 1])\n', (9850, 9883), True, 'import numpy as np\n'), ((10415, 10480), 'numpy.reshape', 'np.reshape', (['each_data', '(1, each_data.shape[0], joints, coords, 1)'], {}), '(each_data, (1, each_data.shape[0], joints, coords, 1))\n', (10425, 10480), True, 'import numpy as np\n'), ((10505, 10549), 'numpy.moveaxis', 'np.moveaxis', (['each_data', '[1, 2, 3]', '[2, 3, 1]'], {}), '(each_data, [1, 2, 3], [2, 3, 1])\n', (10516, 10549), True, 'import numpy as np\n'), ((2446, 2468), 'numpy.max', 'np.max', (['self.args.topk'], {}), '(self.args.topk)\n', (2452, 2468), True, 'import numpy as np\n'), ((6404, 6419), 'torch.no_grad', 'torch.no_grad', ([], {}), '()\n', (6417, 6419), False, 'import torch\n'), ((10004, 10019), 'torch.no_grad', 'torch.no_grad', ([], {}), '()\n', (10017, 10019), False, 'import torch\n'), ((10670, 10685), 'torch.no_grad', 'torch.no_grad', ([], {}), '()\n', (10683, 10685), False, 'import torch\n'), ((3346, 3372), 'numpy.array', 'np.array', (['self.step_epochs'], {}), '(self.step_epochs)\n', (3354, 3372), True, 'import numpy as np\n'), ((9908, 9935), 'torch.from_numpy', 'torch.from_numpy', (['each_data'], {}), '(each_data)\n', (9924, 9935), False, 'import torch\n'), ((10225, 10242), 'numpy.array', 'np.array', (['feature'], {}), '(feature)\n', (10233, 10242), True, 'import numpy as np\n'), ((10574, 10601), 'torch.from_numpy', 'torch.from_numpy', (['each_data'], {}), '(each_data)\n', (10590, 10601), False, 'import torch\n'), ((10892, 10909), 'numpy.array', 'np.array', (['feature'], {}), '(feature)\n', (10900, 10909), True, 'import numpy as np\n')] |
# Built in python libs
from typing import List, Tuple, Any
# Additional libs
import numpy as np
import cv2
from numba import jit, njit, prange
# Custom imports
# performs the ratio test on a set of matched keypoints
# the ratio test filters matched keypoints out if they are greater than the minimum seperation between matched
# by a set amount. The usual value is 3. Thus, if the distance between two matched features is 3 times larger than the
# minimum pixel distance between 2 matched features, than we can say it is invalid and in outlier.
# This is because the typical image to image comparision will have a small change between frames
# takes a list of keypoints, a minimum distance, and a given ratio
# kpMatches must be SORTED for optimization purposes
@jit(forceobj=True)
def ratioTest(kpMatches: np.ndarray, ratio: float) -> List:
if len(kpMatches) > 0:
minDist = kpMatches[0].distance
else:
minDist = 0.0
goodDistanceDiffs = []
for m in kpMatches:
if m.distance < ratio * minDist:
goodDistanceDiffs.append(m)
else:
break
return goodDistanceDiffs
@jit(forceobj=True)
def adaptiveRatioTest(kpMatches: np.ndarray, startingRatio: float, targetFeatureRatio: float, stepSize: float,
timeout=1000) -> List:
ratioPoints: List = list()
currentFeatureRatio = 0.0
counter = 0
while currentFeatureRatio < targetFeatureRatio:
ratioPoints = ratioTest(kpMatches, startingRatio + stepSize * counter)
currentFeatureRatio = len(ratioPoints) / len(kpMatches)
counter += 1
if counter > timeout:
break
# print(f"CurrentFeatureRatio: {currentFeatureRatio} with ratio: {startingRatio + (counter - 1) * stepSize}")
return ratioPoints
@jit(forceobj=True)
def getSrcDstPointsFromMatches(matchedKp, prevKp, currKp):
# (x, y) coordinates from the first image.
prevPts = np.float32([prevKp[m.trainIdx].pt for m in matchedKp]).reshape(-1, 1, 2)
# (x, y) coordinates from the second image.
currPts = np.float32([currKp[m.trainIdx].pt for m in matchedKp]).reshape(-1, 1, 2)
# converts to np.arrays
return np.array(prevPts), np.array(currPts)
# actually gets average X
@jit(nopython=True)
def getAvgCoordinate(array):
x_val_sum = 0
for element in array:
x_val_sum += element
return x_val_sum / len(array)
@jit(forceobj=True)
def getCoordinateAverage(array):
# Grabs only the first column (x values)
avgX = getAvgCoordinate(array[:, 0][:, 0])
# Grabs only the second column (y values)
avgY = getAvgCoordinate(array[0, :][0, :])
return avgX, avgY
@jit(forceobj=True)
def getTranslationXY(matchedKp: np.ndarray, prevKp: np.ndarray, currKp: np.ndarray) -> Tuple[float, float]:
prevPts, currPts = getSrcDstPointsFromMatches(matchedKp, prevKp, currKp)
prevAvgX, prevAvgY = getCoordinateAverage(prevPts)
currAvgX, currAvgY = getCoordinateAverage(currPts)
transX = currAvgX - prevAvgX
transY = currAvgY - prevAvgY
return transX, transY
def getAvgTranslationXY(leftMatches: np.ndarray, prevLeftKp: np.ndarray, leftKp: np.ndarray, rightMatches: np.ndarray,
prevRightKp: np.ndarray, rightKp: np.ndarray) -> Tuple[float, float]:
leftX, leftY = 0.0, 0.0
rightX, rightY = 0.0, 0.0
numErrors = 0
try:
leftX, leftY = getTranslationXY(leftMatches, prevLeftKp, leftKp)
except IndexError:
numErrors += 1
except ZeroDivisionError:
numErrors += 1
try:
rightX, rightY = getTranslationXY(rightMatches, prevRightKp, rightKp)
except IndexError:
numErrors += 1
except ZeroDivisionError:
numErrors += 1
if numErrors >= 2:
return 0.0, 0.0
return (leftX + rightX) / (2 - numErrors), (leftY + rightY) / (2 - numErrors)
| [
"numpy.array",
"numpy.float32",
"numba.jit"
] | [((768, 786), 'numba.jit', 'jit', ([], {'forceobj': '(True)'}), '(forceobj=True)\n', (771, 786), False, 'from numba import jit, njit, prange\n'), ((1142, 1160), 'numba.jit', 'jit', ([], {'forceobj': '(True)'}), '(forceobj=True)\n', (1145, 1160), False, 'from numba import jit, njit, prange\n'), ((1798, 1816), 'numba.jit', 'jit', ([], {'forceobj': '(True)'}), '(forceobj=True)\n', (1801, 1816), False, 'from numba import jit, njit, prange\n'), ((2250, 2268), 'numba.jit', 'jit', ([], {'nopython': '(True)'}), '(nopython=True)\n', (2253, 2268), False, 'from numba import jit, njit, prange\n'), ((2408, 2426), 'numba.jit', 'jit', ([], {'forceobj': '(True)'}), '(forceobj=True)\n', (2411, 2426), False, 'from numba import jit, njit, prange\n'), ((2670, 2688), 'numba.jit', 'jit', ([], {'forceobj': '(True)'}), '(forceobj=True)\n', (2673, 2688), False, 'from numba import jit, njit, prange\n'), ((2184, 2201), 'numpy.array', 'np.array', (['prevPts'], {}), '(prevPts)\n', (2192, 2201), True, 'import numpy as np\n'), ((2203, 2220), 'numpy.array', 'np.array', (['currPts'], {}), '(currPts)\n', (2211, 2220), True, 'import numpy as np\n'), ((1937, 1991), 'numpy.float32', 'np.float32', (['[prevKp[m.trainIdx].pt for m in matchedKp]'], {}), '([prevKp[m.trainIdx].pt for m in matchedKp])\n', (1947, 1991), True, 'import numpy as np\n'), ((2072, 2126), 'numpy.float32', 'np.float32', (['[currKp[m.trainIdx].pt for m in matchedKp]'], {}), '([currKp[m.trainIdx].pt for m in matchedKp])\n', (2082, 2126), True, 'import numpy as np\n')] |
import numpy
from distutils.core import setup
from Cython.Build import cythonize
setup(
name='features_labels',
ext_modules=cythonize('features_labels.pyx', include_dirs=[numpy.get_include()])
)
| [
"numpy.get_include"
] | [((180, 199), 'numpy.get_include', 'numpy.get_include', ([], {}), '()\n', (197, 199), False, 'import numpy\n')] |
"""
Script for extracting particles from membranes by looking for low pass filtering + non-maximum suppression
Input: - Directory with the _imod.csv files with reference picked particles for each microsome
- Directory with offsets for each microsome in a tomogram
- Directory with the density maps
Output: - A STAR file and a list with the coordinates pickled per microsome
- Additional files for visualization
"""
__author__ = '<NAME>'
import pyseg as ps
import scipy as sp
########################################################################################
# GLOBAL VARIABLES
########################################################################################
MB_LBL = 1
########################################################################################
# PARAMETERS
########################################################################################
ROOT_PATH = '/fs/pool/pool-lucic2'
# Input STAR files
in_dir = ROOT_PATH + '/antonio/pick_test/cont/160614' # '/johannes/tomograms/stefan/090614/reconstruct/coords'
in_dir_seg = ROOT_PATH + '/johannes/tomograms/stefan/160614/graph_ribo'
in_dir_den = ROOT_PATH + '/antonio/pick_test/tm/cc/tomos_nobeads/160614'
in_ctf = ROOT_PATH + '/antonio/pick_test/recontruct/ctf/CTF_model_160.mrc'
####### Output data
out_dir = ROOT_PATH + '/antonio/pick_test/pick/den/160614'
out_dir_parts = ROOT_PATH + '/antonio/pick_test/recontruct/den/160614/protein_160'
###### Low pass filter settings
lp_res = 1.048 # nm
lp_sg = 0.8 # Gaussian: min_feat=(1/fc), fc=Fs/(2*pi*sg), Fs=1/vx_size
###### Peaks configuration
peak_side = 2 # 3
peak_dst_rg = (0, 20) # vx # None
###### Advanced peaks configuration
peak_prop_pt = 'pt_normal'
peak_prop_norm = 'smb_normal'
peak_prop_rot = 'norm_rot'
########################################################################################
# MAIN ROUTINE
########################################################################################
################# Package import
import os
import gc
import csv
import time
import math
import numpy as np
from pyseg.sub import TomoPeaks
from skimage.feature import peak_local_max
from pyorg.surf import points_to_poly
########## Global variables
########## Print initial message
print('Extracting particles randomly within segmented membranes.')
print('\tAuthor: ' + __author__)
print('\tDate: ' + time.strftime("%c") + '\n')
print('Options:')
print('\tDirectory for reference picked particles: ' + str(in_dir))
print('\tDirectory for microsomes offset per tomogram: ' + str(in_dir_seg))
print('\tDirectory for density maps: ' + str(in_dir_den))
print('\tOutput directory: ' + str(out_dir))
print('\tLow pass filter settings (Gaussian):')
print('\t\t-Voxel size (nm/vx): ' + str(lp_res))
print('\t\t-Gaussian sigma (nm): ' + str(lp_sg))
sampling_rate = 1. / lp_res
print('\t\t-Sampling rate (1/nm): ' + str(sampling_rate))
freq_cutoff = sampling_rate / (2.*np.pi*lp_sg)
print('\t\t-Cut-off frequency at level 0.5 (1/nm): ' + str(freq_cutoff))
min_feature = 1. / freq_cutoff
print('\t\t-Min. feature size (nm): ' + str(min_feature))
print('\tPeaks configuration:')
print('\t\t-Peak side: ' + str(peak_side))
if peak_dst_rg is None:
print('\t\t-Distances range: [0, 0]')
else:
print('\t\t-Distances range: (' + str(peak_dst_rg[0]) + ', ' + str(peak_dst_rg[1]) + ')')
print('')
######### Process
print('\tSearching for *_imod.csv files in input folder...')
ref_files = list()
for fname in os.listdir(in_dir):
if fname.endswith('_imod.csv'):
ref_files.append(os.path.split(fname)[1])
print('\t\t-File found: ' + ref_files[-1])
print('\tPairing files found with their segmentation offset...')
mic_dic = dict()
for fname in ref_files:
names = fname.split('_')
fname_off = in_dir_seg + '/' + names[0] + '_' + names[1] + '_mb_graph.star'
ves_id = names[3]
print('\t\t-Tomogram offset found: ' + fname_off)
star = ps.sub.Star()
star.load(fname_off)
for row in range(star.get_nrows()):
path_seg = star.get_element('_psSegImage', row)
path_ref = star.get_element('_rlnMicrographName', row)
fname_seg = os.path.split(path_seg)[1]
names_seg = fname_seg.split('_')
if ves_id == names_seg[3]:
offx = star.get_element('_psSegOffX', row)
offy = star.get_element('_psSegOffY', row)
offz = star.get_element('_psSegOffZ', row)
n_parts = sum(1 for line in open(in_dir + '/' + fname))
mic_dic[fname] = (path_ref, path_seg, offx, offy, offz, n_parts)
print('\t\t\t-Offset found for microsome ' + fname + ' with ' + str(n_parts) + ' particles.')
print('\tPreparing output particles STAR file...')
star_parts = ps.sub.Star()
star_parts.add_column('_rlnMicrographName')
star_parts.add_column('_rlnImageName')
star_parts.add_column('_rlnCtfImage')
star_parts.add_column('_rlnCoordinateX')
star_parts.add_column('_rlnCoordinateY')
star_parts.add_column('_rlnCoordinateZ')
star_parts.add_column('_rlnAngleRot')
star_parts.add_column('_rlnAngleTilt')
star_parts.add_column('_rlnAnglePsi')
part_row = 0
print('\tPICKING LOOP:')
for in_mic in mic_dic.keys():
path_seg, path_ref = mic_dic[in_mic][1], mic_dic[in_mic][0]
names = (os.path.splitext(os.path.split(path_seg)[1])[0]).split('_')
stem_mic = names[0] + '_' + names[1] + '_' + names[2] + '_' + names[3]
print('\t-Processing stem: ' + stem_mic)
path_cc = in_dir_den + '/' + stem_mic + '.mrc'
print('\t\t-Loading corresponding density map: ' + path_cc)
try:
tomo_cc = ps.disperse_io.load_tomo(path_cc, mmap=True)
except IOError:
print('\t\t\t-WARNING: the corresponding CC map cannot be loaded, continuing...')
gc.collect()
continue
print('\t\t-Low pass filtering...')
tomo_cc = sp.ndimage.filters.gaussian_filter(ps.globals.lin_map(tomo_cc, lb=1, ub=0), lp_sg)
print('\t\t-Loading microsome segmentation file: ' + path_seg)
seg = ps.disperse_io.load_tomo(path_seg, mmap=True)
mb_mask, seg_mask = seg == MB_LBL, seg == peak_side
del seg
n_samp = mic_dic[in_mic][5]
if peak_dst_rg is not None:
mb_dsts = sp.ndimage.morphology.distance_transform_edt(~mb_mask)
mb_mask = (mb_dsts > peak_dst_rg[0]) & (mb_dsts < peak_dst_rg[1])
del mb_dsts
mb_mask *= seg_mask
seg_mask[mb_mask] = False
# ps.disperse_io.save_numpy(mb_mask, out_dir + '/hold1.mrc')
# ps.disperse_io.save_numpy(seg_mask, out_dir + '/hold2.mrc')
print('\t\t-Finding peaks (local maxima) in the cross-correlation map: ')
tomo_peaks = peak_local_max(tomo_cc, min_distance=int(math.ceil(min_feature)), indices=False)
# tomo_peaks = ski.morphology.local_maxima(tomo_cc, indices=False)
peaks = np.where(tomo_peaks * mb_mask)
n_peaks = len(peaks[0])
hold_peaks = n_samp
if n_peaks < hold_peaks:
hold_peaks = n_peaks
print('\t\t\t-Number of peaks found ' + str(n_peaks) + ', ' + str(hold_peaks) + ' are going to be picked.')
peaks_cc = np.zeros(shape=n_peaks, dtype=np.float)
for i in range(n_peaks):
peaks_cc[i] = tomo_cc[peaks[0][i], peaks[1][i], peaks[2][i]]
peaks_sorted = np.argsort(peaks_cc)[::-1]
coords = list()
for id in peaks_sorted[:hold_peaks]:
coords.append((peaks[0][id], peaks[1][id], peaks[2][id]))
print('\t\t-Creating the peaks container...')
out_seg = out_dir + '/' + stem_mic + '_mb.mrc'
tomo_peaks = TomoPeaks(shape=mb_mask.shape, name=out_seg, mask=mb_mask)
tomo_peaks.add_peaks(coords)
tomo_peaks.seg_shortest_pt(seg_mask, peak_prop_pt)
print('\t\t\t-Number of peaks found: ' + str(tomo_peaks.get_num_peaks()))
out_imod_csv = out_dir + '/' + stem_mic + '_den_imod.csv'
if not os.path.exists(out_imod_csv):
print('\t\t-Creating output IMOD CSV file: ' + out_imod_csv)
with open(out_imod_csv, 'w') as imod_csv_file:
writer = csv.DictWriter(imod_csv_file, dialect=csv.excel_tab, fieldnames=('X', 'Y', 'Z'))
out_rln_coords = out_dir + '/' + stem_mic + '_den_rln.coords'
if not os.path.exists(out_rln_coords):
print('\t\t-Creating output RELION COORDS file: ' + out_rln_coords)
with open(out_rln_coords, 'w') as rln_coords_file:
writer = csv.DictWriter(rln_coords_file, dialect=csv.excel_tab, fieldnames=('X', 'Y', 'Z', 'Rho', 'Tilt', 'Psi'))
print('\t\tParticles loop..')
part_seg_row = 1
coords_noff, normals_imod = list(), list()
gcrop_off = mic_dic[in_mic][2], mic_dic[in_mic][3], mic_dic[in_mic][4]
gcrop_off_rln = np.asarray((gcrop_off[1], gcrop_off[0], gcrop_off[2]), dtype=np.float32)
coords, coords_pt = tomo_peaks.get_prop_vals(ps.sub.PK_COORDS), tomo_peaks.get_prop_vals(peak_prop_pt)
for coord, pt_coord in zip(coords, coords_pt):
# Coordinate transformation for IMOD
coords_noff.append(coord)
coord_imod, pt_coord_imod = coord + gcrop_off, pt_coord + gcrop_off
vec_imod = pt_coord_imod - coord_imod
hold_norm = math.sqrt((vec_imod * vec_imod).sum())
if hold_norm <= 0:
vec_imod = np.asarray((0., 0., 0.))
else:
vec_imod /= hold_norm
normals_imod.append(vec_imod)
out_imod_csv = out_dir + '/' + stem_mic + '_den_imod.csv'
with open(out_imod_csv, 'a') as imod_csv_file:
writer = csv.DictWriter(imod_csv_file, dialect=csv.excel_tab, fieldnames=('X', 'Y', 'Z'))
writer.writerow({'X':coord_imod[0], 'Y':coord_imod[1], 'Z':coord_imod[2]})
# Coordinate transformation for RELION
coord_rln = np.asarray((coord[1], coord[0], coord[2]), dtype=np.float32)
pt_coord_rln = np.asarray((pt_coord[1], pt_coord[0], pt_coord[2]), dtype=np.float32)
coord_rln, pt_coord_rln = coord_rln + gcrop_off_rln, pt_coord_rln + gcrop_off_rln
vec_rln = pt_coord_rln - coord_rln
# hold_norm = math.sqrt((vec_rln * vec_rln).sum())
# if hold_norm <= 0:
# vec_rln = np.asarray((0., 0., 0.))
# else:
# vec_rln /= hold_norm
rho, tilt, psi = ps.globals.vect_to_zrelion(vec_rln)
out_rln_coords = out_dir + '/' + stem_mic + '_den_rln.coords'
with open(out_rln_coords, 'a') as rln_coords_file:
writer = csv.DictWriter(rln_coords_file, dialect=csv.excel_tab, fieldnames=('X', 'Y', 'Z', 'Rho', 'Tilt', 'Psi'))
writer.writerow({'X':coord_rln[0], 'Y':coord_rln[1], 'Z':coord_rln[2], 'Rho':rho, 'Tilt':tilt, 'Psi':psi})
part_path = out_dir_parts + '/' + stem_mic + '_' + str(part_seg_row) + '.mrc'
star_row = {'_rlnMicrographName':path_seg, '_rlnImageName':part_path, '_rlnCtfImage':in_ctf,
'_rlnCoordinateX':coord_rln[0], '_rlnCoordinateY':coord_rln[1], '_rlnCoordinateZ':coord_rln[2],
'_rlnAngleRot':rho, '_rlnAngleTilt':tilt, '_rlnAnglePsi':psi}
star_parts.add_row(**star_row)
part_row += 1
part_seg_row += 1
out_vtp = out_dir + '/' + stem_mic + '.vtp'
print('\t\t-Storing the vtp file: ' + out_vtp)
coords_vtp = points_to_poly(coords_noff, normals=normals_imod, n_name='n_normal')
ps.disperse_io.save_vtp(coords_vtp, out_vtp)
gc.collect()
print('\tNumber of particles found: ' + str(part_row))
out_star = out_dir + '/particles_160614_den.star'
print('\tStoring particles STAR file in: ' + out_star)
star_parts.store(out_star)
print('Terminated. (' + time.strftime("%c") + ')') | [
"pyseg.globals.lin_map",
"time.strftime",
"numpy.argsort",
"gc.collect",
"pyseg.globals.vect_to_zrelion",
"pyseg.sub.Star",
"csv.DictWriter",
"pyseg.sub.TomoPeaks",
"os.path.exists",
"pyseg.disperse_io.load_tomo",
"pyorg.surf.points_to_poly",
"math.ceil",
"numpy.asarray",
"scipy.ndimage.mo... | [((3509, 3527), 'os.listdir', 'os.listdir', (['in_dir'], {}), '(in_dir)\n', (3519, 3527), False, 'import os\n'), ((4771, 4784), 'pyseg.sub.Star', 'ps.sub.Star', ([], {}), '()\n', (4782, 4784), True, 'import pyseg as ps\n'), ((3969, 3982), 'pyseg.sub.Star', 'ps.sub.Star', ([], {}), '()\n', (3980, 3982), True, 'import pyseg as ps\n'), ((6023, 6068), 'pyseg.disperse_io.load_tomo', 'ps.disperse_io.load_tomo', (['path_seg'], {'mmap': '(True)'}), '(path_seg, mmap=True)\n', (6047, 6068), True, 'import pyseg as ps\n'), ((6830, 6860), 'numpy.where', 'np.where', (['(tomo_peaks * mb_mask)'], {}), '(tomo_peaks * mb_mask)\n', (6838, 6860), True, 'import numpy as np\n'), ((7098, 7137), 'numpy.zeros', 'np.zeros', ([], {'shape': 'n_peaks', 'dtype': 'np.float'}), '(shape=n_peaks, dtype=np.float)\n', (7106, 7137), True, 'import numpy as np\n'), ((7528, 7586), 'pyseg.sub.TomoPeaks', 'TomoPeaks', ([], {'shape': 'mb_mask.shape', 'name': 'out_seg', 'mask': 'mb_mask'}), '(shape=mb_mask.shape, name=out_seg, mask=mb_mask)\n', (7537, 7586), False, 'from pyseg.sub import TomoPeaks\n'), ((8651, 8723), 'numpy.asarray', 'np.asarray', (['(gcrop_off[1], gcrop_off[0], gcrop_off[2])'], {'dtype': 'np.float32'}), '((gcrop_off[1], gcrop_off[0], gcrop_off[2]), dtype=np.float32)\n', (8661, 8723), True, 'import numpy as np\n'), ((11181, 11249), 'pyorg.surf.points_to_poly', 'points_to_poly', (['coords_noff'], {'normals': 'normals_imod', 'n_name': '"""n_normal"""'}), "(coords_noff, normals=normals_imod, n_name='n_normal')\n", (11195, 11249), False, 'from pyorg.surf import points_to_poly\n'), ((11254, 11298), 'pyseg.disperse_io.save_vtp', 'ps.disperse_io.save_vtp', (['coords_vtp', 'out_vtp'], {}), '(coords_vtp, out_vtp)\n', (11277, 11298), True, 'import pyseg as ps\n'), ((11304, 11316), 'gc.collect', 'gc.collect', ([], {}), '()\n', (11314, 11316), False, 'import gc\n'), ((5614, 5658), 'pyseg.disperse_io.load_tomo', 'ps.disperse_io.load_tomo', (['path_cc'], {'mmap': '(True)'}), '(path_cc, mmap=True)\n', (5638, 5658), True, 'import pyseg as ps\n'), ((5897, 5936), 'pyseg.globals.lin_map', 'ps.globals.lin_map', (['tomo_cc'], {'lb': '(1)', 'ub': '(0)'}), '(tomo_cc, lb=1, ub=0)\n', (5915, 5936), True, 'import pyseg as ps\n'), ((6220, 6274), 'scipy.ndimage.morphology.distance_transform_edt', 'sp.ndimage.morphology.distance_transform_edt', (['(~mb_mask)'], {}), '(~mb_mask)\n', (6264, 6274), True, 'import scipy as sp\n'), ((7255, 7275), 'numpy.argsort', 'np.argsort', (['peaks_cc'], {}), '(peaks_cc)\n', (7265, 7275), True, 'import numpy as np\n'), ((7827, 7855), 'os.path.exists', 'os.path.exists', (['out_imod_csv'], {}), '(out_imod_csv)\n', (7841, 7855), False, 'import os\n'), ((8160, 8190), 'os.path.exists', 'os.path.exists', (['out_rln_coords'], {}), '(out_rln_coords)\n', (8174, 8190), False, 'import os\n'), ((9682, 9742), 'numpy.asarray', 'np.asarray', (['(coord[1], coord[0], coord[2])'], {'dtype': 'np.float32'}), '((coord[1], coord[0], coord[2]), dtype=np.float32)\n', (9692, 9742), True, 'import numpy as np\n'), ((9766, 9835), 'numpy.asarray', 'np.asarray', (['(pt_coord[1], pt_coord[0], pt_coord[2])'], {'dtype': 'np.float32'}), '((pt_coord[1], pt_coord[0], pt_coord[2]), dtype=np.float32)\n', (9776, 9835), True, 'import numpy as np\n'), ((10182, 10217), 'pyseg.globals.vect_to_zrelion', 'ps.globals.vect_to_zrelion', (['vec_rln'], {}), '(vec_rln)\n', (10208, 10217), True, 'import pyseg as ps\n'), ((2411, 2430), 'time.strftime', 'time.strftime', (['"""%c"""'], {}), "('%c')\n", (2424, 2430), False, 'import time\n'), ((4187, 4210), 'os.path.split', 'os.path.split', (['path_seg'], {}), '(path_seg)\n', (4200, 4210), False, 'import os\n'), ((5777, 5789), 'gc.collect', 'gc.collect', ([], {}), '()\n', (5787, 5789), False, 'import gc\n'), ((8002, 8087), 'csv.DictWriter', 'csv.DictWriter', (['imod_csv_file'], {'dialect': 'csv.excel_tab', 'fieldnames': "('X', 'Y', 'Z')"}), "(imod_csv_file, dialect=csv.excel_tab, fieldnames=('X', 'Y', 'Z')\n )\n", (8016, 8087), False, 'import csv\n'), ((8348, 8456), 'csv.DictWriter', 'csv.DictWriter', (['rln_coords_file'], {'dialect': 'csv.excel_tab', 'fieldnames': "('X', 'Y', 'Z', 'Rho', 'Tilt', 'Psi')"}), "(rln_coords_file, dialect=csv.excel_tab, fieldnames=('X', 'Y',\n 'Z', 'Rho', 'Tilt', 'Psi'))\n", (8362, 8456), False, 'import csv\n'), ((9193, 9220), 'numpy.asarray', 'np.asarray', (['(0.0, 0.0, 0.0)'], {}), '((0.0, 0.0, 0.0))\n', (9203, 9220), True, 'import numpy as np\n'), ((9446, 9531), 'csv.DictWriter', 'csv.DictWriter', (['imod_csv_file'], {'dialect': 'csv.excel_tab', 'fieldnames': "('X', 'Y', 'Z')"}), "(imod_csv_file, dialect=csv.excel_tab, fieldnames=('X', 'Y', 'Z')\n )\n", (9460, 9531), False, 'import csv\n'), ((10368, 10476), 'csv.DictWriter', 'csv.DictWriter', (['rln_coords_file'], {'dialect': 'csv.excel_tab', 'fieldnames': "('X', 'Y', 'Z', 'Rho', 'Tilt', 'Psi')"}), "(rln_coords_file, dialect=csv.excel_tab, fieldnames=('X', 'Y',\n 'Z', 'Rho', 'Tilt', 'Psi'))\n", (10382, 10476), False, 'import csv\n'), ((11530, 11549), 'time.strftime', 'time.strftime', (['"""%c"""'], {}), "('%c')\n", (11543, 11549), False, 'import time\n'), ((3590, 3610), 'os.path.split', 'os.path.split', (['fname'], {}), '(fname)\n', (3603, 3610), False, 'import os\n'), ((6707, 6729), 'math.ceil', 'math.ceil', (['min_feature'], {}), '(min_feature)\n', (6716, 6729), False, 'import math\n'), ((5308, 5331), 'os.path.split', 'os.path.split', (['path_seg'], {}), '(path_seg)\n', (5321, 5331), False, 'import os\n')] |
import napari
import os
import shutil
import squidpy as sq
from ngff_tables_prototype.writer import write_spatial_anndata
from ngff_tables_prototype.reader import load_to_napari_viewer
import numpy as np
output_fpath = "test_segment.zarr"
def write_segmentation_adata() -> None:
adata = sq.datasets.mibitof()
lib_id = "point8"
spatial_key = "spatial"
adata = adata[adata.obs.library_id == lib_id].copy()
image = adata.uns[spatial_key][lib_id]["images"]["hires"]
segment = adata.uns[spatial_key][lib_id]["images"]["segmentation"]
adata.X = adata.X.A.copy()
tables_adata = adata.copy()
if True:
if os.path.isdir(output_fpath):
shutil.rmtree(output_fpath)
if not os.path.isdir(output_fpath):
write_spatial_anndata(
file_path=output_fpath,
image_axes=["c", "y", "x"],
image=np.swapaxes(image, 2, 0),
label_image=segment,
tables_adata=tables_adata,
tables_region="labels/label_image",
tables_instance_key="cell_id",
)
return
if __name__ == "__main__":
write_segmentation_adata()
viewer = load_to_napari_viewer(
file_path=output_fpath,
groups=["labels/label_image", "tables/regions_table"],
)
napari.run()
| [
"os.path.isdir",
"squidpy.datasets.mibitof",
"numpy.swapaxes",
"ngff_tables_prototype.reader.load_to_napari_viewer",
"shutil.rmtree",
"napari.run"
] | [((294, 315), 'squidpy.datasets.mibitof', 'sq.datasets.mibitof', ([], {}), '()\n', (313, 315), True, 'import squidpy as sq\n'), ((1163, 1267), 'ngff_tables_prototype.reader.load_to_napari_viewer', 'load_to_napari_viewer', ([], {'file_path': 'output_fpath', 'groups': "['labels/label_image', 'tables/regions_table']"}), "(file_path=output_fpath, groups=['labels/label_image',\n 'tables/regions_table'])\n", (1184, 1267), False, 'from ngff_tables_prototype.reader import load_to_napari_viewer\n'), ((1291, 1303), 'napari.run', 'napari.run', ([], {}), '()\n', (1301, 1303), False, 'import napari\n'), ((644, 671), 'os.path.isdir', 'os.path.isdir', (['output_fpath'], {}), '(output_fpath)\n', (657, 671), False, 'import os\n'), ((725, 752), 'os.path.isdir', 'os.path.isdir', (['output_fpath'], {}), '(output_fpath)\n', (738, 752), False, 'import os\n'), ((685, 712), 'shutil.rmtree', 'shutil.rmtree', (['output_fpath'], {}), '(output_fpath)\n', (698, 712), False, 'import shutil\n'), ((879, 903), 'numpy.swapaxes', 'np.swapaxes', (['image', '(2)', '(0)'], {}), '(image, 2, 0)\n', (890, 903), True, 'import numpy as np\n')] |
from __future__ import division
import numpy as np
#----------------------------- COSMOS galaxy bias (arXiv:1205.1064) ----------------------------------------
def bias_Amara(z, bias_type, zcutoff):
y = z/(1+z)
ycutoff = zcutoff/(1+zcutoff)
#bias_type --> nn:fifth nearest neighbor or gs: gaussian smoothing
if(bias_type=='nn'): b0=1.25; f=-1.4
if(bias_type=='gs'): b0=0.7 ; f=-1.615
bmax = b0/(1 + (f*ycutoff) )
if(z<=zcutoff): return b0/(1+ (f*y))
if(z>zcutoff): return bmax
#----------------- Galaxy bias (1.> arXiv:1101.2453 (pg. 4), 2.> arXiv:0810.0003 (pg. 7, Table 3)) ---------
def bias_Porto(z):
return np.sqrt(1+z) | [
"numpy.sqrt"
] | [((629, 643), 'numpy.sqrt', 'np.sqrt', (['(1 + z)'], {}), '(1 + z)\n', (636, 643), True, 'import numpy as np\n')] |
import os
import json
from collections import OrderedDict
import numpy as np
import tensorflow as tf
cur_path = os.path.realpath(__file__)
ROOT_PATH = os.path.dirname(cur_path)
# add any new ops under the following
pose_to_heatmap_fn = tf.load_op_library(
os.path.join(ROOT_PATH, 'pose_to_heatmap.so')).pose_to_heatmap
zero_out_channels_fn = tf.load_op_library(
os.path.join(ROOT_PATH, 'zero_out_channels.so')).zero_out_channels
render_pose_fn = tf.load_op_library(
os.path.join(ROOT_PATH, 'render_pose.so')).render_pose
render_objects_fn = tf.load_op_library(
os.path.join(ROOT_PATH, 'render_objects.so')).render_objects
def pose_to_heatmap(*args, **kwargs):
with tf.variable_scope('pose_to_heatmap_pyWrapper'):
pose_img, pose_valid = pose_to_heatmap_fn(*args, **kwargs)
out_channels = kwargs['out_channels']
pose_img.set_shape((None, None, out_channels))
pose_valid.set_shape((out_channels,))
pose_img *= 255.0
pose_img = tf.cast(pose_img, tf.uint8)
return pose_img, pose_valid
def zero_out_channels(*args, **kwargs):
with tf.variable_scope('zero_out_channels_pyWrapper'):
return zero_out_channels_fn(*args, **kwargs)
def render_pose(*args, **kwargs):
with tf.variable_scope('render_pose_pyWrapper'):
out_channels = 3
if kwargs['out_type'] == 'rgb':
kwargs['out_type'] = 1
out_channels = 3
elif kwargs['out_type'] == 'split-channel':
kwargs['out_type'] = 2
out_channels = 18 # number of limbs
img = render_pose_fn(*args, **kwargs)
img *= 255.0
img = tf.cast(img, tf.uint8)
img.set_shape((None, None, out_channels))
return img
# from render_pose.cc
mpii_to_coco = OrderedDict([
(9, 0),
(8, 1),
(12, 2),
(11, 3),
(10, 4),
(13, 5),
(14, 6),
(15, 7),
(2, 8),
(1, 9),
(0, 10),
(3, 11),
(4, 12),
(5, 13),
])
def read_json_pose_fn(fpath):
try:
with open(fpath, 'r') as fin:
data = json.load(fin)
except:
print('Unable to open file {}'.format(fpath))
return -np.ones((16*3,)).astype('int64')
res = []
for body in data['bodies']:
mpii_joints = -np.ones((16, 3))
joints = np.array(body['joints'])
joints = np.reshape(joints, (-1, 3))
joints[joints[..., :] <= 0] = -1
mpii_joints[np.array(mpii_to_coco.keys()), :] = \
joints[np.array(mpii_to_coco.values()), :]
res += mpii_joints.reshape((-1,)).tolist()
res = np.array(res).astype('int64')
return res
def read_json_pose(*args):
return tf.py_func(read_json_pose_fn, args, tf.int64)
def render_objects(*args, **kwargs):
with tf.variable_scope('render_objects_pyWrapper'):
img = render_objects_fn(*args, **kwargs)
img *= 255.0
img = tf.cast(img, tf.uint8)
img.set_shape((None, None, kwargs['out_channels']))
return img
def extract_glimpse(image, pose_label, orig_im_ht, orig_im_wd,
out_side, pad_ratio, parts_keep):
# pose label is a [3x16xn,] vector
# for now just take the first pose and crop out the human
with tf.name_scope('ExtractGlimpse'):
pose_label = pose_label[:16*3]
pose_label = tf.reshape(pose_label, [16, 3])
if len(parts_keep) > 0:
pose_label = tf.gather(pose_label, parts_keep)
if len(parts_keep) == 1:
# now only one point, but need at least two to make a crop region
delta = tf.to_int64(
[tf.to_float(tf.shape(image)[-2]) * 0.1,
tf.to_float(tf.shape(image)[-3]) * 0.1, 0])
pose_label = tf.stack([
pose_label[0] - delta, pose_label[0] + delta])
pose_label_x = tf.to_float(pose_label[:, 0]) * \
tf.to_float(tf.shape(image)[-2]) / tf.to_float(orig_im_wd)
pose_label_y = tf.to_float(pose_label[:, 1]) * \
tf.to_float(tf.shape(image)[-3]) / tf.to_float(orig_im_ht)
pose_label = tf.stack([pose_label_y, pose_label_x])
mx_pts = tf.to_int32(tf.reduce_max(pose_label, axis=1))
mn_pts = tf.to_int32(tf.reduce_min(
tf.where(tf.greater_equal(pose_label, 0), pose_label,
tf.ones(pose_label.get_shape()) * 999999), axis=1))
delta_0 = tf.to_int32(tf.to_float((mx_pts[0] - mn_pts[0])) * pad_ratio)
delta_1 = tf.to_int32(tf.to_float((mx_pts[1] - mn_pts[1])) * pad_ratio)
mx_pts = mx_pts + [delta_0, delta_1]
mn_pts = mn_pts - [delta_0, delta_1]
offset_ht = tf.maximum(mn_pts[0], 0)
offset_wd = tf.maximum(mn_pts[1], 0)
target_ht = tf.minimum(mx_pts[0]-offset_ht, tf.shape(image)[-3]-offset_ht-1)
target_wd = tf.minimum(mx_pts[1]-offset_wd, tf.shape(image)[-2]-offset_wd-1)
# image = tf.Print(image, [offset_ht, offset_wd, target_ht, target_wd,
# tf.shape(image)], "stuff:")
image = tf.cond(tf.logical_and(
tf.greater(mx_pts[1], mn_pts[1]),
tf.greater(mx_pts[0], mn_pts[0])),
lambda: tf.image.crop_to_bounding_box(
image, offset_ht, offset_wd, target_ht, target_wd),
lambda: image)
if out_side > 0:
image = tf.image.resize_images(
image, [out_side, out_side])
return image
def read_sparse_label_fn(sparse_label, nclasses):
"""sparse_label is a string and return a 1D vector with the dense label
"""
res = np.zeros((nclasses,), dtype='int32')
res[np.array([int(el.split(':')[0]) for el in sparse_label.split(',')])] = \
np.array([int(el.split(':')[1]) for el in sparse_label.split(',')])
res[res < 0] = 0 # get rid of -1 label for now
return res
def read_sparse_label(*args):
return tf.py_func(read_sparse_label_fn, args, tf.int32)
| [
"tensorflow.maximum",
"tensorflow.reshape",
"numpy.ones",
"tensorflow.greater_equal",
"tensorflow.reduce_max",
"tensorflow.greater",
"os.path.join",
"tensorflow.image.crop_to_bounding_box",
"tensorflow.gather",
"os.path.dirname",
"tensorflow.variable_scope",
"tensorflow.stack",
"tensorflow.c... | [((113, 139), 'os.path.realpath', 'os.path.realpath', (['__file__'], {}), '(__file__)\n', (129, 139), False, 'import os\n'), ((152, 177), 'os.path.dirname', 'os.path.dirname', (['cur_path'], {}), '(cur_path)\n', (167, 177), False, 'import os\n'), ((1670, 1810), 'collections.OrderedDict', 'OrderedDict', (['[(9, 0), (8, 1), (12, 2), (11, 3), (10, 4), (13, 5), (14, 6), (15, 7), (2, \n 8), (1, 9), (0, 10), (3, 11), (4, 12), (5, 13)]'], {}), '([(9, 0), (8, 1), (12, 2), (11, 3), (10, 4), (13, 5), (14, 6), (\n 15, 7), (2, 8), (1, 9), (0, 10), (3, 11), (4, 12), (5, 13)])\n', (1681, 1810), False, 'from collections import OrderedDict\n'), ((2472, 2517), 'tensorflow.py_func', 'tf.py_func', (['read_json_pose_fn', 'args', 'tf.int64'], {}), '(read_json_pose_fn, args, tf.int64)\n', (2482, 2517), True, 'import tensorflow as tf\n'), ((5140, 5176), 'numpy.zeros', 'np.zeros', (['(nclasses,)'], {'dtype': '"""int32"""'}), "((nclasses,), dtype='int32')\n", (5148, 5176), True, 'import numpy as np\n'), ((5433, 5481), 'tensorflow.py_func', 'tf.py_func', (['read_sparse_label_fn', 'args', 'tf.int32'], {}), '(read_sparse_label_fn, args, tf.int32)\n', (5443, 5481), True, 'import tensorflow as tf\n'), ((260, 305), 'os.path.join', 'os.path.join', (['ROOT_PATH', '"""pose_to_heatmap.so"""'], {}), "(ROOT_PATH, 'pose_to_heatmap.so')\n", (272, 305), False, 'import os\n'), ((368, 415), 'os.path.join', 'os.path.join', (['ROOT_PATH', '"""zero_out_channels.so"""'], {}), "(ROOT_PATH, 'zero_out_channels.so')\n", (380, 415), False, 'import os\n'), ((474, 515), 'os.path.join', 'os.path.join', (['ROOT_PATH', '"""render_pose.so"""'], {}), "(ROOT_PATH, 'render_pose.so')\n", (486, 515), False, 'import os\n'), ((571, 615), 'os.path.join', 'os.path.join', (['ROOT_PATH', '"""render_objects.so"""'], {}), "(ROOT_PATH, 'render_objects.so')\n", (583, 615), False, 'import os\n'), ((678, 724), 'tensorflow.variable_scope', 'tf.variable_scope', (['"""pose_to_heatmap_pyWrapper"""'], {}), "('pose_to_heatmap_pyWrapper')\n", (695, 724), True, 'import tensorflow as tf\n'), ((961, 988), 'tensorflow.cast', 'tf.cast', (['pose_img', 'tf.uint8'], {}), '(pose_img, tf.uint8)\n', (968, 988), True, 'import tensorflow as tf\n'), ((1067, 1115), 'tensorflow.variable_scope', 'tf.variable_scope', (['"""zero_out_channels_pyWrapper"""'], {}), "('zero_out_channels_pyWrapper')\n", (1084, 1115), True, 'import tensorflow as tf\n'), ((1208, 1250), 'tensorflow.variable_scope', 'tf.variable_scope', (['"""render_pose_pyWrapper"""'], {}), "('render_pose_pyWrapper')\n", (1225, 1250), True, 'import tensorflow as tf\n'), ((1550, 1572), 'tensorflow.cast', 'tf.cast', (['img', 'tf.uint8'], {}), '(img, tf.uint8)\n', (1557, 1572), True, 'import tensorflow as tf\n'), ((2131, 2155), 'numpy.array', 'np.array', (["body['joints']"], {}), "(body['joints'])\n", (2139, 2155), True, 'import numpy as np\n'), ((2169, 2196), 'numpy.reshape', 'np.reshape', (['joints', '(-1, 3)'], {}), '(joints, (-1, 3))\n', (2179, 2196), True, 'import numpy as np\n'), ((2563, 2608), 'tensorflow.variable_scope', 'tf.variable_scope', (['"""render_objects_pyWrapper"""'], {}), "('render_objects_pyWrapper')\n", (2580, 2608), True, 'import tensorflow as tf\n'), ((2682, 2704), 'tensorflow.cast', 'tf.cast', (['img', 'tf.uint8'], {}), '(img, tf.uint8)\n', (2689, 2704), True, 'import tensorflow as tf\n'), ((2996, 3027), 'tensorflow.name_scope', 'tf.name_scope', (['"""ExtractGlimpse"""'], {}), "('ExtractGlimpse')\n", (3009, 3027), True, 'import tensorflow as tf\n'), ((3081, 3112), 'tensorflow.reshape', 'tf.reshape', (['pose_label', '[16, 3]'], {}), '(pose_label, [16, 3])\n', (3091, 3112), True, 'import tensorflow as tf\n'), ((3766, 3804), 'tensorflow.stack', 'tf.stack', (['[pose_label_y, pose_label_x]'], {}), '([pose_label_y, pose_label_x])\n', (3774, 3804), True, 'import tensorflow as tf\n'), ((4283, 4307), 'tensorflow.maximum', 'tf.maximum', (['mn_pts[0]', '(0)'], {}), '(mn_pts[0], 0)\n', (4293, 4307), True, 'import tensorflow as tf\n'), ((4324, 4348), 'tensorflow.maximum', 'tf.maximum', (['mn_pts[1]', '(0)'], {}), '(mn_pts[1], 0)\n', (4334, 4348), True, 'import tensorflow as tf\n'), ((1921, 1935), 'json.load', 'json.load', (['fin'], {}), '(fin)\n', (1930, 1935), False, 'import json\n'), ((2101, 2117), 'numpy.ones', 'np.ones', (['(16, 3)'], {}), '((16, 3))\n', (2108, 2117), True, 'import numpy as np\n'), ((2392, 2405), 'numpy.array', 'np.array', (['res'], {}), '(res)\n', (2400, 2405), True, 'import numpy as np\n'), ((3160, 3193), 'tensorflow.gather', 'tf.gather', (['pose_label', 'parts_keep'], {}), '(pose_label, parts_keep)\n', (3169, 3193), True, 'import tensorflow as tf\n'), ((3443, 3499), 'tensorflow.stack', 'tf.stack', (['[pose_label[0] - delta, pose_label[0] + delta]'], {}), '([pose_label[0] - delta, pose_label[0] + delta])\n', (3451, 3499), True, 'import tensorflow as tf\n'), ((3605, 3628), 'tensorflow.to_float', 'tf.to_float', (['orig_im_wd'], {}), '(orig_im_wd)\n', (3616, 3628), True, 'import tensorflow as tf\n'), ((3725, 3748), 'tensorflow.to_float', 'tf.to_float', (['orig_im_ht'], {}), '(orig_im_ht)\n', (3736, 3748), True, 'import tensorflow as tf\n'), ((3830, 3863), 'tensorflow.reduce_max', 'tf.reduce_max', (['pose_label'], {'axis': '(1)'}), '(pose_label, axis=1)\n', (3843, 3863), True, 'import tensorflow as tf\n'), ((4923, 4974), 'tensorflow.image.resize_images', 'tf.image.resize_images', (['image', '[out_side, out_side]'], {}), '(image, [out_side, out_side])\n', (4945, 4974), True, 'import tensorflow as tf\n'), ((3528, 3557), 'tensorflow.to_float', 'tf.to_float', (['pose_label[:, 0]'], {}), '(pose_label[:, 0])\n', (3539, 3557), True, 'import tensorflow as tf\n'), ((3648, 3677), 'tensorflow.to_float', 'tf.to_float', (['pose_label[:, 1]'], {}), '(pose_label[:, 1])\n', (3659, 3677), True, 'import tensorflow as tf\n'), ((4058, 4092), 'tensorflow.to_float', 'tf.to_float', (['(mx_pts[0] - mn_pts[0])'], {}), '(mx_pts[0] - mn_pts[0])\n', (4069, 4092), True, 'import tensorflow as tf\n'), ((4134, 4168), 'tensorflow.to_float', 'tf.to_float', (['(mx_pts[1] - mn_pts[1])'], {}), '(mx_pts[1] - mn_pts[1])\n', (4145, 4168), True, 'import tensorflow as tf\n'), ((4687, 4719), 'tensorflow.greater', 'tf.greater', (['mx_pts[1]', 'mn_pts[1]'], {}), '(mx_pts[1], mn_pts[1])\n', (4697, 4719), True, 'import tensorflow as tf\n'), ((4727, 4759), 'tensorflow.greater', 'tf.greater', (['mx_pts[0]', 'mn_pts[0]'], {}), '(mx_pts[0], mn_pts[0])\n', (4737, 4759), True, 'import tensorflow as tf\n'), ((4776, 4861), 'tensorflow.image.crop_to_bounding_box', 'tf.image.crop_to_bounding_box', (['image', 'offset_ht', 'offset_wd', 'target_ht', 'target_wd'], {}), '(image, offset_ht, offset_wd, target_ht, target_wd\n )\n', (4805, 4861), True, 'import tensorflow as tf\n'), ((3920, 3951), 'tensorflow.greater_equal', 'tf.greater_equal', (['pose_label', '(0)'], {}), '(pose_label, 0)\n', (3936, 3951), True, 'import tensorflow as tf\n'), ((2008, 2026), 'numpy.ones', 'np.ones', (['(16 * 3,)'], {}), '((16 * 3,))\n', (2015, 2026), True, 'import numpy as np\n'), ((3582, 3597), 'tensorflow.shape', 'tf.shape', (['image'], {}), '(image)\n', (3590, 3597), True, 'import tensorflow as tf\n'), ((3702, 3717), 'tensorflow.shape', 'tf.shape', (['image'], {}), '(image)\n', (3710, 3717), True, 'import tensorflow as tf\n'), ((4397, 4412), 'tensorflow.shape', 'tf.shape', (['image'], {}), '(image)\n', (4405, 4412), True, 'import tensorflow as tf\n'), ((4478, 4493), 'tensorflow.shape', 'tf.shape', (['image'], {}), '(image)\n', (4486, 4493), True, 'import tensorflow as tf\n'), ((3343, 3358), 'tensorflow.shape', 'tf.shape', (['image'], {}), '(image)\n', (3351, 3358), True, 'import tensorflow as tf\n'), ((3392, 3407), 'tensorflow.shape', 'tf.shape', (['image'], {}), '(image)\n', (3400, 3407), True, 'import tensorflow as tf\n')] |
"""
Authors: <NAME>
Contact : https://adityajain.me
"""
import numpy as np
class KNeighborsClassifier():
"""
K Nearest Neighbors Classifier which classifies sample point based on majority of k nearby sample classes
Parameters
----------
n_neighbors : integer (Default 5), number of neighbors to consider
metric : str ( 'minkowski', 'euclidean', 'manhatten' ) (Default : 'minkowski')
p : integer (Default 2), metric used in minkowski distance
normalize : boolean, normalize data before calculating distance
"""
def __init__(self, n_neighbors=5, metric='minkowski', p=2, normalize=False):
self.__n_neighbors = n_neighbors
self.__X = None
self.__y = None
self.__normalize = normalize
self.__metric = { 'minkowski': self.__minkowski, 'euclidean':self.__euclidean, 'manhatten':self.__manhatten }[metric]
self.__p = p
self.__n_classes = None
self.__means = None
self.__std = None
def __euclidean(self,X1,X2):
return np.sqrt(np.sum(np.square(X1-X2),axis=1))
def __manhatten(self,X1,X2):
return np.sum(np.abs(X1-X2),axis=1)
def __minkowski(self,X1,X2):
return np.power(np.sum(np.power(np.abs(X1-X2),self.__p),axis=1),1/self.__p)
def __normalizeX(self,X):
return (X-self.__means)/self.__std
def fit(self,X,y):
"""
Fit X using y
Parameters
----------
X : 2D numpy array, independent variables
y : 1D numpy array, dependent variable
"""
self.__y, self.__n_classes = y, len(np.unique(y))
if self.__normalize:
self.__means, self.__std = X.mean(axis=0), X.std(axis=0)
self.__X = self.__normalizeX(X)
else:
self.__X = X
def predict_proba(self,X):
"""
Predict probability of all classes
Parameters
----------
X : numpy array, independent variables
Returns
-------
predicted probabilities
"""
if self.__normalize: X = self.__normalizeX(X)
probs = []
for sample in X:
x = np.expand_dims(sample,axis=0)
distances = self.__metric(self.__X, x)
top_k_index = distances.argsort()[:self.__n_neighbors]
prob = np.zeros(self.__n_classes)
cls,cnts = np.unique(self.__y[top_k_index], return_counts=True)
for cl,cn in zip(cls,cnts): prob[cl] = cn/sum(cnts)
probs.append(prob)
return np.array(probs)
def predict(self,X):
"""
Predict dependent variable
Parameters
---------
X : numpy array, independent variables
Returns
-------
Predicted classes
"""
return np.argmax( self.predict_proba(X), axis=1 )
def score(self,X,y):
"""
Calculate accuracy from independent variables
Parameters
----------
X : numpy array, independent variables
y : numpy array, dependent variable
Returns
-------
accuracy score
"""
return (y==self.predict(X)).sum()/len(y)
| [
"numpy.abs",
"numpy.square",
"numpy.zeros",
"numpy.expand_dims",
"numpy.array",
"numpy.unique"
] | [((2204, 2219), 'numpy.array', 'np.array', (['probs'], {}), '(probs)\n', (2212, 2219), True, 'import numpy as np\n'), ((1039, 1054), 'numpy.abs', 'np.abs', (['(X1 - X2)'], {}), '(X1 - X2)\n', (1045, 1054), True, 'import numpy as np\n'), ((1883, 1913), 'numpy.expand_dims', 'np.expand_dims', (['sample'], {'axis': '(0)'}), '(sample, axis=0)\n', (1897, 1913), True, 'import numpy as np\n'), ((2024, 2050), 'numpy.zeros', 'np.zeros', (['self.__n_classes'], {}), '(self.__n_classes)\n', (2032, 2050), True, 'import numpy as np\n'), ((2065, 2117), 'numpy.unique', 'np.unique', (['self.__y[top_k_index]'], {'return_counts': '(True)'}), '(self.__y[top_k_index], return_counts=True)\n', (2074, 2117), True, 'import numpy as np\n'), ((966, 984), 'numpy.square', 'np.square', (['(X1 - X2)'], {}), '(X1 - X2)\n', (975, 984), True, 'import numpy as np\n'), ((1442, 1454), 'numpy.unique', 'np.unique', (['y'], {}), '(y)\n', (1451, 1454), True, 'import numpy as np\n'), ((1126, 1141), 'numpy.abs', 'np.abs', (['(X1 - X2)'], {}), '(X1 - X2)\n', (1132, 1141), True, 'import numpy as np\n')] |
"""TFRecords data-loader for audiovisual datasets."""
import functools
from typing import Dict, Iterator, List, Optional, Text, Tuple, Union
from absl import logging
from dmvr import modalities as load_modalities
from flax import jax_utils
import jax
import jax.numpy as jnp
import ml_collections
import numpy as np
from scenic.dataset_lib import dataset_utils
from scenic.dataset_lib import datasets
from scenic.dataset_lib import video_ops
from scenic.projects.mbt.datasets.dataset_utils import add_spectrogram
from scenic.projects.vivit.data import video_tfrecord_dataset
import tensorflow as tf
# Aliases for custom types:
Batch = Dict[str, jnp.ndarray]
def maybe_pad_batch(batch, train, batch_size, return_as_dict):
"""Zero pad the batch on the right to the batch_size."""
if not return_as_dict:
return dataset_utils.maybe_pad_batch(batch, train, batch_size)
assert 'batch_mask' not in batch
if 'rgb' in batch['inputs']:
unpadded_mask_shape = batch['inputs']['rgb'].shape[0]
batch_pad = batch_size - unpadded_mask_shape
elif 'spectrogram' in batch['inputs']:
unpadded_mask_shape = batch['inputs']['spectrogram'].shape[0]
batch_pad = batch_size - unpadded_mask_shape
else:
raise ValueError('invalid input batch')
if train and batch_pad != 0:
raise ValueError('In this codebase, we assumed that we always drop the '
'last partial batch of the train set. Please use '
'` drop_remainder=True` for the training set.')
# Most batches will not need padding so we quickly return to avoid slowdown.
if train or batch_pad == 0:
if 'batch_mask' not in batch:
batch['batch_mask'] = np.ones(unpadded_mask_shape, dtype=np.float32)
return batch
def zero_pad(array):
pad_with = [(0, batch_pad)] + [(0, 0)] * (array.ndim - 1)
return np.pad(array, pad_with, mode='constant')
padded_batch = jax.tree_map(zero_pad, batch)
padded_batch_mask = zero_pad(np.ones(unpadded_mask_shape, dtype=np.float32))
padded_batch['batch_mask'] = padded_batch_mask
return padded_batch
class AVTFRecordDatasetFactory(video_tfrecord_dataset.TFRecordDatasetFactory):
"""Reader for TFRecords using the MediaSequence format.
The TFrecords already contain images and spectrograms. Spectrograms are
extracted per second and stored with size 128x100 for each second of audio.
"""
_MODALITIES = ('rgb', 'spectrogram')
def __init__(self,
base_dir: str,
tables: Dict[str, Union[str, List[str]]],
num_classes: int,
examples_per_subset: Dict[str, int],
subset: str = 'train',
modalities: Tuple[str] = ('rgb',),
prop_data: float = 1.0,
num_groups: Optional[int] = None,
group_index: Optional[int] = None):
"""Initializes the instance of TFRecordDatasetFactory.
Initializes a data-loader using DeepMind Video Reader (DMVR) pre-processing
(https://github.com/deepmind/dmvr).
TFRecords are assumed to consist of tf.SequenceExample protocol buffers in
the MediaSequence
(https://github.com/google/mediapipe/tree/master/mediapipe/util/sequence)
format.
Args:
base_dir: The base directory of the TFRecords.
tables: A dictionary mapping the subset name (train, val or test) to the
relative path of the SSTable containing them. Follows DMVR convention.
The values of the dictionary can either be a string or a list. If it is
a string, it specifies all the shards in the SSTable. Example -
"/path/to/sstable@10". If passing a list, each entry is a shard of the
SSTable. Example - "[/path/to/sstable_shard_1_of_10, ...,
/path/to/sstabble_shard_10_of_10]." The latter scenario is useful for
debugging.
num_classes: The number of classes in the dataset.
examples_per_subset: A dictionary mapping the subset name (train, val or
test) to the number of examples in the dataset for that subset.
subset: The subset of the dataset to load. Must be a key of "tables"
modalities: Which modality to load. Currently supports 'rgb' and
'spectrogram'
prop_data: The proportion of the data to load. If less than 1.0, this
proportion of the total TFRecord shards are read.
num_groups: If specified will reshard the data according to `num_groups`.
A `group_index` should be specified if using `num_groups`.
group_index: Index of the shard to return after resharding. `num_groups`
should be specified if using `group_index`. This is useful in
distributed setting where one wants to ensure that different data is
read by different workers.
"""
for modality in modalities:
if modality not in AVTFRecordDatasetFactory._MODALITIES:
raise ValueError('Invalid modality %s.' % modality)
self._modalities = modalities
super().__init__(
base_dir=base_dir,
tables=tables,
examples_per_subset=examples_per_subset,
subset=subset,
num_classes=num_classes,
fraction_data=prop_data,
num_groups=num_groups,
group_index=group_index)
def _build(
self,
is_training: bool = True,
# Video related parameters.
num_frames: int = 32,
stride: int = 1,
num_spec_frames: int = 5,
spec_stride: int = 1,
dataset_spec_mean: float = 0.,
dataset_spec_stddev: float = 1.,
num_test_clips: int = 1,
min_resize: int = 256,
crop_size: int = 224,
# Audio related parameters.
spec_shape: Tuple[int, int] = (100, 128),
spec_augment: bool = False,
spec_augment_params=None,
zero_centering_image: bool = False,
# Label related parameters.
one_hot_label: bool = True,
get_label_str: bool = False):
"""Adds DMVR pre-processors to the dataset.
Args:
is_training: whether or not in training mode.
num_frames: number of frames per subclip.
stride: temporal stride to sample frames.
num_spec_frames: number of spectrogram frames.
spec_stride: stride to sample spectrogram.
dataset_spec_mean: Mean of spectrograms in the dataset.
dataset_spec_stddev: Std dev of spectrograms in the dataset.
num_test_clips: number of test clip (1 by default). If more than one, this
will sample multiple linearly spaced clips within each video at test
time. If 1, then a single clip in the middle of the video is sampled.
min_resize: frames are resized so that min width/height is min_resize.
crop_size: final size of the frame after cropping the resized frames.
spec_shape: input size of spectrogram per frame.
spec_augment: whether to apply augmentation using SpecAugment.
spec_augment_params: parameters for SpecAugment.
zero_centering_image: whether to have images between [-1, 1] or [0, 1].
one_hot_label: whether or not to return one hot version of labels.
get_label_str: whether or not to return label as text.
"""
# We set sync_random_state to True so that sample_offset_proportion is
# the same for all modalities.
if 'rgb' in self._modalities:
load_modalities.add_image(
parser_builder=self.parser_builder,
sampler_builder=self.sampler_builder,
decoder_builder=self.decoder_builder,
preprocessor_builder=self.preprocessor_builder,
postprocessor_builder=self.postprocessor_builder,
is_training=is_training,
num_frames=num_frames,
stride=stride,
num_test_clips=num_test_clips,
min_resize=min_resize,
crop_size=crop_size,
zero_centering_image=zero_centering_image,
sync_random_state=True)
if 'spectrogram' in self._modalities:
add_spectrogram(
parser_builder=self.parser_builder,
sampler_builder=self.sampler_builder,
decoder_builder=self.decoder_builder,
preprocessor_builder=self.preprocessor_builder,
postprocessor_builder=self.postprocessor_builder,
input_shape=spec_shape,
is_training=is_training,
num_frames=num_spec_frames,
stride=spec_stride,
num_test_clips=num_test_clips,
spec_augment=spec_augment,
spec_augment_params=spec_augment_params,
zero_centering_image=zero_centering_image,
dataset_mean=dataset_spec_mean,
dataset_stddev=dataset_spec_stddev,
sync_random_state=True)
load_modalities.add_label(
parser_builder=self.parser_builder,
decoder_builder=self.decoder_builder,
preprocessor_builder=self.preprocessor_builder,
is_multi_label=False,
one_hot_label=True,
num_classes=self.num_classes,
add_label_name=False)
def load_split_from_dmvr(ds_factory,
batch_size,
subset='train',
modalities=('rgb'),
num_frames=32,
stride=2,
num_spec_frames=5,
spec_stride=1,
num_test_clips=1,
min_resize=256,
crop_size=224,
spec_shape=(100, 128),
dataset_spec_mean=0.,
dataset_spec_stddev=1.,
spec_augment=False,
spec_augment_params=None,
one_hot_label=True,
zero_centering=True,
get_label_str=False,
augmentation_params=None,
keep_key=False):
"""Loads dataset using DMVR for pre-processing.
DMVR dataset loader already does basic augmentation (random crop and flip in
train mode. It also already shuffles and batches the data.
Args:
ds_factory: A DMVR factory to instantiate with the subset.
batch_size: The batch_size to use.
subset: train, validation or test.
modalities: list of input modalities.
num_frames: Number of RGB frames per subclip.
stride: Temporal stride to sample RGB frames.
num_spec_frames: Number of spectrogram frames per subclip.
spec_stride: Temporal stride to sample spectrogram.
num_test_clips: Number of test clips (1 by default). If more than 1, this
will sample multiple linearly spaced clips within each video at test time.
If 1, then a single clip in the middle of the video is sampled. The clips
are aggreagated in the batch dimension.
min_resize: Frames are resized so that min(height, width) is min_resize.
crop_size: Final size of the frame after cropping the resized frames. Both
height and width are the same.
spec_shape: Input size of spectrogram per frame.
dataset_spec_mean: Mean of spectrograms in the dataset.
dataset_spec_stddev: Std dev of spectrograms in the dataset.
spec_augment: whether to apply augmentation using SpecAugment.
spec_augment_params: dict; augmentation configurations for SpecAugment
one_hot_label: If True, return one-hot version of the labels (ie [N, C])
array. Otherwise, return [N]-dimensional array of labels.
zero_centering: If True, frames are normalized to values in [-1, 1]. If
False, values in [0, 1].
get_label_str: whether or not to return label as text. This does not work on
TPU!.
augmentation_params: dict; augmentation configurations in train mode.
keep_key: bool; If true, also return the key for each example.
Returns:
A pair `(ds, num_examples)` with
ds: A `tf.data.Dataset` object
num_examples: Number of examples in the dataset.
"""
is_training = (subset == 'train')
ds_factory = ds_factory(
subset=subset, modalities=modalities).configure(
is_training=is_training,
num_frames=num_frames,
stride=stride,
num_spec_frames=num_spec_frames,
spec_stride=spec_stride,
num_test_clips=num_test_clips,
min_resize=min_resize,
crop_size=crop_size,
spec_shape=spec_shape,
dataset_spec_mean=dataset_spec_mean,
dataset_spec_stddev=dataset_spec_stddev,
spec_augment=spec_augment,
spec_augment_params=spec_augment_params,
zero_centering_image=zero_centering,
one_hot_label=one_hot_label,
get_label_str=get_label_str)
if 'rgb' in modalities and is_training and augmentation_params:
# additional augmentation for the RGB features.
ds_factory = video_ops.additional_augmentations(ds_factory,
augmentation_params,
crop_size, num_frames,
zero_centering)
logging.info('Preprocessing graph: %s',
ds_factory.preprocessor_builder.get_summary())
logging.info('Postprocessing graph: %s',
ds_factory.postprocessor_builder.get_summary())
num_examples = ds_factory.num_examples
ds = ds_factory.make_dataset(
batch_size=batch_size,
shuffle=is_training,
num_epochs=None if is_training else 1,
drop_remainder=is_training,
keep_key=(not is_training and keep_key))
if not is_training:
ds = ds.repeat(None)
options = tf.data.Options()
options.experimental_threading.private_threadpool_size = 48
ds = ds.with_options(options)
return ds, num_examples
def map_keys(batch, modalities=('rgb'), return_as_dict=False):
"""DMVR dataset returns 'image' and 'label'. We want 'inputs' and 'label'."""
if not return_as_dict:
if len(modalities) == 1 and modalities[0] == 'rgb':
batch['inputs'] = batch['image']
elif len(modalities) == 1 and modalities[0] == 'spectrogram':
batch['inputs'] = batch['spectrogram']
else:
raise NotImplementedError('modality not supported by map_keys.')
else:
batch['inputs'] = {}
if 'rgb' in modalities:
batch['inputs']['rgb'] = batch['image']
if 'spectrogram' in modalities:
batch['inputs']['spectrogram'] = batch['spectrogram']
return batch
def tile_label_key(batch, return_as_dict=False):
"""Tile labels and keys to match input videos when num_test_clips > 1.
When multiple test crops are used (ie num_test_clips > 1), the batch dimension
of batch['inputs'] = test_batch_size * num_test_clips.
However, labels and keys remain of size [test_batch_size].
This function repeats label and key to match the inputs.
Args:
batch: Batch from iterator
return_as_dict: Whether to return multimodal inputs as a dictionary.
Returns:
batch: Batch with 'label' and 'key' tiled to match 'inputs'.
"""
if not return_as_dict:
n_repeats = batch['inputs'].shape[0] // batch['label'].shape[0]
elif 'rgb' in batch['inputs']:
n_repeats = batch['inputs']['rgb'].shape[0] // batch['label'].shape[0]
elif 'spectrogram' in batch['inputs']:
n_repeats = (
batch['inputs']['spectrogram'].shape[0] // batch['label'].shape[0])
batch['label'] = np.repeat(batch['label'], n_repeats, axis=0)
if 'key' in batch:
batch['key'] = np.repeat(batch['key'], n_repeats, axis=0)
return batch
@datasets.add_dataset('audiovisual_tfrecord_dataset')
def get_dataset(
*,
batch_size,
eval_batch_size,
num_shards,
dtype_str='float32',
shuffle_seed=0, # pylint:disable=unused-argument
rng=None,
dataset_configs: ml_collections.ConfigDict,
dataset_service_address: Optional[str] = None):
"""Returns a generator for the audiovisual dataset."""
del rng
modalities = dataset_configs.get('modalities', ['rgb'])
return_as_dict = dataset_configs.get('return_as_dict', False)
# RGB related configs.
num_frames = dataset_configs.get('num_frames', 32)
stride = dataset_configs.get('stride', 2)
min_resize = dataset_configs.get('min_resize', 256)
crop_size = dataset_configs.get('crop_size', 224)
# Spectrogram related configs.
num_spec_frames = dataset_configs.get('num_spec_frames', 5)
spec_stride = dataset_configs.get('spec_stride', 1)
spec_shape = dataset_configs.get('spec_shape', (100, 128))
spec_augment = dataset_configs.get('spec_augment', False)
spec_augment_params = dataset_configs.get('spec_augment_params', None)
dataset_spec_mean = dataset_configs.get('spec_mean', 0.)
dataset_spec_stddev = dataset_configs.get('spec_stddev', 1.)
# General configs.
num_test_clips = dataset_configs.get('num_test_clips', 1)
one_hot_label = dataset_configs.get('one_hot_label', True)
zero_centre_data = dataset_configs.get('zero_centering', True)
augmentation_params = dataset_configs.get('augmentation_params', None)
num_train_val_clips = dataset_configs.get('num_train_val_clips', 1)
do_three_spatial_crops = dataset_configs.get('do_three_spatial_crops', False)
num_spatial_crops = 3 if do_three_spatial_crops else 1
keep_test_key = dataset_configs.get('keep_test_key', False)
test_split = dataset_configs.get('test_split', 'test')
# For the test set, the actual batch size is
# test_batch_size * num_test_clips
test_batch_size = dataset_configs.get('test_batch_size', eval_batch_size)
def validate_config(field):
if dataset_configs.get(field) is None:
raise ValueError(f'{field} must be specified for TFRecord dataset.')
validate_config('base_dir')
validate_config('tables')
validate_config('examples_per_subset')
validate_config('num_classes')
ds_factory = functools.partial(
AVTFRecordDatasetFactory,
base_dir=dataset_configs.base_dir,
tables=dataset_configs.tables,
examples_per_subset=dataset_configs.examples_per_subset,
num_classes=dataset_configs.num_classes,
num_groups=jax.process_count(),
group_index=jax.process_index())
def create_dataset_iterator(
subset: Text,
batch_size_local: int,
num_clips: int,
keep_key_local: bool = False) -> Tuple[Iterator[Batch], int]:
is_training = subset == 'train'
is_test = subset == 'test'
logging.info('Loading split %s', subset)
dataset, num_examples = load_split_from_dmvr(
ds_factory,
batch_size=batch_size_local,
subset=subset,
modalities=modalities,
num_frames=num_frames,
stride=stride,
num_spec_frames=num_spec_frames,
spec_stride=spec_stride,
num_test_clips=num_clips,
min_resize=min_resize,
crop_size=crop_size,
spec_shape=spec_shape,
dataset_spec_mean=dataset_spec_mean,
dataset_spec_stddev=dataset_spec_stddev,
spec_augment=spec_augment,
spec_augment_params=spec_augment_params,
one_hot_label=one_hot_label,
zero_centering=zero_centre_data,
augmentation_params=augmentation_params,
keep_key=keep_key_local)
if dataset_service_address and is_training:
if shuffle_seed is not None:
raise ValueError('Using dataset service with a random seed causes each '
'worker to produce exactly the same data. Add '
'config.shuffle_seed = None to your config if you '
'want to run with dataset service.')
logging.info('Using the tf.data service at %s', dataset_service_address)
dataset = dataset_utils.distribute(dataset, dataset_service_address)
pad_batch_size = batch_size_local
if is_test:
pad_batch_size = batch_size_local * num_clips * num_spatial_crops
maybe_pad_batches = functools.partial(
maybe_pad_batch,
train=is_training,
batch_size=pad_batch_size,
return_as_dict=return_as_dict)
shard_batches = functools.partial(dataset_utils.shard, n_devices=num_shards)
current_iter = iter(dataset)
current_iter = map(dataset_utils.tf_to_numpy, current_iter)
current_iter = map(
functools.partial(
map_keys, modalities=modalities, return_as_dict=return_as_dict),
current_iter)
current_iter = map(
functools.partial(
tile_label_key, return_as_dict=return_as_dict),
current_iter)
current_iter = map(maybe_pad_batches, current_iter)
if augmentation_params and augmentation_params.get('do_mixup', False):
raise ValueError('mixup should be done in the trainer.')
current_iter = map(shard_batches, current_iter)
if is_training and dataset_configs.get('prefetch_to_device'):
# Async bind batch to device which speeds up training.
current_iter = jax_utils.prefetch_to_device(
current_iter, dataset_configs.get('prefetch_to_device'))
return current_iter, num_examples
train_iter, n_train_examples = create_dataset_iterator(
'train', batch_size, num_train_val_clips)
eval_iter, n_eval_examples = create_dataset_iterator('validation',
eval_batch_size,
num_train_val_clips)
test_iter, n_test_examples = create_dataset_iterator(test_split,
test_batch_size,
num_test_clips,
keep_test_key)
meta_data = {
'num_classes': dataset_configs.num_classes, # pylint:disable=protected-access
'num_train_examples': (n_train_examples * num_train_val_clips),
'num_eval_examples': (n_eval_examples * num_train_val_clips),
'num_test_examples':
(n_test_examples * num_test_clips * num_spatial_crops),
'input_dtype': getattr(jnp, dtype_str),
'target_is_onehot': True,
}
if return_as_dict:
meta_data['input_shape'] = {
'rgb': (-1, num_frames, crop_size, crop_size, 3),
'spectrogram': (-1, num_spec_frames * spec_shape[0], spec_shape[1], 3)
}
elif len(modalities) == 1 and modalities[0] == 'rgb':
meta_data['input_shape'] = (-1, num_frames, crop_size, crop_size, 3)
elif len(modalities) == 1 and modalities[0] == 'spectrogram':
meta_data['input_shape'] = (-1, num_spec_frames * spec_shape[0],
spec_shape[1], 3)
else:
raise NotImplementedError('modality not supported')
logging.info('Dataset metadata:\n%s', meta_data)
return dataset_utils.Dataset(train_iter, eval_iter, test_iter, meta_data)
| [
"numpy.pad",
"functools.partial",
"scenic.projects.mbt.datasets.dataset_utils.add_spectrogram",
"scenic.dataset_lib.datasets.add_dataset",
"dmvr.modalities.add_label",
"numpy.ones",
"scenic.dataset_lib.dataset_utils.distribute",
"absl.logging.info",
"jax.process_count",
"tensorflow.data.Options",
... | [((15405, 15457), 'scenic.dataset_lib.datasets.add_dataset', 'datasets.add_dataset', (['"""audiovisual_tfrecord_dataset"""'], {}), "('audiovisual_tfrecord_dataset')\n", (15425, 15457), False, 'from scenic.dataset_lib import datasets\n'), ((1900, 1929), 'jax.tree_map', 'jax.tree_map', (['zero_pad', 'batch'], {}), '(zero_pad, batch)\n', (1912, 1929), False, 'import jax\n'), ((13511, 13528), 'tensorflow.data.Options', 'tf.data.Options', ([], {}), '()\n', (13526, 13528), True, 'import tensorflow as tf\n'), ((15259, 15303), 'numpy.repeat', 'np.repeat', (["batch['label']", 'n_repeats'], {'axis': '(0)'}), "(batch['label'], n_repeats, axis=0)\n", (15268, 15303), True, 'import numpy as np\n'), ((22441, 22492), 'absl.logging.info', 'logging.info', (['"""Dataset metadata:\n%s"""', 'meta_data'], {}), '("""Dataset metadata:\n%s""", meta_data)\n', (22453, 22492), False, 'from absl import logging\n'), ((22500, 22566), 'scenic.dataset_lib.dataset_utils.Dataset', 'dataset_utils.Dataset', (['train_iter', 'eval_iter', 'test_iter', 'meta_data'], {}), '(train_iter, eval_iter, test_iter, meta_data)\n', (22521, 22566), False, 'from scenic.dataset_lib import dataset_utils\n'), ((820, 875), 'scenic.dataset_lib.dataset_utils.maybe_pad_batch', 'dataset_utils.maybe_pad_batch', (['batch', 'train', 'batch_size'], {}), '(batch, train, batch_size)\n', (849, 875), False, 'from scenic.dataset_lib import dataset_utils\n'), ((1841, 1881), 'numpy.pad', 'np.pad', (['array', 'pad_with'], {'mode': '"""constant"""'}), "(array, pad_with, mode='constant')\n", (1847, 1881), True, 'import numpy as np\n'), ((1961, 2007), 'numpy.ones', 'np.ones', (['unpadded_mask_shape'], {'dtype': 'np.float32'}), '(unpadded_mask_shape, dtype=np.float32)\n', (1968, 2007), True, 'import numpy as np\n'), ((8599, 8853), 'dmvr.modalities.add_label', 'load_modalities.add_label', ([], {'parser_builder': 'self.parser_builder', 'decoder_builder': 'self.decoder_builder', 'preprocessor_builder': 'self.preprocessor_builder', 'is_multi_label': '(False)', 'one_hot_label': '(True)', 'num_classes': 'self.num_classes', 'add_label_name': '(False)'}), '(parser_builder=self.parser_builder,\n decoder_builder=self.decoder_builder, preprocessor_builder=self.\n preprocessor_builder, is_multi_label=False, one_hot_label=True,\n num_classes=self.num_classes, add_label_name=False)\n', (8624, 8853), True, 'from dmvr import modalities as load_modalities\n'), ((12719, 12829), 'scenic.dataset_lib.video_ops.additional_augmentations', 'video_ops.additional_augmentations', (['ds_factory', 'augmentation_params', 'crop_size', 'num_frames', 'zero_centering'], {}), '(ds_factory, augmentation_params,\n crop_size, num_frames, zero_centering)\n', (12753, 12829), False, 'from scenic.dataset_lib import video_ops\n'), ((15344, 15386), 'numpy.repeat', 'np.repeat', (["batch['key']", 'n_repeats'], {'axis': '(0)'}), "(batch['key'], n_repeats, axis=0)\n", (15353, 15386), True, 'import numpy as np\n'), ((18232, 18272), 'absl.logging.info', 'logging.info', (['"""Loading split %s"""', 'subset'], {}), "('Loading split %s', subset)\n", (18244, 18272), False, 'from absl import logging\n'), ((19707, 19823), 'functools.partial', 'functools.partial', (['maybe_pad_batch'], {'train': 'is_training', 'batch_size': 'pad_batch_size', 'return_as_dict': 'return_as_dict'}), '(maybe_pad_batch, train=is_training, batch_size=\n pad_batch_size, return_as_dict=return_as_dict)\n', (19724, 19823), False, 'import functools\n'), ((19873, 19933), 'functools.partial', 'functools.partial', (['dataset_utils.shard'], {'n_devices': 'num_shards'}), '(dataset_utils.shard, n_devices=num_shards)\n', (19890, 19933), False, 'import functools\n'), ((1680, 1726), 'numpy.ones', 'np.ones', (['unpadded_mask_shape'], {'dtype': 'np.float32'}), '(unpadded_mask_shape, dtype=np.float32)\n', (1687, 1726), True, 'import numpy as np\n'), ((7256, 7723), 'dmvr.modalities.add_image', 'load_modalities.add_image', ([], {'parser_builder': 'self.parser_builder', 'sampler_builder': 'self.sampler_builder', 'decoder_builder': 'self.decoder_builder', 'preprocessor_builder': 'self.preprocessor_builder', 'postprocessor_builder': 'self.postprocessor_builder', 'is_training': 'is_training', 'num_frames': 'num_frames', 'stride': 'stride', 'num_test_clips': 'num_test_clips', 'min_resize': 'min_resize', 'crop_size': 'crop_size', 'zero_centering_image': 'zero_centering_image', 'sync_random_state': '(True)'}), '(parser_builder=self.parser_builder,\n sampler_builder=self.sampler_builder, decoder_builder=self.\n decoder_builder, preprocessor_builder=self.preprocessor_builder,\n postprocessor_builder=self.postprocessor_builder, is_training=\n is_training, num_frames=num_frames, stride=stride, num_test_clips=\n num_test_clips, min_resize=min_resize, crop_size=crop_size,\n zero_centering_image=zero_centering_image, sync_random_state=True)\n', (7281, 7723), True, 'from dmvr import modalities as load_modalities\n'), ((7876, 8469), 'scenic.projects.mbt.datasets.dataset_utils.add_spectrogram', 'add_spectrogram', ([], {'parser_builder': 'self.parser_builder', 'sampler_builder': 'self.sampler_builder', 'decoder_builder': 'self.decoder_builder', 'preprocessor_builder': 'self.preprocessor_builder', 'postprocessor_builder': 'self.postprocessor_builder', 'input_shape': 'spec_shape', 'is_training': 'is_training', 'num_frames': 'num_spec_frames', 'stride': 'spec_stride', 'num_test_clips': 'num_test_clips', 'spec_augment': 'spec_augment', 'spec_augment_params': 'spec_augment_params', 'zero_centering_image': 'zero_centering_image', 'dataset_mean': 'dataset_spec_mean', 'dataset_stddev': 'dataset_spec_stddev', 'sync_random_state': '(True)'}), '(parser_builder=self.parser_builder, sampler_builder=self.\n sampler_builder, decoder_builder=self.decoder_builder,\n preprocessor_builder=self.preprocessor_builder, postprocessor_builder=\n self.postprocessor_builder, input_shape=spec_shape, is_training=\n is_training, num_frames=num_spec_frames, stride=spec_stride,\n num_test_clips=num_test_clips, spec_augment=spec_augment,\n spec_augment_params=spec_augment_params, zero_centering_image=\n zero_centering_image, dataset_mean=dataset_spec_mean, dataset_stddev=\n dataset_spec_stddev, sync_random_state=True)\n', (7891, 8469), False, 'from scenic.projects.mbt.datasets.dataset_utils import add_spectrogram\n'), ((17929, 17948), 'jax.process_count', 'jax.process_count', ([], {}), '()\n', (17946, 17948), False, 'import jax\n'), ((17968, 17987), 'jax.process_index', 'jax.process_index', ([], {}), '()\n', (17985, 17987), False, 'import jax\n'), ((19408, 19480), 'absl.logging.info', 'logging.info', (['"""Using the tf.data service at %s"""', 'dataset_service_address'], {}), "('Using the tf.data service at %s', dataset_service_address)\n", (19420, 19480), False, 'from absl import logging\n'), ((19497, 19555), 'scenic.dataset_lib.dataset_utils.distribute', 'dataset_utils.distribute', (['dataset', 'dataset_service_address'], {}), '(dataset, dataset_service_address)\n', (19521, 19555), False, 'from scenic.dataset_lib import dataset_utils\n'), ((20063, 20149), 'functools.partial', 'functools.partial', (['map_keys'], {'modalities': 'modalities', 'return_as_dict': 'return_as_dict'}), '(map_keys, modalities=modalities, return_as_dict=\n return_as_dict)\n', (20080, 20149), False, 'import functools\n'), ((20213, 20277), 'functools.partial', 'functools.partial', (['tile_label_key'], {'return_as_dict': 'return_as_dict'}), '(tile_label_key, return_as_dict=return_as_dict)\n', (20230, 20277), False, 'import functools\n')] |
# -*- coding: utf-8 -*-
# File generated according to Generator/ClassesRef/Simulation/InputFlux.csv
# WARNING! All changes made in this file will be lost!
"""Method code available at https://github.com/Eomys/pyleecan/tree/master/pyleecan/Methods/Simulation/InputFlux
"""
from os import linesep
from sys import getsizeof
from logging import getLogger
from ._check import set_array, check_var, raise_
from ..Functions.get_logger import get_logger
from ..Functions.save import save
from ..Functions.copy import copy
from ..Functions.load import load_init_dict
from ..Functions.Load.import_class import import_class
from .InputCurrent import InputCurrent
# Import all class method
# Try/catch to remove unnecessary dependencies in unused method
try:
from ..Methods.Simulation.InputFlux.gen_input import gen_input
except ImportError as error:
gen_input = error
from ..Classes.ImportMatrixVal import ImportMatrixVal
from numpy import ndarray
from numpy import array, array_equal
from ._check import InitUnKnowClassError
from .ImportMatrix import ImportMatrix
from .ImportData import ImportData
from .ImportGenPWM import ImportGenPWM
from .OP import OP
class InputFlux(InputCurrent):
"""Input to skip the magnetic module and start with the structural one"""
VERSION = 1
# cf Methods.Simulation.InputFlux.gen_input
if isinstance(gen_input, ImportError):
gen_input = property(
fget=lambda x: raise_(
ImportError("Can't use InputFlux method gen_input: " + str(gen_input))
)
)
else:
gen_input = gen_input
# save and copy methods are available in all object
save = save
copy = copy
# get_logger method is available in all object
get_logger = get_logger
def __init__(
self,
per_a=1,
per_t=1,
is_antiper_a=False,
is_antiper_t=False,
B_dict=None,
unit=None,
slice=None,
B_enforced=None,
Is=None,
Ir=None,
Is_harm=None,
rot_dir=None,
angle_rotor_initial=0,
PWM=None,
phase_dir=None,
current_dir=None,
is_periodicity_t=False,
is_periodicity_a=False,
time=None,
angle=None,
Nt_tot=2048,
Nrev=None,
Na_tot=2048,
OP=None,
t_final=None,
init_dict=None,
init_str=None,
):
"""Constructor of the class. Can be use in three ways :
- __init__ (arg1 = 1, arg3 = 5) every parameters have name and default values
for pyleecan type, -1 will call the default constructor
- __init__ (init_dict = d) d must be a dictionary with property names as keys
- __init__ (init_str = s) s must be a string
s is the file path to load
ndarray or list can be given for Vector and Matrix
object or dict can be given for pyleecan Object"""
if init_str is not None: # Load from a file
init_dict = load_init_dict(init_str)[1]
if init_dict is not None: # Initialisation by dict
assert type(init_dict) is dict
# Overwrite default value with init_dict content
if "per_a" in list(init_dict.keys()):
per_a = init_dict["per_a"]
if "per_t" in list(init_dict.keys()):
per_t = init_dict["per_t"]
if "is_antiper_a" in list(init_dict.keys()):
is_antiper_a = init_dict["is_antiper_a"]
if "is_antiper_t" in list(init_dict.keys()):
is_antiper_t = init_dict["is_antiper_t"]
if "B_dict" in list(init_dict.keys()):
B_dict = init_dict["B_dict"]
if "unit" in list(init_dict.keys()):
unit = init_dict["unit"]
if "slice" in list(init_dict.keys()):
slice = init_dict["slice"]
if "B_enforced" in list(init_dict.keys()):
B_enforced = init_dict["B_enforced"]
if "Is" in list(init_dict.keys()):
Is = init_dict["Is"]
if "Ir" in list(init_dict.keys()):
Ir = init_dict["Ir"]
if "Is_harm" in list(init_dict.keys()):
Is_harm = init_dict["Is_harm"]
if "rot_dir" in list(init_dict.keys()):
rot_dir = init_dict["rot_dir"]
if "angle_rotor_initial" in list(init_dict.keys()):
angle_rotor_initial = init_dict["angle_rotor_initial"]
if "PWM" in list(init_dict.keys()):
PWM = init_dict["PWM"]
if "phase_dir" in list(init_dict.keys()):
phase_dir = init_dict["phase_dir"]
if "current_dir" in list(init_dict.keys()):
current_dir = init_dict["current_dir"]
if "is_periodicity_t" in list(init_dict.keys()):
is_periodicity_t = init_dict["is_periodicity_t"]
if "is_periodicity_a" in list(init_dict.keys()):
is_periodicity_a = init_dict["is_periodicity_a"]
if "time" in list(init_dict.keys()):
time = init_dict["time"]
if "angle" in list(init_dict.keys()):
angle = init_dict["angle"]
if "Nt_tot" in list(init_dict.keys()):
Nt_tot = init_dict["Nt_tot"]
if "Nrev" in list(init_dict.keys()):
Nrev = init_dict["Nrev"]
if "Na_tot" in list(init_dict.keys()):
Na_tot = init_dict["Na_tot"]
if "OP" in list(init_dict.keys()):
OP = init_dict["OP"]
if "t_final" in list(init_dict.keys()):
t_final = init_dict["t_final"]
# Set the properties (value check and convertion are done in setter)
self.per_a = per_a
self.per_t = per_t
self.is_antiper_a = is_antiper_a
self.is_antiper_t = is_antiper_t
self.B_dict = B_dict
self.unit = unit
self.slice = slice
self.B_enforced = B_enforced
# Call InputCurrent init
super(InputFlux, self).__init__(
Is=Is,
Ir=Ir,
Is_harm=Is_harm,
rot_dir=rot_dir,
angle_rotor_initial=angle_rotor_initial,
PWM=PWM,
phase_dir=phase_dir,
current_dir=current_dir,
is_periodicity_t=is_periodicity_t,
is_periodicity_a=is_periodicity_a,
time=time,
angle=angle,
Nt_tot=Nt_tot,
Nrev=Nrev,
Na_tot=Na_tot,
OP=OP,
t_final=t_final,
)
# The class is frozen (in InputCurrent init), for now it's impossible to
# add new properties
def __str__(self):
"""Convert this object in a readeable string (for print)"""
InputFlux_str = ""
# Get the properties inherited from InputCurrent
InputFlux_str += super(InputFlux, self).__str__()
InputFlux_str += "per_a = " + str(self.per_a) + linesep
InputFlux_str += "per_t = " + str(self.per_t) + linesep
InputFlux_str += "is_antiper_a = " + str(self.is_antiper_a) + linesep
InputFlux_str += "is_antiper_t = " + str(self.is_antiper_t) + linesep
InputFlux_str += "B_dict = " + str(self.B_dict) + linesep
InputFlux_str += 'unit = "' + str(self.unit) + '"' + linesep
InputFlux_str += (
"slice = "
+ linesep
+ str(self.slice).replace(linesep, linesep + "\t")
+ linesep
+ linesep
)
InputFlux_str += "B_enforced = " + str(self.B_enforced) + linesep + linesep
return InputFlux_str
def __eq__(self, other):
"""Compare two objects (skip parent)"""
if type(other) != type(self):
return False
# Check the properties inherited from InputCurrent
if not super(InputFlux, self).__eq__(other):
return False
if other.per_a != self.per_a:
return False
if other.per_t != self.per_t:
return False
if other.is_antiper_a != self.is_antiper_a:
return False
if other.is_antiper_t != self.is_antiper_t:
return False
if other.B_dict != self.B_dict:
return False
if other.unit != self.unit:
return False
if not array_equal(other.slice, self.slice):
return False
if other.B_enforced != self.B_enforced:
return False
return True
def compare(self, other, name="self", ignore_list=None):
"""Compare two objects and return list of differences"""
if ignore_list is None:
ignore_list = list()
if type(other) != type(self):
return ["type(" + name + ")"]
diff_list = list()
# Check the properties inherited from InputCurrent
diff_list.extend(super(InputFlux, self).compare(other, name=name))
if other._per_a != self._per_a:
diff_list.append(name + ".per_a")
if other._per_t != self._per_t:
diff_list.append(name + ".per_t")
if other._is_antiper_a != self._is_antiper_a:
diff_list.append(name + ".is_antiper_a")
if other._is_antiper_t != self._is_antiper_t:
diff_list.append(name + ".is_antiper_t")
if other._B_dict != self._B_dict:
diff_list.append(name + ".B_dict")
if other._unit != self._unit:
diff_list.append(name + ".unit")
if not array_equal(other.slice, self.slice):
diff_list.append(name + ".slice")
if (other.B_enforced is None and self.B_enforced is not None) or (
other.B_enforced is not None and self.B_enforced is None
):
diff_list.append(name + ".B_enforced None mismatch")
elif self.B_enforced is not None:
diff_list.extend(
self.B_enforced.compare(other.B_enforced, name=name + ".B_enforced")
)
# Filter ignore differences
diff_list = list(filter(lambda x: x not in ignore_list, diff_list))
return diff_list
def __sizeof__(self):
"""Return the size in memory of the object (including all subobject)"""
S = 0 # Full size of the object
# Get size of the properties inherited from InputCurrent
S += super(InputFlux, self).__sizeof__()
S += getsizeof(self.per_a)
S += getsizeof(self.per_t)
S += getsizeof(self.is_antiper_a)
S += getsizeof(self.is_antiper_t)
if self.B_dict is not None:
for key, value in self.B_dict.items():
S += getsizeof(value) + getsizeof(key)
S += getsizeof(self.unit)
S += getsizeof(self.slice)
S += getsizeof(self.B_enforced)
return S
def as_dict(self, type_handle_ndarray=0, keep_function=False, **kwargs):
"""
Convert this object in a json serializable dict (can be use in __init__).
type_handle_ndarray: int
How to handle ndarray (0: tolist, 1: copy, 2: nothing)
keep_function : bool
True to keep the function object, else return str
Optional keyword input parameter is for internal use only
and may prevent json serializability.
"""
# Get the properties inherited from InputCurrent
InputFlux_dict = super(InputFlux, self).as_dict(
type_handle_ndarray=type_handle_ndarray,
keep_function=keep_function,
**kwargs
)
InputFlux_dict["per_a"] = self.per_a
InputFlux_dict["per_t"] = self.per_t
InputFlux_dict["is_antiper_a"] = self.is_antiper_a
InputFlux_dict["is_antiper_t"] = self.is_antiper_t
InputFlux_dict["B_dict"] = (
self.B_dict.copy() if self.B_dict is not None else None
)
InputFlux_dict["unit"] = self.unit
if self.slice is None:
InputFlux_dict["slice"] = None
else:
if type_handle_ndarray == 0:
InputFlux_dict["slice"] = self.slice.tolist()
elif type_handle_ndarray == 1:
InputFlux_dict["slice"] = self.slice.copy()
elif type_handle_ndarray == 2:
InputFlux_dict["slice"] = self.slice
else:
raise Exception(
"Unknown type_handle_ndarray: " + str(type_handle_ndarray)
)
if self.B_enforced is None:
InputFlux_dict["B_enforced"] = None
else:
InputFlux_dict["B_enforced"] = self.B_enforced.as_dict(
type_handle_ndarray=type_handle_ndarray,
keep_function=keep_function,
**kwargs
)
# The class name is added to the dict for deserialisation purpose
# Overwrite the mother class name
InputFlux_dict["__class__"] = "InputFlux"
return InputFlux_dict
def _set_None(self):
"""Set all the properties to None (except pyleecan object)"""
self.per_a = None
self.per_t = None
self.is_antiper_a = None
self.is_antiper_t = None
self.B_dict = None
self.unit = None
self.slice = None
self.B_enforced = None
# Set to None the properties inherited from InputCurrent
super(InputFlux, self)._set_None()
def _get_per_a(self):
"""getter of per_a"""
return self._per_a
def _set_per_a(self, value):
"""setter of per_a"""
check_var("per_a", value, "int")
self._per_a = value
per_a = property(
fget=_get_per_a,
fset=_set_per_a,
doc=u"""Angle periodicity
:Type: int
""",
)
def _get_per_t(self):
"""getter of per_t"""
return self._per_t
def _set_per_t(self, value):
"""setter of per_t"""
check_var("per_t", value, "int")
self._per_t = value
per_t = property(
fget=_get_per_t,
fset=_set_per_t,
doc=u"""Time periodicity
:Type: int
""",
)
def _get_is_antiper_a(self):
"""getter of is_antiper_a"""
return self._is_antiper_a
def _set_is_antiper_a(self, value):
"""setter of is_antiper_a"""
check_var("is_antiper_a", value, "bool")
self._is_antiper_a = value
is_antiper_a = property(
fget=_get_is_antiper_a,
fset=_set_is_antiper_a,
doc=u"""If angle is antiperiodic
:Type: bool
""",
)
def _get_is_antiper_t(self):
"""getter of is_antiper_t"""
return self._is_antiper_t
def _set_is_antiper_t(self, value):
"""setter of is_antiper_t"""
check_var("is_antiper_t", value, "bool")
self._is_antiper_t = value
is_antiper_t = property(
fget=_get_is_antiper_t,
fset=_set_is_antiper_t,
doc=u"""If time is antiperiodic
:Type: bool
""",
)
def _get_B_dict(self):
"""getter of B_dict"""
return self._B_dict
def _set_B_dict(self, value):
"""setter of B_dict"""
if type(value) is int and value == -1:
value = dict()
check_var("B_dict", value, "dict")
self._B_dict = value
B_dict = property(
fget=_get_B_dict,
fset=_set_B_dict,
doc=u"""Dict of Import objects or lists for each component of the flux
:Type: dict
""",
)
def _get_unit(self):
"""getter of unit"""
return self._unit
def _set_unit(self, value):
"""setter of unit"""
check_var("unit", value, "str")
self._unit = value
unit = property(
fget=_get_unit,
fset=_set_unit,
doc=u"""Unit of the flux if not T
:Type: str
""",
)
def _get_slice(self):
"""getter of slice"""
return self._slice
def _set_slice(self, value):
"""setter of slice"""
if type(value) is int and value == -1:
value = array([])
elif type(value) is list:
try:
value = array(value)
except:
pass
check_var("slice", value, "ndarray")
self._slice = value
slice = property(
fget=_get_slice,
fset=_set_slice,
doc=u"""Slice axis values
:Type: ndarray
""",
)
def _get_B_enforced(self):
"""getter of B_enforced"""
return self._B_enforced
def _set_B_enforced(self, value):
"""setter of B_enforced"""
if isinstance(value, str): # Load from file
value = load_init_dict(value)[1]
if isinstance(value, dict) and "__class__" in value:
class_obj = import_class(
"SciDataTool.Classes", value.get("__class__"), "B_enforced"
)
value = class_obj(init_dict=value)
elif type(value) is int and value == -1: # Default constructor
value = VectorField()
check_var("B_enforced", value, "VectorField")
self._B_enforced = value
B_enforced = property(
fget=_get_B_enforced,
fset=_set_B_enforced,
doc=u"""Airgap flux density as VectorField object
:Type: SciDataTool.Classes.VectorField.VectorField
""",
)
| [
"numpy.array_equal",
"numpy.array",
"sys.getsizeof"
] | [((10431, 10452), 'sys.getsizeof', 'getsizeof', (['self.per_a'], {}), '(self.per_a)\n', (10440, 10452), False, 'from sys import getsizeof\n'), ((10466, 10487), 'sys.getsizeof', 'getsizeof', (['self.per_t'], {}), '(self.per_t)\n', (10475, 10487), False, 'from sys import getsizeof\n'), ((10501, 10529), 'sys.getsizeof', 'getsizeof', (['self.is_antiper_a'], {}), '(self.is_antiper_a)\n', (10510, 10529), False, 'from sys import getsizeof\n'), ((10543, 10571), 'sys.getsizeof', 'getsizeof', (['self.is_antiper_t'], {}), '(self.is_antiper_t)\n', (10552, 10571), False, 'from sys import getsizeof\n'), ((10727, 10747), 'sys.getsizeof', 'getsizeof', (['self.unit'], {}), '(self.unit)\n', (10736, 10747), False, 'from sys import getsizeof\n'), ((10761, 10782), 'sys.getsizeof', 'getsizeof', (['self.slice'], {}), '(self.slice)\n', (10770, 10782), False, 'from sys import getsizeof\n'), ((10796, 10822), 'sys.getsizeof', 'getsizeof', (['self.B_enforced'], {}), '(self.B_enforced)\n', (10805, 10822), False, 'from sys import getsizeof\n'), ((8378, 8414), 'numpy.array_equal', 'array_equal', (['other.slice', 'self.slice'], {}), '(other.slice, self.slice)\n', (8389, 8414), False, 'from numpy import array, array_equal\n'), ((9542, 9578), 'numpy.array_equal', 'array_equal', (['other.slice', 'self.slice'], {}), '(other.slice, self.slice)\n', (9553, 9578), False, 'from numpy import array, array_equal\n'), ((16084, 16093), 'numpy.array', 'array', (['[]'], {}), '([])\n', (16089, 16093), False, 'from numpy import array, array_equal\n'), ((10680, 10696), 'sys.getsizeof', 'getsizeof', (['value'], {}), '(value)\n', (10689, 10696), False, 'from sys import getsizeof\n'), ((10699, 10713), 'sys.getsizeof', 'getsizeof', (['key'], {}), '(key)\n', (10708, 10713), False, 'from sys import getsizeof\n'), ((16169, 16181), 'numpy.array', 'array', (['value'], {}), '(value)\n', (16174, 16181), False, 'from numpy import array, array_equal\n')] |
from __future__ import print_function, absolute_import
import stimela
import stimela.dismissable as sdm
from pyrap.tables import table as tbl
import os
import sys
import argparse
import numpy as np
from collections import OrderedDict
import shutil
import vermeerkat
parser = argparse.ArgumentParser("MeerKAT BasicApplyTransfer (BAT) pipeline")
parser.add_argument('--input_dir', dest='input_directory', metavar='<input directory>',
action='store', default='input',
help='directory to read input data with e.g. antenna table and beam files')
parser.add_argument('--output_dir', dest='output_directory', metavar='<output directory>',
action='store', default='output', help='directory to write output data')
parser.add_argument('--msdir_dir', dest='msdir_directory', metavar='<msdir directory>',
action='store', default='msdir', help='directory to store measurement sets')
parser.add_argument('--bp', dest='bandpass_field', metavar='<bandpass field>',
help='Bandpass fields')
parser.add_argument('--gc', dest='gain_field', metavar='<gain calibrator fields>', default=[],
action="append", nargs="+",
help='Gain fields. This switch can be used multiple times for more than 1 field')
parser.add_argument('--altcal', dest='alt_cal_field', metavar='<alternative calibrator fields>', default=[],
action="append", nargs="+",
help='Alternative calibrator. Phase corrections will be applied to this field for further '
'diagnostic or calibration procedures. This switch can be used multiple times for '
'more than 1 field. This field has no impact on target field calibration.')
parser.add_argument('--tar', dest='target_field', metavar='<target fields>', type=str, default=[],
action="append", nargs="+",
help='Target fields. This switch can be used multiple times for more than 1 field')
parser.add_argument('--no_delay_with_gcal', dest='delay_with_gcal', action='store_false', default=True,
help='DON''t use gain calibrators for delay calibration')
parser.add_argument('--flag_antenna', dest='flag_antenna', action='append', default=[],
help="Flag antenna. Can be specified more than once to flag more than one antenna.")
parser.add_argument('--skip_prepdata', dest='skip_prepdata', action='store_true',
help="Skip prepdata")
parser.add_argument('--skip_prelim_flagging', dest='skip_prelim_flagging', action='store_true',
help="Skip preliminary flagging")
parser.add_argument('--skip_prelim_1GC', dest='skip_prelim_1GC', action='store_true',
help="Skip preliminary 1GC")
parser.add_argument('--skip_final_flagging', dest='skip_final_flagging', action='store_true',
help="Skip final flagging")
parser.add_argument('--skip_final_1GC', dest='skip_final_1GC', action='store_true',
help="Skip final 1GC")
parser.add_argument('--skip_flag_targets', dest='skip_flag_targets', action='store_true',
help="Skip flag targets")
parser.add_argument('--skip_transfer_to_targets', dest='skip_transfer_to_targets', action='store_true',
help="Skip transfer of solutions to target")
parser.add_argument('--skip_final_split', dest='skip_final_split', action='store_true',
help="Skip final split")
parser.add_argument('msprefix', metavar='<measurement set name prefix>',
help='Prefix of measurement set name as it appears relative to msdir. This must NOT be a '
'path prefix')
parser.add_argument('--time_sol_interval', dest='time_sol_interval', default="inf",
help="Time (gain) solutions interval (default one per scan)")
parser.add_argument('--freq_sol_interval', dest='freq_sol_interval', default="inf",
help="Frequency time-invariant solutions interval (default one per observation)")
parser.add_argument('--clip_delays', dest='clip_delays', default=1, type=float,
help="Clip delays above this absolute in nanoseconds")
parser.add_argument('--cal_model', dest='cal_model', default='pks1934-638.lsm',
help="Calibrator apparent sky model (tigger lsm format)")
parser.add_argument('--ref_ant', dest='ref_ant', default='m037',
help="Reference antenna to use throughout")
parser.add_argument("--containerization", dest="containerization", default="docker",
help="Containerization technology to use. See your stimela installation for options")
parser.add_argument("--image_gaincalibrators", dest="image_gaincalibrators", action="store_true",
help="Image gain calibrators")
parser.add_argument("--dont_prompt", dest="dont_prompt",
action="store_true",
help="Don't prompt the user for confirmation of parameters")
args = parser.parse_args(sys.argv[2:])
INPUT = args.input_directory
MSDIR = args.msdir_directory
OUTPUT = args.output_directory
vermeerkat.log.info("Directory '{0:s}' is used as input directory".format(INPUT))
vermeerkat.log.info("Directory '{0:s}' is used as output directory".format(OUTPUT))
vermeerkat.log.info("Directory '{0:s}' is used as msdir directory".format(MSDIR))
PREFIX = args.msprefix
ZEROGEN_DATA = PREFIX + ".ms"
vermeerkat.log.info("Dataset '{0:s}' to be used throughout".format(ZEROGEN_DATA))
with tbl(os.path.join(MSDIR, ZEROGEN_DATA)+"::FIELD", ack=False) as t:
field_names = t.getcol("NAME")
FDB = {fn: str(fni) for fni, fn in enumerate(field_names)}
def flistr():
vermeerkat.log.info("The following fields are available:")
for f in FDB:
vermeerkat.log.info("\t '{0:s}' index {1:s}".format(f, FDB[f]))
sys.exit(0)
def __merge_input():
mod_path = os.path.dirname(vermeerkat.__file__)
data_dir = os.path.join(mod_path, "data", "input")
shutil.copytree(data_dir, INPUT)
if not os.path.exists(INPUT):
__merge_input()
elif os.path.isdir(INPUT):
shutil.rmtree(INPUT)
__merge_input()
else:
raise RuntimeError("A file called {} already exists, but is not a input directory".format(INPUT))
vermeerkat.log.info("Time invariant solution time interval: {0:s}".format(args.time_sol_interval))
vermeerkat.log.info("Frequency invariant solution frequency interval: {0:s}".format(args.freq_sol_interval))
vermeerkat.log.info("Will clip absolute delays over {0:.2f}ns".format(args.clip_delays))
vermeerkat.log.info("Will use '{}' as flux calibrator full sky model".format(args.cal_model))
FLAGANT = [f[0] if isinstance(f, list) else f for f in args.flag_antenna]
if len(FLAGANT) != 0:
vermeerkat.log.info("Will flag antenna {}".format(", ".join(["'{}'".format(a) for a in FLAGANT])))
BPCALIBRATOR = args.bandpass_field
if BPCALIBRATOR is None: raise ValueError("No bandpass calibrator specified")
GCALIBRATOR = [f[0] if isinstance(f, list) else f for f in args.gain_field]
ALTCAL = [f[0] if isinstance(f, list) else f for f in args.alt_cal_field]
TARGET = [f[0] if isinstance(f, list) else f for f in args.target_field]
if len(TARGET) < 1: raise ValueError("No target specified")
DO_USE_GAINCALIBRATOR = len(GCALIBRATOR) > 0
if not DO_USE_GAINCALIBRATOR:
vermeerkat.log.info("*NO* gain calibrator specified")
DO_USE_GAINCALIBRATOR_DELAY = args.delay_with_gcal and DO_USE_GAINCALIBRATOR
if DO_USE_GAINCALIBRATOR_DELAY:
vermeerkat.log.info("Will transfer rate calibraton using gain calibrator")
else:
vermeerkat.log.info("Will *NOT* transfer rate calibraton from gain calibrator")
REFANT = args.ref_ant
vermeerkat.log.info("Reference antenna {0:s} to be used throughout".format(REFANT))
## DO NOT EDIT ANY OF THESE PREDEFINED WORKERS UNLESS YOU KNOW WHAT YOU'RE DOING
K0 = PREFIX + ".K0"
B0 = PREFIX + ".B0"
G0 = PREFIX + ".G0"
GA = PREFIX + ".GAlt"
G1 = PREFIX + ".G1"
F0 = PREFIX + ".F0"
B1 = PREFIX + ".B1"
MANUAL_FLAG_LIST = []
FIRSTGEN_DATA = ["{}.{}.1gc.ms".format(t, PREFIX) for t in TARGET]
vermeerkat.log.info("The following fields are available:")
for f in FDB:
vermeerkat.log.info("\t '{0:s}' index {1:s}{2:s}".format(
f, FDB[f],
" selected as 'BP'" if f == BPCALIBRATOR else
" selected as 'GC'" if f in GCALIBRATOR else
" selected as 'ALTCAL'" if f in ALTCAL else
" selected as 'TARGET'" if f in TARGET else
" not selected"))
try:
input = raw_input
except NameError:
pass
while not args.dont_prompt:
r = input("Is this configuration correct? (Y/N) >> ").lower()
if r == "y":
break
elif r == "n":
sys.exit(0)
else:
continue
stimela.register_globals()
recipe = stimela.Recipe('MEERKAT: basic transfer calibration',
ms_dir=MSDIR,
singularity_image_dir=os.environ.get("SINGULARITY_PULLFOLDER", ""),
JOB_TYPE=args.containerization)
def addmanualflags(recipe, reason, antenna="", spw="", scan="", uvrange="", field=""):
""" Read CASA flagdata docs before using """
recipe.add("cab/casa_flagdata", "handflags", {
"vis": ZEROGEN_DATA,
"mode": "manual",
"antenna": antenna,
"spw": spw,
"scan": scan,
"uvrange": uvrange,
"field": field
},
input=INPUT, output=OUTPUT, label=reason)
MANUAL_FLAG_LIST.append(reason)
return [reason]
def prepare_data():
recipe.add("cab/casa_flagmanager", "backup_CAM_SP_flags", {
"vis": ZEROGEN_DATA,
"mode": "save",
"versionname": "SP_ORIGINAL",
},
input=INPUT, output=OUTPUT, label="backup_CAM_SP_flags")
recipe.add("cab/casa_listobs", "listobs", {
"vis": ZEROGEN_DATA,
},
input=INPUT, output=OUTPUT, label="listobs")
recipe.add("cab/casa_flagdata", "flag_reset", {
"vis": ZEROGEN_DATA,
"mode": "unflag",
},
input=INPUT, output=OUTPUT, label="reset_flags")
recipe.add("cab/politsiyakat_autocorr_amp", "flag_autopower", {
"msname": ZEROGEN_DATA,
"field": ",".join(str(FDB[f]) for f in FDB),
"cal_field": ",".join([FDB[BPCALIBRATOR]] + [FDB[f] for f in GCALIBRATOR + ALTCAL]),
"nrows_chunk": 15000,
"scan_to_scan_threshold": 1.5,
"antenna_to_group_threshold": 4,
"output_dir": "./:output",
"nio_threads": 1,
"nproc_threads": 32,
"dpi": 80,
},input=INPUT, output=OUTPUT, label="flag_autopower")
recipe.add("cab/casa_flagdata", "flag_autocorrelations", {
"vis": ZEROGEN_DATA,
"mode": "manual",
"autocorr": True,
},
input=INPUT, output=OUTPUT, label="flagging_auto_correlations")
recipe.add("cab/casa_flagdata", "flag_rolloff", {
"vis": ZEROGEN_DATA,
"mode": "manual",
"spw": "*:850~980MHz,*:1658~1800MHz,*:1419.8~1421.3MHz", #band-rolloffs and Milkyway HI line
},
input=INPUT, output=OUTPUT, label="flagging_rolloff")
recipe.add("cab/casa_clearcal", "clear_calibration", {
"vis": ZEROGEN_DATA,
"addmodel": True,
},
input=INPUT, output=OUTPUT, label="clear_calibration")
return [
"backup_CAM_SP_flags",
"listobs",
####"reset_flags",
"flag_autopower",
"flagging_auto_correlations",
"flagging_rolloff",
"clear_calibration",
]
def rfiflag_data(do_flag_targets=False, steplabel="flagpass1", exec_strategy="mk_rfi_flagging_calibrator_fields_firstpass.yaml", on_corr_residuals=False, dc="DATA"):
recipe.add('cab/tricolour', steplabel,
{
"ms" : ZEROGEN_DATA,
"data-column" : dc,
"window-backend" : 'numpy',
"field-names" : [FDB[BPCALIBRATOR]],
"flagging-strategy" : "total_power" if not do_flag_targets else "polarisation",
"config" : exec_strategy,
"subtract-model-column": sdm.dismissable("MODEL_DATA" if on_corr_residuals else None),
"dilate-masks": sdm.dismissable(None),
"ignore-flags": sdm.dismissable(None),
"scan-numbers": sdm.dismissable(None),
},
input=INPUT, output=OUTPUT, label=steplabel)
recipe.add('cab/tricolour', steplabel + "_gc",
{
"ms" : ZEROGEN_DATA,
"data-column" : dc,
"window-backend" : 'numpy',
"field-names" : [FDB[t] for t in TARGET] if do_flag_targets else [FDB[t] for t in GCALIBRATOR + ALTCAL],
"flagging-strategy" : "total_power" if not do_flag_targets else "polarisation",
"subtract-model-column": sdm.dismissable("MODEL_DATA" if on_corr_residuals else None),
"config" : exec_strategy,
"dilate-masks": sdm.dismissable(None),
"ignore-flags": sdm.dismissable(None),
"scan-numbers": sdm.dismissable(None),
},
input=INPUT, output=OUTPUT, label=steplabel + ".gc" if not do_flag_targets else steplabel + ".targets")
recipe.add("cab/casa_flagdata", "flag_summary_{}".format(steplabel), {
"vis": ZEROGEN_DATA,
"mode": "summary"
},
input=INPUT, output=OUTPUT, label="flagging_summary_{}".format(steplabel))
return (([steplabel, steplabel + ".gc"] if len(ALTCAL) > 0 or DO_USE_GAINCALIBRATOR else [steplabel])
if not do_flag_targets else [steplabel + ".targets"]) + \
[
"flagging_summary_{}".format(steplabel)
]
def image_calibrator(recipe, label="prelim"):
imfields = [FDB[a] for a in ALTCAL] + \
([FDB[t] for t in GCALIBRATOR] if (DO_USE_GAINCALIBRATOR and
DO_USE_GAINCALIBRATOR_DELAY) else [])
steps = []
for f in imfields:
imopts={
"msname": ZEROGEN_DATA,
"join-channels": True,
"channels-out": 9,
"size": 4096,
"scale": "1.6asec",
"mgain": 0.8,
"gain": 0.1,
"niter": 3000,
"name": "calfield-{}-{}".format(f, label),
"field": f,
"fits-mask": sdm.dismissable(None),
###"save-source-list": True,
"fit-spectral-pol": 3,
}
maskname = "MASK-{}-{}.fits".format(f, label)
recipe.add("cab/wsclean", "image_{}_field{}".format(label, f),
imopts,
input=INPUT, output=OUTPUT, label="image_calfield_{}_{}".format(f, label))
recipe.add("cab/cleanmask", "mask_{}_{}".format(label, f), {
'image': "calfield-{}-{}-MFS-image.fits:output".format(f, label),
'output': maskname,
'sigma': 35,
'boxes': 9,
'iters': 20,
'overlap': 0.3,
'no-negative': True,
'tolerance': 0.75,
}, input=INPUT, output=OUTPUT, label='mask_{}_{}'.format(label, f))
imopts2 = {k: v for k, v in list(imopts.items())}
imopts2["fits-mask"] = maskname + ":output"
imopts2["local-rms"] = True
imopts2["auto-threshold"] = 5
recipe.add("cab/wsclean", "image_{}_field{}_rnd2".format(label, f),
imopts2,
input=INPUT, output=OUTPUT, label="image_calfield_{}_{}_rnd2".format(f, label))
steps += ["image_calfield_{}_{}".format(f, label),
'mask_{}_{}'.format(label, f),
"image_calfield_{}_{}_rnd2".format(f, label)]
return steps
def do_1GC(recipe, label="prelim", do_apply_target=False, do_predict=True, applyonly=False):
recipe.add("cab/casa_flagmanager", "backup_flags_prior_1gc_%s" % label, {
"vis": ZEROGEN_DATA,
"mode": "save",
"versionname": "prior_%s_1GC" % label,
},
input=INPUT, output=OUTPUT, label="backup_flags_prior_1gc_%s" % label)
recipe.add("cab/simulator", "predict_fluxcalibrator_%s" % label, {
"skymodel": args.cal_model, # we are using 1934-638 as flux scale reference
"msname": ZEROGEN_DATA,
"threads": 24,
"mode": "simulate",
"column": "MODEL_DATA",
"Ejones": False, # first bandpass calibration is normal calibration then we absorb coefficients into another table
"beam-files-pattern": "meerkat_pb_jones_cube_95channels_$(xy)_$(reim).fits",
"beam-l-axis": "X",
"beam-m-axis": "-Y", #[OMS] flipped in code: southern hemisphere
"parallactic-angle-rotation": True,
"field-id": int(FDB[BPCALIBRATOR]),
},
input=INPUT, output=OUTPUT, label="set_flux_reference_%s" % label)
recipe.add("cab/casa47_gaincal", "delaycal_%s" % label, {
"vis": ZEROGEN_DATA,
"caltable": K0,
"field": ",".join([FDB[BPCALIBRATOR]]),
"refant": REFANT,
"solint": args.time_sol_interval,
"combine": "",
"minsnr": 3,
"minblperant": 4,
"gaintype": "K",
},
input=INPUT, output=OUTPUT, label="delay_calibration_bp_%s" % label)
def clip_delays(vis, clipminmax):
with tbl(os.path.join(OUTPUT, vis), readonly=False) as t:
fl = t.getcol("FLAG")
d = t.getcol("FPARAM")
prevflagged = np.sum(fl) * 100.0 / fl.size
fl = np.logical_or(fl,
np.logical_or(d.real > np.max(clipminmax),
d.real < np.min(clipminmax)))
t.putcol("FLAG", fl)
currentflagged = np.sum(fl) * 100.0 / fl.size
vermeerkat.log.info("Flagged {0:.2f}%, up from previous {1:.2f}%".format(currentflagged,
prevflagged))
recipe.add(clip_delays, "clipdel_%s" % label, {
"vis": K0,
"clipminmax": [-args.clip_delays, +args.clip_delays],
},
input=INPUT, output=OUTPUT, label="clip_delay_%s" % label)
##capture time drift of bandpass
recipe.add("cab/casa47_gaincal", "bandpassgain_%s" % label, {
"vis": ZEROGEN_DATA,
"caltable": G0,
"field": ",".join([FDB[BPCALIBRATOR]]),
"solint": args.time_sol_interval,
"combine": "",
"gaintype": "G",
"uvrange": "150~10000m", # EXCLUDE RFI INFESTATION!
##"spw": "0:1.3~1.5GHz",
"gaintable": ["%s:output" % ct for ct in [K0]],
"gainfield": [FDB[BPCALIBRATOR]],
"interp":["nearest"],
"refant": REFANT,
},
input=INPUT, output=OUTPUT, label="remove_bp_average_%s" % label)
# average as much as possible to get as good SNR on bandpass as possible
recipe.add("cab/casa47_bandpass", "bandpasscal_%s" % label, {
"vis": ZEROGEN_DATA,
"caltable": "%s:output" % B0,
"field": ",".join([FDB[BPCALIBRATOR]]),
"solint": args.freq_sol_interval,
"combine": "scan",
"minsnr": 3.0,
"uvrange": "150~10000m", # EXCLUDE RFI INFESTATION!
#"fillgaps": 100000000, # LERP!
"gaintable": ["%s:output" % ct for ct in [K0, G0]],
"gainfield": [FDB[BPCALIBRATOR], FDB[BPCALIBRATOR]],
"interp": ["nearest", "nearest"],
"refant": REFANT,
},
input=INPUT, output=OUTPUT, label="bp_freq_calibration_%s" % label)
recipe.add("cab/casa47_applycal", "apply_sols_bp_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[BPCALIBRATOR]] +
[FDB[a] for a in ALTCAL] +
([FDB[t] for t in GCALIBRATOR] if (DO_USE_GAINCALIBRATOR and DO_USE_GAINCALIBRATOR_DELAY) else
[])),
"gaintable": ["%s:output" % ct for ct in [K0,G0,B0]],
"gainfield": [",".join([FDB[BPCALIBRATOR]]),
",".join([FDB[BPCALIBRATOR]]),
",".join([FDB[BPCALIBRATOR]])],
"interp": ["nearest","nearest","linear,linear"]
},
input=INPUT, output=OUTPUT, label="apply_sols_bp_%s" % label)
#create basic model for secondaries
cal_im_steps = image_calibrator(recipe=recipe, label=label) if args.image_gaincalibrators else []
##capture time drift of gain and alternative calibrators
recipe.add("cab/casa47_gaincal", "delaycal_gc_%s" % label, {
"vis": ZEROGEN_DATA,
"caltable": K0,
"field": ",".join([FDB[a] for a in ALTCAL] +
[FDB[t] for t in GCALIBRATOR]) if (DO_USE_GAINCALIBRATOR and
DO_USE_GAINCALIBRATOR_DELAY) else
",".join([FDB[a] for a in ALTCAL]),
"refant": REFANT,
"solint": args.time_sol_interval,
"combine": "",
"minsnr": 3,
"minblperant": 4,
"gaintype": "K",
"gaintable": ["%s:output" % ct for ct in [B0]],
"gainfield": [",".join([FDB[BPCALIBRATOR]])],
"interp": ["linear,linear"],
##"spw": "0:1.3~1.5GHz",
"uvrange": "150~10000m", # EXCLUDE RFI INFESTATION!
"append": True,
"refant": REFANT,
},
input=INPUT, output=OUTPUT, label="delay_calibration_gc_%s" % label)
recipe.add(clip_delays, "clipdel_%s" % label, {
"vis": K0,
"clipminmax": [-args.clip_delays, +args.clip_delays],
},
input=INPUT, output=OUTPUT, label="clip_delay_gc_%s" % label)
recipe.add("cab/msutils", "delaycal_plot_%s" % label, {
"command": "plot_gains",
"ctable": "%s:output" % K0,
"tabtype": "delay",
"plot_file": "{0:s}.{1:s}.K0.png".format(PREFIX, label),
"subplot_scale": 4,
"plot_dpi": 180
},
input=INPUT, output=OUTPUT, label="plot_delays_%s" % label)
if DO_USE_GAINCALIBRATOR:
recipe.add("cab/casa47_gaincal", "apgain_%s" % label, {
"vis": ZEROGEN_DATA,
"caltable": G0,
"field": ",".join([FDB[t] for t in GCALIBRATOR]),
"solint": args.time_sol_interval,
"combine": "",
"gaintype": "G",
"uvrange": "150~10000m", # EXCLUDE RFI INFESTATION!
##"spw": "0:1.3~1.5GHz",
"gaintable": ["%s:output" % ct for ct in [B0, K0]],
"gainfield": [FDB[BPCALIBRATOR],
",".join([FDB[a] for a in ALTCAL] +
[FDB[t] for t in GCALIBRATOR]) if (DO_USE_GAINCALIBRATOR and
DO_USE_GAINCALIBRATOR_DELAY) else
",".join([FDB[a] for a in ALTCAL]),
],
"interp":["linear,linear", "nearest"],
"append": True,
"refant": REFANT,
},
input=INPUT, output=OUTPUT, label="apgain_%s" % label)
recipe.add("cab/casa_fluxscale", "fluxscale_%s" % label, {
"vis": ZEROGEN_DATA,
"caltable": "%s:output" % G0,
"fluxtable": "%s:output" % F0,
"reference": ",".join([FDB[BPCALIBRATOR]]),
"transfer": ",".join([FDB[t] for t in GCALIBRATOR]),
},
input=INPUT, output=OUTPUT, label="fluxscale_%s" % label)
recipe.add("cab/msutils", "gain_plot_%s" % label, {
"command": "plot_gains",
"ctable": "%s:output" % (F0 if DO_USE_GAINCALIBRATOR else G0),
"tabtype": "gain",
"plot_file": "{0:s}.{1:s}.F0.png".format(PREFIX, label),
"subplot_scale": 4,
"plot_dpi": 180
},
input=INPUT, output=OUTPUT, label="plot_gain_%s" % label)
recipe.add("cab/msutils", "bpgain_plot_%s" % label, {
"command": "plot_gains",
"ctable": "%s:output" % B0,
"tabtype": "bandpass",
"plot_file": "{0:s}.{1:s}.B0.png".format(PREFIX, label),
"subplot_scale": 4,
"plot_dpi": 180
},
input=INPUT, output=OUTPUT, label="plot_bpgain_%s" % label)
if len(ALTCAL) > 0:
# no model of alternatives, don't adjust amp
recipe.add("cab/casa47_gaincal", "altcalgain_%s" % label, {
"vis": ZEROGEN_DATA,
"caltable": GA,
"field": ",".join([FDB[a] for a in ALTCAL]),
"solint": args.time_sol_interval,
"combine": "",
"gaintype": "G",
"calmode": "p",
"uvrange": "150~10000m", # EXCLUDE RFI INFESTATION!
##"spw": "0:1.3~1.5GHz",
"gaintable": ["%s:output" % ct for ct in [K0, G0, B0]],
"gainfield": [
",".join([FDB[a] for a in ALTCAL]),
FDB[BPCALIBRATOR],
FDB[BPCALIBRATOR]],
"interp":["linear,linear","nearest"],
"refant": REFANT,
},
input=INPUT, output=OUTPUT, label="remove_altcal_average_%s" % label)
recipe.add("cab/msutils", "altgain_plot_%s" % label, {
"command": "plot_gains",
"ctable": "%s:output" % GA,
"tabtype": "gain",
"plot_file": "{0:s}.{1:s}.GA.png".format(PREFIX, label),
"subplot_scale": 4,
"plot_dpi": 180
},
input=INPUT, output=OUTPUT, label="plot_altgains_%s" % label)
for a in ALTCAL:
recipe.add("cab/casa47_applycal", "apply_sols_ac_%s_%s" % (FDB[a], label), {
"vis": ZEROGEN_DATA,
"field": FDB[a],
"gaintable": ["%s:output" % ct for ct in [K0,G0,B0,GA]],
"gainfield": [
",".join([FDB[a]]),
",".join([FDB[BPCALIBRATOR]]),
",".join([FDB[BPCALIBRATOR]]),
",".join([FDB[a]]),
],
"interp": ["linear,linear","nearest","nearest"]
},
input=INPUT, output=OUTPUT, label="apply_sols_ac_%s_%s" % (FDB[a], label))
if do_apply_target or DO_USE_GAINCALIBRATOR:
recipe.add("cab/casa47_applycal", "apply_sols_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[t] for t in GCALIBRATOR] +
([FDB[t] for t in TARGET] if do_apply_target else [])),
"gaintable": ["%s:output" % ct for ct in [B0,K0,F0]] if DO_USE_GAINCALIBRATOR else ["%s:output" % ct for ct in [B0,K0,G0]],
"gainfield": [FDB[BPCALIBRATOR],
",".join([FDB[t] for t in GCALIBRATOR])
if (DO_USE_GAINCALIBRATOR and DO_USE_GAINCALIBRATOR_DELAY) else FDB[BPCALIBRATOR],
",".join([FDB[t] for t in GCALIBRATOR])
if DO_USE_GAINCALIBRATOR else FDB[BPCALIBRATOR],
],
"interp": ["linear,linear","nearest","nearest"]
},
input=INPUT, output=OUTPUT, label="apply_1GC_solutions_%s" % label)
recipe.add("cab/casa_plotms", "plot_pa_bp_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[BPCALIBRATOR]]),
"correlation": "XX,YY",
"xaxis": "amp",
"xdatacolumn": "corrected/model_vector",
"yaxis": "phase",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"overwrite": True,
"showgui": False,
"avgtime": "32",
"avgchannel": "32",
"plotfile": "{}.{}.bp.ampphase.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="phaseamp_plot_for_bandpass_%s" % label)
recipe.add("cab/casa_plotms", "plot_pa_gcal_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[t] for t in GCALIBRATOR]),
"correlation": "XX,YY",
"xaxis": "amp",
"xdatacolumn": "corrected/model_vector",
"yaxis": "phase",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"iteraxis": "field",
"overwrite": True,
"showgui": False,
"avgtime": "32",
"avgchannel": "32",
"plotfile": "{}.{}.gc.ampphase.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="phaseamp_plot_for_gain_%s" % label)
recipe.add("cab/casa_plotms", "plot_ri_gcal_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[t] for t in GCALIBRATOR + ALTCAL]),
"correlation": "XX,YY",
"xaxis": "real",
"xdatacolumn": "corrected/model_vector",
"yaxis": "imag",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"iteraxis": "field",
"overwrite": True,
"showgui": False,
"avgtime": "32",
"avgchannel": "32",
"plotfile": "{}.{}.gc.realimag.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="reim_plot_for_gain_%s" % label)
recipe.add("cab/casa_plotms", "plot_ri_bpcal_%s" % label, {
"vis": ZEROGEN_DATA,
"field": FDB[BPCALIBRATOR],
"correlation": "XX,YY",
"xaxis": "real",
"xdatacolumn": "corrected/model_vector",
"yaxis": "imag",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"iteraxis": "field",
"overwrite": True,
"showgui": False,
"avgtime": "32",
"avgchannel": "32",
"plotfile": "{}.{}.bp.realimag.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="reim_plot_for_bp_%s" % label)
recipe.add("cab/casa_plotms", "plot_afreq_gcal_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[t] for t in GCALIBRATOR + ALTCAL]),
"correlation": "XX,YY",
"xaxis": "freq",
"yaxis": "amp",
"xdatacolumn": "corrected/model_vector",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"iteraxis": "field",
"overwrite": True,
"showgui": False,
"avgtime": "64",
"plotfile": "{}.{}.gc.ampfreq.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="afreq_for_gain_%s" % label)
recipe.add("cab/casa_plotms", "plot_pfreq_gcal_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[t] for t in GCALIBRATOR + ALTCAL]),
"correlation": "XX,YY",
"xaxis": "freq",
"yaxis": "phase",
"xdatacolumn": "corrected/model_vector",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"iteraxis": "field",
"overwrite": True,
"showgui": False,
"avgtime": "64",
"plotfile": "{}.{}.gc.phasefreq.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="pfreq_for_gain_%s" % label)
recipe.add("cab/casa_plotms", "plot_ascan_gcal_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[t] for t in GCALIBRATOR + ALTCAL]),
"correlation": "XX,YY",
"xaxis": "scan",
"yaxis": "amp",
"xdatacolumn": "corrected/model_vector",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"iteraxis": "field",
"overwrite": True,
"showgui": False,
"avgtime": "64",
"avgchannel": "32",
"plotfile": "{}.{}.gc.ampscan.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="ascan_for_gain_%s" % label)
recipe.add("cab/casa_plotms", "plot_pscan_gcal_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[t] for t in GCALIBRATOR + ALTCAL]),
"correlation": "XX,YY",
"xaxis": "scan",
"yaxis": "phase",
"xdatacolumn": "corrected/model_vector",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"iteraxis": "field",
"overwrite": True,
"showgui": False,
"avgtime": "64",
"avgchannel": "32",
"plotfile": "{}.{}.gc.phasescan.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="pscan_for_gain_%s" % label)
recipe.add("cab/casa_plotms", "plot_afreq_bpcal_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[BPCALIBRATOR]]),
"correlation": "XX,YY",
"xaxis": "freq",
"yaxis": "amp",
"xdatacolumn": "corrected/model_vector",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"overwrite": True,
"showgui": False,
"avgtime": "64",
"plotfile": "{}.{}.bp.ampfreq.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="afreq_for_bp_%s" % label)
recipe.add("cab/casa_plotms", "plot_pfreq_bpcal_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[BPCALIBRATOR]]),
"correlation": "XX,YY",
"xaxis": "freq",
"yaxis": "phase",
"xdatacolumn": "corrected/model_vector",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"overwrite": True,
"showgui": False,
"avgtime": "64",
"plotfile": "{}.{}.bp.phasefreq.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="pfreq_for_bp_%s" % label)
recipe.add("cab/casa_plotms", "plot_ascan_bpcal_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[BPCALIBRATOR]]),
"correlation": "XX,YY",
"xaxis": "scan",
"yaxis": "amp",
"xdatacolumn": "corrected/model_vector",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"overwrite": True,
"showgui": False,
"avgtime": "64",
"avgchannel": "32",
"plotfile": "{}.{}.bp.ampscan.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="ascan_for_bp_%s" % label)
recipe.add("cab/casa_plotms", "plot_pscan_bpcal_%s" % label, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[BPCALIBRATOR]]),
"correlation": "XX,YY",
"xaxis": "scan",
"yaxis": "phase",
"xdatacolumn": "corrected/model_vector",
"ydatacolumn": "corrected/model_vector",
"coloraxis": "baseline",
"expformat": "png",
"exprange": "all",
"overwrite": True,
"showgui": False,
"avgtime": "64",
"avgchannel": "32",
"plotfile": "{}.{}.bp.phasescan.png".format(PREFIX, label)
},
input=INPUT, output=OUTPUT, label="pscan_for_bp_%s" % label)
return ([
"backup_flags_prior_1gc_{}".format(label)
] +
[
"set_flux_reference_{}".format(label)
] if do_predict else []) + ([
"delay_calibration_bp_{}".format(label),
"clip_delay_{}".format(label),
"remove_bp_average_{}".format(label),
"bp_freq_calibration_{}".format(label),
"apply_sols_bp_{}".format(label),
] + cal_im_steps +
([
"delay_calibration_gc_{}".format(label),
"clip_delay_gc_{}".format(label),
] if len(ALTCAL) > 0 or DO_USE_GAINCALIBRATOR_DELAY else []) + ([
"apgain_{}".format(label),
"fluxscale_{}".format(label),
] if DO_USE_GAINCALIBRATOR else []) + ([
"remove_altcal_average_{}".format(label),
"plot_altgains_{}".format(label),
] if len(ALTCAL) > 0 else [])
if not applyonly else [
"apply_sols_bp_{}".format(label)
]) +\
[
"plot_delays_{}".format(label),
"plot_gain_{}".format(label),
"plot_bpgain_{}".format(label),
"apply_sols_bp_{}".format(label),
] + ([
"apply_1GC_solutions_{}".format(label)
] if do_apply_target or DO_USE_GAINCALIBRATOR else []) +\
[
"apply_sols_ac_{0:s}_{1:s}".format(FDB[a], label) for a in ALTCAL
] +\
[
"phaseamp_plot_for_bandpass_{}".format(label),
"reim_plot_for_bp_{}".format(label),
"afreq_for_bp_{}".format(label),
"pfreq_for_bp_{}".format(label),
"ascan_for_bp_{}".format(label),
"pscan_for_bp_{}".format(label)
] + ([
"afreq_for_gain_{}".format(label),
"pfreq_for_gain_{}".format(label),
"ascan_for_gain_{}".format(label),
"pscan_for_gain_{}".format(label),
"phaseamp_plot_for_gain_{}".format(label),
"reim_plot_for_gain_{}".format(label),
] if len(ALTCAL) > 0 or DO_USE_GAINCALIBRATOR else [])
def finalize_and_split():
for ti, t in enumerate(TARGET):
recipe.add("cab/casa_split", "split_%d" % ti, {
"vis": ZEROGEN_DATA,
"field": ",".join([FDB[t]]),
"outputvis": FIRSTGEN_DATA[ti]
},
input=INPUT, output=OUTPUT, label="split_%d" % ti)
recipe.add("cab/casa_flagmanager", "backup_1GC_flags_%d" % ti, {
"vis": FIRSTGEN_DATA[ti],
"mode": "save",
"versionname": "1GC_LEGACY",
},
input=INPUT, output=OUTPUT, label="backup_1GC_flags_%d" % ti)
return ["split_%d" % ti for ti, t in enumerate(TARGET)] + \
["backup_1GC_flags_%d" % ti for ti, t in enumerate(TARGET)]
def define_steps():
STEPS = []
if not args.skip_prepdata:
STEPS += prepare_data()
for a in FLAGANT:
STEPS += addmanualflags(recipe, "Pointing issue {}".format(a), antenna=a, spw="", scan="", uvrange="", field="")
if not args.skip_prelim_flagging:
STEPS += rfiflag_data(do_flag_targets=False, steplabel="flagpass1", exec_strategy="mk_rfi_flagging_calibrator_fields_firstpass.yaml", on_corr_residuals=False, dc="DATA")
if not args.skip_prelim_1GC:
STEPS += do_1GC(recipe, label="prelim", do_predict=True)
if not args.skip_final_flagging:
STEPS += rfiflag_data(do_flag_targets=False, steplabel="flagpass2", exec_strategy="mk_rfi_flagging_calibrator_fields_secondpass.yaml", on_corr_residuals=True, dc="CORRECTED_DATA")
if not args.skip_final_1GC:
STEPS += do_1GC(recipe, label="second_round", do_predict=False, do_apply_target=False)
if not args.skip_transfer_to_targets:
STEPS += do_1GC(recipe, label="apply_only", do_predict=False, do_apply_target=True, applyonly=True)
if not args.skip_flag_targets:
STEPS += rfiflag_data(do_flag_targets=True, steplabel="flagfinal", exec_strategy="mk_rfi_flagging_target_fields_firstpass.yaml", on_corr_residuals=False, dc="CORRECTED_DATA")
if not args.skip_final_split:
STEPS += finalize_and_split()
checked_opts = OrderedDict()
for o in STEPS: checked_opts[o] = True
return checked_opts
def compile_and_run(STEPS):
if len(STEPS) != 0:
recipe.run(STEPS)
def main():
steps = define_steps()
compile_and_run(list(steps.keys()))
if __name__ == "__main__":
main()
| [
"stimela.dismissable.dismissable",
"numpy.sum",
"argparse.ArgumentParser",
"shutil.rmtree",
"vermeerkat.log.info",
"os.path.isdir",
"os.path.dirname",
"os.path.exists",
"stimela.register_globals",
"os.environ.get",
"numpy.max",
"numpy.min",
"collections.OrderedDict",
"shutil.copytree",
"... | [((277, 345), 'argparse.ArgumentParser', 'argparse.ArgumentParser', (['"""MeerKAT BasicApplyTransfer (BAT) pipeline"""'], {}), "('MeerKAT BasicApplyTransfer (BAT) pipeline')\n", (300, 345), False, 'import argparse\n'), ((8170, 8228), 'vermeerkat.log.info', 'vermeerkat.log.info', (['"""The following fields are available:"""'], {}), "('The following fields are available:')\n", (8189, 8228), False, 'import vermeerkat\n'), ((8809, 8835), 'stimela.register_globals', 'stimela.register_globals', ([], {}), '()\n', (8833, 8835), False, 'import stimela\n'), ((5782, 5840), 'vermeerkat.log.info', 'vermeerkat.log.info', (['"""The following fields are available:"""'], {}), "('The following fields are available:')\n", (5801, 5840), False, 'import vermeerkat\n'), ((5935, 5946), 'sys.exit', 'sys.exit', (['(0)'], {}), '(0)\n', (5943, 5946), False, 'import sys\n'), ((5984, 6020), 'os.path.dirname', 'os.path.dirname', (['vermeerkat.__file__'], {}), '(vermeerkat.__file__)\n', (5999, 6020), False, 'import os\n'), ((6036, 6075), 'os.path.join', 'os.path.join', (['mod_path', '"""data"""', '"""input"""'], {}), "(mod_path, 'data', 'input')\n", (6048, 6075), False, 'import os\n'), ((6080, 6112), 'shutil.copytree', 'shutil.copytree', (['data_dir', 'INPUT'], {}), '(data_dir, INPUT)\n', (6095, 6112), False, 'import shutil\n'), ((6121, 6142), 'os.path.exists', 'os.path.exists', (['INPUT'], {}), '(INPUT)\n', (6135, 6142), False, 'import os\n'), ((6169, 6189), 'os.path.isdir', 'os.path.isdir', (['INPUT'], {}), '(INPUT)\n', (6182, 6189), False, 'import os\n'), ((7413, 7466), 'vermeerkat.log.info', 'vermeerkat.log.info', (['"""*NO* gain calibrator specified"""'], {}), "('*NO* gain calibrator specified')\n", (7432, 7466), False, 'import vermeerkat\n'), ((7581, 7655), 'vermeerkat.log.info', 'vermeerkat.log.info', (['"""Will transfer rate calibraton using gain calibrator"""'], {}), "('Will transfer rate calibraton using gain calibrator')\n", (7600, 7655), False, 'import vermeerkat\n'), ((7666, 7745), 'vermeerkat.log.info', 'vermeerkat.log.info', (['"""Will *NOT* transfer rate calibraton from gain calibrator"""'], {}), "('Will *NOT* transfer rate calibraton from gain calibrator')\n", (7685, 7745), False, 'import vermeerkat\n'), ((41749, 41762), 'collections.OrderedDict', 'OrderedDict', ([], {}), '()\n', (41760, 41762), False, 'from collections import OrderedDict\n'), ((6195, 6215), 'shutil.rmtree', 'shutil.rmtree', (['INPUT'], {}), '(INPUT)\n', (6208, 6215), False, 'import shutil\n'), ((8983, 9027), 'os.environ.get', 'os.environ.get', (['"""SINGULARITY_PULLFOLDER"""', '""""""'], {}), "('SINGULARITY_PULLFOLDER', '')\n", (8997, 9027), False, 'import os\n'), ((5603, 5636), 'os.path.join', 'os.path.join', (['MSDIR', 'ZEROGEN_DATA'], {}), '(MSDIR, ZEROGEN_DATA)\n', (5615, 5636), False, 'import os\n'), ((8769, 8780), 'sys.exit', 'sys.exit', (['(0)'], {}), '(0)\n', (8777, 8780), False, 'import sys\n'), ((12594, 12654), 'stimela.dismissable.dismissable', 'sdm.dismissable', (["('MODEL_DATA' if on_corr_residuals else None)"], {}), "('MODEL_DATA' if on_corr_residuals else None)\n", (12609, 12654), True, 'import stimela.dismissable as sdm\n'), ((12690, 12711), 'stimela.dismissable.dismissable', 'sdm.dismissable', (['None'], {}), '(None)\n', (12705, 12711), True, 'import stimela.dismissable as sdm\n'), ((12747, 12768), 'stimela.dismissable.dismissable', 'sdm.dismissable', (['None'], {}), '(None)\n', (12762, 12768), True, 'import stimela.dismissable as sdm\n'), ((12804, 12825), 'stimela.dismissable.dismissable', 'sdm.dismissable', (['None'], {}), '(None)\n', (12819, 12825), True, 'import stimela.dismissable as sdm\n'), ((13383, 13443), 'stimela.dismissable.dismissable', 'sdm.dismissable', (["('MODEL_DATA' if on_corr_residuals else None)"], {}), "('MODEL_DATA' if on_corr_residuals else None)\n", (13398, 13443), True, 'import stimela.dismissable as sdm\n'), ((13536, 13557), 'stimela.dismissable.dismissable', 'sdm.dismissable', (['None'], {}), '(None)\n', (13551, 13557), True, 'import stimela.dismissable as sdm\n'), ((13593, 13614), 'stimela.dismissable.dismissable', 'sdm.dismissable', (['None'], {}), '(None)\n', (13608, 13614), True, 'import stimela.dismissable as sdm\n'), ((13650, 13671), 'stimela.dismissable.dismissable', 'sdm.dismissable', (['None'], {}), '(None)\n', (13665, 13671), True, 'import stimela.dismissable as sdm\n'), ((14963, 14984), 'stimela.dismissable.dismissable', 'sdm.dismissable', (['None'], {}), '(None)\n', (14978, 14984), True, 'import stimela.dismissable as sdm\n'), ((18001, 18026), 'os.path.join', 'os.path.join', (['OUTPUT', 'vis'], {}), '(OUTPUT, vis)\n', (18013, 18026), False, 'import os\n'), ((18145, 18155), 'numpy.sum', 'np.sum', (['fl'], {}), '(fl)\n', (18151, 18155), True, 'import numpy as np\n'), ((18420, 18430), 'numpy.sum', 'np.sum', (['fl'], {}), '(fl)\n', (18426, 18430), True, 'import numpy as np\n'), ((18263, 18281), 'numpy.max', 'np.max', (['clipminmax'], {}), '(clipminmax)\n', (18269, 18281), True, 'import numpy as np\n'), ((18337, 18355), 'numpy.min', 'np.min', (['clipminmax'], {}), '(clipminmax)\n', (18343, 18355), True, 'import numpy as np\n')] |
#! /usr/bin/python
# -*- coding: utf-8 -*-
"""NAO tasks."""
import itertools
import os
import numpy as np
import torch
import torch.nn.functional as F
from fairseq import utils
from fairseq.data import ConcatDataset, NaoLanguagePairDataset
from fairseq.data.nao_dataset import SingleTensorDataset
from . import register_task
from .translation import TranslationTask
# TODO: Move it to molecule related folder.
DEFAULT_BOUNDS = {
'01': (0.0, 1.0),
'default': (0.0, 1.0),
'drd2-m1': (0.0, 0.05),
'qed-m1': (0.7, 0.8),
'logp04-m1': (-10.0, 2.0),
'logp06-m1': (-10.0, 4.0),
'drd2-default': (0.0, 1.0),
'qed-default': (0.0, 1.0),
'logp-default': (-10.0, 5.0),
}
def _get_bound(bound_str: str):
if bound_str is None:
return None
if ',' in bound_str:
low, high = bound_str.split(',')
return float(low), float(high)
bound = DEFAULT_BOUNDS.get(bound_str, None)
if bound is None:
print('| WARNING: default bound name {!r} not found, fall back to [0, 1].'.format(bound_str))
bound = (0.0, 1.0)
return bound
def load_score(
data_path, split, src, props, bounds,
combine, upsample_primary,
):
if len(props) > 1:
raise NotImplementedError('multiple property scores not supported now')
if len(props) != len(bounds):
raise RuntimeError('property and bound length mismatch')
prop = props[0]
bound = _get_bound(bounds[0])
prop_name = '' if prop is None else '-{}'.format(prop)
score_datasets = []
for k in itertools.count():
split_k = split + (str(k) if k > 0 else '')
filename = os.path.join(data_path, '{}.score{}.{}.npz'.format(split_k, prop_name, src))
if not os.path.exists(filename):
if k > 0:
break
else:
raise FileNotFoundError('Score dataset not found: {}, {} ({})'.format(split, prop, data_path))
data = np.load(filename)['arr_0']
if bound is None:
if np.any(data > 1.0) and np.any(data < 0.0):
raise RuntimeError('scores must be scaled to [0, 1]')
else:
assert bound[0] < bound[1]
data = np.maximum(data, bound[0])
data = np.minimum(data, bound[1])
data = (data - bound[0]) / (bound[1] - bound[0])
dataset = SingleTensorDataset(torch.from_numpy(data).to(dtype=torch.float32))
score_datasets.append(dataset)
print('| {} {} {}-score{} {} examples'.format(data_path, split_k, src, prop_name, len(score_datasets[-1])))
if not combine:
break
if len(score_datasets) == 1:
score_dataset = score_datasets[0]
else:
sample_ratios = [1] * len(score_datasets)
sample_ratios[0] = upsample_primary
score_dataset = ConcatDataset(score_datasets, sample_ratios)
return score_dataset
@register_task('nao_translation')
class NaoTranslationTask(TranslationTask):
"""Translation task with NAO prediction.
Include sources, targets and scores.
"""
def __init__(self, args, src_dict, tgt_dict):
super().__init__(args, src_dict, tgt_dict)
@staticmethod
def add_args(parser):
TranslationTask.add_args(parser)
parser.add_argument('--disable-score', action='store_true', default=False,
help='disable score dataset, train as common translation tasks')
parser.add_argument('--score-prop', default=None,
help='colon separated score property list, will use "score" name by default')
parser.add_argument('--score-bound', default=None,
help='colon separated score bound list in format "<LOW>,<HIGH>" or string name, '
'default is no scaling')
# Evaluation arguments.
parser.add_argument('--eval-score-only', action='store_true',
help='Only evaluate predicted scores for NAO tasks')
parser.add_argument('--nao-gen-step', action='store_true',
help='Generate new target sequences for NAO tasks')
parser.add_argument('--nao-lambda-max', type=float, default=1000.0,
help='Max value of NAO predict lambda, default is %(default)r')
@classmethod
def setup_task(cls, args, **kwargs):
instance = super().setup_task(args, **kwargs)
if instance.args.eval_score_only:
print('| NAO evaluate score only')
if instance.args.nao_gen_step:
print('| NAO generate | max-lambda {:6.1f}'.format(instance.args.nao_lambda_max))
return instance
def load_dataset(self, split, epoch=0, combine=False, **kwargs):
super().load_dataset(split, epoch=epoch, combine=combine, **kwargs)
if self.args.disable_score:
score = None
else:
paths = self.args.data.split(':')
assert len(paths) > 0
data_path = paths[epoch % len(paths)]
# infer langcode
src, tgt = self.args.source_lang, self.args.target_lang
if self.args.score_prop is None:
props = [None]
else:
props = self.args.score_prop.split(':')
if self.args.score_bound is None:
bounds = [None]
else:
bounds = self.args.score_bound.split(':')
score = load_score(
data_path, split, src, props, bounds,
combine=combine, upsample_primary=self.args.upsample_primary,
)
self.datasets[split] = NaoLanguagePairDataset.from_base_dataset(self.datasets[split], score)
def build_dataset_for_inference(self, src_tokens, src_lengths):
return NaoLanguagePairDataset(src_tokens, src_lengths, self.source_dictionary)
def valid_step(self, sample, model, criterion):
if self.args.eval_score_only:
assert hasattr(criterion, 'eval_score_only'), 'the criterion does not support --eval-score-only mode'
old_flag = criterion.eval_score_only
criterion.eval_score_only = True
loss, sample_size, logging_output = super().valid_step(sample, model, criterion)
criterion.eval_score_only = old_flag
return loss, sample_size, logging_output
else:
return super().valid_step(sample, model, criterion)
def predict_step(self, sample, model):
model.eval()
with torch.no_grad():
predict_value = model.encode_and_predict(**sample['net_input'])
return predict_value
def inference_step(self, generator, models, sample, prefix_tokens=None):
if self.args.nao_gen_step:
# TODO
pass
else:
return super().inference_step(generator, models, sample, prefix_tokens=prefix_tokens)
def generate_new_seq_step(self):
# TODO: Add new sequence generation step of NAO tasks.
pass
| [
"numpy.load",
"numpy.minimum",
"numpy.maximum",
"fairseq.data.ConcatDataset",
"os.path.exists",
"itertools.count",
"numpy.any",
"fairseq.data.NaoLanguagePairDataset.from_base_dataset",
"torch.no_grad",
"fairseq.data.NaoLanguagePairDataset",
"torch.from_numpy"
] | [((1550, 1567), 'itertools.count', 'itertools.count', ([], {}), '()\n', (1565, 1567), False, 'import itertools\n'), ((2826, 2870), 'fairseq.data.ConcatDataset', 'ConcatDataset', (['score_datasets', 'sample_ratios'], {}), '(score_datasets, sample_ratios)\n', (2839, 2870), False, 'from fairseq.data import ConcatDataset, NaoLanguagePairDataset\n'), ((5653, 5722), 'fairseq.data.NaoLanguagePairDataset.from_base_dataset', 'NaoLanguagePairDataset.from_base_dataset', (['self.datasets[split]', 'score'], {}), '(self.datasets[split], score)\n', (5693, 5722), False, 'from fairseq.data import ConcatDataset, NaoLanguagePairDataset\n'), ((5807, 5878), 'fairseq.data.NaoLanguagePairDataset', 'NaoLanguagePairDataset', (['src_tokens', 'src_lengths', 'self.source_dictionary'], {}), '(src_tokens, src_lengths, self.source_dictionary)\n', (5829, 5878), False, 'from fairseq.data import ConcatDataset, NaoLanguagePairDataset\n'), ((1734, 1758), 'os.path.exists', 'os.path.exists', (['filename'], {}), '(filename)\n', (1748, 1758), False, 'import os\n'), ((1949, 1966), 'numpy.load', 'np.load', (['filename'], {}), '(filename)\n', (1956, 1966), True, 'import numpy as np\n'), ((2202, 2228), 'numpy.maximum', 'np.maximum', (['data', 'bound[0]'], {}), '(data, bound[0])\n', (2212, 2228), True, 'import numpy as np\n'), ((2248, 2274), 'numpy.minimum', 'np.minimum', (['data', 'bound[1]'], {}), '(data, bound[1])\n', (2258, 2274), True, 'import numpy as np\n'), ((6532, 6547), 'torch.no_grad', 'torch.no_grad', ([], {}), '()\n', (6545, 6547), False, 'import torch\n'), ((2017, 2035), 'numpy.any', 'np.any', (['(data > 1.0)'], {}), '(data > 1.0)\n', (2023, 2035), True, 'import numpy as np\n'), ((2040, 2058), 'numpy.any', 'np.any', (['(data < 0.0)'], {}), '(data < 0.0)\n', (2046, 2058), True, 'import numpy as np\n'), ((2375, 2397), 'torch.from_numpy', 'torch.from_numpy', (['data'], {}), '(data)\n', (2391, 2397), False, 'import torch\n')] |
#! -*- coding: utf-8 -*-
# ็จGlobalPointerๅไธญๆๅฝๅๅฎไฝ่ฏๅซ
# ๆฐๆฎ้ https://github.com/CLUEbenchmark/CLUENER2020
import json
import numpy as np
from snippets import *
from bert4keras.backend import keras
from bert4keras.backend import multilabel_categorical_crossentropy
from bert4keras.layers import EfficientGlobalPointer as GlobalPointer
from bert4keras.snippets import sequence_padding, DataGenerator
from bert4keras.snippets import open
from tqdm import tqdm
maxlen = 256
epochs = 10
batch_size = 32
categories = set()
def load_data(filename):
"""ๅ ่ฝฝๆฐๆฎ
ๅๆกๆ ผๅผ๏ผ[text, (start, end, label), (start, end, label), ...]๏ผ
ๆๅณ็text[start:end + 1]ๆฏ็ฑปๅไธบlabel็ๅฎไฝใ
"""
D = []
with open(filename, encoding='utf-8') as f:
for l in f:
l = json.loads(l)
d = [l['text']]
for k, v in l.get('label', {}).items():
categories.add(k)
for spans in v.values():
for start, end in spans:
d.append((start, end, k))
D.append(d)
return D
# ๆ ๆณจๆฐๆฎ
train_data = load_data(data_path + 'cluener/train.json')
valid_data = load_data(data_path + 'cluener/dev.json')
categories = list(sorted(categories))
num_classes = len(categories)
class data_generator(DataGenerator):
"""ๆฐๆฎ็ๆๅจ
"""
def __iter__(self, random=False):
batch_token_ids, batch_segment_ids, batch_labels = [], [], []
for is_end, d in self.sample(random):
tokens = tokenizer.tokenize(d[0], maxlen=maxlen)
mapping = tokenizer.rematch(d[0], tokens)
start_mapping = {j[0]: i for i, j in enumerate(mapping) if j}
end_mapping = {j[-1]: i for i, j in enumerate(mapping) if j}
token_ids = tokenizer.tokens_to_ids(tokens)
segment_ids = [0] * len(token_ids)
labels = np.zeros((len(categories), maxlen, maxlen))
for start, end, label in d[1:]:
if start in start_mapping and end in end_mapping:
start = start_mapping[start]
end = end_mapping[end]
label = categories.index(label)
labels[label, start, end] = 1
batch_token_ids.append(token_ids)
batch_segment_ids.append(segment_ids)
batch_labels.append(labels[:, :len(token_ids), :len(token_ids)])
if len(batch_token_ids) == self.batch_size or is_end:
batch_token_ids = sequence_padding(batch_token_ids)
batch_segment_ids = sequence_padding(batch_segment_ids)
batch_labels = sequence_padding(batch_labels, seq_dims=3)
yield [batch_token_ids, batch_segment_ids], batch_labels
batch_token_ids, batch_segment_ids, batch_labels = [], [], []
# ่ฝฌๆขๆฐๆฎ้
train_generator = data_generator(train_data, batch_size)
valid_generator = data_generator(valid_data, batch_size)
def globalpointer_crossentropy(y_true, y_pred):
"""็ปGlobalPointer่ฎพ่ฎก็ไบคๅ็ต
"""
bh = K.prod(K.shape(y_pred)[:2])
y_true = K.reshape(y_true, (bh, -1))
y_pred = K.reshape(y_pred, (bh, -1))
return K.mean(multilabel_categorical_crossentropy(y_true, y_pred))
def globalpointer_f1score(y_true, y_pred):
"""็ปGlobalPointer่ฎพ่ฎก็F1
"""
y_pred = K.cast(K.greater(y_pred, 0), K.floatx())
return 2 * K.sum(y_true * y_pred) / K.sum(y_true + y_pred)
# ๆๅปบๆจกๅ
output = base.model.output
output = GlobalPointer(
heads=num_classes,
head_size=base.attention_head_size,
use_bias=False,
kernel_initializer=base.initializer
)(output)
model = keras.models.Model(base.model.input, output)
model.summary()
model.compile(
loss=globalpointer_crossentropy,
optimizer=optimizer,
metrics=[globalpointer_f1score]
)
class Evaluator(keras.callbacks.Callback):
"""ไฟๅญ้ช่ฏ้f1ๆๅฅฝ็ๆจกๅ
"""
def __init__(self):
self.best_val_f1 = 0
def on_epoch_end(self, epoch, logs=None):
f1, precision, recall = self.evaluate(valid_generator)
# ไฟๅญๆไผ
if f1 >= self.best_val_f1:
self.best_val_f1 = f1
model.save_weights('weights/cluener.weights')
print(
'valid: f1: %.5f, precision: %.5f, recall: %.5f, best f1: %.5f\n' %
(f1, precision, recall, self.best_val_f1)
)
def evaluate(self, data):
X, Y, Z = 1e-10, 1e-10, 1e-10
for x_true, y_true in data:
y_pred = (model.predict(x_true) > 0).astype(int)
X += (y_pred * y_true).sum()
Y += y_pred.sum()
Z += y_true.sum()
f1, precision, recall = 2 * X / (Y + Z), X / Y, X / Z
return f1, precision, recall
def test_predict(in_file, out_file):
"""่พๅบๆต่ฏ็ปๆๅฐๆไปถ
็ปๆๆไปถๅฏไปฅๆไบคๅฐ https://www.cluebenchmarks.com ่ฏๆตใ
"""
test_data = load_data(in_file)
test_generator = data_generator(test_data, batch_size)
results = []
for x_true, _ in tqdm(test_generator, ncols=0):
y_pred = model.predict(x_true)
for y in y_pred:
results.append(np.where(y > 0))
fw = open(out_file, 'w', encoding='utf-8')
with open(in_file) as fr:
for l, r in zip(fr, results):
l = json.loads(l)
l['label'] = {}
tokens = tokenizer.tokenize(l['text'], maxlen=maxlen)
mapping = tokenizer.rematch(l['text'], tokens)
for label, start, end in zip(*r):
label = categories[label]
start, end = mapping[start][0], mapping[end][-1]
if label not in l['label']:
l['label'][label] = {}
entity = l['text'][start:end + 1]
if entity not in l['label'][label]:
l['label'][label][entity] = []
l['label'][label][entity].append([start, end])
l = json.dumps(l, ensure_ascii=False)
fw.write(l + '\n')
fw.close()
if __name__ == '__main__':
evaluator = Evaluator()
model.fit_generator(
train_generator.forfit(),
steps_per_epoch=len(train_generator),
epochs=epochs,
callbacks=[evaluator]
)
model.load_weights('weights/cluener.weights')
test_predict(
in_file=data_path + 'cluener/test.json',
out_file='results/cluener_predict.json'
)
else:
model.load_weights('weights/cluener.weights')
| [
"tqdm.tqdm",
"json.loads",
"bert4keras.backend.multilabel_categorical_crossentropy",
"bert4keras.backend.keras.models.Model",
"json.dumps",
"bert4keras.snippets.open",
"numpy.where",
"bert4keras.layers.EfficientGlobalPointer",
"bert4keras.snippets.sequence_padding"
] | [((3612, 3656), 'bert4keras.backend.keras.models.Model', 'keras.models.Model', (['base.model.input', 'output'], {}), '(base.model.input, output)\n', (3630, 3656), False, 'from bert4keras.backend import keras\n'), ((3455, 3580), 'bert4keras.layers.EfficientGlobalPointer', 'GlobalPointer', ([], {'heads': 'num_classes', 'head_size': 'base.attention_head_size', 'use_bias': '(False)', 'kernel_initializer': 'base.initializer'}), '(heads=num_classes, head_size=base.attention_head_size,\n use_bias=False, kernel_initializer=base.initializer)\n', (3468, 3580), True, 'from bert4keras.layers import EfficientGlobalPointer as GlobalPointer\n'), ((4939, 4968), 'tqdm.tqdm', 'tqdm', (['test_generator'], {'ncols': '(0)'}), '(test_generator, ncols=0)\n', (4943, 4968), False, 'from tqdm import tqdm\n'), ((5088, 5125), 'bert4keras.snippets.open', 'open', (['out_file', '"""w"""'], {'encoding': '"""utf-8"""'}), "(out_file, 'w', encoding='utf-8')\n", (5092, 5125), False, 'from bert4keras.snippets import open\n'), ((696, 728), 'bert4keras.snippets.open', 'open', (['filename'], {'encoding': '"""utf-8"""'}), "(filename, encoding='utf-8')\n", (700, 728), False, 'from bert4keras.snippets import open\n'), ((3160, 3211), 'bert4keras.backend.multilabel_categorical_crossentropy', 'multilabel_categorical_crossentropy', (['y_true', 'y_pred'], {}), '(y_true, y_pred)\n', (3195, 3211), False, 'from bert4keras.backend import multilabel_categorical_crossentropy\n'), ((5135, 5148), 'bert4keras.snippets.open', 'open', (['in_file'], {}), '(in_file)\n', (5139, 5148), False, 'from bert4keras.snippets import open\n'), ((771, 784), 'json.loads', 'json.loads', (['l'], {}), '(l)\n', (781, 784), False, 'import json\n'), ((5210, 5223), 'json.loads', 'json.loads', (['l'], {}), '(l)\n', (5220, 5223), False, 'import json\n'), ((5849, 5882), 'json.dumps', 'json.dumps', (['l'], {'ensure_ascii': '(False)'}), '(l, ensure_ascii=False)\n', (5859, 5882), False, 'import json\n'), ((2482, 2515), 'bert4keras.snippets.sequence_padding', 'sequence_padding', (['batch_token_ids'], {}), '(batch_token_ids)\n', (2498, 2515), False, 'from bert4keras.snippets import sequence_padding, DataGenerator\n'), ((2552, 2587), 'bert4keras.snippets.sequence_padding', 'sequence_padding', (['batch_segment_ids'], {}), '(batch_segment_ids)\n', (2568, 2587), False, 'from bert4keras.snippets import sequence_padding, DataGenerator\n'), ((2619, 2661), 'bert4keras.snippets.sequence_padding', 'sequence_padding', (['batch_labels'], {'seq_dims': '(3)'}), '(batch_labels, seq_dims=3)\n', (2635, 2661), False, 'from bert4keras.snippets import sequence_padding, DataGenerator\n'), ((5061, 5076), 'numpy.where', 'np.where', (['(y > 0)'], {}), '(y > 0)\n', (5069, 5076), True, 'import numpy as np\n')] |
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import csv
array=[['wrong_covid_normal','wrong_covid_pneu','correct_covid'],
['wrong_pneu_normal','correct_pneu','wrong_pneu_covid'],
['correct_normal','wrong_normal_pneu','wrong_normal_covid']]
results={1:{},2:{},3:{},4:{},5:{}}
for fo in range(1,6):
report=pd.read_excel('results/selected_fold{}.xlsx'.format(fo))
data={}
for index,item in report.iterrows():
images_num=item['tp']+item['fp']
data[item['name']]=[[],[],[]]
acc=item['tp']/(item['tp']+item['fp'])
covid_recall=item['correct_covid']/item['covid_num']
covid_Specificity=(images_num-item['covid_num']-item['wrong_covid'])/(images_num-item['covid_num']-item['wrong_covid']+item['wrong_covid'])
covid_accuracy=(images_num-item['covid_num']-item['wrong_covid']+item['correct_covid'])/(images_num-item['covid_num']-item['wrong_covid']+item['correct_covid']+item['wrong_covid']+item['not_detected_covid'])
pneu_recall=item['correct_pneu']/item['pneu_num']
pneu_Specificity=(images_num-item['pneu_num']-item['wrong_pneu'])/(images_num-item['pneu_num']-item['wrong_pneu']+item['wrong_pneu'])
pneu_accuracy=(images_num-item['pneu_num']-item['wrong_pneu']+item['correct_pneu'])/(images_num-item['pneu_num']-item['wrong_pneu']+item['correct_pneu']+item['wrong_pneu']+item['not_detected_pneu'])
normal_recall=item['correct_normal']/item['normal_num']
normal_Specificity=(images_num-item['normal_num']-item['wrong_normal'])/(images_num-item['normal_num']-item['wrong_normal']+item['wrong_normal'])
normal_accuracy=(images_num-item['normal_num']-item['wrong_normal']+item['correct_normal'])/(images_num-item['normal_num']-item['wrong_normal']+item['correct_normal']+item['wrong_normal']+item['not_detected_normal'])
results[fo][item['name']]={'acc':acc,'covid_recall':covid_recall,'covid_Specificity':covid_Specificity,
'covid_accuracy':covid_accuracy,'pneu_recall':pneu_recall,'pneu_Specificity':pneu_Specificity,
'pneu_accuracy':pneu_accuracy,
'normal_recall':normal_recall,'normal_Specificity':normal_Specificity,
'normal_accuracy':normal_accuracy}
for nn,aa in enumerate(array):
for a in aa:
data[item['name']][nn].append(item[a])
for key in data:
gt = ['NORMAL','PNEUMONIA','COVID-19']
preds = ["COVID-19", "PNEUMONIA", "NORMAL",]
fig, ax = plt.subplots()
im = ax.imshow(np.array(data[key]), interpolation='nearest', cmap=plt.cm.Blues)
index=key.find('-')
if 'concatenat' in key:
ax.set(xticks=np.arange(np.array(data[key]).shape[1]),
yticks=np.arange(np.array(data[key]).shape[0]),
# ... and label them with the respective list entries
xticklabels=gt, yticklabels=preds,
title='Confusion Matrix for the concatenated network-fold{}'.format(fo),
ylabel='Ground Truth Label',
xlabel='Predicted Label')
else:
ax.set(xticks=np.arange(np.array(data[key]).shape[1]),
yticks=np.arange(np.array(data[key]).shape[0]),
# ... and label them with the respective list entries
xticklabels=gt, yticklabels=preds,
title='Confusion Matrix for {}-fold{}'.format(key[:index],fo),
ylabel='Ground Truth Label',
xlabel='Predicted Label')
# Rotate the tick labels and set their alignment.
plt.setp(ax.get_xticklabels(), rotation=45, ha="right",
rotation_mode="anchor")
fmt = '.2f'
thresh = 1000000.
# Loop over data dimensions and create text annotations.
for i in range(len(gt)):
for j in range(len(preds)):
ax.text(j, i, format(np.array(data[key])[i, j]),
ha="center", va="center",
color="white" if np.array(data[key])[i, j] > thresh else "black")
fig.tight_layout()
#plt.show()
dash=key.find('-')
plt.savefig('{}-fold{}-confusion_matrix.pdf'.format(key[:dash],fo))
results['Full']={'Xception':{}, 'concatenate':{},'ResNet50V2':{}}
results['average']={'Xception':{}, 'concatenate':{},'ResNet50V2':{}}
nets=['Xception','ResNet50V2','concatenate']
for net in nets:
for fokey in results:
for netkey in results[fokey]:
if net in netkey:
for param in results[fokey][netkey]:
if param not in results['Full'][net]:
results['Full'][net][param]=[]
results['Full'][net][param].append(results[fokey][netkey][param])
for net in results['Full']:
for param in results['Full'][net]:
results['average'][net][param]=np.average(results['Full'][net][param][:-1])
temp_data=[]
for fo in [1,2,3,4,5,'average']:
for net in results[fo]:
if 'Xception' in net:
temp_data.append([results[fo][net]['covid_Specificity'],
results[fo][net]['pneu_Specificity'],
results[fo][net]['normal_Specificity'],
results[fo][net]['covid_accuracy'],
results[fo][net]['pneu_accuracy'],
results[fo][net]['normal_accuracy']])
for net in results[fo]:
if 'ResNet' in net:
temp_data.append([results[fo][net]['covid_Specificity'],
results[fo][net]['pneu_Specificity'],
results[fo][net]['normal_Specificity'],
results[fo][net]['covid_accuracy'],
results[fo][net]['pneu_accuracy'],
results[fo][net]['normal_accuracy']])
for net in results[fo]:
if 'oncatenat' in net:
temp_data.append([results[fo][net]['covid_Specificity'],
results[fo][net]['pneu_Specificity'],
results[fo][net]['normal_Specificity'],
results[fo][net]['covid_accuracy'],
results[fo][net]['pneu_accuracy'],
results[fo][net]['normal_accuracy']])
| [
"numpy.array",
"numpy.average",
"matplotlib.pyplot.subplots"
] | [((2672, 2686), 'matplotlib.pyplot.subplots', 'plt.subplots', ([], {}), '()\n', (2684, 2686), True, 'import matplotlib.pyplot as plt\n'), ((5142, 5186), 'numpy.average', 'np.average', (["results['Full'][net][param][:-1]"], {}), "(results['Full'][net][param][:-1])\n", (5152, 5186), True, 'import numpy as np\n'), ((2711, 2730), 'numpy.array', 'np.array', (['data[key]'], {}), '(data[key])\n', (2719, 2730), True, 'import numpy as np\n'), ((4123, 4142), 'numpy.array', 'np.array', (['data[key]'], {}), '(data[key])\n', (4131, 4142), True, 'import numpy as np\n'), ((2875, 2894), 'numpy.array', 'np.array', (['data[key]'], {}), '(data[key])\n', (2883, 2894), True, 'import numpy as np\n'), ((2939, 2958), 'numpy.array', 'np.array', (['data[key]'], {}), '(data[key])\n', (2947, 2958), True, 'import numpy as np\n'), ((3320, 3339), 'numpy.array', 'np.array', (['data[key]'], {}), '(data[key])\n', (3328, 3339), True, 'import numpy as np\n'), ((3384, 3403), 'numpy.array', 'np.array', (['data[key]'], {}), '(data[key])\n', (3392, 3403), True, 'import numpy as np\n'), ((4252, 4271), 'numpy.array', 'np.array', (['data[key]'], {}), '(data[key])\n', (4260, 4271), True, 'import numpy as np\n')] |
# Python 3.7.6
# -*- coding: utf-8 -*-
# Author: <NAME>
import os
import numpy as np
import torch
from torch.nn.utils.rnn import pad_sequence
char_list = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789'
len_char_list = len(char_list)
def pad_labels():
path = os.getcwd()
labels = []
max_text_length = 0
np_data = np.load(path+'/val/val_labels.npy', allow_pickle=True)
np_data1 = np.load(path+'/test/test_labels.npy', allow_pickle=True)
np_data2 = np.load(path+'/train/train_labels.npy', allow_pickle=True)
datasets = ['train', 'val', 'test']
sizes = [160000, 20000, 20000]
for dataset in datasets:
dset = np.load(path+'/'+dataset+'/'+dataset+'_labels.npy', allow_pickle=True).tolist()
new_dset = []
for arr in dset:
#new_arr = arr.tolist()
new_arr = torch.from_numpy(arr).long().cuda()
if len(new_arr) > max_text_length:
max_text_length = len(new_arr)
new_dset.append(new_arr)
labels.extend(new_dset)
new_labels = pad_sequence(labels, padding_value=len_char_list).transpose(0,1).tolist()
tmp = 0
for dataset, size in zip(datasets, sizes):
new_dataset = new_labels[tmp:tmp+size]
res = torch.stack([torch.LongTensor(x) for x in new_dataset])
torch.save(res, path+'/'+dataset+'/'+dataset+'_labels.pt')
tmp += size
if __name__ == '__main__':
pad_labels() | [
"numpy.load",
"torch.LongTensor",
"os.getcwd",
"torch.save",
"torch.nn.utils.rnn.pad_sequence",
"torch.from_numpy"
] | [((301, 312), 'os.getcwd', 'os.getcwd', ([], {}), '()\n', (310, 312), False, 'import os\n'), ((388, 444), 'numpy.load', 'np.load', (["(path + '/val/val_labels.npy')"], {'allow_pickle': '(True)'}), "(path + '/val/val_labels.npy', allow_pickle=True)\n", (395, 444), True, 'import numpy as np\n'), ((459, 517), 'numpy.load', 'np.load', (["(path + '/test/test_labels.npy')"], {'allow_pickle': '(True)'}), "(path + '/test/test_labels.npy', allow_pickle=True)\n", (466, 517), True, 'import numpy as np\n'), ((532, 592), 'numpy.load', 'np.load', (["(path + '/train/train_labels.npy')"], {'allow_pickle': '(True)'}), "(path + '/train/train_labels.npy', allow_pickle=True)\n", (539, 592), True, 'import numpy as np\n'), ((1421, 1489), 'torch.save', 'torch.save', (['res', "(path + '/' + dataset + '/' + dataset + '_labels.pt')"], {}), "(res, path + '/' + dataset + '/' + dataset + '_labels.pt')\n", (1431, 1489), False, 'import torch\n'), ((736, 821), 'numpy.load', 'np.load', (["(path + '/' + dataset + '/' + dataset + '_labels.npy')"], {'allow_pickle': '(True)'}), "(path + '/' + dataset + '/' + dataset + '_labels.npy', allow_pickle=True\n )\n", (743, 821), True, 'import numpy as np\n'), ((1369, 1388), 'torch.LongTensor', 'torch.LongTensor', (['x'], {}), '(x)\n', (1385, 1388), False, 'import torch\n'), ((1152, 1201), 'torch.nn.utils.rnn.pad_sequence', 'pad_sequence', (['labels'], {'padding_value': 'len_char_list'}), '(labels, padding_value=len_char_list)\n', (1164, 1201), False, 'from torch.nn.utils.rnn import pad_sequence\n'), ((925, 946), 'torch.from_numpy', 'torch.from_numpy', (['arr'], {}), '(arr)\n', (941, 946), False, 'import torch\n')] |
from copy import deepcopy
import numpy as np
import pickle as pkl
import random
from joblib import Parallel, delayed
from tqdm.notebook import tqdm
import colorednoise as cn
import mne
from time import time
from . import util
DEFAULT_SETTINGS = {
'number_of_sources': (1, 20),
'extents': (1, 50),
'amplitudes': (1, 10),
'shapes': 'both',
'duration_of_trial': 0,
'sample_frequency': 100,
'target_snr': 4,
'beta': (0, 3),
}
class Simulation:
''' Simulate and hold source and M/EEG data.
Attributes
----------
settings : dict
The Settings for the simulation. Keys:
number_of_sources : int/tuple/list
number of sources. Can be a single number or a list of two numbers
specifying a range.
extents : int/float/tuple/list
size of sources in mm. Can be a single number or a list of two
numbers specifying a range.
amplitudes : int/float/tuple/list
the current of the source in nAm
shapes : str
How the amplitudes evolve over space. Can be 'gaussian' or 'flat'
(i.e. uniform) or 'both'.
duration_of_trial : int/float
specifies the duration of a trial.
sample_frequency : int
specifies the sample frequency of the data.
target_snr : float/tuple/list
The desired average SNR of the simulation(s)
beta : float/tuple/list
The desired frequency spectrum slope (1/f**beta) of the noise.
fwd : mne.Forward
The mne-python Forward object that contains the forward model
source_data : mne.sourceEstimate
A source estimate object from mne-python which contains the source
data.
eeg_data : mne.Epochs
A mne.Epochs object which contains the EEG data.
n_jobs : int
The number of jobs/cores to utilize.
Methods
-------
simulate : Simulate source and EEG data
plot : plot a random sample source and EEG
'''
def __init__(self, fwd, info, settings=DEFAULT_SETTINGS, n_jobs=-1,
parallel=False, verbose=False):
settings['sample_frequency'] = info['sfreq']
self.settings = settings
self.check_settings()
self.source_data = None
self.eeg_data = None
self.fwd = deepcopy(fwd)
self.fwd.pick_channels(info['ch_names'])
self.check_info(deepcopy(info))
self.info['sfreq'] = self.settings['sample_frequency']
self.subject = self.fwd['src'][0]['subject_his_id']
self.n_jobs = n_jobs
self.parallel = parallel
self.verbose = verbose
_, _, self.pos, _ = util.unpack_fwd(self.fwd)
def check_info(self, info):
self.info = info.pick_channels(self.fwd.ch_names, ordered=True)
def simulate(self, n_samples=10000):
''' Simulate sources and EEG data'''
self.source_data = self.simulate_sources(n_samples)
self.eeg_data = self.simulate_eeg()
return self
def plot(self):
pass
def simulate_sources(self, n_samples):
if self.verbose:
print(f'Simulate Source')
if self.parallel:
source_data = np.stack(Parallel(n_jobs=self.n_jobs, backend='loky')
(delayed(self.simulate_source)()
for _ in range(n_samples)))
else:
source_data = np.stack([self.simulate_source()
for _ in tqdm(range(n_samples))], axis=0)
# Convert to mne.SourceEstimate
if self.verbose:
print(f'Converting Source Data to mne.SourceEstimate object')
if self.settings['duration_of_trial'] == 0:
sources = util.source_to_sourceEstimate(source_data, self.fwd,
sfreq=self.settings['sample_frequency'], subject=self.subject)
else:
sources = self.sources_to_sourceEstimates(source_data)
return sources
def sources_to_sourceEstimates(self, source_data):
template = util.source_to_sourceEstimate(source_data[0],
self.fwd, sfreq=self.settings['sample_frequency'],
subject=self.subject)
sources = []
for source in tqdm(source_data):
tmp = deepcopy(template)
tmp.data = source
sources.append(tmp)
return sources
def simulate_source(self):
''' Returns a vector containing the dipole currents. Requires only a
dipole position list and the simulation settings.
Parameters
----------
pos : numpy.ndarray
(n_dipoles x 3), list of dipole positions.
number_of_sources : int/tuple/list
number of sources. Can be a single number or a list of two
numbers specifying a range.
extents : int/float/tuple/list
diameter of sources (in mm). Can be a single number or a list of
two numbers specifying a range.
amplitudes : int/float/tuple/list
the current of the source in nAm
shapes : str
How the amplitudes evolve over space. Can be 'gaussian' or 'flat'
(i.e. uniform) or 'both'.
duration_of_trial : int/float
specifies the duration of a trial.
sample_frequency : int
specifies the sample frequency of the data.
Return
------
source : numpy.ndarray, (n_dipoles x n_timepoints), the simulated
source signal
simSettings : dict, specifications about the source.
<NAME>., <NAME>., <NAME>., <NAME>., <NAME>., &
<NAME>. (2006). Evaluation of EEG localization methods using
realistic simulations of interictal spikes. Neuroimage, 29(3),
734-753.
'''
###########################################
# Select ranges and prepare some variables
# Get number of sources is a range:
number_of_sources = self.get_from_range(
self.settings['number_of_sources'], dtype=int)
# Get amplitudes for each source
extents = [self.get_from_range(self.settings['extents'], dtype=float)
for _ in range(number_of_sources)]
# Decide shape of sources
if self.settings['shapes'] == 'both':
shapes = ['gaussian', 'flat']*number_of_sources
np.random.shuffle(shapes)
shapes = shapes[:number_of_sources]
if type(shapes) == str:
shapes = [shapes]
elif self.settings['shapes'] == 'gaussian' or self.settings['shapes'] == 'flat':
shapes = [self.settings['shapes']] * number_of_sources
# Get amplitude gain for each source (amplitudes come in nAm)
amplitudes = [self.get_from_range(self.settings['amplitudes'], dtype=float) * 1e-9 for _ in range(number_of_sources)]
src_centers = np.random.choice(np.arange(self.pos.shape[0]), \
number_of_sources, replace=False)
if self.settings['duration_of_trial'] > 0:
signal_length = int(self.settings['sample_frequency']*self.settings['duration_of_trial'])
pulselen = self.settings['sample_frequency']/10
# pulse = self.get_pulse(pulselen)
signals = []
for _ in range(number_of_sources):
signal = cn.powerlaw_psd_gaussian(self.get_from_range(self.settings['beta'], dtype=float), signal_length)
# Old: have positive source values
# signal += np.abs(np.min(signal))
# signal /= np.max(signal)
# New:
signal /= np.max(np.abs(signal))
signals.append(signal)
sample_frequency = self.settings['sample_frequency']
else: # else its a single instance
sample_frequency = 0
signal_length = 1
signals = [np.array([1])]*number_of_sources
# sourceMask = np.zeros((self.pos.shape[0]))
source = np.zeros((self.pos.shape[0], signal_length))
##############################################
# Loop through source centers (i.e. seeds of source positions)
for i, (src_center, shape, amplitude, signal) in enumerate(zip(src_centers, shapes, amplitudes, signals)):
dists = np.sqrt(np.sum((self.pos - self.pos[src_center, :])**2, axis=1))
d = np.where(dists<extents[i]/2)[0]
if shape == 'gaussian':
sd = np.clip(np.max(dists[d]) / 2, a_min=0.1, a_max=np.inf) # <- works better
activity = np.expand_dims(util.gaussian(dists, 0, sd) * amplitude, axis=1) * signal
source += activity
elif shape == 'flat':
activity = util.repeat_newcol(amplitude * signal, len(d)).T
if len(activity.shape) == 1:
if len(d) == 1:
activity = np.expand_dims(activity, axis=0)
else:
activity = np.expand_dims(activity, axis=1)
source[d, :] += activity
else:
msg = BaseException("shape must be of type >string< and be either >gaussian< or >flat<.")
raise(msg)
# sourceMask[d] = 1
# if durOfTrial > 0:
# n = np.clip(int(sample_frequency * self.settings['duration_of_trial']), a_min=1, a_max=None)
# sourceOverTime = util.repeat_newcol(source, n)
# source = np.squeeze(sourceOverTime * signal)
# if len(source.shape) == 1:
# source = np.expand_dims(source, axis=1)
return source
def simulate_eeg(self):
''' Create EEG of specified number of trials based on sources and some SNR.
Parameters
-----------
sourceEstimates : list
list containing mne.SourceEstimate objects
fwd : mne.Forward
the mne.Forward object
target_snr : tuple/list/float,
desired signal to noise ratio. Can be a list or tuple of two
floats specifying a range.
beta : float
determines the frequency spectrum of the noise added to the signal:
power = 1/f^beta.
0 will yield white noise, 1 will yield pink noise (1/f spectrum)
n_jobs : int
Number of jobs to run in parallel. -1 will utilize all cores.
return_raw_data : bool
if True the function returns a list of mne.SourceEstimate
objects, otherwise it returns raw data
Return
-------
epochs : list
list of either mne.Epochs objects or list of raw EEG data
(see argument <return_raw_data> to change output)
'''
n_simulation_trials = 20
# Desired Dim of sources: (samples x dipoles x time points)
# unpack numpy array of source data
if isinstance(self.source_data, (list, tuple)):
sources = np.stack([source.data for source in self.source_data], axis=0)
else:
sources = self.source_data.data.T
# if there is no temporal dimension...
if len(sources.shape) < 3:
# ...add empty temporal dimension
sources = np.expand_dims(sources, axis=2)
# Load some forward model objects
fwd_fixed, leadfield = util.unpack_fwd(self.fwd)[:2]
n_samples, _, _ = sources.shape
n_elec = leadfield.shape[0]
# Desired Dim for eeg_clean: (samples, electrodes, time points)
if self.verbose:
print(f'\nProject sources to EEG...')
eeg_clean = self.project_sources(sources)
if self.verbose:
print(f'\nCreate EEG trials with noise...')
if self.parallel:
eeg_trials_noisy = np.stack(Parallel(n_jobs=self.n_jobs, backend='loky')
(delayed(self.create_eeg_helper)(eeg_clean[sample], n_simulation_trials,
self.settings['target_snr'], self.settings['beta'])
for sample in tqdm(range(n_samples))), axis=0)
else:
eeg_trials_noisy = np.stack(
[self.create_eeg_helper(eeg_clean[sample], n_simulation_trials,
self.settings['target_snr'], self.settings['beta'])
for sample in tqdm(range(n_samples))],
axis=0)
if n_simulation_trials == 1 and len(eeg_trials_noisy.shape) == 2:
# Add empty dimension to contain the single trial
eeg_trials_noisy = np.expand_dims(eeg_trials_noisy, axis=1)
if len(eeg_trials_noisy.shape) == 3:
eeg_trials_noisy = np.expand_dims(eeg_trials_noisy, axis=-1)
if eeg_trials_noisy.shape[2] != n_elec:
eeg_trials_noisy = np.swapaxes(eeg_trials_noisy, 1, 2)
if self.verbose:
print(f'\nConvert EEG matrices to a single instance of mne.Epochs...')
ERP_samples_noisy = np.mean(eeg_trials_noisy, axis=1)
epochs = util.eeg_to_Epochs(ERP_samples_noisy, fwd_fixed, info=self.info)
return epochs
def create_eeg_helper(self, eeg_sample, n_simulation_trials, target_snr, beta):
''' Helper function for EEG simulation that transforms a clean
M/EEG signal to a bunch of noisy trials.
Parameters
----------
eeg_sample : numpy.ndarray
data sample with dimension (time_points, electrodes)
n_simulation_trials : int
The number of trials desired
target_snr : float/list/tuple
The target signal-to-noise ratio, is converted to
single-trial SNR based on number of trials
beta : float/list/tuple
The beta exponent of the 1/f**beta noise
'''
target_snr = self.get_from_range(target_snr, dtype=float)
beta = self.get_from_range(beta, dtype=float)
assert len(eeg_sample.shape) == 2, 'Length of eeg_sample must be 2 (time_points, electrodes)'
eeg_sample = np.repeat(np.expand_dims(eeg_sample, 0), n_simulation_trials, axis=0)
snr = target_snr / np.sqrt(n_simulation_trials)
# Before: Add noise based on the GFP of all channels
# noise_trial = self.add_noise(eeg_sample, snr, beta=beta)
# NEW: ADD noise for different types of channels, separately
# since they can have entirely different scales.
coil_types = [ch['coil_type'] for ch in self.info['chs']]
coil_types_set = list(set(coil_types))
if len(coil_types_set)>1:
msg = f'Simulations attempted with more than one channel type \
({coil_types_set}) may result in unexpected behavior. Please \
select one channel type in your data only'
raise ValueError(msg)
coil_types_set = np.array([int(i) for i in coil_types_set])
coil_type_assignments = np.array(
[np.where(coil_types_set==coil_type)[0][0]
for coil_type in coil_types]
)
noise_trial = np.zeros(
(eeg_sample.shape[0], eeg_sample.shape[1], eeg_sample.shape[2])
)
for i, coil_type in enumerate(coil_types_set):
channel_indices = np.where(coil_type_assignments==i)[0]
eeg_sample_temp = eeg_sample[:, channel_indices, :]
noise_trial_subtype = self.add_noise(eeg_sample_temp, snr, beta=beta)
noise_trial[:, channel_indices, :] = noise_trial_subtype
return noise_trial
def project_sources(self, sources):
''' Project sources through the leadfield to obtain the EEG data.
Parameters
----------
sources : numpy.ndarray
3D array of shape (samples, dipoles, time points)
Return
------
'''
fwd_fixed, leadfield = util.unpack_fwd(self.fwd)[:2]
n_samples, n_dipoles, n_timepoints = sources.shape
n_elec = leadfield.shape[0]
eeg = np.zeros((n_samples, n_elec, n_timepoints))
# Swap axes to dipoles, samples, time_points
sources_tmp = np.swapaxes(sources, 0,1)
# Collapse last two dims into one
short_shape = (sources_tmp.shape[0],
sources_tmp.shape[1]*sources_tmp.shape[2])
sources_tmp = sources_tmp.reshape(short_shape)
# Scale to allow for lower precision
scaler = 1/sources_tmp.max()
sources_tmp *= scaler
# Perform Matmul
result = np.matmul(
leadfield.astype(np.float32), sources_tmp.astype(np.float32))
# Reshape result
result = result.reshape(result.shape[0], n_samples, n_timepoints)
# swap axes to correct order
result = np.swapaxes(result,0,1)
# Rescale
result /= scaler
return result
def add_noise(self, x, snr, beta=0):
""" Add noise of given SNR to signal x.
Parameters:
-----------
x : numpy.ndarray, 3-dimensional numpy array of dims (trials, channels, timepoints)
Return:
-------
"""
# This looks inconvenient but we need to make sure that there is no empty dimension for the powerlaw noise function.
x_shape = (x.shape[0], x.shape[1], np.clip(x.shape[2], a_min=2, a_max=np.inf).astype(int))
noise = cn.powerlaw_psd_gaussian(beta, x_shape)
# In case we added another entry in the 2nd dimension we have to remove it here again.
if x_shape[2] != x.shape[2]:
noise=noise[:, :, :1]
noise_gfp = np.std(noise, axis=1)
rms_noise = np.mean(noise_gfp) # rms(noise)
x_gfp = np.std(x, axis=1)
rms_x = np.mean(np.max(np.abs(x_gfp), axis=1)) # x.max()
# rms_noise = rms(noise-np.mean(noise))
noise_scaler = rms_x / (rms_noise*snr)
out = x + noise*noise_scaler
return out
def check_settings(self):
''' Check if settings are complete and insert missing
entries if there are any.
'''
# Check for wrong keys:
for key in self.settings.keys():
if not key in DEFAULT_SETTINGS.keys():
msg = f'key {key} is not part of allowed settings. See DEFAULT_SETTINGS for reference: {DEFAULT_SETTINGS}'
raise AttributeError(msg)
# Check for missing keys and replace them from the DEFAULT_SETTINGS
for key in DEFAULT_SETTINGS.keys():
# Check if setting exists and is not None
if not (key in self.settings.keys() and self.settings[key] is not None):
self.settings[key] = DEFAULT_SETTINGS[key]
if self.settings['duration_of_trial'] == 0:
self.temporal = False
else:
self.temporal = True
@staticmethod
def get_pulse(pulse_len):
''' Returns a pulse of given length. A pulse is defined as
half a revolution of a sine.
Parameters
----------
x : int
the number of data points
'''
pulse_len = int(pulse_len)
freq = (1/pulse_len) / 2
time = np.arange(pulse_len)
signal = np.sin(2*np.pi*freq*time)
return signal
@staticmethod
def get_from_range(val, dtype=int):
''' If list of two integers/floats is given this method outputs a value in between the two values.
Otherwise, it returns the value.
Parameters
----------
val : list/tuple/int/float
Return
------
out : int/float
'''
if dtype==int:
rng = random.randrange
elif dtype==float:
rng = random.uniform
else:
msg = f'dtype must be int or float, got {type(dtype)} instead'
raise AttributeError(msg)
if isinstance(val, (list, tuple, np.ndarray)):
out = rng(*val)
elif isinstance(val, (int, float)):
out = val
return out
def save(self, file_name):
''' Store the simulation object.
Parameters
----------
file_name : str
Filename or full path to store the object to.
Example
-------
sim = Simulation().simulate()
sim.save('C/Users/User/Desktop/simulation.pkl')
'''
with open(file_name, 'wb') as f:
pkl.dump(self, f)
def to_nontemporal(self):
''' Converts the internal data representation from temporal to
non-temporal.
Specifically, this changes the shape of sources from a
list of mne.sourceEstimate to a single mne.sourceEstimate in which the
time dimension holds a concatenation of timepoints and samples.
The eeg data is reshaped from (samples, channels, time points) to
(samples*time points, channels, 1).
Parameters
----------
Return
------
self : esinet.Simulation
Method returns itself for convenience
'''
if not self.temporal:
print('This Simulation() instance is already non-temporal')
return self
self.temporal = False
self.settings['duration_of_trial'] = 0
eeg_data_lstm = self.eeg_data.get_data()
# Reshape EEG data
eeg_data_single = np.expand_dims(np.vstack(np.swapaxes(eeg_data_lstm, 1,2)), axis=-1)
# Pack into mne.EpochsArray object
epochs_single = mne.EpochsArray(eeg_data_single, self.eeg_data.info,
tmin=self.eeg_data.tmin, verbose=0)
# Store the newly shaped data
self.eeg_data = epochs_single
# Reshape Source data
source_data = np.vstack(np.swapaxes(np.stack(
[source.data for source in self.source_data], axis=0), 1,2)).T
# Pack into mne.SourceEstimate object
source_single = deepcopy(self.source_data[0])
source_single.data = source_data
self.source_data = source_single
return self
| [
"pickle.dump",
"numpy.sum",
"numpy.abs",
"tqdm.notebook.tqdm",
"numpy.clip",
"numpy.mean",
"numpy.arange",
"numpy.sin",
"numpy.std",
"numpy.max",
"numpy.swapaxes",
"numpy.random.shuffle",
"numpy.stack",
"copy.deepcopy",
"colorednoise.powerlaw_psd_gaussian",
"numpy.zeros",
"numpy.expa... | [((2411, 2424), 'copy.deepcopy', 'deepcopy', (['fwd'], {}), '(fwd)\n', (2419, 2424), False, 'from copy import deepcopy\n'), ((4332, 4349), 'tqdm.notebook.tqdm', 'tqdm', (['source_data'], {}), '(source_data)\n', (4336, 4349), False, 'from tqdm.notebook import tqdm\n'), ((8199, 8243), 'numpy.zeros', 'np.zeros', (['(self.pos.shape[0], signal_length)'], {}), '((self.pos.shape[0], signal_length))\n', (8207, 8243), True, 'import numpy as np\n'), ((13256, 13289), 'numpy.mean', 'np.mean', (['eeg_trials_noisy'], {'axis': '(1)'}), '(eeg_trials_noisy, axis=1)\n', (13263, 13289), True, 'import numpy as np\n'), ((15396, 15469), 'numpy.zeros', 'np.zeros', (['(eeg_sample.shape[0], eeg_sample.shape[1], eeg_sample.shape[2])'], {}), '((eeg_sample.shape[0], eeg_sample.shape[1], eeg_sample.shape[2]))\n', (15404, 15469), True, 'import numpy as np\n'), ((16343, 16386), 'numpy.zeros', 'np.zeros', (['(n_samples, n_elec, n_timepoints)'], {}), '((n_samples, n_elec, n_timepoints))\n', (16351, 16386), True, 'import numpy as np\n'), ((16463, 16489), 'numpy.swapaxes', 'np.swapaxes', (['sources', '(0)', '(1)'], {}), '(sources, 0, 1)\n', (16474, 16489), True, 'import numpy as np\n'), ((17079, 17104), 'numpy.swapaxes', 'np.swapaxes', (['result', '(0)', '(1)'], {}), '(result, 0, 1)\n', (17090, 17104), True, 'import numpy as np\n'), ((17685, 17724), 'colorednoise.powerlaw_psd_gaussian', 'cn.powerlaw_psd_gaussian', (['beta', 'x_shape'], {}), '(beta, x_shape)\n', (17709, 17724), True, 'import colorednoise as cn\n'), ((17925, 17946), 'numpy.std', 'np.std', (['noise'], {'axis': '(1)'}), '(noise, axis=1)\n', (17931, 17946), True, 'import numpy as np\n'), ((17967, 17985), 'numpy.mean', 'np.mean', (['noise_gfp'], {}), '(noise_gfp)\n', (17974, 17985), True, 'import numpy as np\n'), ((18025, 18042), 'numpy.std', 'np.std', (['x'], {'axis': '(1)'}), '(x, axis=1)\n', (18031, 18042), True, 'import numpy as np\n'), ((19540, 19560), 'numpy.arange', 'np.arange', (['pulse_len'], {}), '(pulse_len)\n', (19549, 19560), True, 'import numpy as np\n'), ((19579, 19610), 'numpy.sin', 'np.sin', (['(2 * np.pi * freq * time)'], {}), '(2 * np.pi * freq * time)\n', (19585, 19610), True, 'import numpy as np\n'), ((21897, 21990), 'mne.EpochsArray', 'mne.EpochsArray', (['eeg_data_single', 'self.eeg_data.info'], {'tmin': 'self.eeg_data.tmin', 'verbose': '(0)'}), '(eeg_data_single, self.eeg_data.info, tmin=self.eeg_data.\n tmin, verbose=0)\n', (21912, 21990), False, 'import mne\n'), ((22313, 22342), 'copy.deepcopy', 'deepcopy', (['self.source_data[0]'], {}), '(self.source_data[0])\n', (22321, 22342), False, 'from copy import deepcopy\n'), ((2498, 2512), 'copy.deepcopy', 'deepcopy', (['info'], {}), '(info)\n', (2506, 2512), False, 'from copy import deepcopy\n'), ((4369, 4387), 'copy.deepcopy', 'deepcopy', (['template'], {}), '(template)\n', (4377, 4387), False, 'from copy import deepcopy\n'), ((6512, 6537), 'numpy.random.shuffle', 'np.random.shuffle', (['shapes'], {}), '(shapes)\n', (6529, 6537), True, 'import numpy as np\n'), ((7066, 7094), 'numpy.arange', 'np.arange', (['self.pos.shape[0]'], {}), '(self.pos.shape[0])\n', (7075, 7094), True, 'import numpy as np\n'), ((11242, 11304), 'numpy.stack', 'np.stack', (['[source.data for source in self.source_data]'], {'axis': '(0)'}), '([source.data for source in self.source_data], axis=0)\n', (11250, 11304), True, 'import numpy as np\n'), ((11516, 11547), 'numpy.expand_dims', 'np.expand_dims', (['sources'], {'axis': '(2)'}), '(sources, axis=2)\n', (11530, 11547), True, 'import numpy as np\n'), ((12814, 12854), 'numpy.expand_dims', 'np.expand_dims', (['eeg_trials_noisy'], {'axis': '(1)'}), '(eeg_trials_noisy, axis=1)\n', (12828, 12854), True, 'import numpy as np\n'), ((12941, 12982), 'numpy.expand_dims', 'np.expand_dims', (['eeg_trials_noisy'], {'axis': '(-1)'}), '(eeg_trials_noisy, axis=-1)\n', (12955, 12982), True, 'import numpy as np\n'), ((13075, 13110), 'numpy.swapaxes', 'np.swapaxes', (['eeg_trials_noisy', '(1)', '(2)'], {}), '(eeg_trials_noisy, 1, 2)\n', (13086, 13110), True, 'import numpy as np\n'), ((14348, 14377), 'numpy.expand_dims', 'np.expand_dims', (['eeg_sample', '(0)'], {}), '(eeg_sample, 0)\n', (14362, 14377), True, 'import numpy as np\n'), ((14435, 14463), 'numpy.sqrt', 'np.sqrt', (['n_simulation_trials'], {}), '(n_simulation_trials)\n', (14442, 14463), True, 'import numpy as np\n'), ((20792, 20809), 'pickle.dump', 'pkl.dump', (['self', 'f'], {}), '(self, f)\n', (20800, 20809), True, 'import pickle as pkl\n'), ((8522, 8579), 'numpy.sum', 'np.sum', (['((self.pos - self.pos[src_center, :]) ** 2)'], {'axis': '(1)'}), '((self.pos - self.pos[src_center, :]) ** 2, axis=1)\n', (8528, 8579), True, 'import numpy as np\n'), ((8595, 8627), 'numpy.where', 'np.where', (['(dists < extents[i] / 2)'], {}), '(dists < extents[i] / 2)\n', (8603, 8627), True, 'import numpy as np\n'), ((15578, 15614), 'numpy.where', 'np.where', (['(coil_type_assignments == i)'], {}), '(coil_type_assignments == i)\n', (15586, 15614), True, 'import numpy as np\n'), ((18074, 18087), 'numpy.abs', 'np.abs', (['x_gfp'], {}), '(x_gfp)\n', (18080, 18087), True, 'import numpy as np\n'), ((21787, 21819), 'numpy.swapaxes', 'np.swapaxes', (['eeg_data_lstm', '(1)', '(2)'], {}), '(eeg_data_lstm, 1, 2)\n', (21798, 21819), True, 'import numpy as np\n'), ((3321, 3365), 'joblib.Parallel', 'Parallel', ([], {'n_jobs': 'self.n_jobs', 'backend': '"""loky"""'}), "(n_jobs=self.n_jobs, backend='loky')\n", (3329, 3365), False, 'from joblib import Parallel, delayed\n'), ((7814, 7828), 'numpy.abs', 'np.abs', (['signal'], {}), '(signal)\n', (7820, 7828), True, 'import numpy as np\n'), ((8078, 8091), 'numpy.array', 'np.array', (['[1]'], {}), '([1])\n', (8086, 8091), True, 'import numpy as np\n'), ((12083, 12127), 'joblib.Parallel', 'Parallel', ([], {'n_jobs': 'self.n_jobs', 'backend': '"""loky"""'}), "(n_jobs=self.n_jobs, backend='loky')\n", (12091, 12127), False, 'from joblib import Parallel, delayed\n'), ((17613, 17655), 'numpy.clip', 'np.clip', (['x.shape[2]'], {'a_min': '(2)', 'a_max': 'np.inf'}), '(x.shape[2], a_min=2, a_max=np.inf)\n', (17620, 17655), True, 'import numpy as np\n'), ((22158, 22220), 'numpy.stack', 'np.stack', (['[source.data for source in self.source_data]'], {'axis': '(0)'}), '([source.data for source in self.source_data], axis=0)\n', (22166, 22220), True, 'import numpy as np\n'), ((8693, 8709), 'numpy.max', 'np.max', (['dists[d]'], {}), '(dists[d])\n', (8699, 8709), True, 'import numpy as np\n'), ((15276, 15313), 'numpy.where', 'np.where', (['(coil_types_set == coil_type)'], {}), '(coil_types_set == coil_type)\n', (15284, 15313), True, 'import numpy as np\n'), ((3383, 3412), 'joblib.delayed', 'delayed', (['self.simulate_source'], {}), '(self.simulate_source)\n', (3390, 3412), False, 'from joblib import Parallel, delayed\n'), ((9120, 9152), 'numpy.expand_dims', 'np.expand_dims', (['activity'], {'axis': '(0)'}), '(activity, axis=0)\n', (9134, 9152), True, 'import numpy as np\n'), ((9218, 9250), 'numpy.expand_dims', 'np.expand_dims', (['activity'], {'axis': '(1)'}), '(activity, axis=1)\n', (9232, 9250), True, 'import numpy as np\n'), ((12145, 12176), 'joblib.delayed', 'delayed', (['self.create_eeg_helper'], {}), '(self.create_eeg_helper)\n', (12152, 12176), False, 'from joblib import Parallel, delayed\n')] |
# -*- coding: utf-8 -*-
import numpy as np
NUMPY_COMPLEX128_MAX = np.finfo(np.complex128).max
NUMPY_LOG_COMPLEX128_MAX = np.log(NUMPY_COMPLEX128_MAX)
class HestonModel:
def __init__(self, forward, vol, kappa, theta, sigma, rho, rate):
self.forward = forward
self.vol = vol
self.kappa = kappa
self.theta = theta
self.sigma = sigma
self.rho = rho
self.rate = rate
def __str__(self):
out_str = f"forward: {self.forward}\n\r" +\
f"vol: {self.vol}\n\r" +\
f"kappa: {self.kappa}\n\r" +\
f"theta: {self.theta}\n\r" +\
f"sigma: {self.sigma}\n\r" +\
f"rho: {self.rho}\n\r" + \
f"rate: {self.rate}\n\r"
return out_str
def cf(self, z, tau) -> complex:
beta = self.kappa - 1j * self.sigma * self.rho * z
sigma_sq = self.sigma * self.sigma
D = np.sqrt(beta * beta + sigma_sq * z * (z + 1j))
if beta.real * D.real + beta.imag * D.imag > 0:
r = - sigma_sq * z * (z + 1j) / (beta + D)
else:
r = beta - D
if D != 0:
y = np.expm1(-D * tau) / (2 * D)
else:
y = -tau / 2
A = self.kappa * self.theta / sigma_sq * \
(r * tau - 2 * np.log1p(- r * y))
B = z * (z + 1j) * y / (1 - r * y)
exponent = A + B * self.vol
if exponent > NUMPY_LOG_COMPLEX128_MAX:
raise OverflowError("too large exponent in characteristic function")
return np.exp(exponent)
def log_cf_real(self, alpha, tau) -> float:
# Evaluation of ln HestomModel.cf(-1j * (1 + alpha))
beta = self.kappa - self.rho * self.sigma * (1 + alpha)
Dsq = beta**2 - self.sigma**2 * (1 + alpha) * alpha
if Dsq > 0:
D = np.sqrt(Dsq)
coshdt = np.cosh(D * tau / 2)
sinhdt = np.sinh(D * tau / 2) / D
nume = coshdt + beta * sinhdt
else:
# D = 1j * x
x = np.sqrt(-Dsq)
coshdt = np.cos(x * tau / 2)
sinhdt = np.sin(x * tau / 2) / x
nume = coshdt + beta * sinhdt
A = self.kappa * self.theta / self.sigma**2 *\
(beta * tau - np.log(nume**2))
B = alpha * (1 + alpha) * sinhdt / nume
return A + B * self.vol
class BlackScholesModel():
def __init__(self, forward, vol, rate):
self.forward = forward
self.vol = vol
self.rate = rate
def cf(self, z, tau):
return np.exp(-0.5 * self.vol * tau * z * (z + 1j))
| [
"numpy.log",
"numpy.finfo",
"numpy.expm1",
"numpy.sin",
"numpy.exp",
"numpy.cos",
"numpy.cosh",
"numpy.sinh",
"numpy.log1p",
"numpy.sqrt"
] | [((122, 150), 'numpy.log', 'np.log', (['NUMPY_COMPLEX128_MAX'], {}), '(NUMPY_COMPLEX128_MAX)\n', (128, 150), True, 'import numpy as np\n'), ((67, 90), 'numpy.finfo', 'np.finfo', (['np.complex128'], {}), '(np.complex128)\n', (75, 90), True, 'import numpy as np\n'), ((952, 1000), 'numpy.sqrt', 'np.sqrt', (['(beta * beta + sigma_sq * z * (z + 1.0j))'], {}), '(beta * beta + sigma_sq * z * (z + 1.0j))\n', (959, 1000), True, 'import numpy as np\n'), ((1579, 1595), 'numpy.exp', 'np.exp', (['exponent'], {}), '(exponent)\n', (1585, 1595), True, 'import numpy as np\n'), ((2601, 2647), 'numpy.exp', 'np.exp', (['(-0.5 * self.vol * tau * z * (z + 1.0j))'], {}), '(-0.5 * self.vol * tau * z * (z + 1.0j))\n', (2607, 2647), True, 'import numpy as np\n'), ((1875, 1887), 'numpy.sqrt', 'np.sqrt', (['Dsq'], {}), '(Dsq)\n', (1882, 1887), True, 'import numpy as np\n'), ((1909, 1929), 'numpy.cosh', 'np.cosh', (['(D * tau / 2)'], {}), '(D * tau / 2)\n', (1916, 1929), True, 'import numpy as np\n'), ((2086, 2099), 'numpy.sqrt', 'np.sqrt', (['(-Dsq)'], {}), '(-Dsq)\n', (2093, 2099), True, 'import numpy as np\n'), ((2121, 2140), 'numpy.cos', 'np.cos', (['(x * tau / 2)'], {}), '(x * tau / 2)\n', (2127, 2140), True, 'import numpy as np\n'), ((1186, 1204), 'numpy.expm1', 'np.expm1', (['(-D * tau)'], {}), '(-D * tau)\n', (1194, 1204), True, 'import numpy as np\n'), ((1951, 1971), 'numpy.sinh', 'np.sinh', (['(D * tau / 2)'], {}), '(D * tau / 2)\n', (1958, 1971), True, 'import numpy as np\n'), ((2162, 2181), 'numpy.sin', 'np.sin', (['(x * tau / 2)'], {}), '(x * tau / 2)\n', (2168, 2181), True, 'import numpy as np\n'), ((2310, 2327), 'numpy.log', 'np.log', (['(nume ** 2)'], {}), '(nume ** 2)\n', (2316, 2327), True, 'import numpy as np\n'), ((1333, 1349), 'numpy.log1p', 'np.log1p', (['(-r * y)'], {}), '(-r * y)\n', (1341, 1349), True, 'import numpy as np\n')] |
import numpy as np
import unittest
from laika.gps_time import GPSTime
from laika import AstroDog
gps_times_list = [[1950, 415621.0],
[1895, 455457.0],
[1885, 443787.0]]
svIds = ['G01', 'G31', 'R08']
gps_times = [GPSTime(*gps_time_list) for gps_time_list in gps_times_list]
class TestAstroDog(unittest.TestCase):
'''
def test_nav_vs_orbit_now(self):
dog_orbit = AstroDog(pull_orbit=True)
dog_nav = AstroDog(pull_orbit=False)
gps_time = GPSTime.from_datetime(datetime.utcnow()) - SECS_IN_DAY*2
for svId in svIds:
sat_info_nav = dog_nav.get_sat_info(svId, gps_time)
sat_info_orbit = dog_orbit.get_sat_info(svId, gps_time)
np.testing.assert_allclose(sat_info_nav[0], sat_info_orbit[0], rtol=0, atol=5)
np.testing.assert_allclose(sat_info_nav[1], sat_info_orbit[1], rtol=0, atol=.1)
np.testing.assert_allclose(sat_info_nav[2], sat_info_orbit[2], rtol=0, atol=1e-7)
np.testing.assert_allclose(sat_info_nav[3], sat_info_orbit[3], rtol=0, atol=1e-11)
'''
def test_nav_vs_orbit__old(self):
dog_orbit = AstroDog(pull_orbit=True)
dog_nav = AstroDog(pull_orbit=False)
for gps_time in gps_times:
for svId in svIds:
sat_info_nav = dog_nav.get_sat_info(svId, gps_time)
sat_info_orbit = dog_orbit.get_sat_info(svId, gps_time)
np.testing.assert_allclose(sat_info_nav[0], sat_info_orbit[0], rtol=0, atol=5)
np.testing.assert_allclose(sat_info_nav[1], sat_info_orbit[1], rtol=0, atol=.1)
np.testing.assert_allclose(sat_info_nav[2], sat_info_orbit[2], rtol=0, atol=1e-7)
np.testing.assert_allclose(sat_info_nav[3], sat_info_orbit[3], rtol=0, atol=1e-11)
if __name__ == "__main__":
unittest.main()
| [
"unittest.main",
"numpy.testing.assert_allclose",
"laika.AstroDog",
"laika.gps_time.GPSTime"
] | [((223, 246), 'laika.gps_time.GPSTime', 'GPSTime', (['*gps_time_list'], {}), '(*gps_time_list)\n', (230, 246), False, 'from laika.gps_time import GPSTime\n'), ((1704, 1719), 'unittest.main', 'unittest.main', ([], {}), '()\n', (1717, 1719), False, 'import unittest\n'), ((1070, 1095), 'laika.AstroDog', 'AstroDog', ([], {'pull_orbit': '(True)'}), '(pull_orbit=True)\n', (1078, 1095), False, 'from laika import AstroDog\n'), ((1110, 1136), 'laika.AstroDog', 'AstroDog', ([], {'pull_orbit': '(False)'}), '(pull_orbit=False)\n', (1118, 1136), False, 'from laika import AstroDog\n'), ((1325, 1403), 'numpy.testing.assert_allclose', 'np.testing.assert_allclose', (['sat_info_nav[0]', 'sat_info_orbit[0]'], {'rtol': '(0)', 'atol': '(5)'}), '(sat_info_nav[0], sat_info_orbit[0], rtol=0, atol=5)\n', (1351, 1403), True, 'import numpy as np\n'), ((1412, 1497), 'numpy.testing.assert_allclose', 'np.testing.assert_allclose', (['sat_info_nav[1]', 'sat_info_orbit[1]'], {'rtol': '(0)', 'atol': '(0.1)'}), '(sat_info_nav[1], sat_info_orbit[1], rtol=0, atol=0.1\n )\n', (1438, 1497), True, 'import numpy as np\n'), ((1500, 1587), 'numpy.testing.assert_allclose', 'np.testing.assert_allclose', (['sat_info_nav[2]', 'sat_info_orbit[2]'], {'rtol': '(0)', 'atol': '(1e-07)'}), '(sat_info_nav[2], sat_info_orbit[2], rtol=0, atol\n =1e-07)\n', (1526, 1587), True, 'import numpy as np\n'), ((1590, 1677), 'numpy.testing.assert_allclose', 'np.testing.assert_allclose', (['sat_info_nav[3]', 'sat_info_orbit[3]'], {'rtol': '(0)', 'atol': '(1e-11)'}), '(sat_info_nav[3], sat_info_orbit[3], rtol=0, atol\n =1e-11)\n', (1616, 1677), True, 'import numpy as np\n')] |
import numpy as np
a = np.array([1, 2, 3, 4, 5])
np.sum(a ** 2)
| [
"numpy.array",
"numpy.sum"
] | [((23, 48), 'numpy.array', 'np.array', (['[1, 2, 3, 4, 5]'], {}), '([1, 2, 3, 4, 5])\n', (31, 48), True, 'import numpy as np\n'), ((49, 63), 'numpy.sum', 'np.sum', (['(a ** 2)'], {}), '(a ** 2)\n', (55, 63), True, 'import numpy as np\n')] |
# -*- coding: utf-8 -*-
"""
Created on Thu Dec 5 12:43:51 2019
@author: Blackr
"""
"""Cyclic Voltammetry (CV) technique class.
The CV technique returns data on fields (in order):
* time (float)
* Ec (float)
* I (float)
* Ewe (float)
* cycle (int)
"""
''' E_we
^
| E_1
| /\
| / \
| / \ E_f
| E_i/ \ /
| \ /
| \/
| E_2
+----------------------> t
Args:
vs_initial (list): List (or tuple) of 5 booleans indicating
whether the current step is vs. the initial one
voltage_step (list): List (or tuple) of 5 floats (Ei, E1, E2, Ei,
Ef) indicating the voltage steps (V)
scan_rate (list): List (or tuple) of 5 floats indicating the scan
rates (mV/s)
record_every_dE (float): Record every dE (V)
average_over_dE (bool): Whether averaging should be performed over
dE
N_cycles (int): The number of cycles
begin_measuring_I (float): Begin step accumulation, 1 is 100%
end_measuring_I (float): Begin step accumulation, 1 is 100%
I_Range (str): A string describing the I range, see the
:data:`I_RANGES` module variable for possible values
E_range (str): A string describing the E range to use, see the
:data:`E_RANGES` module variable for possible values
Bandwidth (str): A string describing the bandwidth setting, see the
:data:`BANDWIDTHS` module variable for possible values'''
"""CV example"""
'''A program to run a typical CV experiment and export the data to a .csv file'''
'''Currently have 32-bit vs. 64-bit interpreter problems with pandas library, so dump to .csv and use other to put into pandas database'''
import time
import numpy
from bio_logic import SP150, CV
def run_cv():
"""Test the CV technique"""
ip_address = 'USB0' # REPLACE THIS WITH A VALID IP
# Instantiate the instrument and connect to it
sp150 = SP150(ip_address, 'C:\\EC-Lab Development Package\\EC-Lab Development Package\\EClib.dll')
sp150.connect()
sp150.load_firmware([1])
# Instantiate the technique. Make sure to give values for all the
# arguments where the default values does not fit your purpose. The
# default values can be viewed in the API documentation for the
# technique.
cv = CV(vs_initial=(False,) * 5,
voltage_step=(2, 0.5, -0.7, 0.0, 0.0),
scan_rate=(10.0,) * 5,
record_every_dE=0.01,
N_cycles=3)
# Load the technique onto channel 0 of the potentiostat and start it
sp150.load_technique(0, cv)
sp150.start_channel(0)
Time = numpy.array([])
Ewe = numpy.array([])
Ec = numpy.array([])
I = numpy.array([])
cycle = numpy.array([])
while True:
# Get the currently available data on channel 0 (only what has
# been gathered since last get_data)
data_out = sp150.get_data(0)
# If there is none, assume the technique has finished
if data_out is None:
break
# The data is available in lists as attributes on the data
# object. The available data fields are listed in the API
# documentation for the technique.
# print("Time:", data_out.time)
# print("Ewe:", data_out.Ewe)
# If numpy is installed, the data can also be retrieved as
# numpy arrays
# printing the values to follow for testing
print('Time:', data_out.time_numpy)
print('Ewe:', data_out.Ewe_numpy)
print('Ec', data_out.Ec_numpy)
print('I', data_out.I_numpy)
print('cycle', data_out.cycle_numpy)
# Updating the variables with the appended data per data call
Ewe = numpy.append(Ewe, data_out.Ewe_numpy_numpy)
Time = numpy.append(Time, data_out.time_numpy)
Ec = numpy.append(Ec, data_out.Ec_numpy)
I = numpy.append(I, data_out.I_numpy)
cycle = numpy.append(cycle, data_out.cycle_numpy)
# Sleep
# dataframe of each variable
df = (Time, Ewe, Ec, I, cycle)
#Due to compatibility issues (in my head, this can be fixed), writing data to a .csv for importing into pandas
# Note the order of header and the df as indicated
numpy.savetxt("testCV.csv", numpy.transpose(df), delimiter=",", header = 'Time,Ewe,Ec,I,cycle', comments = '')
sp150.stop_channel(0)
sp150.disconnect()
if __name__ == '__main__':
run_cv()
| [
"bio_logic.CV",
"numpy.transpose",
"numpy.append",
"numpy.array",
"bio_logic.SP150"
] | [((2207, 2301), 'bio_logic.SP150', 'SP150', (['ip_address', '"""C:\\\\EC-Lab Development Package\\\\EC-Lab Development Package\\\\EClib.dll"""'], {}), "(ip_address,\n 'C:\\\\EC-Lab Development Package\\\\EC-Lab Development Package\\\\EClib.dll')\n", (2212, 2301), False, 'from bio_logic import SP150, CV\n'), ((2590, 2717), 'bio_logic.CV', 'CV', ([], {'vs_initial': '((False,) * 5)', 'voltage_step': '(2, 0.5, -0.7, 0.0, 0.0)', 'scan_rate': '((10.0,) * 5)', 'record_every_dE': '(0.01)', 'N_cycles': '(3)'}), '(vs_initial=(False,) * 5, voltage_step=(2, 0.5, -0.7, 0.0, 0.0),\n scan_rate=(10.0,) * 5, record_every_dE=0.01, N_cycles=3)\n', (2592, 2717), False, 'from bio_logic import SP150, CV\n'), ((2921, 2936), 'numpy.array', 'numpy.array', (['[]'], {}), '([])\n', (2932, 2936), False, 'import numpy\n'), ((2948, 2963), 'numpy.array', 'numpy.array', (['[]'], {}), '([])\n', (2959, 2963), False, 'import numpy\n'), ((2974, 2989), 'numpy.array', 'numpy.array', (['[]'], {}), '([])\n', (2985, 2989), False, 'import numpy\n'), ((2999, 3014), 'numpy.array', 'numpy.array', (['[]'], {}), '([])\n', (3010, 3014), False, 'import numpy\n'), ((3028, 3043), 'numpy.array', 'numpy.array', (['[]'], {}), '([])\n', (3039, 3043), False, 'import numpy\n'), ((4037, 4080), 'numpy.append', 'numpy.append', (['Ewe', 'data_out.Ewe_numpy_numpy'], {}), '(Ewe, data_out.Ewe_numpy_numpy)\n', (4049, 4080), False, 'import numpy\n'), ((4097, 4136), 'numpy.append', 'numpy.append', (['Time', 'data_out.time_numpy'], {}), '(Time, data_out.time_numpy)\n', (4109, 4136), False, 'import numpy\n'), ((4151, 4186), 'numpy.append', 'numpy.append', (['Ec', 'data_out.Ec_numpy'], {}), '(Ec, data_out.Ec_numpy)\n', (4163, 4186), False, 'import numpy\n'), ((4200, 4233), 'numpy.append', 'numpy.append', (['I', 'data_out.I_numpy'], {}), '(I, data_out.I_numpy)\n', (4212, 4233), False, 'import numpy\n'), ((4251, 4292), 'numpy.append', 'numpy.append', (['cycle', 'data_out.cycle_numpy'], {}), '(cycle, data_out.cycle_numpy)\n', (4263, 4292), False, 'import numpy\n'), ((4587, 4606), 'numpy.transpose', 'numpy.transpose', (['df'], {}), '(df)\n', (4602, 4606), False, 'import numpy\n')] |
"""
Matplotlib volumetric benchmarking plotting routines.
"""
#***************************************************************************************************
# Copyright 2015, 2019 National Technology & Engineering Solutions of Sandia, LLC (NTESS).
# Under the terms of Contract DE-NA0003525 with NTESS, the U.S. Government retains certain rights
# in this software.
# Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
# in compliance with the License. You may obtain a copy of the License at
# http://www.apache.org/licenses/LICENSE-2.0 or in the LICENSE file in the root pyGSTi directory.
#***************************************************************************************************
import numpy as _np
try:
import matplotlib.pyplot as _plt
from matplotlib.colors import ListedColormap as _ListedColormap
from matplotlib import cm as _cm
import seaborn as _sns
_sns.set_style('white')
_sns.set_style('ticks')
# Utility color maps.
blues = _sns.color_palette(_sns.color_palette("Blues", 200)).as_hex()
blues[0] = '#ffffff'
blues = _ListedColormap(blues)
reds = _sns.color_palette(_sns.color_palette("Reds", 200)).as_hex()
reds[0] = '#ffffff'
reds = _ListedColormap(reds)
greens = _sns.color_palette(_sns.color_palette("Greens", 200)).as_hex()
greens[0] = '#ffffff'
greens = _ListedColormap(greens)
binary_blue = _sns.color_palette(_sns.color_palette("Blues", 200)).as_hex()
binary_blue[0] = '#ffffff'
binary_blue = _ListedColormap([binary_blue[0], binary_blue[50]])
spectral = _cm.get_cmap('Spectral')
# The default color map.
my_cmap = blues
except ImportError:
_plt = None
_sns = None
my_cmap = None
def empty_volumetric_plot(figsize=None, y_values=None, x_values=None, title=None, xlabel='Depth', ylabel='Width'):
"""
Creates an empty volumetric plot with just the axes set.
Parameters
----------
figsize : tuple or None, optional
The figure size.
y_values : list or None, optional
The y-axis values, typically corresponding to circuit widths.
x_values : list or None, optional
The x-axis values, typically corresponding to circuit depths.
title : string or None, optional
Plot title
xlabel : string, optional
x-axis label
ylabel : string, optional
y-axis label.
Return
------
fig, ax : matplolib fig and ax.
"""
if _plt is None or _sns is None:
raise ValueError(("While not a core requirement of pyGSTi, Matplotlib and Seaborn are "
"required to generate VB plots. It looks like you "
"don't have them installed on your system (it failed to import)."))
fig, ax = _plt.subplots(figsize=figsize)
ax.set_aspect('equal')
_plt.xlabel(xlabel, fontsize=20)
_plt.ylabel(ylabel, fontsize=20)
_plt.title(title, fontsize=24, y=1.02)
_plt.xlim(-1, len(x_values))
_plt.ylim(-1, len(y_values))
depth_labels = [str(d)[0:len(str(d)) - ((len(str(d)) - 1) // 3) * 3]
+ ['', 'k', 'M', 'G'][(len(str(d)) - 1) // 3] for d in x_values]
_plt.xticks(range(len(x_values)), depth_labels, rotation=-60, fontsize=14)
_plt.yticks(range(len(y_values)), y_values, fontsize=14)
_sns.despine()
return fig, ax
def _get_xy(data, y_values=None, x_values=None):
# Helper function for setting the x and y axes of VB plots.
if x_values is None:
x_values = list(set([shape[0] for shape in data.keys()]))
x_values.sort()
if y_values is None:
y_values = list(set([shape[1] for shape in data.keys()]))
y_values.sort()
return y_values, x_values
def volumetric_plot(data, y_values=None, x_values=None, title=None, fig=None, ax=None,
cmap=my_cmap, color=None, flagQV=False, qv_threshold=None,
figsize=(10, 10), scale=1., centerscale=1., linescale=1.,
pass_threshold=0, show_threshold=0):
"""
Creates a volumetric benchmarking plot.
"""
y_values, x_values = _get_xy(data, y_values, x_values)
if fig is None:
fig, ax = empty_volumetric_plot(figsize=figsize, y_values=y_values, x_values=x_values, title=title)
if qv_threshold is None:
qv_threshold = pass_threshold
if color is not None:
cmap = None
point_color = color
for indw, w in enumerate(y_values):
for indd, d in enumerate(x_values):
edgecolor = 'k'
linewidth = 1 * linescale
datapoint = data.get((d, w), None)
if (datapoint is not None) and (not _np.isnan(datapoint)):
if w == d and flagQV:
if datapoint > qv_threshold:
edgecolor = 'r'
linewidth = 5 * scale * linescale
if datapoint >= show_threshold:
if datapoint < pass_threshold:
datapoint = 0
if color is None:
point_color = [datapoint]
ax.scatter([indd], [indw], marker="s", s=280 * scale - 30 * linewidth, c=point_color,
cmap=cmap, vmin=0, vmax=1, edgecolor=edgecolor, linewidth=linewidth)
return fig, ax
def volumetric_boundary_plot(data, y_values=None, x_values=None, boundary=None, threshold=.5,
missing_data_action='continue', monotonic=True, color='k', linewidth=4,
linestyle='-', dashing=None, fig=None, ax=None, figsize=None, title=None,
label=None):
"""
Creates a volumetric benchmarking boundary plot, that displays boundary at which the given data
drops below the specified threshold
"""
y_values, x_values = _get_xy(data, y_values, x_values)
if fig is None:
fig, ax = empty_volumetric_plot(figsize=figsize, y_values=y_values, x_values=x_values, title=title)
if boundary is not None:
boundaries = _np.array([-1 if boundary[d] == 0 else y_values.index(boundary[d]) for d in x_values])
# x-values for a jagged line that outlines the boxes (one pair for each box)
xvals = [y for x in range(len(x_values)) for y in [x - .5, x + .5]]
# y-values for a jagged line that outlines the boxes (one pair for each box)
yvals = [y + .5 for boundary in boundaries for y in [boundary, boundary]]
else:
# For each depth, find the widest circuit that achieves the threshold performance (return -1 if none)
if missing_data_action == 'none':
boundaries = _np.array([_np.max([-1] + [y_values.index(w) for w in y_values if (d, w) in data.keys()
and data[d, w] >= threshold]) for d in x_values])
# x-values for a jagged line that outlines the boxes (one pair for each box)
xvals = [y for x in range(len(x_values)) for y in [x - .5, x + .5]]
# y-values for a jagged line that outlines the boxes (one pair for each box)
yvals = [y + .5 for boundary in boundaries for y in [boundary, boundary]]
elif missing_data_action == 'continue' or missing_data_action == 'hedge':
boundaries = []
d = x_values[0]
boundary_at_d = _np.max([-1] + [y_values.index(w) for w in y_values if (d, w) in data.keys()
and data[d, w] >= threshold])
boundaries.append(boundary_at_d)
previous_boundary = boundary_at_d
hedged_x_values = []
for i, d in enumerate(x_values[1:]):
max_width_at_depth = _np.max([-1] + [w for w in y_values if (d, w) in data.keys()])
if max_width_at_depth < previous_boundary:
boundary_at_d = previous_boundary
hedged_x_values.append(d)
else:
boundary_at_d = _np.max([-1] + [y_values.index(w) for w in y_values if (d, w) in data.keys()
and data[d, w] >= threshold])
boundaries.append(boundary_at_d)
previous_boundary = boundary_at_d
if missing_data_action == 'continue':
# x-values for a jagged line that outlines the boxes (one pair for each box)
xvals = [y for x in range(len(x_values)) for y in [x - .5, x + .5]]
# y-values for a jagged line that outlines the boxes (one pair for each box)
yvals = [y + .5 for boundary in boundaries for y in [boundary, boundary]]
elif missing_data_action == 'hedge':
# x-values for a jagged line that outlines the boxes (one pair for each box)
xvals = []
yvals = []
last_xval = -0.5
for x, boundary in zip(range(len(x_values)), boundaries):
d = x_values[x]
if d in hedged_x_values:
# Only hedge when there's actually some data at larger x_values.
if not all([d in hedged_x_values for d in x_values[x:]]):
xvals += [last_xval, x]
yvals += [boundary + .5, boundary + .5]
else:
xvals += [last_xval, x + .5]
yvals += [boundary + .5, boundary + .5]
last_xval = xvals[-1]
if monotonic:
monotonic_yvals = [yvals[0]]
for y in yvals[1:]:
if y > monotonic_yvals[-1]:
monotonic_yvals.append(monotonic_yvals[-1])
else:
monotonic_yvals.append(y)
yvals = monotonic_yvals
line, = ax.plot(xvals, yvals, color, linewidth=linewidth, label=label, linestyle=linestyle)
if dashing is not None:
line.set_dashes(dashing)
return fig, ax
def capability_region_plot(vbdataframe, metric='polarization', threshold=1 / _np.e, significance=0.05, figsize=(10, 10),
scale=1., title=None, colors=None):
"""
Creates a capability regions plot from a VBDataFrame. Default options creates plots like those shown
in Fig. 3 of "Measuring the Capabilities of Quantum Computers" arXiv:2008.11294.
"""
x_values = vbdataframe.x_values
y_values = vbdataframe.y_values
fig, ax = empty_volumetric_plot(figsize=figsize, y_values=y_values, x_values=x_values, title=title)
creg = vbdataframe.capability_regions(metric=metric, threshold=threshold, significance=significance, monotonic=True)
# Split the data up into dicts for the three different regions: 'success', 'indeterminate' and 'fail'.
creg_split = {}
creg_split['success'] = {(w, d): 1 for (w, d), val in creg.items() if val == 2}
creg_split['indeterminate'] = {(w, d): 1 for (w, d), val in creg.items() if val == 1}
creg_split['fail'] = {(w, d): 1 for (w, d), val in creg.items() if val == 0}
if colors is None:
colors = {'success': [(0.2, 0.6274509803921569, 0.17254901960784313)],
'indeterminate': [(0.9921568627450981, 0.7490196078431373, 0.43529411764705883)],
'fail': 'w'}
for region in ('success', 'indeterminate', 'fail'):
fig, ax = volumetric_plot(creg_split[region], y_values=y_values, x_values=x_values, scale=scale, fig=fig, ax=ax,
color=colors[region])
return fig, ax
def volumetric_distribution_plot(vbdataframe, metric='polarization', threshold=1 / _np.e, hypothesis_test='standard',
significance=0.05, figsize=(10, 10), scale={'min': 1.95, 'mean': 1, 'max': 0.13},
title=None, cmap=my_cmap):
"""
Creates volumetric benchmarking plots that display the maximum, mean and minimum of a given figure-of-merit (by
default, circuit polarization) as a function of circuit shape. This function can be used to create figures like
those shown in Fig. 1 of "Measuring the Capabilities of Quantum Computers" arXiv:2008.11294.
Parameters
----------
vbdataframe : VBDataFrame
A VBDataFrame object containing the data to be plotted in a VB plot.
metric : string, optional
The quantity to plot. Default is 'polarization' as used and defined in arXiv:2008.11294. The plot
will show the maximum, mean, and minimum of this metric at each circuit shape.
threshold : float, optional
The threshold for "success" for the figure-of-merit defined by `metric`. This threshold is used
to compute the three "success" boundaries that are shown in the plot.
hypothesis_test : string, optional
The type of statistical significance adjustment to apply to the boundaries. The options are
- 'standard': this reproduces the method used and described in arXiv:2008.11294 (see the
appendices for details). With this option, there will be a difference between the
boundary for the minimum and maximum polarization only if there is statistically significant
evidence in the data for this.
- 'none': no statistical significance adjustment: all three boundaries show the point at which
relevant statistic (maximum, mean, minimum) drops below the threshold.
significance : float, optional
The statistical significance in the hypothesis tests. Only used in `hypothesis_test` is not 'none'.
figsize : tuple, optional
The figure size
scale : dict, optional
The scale for the three concentric squares, showing the maximum, mean and minimum.
title : sting, optional
The figure title.
cmap : ColorMap, optional
A matplotlib colormap.
Return
------
fig, ax : matplolib fig and ax.
"""
linescale = {'min': 1, 'mean': 0, 'max': 0}
boundary_color = {'min': '#ff0000', 'mean': '#000000', 'max': '#2ecc71'}
boundary_dashing = {'min': [1, 1], 'mean': None, 'max': [0.5, 0.5]}
boundary_linewidth = {'min': 3, 'mean': 6, 'max': 5}
x_values = vbdataframe.x_values
y_values = vbdataframe.y_values
fig, ax = empty_volumetric_plot(figsize=figsize, y_values=y_values, x_values=x_values, title=title)
# Dictionary containing the three types of VB data that are used in this plot.
vb_data = {stat: vbdataframe.vb_data(metric=metric, statistic=stat, no_data_action='discard')
for stat in ('min', 'mean', 'max')}
# Used to find the min and max boundaries if they are adjusted for statistical significance.
capability_regions = vbdataframe.capability_regions(metric=metric, threshold=threshold, significance=significance,
monotonic=True)
if hypothesis_test == 'standard':
adjusted_boundaries = ('max', 'min')
unadjusted_boundaries = ('mean',)
elif hypothesis_test == 'none':
adjusted_boundaries = ()
unadjusted_boundaries = ('max', 'mean', 'min',)
else:
raise ValueError("`hypothesis_test` must be 'standard' or 'none'!")
# Plots the data.
for statistic in ('min', 'mean', 'max'):
fig, ax = volumetric_plot(vb_data[statistic], y_values=y_values, x_values=x_values, fig=fig, ax=ax,
scale=scale[statistic], linescale=linescale[statistic], cmap=cmap)
# Plots the boundaries that have been adjusted for statistical significance.
for statistic in adjusted_boundaries:
if statistic == 'max': effective_threshold = 0.99
elif statistic == 'min': effective_threshold = 1.99
volumetric_boundary_plot(capability_regions, y_values=y_values, x_values=x_values, threshold=effective_threshold,
missing_data_action='hedge', fig=fig, ax=ax, linestyle='-',
color=boundary_color[statistic], linewidth=boundary_linewidth[statistic],
dashing=boundary_dashing[statistic])
# Plots the boundaries that are not adjusted for statistical significance.
for statistic in unadjusted_boundaries:
volumetric_boundary_plot(vb_data[statistic], y_values=y_values, x_values=x_values, threshold=threshold,
monotonic=False, missing_data_action='hedge', fig=fig, ax=ax, linestyle='-',
color=boundary_color[statistic], linewidth=boundary_linewidth[statistic],
dashing=boundary_dashing[statistic])
return fig, ax
| [
"matplotlib.pyplot.title",
"seaborn.set_style",
"matplotlib.cm.get_cmap",
"seaborn.despine",
"numpy.isnan",
"seaborn.color_palette",
"matplotlib.pyplot.ylabel",
"matplotlib.pyplot.xlabel",
"matplotlib.pyplot.subplots",
"matplotlib.colors.ListedColormap"
] | [((946, 969), 'seaborn.set_style', '_sns.set_style', (['"""white"""'], {}), "('white')\n", (960, 969), True, 'import seaborn as _sns\n'), ((974, 997), 'seaborn.set_style', '_sns.set_style', (['"""ticks"""'], {}), "('ticks')\n", (988, 997), True, 'import seaborn as _sns\n'), ((1136, 1158), 'matplotlib.colors.ListedColormap', '_ListedColormap', (['blues'], {}), '(blues)\n', (1151, 1158), True, 'from matplotlib.colors import ListedColormap as _ListedColormap\n'), ((1267, 1288), 'matplotlib.colors.ListedColormap', '_ListedColormap', (['reds'], {}), '(reds)\n', (1282, 1288), True, 'from matplotlib.colors import ListedColormap as _ListedColormap\n'), ((1405, 1428), 'matplotlib.colors.ListedColormap', '_ListedColormap', (['greens'], {}), '(greens)\n', (1420, 1428), True, 'from matplotlib.colors import ListedColormap as _ListedColormap\n'), ((1559, 1609), 'matplotlib.colors.ListedColormap', '_ListedColormap', (['[binary_blue[0], binary_blue[50]]'], {}), '([binary_blue[0], binary_blue[50]])\n', (1574, 1609), True, 'from matplotlib.colors import ListedColormap as _ListedColormap\n'), ((1626, 1650), 'matplotlib.cm.get_cmap', '_cm.get_cmap', (['"""Spectral"""'], {}), "('Spectral')\n", (1638, 1650), True, 'from matplotlib import cm as _cm\n'), ((2821, 2851), 'matplotlib.pyplot.subplots', '_plt.subplots', ([], {'figsize': 'figsize'}), '(figsize=figsize)\n', (2834, 2851), True, 'import matplotlib.pyplot as _plt\n'), ((2883, 2915), 'matplotlib.pyplot.xlabel', '_plt.xlabel', (['xlabel'], {'fontsize': '(20)'}), '(xlabel, fontsize=20)\n', (2894, 2915), True, 'import matplotlib.pyplot as _plt\n'), ((2920, 2952), 'matplotlib.pyplot.ylabel', '_plt.ylabel', (['ylabel'], {'fontsize': '(20)'}), '(ylabel, fontsize=20)\n', (2931, 2952), True, 'import matplotlib.pyplot as _plt\n'), ((2957, 2995), 'matplotlib.pyplot.title', '_plt.title', (['title'], {'fontsize': '(24)', 'y': '(1.02)'}), '(title, fontsize=24, y=1.02)\n', (2967, 2995), True, 'import matplotlib.pyplot as _plt\n'), ((3365, 3379), 'seaborn.despine', '_sns.despine', ([], {}), '()\n', (3377, 3379), True, 'import seaborn as _sns\n'), ((1056, 1088), 'seaborn.color_palette', '_sns.color_palette', (['"""Blues"""', '(200)'], {}), "('Blues', 200)\n", (1074, 1088), True, 'import seaborn as _sns\n'), ((1190, 1221), 'seaborn.color_palette', '_sns.color_palette', (['"""Reds"""', '(200)'], {}), "('Reds', 200)\n", (1208, 1221), True, 'import seaborn as _sns\n'), ((1322, 1355), 'seaborn.color_palette', '_sns.color_palette', (['"""Greens"""', '(200)'], {}), "('Greens', 200)\n", (1340, 1355), True, 'import seaborn as _sns\n'), ((1467, 1499), 'seaborn.color_palette', '_sns.color_palette', (['"""Blues"""', '(200)'], {}), "('Blues', 200)\n", (1485, 1499), True, 'import seaborn as _sns\n'), ((4750, 4770), 'numpy.isnan', '_np.isnan', (['datapoint'], {}), '(datapoint)\n', (4759, 4770), True, 'import numpy as _np\n')] |
import os
import numpy as np
opj = os.path.join
from astropy.io import fits
data_path = '../../data/cosmology'
def downsample(parameter_file, root_dir, resize=64, nsamples=30000, ncosmo=10):
'''
downsample cosmolgy image
'''
print('preprocessing...')
img_size = 256
params_ = np.loadtxt(parameter_file)[:ncosmo]
image = []
params = []
for idx in range(nsamples):
img_name = opj(root_dir, 'model%03d/WLconv_z1.00_%04dr.fits' % (idx % len(params_), idx // len(params_)))
start1 = np.random.randint(0, img_size - resize - 1, 1)[0]
start2 = np.random.randint(0, img_size - resize - 1, 1)[0]
end1 = start1 + resize
end2 = start2 + resize
hdu_list = fits.open(img_name, memmap=False)
img_data = hdu_list[0].data
image.append(img_data[start1:end1, start2:end2])
hdu_list.close()
params.append(params_[idx % len(params_), 1:-1])
print('\r idx: {}/{}'.format(idx, nsamples), end='')
image = np.stack(image, axis=0)
params = np.stack(params, axis=0)
# save
np.savez(opj(data_path, 'cosmo_resize_{}.npz'.format(resize)), imgs=image, params=params)
if __name__ == '__main__':
parameter_file = opj(data_path, 'cosmological_parameters.txt')
root_dir = opj(data_path, 'z1_256')
resize = 64
# save
downsample(parameter_file, root_dir, resize)
# load
dataset_zip = np.load(opj(data_path, 'cosmo_resize_{}.npz'.format(resize)))
imgs = dataset_zip['imgs']
params = dataset_zip['params']
print(imgs.shape, params.shape)
| [
"numpy.stack",
"numpy.random.randint",
"astropy.io.fits.open",
"numpy.loadtxt"
] | [((1017, 1040), 'numpy.stack', 'np.stack', (['image'], {'axis': '(0)'}), '(image, axis=0)\n', (1025, 1040), True, 'import numpy as np\n'), ((1054, 1078), 'numpy.stack', 'np.stack', (['params'], {'axis': '(0)'}), '(params, axis=0)\n', (1062, 1078), True, 'import numpy as np\n'), ((305, 331), 'numpy.loadtxt', 'np.loadtxt', (['parameter_file'], {}), '(parameter_file)\n', (315, 331), True, 'import numpy as np\n'), ((734, 767), 'astropy.io.fits.open', 'fits.open', (['img_name'], {'memmap': '(False)'}), '(img_name, memmap=False)\n', (743, 767), False, 'from astropy.io import fits\n'), ((536, 582), 'numpy.random.randint', 'np.random.randint', (['(0)', '(img_size - resize - 1)', '(1)'], {}), '(0, img_size - resize - 1, 1)\n', (553, 582), True, 'import numpy as np\n'), ((603, 649), 'numpy.random.randint', 'np.random.randint', (['(0)', '(img_size - resize - 1)', '(1)'], {}), '(0, img_size - resize - 1, 1)\n', (620, 649), True, 'import numpy as np\n')] |
#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""Core IO, DSP and utility functions."""
import os
import six
import audioread
import numpy as np
import scipy.signal
import scipy.fftpack as fft
import resampy
from .time_frequency import frames_to_samples, time_to_samples
from .. import cache
from .. import util
from ..util.exceptions import ParameterError
__all__ = ['load', 'to_mono', 'resample', 'get_duration',
'autocorrelate', 'zero_crossings', 'clicks', 'tone', 'chirp']
# Resampling bandwidths as percentage of Nyquist
BW_BEST = resampy.filters.get_filter('kaiser_best')[2]
BW_FASTEST = resampy.filters.get_filter('kaiser_fast')[2]
# -- CORE ROUTINES --#
# Load should never be cached, since we cannot verify that the contents of
# 'path' are unchanged across calls.
def load(path, sr=22050, mono=True, offset=0.0, duration=None,
dtype=np.float32, res_type='kaiser_best'):
"""Load an audio file as a floating point time series.
Audio will be automatically resampled to the given rate
(default `sr=22050`).
To preserve the native sampling rate of the file, use `sr=None`.
Parameters
----------
path : string
path to the input file.
Any format supported by `audioread` will work.
sr : number > 0 [scalar]
target sampling rate
'None' uses the native sampling rate
mono : bool
convert signal to mono
offset : float
start reading after this time (in seconds)
duration : float
only load up to this much audio (in seconds)
dtype : numeric type
data type of `y`
res_type : str
resample type (see note)
.. note::
By default, this uses `resampy`'s high-quality mode ('kaiser_best').
To use a faster method, set `res_type='kaiser_fast'`.
To use `scipy.signal.resample`, set `res_type='scipy'`.
Returns
-------
y : np.ndarray [shape=(n,) or (2, n)]
audio time series
sr : number > 0 [scalar]
sampling rate of `y`
Examples
--------
>>> # Load a wav file
>>> filename = librosa.util.example_audio_file()
>>> y, sr = librosa.load(filename)
>>> y
array([ -4.756e-06, -6.020e-06, ..., -1.040e-06, 0.000e+00], dtype=float32)
>>> sr
22050
>>> # Load a wav file and resample to 11 KHz
>>> filename = librosa.util.example_audio_file()
>>> y, sr = librosa.load(filename, sr=11025)
>>> y
array([ -2.077e-06, -2.928e-06, ..., -4.395e-06, 0.000e+00], dtype=float32)
>>> sr
11025
>>> # Load 5 seconds of a wav file, starting 15 seconds in
>>> filename = librosa.util.example_audio_file()
>>> y, sr = librosa.load(filename, offset=15.0, duration=5.0)
>>> y
array([ 0.069, 0.1 , ..., -0.101, 0. ], dtype=float32)
>>> sr
22050
"""
y = []
with audioread.audio_open(os.path.realpath(path)) as input_file:
sr_native = input_file.samplerate
n_channels = input_file.channels
s_start = int(np.round(sr_native * offset)) * n_channels
if duration is None:
s_end = np.inf
else:
s_end = s_start + (int(np.round(sr_native * duration))
* n_channels)
n = 0
for frame in input_file:
frame = util.buf_to_float(frame, dtype=dtype)
n_prev = n
n = n + len(frame)
if n < s_start:
# offset is after the current frame
# keep reading
continue
if s_end < n_prev:
# we're off the end. stop reading
break
if s_end < n:
# the end is in this frame. crop.
frame = frame[:s_end - n_prev]
if n_prev <= s_start <= n:
# beginning is in this frame
frame = frame[(s_start - n_prev):]
# tack on the current frame
y.append(frame)
if y:
y = np.concatenate(y)
if n_channels > 1:
y = y.reshape((-1, n_channels)).T
if mono:
y = to_mono(y)
if sr is not None:
y = resample(y, sr_native, sr, res_type=res_type)
else:
sr = sr_native
# Final cleanup for dtype and contiguity
y = np.ascontiguousarray(y, dtype=dtype)
return (y, sr)
@cache(level=20)
def to_mono(y):
'''Force an audio signal down to mono.
Parameters
----------
y : np.ndarray [shape=(2,n) or shape=(n,)]
audio time series, either stereo or mono
Returns
-------
y_mono : np.ndarray [shape=(n,)]
`y` as a monophonic time-series
Notes
-----
This function caches at level 20.
Examples
--------
>>> y, sr = librosa.load(librosa.util.example_audio_file(), mono=False)
>>> y.shape
(2, 1355168)
>>> y_mono = librosa.to_mono(y)
>>> y_mono.shape
(1355168,)
'''
# Validate the buffer. Stereo is ok here.
util.valid_audio(y, mono=False)
if y.ndim > 1:
y = np.mean(y, axis=0)
return y
@cache(level=20)
def resample(y, orig_sr, target_sr, res_type='kaiser_best', fix=True, scale=False, **kwargs):
"""Resample a time series from orig_sr to target_sr
Parameters
----------
y : np.ndarray [shape=(n,) or shape=(2, n)]
audio time series. Can be mono or stereo.
orig_sr : number > 0 [scalar]
original sampling rate of `y`
target_sr : number > 0 [scalar]
target sampling rate
res_type : str
resample type (see note)
.. note::
By default, this uses `resampy`'s high-quality mode ('kaiser_best').
To use a faster method, set `res_type='kaiser_fast'`.
To use `scipy.signal.resample`, set `res_type='scipy'`.
fix : bool
adjust the length of the resampled signal to be of size exactly
`ceil(target_sr * len(y) / orig_sr)`
scale : bool
Scale the resampled signal so that `y` and `y_hat` have approximately
equal total energy.
kwargs : additional keyword arguments
If `fix==True`, additional keyword arguments to pass to
`librosa.util.fix_length`.
Returns
-------
y_hat : np.ndarray [shape=(n * target_sr / orig_sr,)]
`y` resampled from `orig_sr` to `target_sr`
See Also
--------
librosa.util.fix_length
scipy.signal.resample
resampy.resample
Notes
-----
This function caches at level 20.
Examples
--------
Downsample from 22 KHz to 8 KHz
>>> y, sr = librosa.load(librosa.util.example_audio_file(), sr=22050)
>>> y_8k = librosa.resample(y, sr, 8000)
>>> y.shape, y_8k.shape
((1355168,), (491671,))
"""
# First, validate the audio buffer
util.valid_audio(y, mono=False)
if orig_sr == target_sr:
return y
ratio = float(target_sr) / orig_sr
n_samples = int(np.ceil(y.shape[-1] * ratio))
if res_type == 'scipy':
y_hat = scipy.signal.resample(y, n_samples, axis=-1)
else:
y_hat = resampy.resample(y, orig_sr, target_sr, filter=res_type, axis=-1)
if fix:
y_hat = util.fix_length(y_hat, n_samples, **kwargs)
if scale:
y_hat /= np.sqrt(ratio)
return np.ascontiguousarray(y_hat, dtype=y.dtype)
def get_duration(y=None, sr=22050, S=None, n_fft=2048, hop_length=512,
center=True, filename=None):
"""Compute the duration (in seconds) of an audio time series,
feature matrix, or filename.
Examples
--------
>>> # Load the example audio file
>>> y, sr = librosa.load(librosa.util.example_audio_file())
>>> librosa.get_duration(y=y, sr=sr)
61.45886621315193
>>> # Or directly from an audio file
>>> librosa.get_duration(filename=librosa.util.example_audio_file())
61.4
>>> # Or compute duration from an STFT matrix
>>> y, sr = librosa.load(librosa.util.example_audio_file())
>>> S = librosa.stft(y)
>>> librosa.get_duration(S=S, sr=sr)
61.44
>>> # Or a non-centered STFT matrix
>>> S_left = librosa.stft(y, center=False)
>>> librosa.get_duration(S=S_left, sr=sr)
61.3471201814059
Parameters
----------
y : np.ndarray [shape=(n,), (2, n)] or None
audio time series
sr : number > 0 [scalar]
audio sampling rate of `y`
S : np.ndarray [shape=(d, t)] or None
STFT matrix, or any STFT-derived matrix (e.g., chromagram
or mel spectrogram).
Durations calculated from spectrogram inputs are only accurate
up to the frame resolution. If high precision is required,
it is better to use the audio time series directly.
n_fft : int > 0 [scalar]
FFT window size for `S`
hop_length : int > 0 [ scalar]
number of audio samples between columns of `S`
center : boolean
- If `True`, `S[:, t]` is centered at `y[t * hop_length]`
- If `False`, then `S[:, t]` begins at `y[t * hop_length]`
filename : str
If provided, all other parameters are ignored, and the
duration is calculated directly from the audio file.
Note that this avoids loading the contents into memory,
and is therefore useful for querying the duration of
long files.
Returns
-------
d : float >= 0
Duration (in seconds) of the input time series or spectrogram.
Raises
------
ParameterError
if none of `y`, `S`, or `filename` are provided.
Notes
-----
`get_duration` can be applied to a file (`filename`), a spectrogram (`S`),
or audio buffer (`y, sr`). Only one of these three options should be
provided. If you do provide multiple options (e.g., `filename` and `S`),
then `filename` takes precedence over `S`, and `S` takes precedence over
`(y, sr)`.
"""
if filename is not None:
with audioread.audio_open(filename) as fdesc:
return fdesc.duration
if y is None:
if S is None:
raise ParameterError('At least one of (y, sr), S, or filename must be provided')
n_frames = S.shape[1]
n_samples = n_fft + hop_length * (n_frames - 1)
# If centered, we lose half a window from each end of S
if center:
n_samples = n_samples - 2 * int(n_fft / 2)
else:
# Validate the audio buffer. Stereo is okay here.
util.valid_audio(y, mono=False)
if y.ndim == 1:
n_samples = len(y)
else:
n_samples = y.shape[-1]
return float(n_samples) / sr
@cache(level=20)
def autocorrelate(y, max_size=None, axis=-1):
"""Bounded auto-correlation
Parameters
----------
y : np.ndarray
array to autocorrelate
max_size : int > 0 or None
maximum correlation lag.
If unspecified, defaults to `y.shape[axis]` (unbounded)
axis : int
The axis along which to autocorrelate.
By default, the last axis (-1) is taken.
Returns
-------
z : np.ndarray
truncated autocorrelation `y*y` along the specified axis.
If `max_size` is specified, then `z.shape[axis]` is bounded
to `max_size`.
Notes
-----
This function caches at level 20.
Examples
--------
Compute full autocorrelation of y
>>> y, sr = librosa.load(librosa.util.example_audio_file(), offset=20, duration=10)
>>> librosa.autocorrelate(y)
array([ 3.226e+03, 3.217e+03, ..., 8.277e-04, 3.575e-04], dtype=float32)
Compute onset strength auto-correlation up to 4 seconds
>>> import matplotlib.pyplot as plt
>>> odf = librosa.onset.onset_strength(y=y, sr=sr, hop_length=512)
>>> ac = librosa.autocorrelate(odf, max_size=4* sr / 512)
>>> plt.plot(ac)
>>> plt.title('Auto-correlation')
>>> plt.xlabel('Lag (frames)')
"""
if max_size is None:
max_size = y.shape[axis]
max_size = int(min(max_size, y.shape[axis]))
# Compute the power spectrum along the chosen axis
# Pad out the signal to support full-length auto-correlation.
powspec = np.abs(fft.fft(y, n=2 * y.shape[axis] + 1, axis=axis))**2
# Convert back to time domain
autocorr = fft.ifft(powspec, axis=axis, overwrite_x=True)
# Slice down to max_size
subslice = [slice(None)] * autocorr.ndim
subslice[axis] = slice(max_size)
autocorr = autocorr[tuple(subslice)]
if not np.iscomplexobj(y):
autocorr = autocorr.real
return autocorr
@cache(level=20)
def zero_crossings(y, threshold=1e-10, ref_magnitude=None, pad=True,
zero_pos=True, axis=-1):
'''Find the zero-crossings of a signal `y`: indices `i` such that
`sign(y[i]) != sign(y[j])`.
If `y` is multi-dimensional, then zero-crossings are computed along
the specified `axis`.
Parameters
----------
y : np.ndarray
The input array
threshold : float > 0 or None
If specified, values where `-threshold <= y <= threshold` are
clipped to 0.
ref_magnitude : float > 0 or callable
If numeric, the threshold is scaled relative to `ref_magnitude`.
If callable, the threshold is scaled relative to
`ref_magnitude(np.abs(y))`.
pad : boolean
If `True`, then `y[0]` is considered a valid zero-crossing.
zero_pos : boolean
If `True` then the value 0 is interpreted as having positive sign.
If `False`, then 0, -1, and +1 all have distinct signs.
axis : int
Axis along which to compute zero-crossings.
Returns
-------
zero_crossings : np.ndarray [shape=y.shape, dtype=boolean]
Indicator array of zero-crossings in `y` along the selected axis.
Notes
-----
This function caches at level 20.
Examples
--------
>>> # Generate a time-series
>>> y = np.sin(np.linspace(0, 4 * 2 * np.pi, 20))
>>> y
array([ 0.000e+00, 9.694e-01, 4.759e-01, -7.357e-01,
-8.372e-01, 3.247e-01, 9.966e-01, 1.646e-01,
-9.158e-01, -6.142e-01, 6.142e-01, 9.158e-01,
-1.646e-01, -9.966e-01, -3.247e-01, 8.372e-01,
7.357e-01, -4.759e-01, -9.694e-01, -9.797e-16])
>>> # Compute zero-crossings
>>> z = librosa.zero_crossings(y)
>>> z
array([ True, False, False, True, False, True, False, False,
True, False, True, False, True, False, False, True,
False, True, False, True], dtype=bool)
>>> # Stack y against the zero-crossing indicator
>>> np.vstack([y, z]).T
array([[ 0.000e+00, 1.000e+00],
[ 9.694e-01, 0.000e+00],
[ 4.759e-01, 0.000e+00],
[ -7.357e-01, 1.000e+00],
[ -8.372e-01, 0.000e+00],
[ 3.247e-01, 1.000e+00],
[ 9.966e-01, 0.000e+00],
[ 1.646e-01, 0.000e+00],
[ -9.158e-01, 1.000e+00],
[ -6.142e-01, 0.000e+00],
[ 6.142e-01, 1.000e+00],
[ 9.158e-01, 0.000e+00],
[ -1.646e-01, 1.000e+00],
[ -9.966e-01, 0.000e+00],
[ -3.247e-01, 0.000e+00],
[ 8.372e-01, 1.000e+00],
[ 7.357e-01, 0.000e+00],
[ -4.759e-01, 1.000e+00],
[ -9.694e-01, 0.000e+00],
[ -9.797e-16, 1.000e+00]])
>>> # Find the indices of zero-crossings
>>> np.nonzero(z)
(array([ 0, 3, 5, 8, 10, 12, 15, 17, 19]),)
'''
# Clip within the threshold
if threshold is None:
threshold = 0.0
if six.callable(ref_magnitude):
threshold = threshold * ref_magnitude(np.abs(y))
elif ref_magnitude is not None:
threshold = threshold * ref_magnitude
if threshold > 0:
y = y.copy()
y[np.abs(y) <= threshold] = 0
# Extract the sign bit
if zero_pos:
y_sign = np.signbit(y)
else:
y_sign = np.sign(y)
# Find the change-points by slicing
slice_pre = [slice(None)] * y.ndim
slice_pre[axis] = slice(1, None)
slice_post = [slice(None)] * y.ndim
slice_post[axis] = slice(-1)
# Since we've offset the input by one, pad back onto the front
padding = [(0, 0)] * y.ndim
padding[axis] = (1, 0)
return np.pad((y_sign[tuple(slice_post)] != y_sign[tuple(slice_pre)]),
padding,
mode='constant',
constant_values=pad)
def clicks(times=None, frames=None, sr=22050, hop_length=512,
click_freq=1000.0, click_duration=0.1, click=None, length=None):
"""Returns a signal with the signal `click` placed at each specified time
Parameters
----------
times : np.ndarray or None
times to place clicks, in seconds
frames : np.ndarray or None
frame indices to place clicks
sr : number > 0
desired sampling rate of the output signal
hop_length : int > 0
if positions are specified by `frames`, the number of samples between frames.
click_freq : float > 0
frequency (in Hz) of the default click signal. Default is 1KHz.
click_duration : float > 0
duration (in seconds) of the default click signal. Default is 100ms.
click : np.ndarray or None
optional click signal sample to use instead of the default blip.
length : int > 0
desired number of samples in the output signal
Returns
-------
click_signal : np.ndarray
Synthesized click signal
Raises
------
ParameterError
- If neither `times` nor `frames` are provided.
- If any of `click_freq`, `click_duration`, or `length` are out of range.
Examples
--------
>>> # Sonify detected beat events
>>> y, sr = librosa.load(librosa.util.example_audio_file())
>>> tempo, beats = librosa.beat.beat_track(y=y, sr=sr)
>>> y_beats = librosa.clicks(frames=beats, sr=sr)
>>> # Or generate a signal of the same length as y
>>> y_beats = librosa.clicks(frames=beats, sr=sr, length=len(y))
>>> # Or use timing instead of frame indices
>>> times = librosa.frames_to_time(beats, sr=sr)
>>> y_beat_times = librosa.clicks(times=times, sr=sr)
>>> # Or with a click frequency of 880Hz and a 500ms sample
>>> y_beat_times880 = librosa.clicks(times=times, sr=sr,
... click_freq=880, click_duration=0.5)
Display click waveform next to the spectrogram
>>> import matplotlib.pyplot as plt
>>> plt.figure()
>>> S = librosa.feature.melspectrogram(y=y, sr=sr)
>>> ax = plt.subplot(2,1,2)
>>> librosa.display.specshow(librosa.power_to_db(S, ref=np.max),
... x_axis='time', y_axis='mel')
>>> plt.subplot(2,1,1, sharex=ax)
>>> librosa.display.waveplot(y_beat_times, sr=sr, label='Beat clicks')
>>> plt.legend()
>>> plt.xlim(15, 30)
>>> plt.tight_layout()
"""
# Compute sample positions from time or frames
if times is None:
if frames is None:
raise ParameterError('either "times" or "frames" must be provided')
positions = frames_to_samples(frames, hop_length=hop_length)
else:
# Convert times to positions
positions = time_to_samples(times, sr=sr)
if click is not None:
# Check that we have a well-formed audio buffer
util.valid_audio(click, mono=True)
else:
# Create default click signal
if click_duration <= 0:
raise ParameterError('click_duration must be strictly positive')
if click_freq <= 0:
raise ParameterError('click_freq must be strictly positive')
angular_freq = 2 * np.pi * click_freq / float(sr)
click = np.logspace(0, -10,
num=int(np.round(sr * click_duration)),
base=2.0)
click *= np.sin(angular_freq * np.arange(len(click)))
# Set default length
if length is None:
length = positions.max() + click.shape[0]
else:
if length < 1:
raise ParameterError('length must be a positive integer')
# Filter out any positions past the length boundary
positions = positions[positions < length]
# Pre-allocate click signal
click_signal = np.zeros(length, dtype=np.float32)
# Place clicks
for start in positions:
# Compute the end-point of this click
end = start + click.shape[0]
if end >= length:
click_signal[start:] += click[:length - start]
else:
# Normally, just add a click here
click_signal[start:end] += click
return click_signal
def tone(frequency, sr=22050, length=None, duration=None, phi=None):
"""Returns a pure tone signal. The signal generated is a cosine wave.
Parameters
----------
frequency : float > 0
frequency
sr : number > 0
desired sampling rate of the output signal
length : int > 0
desired number of samples in the output signal. When both `duration` and `length` are defined, `length` would take priority.
duration : float > 0
desired duration in seconds. When both `duration` and `length` are defined, `length` would take priority.
phi : float or None
phase offset, in radians. If unspecified, defaults to `-np.pi * 0.5`.
Returns
-------
tone_signal : np.ndarray [shape=(length,), dtype=float64]
Synthesized pure sine tone signal
Raises
------
ParameterError
- If `frequency` is not provided.
- If neither `length` nor `duration` are provided.
Examples
--------
>>> # Generate a pure sine tone A4
>>> tone = librosa.tone(440, duration=1)
>>> # Or generate the same signal using `length`
>>> tone = librosa.tone(440, sr=22050, length=22050)
Display spectrogram
>>> import matplotlib.pyplot as plt
>>> plt.figure()
>>> S = librosa.feature.melspectrogram(y=tone)
>>> librosa.display.specshow(librosa.power_to_db(S, ref=np.max),
... x_axis='time', y_axis='mel')
"""
if frequency is None:
raise ParameterError('"frequency" must be provided')
# Compute signal length
if length is None:
if duration is None:
raise ParameterError('either "length" or "duration" must be provided')
length = duration * sr
if phi is None:
phi = -np.pi * 0.5
step = 1.0 / sr
return np.cos(2 * np.pi * frequency * (np.arange(step * length, step=step)) + phi)
def chirp(fmin, fmax, sr=22050, length=None, duration=None, linear=False, phi=None):
"""Returns a chirp signal that goes from frequency `fmin` to frequency `fmax`
Parameters
----------
fmin : float > 0
initial frequency
fmax : float > 0
final frequency
sr : number > 0
desired sampling rate of the output signal
length : int > 0
desired number of samples in the output signal.
When both `duration` and `length` are defined, `length` would take priority.
duration : float > 0
desired duration in seconds.
When both `duration` and `length` are defined, `length` would take priority.
linear : boolean
- If `True`, use a linear sweep, i.e., frequency changes linearly with time
- If `False`, use a exponential sweep.
Default is `False`.
phi : float or None
phase offset, in radians.
If unspecified, defaults to `-np.pi * 0.5`.
Returns
-------
chirp_signal : np.ndarray [shape=(length,), dtype=float64]
Synthesized chirp signal
Raises
------
ParameterError
- If either `fmin` or `fmax` are not provided.
- If neither `length` nor `duration` are provided.
See Also
--------
scipy.signal.chirp
Examples
--------
>>> # Generate a exponential chirp from A4 to A5
>>> exponential_chirp = librosa.chirp(440, 880, duration=1)
>>> # Or generate the same signal using `length`
>>> exponential_chirp = librosa.chirp(440, 880, sr=22050, length=22050)
>>> # Or generate a linear chirp instead
>>> linear_chirp = librosa.chirp(440, 880, duration=1, linear=True)
Display spectrogram for both exponential and linear chirps
>>> import matplotlib.pyplot as plt
>>> plt.figure()
>>> S_exponential = librosa.feature.melspectrogram(y=exponential_chirp)
>>> ax = plt.subplot(2,1,1)
>>> librosa.display.specshow(librosa.power_to_db(S_exponential, ref=np.max),
... x_axis='time', y_axis='mel')
>>> plt.subplot(2,1,2, sharex=ax)
>>> S_linear = librosa.feature.melspectrogram(y=linear_chirp)
>>> librosa.display.specshow(librosa.power_to_db(S_linear, ref=np.max),
... x_axis='time', y_axis='mel')
>>> plt.tight_layout()
"""
if fmin is None or fmax is None:
raise ParameterError('both "fmin" and "fmax" must be provided')
# Compute signal duration
period = 1.0 / sr
if length is None:
if duration is None:
raise ParameterError('either "length" or "duration" must be provided')
else:
duration = period * length
if phi is None:
phi = -np.pi * 0.5
method = 'linear' if linear else 'logarithmic'
return scipy.signal.chirp(
np.arange(duration, step=period),
fmin,
duration,
fmax,
method=method,
phi=phi / np.pi * 180, # scipy.signal.chirp uses degrees for phase offset
)
| [
"numpy.abs",
"resampy.filters.get_filter",
"numpy.mean",
"numpy.arange",
"numpy.sqrt",
"numpy.round",
"resampy.resample",
"six.callable",
"numpy.ceil",
"os.path.realpath",
"audioread.audio_open",
"scipy.fftpack.ifft",
"scipy.fftpack.fft",
"numpy.concatenate",
"numpy.iscomplexobj",
"num... | [((552, 593), 'resampy.filters.get_filter', 'resampy.filters.get_filter', (['"""kaiser_best"""'], {}), "('kaiser_best')\n", (578, 593), False, 'import resampy\n'), ((610, 651), 'resampy.filters.get_filter', 'resampy.filters.get_filter', (['"""kaiser_fast"""'], {}), "('kaiser_fast')\n", (636, 651), False, 'import resampy\n'), ((4373, 4409), 'numpy.ascontiguousarray', 'np.ascontiguousarray', (['y'], {'dtype': 'dtype'}), '(y, dtype=dtype)\n', (4393, 4409), True, 'import numpy as np\n'), ((7361, 7403), 'numpy.ascontiguousarray', 'np.ascontiguousarray', (['y_hat'], {'dtype': 'y.dtype'}), '(y_hat, dtype=y.dtype)\n', (7381, 7403), True, 'import numpy as np\n'), ((12333, 12379), 'scipy.fftpack.ifft', 'fft.ifft', (['powspec'], {'axis': 'axis', 'overwrite_x': '(True)'}), '(powspec, axis=axis, overwrite_x=True)\n', (12341, 12379), True, 'import scipy.fftpack as fft\n'), ((15692, 15719), 'six.callable', 'six.callable', (['ref_magnitude'], {}), '(ref_magnitude)\n', (15704, 15719), False, 'import six\n'), ((20414, 20448), 'numpy.zeros', 'np.zeros', (['length'], {'dtype': 'np.float32'}), '(length, dtype=np.float32)\n', (20422, 20448), True, 'import numpy as np\n'), ((4043, 4060), 'numpy.concatenate', 'np.concatenate', (['y'], {}), '(y)\n', (4057, 4060), True, 'import numpy as np\n'), ((5129, 5147), 'numpy.mean', 'np.mean', (['y'], {'axis': '(0)'}), '(y, axis=0)\n', (5136, 5147), True, 'import numpy as np\n'), ((7017, 7045), 'numpy.ceil', 'np.ceil', (['(y.shape[-1] * ratio)'], {}), '(y.shape[-1] * ratio)\n', (7024, 7045), True, 'import numpy as np\n'), ((7163, 7228), 'resampy.resample', 'resampy.resample', (['y', 'orig_sr', 'target_sr'], {'filter': 'res_type', 'axis': '(-1)'}), '(y, orig_sr, target_sr, filter=res_type, axis=-1)\n', (7179, 7228), False, 'import resampy\n'), ((7334, 7348), 'numpy.sqrt', 'np.sqrt', (['ratio'], {}), '(ratio)\n', (7341, 7348), True, 'import numpy as np\n'), ((12546, 12564), 'numpy.iscomplexobj', 'np.iscomplexobj', (['y'], {}), '(y)\n', (12561, 12564), True, 'import numpy as np\n'), ((16005, 16018), 'numpy.signbit', 'np.signbit', (['y'], {}), '(y)\n', (16015, 16018), True, 'import numpy as np\n'), ((16046, 16056), 'numpy.sign', 'np.sign', (['y'], {}), '(y)\n', (16053, 16056), True, 'import numpy as np\n'), ((25526, 25558), 'numpy.arange', 'np.arange', (['duration'], {'step': 'period'}), '(duration, step=period)\n', (25535, 25558), True, 'import numpy as np\n'), ((2916, 2938), 'os.path.realpath', 'os.path.realpath', (['path'], {}), '(path)\n', (2932, 2938), False, 'import os\n'), ((10007, 10037), 'audioread.audio_open', 'audioread.audio_open', (['filename'], {}), '(filename)\n', (10027, 10037), False, 'import audioread\n'), ((12232, 12278), 'scipy.fftpack.fft', 'fft.fft', (['y'], {'n': '(2 * y.shape[axis] + 1)', 'axis': 'axis'}), '(y, n=2 * y.shape[axis] + 1, axis=axis)\n', (12239, 12278), True, 'import scipy.fftpack as fft\n'), ((3061, 3089), 'numpy.round', 'np.round', (['(sr_native * offset)'], {}), '(sr_native * offset)\n', (3069, 3089), True, 'import numpy as np\n'), ((15767, 15776), 'numpy.abs', 'np.abs', (['y'], {}), '(y)\n', (15773, 15776), True, 'import numpy as np\n'), ((15915, 15924), 'numpy.abs', 'np.abs', (['y'], {}), '(y)\n', (15921, 15924), True, 'import numpy as np\n'), ((22655, 22690), 'numpy.arange', 'np.arange', (['(step * length)'], {'step': 'step'}), '(step * length, step=step)\n', (22664, 22690), True, 'import numpy as np\n'), ((19916, 19945), 'numpy.round', 'np.round', (['(sr * click_duration)'], {}), '(sr * click_duration)\n', (19924, 19945), True, 'import numpy as np\n'), ((3210, 3240), 'numpy.round', 'np.round', (['(sr_native * duration)'], {}), '(sr_native * duration)\n', (3218, 3240), True, 'import numpy as np\n')] |
import tkinter as tk
from collections import deque
from tkinter.constants import BUTT, END, GROOVE, NW, RAISED, RIDGE, S, SUNKEN
import numpy as np
import cv2
from PIL import Image,ImageTk
import os
import face_recognition
import time
window =tk.Tk()
window.option_add("*Font","Helvetica 14")
window.geometry("300x350+100+20")
window.title("Face Recognition System")
window.resizable(False, False)
window.configure(bg="#7FACD6")
facerecog_Mainbutton= ImageTk.PhotoImage((Image.open('./Asset/savemain.png')).resize((250,100), Image.ANTIALIAS))
facerecog_Mainbutton_change= ImageTk.PhotoImage((Image.open('./Asset/savemain_change.png')).resize((250,100), Image.ANTIALIAS))
facedetect_Mainbutton= ImageTk.PhotoImage((Image.open('./Asset/facerecognitionmain.png')).resize((250,100), Image.ANTIALIAS))
facedetect_Mainbutton_change=ImageTk.PhotoImage((Image.open('./Asset/facerecognitionmain_change.png')).resize((250,100), Image.ANTIALIAS))
video_capture=cv2.VideoCapture(0+cv2.CAP_DSHOW)
def recog_enter(a):
facerecog.configure(image=facerecog_Mainbutton_change)
def recog_leave(a):
facerecog.configure(image=facerecog_Mainbutton)
def detect_enter(a):
facedetect.configure(image=facedetect_Mainbutton_change)
def detect_leave(a):
facedetect.configure(image=facedetect_Mainbutton)
def save_file():
Name=str(save_Entry.get())
print("Your name is "+Name)
newpath = f'./Known/{Name}'
ret,capture=video_capture.read()
if not os.path.exists(newpath):
os.makedirs(newpath)
print("make new floder")
time.sleep(1)
print("processing.......")
time.sleep(2)
filename=Name+"1.jpg"
cv2.imwrite(f'./Known/{Name}/{filename}',capture)
Path, dirs, files_now = next(os.walk("./Known/{}".format(Name)))
file_count_now = len(files_now)
print("now you have {} picture in floder".format(file_count_now))
print(os.listdir(f'./Known/{Name}'))
else :
Path, dirs, files = next(os.walk("./Known/{}".format(Name)))
file_count_beta = len(files)
print("Before : you have {} picture in floder ".format(file_count_beta))
print(os.listdir(f'./Known/{Name}'))
time.sleep(1)
print("processing.......")
time.sleep(2)
filename=Name+str(file_count_beta+1)+".jpg"
cv2.imwrite('./Known/{}/{}'.format(Name,filename),capture)
Path, dirs, files_now = next(os.walk("./Known/{}".format(Name)))
file_count_now = len(files_now)
print("After : you have {} picture in floder".format(file_count_now))
print(os.listdir(f'./Known/{Name}'))
def facerecognition():
def reset():
fps_label.pack_forget()
image_label.place_forget()
def Entry_Callback(event):
save_Entry.selection_range(0, END)
reset()
new_tab.title("เธเธเธเธณเนเธเธซเธเนเธฒ")
window.geometry("300x490+100+20")
image_label.place(x = 0, y = 0)
save_Label.configure(text='เธฅเธเธเธทเนเธญเธเธนเนเนเธเนเธฃเธฐเธเธ')
save_Label.pack(pady=20)
save_Entry.pack()
save_Entry.bind("<FocusIn>",Entry_Callback)
save_Button.pack(pady=10)
fps_label._frame_times = deque([0]*5)
fps_label.pack()
cascPathface = os.path.dirname(cv2.__file__) + "/data/haarcascade_frontalface_alt2.xml"
faceCascade = cv2.CascadeClassifier(cascPathface)
video_capture= cv2.VideoCapture(0+cv2.CAP_DSHOW)
video_capture.set(3, 640)
video_capture.set(4, 480)
def all_update(new_tab, image_label, video_capture,fps_label):
show_frames(image_label, video_capture)
update_fps(fps_label)
new_tab.after(0, func=lambda: all_update(new_tab, image_label, video_capture,fps_label))
def update_fps(fps_label):
frame_times = fps_label._frame_times
frame_times.rotate()
frame_times[0] = time.time()
sum_of_deltas = frame_times[0] - frame_times[-1]
count_of_deltas = len(frame_times) - 1
try:
fps = int(float(count_of_deltas) / sum_of_deltas)
except ZeroDivisionError:
fps = 0
fps_label.configure(text=("FPS: {}".format(fps)))
def show_frames(image_label, video_capture):
ret,frame = video_capture.read(0)
gray=cv2.cvtColor(frame,cv2.COLOR_BGR2GRAY)
Face = faceCascade.detectMultiScale(gray,scaleFactor=1.1,minNeighbors=5,minSize=(60, 60),flags=cv2.CASCADE_SCALE_IMAGE)
for (x, y, w, h) in Face:
cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 2)
cv2.waitKey(0)
img = ImageTk.PhotoImage(Image.fromarray(cv2.cvtColor(cv2.resize(frame,(0,0),fx=1.35,fy=1.35), cv2.COLOR_BGR2RGB)))
image_label.configure(image=img)
image_label._image_cache = img
(new_tab).update()
new_tab.after(0,func=lambda:all_update(new_tab, image_label, video_capture,fps_label))
new_tab.mainloop()
def facedetect():
def reset():
fps_label.pack_forget()
image_label.place_forget()
save_Label.pack_forget()
save_Entry.pack_forget()
save_Button.pack_forget()
reset()
new_tab.title("เธเธฃเธงเธเธชเธญเธเนเธเธซเธเนเธฒ")
window.geometry("300x370+100+20")
image_label.place(x = 0, y = 0)
fps_label._frame_times = deque([0]*5)
fps_label.pack()
video_capture = cv2.VideoCapture(0+cv2.CAP_DSHOW)
video_capture.set(3, 640)
video_capture.set(4, 480)
known_faces = []
known_names = []
try:
for name in os.listdir('Known'):
for filename in os.listdir(f'Known/{name}'):
image = face_recognition.load_image_file(f'Known/{name}/{filename}')
test_encoding = face_recognition.face_encodings(image)
if len(test_encoding) > 0 :
encoding = test_encoding[0]
else:
continue
known_faces.append(encoding)
known_names.append(name)
except:
known_faces.append(None)
known_names.append("Unknown")
def all_update(new_tab, image_label, video_capture,fps_label):
show_frames(image_label, video_capture)
update_fps(fps_label)
new_tab.after(0, func=lambda: all_update(new_tab, image_label, video_capture,fps_label))
def update_fps(fps_label):
frame_times = fps_label._frame_times
frame_times.rotate()
frame_times[0] = time.time()
sum_of_deltas = frame_times[0] - frame_times[-1]
count_of_deltas = len(frame_times) - 1
try:
fps = int(float(count_of_deltas) / sum_of_deltas)
except ZeroDivisionError:
fps = 0
fps_label.configure(text=("FPS: {}".format(fps)))
def show_frames(image_label, video_capture):
face_locations = []
face_encodings = []
face_name = []
face_percent=[]
ret,frame = video_capture.read(0)
if not ret:
new_tab.destroy()
rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
small_frame = cv2.resize(rgb, (0, 0), fx=1/4, fy=1/4)
try:
face_locations = face_recognition.face_locations(small_frame)
face_encodings = face_recognition.face_encodings(small_frame,face_locations)
for encoding in face_encodings:
face_distance=face_recognition.face_distance(known_faces,encoding)
best_match = np.argmin(face_distance)
value_percent = 1-face_distance[best_match]
if value_percent >=0.5:
name=known_names[best_match]
percent=round(value_percent*100,0)
face_percent.append(int(percent))
else:
name='Unknown'
face_percent.append(0)
face_name.append(name)
for ( (TOP,RIGHT,BOTTOM,LEFT), name,percent) in zip( face_locations, face_name,face_percent):
if name == 'Unknown':
color_rectangle= [46,2,209]
color_Match=[0, 0, 255]
else:
color_rectangle=[255,102,51]
color_Match=[0,255,0]
cv2.rectangle(frame, (LEFT*4, TOP*4), (RIGHT*4,BOTTOM*4), color_rectangle, 2)
cv2.putText(frame, name, (LEFT*4, TOP*4), cv2.FONT_HERSHEY_COMPLEX,1, (255, 255, 255), 2)
cv2.putText(frame,'Matches '+str(percent)+' %',(LEFT*4,(BOTTOM*4)+16),cv2.FONT_HERSHEY_COMPLEX,0.5, color_Match, 1)
except:
face_locations = face_recognition.face_locations(small_frame)
for (TOP,RIGHT,BOTTOM,LEFT) in face_locations:
name = 'Unknown'
color_rectangle= [46,2,209]
color_Match=[0, 0, 255]
cv2.rectangle(frame, (LEFT*4, TOP*4), (RIGHT*4,BOTTOM*4), color_rectangle, 2)
cv2.putText(frame, name, (LEFT*4, TOP*4), cv2.FONT_HERSHEY_COMPLEX,1, (255, 255, 255), 2)
cv2.putText(frame,'Matches '+'0'+' %',(LEFT*4,(BOTTOM*4)+16),cv2.FONT_HERSHEY_COMPLEX,0.5, color_Match, 1)
cv2.waitKey(0)
img = ImageTk.PhotoImage(Image.fromarray(cv2.cvtColor(cv2.resize(frame,(0,0),fx=1.35,fy=1.35), cv2.COLOR_BGR2RGB)))
image_label.configure(image=img)
image_label._image_cache = img
(new_tab).update()
new_tab.after(0,func=lambda:all_update(new_tab, image_label, video_capture,fps_label))
new_tab.mainloop()
label_main = (tk.Label(master=window,text="Face Recognition System",bg="#7FACD6")).pack(pady=30)
facerecog=tk.Button(master=window,image=facerecog_Mainbutton,command=facerecognition,background="#7FACD6",activebackground="#7FACD6",borderwidth=0)
facerecog.pack()
facerecog.bind("<Enter>", recog_enter)
facerecog.bind("<Leave>", recog_leave)
facedetect=tk.Button(master=window,image=facedetect_Mainbutton,command=facedetect,background="#7FACD6",activebackground="#7FACD6",borderwidth=0)
facedetect.pack(pady=10)
facedetect.bind("<Enter>", detect_enter)
facedetect.bind("<Leave>", detect_leave)
new_tab=tk.Toplevel()
new_tab.option_add("*Font","Helvetica 14")
new_tab.configure(bg='black')
new_tab.title("เธชเธงเธฑเธชเธเธตเธเธฃเธฑเธ")
new_tab.geometry("%dx%d+%d+%d" % (864,648, 300+100, 20))
new_tab.resizable(False, False)
image_label=tk.Label(master=new_tab,borderwidth=0)
save_Label =tk.Label(master=window,bg="#7FACD6",borderwidth=0)
save_Entry=tk.Entry(master=window,borderwidth=1,width=16,relief=SUNKEN,fg="#6E3CBC")
save_Button=tk.Button(master=window,command=save_file,bg="#B8E4F0",activebackground="#B8E4F0",text='เธเธฑเธเธเธถเธเธฃเธนเธเธ เธฒเธ',relief=RAISED,borderwidth=1)
fps_label = tk.Label(master=window,background="#7FACD6",foreground="#225140",borderwidth=0)
new_tab.mainloop()
window.mainloop() | [
"numpy.argmin",
"cv2.rectangle",
"tkinter.Label",
"collections.deque",
"tkinter.Button",
"cv2.imwrite",
"os.path.dirname",
"tkinter.Entry",
"os.path.exists",
"cv2.cvtColor",
"face_recognition.face_encodings",
"tkinter.Toplevel",
"tkinter.Tk",
"cv2.resize",
"face_recognition.face_distance... | [((254, 261), 'tkinter.Tk', 'tk.Tk', ([], {}), '()\n', (259, 261), True, 'import tkinter as tk\n'), ((971, 1006), 'cv2.VideoCapture', 'cv2.VideoCapture', (['(0 + cv2.CAP_DSHOW)'], {}), '(0 + cv2.CAP_DSHOW)\n', (987, 1006), False, 'import cv2\n'), ((9862, 10013), 'tkinter.Button', 'tk.Button', ([], {'master': 'window', 'image': 'facerecog_Mainbutton', 'command': 'facerecognition', 'background': '"""#7FACD6"""', 'activebackground': '"""#7FACD6"""', 'borderwidth': '(0)'}), "(master=window, image=facerecog_Mainbutton, command=\n facerecognition, background='#7FACD6', activebackground='#7FACD6',\n borderwidth=0)\n", (9871, 10013), True, 'import tkinter as tk\n'), ((10110, 10252), 'tkinter.Button', 'tk.Button', ([], {'master': 'window', 'image': 'facedetect_Mainbutton', 'command': 'facedetect', 'background': '"""#7FACD6"""', 'activebackground': '"""#7FACD6"""', 'borderwidth': '(0)'}), "(master=window, image=facedetect_Mainbutton, command=facedetect,\n background='#7FACD6', activebackground='#7FACD6', borderwidth=0)\n", (10119, 10252), True, 'import tkinter as tk\n'), ((10363, 10376), 'tkinter.Toplevel', 'tk.Toplevel', ([], {}), '()\n', (10374, 10376), True, 'import tkinter as tk\n'), ((10586, 10625), 'tkinter.Label', 'tk.Label', ([], {'master': 'new_tab', 'borderwidth': '(0)'}), '(master=new_tab, borderwidth=0)\n', (10594, 10625), True, 'import tkinter as tk\n'), ((10638, 10690), 'tkinter.Label', 'tk.Label', ([], {'master': 'window', 'bg': '"""#7FACD6"""', 'borderwidth': '(0)'}), "(master=window, bg='#7FACD6', borderwidth=0)\n", (10646, 10690), True, 'import tkinter as tk\n'), ((10701, 10778), 'tkinter.Entry', 'tk.Entry', ([], {'master': 'window', 'borderwidth': '(1)', 'width': '(16)', 'relief': 'SUNKEN', 'fg': '"""#6E3CBC"""'}), "(master=window, borderwidth=1, width=16, relief=SUNKEN, fg='#6E3CBC')\n", (10709, 10778), True, 'import tkinter as tk\n'), ((10788, 10929), 'tkinter.Button', 'tk.Button', ([], {'master': 'window', 'command': 'save_file', 'bg': '"""#B8E4F0"""', 'activebackground': '"""#B8E4F0"""', 'text': '"""เธเธฑเธเธเธถเธเธฃเธนเธเธ เธฒเธ"""', 'relief': 'RAISED', 'borderwidth': '(1)'}), "(master=window, command=save_file, bg='#B8E4F0', activebackground=\n '#B8E4F0', text='เธเธฑเธเธเธถเธเธฃเธนเธเธ เธฒเธ', relief=RAISED, borderwidth=1)\n", (10797, 10929), True, 'import tkinter as tk\n'), ((10932, 11018), 'tkinter.Label', 'tk.Label', ([], {'master': 'window', 'background': '"""#7FACD6"""', 'foreground': '"""#225140"""', 'borderwidth': '(0)'}), "(master=window, background='#7FACD6', foreground='#225140',\n borderwidth=0)\n", (10940, 11018), True, 'import tkinter as tk\n'), ((3323, 3337), 'collections.deque', 'deque', (['([0] * 5)'], {}), '([0] * 5)\n', (3328, 3337), False, 'from collections import deque\n'), ((3472, 3507), 'cv2.CascadeClassifier', 'cv2.CascadeClassifier', (['cascPathface'], {}), '(cascPathface)\n', (3493, 3507), False, 'import cv2\n'), ((3528, 3563), 'cv2.VideoCapture', 'cv2.VideoCapture', (['(0 + cv2.CAP_DSHOW)'], {}), '(0 + cv2.CAP_DSHOW)\n', (3544, 3563), False, 'import cv2\n'), ((5445, 5459), 'collections.deque', 'deque', (['([0] * 5)'], {}), '([0] * 5)\n', (5450, 5459), False, 'from collections import deque\n'), ((5503, 5538), 'cv2.VideoCapture', 'cv2.VideoCapture', (['(0 + cv2.CAP_DSHOW)'], {}), '(0 + cv2.CAP_DSHOW)\n', (5519, 5538), False, 'import cv2\n'), ((1512, 1535), 'os.path.exists', 'os.path.exists', (['newpath'], {}), '(newpath)\n', (1526, 1535), False, 'import os\n'), ((1550, 1570), 'os.makedirs', 'os.makedirs', (['newpath'], {}), '(newpath)\n', (1561, 1570), False, 'import os\n'), ((1622, 1635), 'time.sleep', 'time.sleep', (['(1)'], {}), '(1)\n', (1632, 1635), False, 'import time\n'), ((1689, 1702), 'time.sleep', 'time.sleep', (['(2)'], {}), '(2)\n', (1699, 1702), False, 'import time\n'), ((1751, 1801), 'cv2.imwrite', 'cv2.imwrite', (['f"""./Known/{Name}/{filename}"""', 'capture'], {}), "(f'./Known/{Name}/{filename}', capture)\n", (1762, 1801), False, 'import cv2\n'), ((2334, 2347), 'time.sleep', 'time.sleep', (['(1)'], {}), '(1)\n', (2344, 2347), False, 'import time\n'), ((2401, 2414), 'time.sleep', 'time.sleep', (['(2)'], {}), '(2)\n', (2411, 2414), False, 'import time\n'), ((3380, 3409), 'os.path.dirname', 'os.path.dirname', (['cv2.__file__'], {}), '(cv2.__file__)\n', (3395, 3409), False, 'import os\n'), ((4009, 4020), 'time.time', 'time.time', ([], {}), '()\n', (4018, 4020), False, 'import time\n'), ((4426, 4465), 'cv2.cvtColor', 'cv2.cvtColor', (['frame', 'cv2.COLOR_BGR2GRAY'], {}), '(frame, cv2.COLOR_BGR2GRAY)\n', (4438, 4465), False, 'import cv2\n'), ((4712, 4726), 'cv2.waitKey', 'cv2.waitKey', (['(0)'], {}), '(0)\n', (4723, 4726), False, 'import cv2\n'), ((5674, 5693), 'os.listdir', 'os.listdir', (['"""Known"""'], {}), "('Known')\n", (5684, 5693), False, 'import os\n'), ((6638, 6649), 'time.time', 'time.time', ([], {}), '()\n', (6647, 6649), False, 'import time\n'), ((7215, 7253), 'cv2.cvtColor', 'cv2.cvtColor', (['frame', 'cv2.COLOR_BGR2RGB'], {}), '(frame, cv2.COLOR_BGR2RGB)\n', (7227, 7253), False, 'import cv2\n'), ((7277, 7320), 'cv2.resize', 'cv2.resize', (['rgb', '(0, 0)'], {'fx': '(1 / 4)', 'fy': '(1 / 4)'}), '(rgb, (0, 0), fx=1 / 4, fy=1 / 4)\n', (7287, 7320), False, 'import cv2\n'), ((9386, 9400), 'cv2.waitKey', 'cv2.waitKey', (['(0)'], {}), '(0)\n', (9397, 9400), False, 'import cv2\n'), ((9768, 9837), 'tkinter.Label', 'tk.Label', ([], {'master': 'window', 'text': '"""Face Recognition System"""', 'bg': '"""#7FACD6"""'}), "(master=window, text='Face Recognition System', bg='#7FACD6')\n", (9776, 9837), True, 'import tkinter as tk\n'), ((488, 522), 'PIL.Image.open', 'Image.open', (['"""./Asset/savemain.png"""'], {}), "('./Asset/savemain.png')\n", (498, 522), False, 'from PIL import Image, ImageTk\n'), ((610, 651), 'PIL.Image.open', 'Image.open', (['"""./Asset/savemain_change.png"""'], {}), "('./Asset/savemain_change.png')\n", (620, 651), False, 'from PIL import Image, ImageTk\n'), ((733, 778), 'PIL.Image.open', 'Image.open', (['"""./Asset/facerecognitionmain.png"""'], {}), "('./Asset/facerecognitionmain.png')\n", (743, 778), False, 'from PIL import Image, ImageTk\n'), ((866, 918), 'PIL.Image.open', 'Image.open', (['"""./Asset/facerecognitionmain_change.png"""'], {}), "('./Asset/facerecognitionmain_change.png')\n", (876, 918), False, 'from PIL import Image, ImageTk\n'), ((2022, 2051), 'os.listdir', 'os.listdir', (['f"""./Known/{Name}"""'], {}), "(f'./Known/{Name}')\n", (2032, 2051), False, 'import os\n'), ((2290, 2319), 'os.listdir', 'os.listdir', (['f"""./Known/{Name}"""'], {}), "(f'./Known/{Name}')\n", (2300, 2319), False, 'import os\n'), ((2769, 2798), 'os.listdir', 'os.listdir', (['f"""./Known/{Name}"""'], {}), "(f'./Known/{Name}')\n", (2779, 2798), False, 'import os\n'), ((4642, 4702), 'cv2.rectangle', 'cv2.rectangle', (['frame', '(x, y)', '(x + w, y + h)', '(0, 255, 0)', '(2)'], {}), '(frame, (x, y), (x + w, y + h), (0, 255, 0), 2)\n', (4655, 4702), False, 'import cv2\n'), ((5724, 5751), 'os.listdir', 'os.listdir', (['f"""Known/{name}"""'], {}), "(f'Known/{name}')\n", (5734, 5751), False, 'import os\n'), ((7361, 7405), 'face_recognition.face_locations', 'face_recognition.face_locations', (['small_frame'], {}), '(small_frame)\n', (7392, 7405), False, 'import face_recognition\n'), ((7436, 7496), 'face_recognition.face_encodings', 'face_recognition.face_encodings', (['small_frame', 'face_locations'], {}), '(small_frame, face_locations)\n', (7467, 7496), False, 'import face_recognition\n'), ((5778, 5838), 'face_recognition.load_image_file', 'face_recognition.load_image_file', (['f"""Known/{name}/{filename}"""'], {}), "(f'Known/{name}/{filename}')\n", (5810, 5838), False, 'import face_recognition\n'), ((5872, 5910), 'face_recognition.face_encodings', 'face_recognition.face_encodings', (['image'], {}), '(image)\n', (5903, 5910), False, 'import face_recognition\n'), ((7572, 7625), 'face_recognition.face_distance', 'face_recognition.face_distance', (['known_faces', 'encoding'], {}), '(known_faces, encoding)\n', (7602, 7625), False, 'import face_recognition\n'), ((7655, 7679), 'numpy.argmin', 'np.argmin', (['face_distance'], {}), '(face_distance)\n', (7664, 7679), True, 'import numpy as np\n'), ((8459, 8549), 'cv2.rectangle', 'cv2.rectangle', (['frame', '(LEFT * 4, TOP * 4)', '(RIGHT * 4, BOTTOM * 4)', 'color_rectangle', '(2)'], {}), '(frame, (LEFT * 4, TOP * 4), (RIGHT * 4, BOTTOM * 4),\n color_rectangle, 2)\n', (8472, 8549), False, 'import cv2\n'), ((8554, 8652), 'cv2.putText', 'cv2.putText', (['frame', 'name', '(LEFT * 4, TOP * 4)', 'cv2.FONT_HERSHEY_COMPLEX', '(1)', '(255, 255, 255)', '(2)'], {}), '(frame, name, (LEFT * 4, TOP * 4), cv2.FONT_HERSHEY_COMPLEX, 1,\n (255, 255, 255), 2)\n', (8565, 8652), False, 'import cv2\n'), ((8824, 8868), 'face_recognition.face_locations', 'face_recognition.face_locations', (['small_frame'], {}), '(small_frame)\n', (8855, 8868), False, 'import face_recognition\n'), ((4790, 4833), 'cv2.resize', 'cv2.resize', (['frame', '(0, 0)'], {'fx': '(1.35)', 'fy': '(1.35)'}), '(frame, (0, 0), fx=1.35, fy=1.35)\n', (4800, 4833), False, 'import cv2\n'), ((9068, 9158), 'cv2.rectangle', 'cv2.rectangle', (['frame', '(LEFT * 4, TOP * 4)', '(RIGHT * 4, BOTTOM * 4)', 'color_rectangle', '(2)'], {}), '(frame, (LEFT * 4, TOP * 4), (RIGHT * 4, BOTTOM * 4),\n color_rectangle, 2)\n', (9081, 9158), False, 'import cv2\n'), ((9163, 9261), 'cv2.putText', 'cv2.putText', (['frame', 'name', '(LEFT * 4, TOP * 4)', 'cv2.FONT_HERSHEY_COMPLEX', '(1)', '(255, 255, 255)', '(2)'], {}), '(frame, name, (LEFT * 4, TOP * 4), cv2.FONT_HERSHEY_COMPLEX, 1,\n (255, 255, 255), 2)\n', (9174, 9261), False, 'import cv2\n'), ((9270, 9393), 'cv2.putText', 'cv2.putText', (['frame', "('Matches ' + '0' + ' %')", '(LEFT * 4, BOTTOM * 4 + 16)', 'cv2.FONT_HERSHEY_COMPLEX', '(0.5)', 'color_Match', '(1)'], {}), "(frame, 'Matches ' + '0' + ' %', (LEFT * 4, BOTTOM * 4 + 16),\n cv2.FONT_HERSHEY_COMPLEX, 0.5, color_Match, 1)\n", (9281, 9393), False, 'import cv2\n'), ((9464, 9507), 'cv2.resize', 'cv2.resize', (['frame', '(0, 0)'], {'fx': '(1.35)', 'fy': '(1.35)'}), '(frame, (0, 0), fx=1.35, fy=1.35)\n', (9474, 9507), False, 'import cv2\n')] |
import numpy as np
import cv2 as cv
import cmath
#cap = cv.VideoCapture(0)
cap = cv.VideoCapture('udpsrc port=5004 ! application/x-rtp,encoding-name=H264,payload=96 ! rtph264depay ! avdec_h264 ! videoconvert ! appsink', cv.CAP_GSTREAMER)
PI = 3.14159
while(1):
# read the video capture frame
_, frame = cap.read()
#cv.imshow('frame',frame)
#break
# blur for better edge finding
blur = cv.GaussianBlur(frame,(5,5),0)
frameGray = cv.cvtColor(blur, cv.COLOR_BGR2GRAY)
# create threshold for edge finding
ret, thresh = cv.threshold(frameGray, 120, 255, cv.THRESH_BINARY)
contours, _ = cv.findContours(thresh, cv.RETR_TREE, cv.CHAIN_APPROX_NONE)
count = 0
tri = 0
sqr = 0
rect = 0
circ = 0
for contour in contours:
area = cv.contourArea(contour)
if area > 1000 and area < 30000:
M = cv.moments(contour)
cX = int(M["m10"]/M["m00"])
cY = int(M["m01"]/M["m00"])
if(frame[cY,cX][0] < 50 and frame[cY,cX][1] < 50 and frame[cY,cX][2] < 50):
cv.circle(frame, (cX,cY), 7, (255,255,0), -1)
#cv.drawContours(frame, contour, -1, (0,255,0), 3)
count += 1
(x,y), (MA, ma), angle = cv.fitEllipse(contour)
areaEllipse = PI/4 * MA * ma
if(abs(areaEllipse - area) < 100):
#is circle
circ += 1
cv.drawContours(frame, contour, -1, (0,255,255), 3)
else:
((x,y), (w,h), rot) = cv.minAreaRect(contour)
if(float(w) > 0.0 and float(h) > 0.0):
ratio = w / float(h)
#font = cv.FONT_HERSHEY_COMPLEX_SMALL
#cv.putText(frame, str(ratio), (cX, cY - 40), font, 2, (0, 0, 255), 2, cv.LINE_AA)
if ratio <= 0.6 or ratio >= 2.8:
#is rect
cv.drawContours(frame, contour, -1, (0,255,0), 3)
rect += 1
else:
#peri = cv.arcLength(contour, True)
#approx = cv.approxPolyDP(contour, 0.04 * peri, True)
#if len(approx) == 3:
areaAdj = 1400
#font = cv.FONT_HERSHEY_COMPLEX_SMALL
#cv.putText(frame, str(int(area)), (cX, cY - 40), font, 2, (0, 0, 255), 2, cv.LINE_AA)
#cv.putText(frame, str(int(w*h/2)), (cX, cY - 60), font, 2, (0, 0, 255), 2, cv.LINE_AA)
if(w*h/2 > area - areaAdj and w*h/2 < area + areaAdj):
#is triangle
cv.drawContours(frame, contour, -1, (255,0,0), 3)
tri += 1
else:
#is square
sqr += 1
cv.drawContours(frame, contour, -1, (0,0,255), 3)
cv.circle(frame, (70, 300), 20, (0,0,255), -1)
pts = np.array([[70, 330], [50, 360], [90, 360]], np.int32)
pts = pts.reshape((-1,1,2))
cv.fillPoly(frame, [pts], (0, 0, 255))
cv.rectangle(frame, (50, 381), (90, 389), (0,0,255), -1)
cv.rectangle(frame, (50, 410), (90, 450), (0,0,255), -1)
font = cv.FONT_HERSHEY_COMPLEX_SMALL
cv.putText(frame, str(circ), (10, 310), font, 2, (0, 0, 255), 2, cv.LINE_AA)
cv.putText(frame, str(tri), (10, 355), font, 2, (0, 0, 255), 2, cv.LINE_AA)
cv.putText(frame, str(rect), (10, 400), font, 2, (0, 0, 255), 2, cv.LINE_AA)
cv.putText(frame, str(sqr), (10, 445), font, 2, (0, 0, 255), 2, cv.LINE_AA)
cv.imshow('frame',frame)
#cv.imshow('thresh', thresh)
k = cv.waitKey(5) & 0xFF
if k == 27:
break
cv.destroyAllWindows()
cap.release()
| [
"cv2.minAreaRect",
"cv2.GaussianBlur",
"cv2.contourArea",
"cv2.circle",
"cv2.cvtColor",
"cv2.waitKey",
"cv2.threshold",
"cv2.moments",
"cv2.imshow",
"cv2.fillPoly",
"cv2.VideoCapture",
"cv2.fitEllipse",
"numpy.array",
"cv2.rectangle",
"cv2.drawContours",
"cv2.destroyAllWindows",
"cv2... | [((82, 248), 'cv2.VideoCapture', 'cv.VideoCapture', (['"""udpsrc port=5004 ! application/x-rtp,encoding-name=H264,payload=96 ! rtph264depay ! avdec_h264 ! videoconvert ! appsink"""', 'cv.CAP_GSTREAMER'], {}), "(\n 'udpsrc port=5004 ! application/x-rtp,encoding-name=H264,payload=96 ! rtph264depay ! avdec_h264 ! videoconvert ! appsink'\n , cv.CAP_GSTREAMER)\n", (97, 248), True, 'import cv2 as cv\n'), ((4034, 4056), 'cv2.destroyAllWindows', 'cv.destroyAllWindows', ([], {}), '()\n', (4054, 4056), True, 'import cv2 as cv\n'), ((421, 454), 'cv2.GaussianBlur', 'cv.GaussianBlur', (['frame', '(5, 5)', '(0)'], {}), '(frame, (5, 5), 0)\n', (436, 454), True, 'import cv2 as cv\n'), ((468, 504), 'cv2.cvtColor', 'cv.cvtColor', (['blur', 'cv.COLOR_BGR2GRAY'], {}), '(blur, cv.COLOR_BGR2GRAY)\n', (479, 504), True, 'import cv2 as cv\n'), ((563, 614), 'cv2.threshold', 'cv.threshold', (['frameGray', '(120)', '(255)', 'cv.THRESH_BINARY'], {}), '(frameGray, 120, 255, cv.THRESH_BINARY)\n', (575, 614), True, 'import cv2 as cv\n'), ((633, 692), 'cv2.findContours', 'cv.findContours', (['thresh', 'cv.RETR_TREE', 'cv.CHAIN_APPROX_NONE'], {}), '(thresh, cv.RETR_TREE, cv.CHAIN_APPROX_NONE)\n', (648, 692), True, 'import cv2 as cv\n'), ((3239, 3287), 'cv2.circle', 'cv.circle', (['frame', '(70, 300)', '(20)', '(0, 0, 255)', '(-1)'], {}), '(frame, (70, 300), 20, (0, 0, 255), -1)\n', (3248, 3287), True, 'import cv2 as cv\n'), ((3296, 3349), 'numpy.array', 'np.array', (['[[70, 330], [50, 360], [90, 360]]', 'np.int32'], {}), '([[70, 330], [50, 360], [90, 360]], np.int32)\n', (3304, 3349), True, 'import numpy as np\n'), ((3386, 3424), 'cv2.fillPoly', 'cv.fillPoly', (['frame', '[pts]', '(0, 0, 255)'], {}), '(frame, [pts], (0, 0, 255))\n', (3397, 3424), True, 'import cv2 as cv\n'), ((3429, 3487), 'cv2.rectangle', 'cv.rectangle', (['frame', '(50, 381)', '(90, 389)', '(0, 0, 255)', '(-1)'], {}), '(frame, (50, 381), (90, 389), (0, 0, 255), -1)\n', (3441, 3487), True, 'import cv2 as cv\n'), ((3490, 3548), 'cv2.rectangle', 'cv.rectangle', (['frame', '(50, 410)', '(90, 450)', '(0, 0, 255)', '(-1)'], {}), '(frame, (50, 410), (90, 450), (0, 0, 255), -1)\n', (3502, 3548), True, 'import cv2 as cv\n'), ((3915, 3940), 'cv2.imshow', 'cv.imshow', (['"""frame"""', 'frame'], {}), "('frame', frame)\n", (3924, 3940), True, 'import cv2 as cv\n'), ((802, 825), 'cv2.contourArea', 'cv.contourArea', (['contour'], {}), '(contour)\n', (816, 825), True, 'import cv2 as cv\n'), ((3982, 3995), 'cv2.waitKey', 'cv.waitKey', (['(5)'], {}), '(5)\n', (3992, 3995), True, 'import cv2 as cv\n'), ((883, 902), 'cv2.moments', 'cv.moments', (['contour'], {}), '(contour)\n', (893, 902), True, 'import cv2 as cv\n'), ((1087, 1135), 'cv2.circle', 'cv.circle', (['frame', '(cX, cY)', '(7)', '(255, 255, 0)', '(-1)'], {}), '(frame, (cX, cY), 7, (255, 255, 0), -1)\n', (1096, 1135), True, 'import cv2 as cv\n'), ((1285, 1307), 'cv2.fitEllipse', 'cv.fitEllipse', (['contour'], {}), '(contour)\n', (1298, 1307), True, 'import cv2 as cv\n'), ((1485, 1538), 'cv2.drawContours', 'cv.drawContours', (['frame', 'contour', '(-1)', '(0, 255, 255)', '(3)'], {}), '(frame, contour, -1, (0, 255, 255), 3)\n', (1500, 1538), True, 'import cv2 as cv\n'), ((1601, 1624), 'cv2.minAreaRect', 'cv.minAreaRect', (['contour'], {}), '(contour)\n', (1615, 1624), True, 'import cv2 as cv\n'), ((2046, 2097), 'cv2.drawContours', 'cv.drawContours', (['frame', 'contour', '(-1)', '(0, 255, 0)', '(3)'], {}), '(frame, contour, -1, (0, 255, 0), 3)\n', (2061, 2097), True, 'import cv2 as cv\n'), ((2943, 2994), 'cv2.drawContours', 'cv.drawContours', (['frame', 'contour', '(-1)', '(255, 0, 0)', '(3)'], {}), '(frame, contour, -1, (255, 0, 0), 3)\n', (2958, 2994), True, 'import cv2 as cv\n'), ((3184, 3235), 'cv2.drawContours', 'cv.drawContours', (['frame', 'contour', '(-1)', '(0, 0, 255)', '(3)'], {}), '(frame, contour, -1, (0, 0, 255), 3)\n', (3199, 3235), True, 'import cv2 as cv\n')] |
import sys
import sysconfig
import warnings
import numpy as nu
import ctypes
import ctypes.util
from numpy.ctypeslib import ndpointer
import os
from galpy import potential
from galpy.util import galpyWarning
from galpy.orbit_src.integratePlanarOrbit import _parse_integrator, _parse_tol
#Find and load the library
_lib= None
outerr= None
PY3= sys.version > '3'
if PY3: #pragma: no cover
_ext_suffix= sysconfig.get_config_var('EXT_SUFFIX')
else:
_ext_suffix= '.so'
for path in sys.path:
try:
_lib = ctypes.CDLL(os.path.join(path,'galpy_integrate_c%s' % _ext_suffix))
except OSError as e:
if os.path.exists(os.path.join(path,'galpy_integrate_c%s' % _ext_suffix)): #pragma: no cover
outerr= e
_lib = None
else:
break
if _lib is None: #pragma: no cover
if not outerr is None:
warnings.warn("integrateFullOrbit_c extension module not loaded, because of error '%s' " % outerr,
galpyWarning)
else:
warnings.warn("integrateFullOrbit_c extension module not loaded, because galpy_integrate_c%s image was not found" % _ext_suffix,
galpyWarning)
_ext_loaded= False
else:
_ext_loaded= True
def _parse_pot(pot,potforactions=False):
"""Parse the potential so it can be fed to C"""
#Figure out what's in pot
if not isinstance(pot,list):
pot= [pot]
#Initialize everything
pot_type= []
pot_args= []
npot= len(pot)
for p in pot:
if isinstance(p,potential.LogarithmicHaloPotential):
pot_type.append(0)
pot_args.extend([p._amp,p._q,p._core2])
elif isinstance(p,potential.MiyamotoNagaiPotential):
pot_type.append(5)
pot_args.extend([p._amp,p._a,p._b])
elif isinstance(p,potential.PowerSphericalPotential):
pot_type.append(7)
pot_args.extend([p._amp,p.alpha])
elif isinstance(p,potential.HernquistPotential):
pot_type.append(8)
pot_args.extend([p._amp,p.a])
elif isinstance(p,potential.FlattenedNFWPotential):
pot_type.append(91)
pot_args.extend([p._amp,p.a,p.q])
elif isinstance(p,potential.NFWPotential):
pot_type.append(9)
pot_args.extend([p._amp,p.a])
elif isinstance(p,potential.JaffePotential):
pot_type.append(10)
pot_args.extend([p._amp,p.a])
elif isinstance(p,potential.DoubleExponentialDiskPotential):
pot_type.append(11)
pot_args.extend([p._amp,p._alpha,p._beta,p._kmaxFac,
p._nzeros,p._glorder])
pot_args.extend([p._glx[ii] for ii in range(p._glorder)])
pot_args.extend([p._glw[ii] for ii in range(p._glorder)])
pot_args.extend([p._j0zeros[ii] for ii in range(p._nzeros+1)])
pot_args.extend([p._dj0zeros[ii] for ii in range(p._nzeros+1)])
pot_args.extend([p._j1zeros[ii] for ii in range(p._nzeros+1)])
pot_args.extend([p._dj1zeros[ii] for ii in range(p._nzeros+1)])
pot_args.extend([p._kp._amp,p._kp.alpha])
elif isinstance(p,potential.FlattenedPowerPotential):
pot_type.append(12)
pot_args.extend([p._amp,p.alpha,p.q2,p.core2])
elif isinstance(p,potential.interpRZPotential):
pot_type.append(13)
pot_args.extend([len(p._rgrid),len(p._zgrid)])
if p._logR:
pot_args.extend([p._logrgrid[ii] for ii in range(len(p._rgrid))])
else:
pot_args.extend([p._rgrid[ii] for ii in range(len(p._rgrid))])
pot_args.extend([p._zgrid[ii] for ii in range(len(p._zgrid))])
if potforactions:
pot_args.extend([x for x in p._potGrid_splinecoeffs.flatten(order='C')])
else:
pot_args.extend([x for x in p._rforceGrid_splinecoeffs.flatten(order='C')])
pot_args.extend([x for x in p._zforceGrid_splinecoeffs.flatten(order='C')])
pot_args.extend([p._amp,int(p._logR)])
elif isinstance(p,potential.IsochronePotential):
pot_type.append(14)
pot_args.extend([p._amp,p.b])
elif isinstance(p,potential.PowerSphericalPotentialwCutoff):
pot_type.append(15)
pot_args.extend([p._amp,p.alpha,p.rc])
elif isinstance(p,potential.MN3ExponentialDiskPotential):
# Three Miyamoto-Nagai disks
npot+= 2
pot_type.extend([5,5,5])
pot_args.extend([p._amp*p._mn3[0]._amp,
p._mn3[0]._a,p._mn3[0]._b,
p._amp*p._mn3[1]._amp,
p._mn3[1]._a,p._mn3[1]._b,
p._amp*p._mn3[2]._amp,
p._mn3[2]._a,p._mn3[2]._b])
elif isinstance(p,potential.KuzminKutuzovStaeckelPotential):
pot_type.append(16)
pot_args.extend([p._amp,p._ac,p._Delta])
elif isinstance(p,potential.PlummerPotential):
pot_type.append(17)
pot_args.extend([p._amp,p._b])
elif isinstance(p,potential.PseudoIsothermalPotential):
pot_type.append(18)
pot_args.extend([p._amp,p._a])
pot_type= nu.array(pot_type,dtype=nu.int32,order='C')
pot_args= nu.array(pot_args,dtype=nu.float64,order='C')
return (npot,pot_type,pot_args)
def integrateFullOrbit_c(pot,yo,t,int_method,rtol=None,atol=None,dt=None):
"""
NAME:
integrateFullOrbit_c
PURPOSE:
C integrate an ode for a FullOrbit
INPUT:
pot - Potential or list of such instances
yo - initial condition [q,p]
t - set of times at which one wants the result
int_method= 'leapfrog_c', 'rk4_c', 'rk6_c', 'symplec4_c'
rtol, atol
dt= (None) force integrator to use this stepsize (default is to automatically determine one))
OUTPUT:
(y,err)
y : array, shape (len(y0), len(t))
Array containing the value of y for each desired time in t, \
with the initial value y0 in the first row.
err: error message, if not zero: 1 means maximum step reduction happened for adaptive integrators
HISTORY:
2011-11-13 - Written - Bovy (IAS)
"""
rtol, atol= _parse_tol(rtol,atol)
npot, pot_type, pot_args= _parse_pot(pot)
int_method_c= _parse_integrator(int_method)
if dt is None:
dt= -9999.99
#Set up result array
result= nu.empty((len(t),6))
err= ctypes.c_int(0)
#Set up the C code
ndarrayFlags= ('C_CONTIGUOUS','WRITEABLE')
integrationFunc= _lib.integrateFullOrbit
integrationFunc.argtypes= [ndpointer(dtype=nu.float64,flags=ndarrayFlags),
ctypes.c_int,
ndpointer(dtype=nu.float64,flags=ndarrayFlags),
ctypes.c_int,
ndpointer(dtype=nu.int32,flags=ndarrayFlags),
ndpointer(dtype=nu.float64,flags=ndarrayFlags),
ctypes.c_double,
ctypes.c_double,
ctypes.c_double,
ndpointer(dtype=nu.float64,flags=ndarrayFlags),
ctypes.POINTER(ctypes.c_int),
ctypes.c_int]
#Array requirements, first store old order
f_cont= [yo.flags['F_CONTIGUOUS'],
t.flags['F_CONTIGUOUS']]
yo= nu.require(yo,dtype=nu.float64,requirements=['C','W'])
t= nu.require(t,dtype=nu.float64,requirements=['C','W'])
result= nu.require(result,dtype=nu.float64,requirements=['C','W'])
#Run the C code
integrationFunc(yo,
ctypes.c_int(len(t)),
t,
ctypes.c_int(npot),
pot_type,
pot_args,
ctypes.c_double(dt),
ctypes.c_double(rtol),ctypes.c_double(atol),
result,
ctypes.byref(err),
ctypes.c_int(int_method_c))
#Reset input arrays
if f_cont[0]: yo= nu.asfortranarray(yo)
if f_cont[1]: t= nu.asfortranarray(t)
return (result,err.value)
def integrateFullOrbit_dxdv_c(pot,yo,dyo,t,int_method,rtol=None,atol=None): #pragma: no cover because not included in v1, uncover when included
"""
NAME:
integrateFullOrbit_dxdv_c
PURPOSE:
C integrate an ode for a planarOrbit+phase space volume dxdv
INPUT:
pot - Potential or list of such instances
yo - initial condition [q,p]
dyo - initial condition [dq,dp]
t - set of times at which one wants the result
int_method= 'leapfrog_c', 'rk4_c', 'rk6_c', 'symplec4_c'
rtol, atol
OUTPUT:
(y,err)
y : array, shape (len(y0), len(t))
Array containing the value of y for each desired time in t, \
with the initial value y0 in the first row.
err: error message if not zero, 1: maximum step reduction happened for adaptive integrators
HISTORY:
2011-11-13 - Written - Bovy (IAS)
"""
rtol, atol= _parse_tol(rtol,atol)
npot, pot_type, pot_args= _parse_pot(pot)
int_method_c= _parse_integrator(int_method)
yo= nu.concatenate((yo,dyo))
#Set up result array
result= nu.empty((len(t),12))
err= ctypes.c_int(0)
#Set up the C code
ndarrayFlags= ('C_CONTIGUOUS','WRITEABLE')
integrationFunc= _lib.integrateFullOrbit_dxdv
integrationFunc.argtypes= [ndpointer(dtype=nu.float64,flags=ndarrayFlags),
ctypes.c_int,
ndpointer(dtype=nu.float64,flags=ndarrayFlags),
ctypes.c_int,
ndpointer(dtype=nu.int32,flags=ndarrayFlags),
ndpointer(dtype=nu.float64,flags=ndarrayFlags),
ctypes.c_double,
ctypes.c_double,
ndpointer(dtype=nu.float64,flags=ndarrayFlags),
ctypes.POINTER(ctypes.c_int),
ctypes.c_int]
#Array requirements, first store old order
f_cont= [yo.flags['F_CONTIGUOUS'],
t.flags['F_CONTIGUOUS']]
yo= nu.require(yo,dtype=nu.float64,requirements=['C','W'])
t= nu.require(t,dtype=nu.float64,requirements=['C','W'])
result= nu.require(result,dtype=nu.float64,requirements=['C','W'])
#Run the C code
integrationFunc(yo,
ctypes.c_int(len(t)),
t,
ctypes.c_int(npot),
pot_type,
pot_args,
ctypes.c_double(rtol),ctypes.c_double(atol),
result,
ctypes.byref(err),
ctypes.c_int(int_method_c))
#Reset input arrays
if f_cont[0]: yo= nu.asfortranarray(yo)
if f_cont[1]: t= nu.asfortranarray(t)
return (result,err.value)
| [
"numpy.ctypeslib.ndpointer",
"ctypes.c_int",
"ctypes.c_double",
"galpy.orbit_src.integratePlanarOrbit._parse_tol",
"ctypes.byref",
"sysconfig.get_config_var",
"numpy.asfortranarray",
"numpy.require",
"numpy.array",
"warnings.warn",
"galpy.orbit_src.integratePlanarOrbit._parse_integrator",
"os.... | [((404, 442), 'sysconfig.get_config_var', 'sysconfig.get_config_var', (['"""EXT_SUFFIX"""'], {}), "('EXT_SUFFIX')\n", (428, 442), False, 'import sysconfig\n'), ((5307, 5352), 'numpy.array', 'nu.array', (['pot_type'], {'dtype': 'nu.int32', 'order': '"""C"""'}), "(pot_type, dtype=nu.int32, order='C')\n", (5315, 5352), True, 'import numpy as nu\n'), ((5365, 5412), 'numpy.array', 'nu.array', (['pot_args'], {'dtype': 'nu.float64', 'order': '"""C"""'}), "(pot_args, dtype=nu.float64, order='C')\n", (5373, 5412), True, 'import numpy as nu\n'), ((6329, 6351), 'galpy.orbit_src.integratePlanarOrbit._parse_tol', '_parse_tol', (['rtol', 'atol'], {}), '(rtol, atol)\n', (6339, 6351), False, 'from galpy.orbit_src.integratePlanarOrbit import _parse_integrator, _parse_tol\n'), ((6415, 6444), 'galpy.orbit_src.integratePlanarOrbit._parse_integrator', '_parse_integrator', (['int_method'], {}), '(int_method)\n', (6432, 6444), False, 'from galpy.orbit_src.integratePlanarOrbit import _parse_integrator, _parse_tol\n'), ((6554, 6569), 'ctypes.c_int', 'ctypes.c_int', (['(0)'], {}), '(0)\n', (6566, 6569), False, 'import ctypes\n'), ((7581, 7638), 'numpy.require', 'nu.require', (['yo'], {'dtype': 'nu.float64', 'requirements': "['C', 'W']"}), "(yo, dtype=nu.float64, requirements=['C', 'W'])\n", (7591, 7638), True, 'import numpy as nu\n'), ((7643, 7699), 'numpy.require', 'nu.require', (['t'], {'dtype': 'nu.float64', 'requirements': "['C', 'W']"}), "(t, dtype=nu.float64, requirements=['C', 'W'])\n", (7653, 7699), True, 'import numpy as nu\n'), ((7709, 7770), 'numpy.require', 'nu.require', (['result'], {'dtype': 'nu.float64', 'requirements': "['C', 'W']"}), "(result, dtype=nu.float64, requirements=['C', 'W'])\n", (7719, 7770), True, 'import numpy as nu\n'), ((9255, 9277), 'galpy.orbit_src.integratePlanarOrbit._parse_tol', '_parse_tol', (['rtol', 'atol'], {}), '(rtol, atol)\n', (9265, 9277), False, 'from galpy.orbit_src.integratePlanarOrbit import _parse_integrator, _parse_tol\n'), ((9341, 9370), 'galpy.orbit_src.integratePlanarOrbit._parse_integrator', '_parse_integrator', (['int_method'], {}), '(int_method)\n', (9358, 9370), False, 'from galpy.orbit_src.integratePlanarOrbit import _parse_integrator, _parse_tol\n'), ((9379, 9404), 'numpy.concatenate', 'nu.concatenate', (['(yo, dyo)'], {}), '((yo, dyo))\n', (9393, 9404), True, 'import numpy as nu\n'), ((9473, 9488), 'ctypes.c_int', 'ctypes.c_int', (['(0)'], {}), '(0)\n', (9485, 9488), False, 'import ctypes\n'), ((10457, 10514), 'numpy.require', 'nu.require', (['yo'], {'dtype': 'nu.float64', 'requirements': "['C', 'W']"}), "(yo, dtype=nu.float64, requirements=['C', 'W'])\n", (10467, 10514), True, 'import numpy as nu\n'), ((10519, 10575), 'numpy.require', 'nu.require', (['t'], {'dtype': 'nu.float64', 'requirements': "['C', 'W']"}), "(t, dtype=nu.float64, requirements=['C', 'W'])\n", (10529, 10575), True, 'import numpy as nu\n'), ((10585, 10646), 'numpy.require', 'nu.require', (['result'], {'dtype': 'nu.float64', 'requirements': "['C', 'W']"}), "(result, dtype=nu.float64, requirements=['C', 'W'])\n", (10595, 10646), True, 'import numpy as nu\n'), ((848, 970), 'warnings.warn', 'warnings.warn', (['("integrateFullOrbit_c extension module not loaded, because of error \'%s\' " %\n outerr)', 'galpyWarning'], {}), '(\n "integrateFullOrbit_c extension module not loaded, because of error \'%s\' "\n % outerr, galpyWarning)\n', (861, 970), False, 'import warnings\n'), ((1001, 1153), 'warnings.warn', 'warnings.warn', (["('integrateFullOrbit_c extension module not loaded, because galpy_integrate_c%s image was not found'\n % _ext_suffix)", 'galpyWarning'], {}), "(\n 'integrateFullOrbit_c extension module not loaded, because galpy_integrate_c%s image was not found'\n % _ext_suffix, galpyWarning)\n", (1014, 1153), False, 'import warnings\n'), ((6717, 6764), 'numpy.ctypeslib.ndpointer', 'ndpointer', ([], {'dtype': 'nu.float64', 'flags': 'ndarrayFlags'}), '(dtype=nu.float64, flags=ndarrayFlags)\n', (6726, 6764), False, 'from numpy.ctypeslib import ndpointer\n'), ((6870, 6917), 'numpy.ctypeslib.ndpointer', 'ndpointer', ([], {'dtype': 'nu.float64', 'flags': 'ndarrayFlags'}), '(dtype=nu.float64, flags=ndarrayFlags)\n', (6879, 6917), False, 'from numpy.ctypeslib import ndpointer\n'), ((6994, 7039), 'numpy.ctypeslib.ndpointer', 'ndpointer', ([], {'dtype': 'nu.int32', 'flags': 'ndarrayFlags'}), '(dtype=nu.int32, flags=ndarrayFlags)\n', (7003, 7039), False, 'from numpy.ctypeslib import ndpointer\n'), ((7071, 7118), 'numpy.ctypeslib.ndpointer', 'ndpointer', ([], {'dtype': 'nu.float64', 'flags': 'ndarrayFlags'}), '(dtype=nu.float64, flags=ndarrayFlags)\n', (7080, 7118), False, 'from numpy.ctypeslib import ndpointer\n'), ((7294, 7341), 'numpy.ctypeslib.ndpointer', 'ndpointer', ([], {'dtype': 'nu.float64', 'flags': 'ndarrayFlags'}), '(dtype=nu.float64, flags=ndarrayFlags)\n', (7303, 7341), False, 'from numpy.ctypeslib import ndpointer\n'), ((7373, 7401), 'ctypes.POINTER', 'ctypes.POINTER', (['ctypes.c_int'], {}), '(ctypes.c_int)\n', (7387, 7401), False, 'import ctypes\n'), ((7898, 7916), 'ctypes.c_int', 'ctypes.c_int', (['npot'], {}), '(npot)\n', (7910, 7916), False, 'import ctypes\n'), ((7998, 8017), 'ctypes.c_double', 'ctypes.c_double', (['dt'], {}), '(dt)\n', (8013, 8017), False, 'import ctypes\n'), ((8039, 8060), 'ctypes.c_double', 'ctypes.c_double', (['rtol'], {}), '(rtol)\n', (8054, 8060), False, 'import ctypes\n'), ((8061, 8082), 'ctypes.c_double', 'ctypes.c_double', (['atol'], {}), '(atol)\n', (8076, 8082), False, 'import ctypes\n'), ((8132, 8149), 'ctypes.byref', 'ctypes.byref', (['err'], {}), '(err)\n', (8144, 8149), False, 'import ctypes\n'), ((8171, 8197), 'ctypes.c_int', 'ctypes.c_int', (['int_method_c'], {}), '(int_method_c)\n', (8183, 8197), False, 'import ctypes\n'), ((8246, 8267), 'numpy.asfortranarray', 'nu.asfortranarray', (['yo'], {}), '(yo)\n', (8263, 8267), True, 'import numpy as nu\n'), ((8289, 8309), 'numpy.asfortranarray', 'nu.asfortranarray', (['t'], {}), '(t)\n', (8306, 8309), True, 'import numpy as nu\n'), ((9641, 9688), 'numpy.ctypeslib.ndpointer', 'ndpointer', ([], {'dtype': 'nu.float64', 'flags': 'ndarrayFlags'}), '(dtype=nu.float64, flags=ndarrayFlags)\n', (9650, 9688), False, 'from numpy.ctypeslib import ndpointer\n'), ((9794, 9841), 'numpy.ctypeslib.ndpointer', 'ndpointer', ([], {'dtype': 'nu.float64', 'flags': 'ndarrayFlags'}), '(dtype=nu.float64, flags=ndarrayFlags)\n', (9803, 9841), False, 'from numpy.ctypeslib import ndpointer\n'), ((9918, 9963), 'numpy.ctypeslib.ndpointer', 'ndpointer', ([], {'dtype': 'nu.int32', 'flags': 'ndarrayFlags'}), '(dtype=nu.int32, flags=ndarrayFlags)\n', (9927, 9963), False, 'from numpy.ctypeslib import ndpointer\n'), ((9995, 10042), 'numpy.ctypeslib.ndpointer', 'ndpointer', ([], {'dtype': 'nu.float64', 'flags': 'ndarrayFlags'}), '(dtype=nu.float64, flags=ndarrayFlags)\n', (10004, 10042), False, 'from numpy.ctypeslib import ndpointer\n'), ((10170, 10217), 'numpy.ctypeslib.ndpointer', 'ndpointer', ([], {'dtype': 'nu.float64', 'flags': 'ndarrayFlags'}), '(dtype=nu.float64, flags=ndarrayFlags)\n', (10179, 10217), False, 'from numpy.ctypeslib import ndpointer\n'), ((10249, 10277), 'ctypes.POINTER', 'ctypes.POINTER', (['ctypes.c_int'], {}), '(ctypes.c_int)\n', (10263, 10277), False, 'import ctypes\n'), ((10774, 10792), 'ctypes.c_int', 'ctypes.c_int', (['npot'], {}), '(npot)\n', (10786, 10792), False, 'import ctypes\n'), ((10874, 10895), 'ctypes.c_double', 'ctypes.c_double', (['rtol'], {}), '(rtol)\n', (10889, 10895), False, 'import ctypes\n'), ((10896, 10917), 'ctypes.c_double', 'ctypes.c_double', (['atol'], {}), '(atol)\n', (10911, 10917), False, 'import ctypes\n'), ((10967, 10984), 'ctypes.byref', 'ctypes.byref', (['err'], {}), '(err)\n', (10979, 10984), False, 'import ctypes\n'), ((11006, 11032), 'ctypes.c_int', 'ctypes.c_int', (['int_method_c'], {}), '(int_method_c)\n', (11018, 11032), False, 'import ctypes\n'), ((11081, 11102), 'numpy.asfortranarray', 'nu.asfortranarray', (['yo'], {}), '(yo)\n', (11098, 11102), True, 'import numpy as nu\n'), ((11124, 11144), 'numpy.asfortranarray', 'nu.asfortranarray', (['t'], {}), '(t)\n', (11141, 11144), True, 'import numpy as nu\n'), ((530, 585), 'os.path.join', 'os.path.join', (['path', "('galpy_integrate_c%s' % _ext_suffix)"], {}), "(path, 'galpy_integrate_c%s' % _ext_suffix)\n", (542, 585), False, 'import os\n'), ((637, 692), 'os.path.join', 'os.path.join', (['path', "('galpy_integrate_c%s' % _ext_suffix)"], {}), "(path, 'galpy_integrate_c%s' % _ext_suffix)\n", (649, 692), False, 'import os\n')] |
import numpy as np
import pandas as pd
from sklearn.preprocessing import LabelBinarizer
from sklearn.model_selection import StratifiedKFold
from sklearn.preprocessing import StandardScaler
from joblib import parallel_backend
from multiprocessing import cpu_count
import os, gc, joblib
from tqdm import tqdm
from collections import defaultdict
import torch
import warnings
warnings.filterwarnings('ignore')
pd.set_option("display.max_colwidth", 100)
pd.set_option("display.max_rows", 20)
osj = os.path.join; osl = os.listdir
n_cpus = cpu_count()
class ViralDataset(torch.utils.data.Dataset):
def __init__(self, df: pd.DataFrame, feat_cols: list, mode: str):
self.X = df[feat_cols].values # [:,np.newaxis,:]
self.mode = mode
if mode != 'test':
self.targets = df['virality'].values # [:,np.newaxis] # - 1
# assert np.sum(~df['virality'].isin(list(range(5))))==0
def __len__(self):
return len(self.X)
def __getitem__(self, idx):
if self.mode=='test':
return torch.tensor(self.X[idx], dtype=torch.float32)
else:
return (torch.tensor(self.X[idx], dtype=torch.float32),
torch.tensor(self.targets[idx], dtype=torch.long)) # long))
class ExtractFeatsDataset(torch.utils.data.Dataset):
def __init__(self, df: pd.DataFrame, feat_cols: list, target_cols: list, mode: str):
self.X = df[feat_cols].values # [:,np.newaxis,:]
# self.target_cols = target_cols
self.mode = mode
if mode != 'test':
if len(target_cols)==1:
self.targets = df[target_cols[0]].values # [:,np.newaxis] # - 1
self.target_dtype = torch.long
else:
self.targets = df[target_cols].values # [:,np.newaxis] # - 1
self.target_dtype = torch.float32
# assert np.sum(~df['virality'].isin(list(range(5))))==0
def __len__(self):
return len(self.X)
def __getitem__(self, idx):
if self.mode=='test':
return torch.tensor(self.X[idx], dtype=torch.float32)
else:
return (torch.tensor(self.X[idx], dtype=torch.float32),
torch.tensor(self.targets[idx], dtype=self.target_dtype)) # long))
def to_binary_categories(df, cat_col='tweet_language_id'):
df.loc[:, cat_col] = (df[cat_col]!=0).astype(np.int8)
return df
def freq_encoding(df, freq_cols: list, main_col='tweet_id'):
for c in freq_cols:
count_df = df.groupby([c])[main_col].count().reset_index()
count_df.columns = [c, '{}_freq'.format(c)]
df = df.merge(count_df, how='left', on=c)
return df
def bin_feats(df, feats=[], n_bins_default=20):
bin_counts = defaultdict(lambda: n_bins_default)
bin_counts['user_tweet_count'] = 20
for feature in feats:
if '_binned' in feature:
continue
n_bins = bin_counts[feature]
if n_bins:
bins = np.unique(df[feature].quantile(np.linspace(0, 1, n_bins)).values)
df[feature + '_binned'] = pd.cut(
df[feature], bins=bins, duplicates='drop'
).cat.codes
return df
def to_categorical(df):
cat_cols = ['tweet_has_attachment', 'user_has_location', 'user_has_url', 'user_verified', ]
df[cat_cols] = df[cat_cols].astype('category')
return df
def change2float32(df):
float_cols = df.select_dtypes('float64').columns
df[float_cols] = df[float_cols].astype(np.float32)
return df
def merge_df2media(df, df_media):
num_media = (df_media.groupby('tweet_id')['media_id']
.nunique()
.reset_index())
df_media.drop('media_id', axis=1, inplace=True)
num_media.columns = ['tweet_id', 'num_media']
df_media = df_media.merge(num_media, how='left', on='tweet_id')
media_cols = [col for col in df_media if col not in ['tweet_id','media_id']]
df_media = df_media.groupby('tweet_id')[media_cols].mean().reset_index()
# df_media = mean_feats.merge(df_media[['tweet_id']], how='left', on='tweet_id')
# del mean_feats; _ = gc.collect()
df_media['tweet_has_media'] = True
df = df.merge(df_media, how='left', on='tweet_id')
# fillna False if tweet has no media
df['tweet_has_media'] = df['tweet_has_media'].fillna(False)
# the same for the count of number of media per tweet
df['num_media'] = df['num_media'].fillna(0).astype(np.int8)
return df
# def add_num_media_user(df):
# # todo when not debug: df['num_media'].equals(df['num_media_user'])
# num_media_user = df.groupby('tweet_id')['num_media'].sum().reset_index()
# num_media_user.columns = ['tweet_id','num_media_user']
# df = df.merge(num_media_user, how='left', on='tweet_id')
# df['num_media_user'] = df['num_media_user'].astype(np.int8)
# return df
def tweets_user_created_date(df):
for feat_ in ['tweet_created_at_year', 'tweet_created_at_month', 'tweet_created_at_day',
'tweet_created_at_hour']:
# counts_df_cols = ['tweet_user_id']+[f"tweets_in_{feat_.split('_')[-1]}_{time_}" for time_ in np.sort(df[feat_].unique())]
# tweet_user_ids = np.sort(df['tweet_user_id'].unique())
# counts_df = pd.DataFrame(index=range(tweet_user_ids), columns=counts_df_cols)
# counts_df['tweet_user_id'] = tweet_user_ids
counts_map = df.groupby('tweet_user_id')[feat_].apply(lambda x: x.value_counts())
counts_map = counts_map.unstack(level=1)
counts_map.columns = [f"tweets_in_{feat_.split('_')[-1]}_"+str(col) for col in counts_map.columns]
counts_map = counts_map.fillna(0).reset_index()
df = df.merge(counts_map, how='left', on='tweet_user_id')
return df
# n_tweets_time_user = df.groupby('tweet_user_id')[feat_].count().reset_index()
# n_tweets_time_user.columns = ['tweet_user_id', f"n_tweets_{feat_.split('_')[-1]}_user_count"]
# df = df.merge(n_tweets_time_user, how='left', on='tweet_user_id')
def create_date_col(df):
tweet_date_cols = ['tweet_created_at_year', 'tweet_created_at_month', 'tweet_created_at_day']
df['date'] = df[tweet_date_cols].apply(lambda x:
str(x['tweet_created_at_month']).strip() + '/' +
str(x['tweet_created_at_day']).strip() + '/' +
str(x['tweet_created_at_year']).strip(), axis=1)
df['date'] = pd.to_datetime(df['date'])
return df
def add_sincos(df):
hour_sine = np.sin(2 * np.pi * df['tweet_created_at_hour'] / 24.0)
hour_sine.name = 'sin_hour'
hour_cosine = np.cos(2 * np.pi * df['tweet_created_at_hour'] / 24.0)
hour_cosine.name = 'cos_hour'
df = df.join([hour_sine, hour_cosine])
return df
def add_dummy_dates(df):
year = pd.get_dummies(df.tweet_created_at_year, prefix='ohe_year')
month = pd.get_dummies(df.tweet_created_at_month, prefix='ohe_month')
day = pd.get_dummies(df.tweet_created_at_day, prefix='ohe_day')
user_year = pd.get_dummies(df.user_created_at_year, prefix='ohe_user_year')
user_month = pd.get_dummies(df.user_created_at_month, prefix='ohe_user_month')
df = df.join([year, month, day, user_year, user_month])
return df
def add_date_feats(df):
# todo OHE date
# todo to sin, cos(date)
#df_old_index = df.index
df = create_date_col(df)
df = add_sincos(df)
df = add_dummy_dates(df)
cols_resample = ['tweet_hashtag_count', 'tweet_url_count', 'tweet_mention_count',
]
date_freqs = ['1Q'] # ,'1M']
# todo DON't use _func_min if does not affect CV (low feat importance)
stats = ['sum','mean','std','max'] # ['mean', 'max', 'min', 'median', 'std']
for freq_ in date_freqs:
for stat_ in stats:
df.set_index('date', inplace=True)
g = (df.groupby('tweet_user_id').resample(freq_, closed='left')
[cols_resample].agg(stat_)
.astype(np.float32)
) # .set_index('date'))
g = g.unstack('date').fillna(0)
g.columns = [col1 + f'_func_{stat_}_' + col2.strftime('%Y-%m-%d') for (col1, col2) in g.columns]
g.reset_index(inplace=True)
# g = g.rename(columns ={col: f"{col}_rsmpl_{freq_}_func_{stat_}"
# for col in g.columns if col not in ['tweet_user_id','date']})
#df = df.reset_index().merge(g, how='left', on='tweet_user_id')
df = df.reset_index().merge(g, how='left', on='tweet_user_id')
# df.reset_index(drop=False, inplace=True)
# todo count 'tweet_id' for each period for user
today = pd.to_datetime('7/1/2021')
df['days_since_tweet'] = (today - df['date']).dt.days # .astype(int)
df['user_followers_count_2days'] = df['user_followers_count'] / df['days_since_tweet']
df['user_following_count_2days'] = df['user_following_count'] / df['days_since_tweet']
df['user_listed_on_count_2days'] = df['user_listed_on_count'] / df['days_since_tweet']
df['user_tweet_count_2days'] = df['user_tweet_count'] / df['days_since_tweet']
df['tweet_hashtag_count_2days'] = df['tweet_hashtag_count'] / df['days_since_tweet']
df['tweet_mention_count_2days'] = df['tweet_mention_count'] / df['days_since_tweet']
df['tweet_url_count_2days'] = df['tweet_url_count'] / df['days_since_tweet']
# todo not a date related functions:
df['tweet_mention_count_div_followers'] = df['tweet_mention_count'].divide(df['user_followers_count']+1)
df['tweet_url_count_div_followers'] = df['tweet_url_count'].divide(df['user_followers_count']+1)
df['tweet_hashtag_count_div_followers'] = df['tweet_hashtag_count'].divide(df['user_followers_count']+1)
df['tweet_mention_count_div_followers'] = df['tweet_mention_count'].divide(df['user_followers_count']+1)
df['tweet_mention_count_div_n_tweets'] = df['tweet_mention_count'].divide(df['user_tweet_count']+1)
df['tweet_url_count_div_n_tweets'] = df['tweet_url_count'].divide(df['user_tweet_count']+1)
df['tweet_hashtag_count_div_n_tweets'] = df['tweet_hashtag_count'].divide(df['user_tweet_count']+1)
df['tweet_mention_count_div_n_tweets'] = df['tweet_mention_count'].divide(df['user_tweet_count']+1)
df['tweet_mention_count_div_likes'] = df['tweet_mention_count'].divide(df['user_like_count']+1)
df['tweet_url_count_div_likes'] = df['tweet_url_count'].divide(df['user_like_count']+1)
df['tweet_hashtag_count_div_likes'] = df['tweet_hashtag_count'].divide(df['user_like_count']+1)
df['tweet_mention_count_div_likes'] = df['tweet_mention_count'].divide(df['user_like_count']+1)
cols_drop = ['date', 'tweet_created_at_year', 'tweet_created_at_month',
'tweet_created_at_day',
'user_created_at_year', 'user_created_at_month']
df.drop(cols_drop, axis=1, inplace=True)
return df
def ohe_func(df, cat_col, ohe_tfm=LabelBinarizer(), prefix=None):
""" OHE one categorical column of df, and return df with columns 'label_{range(1,x}' added
"""
# ohe.iloc[:, df['tweet_language_id'].tolist()]
ohe_tfm.fit(df[cat_col])
ohe_transformed = ohe_tfm.transform(df[cat_col])
if prefix:
cat_cols = [f'{prefix}_{cat_col}_{i}' for i in range(ohe_transformed.shape[1])]
else:
cat_cols = [f'{cat_col}_{i}' for i in range(ohe_transformed.shape[1])]
ohe_df = pd.DataFrame(ohe_transformed, index=df.index, columns=cat_cols)
df = pd.concat([df, ohe_df], axis=1)
df.drop(cat_col, axis=1, inplace=True)
return df
def drop_unnecessary_cols(cfg, df):
cols_drop = [] # 'tweet_created_at_year', 'tweet_created_at_month',
# 'tweet_created_at_day']
# 'days_since_user', 'user_created_at_year', 'user_created_at_month',
# 'user_verified', 'user_has_url']
if cfg.drop_rare_ohe_language_ids and cfg.one_hot_encode:
lang_leave_ids = [0, 1, 3]
cols_drop += [f'tweet_language_id_{i}' for i in range(31)
if i not in lang_leave_ids
]
for col in cols_drop:
if col in df.columns:
df.drop(col, axis=1, inplace=True)
# print(f"Dropped col: {col}")
return df
class Features():
def __init__(self,):
self.transformers = {}
self.impute_img_feature_nulls = -1
self.media_img_feat_cols = []
self.text_feat_cols = []
self.user_des_feat_cols = []
self.user_img_feat_cols = []
# union of topic ids in train and test , 0 - nan value, min=36, max=172
# xor train, test = [ 38, 117, 123, 165]
# in test but not in train = [ 38, 117, 123]
self.unique_topic_ids = [ 0, 36, 37, 38, 39, 43, 44, 45, 52, 58, 59, 60, 61,
63, 68, 71, 72, 73, 78, 79, 80, 81, 82, 87, 88, 89,
91, 93, 98, 99, 100, 101, 104, 111, 112, 117, 118, 119, 120,
121, 122, 123, 125, 126, 127, 147, 148, 149, 150, 151, 152, 153,
155, 156, 163, 165, 169, 170, 171, 172]
self.cols2int8 = ['fold', 'user_created_at_month', 'tweet_created_at_day', 'tweet_created_at_hour',
'tweet_hashtag_count', 'tweet_url_count', 'tweet_mention_count', 'tweet_has_attachment',
'virality', 'tweet_has_media', 'user_has_url', 'user_verified', 'num_media',
'user_id', 'tweet_user_id']
# 'tweet_created_at_year', 'user_created_at_year',
self.cols2int8 += [f'tweet_language_id_{i}' for i in range(30)]
def get_data_stage1(self, cfg, base_dir, n_samples=int(1e10)):
df = pd.read_csv(osj(base_dir, 'Tweets',f'train_tweets.csv'), nrows=n_samples)
test = pd.read_csv(osj(base_dir, 'Tweets',f'test_tweets.csv'), nrows=n_samples)
# test_tweet_ids = test['tweet_id'].to_list()
# self.tabular_feats.append()
df = pd.concat([df, test])
del test; _ = gc.collect()
df = change2float32(df)
df = self.optimize_ints(df)
#df.drop('tweet_attachment_class', axis=1, inplace=True)
# try using 'media_id' columns
df_media = pd.read_csv(osj(base_dir, 'Tweets',f'train_tweets_vectorized_media.csv'))
df_media_test = pd.read_csv(osj(base_dir, 'Tweets',f'test_tweets_vectorized_media.csv'))
df_media = pd.concat([df_media, df_media_test])
df_media = change2float32(df_media)
df = merge_df2media(df, df_media)
del df_media, df_media_test; _ = gc.collect()
df_text = pd.read_csv(osj(base_dir, 'Tweets',f'train_tweets_vectorized_text.csv'))
df_text_test = pd.read_csv(osj(base_dir, 'Tweets',f'test_tweets_vectorized_text.csv'))
df_text = pd.concat([df_text, df_text_test])
text_feat_cols = ['text_'+ col for col in df_text.columns if col.startswith('feature_')]
df_text.columns = ['tweet_id'] + text_feat_cols
df_text.loc[:, text_feat_cols] = np.log(df_text[text_feat_cols] + 13)
df_text = change2float32(df_text)
df = df.merge(df_text, how='left', on='tweet_id')
del df_text, df_text_test; _ = gc.collect()
users = pd.read_csv(osj(base_dir, 'Users','users.csv'))
# log of _count feats
users_des = pd.read_csv(osj(base_dir, 'Users','user_vectorized_descriptions.csv'))
# for col in ['tweet_hashtag_count','tweet_url_count','tweet_mention_count']:
# users[col] = users[col].astype(int)
users_img = pd.read_csv(osj(base_dir, 'Users','user_vectorized_profile_images.csv'))
user_des_feat_cols = ['user_des_'+col for col in users_des.columns if col.startswith('feature')]
users_des.columns = ['user_id'] + user_des_feat_cols
user_img_feat_cols = ['user_img_'+col for col in users_img.columns if col.startswith('feature')]
users_img.columns = ['user_id'] + user_img_feat_cols
# user_data = users # .merge(users, how='left', on='user_id')
user_data = users.merge(users_des, how='left', on='user_id')
user_data = user_data.merge(users_img, how='left', on='user_id')
user_data = change2float32(user_data)
user_data = self.optimize_ints(user_data) # # no nulls in user_data 25-may
df = df.merge(user_data, how='left', left_on='tweet_user_id', right_on='user_id')
df.drop('user_id', axis=1, inplace=True)
df = cond_drop_imgtext(cfg, df)
# df = add_num_media_user(df)
del users_des, users_img, user_data;
_ = gc.collect()
return df # , test_tweet_ids
def get_data_stage2(self, cfg, df):
df = tweets_user_created_date(df) # add feats: number of user tweets in time period (year, month, day, hour)
df = add_date_feats(df)
df = bin_feats(df, feats=['tweet_mention_count','user_tweet_count',
'user_followers_count','user_following_count',
'user_listed_on_count'])
df = add_topic_count(df)
df = add_topic_ids(df)
bool_cols = df.select_dtypes(include='bool').columns
df[bool_cols] = df[bool_cols].astype(np.int8)
if cfg.one_hot_encode:
df = ohe_func(df, cat_col='tweet_language_id', ohe_tfm=LabelBinarizer())
df = ohe_func(df, cat_col='tweet_attachment_class', ohe_tfm=LabelBinarizer())
else:
df['tweet_attachment_class'] = df['tweet_attachment_class'].astype('category').cat.codes
# df = to_binary_categories(df, cat_col='tweet_language_id')
media_img_feat_cols = [col for col in df.columns if col.startswith('img_feature_')]
if cfg.impute_nulls:
df.loc[:,media_img_feat_cols] = df[media_img_feat_cols].fillna(self.impute_img_feature_nulls)
if cfg.add_user_virality:
df = self.add_virality_feature(df)
df = freq_encoding(df, freq_cols=['tweet_user_id'], main_col='tweet_id')
df = drop_unnecessary_cols(cfg, df)
# log (feats) :
cols2log = ['user_like_count','user_followers_count',
'user_following_count', 'user_listed_on_count',
'user_tweet_count']
# 'tweet_hashtag_count' , 'tweet_url_count', 'tweet_mention_count'
cols2log = [col for col in df.columns if col in cols2log]
df = logtransform(df, cols2log)
# print("df.shape after merging all csv files:", df.shape)
# print("df.dtypes.value_counts():\n", df.dtypes.value_counts())
# train = df[~df['tweet_id'].isin(test_tweet_ids)]
# test = df[df['tweet_id'].isin(test_tweet_ids)]
train = df[~df['virality'].isnull()]
test = df[df['virality'].isnull()]
del test['virality']; _ = gc.collect()
print(f"train.shape = {train.shape}, test.shape = {test.shape}")
return train, test
# end of def get_data
def add_virality_feature(self, df):
df_train = df[~df['virality'].isnull()]
viral_user = df_train.groupby('tweet_user_id')['virality'].mean().reset_index()
viral_user.columns = ['tweet_user_id', 'user_virality']
df = df.merge(viral_user, how='left', on='tweet_user_id')
return df
def optimize_ints(self, df):
int8_candidates = self.cols2int8
# for col in ['tweet_created_at_year', 'user_created_at_year']:
# if col in df.columns:
# df.loc[:, col] = df.loc[:, col] - 2000
# df.loc[:, col] = df.loc[:, col].astype(np.int8)
for col in int8_candidates:
if (col in df.columns) and (df[col].isnull().sum()==0):
df.loc[:, col] = df.loc[:, col].astype(np.int8)
return df
# end of class Features
def logtransform(df, cols2log):
df.loc[:, cols2log] = np.log(df[cols2log] + 2)
return df
class NormalizeFeats_Parallel():
""" https://scikit-learn.org/stable/computing/parallelism.html
from joblib import parallel_backend
with parallel_backend('threading', n_jobs=2):
# Your scikit-learn code here
"""
def __init__(self, feat_cols: list):
self.feat_cols = feat_cols
self.scalers_dict = {}
def normalize_data(self, df, mode='train', scaler=StandardScaler()):
if mode =='train':
for col in self.feat_cols:
with parallel_backend('threading', n_jobs=n_cpus):
scaler.fit(df[col].values.reshape(-1,1))
self.scalers_dict[col] = scaler
# scaler.fit(df[feat_cols].values)
df.loc[:,col] = self.scalers_dict[col].transform(df[col].values.reshape(-1,1))
else:
for col in self.feat_cols:
with parallel_backend('threading', n_jobs=n_cpus):
df.loc[:,col] = self.scalers_dict[col].transform(df[col].values.reshape(-1,1))
return df
# end of NormalizeFeats class
class NormalizeFeats():
def __init__(self, feat_cols: list):
self.feat_cols = feat_cols
self.scalers_dict = {}
def normalize_data(self, df, mode='train', scaler=StandardScaler()):
if mode =='train':
for col in self.feat_cols:
scaler.fit(df[col].values.reshape(-1,1))
self.scalers_dict[col] = scaler
# scaler.fit(df[feat_cols].values)
df.loc[:,col] = self.scalers_dict[col].transform(df[col].values.reshape(-1,1))
else:
for col in self.feat_cols:
df.loc[:,col] = self.scalers_dict[col].transform(df[col].values.reshape(-1,1))
return df
# end of NormalizeFeats class
def transform_joint(train, test=None, norm_cols=None, tfm = StandardScaler()):
# normalize joint train test data in chunks by columns
l_train = len(train)
if len(norm_cols) < 1000:
if isinstance(test, pd.DataFrame):
assert train[norm_cols].columns.equals(test[norm_cols].columns)
data = pd.concat([train[norm_cols], test[norm_cols]]).values
else:
data = train[norm_cols].values
with parallel_backend('threading', n_jobs=n_cpus):
tfm.fit(data)
data = tfm.transform(data)
train.loc[:, norm_cols] = data[:l_train]
if isinstance(test, pd.DataFrame):
test.loc[:, norm_cols] = data[l_train:]
else: # len(norm_cols) >= 1000
all_col_chunks = [norm_cols[i:i+1000] for i in range(0, len(norm_cols), 1000)]
for cols_chunk in all_col_chunks:
if isinstance(test, pd.DataFrame):
assert train[norm_cols].columns.equals(test[norm_cols].columns)
data_chunk = pd.concat([train[cols_chunk], test[cols_chunk]]).values
else:
data_chunk = train[cols_chunk]
scaler = StandardScaler()
with parallel_backend('threading', n_jobs=n_cpus):
tfm.fit(data_chunk)
data_chunk = tfm.transform(data_chunk)
train.loc[:, cols_chunk] = data_chunk[:l_train] # todo LONGEST RUNTIME and memory
if isinstance(test, pd.DataFrame):
test.loc[:, cols_chunk] = data_chunk[l_train:] # todo LONGEST RUNTIME and memory
return train, test # test cab be None
def normalize_npnan(train, test=None, norm_cols=[]):
if len(norm_cols)==0:
raise NotImplementedError
l_train = len(train)
if len(norm_cols) < 1000:
if isinstance(test, pd.DataFrame):
# assert train[norm_cols].columns.equals(test[norm_cols].columns)
data = pd.concat([train[norm_cols], test[norm_cols]]).values
else:
data = train[norm_cols].values
data = (data - np.nanmean(data, axis=0))/np.nanstd(data, axis=0)
train.loc[:, norm_cols] = data[:l_train]
if isinstance(test, pd.DataFrame):
test.loc[:, norm_cols] = data[l_train:]
else: # len(norm_cols) >= 1000
all_col_chunks = [norm_cols[i:i + 1000] for i in range(0, len(norm_cols), 1000)]
for cols_chunk in all_col_chunks:
if isinstance(test, pd.DataFrame):
# assert train[norm_cols].columns.equals(test[norm_cols].columns)
data_chunk = pd.concat([train[cols_chunk], test[cols_chunk]]).values
else:
data_chunk = train[cols_chunk]
data_chunk = (data_chunk - np.nanmean(data_chunk, axis=0))/np.nanstd(data_chunk, axis=0)
train.loc[:, cols_chunk] = data_chunk[:l_train] # todo LONGEST RUNTIME and memory
if isinstance(test, pd.DataFrame):
test.loc[:, cols_chunk] = data_chunk[l_train:] # todo LONGEST RUNTIME and memory
return train, test # test cab be None
def normalize_joint(train, test=None, norm_cols=None):
# normalize joint train test data in chunks by columns
l_train = len(train)
if len(norm_cols) < 1000:
if isinstance(test, pd.DataFrame):
assert train[norm_cols].columns.equals(test[norm_cols].columns)
data = pd.concat([train[norm_cols], test[norm_cols]]).values
else:
data = train[norm_cols].values
scaler = StandardScaler()
with parallel_backend('threading', n_jobs=n_cpus):
scaler.fit(data)
data = scaler.transform(data)
train.loc[:, norm_cols] = data[:l_train]
if isinstance(test, pd.DataFrame):
test.loc[:, norm_cols] = data[l_train:]
else: # len(norm_cols) >= 1000
all_col_chunks = [norm_cols[i:i+1000] for i in range(0, len(norm_cols), 1000)]
for cols_chunk in all_col_chunks:
if isinstance(test, pd.DataFrame):
assert train[norm_cols].columns.equals(test[norm_cols].columns)
data_chunk = pd.concat([train[cols_chunk], test[cols_chunk]]).values
else:
data_chunk = train[cols_chunk]
scaler = StandardScaler()
with parallel_backend('threading', n_jobs=n_cpus):
scaler.fit(data_chunk)
data_chunk = scaler.transform(data_chunk)
train.loc[:, cols_chunk] = data_chunk[:l_train] # todo LONGEST RUNTIME and memory
if isinstance(test, pd.DataFrame):
test.loc[:, cols_chunk] = data_chunk[l_train:] # todo LONGEST RUNTIME and memory
return train, test # test cab be None
def normalize_joint_parallel(train, test, norm_cols, num_workers=6):
# normalize joint train test data in chunks by columns
from joblib import Parallel, delayed
l_train = len(train)
assert train[norm_cols].columns.equals(test[norm_cols].columns)
if len(norm_cols) < 1000:
data = pd.concat([train[norm_cols], test[norm_cols]]).values
scaler = StandardScaler()
with parallel_backend('threading', n_jobs=n_cpus):
scaler.fit(data)
data = scaler.transform(data)
train.loc[:, norm_cols] = data[:l_train]
test.loc[:, norm_cols] = data[l_train:]
else: # len(norm_cols) >= 1000
all_col_chunks = [norm_cols[i:i+1000] for i in range(0, len(norm_cols), 1000)]
for cols_chunk in all_col_chunks:
data_chunk = pd.concat([train[cols_chunk], test[cols_chunk]]).values
scaler = StandardScaler()
with parallel_backend('threading', n_jobs=n_cpus):
scaler.fit(data_chunk)
data_chunk = scaler.transform(data_chunk)
train.loc[:, cols_chunk] = data_chunk[:l_train] # todo LONGEST RUNTIME and memory
test.loc[:, cols_chunk] = data_chunk[l_train:] # todo LONGEST RUNTIME and memory
return train, test
def split2folds_user_viral(df, n_folds, seed_folds, label_cols=None, foldnum_col='fold'):
# df is added foldnum_col='fold' column based on KFoldMethod = StratifiedKFold class
# applied to label_col='label' column
temp_col = label_cols[0] + "_" + label_cols[1]
df[temp_col] = df[label_cols[0]].astype(str) + "_" + df[label_cols[1]].astype(str)
df[temp_col] =df[temp_col].astype('category')
skf = StratifiedKFold(n_splits=n_folds, shuffle=True, random_state=seed_folds)
df[foldnum_col] = np.nan
for fold, (train_idx, val_idx) in enumerate(skf.split(np.zeros(df.shape[0]), df[temp_col])):
df.iloc[val_idx, df.columns.get_loc(foldnum_col)] = fold
df[foldnum_col] = df[foldnum_col].astype(int)
# assert df.isnull().sum().sum() == 0, "Error: null values in df"
del df[temp_col]
return df
def split2folds_viral_only(df, n_folds, seed_folds, label_col='label', foldnum_col='fold'):
# df is added foldnum_col='fold' column based on KFoldMethod = StratifiedKFold class
# applied to label_col='label' column
skf = StratifiedKFold(n_splits=n_folds, shuffle=True, random_state=seed_folds)
df[foldnum_col] = np.nan
for fold, (train_idx, val_idx) in enumerate(skf.split(df.values[:,:1], df[label_col])):
df.iloc[val_idx, df.columns.get_loc(foldnum_col)] = fold
df[foldnum_col] = df[foldnum_col].astype(int)
# assert df.isnull().sum().sum() == 0, "Error: null values in df"
return df
def split2folds_simple(df, n_folds, seed_folds, foldnum_col='fold'):
# df is added foldnum_col='fold' column based on KFoldMethod = StratifiedKFold class
# applied to label_col='label' column
skf = KFold(n_splits=n_folds, shuffle=True, random_state=seed_folds)
df[foldnum_col] = np.nan
for fold, (train_idx, val_idx) in enumerate(skf.split(df.values[:,:1])):
df.iloc[val_idx, df.columns.get_loc(foldnum_col)] = fold
df[foldnum_col] = df[foldnum_col].astype(int)
# assert df.isnull().sum().sum() == 0, "Error: null values in df"
return df
def get_folds(cfg, train, default_seed_folds=24):
# if cfg.seed_folds == default_seed_folds:
# folds = pd.read_csv(cfg.folds_split_path)
# if 'fold' in train.columns:
# del train['fold']
# train = train.merge(folds, how='left', on='tweet_id')
# train.dropna(0, subset=['fold'], inplace=True)
# else:
if cfg.folds_split_method == 'user_viral':
train = split2folds_user_viral(train, cfg.n_folds, cfg.seed_folds,
label_cols=['tweet_user_id', 'virality'], foldnum_col='fold')
return train
def get_feat_cols(train):
feat_cols = [col for col in train.columns if (col not in ['virality', 'tweet_id',
'fold','is_test'])
and not col.startswith('target_')]
media_img_feat_cols = [col for col in train.columns if col.startswith('img_feature')]
text_feat_cols = [col for col in train.columns if col.startswith('text_feature')]
user_des_feat_cols = [col for col in train.columns if col.startswith('user_des_feature')]
user_img_feat_cols = [col for col in train.columns if col.startswith('user_img_feature')]
feats_some = [col for col in feat_cols if not col in media_img_feat_cols +
text_feat_cols + user_img_feat_cols + user_des_feat_cols]
# print(f"Null values:\n{train[feat_cols].isnull().sum().sort_values(ascending=False).head(2)}")
return (feat_cols, media_img_feat_cols, text_feat_cols,
user_des_feat_cols, user_img_feat_cols, feats_some)
def cond_drop_imgtext(cfg, df):
(feat_cols, media_img_feat_cols, text_feat_cols,
user_des_feat_cols, user_img_feat_cols, feats_some) = get_feat_cols(df)
if cfg.drop_media_img_feats:
df.drop(media_img_feat_cols, axis=1, inplace=True)
if cfg.drop_text_feats:
df.drop(text_feat_cols, axis=1, inplace=True)
if cfg.drop_user_des_feats:
df.drop(user_des_feat_cols, axis=1, inplace=True)
if cfg.drop_user_img_feats:
df.drop(user_img_feat_cols, axis=1, inplace=True)
return df
def add_topic_count(df):
# and drop the column
nan_replace = '0'
topics = df['tweet_topic_ids'].fillna(f'[{nan_replace}]')
# topics_xnan = train['tweet_topic_ids'].dropna()
# fill_value = topicsx_xnan.apply(lambda x: len(eval(x))).mean()
# fill_value = topicx_xnan.apply(lambda x: len(eval(x))).median()
n_topics = topics.apply(lambda x: len(eval(x)))
n_topics_mean = n_topics.mean()
n_topics = np.where(topics == nan_replace, n_topics_mean, n_topics)
df['n_topics'] = n_topics.astype(int)
return df
def add_topic_ids(df):
df.fillna({'tweet_topic_ids': "['0']"}, inplace=True)
topic_ids = (
df['tweet_topic_ids'].str.strip('[]').str.split('\s*,\s*').explode()
.str.get_dummies().sum(level=0).add_prefix('topic_id_')
)
topic_ids.rename(columns=lambda x: x.replace("'", ""), inplace=True)
if 'tweet_topic_ids' in df.columns:
df.drop('tweet_topic_ids', 1)
df = df.join(topic_ids) # , how='left', on='tweet_id')
for col_ in topic_ids.columns:
if df[col_].max() > 1:
df[f"{col_}_hthan1_binary"] = (df[col_] > 0).astype(np.int8)
df.drop('tweet_topic_ids',1, inplace=True)
return df
#
# def replace_add_new_topic_ids(train, test):
# # add topic_id cols (57) with number of times the topic is in the sample
# # add _binary cols (45) where =1 if topic_id is more than once
# old_topic_id_cols = [col for col in train.columns if 'topic_id' in col]
# print(f"old_topic_id_cols: {old_topic_id_cols}")
# len_train = train.shape[0]
# train = pd.concat([train, test]).reset_index(drop=True)
# del test;
# _ = gc.collect()
# train.drop(old_topic_id_cols, axis=1, inplace=True)
# train, new_topic_id_cols = add_new_topic_ids(base_dir, train, df_name='train_test')
# # todo cols ['topic_id_117' 'topic_id_123' 'topic_id_38'] are not in new_topic_id_cols
# # done: only one sample==1 for each topic_id_117 topic_id_123 topic_id_38 [0 42274 42274 42274] [1 1 1 1]
# for col_ in new_topic_id_cols:
# if train[col_].max() > 1:
# train[f"{col_}_hthan1_binary"] = (train[col_] > 0).astype(np.int8)
# # train.drop(col_, axis=1, inplace=True)
# test = train.iloc[len_train:, :].reset_index(drop=True)
# train = train.iloc[:len_train, :]
# return train, test
def extract_feats_media_text(df):
# todo extact_feats_media_text
# Set the target as well as dependent variables from image data.
y = vectorized_media_df['virality']
x = vectorized_media_df.loc[:, vectorized_media_df.columns.str.contains("img_")]
# Run Lasso regression for feature selection.
sel_model = SelectFromModel(LogisticRegression(C=1, penalty='l1', solver='liblinear'))
# time the model fitting
start = timeit.default_timer()
# Fit the trained model on our data
sel_model.fit(x, y)
stop = timeit.default_timer()
print('Time: ', stop - start)
# get index of good features
sel_index = sel_model.get_support()
# count the no of columns selected
counter = collections.Counter(sel_model.get_support())
print(counter)
def save_preprocessed(cfg, train, test, path_train, path_test):
# p_train = cfg.train_preprocessed_path
# p_test = cfg.test_preprocessed_path
# if cfg.debug:
# path_train = osj(os.path.dirname(p_train), 'debug_' + os.path.basename(path_train))
# path_test = osj(os.path.dirname(p_test), 'debug_' + os.path.basename(path_test))
assert not os.path.isfile(path_train), f"WON'T OVERWRITE/SAVE: file exists {os.path.basename(path_train)}"
assert not os.path.isfile(path_test), f"WON'T OVERWRITE/SAVE: file exists {os.path.basename(path_test)}"
train.to_csv(path_train, index=False)
test.to_csv(path_test, index=False)
def get_raw_train_tweet_cols(df):
# get cols from train_tweets.csv and users.csv
init_tweets_cols = ['tweet_id', 'tweet_user_id', 'tweet_created_at_year',
'tweet_created_at_month', 'tweet_created_at_day',
'tweet_created_at_hour', 'tweet_hashtag_count', 'tweet_url_count',
'tweet_mention_count', 'tweet_has_attachment', 'tweet_attachment_class',
'tweet_language_id', 'tweet_topic_ids', 'virality']
init_users_cols = ['user_id', 'user_like_count', 'user_followers_count',
'user_following_count', 'user_listed_on_count', 'user_has_location',
'user_tweet_count', 'user_has_url', 'user_verified',
'user_created_at_year', 'user_created_at_month']
def add_new_topic_ids(base_dir, df, df_name='train'):
if df_name=='train_test':
df_tweets = pd.read_csv(osj(base_dir, 'Tweets', f'train_tweets.csv'),
usecols=['tweet_id', 'tweet_topic_ids']
)
df_tweets_test = pd.read_csv(osj(base_dir, 'Tweets', f'test_tweets.csv'),
usecols=['tweet_id', 'tweet_topic_ids']
)
df_tweets = pd.concat([df_tweets, df_tweets_test]).reset_index(drop=True)
# df_tweets = df.reindex(df.index)
else:
df_tweets = pd.read_csv(osj(base_dir, 'Tweets', f'{df_name}_tweets.csv'),
usecols=['tweet_id', 'tweet_topic_ids']
)
df_tweets.fillna({'tweet_topic_ids': "['0']"}, inplace=True)
topic_ids = (
df_tweets['tweet_topic_ids'].str.strip('[]').str.split('\s*,\s*').explode()
.str.get_dummies().sum(level=0).add_prefix('topic_id_')
)
topic_ids.rename(columns=lambda x: x.replace("'", ""), inplace=True)
topic_ids['tweet_id'] = df_tweets['tweet_id']
if 'tweet_topic_ids' in df.columns:
df.drop('tweet_topic_ids')
df = df.merge(topic_ids, how='left', on='tweet_id')
return df, list(topic_ids.columns)
| [
"sklearn.preprocessing.LabelBinarizer",
"sklearn.preprocessing.StandardScaler",
"collections.defaultdict",
"gc.collect",
"os.path.isfile",
"numpy.sin",
"joblib.parallel_backend",
"pandas.set_option",
"multiprocessing.cpu_count",
"pandas.DataFrame",
"numpy.nanmean",
"numpy.linspace",
"pandas.... | [((375, 408), 'warnings.filterwarnings', 'warnings.filterwarnings', (['"""ignore"""'], {}), "('ignore')\n", (398, 408), False, 'import warnings\n'), ((409, 451), 'pandas.set_option', 'pd.set_option', (['"""display.max_colwidth"""', '(100)'], {}), "('display.max_colwidth', 100)\n", (422, 451), True, 'import pandas as pd\n'), ((452, 489), 'pandas.set_option', 'pd.set_option', (['"""display.max_rows"""', '(20)'], {}), "('display.max_rows', 20)\n", (465, 489), True, 'import pandas as pd\n'), ((537, 548), 'multiprocessing.cpu_count', 'cpu_count', ([], {}), '()\n', (546, 548), False, 'from multiprocessing import cpu_count\n'), ((2758, 2794), 'collections.defaultdict', 'defaultdict', (['(lambda : n_bins_default)'], {}), '(lambda : n_bins_default)\n', (2769, 2794), False, 'from collections import defaultdict\n'), ((6491, 6517), 'pandas.to_datetime', 'pd.to_datetime', (["df['date']"], {}), "(df['date'])\n", (6505, 6517), True, 'import pandas as pd\n'), ((6570, 6624), 'numpy.sin', 'np.sin', (["(2 * np.pi * df['tweet_created_at_hour'] / 24.0)"], {}), "(2 * np.pi * df['tweet_created_at_hour'] / 24.0)\n", (6576, 6624), True, 'import numpy as np\n'), ((6675, 6729), 'numpy.cos', 'np.cos', (["(2 * np.pi * df['tweet_created_at_hour'] / 24.0)"], {}), "(2 * np.pi * df['tweet_created_at_hour'] / 24.0)\n", (6681, 6729), True, 'import numpy as np\n'), ((6858, 6917), 'pandas.get_dummies', 'pd.get_dummies', (['df.tweet_created_at_year'], {'prefix': '"""ohe_year"""'}), "(df.tweet_created_at_year, prefix='ohe_year')\n", (6872, 6917), True, 'import pandas as pd\n'), ((6930, 6991), 'pandas.get_dummies', 'pd.get_dummies', (['df.tweet_created_at_month'], {'prefix': '"""ohe_month"""'}), "(df.tweet_created_at_month, prefix='ohe_month')\n", (6944, 6991), True, 'import pandas as pd\n'), ((7002, 7059), 'pandas.get_dummies', 'pd.get_dummies', (['df.tweet_created_at_day'], {'prefix': '"""ohe_day"""'}), "(df.tweet_created_at_day, prefix='ohe_day')\n", (7016, 7059), True, 'import pandas as pd\n'), ((7076, 7139), 'pandas.get_dummies', 'pd.get_dummies', (['df.user_created_at_year'], {'prefix': '"""ohe_user_year"""'}), "(df.user_created_at_year, prefix='ohe_user_year')\n", (7090, 7139), True, 'import pandas as pd\n'), ((7157, 7222), 'pandas.get_dummies', 'pd.get_dummies', (['df.user_created_at_month'], {'prefix': '"""ohe_user_month"""'}), "(df.user_created_at_month, prefix='ohe_user_month')\n", (7171, 7222), True, 'import pandas as pd\n'), ((8761, 8787), 'pandas.to_datetime', 'pd.to_datetime', (['"""7/1/2021"""'], {}), "('7/1/2021')\n", (8775, 8787), True, 'import pandas as pd\n'), ((11030, 11046), 'sklearn.preprocessing.LabelBinarizer', 'LabelBinarizer', ([], {}), '()\n', (11044, 11046), False, 'from sklearn.preprocessing import LabelBinarizer\n'), ((11505, 11568), 'pandas.DataFrame', 'pd.DataFrame', (['ohe_transformed'], {'index': 'df.index', 'columns': 'cat_cols'}), '(ohe_transformed, index=df.index, columns=cat_cols)\n', (11517, 11568), True, 'import pandas as pd\n'), ((11578, 11609), 'pandas.concat', 'pd.concat', (['[df, ohe_df]'], {'axis': '(1)'}), '([df, ohe_df], axis=1)\n', (11587, 11609), True, 'import pandas as pd\n'), ((19868, 19892), 'numpy.log', 'np.log', (['(df[cols2log] + 2)'], {}), '(df[cols2log] + 2)\n', (19874, 19892), True, 'import numpy as np\n'), ((21780, 21796), 'sklearn.preprocessing.StandardScaler', 'StandardScaler', ([], {}), '()\n', (21794, 21796), False, 'from sklearn.preprocessing import StandardScaler\n'), ((28169, 28241), 'sklearn.model_selection.StratifiedKFold', 'StratifiedKFold', ([], {'n_splits': 'n_folds', 'shuffle': '(True)', 'random_state': 'seed_folds'}), '(n_splits=n_folds, shuffle=True, random_state=seed_folds)\n', (28184, 28241), False, 'from sklearn.model_selection import StratifiedKFold\n'), ((28828, 28900), 'sklearn.model_selection.StratifiedKFold', 'StratifiedKFold', ([], {'n_splits': 'n_folds', 'shuffle': '(True)', 'random_state': 'seed_folds'}), '(n_splits=n_folds, shuffle=True, random_state=seed_folds)\n', (28843, 28900), False, 'from sklearn.model_selection import StratifiedKFold\n'), ((32400, 32456), 'numpy.where', 'np.where', (['(topics == nan_replace)', 'n_topics_mean', 'n_topics'], {}), '(topics == nan_replace, n_topics_mean, n_topics)\n', (32408, 32456), True, 'import numpy as np\n'), ((13985, 14006), 'pandas.concat', 'pd.concat', (['[df, test]'], {}), '([df, test])\n', (13994, 14006), True, 'import pandas as pd\n'), ((14029, 14041), 'gc.collect', 'gc.collect', ([], {}), '()\n', (14039, 14041), False, 'import os, gc, joblib\n'), ((14424, 14460), 'pandas.concat', 'pd.concat', (['[df_media, df_media_test]'], {}), '([df_media, df_media_test])\n', (14433, 14460), True, 'import pandas as pd\n'), ((14588, 14600), 'gc.collect', 'gc.collect', ([], {}), '()\n', (14598, 14600), False, 'import os, gc, joblib\n'), ((14806, 14840), 'pandas.concat', 'pd.concat', (['[df_text, df_text_test]'], {}), '([df_text, df_text_test])\n', (14815, 14840), True, 'import pandas as pd\n'), ((15035, 15071), 'numpy.log', 'np.log', (['(df_text[text_feat_cols] + 13)'], {}), '(df_text[text_feat_cols] + 13)\n', (15041, 15071), True, 'import numpy as np\n'), ((15211, 15223), 'gc.collect', 'gc.collect', ([], {}), '()\n', (15221, 15223), False, 'import os, gc, joblib\n'), ((16597, 16609), 'gc.collect', 'gc.collect', ([], {}), '()\n', (16607, 16609), False, 'import os, gc, joblib\n'), ((18821, 18833), 'gc.collect', 'gc.collect', ([], {}), '()\n', (18831, 18833), False, 'import os, gc, joblib\n'), ((20305, 20321), 'sklearn.preprocessing.StandardScaler', 'StandardScaler', ([], {}), '()\n', (20319, 20321), False, 'from sklearn.preprocessing import StandardScaler\n'), ((21182, 21198), 'sklearn.preprocessing.StandardScaler', 'StandardScaler', ([], {}), '()\n', (21196, 21198), False, 'from sklearn.preprocessing import StandardScaler\n'), ((25252, 25268), 'sklearn.preprocessing.StandardScaler', 'StandardScaler', ([], {}), '()\n', (25266, 25268), False, 'from sklearn.preprocessing import StandardScaler\n'), ((26843, 26859), 'sklearn.preprocessing.StandardScaler', 'StandardScaler', ([], {}), '()\n', (26857, 26859), False, 'from sklearn.preprocessing import StandardScaler\n'), ((35557, 35583), 'os.path.isfile', 'os.path.isfile', (['path_train'], {}), '(path_train)\n', (35571, 35583), False, 'import os, gc, joblib\n'), ((35668, 35693), 'os.path.isfile', 'os.path.isfile', (['path_test'], {}), '(path_test)\n', (35682, 35693), False, 'import os, gc, joblib\n'), ((1053, 1099), 'torch.tensor', 'torch.tensor', (['self.X[idx]'], {'dtype': 'torch.float32'}), '(self.X[idx], dtype=torch.float32)\n', (1065, 1099), False, 'import torch\n'), ((2072, 2118), 'torch.tensor', 'torch.tensor', (['self.X[idx]'], {'dtype': 'torch.float32'}), '(self.X[idx], dtype=torch.float32)\n', (2084, 2118), False, 'import torch\n'), ((22176, 22220), 'joblib.parallel_backend', 'parallel_backend', (['"""threading"""'], {'n_jobs': 'n_cpus'}), "('threading', n_jobs=n_cpus)\n", (22192, 22220), False, 'from joblib import parallel_backend\n'), ((22893, 22909), 'sklearn.preprocessing.StandardScaler', 'StandardScaler', ([], {}), '()\n', (22907, 22909), False, 'from sklearn.preprocessing import StandardScaler\n'), ((23815, 23838), 'numpy.nanstd', 'np.nanstd', (['data'], {'axis': '(0)'}), '(data, axis=0)\n', (23824, 23838), True, 'import numpy as np\n'), ((25282, 25326), 'joblib.parallel_backend', 'parallel_backend', (['"""threading"""'], {'n_jobs': 'n_cpus'}), "('threading', n_jobs=n_cpus)\n", (25298, 25326), False, 'from joblib import parallel_backend\n'), ((26005, 26021), 'sklearn.preprocessing.StandardScaler', 'StandardScaler', ([], {}), '()\n', (26019, 26021), False, 'from sklearn.preprocessing import StandardScaler\n'), ((26772, 26818), 'pandas.concat', 'pd.concat', (['[train[norm_cols], test[norm_cols]]'], {}), '([train[norm_cols], test[norm_cols]])\n', (26781, 26818), True, 'import pandas as pd\n'), ((26873, 26917), 'joblib.parallel_backend', 'parallel_backend', (['"""threading"""'], {'n_jobs': 'n_cpus'}), "('threading', n_jobs=n_cpus)\n", (26889, 26917), False, 'from joblib import parallel_backend\n'), ((27354, 27370), 'sklearn.preprocessing.StandardScaler', 'StandardScaler', ([], {}), '()\n', (27368, 27370), False, 'from sklearn.preprocessing import StandardScaler\n'), ((28330, 28351), 'numpy.zeros', 'np.zeros', (['df.shape[0]'], {}), '(df.shape[0])\n', (28338, 28351), True, 'import numpy as np\n'), ((35622, 35650), 'os.path.basename', 'os.path.basename', (['path_train'], {}), '(path_train)\n', (35638, 35650), False, 'import os, gc, joblib\n'), ((35732, 35759), 'os.path.basename', 'os.path.basename', (['path_test'], {}), '(path_test)\n', (35748, 35759), False, 'import os, gc, joblib\n'), ((1134, 1180), 'torch.tensor', 'torch.tensor', (['self.X[idx]'], {'dtype': 'torch.float32'}), '(self.X[idx], dtype=torch.float32)\n', (1146, 1180), False, 'import torch\n'), ((1202, 1251), 'torch.tensor', 'torch.tensor', (['self.targets[idx]'], {'dtype': 'torch.long'}), '(self.targets[idx], dtype=torch.long)\n', (1214, 1251), False, 'import torch\n'), ((2153, 2199), 'torch.tensor', 'torch.tensor', (['self.X[idx]'], {'dtype': 'torch.float32'}), '(self.X[idx], dtype=torch.float32)\n', (2165, 2199), False, 'import torch\n'), ((2221, 2277), 'torch.tensor', 'torch.tensor', (['self.targets[idx]'], {'dtype': 'self.target_dtype'}), '(self.targets[idx], dtype=self.target_dtype)\n', (2233, 2277), False, 'import torch\n'), ((22052, 22098), 'pandas.concat', 'pd.concat', (['[train[norm_cols], test[norm_cols]]'], {}), '([train[norm_cols], test[norm_cols]])\n', (22061, 22098), True, 'import pandas as pd\n'), ((22927, 22971), 'joblib.parallel_backend', 'parallel_backend', (['"""threading"""'], {'n_jobs': 'n_cpus'}), "('threading', n_jobs=n_cpus)\n", (22943, 22971), False, 'from joblib import parallel_backend\n'), ((23655, 23701), 'pandas.concat', 'pd.concat', (['[train[norm_cols], test[norm_cols]]'], {}), '([train[norm_cols], test[norm_cols]])\n', (23664, 23701), True, 'import pandas as pd\n'), ((23789, 23813), 'numpy.nanmean', 'np.nanmean', (['data'], {'axis': '(0)'}), '(data, axis=0)\n', (23799, 23813), True, 'import numpy as np\n'), ((24501, 24530), 'numpy.nanstd', 'np.nanstd', (['data_chunk'], {'axis': '(0)'}), '(data_chunk, axis=0)\n', (24510, 24530), True, 'import numpy as np\n'), ((25124, 25170), 'pandas.concat', 'pd.concat', (['[train[norm_cols], test[norm_cols]]'], {}), '([train[norm_cols], test[norm_cols]])\n', (25133, 25170), True, 'import pandas as pd\n'), ((26039, 26083), 'joblib.parallel_backend', 'parallel_backend', (['"""threading"""'], {'n_jobs': 'n_cpus'}), "('threading', n_jobs=n_cpus)\n", (26055, 26083), False, 'from joblib import parallel_backend\n'), ((27277, 27325), 'pandas.concat', 'pd.concat', (['[train[cols_chunk], test[cols_chunk]]'], {}), '([train[cols_chunk], test[cols_chunk]])\n', (27286, 27325), True, 'import pandas as pd\n'), ((27388, 27432), 'joblib.parallel_backend', 'parallel_backend', (['"""threading"""'], {'n_jobs': 'n_cpus'}), "('threading', n_jobs=n_cpus)\n", (27404, 27432), False, 'from joblib import parallel_backend\n'), ((37026, 37064), 'pandas.concat', 'pd.concat', (['[df_tweets, df_tweets_test]'], {}), '([df_tweets, df_tweets_test])\n', (37035, 37064), True, 'import pandas as pd\n'), ((3093, 3142), 'pandas.cut', 'pd.cut', (['df[feature]'], {'bins': 'bins', 'duplicates': '"""drop"""'}), "(df[feature], bins=bins, duplicates='drop')\n", (3099, 3142), True, 'import pandas as pd\n'), ((17336, 17352), 'sklearn.preprocessing.LabelBinarizer', 'LabelBinarizer', ([], {}), '()\n', (17350, 17352), False, 'from sklearn.preprocessing import LabelBinarizer\n'), ((17426, 17442), 'sklearn.preprocessing.LabelBinarizer', 'LabelBinarizer', ([], {}), '()\n', (17440, 17442), False, 'from sklearn.preprocessing import LabelBinarizer\n'), ((20411, 20455), 'joblib.parallel_backend', 'parallel_backend', (['"""threading"""'], {'n_jobs': 'n_cpus'}), "('threading', n_jobs=n_cpus)\n", (20427, 20455), False, 'from joblib import parallel_backend\n'), ((20798, 20842), 'joblib.parallel_backend', 'parallel_backend', (['"""threading"""'], {'n_jobs': 'n_cpus'}), "('threading', n_jobs=n_cpus)\n", (20814, 20842), False, 'from joblib import parallel_backend\n'), ((22751, 22799), 'pandas.concat', 'pd.concat', (['[train[cols_chunk], test[cols_chunk]]'], {}), '([train[cols_chunk], test[cols_chunk]])\n', (22760, 22799), True, 'import pandas as pd\n'), ((24308, 24356), 'pandas.concat', 'pd.concat', (['[train[cols_chunk], test[cols_chunk]]'], {}), '([train[cols_chunk], test[cols_chunk]])\n', (24317, 24356), True, 'import pandas as pd\n'), ((24469, 24499), 'numpy.nanmean', 'np.nanmean', (['data_chunk'], {'axis': '(0)'}), '(data_chunk, axis=0)\n', (24479, 24499), True, 'import numpy as np\n'), ((25863, 25911), 'pandas.concat', 'pd.concat', (['[train[cols_chunk], test[cols_chunk]]'], {}), '([train[cols_chunk], test[cols_chunk]])\n', (25872, 25911), True, 'import pandas as pd\n'), ((3020, 3045), 'numpy.linspace', 'np.linspace', (['(0)', '(1)', 'n_bins'], {}), '(0, 1, n_bins)\n', (3031, 3045), True, 'import numpy as np\n')] |
import tensorflow as tf
import numpy as np
def _tf_fspecial_gauss(size, sigma, ch=1):
"""Function to mimic the 'fspecial' gaussian MATLAB function
"""
x_data, y_data = np.mgrid[-size//2 + 1:size//2 + 1, -size//2 + 1:size//2 + 1]
x_data = np.expand_dims(x_data, axis=-1)
x_data = np.expand_dims(x_data, axis=-1)
y_data = np.expand_dims(y_data, axis=-1)
y_data = np.expand_dims(y_data, axis=-1)
x = tf.constant(x_data, dtype=tf.float32)
y = tf.constant(y_data, dtype=tf.float32)
g = tf.exp(-((x**2 + y**2)/(2.0*sigma**2)))
g = tf.tile(g, [1, 1, ch, 1])
return g / tf.reduce_sum(g)
def tf_ssim(img1, img2, cs_map=False, mean_metric=True, size=11, sigma=0.5):
img1 = tf.image.rgb_to_grayscale(img1)
img2 = tf.image.rgb_to_grayscale(img2)
window = _tf_fspecial_gauss(size, sigma,
ch=img1.get_shape().as_list()[-1]) # window shape [size, size]
K1 = 0.01
K2 = 0.03
L = 1 # depth of image (255 in case the image has a differnt scale)
C1 = (K1*L)**2
C2 = (K2*L)**2
mu1 = tf.nn.conv2d(img1, window, strides=[1, 1, 1, 1], padding='VALID')
mu2 = tf.nn.conv2d(img2, window, strides=[1, 1, 1, 1], padding='VALID')
mu1_sq = mu1*mu1
mu2_sq = mu2*mu2
mu1_mu2 = mu1*mu2
sigma1_sq = tf.nn.conv2d(img1*img1, window, strides=[1, 1, 1, 1],
padding='VALID') - mu1_sq
sigma2_sq = tf.nn.conv2d(img2*img2, window, strides=[1, 1, 1, 1],
padding='VALID') - mu2_sq
sigma12 = tf.nn.conv2d(img1*img2, window, strides=[1, 1, 1, 1],
padding='VALID') - mu1_mu2
if cs_map:
value = (
((2*mu1_mu2 + C1) * (2*sigma12 + C2)) / (
(mu1_sq + mu2_sq + C1) * (sigma1_sq + sigma2_sq + C2)
), (2.0*sigma12 + C2)/(sigma1_sq + sigma2_sq + C2)
)
else:
value = ((2*mu1_mu2 + C1)*(2*sigma12 + C2)) / (
(mu1_sq + mu2_sq + C1) * (sigma1_sq + sigma2_sq + C2))
if mean_metric:
value = tf.reduce_mean(value)
return value
def tf_ms_ssim(img1, img2, mean_metric=True, level=5):
weight = tf.constant([0.0448, 0.2856, 0.3001, 0.2363, 0.1333], dtype=tf.float32)
mssim = []
mcs = []
for l in range(level):
ssim_map, cs_map = tf_ssim(img1, img2, cs_map=True, mean_metric=False)
mssim.append(tf.reduce_mean(ssim_map))
mcs.append(tf.reduce_mean(cs_map))
filtered_im1 = tf.nn.avg_pool(img1, [1, 2, 2, 1], [1, 2, 2, 1], padding='SAME')
filtered_im2 = tf.nn.avg_pool(img2, [1, 2, 2, 1], [1, 2, 2, 1], padding='SAME')
img1 = filtered_im1
img2 = filtered_im2
# list to tensor of dim D+1
mssim = tf.pack(mssim, axis=0)
mcs = tf.pack(mcs, axis=0)
value = (tf.reduce_prod(
mcs[0:level-1]**weight[0:level-1]) * (mssim[level-1]**weight[level-1]))
if mean_metric:
value = tf.reduce_mean(value)
return value
| [
"tensorflow.image.rgb_to_grayscale",
"tensorflow.reduce_sum",
"tensorflow.reduce_mean",
"numpy.expand_dims",
"tensorflow.constant",
"tensorflow.tile",
"tensorflow.nn.avg_pool",
"tensorflow.exp",
"tensorflow.nn.conv2d",
"tensorflow.reduce_prod",
"tensorflow.pack"
] | [((257, 288), 'numpy.expand_dims', 'np.expand_dims', (['x_data'], {'axis': '(-1)'}), '(x_data, axis=-1)\n', (271, 288), True, 'import numpy as np\n'), ((302, 333), 'numpy.expand_dims', 'np.expand_dims', (['x_data'], {'axis': '(-1)'}), '(x_data, axis=-1)\n', (316, 333), True, 'import numpy as np\n'), ((348, 379), 'numpy.expand_dims', 'np.expand_dims', (['y_data'], {'axis': '(-1)'}), '(y_data, axis=-1)\n', (362, 379), True, 'import numpy as np\n'), ((393, 424), 'numpy.expand_dims', 'np.expand_dims', (['y_data'], {'axis': '(-1)'}), '(y_data, axis=-1)\n', (407, 424), True, 'import numpy as np\n'), ((434, 471), 'tensorflow.constant', 'tf.constant', (['x_data'], {'dtype': 'tf.float32'}), '(x_data, dtype=tf.float32)\n', (445, 471), True, 'import tensorflow as tf\n'), ((480, 517), 'tensorflow.constant', 'tf.constant', (['y_data'], {'dtype': 'tf.float32'}), '(y_data, dtype=tf.float32)\n', (491, 517), True, 'import tensorflow as tf\n'), ((527, 576), 'tensorflow.exp', 'tf.exp', (['(-((x ** 2 + y ** 2) / (2.0 * sigma ** 2)))'], {}), '(-((x ** 2 + y ** 2) / (2.0 * sigma ** 2)))\n', (533, 576), True, 'import tensorflow as tf\n'), ((575, 600), 'tensorflow.tile', 'tf.tile', (['g', '[1, 1, ch, 1]'], {}), '(g, [1, 1, ch, 1])\n', (582, 600), True, 'import tensorflow as tf\n'), ((724, 755), 'tensorflow.image.rgb_to_grayscale', 'tf.image.rgb_to_grayscale', (['img1'], {}), '(img1)\n', (749, 755), True, 'import tensorflow as tf\n'), ((767, 798), 'tensorflow.image.rgb_to_grayscale', 'tf.image.rgb_to_grayscale', (['img2'], {}), '(img2)\n', (792, 798), True, 'import tensorflow as tf\n'), ((1089, 1154), 'tensorflow.nn.conv2d', 'tf.nn.conv2d', (['img1', 'window'], {'strides': '[1, 1, 1, 1]', 'padding': '"""VALID"""'}), "(img1, window, strides=[1, 1, 1, 1], padding='VALID')\n", (1101, 1154), True, 'import tensorflow as tf\n'), ((1165, 1230), 'tensorflow.nn.conv2d', 'tf.nn.conv2d', (['img2', 'window'], {'strides': '[1, 1, 1, 1]', 'padding': '"""VALID"""'}), "(img2, window, strides=[1, 1, 1, 1], padding='VALID')\n", (1177, 1230), True, 'import tensorflow as tf\n'), ((2176, 2247), 'tensorflow.constant', 'tf.constant', (['[0.0448, 0.2856, 0.3001, 0.2363, 0.1333]'], {'dtype': 'tf.float32'}), '([0.0448, 0.2856, 0.3001, 0.2363, 0.1333], dtype=tf.float32)\n', (2187, 2247), True, 'import tensorflow as tf\n'), ((2749, 2771), 'tensorflow.pack', 'tf.pack', (['mssim'], {'axis': '(0)'}), '(mssim, axis=0)\n', (2756, 2771), True, 'import tensorflow as tf\n'), ((2782, 2802), 'tensorflow.pack', 'tf.pack', (['mcs'], {'axis': '(0)'}), '(mcs, axis=0)\n', (2789, 2802), True, 'import tensorflow as tf\n'), ((617, 633), 'tensorflow.reduce_sum', 'tf.reduce_sum', (['g'], {}), '(g)\n', (630, 633), True, 'import tensorflow as tf\n'), ((1311, 1383), 'tensorflow.nn.conv2d', 'tf.nn.conv2d', (['(img1 * img1)', 'window'], {'strides': '[1, 1, 1, 1]', 'padding': '"""VALID"""'}), "(img1 * img1, window, strides=[1, 1, 1, 1], padding='VALID')\n", (1323, 1383), True, 'import tensorflow as tf\n'), ((1436, 1508), 'tensorflow.nn.conv2d', 'tf.nn.conv2d', (['(img2 * img2)', 'window'], {'strides': '[1, 1, 1, 1]', 'padding': '"""VALID"""'}), "(img2 * img2, window, strides=[1, 1, 1, 1], padding='VALID')\n", (1448, 1508), True, 'import tensorflow as tf\n'), ((1559, 1631), 'tensorflow.nn.conv2d', 'tf.nn.conv2d', (['(img1 * img2)', 'window'], {'strides': '[1, 1, 1, 1]', 'padding': '"""VALID"""'}), "(img1 * img2, window, strides=[1, 1, 1, 1], padding='VALID')\n", (1571, 1631), True, 'import tensorflow as tf\n'), ((2067, 2088), 'tensorflow.reduce_mean', 'tf.reduce_mean', (['value'], {}), '(value)\n', (2081, 2088), True, 'import tensorflow as tf\n'), ((2495, 2559), 'tensorflow.nn.avg_pool', 'tf.nn.avg_pool', (['img1', '[1, 2, 2, 1]', '[1, 2, 2, 1]'], {'padding': '"""SAME"""'}), "(img1, [1, 2, 2, 1], [1, 2, 2, 1], padding='SAME')\n", (2509, 2559), True, 'import tensorflow as tf\n'), ((2583, 2647), 'tensorflow.nn.avg_pool', 'tf.nn.avg_pool', (['img2', '[1, 2, 2, 1]', '[1, 2, 2, 1]'], {'padding': '"""SAME"""'}), "(img2, [1, 2, 2, 1], [1, 2, 2, 1], padding='SAME')\n", (2597, 2647), True, 'import tensorflow as tf\n'), ((2817, 2872), 'tensorflow.reduce_prod', 'tf.reduce_prod', (['(mcs[0:level - 1] ** weight[0:level - 1])'], {}), '(mcs[0:level - 1] ** weight[0:level - 1])\n', (2831, 2872), True, 'import tensorflow as tf\n'), ((2950, 2971), 'tensorflow.reduce_mean', 'tf.reduce_mean', (['value'], {}), '(value)\n', (2964, 2971), True, 'import tensorflow as tf\n'), ((2403, 2427), 'tensorflow.reduce_mean', 'tf.reduce_mean', (['ssim_map'], {}), '(ssim_map)\n', (2417, 2427), True, 'import tensorflow as tf\n'), ((2448, 2470), 'tensorflow.reduce_mean', 'tf.reduce_mean', (['cs_map'], {}), '(cs_map)\n', (2462, 2470), True, 'import tensorflow as tf\n')] |
import train_keras
from keras.models import load_model
import os
import numpy as np
import pandas as pd
from tqdm import tqdm
from keras.callbacks import ModelCheckpoint
import sys
TF_CPP_MIN_LOG_LEVEL=2
TEST_BATCH = 128
def load_params():
X_test = os.listdir('./test-jpg')
X_test = [fn.replace('.jpg', '') for fn in X_test]
model = load_model('model_amazon6.h5', custom_objects={'fbeta': train_keras.fbeta})
with open('tag_columns.txt', 'r') as f:
tag_columns = f.read().split('\n')
return X_test, model, tag_columns
def prediction(X_test, model, tag_columns, test_folder):
result = []
for i in tqdm(range(0, len(X_test), TEST_BATCH)):
X_batch = X_test[i:i+TEST_BATCH]
X_batch = np.array([train_keras.preprocess(train_keras.load_image(fn, folder=test_folder)) for fn in X_batch])
p = model.predict(X_batch)
result.append(p)
r = np.concatenate(result)
r = r > 0.5
table = []
for row in r:
t = []
for b, v in zip(row, tag_columns):
if b:
t.append(v.replace('tag_', ''))
table.append(' '.join(t))
print('Prediction done !')
return table
def launch(test_folder):
X_test, model, tag_columns = load_params()
table = prediction(X_test, model, tag_columns, test_folder)
try:
df_pred = pd.DataFrame.from_dict({'image_name': X_test, 'tags': table})
df_pred.to_csv('submission9.csv', index=False)
except:
np.save('image_name', X_test)
np.save('table', table)
if __name__ == '__main__':
if len(sys.argv) > 1:
test_folder = sys.argv[1]
else:
test_folder='test-jpg'
launch(test_folder) | [
"keras.models.load_model",
"numpy.save",
"train_keras.load_image",
"pandas.DataFrame.from_dict",
"os.listdir",
"numpy.concatenate"
] | [((255, 279), 'os.listdir', 'os.listdir', (['"""./test-jpg"""'], {}), "('./test-jpg')\n", (265, 279), False, 'import os\n'), ((347, 422), 'keras.models.load_model', 'load_model', (['"""model_amazon6.h5"""'], {'custom_objects': "{'fbeta': train_keras.fbeta}"}), "('model_amazon6.h5', custom_objects={'fbeta': train_keras.fbeta})\n", (357, 422), False, 'from keras.models import load_model\n'), ((905, 927), 'numpy.concatenate', 'np.concatenate', (['result'], {}), '(result)\n', (919, 927), True, 'import numpy as np\n'), ((1347, 1408), 'pandas.DataFrame.from_dict', 'pd.DataFrame.from_dict', (["{'image_name': X_test, 'tags': table}"], {}), "({'image_name': X_test, 'tags': table})\n", (1369, 1408), True, 'import pandas as pd\n'), ((1484, 1513), 'numpy.save', 'np.save', (['"""image_name"""', 'X_test'], {}), "('image_name', X_test)\n", (1491, 1513), True, 'import numpy as np\n'), ((1522, 1545), 'numpy.save', 'np.save', (['"""table"""', 'table'], {}), "('table', table)\n", (1529, 1545), True, 'import numpy as np\n'), ((768, 814), 'train_keras.load_image', 'train_keras.load_image', (['fn'], {'folder': 'test_folder'}), '(fn, folder=test_folder)\n', (790, 814), False, 'import train_keras\n')] |
# -*- coding: utf-8 -*-
#
# BRAINS
# (B)LR (R)everberation-mapping (A)nalysis (I)n AGNs with (N)ested (S)ampling
# <NAME>, <EMAIL>
# Thu, Aug 4, 2016
#
import os
import sys
import corner
import numpy as np
import configparser as cp
import matplotlib.pyplot as plt
__all__ = ['plotbackend']
class plotbackend:
"""
plot backend for BRAINS
"""
def __init__(self, fname="../src/param", fopt=""):
# setup param
self.param_file = fname
self._param_parser(self.param_file)
self.file_dir = self.param['filedir']+"/"
# setup options
if fopt == "":
if self.param['flagdim'] == '-2':
self.option_file = ""
elif self.param['flagdim'] == '-1':
self.option_file = self.file_dir + "/src/OPTIONSCON"
elif self.param['flagdim'] == '0':
self.option_file = self.file_dir + "/src/OPTIONSCON"
elif self.param['flagdim'] == '1':
self.option_file = self.file_dir + "/src/OPTIONS1D"
elif self.param['flagdim'] == '2':
self.option_file = self.file_dir + "/src/OPTIONS2D"
elif self.param['flagdim'] == '3':
self.option_file = self.file_dir + "/src/OPTIONSSA"
elif self.param['flagdim'] == '4':
self.option_file = self.file_dir + "/src/OPTIONSSA1D"
elif self.param['flagdim'] == '5':
self.option_file = self.file_dir + "/src/OPTIONSSA2D"
if self.option_file != "":
self._option_load(self.option_file)
self._get_sample_size()
else:
self.sample_size = 0
else:
self.set_option_file(fopt)
def _option_load(self, fname):
"""
load option file
"""
with open(fname, "r") as f:
lines = f.readlines()
# negect comments
i=0
for line in lines:
if line[0] == '#' or len(line.strip()) == 0:
i+=1
option={}
option['num_particles'] = int(lines[i].split()[0])
i+=1
option['new_level_interval'] = int(lines[i].split()[0])
i+=1
option['save_interval'] = int(lines[i].split()[0])
i+=1
option['thread_step'] = int(lines[i].split()[0])
i+=1
option['num_levels'] = int(lines[i].split()[0])
i+=1
option['lambda'] = int(lines[i].split()[0])
i+=1
option['beta'] = int(lines[i].split()[0])
i+=1
option['num_saves'] = int(lines[i].split()[0])
i+=1
option['file_sample'] = lines[i].split()[0]
i+=1
option['file_sample_info'] = lines[i].split()[0]
i+=1
option['file_levels'] = lines[i].split()[0]
i+=1
option['file_sampler_state'] = lines[i].split()[0]
i+=1
option['file_post_sample'] = lines[i].split()[0]
i+=1
option['file_post_sample_info'] = lines[i].split()[0]
i+=1
option['file_limits'] = lines[i].split()[0]
self.option = option
def _param_parser(self, fname):
"""
parse parameter file
"""
config = cp.RawConfigParser(delimiters=' ', comment_prefixes='%', inline_comment_prefixes='%',
default_section=cp.DEFAULTSECT, empty_lines_in_values=False)
with open(fname) as f:
file_content = '[dump]\n' + f.read()
config.read_string(file_content)
# check the absolute path
if os.path.isabs(config['dump']['filedir']) == False:
raise Exception("FileDir in %s is not an absoulte path.\n"%self.param_file)
self.param = config['dump']
def _get_sample_size(self):
"""
load results
"""
with open(self.file_dir+self.option['file_post_sample']) as f:
self.sample_size = int(f.readline().split()[1])
def set_param_file(self, fname):
"""
set parameter file
"""
self.param_file = fname
self._param_parser(fname)
self.file_dir = self.param['filedir']+"/"
return
def set_option_file(self, fname):
"""
set option file
"""
self.option_file = fname
self._option_load(fname)
self._get_sample_size()
return
def load_results(self):
"""
load results
"""
self.results={}
if self.param['flagdim'] == '-2':
self.results['con_sim'] = np.loadtxt(self.file_dir+"data/sim_con.txt")
self.results['line_sim'] = np.loadtxt(self.file_dir+"data/sim_hb.txt")
self.results['line2d_sim'] = np.loadtxt(self.file_dir+"data/sim_hb2d.txt")
elif self.param['flagdim'] == '-1':
self.results['con_data'] = np.loadtxt(self.file_dir+self.param['continuumfile'])
self.results['con_sim'] = np.loadtxt(self.file_dir+"data/sim_con.txt")
self.results['line_sim'] = np.loadtxt(self.file_dir+"data/sim_hb.txt")
self.results['line2d_sim'] = np.loadtxt(self.file_dir+"data/sim_hb2d.txt")
elif self.param['flagdim'] == '0':
self.results['sample'] = np.loadtxt(self.file_dir + self.option['file_post_sample'])
self.results['con_data'] = np.loadtxt(self.file_dir+self.param['continuumfile'])
self.results['con_rec'] = np.loadtxt(self.file_dir+"data/con_rec.txt")
return
def plot_drw_parameters(self):
if self.param['flagdim'] == '3' or self.param['flagdim'] == '-2':
raise Exception("FlagDim=%d, no DRW parameters.\n"%self.param['flagdim'])
sample = self.results['sample']
fig = corner.corner(sample[:, 1:3], smooth=True, smooth1d=True, labels=[r"$\ln(\hat\sigma)$", r"$\ln(\tau)$"])
return fig
def plot_con_rec(self):
if self.param['flagdim'] == '3' or self.param['flagdim'] == '-2':
raise Exception("FlagDim=%d, no continuum reconstruction.\n"%self.param['flagdim'])
con_data = self.results['con_data']
con = self.results['con_rec']
offset = int(con.shape[0]/self.sample_size)
fig, ax = plt.subplots(1,1)
ax.errorbar(con_data[:, 0], con_data[:, 1], yerr = con_data[:, 2], ls='none', marker='o', label='Data')
for i in np.random.randint(self.sample_size, size=np.min((100, self.sample_size))):
plt.plot(con[i*offset:(i+1)*offset, 0], con[i*offset:(i+1)*offset, 1], lw=0.2)
ax.set_xlabel('Time')
ax.set_ylabel('Flux')
ax.legend()
return fig
| [
"os.path.isabs",
"corner.corner",
"matplotlib.pyplot.plot",
"configparser.RawConfigParser",
"numpy.min",
"numpy.loadtxt",
"matplotlib.pyplot.subplots"
] | [((2873, 3027), 'configparser.RawConfigParser', 'cp.RawConfigParser', ([], {'delimiters': '""" """', 'comment_prefixes': '"""%"""', 'inline_comment_prefixes': '"""%"""', 'default_section': 'cp.DEFAULTSECT', 'empty_lines_in_values': '(False)'}), "(delimiters=' ', comment_prefixes='%',\n inline_comment_prefixes='%', default_section=cp.DEFAULTSECT,\n empty_lines_in_values=False)\n", (2891, 3027), True, 'import configparser as cp\n'), ((5179, 5291), 'corner.corner', 'corner.corner', (['sample[:, 1:3]'], {'smooth': '(True)', 'smooth1d': '(True)', 'labels': "['$\\\\ln(\\\\hat\\\\sigma)$', '$\\\\ln(\\\\tau)$']"}), "(sample[:, 1:3], smooth=True, smooth1d=True, labels=[\n '$\\\\ln(\\\\hat\\\\sigma)$', '$\\\\ln(\\\\tau)$'])\n", (5192, 5291), False, 'import corner\n'), ((5631, 5649), 'matplotlib.pyplot.subplots', 'plt.subplots', (['(1)', '(1)'], {}), '(1, 1)\n', (5643, 5649), True, 'import matplotlib.pyplot as plt\n'), ((3178, 3218), 'os.path.isabs', 'os.path.isabs', (["config['dump']['filedir']"], {}), "(config['dump']['filedir'])\n", (3191, 3218), False, 'import os\n'), ((4059, 4105), 'numpy.loadtxt', 'np.loadtxt', (["(self.file_dir + 'data/sim_con.txt')"], {}), "(self.file_dir + 'data/sim_con.txt')\n", (4069, 4105), True, 'import numpy as np\n'), ((4137, 4182), 'numpy.loadtxt', 'np.loadtxt', (["(self.file_dir + 'data/sim_hb.txt')"], {}), "(self.file_dir + 'data/sim_hb.txt')\n", (4147, 4182), True, 'import numpy as np\n'), ((4216, 4263), 'numpy.loadtxt', 'np.loadtxt', (["(self.file_dir + 'data/sim_hb2d.txt')"], {}), "(self.file_dir + 'data/sim_hb2d.txt')\n", (4226, 4263), True, 'import numpy as np\n'), ((5851, 5945), 'matplotlib.pyplot.plot', 'plt.plot', (['con[i * offset:(i + 1) * offset, 0]', 'con[i * offset:(i + 1) * offset, 1]'], {'lw': '(0.2)'}), '(con[i * offset:(i + 1) * offset, 0], con[i * offset:(i + 1) *\n offset, 1], lw=0.2)\n', (5859, 5945), True, 'import matplotlib.pyplot as plt\n'), ((4336, 4391), 'numpy.loadtxt', 'np.loadtxt', (["(self.file_dir + self.param['continuumfile'])"], {}), "(self.file_dir + self.param['continuumfile'])\n", (4346, 4391), True, 'import numpy as np\n'), ((4422, 4468), 'numpy.loadtxt', 'np.loadtxt', (["(self.file_dir + 'data/sim_con.txt')"], {}), "(self.file_dir + 'data/sim_con.txt')\n", (4432, 4468), True, 'import numpy as np\n'), ((4500, 4545), 'numpy.loadtxt', 'np.loadtxt', (["(self.file_dir + 'data/sim_hb.txt')"], {}), "(self.file_dir + 'data/sim_hb.txt')\n", (4510, 4545), True, 'import numpy as np\n'), ((4579, 4626), 'numpy.loadtxt', 'np.loadtxt', (["(self.file_dir + 'data/sim_hb2d.txt')"], {}), "(self.file_dir + 'data/sim_hb2d.txt')\n", (4589, 4626), True, 'import numpy as np\n'), ((5811, 5842), 'numpy.min', 'np.min', (['(100, self.sample_size)'], {}), '((100, self.sample_size))\n', (5817, 5842), True, 'import numpy as np\n'), ((4700, 4759), 'numpy.loadtxt', 'np.loadtxt', (["(self.file_dir + self.option['file_post_sample'])"], {}), "(self.file_dir + self.option['file_post_sample'])\n", (4710, 4759), True, 'import numpy as np\n'), ((4793, 4848), 'numpy.loadtxt', 'np.loadtxt', (["(self.file_dir + self.param['continuumfile'])"], {}), "(self.file_dir + self.param['continuumfile'])\n", (4803, 4848), True, 'import numpy as np\n'), ((4879, 4925), 'numpy.loadtxt', 'np.loadtxt', (["(self.file_dir + 'data/con_rec.txt')"], {}), "(self.file_dir + 'data/con_rec.txt')\n", (4889, 4925), True, 'import numpy as np\n')] |
import sys
import os
import numpy as np
import random
from collections import OrderedDict
import pickle
import datetime
from tqdm import tqdm
from recordclass import recordclass
import math
import torch
import torch.autograd as autograd
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
import json
# Helper funcs
def custom_print(*msg):
for i in range(0, len(msg)):
if i == len(msg) - 1:
print(msg[i])
logger.write(str(msg[i]) + '\n')
else:
print(msg[i], ' ', end='')
logger.write(str(msg[i]))
def load_word_embedding(embed_file, vocab):
custom_print('vocab length:', len(vocab))
embed_vocab = OrderedDict()
rev_embed_vocab = OrderedDict()
embed_matrix = list()
embed_vocab['<PAD>'] = 0
rev_embed_vocab[0] = '<PAD>'
embed_matrix.append(np.zeros(word_embed_dim, dtype=np.float32))
embed_vocab['<UNK>'] = 1
rev_embed_vocab[1] = '<UNK>'
embed_matrix.append(np.random.uniform(-0.25, 0.25, word_embed_dim))
embed_vocab['<SOS>'] = 2
rev_embed_vocab[2] = '<SOS>'
embed_matrix.append(np.random.uniform(-0.25, 0.25, word_embed_dim))
embed_vocab['<EOS>'] = 3
rev_embed_vocab[3] = '<EOS>'
embed_matrix.append(np.random.uniform(-0.25, 0.25, word_embed_dim))
word_idx = 4
with open(embed_file, "r") as f:
for line in f:
parts = line.split()
if len(parts) < word_embed_dim + 1:
continue
word = parts[0]
if word in vocab and vocab[word] >= word_min_freq:
vec = [np.float32(val) for val in parts[1:]]
embed_matrix.append(vec)
embed_vocab[word] = word_idx
rev_embed_vocab[word_idx] = word
word_idx += 1
for word in vocab:
if word not in embed_vocab and vocab[word] >= word_min_freq:
embed_matrix.append(np.random.uniform(-0.25, 0.25, word_embed_dim))
embed_vocab[word] = word_idx
rev_embed_vocab[word_idx] = word
word_idx += 1
custom_print('embed dictionary length:', len(embed_vocab))
return embed_vocab, rev_embed_vocab, np.array(embed_matrix, dtype=np.float32)
def build_vocab(data, events, arguments, roles, vocab_file, embed_file):
vocab = OrderedDict()
char_v = OrderedDict()
char_v['<PAD>'] = 0
char_v['<UNK>'] = 1
char_v[';'] = 2
char_v['|'] = 3
char_idx = 4
for d in data:
for word in d.SrcWords:
if word not in vocab:
vocab[word] = 1
else:
vocab[word] += 1
for c in word:
if c not in char_v:
char_v[c] = char_idx
char_idx += 1
for event in events:
vocab[event] = word_min_freq
for argument in arguments:
vocab[argument] = word_min_freq
for role in roles:
vocab[role] = word_min_freq
vocab[';'] = word_min_freq
vocab['|'] = word_min_freq
word_v, rev_word_v, embed_matrix = load_word_embedding(embed_file, vocab)
output = open(vocab_file, 'wb')
pickle.dump([word_v, char_v], output)
output.close()
return word_v, rev_word_v, char_v, embed_matrix
def load_vocab(vocab_file):
with open(vocab_file, 'rb') as f:
word_v, char_v = pickle.load(f)
return word_v, char_v
def get_adj_mat(amat):
K = 5
adj_mat = np.zeros((len(amat), len(amat)), np.float32)
for i in range(len(amat)):
for j in range(len(amat)):
if 0 <= amat[i][j] <= K:
adj_mat[i][j] = 1.0 / math.pow(2, amat[i][j])
else:
adj_mat[i][j] = 0
return adj_mat
def get_data(src_lines, trg_lines, datatype):
samples = []
uid = 1
src_len = -1
trg_len = -1
for i in range(0, len(src_lines)):
src_line = src_lines[i].strip()
trg_line = trg_lines[i].strip()
src_words = src_line.split()
if datatype == 1:
tuples = trg_line.strip().split('|')
random.shuffle(tuples)
new_trg_line = ' | '.join(tuples)
assert len(trg_line.split()) == len(new_trg_line.split())
trg_line = new_trg_line
trg_words = list()
trg_words.append('<SOS>')
trg_words += trg_line.split()
trg_words.append('<EOS>')
if datatype == 1 and (len(src_words) > max_src_len or len(trg_words) > max_trg_len + 1):
continue
if len(src_words) > src_len:
src_len = len(src_words)
if len(trg_words) > trg_len:
trg_len = len(trg_words)
sample = Sample(Id=uid, SrcLen=len(src_words), SrcWords=src_words, TrgLen=len(trg_words),
TrgWords=trg_words) #c
samples.append(sample)
uid += 1
print(src_len)
print(trg_len)
return samples
def read_data(src_file, trg_file, datatype):
reader = open(src_file)
src_lines = reader.readlines()
reader.close()
reader = open(trg_file)
trg_lines = reader.readlines()
reader.close()
# tot_len = 100
# src_lines = src_lines[0:min(tot_len, len(src_lines))]
# trg_lines = trg_lines[0:min(tot_len, len(trg_lines))]
# adj_lines = adj_lines[0:min(tot_len, len(adj_lines))]
data = get_data(src_lines, trg_lines, datatype)
return data
#event_lines, argument_lines, roles_lines
# to add option for less detailed checks
def check_event_trigger(ref_string, pred_string):
return (ref_string == pred_string)
pass
def check_event_type(ref_string, pred_string, event_lines):
if granular_mode == 0:
if pred_string in event_lines:
return (ref_string == pred_string)
else:
# print("invalid prediction")
return False
pass
if granular_mode == 1:
pred_token = pred_string.split(":")[0]
ref_token = ref_string.split(":")[0]
return (pred_token == ref_token)
pass
def check_event_argument(ref_string, pred_string):
return (ref_string == pred_string)
pass
def check_argument_type(ref_string, pred_string, argument_lines):
if granular_mode == 0:
if pred_string in argument_lines:
return (ref_string == pred_string)
else:
# print("invalid prediction")
return False
pass
if granular_mode == 1:
pred_token = pred_string.split(":")[0]
ref_token = ref_string.split(":")[0]
return (pred_token == ref_token)
pass
def check_argument_role(ref_string, pred_string, roles_lines):
if pred_string in roles_lines:
return (ref_string == pred_string)
else:
# print("invalid prediction")
return False
pass
def calculate_f1(ref_lines, pred_lines, event_lines, argument_lines, roles_lines):
list_of_tracking_metrics = ['predicted_tuples',
'ground_truth_tuples',
'correct_predictions',
'events_count',
'correct_events',
'correct_event_type',
'correct_arguments',
'correct_argment_types',
'correct_argument_roles'
]
metric_counts = dict.fromkeys(list_of_tracking_metrics, 0)
for i in range(0, min(len(ref_lines), len(pred_lines))):
ref_line = ref_lines[i].strip()
pred_line = pred_lines[i].strip()
ref_tuples = ref_line.split('|')
pred_tuples = pred_line.split('|')
# find a way to compare multiple tuples
# correct - t1 | t2 | t3
# pred - p1 | p2
# postives = 3 [number of ground truths minus nones]
# predicted_pos = 2 [number of preds minus nones]
# TP = correct preds
# TP + FP = predicted
# TP + FN = positives
# Precision = correct / predicted_pos
# Recall = correct / positives
# f = pr/p+r
# handling repeated predictions
# set_of_preds = set()
# for pred_tuple in pred_tuples:
# set_of_preds.add(pred_tuple.strip())
# pred_tuples = list(set_of_preds)
for pred_tuple in pred_tuples:
pred_strings = pred_tuple.split(';')
if(len(pred_strings) < 3):
continue
# in the case of no argument detection, we only calculate the event trigger scores
if(pred_strings[2].strip().lower()) == 'none':
max_matches = 0
part_matches = []
for ref_tuple in ref_tuples:
# ssss
ev1, ev2 = cal_f1_for_pair(ref_tuple, pred_tuple, event_lines)
pair_score = ev1+ev2
if pair_score > max_matches:
max_matches = pair_score
part_matches = (ev1, ev2)
pass
pass
metric_counts['events_count'] += 1
if ev1 == 1:
metric_counts['correct_events'] += 1
if ev2 == 1:
metric_counts['correct_event_type'] += 1
continue
max_matches = 0
part_matches = cal_f1_for_tuple(ref_tuples[0], pred_tuple, event_lines, argument_lines, roles_lines)
for ref_tuple in ref_tuples:
res = cal_f1_for_tuple(ref_tuple, pred_tuple, event_lines, argument_lines, roles_lines)
tuple_score = sum(res)
if tuple_score >= max_matches:
max_matches = tuple_score
part_matches = res
pass
pass
metric_counts['predicted_tuples'] += 1
metric_counts['events_count'] += 1
if max_matches >= 4:
metric_counts['correct_predictions'] += 1
if part_matches[0] == 1:
metric_counts['correct_events'] += 1
if part_matches[1] == 1:
metric_counts['correct_event_type'] += 1
if part_matches[2] == 1:
metric_counts['correct_arguments'] += 1
if part_matches[3] == 1:
metric_counts['correct_argment_types'] += 1
if part_matches[4] == 1:
metric_counts['correct_argument_roles'] += 1
pass
for ref_tuple in ref_tuples:
if(ref_tuple.split(';')[2].strip().lower()) != 'none':
metric_counts['ground_truth_tuples'] += 1
pass
print(metric_counts)
precision = float(metric_counts['correct_predictions'] / (metric_counts['predicted_tuples'] + 1e-08))
recall = float(metric_counts['correct_predictions'] / (metric_counts['ground_truth_tuples'] + 1e-08))
f1 = 2 * precision * recall / (precision + recall + 1e-08)
precision = round(precision, 3)
recall = round(recall, 3)
f1 = round(f1, 3)
print("Partwise Results")
event_acc = metric_counts['correct_events']/ (metric_counts['events_count'] + 1e-08)
evtype_acc = metric_counts['correct_event_type']/ (metric_counts['events_count'] + 1e-08)
argument_acc = metric_counts['correct_arguments']/ (metric_counts['predicted_tuples'] + 1e-08)
argtype_acc = metric_counts['correct_argment_types']/ (metric_counts['predicted_tuples'] + 1e-08)
role_acc = metric_counts['correct_argument_roles']/ (metric_counts['predicted_tuples'] + 1e-08)
print(f'Event Trigger Word Accuracy: {event_acc}')
print(f'Event Type Accuracy: {evtype_acc}')
print(f'Argument Identification Accuracy: {argument_acc}')
print(f'Argument Type Accuracy: {argtype_acc}')
print(f'Argument Role Accuracy: {role_acc}')
print(f'Macro f-score: {f1}')
targ_file = os.path.join(trg_data_folder, 'Results_logger.txt')
f = open(targ_file, "a")
f.write(f'Event Trigger Word Accuracy: {event_acc}')
f.write("\n")
f.write(f'Event Type Accuracy: {evtype_acc}')
f.write("\n")
f.write(f'Argument Identification Accuracy: {argument_acc}')
f.write("\n")
f.write(f'Argument Type Accuracy: {argtype_acc}')
f.write("\n")
f.write(f'Argument Role Accuracy: {role_acc}')
f.write("\n")
f.write(f'Macro f-score: {f1}')
f.write("\n")
f.close()
return f1
def cal_f1_for_pair(ref_tuple: str ,
pred_tuple: str,
event_lines: list
) -> list:
ref_strings = ref_tuple.split(';')
pred_strings = pred_tuple.split(';')
ev1 = int( check_event_trigger(ref_strings[0].strip(), pred_strings[0].strip()) )
ev2 = int( check_event_type(ref_strings[1].strip(), pred_strings[1].strip(), event_lines) )
return ev1, ev2
def cal_f1_for_tuple(ref_tuple: str ,
pred_tuple: str,
event_lines: list,
argument_lines: list,
roles_lines: list
) -> list:
ref_strings = ref_tuple.split(';')
pred_strings = pred_tuple.split(';')
if (len (pred_strings) != 5 ):
if (len (pred_strings) >= 2 ):
ev1 = int( check_event_trigger(ref_strings[0].strip(), pred_strings[0].strip()) )
ev2 = int( check_event_type(ref_strings[1].strip(), pred_strings[1].strip(), event_lines) )
return [ev1, ev2, 0, 0, 0]
return list([0,0,0,0,0])
ev1 = int( check_event_trigger(ref_strings[0].strip(), pred_strings[0].strip()) )
ev2 = int( check_event_type(ref_strings[1].strip(), pred_strings[1].strip(), event_lines) )
ev3 = int( check_event_argument(ref_strings[2].strip(), pred_strings[2].strip()) )
ev4 = int( check_argument_type(ref_strings[3].strip(), pred_strings[3].strip(), argument_lines) )
ev5 = int( check_argument_role(ref_strings[4].strip(), pred_strings[4].strip(), roles_lines) )
ret = [ev1, ev2, ev3, ev4, ev5]
return ret
def get_model(model_id):
if model_id == 1:
return SeqToSeqModel()
def write_test_res(data, preds, attns, outfile):
writer = open(outfile, 'w')
for i in range(0, len(data)):
pred_words = get_pred_words(preds[i], attns[i], data[i].SrcWords)[:-1]
writer.write(' '.join(pred_words) + '\n')
writer.close()
def set_random_seeds(seed):
random.seed(seed)
np.random.seed(seed)
torch.manual_seed(seed)
if n_gpu > 1:
torch.cuda.manual_seed_all(seed)
def get_max_len(sample_batch):
src_max_len = len(sample_batch[0].SrcWords)
for idx in range(1, len(sample_batch)):
if len(sample_batch[idx].SrcWords) > src_max_len:
src_max_len = len(sample_batch[idx].SrcWords)
trg_max_len = len(sample_batch[0].TrgWords)
for idx in range(1, len(sample_batch)):
if len(sample_batch[idx].TrgWords) > trg_max_len:
trg_max_len = len(sample_batch[idx].TrgWords)
return src_max_len, trg_max_len
def get_words_index_seq(words, max_len):
seq = list()
for word in words:
if word in word_vocab:
seq.append(word_vocab[word])
else:
seq.append(word_vocab['<UNK>'])
pad_len = max_len - len(words)
for i in range(0, pad_len):
seq.append(word_vocab['<PAD>'])
return seq
def get_target_words_index_seq(words, max_len):
seq = list()
for word in words:
if word in word_vocab:
seq.append(word_vocab[word])
else:
seq.append(word_vocab['<UNK>'])
pad_len = max_len - len(words)
for i in range(0, pad_len):
seq.append(word_vocab['<EOS>'])
return seq
def get_padded_mask(cur_len, max_len):
mask_seq = list()
for i in range(0, cur_len):
mask_seq.append(0)
pad_len = max_len - cur_len
for i in range(0, pad_len):
mask_seq.append(1)
return mask_seq
def get_target_vocab_mask(src_words):
mask = []
for i in range(0, len(word_vocab)):
mask.append(1)
for word in src_words:
if word in word_vocab:
mask[word_vocab[word]] = 0
# events, arguments, roles
for event in events:
mask[word_vocab[event]] = 0
for argument in arguments:
mask[word_vocab[argument]] = 0
for role in roles:
mask[word_vocab[role]] = 0
mask[word_vocab['<UNK>']] = 0
mask[word_vocab['<EOS>']] = 0
mask[word_vocab[';']] = 0
mask[word_vocab['|']] = 0
return mask
def get_rel_mask(trg_words, max_len):
mask_seq = list()
for word in trg_words:
mask_seq.append(0)
# if word in relations:
# mask_seq.append(0)
# else:
# mask_seq.append(1)
pad_len = max_len - len(trg_words)
for i in range(0, pad_len):
mask_seq.append(1)
return mask_seq
def get_char_seq(words, max_len):
char_seq = list()
for i in range(0, conv_filter_size - 1):
char_seq.append(char_vocab['<PAD>'])
for word in words:
for c in word[0:min(len(word), max_word_len)]:
if c in char_vocab:
char_seq.append(char_vocab[c])
else:
char_seq.append(char_vocab['<UNK>'])
pad_len = max_word_len - len(word)
for i in range(0, pad_len):
char_seq.append(char_vocab['<PAD>'])
for i in range(0, conv_filter_size - 1):
char_seq.append(char_vocab['<PAD>'])
pad_len = max_len - len(words)
for i in range(0, pad_len):
for i in range(0, max_word_len + conv_filter_size - 1):
char_seq.append(char_vocab['<PAD>'])
return char_seq
def get_relations(file_name):
rels = []
reader = open(file_name)
lines = reader.readlines()
reader.close()
for line in lines:
rels.append(line.strip())
return rels
def get_batch_data(cur_samples, is_training=False):
"""
Returns the training samples and labels as numpy array
"""
batch_src_max_len, batch_trg_max_len = get_max_len(cur_samples)
src_words_list = list()
src_words_mask_list = list()
src_char_seq = list()
trg_words_list = list()
trg_vocab_mask = list()
adj_lst = []
target = list()
cnt = 0
for sample in cur_samples:
src_words_list.append(get_words_index_seq(sample.SrcWords, batch_src_max_len))
src_words_mask_list.append(get_padded_mask(sample.SrcLen, batch_src_max_len))
src_char_seq.append(get_char_seq(sample.SrcWords, batch_src_max_len))
trg_vocab_mask.append(get_target_vocab_mask(sample.SrcWords))
# cur_masked_adj = np.zeros((batch_src_max_len, batch_src_max_len), dtype=np.float32)
# cur_masked_adj[:len(sample.SrcWords), :len(sample.SrcWords)] = sample.AdjMat
# adj_lst.append(cur_masked_adj)
if is_training:
padded_trg_words = get_words_index_seq(sample.TrgWords, batch_trg_max_len)
trg_words_list.append(padded_trg_words)
target.append(padded_trg_words[1:])
else:
trg_words_list.append(get_words_index_seq(['<SOS>'], 1))
cnt += 1
return {'src_words': np.array(src_words_list, dtype=np.float32),
'src_chars': np.array(src_char_seq),
'src_words_mask': np.array(src_words_mask_list),
'adj': np.array(adj_lst),
'trg_vocab_mask': np.array(trg_vocab_mask),
'trg_words': np.array(trg_words_list, dtype=np.int32),
'target': np.array(target)}
def shuffle_data(data):
custom_print(len(data))
data.sort(key=lambda x: x.SrcLen)
num_batch = int(len(data) / batch_size)
rand_idx = random.sample(range(num_batch), num_batch)
new_data = []
for idx in rand_idx:
new_data += data[batch_size * idx: batch_size * (idx + 1)]
if len(new_data) < len(data):
new_data += data[num_batch * batch_size:]
return new_data
def get_pred_words(preds, attns, src_words):
pred_words = []
for i in range(0, max_trg_len):
word_idx = preds[i]
if word_vocab['<EOS>'] == word_idx:
pred_words.append('<EOS>')
break
elif att_type != 'None' and copy_on and word_vocab['<UNK>'] == word_idx:
word_idx = attns[i]
pred_words.append(src_words[word_idx])
else:
pred_words.append(rev_word_vocab[word_idx])
return pred_words
class WordEmbeddings(nn.Module):
def __init__(self, vocab_size, embed_dim, pre_trained_embed_matrix, drop_out_rate):
super(WordEmbeddings, self).__init__()
self.embeddings = nn.Embedding(vocab_size, embed_dim, padding_idx=0)
self.embeddings.weight.data.copy_(torch.from_numpy(pre_trained_embed_matrix))
self.dropout = nn.Dropout(drop_out_rate)
def forward(self, words_seq):
word_embeds = self.embeddings(words_seq)
word_embeds = self.dropout(word_embeds)
return word_embeds
def weight(self):
return self.embeddings.weight
# Potentially use a pretrained BERT - 509
class CharEmbeddings(nn.Module):
def __init__(self, vocab_size, embed_dim, drop_out_rate):
super(CharEmbeddings, self).__init__()
# Layers
self.embeddings = nn.Embedding(vocab_size, embed_dim, padding_idx=0)
self.dropout = nn.Dropout(drop_out_rate)
def forward(self, words_seq):
char_embeds = self.embeddings(words_seq)
char_embeds = self.dropout(char_embeds)
return char_embeds
# DONT CHANGE CLASSES
# 543
class Encoder(nn.Module):
def __init__(self, input_dim, hidden_dim, layers, is_bidirectional, drop_out_rate):
super(Encoder, self).__init__()
self.input_dim = input_dim
self.hidden_dim = hidden_dim
self.layers = layers
self.is_bidirectional = is_bidirectional
self.drop_rate = drop_out_rate
self.char_embeddings = CharEmbeddings(len(char_vocab), char_embed_dim, drop_rate)
# Remove In case we want to BERT
self.lstm = nn.LSTM(self.input_dim, self.hidden_dim, self.layers, batch_first=True,
bidirectional=self.is_bidirectional)
self.dropout = nn.Dropout(self.drop_rate)
self.conv1d = nn.Conv1d(char_embed_dim, char_feature_size, conv_filter_size)
self.max_pool = nn.MaxPool1d(max_word_len + conv_filter_size - 1, max_word_len + conv_filter_size - 1)
def forward(self, words_input, char_seq, adj, is_training=False):
char_embeds = self.char_embeddings(char_seq)
char_embeds = char_embeds.permute(0, 2, 1)
char_feature = torch.tanh(self.max_pool(self.conv1d(char_embeds)))
char_feature = char_feature.permute(0, 2, 1)
words_input = torch.cat((words_input, char_feature), -1)
outputs, hc = self.lstm(words_input)
outputs = self.dropout(outputs)
return outputs
# 597
class Attention(nn.Module):
def __init__(self, input_dim):
super(Attention, self).__init__()
self.input_dim = input_dim
self.linear_ctx = nn.Linear(self.input_dim, self.input_dim, bias=False)
self.linear_query = nn.Linear(self.input_dim, self.input_dim, bias=True)
self.v = nn.Linear(self.input_dim, 1)
def forward(self, s_prev, enc_hs, src_mask):
uh = self.linear_ctx(enc_hs)
wq = self.linear_query(s_prev)
wquh = torch.tanh(wq + uh)
attn_weights = self.v(wquh).squeeze()
attn_weights.data.masked_fill_(src_mask.data, -float('inf'))
attn_weights = F.softmax(attn_weights, dim=-1)
ctx = torch.bmm(attn_weights.unsqueeze(1), enc_hs).squeeze()
return ctx, attn_weights
# 617
class NGram_Attention(nn.Module):
def __init__(self, input_dim, N):
super(NGram_Attention, self).__init__()
self.input_dim = input_dim
self.layers = N
self.V_layers = nn.ModuleList()
self.W_layers = nn.ModuleList()
for i in range(N):
self.V_layers.append(nn.Linear(input_dim, input_dim))
self.W_layers.append(nn.Linear(input_dim, input_dim))
def forward(self, s_prev, enc_hs, src_mask):
att = torch.bmm(s_prev.unsqueeze(1), self.V_layers[0](enc_hs).transpose(1, 2)).squeeze()
att.data.masked_fill_(src_mask.data, -float('inf'))
att = F.softmax(att, dim=-1)
ctx = self.W_layers[0](torch.bmm(att.unsqueeze(1), enc_hs).squeeze())
for i in range(1, self.layers):
enc_hs_ngram = torch.nn.AvgPool1d(i+1, 1)(enc_hs.transpose(1, 2)).transpose(1, 2)
n_mask = src_mask.unsqueeze(1).float()
n_mask = torch.nn.AvgPool1d(i+1, 1)(n_mask).squeeze()
n_mask[n_mask > 0] = 1
n_mask = n_mask.byte()
n_att = torch.bmm(s_prev.unsqueeze(1), self.V_layers[i](enc_hs_ngram).transpose(1, 2)).squeeze()
n_att.data.masked_fill_(n_mask.data, -float('inf'))
n_att = F.softmax(n_att, dim=-1)
ctx += self.W_layers[i](torch.bmm(n_att.unsqueeze(1), enc_hs_ngram).squeeze())
return ctx, att
# 588
def mean_over_time(x, mask):
x.data.masked_fill_(mask.unsqueeze(2).data, 0)
x = torch.sum(x, dim=1)
time_steps = torch.sum(mask.eq(0), dim=1, keepdim=True).float()
x /= time_steps
return x
# 645
class Decoder(nn.Module):
def __init__(self, input_dim, hidden_dim, layers, drop_out_rate, max_length):
super(Decoder, self).__init__()
self.input_dim = input_dim
self.hidden_dim = hidden_dim
self.layers = layers
self.drop_rate = drop_out_rate
self.max_length = max_length
if att_type == 'None':
self.lstm = nn.LSTMCell(2 * self.input_dim, self.hidden_dim, self.layers)
elif att_type == 'Unigram':
self.attention = Attention(input_dim)
self.lstm = nn.LSTMCell(2 * self.input_dim, self.hidden_dim, self.layers)
else:
self.attention = NGram_Attention(input_dim, 3)
self.lstm = nn.LSTMCell(3 * self.input_dim, self.hidden_dim, self.layers)
self.dropout = nn.Dropout(self.drop_rate)
self.ent_out = nn.Linear(self.input_dim, len(word_vocab))
def forward(self, y_prev, h_prev, enc_hs, src_word_embeds, src_mask, is_training=False):
src_time_steps = enc_hs.size()[1]
if att_type == 'None':
ctx = mean_over_time(enc_hs, src_mask)
attn_weights = torch.zeros(src_mask.size()).cuda()
elif att_type == 'Unigram':
s_prev = h_prev[0]
s_prev = s_prev.unsqueeze(1)
s_prev = s_prev.repeat(1, src_time_steps, 1)
ctx, attn_weights = self.attention(s_prev, enc_hs, src_mask)
else:
last_index = src_mask.size()[1] - torch.sum(src_mask, dim=-1).long() - 1
last_index = last_index.unsqueeze(1).unsqueeze(1).repeat(1, 1, enc_hs.size()[-1])
enc_last = torch.gather(enc_hs, 1, last_index).squeeze()
ctx, attn_weights = self.attention(enc_last, src_word_embeds, src_mask)
ctx = torch.cat((enc_last, ctx), -1)
y_prev = y_prev.squeeze()
s_cur = torch.cat((y_prev, ctx), 1)
hidden, cell_state = self.lstm(s_cur, h_prev)
hidden = self.dropout(hidden)
output = self.ent_out(hidden)
return output, (hidden, cell_state), attn_weights
# 690
class SeqToSeqModel(nn.Module):
def __init__(self):
super(SeqToSeqModel, self).__init__()
self.word_embeddings = WordEmbeddings(len(word_vocab), word_embed_dim, word_embed_matrix, drop_rate)
self.encoder = Encoder(enc_inp_size, int(enc_hidden_size/2), layers, True, drop_rate)
self.decoder = Decoder(dec_inp_size, dec_hidden_size, layers, drop_rate, max_trg_len)
def forward(self, src_words_seq, src_chars_seq, src_mask, trg_words_seq, trg_vocab_mask, adj, is_training=False):
src_word_embeds = self.word_embeddings(src_words_seq)
trg_word_embeds = self.word_embeddings(trg_words_seq)
batch_len = src_word_embeds.size()[0]
if is_training:
time_steps = trg_word_embeds.size()[1] - 1
else:
time_steps = max_trg_len
encoder_output = self.encoder(src_word_embeds, src_chars_seq, adj, is_training)
h0 = autograd.Variable(torch.FloatTensor(torch.zeros(batch_len, word_embed_dim)))
h0 = h0.cuda()
c0 = autograd.Variable(torch.FloatTensor(torch.zeros(batch_len, word_embed_dim)))
c0 = c0.cuda()
dec_hid = (h0, c0)
if is_training:
dec_inp = trg_word_embeds[:, 0, :]
dec_out, dec_hid, dec_attn = self.decoder(dec_inp, dec_hid, encoder_output, src_word_embeds,
src_mask, is_training)
dec_out = dec_out.view(-1, len(word_vocab))
dec_out = F.log_softmax(dec_out, dim=-1)
dec_out = dec_out.unsqueeze(1)
for t in range(1, time_steps):
dec_inp = trg_word_embeds[:, t, :]
cur_dec_out, dec_hid, dec_attn = self.decoder(dec_inp, dec_hid, encoder_output, src_word_embeds,
src_mask, is_training)
cur_dec_out = cur_dec_out.view(-1, len(word_vocab))
dec_out = torch.cat((dec_out, F.log_softmax(cur_dec_out, dim=-1).unsqueeze(1)), 1)
else:
dec_inp = trg_word_embeds[:, 0, :]
dec_out, dec_hid, dec_attn = self.decoder(dec_inp, dec_hid, encoder_output, src_word_embeds,
src_mask, is_training)
dec_out = dec_out.view(-1, len(word_vocab))
if copy_on:
dec_out.data.masked_fill_(trg_vocab_mask.data, -float('inf'))
dec_out = F.log_softmax(dec_out, dim=-1)
topv, topi = dec_out.topk(1)
dec_out_v, dec_out_i = dec_out.topk(1)
dec_attn_v, dec_attn_i = dec_attn.topk(1)
for t in range(1, time_steps):
dec_inp = self.word_embeddings(topi.squeeze().detach())
cur_dec_out, dec_hid, cur_dec_attn = self.decoder(dec_inp, dec_hid, encoder_output, src_word_embeds,
src_mask, is_training)
cur_dec_out = cur_dec_out.view(-1, len(word_vocab))
if copy_on:
cur_dec_out.data.masked_fill_(trg_vocab_mask.data, -float('inf'))
cur_dec_out = F.log_softmax(cur_dec_out, dim=-1)
topv, topi = cur_dec_out.topk(1)
cur_dec_out_v, cur_dec_out_i = cur_dec_out.topk(1)
dec_out_i = torch.cat((dec_out_i, cur_dec_out_i), 1)
cur_dec_attn_v, cur_dec_attn_i = cur_dec_attn.topk(1)
dec_attn_i = torch.cat((dec_attn_i, cur_dec_attn_i), 1)
if is_training:
dec_out = dec_out.view(-1, len(word_vocab))
return dec_out
else:
return dec_out_i, dec_attn_i
def predict(samples, model, model_id):
pred_batch_size = batch_size
batch_count = math.ceil(len(samples) / pred_batch_size)
move_last_batch = False
if len(samples) - batch_size * (batch_count - 1) == 1:
move_last_batch = True
batch_count -= 1
preds = list()
attns = list()
model.eval()
set_random_seeds(random_seed)
start_time = datetime.datetime.now()
for batch_idx in tqdm(range(0, batch_count)):
batch_start = batch_idx * pred_batch_size
batch_end = min(len(samples), batch_start + pred_batch_size)
if batch_idx == batch_count - 1 and move_last_batch:
batch_end = len(samples)
cur_batch = samples[batch_start:batch_end]
cur_samples_input = get_batch_data(cur_batch, False)
src_words_seq = torch.from_numpy(cur_samples_input['src_words'].astype('long'))
src_words_mask = torch.from_numpy(cur_samples_input['src_words_mask'].astype('uint8'))
trg_vocab_mask = torch.from_numpy(cur_samples_input['trg_vocab_mask'].astype('uint8'))
trg_words_seq = torch.from_numpy(cur_samples_input['trg_words'].astype('long'))
adj = torch.from_numpy(cur_samples_input['adj'].astype('float32'))
src_chars_seq = torch.from_numpy(cur_samples_input['src_chars'].astype('long'))
if torch.cuda.is_available():
src_words_seq = src_words_seq.cuda()
src_words_mask = src_words_mask.cuda()
trg_vocab_mask = trg_vocab_mask.cuda()
trg_words_seq = trg_words_seq.cuda()
adj = adj.cuda()
src_chars_seq = src_chars_seq.cuda()
src_words_seq = autograd.Variable(src_words_seq)
src_words_mask = autograd.Variable(src_words_mask)
trg_vocab_mask = autograd.Variable(trg_vocab_mask)
adj = autograd.Variable(adj)
src_chars_seq = autograd.Variable(src_chars_seq)
trg_words_seq = autograd.Variable(trg_words_seq)
with torch.no_grad():
outputs = model(src_words_seq, src_chars_seq, src_words_mask, trg_words_seq, trg_vocab_mask, adj,False)
preds += list(outputs[0].data.cpu().numpy())
attns += list(outputs[1].data.cpu().numpy())
model.zero_grad()
end_time = datetime.datetime.now()
custom_print('Prediction time:', end_time - start_time)
return preds, attns
def train_model(model_id, train_samples, dev_samples, best_model_file):
train_size = len(train_samples)
batch_count = int(math.ceil(train_size/batch_size))
move_last_batch = False
if len(train_samples) - batch_size * (batch_count - 1) == 1:
move_last_batch = True
batch_count -= 1
custom_print(batch_count)
# model = get_model(model_id)
model = SeqToSeqModel()
pytorch_total_params = sum(p.numel() for p in model.parameters() if p.requires_grad)
custom_print('Parameters size:', pytorch_total_params)
custom_print(model)
if torch.cuda.is_available():
model.cuda()
if n_gpu > 1:
model = torch.nn.DataParallel(model)
criterion = nn.NLLLoss(ignore_index=0)
optimizer = optim.Adam(model.parameters())
custom_print(optimizer)
best_dev_acc = -1.0
best_epoch_idx = -1
best_epoch_seed = -1
for epoch_idx in range(0, num_epoch):
model.train()
model.zero_grad()
custom_print('Epoch:', epoch_idx + 1)
cur_seed = random_seed + epoch_idx + 1
set_random_seeds(cur_seed)
cur_shuffled_train_data = shuffle_data(train_samples)
start_time = datetime.datetime.now()
train_loss_val = 0.0
for batch_idx in tqdm(range(0, batch_count)):
batch_start = batch_idx * batch_size
batch_end = min(len(cur_shuffled_train_data), batch_start + batch_size)
if batch_idx == batch_count - 1 and move_last_batch:
batch_end = len(cur_shuffled_train_data)
cur_batch = cur_shuffled_train_data[batch_start:batch_end]
cur_samples_input = get_batch_data(cur_batch, True)
# np arrays to tensors
src_words_seq = torch.from_numpy(cur_samples_input['src_words'].astype('long'))
src_words_mask = torch.from_numpy(cur_samples_input['src_words_mask'].astype('uint8'))
trg_vocab_mask = torch.from_numpy(cur_samples_input['trg_vocab_mask'].astype('uint8'))
trg_words_seq = torch.from_numpy(cur_samples_input['trg_words'].astype('long'))
adj = torch.from_numpy(cur_samples_input['adj'].astype('float32'))
src_chars_seq = torch.from_numpy(cur_samples_input['src_chars'].astype('long'))
target = torch.from_numpy(cur_samples_input['target'].astype('long'))
if torch.cuda.is_available():
src_words_seq = src_words_seq.cuda()
src_words_mask = src_words_mask.cuda()
trg_vocab_mask = trg_vocab_mask.cuda()
trg_words_seq = trg_words_seq.cuda()
adj = adj.cuda()
src_chars_seq = src_chars_seq.cuda()
target = target.cuda()
src_words_seq = autograd.Variable(src_words_seq)
src_words_mask = autograd.Variable(src_words_mask)
trg_vocab_mask = autograd.Variable(trg_vocab_mask)
trg_words_seq = autograd.Variable(trg_words_seq)
adj = autograd.Variable(adj)
src_chars_seq = autograd.Variable(src_chars_seq)
target = autograd.Variable(target)
outputs = model(src_words_seq, src_chars_seq, src_words_mask, trg_words_seq, trg_vocab_mask, adj, True)
target = target.view(-1, 1).squeeze()
loss = criterion(outputs, target)
loss.backward()
torch.nn.utils.clip_grad_norm_(model.parameters(), 10.0)
if (batch_idx + 1) % update_freq == 0:
optimizer.step()
model.zero_grad()
train_loss_val += loss.item()
train_loss_val /= batch_count
end_time = datetime.datetime.now()
custom_print('Training loss:', train_loss_val)
custom_print('Training time:', end_time - start_time)
custom_print('\nDev Results\n')
set_random_seeds(random_seed)
dev_preds, dev_attns = predict(dev_samples, model, model_id)
write_test_res(dev_samples, dev_preds, dev_attns, os.path.join(trg_data_folder, 'dev.out'))
ref_lines = open(trg_dev_file).read().splitlines()
pred_lines = open(os.path.join(trg_data_folder, 'dev.out')).read().splitlines()
event_lines = open(events_file).read().splitlines()
argument_lines = open(arguments_file).read().splitlines()
roles_lines = open(roles_file).read().splitlines()
dev_acc = calculate_f1(ref_lines, pred_lines, event_lines, argument_lines, roles_lines)
# pred_pos, gt_pos, correct_pos = get_F1(dev_samples, dev_preds, dev_attns)
# custom_print(pred_pos, '\t', gt_pos, '\t', correct_pos)
# p = float(correct_pos) / (pred_pos + 1e-8)
# r = float(correct_pos) / (gt_pos + 1e-8)
# dev_acc = (2 * p * r) / (p + r + 1e-8)
# custom_print('F1:', dev_acc)
if dev_acc >= best_dev_acc:
best_epoch_idx = epoch_idx + 1
best_epoch_seed = cur_seed
custom_print('model saved......')
best_dev_acc = dev_acc
torch.save(model.state_dict(), best_model_file)
custom_print('\n\n')
if epoch_idx + 1 - best_epoch_idx >= early_stop_cnt:
break
custom_print('*******')
custom_print('Best Epoch:', best_epoch_idx)
custom_print('Best Epoch Seed:', best_epoch_seed)
if __name__ == "__main__":
os.environ['CUDA_VISIBLE_DEVICES'] = sys.argv[1]
random_seed = int(sys.argv[2])
src_data_folder = sys.argv[3]
trg_data_folder = sys.argv[4]
job_mode = sys.argv[5]
embedding_type = sys.argv[6]
granular_mode = 1
n_gpu = torch.cuda.device_count()
set_random_seeds(random_seed)
if not os.path.exists(trg_data_folder):
os.mkdir(trg_data_folder)
model_name = 1
#Tunable Hyperparameters
batch_size = 32
num_epoch = 30
max_src_len = 100
max_trg_len = 50
if embedding_type == 'w2v':
embedding_file = os.path.join(src_data_folder, 'w2v.txt')
else:
embedding_file = os.path.join(src_data_folder, 'Bert_embeddings.txt')
update_freq = 1
enc_type = ['LSTM', 'GCN', 'LSTM-GCN'][0]
att_type = ['None', 'Unigram', 'N-Gram-Enc'][1]
copy_on = True
gcn_num_layers = 3
if embedding_type == 'w2v':
word_embed_dim = 300
else:
word_embed_dim = 768
word_min_freq = 2
char_embed_dim = 50
char_feature_size = 50
conv_filter_size = 3
max_word_len = 10
enc_inp_size = word_embed_dim + char_feature_size
enc_hidden_size = word_embed_dim
dec_inp_size = enc_hidden_size
dec_hidden_size = dec_inp_size
drop_rate = 0.3
layers = 1
early_stop_cnt = 20
sample_cnt = 0
Sample = recordclass("Sample", "Id SrcLen SrcWords TrgLen TrgWords")
events_file = os.path.join(src_data_folder, 'event_types.txt')
arguments_file = os.path.join(src_data_folder, 'arguments.txt')
roles_file = os.path.join(src_data_folder, 'roles.txt')
events = get_relations(events_file)
arguments = get_relations(arguments_file)
roles = get_relations(roles_file)
# train a model
if job_mode == 'train':
logger = open(os.path.join(trg_data_folder, 'training.log'), 'w')
custom_print(sys.argv)
custom_print(max_src_len, max_trg_len, drop_rate, layers)
custom_print('loading data......')
model_file_name = os.path.join(trg_data_folder, 'model.h5py')
src_train_file = os.path.join(src_data_folder, 'train.sent')
trg_train_file = os.path.join(src_data_folder, 'train.tup')
train_data = read_data(src_train_file, trg_train_file, 1)
src_dev_file = os.path.join(src_data_folder, 'dev.sent')
trg_dev_file = os.path.join(src_data_folder, 'dev.tup')
dev_data = read_data(src_dev_file, trg_dev_file, 2)
custom_print('Training data size:', len(train_data))
custom_print('Development data size:', len(dev_data))
custom_print("preparing vocabulary......")
save_vocab = os.path.join(trg_data_folder, 'vocab.pkl')
word_vocab, rev_word_vocab, char_vocab, word_embed_matrix = build_vocab(train_data, events, arguments, roles, save_vocab,
embedding_file)
custom_print("Training started......")
train_model(model_name, train_data, dev_data, model_file_name)
logger.close()
if job_mode == 'test':
logger = open(os.path.join(trg_data_folder, 'test.log'), 'w')
custom_print(sys.argv)
custom_print("loading word vectors......")
vocab_file_name = os.path.join(trg_data_folder, 'vocab.pkl')
word_vocab, char_vocab = load_vocab(vocab_file_name)
rev_word_vocab = OrderedDict()
for word in word_vocab:
idx = word_vocab[word]
rev_word_vocab[idx] = word
word_embed_matrix = np.zeros((len(word_vocab), word_embed_dim), dtype=np.float32)
custom_print('vocab size:', len(word_vocab))
src_test_file = os.path.join(src_data_folder, 'test.sent')
trg_test_file = os.path.join(src_data_folder, 'test.tup')
test_data = read_data(src_test_file, trg_test_file, 3)
custom_print('Test data size:', len(test_data))
custom_print('seed:', random_seed)
model_file = os.path.join(trg_data_folder, 'model.h5py')
best_model = get_model(model_name)
custom_print(best_model)
if torch.cuda.is_available():
best_model.cuda()
if n_gpu > 1:
best_model = torch.nn.DataParallel(best_model)
best_model.load_state_dict(torch.load(model_file))
custom_print('\nTest Results\n')
set_random_seeds(random_seed)
test_preds, test_attns = predict(test_data, best_model, model_name)
custom_print('Copy On')
write_test_res(test_data, test_preds, test_attns, os.path.join(trg_data_folder, 'test.out'))
# ref_lines = open(trg_test_file).readlines()
# pred_lines = open(os.path.join(trg_data_folder, 'test.out')).readlines()
# event_lines = open(events_file).readlines()
# argument_lines = open(arguments_file).readlines()
# roles_lines = open(roles_file).readlines()
ref_lines = open(trg_test_file).read().splitlines()
pred_lines = open(os.path.join(trg_data_folder, 'test.out')).read().splitlines()
event_lines = open(events_file).read().splitlines()
argument_lines = open(arguments_file).read().splitlines()
roles_lines = open(roles_file).read().splitlines()
mode = 1
custom_print('Overall F1')
# custom_print(cal_f1(ref_lines, pred_lines, event_lines, argument_lines, roles_lines, mode))
calculate_f1(ref_lines, pred_lines, event_lines, argument_lines, roles_lines)
copy_on = False
custom_print('Copy Off')
set_random_seeds(random_seed)
test_preds, test_attns = predict(test_data, best_model, model_name)
write_test_res(test_data, test_preds, test_attns, os.path.join(trg_data_folder, 'test_without_copy.out'))
# ref_lines = open(trg_test_file).readlines()
# pred_lines = open(os.path.join(trg_data_folder, 'test_without_copy.out')).readlines()
# event_lines = open(events_file).readlines()
# argument_lines = open(arguments_file).readlines()
# roles_lines = open(roles_file).readlines()
ref_lines = open(trg_test_file).read().splitlines()
pred_lines = open(os.path.join(trg_data_folder, 'test_without_copy.out')).read().splitlines()
event_lines = open(events_file).read().splitlines()
argument_lines = open(arguments_file).read().splitlines()
roles_lines = open(roles_file).read().splitlines()
mode = 1
custom_print('Overall F1')
# custom_print(cal_f1(ref_lines, pred_lines, event_lines, argument_lines, roles_lines, mode))
calculate_f1(ref_lines, pred_lines, event_lines, argument_lines, roles_lines)
logger.close()
| [
"torch.nn.Dropout",
"recordclass.recordclass",
"pickle.dump",
"os.mkdir",
"numpy.random.seed",
"random.shuffle",
"torch.nn.Embedding",
"torch.nn.MaxPool1d",
"torch.nn.LSTMCell",
"torch.cat",
"torch.cuda.device_count",
"torch.nn.NLLLoss",
"pickle.load",
"torch.no_grad",
"os.path.join",
... | [((707, 720), 'collections.OrderedDict', 'OrderedDict', ([], {}), '()\n', (718, 720), False, 'from collections import OrderedDict\n'), ((743, 756), 'collections.OrderedDict', 'OrderedDict', ([], {}), '()\n', (754, 756), False, 'from collections import OrderedDict\n'), ((2338, 2351), 'collections.OrderedDict', 'OrderedDict', ([], {}), '()\n', (2349, 2351), False, 'from collections import OrderedDict\n'), ((2365, 2378), 'collections.OrderedDict', 'OrderedDict', ([], {}), '()\n', (2376, 2378), False, 'from collections import OrderedDict\n'), ((3166, 3203), 'pickle.dump', 'pickle.dump', (['[word_v, char_v]', 'output'], {}), '([word_v, char_v], output)\n', (3177, 3203), False, 'import pickle\n'), ((11979, 12030), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""Results_logger.txt"""'], {}), "(trg_data_folder, 'Results_logger.txt')\n", (11991, 12030), False, 'import os\n'), ((14512, 14529), 'random.seed', 'random.seed', (['seed'], {}), '(seed)\n', (14523, 14529), False, 'import random\n'), ((14534, 14554), 'numpy.random.seed', 'np.random.seed', (['seed'], {}), '(seed)\n', (14548, 14554), True, 'import numpy as np\n'), ((14559, 14582), 'torch.manual_seed', 'torch.manual_seed', (['seed'], {}), '(seed)\n', (14576, 14582), False, 'import torch\n'), ((25290, 25309), 'torch.sum', 'torch.sum', (['x'], {'dim': '(1)'}), '(x, dim=1)\n', (25299, 25309), False, 'import torch\n'), ((31644, 31667), 'datetime.datetime.now', 'datetime.datetime.now', ([], {}), '()\n', (31665, 31667), False, 'import datetime\n'), ((33522, 33545), 'datetime.datetime.now', 'datetime.datetime.now', ([], {}), '()\n', (33543, 33545), False, 'import datetime\n'), ((34229, 34254), 'torch.cuda.is_available', 'torch.cuda.is_available', ([], {}), '()\n', (34252, 34254), False, 'import torch\n'), ((34357, 34383), 'torch.nn.NLLLoss', 'nn.NLLLoss', ([], {'ignore_index': '(0)'}), '(ignore_index=0)\n', (34367, 34383), True, 'import torch.nn as nn\n'), ((39283, 39308), 'torch.cuda.device_count', 'torch.cuda.device_count', ([], {}), '()\n', (39306, 39308), False, 'import torch\n'), ((40385, 40444), 'recordclass.recordclass', 'recordclass', (['"""Sample"""', '"""Id SrcLen SrcWords TrgLen TrgWords"""'], {}), "('Sample', 'Id SrcLen SrcWords TrgLen TrgWords')\n", (40396, 40444), False, 'from recordclass import recordclass\n'), ((40464, 40512), 'os.path.join', 'os.path.join', (['src_data_folder', '"""event_types.txt"""'], {}), "(src_data_folder, 'event_types.txt')\n", (40476, 40512), False, 'import os\n'), ((40534, 40580), 'os.path.join', 'os.path.join', (['src_data_folder', '"""arguments.txt"""'], {}), "(src_data_folder, 'arguments.txt')\n", (40546, 40580), False, 'import os\n'), ((40598, 40640), 'os.path.join', 'os.path.join', (['src_data_folder', '"""roles.txt"""'], {}), "(src_data_folder, 'roles.txt')\n", (40610, 40640), False, 'import os\n'), ((870, 912), 'numpy.zeros', 'np.zeros', (['word_embed_dim'], {'dtype': 'np.float32'}), '(word_embed_dim, dtype=np.float32)\n', (878, 912), True, 'import numpy as np\n'), ((1001, 1047), 'numpy.random.uniform', 'np.random.uniform', (['(-0.25)', '(0.25)', 'word_embed_dim'], {}), '(-0.25, 0.25, word_embed_dim)\n', (1018, 1047), True, 'import numpy as np\n'), ((1136, 1182), 'numpy.random.uniform', 'np.random.uniform', (['(-0.25)', '(0.25)', 'word_embed_dim'], {}), '(-0.25, 0.25, word_embed_dim)\n', (1153, 1182), True, 'import numpy as np\n'), ((1271, 1317), 'numpy.random.uniform', 'np.random.uniform', (['(-0.25)', '(0.25)', 'word_embed_dim'], {}), '(-0.25, 0.25, word_embed_dim)\n', (1288, 1317), True, 'import numpy as np\n'), ((2210, 2250), 'numpy.array', 'np.array', (['embed_matrix'], {'dtype': 'np.float32'}), '(embed_matrix, dtype=np.float32)\n', (2218, 2250), True, 'import numpy as np\n'), ((3368, 3382), 'pickle.load', 'pickle.load', (['f'], {}), '(f)\n', (3379, 3382), False, 'import pickle\n'), ((14609, 14641), 'torch.cuda.manual_seed_all', 'torch.cuda.manual_seed_all', (['seed'], {}), '(seed)\n', (14635, 14641), False, 'import torch\n'), ((19269, 19311), 'numpy.array', 'np.array', (['src_words_list'], {'dtype': 'np.float32'}), '(src_words_list, dtype=np.float32)\n', (19277, 19311), True, 'import numpy as np\n'), ((19338, 19360), 'numpy.array', 'np.array', (['src_char_seq'], {}), '(src_char_seq)\n', (19346, 19360), True, 'import numpy as np\n'), ((19392, 19421), 'numpy.array', 'np.array', (['src_words_mask_list'], {}), '(src_words_mask_list)\n', (19400, 19421), True, 'import numpy as np\n'), ((19442, 19459), 'numpy.array', 'np.array', (['adj_lst'], {}), '(adj_lst)\n', (19450, 19459), True, 'import numpy as np\n'), ((19491, 19515), 'numpy.array', 'np.array', (['trg_vocab_mask'], {}), '(trg_vocab_mask)\n', (19499, 19515), True, 'import numpy as np\n'), ((19542, 19582), 'numpy.array', 'np.array', (['trg_words_list'], {'dtype': 'np.int32'}), '(trg_words_list, dtype=np.int32)\n', (19550, 19582), True, 'import numpy as np\n'), ((19606, 19622), 'numpy.array', 'np.array', (['target'], {}), '(target)\n', (19614, 19622), True, 'import numpy as np\n'), ((20715, 20765), 'torch.nn.Embedding', 'nn.Embedding', (['vocab_size', 'embed_dim'], {'padding_idx': '(0)'}), '(vocab_size, embed_dim, padding_idx=0)\n', (20727, 20765), True, 'import torch.nn as nn\n'), ((20875, 20900), 'torch.nn.Dropout', 'nn.Dropout', (['drop_out_rate'], {}), '(drop_out_rate)\n', (20885, 20900), True, 'import torch.nn as nn\n'), ((21350, 21400), 'torch.nn.Embedding', 'nn.Embedding', (['vocab_size', 'embed_dim'], {'padding_idx': '(0)'}), '(vocab_size, embed_dim, padding_idx=0)\n', (21362, 21400), True, 'import torch.nn as nn\n'), ((21424, 21449), 'torch.nn.Dropout', 'nn.Dropout', (['drop_out_rate'], {}), '(drop_out_rate)\n', (21434, 21449), True, 'import torch.nn as nn\n'), ((22135, 22247), 'torch.nn.LSTM', 'nn.LSTM', (['self.input_dim', 'self.hidden_dim', 'self.layers'], {'batch_first': '(True)', 'bidirectional': 'self.is_bidirectional'}), '(self.input_dim, self.hidden_dim, self.layers, batch_first=True,\n bidirectional=self.is_bidirectional)\n', (22142, 22247), True, 'import torch.nn as nn\n'), ((22295, 22321), 'torch.nn.Dropout', 'nn.Dropout', (['self.drop_rate'], {}), '(self.drop_rate)\n', (22305, 22321), True, 'import torch.nn as nn\n'), ((22344, 22406), 'torch.nn.Conv1d', 'nn.Conv1d', (['char_embed_dim', 'char_feature_size', 'conv_filter_size'], {}), '(char_embed_dim, char_feature_size, conv_filter_size)\n', (22353, 22406), True, 'import torch.nn as nn\n'), ((22431, 22521), 'torch.nn.MaxPool1d', 'nn.MaxPool1d', (['(max_word_len + conv_filter_size - 1)', '(max_word_len + conv_filter_size - 1)'], {}), '(max_word_len + conv_filter_size - 1, max_word_len +\n conv_filter_size - 1)\n', (22443, 22521), True, 'import torch.nn as nn\n'), ((22845, 22887), 'torch.cat', 'torch.cat', (['(words_input, char_feature)', '(-1)'], {}), '((words_input, char_feature), -1)\n', (22854, 22887), False, 'import torch\n'), ((23180, 23233), 'torch.nn.Linear', 'nn.Linear', (['self.input_dim', 'self.input_dim'], {'bias': '(False)'}), '(self.input_dim, self.input_dim, bias=False)\n', (23189, 23233), True, 'import torch.nn as nn\n'), ((23262, 23314), 'torch.nn.Linear', 'nn.Linear', (['self.input_dim', 'self.input_dim'], {'bias': '(True)'}), '(self.input_dim, self.input_dim, bias=True)\n', (23271, 23314), True, 'import torch.nn as nn\n'), ((23332, 23360), 'torch.nn.Linear', 'nn.Linear', (['self.input_dim', '(1)'], {}), '(self.input_dim, 1)\n', (23341, 23360), True, 'import torch.nn as nn\n'), ((23502, 23521), 'torch.tanh', 'torch.tanh', (['(wq + uh)'], {}), '(wq + uh)\n', (23512, 23521), False, 'import torch\n'), ((23660, 23691), 'torch.nn.functional.softmax', 'F.softmax', (['attn_weights'], {'dim': '(-1)'}), '(attn_weights, dim=-1)\n', (23669, 23691), True, 'import torch.nn.functional as F\n'), ((24004, 24019), 'torch.nn.ModuleList', 'nn.ModuleList', ([], {}), '()\n', (24017, 24019), True, 'import torch.nn as nn\n'), ((24044, 24059), 'torch.nn.ModuleList', 'nn.ModuleList', ([], {}), '()\n', (24057, 24059), True, 'import torch.nn as nn\n'), ((24440, 24462), 'torch.nn.functional.softmax', 'F.softmax', (['att'], {'dim': '(-1)'}), '(att, dim=-1)\n', (24449, 24462), True, 'import torch.nn.functional as F\n'), ((26225, 26251), 'torch.nn.Dropout', 'nn.Dropout', (['self.drop_rate'], {}), '(self.drop_rate)\n', (26235, 26251), True, 'import torch.nn as nn\n'), ((27283, 27310), 'torch.cat', 'torch.cat', (['(y_prev, ctx)', '(1)'], {}), '((y_prev, ctx), 1)\n', (27292, 27310), False, 'import torch\n'), ((32595, 32620), 'torch.cuda.is_available', 'torch.cuda.is_available', ([], {}), '()\n', (32618, 32620), False, 'import torch\n'), ((32925, 32957), 'torch.autograd.Variable', 'autograd.Variable', (['src_words_seq'], {}), '(src_words_seq)\n', (32942, 32957), True, 'import torch.autograd as autograd\n'), ((32983, 33016), 'torch.autograd.Variable', 'autograd.Variable', (['src_words_mask'], {}), '(src_words_mask)\n', (33000, 33016), True, 'import torch.autograd as autograd\n'), ((33042, 33075), 'torch.autograd.Variable', 'autograd.Variable', (['trg_vocab_mask'], {}), '(trg_vocab_mask)\n', (33059, 33075), True, 'import torch.autograd as autograd\n'), ((33090, 33112), 'torch.autograd.Variable', 'autograd.Variable', (['adj'], {}), '(adj)\n', (33107, 33112), True, 'import torch.autograd as autograd\n'), ((33137, 33169), 'torch.autograd.Variable', 'autograd.Variable', (['src_chars_seq'], {}), '(src_chars_seq)\n', (33154, 33169), True, 'import torch.autograd as autograd\n'), ((33195, 33227), 'torch.autograd.Variable', 'autograd.Variable', (['trg_words_seq'], {}), '(trg_words_seq)\n', (33212, 33227), True, 'import torch.autograd as autograd\n'), ((33761, 33795), 'math.ceil', 'math.ceil', (['(train_size / batch_size)'], {}), '(train_size / batch_size)\n', (33770, 33795), False, 'import math\n'), ((34311, 34339), 'torch.nn.DataParallel', 'torch.nn.DataParallel', (['model'], {}), '(model)\n', (34332, 34339), False, 'import torch\n'), ((34839, 34862), 'datetime.datetime.now', 'datetime.datetime.now', ([], {}), '()\n', (34860, 34862), False, 'import datetime\n'), ((37326, 37349), 'datetime.datetime.now', 'datetime.datetime.now', ([], {}), '()\n', (37347, 37349), False, 'import datetime\n'), ((39356, 39387), 'os.path.exists', 'os.path.exists', (['trg_data_folder'], {}), '(trg_data_folder)\n', (39370, 39387), False, 'import os\n'), ((39397, 39422), 'os.mkdir', 'os.mkdir', (['trg_data_folder'], {}), '(trg_data_folder)\n', (39405, 39422), False, 'import os\n'), ((39613, 39653), 'os.path.join', 'os.path.join', (['src_data_folder', '"""w2v.txt"""'], {}), "(src_data_folder, 'w2v.txt')\n", (39625, 39653), False, 'import os\n'), ((39689, 39741), 'os.path.join', 'os.path.join', (['src_data_folder', '"""Bert_embeddings.txt"""'], {}), "(src_data_folder, 'Bert_embeddings.txt')\n", (39701, 39741), False, 'import os\n'), ((41056, 41099), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""model.h5py"""'], {}), "(trg_data_folder, 'model.h5py')\n", (41068, 41099), False, 'import os\n'), ((41125, 41168), 'os.path.join', 'os.path.join', (['src_data_folder', '"""train.sent"""'], {}), "(src_data_folder, 'train.sent')\n", (41137, 41168), False, 'import os\n'), ((41194, 41236), 'os.path.join', 'os.path.join', (['src_data_folder', '"""train.tup"""'], {}), "(src_data_folder, 'train.tup')\n", (41206, 41236), False, 'import os\n'), ((41327, 41368), 'os.path.join', 'os.path.join', (['src_data_folder', '"""dev.sent"""'], {}), "(src_data_folder, 'dev.sent')\n", (41339, 41368), False, 'import os\n'), ((41392, 41432), 'os.path.join', 'os.path.join', (['src_data_folder', '"""dev.tup"""'], {}), "(src_data_folder, 'dev.tup')\n", (41404, 41432), False, 'import os\n'), ((41690, 41732), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""vocab.pkl"""'], {}), "(trg_data_folder, 'vocab.pkl')\n", (41702, 41732), False, 'import os\n'), ((42307, 42349), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""vocab.pkl"""'], {}), "(trg_data_folder, 'vocab.pkl')\n", (42319, 42349), False, 'import os\n'), ((42437, 42450), 'collections.OrderedDict', 'OrderedDict', ([], {}), '()\n', (42448, 42450), False, 'from collections import OrderedDict\n'), ((42726, 42768), 'os.path.join', 'os.path.join', (['src_data_folder', '"""test.sent"""'], {}), "(src_data_folder, 'test.sent')\n", (42738, 42768), False, 'import os\n'), ((42793, 42834), 'os.path.join', 'os.path.join', (['src_data_folder', '"""test.tup"""'], {}), "(src_data_folder, 'test.tup')\n", (42805, 42834), False, 'import os\n'), ((43020, 43063), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""model.h5py"""'], {}), "(trg_data_folder, 'model.h5py')\n", (43032, 43063), False, 'import os\n'), ((43152, 43177), 'torch.cuda.is_available', 'torch.cuda.is_available', ([], {}), '()\n', (43175, 43177), False, 'import torch\n'), ((4094, 4116), 'random.shuffle', 'random.shuffle', (['tuples'], {}), '(tuples)\n', (4108, 4116), False, 'import random\n'), ((20808, 20850), 'torch.from_numpy', 'torch.from_numpy', (['pre_trained_embed_matrix'], {}), '(pre_trained_embed_matrix)\n', (20824, 20850), False, 'import torch\n'), ((25055, 25079), 'torch.nn.functional.softmax', 'F.softmax', (['n_att'], {'dim': '(-1)'}), '(n_att, dim=-1)\n', (25064, 25079), True, 'import torch.nn.functional as F\n'), ((25808, 25869), 'torch.nn.LSTMCell', 'nn.LSTMCell', (['(2 * self.input_dim)', 'self.hidden_dim', 'self.layers'], {}), '(2 * self.input_dim, self.hidden_dim, self.layers)\n', (25819, 25869), True, 'import torch.nn as nn\n'), ((29049, 29079), 'torch.nn.functional.log_softmax', 'F.log_softmax', (['dec_out'], {'dim': '(-1)'}), '(dec_out, dim=-1)\n', (29062, 29079), True, 'import torch.nn.functional as F\n'), ((30006, 30036), 'torch.nn.functional.log_softmax', 'F.log_softmax', (['dec_out'], {'dim': '(-1)'}), '(dec_out, dim=-1)\n', (30019, 30036), True, 'import torch.nn.functional as F\n'), ((33241, 33256), 'torch.no_grad', 'torch.no_grad', ([], {}), '()\n', (33254, 33256), False, 'import torch\n'), ((36027, 36052), 'torch.cuda.is_available', 'torch.cuda.is_available', ([], {}), '()\n', (36050, 36052), False, 'import torch\n'), ((36425, 36457), 'torch.autograd.Variable', 'autograd.Variable', (['src_words_seq'], {}), '(src_words_seq)\n', (36442, 36457), True, 'import torch.autograd as autograd\n'), ((36487, 36520), 'torch.autograd.Variable', 'autograd.Variable', (['src_words_mask'], {}), '(src_words_mask)\n', (36504, 36520), True, 'import torch.autograd as autograd\n'), ((36550, 36583), 'torch.autograd.Variable', 'autograd.Variable', (['trg_vocab_mask'], {}), '(trg_vocab_mask)\n', (36567, 36583), True, 'import torch.autograd as autograd\n'), ((36612, 36644), 'torch.autograd.Variable', 'autograd.Variable', (['trg_words_seq'], {}), '(trg_words_seq)\n', (36629, 36644), True, 'import torch.autograd as autograd\n'), ((36663, 36685), 'torch.autograd.Variable', 'autograd.Variable', (['adj'], {}), '(adj)\n', (36680, 36685), True, 'import torch.autograd as autograd\n'), ((36714, 36746), 'torch.autograd.Variable', 'autograd.Variable', (['src_chars_seq'], {}), '(src_chars_seq)\n', (36731, 36746), True, 'import torch.autograd as autograd\n'), ((36769, 36794), 'torch.autograd.Variable', 'autograd.Variable', (['target'], {}), '(target)\n', (36786, 36794), True, 'import torch.autograd as autograd\n'), ((37682, 37722), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""dev.out"""'], {}), "(trg_data_folder, 'dev.out')\n", (37694, 37722), False, 'import os\n'), ((40838, 40883), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""training.log"""'], {}), "(trg_data_folder, 'training.log')\n", (40850, 40883), False, 'import os\n'), ((42151, 42192), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""test.log"""'], {}), "(trg_data_folder, 'test.log')\n", (42163, 42192), False, 'import os\n'), ((43256, 43289), 'torch.nn.DataParallel', 'torch.nn.DataParallel', (['best_model'], {}), '(best_model)\n', (43277, 43289), False, 'import torch\n'), ((43325, 43347), 'torch.load', 'torch.load', (['model_file'], {}), '(model_file)\n', (43335, 43347), False, 'import torch\n'), ((43596, 43637), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""test.out"""'], {}), "(trg_data_folder, 'test.out')\n", (43608, 43637), False, 'import os\n'), ((44750, 44804), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""test_without_copy.out"""'], {}), "(trg_data_folder, 'test_without_copy.out')\n", (44762, 44804), False, 'import os\n'), ((1945, 1991), 'numpy.random.uniform', 'np.random.uniform', (['(-0.25)', '(0.25)', 'word_embed_dim'], {}), '(-0.25, 0.25, word_embed_dim)\n', (1962, 1991), True, 'import numpy as np\n'), ((24120, 24151), 'torch.nn.Linear', 'nn.Linear', (['input_dim', 'input_dim'], {}), '(input_dim, input_dim)\n', (24129, 24151), True, 'import torch.nn as nn\n'), ((24186, 24217), 'torch.nn.Linear', 'nn.Linear', (['input_dim', 'input_dim'], {}), '(input_dim, input_dim)\n', (24195, 24217), True, 'import torch.nn as nn\n'), ((25980, 26041), 'torch.nn.LSTMCell', 'nn.LSTMCell', (['(2 * self.input_dim)', 'self.hidden_dim', 'self.layers'], {}), '(2 * self.input_dim, self.hidden_dim, self.layers)\n', (25991, 26041), True, 'import torch.nn as nn\n'), ((26139, 26200), 'torch.nn.LSTMCell', 'nn.LSTMCell', (['(3 * self.input_dim)', 'self.hidden_dim', 'self.layers'], {}), '(3 * self.input_dim, self.hidden_dim, self.layers)\n', (26150, 26200), True, 'import torch.nn as nn\n'), ((27201, 27231), 'torch.cat', 'torch.cat', (['(enc_last, ctx)', '(-1)'], {}), '((enc_last, ctx), -1)\n', (27210, 27231), False, 'import torch\n'), ((28474, 28512), 'torch.zeros', 'torch.zeros', (['batch_len', 'word_embed_dim'], {}), '(batch_len, word_embed_dim)\n', (28485, 28512), False, 'import torch\n'), ((28587, 28625), 'torch.zeros', 'torch.zeros', (['batch_len', 'word_embed_dim'], {}), '(batch_len, word_embed_dim)\n', (28598, 28625), False, 'import torch\n'), ((30717, 30751), 'torch.nn.functional.log_softmax', 'F.log_softmax', (['cur_dec_out'], {'dim': '(-1)'}), '(cur_dec_out, dim=-1)\n', (30730, 30751), True, 'import torch.nn.functional as F\n'), ((30896, 30936), 'torch.cat', 'torch.cat', (['(dec_out_i, cur_dec_out_i)', '(1)'], {}), '((dec_out_i, cur_dec_out_i), 1)\n', (30905, 30936), False, 'import torch\n'), ((31036, 31078), 'torch.cat', 'torch.cat', (['(dec_attn_i, cur_dec_attn_i)', '(1)'], {}), '((dec_attn_i, cur_dec_attn_i), 1)\n', (31045, 31078), False, 'import torch\n'), ((1617, 1632), 'numpy.float32', 'np.float32', (['val'], {}), '(val)\n', (1627, 1632), True, 'import numpy as np\n'), ((3643, 3666), 'math.pow', 'math.pow', (['(2)', 'amat[i][j]'], {}), '(2, amat[i][j])\n', (3651, 3666), False, 'import math\n'), ((24608, 24636), 'torch.nn.AvgPool1d', 'torch.nn.AvgPool1d', (['(i + 1)', '(1)'], {}), '(i + 1, 1)\n', (24626, 24636), False, 'import torch\n'), ((24747, 24775), 'torch.nn.AvgPool1d', 'torch.nn.AvgPool1d', (['(i + 1)', '(1)'], {}), '(i + 1, 1)\n', (24765, 24775), False, 'import torch\n'), ((27053, 27088), 'torch.gather', 'torch.gather', (['enc_hs', '(1)', 'last_index'], {}), '(enc_hs, 1, last_index)\n', (27065, 27088), False, 'import torch\n'), ((37810, 37850), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""dev.out"""'], {}), "(trg_data_folder, 'dev.out')\n", (37822, 37850), False, 'import os\n'), ((44031, 44072), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""test.out"""'], {}), "(trg_data_folder, 'test.out')\n", (44043, 44072), False, 'import os\n'), ((45211, 45265), 'os.path.join', 'os.path.join', (['trg_data_folder', '"""test_without_copy.out"""'], {}), "(trg_data_folder, 'test_without_copy.out')\n", (45223, 45265), False, 'import os\n'), ((26897, 26924), 'torch.sum', 'torch.sum', (['src_mask'], {'dim': '(-1)'}), '(src_mask, dim=-1)\n', (26906, 26924), False, 'import torch\n'), ((29530, 29564), 'torch.nn.functional.log_softmax', 'F.log_softmax', (['cur_dec_out'], {'dim': '(-1)'}), '(cur_dec_out, dim=-1)\n', (29543, 29564), True, 'import torch.nn.functional as F\n')] |
import numpy as np
import cv2
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import pickle
class Line():
def __init__(self,n):
self.n=n
self.detected =False
#Polynomial coefficients of the lines
self.A=[]
self.B=[]
self.C=[]
#Running average of coefficients
self.A_avg=0.
self.B_avg=0.
self.C_avg=0.
def obtain_fit(self):
return (self.A_avg,self.B_avg,self.C_avg)
def update_fit(self,fit_coeffs):
"""Obtain the fit coefficients from the latest frame and apply over each of 2nd polynomial coefficients
for the purpose of smoothing
"""
full_Q= len(self.A) >= self.n
#Append line fit coefficients
self.A.append(fit_coeffs[0])
self.B.append(fit_coeffs[1])
self.C.append(fit_coeffs[2])
if full_Q:
_=self.A.pop(0)
_=self.B.pop(0)
_=self.C.pop(0)
# Compute the average of the polynomial coefficients
self.A_avg = np.mean(self.A)
self.B_avg = np.mean(self.B)
self.C_avg = np.mean(self.C)
return (self.A_avg,self.B_avg,self.C_avg)
| [
"numpy.mean"
] | [((950, 965), 'numpy.mean', 'np.mean', (['self.A'], {}), '(self.A)\n', (957, 965), True, 'import numpy as np\n'), ((982, 997), 'numpy.mean', 'np.mean', (['self.B'], {}), '(self.B)\n', (989, 997), True, 'import numpy as np\n'), ((1014, 1029), 'numpy.mean', 'np.mean', (['self.C'], {}), '(self.C)\n', (1021, 1029), True, 'import numpy as np\n')] |
# Copyright 2016 <NAME>, alexggmatthews
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ------------------------------------------
# Modification notice:
# This file was modified by <NAME>
# ------------------------------------------
import tensorflow as tf
from settings import float_type
from quadrature import hermgauss
import numpy as np
def eye(N):
"""
An identitiy matrix
"""
return tf.diag(tf.ones(tf.stack([N, ]), dtype=float_type))
def variational_expectations( Fmu, Fvar, phi, num_gauss_hermite_points=20):
"""
Compute the expected value of a function phi, given a Gaussian
distribution for the input values.
if
q(f) = N(Fmu, Fvar)
then this method computes
\int phi(f) q(f) df.
Here, we implement a default Gauss-Hermite quadrature routine
"""
gh_x, gh_w = hermgauss(num_gauss_hermite_points)
gh_x = gh_x.reshape(1, -1)
gh_w = gh_w.reshape(-1, 1) / np.sqrt(np.pi)
shape = tf.shape(Fmu)
Fmu, Fvar = [tf.reshape(e, (-1, 1)) for e in (Fmu, Fvar)]
X = gh_x * tf.sqrt(2.0 * Fvar) + Fmu
logp = phi(X)
return tf.reshape(tf.matmul(logp, gh_w), shape)
import tensorflow as tf
def block_diagonal(matrices, dtype=tf.float32):
"""Constructs block-diagonal matrices from a list of batched 2D tensors.
Args:
matrices: A list of Tensors with shape [..., N_i, M_i] (i.e. a list of
matrices with the same batch dimension).
dtype: Data type to use. The Tensors in `matrices` must match this dtype.
Returns:
A matrix with the input matrices stacked along its main diagonal, having
shape [..., \sum_i N_i, \sum_i M_i].
"""
matrices = [tf.convert_to_tensor(matrix, dtype=dtype) for matrix in matrices]
blocked_rows = tf.Dimension(0)
blocked_cols = tf.Dimension(0)
batch_shape = tf.TensorShape(None)
for matrix in matrices:
full_matrix_shape = matrix.get_shape().with_rank_at_least(2)
batch_shape = batch_shape.merge_with(full_matrix_shape[:-2])
blocked_rows += full_matrix_shape[-2]
blocked_cols += full_matrix_shape[-1]
ret_columns_list = []
for matrix in matrices:
matrix_shape = tf.shape(matrix)
ret_columns_list.append(matrix_shape[-1])
ret_columns = tf.add_n(ret_columns_list)
row_blocks = []
current_column = 0
for matrix in matrices:
matrix_shape = tf.shape(matrix)
row_before_length = current_column
current_column += matrix_shape[-1]
row_after_length = ret_columns - current_column
row_blocks.append(tf.pad(
tensor=matrix,
paddings=tf.concat(
[tf.zeros([tf.rank(matrix) - 1, 2], dtype=tf.int32),
[(row_before_length, row_after_length)]],
axis=0)))
blocked = tf.concat(row_blocks, -2)
blocked.set_shape(batch_shape.concatenate((blocked_rows, blocked_cols)))
return blocked | [
"tensorflow.Dimension",
"tensorflow.add_n",
"quadrature.hermgauss",
"tensorflow.convert_to_tensor",
"tensorflow.reshape",
"tensorflow.TensorShape",
"tensorflow.concat",
"tensorflow.stack",
"tensorflow.matmul",
"tensorflow.shape",
"tensorflow.sqrt",
"tensorflow.rank",
"numpy.sqrt"
] | [((1333, 1368), 'quadrature.hermgauss', 'hermgauss', (['num_gauss_hermite_points'], {}), '(num_gauss_hermite_points)\n', (1342, 1368), False, 'from quadrature import hermgauss\n'), ((1460, 1473), 'tensorflow.shape', 'tf.shape', (['Fmu'], {}), '(Fmu)\n', (1468, 1473), True, 'import tensorflow as tf\n'), ((2263, 2278), 'tensorflow.Dimension', 'tf.Dimension', (['(0)'], {}), '(0)\n', (2275, 2278), True, 'import tensorflow as tf\n'), ((2298, 2313), 'tensorflow.Dimension', 'tf.Dimension', (['(0)'], {}), '(0)\n', (2310, 2313), True, 'import tensorflow as tf\n'), ((2332, 2352), 'tensorflow.TensorShape', 'tf.TensorShape', (['None'], {}), '(None)\n', (2346, 2352), True, 'import tensorflow as tf\n'), ((2773, 2799), 'tensorflow.add_n', 'tf.add_n', (['ret_columns_list'], {}), '(ret_columns_list)\n', (2781, 2799), True, 'import tensorflow as tf\n'), ((3314, 3339), 'tensorflow.concat', 'tf.concat', (['row_blocks', '(-2)'], {}), '(row_blocks, -2)\n', (3323, 3339), True, 'import tensorflow as tf\n'), ((1433, 1447), 'numpy.sqrt', 'np.sqrt', (['np.pi'], {}), '(np.pi)\n', (1440, 1447), True, 'import numpy as np\n'), ((1491, 1513), 'tensorflow.reshape', 'tf.reshape', (['e', '(-1, 1)'], {}), '(e, (-1, 1))\n', (1501, 1513), True, 'import tensorflow as tf\n'), ((1617, 1638), 'tensorflow.matmul', 'tf.matmul', (['logp', 'gh_w'], {}), '(logp, gh_w)\n', (1626, 1638), True, 'import tensorflow as tf\n'), ((2178, 2219), 'tensorflow.convert_to_tensor', 'tf.convert_to_tensor', (['matrix'], {'dtype': 'dtype'}), '(matrix, dtype=dtype)\n', (2198, 2219), True, 'import tensorflow as tf\n'), ((2688, 2704), 'tensorflow.shape', 'tf.shape', (['matrix'], {}), '(matrix)\n', (2696, 2704), True, 'import tensorflow as tf\n'), ((2894, 2910), 'tensorflow.shape', 'tf.shape', (['matrix'], {}), '(matrix)\n', (2902, 2910), True, 'import tensorflow as tf\n'), ((921, 934), 'tensorflow.stack', 'tf.stack', (['[N]'], {}), '([N])\n', (929, 934), True, 'import tensorflow as tf\n'), ((1551, 1570), 'tensorflow.sqrt', 'tf.sqrt', (['(2.0 * Fvar)'], {}), '(2.0 * Fvar)\n', (1558, 1570), True, 'import tensorflow as tf\n'), ((3173, 3188), 'tensorflow.rank', 'tf.rank', (['matrix'], {}), '(matrix)\n', (3180, 3188), True, 'import tensorflow as tf\n')] |
# Copyright 2019 DeepMind Technologies Limited
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Tests for open_spiel.python.algorithms.double_oracle."""
from absl.testing import absltest
import numpy as np
from open_spiel.python.algorithms import double_oracle
import pyspiel
class DoubleOracleTest(absltest.TestCase):
def test_rock_paper_scissors(self):
game = pyspiel.load_matrix_game("matrix_rps")
solver = double_oracle.DoubleOracleSolver(game)
solution, iteration, value = solver.solve(initial_strategies=[[0], [0]])
np.testing.assert_allclose(solution[0], np.ones(3)/3.)
np.testing.assert_allclose(solution[1], np.ones(3)/3.)
self.assertEqual(iteration, 3)
self.assertAlmostEqual(value, 0.0)
def test_single_step(self):
game = pyspiel.load_matrix_game("matrix_rps")
solver = double_oracle.DoubleOracleSolver(game)
solver.subgame_strategies = [[0], [0]]
best_response, best_response_utility = solver.step()
self.assertListEqual(best_response, [1, 1])
self.assertListEqual(best_response_utility, [1.0, 1.0])
def test_kuhn_poker(self):
game = pyspiel.extensive_to_matrix_game(pyspiel.load_game("kuhn_poker"))
solver = double_oracle.DoubleOracleSolver(game)
solution, iteration, value = solver.solve(initial_strategies=[[0], [0]])
# check if solution is Nash
exp_utilty = solution[0] @ solver.payoffs @ solution[1]
self.assertAlmostEqual(max(solver.payoffs[0] @ solution[1]), exp_utilty[0])
self.assertAlmostEqual(max(solution[0] @ solver.payoffs[1]), exp_utilty[1])
self.assertEqual(iteration, 8)
self.assertAlmostEqual(value, 0.0)
if __name__ == "__main__":
absltest.main()
| [
"pyspiel.load_game",
"absl.testing.absltest.main",
"numpy.ones",
"pyspiel.load_matrix_game",
"open_spiel.python.algorithms.double_oracle.DoubleOracleSolver"
] | [((2171, 2186), 'absl.testing.absltest.main', 'absltest.main', ([], {}), '()\n', (2184, 2186), False, 'from absl.testing import absltest\n'), ((875, 913), 'pyspiel.load_matrix_game', 'pyspiel.load_matrix_game', (['"""matrix_rps"""'], {}), "('matrix_rps')\n", (899, 913), False, 'import pyspiel\n'), ((927, 965), 'open_spiel.python.algorithms.double_oracle.DoubleOracleSolver', 'double_oracle.DoubleOracleSolver', (['game'], {}), '(game)\n', (959, 965), False, 'from open_spiel.python.algorithms import double_oracle\n'), ((1277, 1315), 'pyspiel.load_matrix_game', 'pyspiel.load_matrix_game', (['"""matrix_rps"""'], {}), "('matrix_rps')\n", (1301, 1315), False, 'import pyspiel\n'), ((1329, 1367), 'open_spiel.python.algorithms.double_oracle.DoubleOracleSolver', 'double_oracle.DoubleOracleSolver', (['game'], {}), '(game)\n', (1361, 1367), False, 'from open_spiel.python.algorithms import double_oracle\n'), ((1696, 1734), 'open_spiel.python.algorithms.double_oracle.DoubleOracleSolver', 'double_oracle.DoubleOracleSolver', (['game'], {}), '(game)\n', (1728, 1734), False, 'from open_spiel.python.algorithms import double_oracle\n'), ((1650, 1681), 'pyspiel.load_game', 'pyspiel.load_game', (['"""kuhn_poker"""'], {}), "('kuhn_poker')\n", (1667, 1681), False, 'import pyspiel\n'), ((1087, 1097), 'numpy.ones', 'np.ones', (['(3)'], {}), '(3)\n', (1094, 1097), True, 'import numpy as np\n'), ((1146, 1156), 'numpy.ones', 'np.ones', (['(3)'], {}), '(3)\n', (1153, 1156), True, 'import numpy as np\n')] |
import numpy as np
import torch
from torch import nn
import torch.nn.functional as F
# open text file and read in data as `text`
with open('data/anna.txt', 'r') as f:
text = f.read()
# print(text[:100])
# encode the text and map each character to an integer and vice versa
# we create two dictionaries:
# 1. int2char, which maps integers to characters
# 2. char2int, which maps characters to unique integers
# text = text[:100]
chars = tuple(set(text)) #(1', 'v', 'H', '.', 'i', 'E', 'a', 'r', 'C', 'p',...)
int2char = dict(enumerate(chars))
char2int = {ch: ii for ii, ch in int2char.items()}
# encode the text
encoded = np.array([char2int[ch] for ch in text])
print(encoded[:100])
def one_hot_encode(arr, n_labels):
# Initialize the the encoded array
# arr is a multi-dim array
one_hot = np.zeros((np.multiply(*arr.shape), n_labels), dtype=np.float32)
# Fill the appropriate elements with ones
one_hot[np.arange(one_hot.shape[0]), arr.flatten()] = 1.
# Finally reshape it to get back to the original array
one_hot = one_hot.reshape((*arr.shape, n_labels))
return one_hot
# check that the function works as expected
test_seq = np.array([[3, 5, 1]])
one_hot = np.zeros((np.multiply(*test_seq.shape), 8), dtype=np.float32)
# one_hot = one_hot_encode(test_seq, 8)
print(one_hot)
def get_batches(arr, batch_size, seq_length):
'''Create a generator that returns batches of size
batch_size x seq_length from arr.
Arguments
---------
arr: Array you want to make batches from
batch_size: Batch size, the number of sequences per batch
seq_length: Number of encoded chars in a sequence
'''
total = batch_size * seq_length
n_batches = len(arr) // total
arr = arr[:n_batches * total]
arr = arr.reshape(batch_size, -1)
for n in range(0, arr.shape[1], seq_length):
# The features
x = arr[:,n:n+seq_length]
# The targets, shifted by one
y = np.zeros_like(x)
try:
y[:, :-1], y[:, -1] = x[:, 1:], arr[:, n+seq_length]
except IndexError:
y[:, :-1], y[:, -1] = x[:, 1:], arr[:, 0]
yield x, y
batches = get_batches(encoded, 8, 50)
x, y = next(batches)
# printing out the first 10 items in a sequence
print('x\n', x[:10, :10])
print('\ny\n', y[:10, :10])
# check if GPU is available
train_on_gpu = torch.cuda.is_available()
if(train_on_gpu):
print('Training on GPU!')
else:
print('No GPU available, training on CPU; consider making n_epochs very small.')
class CharRNN(nn.Module):
def __init__(self, tokens, n_hidden=256, n_layers=2,
drop_prob=0.5, lr=0.001):
super().__init__()
self.drop_prob = drop_prob
self.n_layers = n_layers
self.n_hidden = n_hidden
self.lr = lr
# creating character dictionaries
self.chars = tokens
self.int2char = dict(enumerate(self.chars))
self.char2int = {ch: ii for ii, ch in self.int2char.items()}
## TODO: define the layers of the model
self.lstm = nn.LSTM(len(tokens), n_hidden, n_layers, dropout=drop_prob, batch_first=True)
self.dropout = nn.Dropout(drop_prob)
self.fc = nn.Linear (n_hidden, len(tokens))
def forward(self, x, hidden):
''' Forward pass through the network.
These inputs are x, and the hidden/cell state `hidden`. '''
## TODO: Get the outputs and the new hidden state from the lstm
x, hidden = self.lstm(x,hidden)
x = self.dropout(x)
x = x.contiguous().view(-1, n_hidden)
x = self.fc(x)
# return the final output and the hidden state
return x, hidden
def init_hidden(self, batch_size):
''' Initializes hidden state '''
# Create two new tensors with sizes n_layers x batch_size x n_hidden,
# initialized to zero, for hidden state and cell state of LSTM
weight = next(self.parameters()).data
if (train_on_gpu):
hidden = (weight.new(self.n_layers, batch_size, self.n_hidden).zero_().cuda(),
weight.new(self.n_layers, batch_size, self.n_hidden).zero_().cuda())
else:
hidden = (weight.new(self.n_layers, batch_size, self.n_hidden).zero_(),
weight.new(self.n_layers, batch_size, self.n_hidden).zero_())
return hidden
def train(net, data, epochs=10, batch_size=10, seq_length=50, lr=0.001, clip=5, val_frac=0.1, print_every=10):
''' Training a network
Arguments
---------
net: CharRNN network
data: text data to train the network
epochs: Number of epochs to train
batch_size: Number of mini-sequences per mini-batch, aka batch size
seq_length: Number of character steps per mini-batch
lr: learning rate
clip: gradient clipping
val_frac: Fraction of data to hold out for validation
print_every: Number of steps for printing training and validation loss
'''
net.train()
opt = torch.optim.Adam(net.parameters(), lr=lr)
criterion = nn.CrossEntropyLoss()
# create training and validation data
val_idx = int(len(data)*(1-val_frac))
data, val_data = data[:val_idx], data[val_idx:]
if(train_on_gpu):
net.cuda()
counter = 0
n_chars = len(net.chars)
for e in range(epochs):
# initialize hidden state
h = net.init_hidden(batch_size)
for x, y in get_batches(data, batch_size, seq_length):
counter += 1
# One-hot encode our data and make them Torch tensors
x = one_hot_encode(x, n_chars)
inputs, targets = torch.from_numpy(x), torch.from_numpy(y)
if(train_on_gpu):
inputs, targets = inputs.cuda(), targets.cuda()
# Creating new variables for the hidden state, otherwise
# we'd backprop through the entire training history
h = tuple([each.data for each in h])
# zero accumulated gradients
net.zero_grad()
# get the output from the model
output, h = net(inputs, h)
# calculate the loss and perform backprop
loss = criterion(output, targets.view(batch_size*seq_length))
loss.backward()
# `clip_grad_norm` helps prevent the exploding gradient problem in RNNs / LSTMs.
nn.utils.clip_grad_norm_(net.parameters(), clip)
opt.step()
# loss stats
if counter % print_every == 0:
# Get validation loss
val_h = net.init_hidden(batch_size)
val_losses = []
net.eval()
for x, y in get_batches(val_data, batch_size, seq_length):
# One-hot encode our data and make them Torch tensors
x = one_hot_encode(x, n_chars)
x, y = torch.from_numpy(x), torch.from_numpy(y)
# Creating new variables for the hidden state, otherwise
# we'd backprop through the entire training history
val_h = tuple([each.data for each in val_h])
inputs, targets = x, y
if(train_on_gpu):
inputs, targets = inputs.cuda(), targets.cuda()
output, val_h = net(inputs, val_h)
val_loss = criterion(output, targets.view(batch_size*seq_length))
val_losses.append(val_loss.item())
net.train() # reset to train mode after iterationg through validation data
print("Epoch: {}/{}...".format(e+1, epochs),
"Step: {}...".format(counter),
"Loss: {:.4f}...".format(loss.item()),
"Val Loss: {:.4f}".format(np.mean(val_losses)))
n_hidden=512
n_layers=2
net = CharRNN(chars, n_hidden, n_layers)
print(net)
batch_size = 128
seq_length = 100
n_epochs = 20 # start smaller if you are just testing initial behavior
# train the model
train(net, encoded, epochs=n_epochs, batch_size=batch_size, seq_length=seq_length, lr=0.001, print_every=10)
# change the name, for saving multiple files
model_name = 'rnn_20_epoch.net'
checkpoint = {'n_hidden': net.n_hidden,
'n_layers': net.n_layers,
'state_dict': net.state_dict(),
'tokens': net.chars}
with open(model_name, 'wb') as f:
torch.save(checkpoint, f)
## Making Predictions
def predict(net, char, h=None, top_k=None):
''' Given a character, predict the next character.
Returns the predicted character and the hidden state.
'''
# tensor inputs
x = np.array([[net.char2int[char]]])
x = one_hot_encode(x, len(net.chars))
inputs = torch.from_numpy(x)
if(train_on_gpu):
inputs = inputs.cuda()
# detach hidden state from history
h = tuple([each.data for each in h])
# get the output of the model
out, h = net(inputs, h)
# get the character probabilities
p = F.softmax(out, dim=1).data
if(train_on_gpu):
p = p.cpu() # move to cpu
# get top characters
if top_k is None:
top_ch = np.arange(len(net.chars))
else:
p, top_ch = p.topk(top_k)
top_ch = top_ch.numpy().squeeze()
# select the likely next character with some element of randomness
p = p.numpy().squeeze()
char = np.random.choice(top_ch, p=p/p.sum())
# return the encoded value of the predicted char and the hidden state
return net.int2char[char], h
def sample(net, size, prime='The', top_k=None):
#prime is the arg that we want to start our model with
if(train_on_gpu):
net.cuda()
else:
net.cpu()
net.eval() # eval mode
# First off, run through the prime characters
chars = [ch for ch in prime]
h = net.init_hidden(1)
for ch in prime:
char, h = predict(net, ch, h, top_k=top_k)
chars.append(char)
# Now pass in the previous character and get a new one
for ii in range(size):
char, h = predict(net, chars[-1], h, top_k=top_k)
chars.append(char)
return ''.join(chars)
print(sample(net, 1000, prime='Anna', top_k=5))
## Loading a checkpoint
with open('rnn_20_epoch.net', 'rb') as f:
checkpoint = torch.load(f)
loaded = CharRNN(checkpoint['tokens'], n_hidden=checkpoint['n_hidden'], n_layers=checkpoint['n_layers'])
loaded.load_state_dict(checkpoint['state_dict'])
# Sample using a loaded model
print(sample(loaded, 2000, top_k=5, prime="And Levin said")) | [
"torch.nn.Dropout",
"numpy.zeros_like",
"numpy.multiply",
"torch.load",
"torch.nn.CrossEntropyLoss",
"torch.nn.functional.softmax",
"torch.save",
"numpy.mean",
"torch.cuda.is_available",
"numpy.array",
"numpy.arange",
"torch.from_numpy"
] | [((630, 669), 'numpy.array', 'np.array', (['[char2int[ch] for ch in text]'], {}), '([char2int[ch] for ch in text])\n', (638, 669), True, 'import numpy as np\n'), ((1170, 1191), 'numpy.array', 'np.array', (['[[3, 5, 1]]'], {}), '([[3, 5, 1]])\n', (1178, 1191), True, 'import numpy as np\n'), ((2383, 2408), 'torch.cuda.is_available', 'torch.cuda.is_available', ([], {}), '()\n', (2406, 2408), False, 'import torch\n'), ((5229, 5250), 'torch.nn.CrossEntropyLoss', 'nn.CrossEntropyLoss', ([], {}), '()\n', (5248, 5250), False, 'from torch import nn\n'), ((8780, 8805), 'torch.save', 'torch.save', (['checkpoint', 'f'], {}), '(checkpoint, f)\n', (8790, 8805), False, 'import torch\n'), ((9060, 9092), 'numpy.array', 'np.array', (['[[net.char2int[char]]]'], {}), '([[net.char2int[char]]])\n', (9068, 9092), True, 'import numpy as np\n'), ((9156, 9175), 'torch.from_numpy', 'torch.from_numpy', (['x'], {}), '(x)\n', (9172, 9175), False, 'import torch\n'), ((10830, 10843), 'torch.load', 'torch.load', (['f'], {}), '(f)\n', (10840, 10843), False, 'import torch\n'), ((1212, 1240), 'numpy.multiply', 'np.multiply', (['*test_seq.shape'], {}), '(*test_seq.shape)\n', (1223, 1240), True, 'import numpy as np\n'), ((1982, 1998), 'numpy.zeros_like', 'np.zeros_like', (['x'], {}), '(x)\n', (1995, 1998), True, 'import numpy as np\n'), ((3226, 3247), 'torch.nn.Dropout', 'nn.Dropout', (['drop_prob'], {}), '(drop_prob)\n', (3236, 3247), False, 'from torch import nn\n'), ((9468, 9489), 'torch.nn.functional.softmax', 'F.softmax', (['out'], {'dim': '(1)'}), '(out, dim=1)\n', (9477, 9489), True, 'import torch.nn.functional as F\n'), ((821, 844), 'numpy.multiply', 'np.multiply', (['*arr.shape'], {}), '(*arr.shape)\n', (832, 844), True, 'import numpy as np\n'), ((933, 960), 'numpy.arange', 'np.arange', (['one_hot.shape[0]'], {}), '(one_hot.shape[0])\n', (942, 960), True, 'import numpy as np\n'), ((5839, 5858), 'torch.from_numpy', 'torch.from_numpy', (['x'], {}), '(x)\n', (5855, 5858), False, 'import torch\n'), ((5860, 5879), 'torch.from_numpy', 'torch.from_numpy', (['y'], {}), '(y)\n', (5876, 5879), False, 'import torch\n'), ((7139, 7158), 'torch.from_numpy', 'torch.from_numpy', (['x'], {}), '(x)\n', (7155, 7158), False, 'import torch\n'), ((7160, 7179), 'torch.from_numpy', 'torch.from_numpy', (['y'], {}), '(y)\n', (7176, 7179), False, 'import torch\n'), ((8151, 8170), 'numpy.mean', 'np.mean', (['val_losses'], {}), '(val_losses)\n', (8158, 8170), True, 'import numpy as np\n')] |
"""
Dataset classes for variable number of speakers
Author: <NAME>
"""
import numpy as np
import torch
import torch.utils.data as data
from librosa import load
from time import time
import glob
import os
import random
import json
from tqdm import tqdm
def load_json(filename):
with open(filename) as f:
data = json.load(f)
return data
def pad_audio(audio, len_samples=4*8000):
if len(audio) < len_samples:
audio = np.concatenate([audio, np.zeros(len_samples - len(audio))])
return audio
class MixtureDataset(data.Dataset):
def __init__(self, root, json_folders, sr=8000, seglen=4.0, minlen=2.0, debug=False): # segment and cv_maxlen not implemented
"""
each line of textfile comes in the form of:
filename1, dB1, filename2, dB2, ...
args:
root: folder where dataset/ is located
json_folders: folders containing json files, **/dataset/#speakers/wav8k/min/tr/**
sr: sample rate
seglen: length of each segment in seconds
minlen: minimum segment length
"""
str_tmp = '_debug' if debug else ''
seglen = int(seglen * sr)
minlen = int(minlen * sr)
self.sr = sr
self.mixes = []
for json_folder in json_folders:
mixfiles, wavlens = list(zip(*load_json(os.path.join(root + str_tmp, json_folder, 'mix.json')))) # list of 20000 filenames, and 20000 lengths
mixfiles = [os.path.join(root, mixfile.split('dataset/')[1]) for mixfile in mixfiles]
sig_json = [load_json(file) for file in sorted(glob.glob(os.path.join(root + str_tmp, json_folder, 's*.json')))] # list C, each have 20000 filenames
for i, spkr_json in enumerate(sig_json):
sig_json[i] = [os.path.join(root, line[0].split('dataset/')[1]) for line in spkr_json] # list C, each have 20000 filenames
siglists = list(zip(*sig_json)) # list of 20000, each have C filenames
self.mixes += list(zip(mixfiles, siglists, wavlens))
self.examples = []
for i, mix in enumerate(self.mixes):
if mix[2] < minlen:
continue
start = 0
while start + minlen <= mix[2]:
end = min(start + seglen, mix[2])
self.examples.append({'mixfile': mix[0], 'sourcefiles': mix[1], 'start': start, 'end':end})
start += minlen
random.seed(0)
self.examples = random.sample(self.examples, len(self.examples))
# Count.
example_source_files_len = [len(tmp['sourcefiles'] )for tmp in self.examples]
unique, counts = np.unique(np.array(example_source_files_len), return_counts=True)
self.example_weights =[]
for tmp in example_source_files_len:
self.example_weights.append(1./counts[tmp-2])
self.example_weights = torch.Tensor(self.example_weights)
def __len__(self):
return len(self.examples)
def __getitem__(self, idx):
"""
Returns:
mixture: [T]
sources: list of C, each [T]
"""
example = self.examples[idx]
mixfile, sourcefiles, start, end = example['mixfile'], example['sourcefiles'], example['start'], example['end']
mixture, sr = load(mixfile, sr=self.sr)
assert sr == self.sr, 'need to resample'
mixture = mixture[start:end]
sources = [load(sourcefile, sr=sr)[0][start:end] for sourcefile in sourcefiles]
return mixture, sources
def _collate_fn(batch):
"""
Args:
batch: list, len(batch) = batch_size, each entry is a tuple of (mixture, sources)
Returns:
mixtures_list: B x T, torch.Tensor, padded mixtures
ilens : B, torch.Tensor, length of each mixture before padding
sources_list: list of B Tensors, each C x T, where C is (variable) number of source audios
"""
ilens = [] # shape of mixtures
mixtures = [] # mixtures, same length as longest source in whole batch
sources_list = [] # padded sources, same length as mixtures
for mixture, sources in batch: # compute length to pad to
assert len(mixture) == len(sources[0])
assert len(mixture) <= 32000
ilens.append(len(mixture))
mixtures.append(pad_audio(mixture))
sources = torch.Tensor(np.stack([pad_audio(source) for source in sources], axis=0)).float()
sources_list.append(sources)
mixtures = torch.Tensor(np.stack(mixtures, axis=0)).float()
ilens = torch.Tensor(np.stack(ilens)).int()
return mixtures, ilens, sources_list
class TestDataset(data.Dataset):
def __init__(self, root, json_folders, sr=8000): # segment and cv_maxlen not implemented
"""
each line of textfile comes in the form of:
filename1, dB1, filename2, dB2, ...
args:
root: folder where dataset/ is located
json_folders: folders containing json files, **/dataset/#speakers/wav8k/min/tr/**
sr: sample rate
seglen: length of each segment in seconds
minlen: minimum segment length
"""
self.sr = sr
self.mixes = []
for json_folder in json_folders:
mixfiles, wavlens = list(zip(*load_json(os.path.join(root, json_folder, 'mix.json')))) # list of 20000 filenames, and 20000 lengths
mixfiles = [os.path.join(root, mixfile.split('dataset/')[1]) for mixfile in mixfiles]
sig_json = [load_json(file) for file in sorted(glob.glob(os.path.join(root, json_folder, 's*.json')))] # list C, each have 20000 filenames
for i, spkr_json in enumerate(sig_json):
sig_json[i] = [os.path.join(root, line[0].split('dataset/')[1]) for line in spkr_json] # list C, each have 20000 filenames
siglists = list(zip(*sig_json)) # list of 20000, each have C filenames
self.mixes += list(zip(mixfiles, siglists, wavlens))
#printlist(self.mixes)
self.examples = []
for i, mix in enumerate(self.mixes):
self.examples.append({'mixfile': mix[0], 'sourcefiles': mix[1], 'start': 0, 'end': mix[2]})
random.seed(0)
self.examples = random.sample(self.examples, len(self.examples))
def __len__(self):
return len(self.examples)
def __getitem__(self, idx):
"""
Returns:
mixture: [T]
sources: list of C, each [T]
"""
example = self.examples[idx]
mixfile, sourcefiles, start, end = example['mixfile'], example['sourcefiles'], example['start'], example['end']
mixture, sr = load(mixfile, sr=self.sr)
assert sr == self.sr, 'need to resample'
mixture = mixture[start:end]
sources = [load(sourcefile, sr=sr)[0][start:end] for sourcefile in sourcefiles]
return mixture, sources
if __name__ == "__main__":
root = "/ws/ifp-10_3/hasegawa/junzhez2/Baseline_Model/dataset"
tr_json = ["2spkr_json/tr/",
"3spkr_json/tr/",
"4spkr_json/tr/",
"5spkr_json/tr/"]
val_json = ["2spkr_json/cv/",
"3spkr_json/cv/",
"4spkr_json/cv/",
"5spkr_json/cv/"]
test_json = ["2spkr_json/tt",
"3spkr_json/tt",
"4spkr_json/tt",
"5spkr_json/tt"]
dataset = MixtureDataset(root, tr_json)
dataloader = torch.utils.data.DataLoader(dataset, batch_size=3, collate_fn=_collate_fn)
print(len(dataset))
for mixtures, ilens, sources_list in tqdm(dataloader):
start = time()
print(mixtures.shape, ilens.shape, [len(sources) for sources in sources_list])
print(time() - start)
| [
"numpy.stack",
"tqdm.tqdm",
"json.load",
"torch.utils.data.DataLoader",
"time.time",
"torch.Tensor",
"random.seed",
"librosa.load",
"numpy.array",
"os.path.join"
] | [((7485, 7559), 'torch.utils.data.DataLoader', 'torch.utils.data.DataLoader', (['dataset'], {'batch_size': '(3)', 'collate_fn': '_collate_fn'}), '(dataset, batch_size=3, collate_fn=_collate_fn)\n', (7512, 7559), False, 'import torch\n'), ((7625, 7641), 'tqdm.tqdm', 'tqdm', (['dataloader'], {}), '(dataloader)\n', (7629, 7641), False, 'from tqdm import tqdm\n'), ((322, 334), 'json.load', 'json.load', (['f'], {}), '(f)\n', (331, 334), False, 'import json\n'), ((2464, 2478), 'random.seed', 'random.seed', (['(0)'], {}), '(0)\n', (2475, 2478), False, 'import random\n'), ((2914, 2948), 'torch.Tensor', 'torch.Tensor', (['self.example_weights'], {}), '(self.example_weights)\n', (2926, 2948), False, 'import torch\n'), ((3324, 3349), 'librosa.load', 'load', (['mixfile'], {'sr': 'self.sr'}), '(mixfile, sr=self.sr)\n', (3328, 3349), False, 'from librosa import load\n'), ((6222, 6236), 'random.seed', 'random.seed', (['(0)'], {}), '(0)\n', (6233, 6236), False, 'import random\n'), ((6685, 6710), 'librosa.load', 'load', (['mixfile'], {'sr': 'self.sr'}), '(mixfile, sr=self.sr)\n', (6689, 6710), False, 'from librosa import load\n'), ((7659, 7665), 'time.time', 'time', ([], {}), '()\n', (7663, 7665), False, 'from time import time\n'), ((2691, 2725), 'numpy.array', 'np.array', (['example_source_files_len'], {}), '(example_source_files_len)\n', (2699, 2725), True, 'import numpy as np\n'), ((4504, 4530), 'numpy.stack', 'np.stack', (['mixtures'], {'axis': '(0)'}), '(mixtures, axis=0)\n', (4512, 4530), True, 'import numpy as np\n'), ((4565, 4580), 'numpy.stack', 'np.stack', (['ilens'], {}), '(ilens)\n', (4573, 4580), True, 'import numpy as np\n'), ((7767, 7773), 'time.time', 'time', ([], {}), '()\n', (7771, 7773), False, 'from time import time\n'), ((3455, 3478), 'librosa.load', 'load', (['sourcefile'], {'sr': 'sr'}), '(sourcefile, sr=sr)\n', (3459, 3478), False, 'from librosa import load\n'), ((6816, 6839), 'librosa.load', 'load', (['sourcefile'], {'sr': 'sr'}), '(sourcefile, sr=sr)\n', (6820, 6839), False, 'from librosa import load\n'), ((1369, 1422), 'os.path.join', 'os.path.join', (['(root + str_tmp)', 'json_folder', '"""mix.json"""'], {}), "(root + str_tmp, json_folder, 'mix.json')\n", (1381, 1422), False, 'import os\n'), ((1638, 1690), 'os.path.join', 'os.path.join', (['(root + str_tmp)', 'json_folder', '"""s*.json"""'], {}), "(root + str_tmp, json_folder, 's*.json')\n", (1650, 1690), False, 'import os\n'), ((5326, 5369), 'os.path.join', 'os.path.join', (['root', 'json_folder', '"""mix.json"""'], {}), "(root, json_folder, 'mix.json')\n", (5338, 5369), False, 'import os\n'), ((5585, 5627), 'os.path.join', 'os.path.join', (['root', 'json_folder', '"""s*.json"""'], {}), "(root, json_folder, 's*.json')\n", (5597, 5627), False, 'import os\n')] |
import os
from pathlib import Path
import pytest
from flopy.utils import binaryfile as bf
import numpy as np
import fiona
import rasterio
from shapely.geometry import box
import pytest
from ..grid import load_modelgrid
from ..results import export_cell_budget, export_heads, export_drawdown, export_sfr_results
@pytest.fixture(scope='module')
def lpr_output_path(test_output_folder):
return os.path.join(test_output_folder, 'lpr')
def check_files(outfiles, variables, kstpkper=None, layers=None):
replace = [('model_top', 'top')]
variables = set(variables)
if kstpkper is not None and np.isscalar(kstpkper[0]):
kstpkper = [kstpkper]
written = set()
for f in outfiles:
assert os.path.getsize(f) > 0
fname = os.path.split(f)[1]
for pair in replace:
fname = fname.replace(*pair)
props = parse_fname(fname)
assert props['var'] in variables
written.add(props['var'])
if kstpkper is not None:
assert (props['stp'], props['per']) in kstpkper
if props['lay'] is not None:
assert props['lay'] in layers
# verify that all variables were exported
assert len(written.difference(variables)) == 0
def parse_fname(fname):
props = {'var': None,
'lay': None,
'per': None,
'stp': None,
'suffix': None}
if 'stress_period_data' in fname:
props['var'] = os.path.splitext(fname)[0]
return props
info = os.path.splitext(fname)[0].split('_')
props['var'] = info.pop(0)
for i in range(len(info)):
item = info.pop(0)
if 'ctr' in item:
continue
for p in ['lay', 'per', 'stp']:
if p in item:
props[p] = int(item.strip(p))
return props
def compare_polygons(p1, p2, **kwargs):
"""Check that two polygons have the same extent"""
assert np.allclose(p1.area, p2.area, **kwargs)
assert np.allclose(p1.intersection(p2).area, p1.area, **kwargs)
def test_cell_budget_export(model):
m, grid, output_path = model
precision = 'single'
binary_grid_file = None
skip = []
if m.version == 'mf6':
precision = 'double'
binary_grid_file = os.path.join(m.model_ws, '{}.dis.grb'.format(m.name))
skip = ['WEL']
file = os.path.join(m.model_ws, '{}.cbc'.format(m.name))
#file = 'Examples/data/lpr/lpr_inset.cbc'
assert os.path.exists(file)
cbobj = bf.CellBudgetFile(file, precision=precision)
layers = list(range(cbobj.nlay))
kstpkper = cbobj.get_kstpkper()[0]
variables = [bs.decode().strip() for bs in cbobj.textlist
if bs.decode().strip() not in skip]
nrow, ncol = cbobj.nrow, cbobj.ncol
cbobj.close()
outfiles = export_cell_budget(file, grid,
binary_grid_file=binary_grid_file,
kstpkper=kstpkper,
precision=precision,
output_path=output_path)
check_files(outfiles, variables, kstpkper)
tifs = [f for f in outfiles if f.endswith('.tif')]
for f in tifs:
with rasterio.open(f) as src:
assert src.width == ncol
assert src.height == nrow
compare_polygons(grid.bbox, box(*src.bounds))
@pytest.mark.parametrize(('export_depth_to_water,export_layers,'
'export_water_table'),
((True, False, True),
(False, True, False)
))
def test_heads_export(model, export_depth_to_water, export_layers,
export_water_table):
m, grid, output_path = model
file = os.path.join(m.model_ws, '{}.hds'.format(m.name))
#file = 'Examples/data/lpr/lpr_inset.hds'
variables = ['hds']
if export_depth_to_water:
variables += ['wt', 'dtw', 'op']
if export_water_table and 'wt' not in variables:
variables.append('wt')
hdsobj = bf.HeadFile(file)
kstpkper = hdsobj.get_kstpkper()[-1:]
layers = list(range(hdsobj.nlay))
nrow, ncol = hdsobj.nrow, hdsobj.ncol
hdsobj.close()
outfiles = export_heads(file, grid, -1e4, -9999,
kstpkper=kstpkper,
export_depth_to_water=export_depth_to_water,
export_water_table=export_water_table,
export_layers=export_layers,
land_surface_elevations=m.dis.top.array,
output_path=output_path)
check_files(outfiles, variables, kstpkper, layers)
tifs = [f for f in outfiles if f.endswith('.tif')]
for f in tifs:
with rasterio.open(f) as src:
assert src.width == ncol
assert src.height == nrow
compare_polygons(grid.bbox, box(*src.bounds))
shps = [f for f in outfiles if f.endswith('.shp')]
for f in shps:
with fiona.open(f) as src:
assert box(*src.bounds).within(grid.bbox)
#compare_polygons(grid.bbox, box(*src.bounds), rtol=0.1)
def test_drawdown_export(model):
m, grid, output_path = model
file = os.path.join(m.model_ws, '{}.hds'.format(m.name))
#file = 'Examples/data/lpr/lpr_inset.hds'
variables = ['ddn', 'wt-ddn']
hdsobj = bf.HeadFile(file)
kstpkper0 = hdsobj.get_kstpkper()[1]
kstpkper1 = hdsobj.get_kstpkper()[-1]
layers = list(range(hdsobj.nlay))
nrow, ncol = hdsobj.nrow, hdsobj.ncol
hdsobj.close()
outfiles = export_drawdown(file, grid, -1e4, -9999,
kstpkper0=kstpkper0,
kstpkper1=kstpkper1,
output_path=output_path)
check_files(outfiles, variables, [kstpkper1], layers)
tifs = [f for f in outfiles if f.endswith('.tif')]
for f in tifs:
with rasterio.open(f) as src:
assert src.width == ncol
assert src.height == nrow
compare_polygons(grid.bbox, box(*src.bounds))
shps = [f for f in outfiles if f.endswith('.shp')]
for f in shps:
with fiona.open(f) as src:
assert box(*src.bounds).within(grid.bbox)
def test_sfr_results_export(lpr_model, lpr_modelgrid, lpr_output_path):
mf2005_sfr_outputfile = 'Examples/data/lpr/lpr_inset.sfr.out'
kstpkper = [(4, 0)]
variables = ['sfrout', 'baseflow', 'qaquifer']
outfiles = export_sfr_results(mf2005_sfr_outputfile=mf2005_sfr_outputfile,
model=lpr_model,
grid=lpr_modelgrid,
kstpkper=kstpkper,
output_length_units='feet',
output_time_units='seconds',
output_path=lpr_output_path
)
check_files(outfiles, variables, kstpkper)
@pytest.mark.parametrize('use_flopy', (False, True))
def test_mf6sfr_results_export(shellmound_model, shellmound_modelgrid, shellmound_output_path,
use_flopy):
mf6_sfr_stage_file = os.path.join(shellmound_model.model_ws, '{}.sfr.stage.bin'
.format(shellmound_model.name))
mf6_sfr_budget_file = os.path.join(shellmound_model.model_ws, '{}.sfr.out.bin'
.format(shellmound_model.name))
model_ws = Path(shellmound_model.model_ws)
if use_flopy:
model = shellmound_model
package_data_file=None
else:
package_data_file = model_ws / f'external/{shellmound_model.name}_packagedata.dat'
model = None
hdsobj = bf.HeadFile(mf6_sfr_stage_file, text='stage')
kstpkper = hdsobj.get_kstpkper()[:1] + hdsobj.get_kstpkper()[-1:]
variables = ['sfrout', 'baseflow', 'qaquifer']
outfiles = export_sfr_results(mf6_sfr_stage_file=mf6_sfr_stage_file,
mf6_sfr_budget_file=mf6_sfr_budget_file,
model=model,
mf6_package_data=package_data_file,
grid=shellmound_modelgrid,
kstpkper=kstpkper,
output_length_units='feet',
output_time_units='seconds',
output_path=shellmound_output_path
)
check_files(outfiles, variables, kstpkper)
def test_parse_fname():
fname = 'wel0_stress_period_data.shp'
result = parse_fname(fname)
assert result['var'] == os.path.splitext(fname)[0] | [
"rasterio.open",
"fiona.open",
"numpy.isscalar",
"os.path.getsize",
"numpy.allclose",
"pytest.fixture",
"os.path.exists",
"flopy.utils.binaryfile.HeadFile",
"pathlib.Path",
"os.path.splitext",
"flopy.utils.binaryfile.CellBudgetFile",
"pytest.mark.parametrize",
"os.path.split",
"os.path.joi... | [((314, 344), 'pytest.fixture', 'pytest.fixture', ([], {'scope': '"""module"""'}), "(scope='module')\n", (328, 344), False, 'import pytest\n'), ((3345, 3480), 'pytest.mark.parametrize', 'pytest.mark.parametrize', (['"""export_depth_to_water,export_layers,export_water_table"""', '((True, False, True), (False, True, False))'], {}), "(\n 'export_depth_to_water,export_layers,export_water_table', ((True, False,\n True), (False, True, False)))\n", (3368, 3480), False, 'import pytest\n'), ((6899, 6950), 'pytest.mark.parametrize', 'pytest.mark.parametrize', (['"""use_flopy"""', '(False, True)'], {}), "('use_flopy', (False, True))\n", (6922, 6950), False, 'import pytest\n'), ((397, 436), 'os.path.join', 'os.path.join', (['test_output_folder', '"""lpr"""'], {}), "(test_output_folder, 'lpr')\n", (409, 436), False, 'import os\n'), ((1917, 1956), 'numpy.allclose', 'np.allclose', (['p1.area', 'p2.area'], {}), '(p1.area, p2.area, **kwargs)\n', (1928, 1956), True, 'import numpy as np\n'), ((2441, 2461), 'os.path.exists', 'os.path.exists', (['file'], {}), '(file)\n', (2455, 2461), False, 'import os\n'), ((2474, 2518), 'flopy.utils.binaryfile.CellBudgetFile', 'bf.CellBudgetFile', (['file'], {'precision': 'precision'}), '(file, precision=precision)\n', (2491, 2518), True, 'from flopy.utils import binaryfile as bf\n'), ((4024, 4041), 'flopy.utils.binaryfile.HeadFile', 'bf.HeadFile', (['file'], {}), '(file)\n', (4035, 4041), True, 'from flopy.utils import binaryfile as bf\n'), ((5292, 5309), 'flopy.utils.binaryfile.HeadFile', 'bf.HeadFile', (['file'], {}), '(file)\n', (5303, 5309), True, 'from flopy.utils import binaryfile as bf\n'), ((7413, 7444), 'pathlib.Path', 'Path', (['shellmound_model.model_ws'], {}), '(shellmound_model.model_ws)\n', (7417, 7444), False, 'from pathlib import Path\n'), ((7662, 7707), 'flopy.utils.binaryfile.HeadFile', 'bf.HeadFile', (['mf6_sfr_stage_file'], {'text': '"""stage"""'}), "(mf6_sfr_stage_file, text='stage')\n", (7673, 7707), True, 'from flopy.utils import binaryfile as bf\n'), ((605, 629), 'numpy.isscalar', 'np.isscalar', (['kstpkper[0]'], {}), '(kstpkper[0])\n', (616, 629), True, 'import numpy as np\n'), ((719, 737), 'os.path.getsize', 'os.path.getsize', (['f'], {}), '(f)\n', (734, 737), False, 'import os\n'), ((758, 774), 'os.path.split', 'os.path.split', (['f'], {}), '(f)\n', (771, 774), False, 'import os\n'), ((1447, 1470), 'os.path.splitext', 'os.path.splitext', (['fname'], {}), '(fname)\n', (1463, 1470), False, 'import os\n'), ((3184, 3200), 'rasterio.open', 'rasterio.open', (['f'], {}), '(f)\n', (3197, 3200), False, 'import rasterio\n'), ((4680, 4696), 'rasterio.open', 'rasterio.open', (['f'], {}), '(f)\n', (4693, 4696), False, 'import rasterio\n'), ((4925, 4938), 'fiona.open', 'fiona.open', (['f'], {}), '(f)\n', (4935, 4938), False, 'import fiona\n'), ((5853, 5869), 'rasterio.open', 'rasterio.open', (['f'], {}), '(f)\n', (5866, 5869), False, 'import rasterio\n'), ((6098, 6111), 'fiona.open', 'fiona.open', (['f'], {}), '(f)\n', (6108, 6111), False, 'import fiona\n'), ((8613, 8636), 'os.path.splitext', 'os.path.splitext', (['fname'], {}), '(fname)\n', (8629, 8636), False, 'import os\n'), ((1506, 1529), 'os.path.splitext', 'os.path.splitext', (['fname'], {}), '(fname)\n', (1522, 1529), False, 'import os\n'), ((3324, 3340), 'shapely.geometry.box', 'box', (['*src.bounds'], {}), '(*src.bounds)\n', (3327, 3340), False, 'from shapely.geometry import box\n'), ((4820, 4836), 'shapely.geometry.box', 'box', (['*src.bounds'], {}), '(*src.bounds)\n', (4823, 4836), False, 'from shapely.geometry import box\n'), ((5993, 6009), 'shapely.geometry.box', 'box', (['*src.bounds'], {}), '(*src.bounds)\n', (5996, 6009), False, 'from shapely.geometry import box\n'), ((4966, 4982), 'shapely.geometry.box', 'box', (['*src.bounds'], {}), '(*src.bounds)\n', (4969, 4982), False, 'from shapely.geometry import box\n'), ((6139, 6155), 'shapely.geometry.box', 'box', (['*src.bounds'], {}), '(*src.bounds)\n', (6142, 6155), False, 'from shapely.geometry import box\n')] |
## Compiled from NodeLoads.ipynb on Sun Dec 10 12:51:11 2017
## DO NOT EDIT THIS FILE. YOUR CHANGES WILL BE LOST!!
## In [1]:
import numpy as np
from salib import extend
## In [9]:
class NodeLoad(object):
def __init__(self,fx=0.,fy=0.,mz=0.):
if np.isscalar(fx):
self.forces = np.matrix([fx,fy,mz],dtype=np.float64).T
else:
self.forces= fx.copy()
def __mul__(self,scale):
if scale == 1.0:
return self
return self.__class__(self.forces*scale)
__rmul__ = __mul__
def __repr__(self):
return "{}({},{},{})".format(self.__class__.__name__,*list(np.array(self.forces.T)[0]))
def __getitem__(self,ix):
return self.forces[ix,0]
## In [11]:
def makeNodeLoad(data):
G = data.get
return NodeLoad(G('FX',0),G('FY',0),G('MZ',0))
## In [13]:
id(NodeLoad)
## In [17]:
@extend
class NodeLoad:
@property
def fx(self):
return self.forces[0,0]
@fx.setter
def fx(self,v):
self.forces[0,0] = v
@property
def fy(self):
return self.forces[1,0]
@fy.setter
def fy(self,v):
self.forces[1,0] = v
@property
def mz(self):
return self.forces[2,0]
@mz.setter
def mz(self,v):
self.forces[2,0] = v
## In [ ]:
| [
"numpy.isscalar",
"numpy.matrix",
"numpy.array"
] | [((265, 280), 'numpy.isscalar', 'np.isscalar', (['fx'], {}), '(fx)\n', (276, 280), True, 'import numpy as np\n'), ((308, 349), 'numpy.matrix', 'np.matrix', (['[fx, fy, mz]'], {'dtype': 'np.float64'}), '([fx, fy, mz], dtype=np.float64)\n', (317, 349), True, 'import numpy as np\n'), ((662, 685), 'numpy.array', 'np.array', (['self.forces.T'], {}), '(self.forces.T)\n', (670, 685), True, 'import numpy as np\n')] |
import os
filename = 'seg-0_0_0.npz'
outputdir = os.getcwd() + os.sep + 'inferred_segmentation'
inputdir = os.getcwd()
import numpy as np
import h5py
import PIL
import PIL.Image
import cv2
import png
def save_tif8(id_data, filename):
cv2.imwrite(filename, id_data.astype('uint8'))
def save_tifc(id_data, filename, colordata):
pilOUT = gen_col_pil(id_data, colordata)
pilOUT.save(filename)
def save_png16(id_data, filename):
# Use pypng to write zgray as a grayscale PNG.
with open(filename, 'wb') as f:
writer = png.Writer(width=id_data.shape[1], height=id_data.shape[0], bitdepth=16, greyscale=True)
id_data_list = id_data.astype('uint16').tolist()
writer.write(f, id_data_list)
def save_png8(id_data, filename):
# Use pypng to write zgray as a grayscale PNG.
with open(filename, 'wb') as f:
writer = png.Writer(width=id_data.shape[1], height=id_data.shape[0], bitdepth=8, greyscale=True)
id_data_list = id_data.astype('uint8').tolist()
writer.write(f, id_data_list)
def save_pngc(id_data, filename, colordata):
pilOUT = gen_col_pil(id_data, colordata)
pilOUT.save(filename)
def save_npy(id_data, filename):
np.save(filename, id_data)
inputdir = os.getcwd()
data = np.load(inputdir+ os.sep+filename)
# print data.files
# print data['segmentation'].shape
num_z = data['segmentation'].shape[0]
num_y = data['segmentation'].shape[1]
num_x = data['segmentation'].shape[2]
for idz in range(num_z):
tmp = outputdir + os.sep + 'z' + '%04d' % (idz) + '.png'
save_png8(data['segmentation'][idz,:,:].transpose(), tmp)
| [
"os.getcwd",
"numpy.load",
"numpy.save",
"png.Writer"
] | [((110, 121), 'os.getcwd', 'os.getcwd', ([], {}), '()\n', (119, 121), False, 'import os\n'), ((1251, 1262), 'os.getcwd', 'os.getcwd', ([], {}), '()\n', (1260, 1262), False, 'import os\n'), ((1272, 1309), 'numpy.load', 'np.load', (['(inputdir + os.sep + filename)'], {}), '(inputdir + os.sep + filename)\n', (1279, 1309), True, 'import numpy as np\n'), ((1210, 1236), 'numpy.save', 'np.save', (['filename', 'id_data'], {}), '(filename, id_data)\n', (1217, 1236), True, 'import numpy as np\n'), ((50, 61), 'os.getcwd', 'os.getcwd', ([], {}), '()\n', (59, 61), False, 'import os\n'), ((550, 642), 'png.Writer', 'png.Writer', ([], {'width': 'id_data.shape[1]', 'height': 'id_data.shape[0]', 'bitdepth': '(16)', 'greyscale': '(True)'}), '(width=id_data.shape[1], height=id_data.shape[0], bitdepth=16,\n greyscale=True)\n', (560, 642), False, 'import png\n'), ((873, 964), 'png.Writer', 'png.Writer', ([], {'width': 'id_data.shape[1]', 'height': 'id_data.shape[0]', 'bitdepth': '(8)', 'greyscale': '(True)'}), '(width=id_data.shape[1], height=id_data.shape[0], bitdepth=8,\n greyscale=True)\n', (883, 964), False, 'import png\n')] |
"""
A. Long-term future prediction (model rollout)
1. encoder-decoder (0, 1 -> 8192-dim latent -> 2', 3'):
- feed (2', 3') images as input to predict (4', 5') images ...
2. encoder-decoder-64 (0, 1 -> 64-dim latent -> 2', 3'):
- feed (2', 3') images as input to predict (4', 5') images ...
3. encoder-decoder-64 & refine-64 ๏ผ0, 1 -> id-dim latent -> 2', 3')
- feed (2', 3') images as input to predict (4', 5') images ...
4. encoder-decoder-64 & refine-64 hybrid:
- use refine-64 model at certain prediction steps
B. Long-term future prediction with perturbation (model rollout)
"""
import os
import sys
import glob
import yaml
import json
import torch
import pprint
import shutil
import numpy as np
from PIL import Image
from tqdm import tqdm
from munch import munchify
from torchvision import transforms
from collections import OrderedDict
from models import VisDynamicsModel
from models_latentpred import VisLatentDynamicsModel
from dataset import NeuralPhysDataset
from pytorch_lightning import Trainer, seed_everything
from pytorch_lightning.loggers import TensorBoardLogger
def mkdir(folder):
if os.path.exists(folder):
shutil.rmtree(folder)
os.makedirs(folder)
def load_config(filepath):
with open(filepath, 'r') as stream:
try:
trainer_params = yaml.safe_load(stream)
return trainer_params
except yaml.YAMLError as exc:
print(exc)
def seed(cfg):
torch.manual_seed(cfg.seed)
if cfg.if_cuda:
torch.cuda.manual_seed(cfg.seed)
# uncomment for strict reproducibility
# torch.set_deterministic(True)
def model_rollout():
config_filepath = str(sys.argv[2])
cfg = load_config(filepath=config_filepath)
pprint.pprint(cfg)
cfg = munchify(cfg)
seed(cfg)
seed_everything(cfg.seed)
log_dir = '_'.join([cfg.log_dir,
cfg.dataset,
cfg.model_name,
str(cfg.seed)])
model = VisDynamicsModel(lr=cfg.lr,
seed=cfg.seed,
if_cuda=cfg.if_cuda,
if_test=True,
gamma=cfg.gamma,
log_dir=log_dir,
train_batch=cfg.train_batch,
val_batch=cfg.val_batch,
test_batch=cfg.test_batch,
num_workers=cfg.num_workers,
model_name=cfg.model_name,
data_filepath=cfg.data_filepath,
dataset=cfg.dataset,
lr_schedule=cfg.lr_schedule)
# load model
if cfg.model_name == 'encoder-decoder' or cfg.model_name == 'encoder-decoder-64':
checkpoint_filepath = str(sys.argv[3])
checkpoint_filepath = glob.glob(os.path.join(checkpoint_filepath, '*.ckpt'))[0]
ckpt = torch.load(checkpoint_filepath)
model.load_state_dict(ckpt['state_dict'])
if 'refine' in cfg.model_name:
checkpoint_filepath = str(sys.argv[4])
checkpoint_filepath = glob.glob(os.path.join(checkpoint_filepath, '*.ckpt'))[0]
ckpt = torch.load(checkpoint_filepath)
ckpt = rename_ckpt_for_multi_models(ckpt)
model.model.load_state_dict(ckpt)
high_dim_checkpoint_filepath = str(sys.argv[3])
high_dim_checkpoint_filepath = glob.glob(os.path.join(high_dim_checkpoint_filepath, '*.ckpt'))[0]
ckpt = torch.load(high_dim_checkpoint_filepath)
ckpt = rename_ckpt_for_multi_models(ckpt)
model.high_dim_model.load_state_dict(ckpt)
model = model.to('cuda')
model.eval()
model.freeze()
# get all the test video ids
data_filepath_base = os.path.join(cfg.data_filepath, cfg.dataset)
with open(os.path.join('../datainfo', cfg.dataset, f'data_split_dict_{cfg.seed}.json'), 'r') as file:
seq_dict = json.load(file)
test_vid_ids = seq_dict['test']
pred_len = int(sys.argv[6])
long_term_folder = os.path.join(log_dir, 'prediction_long_term', 'model_rollout')
loss_dict = {}
if cfg.model_name == 'encoder-decoder' or cfg.model_name == 'encoder-decoder-64':
for p_vid_idx in tqdm(test_vid_ids):
vid_filepath = os.path.join(data_filepath_base, str(p_vid_idx))
total_num_frames = len(os.listdir(vid_filepath))
suf = os.listdir(vid_filepath)[0].split('.')[-1]
data = None
saved_folder = os.path.join(long_term_folder, str(p_vid_idx))
mkdir(saved_folder)
loss_lst = []
for start_frame_idx in range(total_num_frames - 3):
if start_frame_idx % 2 != 0:
continue
# take the initial input from ground truth data
if start_frame_idx % pred_len == 0:
data = [get_data(os.path.join(vid_filepath, f'{start_frame_idx}.{suf}')),
get_data(os.path.join(vid_filepath, f'{start_frame_idx+1}.{suf}'))]
data = (torch.cat(data, 2)).unsqueeze(0)
# get the target
target = [get_data(os.path.join(vid_filepath, f'{start_frame_idx+2}.{suf}')),
get_data(os.path.join(vid_filepath, f'{start_frame_idx+3}.{suf}'))]
target = (torch.cat(target, 2)).unsqueeze(0)
# feed into the model
if cfg.model_name == 'encoder-decoder':
output, latent = model.model(data.cuda())
if cfg.model_name == 'encoder-decoder-64':
output, latent = model.model(data.cuda(), data.cuda(), False)
# compute loss
loss_lst.append(float(model.loss_func(output, target.cuda()).cpu().detach().numpy()))
# save (2', 3'), (4', 5'), ...
img = tensor_to_img(output[0, :, :, :128])
img.save(os.path.join(saved_folder, f'{start_frame_idx+2}.{suf}'))
img = tensor_to_img(output[0, :, :, 128:])
img.save(os.path.join(saved_folder, f'{start_frame_idx+3}.{suf}'))
# the output becomes the input data in the next iteration
data = torch.tensor(output.cpu().detach().numpy()).float()
loss_dict[p_vid_idx] = loss_lst
# save the test loss for all the testing videos
with open(os.path.join(long_term_folder, 'test_loss.json'), 'w') as file:
json.dump(loss_dict, file, indent=4)
if 'refine' in cfg.model_name:
for p_vid_idx in tqdm(test_vid_ids):
vid_filepath = os.path.join(data_filepath_base, str(p_vid_idx))
total_num_frames = len(os.listdir(vid_filepath))
suf = os.listdir(vid_filepath)[0].split('.')[-1]
data = None
saved_folder = os.path.join(long_term_folder, str(p_vid_idx))
mkdir(saved_folder)
loss_lst = []
for start_frame_idx in range(total_num_frames - 3):
if start_frame_idx % 2 != 0:
continue
# take the initial input from ground truth data
if start_frame_idx % pred_len == 0:
data = [get_data(os.path.join(vid_filepath, f'{start_frame_idx}.{suf}')),
get_data(os.path.join(vid_filepath, f'{start_frame_idx+1}.{suf}'))]
data = (torch.cat(data, 2)).unsqueeze(0)
# get the target
target = [get_data(os.path.join(vid_filepath, f'{start_frame_idx+2}.{suf}')),
get_data(os.path.join(vid_filepath, f'{start_frame_idx+3}.{suf}'))]
target = (torch.cat(target, 2)).unsqueeze(0)
# feed into the model
_, latent = model.high_dim_model(data.cuda(), data.cuda(), False)
latent = latent.squeeze(-1).squeeze(-1)
latent_reconstructed, latent_latent = model.model(latent)
output, _ = model.high_dim_model(data.cuda(), latent_reconstructed.unsqueeze(2).unsqueeze(3), True)
# compute loss
loss_lst.append(float(model.loss_func(output, target.cuda()).cpu().detach().numpy()))
# save (2', 3'), (4', 5'), ...
img = tensor_to_img(output[0, :, :, :128])
img.save(os.path.join(saved_folder, f'{start_frame_idx+2}.{suf}'))
img = tensor_to_img(output[0, :, :, 128:])
img.save(os.path.join(saved_folder, f'{start_frame_idx+3}.{suf}'))
# the output becomes the input data in the next iteration
data = torch.tensor(output.cpu().detach().numpy()).float()
loss_dict[p_vid_idx] = loss_lst
# save the test loss for all the testing videos
with open(os.path.join(long_term_folder, 'test_loss.json'), 'w') as file:
json.dump(loss_dict, file, indent=4)
def model_rollout_hybrid(step):
config_filepath = str(sys.argv[2])
cfg = load_config(filepath=config_filepath)
pprint.pprint(cfg)
cfg = munchify(cfg)
seed(cfg)
seed_everything(cfg.seed)
if 'refine' not in cfg.model_name:
assert False, "the hybrid scheme is only supported with refine model..."
log_dir = '_'.join([cfg.log_dir,
cfg.dataset,
cfg.model_name,
str(cfg.seed)])
model = VisDynamicsModel(lr=cfg.lr,
seed=cfg.seed,
if_cuda=cfg.if_cuda,
if_test=True,
gamma=cfg.gamma,
log_dir=log_dir,
train_batch=cfg.train_batch,
val_batch=cfg.val_batch,
test_batch=cfg.test_batch,
num_workers=cfg.num_workers,
model_name=cfg.model_name,
data_filepath=cfg.data_filepath,
dataset=cfg.dataset,
lr_schedule=cfg.lr_schedule)
# load model
checkpoint_filepath = str(sys.argv[4])
checkpoint_filepath = glob.glob(os.path.join(checkpoint_filepath, '*.ckpt'))[0]
ckpt = torch.load(checkpoint_filepath)
ckpt = rename_ckpt_for_multi_models(ckpt)
model.model.load_state_dict(ckpt)
high_dim_checkpoint_filepath = str(sys.argv[3])
high_dim_checkpoint_filepath = glob.glob(os.path.join(high_dim_checkpoint_filepath, '*.ckpt'))[0]
ckpt = torch.load(high_dim_checkpoint_filepath)
ckpt = rename_ckpt_for_multi_models(ckpt)
model.high_dim_model.load_state_dict(ckpt)
model = model.to('cuda')
model.eval()
model.freeze()
# get all the test video ids
data_filepath_base = os.path.join(cfg.data_filepath, cfg.dataset)
with open(os.path.join('../datainfo', cfg.dataset, f'data_split_dict_{cfg.seed}.json'), 'r') as file:
seq_dict = json.load(file)
test_vid_ids = seq_dict['test']
pred_len = int(sys.argv[6])
long_term_folder = os.path.join(log_dir, 'prediction_long_term', f'hybrid_rollout_{step}')
loss_dict = {}
for p_vid_idx in tqdm(test_vid_ids):
vid_filepath = os.path.join(data_filepath_base, str(p_vid_idx))
total_num_frames = len(os.listdir(vid_filepath))
suf = os.listdir(vid_filepath)[0].split('.')[-1]
data = None
saved_folder = os.path.join(long_term_folder, str(p_vid_idx))
mkdir(saved_folder)
loss_lst = []
for start_frame_idx in range(total_num_frames - 3):
if start_frame_idx % 2 != 0:
continue
# take the initial input from ground truth data
if start_frame_idx % pred_len == 0:
data = [get_data(os.path.join(vid_filepath, f'{start_frame_idx}.{suf}')),
get_data(os.path.join(vid_filepath, f'{start_frame_idx+1}.{suf}'))]
data = (torch.cat(data, 2)).unsqueeze(0)
# get the target
target = [get_data(os.path.join(vid_filepath, f'{start_frame_idx+2}.{suf}')),
get_data(os.path.join(vid_filepath, f'{start_frame_idx+3}.{suf}'))]
target = (torch.cat(target, 2)).unsqueeze(0)
# feed into the model
if (start_frame_idx + 2) % (2 * step + 2) == 0:
_, latent = model.high_dim_model(data.cuda(), data.cuda(), False)
latent = latent.squeeze(-1).squeeze(-1)
latent_reconstructed, latent_latent = model.model(latent)
output, _ = model.high_dim_model(data.cuda(), latent_reconstructed.unsqueeze(2).unsqueeze(3), True)
else:
output, _ = model.high_dim_model(data.cuda(), data.cuda(), False)
# compute loss
loss_lst.append(float(model.loss_func(output, target.cuda()).cpu().detach().numpy()))
# save (2', 3'), (4', 5'), ...
img = tensor_to_img(output[0, :, :, :128])
img.save(os.path.join(saved_folder, f'{start_frame_idx+2}.{suf}'))
img = tensor_to_img(output[0, :, :, 128:])
img.save(os.path.join(saved_folder, f'{start_frame_idx+3}.{suf}'))
# the output becomes the input data in the next iteration
data = torch.tensor(output.cpu().detach().numpy()).float()
loss_dict[p_vid_idx] = loss_lst
# save the test loss for all the testing videos
with open(os.path.join(long_term_folder, 'test_loss.json'), 'w') as file:
json.dump(loss_dict, file, indent=4)
def model_rollout_perturb(perturb_type, perturb_level):
config_filepath = str(sys.argv[2])
cfg = load_config(filepath=config_filepath)
pprint.pprint(cfg)
cfg = munchify(cfg)
seed(cfg)
seed_everything(cfg.seed)
log_dir = '_'.join([cfg.log_dir,
cfg.dataset,
cfg.model_name,
str(cfg.seed)])
model = VisDynamicsModel(lr=cfg.lr,
seed=cfg.seed,
if_cuda=cfg.if_cuda,
if_test=True,
gamma=cfg.gamma,
log_dir=log_dir,
train_batch=cfg.train_batch,
val_batch=cfg.val_batch,
test_batch=cfg.test_batch,
num_workers=cfg.num_workers,
model_name=cfg.model_name,
data_filepath=cfg.data_filepath,
dataset=cfg.dataset,
lr_schedule=cfg.lr_schedule)
# load model
if cfg.model_name == 'encoder-decoder' or cfg.model_name == 'encoder-decoder-64':
checkpoint_filepath = str(sys.argv[3])
checkpoint_filepath = glob.glob(os.path.join(checkpoint_filepath, '*.ckpt'))[0]
ckpt = torch.load(checkpoint_filepath)
model.load_state_dict(ckpt['state_dict'])
if 'refine' in cfg.model_name:
checkpoint_filepath = str(sys.argv[4])
checkpoint_filepath = glob.glob(os.path.join(checkpoint_filepath, '*.ckpt'))[0]
ckpt = torch.load(checkpoint_filepath)
ckpt = rename_ckpt_for_multi_models(ckpt)
model.model.load_state_dict(ckpt)
high_dim_checkpoint_filepath = str(sys.argv[3])
high_dim_checkpoint_filepath = glob.glob(os.path.join(high_dim_checkpoint_filepath, '*.ckpt'))[0]
ckpt = torch.load(high_dim_checkpoint_filepath)
ckpt = rename_ckpt_for_multi_models(ckpt)
model.high_dim_model.load_state_dict(ckpt)
model = model.to('cuda')
model.eval()
model.freeze()
# get all the test video ids
data_filepath_base = os.path.join(cfg.data_filepath, cfg.dataset)
with open(os.path.join('../datainfo', cfg.dataset, f'data_split_dict_{cfg.seed}.json'), 'r') as file:
seq_dict = json.load(file)
test_vid_ids = seq_dict['test']
pred_len = int(sys.argv[6])
long_term_folder = os.path.join(log_dir, 'prediction_long_term', f'model_rollout_perturb_{perturb_type}_{perturb_level}')
loss_dict = {}
if cfg.model_name == 'encoder-decoder' or cfg.model_name == 'encoder-decoder-64':
for p_vid_idx in tqdm(test_vid_ids):
vid_filepath = os.path.join(data_filepath_base, str(p_vid_idx))
total_num_frames = len(os.listdir(vid_filepath))
suf = os.listdir(vid_filepath)[0].split('.')[-1]
data = None
saved_folder = os.path.join(long_term_folder, str(p_vid_idx))
mkdir(saved_folder)
loss_lst = []
for start_frame_idx in range(total_num_frames - 3):
if start_frame_idx % 2 != 0:
continue
# take the initial input from ground truth data
if start_frame_idx % pred_len == 0:
data = [get_data_perturb(os.path.join(vid_filepath, f'{start_frame_idx}.{suf}'), perturb_type, perturb_level),
get_data_perturb(os.path.join(vid_filepath, f'{start_frame_idx+1}.{suf}'), perturb_type, perturb_level)]
data = (torch.cat(data, 2)).unsqueeze(0)
img = tensor_to_img(data[0, :, :, :128])
img.save(os.path.join(saved_folder, f'{start_frame_idx}.{suf}'))
img = tensor_to_img(data[0, :, :, 128:])
img.save(os.path.join(saved_folder, f'{start_frame_idx+1}.{suf}'))
# get the target
target = [get_data(os.path.join(vid_filepath, f'{start_frame_idx+2}.{suf}')),
get_data(os.path.join(vid_filepath, f'{start_frame_idx+3}.{suf}'))]
target = (torch.cat(target, 2)).unsqueeze(0)
# feed into the model
if cfg.model_name == 'encoder-decoder':
output, latent = model.model(data.cuda())
if cfg.model_name == 'encoder-decoder-64':
output, latent = model.model(data.cuda(), data.cuda(), False)
# compute loss
loss_lst.append(float(model.loss_func(output, target.cuda()).cpu().detach().numpy()))
# save (2', 3'), (4', 5'), ...
img = tensor_to_img(output[0, :, :, :128])
img.save(os.path.join(saved_folder, f'{start_frame_idx+2}.{suf}'))
img = tensor_to_img(output[0, :, :, 128:])
img.save(os.path.join(saved_folder, f'{start_frame_idx+3}.{suf}'))
# the output becomes the input data in the next iteration
data = torch.tensor(output.cpu().detach().numpy()).float()
loss_dict[p_vid_idx] = loss_lst
# save the test loss for all the testing videos
with open(os.path.join(long_term_folder, 'test_loss.json'), 'w') as file:
json.dump(loss_dict, file, indent=4)
if 'refine' in cfg.model_name:
for p_vid_idx in tqdm(test_vid_ids):
vid_filepath = os.path.join(data_filepath_base, str(p_vid_idx))
total_num_frames = len(os.listdir(vid_filepath))
suf = os.listdir(vid_filepath)[0].split('.')[-1]
data = None
saved_folder = os.path.join(long_term_folder, str(p_vid_idx))
mkdir(saved_folder)
loss_lst = []
for start_frame_idx in range(total_num_frames - 3):
if start_frame_idx % 2 != 0:
continue
# take the initial input from ground truth data
if start_frame_idx % pred_len == 0:
data = [get_data_perturb(os.path.join(vid_filepath, f'{start_frame_idx}.{suf}'), perturb_type, perturb_level),
get_data_perturb(os.path.join(vid_filepath, f'{start_frame_idx+1}.{suf}'), perturb_type, perturb_level)]
data = (torch.cat(data, 2)).unsqueeze(0)
img = tensor_to_img(data[0, :, :, :128])
img.save(os.path.join(saved_folder, f'{start_frame_idx}.{suf}'))
img = tensor_to_img(data[0, :, :, 128:])
img.save(os.path.join(saved_folder, f'{start_frame_idx+1}.{suf}'))
# get the target
target = [get_data(os.path.join(vid_filepath, f'{start_frame_idx+2}.{suf}')),
get_data(os.path.join(vid_filepath, f'{start_frame_idx+3}.{suf}'))]
target = (torch.cat(target, 2)).unsqueeze(0)
# feed into the model
_, latent = model.high_dim_model(data.cuda(), data.cuda(), False)
latent = latent.squeeze(-1).squeeze(-1)
latent_reconstructed, latent_latent = model.model(latent)
output, _ = model.high_dim_model(data.cuda(), latent_reconstructed.unsqueeze(2).unsqueeze(3), True)
# compute loss
loss_lst.append(float(model.loss_func(output, target.cuda()).cpu().detach().numpy()))
# save (2', 3'), (4', 5'), ...
img = tensor_to_img(output[0, :, :, :128])
img.save(os.path.join(saved_folder, f'{start_frame_idx+2}.{suf}'))
img = tensor_to_img(output[0, :, :, 128:])
img.save(os.path.join(saved_folder, f'{start_frame_idx+3}.{suf}'))
# the output becomes the input data in the next iteration
data = torch.tensor(output.cpu().detach().numpy()).float()
loss_dict[p_vid_idx] = loss_lst
# save the test loss for all the testing videos
with open(os.path.join(long_term_folder, 'test_loss.json'), 'w') as file:
json.dump(loss_dict, file, indent=4)
def rename_ckpt_for_multi_models(ckpt):
renamed_state_dict = OrderedDict()
for k, v in ckpt['state_dict'].items():
if 'high_dim_model' in k:
name = k.replace('high_dim_model.', '')
else:
name = k.replace('model.', '')
renamed_state_dict[name] = v
return renamed_state_dict
def get_data(filepath):
data = Image.open(filepath)
data = data.resize((128, 128))
data = np.array(data)
data = torch.tensor(data / 255.0)
data = data.permute(2, 0, 1).float()
return data
def get_data_perturb(filepath, perturb_type, perturb_level):
data = Image.open(filepath)
data = data.resize((128, 128))
data = np.array(data)
bg_color = np.array([215, 205, 192])
rng = np.random.RandomState(int(filepath.split('/')[-2]))
new_bg_color = rng.randint(256, size=3)
if perturb_type == 'background_replace':
for i in range(2**(perturb_level-1)):
for j in range(2**(perturb_level-1)):
if np.array_equal(data[i, j], bg_color):
data[i, j] = new_bg_color
elif perturb_type == 'background_cover':
for i in range(2**(perturb_level-1)):
for j in range(2**(perturb_level-1)):
data[i, j] = new_bg_color
elif perturb_type == 'white_noise':
sigma = 255.0 * (2**(perturb_level-1) / 128) ** 2
data = data + rng.normal(0, sigma, data.shape)
else:
pass
data = torch.tensor(data / 255.0)
data = data.permute(2, 0, 1).float()
return data
# out_tensor: 3 x 128 x 128 -> 128 x 128 x 3
def tensor_to_img(out_tensor):
return transforms.ToPILImage()(out_tensor).convert("RGB")
if __name__ == '__main__':
if str(sys.argv[1]) == 'model-rollout':
model_rollout()
elif 'hybrid' in str(sys.argv[1]):
step = int(sys.argv[1].split('-')[-1])
model_rollout_hybrid(step)
elif str(sys.argv[1]) == 'latent-prediction':
latent_prediction()
elif 'perturb' in str(sys.argv[1]):
perturb_type = str(sys.argv[1].split('-')[-2])
perturb_level = int(sys.argv[1].split('-')[-1])
model_rollout_perturb(perturb_type, perturb_level)
else:
assert False, "prediction scheme is not supported..." | [
"pytorch_lightning.seed_everything",
"numpy.array_equal",
"torch.cat",
"yaml.safe_load",
"pprint.pprint",
"shutil.rmtree",
"os.path.join",
"models.VisDynamicsModel",
"torch.load",
"os.path.exists",
"torchvision.transforms.ToPILImage",
"json.dump",
"tqdm.tqdm",
"munch.munchify",
"torch.ma... | [((1128, 1150), 'os.path.exists', 'os.path.exists', (['folder'], {}), '(folder)\n', (1142, 1150), False, 'import os\n'), ((1186, 1205), 'os.makedirs', 'os.makedirs', (['folder'], {}), '(folder)\n', (1197, 1205), False, 'import os\n'), ((1454, 1481), 'torch.manual_seed', 'torch.manual_seed', (['cfg.seed'], {}), '(cfg.seed)\n', (1471, 1481), False, 'import torch\n'), ((1736, 1754), 'pprint.pprint', 'pprint.pprint', (['cfg'], {}), '(cfg)\n', (1749, 1754), False, 'import pprint\n'), ((1765, 1778), 'munch.munchify', 'munchify', (['cfg'], {}), '(cfg)\n', (1773, 1778), False, 'from munch import munchify\n'), ((1797, 1822), 'pytorch_lightning.seed_everything', 'seed_everything', (['cfg.seed'], {}), '(cfg.seed)\n', (1812, 1822), False, 'from pytorch_lightning import Trainer, seed_everything\n'), ((1991, 2340), 'models.VisDynamicsModel', 'VisDynamicsModel', ([], {'lr': 'cfg.lr', 'seed': 'cfg.seed', 'if_cuda': 'cfg.if_cuda', 'if_test': '(True)', 'gamma': 'cfg.gamma', 'log_dir': 'log_dir', 'train_batch': 'cfg.train_batch', 'val_batch': 'cfg.val_batch', 'test_batch': 'cfg.test_batch', 'num_workers': 'cfg.num_workers', 'model_name': 'cfg.model_name', 'data_filepath': 'cfg.data_filepath', 'dataset': 'cfg.dataset', 'lr_schedule': 'cfg.lr_schedule'}), '(lr=cfg.lr, seed=cfg.seed, if_cuda=cfg.if_cuda, if_test=\n True, gamma=cfg.gamma, log_dir=log_dir, train_batch=cfg.train_batch,\n val_batch=cfg.val_batch, test_batch=cfg.test_batch, num_workers=cfg.\n num_workers, model_name=cfg.model_name, data_filepath=cfg.data_filepath,\n dataset=cfg.dataset, lr_schedule=cfg.lr_schedule)\n', (2007, 2340), False, 'from models import VisDynamicsModel\n'), ((3799, 3843), 'os.path.join', 'os.path.join', (['cfg.data_filepath', 'cfg.dataset'], {}), '(cfg.data_filepath, cfg.dataset)\n', (3811, 3843), False, 'import os\n'), ((4077, 4139), 'os.path.join', 'os.path.join', (['log_dir', '"""prediction_long_term"""', '"""model_rollout"""'], {}), "(log_dir, 'prediction_long_term', 'model_rollout')\n", (4089, 4139), False, 'import os\n'), ((9160, 9178), 'pprint.pprint', 'pprint.pprint', (['cfg'], {}), '(cfg)\n', (9173, 9178), False, 'import pprint\n'), ((9189, 9202), 'munch.munchify', 'munchify', (['cfg'], {}), '(cfg)\n', (9197, 9202), False, 'from munch import munchify\n'), ((9221, 9246), 'pytorch_lightning.seed_everything', 'seed_everything', (['cfg.seed'], {}), '(cfg.seed)\n', (9236, 9246), False, 'from pytorch_lightning import Trainer, seed_everything\n'), ((9536, 9885), 'models.VisDynamicsModel', 'VisDynamicsModel', ([], {'lr': 'cfg.lr', 'seed': 'cfg.seed', 'if_cuda': 'cfg.if_cuda', 'if_test': '(True)', 'gamma': 'cfg.gamma', 'log_dir': 'log_dir', 'train_batch': 'cfg.train_batch', 'val_batch': 'cfg.val_batch', 'test_batch': 'cfg.test_batch', 'num_workers': 'cfg.num_workers', 'model_name': 'cfg.model_name', 'data_filepath': 'cfg.data_filepath', 'dataset': 'cfg.dataset', 'lr_schedule': 'cfg.lr_schedule'}), '(lr=cfg.lr, seed=cfg.seed, if_cuda=cfg.if_cuda, if_test=\n True, gamma=cfg.gamma, log_dir=log_dir, train_batch=cfg.train_batch,\n val_batch=cfg.val_batch, test_batch=cfg.test_batch, num_workers=cfg.\n num_workers, model_name=cfg.model_name, data_filepath=cfg.data_filepath,\n dataset=cfg.dataset, lr_schedule=cfg.lr_schedule)\n', (9552, 9885), False, 'from models import VisDynamicsModel\n'), ((10401, 10432), 'torch.load', 'torch.load', (['checkpoint_filepath'], {}), '(checkpoint_filepath)\n', (10411, 10432), False, 'import torch\n'), ((10683, 10723), 'torch.load', 'torch.load', (['high_dim_checkpoint_filepath'], {}), '(high_dim_checkpoint_filepath)\n', (10693, 10723), False, 'import torch\n'), ((10942, 10986), 'os.path.join', 'os.path.join', (['cfg.data_filepath', 'cfg.dataset'], {}), '(cfg.data_filepath, cfg.dataset)\n', (10954, 10986), False, 'import os\n'), ((11220, 11291), 'os.path.join', 'os.path.join', (['log_dir', '"""prediction_long_term"""', 'f"""hybrid_rollout_{step}"""'], {}), "(log_dir, 'prediction_long_term', f'hybrid_rollout_{step}')\n", (11232, 11291), False, 'import os\n'), ((11333, 11351), 'tqdm.tqdm', 'tqdm', (['test_vid_ids'], {}), '(test_vid_ids)\n', (11337, 11351), False, 'from tqdm import tqdm\n'), ((13886, 13904), 'pprint.pprint', 'pprint.pprint', (['cfg'], {}), '(cfg)\n', (13899, 13904), False, 'import pprint\n'), ((13915, 13928), 'munch.munchify', 'munchify', (['cfg'], {}), '(cfg)\n', (13923, 13928), False, 'from munch import munchify\n'), ((13947, 13972), 'pytorch_lightning.seed_everything', 'seed_everything', (['cfg.seed'], {}), '(cfg.seed)\n', (13962, 13972), False, 'from pytorch_lightning import Trainer, seed_everything\n'), ((14141, 14490), 'models.VisDynamicsModel', 'VisDynamicsModel', ([], {'lr': 'cfg.lr', 'seed': 'cfg.seed', 'if_cuda': 'cfg.if_cuda', 'if_test': '(True)', 'gamma': 'cfg.gamma', 'log_dir': 'log_dir', 'train_batch': 'cfg.train_batch', 'val_batch': 'cfg.val_batch', 'test_batch': 'cfg.test_batch', 'num_workers': 'cfg.num_workers', 'model_name': 'cfg.model_name', 'data_filepath': 'cfg.data_filepath', 'dataset': 'cfg.dataset', 'lr_schedule': 'cfg.lr_schedule'}), '(lr=cfg.lr, seed=cfg.seed, if_cuda=cfg.if_cuda, if_test=\n True, gamma=cfg.gamma, log_dir=log_dir, train_batch=cfg.train_batch,\n val_batch=cfg.val_batch, test_batch=cfg.test_batch, num_workers=cfg.\n num_workers, model_name=cfg.model_name, data_filepath=cfg.data_filepath,\n dataset=cfg.dataset, lr_schedule=cfg.lr_schedule)\n', (14157, 14490), False, 'from models import VisDynamicsModel\n'), ((15949, 15993), 'os.path.join', 'os.path.join', (['cfg.data_filepath', 'cfg.dataset'], {}), '(cfg.data_filepath, cfg.dataset)\n', (15961, 15993), False, 'import os\n'), ((16227, 16333), 'os.path.join', 'os.path.join', (['log_dir', '"""prediction_long_term"""', 'f"""model_rollout_perturb_{perturb_type}_{perturb_level}"""'], {}), "(log_dir, 'prediction_long_term',\n f'model_rollout_perturb_{perturb_type}_{perturb_level}')\n", (16239, 16333), False, 'import os\n'), ((22024, 22037), 'collections.OrderedDict', 'OrderedDict', ([], {}), '()\n', (22035, 22037), False, 'from collections import OrderedDict\n'), ((22328, 22348), 'PIL.Image.open', 'Image.open', (['filepath'], {}), '(filepath)\n', (22338, 22348), False, 'from PIL import Image\n'), ((22395, 22409), 'numpy.array', 'np.array', (['data'], {}), '(data)\n', (22403, 22409), True, 'import numpy as np\n'), ((22421, 22447), 'torch.tensor', 'torch.tensor', (['(data / 255.0)'], {}), '(data / 255.0)\n', (22433, 22447), False, 'import torch\n'), ((22578, 22598), 'PIL.Image.open', 'Image.open', (['filepath'], {}), '(filepath)\n', (22588, 22598), False, 'from PIL import Image\n'), ((22645, 22659), 'numpy.array', 'np.array', (['data'], {}), '(data)\n', (22653, 22659), True, 'import numpy as np\n'), ((22680, 22705), 'numpy.array', 'np.array', (['[215, 205, 192]'], {}), '([215, 205, 192])\n', (22688, 22705), True, 'import numpy as np\n'), ((23429, 23455), 'torch.tensor', 'torch.tensor', (['(data / 255.0)'], {}), '(data / 255.0)\n', (23441, 23455), False, 'import torch\n'), ((1160, 1181), 'shutil.rmtree', 'shutil.rmtree', (['folder'], {}), '(folder)\n', (1173, 1181), False, 'import shutil\n'), ((1510, 1542), 'torch.cuda.manual_seed', 'torch.cuda.manual_seed', (['cfg.seed'], {}), '(cfg.seed)\n', (1532, 1542), False, 'import torch\n'), ((2958, 2989), 'torch.load', 'torch.load', (['checkpoint_filepath'], {}), '(checkpoint_filepath)\n', (2968, 2989), False, 'import torch\n'), ((3230, 3261), 'torch.load', 'torch.load', (['checkpoint_filepath'], {}), '(checkpoint_filepath)\n', (3240, 3261), False, 'import torch\n'), ((3532, 3572), 'torch.load', 'torch.load', (['high_dim_checkpoint_filepath'], {}), '(high_dim_checkpoint_filepath)\n', (3542, 3572), False, 'import torch\n'), ((3969, 3984), 'json.load', 'json.load', (['file'], {}), '(file)\n', (3978, 3984), False, 'import json\n'), ((4271, 4289), 'tqdm.tqdm', 'tqdm', (['test_vid_ids'], {}), '(test_vid_ids)\n', (4275, 4289), False, 'from tqdm import tqdm\n'), ((6659, 6677), 'tqdm.tqdm', 'tqdm', (['test_vid_ids'], {}), '(test_vid_ids)\n', (6663, 6677), False, 'from tqdm import tqdm\n'), ((11112, 11127), 'json.load', 'json.load', (['file'], {}), '(file)\n', (11121, 11127), False, 'import json\n'), ((13700, 13736), 'json.dump', 'json.dump', (['loss_dict', 'file'], {'indent': '(4)'}), '(loss_dict, file, indent=4)\n', (13709, 13736), False, 'import json\n'), ((15108, 15139), 'torch.load', 'torch.load', (['checkpoint_filepath'], {}), '(checkpoint_filepath)\n', (15118, 15139), False, 'import torch\n'), ((15380, 15411), 'torch.load', 'torch.load', (['checkpoint_filepath'], {}), '(checkpoint_filepath)\n', (15390, 15411), False, 'import torch\n'), ((15682, 15722), 'torch.load', 'torch.load', (['high_dim_checkpoint_filepath'], {}), '(high_dim_checkpoint_filepath)\n', (15692, 15722), False, 'import torch\n'), ((16119, 16134), 'json.load', 'json.load', (['file'], {}), '(file)\n', (16128, 16134), False, 'import json\n'), ((16461, 16479), 'tqdm.tqdm', 'tqdm', (['test_vid_ids'], {}), '(test_vid_ids)\n', (16465, 16479), False, 'from tqdm import tqdm\n'), ((19213, 19231), 'tqdm.tqdm', 'tqdm', (['test_vid_ids'], {}), '(test_vid_ids)\n', (19217, 19231), False, 'from tqdm import tqdm\n'), ((1316, 1338), 'yaml.safe_load', 'yaml.safe_load', (['stream'], {}), '(stream)\n', (1330, 1338), False, 'import yaml\n'), ((3858, 3934), 'os.path.join', 'os.path.join', (['"""../datainfo"""', 'cfg.dataset', 'f"""data_split_dict_{cfg.seed}.json"""'], {}), "('../datainfo', cfg.dataset, f'data_split_dict_{cfg.seed}.json')\n", (3870, 3934), False, 'import os\n'), ((6557, 6593), 'json.dump', 'json.dump', (['loss_dict', 'file'], {'indent': '(4)'}), '(loss_dict, file, indent=4)\n', (6566, 6593), False, 'import json\n'), ((8998, 9034), 'json.dump', 'json.dump', (['loss_dict', 'file'], {'indent': '(4)'}), '(loss_dict, file, indent=4)\n', (9007, 9034), False, 'import json\n'), ((10342, 10385), 'os.path.join', 'os.path.join', (['checkpoint_filepath', '"""*.ckpt"""'], {}), "(checkpoint_filepath, '*.ckpt')\n", (10354, 10385), False, 'import os\n'), ((10615, 10667), 'os.path.join', 'os.path.join', (['high_dim_checkpoint_filepath', '"""*.ckpt"""'], {}), "(high_dim_checkpoint_filepath, '*.ckpt')\n", (10627, 10667), False, 'import os\n'), ((11001, 11077), 'os.path.join', 'os.path.join', (['"""../datainfo"""', 'cfg.dataset', 'f"""data_split_dict_{cfg.seed}.json"""'], {}), "('../datainfo', cfg.dataset, f'data_split_dict_{cfg.seed}.json')\n", (11013, 11077), False, 'import os\n'), ((11456, 11480), 'os.listdir', 'os.listdir', (['vid_filepath'], {}), '(vid_filepath)\n', (11466, 11480), False, 'import os\n'), ((13628, 13676), 'os.path.join', 'os.path.join', (['long_term_folder', '"""test_loss.json"""'], {}), "(long_term_folder, 'test_loss.json')\n", (13640, 13676), False, 'import os\n'), ((16008, 16084), 'os.path.join', 'os.path.join', (['"""../datainfo"""', 'cfg.dataset', 'f"""data_split_dict_{cfg.seed}.json"""'], {}), "('../datainfo', cfg.dataset, f'data_split_dict_{cfg.seed}.json')\n", (16020, 16084), False, 'import os\n'), ((19115, 19151), 'json.dump', 'json.dump', (['loss_dict', 'file'], {'indent': '(4)'}), '(loss_dict, file, indent=4)\n', (19124, 19151), False, 'import json\n'), ((21920, 21956), 'json.dump', 'json.dump', (['loss_dict', 'file'], {'indent': '(4)'}), '(loss_dict, file, indent=4)\n', (21929, 21956), False, 'import json\n'), ((2895, 2938), 'os.path.join', 'os.path.join', (['checkpoint_filepath', '"""*.ckpt"""'], {}), "(checkpoint_filepath, '*.ckpt')\n", (2907, 2938), False, 'import os\n'), ((3167, 3210), 'os.path.join', 'os.path.join', (['checkpoint_filepath', '"""*.ckpt"""'], {}), "(checkpoint_filepath, '*.ckpt')\n", (3179, 3210), False, 'import os\n'), ((3460, 3512), 'os.path.join', 'os.path.join', (['high_dim_checkpoint_filepath', '"""*.ckpt"""'], {}), "(high_dim_checkpoint_filepath, '*.ckpt')\n", (3472, 3512), False, 'import os\n'), ((4402, 4426), 'os.listdir', 'os.listdir', (['vid_filepath'], {}), '(vid_filepath)\n', (4412, 4426), False, 'import os\n'), ((6481, 6529), 'os.path.join', 'os.path.join', (['long_term_folder', '"""test_loss.json"""'], {}), "(long_term_folder, 'test_loss.json')\n", (6493, 6529), False, 'import os\n'), ((6790, 6814), 'os.listdir', 'os.listdir', (['vid_filepath'], {}), '(vid_filepath)\n', (6800, 6814), False, 'import os\n'), ((8922, 8970), 'os.path.join', 'os.path.join', (['long_term_folder', '"""test_loss.json"""'], {}), "(long_term_folder, 'test_loss.json')\n", (8934, 8970), False, 'import os\n'), ((13186, 13244), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 2}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 2}.{suf}')\n", (13198, 13244), False, 'import os\n'), ((13320, 13378), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 3}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 3}.{suf}')\n", (13332, 13378), False, 'import os\n'), ((15045, 15088), 'os.path.join', 'os.path.join', (['checkpoint_filepath', '"""*.ckpt"""'], {}), "(checkpoint_filepath, '*.ckpt')\n", (15057, 15088), False, 'import os\n'), ((15317, 15360), 'os.path.join', 'os.path.join', (['checkpoint_filepath', '"""*.ckpt"""'], {}), "(checkpoint_filepath, '*.ckpt')\n", (15329, 15360), False, 'import os\n'), ((15610, 15662), 'os.path.join', 'os.path.join', (['high_dim_checkpoint_filepath', '"""*.ckpt"""'], {}), "(high_dim_checkpoint_filepath, '*.ckpt')\n", (15622, 15662), False, 'import os\n'), ((16592, 16616), 'os.listdir', 'os.listdir', (['vid_filepath'], {}), '(vid_filepath)\n', (16602, 16616), False, 'import os\n'), ((19039, 19087), 'os.path.join', 'os.path.join', (['long_term_folder', '"""test_loss.json"""'], {}), "(long_term_folder, 'test_loss.json')\n", (19051, 19087), False, 'import os\n'), ((19344, 19368), 'os.listdir', 'os.listdir', (['vid_filepath'], {}), '(vid_filepath)\n', (19354, 19368), False, 'import os\n'), ((21844, 21892), 'os.path.join', 'os.path.join', (['long_term_folder', '"""test_loss.json"""'], {}), "(long_term_folder, 'test_loss.json')\n", (21856, 21892), False, 'import os\n'), ((22973, 23009), 'numpy.array_equal', 'np.array_equal', (['data[i, j]', 'bg_color'], {}), '(data[i, j], bg_color)\n', (22987, 23009), True, 'import numpy as np\n'), ((23601, 23624), 'torchvision.transforms.ToPILImage', 'transforms.ToPILImage', ([], {}), '()\n', (23622, 23624), False, 'from torchvision import transforms\n'), ((6011, 6069), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 2}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 2}.{suf}')\n", (6023, 6069), False, 'import os\n'), ((6153, 6211), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 3}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 3}.{suf}')\n", (6165, 6211), False, 'import os\n'), ((8452, 8510), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 2}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 2}.{suf}')\n", (8464, 8510), False, 'import os\n'), ((8594, 8652), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 3}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 3}.{suf}')\n", (8606, 8652), False, 'import os\n'), ((12212, 12270), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 2}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 2}.{suf}')\n", (12224, 12270), False, 'import os\n'), ((12302, 12360), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 3}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 3}.{suf}')\n", (12314, 12360), False, 'import os\n'), ((12383, 12403), 'torch.cat', 'torch.cat', (['target', '(2)'], {}), '(target, 2)\n', (12392, 12403), False, 'import torch\n'), ((18569, 18627), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 2}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 2}.{suf}')\n", (18581, 18627), False, 'import os\n'), ((18711, 18769), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 3}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 3}.{suf}')\n", (18723, 18769), False, 'import os\n'), ((21374, 21432), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 2}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 2}.{suf}')\n", (21386, 21432), False, 'import os\n'), ((21516, 21574), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 3}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 3}.{suf}')\n", (21528, 21574), False, 'import os\n'), ((5218, 5276), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 2}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 2}.{suf}')\n", (5230, 5276), False, 'import os\n'), ((5312, 5370), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 3}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 3}.{suf}')\n", (5324, 5370), False, 'import os\n'), ((5397, 5417), 'torch.cat', 'torch.cat', (['target', '(2)'], {}), '(target, 2)\n', (5406, 5417), False, 'import torch\n'), ((7606, 7664), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 2}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 2}.{suf}')\n", (7618, 7664), False, 'import os\n'), ((7700, 7758), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 3}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 3}.{suf}')\n", (7712, 7758), False, 'import os\n'), ((7785, 7805), 'torch.cat', 'torch.cat', (['target', '(2)'], {}), '(target, 2)\n', (7794, 7805), False, 'import torch\n'), ((11496, 11520), 'os.listdir', 'os.listdir', (['vid_filepath'], {}), '(vid_filepath)\n', (11506, 11520), False, 'import os\n'), ((11946, 12000), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx}.{suf}')\n", (11958, 12000), False, 'import os\n'), ((12036, 12094), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 1}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 1}.{suf}')\n", (12048, 12094), False, 'import os\n'), ((12119, 12137), 'torch.cat', 'torch.cat', (['data', '(2)'], {}), '(data, 2)\n', (12128, 12137), False, 'import torch\n'), ((17504, 17558), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx}.{suf}')\n", (17516, 17558), False, 'import os\n'), ((17650, 17708), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 1}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 1}.{suf}')\n", (17662, 17708), False, 'import os\n'), ((17776, 17834), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 2}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 2}.{suf}')\n", (17788, 17834), False, 'import os\n'), ((17870, 17928), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 3}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 3}.{suf}')\n", (17882, 17928), False, 'import os\n'), ((17955, 17975), 'torch.cat', 'torch.cat', (['target', '(2)'], {}), '(target, 2)\n', (17964, 17975), False, 'import torch\n'), ((20256, 20310), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx}.{suf}')\n", (20268, 20310), False, 'import os\n'), ((20402, 20460), 'os.path.join', 'os.path.join', (['saved_folder', 'f"""{start_frame_idx + 1}.{suf}"""'], {}), "(saved_folder, f'{start_frame_idx + 1}.{suf}')\n", (20414, 20460), False, 'import os\n'), ((20528, 20586), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 2}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 2}.{suf}')\n", (20540, 20586), False, 'import os\n'), ((20622, 20680), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 3}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 3}.{suf}')\n", (20634, 20680), False, 'import os\n'), ((20707, 20727), 'torch.cat', 'torch.cat', (['target', '(2)'], {}), '(target, 2)\n', (20716, 20727), False, 'import torch\n'), ((4446, 4470), 'os.listdir', 'os.listdir', (['vid_filepath'], {}), '(vid_filepath)\n', (4456, 4470), False, 'import os\n'), ((4936, 4990), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx}.{suf}')\n", (4948, 4990), False, 'import os\n'), ((5030, 5088), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 1}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 1}.{suf}')\n", (5042, 5088), False, 'import os\n'), ((5117, 5135), 'torch.cat', 'torch.cat', (['data', '(2)'], {}), '(data, 2)\n', (5126, 5135), False, 'import torch\n'), ((6834, 6858), 'os.listdir', 'os.listdir', (['vid_filepath'], {}), '(vid_filepath)\n', (6844, 6858), False, 'import os\n'), ((7324, 7378), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx}.{suf}')\n", (7336, 7378), False, 'import os\n'), ((7418, 7476), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 1}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 1}.{suf}')\n", (7430, 7476), False, 'import os\n'), ((7505, 7523), 'torch.cat', 'torch.cat', (['data', '(2)'], {}), '(data, 2)\n', (7514, 7523), False, 'import torch\n'), ((16636, 16660), 'os.listdir', 'os.listdir', (['vid_filepath'], {}), '(vid_filepath)\n', (16646, 16660), False, 'import os\n'), ((17134, 17188), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx}.{suf}')\n", (17146, 17188), False, 'import os\n'), ((17265, 17323), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 1}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 1}.{suf}')\n", (17277, 17323), False, 'import os\n'), ((17381, 17399), 'torch.cat', 'torch.cat', (['data', '(2)'], {}), '(data, 2)\n', (17390, 17399), False, 'import torch\n'), ((19388, 19412), 'os.listdir', 'os.listdir', (['vid_filepath'], {}), '(vid_filepath)\n', (19398, 19412), False, 'import os\n'), ((19886, 19940), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx}.{suf}')\n", (19898, 19940), False, 'import os\n'), ((20017, 20075), 'os.path.join', 'os.path.join', (['vid_filepath', 'f"""{start_frame_idx + 1}.{suf}"""'], {}), "(vid_filepath, f'{start_frame_idx + 1}.{suf}')\n", (20029, 20075), False, 'import os\n'), ((20133, 20151), 'torch.cat', 'torch.cat', (['data', '(2)'], {}), '(data, 2)\n', (20142, 20151), False, 'import torch\n')] |
from keras.utils import Sequence
import os
import signal
import psutil
import gc
import pandas as pd
import numpy as np
import random
import math
import pysam
from ..util import *
import threading
import pickle
import pdb
def kill_child_processes(parent_pid, sig=signal.SIGTERM):
try:
parent = psutil.Process(parent_pid)
except psutil.NoSuchProcess:
return
children = parent.children(recursive=True)
for process in children:
process.send_signal(sig)
def get_weights(data):
w1=[float(data.shape[0])/sum(data.iloc[:,i]==1) for i in range(data.shape[1])]
w0=[float(data.shape[0])/sum(data.iloc[:,i]==0) for i in range(data.shape[1])]
return w1,w0
def open_data_file(data_path=None,tasks=None,chroms_to_use=None):
print("running open_data_file with tasks:"+str(tasks))
if data_path.endswith('.hdf5'):
if tasks==None:
data=pd.read_hdf(data_path)
else:
data=pd.read_hdf(data_path,columns=['CHR','START','END']+tasks)
else:
#treat as bed file
if tasks==None:
data=pd.read_csv(data_path,header=0,sep='\t')
else:
data=pd.read_csv(data_path,header=0,sep='\t',nrows=1)
chrom_col=data.columns[0]
start_col=data.columns[1]
end_col=data.columns[2]
data=pd.read_csv(data_path,header=0,sep='\t',usecols=[chrom_col,start_col,end_col]+tasks)
print("loaded labels")
print("chroms_to_use:"+str(chroms_to_use))
print(data.head())
try:
data=data.set_index(['CHR','START','END'])
print('set index to CHR, START, END')
except:
pass
if chroms_to_use!=None:
data=data[np.in1d(data.index.get_level_values(0), chroms_to_use)]
print("filtered on chroms_to_use")
print("data.shape:"+str(data.shape), data.columns)
return data
#use wrappers for keras Sequence generator class to allow batch shuffling upon epoch end
class DataGenerator(Sequence):
def __init__(self,
index_path,
input_path,
output_path,
num_inputs,
num_outputs,
ref_fasta=None,
batch_size=128,
add_revcomp=False,
index_tasks=None,
tasks=None,
shuffled_ref_negatives=False,
chroms_to_use=None,
get_w1_w0=False,
expand_dims=True,
upsample_thresh_list=None,
upsample_ratio_list=None,
shuffle=True,
return_coords=False):
self.return_coords=return_coords
self.expand_dims=expand_dims
self.shuffle=shuffle
self.batch_size=batch_size
self.ref_fasta=ref_fasta
self.chroms_to_use=chroms_to_use
#decide if reverse complement should be used
self.add_revcomp=add_revcomp
if add_revcomp==True:
self.batch_size=int(batch_size/2)
#determine whether negative set should consist of the shuffled refs.
# If so, split batch size in 2, as each batch will be augmented with shuffled ref negatives
# in ratio equal to positives
self.shuffled_ref_negatives=shuffled_ref_negatives
if self.shuffled_ref_negatives==True:
self.batch_size=int(self.batch_size/2)
#get the index, input, and output files
self.index_tasks=index_tasks
if tasks is None:
tasks=[None]*num_inputs
else:
tasks=[i.split(',') for i in tasks] + [None]*(num_inputs-1)
self.tasks=tasks
print("TASKS:"+str(self.tasks))
self.index_path=index_path
self.input_path=input_path
self.output_path=output_path
self.num_inputs=num_inputs
self.num_outputs=num_outputs
self.file_to_pd=self.get_file_to_pd()
self.indices=self.file_to_pd[self.index_path]
self.num_indices=self.indices.shape[0]
print("indices:"+str(self.indices.head()))
print("num_indices:"+str(self.num_indices))
#handle task-specific weights -- this is a bit outdated and may be removed in the future.
if get_w1_w0==True:
assert self.data is not None
w1,w0=get_weights(self.data)
self.w1=w1
self.w0=w0
#set variables needed for upsampling the positives
self.upsample_thresh_list=upsample_thresh_list
self.upsample_ratio_list=upsample_ratio_list
#generate the upsampled threshold index subgroups
print("creating upsampling logic for generator")
print("self.upsample_thresh_list:"+str(self.upsample_thresh_list))
if self.upsample_thresh_list is not None:
self.get_upsampled_indices()
else:
self.indices=self.indices.index.tolist()
if self.shuffle == True:
np.random.shuffle(self.indices)
self.lock=threading.Lock()
print("generator initialized")
def get_file_to_pd(self):
'''
make sure all input/output/index files are loaded only once in case there's overlap
generate a dictionary of file name to pandas data frame of file contents
'''
file_to_df={}
print(self.index_path)
print(self.tasks)
if self.tasks[0] is not None:
file_to_df[self.index_path]=open_data_file(data_path=self.index_path,tasks=[ti[0] for ti in self.tasks if ti is not None],chroms_to_use=self.chroms_to_use)
else:
file_to_df[self.index_path]=open_data_file(data_path=self.index_path,tasks=self.index_tasks,chroms_to_use=self.chroms_to_use)
print("got index_path df")
for i in range(self.num_inputs):
cur_input=self.input_path[i]
print(cur_input)
if cur_input=="seq":
continue
if cur_input in file_to_df:
continue
file_to_df[self.input_path[i]]=open_data_file(data_path=cur_input,tasks=self.tasks[i],chroms_to_use=self.chroms_to_use)
print('got input')
for i in range(self.num_outputs):
cur_output=self.output_path[i]
print(cur_output)
if cur_output in file_to_df:
print('skipped output reading')
continue
file_to_df[cur_output]=open_data_file(data_path=cur_output,tasks=self.tasks[i],chroms_to_use=self.chroms_to_use)
return file_to_df
def get_upsampled_indices(self):
'''
several levels of upsampling are handled
self.upsample_thresh_list -- list of thresholds for upsampling the dataset
self.upsample_ratio_list -- fraction of batch to be generated at each threshold, length of this list = len(self.upsample_thresh_list)-1
i.e. upsample_thresh_list=[0,0.01,1], upsample_ratio_list=[0.7,0.2] means 70% of samples should be in the range [0,0.1), 20% of samples should be
in the range [0.1,1), and remaining 10% of samples are in the range [1,inf)
'''
self.upsampled_coord_indices = {}
self.upsampled_numerical_indices = {}
self.batch_sizes = []
print("upsample thresh list:"+str(self.upsample_thresh_list))
if self.upsample_ratio_list is None:
#all thresholds represented equally in the batch
self.upsample_ratio = 1 / (len(self.upsample_thresh_list) - 1)
self.upsample_ratio_list = [self.upsample_ratio for i in range(len(self.upsample_thresh_list) - 1)]
print("upsample ratio list: " + str(self.upsample_ratio_list))
for ind in range(len(self.upsample_thresh_list)-1):
lower_thresh_bound=self.upsample_thresh_list[ind]
upper_thresh_bound=self.upsample_thresh_list[ind+1]
#get the sub-batch that contains values within this value threshold
sub_batch_size=int(self.batch_size*self.upsample_ratio_list[ind])
self.batch_sizes.append(sub_batch_size)
#get the coordinates where all values fall in the range [lower_thresh_bound, upper_thresh_bound)
sub_batch_coords=self.indices.loc[(self.indices>=lower_thresh_bound).any(axis=1) & (self.indices < upper_thresh_bound).all(axis=1)].index
len_sub_batch_coords=len(sub_batch_coords)
self.upsampled_coord_indices[ind]=sub_batch_coords
self.upsampled_numerical_indices[ind] = np.arange(len_sub_batch_coords)
#number of times the current sub-set of upsampled indices should be wrapped to get to num_indices values
num_wraps=math.ceil(self.num_indices/len_sub_batch_coords)
self.upsampled_numerical_indices[ind] = np.tile(self.upsampled_numerical_indices[ind], num_wraps)[0:self.num_indices]
#shuffle the sub-set of indices, if specified
if self.shuffle == True:
np.random.shuffle(self.upsampled_numerical_indices[ind])
#handle the final index (i.e. unspecified upper bound)
ind=len(self.upsample_thresh_list)-1
lower_thresh_bound=self.upsample_thresh_list[ind]
sub_batch_size=int(self.batch_size-sum(self.batch_sizes))
self.batch_sizes.append(sub_batch_size)
sub_batch_coords=self.indices.loc[(self.indices>=lower_thresh_bound).any(axis=1)].index
len_sub_batch_coords=len(sub_batch_coords)
self.upsampled_coord_indices[ind]=sub_batch_coords
self.upsampled_numerical_indices[ind] = np.arange(len_sub_batch_coords)
#number of times the current sub-set of upsampled indices should be wrapped to get to num_indices values
num_wraps=math.ceil(self.num_indices/len_sub_batch_coords)
self.upsampled_numerical_indices[ind] = np.tile(self.upsampled_numerical_indices[ind], num_wraps)[0:self.num_indices]
#shuffle the sub-set of indices, if specified
if self.shuffle == True:
np.random.shuffle(self.upsampled_numerical_indices[ind])
del self.indices
return
def __len__(self):
return math.ceil(self.num_indices/self.batch_size)
def get_coords(self,idx):
if self.upsample_thresh_list is not None:
all_bed_entries=[]
for ind,val in enumerate(self.batch_sizes):
batch_indices = self.upsampled_numerical_indices[ind][idx*val:(idx+1)*val]
bed_entries = self.upsampled_coord_indices[ind][batch_indices].tolist()
all_bed_entries+=bed_entries
else:
all_bed_entries=self.indices[idx*self.batch_size:(idx+1)*self.batch_size]
return all_bed_entries
def get_seq(self,coords):
seqs=[self.ref.fetch(i[0],i[1],i[2]) for i in coords]
return seqs
def get_pd_vals(self,coords,io_index):
try:
return self.file_to_pd[io_index].loc[coords].values
except:
raise Exception("could not fetch coords:"+str(coords))
def transform_seq(self,seqs):
if self.add_revcomp==True:
#add in the reverse-complemented sequences for training.
seqs_rc=[revcomp(s) for s in seqs]
seqs=seqs+seqs_rc
if self.shuffled_ref_negatives is True:
#generate the corresponding negative set by dinucleotide-shuffling the sequences
seqs_shuffled=[dinuc_shuffle(s) for s in seqs]
seqs=seqs+seqs_shuffled
return seqs
def transform_vals(self,vals):
if self.add_revcomp==True:
vals=np.concatenate((vals,vals),axis=0)
if self.shuffled_ref_negatives is True:
val_shape=vals.shape
vals=np.concatenate((vals,np.zeros(vals_shape)))
return vals
def __getitem__(self,idx):
try:
gc.unfreeze()
self.lock.acquire()
#print("STARTING")
self.ref=pysam.FastaFile(self.ref_fasta)
#get the coordinates for the current batch
coords=self.get_coords(idx)
#print("GOT COORDS")
#get the inputs
X=[]
for cur_input_index in range(self.num_inputs):
cur_input=self.input_path[cur_input_index]
if cur_input=="seq":
cur_x=one_hot_encode(self.transform_seq(self.get_seq(coords)))
if self.expand_dims==True:
cur_x=np.expand_dims(cur_x,axis=1)
else:
#extract values from pandas df
cur_x=self.transform_vals(self.get_pd_vals(coords,cur_input))
X.append(cur_x)
#get the outputs
y=[]
for cur_output_index in range(self.num_outputs):
cur_output=self.output_path[cur_output_index]
if cur_output=="seq":
cur_y=one_hot_encode(self.transform_seq(self.get_seq(coords)))
if self.expand_dims==True:
cur_y=np.expand_dims(cur_y,axis=1)
else:
cur_y=self.transform_vals(self.get_pd_vals(coords,cur_output))
y.append(cur_y)
self.lock.release()
#return the batch as an X,y tuple
#print("SUCCESS")
if self.return_coords is False:
return (X,y)
else:
return (X,y,coords)
except Exception as e:
print(str(e)+" from id:"+str(idx))
kill_child_processes(os.getpid())
raise
def on_epoch_end(self):
#if upsampling is being used, shuffle the positive and negative indices
if self.shuffle==True:
if self.upsample_thresh_list is not None:
for ind,val in enumerate(self.batch_sizes):
np.random.shuffle(self.upsampled_numerical_indices[ind])
else:
np.random.shuffle(self.indices)
| [
"psutil.Process",
"os.getpid",
"pandas.read_hdf",
"numpy.concatenate",
"math.ceil",
"pandas.read_csv",
"pysam.FastaFile",
"numpy.zeros",
"numpy.expand_dims",
"threading.Lock",
"numpy.arange",
"numpy.tile",
"numpy.random.shuffle",
"gc.unfreeze"
] | [((308, 334), 'psutil.Process', 'psutil.Process', (['parent_pid'], {}), '(parent_pid)\n', (322, 334), False, 'import psutil\n'), ((5035, 5051), 'threading.Lock', 'threading.Lock', ([], {}), '()\n', (5049, 5051), False, 'import threading\n'), ((9662, 9693), 'numpy.arange', 'np.arange', (['len_sub_batch_coords'], {}), '(len_sub_batch_coords)\n', (9671, 9693), True, 'import numpy as np\n'), ((9834, 9884), 'math.ceil', 'math.ceil', (['(self.num_indices / len_sub_batch_coords)'], {}), '(self.num_indices / len_sub_batch_coords)\n', (9843, 9884), False, 'import math\n'), ((10267, 10312), 'math.ceil', 'math.ceil', (['(self.num_indices / self.batch_size)'], {}), '(self.num_indices / self.batch_size)\n', (10276, 10312), False, 'import math\n'), ((903, 925), 'pandas.read_hdf', 'pd.read_hdf', (['data_path'], {}), '(data_path)\n', (914, 925), True, 'import pandas as pd\n'), ((957, 1020), 'pandas.read_hdf', 'pd.read_hdf', (['data_path'], {'columns': "(['CHR', 'START', 'END'] + tasks)"}), "(data_path, columns=['CHR', 'START', 'END'] + tasks)\n", (968, 1020), True, 'import pandas as pd\n'), ((1094, 1136), 'pandas.read_csv', 'pd.read_csv', (['data_path'], {'header': '(0)', 'sep': '"""\t"""'}), "(data_path, header=0, sep='\\t')\n", (1105, 1136), True, 'import pandas as pd\n'), ((1166, 1217), 'pandas.read_csv', 'pd.read_csv', (['data_path'], {'header': '(0)', 'sep': '"""\t"""', 'nrows': '(1)'}), "(data_path, header=0, sep='\\t', nrows=1)\n", (1177, 1217), True, 'import pandas as pd\n'), ((1344, 1439), 'pandas.read_csv', 'pd.read_csv', (['data_path'], {'header': '(0)', 'sep': '"""\t"""', 'usecols': '([chrom_col, start_col, end_col] + tasks)'}), "(data_path, header=0, sep='\\t', usecols=[chrom_col, start_col,\n end_col] + tasks)\n", (1355, 1439), True, 'import pandas as pd\n'), ((8578, 8609), 'numpy.arange', 'np.arange', (['len_sub_batch_coords'], {}), '(len_sub_batch_coords)\n', (8587, 8609), True, 'import numpy as np\n'), ((8763, 8813), 'math.ceil', 'math.ceil', (['(self.num_indices / len_sub_batch_coords)'], {}), '(self.num_indices / len_sub_batch_coords)\n', (8772, 8813), False, 'import math\n'), ((9931, 9988), 'numpy.tile', 'np.tile', (['self.upsampled_numerical_indices[ind]', 'num_wraps'], {}), '(self.upsampled_numerical_indices[ind], num_wraps)\n', (9938, 9988), True, 'import numpy as np\n'), ((10109, 10165), 'numpy.random.shuffle', 'np.random.shuffle', (['self.upsampled_numerical_indices[ind]'], {}), '(self.upsampled_numerical_indices[ind])\n', (10126, 10165), True, 'import numpy as np\n'), ((11740, 11776), 'numpy.concatenate', 'np.concatenate', (['(vals, vals)'], {'axis': '(0)'}), '((vals, vals), axis=0)\n', (11754, 11776), True, 'import numpy as np\n'), ((12012, 12025), 'gc.unfreeze', 'gc.unfreeze', ([], {}), '()\n', (12023, 12025), False, 'import gc\n'), ((12111, 12142), 'pysam.FastaFile', 'pysam.FastaFile', (['self.ref_fasta'], {}), '(self.ref_fasta)\n', (12126, 12142), False, 'import pysam\n'), ((4985, 5016), 'numpy.random.shuffle', 'np.random.shuffle', (['self.indices'], {}), '(self.indices)\n', (5002, 5016), True, 'import numpy as np\n'), ((8864, 8921), 'numpy.tile', 'np.tile', (['self.upsampled_numerical_indices[ind]', 'num_wraps'], {}), '(self.upsampled_numerical_indices[ind], num_wraps)\n', (8871, 8921), True, 'import numpy as np\n'), ((9054, 9110), 'numpy.random.shuffle', 'np.random.shuffle', (['self.upsampled_numerical_indices[ind]'], {}), '(self.upsampled_numerical_indices[ind])\n', (9071, 9110), True, 'import numpy as np\n'), ((14132, 14163), 'numpy.random.shuffle', 'np.random.shuffle', (['self.indices'], {}), '(self.indices)\n', (14149, 14163), True, 'import numpy as np\n'), ((11895, 11915), 'numpy.zeros', 'np.zeros', (['vals_shape'], {}), '(vals_shape)\n', (11903, 11915), True, 'import numpy as np\n'), ((13732, 13743), 'os.getpid', 'os.getpid', ([], {}), '()\n', (13741, 13743), False, 'import os\n'), ((14041, 14097), 'numpy.random.shuffle', 'np.random.shuffle', (['self.upsampled_numerical_indices[ind]'], {}), '(self.upsampled_numerical_indices[ind])\n', (14058, 14097), True, 'import numpy as np\n'), ((12632, 12661), 'numpy.expand_dims', 'np.expand_dims', (['cur_x'], {'axis': '(1)'}), '(cur_x, axis=1)\n', (12646, 12661), True, 'import numpy as np\n'), ((13217, 13246), 'numpy.expand_dims', 'np.expand_dims', (['cur_y'], {'axis': '(1)'}), '(cur_y, axis=1)\n', (13231, 13246), True, 'import numpy as np\n')] |
'''
This script executes 2D FFT convolution on images in grayscale.
Usage:
Run without argument will use builtin Lena image:
python fftconvolve.py
Or, specify an image to use
python fftconvolve.py myimage.jpg
python fftconvolve.py myimage.png
= Getting The Requirements =
For Conda user, run the following to ensure the dependencies are fulfilled:
conda install scipy matplotlib
You may need to install PIL from pip.
conda install pip
pip install PIL
DvG Note: conda install -c numba pyculib (deprecated though)
'''
from __future__ import print_function
import sys
from timeit import default_timer as timer
import numpy as np
from scipy.signal import fftconvolve
from scipy import misc, ndimage
from matplotlib import pyplot as plt
from accelerate.cuda.fft import FFTPlan
from numba import cuda
@cuda.jit('void(complex64[:,:], complex64[:,:])')
def mult_inplace(img, resp):
i, j = cuda.grid(2)
if j < img.shape[0] and i < img.shape[1]:
img[j, i] *= resp[j, i]
def best_grid_size(size, tpb):
bpg = np.ceil(np.array(size, dtype=np.float) / tpb).astype(np.int).tolist()
return tuple(bpg)
def main():
# Build Filter
laplacian_pts = '''
-4 -1 0 -1 -4
-1 2 3 2 -1
0 3 4 3 0
-1 2 3 2 -1
-4 -1 0 -1 -4
'''.split()
laplacian = np.array(laplacian_pts, dtype=np.float32).reshape(5, 5)
# Build Image
try:
filename = sys.argv[1]
image = ndimage.imread(filename, flatten=True).astype(np.float32)
except IndexError:
image = misc.face(gray=True).astype(np.float32)
print("Image size: %s" % (image.shape,))
response = np.zeros_like(image)
response[:5, :5] = laplacian
# CPU
ts = timer()
cvimage_cpu = fftconvolve(image, laplacian, mode='same')
te = timer()
print('CPU: %.2fs' % (te - ts))
# GPU
threadperblock = 32, 8
blockpergrid = best_grid_size(tuple(reversed(image.shape)), threadperblock)
print('kernel config: %s x %s' % (blockpergrid, threadperblock))
# Trigger initialization the cuFFT system.
# This takes significant time for small dataset.
# We should not be including the time wasted here
FFTPlan(shape=image.shape, itype=np.complex64, otype=np.complex64)
# Start GPU timer
ts = timer()
image_complex = image.astype(np.complex64)
response_complex = response.astype(np.complex64)
stream1 = cuda.stream()
stream2 = cuda.stream()
fftplan1 = FFTPlan(shape=image.shape, itype=np.complex64,
otype=np.complex64, stream=stream1)
fftplan2 = FFTPlan(shape=image.shape, itype=np.complex64,
otype=np.complex64, stream=stream2)
# pagelock memory
with cuda.pinned(image_complex, response_complex):
# We can overlap the transfer of response_complex with the forward FFT
# on image_complex.
d_image_complex = cuda.to_device(image_complex, stream=stream1)
d_response_complex = cuda.to_device(response_complex, stream=stream2)
fftplan1.forward(d_image_complex, out=d_image_complex)
fftplan2.forward(d_response_complex, out=d_response_complex)
stream2.synchronize()
mult_inplace[blockpergrid, threadperblock, stream1](d_image_complex,
d_response_complex)
fftplan1.inverse(d_image_complex, out=d_image_complex)
# implicitly synchronizes the streams
cvimage_gpu = d_image_complex.copy_to_host().real / np.prod(image.shape)
te = timer()
print('GPU: %.2fs' % (te - ts))
# Plot the results
plt.subplot(1, 2, 1)
plt.title('CPU')
plt.imshow(cvimage_cpu, cmap=plt.cm.gray)
plt.axis('off')
plt.subplot(1, 2, 2)
plt.title('GPU')
plt.imshow(cvimage_gpu, cmap=plt.cm.gray)
plt.axis('off')
plt.show()
if __name__ == '__main__':
main()
| [
"matplotlib.pyplot.title",
"matplotlib.pyplot.subplot",
"numpy.zeros_like",
"numba.cuda.stream",
"matplotlib.pyplot.show",
"timeit.default_timer",
"accelerate.cuda.fft.FFTPlan",
"matplotlib.pyplot.imshow",
"scipy.misc.face",
"matplotlib.pyplot.axis",
"numba.cuda.to_device",
"numba.cuda.pinned"... | [((837, 885), 'numba.cuda.jit', 'cuda.jit', (['"""void(complex64[:,:], complex64[:,:])"""'], {}), "('void(complex64[:,:], complex64[:,:])')\n", (845, 885), False, 'from numba import cuda\n'), ((926, 938), 'numba.cuda.grid', 'cuda.grid', (['(2)'], {}), '(2)\n', (935, 938), False, 'from numba import cuda\n'), ((1654, 1674), 'numpy.zeros_like', 'np.zeros_like', (['image'], {}), '(image)\n', (1667, 1674), True, 'import numpy as np\n'), ((1728, 1735), 'timeit.default_timer', 'timer', ([], {}), '()\n', (1733, 1735), True, 'from timeit import default_timer as timer\n'), ((1754, 1796), 'scipy.signal.fftconvolve', 'fftconvolve', (['image', 'laplacian'], {'mode': '"""same"""'}), "(image, laplacian, mode='same')\n", (1765, 1796), False, 'from scipy.signal import fftconvolve\n'), ((1806, 1813), 'timeit.default_timer', 'timer', ([], {}), '()\n', (1811, 1813), True, 'from timeit import default_timer as timer\n'), ((2196, 2262), 'accelerate.cuda.fft.FFTPlan', 'FFTPlan', ([], {'shape': 'image.shape', 'itype': 'np.complex64', 'otype': 'np.complex64'}), '(shape=image.shape, itype=np.complex64, otype=np.complex64)\n', (2203, 2262), False, 'from accelerate.cuda.fft import FFTPlan\n'), ((2295, 2302), 'timeit.default_timer', 'timer', ([], {}), '()\n', (2300, 2302), True, 'from timeit import default_timer as timer\n'), ((2418, 2431), 'numba.cuda.stream', 'cuda.stream', ([], {}), '()\n', (2429, 2431), False, 'from numba import cuda\n'), ((2446, 2459), 'numba.cuda.stream', 'cuda.stream', ([], {}), '()\n', (2457, 2459), False, 'from numba import cuda\n'), ((2476, 2563), 'accelerate.cuda.fft.FFTPlan', 'FFTPlan', ([], {'shape': 'image.shape', 'itype': 'np.complex64', 'otype': 'np.complex64', 'stream': 'stream1'}), '(shape=image.shape, itype=np.complex64, otype=np.complex64, stream=\n stream1)\n', (2483, 2563), False, 'from accelerate.cuda.fft import FFTPlan\n'), ((2597, 2684), 'accelerate.cuda.fft.FFTPlan', 'FFTPlan', ([], {'shape': 'image.shape', 'itype': 'np.complex64', 'otype': 'np.complex64', 'stream': 'stream2'}), '(shape=image.shape, itype=np.complex64, otype=np.complex64, stream=\n stream2)\n', (2604, 2684), False, 'from accelerate.cuda.fft import FFTPlan\n'), ((3562, 3569), 'timeit.default_timer', 'timer', ([], {}), '()\n', (3567, 3569), True, 'from timeit import default_timer as timer\n'), ((3634, 3654), 'matplotlib.pyplot.subplot', 'plt.subplot', (['(1)', '(2)', '(1)'], {}), '(1, 2, 1)\n', (3645, 3654), True, 'from matplotlib import pyplot as plt\n'), ((3659, 3675), 'matplotlib.pyplot.title', 'plt.title', (['"""CPU"""'], {}), "('CPU')\n", (3668, 3675), True, 'from matplotlib import pyplot as plt\n'), ((3680, 3721), 'matplotlib.pyplot.imshow', 'plt.imshow', (['cvimage_cpu'], {'cmap': 'plt.cm.gray'}), '(cvimage_cpu, cmap=plt.cm.gray)\n', (3690, 3721), True, 'from matplotlib import pyplot as plt\n'), ((3726, 3741), 'matplotlib.pyplot.axis', 'plt.axis', (['"""off"""'], {}), "('off')\n", (3734, 3741), True, 'from matplotlib import pyplot as plt\n'), ((3747, 3767), 'matplotlib.pyplot.subplot', 'plt.subplot', (['(1)', '(2)', '(2)'], {}), '(1, 2, 2)\n', (3758, 3767), True, 'from matplotlib import pyplot as plt\n'), ((3772, 3788), 'matplotlib.pyplot.title', 'plt.title', (['"""GPU"""'], {}), "('GPU')\n", (3781, 3788), True, 'from matplotlib import pyplot as plt\n'), ((3793, 3834), 'matplotlib.pyplot.imshow', 'plt.imshow', (['cvimage_gpu'], {'cmap': 'plt.cm.gray'}), '(cvimage_gpu, cmap=plt.cm.gray)\n', (3803, 3834), True, 'from matplotlib import pyplot as plt\n'), ((3839, 3854), 'matplotlib.pyplot.axis', 'plt.axis', (['"""off"""'], {}), "('off')\n", (3847, 3854), True, 'from matplotlib import pyplot as plt\n'), ((3860, 3870), 'matplotlib.pyplot.show', 'plt.show', ([], {}), '()\n', (3868, 3870), True, 'from matplotlib import pyplot as plt\n'), ((2735, 2779), 'numba.cuda.pinned', 'cuda.pinned', (['image_complex', 'response_complex'], {}), '(image_complex, response_complex)\n', (2746, 2779), False, 'from numba import cuda\n'), ((2915, 2960), 'numba.cuda.to_device', 'cuda.to_device', (['image_complex'], {'stream': 'stream1'}), '(image_complex, stream=stream1)\n', (2929, 2960), False, 'from numba import cuda\n'), ((2990, 3038), 'numba.cuda.to_device', 'cuda.to_device', (['response_complex'], {'stream': 'stream2'}), '(response_complex, stream=stream2)\n', (3004, 3038), False, 'from numba import cuda\n'), ((1324, 1365), 'numpy.array', 'np.array', (['laplacian_pts'], {'dtype': 'np.float32'}), '(laplacian_pts, dtype=np.float32)\n', (1332, 1365), True, 'import numpy as np\n'), ((3531, 3551), 'numpy.prod', 'np.prod', (['image.shape'], {}), '(image.shape)\n', (3538, 3551), True, 'import numpy as np\n'), ((1455, 1493), 'scipy.ndimage.imread', 'ndimage.imread', (['filename'], {'flatten': '(True)'}), '(filename, flatten=True)\n', (1469, 1493), False, 'from scipy import misc, ndimage\n'), ((1552, 1572), 'scipy.misc.face', 'misc.face', ([], {'gray': '(True)'}), '(gray=True)\n', (1561, 1572), False, 'from scipy import misc, ndimage\n'), ((1068, 1098), 'numpy.array', 'np.array', (['size'], {'dtype': 'np.float'}), '(size, dtype=np.float)\n', (1076, 1098), True, 'import numpy as np\n')] |
"""
Contains class that runs inferencing
"""
import torch
import numpy as np
from networks.RecursiveUNet import UNet
from utils.utils import med_reshape
import torch.nn.functional as F
class UNetInferenceAgent:
"""
Stores model and parameters and some methods to handle inferencing
"""
def __init__(self, parameter_file_path='', model=None, device="cpu", patch_size=64):
self.model = model
self.patch_size = patch_size
self.device = device
if model is None:
self.model = UNet(num_classes=3)
if parameter_file_path:
self.model.load_state_dict(torch.load(parameter_file_path, map_location=self.device))
self.model.to(device)
def single_volume_inference_unpadded(self, volume):
"""
Runs inference on a single volume of arbitrary patch size,
padding it to the conformant size first
Arguments:
volume {Numpy array} -- 3D array representing the volume
Returns:
3D NumPy array with prediction mask
"""
volume =(volume.astype(float))/np.max(volume)
reshaped_image = med_reshape(volume, new_shape=(volume.shape[0], self.patch_size, self.patch_size)).astype(np.single)
return reshaped_image
#raise NotImplementedError
def single_volume_inference(self, volume):
"""
Runs inference on a single volume of conformant patch size
Arguments:
volume {Numpy array} -- 3D array representing the volume
Returns:
3D NumPy array with prediction mask
"""
# Assuming volume is a numpy array of shape [X,Y,Z] and we need to slice X axis
# TASK: Write code that will create mask for each slice across the X (0th) dimension. After
# that, put all slices into a 3D Numpy array. You can verify if your method is
# correct by running it on one of the volumes in your training set and comparing
# with the label in 3D Slicer.
volume=self.single_volume_inference_unpadded(volume)
volume=volume.reshape([volume.shape[0],1,volume.shape[1],volume.shape[2]])
#slices = []
self.model.eval()
with torch.no_grad():
#for batch_ind in range()
data = torch.from_numpy(volume).to(self.device)
print(data.shape)
prediction = self.model(data)
prediction_softmax = F.softmax(prediction, dim=1)
res=(prediction_softmax).squeeze(0).cpu().detach().numpy()
slices=res
return slices | [
"utils.utils.med_reshape",
"torch.load",
"torch.nn.functional.softmax",
"numpy.max",
"networks.RecursiveUNet.UNet",
"torch.no_grad",
"torch.from_numpy"
] | [((535, 554), 'networks.RecursiveUNet.UNet', 'UNet', ([], {'num_classes': '(3)'}), '(num_classes=3)\n', (539, 554), False, 'from networks.RecursiveUNet import UNet\n'), ((1107, 1121), 'numpy.max', 'np.max', (['volume'], {}), '(volume)\n', (1113, 1121), True, 'import numpy as np\n'), ((2217, 2232), 'torch.no_grad', 'torch.no_grad', ([], {}), '()\n', (2230, 2232), False, 'import torch\n'), ((2437, 2465), 'torch.nn.functional.softmax', 'F.softmax', (['prediction'], {'dim': '(1)'}), '(prediction, dim=1)\n', (2446, 2465), True, 'import torch.nn.functional as F\n'), ((627, 684), 'torch.load', 'torch.load', (['parameter_file_path'], {'map_location': 'self.device'}), '(parameter_file_path, map_location=self.device)\n', (637, 684), False, 'import torch\n'), ((1147, 1234), 'utils.utils.med_reshape', 'med_reshape', (['volume'], {'new_shape': '(volume.shape[0], self.patch_size, self.patch_size)'}), '(volume, new_shape=(volume.shape[0], self.patch_size, self.\n patch_size))\n', (1158, 1234), False, 'from utils.utils import med_reshape\n'), ((2291, 2315), 'torch.from_numpy', 'torch.from_numpy', (['volume'], {}), '(volume)\n', (2307, 2315), False, 'import torch\n')] |
# Copyright 2020 The Magenta Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Utilities for fine alignment.
CQT calculations and NoteSequence manipulations are done in Python. For speed,
DTW calculations are done in C++ by calling the 'align' program, which is
specifically intended to be used with this library. Communication between
Python and C++ is done with a protobuf.
"""
import os
import subprocess
import tempfile
from absl import logging
import alignment_pb2
import librosa
from magenta.music import midi_synth
from magenta.music import sequences_lib
import numpy as np
# Constants based on craffel's example alignment script:
# https://github.com/craffel/pretty-midi/blob/master/examples/align_midi.py
SAMPLE_RATE = 22050
CQT_HOP_LENGTH_FINE = 64 # ~3ms
CQT_N_BINS = 48
CQT_BINS_PER_OCTAVE = 12
CQT_FMIN = librosa.midi_to_hz(36)
ALIGN_BINARY = './align'
def extract_cqt(samples, sample_rate, cqt_hop_length):
"""Transforms the contents of a wav/mp3 file into a series of CQT frames."""
cqt = np.abs(librosa.core.cqt(
samples,
sample_rate,
hop_length=cqt_hop_length,
fmin=CQT_FMIN,
n_bins=CQT_N_BINS,
bins_per_octave=CQT_BINS_PER_OCTAVE), dtype=np.float32)
# Compute log-amplitude
cqt = librosa.power_to_db(cqt)
return cqt
def align_cpp(samples,
sample_rate,
ns,
cqt_hop_length,
sf2_path,
penalty_mul=1.0,
band_radius_seconds=.5):
"""Aligns the notesequence to the wav file using C++ DTW.
Args:
samples: Samples to align.
sample_rate: Sample rate for samples.
ns: The source notesequence to align.
cqt_hop_length: Hop length to use for CQT calculations.
sf2_path: Path to SF2 file for synthesis.
penalty_mul: Penalty multiplier to use for non-diagonal moves.
band_radius_seconds: What size of band radius to use for restricting DTW.
Raises:
RuntimeError: If notes are skipped during alignment.
Returns:
samples: The samples used from the wav file.
aligned_ns: The aligned version of the notesequence.
remaining_ns: Any remaining notesequence that extended beyond the length
of the wav file.
"""
logging.info('Synthesizing')
ns_samples = midi_synth.fluidsynth(
ns, sf2_path=sf2_path, sample_rate=sample_rate).astype(np.float32)
# It is critical that ns_samples and samples are the same length because the
# alignment code does not do subsequence alignment.
ns_samples = np.pad(ns_samples,
(0, max(0, samples.shape[0] - ns_samples.shape[0])),
'constant')
# Pad samples too, if needed, because there are some cases where the
# synthesized NoteSequence is actually longer.
samples = np.pad(samples,
(0, max(0, ns_samples.shape[0] - samples.shape[0])),
'constant')
# Note that we skip normalization here becasue it happens in C++.
logging.info('source_cqt')
source_cqt = extract_cqt(ns_samples, sample_rate, cqt_hop_length)
logging.info('dest_cqt')
dest_cqt = extract_cqt(samples, sample_rate, cqt_hop_length)
alignment_task = alignment_pb2.AlignmentTask()
alignment_task.sequence_1.x = source_cqt.shape[0]
alignment_task.sequence_1.y = source_cqt.shape[1]
for c in source_cqt.reshape([-1]):
alignment_task.sequence_1.content.append(c)
alignment_task.sequence_2.x = dest_cqt.shape[0]
alignment_task.sequence_2.y = dest_cqt.shape[1]
for c in dest_cqt.reshape([-1]):
alignment_task.sequence_2.content.append(c)
seconds_per_frame = cqt_hop_length / sample_rate
alignment_task.band_radius = int(band_radius_seconds / seconds_per_frame)
alignment_task.penalty = 0
alignment_task.penalty_mul = penalty_mul
# Write to file.
fh, temp_path = tempfile.mkstemp(suffix='.proto')
os.close(fh)
with open(temp_path, 'w') as f:
f.write(alignment_task.SerializeToString())
# Align with C++ program.
subprocess.check_call([ALIGN_BINARY, temp_path])
# Read file.
with open(temp_path + '.result') as f:
result = alignment_pb2.AlignmentResult.FromString(f.read())
# Clean up.
os.remove(temp_path)
os.remove(temp_path + '.result')
logging.info('Aligning NoteSequence with warp path.')
warp_seconds_i = np.array([i * seconds_per_frame for i in result.i])
warp_seconds_j = np.array([j * seconds_per_frame for j in result.j])
time_diffs = np.abs(warp_seconds_i - warp_seconds_j)
warps = np.abs(time_diffs[1:] - time_diffs[:-1])
stats = {
'alignment_score': result.score,
'warp_mean_s': np.mean(warps),
'warp_median_s': np.median(warps),
'warp_max_s': np.max(warps),
'warp_min_s': np.min(warps),
'time_diff_mean_s': np.mean(time_diffs),
'time_diff_median_s': np.median(time_diffs),
'time_diff_max_s': np.max(time_diffs),
'time_diff_min_s': np.min(time_diffs),
}
for name, value in sorted(stats.iteritems()):
logging.info('%s: %f', name, value)
aligned_ns, skipped_notes = sequences_lib.adjust_notesequence_times(
ns,
lambda t: np.interp(t, warp_seconds_i, warp_seconds_j),
minimum_duration=seconds_per_frame)
if skipped_notes > 0:
raise RuntimeError('Skipped {} notes'.format(skipped_notes))
logging.debug('done')
return aligned_ns, stats
| [
"os.remove",
"numpy.abs",
"tempfile.mkstemp",
"alignment_pb2.AlignmentTask",
"librosa.core.cqt",
"absl.logging.debug",
"numpy.median",
"numpy.interp",
"absl.logging.info",
"numpy.max",
"librosa.power_to_db",
"os.close",
"numpy.array",
"numpy.mean",
"numpy.min",
"librosa.midi_to_hz",
... | [((1334, 1356), 'librosa.midi_to_hz', 'librosa.midi_to_hz', (['(36)'], {}), '(36)\n', (1352, 1356), False, 'import librosa\n'), ((1762, 1786), 'librosa.power_to_db', 'librosa.power_to_db', (['cqt'], {}), '(cqt)\n', (1781, 1786), False, 'import librosa\n'), ((2725, 2753), 'absl.logging.info', 'logging.info', (['"""Synthesizing"""'], {}), "('Synthesizing')\n", (2737, 2753), False, 'from absl import logging\n'), ((3465, 3491), 'absl.logging.info', 'logging.info', (['"""source_cqt"""'], {}), "('source_cqt')\n", (3477, 3491), False, 'from absl import logging\n'), ((3563, 3587), 'absl.logging.info', 'logging.info', (['"""dest_cqt"""'], {}), "('dest_cqt')\n", (3575, 3587), False, 'from absl import logging\n'), ((3671, 3700), 'alignment_pb2.AlignmentTask', 'alignment_pb2.AlignmentTask', ([], {}), '()\n', (3698, 3700), False, 'import alignment_pb2\n'), ((4313, 4346), 'tempfile.mkstemp', 'tempfile.mkstemp', ([], {'suffix': '""".proto"""'}), "(suffix='.proto')\n", (4329, 4346), False, 'import tempfile\n'), ((4349, 4361), 'os.close', 'os.close', (['fh'], {}), '(fh)\n', (4357, 4361), False, 'import os\n'), ((4475, 4523), 'subprocess.check_call', 'subprocess.check_call', (['[ALIGN_BINARY, temp_path]'], {}), '([ALIGN_BINARY, temp_path])\n', (4496, 4523), False, 'import subprocess\n'), ((4662, 4682), 'os.remove', 'os.remove', (['temp_path'], {}), '(temp_path)\n', (4671, 4682), False, 'import os\n'), ((4685, 4717), 'os.remove', 'os.remove', (["(temp_path + '.result')"], {}), "(temp_path + '.result')\n", (4694, 4717), False, 'import os\n'), ((4721, 4774), 'absl.logging.info', 'logging.info', (['"""Aligning NoteSequence with warp path."""'], {}), "('Aligning NoteSequence with warp path.')\n", (4733, 4774), False, 'from absl import logging\n'), ((4795, 4848), 'numpy.array', 'np.array', (['[(i * seconds_per_frame) for i in result.i]'], {}), '([(i * seconds_per_frame) for i in result.i])\n', (4803, 4848), True, 'import numpy as np\n'), ((4866, 4919), 'numpy.array', 'np.array', (['[(j * seconds_per_frame) for j in result.j]'], {}), '([(j * seconds_per_frame) for j in result.j])\n', (4874, 4919), True, 'import numpy as np\n'), ((4934, 4973), 'numpy.abs', 'np.abs', (['(warp_seconds_i - warp_seconds_j)'], {}), '(warp_seconds_i - warp_seconds_j)\n', (4940, 4973), True, 'import numpy as np\n'), ((4984, 5024), 'numpy.abs', 'np.abs', (['(time_diffs[1:] - time_diffs[:-1])'], {}), '(time_diffs[1:] - time_diffs[:-1])\n', (4990, 5024), True, 'import numpy as np\n'), ((5784, 5805), 'absl.logging.debug', 'logging.debug', (['"""done"""'], {}), "('done')\n", (5797, 5805), False, 'from absl import logging\n'), ((1534, 1675), 'librosa.core.cqt', 'librosa.core.cqt', (['samples', 'sample_rate'], {'hop_length': 'cqt_hop_length', 'fmin': 'CQT_FMIN', 'n_bins': 'CQT_N_BINS', 'bins_per_octave': 'CQT_BINS_PER_OCTAVE'}), '(samples, sample_rate, hop_length=cqt_hop_length, fmin=\n CQT_FMIN, n_bins=CQT_N_BINS, bins_per_octave=CQT_BINS_PER_OCTAVE)\n', (1550, 1675), False, 'import librosa\n'), ((5098, 5112), 'numpy.mean', 'np.mean', (['warps'], {}), '(warps)\n', (5105, 5112), True, 'import numpy as np\n'), ((5137, 5153), 'numpy.median', 'np.median', (['warps'], {}), '(warps)\n', (5146, 5153), True, 'import numpy as np\n'), ((5175, 5188), 'numpy.max', 'np.max', (['warps'], {}), '(warps)\n', (5181, 5188), True, 'import numpy as np\n'), ((5210, 5223), 'numpy.min', 'np.min', (['warps'], {}), '(warps)\n', (5216, 5223), True, 'import numpy as np\n'), ((5251, 5270), 'numpy.mean', 'np.mean', (['time_diffs'], {}), '(time_diffs)\n', (5258, 5270), True, 'import numpy as np\n'), ((5300, 5321), 'numpy.median', 'np.median', (['time_diffs'], {}), '(time_diffs)\n', (5309, 5321), True, 'import numpy as np\n'), ((5348, 5366), 'numpy.max', 'np.max', (['time_diffs'], {}), '(time_diffs)\n', (5354, 5366), True, 'import numpy as np\n'), ((5393, 5411), 'numpy.min', 'np.min', (['time_diffs'], {}), '(time_diffs)\n', (5399, 5411), True, 'import numpy as np\n'), ((5470, 5505), 'absl.logging.info', 'logging.info', (['"""%s: %f"""', 'name', 'value'], {}), "('%s: %f', name, value)\n", (5482, 5505), False, 'from absl import logging\n'), ((2769, 2838), 'magenta.music.midi_synth.fluidsynth', 'midi_synth.fluidsynth', (['ns'], {'sf2_path': 'sf2_path', 'sample_rate': 'sample_rate'}), '(ns, sf2_path=sf2_path, sample_rate=sample_rate)\n', (2790, 2838), False, 'from magenta.music import midi_synth\n'), ((5604, 5648), 'numpy.interp', 'np.interp', (['t', 'warp_seconds_i', 'warp_seconds_j'], {}), '(t, warp_seconds_i, warp_seconds_j)\n', (5613, 5648), True, 'import numpy as np\n')] |
# -*- coding: utf-8 -*-
"""
Created on Sun Dec 8 22:37:00 2019
@author: for_y
"""
import numpy as np
from AnnoDomini.hamilton_mc import HMC, describe
import sys
import os
sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
def test_hmc():
def norm_function(x):
var = 1
denom = (2*np.pi*var)**.5
num = np.exp(-x**2/2)
return num/denom
start_point = 0
chain,accepts_ratio = HMC(target_pdf = norm_function, burn_in=0, thinning=1,chain_len=100, q_init=[start_point],epsilon = 0.05)
assert len(chain) == 100
q = chain[:,0]
assert max(q) < 10
assert np.isnan(q).sum() == 0
def neg_log_weibull(lam = 1, k = 0.5):
def w(x):
if x > 0:
return -(np.log(k / lam) + (k-1) * np.log(x/lam) - (x/lam) ** k)
else:
return float('inf')
return w
start_point = 3
func = neg_log_weibull(k = 1.5)
chain,accepts_ratio = HMC(U = func, burn_in=0, thinning=1,chain_len=100, q_init=[start_point],epsilon = 0.02)
assert len(chain) == 100
q = chain[:,0]
assert max(q) < 10
assert min(q) > 0
assert np.isnan(q).sum() == 0
| [
"AnnoDomini.hamilton_mc.HMC",
"numpy.log",
"os.path.dirname",
"numpy.isnan",
"numpy.exp"
] | [((457, 565), 'AnnoDomini.hamilton_mc.HMC', 'HMC', ([], {'target_pdf': 'norm_function', 'burn_in': '(0)', 'thinning': '(1)', 'chain_len': '(100)', 'q_init': '[start_point]', 'epsilon': '(0.05)'}), '(target_pdf=norm_function, burn_in=0, thinning=1, chain_len=100, q_init=\n [start_point], epsilon=0.05)\n', (460, 565), False, 'from AnnoDomini.hamilton_mc import HMC, describe\n'), ((1000, 1089), 'AnnoDomini.hamilton_mc.HMC', 'HMC', ([], {'U': 'func', 'burn_in': '(0)', 'thinning': '(1)', 'chain_len': '(100)', 'q_init': '[start_point]', 'epsilon': '(0.02)'}), '(U=func, burn_in=0, thinning=1, chain_len=100, q_init=[start_point],\n epsilon=0.02)\n', (1003, 1089), False, 'from AnnoDomini.hamilton_mc import HMC, describe\n'), ((365, 384), 'numpy.exp', 'np.exp', (['(-x ** 2 / 2)'], {}), '(-x ** 2 / 2)\n', (371, 384), True, 'import numpy as np\n'), ((223, 248), 'os.path.dirname', 'os.path.dirname', (['__file__'], {}), '(__file__)\n', (238, 248), False, 'import os\n'), ((645, 656), 'numpy.isnan', 'np.isnan', (['q'], {}), '(q)\n', (653, 656), True, 'import numpy as np\n'), ((1192, 1203), 'numpy.isnan', 'np.isnan', (['q'], {}), '(q)\n', (1200, 1203), True, 'import numpy as np\n'), ((786, 801), 'numpy.log', 'np.log', (['(k / lam)'], {}), '(k / lam)\n', (792, 801), True, 'import numpy as np\n'), ((812, 827), 'numpy.log', 'np.log', (['(x / lam)'], {}), '(x / lam)\n', (818, 827), True, 'import numpy as np\n')] |
import librosa
from utils.hparams import Hparams
from jamo import hangul_to_jamo
from tqdm import tqdm
import numpy as np
import os, glob, json, shutil, torch, torchaudio
hparams = Hparams()
class KSSDatasetPath():
def __init__(self, Hparams):
self.Hparams = Hparams
# Original data
self.text_paths = glob.glob(os.path.join(self.Hparams.out_texts_dir, '*.pt'))
self.mel_paths = glob.glob(os.path.join(self.Hparams.out_mels_dir, '*.pt'))
self.original_len = len(self.text_paths)
# Splited Data Length
self.val_len = int(self.original_len * self.Hparams.split_ratio)
if self.val_len % self.Hparams.batch_size != 0:
extra_num = self.Hparams.batch_size - self.val_len % self.Hparams.batch_size
self.val_len = self.val_len + extra_num
self.trn_len = self.original_len - self.val_len
# Train data, Validation data
self.trn_text_paths, self.val_text_paths = (self.text_paths[:self.trn_len],
self.text_paths[self.trn_len:])
self.trn_mel_paths, self.val_mel_paths = (self.mel_paths[:self.trn_len],
self.mel_paths[self.trn_len:])
class KSSTrainDataset(torch.utils.data.Dataset, KSSDatasetPath):
def __init__(self):
KSSDatasetPath.__init__(self, hparams)
def __len__(self):
return self.trn_len
def __getitem__(self, idx):
# Train data
self.trn_texts = torch.LongTensor(torch.load(self.trn_text_paths[idx]))
self.trn_mels = torch.FloatTensor(torch.load(self.trn_mel_paths[idx]))
return (self.trn_texts, self.trn_mels)
class KSSValidateDataset(torch.utils.data.Dataset, KSSDatasetPath):
def __init__(self):
KSSDatasetPath.__init__(self, hparams)
def __len__(self):
return self.val_len
def __getitem__(self, idx):
# Validate data
self.val_texts = torch.LongTensor(torch.load(self.val_text_paths[idx]))
self.val_mels = torch.FloatTensor(torch.load(self.val_mel_paths[idx]))
return (self.val_texts, self.val_mels)
def collate_fn(batch):
texts, mels = zip(*batch)
text_pads = torch.nn.utils.rnn.pad_sequence(texts, batch_first=True)
mel_pads = torch.nn.utils.rnn.pad_sequence(mels, batch_first=True)
text_lengths = torch.LongTensor([text.size(0) for text in texts])
mel_lengths = torch.LongTensor([mel.size(0) for mel in mels])
return (text_pads.contiguous(),
mel_pads.contiguous(),
text_lengths.contiguous(),
mel_lengths.contiguous())
def _normalize(S, hparams):
return hparams.max_abs_value * ((S - hparams.min_level_db) / (-hparams.min_level_db))
def _denormalize(D, hparams):
return ((D * -hparams.min_level_db / hparams.max_abs_value) + hparams.min_level_db)
def preprocessing(hparams, ap, tp, data_length=None):
print("Pre-processing Audio and Text data from Alignments")
alignments_path = os.path.join(hparams.nas_path, 'kss', 'alignments.json')
with open(alignments_path, 'r', encoding='utf-8') as f:
alignments = json.loads(f.read())
# ๋๋ ํ ๋ฆฌ ์์ฑ
data_dirs = [hparams.out_texts_dir,
hparams.out_specs_dir,
hparams.out_mels_dir]
for data_dir in data_dirs:
if os.path.isdir(data_dir): # ์ด๋ฏธ ์์ผ๋ฉด ๊ฐ์ ์ญ์
shutil.rmtree(data_dir)
os.makedirs(data_dir, exist_ok=True) # ๋๋ ํ ๋ฆฌ ์์ฑ
# pre-processing
for idx, (audio_path, text) in tqdm(enumerate(alignments.items())):
# text pre-processing
seq = tp.text_to_sequence(text)
# audio load
audio, _ = librosa.load(audio_path, hparams.sample_rate)
audio, _ = librosa.effects.trim(audio, top_db=hparams.trim_top_db)
# normalizing
audio = ap._normalizing(audio)
# audio pre-processing
mel = _normalize(ap.melspectrogram(audio), hparams)
# padded
mel = ap.pre_padding(mel)
# text save
torch.save(seq, os.path.join(hparams.out_texts_dir, 'kss-text-%05d.pt' % idx))
torch.save(mel, os.path.join(hparams.out_mels_dir, 'kss-mel-%05d.pt' % idx))
if data_length is not None and idx+1 == data_length:
break
def get_pairs_from_source(hparams, save=True):
print("Getting pairs from metadata")
source_dir = os.path.join(hparams.nas_path, 'kss')
if not os.path.isdir(source_dir):
print("There are not exist data directories. Please check directory which is %s" % source_dir)
source_file = glob.glob(os.path.join(source_dir, '*.txt'))[0] # path|text|text|text|digit|text
pairs = {}
with open(source_file, 'r', encoding='utf-8') as f:
lines = f.readlines() # ์ฝ์ด์ค๊ธฐ
metadata = [line.strip('\n') for line in lines] # ๊ฐํ๋ฌธ์ ์ ๊ฑฐ
for line in metadata:
audio_path = line.split('|')[0] # audio path
if '/' in audio_path:
audio_path = audio_path.split('/')
audio_path = os.path.join(audio_path[0], audio_path[1])
text = line.split('|')[1] # text
pairs[os.path.join(source_dir, audio_path)] = text # save as dictionary
if save==True:
with open(os.path.join(source_dir, 'alignments.json'), 'w', encoding='UTF-8') as json_file:
json.dump(pairs, json_file, ensure_ascii=False, indent=4, sort_keys=False)
return pairs
class AudioPreprocessing():
def __init__(self, Hparams):
self.Hparams = Hparams
def spectrogram(self, audio):
D = self._stft(audio)
M = np.abs(D)
S = self._amp_to_db(M) - self.Hparams.ref_level_db
return torch.Tensor(S.T)
def melspectrogram(self, audio):
D = self._stft(audio)
M = np.abs(D)
S = self._amp_to_db(self._linear_to_mel(M)) - self.Hparams.ref_level_db
return torch.Tensor(S.T)
def _stft(self, audio):
return librosa.stft(audio,
n_fft=self.Hparams.n_fft,
hop_length=self.Hparams.hop_length,
win_length=self.Hparams.win_length,
pad_mode='constant')
def _istft(self, spec):
return librosa.istft(spec,
hop_length=self.Hparams.hop_length,
win_length=self.Hparams.win_length)
def _linear_to_mel(self, spec):
_mel_basis = self._build_mel_basis()
return np.dot(_mel_basis, spec)
def _build_mel_basis(self):
if self.Hparams.fmax != self.Hparams.sample_rate // 2:
self.Hparams.fmax = self.Hparams.sample_rate // 2
return librosa.filters.mel(sr=self.Hparams.sample_rate,
n_fft=self.Hparams.n_fft,
n_mels=self.Hparams.n_mels,
fmin=self.Hparams.fmin,
fmax=self.Hparams.fmax)
def _amp_to_db(self, x):
min_level = np.exp(self.Hparams.min_level_db / 20 * np.log(10))
return 20 * np.log10(np.maximum(min_level, x))
def _db_to_amp(self, x):
return np.power(10.0, (x) * 0.05)
def _resample(self, x):
return librosa.resample(x,
orig_sr=self.Hparams.origin_sample_rate,
target_sr=self.Hparams.sample_rate)
def _normalizing(self, x):
return x / np.abs(x).max()
def pre_padding(self, spec):
# pre-padding from reduction factor(teacher forcing) to feed decoder input in training
# matching r-times
if type(spec) != torch.Tensor:
spec = torch.Tensor(spec)
t = spec.size(0)
n_pads = self.Hparams.reduction - (t % self.Hparams.reduction)\
if t % self.Hparams.reduction != 0 else 0
pad = (0, 0, 0, n_pads)
padded_spec = torch.nn.functional.pad(spec, pad, mode='constant', value=0)
return padded_spec
def spec_to_audio(self, spec):
side_pad = np.zeros(int(0.15*self.Hparams.sample_rate))
audio = self._griffin_lim(spec, self.Hparams)
audio = np.concatenate([side_pad, audio, side_pad])
return audio
def _griffin_lim(self, spec, hparams):
spec = self._db_to_amp(spec + self.Hparams.ref_level_db)
if (spec.shape[0] == ((hparams.sample_rate//20)//2)+1):
spec = spec
else:
inv_mel_filter = np.linalg.pinv(self._build_mel_basis())
if (inv_mel_filter.shape[-1] == spec.shape[-1]):
spec = spec.T
spec = np.maximum(1e-10, np.dot(inv_mel_filter, spec))
return librosa.core.griffinlim(spec,
hop_length=hparams.sample_rate//80,
win_length=hparams.sample_rate//20)
class TextPreprocessing():
def __init__(self, Hparams):
self.Hparams = Hparams
# text -> sequence
def text_to_sequence(self, text):
sequence = []
if not 0x1100 <= ord(text[0]) <= 0x1113:
text = ''.join(list(hangul_to_jamo(text)))
for t in text:
if t in self.Hparams._symbol_to_id:
# ์ง์ ํ์ง ์์ ์ด์ธ์ ๋ฌธ์๋ ์ํ์ค์ ํฌํจํ์ง ์์
sequence.append(self.Hparams._symbol_to_id[t])
sequence.append(self.Hparams._symbol_to_id['~'])
return sequence
# sequence -> text without PAD, EOS tokens
def sequence_to_text(self, sequence):
tokens = [self.Hparams._symbol_to_id[self.Hparams.PAD],
self.Hparams._symbol_to_id[self.Hparams.EOS]]
sequence = [s for s in sequence if s not in tokens]
text = ''
for symbol_id in sequence:
if symbol_id in self.Hparams._id_to_symbol:
t = self.Hparams._id_to_symbol[symbol_id]
text += t
return text
| [
"numpy.abs",
"numpy.maximum",
"librosa.filters.mel",
"librosa.istft",
"librosa.resample",
"shutil.rmtree",
"os.path.join",
"torch.nn.functional.pad",
"utils.hparams.Hparams",
"numpy.power",
"librosa.core.griffinlim",
"torch.load",
"torch.Tensor",
"librosa.effects.trim",
"librosa.stft",
... | [((189, 198), 'utils.hparams.Hparams', 'Hparams', ([], {}), '()\n', (196, 198), False, 'from utils.hparams import Hparams\n'), ((2348, 2404), 'torch.nn.utils.rnn.pad_sequence', 'torch.nn.utils.rnn.pad_sequence', (['texts'], {'batch_first': '(True)'}), '(texts, batch_first=True)\n', (2379, 2404), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((2421, 2476), 'torch.nn.utils.rnn.pad_sequence', 'torch.nn.utils.rnn.pad_sequence', (['mels'], {'batch_first': '(True)'}), '(mels, batch_first=True)\n', (2452, 2476), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((3162, 3218), 'os.path.join', 'os.path.join', (['hparams.nas_path', '"""kss"""', '"""alignments.json"""'], {}), "(hparams.nas_path, 'kss', 'alignments.json')\n", (3174, 3218), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((4768, 4805), 'os.path.join', 'os.path.join', (['hparams.nas_path', '"""kss"""'], {}), "(hparams.nas_path, 'kss')\n", (4780, 4805), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((4818, 4843), 'os.path.isdir', 'os.path.isdir', (['source_dir'], {}), '(source_dir)\n', (4831, 4843), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((6043, 6052), 'numpy.abs', 'np.abs', (['D'], {}), '(D)\n', (6049, 6052), True, 'import numpy as np\n'), ((6129, 6146), 'torch.Tensor', 'torch.Tensor', (['S.T'], {}), '(S.T)\n', (6141, 6146), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((6231, 6240), 'numpy.abs', 'np.abs', (['D'], {}), '(D)\n', (6237, 6240), True, 'import numpy as np\n'), ((6338, 6355), 'torch.Tensor', 'torch.Tensor', (['S.T'], {}), '(S.T)\n', (6350, 6355), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((6403, 6546), 'librosa.stft', 'librosa.stft', (['audio'], {'n_fft': 'self.Hparams.n_fft', 'hop_length': 'self.Hparams.hop_length', 'win_length': 'self.Hparams.win_length', 'pad_mode': '"""constant"""'}), "(audio, n_fft=self.Hparams.n_fft, hop_length=self.Hparams.\n hop_length, win_length=self.Hparams.win_length, pad_mode='constant')\n", (6415, 6546), False, 'import librosa\n'), ((6713, 6809), 'librosa.istft', 'librosa.istft', (['spec'], {'hop_length': 'self.Hparams.hop_length', 'win_length': 'self.Hparams.win_length'}), '(spec, hop_length=self.Hparams.hop_length, win_length=self.\n Hparams.win_length)\n', (6726, 6809), False, 'import librosa\n'), ((6968, 6992), 'numpy.dot', 'np.dot', (['_mel_basis', 'spec'], {}), '(_mel_basis, spec)\n', (6974, 6992), True, 'import numpy as np\n'), ((7171, 7325), 'librosa.filters.mel', 'librosa.filters.mel', ([], {'sr': 'self.Hparams.sample_rate', 'n_fft': 'self.Hparams.n_fft', 'n_mels': 'self.Hparams.n_mels', 'fmin': 'self.Hparams.fmin', 'fmax': 'self.Hparams.fmax'}), '(sr=self.Hparams.sample_rate, n_fft=self.Hparams.n_fft,\n n_mels=self.Hparams.n_mels, fmin=self.Hparams.fmin, fmax=self.Hparams.fmax)\n', (7190, 7325), False, 'import librosa\n'), ((7682, 7706), 'numpy.power', 'np.power', (['(10.0)', '(x * 0.05)'], {}), '(10.0, x * 0.05)\n', (7690, 7706), True, 'import numpy as np\n'), ((7760, 7861), 'librosa.resample', 'librosa.resample', (['x'], {'orig_sr': 'self.Hparams.origin_sample_rate', 'target_sr': 'self.Hparams.sample_rate'}), '(x, orig_sr=self.Hparams.origin_sample_rate, target_sr=self\n .Hparams.sample_rate)\n', (7776, 7861), False, 'import librosa\n'), ((8448, 8508), 'torch.nn.functional.pad', 'torch.nn.functional.pad', (['spec', 'pad'], {'mode': '"""constant"""', 'value': '(0)'}), "(spec, pad, mode='constant', value=0)\n", (8471, 8508), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((8712, 8755), 'numpy.concatenate', 'np.concatenate', (['[side_pad, audio, side_pad]'], {}), '([side_pad, audio, side_pad])\n', (8726, 8755), True, 'import numpy as np\n'), ((9256, 9365), 'librosa.core.griffinlim', 'librosa.core.griffinlim', (['spec'], {'hop_length': '(hparams.sample_rate // 80)', 'win_length': '(hparams.sample_rate // 20)'}), '(spec, hop_length=hparams.sample_rate // 80,\n win_length=hparams.sample_rate // 20)\n', (9279, 9365), False, 'import librosa\n'), ((364, 412), 'os.path.join', 'os.path.join', (['self.Hparams.out_texts_dir', '"""*.pt"""'], {}), "(self.Hparams.out_texts_dir, '*.pt')\n", (376, 412), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((450, 497), 'os.path.join', 'os.path.join', (['self.Hparams.out_mels_dir', '"""*.pt"""'], {}), "(self.Hparams.out_mels_dir, '*.pt')\n", (462, 497), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((1614, 1650), 'torch.load', 'torch.load', (['self.trn_text_paths[idx]'], {}), '(self.trn_text_paths[idx])\n', (1624, 1650), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((1695, 1730), 'torch.load', 'torch.load', (['self.trn_mel_paths[idx]'], {}), '(self.trn_mel_paths[idx])\n', (1705, 1730), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((2096, 2132), 'torch.load', 'torch.load', (['self.val_text_paths[idx]'], {}), '(self.val_text_paths[idx])\n', (2106, 2132), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((2177, 2212), 'torch.load', 'torch.load', (['self.val_mel_paths[idx]'], {}), '(self.val_mel_paths[idx])\n', (2187, 2212), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((3545, 3568), 'os.path.isdir', 'os.path.isdir', (['data_dir'], {}), '(data_dir)\n', (3558, 3568), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((3639, 3675), 'os.makedirs', 'os.makedirs', (['data_dir'], {'exist_ok': '(True)'}), '(data_dir, exist_ok=True)\n', (3650, 3675), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((3943, 3988), 'librosa.load', 'librosa.load', (['audio_path', 'hparams.sample_rate'], {}), '(audio_path, hparams.sample_rate)\n', (3955, 3988), False, 'import librosa\n'), ((4013, 4068), 'librosa.effects.trim', 'librosa.effects.trim', (['audio'], {'top_db': 'hparams.trim_top_db'}), '(audio, top_db=hparams.trim_top_db)\n', (4033, 4068), False, 'import librosa\n'), ((4980, 5013), 'os.path.join', 'os.path.join', (['source_dir', '"""*.txt"""'], {}), "(source_dir, '*.txt')\n", (4992, 5013), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((8219, 8237), 'torch.Tensor', 'torch.Tensor', (['spec'], {}), '(spec)\n', (8231, 8237), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((3602, 3625), 'shutil.rmtree', 'shutil.rmtree', (['data_dir'], {}), '(data_dir)\n', (3615, 3625), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((4400, 4461), 'os.path.join', 'os.path.join', (['hparams.out_texts_dir', "('kss-text-%05d.pt' % idx)"], {}), "(hparams.out_texts_dir, 'kss-text-%05d.pt' % idx)\n", (4412, 4461), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((4492, 4551), 'os.path.join', 'os.path.join', (['hparams.out_mels_dir', "('kss-mel-%05d.pt' % idx)"], {}), "(hparams.out_mels_dir, 'kss-mel-%05d.pt' % idx)\n", (4504, 4551), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((5435, 5477), 'os.path.join', 'os.path.join', (['audio_path[0]', 'audio_path[1]'], {}), '(audio_path[0], audio_path[1])\n', (5447, 5477), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((5543, 5579), 'os.path.join', 'os.path.join', (['source_dir', 'audio_path'], {}), '(source_dir, audio_path)\n', (5555, 5579), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((5770, 5844), 'json.dump', 'json.dump', (['pairs', 'json_file'], {'ensure_ascii': '(False)', 'indent': '(4)', 'sort_keys': '(False)'}), '(pairs, json_file, ensure_ascii=False, indent=4, sort_keys=False)\n', (5779, 5844), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((7566, 7576), 'numpy.log', 'np.log', (['(10)'], {}), '(10)\n', (7572, 7576), True, 'import numpy as np\n'), ((7608, 7632), 'numpy.maximum', 'np.maximum', (['min_level', 'x'], {}), '(min_level, x)\n', (7618, 7632), True, 'import numpy as np\n'), ((9196, 9224), 'numpy.dot', 'np.dot', (['inv_mel_filter', 'spec'], {}), '(inv_mel_filter, spec)\n', (9202, 9224), True, 'import numpy as np\n'), ((5671, 5714), 'os.path.join', 'os.path.join', (['source_dir', '"""alignments.json"""'], {}), "(source_dir, 'alignments.json')\n", (5683, 5714), False, 'import os, glob, json, shutil, torch, torchaudio\n'), ((7979, 7988), 'numpy.abs', 'np.abs', (['x'], {}), '(x)\n', (7985, 7988), True, 'import numpy as np\n'), ((9705, 9725), 'jamo.hangul_to_jamo', 'hangul_to_jamo', (['text'], {}), '(text)\n', (9719, 9725), False, 'from jamo import hangul_to_jamo\n')] |
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# pylint: disable=wrong-import-position
"""
Generate a json file summarizing a CLSim table
"""
from __future__ import absolute_import, division, print_function
__all__ = [
'summarize_clsim_table',
'parse_args',
'main'
]
__author__ = '<NAME>'
__license__ = '''Copyright 2017 <NAME>
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.'''
from argparse import ArgumentParser
from collections import OrderedDict
from glob import glob
from itertools import product
from os.path import abspath, basename, dirname, isfile, join, splitext
import sys
from time import time
import numpy as np
if __name__ == '__main__' and __package__ is None:
RETRO_DIR = dirname(dirname(dirname(abspath(__file__))))
if RETRO_DIR not in sys.path:
sys.path.append(RETRO_DIR)
from retro.const import SPEED_OF_LIGHT_M_PER_NS
from retro.utils.misc import expand, mkdir, wstderr
from retro.tables.clsim_tables import (
CLSIM_TABLE_METANAME_PROTO, interpret_clsim_table_fname, load_clsim_table
)
def summarize_clsim_table(table_fpath, table=None, save_summary=True,
outdir=None):
"""
Parameters
----------
table_fpath : string
Path to table (or just the table's filename if `outdir` is specified)
table : mapping, optional
If the table has already been loaded, it can be passed here to avoid
re-loading the table.
save_summary : bool
Whether to save the table summary to disk.
outdir : string, optional
If `save_summary` is True, write the summary to this directory. If
`outdir` is not specified and `save_summary` is True, the summary will
be written to the same directory that contains `table_fpath`.
Returns
-------
table
See `load_clsim_table` for details of the data structure
summary : OrderedDict
"""
t_start = time()
if save_summary:
from pisa.utils.jsons import from_json, to_json
table_fpath = expand(table_fpath)
srcdir, clsim_fname = dirname(table_fpath), basename(table_fpath)
invalid_fname = False
try:
fname_info = interpret_clsim_table_fname(clsim_fname)
except ValueError:
invalid_fname = True
fname_info = {}
if outdir is None:
outdir = srcdir
outdir = expand(outdir)
mkdir(outdir)
if invalid_fname:
metapath = None
else:
metaname = (CLSIM_TABLE_METANAME_PROTO[-1]
.format(hash_val=fname_info['hash_val']))
metapath = join(outdir, metaname)
if metapath and isfile(metapath):
meta = from_json(metapath)
else:
meta = dict()
if table is None:
table = load_clsim_table(table_fpath)
summary = OrderedDict()
for key in table.keys():
if key == 'table':
continue
summary[key] = table[key]
if fname_info:
for key in ('hash_val', 'string', 'depth_idx', 'seed'):
summary[key] = fname_info[key]
# TODO: Add hole ice info when added to tray_kw_to_hash
if meta:
summary['n_events'] = meta['tray_kw_to_hash']['NEvents']
summary['ice_model'] = meta['tray_kw_to_hash']['IceModel']
summary['tilt'] = not meta['tray_kw_to_hash']['DisableTilt']
for key, val in meta.items():
if key.endswith('_binning_kw'):
summary[key] = val
elif 'fname_version' in fname_info and fname_info['fname_version'] == 1:
summary['n_events'] = fname_info['n_events']
summary['ice_model'] = 'spice_mie'
summary['tilt'] = False
summary['r_binning_kw'] = dict(min=0.0, max=400.0, n_bins=200, power=2)
summary['costheta_binning_kw'] = dict(min=-1, max=1, n_bins=40)
summary['t_binning_kw'] = dict(min=0.0, max=3000.0, n_bins=300)
summary['costhetadir_binning_kw'] = dict(min=-1, max=1, n_bins=20)
summary['deltaphidir_binning_kw'] = dict(min=0.0, max=np.pi, n_bins=20)
# Save marginal distributions and info to file
norm = (
1
/ table['n_photons']
/ (SPEED_OF_LIGHT_M_PER_NS / table['phase_refractive_index']
* np.mean(np.diff(table['t_bin_edges'])))
#* table['angular_acceptance_fract']
* (len(table['costheta_bin_edges']) - 1)
)
summary['norm'] = norm
dim_names = ('r', 'costheta', 't', 'costhetadir', 'deltaphidir')
n_dims = len(table['table_shape'])
assert n_dims == len(dim_names)
# Apply norm to underflow and overflow so magnitudes can be compared
# relative to plotted marginal distributions
for flow, idx in product(('underflow', 'overflow'), iter(range(n_dims))):
summary[flow][idx] = summary[flow][idx] * norm
wstderr('Finding marginal distributions...\n')
wstderr(' masking off zeros in table...')
t0 = time()
nonzero_table = np.ma.masked_equal(table['table'], 0)
wstderr(' ({} ms)\n'.format(np.round((time() - t0)*1e3, 3)))
t0_marg = time()
summary['dimensions'] = OrderedDict()
for keep_axis, ax_name in zip(tuple(range(n_dims)), dim_names):
remove_axes = list(range(n_dims))
remove_axes.pop(keep_axis)
remove_axes = tuple(remove_axes)
axis = OrderedDict()
wstderr(' mean across non-{} axes...'.format(ax_name))
t0 = time()
axis['mean'] = norm * np.asarray(
np.mean(table['table'], axis=remove_axes)
)
wstderr(' ({} s)\n'.format(np.round(time() - t0, 3)))
wstderr(' median across non-{} axes...'.format(ax_name))
t0 = time()
axis['median'] = norm * np.asarray(
np.ma.median(nonzero_table, axis=remove_axes)
)
wstderr(' ({} s)\n'.format(np.round(time() - t0, 3)))
wstderr(' max across non-{} axes...'.format(ax_name))
t0 = time()
axis['max'] = norm * np.asarray(
np.max(table['table'], axis=remove_axes)
)
wstderr(' ({} s)\n'.format(np.round(time() - t0, 3)))
summary['dimensions'][ax_name] = axis
wstderr(
' Total time to find marginal distributions: {} s\n'
.format(np.round(time() - t0_marg, 3))
)
if save_summary:
ext = None
base_fname = clsim_fname
while ext not in ('', '.fits'):
base_fname, ext = splitext(base_fname)
ext = ext.lower()
outfpath = join(outdir, base_fname + '_summary.json.bz2')
to_json(summary, outfpath)
print('saved summary to "{}"'.format(outfpath))
wstderr('Time to summarize table: {} s\n'
.format(np.round(time() - t_start, 3)))
return table, summary
def parse_args(description=__doc__):
"""Parse command line args.
Returns
-------
args : Namespace
"""
parser = ArgumentParser(description=description)
parser.add_argument(
'--outdir', default=None,
help='''Directory in which to save summary (if not specified, summary
is saved to same directory as the table)'''
)
parser.add_argument(
'table-fpaths', nargs='+',
help='''Path(s) to CLSim table(s). Note that literal strings are
glob-expanded.'''
)
return parser.parse_args()
def main():
"""Main function for calling summarize_clsim_table as a script"""
t0 = time()
args = parse_args()
kwargs = vars(args)
table_fpaths = []
for fpath in kwargs.pop('table-fpaths'):
table_fpaths.extend(glob(expand(fpath)))
for fpath in table_fpaths:
kwargs['table_fpath'] = fpath
summarize_clsim_table(**kwargs)
total_time = time() - t0
if len(table_fpaths) > 1:
avg = np.round(total_time / len(table_fpaths), 3)
wstderr('Average time to summarize tables: {} s/table\n'.format(avg))
if __name__ == '__main__':
main()
| [
"argparse.ArgumentParser",
"os.path.isfile",
"numpy.mean",
"os.path.join",
"retro.utils.misc.mkdir",
"sys.path.append",
"os.path.abspath",
"pisa.utils.jsons.from_json",
"os.path.dirname",
"numpy.max",
"numpy.ma.median",
"retro.utils.misc.wstderr",
"os.path.basename",
"pisa.utils.jsons.to_j... | [((2394, 2400), 'time.time', 'time', ([], {}), '()\n', (2398, 2400), False, 'from time import time\n'), ((2497, 2516), 'retro.utils.misc.expand', 'expand', (['table_fpath'], {}), '(table_fpath)\n', (2503, 2516), False, 'from retro.utils.misc import expand, mkdir, wstderr\n'), ((2821, 2835), 'retro.utils.misc.expand', 'expand', (['outdir'], {}), '(outdir)\n', (2827, 2835), False, 'from retro.utils.misc import expand, mkdir, wstderr\n'), ((2840, 2853), 'retro.utils.misc.mkdir', 'mkdir', (['outdir'], {}), '(outdir)\n', (2845, 2853), False, 'from retro.utils.misc import expand, mkdir, wstderr\n'), ((3255, 3268), 'collections.OrderedDict', 'OrderedDict', ([], {}), '()\n', (3266, 3268), False, 'from collections import OrderedDict\n'), ((5240, 5286), 'retro.utils.misc.wstderr', 'wstderr', (['"""Finding marginal distributions...\n"""'], {}), "('Finding marginal distributions...\\n')\n", (5247, 5286), False, 'from retro.utils.misc import expand, mkdir, wstderr\n'), ((5291, 5335), 'retro.utils.misc.wstderr', 'wstderr', (['""" masking off zeros in table..."""'], {}), "(' masking off zeros in table...')\n", (5298, 5335), False, 'from retro.utils.misc import expand, mkdir, wstderr\n'), ((5345, 5351), 'time.time', 'time', ([], {}), '()\n', (5349, 5351), False, 'from time import time\n'), ((5372, 5409), 'numpy.ma.masked_equal', 'np.ma.masked_equal', (["table['table']", '(0)'], {}), "(table['table'], 0)\n", (5390, 5409), True, 'import numpy as np\n'), ((5490, 5496), 'time.time', 'time', ([], {}), '()\n', (5494, 5496), False, 'from time import time\n'), ((5525, 5538), 'collections.OrderedDict', 'OrderedDict', ([], {}), '()\n', (5536, 5538), False, 'from collections import OrderedDict\n'), ((7315, 7354), 'argparse.ArgumentParser', 'ArgumentParser', ([], {'description': 'description'}), '(description=description)\n', (7329, 7354), False, 'from argparse import ArgumentParser\n'), ((7839, 7845), 'time.time', 'time', ([], {}), '()\n', (7843, 7845), False, 'from time import time\n'), ((1275, 1301), 'sys.path.append', 'sys.path.append', (['RETRO_DIR'], {}), '(RETRO_DIR)\n', (1290, 1301), False, 'import sys\n'), ((2543, 2563), 'os.path.dirname', 'dirname', (['table_fpath'], {}), '(table_fpath)\n', (2550, 2563), False, 'from os.path import abspath, basename, dirname, isfile, join, splitext\n'), ((2565, 2586), 'os.path.basename', 'basename', (['table_fpath'], {}), '(table_fpath)\n', (2573, 2586), False, 'from os.path import abspath, basename, dirname, isfile, join, splitext\n'), ((2643, 2683), 'retro.tables.clsim_tables.interpret_clsim_table_fname', 'interpret_clsim_table_fname', (['clsim_fname'], {}), '(clsim_fname)\n', (2670, 2683), False, 'from retro.tables.clsim_tables import CLSIM_TABLE_METANAME_PROTO, interpret_clsim_table_fname, load_clsim_table\n'), ((3043, 3065), 'os.path.join', 'join', (['outdir', 'metaname'], {}), '(outdir, metaname)\n', (3047, 3065), False, 'from os.path import abspath, basename, dirname, isfile, join, splitext\n'), ((3086, 3102), 'os.path.isfile', 'isfile', (['metapath'], {}), '(metapath)\n', (3092, 3102), False, 'from os.path import abspath, basename, dirname, isfile, join, splitext\n'), ((3119, 3138), 'pisa.utils.jsons.from_json', 'from_json', (['metapath'], {}), '(metapath)\n', (3128, 3138), False, 'from pisa.utils.jsons import from_json, to_json\n'), ((3210, 3239), 'retro.tables.clsim_tables.load_clsim_table', 'load_clsim_table', (['table_fpath'], {}), '(table_fpath)\n', (3226, 3239), False, 'from retro.tables.clsim_tables import CLSIM_TABLE_METANAME_PROTO, interpret_clsim_table_fname, load_clsim_table\n'), ((5740, 5753), 'collections.OrderedDict', 'OrderedDict', ([], {}), '()\n', (5751, 5753), False, 'from collections import OrderedDict\n'), ((5834, 5840), 'time.time', 'time', ([], {}), '()\n', (5838, 5840), False, 'from time import time\n'), ((6091, 6097), 'time.time', 'time', ([], {}), '()\n', (6095, 6097), False, 'from time import time\n'), ((6351, 6357), 'time.time', 'time', ([], {}), '()\n', (6355, 6357), False, 'from time import time\n'), ((6912, 6958), 'os.path.join', 'join', (['outdir', "(base_fname + '_summary.json.bz2')"], {}), "(outdir, base_fname + '_summary.json.bz2')\n", (6916, 6958), False, 'from os.path import abspath, basename, dirname, isfile, join, splitext\n'), ((6967, 6993), 'pisa.utils.jsons.to_json', 'to_json', (['summary', 'outfpath'], {}), '(summary, outfpath)\n', (6974, 6993), False, 'from pisa.utils.jsons import from_json, to_json\n'), ((8136, 8142), 'time.time', 'time', ([], {}), '()\n', (8140, 8142), False, 'from time import time\n'), ((6842, 6862), 'os.path.splitext', 'splitext', (['base_fname'], {}), '(base_fname)\n', (6850, 6862), False, 'from os.path import abspath, basename, dirname, isfile, join, splitext\n'), ((1212, 1229), 'os.path.abspath', 'abspath', (['__file__'], {}), '(__file__)\n', (1219, 1229), False, 'from os.path import abspath, basename, dirname, isfile, join, splitext\n'), ((5895, 5936), 'numpy.mean', 'np.mean', (["table['table']"], {'axis': 'remove_axes'}), "(table['table'], axis=remove_axes)\n", (5902, 5936), True, 'import numpy as np\n'), ((6154, 6199), 'numpy.ma.median', 'np.ma.median', (['nonzero_table'], {'axis': 'remove_axes'}), '(nonzero_table, axis=remove_axes)\n', (6166, 6199), True, 'import numpy as np\n'), ((6411, 6451), 'numpy.max', 'np.max', (["table['table']"], {'axis': 'remove_axes'}), "(table['table'], axis=remove_axes)\n", (6417, 6451), True, 'import numpy as np\n'), ((7994, 8007), 'retro.utils.misc.expand', 'expand', (['fpath'], {}), '(fpath)\n', (8000, 8007), False, 'from retro.utils.misc import expand, mkdir, wstderr\n'), ((4675, 4704), 'numpy.diff', 'np.diff', (["table['t_bin_edges']"], {}), "(table['t_bin_edges'])\n", (4682, 4704), True, 'import numpy as np\n'), ((6670, 6676), 'time.time', 'time', ([], {}), '()\n', (6674, 6676), False, 'from time import time\n'), ((7126, 7132), 'time.time', 'time', ([], {}), '()\n', (7130, 7132), False, 'from time import time\n'), ((5452, 5458), 'time.time', 'time', ([], {}), '()\n', (5456, 5458), False, 'from time import time\n'), ((5991, 5997), 'time.time', 'time', ([], {}), '()\n', (5995, 5997), False, 'from time import time\n'), ((6254, 6260), 'time.time', 'time', ([], {}), '()\n', (6258, 6260), False, 'from time import time\n'), ((6506, 6512), 'time.time', 'time', ([], {}), '()\n', (6510, 6512), False, 'from time import time\n')] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.