File size: 3,016 Bytes
bc2cdff | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 |
from .coefficient import *
class FunctionCoefficient(Coefficient):
DRJIT_STRUCT = {
'parameters' : dict,
'function_generator' : callable,
}
def __init__(self, name: str, parameters: dict, function_generator: callable,
grad_generator: callable = None, laplacian_generator: callable = None):
self.is_zero = False
self.parameters = parameters
self.function_generator = function_generator
self.grad_generator = grad_generator
self.laplacian_generator = laplacian_generator
self.name = name
self.type = "CustomFunction"
for key in parameters.keys():
dr.make_opaque(parameters[key])
self.update_function()
def update_function(self):
self.function = lambda points : self.function_generator(points, self.parameters)
if ((self.grad_generator is not None) & (self.laplacian_generator is not None)):
self.grad = lambda points : self.grad_generator(points, self.parameters)
self.laplacian = lambda points : self.laplacian_generator(points, self.parameters)
else:
self.grad = None
self.laplacian = None
def get_value(self, points):
return self.function(points)
def get_grad_laplacian(self, points):
if ((self.grad is not None) & (self.laplacian is not None)):
return self.grad(points), self.laplacian(points)
else:
raise Exception(
f"Laplacian or gradient is not defined for the function coefficient\"{self.name}\"!")
def get_opt_params(self, param_dict: dict, opt_params: list):
for i in opt_params:
param_exists = False
for j in self.parameters.keys():
if i == j:
param_dict[f"{self.name}.{self.type}.{i}"] = self.parameters[i]
param_exists = True
if not param_exists:
raise Exception(
f"Function coefficient \"{self.name}\" of type \"{self.type}\" does not have parameter called \"{i}\".")
def update(self, optimizer):
param_exists = False
for key in optimizer.keys():
vals = key.split(".")
name = vals[0]
type = vals[1]
param = vals[2]
if (name == self.name) & (type == self.type):
for p in self.parameters.keys():
if p == param:
param_exists = True
self.parameters[p] = optimizer[key]
if param_exists:
self.update_function()
def zero_grad(self):
for key in self.parameters.keys():
if dr.grad_enabled(self.parameters[key]):
dr.set_grad(self.parameters[key], 0.0)
def copy(self):
new = FunctionCoefficient(self.name, self.parameters, self.function_generator,
self.grad_generator, self.laplacian_generator)
return new
|