|
|
import numpy as np |
|
|
from scipy.signal import convolve2d |
|
|
import matplotlib.pyplot as plt |
|
|
np.random.seed(1) |
|
|
|
|
|
from smooths import s |
|
|
from kernels import circular_kernel_pair |
|
|
|
|
|
def initialize_state(height, width, r_a, factor, mask_ratio, organism_count): |
|
|
b_kernel, s_kernel = circular_kernel_pair(r_a, factor) |
|
|
sim_state = np.zeros((height, width), dtype=np.float16) |
|
|
mask_height, mask_width = int(height*mask_ratio), int(width*mask_ratio) |
|
|
|
|
|
for i in range(organism_count): |
|
|
mask = np.random.rand(mask_height, mask_width) |
|
|
start_x = np.random.randint(0, width-mask_width+1) |
|
|
start_y = np.random.randint(0, height-mask_height+1) |
|
|
sim_state[start_y:start_y+mask_height,\ |
|
|
start_x:start_x+mask_width] += mask |
|
|
return (s_kernel, b_kernel, sim_state) |
|
|
|
|
|
def run_sim(sim_state, s_kernel, b_kernel, B1, D1, B2, D2, alpha_n, alpha_m): |
|
|
n = convolve2d(sim_state, s_kernel, mode='same') |
|
|
N = s_kernel.sum().sum() |
|
|
m = convolve2d(sim_state, b_kernel, mode='same') |
|
|
M = b_kernel.sum().sum() |
|
|
n = n / N |
|
|
m = m / M |
|
|
apply_s_func = np.vectorize(s) |
|
|
q = apply_s_func(n, m, B1 ,D1, B2, D2, alpha_n, alpha_m) |
|
|
sim_state_diff= 2*q-1 |
|
|
return sim_state_diff |
|
|
|
|
|
def get_next_state(height ,width, r_a, factor, mask_ratio, alpha_n, alpha_m, b1, b2, d1, d2, dt, organism_count): |
|
|
s_kernel, b_kernel, sim_state = initialize_state(height, width, r_a, factor, mask_ratio, organism_count) |
|
|
|
|
|
|
|
|
|
|
|
while True: |
|
|
yield sim_state |
|
|
sim_state_diff = run_sim(sim_state, s_kernel, b_kernel, b1, d1, b2, d2, alpha_n, alpha_m) |
|
|
|
|
|
sim_state += dt*sim_state_diff |
|
|
sim_state[sim_state<0] = 0 |
|
|
|
|
|
|
|
|
|
|
|
|
