Last active
April 13, 2024 12:44
-
-
Save Denbergvanthijs/a634d249e8784e340aa8d5d90d527711 to your computer and use it in GitHub Desktop.
Conway's Game of Life using a neural network with Keras and Tensorflow in Python
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| import matplotlib.pyplot as plt | |
| import numpy as np | |
| from matplotlib.animation import FuncAnimation | |
| from tensorflow.keras.layers import Conv2D, InputLayer | |
| from tensorflow.keras.models import Sequential | |
| size = (10, 10) | |
| expanded_size = (1, *size, 1) | |
| n_frames = 120 | |
| glider = ((1, 2), (2, 3), (3, 1), (3, 2), (3, 3)) | |
| # env = np.random.randint(0, 2, expanded_size) | |
| env = np.zeros(expanded_size, dtype=int) | |
| for pos in glider: | |
| env[(0,) + pos] = 1 | |
| model = Sequential([InputLayer(input_shape=expanded_size[1:]), | |
| Conv2D(1, 3, padding="same", activation=None, use_bias=False, kernel_initializer="ones")]) | |
| frames = [] | |
| for i in range(n_frames): | |
| # 2D sliding window of 3x3 including summation TODO: add torus-style wrapping | |
| # NOTE: convolve2d of scipy does support torus-padding but that's obviously not as cool as a neural network | |
| neighbours = model(env) - env # Don't count the cell itself in the number of neighbours | |
| env = np.where((env & np.isin(neighbours, (2, 3))) | ((env == 0) & (neighbours == 3)), 1, 0) | |
| frames.append(env.squeeze()) | |
| fig = plt.figure() | |
| ax = plt.axes(xlim=(0, size[0]), ylim=(0, size[1])) | |
| render = plt.imshow(frames[0], interpolation="none", cmap="binary") | |
| def animate(i: int): | |
| render.set_array(frames[i]) | |
| return [render] | |
| anim = FuncAnimation(fig, animate, frames=n_frames, interval=30, blit=True) | |
| plt.axis("off") | |
| plt.gca().invert_yaxis() | |
| anim.save("glider.gif", fps=30) | |
| plt.show() |
Hey I'm interested in your work. You got this compressed in a way that I haven't seen before. Can we zoom? I'm building a neuron based GOL based on a similar proximity algorithm. You may find it very interesting. Email me at [email protected] and we can schedule a time.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
10x10 glider
100x100 with random initialization