2D generations cellular automata consists of a 2D grid of cells. These cells live, die, or are in a “dying” transition date depending on some simple rules.
For the most part, there is no clear pattern in “interesting” rules with gliders. A user would have to randomly go through hundreds if not thousands of random rules before finding an interesting one.
Even more daunting, there are 2^28 combinations of rules: 2^9 survival rules, 2^9 born rules, and 2^10 states, assuming we are using a Moore neighborhood and a max of 10 possible states.
In this project, we automatically detect interesting rules so a user doesn’t have to manually go through them. We define interesting rules specifically as those that have clear gliders (small patterns that move across the grid) with some individual characteristics.
MLCAforPaper.ipynb contains our data preprocessing function and ML model
Data zip is located here: https://drive.google.com/file/d/1BAEvsQ2XtJNF2hGdhckPLBl0P2VBiTPS/view?usp=sharing . It contains 80k images, 40k "boring" and 40k "interesting" frames. If you want to train the model, download the dataset and provide its file path in the jupyter file.
Use python main.py to generate CA frames with certain rules/states. Specify rules in main.py. We used some helper functions from cellpylib (saving the frames and plotting them) but the core CA algorithm was implemented by us.
Data generation requirements:
- matplotlib
- PyQt5