Genetic learning algorithm written in Rust with the macroquad graphics library.
Extended fork of Shorelark written bottom-up with Learning to Fly.
Added an adverserial system of chasers (carnivores) and herbivores (as opposed to only "herbivores"). The carnivores' goal is to eat the herbivores. The herbivores goal is to not die while eating as many plants as possible. Code and learning architecture rewritten and expanded to accomodate this. Available models are POSITIONAL, CELLULAR and CLOSEST. Settings can be modified/fine-tuned in /app/src/main.rs. Project can be compiled with cargo build or cargo run. Add the --release tag for acceptable performance: cargo run --release.
Window example:
Adjustable Settings:
/py_gen_plotter is used for plotting fitness. Examples:
pos_vis_safe_evolve:
closest_safe_evolve:
Refer to _data_archive/instances for more figures.