EvoSim

An evolution simulator for a foraging genome based on Yale-NUS' SI1 foraging exercise.

Compiling and running (macOS)

Compile by runnnig:

1. g++ main.cpp -o evosim

To use the evolution simulator:

1. ./evosim

You must paste an initial genome that you want the simulator to evolve. You may make changes in the evolution variables by following the instructions below.

Changing Evolution Variables

You may change the following variables that are defined at the start of main.cpp:

• height: height of foraging board (default: 10)
• width: width of foraging board (default: 10)
• n_moves: number of moves per game (default: 200)
• chance: chance of mutating per character of genome (in percentage) (default: 4)
• f_chance 50 chance of square being food (in percentage) (default: 50)
• n_sim: number of simulations per genetic code (default: 100)
• n_gen: number of genetic codes per generation (default: 500)
• speed: speed of real time simulation in milliseconds (default: 200)

Further, you may change the main function with the help of the following guide:

curr_gen.adv_gens(50, false)

• change 50 to number of generations you would like to advance per iteration (before showing best version of that generation)
• change false to true to see the best version of generation simulated in real time (note: I would recommend you reduce speed in the above variables if you do this)

curr_gen.best(true);

• change true to false to not simulate the best version after every 50 generations in real time (this is good if you would like to keep the simulator running for a long time)

Suggested starting genomes

• Useless Genome (this will eventually evolve into Memory Genome) 000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000

• Your own genome

End genomes evolved up till now

• Memory Genome (this yields an average score of ~485, it evolved after ~2000 generations from the useless genome) 250350050250350250150350120150130250150330250150150310350350050350220000350350000050030050000230050050030230050100050000200050050150320000000000000320050000300110150250250150350250150330020150250120250230150110110330330030210020210320240250210

• Better Than Memory Genome (this will not evolve into Memory Genome as it yields a higher average score of around 490) 254354154054354254054354244156356156156354154154054344054354254254254254054054144054354254054354154054354044154354154154254154154354344054354054054254254354354144354354154254354254354354244154354154254354154154154444344244244044244344144344444

This was achieved when only rules which required here to be empty were evolved. That is, rules that had here as food were mapped to the eat action (action 5)