pycuga 0.0.12

Python Cuda Genetic Algorithm Package

📦pycuga

Pycuga (PYthon CUda Genetic Algorithm) provides a simple and easy package for performing island-based genetic algorithm using Python and Cuda.

Motivation

• When I worked on my previous project on Solving Maximum Satisfiability Problem using CUDA, I realised a lot of code could be reused, which save a lot of development time for solving other optimisation problems using genetic algorithm and CUDA.

Parameters

Methods currently supported
Selection	Elitism, Roulette Wheel
Crossover	Single, Double, Uniform
Mutation	Number

pip install pycuga

p1 = PyCUGA( isTime, time , constArr, chromosomeSize, stringPlaceholder,mutationNumber , selectionMode, crossoverMode)
p1. launchKernel(islandSize , blockSize , chromosomeNo, migrationRounds,rounds)

Examples here.

Limitations

• Use multiples of 32 (for chromosome parameters) to avoid bugs and increase efficiency.
• Island migration is limited to 1 item currently
• Lack unit testing

❗Disclaimer

This is a mini project which I've put quite a lot of time and effort into, but I can't take responsibility for any bugs nor errors.