Genetic Aglorithm framework in Python.
Mendel is a general framework for creating genetic algorithms in Python. While the general workflow of genetic algorithms can be generalized (i.e. create a population of set size, represent different parameter values as individuals within the population, assess performance of individuals given some function fitness function, "genetically cross" individuals to create new population, and finally repeat until some stopping criteria), inevitably some problem-specific routines must be considered.
To account for a generalized framework, while also allowing for
problem-specific implementations, Mendel generalizes the genetic algorithm
process in the GeneticAlgorithm and Individual classes. Problem specific
methods can then be implemented by extending the FitnessMixin class. This
structure allows users to specify their own handling of data input, parameter
unpacking, and performance evaluation, without having to re-implement the more
general aspects of a genetic algorithm.
Mendel requires Python 3.6, numpy/scipy for numerical computations,
tqdm for progress bars, and pytest for unit testing.
To run tests, first install pytest, open a terminal, navigate to the
repository head, and issue the command pytest -v tests/mendel_tests.py.
The repo is not currently available on popular package managers, such as pip
and conda. To install, please fork this repository.