Skip to content

Latest commit

 

History

History
75 lines (52 loc) · 2.27 KB

README.md

File metadata and controls

75 lines (52 loc) · 2.27 KB
Raw
    _______  __   __  _______  _______
   |       ||  | |  ||       ||   _   |
   |    _  ||  |_|  ||    ___||  |_|  |
   |   |_| ||       ||   | __ |       |   Python framework for applying
   |    ___||_     _||   ||  ||       |   Genetic Algorithms
   |   |      |   |  |   |_| ||   _   |
   |___|      |___|  |_______||__| |__|

Build Status codecov

Project status: preparing first release

PyGa is a flexible, object-oriented framework for developing genetic algorithms in Python.

Features

  • Pluggable architecture -- source code is decoupled into many independent classes which can be used to build custom solutions.
  • Batteries included -- most common selection strategies and evolutionary operators are already implemented.

Installation

Until first release, package will not be published to pip repository. Currently the only way to obtain code is to do git clone or download zip.

Getting Started

Note: You should be familiar with how genetic algorithms work and what are basic components of the process.

engine = GenerationalEvolutionEngine()
engine.create(candidate_factory, operator, fitness_evaluator, selection_strategy)
population = engine.evolve(population_size, elite_count, termination_condition)

print(population.get_best())

Examples

Some basic examples are available to give You an idea how PyGA works.

Evolve population of random strings to match "Hello World":

python examples/match_string.py
DEBUG:root:generation = 0; best candidate: ooW e WdWod; fitness: 3
DEBUG:root:initialized population
DEBUG:root:generation = 1; best candidate: ooldd WdWod; fitness: 4
// snip
DEBUG:root:generation = 65; best candidate: Hello World; fitness: 11
DEBUG:root:finished processing
28056635-7393-43d9-8e3e-a2ebf6dc02bf

Documentation

Tests

The source code is very well covered by tests. To run all checks, run:

make test

To execute tests and generate code coverage report, run:

make coverage

Credits