Evolutionary computation

This is a collection of algorithms I studied at my university for the Evolutionary Computation course.

This repository includes 7 methods for function optimization:

• Evolutionary strategy
• Differential evolutionary
• Symbiotic optimization
• Artificial bee colony
• Deformed stars
• Fractal structurization
• Simulated Annealing

You can also find Genetic Programming (GP) algorithm and Ant Colony Optimization (ACO) algorithm.

Installation

# Install directly from GitHub
pip install git+https://github.com/lex-unix/evolutionary-computations.git

# Or clone and install in development mode
git clone https://github.com/lex-unix/evolutionary-computations.git
cd evolutionary-computations
pip install -e .

Quick Start

import evocomp

# Basic usage with Differential Evolution
optimizer = evocomp.DifferentialEvolution(f=2, operation='min', epochs=100)
optimizer.optimize(evocomp.Easom())
print(optimizer.best_candidate)

# Using convergence criteria and maximizing function instead
optimizer = evocomp.DifferentialEvolution(
    f=2,
    operation='max',
    halt_criteria=evocomp.FitnessConvergence(e=0.001)
)
optimizer.optimize(evocomp.Easom())

Check examples/ directory for more usage examples.

Features

Parameter Studies

Study algorithm behaviour with different parameters values:

from evocomp.visualization import study, display, plot_histories

results = study(
    param_values=[0.1, 0.2, 0.3],
    setup=lambda x: evocomp.DifferentialEvolution(f=x, operation='min'),
    objective=evocomp.Easom()
)

display(results, param_name='F coefficient')
plot_histories(results)

Test Functions

Built-in test functions for algorithm evaluation:

optimizer.optimize(evocomp.ThreeHumpCamel())
optimizer.optimize(evocomp.Ackley())
optimizer.optimize(evocomp.Easom())

Add your own objective functions by implementing the Objective interface:

from evocomp import Objective
import numpy as np

class CustomFunction(Objective):
    @property
    def bounds(self) -> np.ndarray:
        return np.array([[-10.0, 10.0], [-10.0, 10.0]])

    def evaluate(self, solution: np.ndarray) -> float:
        x, y = solution
        return (x + 2*y - 7)**2 + (2*x + y - 5)**2