Genetic Algorithm Optimization - Rastrigin Function

Overview

This project implements a Genetic Algorithm (GA) to optimize the Rastrigin function, a non-convex multimodal function commonly used as a benchmark for optimization algorithms. The goal is to evaluate and test various GA parameters to minimize the function in the least number of generations.

Objectives

• Test and evaluate the most important GA parameters to find the minimum value of the Rastrigin function efficiently.
• Analyze the impact of individual parameters on the convergence and performance of the algorithm.
• Provide tables summarizing the tested parameters and their results.
• Include graphical representations to justify findings.

Features

• Python implementation using the geneticalgorithm2 library.
• Automated generation of plots for convergence history and parameter comparisons.
• Support for multiple experiments with varying GA configurations.
• Comparison of GA results with Random Search and Gradient Descent methods.

Results

• Tables summarizing parameter configurations and their corresponding results.
  • See the table in output/ga_results.csv for detailed experiment results.
• Graphs illustrating convergence trends and comparisons between optimization methods:

Function Plots

2D Plot

3D Plot

GA Convergence

Linear Scale

Logarithmic Scale

Comparison Plots

Linear Scale

Logarithmic Scale

• Observations and comments on the impact of parameter changes.

Usage

Run the pipeline using:

python genetical_algorithm.py

Generated outputs include:

• Function plots (2D and 3D).
• GA convergence plots (linear and logarithmic scales).
• Comparison plots between GA, Random Search, and Gradient Descent.
• Results saved in output/ga_results.csv.

Reference

[1] https://github.com/PasaOpasen/geneticalgorithm2/tree/main

[2] https://github.com/PasaOpasen/OptimizationTestFunctions