Skip to content

README.md

Latest commit

 

History

History
59 lines (41 loc) · 2.25 KB

README.md

File metadata and controls

59 lines (41 loc) · 2.25 KB
Raw

Genetic Algorithm Solver

A Streamlit app that uses a genetic algorithm to solve linear equations of the form:

a·x + b·y + c·z + ... = Target

Given a set of coefficients and a target value, the algorithm evolves a population of candidate solutions over generations - selecting, combining, and mutating them - until it finds integer values that satisfy the equation.

Live Demo

[Add your Streamlit Cloud URL here once deployed]

Features

  • Custom equations - enter any number of coefficients
  • Adjustable parameters - population size, mutation rate, max generations
  • Elitism - the best candidate found is always carried into the next generation, so progress is never lost
  • Live convergence chart - tracks both best and average population fitness over time
  • Real-time feedback - progress bar and per-generation status updates

How It Works

  1. Initial population - random integer candidates are generated
  2. Fitness function - measures how far each candidate's result is from the target (lower is better, 0 = solved)
  3. Elitism - the best candidate from the current generation is preserved into the next
  4. Selection - the top-performing candidates are kept as parents
  5. Crossover - parent values are combined to produce new candidates
  6. Mutation - occasional random tweaks are introduced to maintain diversity and avoid getting stuck

The process repeats until either a candidate reaches zero error (an exact solution) or the maximum number of generations is reached, at which point the closest solution found is shown.

Installation & Local Setup

Clone the repository:

git clone https://github.com/salamlakhan7/Genetics-Algorithm-Solver.git
cd Genetics-Algorithm-Solver

Install dependencies:

pip install -r requirements.txt

Run the app:

streamlit run GA_app.py

Known Limitations

  • Candidate values are currently sampled from a fixed integer range, so very large or negative-only solutions may not be reachable without adjusting the code
  • Designed for linear equations; not currently built for quadratic or non-linear targets

Author

Built by Abdul Salam as a personal project exploring genetic algorithms with Python and Streamlit.