FamilyFitness

Why I built it

I wanted to understand how different selection models change evolutionary search behavior, then make that behavior visible enough for other people to inspect.

What it does

• Runs genetic algorithm experiments with multiple selection methods.
• Compares classic baselines with a custom FamilyFitness lifecycle model.
• Provides a browser demo where real C++ genetic algorithm code runs through WebAssembly.

Tech stack

Build / Deployment

App type

Standalone genetic algorithm demo and experiment framework

Code location

Separate project repo

Front end

Web demo plus Python / Streamlit dashboard

Engine

C++ genetic algorithm core compiled to WebAssembly

Deployment

genetics.gavinnesom.com

Public URL

https://genetics.gavinnesom.com

Notes

Browser demo runs compiled C++ through WebAssembly; command-line experiment runner should remain intact.

Architecture / How it works

• The project is intentionally split into layers. The central C++ code handles population setup, evaluation, selection, reproduction, mutation, aging, death, and generation-by-generation experiment flow.
• The framework supports classic fixed-size genetic algorithm baselines such as tournament and rank selection. FamilyFitness adds a lifecycle model with age, sex, family and lineage ideas, and population dynamics.
• The framework is moving toward a cleaner structure where a problem, genome representation, decoder, selection method, and lifecycle model can be combined more flexibly.
• The command-line version is used for repeatable experiments. It can run configured experiments, set seeds, generate CSV outputs, and support larger comparison runs.
• The Python / Streamlit dashboard inspects outputs from experiment runs and helps compare fitness curves, population behavior, and run outcomes.
• The web demo is intentionally smaller and more visual. It presents an interactive text-matching genetic algorithm so a visitor can type something in, press run, and immediately see evolutionary search happening.
• The command-line tool is for serious experiment runs and batch output. The web demo is for explanation, portfolio value, and immediate visual feedback. Both should remain possible.

What I built

• A C++ genetic algorithm engine.
• Configurable experiment runs from the command line.
• Multiple selection methods, including tournament, rank, and FamilyFitness.
• Multiple fitness problems, including bitstring-style benchmark problems and the newer text-matching web demo.
• CSV and telemetry output for analysis.
• A Python / Streamlit dashboard for inspecting experiment results.
• A browser demo that runs the C++ code through WebAssembly.

What I learned

• Simple genetic algorithms are easy to start but hard to compare fairly.
• Visualization and telemetry matter as much as the algorithm itself.
• A model can become too realistic and stop being useful.
• Separating the core engine from the UI makes the project more flexible.
• WebAssembly is useful when you want real compiled code running in a browser without rewriting the algorithm in JavaScript.

Current status

The public browser demo is live at genetics.gavinnesom.com, while the deeper experiment framework remains a separate project with command-line and dashboard pieces.

Next steps

• Improve the comparison dashboard.
• Add clearer charts to the web demo.
• Make more problems selectable from the browser.
• Keep the command-line experiment runner clean and useful for larger batch runs.
• Continue refactoring toward a more generic evolutionary-computation framework.

Links