LLM + Evolutionary Search

I tripped over another LLM + Evolutionary search post yesterday on HN:

• OpenEvolve: Teaching LLMs to Discover Algorithms Through Evolution (news.ycombinator.com)
• OpenEvolve: Teaching LLMs to Discover Algorithms Through Evolution (algorithmicsuperintelligence.ai)

I like this area of research. It’s an area I’d dig into, if I was compelled to do research.

I’ve touched on this before: AlphaEvolve and LLM Prompt Optimization and perhaps others.

What is this field of study called? No consensus yet.

What are the main (popular) projects?

I suspect projects like:

• AlphaEvolve (google)
  • AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms
  • AlphaEvolve: A coding agent for scientific and algorithmic discovery, White Paper.
  • AlphaEvolve, Wikipedia.
• OpenEvolve (algorithmic super intelligence)
  • OpenEvolve, GitHib.
  • OpenEvolve: Teaching LLMs to Discover Algorithms Through Evolution
• AlphaResearch (bytedance)
  • AlphaResearch: Accelerating New Algorithm Discovery with Language Models
• ShinkaEvolve (sakanaai)
  • ShinkaEvolve, GitHub
  • ShinkaEvolve: Towards Open-Ended And Sample-Efficient Program Evolution
  • ShinkaEvolve in Action: How a Human-AI Partnership Conquered a Coding Challenge
  • ShinkaEvolve: Evolving New Algorithms with LLMs, Orders of Magnitude More Efficiently
• LLM4AD (optima)
  • LLM4AD: Large Language Model for Algorithm Design, GitHub.
  • LLM4AD: A Platform for Algorithm Design with Large Language Model, Paper
• Darwin Gödel Machine (sakanaai)
  • The Darwin Gödel Machine: AI that improves itself by rewriting its own code
  • Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents, paper
  • Darwin Gödel Machine, GitHub
• AI Scientist (sakanaai)
  • The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery
  • The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery, paper
  • AI Scientist, GitHub
• CodeEvolve
  • CodeEvolve: An open source evolutionary coding agent for algorithm discovery and optimization, Paper
  • CodeEvolve, GitHub
• CALM
  • CALM: Co-evolution of Algorithms and Language Model for Automatic Heuristic Design, Paper
• AI Co-Scientist (google)
  • Towards an AI co-scientist, Paper
  • Accelerating scientific breakthroughs with an AI co-scientist

And many others.

It’s a mix, some are focused on algorithm-gen, others paper/discovery-gen. I may be mixing things too much here.

Perhaps the taxonomy in this (older) survey helps:

• LLM4EC: Interdisciplinary Research on LLM and Evolutionary Computation
  • Evolutionary Computation in the Era of Large Language Model: Survey and Roadmap, paper.