Mathematical discovery and exploration can be done at scale.

AlphaEvolve autonomously rediscovered or improved solutions for 67 mathematical problems, demonstrating large-scale AI-driven mathematical discovery.

• Why it matters: Advancing automated mathematical problem-solving addresses the challenge of exploring vast search spaces and complementing human intuition, potentially transforming mathematical research.
• What they did: The system combines large language models with automated evaluation in an iterative loop, testing and refining solutions across diverse fields like analysis, combinatorics, and geometry, and integrating proof assistants for deeper insights.
• The result: AlphaEvolve often matches or surpasses the best known solutions, generalizes results into formulas, and enables faster, scalable exploration of complex problems, opening new avenues for AI-assisted mathematics.