Show HN: A knowledge graph of 15,941 math states – proof as path-finding

Category: ai-ml

Tags: knowledge-graph, mathematical-proof, llm-orchestration, automated-theorem-proving, path-finding

Score: 7.0/10 (Innovation: 8, Technical: 8, Documentation: 7, Utility: 5)

The Mathematical Discovery Engine reframes mathematical proof as path-finding through a large knowledge graph of 15,941 states and 1,223 proof techniques, using LLMs to search for new theorems. Its innovative combination of structured graph construction, technique clustering, and LLM orchestration with pruning rules is interesting for automating discovery, though practical utility depends on scaling and formal verification.

Target audience: data engineers, ai researchers, mathematicians

Repository: https://ansumandas441.github.io/mathematical-discovery-engine/ · Python · MIT · 30 stars

View on Hacker News