Math-Graph: A Unified Statement-Level Dependency Graph for Informal and Formal Mathematics

Math-Graph is a unified, statement-level dependency graph spanning both informal and formal mathematics. It parses millions of theorem-like environments from mathematics arXiv and extracts typed declaration dependencies across 25 Lean 4 projects. The dataset was created by uw-math-ai and was last updated on 2026-06-28.

Use Cases

• Analyze dependency patterns in mathematical literature based on parsed theorem-like environments from arXiv.
• Train models for theorem discovery or proof suggestion based on the recovered directed dependency edges.
• Study the bridge between informal and formal mathematics based on the unified graph structure.
• Develop tools for formal verification based on the elaborator-level extraction of Lean 4 project dependencies.

Strengths

• Spans millions of theorem-like environments from the mathematics arXiv.
• Includes dependency data from 25 formal Lean 4 projects.
• Creates a unified bridge between informal and formal mathematical statements.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last updated 2026-06-28 21:56:05; freshness should be verified.

Provenance

Source

Mathematics arXiv and 25 Lean 4 projects.

Collection Method

Parsing of theorem-like environments and elaborator-level extraction of typed declaration dependencies.

Freshness

Last updated 2026-06-28 21:56:05.