MechMath BlogsMechMath develops AI systems for mathematical research, combining theorem proving, formal verification, symbolic computation, and reusable mathematical knowledge.https://mechmath.github.io/blogs/2026-08-06T00:00:00+00:00MechGeo: Automatically Formalizing and Proving IMO Geometry in Lean 4https://mechmath.github.io/blogs/mechgeo-imo-geometry/2026-08-06T00:00:00+00:002026-08-06T00:00:00+00:00MechMath Team<p>MechGeo unifies autoformalization and automated proving for trustworthy Euclidean geometry. It produced Lean-checked statements, proofs, and counterexamples for 44 IMO geometry problems from 2000–2026 and achieved faithful coverage of all 14 LEAP geometry problems after validation and repair.</p>
MechMath Agent Team Is Now Open Sourcehttps://mechmath.github.io/blogs/mechmath-agent-team-open-source/2026-08-01T00:00:00+00:002026-08-01T00:00:00+00:00MechMath Team<p>The MechMath Agent Team project template is now publicly available. Learn how to create a research project, install its dependencies, and launch its natural-language, formal-language, and knowledge-management agents.</p>
IMO 2026: From Natural-Language Solutions to Lean Verificationhttps://mechmath.github.io/blogs/imo-2026-progress/2026-07-18T00:00:00+00:002026-07-18T00:00:00+00:00MechMath Team<p>A progress report on complete natural-language and Lean 4 solutions to all six IMO 2026 problems. It presents the public proof artifacts, verification workflow, and recorded NL/FL proof-generation times.</p>
From Hilbert's Decision Problem to Reasoning Agentshttps://mechmath.github.io/blogs/hilbert-to-agent/2026-07-08T00:00:00+00:002026-07-08T00:00:00+00:00Xiao-Shan Gao<p>A review of automated theorem proving from Hilbert’s decision problem to LLM-based research agents, with an introduction to MechMath Agent Team. It traces the shift from symbolic automation and interactive proving to agent systems that support mathematical research.</p>