Get Started
Home
Topics
Search
Library
Research questionHow can LLM agents discover auxiliary constructions that support formally verified geometry proofs?Many geometry proofs depend on an auxiliary construction that is not apparent from the original statement. An agent must propose useful constructions and determine whether they genuinely advance a proof rather than merely generate plausible geometric objects.
AI
AI Agents
Evaluation & Benchmarks
Reasoning
Reinforcement Learning
Latest papersRecent research connected to this question, newest first.Achieving Olympiad-Level Geometry Large Language Model Agent via Complexity Boosting Reinforcement LearningThe source studies an LLM agent that proposes propositions and auxiliary constructions, checks them with a symbolic engine, and uses the resulting feedback for IMO-level geometry problems. Evidence is limited to the reported problems, training examples, and symbolic-engine setting, so broader mathematical domains are not covered.research paper · Sep 4, 2026
Related questions
How can mathematical-reasoning LLMs learn to construct counterexamples that reveal conceptual understanding?How can LLMs interleave reasoning with reliable step-level self-critique without a separate verifier?How can we tell whether LLM hidden-state geometry reflects reasoning operations rather than lexical or positional cues?How can large reasoning models explore complex research problems while maintaining proof rigor?