Gold-Medal-Level Olympiad Geometry Solving with Efficient Heuristic Auxiliary Constructions
By: Boyan Duan , Xiao Liang , Shuai Lu and more
Potential Business Impact:
Solves hard math problems better than computers.
Automated theorem proving in Euclidean geometry, particularly for International Mathematical Olympiad (IMO) level problems, remains a major challenge and an important research focus in Artificial Intelligence. In this paper, we present a highly efficient method for geometry theorem proving that runs entirely on CPUs without relying on neural network-based inference. Our initial study shows that a simple random strategy for adding auxiliary points can achieve silver-medal level human performance on IMO. Building on this, we propose HAGeo, a Heuristic-based method for adding Auxiliary constructions in Geometric deduction that solves 28 of 30 problems on the IMO-30 benchmark, achieving gold-medal level performance and surpassing AlphaGeometry, a competitive neural network-based approach, by a notable margin. To evaluate our method and existing approaches more comprehensively, we further construct HAGeo-409, a benchmark consisting of 409 geometry problems with human-assessed difficulty levels. Compared with the widely used IMO-30, our benchmark poses greater challenges and provides a more precise evaluation, setting a higher bar for geometry theorem proving.
Similar Papers
Achieving Olympia-Level Geometry Large Language Model Agent via Complexity Boosting Reinforcement Learning
Artificial Intelligence
AI solves hard geometry problems like Olympians.
Geoint-R1: Formalizing Multimodal Geometric Reasoning with Dynamic Auxiliary Constructions
Artificial Intelligence
Helps computers solve hard geometry problems.
Optimizing Geometry Problem Sets for Skill Development
History and Overview
Helps computers teach geometry by solving problems.