RMA: an Agentic System for Research-Level Mathematical Problems

Researchers have developed Research Math Agents (RMA), an agentic framework that automates reasoning on research-level mathematical problems by decomposing proof-solving into specialized modules coordinated by multiple agents. In evaluations on the First Proof benchmark of ten expert-contributed problems, RMA solved eight problems, outperforming strong baselines including GPT-5.2R and Aletheia by producing more logically sound and readable proofs. The system's performance improvements stem from the interaction of structured reasoning modules, iterative refinement, and verifier-based feedback rather than any single component.

arXiv:2605.22875v1 Announce Type: new Abstract: We present $\textbf{Research Math Agents RMA }$, an agentic framework for automated reasoning on research-level mathematical problems. Unlike prior studies centered on competition mathematics or formal theorem proving, RMA targets research-level mathematical problems that require long-horizon reasoning, literature grounding, and iterative proof refinement. RMA decomposes research-level proof solving into specialized modules for problem analysis, literature search and understanding, fair comparison, knowledge-bank construction, and proof verification, all coordinated by initializer, proposer, and verifier agents through a shared structured memory. Within this unified framework, these agents operate in a multi-role, multi-round workflow, collaboratively generating, refining, and verifying candidate proofs through iterative feedback. We evaluate RMA on the First Proof benchmark, which consists of ten research-level problems contributed by expert mathematicians across diverse domains. Through comprehensive expert evaluation, RMA outperforms strong baselines on the First Proof benchmark, including GPT-5.2R and Aletheia, solving eight out of ten research problems and producing more logically sound and readable proofs. Our comprehensive ablation studies further show that performance gains arise from the interaction of structured reasoning modules, iterative refinement, and verifier-based feedback, rather than any single component. Our solutions and implementations will be made publicly available upon acceptance.