RMA: an Agentic System for Research-Level Mathematical Problems
Summary
Research Math Agents (RMA) is an agentic framework for automated reasoning on research-level mathematical problems, achieving state-of-the-art results on the First Proof benchmark by solving 8 out of 10 problems, outperforming strong baselines like GPT-5.2R and Aletheia.
View Cached Full Text
Cached at: 05/25/26, 08:54 AM
# RMA: an Agentic System for Research-Level Mathematical Problems
Source: [https://arxiv.org/abs/2605.22875](https://arxiv.org/abs/2605.22875)
[View PDF](https://arxiv.org/pdf/2605.22875)
> 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\.
## Submission history
From: Zelin Zhao \[[view email](https://arxiv.org/show-email/b0f3eb6c/2605.22875)\] **\[v1\]**Wed, 20 May 2026 04:54:22 UTC \(136 KB\)Similar Articles
Evaluating SageMath-Augmented LLM Agents for Computational and Experimental Mathematics
This paper proposes a ReAct-style agentic setup that combines LLM reasoning with verifiable feedback from SageMath, evaluating it on research-level mathematical problems. Results show substantial performance gains across models, with GPT-5.5 achieving the highest solve rate of 75.2%.
AI Co-Mathematician: Accelerating Mathematicians with Agentic AI
This paper introduces the AI Co-Mathematician, a workbench that uses agentic AI to support mathematicians in open-ended research tasks like ideation and theorem proving. Early tests show the system achieving state-of-the-art results on hard problem-solving benchmarks, including a 48% score on FrontierMath Tier 4.
Learning What to Investigate Next: Meta-Reasoning for Long-Horizon Research Agents
The paper introduces MIRA, a hierarchical meta-reasoning architecture for long-horizon research agents that separates research allocation from execution, and MIRA-AC, a generative actor-critic that learns credit assignment at meta-reasoning decision boundaries, improving theorem proving and neural architecture research.
Beyond the Library: An Agentic Framework for Autoformalizing Research Mathematics
Presents an agentic framework using general coding LLMs to autoformalize research-level mathematics into Lean 4 code, evaluated on Putnam problems and STOC conference papers.
DuMate-DeepResearch: An Auditable Multi-Agent System with Recursive Search and Rubric-Grounded Reasoning
This technical report introduces DuMate-DeepResearch, a multi-agent framework for deep research tasks that decouples the agent core from a tool ecosystem, and incorporates graph-based dynamic planning, recursive two-level execution, and rubric-based test-time optimization. The system achieves state-of-the-art results on two deep research benchmarks, demonstrating the value of auditable agent infrastructure.