Skip to content

Stellar Colosseum: A Many-Agent Harness for Long-Horizon Research in Mathematics and Theoretical Computer Science

Sep 2026 · 3 citations · ⚡ 1 influential · 68 references
Computer Science

TL;DR

Stellar Colosseum is introduced, a model-agnostic harness for allocating inference across research in mathematics and theoretical computer science that demonstrates the capabilities of Colosseum through open-ended research and evaluations on theorem-proving and competitive programming benchmarks.

Abstract

Language models can produce plausible short proofs, but may still be unreliable on long-horizon research problems, where progress depends on a sequence of uncertain and interdependent decisions. We introduce Stellar Colosseum, a model-agnostic harness for allocating inference across research in mathematics and theoretical computer science. Colosseum explores alternative strategies before proof construction, uses a readiness gate to decide when a route is mature enough to decompose, represents the proof plan as interdependent section-level subproblems, and routes verifier findings back to the affected part of the argument. Across these stages, it generates candidates in parallel, attacks them with targeted falsification, and combines candidates and their critiques into a single research artifact through overlapping random-sample tree aggregation. The Colosseum workflow has been integrated into Google Antigravity's Teamwork framework as the Long Proof pattern. We demonstrate the capabilities of Colosseum through open-ended research and evaluations on theorem-proving and competitive programming benchmarks. Using Colosseum with Gemini 3.1 Pro, we obtain several new results that address open problems arising from papers published at top venues such as FOCS and JMLR. On TCS-Bench, a benchmark of research-level theorem-proving tasks drawn from papers published at FOCS, STOC, and SODA, Colosseum achieves 71.0% accuracy using Gemini 3.1 Pro and Gemini 3.7 Flash. In a separate Codeforces evaluation using Gemini 3.1 Pro, the proof-oriented pipeline with execution feedback solves 218 of 222 problems.

View source

Similar papers

Review Sep 2026

Long-horizon autoformalization of a core theorem underlying MIP* = RE

This work completed a machine-checked Lean 4 proof of the quantum soundness of the classical low individual-degree test, a core theorem underlying MIP* = RE, and provides a verified foundation for quantum complexity.

Si-Rui Lu, Rui-Xuan Deng, David Zhu et al. · 1 citation

TCSAlgBench: Benchmarking Automated Proving for Research-Level Theoretical Computer Science

TCSAlgBench provides a refreshable testbed for measuring progress in model reasoning and studying how agent workflows support research-level proof discovery, and evaluates ten model configurations from four families under direct inference and prover-verifier discussion, and compares four agent workflows under matched m...

Chu-Tong Yang, Xi-Yuan Zhang, Yu Huang et al. · 0 citations

Cogentic: Multi-Agent Orchestration for Automated Proof Discovery

We present Cogentic, a multi-agent harness for automated proof discovery on open research problems. While frontier language models can generate strong mathematical ideas in a single shot, single-shot generation is often insufficient for open problems that require exploring multiple competing conjectures, overcoming sub...

Yang Cai, Vineet Gupta, Yan-Chen Jiang et al. · 1 citation
#artificial intelligence Preprint Oct 2026

Continual Graph Memory for Mathematical Research Agents

Using frontier agent harnesses to tackle mathematical research problems has emerged as an effective means of advancing mathematics. However, solving frontier problems in mathematics may require a massive number of agents working in parallel for extended periods to construct proofs, thereby generating an enormous volume...

Jun-Yi Zhang, Jin-Xi Yu, E. Jiang et al. · 1 citation

Related blog posts

MIT News · Artificial Intelligence Nov 17, 2026

Learning the language of machine learning

By studying the computational foundations of language learning, Associate Professor Jacob Andreas builds intelligent systems that can communicate effectively with humans.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.