Iteris: Agentic Research Loops for Computational Mathematics
Leheng Chen, Zihao Liu, Wanyi He, Bin Dong · 2026
Read Iteris: Agentic Research Loops for Computational Mathematics by Leheng Chen, Zihao Liu, Wanyi He, Bin Dong (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-02484.txt.
Description
Chen and co-authors present Iteris, an AI research system for open problems in computational mathematics. Such problems need more than proofs. They also need numerical tests, adversarial constructions, and algorithm design. Language model agents have won ground in pure math. Computational math has seen less of that progress.
Iteris runs agentic research loops on open questions. The team tests it on two problems from a Simons Workshop collection. The system produced numerical evidence, constructions, and proof drafts. Expert review and correction turned those drafts into verified results. One result is a phase diagram. It compares conjugate gradient with randomized coordinate descent on power-law spectra. The other is a counterexample for QR factorization with column pivoting. It fails to pick well-conditioned submatrices under low coherence. Human validation stays essential in the loop.
Catalog record
| CREATOR | Leheng Chen, Zihao Liu, Wanyi He, Bin Dong |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 94 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.02484