about
From Expert-Guided Proof Search to Automated Open-Problem Solving (arxiv.org)
2 points by sbulaev 40 days ago | hide | past | pdf | discuss on HN

In plain words: Bolzano runs many proof-searching agents in parallel, checks their work with a separate verifier, and keeps a readable record of progress. Without expert guidance it solved about 200 open problems, including questions from a top theory conference that their authors confirmed.

Abstract

Large language models are increasingly contributing to mathematical research, where progress often depends on efficient proof search, incremental improvements and careful verification. We describe Bolzano, a multi-agent open-source system that uses parallel prover agents with a verifier agent and maintains a human-readable research state. Initial manual use on expert-selected problems yielded 8 results whose proofs were checked by domain experts. Motivated by these case studies, we ran Bolzano without problem-specific human guidance on about 3,800 open problems extracted from four sets of papers, solving about 200 open problems. One experiment used papers accepted to STOC 2026, a top conference in theoretical computer science. There, we answered four questions raised in the papers, as confirmed by their authors.

Adrián Zámečník, Matěj Kripner, Martin Koutecký, Martin Balko, Jan Grebík, Pavel Hubáček, Robert Šámal, Václav Rozhoň
arXiv:2610.09769 · cs.AI, cs.CL · submitted Oct 7, 2026
abstract · pdf · html · Accepted at the 6th Workshop on Mathematical Reasoning and AI (MATH-AI), NeurIPS 2026

add comment on HN