{"id":"arxiv:2604.01754","title":"LiveMathematicianBench: A Live Benchmark for Mathematician-Level Reasoning with Proof Sketches","date":"2026-04-02","section":"Reasoning and the \"physics\" of language models","quality":0.041286,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/","source_url":"https://arxiv.org/abs/2604.01754","text_available":true,"authors":["Linyang He","Qiyao Yu","Hanze Dong","Baohao Liao"],"outgoing":[],"incoming":[],"text_parts":[{"start":0,"end":15455,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-1.html"},{"start":15455,"end":30655,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-2.html"},{"start":30655,"end":46351,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html"},{"start":46351,"end":56570,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html"}],"contents":[{"title":"LiveMathematicianBench : A Live Benchmark for Mathematician-Level Reasoning with Proof Sketches","offset":0,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-1.html#text"},{"title":"Abstract","offset":99,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-1.html#text"},{"title":"1 Introduction","offset":2612,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-1.html#text"},{"title":"2 Methods","offset":10168,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-1.html#text"},{"title":"3 Benchmark","offset":23824,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-2.html#text"},{"title":"4 Evaluation Results","offset":24527,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-2.html#text"},{"title":"Even frontier models suffer.","offset":24552,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-2.html#text"},{"title":"Category-wise pattern and monthly trajectories.","offset":24822,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-2.html#text"},{"title":"Substitution-resistant options sharply reduce accuracy.","offset":25709,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-2.html#text"},{"title":"Proof sketches yields a consistent improvement.","offset":27066,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-2.html#text"},{"title":"5 Endnote and Discussion","offset":27578,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-2.html#text"},{"title":"Appendix A Related Work","offset":31807,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"Standard and Advanced Mathematical Benchmarks","offset":31835,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"Contamination Mitigation, Formal Mathematics, and Frontier-Level Evaluation","offset":33046,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"Appendix B Human Validation Protocol","offset":35123,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"Appendix C Detailed Logical Taxonomy of Theorem Categories","offset":39295,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"1. Algorithmic / Constructive.","offset":39683,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"2. Asymptotic / Limit.","offset":40048,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"3. Biconditional / Equivalence.","offset":40468,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"4. Classification / Bijection.","offset":40877,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"5. Existence.","offset":42994,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"6. Existential–Universal.","offset":43291,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"7. Implication.","offset":43789,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"8. Independence / Consistency.","offset":44141,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"9. Inequality / Bound.","offset":44501,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"10. Nonexistence.","offset":44915,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"11. Uniqueness.","offset":45217,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"12. Universal.","offset":45544,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"13. Universal–Existential.","offset":45798,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"Appendix D Benchmark Composition Across Construction Stages","offset":46287,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-3.html#text"},{"title":"Appendix E Evaluation Implementation Details","offset":46987,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"Benchmark file format.","offset":47601,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"Prompt construction.","offset":48162,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"Answer extraction and scoring.","offset":49093,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"Execution details.","offset":49918,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"Resuming and result serialization.","offset":51674,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"Appendix F Additional Evaluation Results","offset":52825,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"F.1 Sketch-Aware Accuracy with Substitution-Resistant Settings","offset":52870,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"F.2 Cost Analysis","offset":53747,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"Appendix G Prompt Details","offset":55023,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"Stage 4a: Question Stem and Correct Choice Generation.","offset":55487,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"Stage 4b: Sketch-Adversarial Distractor Generation.","offset":55548,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"},{"title":"Instructions for reporting errors","offset":55606,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/part-4.html#text"}],"text_code_points":56570,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/bdd465567c85086267b19e2a2b0d404cb89ac243a1b7318f7ef2265a9818ddb6/source.md"}