{
  "abstract": "Evolutionary coding agents such as AlphaEvolve are one of the notable successes in using Large Language Models (LLMs) to build AI Scientists. These agents tackle open-ended scientific problems by iteratively improving and evolving programs, leveraging the prior knowledge and reasoning capabilities of LLMs. Despite their success, mutations over programs are mostly low-level upon surface heuristics, without a persistent unit of control. In contrast, science compounds through ideas. In this work, we propose ExpEvolve, which injects idea-level extraction and control into evolution: it discovers and leverages the high-level factors from programs that explain scoring differences among different programs. We characterize theoretical conditions on when discovery with idea factors improves sample efficiency. We also build a practical framework to show that the extracted high-level idea factors are helpful for guiding exploitation from experience and exploration when discovery comes to a plateau. Across 11 open-ended mathematical tasks, ExpEvolve significantly improves the final performance and sample efficiency up to 5.0x. The improvements also transfer to 4 other LLMs and to 6 machine learning tasks from MLS-Bench. Analysis shows that ExpEvolve can discover idea-level factors to provide better exploration and exploitation for evolutionary discovery.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2610.29303",
  "author_total": 7,
  "authors": [
    {
      "name": "Yongqiang Chen"
    },
    {
      "name": "Chenxi Liu"
    },
    {
      "name": "Deyu Zou"
    },
    {
      "name": "Zhenhao Chen"
    },
    {
      "name": "Tongliang Liu"
    },
    {
      "name": "Bo Han"
    },
    {
      "name": "Kun Zhang"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/fsrpa5t9gj4as.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.29303",
  "fulltext_url": null,
  "id": "fsrpa5t9gj4as",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.29303",
  "language": null,
  "licence": "CC-BY-SA-4.0",
  "licence_label": "CC BY-SA 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by-sa/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.CL"
    ],
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.ML"
    ],
    [
      "openalex",
      "1702|Evolutionary Algorithms and Applications"
    ]
  ],
  "openalex": "W7223200371",
  "pdf_url": "https://arxiv.org/pdf/2610.29303",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-21",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/fsrpa5t9gj4as.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:34:29Z",
      "native_id": "2610.29303",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.29303"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "ExpEvolve: Structuring and Reusing Experience for Open-Ended Discovery",
  "title_text": "ExpEvolve: Structuring and Reusing Experience for Open-Ended Discovery",
  "type": "preprint",
  "updated_at": "2026-10-30T12:34:29Z",
  "url": "https://research.draconitic.com/paper/fsrpa5t9gj4as",
  "venue": "arXiv",
  "versions": []
}