{
  "abstract": "Mechanistic modeling of chemical transformations offers a compelling basis for understanding reactivity and allows for prediction of reaction outcomes before attempting experiments. Despite progress in machine-learned interatomic potentials (MLIPs), we demonstrate that available models lack the accuracy for diverse reaction modeling. With this motivation, we developed a general MLIP for mechanistic modeling of closed-shell carbon, hydrogen, nitrogen, and oxygen reactions, AIMNet2-rxn, using a dataset of ∼4.7 × 10 6 range-separated density functional theory calculations. AIMNet2-rxn enables reaction modeling ∼10 6 faster than the reference quantum mechanical (QM) methods while substantially outperforming graph-based ML, reaffirming the value using three-dimensional chemical information for training. On a test suite of well-known reaction mechanisms—such as amide formation, proton transfers, and pericyclics—AIMNet2-rxn yields 1 to 2 kilocalories per mole accuracy across reaction coordinates without retraining or system-specific fine-tuning. To exploit graphics processing unit parallelism and AIMNet2-rxn efficiency, we introduce a batched nudged elastic band procedure that readily achieves minimum energy pathway search on a millions-of-reactions scale. To demonstrate complex reaction characterization, the thermodynamics of an 11-step pathway producing hydroxymethylfurfural, the experimentally observed major product of glucose pyrolysis, is evaluated. Overall, the accuracy and efficiency afforded by AIMNet2-rxn create opportunities in high-throughput reaction discovery and deep reaction network analysis that would be infeasible with QM methods.",
  "abstract_shown": true,
  "affiliations": [
    "Department of Chemical Engineering and Materials Science, Michigan State University, East Lansing, MI 48824, USA",
    "Department of Chemistry, Carnegie Mellon University, Pittsburgh, PA 15213, USA",
    "Carnegie Mellon University",
    "Michigan State University",
    "Deep Principle Inc., HangZhou, Zhejiang, China",
    "Department of Chemical Engineering, Purdue University, West Lafayette, IN 47907, USA",
    "Purdue University West Lafayette",
    "Department of Chemical Engineering, University of Notre Dame, Notre Dame, IN 46556, USA",
    "University of Notre Dame"
  ],
  "article_type": null,
  "arxiv": null,
  "author_total": 8,
  "authors": [
    {
      "affs": [
        0,
        1,
        2,
        3
      ],
      "name": "Dylan M. Anstine",
      "orcid": "0000-0002-4458-7080"
    },
    {
      "affs": [
        4,
        5,
        6
      ],
      "name": "Qiyuan Zhao",
      "orcid": "0000-0003-3228-8160"
    },
    {
      "affs": [
        1,
        2
      ],
      "name": "R.I. Zubatyuk",
      "orcid": "0000-0002-7285-9389"
    },
    {
      "affs": [
        1,
        2
      ],
      "name": "Shuhao Zhang",
      "orcid": "0000-0002-4240-3815"
    },
    {
      "affs": [
        7,
        8
      ],
      "name": "Veerupaksh Singla",
      "orcid": "0000-0001-6300-2167"
    },
    {
      "affs": [
        1,
        2
      ],
      "name": "Filipp Nikitin",
      "orcid": "0000-0002-1157-9471"
    },
    {
      "affs": [
        7,
        8
      ],
      "name": "Brett Matthew Savoie",
      "orcid": "0000-0002-7039-4039"
    },
    {
      "affs": [
        1,
        2
      ],
      "name": "Olexandr Isayev",
      "orcid": "0000-0001-7581-8497"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/dxab78ssr1xnp.bib",
  "cited_by": 21,
  "doi": "10.1126/sciadv.aea1557",
  "fulltext_url": null,
  "id": "dxab78ssr1xnp",
  "issns": [
    "2375-2548"
  ],
  "landing_url": "https://www.science.org/doi/10.1126/sciadv.aea1557",
  "language": "en",
  "licence": "CC-BY-NC",
  "licence_label": "CC BY-NC",
  "licence_source": "journal",
  "licence_url": "https://creativecommons.org/licenses/by-nc/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "openalex",
      "2505|Machine Learning in Materials Science"
    ]
  ],
  "openalex": "W4411574945",
  "pdf_url": null,
  "pmcid": "PMC13644750",
  "pmid": "42842670",
  "published": "2026-10-07",
  "publisher": "American Association for the Advancement of Science (AAAS)",
  "ris": "https://research.draconitic.com/paper/dxab78ssr1xnp.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:17:41Z",
      "native_id": "10.1126/sciadv.aea1557",
      "source": "crossref",
      "url": "https://www.science.org/doi/10.1126/sciadv.aea1557"
    },
    {
      "first_seen": "2026-10-30T13:27:00Z",
      "native_id": "W4411574945",
      "source": "openalex",
      "url": "https://doi.org/10.1126/sciadv.aea1557"
    }
  ],
  "subjects": [
    {
      "field": "chemistry",
      "origin": "openalex:2505|Machine Learning in Materials Science",
      "subfield": "materials-polymers"
    }
  ],
  "title": "AIMNet2-rxn: A machine-learned potential for generalized reaction modeling on a millions-of-pathways scale",
  "title_text": "AIMNet2-rxn: A machine-learned potential for generalized reaction modeling on a millions-of-pathways scale",
  "type": "peer-reviewed",
  "updated_at": "2026-11-09T13:21:22Z",
  "url": "https://research.draconitic.com/paper/dxab78ssr1xnp",
  "venue": "Science Advances",
  "versions": [
    {
      "doi": "10.26434/chemrxiv-2025-hpdmg",
      "pid": null,
      "relation": "preprint",
      "title": null,
      "venue": null
    }
  ]
}