{
  "abstract": "Universal machine learned interatomic potentials (uMLIPs) embody a growing area of interest due to their transferability across the periodic table, displaying an error of about 0.6 kcal/mol against the Matbench Discovery test set. However, we show that achieving more accurate predictions on out-of-domain tasks requires fine-tuning. Additionally, we investigate the existence and influence of model biases in molecular dynamics (MD) by examining two approaches for data generation: from multiple MD trajectories in parallel, which we call naive fine-tuning, and from a single MD trajectory with fine-tuning after set intervals, which we call iterative fine-tuning. Our results find that naive fine-tuning generates constrained data sets that fail to represent MD simulations, and thus downstream fine-tuned models fail during extrapolation. In contrast, iterative fine-tuning yields models that are more generalizable and accurate, producing stable dynamics. These findings indicate the role of uMLIP bias in fine-tuning, and highlight the need for multiple fine-tuning steps. Lastly, we relate unphysical behavior to principal component space, and quantify extrapolations through Q-residual analysis, which are useful as a proxy for epistemic uncertainty for larger simulations.",
  "abstract_shown": true,
  "affiliations": [
    "Department of Chemical Engineering",
    "Georgia Institute of Technology"
  ],
  "article_type": null,
  "arxiv": null,
  "author_total": 2,
  "authors": [
    {
      "affs": [
        0,
        1
      ],
      "name": "Nicolas Wong"
    },
    {
      "affs": [
        0,
        1
      ],
      "name": "Julia H. Yang",
      "orcid": "0000-0002-5713-2288"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/bwty19csphvxs.bib",
  "cited_by": 5,
  "doi": "10.1021/acs.jctc.6c00425",
  "fulltext_url": null,
  "id": "bwty19csphvxs",
  "issns": [
    "1549-9618",
    "1549-9626"
  ],
  "landing_url": "https://doi.org/10.1021/acs.jctc.6c00425",
  "language": "en",
  "licence": "CC-BY",
  "licence_label": "CC BY",
  "licence_source": "journal",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "openalex",
      "2505|Machine Learning in Materials Science"
    ]
  ],
  "openalex": "W7165144824",
  "pdf_url": "https://pubs.acs.org/jctcce/article-pdf/22/13/6820/65904517/acs.jctc.6c00425.pdf",
  "pmcid": "PMC13379080",
  "pmid": "42322232",
  "published": "2026-06-20",
  "publisher": "American Chemical Society",
  "ris": "https://research.draconitic.com/paper/bwty19csphvxs.ris",
  "sources": [
    {
      "first_seen": "2026-11-04T13:21:20Z",
      "native_id": "W7165144824",
      "source": "openalex",
      "url": "https://doi.org/10.1021/acs.jctc.6c00425"
    }
  ],
  "subjects": [
    {
      "field": "chemistry",
      "origin": "openalex:2505|Machine Learning in Materials Science",
      "subfield": "materials-polymers"
    }
  ],
  "title": "Bias in Universal Machine-Learned Interatomic Potentials and Its Effects on Fine-Tuning",
  "title_text": "Bias in Universal Machine-Learned Interatomic Potentials and Its Effects on Fine-Tuning",
  "type": "peer-reviewed",
  "updated_at": "2026-11-05T13:21:14Z",
  "url": "https://research.draconitic.com/paper/bwty19csphvxs",
  "venue": "Journal of Chemical Theory and Computation",
  "versions": []
}