{
  "abstract": "Machine-learned interatomic potentials distilled from foundation models enable fast large-scale molecular dynamics (MD), but they can fail locally in chemical environments that their training data did not sample. System-averaged error metrics and small random audits often miss such sparse or isolated failures. Here we use an exact equivariant foundation model, MACE-MPA-0, to check every atom of million-atom production simulations. Reorganizing two-layer MACE inference into a receiver-chunked three-pass schedule that keeps no per-edge intermediates resident reduces the marginal memory per atom 340-fold, evaluates up to $2.86\\times10^{7}$ atoms on a single H200 GPU, and supervises $10^{6}$-atom trajectories in LAMMPS at 1.3% of the run time. Distilled MACE students run four to nine times slower than NEP89 with 0.95-1.28 times the teacher's force error against density functional theory (DFT). In molten SiO$_2$, whole-system checks uncover a persistent tail of under-coordinated atoms hidden by the system-averaged error and an isolated, unphysical O-O contact that a 1% random audit misses 94% of the time. On-the-fly retraining on teacher-labelled 13-Å clusters, with the force loss restricted to their inner 6 Å, removes three quarters of the failure tail and the kilobar-level pressure biases in SiO$_2$ and compressed water; the repair converges to the teacher, not to DFT.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2610.23371",
  "author_total": 3,
  "authors": [
    {
      "name": "Xiaoqing Liu"
    },
    {
      "name": "Yangshuai Wang"
    },
    {
      "name": "Lei Shen"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/fsymny9xj7xka.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.23371",
  "fulltext_url": null,
  "id": "fsymny9xj7xka",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.23371",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "physics.comp-ph"
    ],
    [
      "openalex",
      "2505|Machine Learning in Materials Science"
    ]
  ],
  "openalex": "W7222935636",
  "pdf_url": "https://arxiv.org/pdf/2610.23371",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-16",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/fsymny9xj7xka.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:30:32Z",
      "native_id": "2610.23371",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.23371"
    }
  ],
  "subjects": [
    {
      "field": "physics",
      "origin": "arxiv:physics.comp-ph",
      "subfield": "computational"
    }
  ],
  "title": "Foundation-Model Supervision of Distilled Potentials for Million-Atom Molecular Dynamics",
  "title_text": "Foundation-Model Supervision of Distilled Potentials for Million-Atom Molecular Dynamics",
  "type": "preprint",
  "updated_at": "2026-10-30T12:33:28Z",
  "url": "https://research.draconitic.com/paper/fsymny9xj7xka",
  "venue": "arXiv",
  "versions": []
}