{
  "abstract": "Machine-learned formation enthalpies for inorganic materials approach density functional theory (DFT) accuracy, yet thermodynamic stability predictions derived from them can remain unreliable, a gap commonly associated with error cancellation in DFT that independent per-compound training does not reproduce. We introduce the interference index ξ, a dimensionless measure of how strongly errors reinforce or cancel in linearly derived quantities. Its root-mean-square average, ξrms, provides a reaction-size-independent null reference, with ξrms = 1 for the isotropic null model. Across seven published models benchmarked on Materials Project compounds, error-cancellation behaviour is distinct from per-compound accuracy and becomes progressively less favourable with increasing formation-enthalpy accuracy among the compositional models. We then introduce iiLoss, a reaction-aware objective that favours cancellation of formation-enthalpy errors rather than directly minimising decomposition-enthalpy error. On Materials Project data, iiLoss reduces decomposition-enthalpy error by 21% and improves thermodynamic stability classification by 6.8% in F1 score, outperforming direct supervision on decomposition enthalpies. These results show that error cancellation can be quantified and directly optimised to improve derived thermodynamic predictions.",
  "abstract_shown": true,
  "affiliations": [
    "Department of Chemistry, Biology and Biotechnology",
    "University of Perugia",
    "Dipartimento di Farmacia",
    "Università G. D’Annunzio",
    "Institute of Science and Chemical Technologies \"Giulio Natta\"-SCITEC, National Research Council, via Elce di Sotto, Perugia, Italy, 06123"
  ],
  "article_type": null,
  "arxiv": null,
  "author_total": 5,
  "authors": [
    {
      "affs": [
        0,
        1
      ],
      "name": "Alexander Kanzow"
    },
    {
      "affs": [
        0,
        1
      ],
      "name": "Cesare Boriosi"
    },
    {
      "affs": [
        0,
        1
      ],
      "name": "Carlos R. Jacinto-Mejía"
    },
    {
      "affs": [
        2,
        3,
        4
      ],
      "name": "Loriano Storchi"
    },
    {
      "affs": [
        0,
        1
      ],
      "name": "Giovanni Bistoni",
      "orcid": "0000-0003-4849-1323"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/73y611yxe9zen.bib",
  "cited_by": 0,
  "doi": "10.26434/chemrxiv.15008797/v1",
  "fulltext_url": null,
  "id": "73y611yxe9zen",
  "issns": [],
  "landing_url": "https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008797/v1",
  "language": null,
  "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": "W7213231093",
  "pdf_url": null,
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-14",
  "publisher": "American Chemical Society (ACS)",
  "ris": "https://research.draconitic.com/paper/73y611yxe9zen.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:17:19Z",
      "native_id": "10.26434/chemrxiv.15008797",
      "source": "chemrxiv",
      "url": "https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008797/v1"
    }
  ],
  "subjects": [
    {
      "field": "chemistry",
      "origin": "openalex:2505|Machine Learning in Materials Science",
      "subfield": "materials-polymers"
    }
  ],
  "title": "The Interference Index: Quantifying and Improving Error Cancellation in Machine-Learned Thermodynamic Stability Predictions",
  "title_text": "The Interference Index: Quantifying and Improving Error Cancellation in Machine-Learned Thermodynamic Stability Predictions",
  "type": "preprint",
  "updated_at": "2026-10-30T12:17:19Z",
  "url": "https://research.draconitic.com/paper/73y611yxe9zen",
  "venue": "ChemRxiv",
  "versions": []
}