Skip to content
DraconiticResearch
Search Sources API
201,683 studies31 sourcesUpdated 14 min ago
Peer-reviewedCC BYMachine Learning Science and Technology

Comparative study of ensemble-based uncertainty quantification methods for neural network interatomic potentials

  • Yonatan Kurniawan1
  • Mingjian Wen2
  • Ellad B. Tadmor3
  • Mark K. Transtrum4
4 affiliations
  1. 1Brigham Young University
  2. 2Institute of Fundamental and Frontier Sciences, University of Electronic Science and Technology of China
  3. 3University of Minnesota
  4. 4Cross Stream Consulting
Open at sourceDOICite

Abstract

Machine learning interatomic potentials (MLIPs) enable atomistic simulations with near first-principles accuracy at substantially reduced computational cost, making them powerful tools for large-scale materials modeling. The accuracy of MLIPs is typically validated on a held-out dataset of ab initio energies and atomic forces. However, accuracy on these small-scale properties does not guarantee reliability for emergent, system-level behavior—precisely the regime where atomistic simulations are most needed, but for which direct validation is often computationally prohibitive. As a practical heuristic, predictive precision—quantified as inverse uncertainty—is commonly used as a proxy for accuracy, but its reliability remains poorly understood, particularly for system-level predictions. In this work, we systematically assess the relationship between predictive precision and accuracy in both in-distribution (ID) and out-of-distribution (OOD) regimes, focusing on ensemble-based uncertainty quantification (UQ) methods for neural network potentials, including bootstrap, dropout, random initialization, and snapshot ensembles. We use held-out cross-validation for ID assessment and calculate cold curve energies and phonon dispersion relations for OOD testing. These evaluations are performed across various carbon allotropes as representative test systems. We find that uncertainty estimates can behave counterintuitively in OOD settings, often plateauing or even decreasing as predictive errors grow. These results highlight fundamental limitations of current UQ approaches and underscore the need for caution when using predictive precision as a stand-in for accuracy in large-scale, extrapolative applications.

Related

  1. Towards Out-of-Distribution Uncertainty Quantification for Machine-Learned Interatomic Potentials via Pre-Training
    Stefan P. Schmid, Tobias Setzer, Maximilian Scheurer
    arXivPreprintCC BY 4.0Full textPhysical chemistryarXiv
  2. Bayesian neural networks versus deep ensembles for uncertainty quantification in machine learning interatomic potentials
    Riccardo Farris, Emanuele Telari, Nongnuch Artrith +2
    Machine Learning Science and TechnologyPeer-reviewedCC BYFull text6 citedMaterials & polymersOpenAlex
  3. AdaptNTK: Adaptive Uncertainty Quantification and Active Learning for Neural Network Potentials
    Prajwal Ananth, Shuwen Yue
    arXivPreprintarXiv licenceFull textPhysical chemistryarXiv
  4. Correcting the Liquid Densities of Universal Machine-Learning Interatomic Potentials Using Local-Number-Density Repulsion
    Shunsuke Mieda
    ChemRxivPreprintLicence not statedGeneral chemistryChemRxiv
  5. Foundation interatomic potentials learn the practical chemistry they were never taught
    Alexander A. Kolganov, Mas P. Klein, Evgeny A. Pidko
    ChemRxivPreprintLicence not statedGeneral chemistryChemRxiv

Cite

BibTeXRIS
BibTeX
@article{kurniawan2026comparative,
  title = {{Comparative study of ensemble-based uncertainty quantification methods for neural network interatomic potentials}},
  author = {Yonatan Kurniawan and Mingjian Wen and Ellad B. Tadmor and Mark K. Transtrum},
  journal = {Machine Learning Science and Technology},
  year = {2026},
  month = aug,
  doi = {10.1088/2632-2153/ae9fb4},
  url = {https://doi.org/10.1088/2632-2153/ae9fb4},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - Comparative study of ensemble-based uncertainty quantification methods for neural network interatomic potentials
AU  - Yonatan Kurniawan
AU  - Mingjian Wen
AU  - Ellad B. Tadmor
AU  - Mark K. Transtrum
PY  - 2026
DA  - 2026/08/27
JO  - Machine Learning Science and Technology
DO  - 10.1088/2632-2153/ae9fb4
UR  - https://doi.org/10.1088/2632-2153/ae9fb4
AB  - Machine learning interatomic potentials (MLIPs) enable atomistic simulations with near first-principles accuracy at substantially reduced computational cost, making them powerful tools for large-scale materials modeling. The accuracy of MLIPs is typically validated on a held-out dataset of ab initio energies and atomic forces. However, accuracy on these small-scale properties does not guarantee reliability for emergent, system-level behavior—precisely the regime where atomistic simulations are most needed, but for which direct validation is often computationally prohibitive. As a practical heuristic, predictive precision—quantified as inverse uncertainty—is commonly used as a proxy for accuracy, but its reliability remains poorly understood, particularly for system-level predictions. In this work, we systematically assess the relationship between predictive precision and accuracy in both in-distribution (ID) and out-of-distribution (OOD) regimes, focusing on ensemble-based uncertainty quantification (UQ) methods for neural network potentials, including bootstrap, dropout, random initialization, and snapshot ensembles. We use held-out cross-validation for ID assessment and calculate cold curve energies and phonon dispersion relations for OOD testing. These evaluations are performed across various carbon allotropes as representative test systems. We find that uncertainty estimates can behave counterintuitively in OOD settings, often plateauing or even decreasing as predictive errors grow. These results highlight fundamental limitations of current UQ approaches and underscore the need for caution when using predictive precision as a stand-in for accuracy in large-scale, extrapolative applications.
SN  - 2632-2153
ER  -