A latent-space extrapolation grade built into graph atomic cluster expansion foundation potentials
Abstract
Foundation machine-learning interatomic potentials cover broad configurational and chemical spaces, but their reliability can vary across the atomic environments encountered during a simulation. Here we introduce the calibrated Mahalanobis (CALM) extrapolation grade γ, a piecewise differentiable per-atom quantity integrated into GRACE foundation models and evaluated alongside energies and forces in a single model pass. We define γ from nearest-cluster Mahalanobis distances in latent feature space, setting γ=1 from the training-distance distribution separately for each element and cluster. Controlled tests show that a normalized random projection of the invariant many-body basis detects structural and chemical extrapolation. On different foundation datasets, OMat24 and SMAX, γ correlates with atomic force errors and separates structures with different error distributions. The CALM grade adds percent-level computational cost, and its spatial gradient guides uncertainty-biased data collection toward configurations with larger absolute force errors.
Related
- Chemical site bases and average-atom potentials for the atomic cluster expansion
- FP-DeErr: Application-Oriented Error Decomposition for Foundation Potentials
- Foundation-Model Supervision of Distilled Potentials for Million-Atom Molecular Dynamics
- Diagnosing Latent Energy Decomposition in Machine-Learning Interatomic Potentials via Interacting Quantum Atoms
- Cluster-based Structural Similarity for Dataset Visualization and Data Selection for Machine Learning Interatomic Potentials
BibTeX
@misc{lysogorskiy2026latentspace,
title = {{A latent-space extrapolation grade built into graph atomic cluster expansion foundation potentials}},
author = {Yury Lysogorskiy and Anton Bochkarev and Ralf Drautz},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.40060},
eprint = {2609.40060},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.40060},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - A latent-space extrapolation grade built into graph atomic cluster expansion foundation potentials
AU - Yury Lysogorskiy
AU - Anton Bochkarev
AU - Ralf Drautz
PY - 2026
DA - 2026/09/30
PB - arXiv
DO - 10.48550/arxiv.2609.40060
UR - https://arxiv.org/abs/2609.40060
AB - Foundation machine-learning interatomic potentials cover broad configurational and chemical spaces, but their reliability can vary across the atomic environments encountered during a simulation. Here we introduce the calibrated Mahalanobis (CALM) extrapolation grade γ, a piecewise differentiable per-atom quantity integrated into GRACE foundation models and evaluated alongside energies and forces in a single model pass. We define γ from nearest-cluster Mahalanobis distances in latent feature space, setting γ=1 from the training-distance distribution separately for each element and cluster. Controlled tests show that a normalized random projection of the invariant many-body basis detects structural and chemical extrapolation. On different foundation datasets, OMat24 and SMAX, γ correlates with atomic force errors and separates structures with different error distributions. The CALM grade adds percent-level computational cost, and its spatial gradient guides uncertainty-biased data collection toward configurations with larger absolute force errors.
ER -