Overcoming systematic softening in universal machine learning interatomic potentials by fine-tuning
Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:6GOSTSQLrecord.jsonopen to challenge →
read the original abstract
Machine learning interatomic potentials (MLIPs) have introduced a new paradigm for atomic simulations. Recent advancements have seen the emergence of universal MLIPs (uMLIPs) that are pre-trained on diverse materials datasets, providing opportunities for both ready-to-use universal force fields and robust foundations for downstream machine learning refinements. However, their performance in extrapolating to out-of-distribution complex atomic environments remains unclear. In this study, we highlight a consistent potential energy surface (PES) softening effect in three uMLIPs: M3GNet, CHGNet, and MACE-MP-0, which is characterized by energy and force under-prediction in a series of atomic-modeling benchmarks including surfaces, defects, solid-solution energetics, phonon vibration modes, ion migration barriers, and general high-energy states. We find that the PES softening behavior originates from a systematic underprediction error of the PES curvature, which derives from the biased sampling of near-equilibrium atomic arrangements in uMLIP pre-training datasets. We demonstrate that the PES softening issue can be effectively rectified by fine-tuning with a single additional data point. Our findings suggest that a considerable fraction of uMLIP errors are highly systematic, and can therefore be efficiently corrected. This result rationalizes the data-efficient fine-tuning performance boost commonly observed with foundational MLIPs. We argue for the importance of a comprehensive materials dataset with improved PES sampling for next-generation foundational MLIPs.
This paper has not been read by Pith yet.
Forward citations
Cited by 7 Pith papers
-
Synthesizability, hardness, and stacking order in multicomponent transition metal carbides from machine-learned potentials
Fine-tuned MACE potentials screen 1500 equiatomic multicomponent carbides, showing group number controls stability/hardness, short-range order effects are small, and a new family of stacking-ordered phases appears in ...
-
Bayesian E(3)-Equivariant Interatomic Potential with Iterative Restratification of Many-body Message Passing
Bayesian E(3)-equivariant MLPs with joint energy-force NLL loss achieve competitive accuracy while enabling uncertainty-guided active learning, OOD detection, and calibration.
-
Data-driven atomistic modelling of hybrid halide perovskite passivation
A continual fine-tuning protocol for machine-learned interatomic potentials enables large-scale simulation of amino-silane passivation at hybrid perovskite surfaces, revealing coverage-dependent lattice disruption.
-
Scalable Prediction of Complex Surface Reconstructions under Operating Conditions via Harmony-Search-Based Global Optimization
HASGO combines harmony search with universal MLIPs and multi-head replay fine-tuning to locate operando surface reconstructions, demonstrated by identifying the square-pyramidal O5 subsurface motif on Ag(100) during e...
-
Comparing the latent features of universal machine-learning interatomic potentials
Different uMLIPs encode chemical space in distinct ways, with high cross-model feature reconstruction errors, and fine-tuning preserves strong pre-training bias in the latent features.
-
Comparing fine-tuning strategies of MACE machine learning force field for modeling Li-ion diffusion in LiF for batteries
MACE-MPA-0 predicts Li diffusion Ea of 0.22 eV in LiF, fine-tuned version with 300 points gives 0.20 eV, close to DeePMD reference of 0.24 eV, using far less training data.
-
SLUSCHI-UP: A Web Infrastructure for SLUSCHI Melting-Temperature Calculations Using Universal Machine-Learning Interatomic Potentials
SLUSCHI-UP deploys the SLUSCHI melting-temperature workflow as a web service backed by selectable universal ML interatomic potentials with reported validation errors on benchmark sets.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.