FPBench: Application-Oriented Error Decomposition for Foundation Potentials

RESUMO

Foundation potentials (FPs) have emerged as a new basis for atomistic modeling. While their evaluation using average energy and force errors often indicates near-DFT accuracy, their performance in practical computational studies remains inconsistent. Here, we present FPBench, an application-oriented benchmark that evaluates FPs on representative computational tasks through error decomposition. Multiple state-of-the-art FPs are assessed on three fundamental tasks: force prediction for atomistic simulations, energy ranking for substitutional and vacancy orderings, and ion/vacancy migration. Benchmarking these FPs shows that average force and energy errors often fail to predict task performance because they do not directly resolve the physically consequential quantities and configurations that govern task outcomes. FPBench introduces error-decomposition metrics that resolve force errors among highly accurate, large-error, and far-from-equilibrium atoms; relative-energy errors among competing orderings, phases, and compositions; and endpoint and along-path errors in ion migration. By identifying where FP errors arise, this error decomposition provides targeted guidance for FP development. FPBench also provides an open benchmark, evaluation code, and a public leaderboard for rigorous FP assessment.