Universal machine-learning interatomic potentials(uMLIPs)are emerging as foundation models for atomistic simulation,offering near-ab initio accuracy at far lower cost.Their safe,broad deployment is limited by the abse...Universal machine-learning interatomic potentials(uMLIPs)are emerging as foundation models for atomistic simulation,offering near-ab initio accuracy at far lower cost.Their safe,broad deployment is limited by the absence of reliable,general uncertainty estimates.We present a unified,scalable uncertainty metric,U,built from a heterogeneous ensemble that reuses existing pretrained MLIPs.Across diverse chemistries and structures,U strongly tracks true prediction errors and robustly ranks configuration-level risk.Using U,we perform uncertainty-aware distillation to train system-specific potentials with far fewer labels:for tungsten,we match full density-functional-theory(DFT)training using 4%of the DFT data;for MoNbTaW,a dataset distilled by U supports high-accuracy potential training.By filtering numerical label noise,the distilled models can in some cases exceed the accuracy of the MLIPs trained on DFT data.This framework provides a practical reliability monitor and guides data selection and fine-tuning,enabling cost-efficient,accurate,and safer deployment of foundation models.展开更多
基金sponsored by Nederlandse Organisatie voor WetenschappelijkOnderzoek (The Netherlands Organization for Scientific Research, NWO) domain Science for the use of supercomputer facilities. The authors also acknowledge the use of DelftBlue supercomputer, provided by Delft High Performance Computing Center (https://www.tudelft.nl/dhpc).
文摘Universal machine-learning interatomic potentials(uMLIPs)are emerging as foundation models for atomistic simulation,offering near-ab initio accuracy at far lower cost.Their safe,broad deployment is limited by the absence of reliable,general uncertainty estimates.We present a unified,scalable uncertainty metric,U,built from a heterogeneous ensemble that reuses existing pretrained MLIPs.Across diverse chemistries and structures,U strongly tracks true prediction errors and robustly ranks configuration-level risk.Using U,we perform uncertainty-aware distillation to train system-specific potentials with far fewer labels:for tungsten,we match full density-functional-theory(DFT)training using 4%of the DFT data;for MoNbTaW,a dataset distilled by U supports high-accuracy potential training.By filtering numerical label noise,the distilled models can in some cases exceed the accuracy of the MLIPs trained on DFT data.This framework provides a practical reliability monitor and guides data selection and fine-tuning,enabling cost-efficient,accurate,and safer deployment of foundation models.
