A Lightweight Universal Machine-Learning Interatomic Potential via Knowledge Distillation for Scalable Atomistic Simulations

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Oh, Sangmin, You, Jinmu, Kim, Jaesun, Lee, Jiho, An, Hyungmin, Han, Seungwu, Kang, Youngho
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915933844406272
author Oh, Sangmin
You, Jinmu
Kim, Jaesun
Lee, Jiho
An, Hyungmin
Han, Seungwu
Kang, Youngho
author_facet Oh, Sangmin
You, Jinmu
Kim, Jaesun
Lee, Jiho
An, Hyungmin
Han, Seungwu
Kang, Youngho
contents We introduce a lightweight universal machine-learning interatomic potential (uMLIP), SevenNet-Nano, based on the graph neural network architecture SevenNet and enabled by a knowledge-distillation framework. The model inherits the broad generalization capability of a large multi-task foundation model, SevenNet-Omni, trained on diverse materials datasets across chemical, configurational, and computational spaces. By learning chemical representations from high-quality inference data generated by the teacher model within a unified computational framework, SevenNet-Nano achieves high accuracy and strong transferability despite its compact architecture. The model also accurately captures a wide range of interatomic interactions, enabling reliable simulations under both equilibrium and extreme conditions, including plasma etching of SiO$_2$. Comprehensive benchmarks on static and dynamical properties--such as Li-ion diffusion and liquid densities--demonstrate its broad applicability with minimal fine-tuning. Importantly, SevenNet-Nano significantly reduces computational cost, achieving over an order-of-magnitude speedup and enabling large-scale atomistic simulations involving thousands of atoms.
format Preprint
id arxiv_https___arxiv_org_abs_2604_10887
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle A Lightweight Universal Machine-Learning Interatomic Potential via Knowledge Distillation for Scalable Atomistic Simulations
Oh, Sangmin
You, Jinmu
Kim, Jaesun
Lee, Jiho
An, Hyungmin
Han, Seungwu
Kang, Youngho
Materials Science
We introduce a lightweight universal machine-learning interatomic potential (uMLIP), SevenNet-Nano, based on the graph neural network architecture SevenNet and enabled by a knowledge-distillation framework. The model inherits the broad generalization capability of a large multi-task foundation model, SevenNet-Omni, trained on diverse materials datasets across chemical, configurational, and computational spaces. By learning chemical representations from high-quality inference data generated by the teacher model within a unified computational framework, SevenNet-Nano achieves high accuracy and strong transferability despite its compact architecture. The model also accurately captures a wide range of interatomic interactions, enabling reliable simulations under both equilibrium and extreme conditions, including plasma etching of SiO$_2$. Comprehensive benchmarks on static and dynamical properties--such as Li-ion diffusion and liquid densities--demonstrate its broad applicability with minimal fine-tuning. Importantly, SevenNet-Nano significantly reduces computational cost, achieving over an order-of-magnitude speedup and enabling large-scale atomistic simulations involving thousands of atoms.
title A Lightweight Universal Machine-Learning Interatomic Potential via Knowledge Distillation for Scalable Atomistic Simulations
topic Materials Science
url https://arxiv.org/abs/2604.10887