arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2604.01017cond-mat.mtrl-sciphysics.comp-ph
收藏

机器学习互原子势的参数高效微调用于声子和热性质

Parameter-Efficient Fine-Tuning of Machine-Learning Interatomic Potentials for Phonon and Thermal Properties

Federal Institute for Materials Research and Testing(联邦材料研究与测试研究所);Institute of Condensed Matter Theory and Solid-State Optics, Friedrich Schiller University Jena(耶拿弗里德里希·席勒大学凝聚态理论与固态光学研究所)

Jonas Grandel, Philipp Benner, Janine George

AI总结:

本文研究了不同微调策略对声子带结构、热性质及虚声子模式势能面预测的影响,提出Equitrain框架实现LoRA微调,展示微调模型在53种材料系统中均优于预训练模型和从头训练模型。

摘要:

机器学习互原子势广泛用于原子模拟中作为计算高效的密度泛函理论替代品,使材料系统的大规模长时间建模成为可能。我们研究了不同微调策略对谐振声子带结构、热性质及虚声子模式势能面预测的影响。我们通过最少的额外数据(仅需10个额外训练结构)实现了显著的精度提升。除了现有方法外,我们引入了Equitrain微调框架,实现基于LoRA的适应。在53种材料系统中,我们证明微调模型在各方面均优于预训练模型和从头训练模型。Equitrain实现了最佳整体性能,我们的结果表明微调能够实现准确的声子预测。

英文摘要:

Machine-learning interatomic potentials are widely used as computationally efficient surrogates for density functional theory in atomistic simulations, enabling large-scale, long-time modeling of materials systems. We investigate how different fine-tuning strategies influence the prediction of harmonic phonon band structures, thermal properties, and the potential energy surface along imaginary phonon modes. We achieve substantial accuracy improvements with minimal additional data, with as few as 10 additional training structures already yielding significant gains. In addition to existing approaches, we introduce Equitrain, a finetuning framework that implements LoRA-based adaptation. Across 53 materials systems, we show that fine-tuned models consistently outperform both the underlying pretrained model and models trained from scratch. Equitrain achieves the best overall performance, and our results demonstrate that fine-tuning enables accurate phonon predictions.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑