Abstract:Molecular dynamics (MD) simulations play an important role in understanding and engineering heat transport properties of complex materials. An essential requirement for reliably predicting heat transport properties is the use of accurate and efficient interatomic potentials. Recently, machine-learned potentials (MLPs) have shown great promise in providing the required accuracy for a broad range of materials. In this mini review and tutorial, we delve into the fundamentals of heat transport, explore pertinent MD simulation methods, and survey the applications of MLPs in MD simulations of heat transport. Furthermore, we provide a step-by-step tutorial on developing MLPs for highly efficient and predictive heat transport simulations, utilizing the neuroevolution potentials (NEPs) as implemented in the GPUMD package. Our aim with this mini review and tutorial is to empower researchers with valuable insights into cutting-edge methodologies that can significantly enhance the accuracy and efficiency of MD simulations for heat transport studies.

What problem does this paper attempt to address?

This paper focuses on using Machine-Learned Potentials (MLPs) for Molecular Dynamics (MD) simulations to study thermal transport. Traditionally, MD simulations are limited in accuracy and efficiency when predicting thermal transport properties of materials. MLPs, particularly Neuroevolution Potentials (NEPs), provide a high-precision and efficient method to address this challenge. The paper first introduces the importance of thermal transport and its role in different computational methods, highlighting the advantages of MD simulations in dealing with phonon-phonon scattering and considering scattering due to anharmonicity and defects. It points out that the main hurdle of MD simulations lies in the need for accurate potential energy surface models, while machine-learning potentials like NEPs offer the ability to simulate various materials with higher accuracy. The paper further discusses the GPU-accelerated MD software package GPUMD, which supports efficient NEP implementation and can handle large complex systems. Additionally, it provides a tutorial on constructing and applying NEPs for thermal transport simulations. By discussing concepts such as thermal conductivity, thermal conduction, heat flux, and thermal current, as well as different MD methods (such as equilibrium molecular dynamics, non-equilibrium molecular dynamics, and homogenous non-equilibrium molecular dynamics), the authors provide readers with foundational knowledge of thermal transport simulations. The paper specifically mentions the excellent performance of NEP methods in simulating thermal conductivity, especially in handling strongly anharmonic or highly disordered systems. Finally, the paper reviews research on thermal transport MD simulations using MLPs and provides several successful application examples, including simulations of liquid water, amorphous silicon dioxide, and amorphous silicon. Although MLPs may underestimate the thermal conductivity of crystalline materials in certain cases, they perform well in describing the temperature-dependent thermal conductivity of disordered materials, and correction methods have been proposed. In conclusion, this paper aims to introduce and promote the latest advancements and practical techniques in thermal transport research using MLPs and NEPs, in order to enhance the accuracy and efficiency of MD simulations.

Molecular dynamics simulations of heat transport using machine-learned potentials: A mini review and tutorial on GPUMD with neuroevolution potentials

Molecular dynamics simulations of heat transport using machine-learned potentials: A mini-review and tutorial on GPUMD with neuroevolution potentials

Neuroevolution machine learning potentials: Combining high accuracy and low cost in atomistic simulations and application to heat transport

GPUMD: A package for constructing accurate machine-learned potentials and performing highly efficient atomistic simulations

Highly efficient path-integral molecular dynamics simulations with GPUMD using neuroevolution potentials: Case studies on thermal properties of materials

Molecular Dynamics with Neural-Network Potentials

Combining the D3 dispersion correction with the neuroevolution machine-learned potential

NNP/MM: Accelerating Molecular Dynamics Simulations with Machine Learning Potentials and Molecular Mechanics

NNP/MM: Accelerating molecular dynamics simulations with machine learning potentials and molecular mechanic

Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics

A Generator of Neural Network Potential for Molecular Dynamics: Constructing Robust and Accurate Potentials with Active Learning for Nanosecond-scale Simulations

Revisit Many-body Interaction Heat Current and Thermal Conductivity Calculation in Moment Tensor Potential/LAMMPS Interface

Heat flux for semi-local machine-learning potentials

Machine Learning for Parameter Auto-tuning in Molecular Dynamics Simulations: Efficient Dynamics of Ions near Polarizable Nanoparticles

Combining linear-scaling quantum transport and machine-learning molecular dynamics to study thermal and electronic transports in complex materials

Machine learning heralding a new development phase in molecular dynamics simulations

Global Neural Network Potential with Explicit Many-Body Functions for Improved Descriptions of Complex Potential Energy Surface.

Deep Potentials for Materials Science

Tutorial: How to Train a Neural Network Potential

Learning Interatomic Potentials at Multiple Scales