Abstract:Neural network potentials (NNPs) enable large-scale molecular dynamics (MD) simulations of systems containing >10,000 atoms with the accuracy comparable to ab initio methods and play a crucial role in material studies. Although NNPs are valuable for short-duration MD simulations, maintaining the stability of long-duration MD simulations remains challenging due to the uncharted regions of the potential energy surface (PES). Currently, there is no effective methodology to address this issue. To overcome this challenge, we developed an automatic generator of robust and accurate NNPs based on an active learning (AL) framework. This generator provides a fully integrated solution encompassing initial dataset creation, NNP training, evaluation, sampling of additional structures, screening, and labeling. Crucially, our approach uses a sampling strategy that focuses on generating unstable structures with short interatomic distances, combined with a screening strategy that efficiently samples these configurations based on interatomic distances and structural features. This approach greatly enhances the MD simulation stability, enabling nanosecond-scale simulations. We evaluated the performance of our NNP generator in terms of its MD simulation stability and physical properties by applying it to liquid propylene glycol (PG) and polyethylene glycol (PEG). The generated NNPs enable stable MD simulations of systems with >10,000 atoms for 20 ns. The predicted physical properties, such as the density and self-diffusion coefficient, show excellent agreement with the experimental values. This work represents a remarkable advance in the generation of robust and accurate NNPs for organic materials, paving the way for long-duration MD simulations of complex systems.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **How to generate robust and accurate neural network potentials (NNPs) to achieve the stability of nanosecond - level molecular dynamics (MD) simulations**. Specifically, although existing NNPs perform well in short - time MD simulations, there are still challenges in long - time, large - scale MD simulations. The main reason is that unexplored potential energy surface (PES) regions may lead to simulation instability. At present, there is still a lack of effective solutions to this problem. To solve this problem, the author has developed an automatic NNP generator based on the active learning (AL) framework. This generator improves the stability and accuracy of NNPs in the following ways: 1. **Initial data set creation**: Use AIMD simulations and random displacement sampling to generate initial structures. 2. **NNP training and evaluation**: Train multiple models to estimate uncertainty and select the best model for subsequent steps. 3. **Sampling**: Explore the configuration space through NNP - MD simulations to generate candidate structures. 4. **Screening**: Adopt a two - step screening method based on model uncertainty and structural features to ensure that the selected structures can effectively improve the current NNP. 5. **Labeling**: Perform DFT calculations on the screened structures and update the data set. Through these steps, the generator can generate robust and accurate NNPs suitable for complex systems, especially for long - time MD simulations of organic materials. The experimental results show that the generated NNPs can achieve stable simulations of up to 20 nanoseconds in systems with more than 10,000 atoms, and the predicted physical properties (such as density and self - diffusion coefficient) are highly consistent with the experimental values. In conclusion, this research provides a systematic method for generating high - quality NNPs, solves the stability problem in long - time MD simulations, and provides a powerful tool for the research of complex materials.

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

Neural Network Potential with Multi-Resolution Approach Enables Accurate Prediction of Reaction Free Energies in Solution

ReacNetGenerator: an Automatic Reaction Network Generator for Reactive Molecular Dynamic Simulations

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

ReacNetGenerator: an Automatic Reaction Network Generator for Reactive Molecular Dynamics Simulations.

Scalable Training of Neural Network Potentials for Complex Interfaces Through Data Augmentation

ReacNetGen: an Automatic Reaction Network Generator for Reactive Molecular Dynamic Simulations

A deep learning interatomic potential developed for atomistic simulation of carbon materials

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

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

A Scalable Graph Neural Network Method For Developing An Accurate Force Field Of Large Flexible Organic Molecules

GPU-accelerated artificial neural network potential for molecular dynamics simulation

Automated Construction of Neural Network Potential Energy Surface: The Enhanced Self-Organizing Incremental Neural Network Deep Potential Method

Application of First-Principles-Based Artificial Neural Network Potentials to Multiscale-Shock Dynamics Simulations on Solid Materials

Ab initio Accuracy Neural Network Potential for Drug-like Molecules

Neural Network Interatomic Potentials For Open Surface Nano-mechanics Applications

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

LAMMPS Implementation of Rapid Artificial Neural Network Derived Interatomic Potentials

Molecular Dynamics with Neural-Network Potentials

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

The Potential of Neural Network Potentials