Abstract:The interaction of condensed phase systems with external electric fields is crucial in myriad processes in nature and technology ranging from the field-directed motion of cells (galvanotaxis), to energy storage and conversion systems including supercapacitors, batteries and solar cells. Molecular simulation in the presence of electric fields would give important atomistic insight into these processes but applications of the most accurate methods such as ab-initio molecular dynamics are limited in scope by their computational expense. Here we introduce Perturbed Neural Network Potential Molecular Dynamics (PNNP MD) to push back the accessible time and length scales of such simulations at virtually no loss in accuracy. The total forces on the atoms are expressed in terms of the unperturbed potential energy surface represented by a standard neural network potential and a field-induced perturbation obtained from a series expansion of the field interaction truncated at first order. The latter is represented in terms of an equivariant graph neural network, trained on the atomic polar tensor. PNNP MD is shown to give excellent results for the dielectric relaxation dynamics, the dielectric constant and the field-dependent IR spectrum of liquid water when compared to ab-initio molecular dynamics or experiment, up to surprisingly high field strengths of about 0.2 V/A. This is remarkable because, in contrast to most previous approaches, the two neural networks on which PNNL MD is based are exclusively trained on zero-field molecular configurations demonstrating that the networks not only interpolate but also reliably extrapolate the field response. PNNP MD is based on rigorous theory yet it is simple, general, modular, and systematically improvable allowing us to obtain atomistic insight into the interaction of a wide range of condensed phase systems with external electric fields.

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

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

Improving the accuracy of the neuroevolution machine learning potential for multi-component systems

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

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

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

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

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

Million-atom heat transport simulations of polycrystalline graphene approaching first-principles accuracy enabled by neuroevolution potential on desktop GPUs

Thermodynamics of Water and Ice from a Fast and Scalable First-Principles Neuroevolution Potential

Neuroevolution of Physics-Informed Neural Nets: Benchmark Problems and Comparative Results

Exploring mechanical and thermal properties of high-entropy ceramics via general machine learning potentials

Accurate nuclear quantum statistics on machine-learned classical effective potentials

Molecular Dynamics with Neural-Network Potentials

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

Molecular dynamics simulation with finite electric fields using Perturbed Neural Network Potentials

A Fourth-Generation High-Dimensional Neural Network Potential with Accurate Electrostatics Including Non-local Charge Transfer

Accurate and efficient molecular dynamics based on machine learning and non von Neumann architecture

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

Accurate prediction of heat conductivity of water by a neuroevolution potential