Pushing the Limit of Molecular Dynamics with Ab Initio Accuracy to 100 Million Atoms with Machine Learning

  title={Pushing the Limit of Molecular Dynamics with Ab Initio Accuracy to 100 Million Atoms with Machine Learning},
  author={Weile Jia and Han Wang and Mohan Chen and Denghui Lu and Jiduan Liu and Lin Lin and Roberto Car and E Weinan and Linfeng Zhang},
  journal={SC20: International Conference for High Performance Computing, Networking, Storage and Analysis},
  • Weile Jia, Han Wang, Linfeng Zhang
  • Published 1 May 2020
  • Computer Science
  • SC20: International Conference for High Performance Computing, Networking, Storage and Analysis
For 35 years, ab initio molecular dynamics (AIMD) has been the method of choice for modeling complex atomistic phenomena from first principles. However, most AIMD applications are limited by computational cost to systems with thousands of atoms at most. We report that a machine learningbased simulation protocol (Deep Potential Molecular Dynamics), while retaining ab initio accuracy, can simulate more than 1 nanosecond-long trajectory of over 100 million atoms per day, using a highly optimized… 

Ab Initio Machine Learning in Chemical Compound Space

While state-of-the-art approximations to quantum problems impose severe computational bottlenecks, recent QML based developments indicate the possibility of substantial acceleration without sacrificing the predictive power of quantum mechanics.

Uncertainty estimation by committee models for molecular dynamics and thermodynamic averages

An on-the-fly reweighing scheme is introduced that makes it possible to estimate the uncertainty in the thermodynamic averages extracted from long trajectories, incorporating both the error in the single-point predictions of the properties and the error due to the distortion of the sampling probability.

Deep-learning interatomic potential for irradiation damage simulations in MoS2 with ab initial accuracy

Potentials that could accurately describe the irradiation damage processes are highly desired to figure out the atomic-level response of various newly-discovered materials under irradiation

Integrating Machine Learning with Physics-Based Modeling

This article provides a simple and intuitive explanation for the fundamental reasons behind the success of modern machine learning, as well as an introduction to the concurrent machine learning framework needed for integrating machine learning with physics-based modeling.

Machine Learning and Computational Mathematics

This article describes some of the most important progress that has been made on computational mathematics issues and aims to put things into a perspective that will help to integrate machine learning with computational mathematics.

Retrospective on a decade of machine learning for chemical discovery

This paper considers selected studies of electronic structure, interatomic potentials, and chemical compound space in chronological order to highlight specific achievements of machine learning models in the field of computational chemistry.

HPC AI500: Representative, Repeatable and Simple HPC AI Benchmarking

This paper presents a representative, repeatable and simple HPC AI benchmarking methodology, and presents a new metric named Valid FLOPS, emphasizing both throughput performance and a target quality.

Quantum Machine Learning in Chemical Compound Space

The case is made for quantum machine learning: An inductive molecular modeling approach which can be applied to quantum chemistry problems.

Thermal transport by electrons and ions in warm dense aluminum: A combined density functional theory and deep potential study

We propose an efficient scheme, which combines density functional theory (DFT) with deep potentials (DP), to systematically study the convergence issues of the computed electronic thermal

The Arab world prepares the exascale workforce

The stories in this article grew up around a university-operated supercomputer, but they largely can be replicated with much less investment because the main challenges today are not in coordinating tens of thousands of nodes across a low-latency, high-bandwidth network, Rather, the challenges lie in extracting performance from within increasingly heterogeneous nodes.



Amp: A modular approach to machine learning in atomistic simulations

Deep Potential Molecular Dynamics: a scalable model with the accuracy of quantum mechanics

We introduce a scheme for molecular simulations, the deep potential molecular dynamics (DPMD) method, based on a many-body potential and interatomic forces generated by a carefully crafted deep

Reinforced dynamics for enhanced sampling in large atomic and molecular systems. I. Basic Methodology

A new approach for efficiently exploring the configuration space and computing the free energy of large atomic and molecular systems is proposed, motivated by an analogy with reinforcement learning, which allows for an efficient exploration of the configurationspace by adding an adaptively computed biasing potential to the original dynamics.

Machine learning of accurate energy-conserving molecular force fields

The GDML approach enables quantitative molecular dynamics simulations for molecules at a fraction of cost of explicit AIMD calculations, thereby allowing the construction of efficient force fields with the accuracy and transferability of high-level ab initio methods.

Isotope effects in liquid water via deep potential molecular dynamics

A comprehensive microscopic understanding of ambient liquid water is a major challenge for ab initio simulations as it simultaneously requires an accurate quantum mechanical description of the

End-to-end Symmetry Preserving Inter-atomic Potential Energy Model for Finite and Extended Systems

Deep Potential - Smooth Edition (DeepPot-SE), an end-to-end machine learning-based PES model, which is able to efficiently represent the PES for a wide variety of systems with the accuracy of ab initio quantum mechanics models is developed.

Fast parallel algorithms for short-range molecular dynamics

Comparing the results to the fastest reported vectorized Cray Y-MP and C90 algorithm shows that the current generation of parallel machines is competitive with conventional vector supercomputers even for small problems.

CHARMM: The biomolecular simulation program

An overview of the CHARMM program as it exists today is provided with an emphasis on developments since the publication of the original CHARMM article in 1983.