Machine Learning for Accurate Force Calculations in Molecular Dynamics Simulations

Abstract

The computationally expensive nature of ab initio molecular dynamics simulations severely limits its ability to simulate large system sizes and long time scales, both of which are necessary to imitate experimental conditions. In this work, we explore an approach to make use of the data obtained using the quantum mechanical density functional theory (DFT) on small systems and use deep learning to subsequently simulate large systems by taking liquid argon as a test case. A suitable vector representation was chosen to represent the surrounding environment of each Ar atom, and a DNetFF machine learning model where, the neural network was trained to predict the difference in resultant forces obtained by DFT and classical force fields was introduced. Molecular dynamics simulations were then performed using forces from the neural network for various system sizes and time scales depending on the properties we calculated. A comparison of properties obtained from the classical force field and the neural network model was presented alongside available experimental data to validate the proposed method. File list (2) download file view on ChemRxiv argon_submitted_may2.pdf (7.97 MiB) download file view on ChemRxiv argon_si.pdf (847.95 KiB)

A review on machine learning algorithms for the ionic liquid chemical space

“…and an accurate high level method (DFT, post-HF etc.). 168,[239][240][241][242] Just as important as the simulation itself is the nal step shown in Fig. 10, i.e.…”

MEMES: Machine learning framework for Enhanced MolEcular Screening

“…Recently new methods that use modern deep/reinforcement learning have been proposed to tackle problems in molecular sciences such as physical property prediction, 8,9 drug design tasks, 10 protein structure prediction, 11–13 molecular simulations, 14–16 and de novo molecule generation. 17 Most of the deep learning models that tackle the problem of molecular generation are based on variational autoencoders, 18–21 Generative Adversarial Networks 22–24 and Reinforcement Learning.…”

DeepPocket: Ligand Binding Site Detection and Segmentation using 3D Convolutional Neural Networks

“…In recent years, with the advancement in computer technology and increase in practicable data, the use of data-intensive techniques like machine learning and deep learning has burgeoned in numerous domains. [27][28][29][30][31] Consequently, this has also influenced the field of bio and cheminformatics greatly by providing solutions to a multitude of problems including binding site prediction. [32][33][34][35][36] The improvement in performance with an increase in accuracy of binding site detection forms substantial evidence to the increasing adoption of ML methods for this problem.…”