scholarly journals Artificial neural network correction for density-functional tight-binding molecular dynamics simulations

2019 ◽  
Vol 9 (3) ◽  
pp. 867-873 ◽  
Author(s):  
Junmian Zhu ◽  
Van Quan Vuong ◽  
Bobby G. Sumpter ◽  
Stephan Irle
Keyword(s):  
Neural Network ◽  
Molecular Dynamics ◽  
Artificial Neural Network ◽  
Molecular Dynamics Simulations ◽  
Density Functional ◽  
Tight Binding ◽  
Artificial Neural ◽  
Dynamics Simulations ◽  
Image Position

Abstract

scholarly journals Multiscale QM/MM Molecular Dynamics Simulations of the Trimeric Major Light-Harvesting Complex II

2021 ◽  
Author(s):  
Sayan Maity ◽  
Vangelis Daskalakis ◽  
Marcus Elstner ◽  
Ulrich Kleinekathöfer
Keyword(s):  
Molecular Dynamics ◽  
Molecular Dynamics Simulations ◽  
Chlorophyll A ◽  
Density Functional ◽  
Light Harvesting ◽  
Higher Plants ◽  
Tight Binding ◽  
Spectral Densities ◽  
Light Harvesting Complex ◽  
Dynamics Simulations

Photosynthetic processes are driven by sunlight. Too little of it and the photosynthetic machinery cannot produce the reductive power to drive the anabolic pathways. Too much sunlight and the machinery can get damaged. In higher plants, the major Light Harvesting Complex (LHCII) efficiently absorbs the light energy, but can also dissipate it when in excess (quenching). In order to study the dynamics related to the quenching process but also the exciton dynamics in general, one needs to accurately determine the so-called spectral density which describes the coupling between the relevant pigment modes and the environmental degrees of freedom. To this end, Born–Oppenheimer molecular dynamics simulations in a quantum mechanics/molecular mechanics (QM/MM) fashion utilizing the density functional based tight binding (DFTB) method have been performed for the ground state dynamics. Subsequently, the time-dependent extension of the long-range-corrected DFTB scheme has been employed for the excited state calculations of the individual chlorophyll-a molecules in the LHCII complex. The analysis of this data resulted in spectral densities showing an astonishing agreement with the experimental counterpart in this rather large system. This consistency with an experimental observable also supports the accuracy, robustness, and reliability of the present multi-scale scheme. In addition, the resulting spectral densities and site energies were used to determine the exciton transfer rate within a special pigment pair consisting of a chlorophyll-a and a carotenoid molecule which is assumed to play a role in the balance between the light harvesting and quenching modes.

scholarly journals Artificial Neural Network Potentials for Mechanics and Fracture Dynamics of Materials

2022 ◽  
Author(s):  
Gang Seob Jung ◽  
Hoon Joo Myung ◽  
Stephan Irle
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Crack Propagation ◽  
Density Functional ◽  
Tight Binding ◽  
Training Data ◽  
Data Generation ◽  
Deformation And Failure ◽  
Deformation And Fracture ◽  
Artificial Neural

Abstract Atomistic understanding of mechanics and failure of materials is the key for engineering and applications. Modeling accurately brittle failure with crack propagation in covalent crystals requires a quantum mechanics-based description of individual bond-breaking events for large system sizes. Machine Learned (ML) potentials have emerged to overcome the traditional, physics-based modeling tradeoff between accuracy and accessible time and length scales. Previous studies have shown successful applications of ML potentials for describing the structure and dynamics of molecular systems and amorphous or liquid phases of materials. However, their application to deformation and failure processes in materials is yet uncommon. In this study, we discuss apparent limitations of ML potentials to describe deformation and fracture under loadings and propose a way to generate and select training data for their employment in simulations of deformation and fracture of crystals. We applied the proposed approach to 2D crystal graphene, utilizing the density-functional tight-binding (DFTB) method for more efficient and extensive data generation in place of density functional theory (DFT). Then, we explore how the data selection affects the accuracy of the developed artificial neural network potential (NNP), indicating that only the errors in total energies and atomic forces are insufficient to judge the NNP’s reliability. Therefore, we evaluate and select NNPs based on their performance in describing physical properties, e.g., stress-strain curves and geometric deformation. In sharp contrast to popular reactive bond order potentials, our optimized NNP predicts straight crack propagation in graphene along both armchair and zigzag lattice directions, as well as higher fracture toughness of zigzag edge direction. Our study provides significant insight into crack propagation mechanisms at atomic scales and highlights strategies for NNP developments of broader materials.

Sign in / Sign up

Export Citation Format

Share Document