Title: Efficient and interpretable graph network representation for angle-dependent properties applied to optical spectroscopy

Journal Article · · npj Computational Materials
ORCiD logo; ; ; ; ORCiD logo; ; ; ; ORCiD logo

Abstract Graph neural networks are attractive for learning properties of atomic structures thanks to their intuitive graph encoding of atoms and bonds. However, conventional encoding does not include angular information, which is critical for describing atomic arrangements in disordered systems. In this work, we extend the recently proposed ALIGNN (Atomistic Line Graph Neural Network) encoding, which incorporates bond angles, to also include dihedral angles (ALIGNN-d). This simple extension leads to a memory-efficient graph representation that captures the complete geometry of atomic structures. ALIGNN-d is applied to predict the infrared optical response of dynamically disordered Cu(II) aqua complexes, leveraging the intrinsic interpretability to elucidate the relative contributions of individual structural components. Bond and dihedral angles are found to be critical contributors to the fine structure of the absorption response, with distortions that represent transitions between more common geometries exhibiting the strongest absorption intensity. Future directions for further development of ALIGNN-d are discussed.

Research Organization:
Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States)
Sponsoring Organization:
USDOE; USDOE Laboratory Directed Research and Development (LDRD) Program; USDOE National Nuclear Security Administration (NNSA)
Grant/Contract Number:
AC52-07NA27344
OSTI ID:
1876249
Report Number(s):
LLNL-JRNL-826919; 151; PII: 841
Journal Information:
npj Computational Materials, Journal Name: npj Computational Materials Journal Issue: 1 Vol. 8; ISSN 2057-3960
Publisher:
Nature Publishing GroupCopyright Statement
Country of Publication:
United Kingdom
Language:
English

References (34)

Practical conversion from torsion space to Cartesian space forin silico protein synthesis journal January 2005
Time-dependent density functional theory within the Tamm–Dancoff approximation journal December 1999
Quantitative structural determination of active sites from in situ and operando XANES spectra: From standard ab initio simulations to chemometric and machine learning approaches journal October 2019
Machine learning models for the prediction of energy, forces, and stresses for Platinum journal March 2020
Sigmoid-weighted linear units for neural network function approximation in reinforcement learning journal November 2018
Machine learning approaches to XANES spectra for quantitative 3D structural determination: The case of CO2 adsorption on CPO-27-Ni MOF journal October 2020
Graph Networks as a Universal Machine Learning Framework for Molecules and Crystals journal April 2019
Convolutional Embedding of Attributed Molecular Graphs for Physical Property Prediction journal July 2017
Analyzing Learned Molecular Representations for Property Prediction journal July 2019
ChIMES: A Force Matched Potential with Explicit Three-Body Interactions for Molten Carbon journal November 2017
Understanding Trends in Molecular Bond Angles journal January 2020
Integrating Ab Initio Simulations and X-ray Photoelectron Spectroscopy: Toward A Realistic Description of Oxidized Solid/Liquid Interfaces journal December 2017
“Inverting” X-ray Absorption Spectra of Catalysts by Machine Learning in Search for Activity Descriptors journal September 2019
Minimal Distortion Pathways in Polyhedral Rearrangements journal February 2004
The Solvation Structure of Mg Ions in Dichloro Complex Solutions from First-Principles Molecular Dynamics and Simulated X-ray Absorption Spectra journal October 2014
Atomistic Line Graph Neural Network for improved materials property predictions journal November 2021
Efficient and universal characterization of atomic structures through a topological graph order parameter journal March 2022
AiiDA 1.0, a scalable computational infrastructure for automated reproducible workflows and data provenance journal September 2020
Origins of optical absorption characteristics of Cu 2+ complexes in aqueous solutions journal January 2015
Towards an assessment of the accuracy of density functional theory for first principles simulations of water journal January 2004
SchNet – A deep learning architecture for molecules and materials journal June 2018
Representing local atomic environment using descriptors based on local correlations journal December 2018
NWChem: Past, present, and future journal May 2020
QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials journal September 2009
The atomic simulation environment—a Python library for working with atoms journal June 2017
Soft self-consistent pseudopotentials in a generalized eigenvalue formalism journal April 1990
Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties journal April 2018
Density-Functional Theory for Time-Dependent Systems journal March 1984
Generalized Gradient Approximation Made Simple journal October 1996
Generalized Neural-Network Representation of High-Dimensional Potential-Energy Surfaces journal April 2007
Developing an improved crystal graph convolutional neural network framework for accelerated materials discovery journal June 2020
Electronic structure of aqueous solutions: Bridging the gap between theory and experiments journal June 2017
The structure of interfacial water on gold electrodes studied by x-ray absorption spectroscopy journal October 2014
First Solvation Shell of the Cu(II) Aqua Ion: Evidence for Fivefold Coordination journal February 2001

Similar Records

Related Subjects

36 MATERIALS SCIENCE
computational methods
coordination chemistry