Nonadiabatic Excited-State Dynamics with Machine Learning

We show that machine learning (ML) can be used to accurately reproduce nonadiabatic excited-state dynamics with decoherence-corrected fewest switches surface hopping in a 1-D model system. We propose to use ML to significantly reduce the simulation time of realistic, high-dimensional systems with good reproduction of observables obtained from reference simulations. Our approach is based on creating approximate ML potentials for each adiabatic state using a small number of training points. We investigate the feasibility of this approach by using adiabatic spin-boson Hamiltonian models of various dimensions as reference methods.

[1]  Hermann Stoll,et al.  Results obtained with the correlation energy density functionals of becke and Lee, Yang and Parr , 1989 .

[2]  Markus Meuwly,et al.  Toolkit for the Construction of Reproducing Kernel-Based Representations of Data: Application to Multidimensional Potential Energy Surfaces , 2017, J. Chem. Inf. Model..

[3]  Michael Gastegger,et al.  Machine learning molecular dynamics for the simulation of infrared spectra† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c7sc02267k , 2017, Chemical science.

[4]  Rampi Ramprasad,et al.  Learning scheme to predict atomic forces and accelerate materials simulations , 2015, 1505.02701.

[5]  M. Barbatti,et al.  Recent Advances and Perspectives on Nonadiabatic Mixed Quantum-Classical Dynamics. , 2018, Chemical reviews.

[6]  G. Granucci,et al.  Including quantum decoherence in surface hopping. , 2010, The Journal of chemical physics.

[7]  Jean-Philippe Blaudeau,et al.  Extension of Gaussian-2 (G2) theory to molecules containing third-row atoms K and Ca , 1995 .

[8]  J. Tully.  Mixed quantum–classical dynamics , 1998 .

[9]  John C. Butcher,et al.  A Modified Multistep Method for the Numerical Integration of Ordinary Differential Equations , 1965, JACM.

[10]  Joseph E. Subotnik,et al.  Can we derive Tully's surface-hopping algorithm from the semiclassical quantum Liouville equation? Almost, but only with decoherence. , 2013, The Journal of chemical physics.

[11]  E Weinan,et al.  Deep Potential Molecular Dynamics: a scalable model with the accuracy of quantum mechanics , 2017, Physical review letters.

[12]  Klaus-Robert Müller,et al.  Machine learning of accurate energy-conserving molecular force fields , 2016, Science Advances.

[13]  Hiroki Nakamura,et al.  The two‐state linear curve crossing problems revisited. III. Analytical approximations for Stokes constant and scattering matrix: Nonadiabatic tunneling case , 1993 .

[14]  Mark A. Ratner,et al.  6‐31G* basis set for third‐row atoms , 2001, J. Comput. Chem..

[15]  A. Leggett,et al.  Dynamics of the dissipative two-state system , 1987 .

[16]  J. Pople,et al.  Self-consistent molecular orbital methods. 21. Small split-valence basis sets for first-row elements , 2002 .

[17]  Joseph E. Subotnik,et al.  Communication: The correct interpretation of surface hopping trajectories: how to calculate electronic properties. , 2013, The Journal of chemical physics.

[18]  Mark S. Gordon,et al.  The isomers of silacyclopropane , 1980 .

[19]  J S Smith,et al.  ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost , 2016, Chemical science.

[20]  P. C. Hariharan,et al.  The influence of polarization functions on molecular orbital hydrogenation energies , 1973 .

[21]  Parr,et al.  Development of the Colle-Salvetti correlation-energy formula into a functional of the electron density. , 1988, Physical review. B, Condensed matter.

[22]  Joseph E. Subotnik,et al.  How to recover Marcus theory with fewest switches surface hopping: add just a touch of decoherence. , 2012, The Journal of chemical physics.

[23]  A. Becke.  A New Mixing of Hartree-Fock and Local Density-Functional Theories , 1993 .

[24]  O. Prezhdo,et al.  A Simple Solution to the Trivial Crossing Problem in Surface Hopping. , 2014, The journal of physical chemistry letters.

[25]  W. C. Swope,et al.  A computer simulation method for the calculation of equilibrium constants for the formation of physi , 1981 .

[26]  Klaus-Robert Müller,et al.  Assessment and Validation of Machine Learning Methods for Predicting Molecular Atomization Energies. , 2013, Journal of chemical theory and computation.

[27]  G. Granucci,et al.  Critical appraisal of the fewest switches algorithm for surface hopping. , 2007, The Journal of chemical physics.

[28]  Gábor Csányi,et al.  Gaussian approximation potentials: A brief tutorial introduction , 2015, 1502.01366.

[29]  J. Pople,et al.  Self‐Consistent Molecular‐Orbital Methods. IX. An Extended Gaussian‐Type Basis for Molecular‐Orbital Studies of Organic Molecules , 1971 .

[30]  Jörg Behler,et al.  Automatic selection of atomic fingerprints and reference configurations for machine-learning potentials. , 2018, The Journal of chemical physics.

[31]  Matthias Rupp,et al.  Big Data Meets Quantum Chemistry Approximations: The Δ-Machine Learning Approach. , 2015, Journal of chemical theory and computation.

[32]  I. Sobol,et al.  Construction and Comparison of High-Dimensional Sobol' Generators , 2011 .

[33]  U. Rothlisberger,et al.  Mixed Quantum Mechanical/Molecular Mechanical Molecular Dynamics Simulations of Biological Systems in Ground and Electronically Excited States. , 2015, Chemical reviews.

[34]  D. Ruppert.  The Elements of Statistical Learning: Data Mining, Inference, and Prediction , 2004 .

[35]  Zhenwei Li,et al.  Molecular dynamics with on-the-fly machine learning of quantum-mechanical forces. , 2015, Physical review letters.

[36]  E. Gross,et al.  Mixed quantum-classical dynamics on the exact time-dependent potential energy surface: a fresh look at non-adiabatic processes , 2013, 1307.0351.

[37]  M. Rupp,et al.  Machine Learning for Quantum Mechanical Properties of Atoms in Molecules , 2015, 1505.00350.

[38]  Mark S. Gordon,et al.  Self-consistent molecular-orbital methods. 22. Small split-valence basis sets for second-row elements , 1980 .

[39]  Shun-ichi Amari,et al.  Four Types of Learning Curves , 1992, Neural Computation.

[40]  J. Tully.  Molecular dynamics with electronic transitions , 1990 .

[41]  Li Li,et al.  Bypassing the Kohn-Sham equations with machine learning , 2016, Nature Communications.

[42]  D. Truhlar,et al.  Coherent switching with decay of mixing: an improved treatment of electronic coherence for non-Born-Oppenheimer trajectories. , 2004, The Journal of chemical physics.

[43]  J. Pople,et al.  Self—Consistent Molecular Orbital Methods. XII. Further Extensions of Gaussian—Type Basis Sets for Use in Molecular Orbital Studies of Organic Molecules , 1972 .

[44]  Matthias Rupp,et al.  Machine learning for quantum mechanics in a nutshell , 2015 .

[45]  Christoph Kreisbeck,et al.  Machine learning for quantum dynamics: deep learning of excitation energy transfer properties† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c7sc03542j , 2017, Chemical science.

[46]  A. Becke,et al.  Density-functional exchange-energy approximation with correct asymptotic behavior. , 1988, Physical review. A, General physics.

[47]  W. Thiel,et al.  Non-Hermitian surface hopping. , 2017, Physical review. E.

[48]  L. Curtiss,et al.  Compact contracted basis sets for third‐row atoms: Ga–Kr , 1990 .

[49]  D. Reichman,et al.  On the accuracy of surface hopping dynamics in condensed phase non-adiabatic problems. , 2016, The Journal of chemical physics.

[50]  Deping Hu,et al.  Inclusion of Machine Learning Kernel Ridge Regression Potential Energy Surfaces in On-the-Fly Nonadiabatic Molecular Dynamics Simulation. , 2018, The journal of physical chemistry letters.

[51]  W. Thiel,et al.  Adaptive time steps in trajectory surface hopping simulations. , 2016, The Journal of chemical physics.

[52]  Walter Thiel,et al.  Structure-based sampling and self-correcting machine learning for accurate calculations of potential energy surfaces and vibrational levels. , 2017, The Journal of chemical physics.

[53]  Alexander Denzel,et al.  Gaussian Process Regression for Geometry Optimization , 2018, 2009.05803.

[54]  A. Becke.  Density-functional thermochemistry. III. The role of exact exchange , 1993 .

[55]  A. Akimov,et al.  Large-Scale Computations in Chemistry: A Bird's Eye View of a Vibrant Field. , 2015, Chemical reviews.

[56]  M. Frisch,et al.  Ab Initio Calculation of Vibrational Absorption and Circular Dichroism Spectra Using Density Functional Force Fields , 1994 .

[57]  Aaron Kelly,et al.  Efficient and accurate surface hopping for long time nonadiabatic quantum dynamics. , 2013, The Journal of chemical physics.

[58]  Hans Lischka,et al.  Newton‐X: a surface‐hopping program for nonadiabatic molecular dynamics , 2014 .

[59]  S. Habershon,et al.  Direct Quantum Dynamics Using Grid-Based Wave Function Propagation and Machine-Learned Potential Energy Surfaces. , 2017, Journal of chemical theory and computation.

[60]  Joseph E Subotnik,et al.  Understanding the Surface Hopping View of Electronic Transitions and Decoherence. , 2016, Annual review of physical chemistry.

[61]  Massimo Olivucci,et al.  Theory and Simulation of the Ultrafast Double-Bond Isomerization of Biological Chromophores. , 2017, Chemical reviews.

[62]  O. A. von Lilienfeld,et al.  Electronic spectra from TDDFT and machine learning in chemical space. , 2015, The Journal of chemical physics.

[63]  Hans Lischka,et al.  Surface hopping dynamics using a locally diabatic formalism: charge transfer in the ethylene dimer cation and excited state dynamics in the 2-pyridone dimer. , 2012, The Journal of chemical physics.

[64]  Mark A. Ratner,et al.  6-31G * basis set for atoms K through Zn , 1998 .

[65]  T. Gneiting,et al.  Matérn Cross-Covariance Functions for Multivariate Random Fields , 2010 .