Machine Learning-Assisted Excited State Molecular Dynamics with the State-Interaction State-Averaged Spin-Restricted Ensemble-Referenced Kohn–Sham Approach

Abstract

We present a machine learning-assisted excited state molecular dynamics (ML-ESMD) based on the ensemble density functional theory framework. Since we represent a diabatic Hamiltonian in terms of generalized valence bond ansatz within the state-interaction state-averaged spin-restricted ensemble-referenced Kohn–Sham (SI-SA-REKS) method, we can avoid singularities near conical intersections, which are crucial in excited state molecular dynamics simulations. We train the diabatic Hamiltonian elements and their analytical gradients with the SchNet architecture to construct machine learning models, while the phase freedom of off-diagonal elements of the Hamiltonian is cured by introducing the phase-less loss function. Our machine learning models show reasonable accuracy with mean absolute errors of ∼0.1 kcal/mol and ∼0.5 kcal/mol/Å for the diabatic Hamiltonian elements and their gradients, respectively, for penta-2,4-dieniminium cation. Moreover, by exploiting the diabatic representation, our models can predict correct conical intersection structures and their topologies. In addition, our ML-ESMD simulations give almost identical result with a direct dynamics at the same level of theory.

Ab initio nonadiabatic molecular dynamics of charge carriers in metal halide perovskites

“…103,110 Machine learning (ML) is a promising strategy to accelerate NA-MD simulations by predicting the excited state properties for longer timescales based on costeffective training sets, without involving underlying quantum mechanics calculations. 130,[249][250][251][252][253][254][255][256][257][258][259][260][261][262][263][264][265][266] The overall idea of QSH/ NA-MD is qualitatively similar to the ML/NA-MD simulation. So far, works on ML/NA-MD have focused primarily on molecular systems.…”

Automatic discovery of photoisomerization mechanisms with nanosecond machine learning photodynamics simulations

“… 18 Recently, Ha et al have trained the SchNet model to study the excited-state dynamics of penta-2,4-dieniminium cation. 19 Applying ML-based NAMD on complex molecules continues to challenge theorists because of additional conformational flexibility and increase in available degrees of freedom. Training ML models becomes increasingly difficult with the rapidly growing required size of training datasets, especially when based on atom-wise molecular representations.…”

Molecular excited states through a machine learning lens

“…Often, excited-state wavepacket dynamics is performed in the diabatic representation (Box 5), and ML algorithms have been intensely employed to aid the fitting and diabatization process. [152][153][154][155][156][157][158][159][160][161][162][163][164][165][166][167][168] Several recent works have extended the latter approaches to also use ML for learning transition dipole moments 164 and spin-orbit couplings. 169 All ML works mentioned in this section were dealing with only a small number of excited states.…”