Machine learning Frenkel Hamiltonian parameters to accelerate simulations of exciton dynamics

Abstract

In this manuscript, we develop multiple machine learning (ML) models to accelerate a scheme for parameterizing site-based models of exciton dynamics from all-atom configurations of condensed phase sexithiophene systems. This scheme encodes the details of a system’s specific molecular morphology in the correlated distributions of model parameters through the analysis of many single-molecule excited-state electronic-structure calculations. These calculations yield excitation energies for each molecule in the system and the network of pair-wise intermolecular electronic couplings. Here, we demonstrate that the excitation energies can be accurately predicted using a kernel ridge regression (KRR) model with Coulomb matrix featurization. We present two ML models for predicting intermolecular couplings. The first one utilizes a deep neural network and bi-molecular featurization to predict the coupling directly, which we find to perform poorly. The second one utilizes a KRR model to predict unimolecular transition densities, which can subsequently be analyzed to compute the coupling. We find that the latter approach performs excellently, indicating that an effective, generalizable strategy for predicting simple bimolecular properties is through the indirect application of ML to predict higher-order unimolecular properties. Such an approach necessitates a much smaller feature space and can incorporate the insight of well-established molecular physics.

Simulating Energy Transfer in Molecular Systems with Digital Quantum Computers

“…It is worth noting that even though the computational cost of constructing the exciton Hamiltonian through quantum chemistry calculations scales linearly with the number of subunits (assuming that the inter-molecular couplings are negligible beyond certain distance), the computational cost itself is still high for large systems. However, several statistical and machine-learning methods have been developed to overcome this difficulty [46][47][48]. Once sufficient quantum chemistry data have been generated, the remaining matrix elements in the exciton Hamiltonians can be obtained through these statistical or machine learning methods with very high accuracy.…”

“…All the quantum chemical calculations were performed with the PySCF program [52], and density fitting was used with the heavy-aug-cc-pvdz-jkfit auxiliary basis set implemented in PySCF. In estimating the excitonic coupling between the local excitations on two T2 molecules (m and n), V mn , we computed the Coulomb coupling via [46] V…”

Perspective on integrating machine learning into computational chemistry and materials science

“…Beyond that, it has been shown that the model can represent interaction integrals in localized effective minimal basis representations, which benefits the prediction accuracy for larger systems. 123…”

“…The concept of ML-based Hamiltonian and densityfunctional surrogate models directly leads to the construction of approximate electronic structure models based on ML. Recently reported approaches include an ML-based Hückel model, 141 parametrized Frenkel 102,[142][143][144][145] and Tight-binding (TB) Hamiltonians 146 as well as semi-empirical methods with ML-tuned parameters. 147,148 .…”

Designing excitonic circuits for the Deutsch–Jozsa algorithm: mitigating fidelity loss by merging gate operations

“…For any specific configuration, we compute each value of the intermolecular electronic coupling in an atomistic basis via the point monopole approximation, which has been demonstrated to accurately represent couplings between closely spaced organic dye molecules. 38,39 Specifically, we define the coupling between molecules i and j as,…”