Machine Learning Exciton Hamiltonians in Light-Harvesting Complexes

Abstract

We propose a machine learning (ML)-based strategy for an inexpensive calculation of excitonic properties of lightharvesting complexes (LHCs). The strategy uses classical molecular dynamics simulations of LHCs in their natural environment in combination with ML prediction of the excitonic Hamiltonian of the embedded aggregate of pigments. The proposed ML model can reproduce the effects of geometrical fluctuations together with those due to electrostatic and polarization interactions between the pigments and the protein. The training is performed on the chlorophylls of the major LHC of plants, but we demonstrate that the model is able to extrapolate well beyond the initial training set. Moreover, the accuracy in predicting the effects of the environment is tested on the simulation of the small changes observed in the absorption spectra of the wild-type and a mutant of a minor LHC.

Multifidelity Machine Learning for Molecular Excitation Energies

“…Moreover, some systems contain more than 2000 pigments, 2 and dynamical simulations for the systems are envisioned to obtain the (timedependent) spectroscopic properties. [3][4][5]7,10,33 The present work is a first step in the direction of these applications and benchmarks MFML and its effectiveness on three molecules of growing size, namely, benzene, naphthalene, and anthracene. The challenge to viably cover the chemical space for such an application is beyond the scope of this work but can include various methods such as active learning approaches.…”

“…To understand the energy flow in such aggregates, the excitation energies to the first excited state need to be known accurately due to the shallow energy landscape in some of these systems. Moreover, some systems contain more than 2000 pigments, and dynamical simulations for the systems are envisioned to obtain the (time-dependent) spectroscopic properties. − ,,, The present work is a first step in the direction of these applications and benchmarks MFML and its effectiveness on three molecules of growing size, namely, benzene, naphthalene, and anthracene. The challenge to viably cover the chemical space for such an application is beyond the scope of this work but can include various methods such as active learning approaches. , It must be noted that this is a general challenge in the field of ML for quantum chemistry and is not restricted to the work presented herein.…”

Elucidating the Role of Hydrogen Bonding in the Optical Spectroscopy of the Solvated Green Fluorescent Protein Chromophore: Using Machine Learning to Establish the Importance of High-Level Electronic Structure

“…We then show how this initial low-level ML model that qualitatively captures the character of the electronic excitations, such as which nuclear motions couple most strongly to the electronic excitation, can be enhanced to achieve quantitative accuracy by transfer learning using only hundreds of high-level energies. The data efficiency of this approach allows us to train ML models to high-level electronic structure with the solvation environment explicitly included, expanding on other recent studies which have trained ML models to high-level electronic structure calculations in the gas phase , or with implicit solvation , or to lower-level electronic structure methods such as TDDFT. ,− In particular, here we demonstrate that one can develop an accurate ML model of electronic excitations at the level of EOM-CCSD embedded in DFT by starting from either a TDDFT or a configuration interaction singles (CIS) , low-level treatment of the excited-state electronic structure using only 400 embedded EOM-CCSD energy calculations.…”

QM/AMOEBA description of properties and dynamics of embedded molecules

“…Machine learning (ML) techniques have now emerged as a powerful means to reduce the computational cost of QM calculations, retaining almost the same accuracy as the QM method they are trained on 151–153 . As ML becomes widespread in computational chemistry applications, new frameworks are emerging to couple ML models for molecules with MM descriptions of the environment 154–156 . Further developments are needed, however, if one wants to introduce a polarizable AMOEBA environment seamlessly in a ML/AMOEBA framework.…”

“…[151][152][153] As ML becomes widespread in computational chemistry applications, new frameworks are emerging to couple ML models for molecules with MM descriptions of the environment. [154][155][156] Further developments are needed, however, if one wants to introduce a polarizable AMOEBA environment seamlessly in a ML/AMOEBA framework. In fact, only a model that accounts for all the physical contributions in a polarizable QM/MM embedding will ensure robustness and accuracy of the ML prediction.…”