Promoting collaboration across the theoretical sciences

machine learning in chemistry

Machine learning energy gaps of molecules in the condensed phase for linear and nonlinear optical spectroscopy
Christine Isborn, University of California Merced

One of the greatest successes of machine learning in chemistry is the prediction (interpolation!) of potential energy surfaces, greatly accelerating energy calculations.  Accurately simulating optical spectroscopy requires the prediction of both ground and excited state potential energy surfaces, so machine learning is well-poised to make a meaningful contribution to these simulations and the interpretation of the excited-state dynamics of chromophores that determine a range of biological and energy capture processes.  A powerful method for including both condensed phase explicit environment and vibronic contributions to linear and nonlinear optical spectra involves the computation of the energy gap time correlation function along a molecular dynamics trajectory, requiring the evaluation of tens of thousands of excited-state electronic structure calculations. This talk will present our results leveraging the locality of chromophore excitations to develop machine learning models to predict the excited-state energy gap of chromophores in complex environments for efficiently constructing linear and multidimensional optical spectra. By analyzing the performance of these models, which span a hierarchy of physical approximations, across a range of chromophore–environment interaction strengths, we provide strategies for the construction of machine learning models that greatly accelerate the calculation of multidimensional optical spectra from first principles.    

References:  

https://pubs.acs.org/doi/10.1021/acs.jpclett.0c02168 [pubs.acs.org] 

https://www.annualreviews.org/doi/pdf/10.1146/annurev-physchem-090419-051350 [annualreviews.org]