α-CASSCF: An Efficient, Empirical Correction for SA-CASSCF To Closely Approximate MS-CASPT2 Potential Energy Surfaces

James W Snyder Jr; Robert M Parrish; Todd J Martínez

doi:10.1021/acs.jpclett.7b00940

α-CASSCF: An Efficient, Empirical Correction for SA-CASSCF To Closely Approximate MS-CASPT2 Potential Energy Surfaces

J Phys Chem Lett. 2017 Jun 1;8(11):2432-2437. doi: 10.1021/acs.jpclett.7b00940. Epub 2017 May 18.

Authors

James W Snyder Jr^{1

2}, Robert M Parrish^{1

2}, Todd J Martínez^{1

2}

Affiliations

¹ Department of Chemistry and The PULSE Institute, Stanford University , Stanford, California 94305, United States.
² SLAC National Accelerator Laboratory , Menlo Park, California 94025, United States.

PMID: 28513165
DOI: 10.1021/acs.jpclett.7b00940

Abstract

Because of its computational efficiency, the state-averaged complete active-space self-consistent field (SA-CASSCF) method is commonly employed in nonadiabatic ab initio molecular dynamics. However, SA-CASSCF does not effectively recover dynamical correlation. As a result, there can be qualitative differences between SA-CASSCF potential energy surfaces (PESs) and more accurate reference surfaces computed using multistate complete active space second-order perturbation theory (MS-CASPT2). Here we introduce an empirical correction to SA-CASSCF that scales the splitting between individual states and the state-averaged energy. We call this the α-CASSCF method, and we show here that it significantly improves the accuracy of relative energies and PESs compared with MS-CASPT2 for the chromophores of green fluorescent and photoactive yellow proteins. As such, this method may prove to be quite valuable for nonadiabatic dynamics.