Current theoretical models for excited states in electronic structure are typically less accurate and less reliable than those for ground states. The reason for this discrepancy can often be traced to a difference in treatment. Ground states are treated, at least in part, through the use of a robust variational principle that allows even nonlinear components of a wave function to be optimized for that particular electronic state. Excited states, however, are more often treated by linear response theory, which is far less capable of tailoring the whole wave function to match a given excited state. The result is often a bias in favor of ground states, which is well documented for charge transfer states in configuration interaction singles and for double excitations in equation of motion coupled cluster with singles and doubles. In light of these difficulties, we will present progress on our development of a variational principle that is equally valid for ground and excited states and thus allows the wave function to be fully tailored to each of the states under study. We will also discuss the variational Monte Carlo algorithms that support the practical application of this principle, as well as some wave function innovations that can be leveraged to maximize its effectiveness. In addition to these formal topics, we will present data showing the effectiveness of the approach for discerning both molecular excitations and the band structures of solids.

Progress in excited state variational principles for molecules and solids

254th National Meeting (2017)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.