Size Consistent Excited States via Algorithmic Transformations between Variational Principles

We
demonstrate that a broad class of excited state variational
principles is not size consistent. In light of this difficulty, we
develop and test an approach to excited state optimization that transforms
between variational principles to achieve state selectivity, size
consistency, and compatibility with quantum Monte Carlo. To complement
our formal analysis, we provide numerical examples that confirm these
properties and demonstrate how they contribute to a more black box
approach to excited states in quantum Monte Carlo.