Inverse Kohn-Sham Density Functional Theory: Progress and Challenges

TL;DR

This paper reviews the challenges of inverse Kohn-Sham problems in density functional theory, compares existing methods, and proposes new strategies to improve numerical stability and tractability.

Contribution

It introduces novel modifications to the Wu-Yang and PDE-CO methods, enhancing their stability and applicability in finite basis set scenarios.

Findings

Finite basis sets limit iKS accuracy

Regularization improves numerical stability

New correction methods enhance inversion performance

Abstract

Inverse Kohn-Sham (iKS) problems are needed to fully understand the one-to-one mapping between densities and potentials on which Density Functional Theory is based. They are also important to advance computational schemes that rely on density-to-potential inversions such as the Optimized Effective Potential method and various techniques for density-based embedding. Unlike the forward Kohn-Sham problems, numerical iKS problems are ill-posed and can be unstable. We discuss some of the fundamental and practical difficulties of iKS problems with constrained-optimization methods on finite basis sets. Various factors that affect the performance are systematically compared and discussed, both analytically and numerically, with a focus on two of the most practical methods: the Wu-Yang method (WY) and partial-differential-equation constrained-optimization (PDE-CO). Our analysis of the WY and PDE-CO highlights the limitation of finite basis sets and the importance of regularization. We introduce two new ideas that will hopefully contribute to making iKS problems more tractable: (1) A correction to the WY method that utilizes the null space of the relevant Hessian matrices; and (2) A finite potential basis-set implementation of the PDE-CO method. We provide an overall strategy for performing numerical density-to-potential inversions that can be directly adopted in practice. We also provide an Appendix with several examples that can be used for benchmarking.