Improving Perturbation Theory with the Sum-of-Squares: Third Order

TL;DR

This paper introduces a new sum-of-squares method that improves perturbation theory accuracy and efficiency for certain quantum Hamiltonians, addressing limitations of existing approaches like the 2RDM method.

Contribution

It develops a perturbation-theory analogue for sum-of-squares, providing a way to estimate error and improve solutions, with demonstrated advantages over traditional methods.

Findings

Self-consistent sum-of-squares outperforms 2RDM in speed and accuracy

The method improves perturbation theory results for specific model Hamiltonians

Identifies limitations for quantum chemistry Hamiltonians and proposes future modifications

Abstract

The sum-of-squares method can give rigorous lower bounds on the energy of quantum Hamiltonians. Unfortunately, typically using this method requires solving a semidefinite program, which can be computationally expensive. Further, the typically used degree-$4$ sum-of-squares (also known as the 2RDM method) does not correctly reproduce second order perturbation theory. Here, we give a general method, an analogue of Wigner's $2n+1$ rule for perturbation theory, to compute the order of the error in a given sum-of-squares ansatz. We also give a method for finding solutions of the dual semidefinite program, based on a perturbative ansatz combined with a self-consistent method. As an illustration, we show that for a class of model Hamiltonians (with a gap in the quadratic term and quartic terms chosen as i.i.d. Gaussians), this self-consistent sum-of-squares method significantly improves over the 2RDM method in both speed and accuracy, and also improves over low order perturbation theory. We then explain why the particular ansatz we implement is not suitable for use for quantum chemistry Hamiltonians (due to presence of certain large diagonal terms), but we suggest a modified ansatz that may be suitable, which will be the subject of future work.