# Block product density matrix embedding theory for strongly correlated spin systems

## TL;DR

This paper enhances block product DMET with spin-state optimization to better study strongly correlated spin systems, applying it to models like the Kitaev-Heisenberg and analyzing energy and correlation functions.

## Contribution

It introduces an improved variational Ansatz for BPDMET using spin-state optimization and applies it to complex spin models, providing insights into excited states and spectral functions.

## Key findings

- Improved accuracy in energy calculations for spin models
- Effective analysis of correlation functions
- Insights into excited states via tangent space diagonalization

## Abstract

Density matrix embedding theory (DMET) is a relatively new technique for the calculation of strongly correlated systems. Recently, block product DMET (BPDMET) was introduced for the study of spin systems such as the antiferromagnetic $J_1 - J_2$ model on the square lattice. In this paper, we extend the variational Ansatz of BPDMET using spin-state optimization, yielding improved results. We apply the same techniques to the Kitaev-Heisenberg model on the honeycomb lattice, comparing the results when using several types of clusters. Energy profiles and correlation functions are investigated. A diagonalization in the tangent space of the variational approach yields information on the excited states and the corresponding spectral functions.