Self-Adaptive Tensor Network States with Multi-Site Correlators

TL;DR

This paper presents a self-adaptive tensor network state method that efficiently incorporates important multi-site correlators guided by entanglement measures, improving convergence in complex quantum systems.

Contribution

Introduction of SATNS, a self-adaptive tensor network approach that selectively includes correlators based on entanglement, reducing parameters while enhancing accuracy.

Findings

Successfully applied to manganocene's low-energy states

Resolves convergence issues of previous tensor network methods

Achieves accurate results with fewer parameters

Abstract

We introduce the concept of self-adaptive tensor network states (SATNS) based on multi-site correlators. The SATNS ansatz gradually extends its variational space incorporating the most important next-order correlators into the ansatz for the wave function. The selection of these correlators is guided by entanglement-entropy measures from quantum information theory. By sequentially introducing variational parameters and adjusting them to the system under study, the SATNS ansatz achieves to keep their number significantly smaller than the total number of full-configuration interaction parameters. The SATNS ansatz is studied for manganocene in its lowest-energy sextet and doublet states, the latter of which is known to be difficult to describe. It is shown that the SATNS parametrization solves the convergence issues found for previous correlator-based tensor network states.