An Equilibrium for Frustrated Quantum Spin Systems in the Stochastic State Selection Method

TL;DR

This paper introduces a new stochastic method to accurately compute the lowest eigenvalues of frustrated quantum spin systems, demonstrating its effectiveness on large lattice models with high precision.

Contribution

It extends the stochastic state selection method to frustrated models, establishing the existence of an equilibrium state for eigenvalue estimation in such systems.

Findings

Successfully applied to 20-, 32-, 36-, and 40-site systems

Reproduces known eigenvalues with high accuracy

Achieves less than 0.2% error on large systems

Abstract

We develop a new method to calculate eigenvalues in frustrated quantum spin models. It is based on the stochastic state selection (SSS) method, which is an unconventional Monte Carlo technique we have investigated in recent years. We observe that a kind of equilibrium is realized under some conditions when we repeatedly operate a Hamiltonian and a random choice operator, which is defined by stochastic variables in the SSS method, to a trial state. In this equilibrium, which we call the SSS equilibrium, we can evaluate the lowest eigenvalue of the Hamiltonian using the statistical average of the normalization factor of the generated state. The SSS equilibrium itself has been already observed in un-frustrated models. Our study in this paper shows that we can also see the equilibrium in frustrated models, with some restriction on values of a parameter introduced in the SSS method. As a concrete example, we employ the spin-1/2 frustrated J1-J2 Heisenberg model on the square lattice. We present numerical results on the 20-, 32-, 36-site systems, which demonstrate that statistical averages of the normalization factors reproduce the known exact eigenvalue in good precision. Finally we apply the method to the 40-site system. Then we obtain the value of the lowest energy eigenvalue with an error less than 0.2%.