Clifford Circuits Augmented Time-Dependent Variational Principle

TL;DR

This paper introduces a Clifford circuits augmented TDVP method that reduces entanglement in many-body system simulations, enabling longer and more reliable time evolution studies with lower computational resources.

Contribution

It generalizes the CA-MPS framework to TDVP, applying Clifford circuits during time evolution to efficiently reduce entanglement and improve simulation accuracy.

Findings

Reduces entanglement entropy in time evolution.

Enables longer simulation times with maintained accuracy.

Low overhead due to unchanged Hamiltonian terms.

Abstract

The recently proposed Clifford Circuits Augmented Matrix Product States (CA-MPS) (arXiv:2405.09217) seamlessly augments Density Matrix Renormalization Group with Clifford circuits. In CA-MPS, the entanglement from stabilizers is transferred to the Clifford circuits which can be easily handled according to the Gottesman-Knill theorem. As a result, MPS needs only to deal with the non-stabilizer entanglement, which largely reduce the bond dimension and the resource required for the accurate simulation of many-body systems. In this work, we generalize CA-MPS to the framework of Time-Dependent Variational Principle (TDVP) for time evolution simulations. In this method, we apply Clifford circuits to the resulting MPS in each TDVP step with a two-site sweeping process similar as in DMRG, aiming at reducing the entanglement entropy in the MPS, and the Hamiltonian is transformed accordingly using the chosen Clifford circuits. Similar as in CA-MPS, the Clifford circuits doesn't increase the number of terms in the Hamiltonian which makes the overhead very small in the new method. We test this method in both XXZ chain and two dimensional Heisenberg model. The results show that the Clifford circuits augmented TDVP method can reduce the entanglement entropy in the time evolution process and hence makes the simulation reliable for longer time. The Clifford circuits augmented Time-Dependent Variational Principle provides a useful tool for the simulation of time evolution process of many-body systems in the future.