Chebyshev matrix product state approach for time evolution

TL;DR

This paper introduces a Chebyshev polynomial-based matrix product state algorithm for simulating time evolution in quantum many-body systems, demonstrating its accuracy and efficiency improvements with a spectral-decomposition extension.

Contribution

The paper develops a novel Chebyshev MPS algorithm for quantum dynamics and enhances it with a spectral-decomposition scheme for longer simulation times and better computational efficiency.

Findings

Accurately computes short-time dynamics of hardcore bosons.

Spectral-decomposition extension extends simulation times.

Offers improved data compression and computational efficiency.

Abstract

We present and test a new algorithm for time-evolving quantum many-body systems initially proposed by Holzner et al. [Phys. Rev. B 83, 195115 (2011)]. The approach is based on merging the matrix product state (MPS) formalism with the method of expanding the time-evolution operator in Chebyshev polynomials. We calculate time-dependent observables of a system of hardcore bosons quenched under the Bose-Hubbard Hamiltonian on a one-dimensional lattice. We compare the new algorithm to more standard methods using the MPS architecture. We find that the Chebyshev method gives numerically exact results for small times. However, the reachable times are smaller than the ones obtained with the other state-of-the-art methods. We further extend the new method using a spectral-decomposition-based projective scheme that utilizes an effective bandwidth significantly smaller than the full bandwidth, leading to longer evolution times than the non-projective method and more efficient information storage, data compression, and less computational effort.