English

Approximating dynamical correlation functions with constant depth quantum circuits

Quantum Physics 2025-02-26 v4

Abstract

One of the most important quantities characterizing the microscopic properties of quantum systems are dynamical correlation functions. These correlations are obtained by time-evolving a perturbation of an eigenstate of the system, typically the ground state. In this work, we study approximations of these correlation functions that do not require time dynamics. We show that having access to a circuit that prepares an eigenstate of the Hamiltonian, it is possible to approximate the dynamical correlation functions up to exponential accuracy in the complex frequency domain ω=(ω)+i(ω)\omega=\Re(\omega)+i\Im(\omega), on a strip above the real line (ω)=0\Im(\omega)=0. We achieve this by exploiting the continued fraction representation of the dynamical correlation functions as functions of frequency ω\omega, where the level kk approximant can be obtained by measuring a weight O(k)O(k) operator on the eigenstate of interest. In the complex ω\omega plane, we show how this approach allows to determine approximations to correlation functions with accuracy that increases exponentially with kk. We analyse two algorithms to generate the continuous fraction representation in scalar or matrix form, starting from either one or many initial operators. We prove that these algorithms generate an exponentially accurate approximation of the dynamical correlation functions on a region sufficiently far away from the real frequency axis. We present numerical evidence of these theoretical results through simulations of small lattice systems. We comment on the stability of these algorithms with respect to sampling noise in the context of quantum simulation using quantum computers.

Keywords

Cite

@article{arxiv.2406.03204,
  title  = {Approximating dynamical correlation functions with constant depth quantum circuits},
  author = {Reinis Irmejs and Raul A. Santos},
  journal= {arXiv preprint arXiv:2406.03204},
  year   = {2025}
}

Comments

31 pages, 10 figures

R2 v1 2026-06-28T16:54:26.520Z