Circuit depth versus energy in topologically ordered systems

TL;DR

This paper establishes a fundamental lower bound on the circuit depth required to prepare low-energy states in 2D topologically ordered systems, highlighting limitations of variational quantum algorithms.

Contribution

It provides the first nontrivial circuit-depth lower bound for preparing low-energy states in 2D topologically ordered systems assuming local quantum circuits.

Findings

Lower bound of (/^{(1-lpha)/2}}, \, \, (\, |\u0018|) for preparing low-energy states.

Variational circuits cannot exponentially reduce energy density with circuit depth.

Long-range entanglement can impose circuit-depth constraints even at nonzero energy density.

Abstract

We prove a nontrivial circuit-depth lower bound for preparing a low-energy state of a locally interacting quantum many-body system in two dimensions, assuming the circuit is geometrically local. For preparing any state which has an energy density of at most $\epsilon$ with respect to Kitaev's toric code Hamiltonian on a two dimensional lattice $\Lambda$, we prove a lower bound of $\Omega\left(\min\left(1/\epsilon^{\frac{1-\alpha}{2}}, \sqrt{|\Lambda|}\right)\right)$ for any $\alpha >0$. We discuss two implications. First, our bound implies that the lowest energy density obtainable from a large class of existing variational circuits (e.g., Hamiltonian variational ansatz) cannot, in general, decay exponentially with the circuit depth. Second, if long-range entanglement is present in the ground state, this can lead to a nontrivial circuit-depth lower bound even at nonzero energy density. Unlike previous approaches to prove circuit-depth lower bounds for preparing low energy states, our proof technique does not rely on the ground state to be degenerate.