Fastest local entanglement scrambler, multistage thermalization, and a non-Hermitian phantom

TL;DR

This paper investigates how random quantum circuits generate entanglement, revealing that non-Hermiticity can lead to unexpected effects like faster or slower relaxation rates and the existence of a 'phantom' eigenvalue, with broad implications.

Contribution

It identifies optimal protocols for fastest entanglement generation, analyzes the impact of non-Hermiticity on relaxation rates, and uncovers the 'phantom' eigenvalue phenomenon in non-Hermitian systems.

Findings

Optimal gates and protocols maximize entanglement rate.

Non-Hermiticity causes phase transitions in relaxation dynamics.

Discovery of a 'phantom' eigenvalue due to non-orthogonal eigenvectors.

Abstract

We study random quantum circuits and their rate of producing bipartite entanglement, specifically with respect to the choice of 2-qubit gates and the order (protocol) in which these are applied. The problem is mapped to a Markovian process and proved that there are large spectral equivalence classes -- different configurations have the same spectrum. Optimal gates and the protocol that generate entanglement with the fastest theoretically possible rate are identified. Relaxation towards the asymptotic thermal entanglement proceeds via a series of phase transitions in the local relaxation rate, which is a consequence of non-Hermiticity. In particular, non-Hermiticity can cause the rate to be either faster, or, even more interestingly, slower than predicted by the matrix eigenvalue gap. This is caused by an exponential in system size explosion of expansion coefficient sizes resulting in a 'phantom' eigenvalue, and is due to non-orthogonality of non-Hermitian eigenvectors. We numerically demonstrate that the phenomenon occurs also in random circuits with non-optimal generic gates, random U(4) gates, and also without spatial or temporal randomness, suggesting that it could be of wide importance also in other non-Hermitian settings, including correlations.