Characterizing noisy quantum computation with imperfectly addressed errors

Quantum protocols on hardware are subject to noise that prohibits performance. Protocols for addressing errors, such as error correction or error mitigation, may fail to combat errors in quantum computation if noise violates critical assumptions required for these protocols to be effective. However, tools for characterizing such failures in realistic operating conditions are limited. For example, while brute force simulations may be used to characterize the impact of such failures on a handful of input states, such simulations lack a complete description for how noise transforms state-spaces in the full quantum Hilbert space. In this work, we associate quantum computation subject to realistic noise to an ensemble of random superoperators and study the eigen- and singular spectral distributions over this ensemble. We propose a new theoretical framework to characterize singular values of random complex matrices using matrix Chernoff concentration. Using our framework, we analyze imperfectly addressed errors in error mitigation and error correction. We find that distributions of singular spectra depend on how noise violates critical assumptions of these protocols. Finally, we quantitatively discuss how our work may be applied to understanding limiting behavior of quantum computation, such as establishing spectral gaps and relaxation times for specific families of quantum Markov processes. Our work paves the way for new tools to diagnose when to trust the output of noisy quantum computers.