Mitigating crosstalk errors by randomized compiling: Simulation of the BCS model on a superconducting quantum computer

TL;DR

This paper extends randomized compiling to include neighboring qubit effects, significantly reducing crosstalk errors in superconducting quantum computers, enabling more accurate simulation of complex quantum models like the BCS Hamiltonian.

Contribution

The authors develop a specialized randomized compiling protocol that mitigates crosstalk errors by transforming coherent noise into depolarising noise, improving quantum simulation accuracy.

Findings

Crosstalk errors are a major source of noise in IBMQ superconducting qubits.

The extended RC protocol effectively converts crosstalk-induced coherent errors into depolarising noise.

Application to the BCS model demonstrates improved simulation fidelity in a crosstalk-dominated regime.

Abstract

We develop and apply an extension of the randomized compiling (RC) protocol that includes a special treatment of neighboring qubits and dramatically reduces crosstalk effects caused by the application of faulty gates on superconducting qubits in IBMQ quantum computers (\texttt{ibm\_lagos} and \texttt{ibmq\_ehningen}). Crosstalk errors, stemming from CNOT two-qubit gates, are a crucial source of errors on numerous quantum computing platforms. For the IBMQ machines, their magnitude is often overlooked-9. Our RC protocol turns coherent noise due to crosstalk into a depolarising noise channel that can then be treated using established error mitigation schemes, such as noise estimation circuits. We apply our approach to the quantum simulation of the non-equilibrium dynamics of the Bardeen-Cooper-Schrieffer (BCS) Hamiltonian for superconductivity, a particularly challenging model to simulate on quantum hardware because of the long-range interaction of Cooper pairs. With 135 CNOT gates, we work in a regime where crosstalk, as opposed to either trotterization or qubit decoherence, dominates the error. Our twirling of neighboring qubits is shown to dramatically improve the noise estimation protocol without the need to add new qubits or circuits and allows for a quantitative simulation of the BCS model.