Calibrating quantum gates up to 52 qubits in a superconducting processor

TL;DR

This paper demonstrates benchmarking of large-scale quantum gates up to 52 qubits using global fidelity, introduces an inter-gate correlation metric, and improves gate fidelity through optimized calibration in a superconducting quantum processor.

Contribution

It presents a scalable benchmarking protocol for large quantum gates, introduces a new correlation metric, and applies global fidelity for effective gate optimization.

Findings

Achieved 63.09% fidelity for a 44-qubit parallel CZ gate.

Enhanced 6-qubit gate fidelity from 87.65% to 92.04% using global fidelity-based optimization.

Validated the noise model with experimental results, revealing insights into correlated noise mitigation.

Abstract

Benchmarking large-scale quantum gates, typically involving multiple native two-qubit and singlequbit gates, is crucial in quantum computing. Global fidelity, encompassing information about intergate correlations, offers a comprehensive metric for evaluating and optimizing gate performance, unlike the fidelities of individual local native gates. In this work, utilizing the character-average benchmarking protocol implementable in a shallow circuit, we successfully benchmark gate fidelities up to 52 qubits. Notably, we achieved a fidelity of 63.09$\pm $0.23% for a 44-qubit parallel CZ gate. Utilizing the global fidelity of the parallel CZ gate, we explore the correlations among local CZ gates by introducing an inter-gate correlation metric, enabling one to simultaneously quantify crosstalk error when benchmarking gate fidelity. Finally, we apply our methods in gate optimization. By leveraging global fidelity for optimization, we enhance the fidelity of a 6-qubit parallel CZ gate from 87.65% to 92.04% and decrease the gate correlation from 3.53% to 3.22%, compared to local gate fidelitybased optimization. The experimental results align well with our established composite noise model, incorporating depolarizing and ZZ-coupling noises, and provide valuable insight into further study and mitigation of correlated noise.