Application-level benchmarking of superconducting qubit platforms using non-local games

Non-local games (NLGs) have recently emerged as a novel benchmark family for near-term quantum hardware applications [1]. In this talk, we explore the utility of NLG circuits and their role in bridging high-level application performance with low-level hardware metrics. Specifically, we demonstrate how to infer high-level performance from low-level metrics. And how to extrapolate hardware characteristics based on the observed performance of two-player NLGs. These games are tested on superconducting qubit platforms with varying native gate sets, qubit connectivity, and noise profiles.

The core elements of the game circuits are: (1) robust multi-partite entanglement across four qubits, and (2) the execution of independent, entangled measurements. Through both noisy circuit simulations and hardware experiments, we evaluate how application-level performance is influenced by low-level factors such as correlated noise, gate fidelity, and state fidelity. We also investigate the effectiveness of noise mitigation, circuit optimization, and game strategy design on overall performance. Lastly, we assess the accuracy of predicting application-level performance from volumetric benchmarks of entanglement [2] and hardware.



[1] arXiv:2311.01363 [quant-ph]

[2] arXiv:2209.00678 [quant-ph]

This research used resources of the Oak Ridge Leadership Computing Facility, which is a DOE Office of Science User Facility supported under Contract DE-AC05-00OR22725.