arXiv:2610.24213
Generalized Svetlichny Inequalities as a Cross-Platform Benchmark for Quantum Processors
Andrés Camilo Granda Arango, Giuseppe Sergioli, Luigi Iapichino, Tobias Valentin Bauer, Carlo Cuccu, Roberto Giuntini
Abstract
A cross-platform benchmark should be device-independent yet sensitive to imperfections accumulated during execution. Building on prior work, we formulate and validate a protocol through demonstrations based on generalized Svetlichny inequality violations (a simulation, rather than a loophole-free certification, of multipartite nonlocality) for larger subsystem sizes. We apply it unchanged to three processors, the trapped-ion Marmot and IBEX Q1 (both manufactured by AQT) and the superconducting Euro-Q-Exa System 1 (IQM), using identical GHZ-state circuits with fixed theoretical measurement settings from N=2 to N=10 qubits. Both trapped-ion processors sustain violations up to ten qubits, while the superconducting processor loses them beyond six qubits under unmitigated (raw) conditions; readout-error mitigation recovers violations up to eight qubits, with a substantial fraction of the degradation attributable to measurement imperfections. These processors provide a heterogeneous testbed for validating the protocol's cross-platform applicability and discriminating power, demonstrating its practicality as a common framework for assessing collective quantum performance.
AI Summary
The paper asks whether generalized Svetlichny inequalities can provide a common benchmark of multipartite correlations across quantum processors. It implements identical logical GHZ-preparation circuits and fixed, theoretically optimal measurement settings on two AQT trapped-ion processors, Marmot and IBEX Q1, and IQM’s superconducting Euro-Q-Exa System 1, for $N=2$–$10$ qubits. Performance is measured by $\mathcal V_N=S_N/2^{N-1}$, with hybrid-local-realistic bound $\mathcal V_N\leq1$ and ideal GHZ value $\sqrt2$. Every one of the $2^N$ correlators is measured directly, reaching 1,024 settings at ten qubits. Each setting uses 2,000 shots per repetition, with five repetitions; reported 95% confidence intervals use repetition-level variability and Student’s $t$ distribution. Both trapped-ion devices retain statistically significant violations through ten qubits. The superconducting device retains raw violations only to approximately six qubits; local readout mitigation extends this to approximately eight. Mitigation changes trapped-ion results by less than approximately 9%, whereas its relative gain reaches approximately 27% at ten qubits on the superconducting device. Calibrated noise-model simulations provide additional comparisons. The benchmark measures combined state-preparation, gate, compilation, decoherence, and measurement effects rather than isolating their individual contributions. Co-located qubits preclude loophole-free nonlocality certification, so the tests are described as simulating nonlocal correlations. Full correlator evaluation costs exponentially many settings. Correlated readout mitigation likewise has exponential calibration cost and is restricted to $N\leq6$; local mitigation assumes independent readout errors. Further scaling requires sampling or reduced-measurement formulations.
Error found: The paper inconsistently describes whether the raw superconducting violation disappears at six qubits or only beyond six, without numerical results in the supplied text that resolve the boundary. The abstract invokes device independence, whereas the demonstrated protocol establishes a hardware-independent logical workload and explicitly does not provide loophole-free nonlocality certification.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 12 pages, 5 figures, 2 tables