Skip to content
aiqbench takeQuantum Computing

aiqbench analysis

This research introduces round-trip state echo (RTSE) as a workload-specific diagnostic metric for evaluating quantum hardware performance, distinguishing it from static calibration scores. By analyzing execution snapshots across superconducting and trapped-ion systems, the study demonstrates that hardware reliability metrics are highly sensitive to compilation details and specific task batches rather than being universally transferable. These findings provide a rigorous framework for benchmarking quantum processors based on actual execution outcomes rather than theoretical error suppression claims.
Source coveringarXiv
What this isA model's reading of the article and related coverage, written for aiqbench. Not the publisher's words.

From Round-Trip State Echo to Error Recovery: Snapshot-Resolved Quantum-Hardware Diagnostics

arXiv3 min read

arXiv:2608.26010v1 Announce Type: new Abstract: End-to-end quantum-hardware scores need not transfer across workloads, compilations, or execution times. We specify a compilation-explicit screen-and-stress profile whose opening diagnostic is round-trip state echo (RTSE): prepare one of four tetrahedral qubit states at a route root, move it out and back by swaps, apply inverse preparation at the root, and record zero. An execution snapshot means a dated submitted task batch together with its captured capability document where available, not a certified calibration epoch. On sparse superconducting hardware, a byte-identical communication rerun changed route-level contrasts although the aggregate RTSE estimates differed by only 0.00125. In a separate prospectively frozen two-window length study, RTSE and the remote-inverse do-nothing predecessor's root marginal both fell from length 2 to length 10; the prespecified interaction did not support superior RTSE retention. The mean selected-output return probability across 64 deletion-recovery cells changed from 0.738 to 0.624 between IQM execution snapshots. On a trapped-ion service advertising all-to-all connectivity among five submitted virtual wires, recovery was 0.911 and 0.923 in two windows, exceeding the frozen two-thirds reference; recovery-minus-adjoint-control differences were 0.446 and 0.443. These are execution-workload diagnostics, not coding-gain, error-suppression, physical-loss, fault-tolerance, or architecture-ranking claims. The results support assessment indexed by workload, placement or virtual-wire contract, compilation, architecture, and execution snapshot.

Abstract:End-to-end quantum-hardware scores need not transfer across workloads, compilations, or execution times. We specify a compilation-explicit screen-and-stress profile whose opening diagnostic is round-trip state echo (RTSE): prepare one of four tetrahedral qubit states at a route root, move it out and back by swaps, apply inverse preparation at the root, and record zero. An execution snapshot means a dated submitted task batch together with its captured capability document where available, not a certified calibration epoch. On sparse superconducting hardware, a byte-identical communication rerun changed route-level contrasts although the aggregate RTSE estimates differed by only 0.00125. In a separate prospectively frozen two-window length study, RTSE and the remote-inverse do-nothing predecessor's root marginal both fell from length 2 to length 10; the prespecified interaction did not support superior RTSE retention. The mean selected-output return probability across 64 deletion-recovery cells changed from 0.738 to 0.624 between IQM execution snapshots. On a trapped-ion service advertising all-to-all connectivity among five submitted virtual wires, recovery was 0.911 and 0.923 in two windows, exceeding the frozen two-thirds reference; recovery-minus-adjoint-control differences were 0.446 and 0.443. These are execution-workload diagnostics, not coding-gain, error-suppression, physical-loss, fault-tolerance, or architecture-ranking claims. The results support assessment indexed by workload, placement or virtual-wire contract, compilation, architecture, and execution snapshot.

43 pages, 11 figures, 10 tables. Reproducibility package: doi:https://doi.org/10.5281/zenodo.21969397

arXiv:2608.26010 [quant-ph]

(or arXiv:2608.26010v1 [quant-ph] for this version)

https://doi.org/10.48550/arXiv.2608.26010 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Full text under the publisher's open licence · 305 words

Related coverage