2026-10-08T00:45:58Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33756607 2026-09-14T20:27:15Z category_30232 item_type_3 month_year_09_2026

Verification Evidence and Source for Readout-Dependent Prediction Rankings and Certified Noncommuting Quantum Memory Satoshi Kawasaki (24726982) Quantum information, computation and communication quantum measurement quantum memory quantum prediction Brier loss open quantum systems matched comparison noncommuting states computer-assisted proof interval arithmetic reproducibility <p dir="ltr">This dataset accompanies the preprint Readout-Dependent Prediction Rankings and Certified Noncommuting Quantum Memory. It contains immutable certificate archives and their specifications, exact dyadic endpoints, standard-library rational replay code, model and readout formulas, manifests and SHA-256 identifiers, review originals with errata and corrected supplements, and editable manuscript and figure sources.</p><p><br></p><p dir="ltr">The two certified results are a readout-dependent D/MATCH ranking reversal at B04 and nine D-only points with noncommuting target-conditioned memory states and positive gain over fixed-Z readout. Their comparison classes and numerical guarantees are stated in the preprint. The nine-point common lower bounds are post-certification extractions, not preregistered acceptance margins.</p><p><br></p><p dir="ltr">Start with the evidence README, archive map, and checksum verifier. The supplied preparation helper reconstructs separate replay directories with the required pre-execution receipts. Archive hashing establishes byte identity; mathematical replay additionally verifies the exponential, model, stationary-system, readout, positivity, and downstream certificate chain. Selected pilot and origin-response records explain point selection and stopped-run history, but are not a complete independent replay bundle for all diagnostic experiments.</p><p><br></p><p dir="ltr">All checks were AI-assisted; there is no claim of human or institutional independent validation, journal peer review, or proof-assistant formalization. The independent Claude route [B] is documented by its submitted code and report; its full exact endpoint collection was not delivered. Original review defects and their corrections are preserved. No target model was newly evaluated during manuscript packaging. Embedded third-party materials retain their included licenses.</p><p><br></p><p dir="ltr">Companion preprint: https://doi.org/10.6084/m9.figshare.33756463</p> 2026-09-14T20:27:15Z Dataset Dataset 10.6084/m9.figshare.33756607.v1 https://figshare.com/articles/dataset/Verification_Evidence_and_Source_for_Readout-Dependent_Prediction_Rankings_and_Certified_Noncommuting_Quantum_Memory/33756607 CC BY 4.0