2026-10-09T18:50:32Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33703660 2026-09-13T15:20:48Z category_29884 item_type_3 month_year_09_2026

Verification Evidence and Source for Certified Prediction Improvement without Net Contact-Work Recovery Satoshi Kawasaki (24726982) Probability theory stochastic thermodynamics prediction Brier loss continuous-time Markov chains entropy production kinetic feedback matched comparison computer-assisted proof interval arithmetic constrained response <p dir="ltr">This dataset contains verification evidence and editable source for “Certified Prediction Improvement without Net Contact-Work Recovery” (preprint repository identifier: https://doi.org/10.6084/m9.figshare.33702808). Its repository identifier is https://doi.org/10.6084/m9.figshare.33703660. Both identifiers were reserved during publication preparation.</p><p dir="ltr"><br></p><p dir="ltr">The files comprise the complete, unchanged scientific evidence archive and the final editable manuscript source archive, including review records and errata. They preserve specifications, generation and verification code, rational interval endpoints, linear-system certificates, attempted evaluation records, exact coverage structures, manifests, and replay entry points and instructions. Archive hashing and mathematical replay are distinct checks; a partial review package cannot replace the complete evidence.</p><p dir="ltr"><br></p><p dir="ltr">In the specified 32-state continuous-time Markov model, an origin-connected analytic branch has equal stationary entropy-production rates and Bayes Brier-loss difference L_D - L_M <= -t^2/5000 for every 0 < t <= 1/128. The comparator is reconstructed at each point to match contact occupancy and directional switching fluxes. This is a pairwise DIF comparison, not fixed baseline statistics or production.</p><p dir="ltr"><br></p><p dir="ltr">Dynamic contact work is zero edgewise; MATCH contact work is zero in the stationary net sum. The A0 ledger excludes additional controller measurement, storage, erasure, and implementation costs. OpenAI (ChatGPT/Codex) and Anthropic (Claude) systems assisted the work. AI checks do not constitute human or institutional independent validation or journal peer review.</p><p dir="ltr"><br></p> 2026-09-13T15:20:48Z Dataset Dataset 10.6084/m9.figshare.33703660.v1 https://figshare.com/articles/dataset/Verification_Evidence_and_Source_for_Certified_Prediction_Improvement_without_Net_Contact-Work_Recovery/33703660 CC BY 4.0