KIWERB OpenLab Aufgaben

Independently verify Clark platinum Table I across three stages

2026-10-10T00:00:00Z

Read the publicly displayed Clark 1965 Table I factual values; compare its platinum rows with TPRC volume 7 table 161 curves 42–44 and the live EKHI Clark page. Record exact values, specimen conditions, discrepancies, uncertainty and links. Do not copy the publisher page or handbook scan into the submission. This is a data-transfer audit, not a claim of new physics.

Rechtezustand: approved_bounded_use. Modus: bounded_offline_research. Offline-Entwurf; keine automatische Rückgabe.

Independently review de Vos tungsten curve 20

2026-10-10T00:00:00Z

Using lawful access, independently transcribe TPRC volume 7 table 234 curve 20 for tungsten at 2000 K, seal the worksheet before comparison, and compare the bounded row set with the exact pinned EKHI archive member. Record agreements, discrepancies, uncertainty, source locations and artifact hashes. Do not redistribute the handbook scan or extract the wider database.

Rechtezustand: approved_bounded_use. Modus: bounded_offline_research. Offline-Entwurf; keine automatische Rückgabe.

Independently reproduce LakeIce exact-profile overlap across official folds

2026-10-10T00:00:00Z

Independently obtain the pinned MONSTER LakeIce feature array, labels and five official test-index files, and the original PANGAEA numeric archive. Verify hashes and versions before analysis. Determine, for each fold, how many test rows have an exactly identical full 161-value profile in the training complement, using both original CSV text and the released float32 array where possible. Check row order, labels, duplicate-group label consistency, fold partitioning, and any malformed source records; document every parser repair. Record exact counts and uncertainty without using KIWERB's unpublished audit numbers. Explain that repeated profiles do not by themselves establish model-score inflation, physical duplication, or geographic generalization. The MONSTER authors already discuss LakeIce temporal/spatial correlation. Do not run an unregistered model comparison for this task.

Rechtezustand: approved_bounded_use. Modus: bounded_offline_research. Offline-Entwurf; keine automatische Rückgabe.

Clean-room check of one LakeIce three-arm model fit

2026-10-10T00:00:00Z

Independently obtain the pinned public LakeIce arrays and official fold indices, and implement the following from scratch without KIWERB fitting/audit code or scores. Define exact profile identity from all 161 float32 feature bytes, not the class label; independently inspect the original PANGAEA CSV at its displayed precision, and confirm the five disjoint test folds partition all rows. For fold 0 use its test indices and all other rows for official training. Purged training removes every exact test-profile match. Control training removes the same class-specific numbers from other nonmatching training rows: use NumPy default_rng(0), eligible candidate indices in increasing original-row order, and one rng.choice(pool, size=count, replace=False) call per increasing class label. Fit all three arms with scikit-learn ExtraTreesClassifier(n_estimators=300, criterion="gini", max_features=1.0, min_samples_leaf=1, class_weight="balanced", random_state=10000); document Python/NumPy/sklearn versions and CPU threads. Score one first-occurring representative per exact test profile using mean class recall, plus ordinary row accuracy, per-class recall and confusion matrices. Record class counts, exact exposure and selected control removals; keep row-level predictions locally, submit their hashes, and arrange a separately rights-reviewed transfer before sharing files. Seal your own report before KIWERB comparison; disclose prior result/code exposure and deviations. This is one-fit method verification, not the full 25-pair effect or a new physical discovery.

Rechtezustand: approved_bounded_use. Modus: bounded_offline_research. Offline-Entwurf; keine automatische Rückgabe.

Independently audit conflicting Pedestrian profiles and trace historical sensor streams

2026-10-10T00:00:00Z

Research source-specific license, licensor scope and database-rights evidence for the blocked source labels: city_counts, city_historic_archive, city_locations, fold_0, fold_1, fold_2, fold_3, fold_4, monster_x, monster_y, zenodo_archive. Do not obtain, access or process the datasets while this catalog status remains blocked.

Rechtezustand: blocked_pending_source_rights. Modus: rights_evidence_only_no_dataset_access. Offline-Entwurf; keine automatische Rückgabe.

Check a theorem-derived answer and explicit witnesses for Mahmoud Question 5.4

2026-10-10T00:00:00Z

Independently transcribe the printed generators in Mahmoud's Theorem 5.3 and check the exact five- and six-qubit reduced-projector purity witnesses. Also evaluate a likely prior-theorem deduction: vectorize each rank-d stabilizer-code projector P as vec(P)/sqrt(d) on 2n qubits, use Pauli postselection to show it is a stabilizer state, and test whether a physical product LU maps these states by U tensor conjugate(U). Determine whether Claudet's pre-existing LU=LC theorem through 26 stabilizer-state qubits plus Mahmoud's LC-plus-permutation inequivalence already rules out an LU-plus-permutation map for the two printed pairs. Report any gap or counterexample, and distinguish this specific-pair answer from Mahmoud's broader question. KIWERB has an internally checked direct witness and a theorem-based inference; neither has outside specialist review. This is an educational verification fixture, not an open discovery or priority challenge. Do not use the unpublished KIWERB certificate as input to the numerical check.

Rechtezustand: approved_bounded_use. Modus: bounded_offline_research. Offline-Entwurf; keine automatische Rückgabe.

Verify Strindehag platinum contact-coating curve 10 against TPRC and EKHI

2026-10-10T00:00:00Z

Independently read TPRC volume 9 table 304 curve 10, including its printed specimen and measurement notes, and compare all four wavelength/reflectance pairs with the version-pinned EKHI record and the current live display. Record the exact values, mapping, any discrepancies, observation date and limits. This checks the handbook-to-database transfer only: do not infer a defect in the original experiment, a new physical result, or a database-wide error rate. Do not copy or upload the handbook scan or article.

Rechtezustand: approved_bounded_use. Modus: bounded_offline_research. Offline-Entwurf; keine automatische Rückgabe.