KLT-MATDES source update and calculation archive audit

CP2B-B3.8A · 30 September 2026 · LOGIC OF KURPISHEV-2

Ivan Borisovich Kurpishev / Курпишев Иван Борисович

1. Baseline and document updates

The update review is complete. KLT-MATDES continues from B3.7 and site version 27 before this change. RUN042, its chemistry source addendum and the newer Analyzer 1.05.5 branch are recorded. The scientific B3.8 objective remains open: no original pilot calculation protocols were obtained.

SourceAdoption decision
B3.7, 26 SeptemberResearch archive SHA-256 matches its checkpoint. Historical files are preserved.
RUN042, 27 SeptemberAll 36 synthetic tests rerun successfully; no real experiment.
RUN042-CHEM, 27 September35 integrity checks pass, including the final manifest; the pre-packaging report counted 34.
Analyzer, 29 SeptemberCurrent site version 55; Android 1.05.5 / Windows 1.05.05. Physical acceptance remains NOT_RUN.

The active compilation has 55,068,559 bytes and 2,197 verified sections. SHA-256: 782371776cf8d3dc9ec87cc4c8652614772ab67d4521d8388c5aa18473f23a0c. This run checks the bytes inside the recovered addendum. The separate current Drive TXT was not rehashed. The compilation is not claimed to recover the historical A7 attachment.

RU/EN/ZH B3.7 PDF copies have different sizes and SHA-256 values across the two stores. For each language, pdftotext -layout extraction is identical. Byte identity and visual equivalence are not asserted. Neither historical copy was replaced.

Review scope: metadata of 109 available files modified after 26 September 20:50 UTC, with selected checkpoints read in full. Drive folder responses contain at most 100 direct children. This is not an exhaustive review of every project file or conversation.

2. Evidence for the three calculation channels

The JVASP-114724 record was retrieved again on 30 September: HTTP 200, 1,828,715 bytes. SHA-256 bc8d103827a6926e74b1fa20279096fecf37f01c9e96731612a6496d04cb40e0 matches the B3.7 snapshot. Freshness is checked at retrieval time. An empty download_files field does not establish that the provider has no raw files.

Catalogue / channelBounded discovery result
SCF / #11 / 8258069 items; 67 parsed ranges; 2 unparsed titles.
TBmBJ / #12 / 8402033 items; 32 parsed ranges; 1 unparsed title.
Elasticity / #13 / 8218132 items; 30 parsed ranges; 2 unparsed titles.

The 134 metadata items contain 129 ranges accepted by the strict parser; none contains ID 114724. Five titles with typos, reversed endpoints or nonstandard spacing remain unresolved. The second pages of all three queries are empty. Multi-gigabyte archive contents and member lists were not inspected. This does not prove that the original files do not exist elsewhere.

When the historical conclusion is identifiable

h = g ∘ O ⇔ [O(p) = O(q) ⇒ h(p) = h(q)] for all p,q ∈ Π.

Π is the admissible protocol set, O the available record and h the requested historical attribute. Necessity follows by substitution. For sufficiency, define g on the image of O by the common h value on each fibre. Constancy makes this well defined. This is a standard factorization criterion, not a new physical law.

For the actual pilot, it has not been shown that the published record uniquely identifies the historical software version, ISPIN and extraction branch. Matching record bytes authenticate the copy; neither the chemistry compilation nor present source code substitutes for the original protocol.

3. Reproduction and continuation

audit_sources.py verifies captured-byte hashes, unique identifiers, the selected pagination and strict title ranges. It never silently repairs titles or grants scientific admission. All 12 tests pass. PUBLIC reproduces catalogue checks and explicitly skips private XML. RESEARCH additionally checks the fresh XML.

science=ABSTAIN; DHC_chem=OPEN-NOT-CONFIRMED; EXT-RUN=NOT_RUN; truth_layer_promotion=0; admission=BLOCKED; training=CLOSED.

T6-RA5=NOT_ISSUED; axiom_promotion=DISABLED; real_corpus_forecast=NOT_RUN; physical_reachability=NOT_ESTABLISHED; parameter_identification=NOT_PERFORMED. The 37 historical MATDES blockers remain; none is removed. RUN042 has a separate 12-blocker register, not an additive count.

Next, B3.8B: obtain #11 settings, eigenvalues, occupations and ISPIN; a separate MBJ protocol for #12; strains, stresses and the original tensor for #13. Establish pilot linkage, historical software/branch, provenance and applicable usage terms. Then run the unchanged B3.6 intake and the B3.7 extraction audit. B3.6 intake in this run: NOT_RUN_NO_ORIGINAL_PROTOCOLS.

No messages were sent to providers and no Analyzer installers were rebuilt. Tomes I-V, the Tome VI freeze and original RUN041/RUN042 remain unchanged. Site rollback: version 27, commit aa6829d6fbc5e902bd07be376839a94ba3ad307a.

Sources and reproduction

S1 · B3.7 checkpoint, 26.09.2026

S2 · RUN042 continuation, 27.09.2026

S3 · Active chemistry source

S4 · JARVIS official download catalogue

S5 · SCF project 82580

S6 · TBmBJ project 84020

S7 · Elasticity project 82181

The package includes registries, pinned catalogue captures, code, tests and reproduction commands. RESEARCH includes unchanged baseline ZIPs to verify the chemistry corpus and RUN042 offline. The separate site PUBLIC package excludes the full chemistry corpus and XML.