Files
CelerisLab/validation-evidence/2026-10-18/README.txt
T
Frank14f 3b9efd9696 test: retain matched inlet and bounded space-time diagnostic evidence
Evidence transport only after Drive blob-upload network failures.
Intact equilibrium comparator,shifted/unshifted tiny,and independent FP64 predictor receipts.
No source/default/gate changes;original scientific verdicts retained.
2026-10-19 02:51:36 +08:00

13 lines
1.3 KiB
Plaintext

CelerisLab reviewed scalar evidence,2026-10-18.
The two archives are intact originals. SHA256 and exact sizes are in transport_manifest.json.
Equilibrium comparator92da2e8:exit0,2400000steps healthy;ascending-zero v2 qualification passes. Frozen result.json remains periodic=false/accuracy_qualified=false;qualification_v2.json is separate and binds both original receipts.
Shifted/unshifted tiny e8b8c58:exit0,23methods pass,one N64 interval each mode only;no shifted full-N64 trajectory.
No production/default/dev/main change.
Both archives retain all their original member bytes,including gzip protection inventories. Metadata were scanned for credentials/private keys/access tokens and unrelated private paths. Only the authorized CelerisLab project/interpreter/system metadata and a reviewed literal official-reference fixture path were found. No credentials or unrelated private paths.
Original larger archives,all scientific receipts and any successful Drive transfers are retained. This single existing-validation-branch commit is evidence transport only.
CPU covered-predictor pair5468695:44tests pass;one FP64 raw/stress pair exit0;raw4samples exact to07879bc;both unchanged campaign/comparative/conservation screens pass,historical uniform pressure budget fails separately for both.No GPU pin or production/default/main/dev merge.