Skip to content

DANDI 000351 raw-to-archived parity audit v0.1

Status: completed; retained structural failure (26 July 2026)

The protocol froze direct sample alignment before aggregate execution. All four draft assets matched their pinned sizes and SHA-256 digests and exposed the expected raw405,raw470 data, but none passed the timestamp/shape contract:

  • two standard-FP sessions had equal raw and archived lengths but accumulated approximately 0.23–0.26 seconds of clock difference;
  • WT-stGtACR and datHT-stGtACR sessions stored approximately 130-Hz raw data and 100-Hz archived dF/F, producing 85,515 and 79,888 fewer archived samples.

Consequently no numerical parity cases were executed, and all scientific gates remain failed rather than being computed after an unregistered alignment choice. The machine-readable result retains all four failures.

This is not evidence that the archived dF/F is incorrect. It establishes that the archive does not support index-wise raw-to-processed reproduction without an explicit time-alignment policy. Protocol v0.2 freezes timestamp interpolation as a separate follow-up rather than changing v0.1 after seeing the mismatch.