Prepare the reference chain.
Keep unmeasured bounds open.
An unexecuted calibration/preparation protocol now connects the twelve measurement requirements to witness records, timing and leakage feasibility.
Timing feasibility is the first gate.
The proposed 0.2 ns budget is split into four 50 ps shares: reference epoch, path skew, within-run variation and clock drift. The drift share over the 8.286 ms record implies about 6.03 parts per billion fractional clock error under the declared linear-drift model.
Repeated RMS jitter alone cannot supply the required deterministic bound. The protocol requires applicable reference uncertainty, before/after checks and justified stability across every displaced sampling interval.
A 1 ns timing alternative fits the mathematical total budget, but fails the current 0.2 ns per-role requirement. It remains a separately identified design alternative; no frozen contract is replaced.
A conditional tail witness has measurable targets.
The proposed reference envelope has three lines at −25, 0 and +25 Hz with magnitudes 1/16, 3/4 and 1/16 V. Its in-band spectral variation is 7/8 V; a 1/1000 V tail leaves an amplitude floor of 0.624 V under the declared model.
A finite-mode theorem gives tail variation ≤√K times the sum of measured complex RMS and coefficient error, plus outside-model variation. The illustrative 100-mode witness needs 60 µV measured RMS, 20 µV total coefficient error and 200 µV outside-model variation to reach 1 mV.
Mode completeness, orthogonality, stationarity and outside-mode contamination remain independent premises. Source command values do not certify a nonlinear simulator’s output.
Finite-mode witness theorem →Blank receipts cannot admit a run.
Each bound template separates its required ceiling from its unmeasured value. Review and witness templates preserve apparatus/run, epoch, channel, raw-file hashes, methods, uncertainty and dependencies. All twelve blank-bound software controls return HOLD.
Next extend the finite-mode witness to frequency-offset and Gram-matrix uncertainty, so a monitor’s clock and off-grid effects have explicit tolerances. Full energy, absolute native support, primitive state clock and action calibration remain open.
Blank review receipt ↓Blank method/witness record ↓Arithmetic and HOLD controls →Evidence-admission contract →