A readout threshold is trusted only after low, equality, and high signals are all checked by the same rule. Exact arithmetic here means exact results for the stated model inputs; measured inputs still carry uncertainty and significant-figure limits.
highlighted = computed this step
The boundary scan starts from calibrated endpoints
The calibrated zero signal is 2, the threshold is 5, and the calibrated one signal is 8.
szero=2,T=5,sone=8
Three signals expose low, boundary, and high decisions
The scan checks a low endpoint, the exact boundary, and a high endpoint. The result column is the helper decision, not a label chosen after the fact.
szero222s258T555sone888resultzerorepeatone
Equality repeats instead of guessing a state
For the rendered row, the signal equals the threshold at 5. The checked result is repeat, so no zero or one decision is invented.
5=5⇒repeat
A high endpoint is the only one-decision in the scan
The high row uses signal 8, which is above threshold 5. That row reads one.
8>5⇒∣1⟩
The boundary model is not a noise model
This audit claims a finite threshold rule only. It does not claim detector noise, stochastic trajectories, or hardware drift.