# Terminal uniform5 physical-cut diagnostic

Preregister before execution. Consume only completed, passing u5_dt00625
with sealed independent saved-step audits. Do not read the live fine case.
Use the UNMODIFIED prepared_motion/analyze.py physical_cut at x=240um,
y=0..160um, t=10us; the original function uses stage0/1/2 and native faces.
Reproduce both frozen uniform10 cut records numerically before processing
uniform5. Preserve the original uniform20/10 observations, including failures.
Compare the SAME dt=.0625us for dx20/10/5; the .03125us uniform10 reference
is reproduced as a control, not mixed into this resolution sequence.

Report signed liquid relative/material flux and surface-material transport,
and abs(recovered-ledger)/abs(ledger), with the inherited 1e-60 zero guard.
No new acceptance threshold, asymptotic rate fit, solver correction or C2.
This is a nonconforming constitutive reconstruction, not independent continuum
truth; stage-lag and finite moving-cell crossing differ. A smaller discrepancy
does not prove physical flux, traction, dynamic space/time convergence, or C1.
Do not infer a cause for any nonmonotonicity from this one saved step.

One Python consumer, 120s cap, no new native steps. Record all terminal raw
hashes, original source hashes and audit seals; retain all failures. White
Nature red/blue PNG and vector PDF, standalone readable Chinese HTML.
Original10/50/200us, physical traction, effective exchange coefficient,
thin-element thickness ratio, global fringes and heldout experiments stay open.