M02 independent computational review — prospective scope

This is a separate computational implementation within the same project, not
an outside laboratory, hardware test or external replication.

Before execution, review the fixed 3-eigensolve/36-direct-reference matrix,
parent fingerprints, inherited criteria, nested modal truncation semantics,
parent replay stop gate and resource enforcement. Preserve source reviews and
negative outcomes in separately named records. Do not edit author sources.

Before seeing outcomes, the selected independent new direct solution is:
structural steel, n=10, 100000 Hz, 1 N peak downward uniform top traction,
20 mm cube, bottom fixed, original Rayleigh damping. It is already one of the
36 planned probes. Retaining its full displacement is an evidence request,
not a new author solve. Independent reconstruction will use the prior reviewed
Cartesian tensor TET4 reference, never the author's FEM kernel.

After the author freezes completed or partial evidence, check every planned
identifier/status, all source and input hashes, exact CSV modal responses,
parent n8 replay, mesh and cutoff comparisons, direct-reference counters and
reported acceptance flags. Check the selected new direct solution with the
independent assembly. If it is unrun or unavailable, record that limitation;
do not replace it silently with a favorable result.

Fixed-144-mode mesh agreement still contains modal truncation. The separate
modal and direct-reference checks must remain visible. The direct sample
comparisons isolate spatial differences only at their named frequencies.
No sampled grid establishes a continuous-frequency maximum or convergence
to measured specimen behavior. Failed adequacy screens remain failed even
if all evidence-fidelity and equation-identity checks pass.

Windows hard commit limits, sampled working-set observations and controller
budget accounting must be distinguished. Tests/serialization count against
the execution clock; unknown or interrupted outcomes stay explicit. A sampled
memory observation is not an instantaneous peak or a hard resident-memory cap.

No physical trials, fracture model, new material selection, apparatus operation
or publication is part of this independent review. Portable public material
must exclude private absolute paths and administrative records.
