M01 thin source and numerical-evidence package, delivery revision 3
============================================

This is a packaging-only revision; both previous packaging editions are preserved. No model, inputs, tests, scientific results,
failed-attempt records or acceptance thresholds were changed. The original
publication/m01-material3d-core.zip remains preserved by the project.

Start here for this download. The older LATEST-EDITION.txt and README.txt are
retained as exact historical source/evidence. Their old package-verifier command
refers to the original fuller delivery layout. For this thin edition use:

  python -m unittest -v test_fem3d.py
  python publication-core-v3/verify_package.py
  python run_study.py --output fresh-results
  python make_figures.py --results fresh-results --output fresh-figures

NumPy and SciPy are required for calculations; Matplotlib is required for figures.
Use a new output directory. These commands do not control physical hardware.

The thin core includes all declared inputs, solver/test/plot source, current
summary and per-run reports, all 18 response CSVs, test evidence, figure-data CSV,
figure provenance, and the first attempt's serialization-failure evidence. It
omits duplicate PNG figures, display JSON and the three full-volume NPZ files.
Both SVG figures are inside this core. The other unchanged files are separate downloads listed by exact path, size and SHA256
in EXTERNAL-ASSETS.json. Paths are relative to the material3d data root used on the
project page. Place those files at their listed relative paths to check all data:

  python publication-core-v3/verify_package.py --require-external

Omitted display/volume data can also be regenerated from the included inputs with
the run and plot commands above. A new run is a reproduction of the same matrix,
not a new physical trial or an independent experiment. Modal eigenvectors may
rotate within repeated eigenspaces across numerical-library versions.

The current results are genuine volumetric linear elasticity calculations, with
unresolved mesh and modal-response adequacy. All 3 first-six-frequency mesh
screens, all 3 whole-band mesh screens, all 12 retained-mode screens and 5 of 12
direct-versus-modal checks failed their original thresholds. The 3 low-frequency
mesh compliance and 3 global time/volume RMS stress screens passed. No hardness,
fracture, specimen damage, acoustic delivery or hardware capability is predicted.

This packaging revision performs zero matrix reruns and zero physical trials.
