S3 Data and Code | SMM definitive manuscript | 16 October 2026

Run from the extracted ZIP's root:
  python code/spatial_prediction.py
  python code/verify_spatial_qif_k.py
  python code/build_figures.py

Dependencies: Python 3 with numpy, scipy, matplotlib. The original EEG frozen-code audit uses the dependencies stated by its historical implementation; it must not be confused with a 535-participant independent refit. All derived values in the main article are linked to sources/*.json and sources/amended_subject_results.csv.

Important provenance:
- Main positive effect: conditional seven-patch 56-state SMM-K *linearized model*; source-normalized astroglial forcing.
- Uniform-forcing invariance: exact within the homogeneous graph Laplacian model (numerical differences from roundoff).
- Negative EEG: original primary NOT EVALUABLE; separately documented POST_OPEN_AMENDED n=535 comparison M3<M2.
- Original nonlinear six-ion FEniCS R1/R2: not independently executed.
- archive_prior_work/ contains V8 and V9 scientific evidence bundles, not independent replication experiments.

No new license has been asserted in this release; author must confirm reuse terms and public stable release prior to formal PLOS submission.
