# NORMALIZATION_REVIEW_INDEX.txt
# STELLAR_BOUNDARY_DYNAMICS_I
# Phase B — Alpha Normalization Review

PHASE:
B_supernova_light_curves

STAGE:
alpha_application / normalization_review

PURPOSE:
Review the first-pass alpha-application outputs before using Phase B 5D vectors
for A–B comparison.

This step protects the pipeline from over-interpreting raw slope/curvature
dominance in the first α run.

WHY THIS REVIEW EXISTS:
The first Phase B α-application completed successfully, but the resulting
anisotropic_persistence values were extremely large for some objects. This
indicates that one or more slope/curvature-derived features may dominate the
5D vector scale.

The review preserves all original v1 outputs, then produces bounded v2 vectors
for comparison.

INPUTS:
../grids/B_SN1987A_alpha_grid.csv
../grids/B_SN1993J_alpha_grid.csv
../grids/B_SN1999em_alpha_grid.csv
../grids/B_SN2011dh_alpha_grid.csv
../grids/B_SN2012aw_alpha_grid.csv
../grids/B_SN2013ej_alpha_grid.csv

../summaries/B_5D_VECTOR_SUMMARY.csv

SCRIPT:
../../tools/b_alpha_normalization_review.py

OUTPUTS:
B_ALPHA_NORMALIZATION_REVIEW.csv
B_5D_VECTOR_SUMMARY_v2.csv
B_NORMALIZATION_REVIEW_INTERPRETATION.txt

REVIEW TASKS:
1. Detect whether slope_response_norm or curvature_response_norm dominates
   structural_response.

2. Compare response channel means:
   gap_response_norm
   duration_response_norm
   slope_response_norm
   curvature_response_norm

3. Compute dominance ratios:
   slope_to_gap_ratio
   curvature_to_gap_ratio
   max_channel_to_gap_ratio

4. Bound anisotropic_persistence using a stable transform:
   anisotropic_persistence_bounded = anisotropic_persistence / (1 + anisotropic_persistence)

5. Recompute a comparable v2 vector summary that preserves:
   mean_GR
   var_GR
   admissibility_persistence
   collapse_onset_radius
   collapse_observed
   kappa_connect_reference
   tail_dominance_reference

6. Add review flags:
   slope_dominant
   curvature_dominant
   scale_review_needed
   high_tail_attention

INTERPRETATION RULE:
The v1 vectors remain valid as first-pass diagnostic outputs.
The v2 vectors are the safer comparison layer for A–B analysis.

The normalization review does not erase the original result. It records whether
the original α features were comparable enough for cross-object interpretation.

SPECIAL ATTENTION:
SN2012aw should remain marked as a special object because it showed:
κ_connect ≈ 4.4
tailDominance ≈ 0.329
FULL_PERCOLATION under STRUC-PERC-I, but high deformation sensitivity.

STATUS:
Initialized.
Ready for b_alpha_normalization_review.py.
