============================================================================================
INDUCTIVE-BIAS STUDY  (repeats=1; shared splits; symmetric nested-CV; gamma grid 12 pts in [1e-3,1e2])
============================================================================================
  running haberman ...
  running blood ...
  running diabetes ...
  running ilpd ...
  running breast_cancer ...
  running ionosphere ...
  running qsar_biodeg ...
  running sonar ...

Phase A -- per-dataset (acc: classical RBF | PQK | Fourier twin; KTA: RBF | PQK)
--------------------------------------------------------------------------------------------
dataset        qb     rbf     pqk    four  pqk-rbf       p  ktaRBF  ktaPQK  twinAln        g
haberman        2   0.699   0.716   0.716    0.017   0.106   0.059   0.037   1.0000    122.7
blood           2   0.759   0.758   0.758   -0.001   0.601   0.090   0.062   1.0000    799.1
diabetes        3   0.736   0.665   0.665   -0.070   0.995   0.153   0.038   1.0000     73.4
ilpd            2   0.705   0.705   0.705    0.000   0.496   0.104   0.096   1.0000    306.9
breast_cancer   2   0.939   0.859   0.859   -0.079   1.000   0.739   0.362   1.0000   2217.2
ionosphere      4   0.917   0.874   0.874   -0.043   0.976   0.319   0.315   1.0000     41.1
qsar_biodeg     2   0.766   0.672   0.672   -0.094   1.000   0.229   0.028   1.0000   2537.8
sonar           3   0.749   0.538   0.538   -0.210   0.999   0.176   0.072   1.0000      9.6

  classical KTA >= quantum KTA on 8/8 datasets; PQK==Fourier twin to 3.2e-15, twin-alignment min 1.0000

Phase B -- feature-map calibration (trig-structured labels at rising input frequency)
--------------------------------------------------------------------------------------------
base       freq     rbf     pqk    four  four-rbf  pqk-four  ktaPQK  ktaRBF
blood       0.5   0.977   0.977   0.977     0.000     0e+00   0.314   0.312
blood       1.0   0.963   0.982   0.982     0.020     0e+00   0.340   0.283
blood       2.0   0.925   0.982   0.982     0.057     0e+00   0.337   0.182
blood       4.0   0.835   0.975   0.975     0.140     0e+00   0.279   0.111
blood       8.0   0.660   0.903   0.903     0.243     0e+00   0.307   0.106
gaussian    0.5   0.975   0.972   0.972    -0.003     0e+00   0.401   0.397
gaussian    1.0   0.915   0.930   0.930     0.015     0e+00   0.233   0.223
gaussian    2.0   0.810   0.965   0.965     0.155     0e+00   0.271   0.102
gaussian    4.0   0.698   0.927   0.927     0.230     0e+00   0.291   0.066
gaussian    8.0   0.565   0.935   0.935     0.370     0e+00   0.275   0.054
  [blood] calibration PASS: feature-map advantage +0.000 (freq 0.5) -> +0.243 (freq 8.0); PQK==Fourier to 0e+00
  [gaussian] calibration PASS: feature-map advantage -0.003 (freq 0.5) -> +0.370 (freq 8.0); PQK==Fourier to 0e+00

Saved -> results/study_alignment.csv , results/study_control.csv
