#!/usr/bin/env python3
"""Render recorded full-model numerical results, without modifying data."""
from pathlib import Path
import json
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import numpy as np

ROOT=Path(__file__).resolve().parents[1]
data=lambda n:json.loads((ROOT/'results'/f'{n}.json').read_text())
plt.rcParams.update({'font.family':'DejaVu Sans','font.size':9,
                     'axes.spines.top':False,'axes.spines.right':False})
fig,axes=plt.subplots(3,1,figsize=(6.3,7.3),constrained_layout=True)
nx=data('nexus')
for kind,label,color in [('coherent_difference','Difference port','#136d87'),
                         ('coherent_sum','Sum port','#9b425a')]:
    rows=[r for r in nx['rows'] if r['kind']==kind]
    # Only the main sequence: refinement runs carry their own kind labels.
    axes[0].loglog([r['epsilon'] for r in rows],[r['impedance'] for r in rows],
                   'o-',label=label,color=color,ms=4)
axes[0].axhline(467/120,ls='--',color='#9b425a',alpha=.6)
axes[0].invert_xaxis();axes[0].set_xlabel(r'$\varepsilon$')
axes[0].set_ylabel(r'$Z_{22}$');axes[0].set_title('Same kernel spectra; different fate')
axes[0].legend(fontsize=7)
lp=data('logperiodic')
rows=[r for r in lp['full_model_rows'] if abs(r['eta']-1/32)<1e-12 and r['cycle']==2]
rows=sorted(rows,key=lambda r:r['target_phase'])
xx=[r['target_phase'] for r in rows];yy=[r['scaled_capacity'] for r in rows]
zz=[r['limiting_constant'] for r in rows]
axes[1].plot(xx,zz,'o-',label='Scalar phase limit',color='#67499c',ms=4)
axes[1].plot(xx,yy,'x',label='Full model, cycle 2',color='#136d87',ms=6)
axes[1].set_xlabel(r'Logarithmic phase $\phi$');axes[1].set_ylabel(r'$\kappa_\varepsilon/\varepsilon$')
axes[1].set_title(r'Persistent phase, $\eta=1/32$');axes[1].legend(fontsize=7)
un=data('universality')['residual_gram_correction']['rows']
for v,c in [(.7,'#136d87'),(1.8,'#67499c')]:
    rows=[r for r in un if r['target_eigenvalue']==v]
    axes[2].semilogx([r['penalty'] for r in rows],
                     [r['scaled_defect_to_gram_ratio'] for r in rows],'o-',color=c,ms=4,label=f'Target v={v}')
axes[2].axhline(1,ls='--',color='#17243c',lw=1)
axes[2].set_xlabel(r'Drainage strength $\rho$')
axes[2].set_ylabel(r'$\rho(Y_\infty-Y_\rho)/W$')
axes[2].set_title('Sharp first correction');axes[2].legend(fontsize=7)
for ax in axes:ax.grid(alpha=.15)
out=ROOT/'figures';out.mkdir(exist_ok=True)
fig.savefig(out/'research_results.pdf');fig.savefig(out/'research_results.png',dpi=180)
print(out/'research_results.pdf')