dsfb-rf 1.0.1

DSFB-RF Structural Semiotics Engine for RF Signal Monitoring - A Deterministic, Non-Intrusive Observer Layer for Typed Structural Interpretation of IQ Residual Streams in Electronic Warfare, Spectrum Monitoring, and Cognitive Radio
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
#!/usr/bin/env python3
"""
DSFB-RF Publication Figure Generator
=====================================
Reads  paper/figure_data.json  (produced by  cargo run --features std,serde
--example generate_figures) and renders 20 publication-quality figures for the
paper and the elite panel briefing.

Usage:
    cd paper
    python3 figures.py                    # all 20 figures
    python3 figures.py --fig 2 8 13       # specific figures only
    python3 figures.py --dpi 300          # print-resolution (default 150)

Output: paper/figs/fig_XX_*.pdf  +  paper/figs/fig_XX_*.png
"""

# ─── stdlib / third-party ─────────────────────────────────────────────────
import argparse
import json
import math
import os
import sys
from pathlib import Path

import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
import matplotlib.patheffects as pe
import matplotlib.ticker as ticker
from matplotlib.colors import ListedColormap, LinearSegmentedColormap
from matplotlib.gridspec import GridSpec
from matplotlib.lines import Line2D
import matplotlib.transforms as transforms
import numpy as np
from mpl_toolkits.mplot3d import Axes3D  # noqa: F401

# ─── global style (IEEE-friendly) ─────────────────────────────────────────
plt.rcParams.update({
    "font.family":        "serif",
    "font.serif":         ["Times New Roman", "DejaVu Serif"],
    "font.size":          9,
    "axes.titlesize":     9,
    "axes.labelsize":     9,
    "legend.fontsize":    8,
    "xtick.labelsize":    8,
    "ytick.labelsize":    8,
    "axes.linewidth":     0.8,
    "lines.linewidth":    1.2,
    "grid.linewidth":     0.4,
    "grid.alpha":         0.4,
    "axes.grid":          True,
    "figure.dpi":         150,
    "savefig.dpi":        150,
    "savefig.bbox":       "tight",
    "savefig.pad_inches": 0.04,
    "text.usetex":        False,
})

# Palette — colour-blind safe + IEEE-print compatible
C_ADMISSIBLE  = "#2ca02c"   # green
C_BOUNDARY    = "#ff7f0e"   # orange
C_VIOLATION   = "#d62728"   # red
C_DSFB        = "#1f77b4"   # blue
C_COMPARATOR  = "#9467bd"   # purple
C_NEUTRAL     = "#7f7f7f"   # grey
C_HIGHLIGHT   = "#e377c2"   # pink

GRAMMAR_COLOR = {
    "Admissible":                   C_ADMISSIBLE,
    "Boundary(SustainedOutwardDrift)": C_BOUNDARY,
    "Boundary(AbruptSlewViolation)":   C_BOUNDARY,
    "Boundary(RecurrentBoundaryGrazing)": C_BOUNDARY,
    "Violation":                    C_VIOLATION,
}


def grammar_color(state: str) -> str:
    for k, v in GRAMMAR_COLOR.items():
        if k in state:
            return v
    return C_NEUTRAL


def save(fig, idx: int, name: str, out_dir: Path, dpi: int = 150):
    out_dir.mkdir(parents=True, exist_ok=True)
    stub = f"fig_{idx:02d}_{name}"
    for ext in ("pdf", "png"):
        path = out_dir / f"{stub}.{ext}"
        fig.savefig(path, dpi=dpi, format=ext)
    print(f"  Saved fig {idx:02d}: {stub}.*")
    plt.close(fig)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 1 — Semiotic Manifold Partition
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig01(d, out_dir, dpi):
    pts  = d["fig01_semiotic_manifold"]["points"]
    norm  = np.array([p["norm"]  for p in pts], dtype=float)
    drift = np.array([p["drift"] for p in pts], dtype=float)
    slew  = np.array([p["slew"]  for p in pts], dtype=float)
    regime = [p["regime"] for p in pts]
    colors = [grammar_color(r) for r in regime]
    labels = sorted(set(regime), key=lambda x: ["Admissible","Boundary","Violation"].index(x) if x in ["Admissible","Boundary","Violation"] else 3)

    fig = plt.figure(figsize=(4.5, 3.8))
    ax  = fig.add_subplot(111, projection="3d")

    for lab in labels:
        mask = np.array([r == lab for r in regime])
        ax.scatter(norm[mask], drift[mask], slew[mask],
                   c=grammar_color(lab), s=14, alpha=0.80,
                   label=lab, depthshade=False)

    # Envelope plane (ρ = 0.10)
    rho = 0.10
    _s = np.linspace(-0.015, 0.015, 4)
    _d = np.linspace(-0.015, 0.015, 4)
    S, Dr = np.meshgrid(_s, _d)
    R = np.full_like(S, rho)
    ax.plot_surface(R, Dr, S, alpha=0.12, color=C_VIOLATION, zorder=0)

    ax.set_xlabel("‖r(k)‖", labelpad=4)
    ax.set_ylabel("ṙ(k) drift", labelpad=4)
    ax.set_zlabel("r̈(k) slew", labelpad=4)
    ax.set_title("Fig. 1 — Semiotic Manifold $\\mathcal{M}_{\\mathrm{sem}}$\nGrammar Partition in $(\\|r\\|, \\dot r, \\ddot r)$ space")
    ax.legend(loc="upper left", framealpha=0.7)
    fig.tight_layout()
    save(fig, 1, "semiotic_manifold", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 2 — Review Surface Compression
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig02(d, out_dir, dpi):
    datasets = d["fig02_compression"]["datasets"]
    fig, axes = plt.subplots(1, 2, figsize=(7.0, 3.5))

    for ax, ds in zip(axes, datasets):
        comps = ds["comparators"]
        names     = [c["name"].replace("(","(\n") for c in comps]
        precision = np.array([c["precision"] for c in comps])
        episodes  = np.array([c["episodes"]  for c in comps])

        colors = [C_DSFB if "DSFB" in n else C_NEUTRAL for n in names]

        x = np.arange(len(names))
        bars = ax.bar(x, precision, color=colors, edgecolor="black", linewidth=0.6, zorder=3)

        # Annotate episode count
        for bar, ep in zip(bars, episodes):
            ax.text(bar.get_x() + bar.get_width()/2, bar.get_height() + 0.01,
                    f"{ep:,}", ha="center", va="bottom", fontsize=6.5, rotation=0)

        ax.set_xticks(x)
        ax.set_xticklabels(names, fontsize=6.5)
        ax.set_ylim(0, 1.05)
        ax.set_ylabel("Episode Precision")
        label = ds["dataset"][:30]
        ax.set_title(f"Fig. 2 — {label}\n"
                     f"Compression {ds['compression_ratio']:.0f}× | "
                     f"Precision gain {ds['precision_gain']:.1f}× | Recall {ds['recall']*100:.1f}%",
                     fontsize=7.5)
        ax.axhline(0.73, color=C_DSFB, lw=0.8, ls="--", label="DSFB precision")
        ax.grid(axis="y", zorder=0)

    fig.suptitle("Review Surface Compression — DSFB vs. Scalar Comparators\n"
                 "(RadioML 2018.01a synthetic  +  ORACLE real USRP B200)",
                 fontsize=9, fontweight="bold")
    fig.tight_layout(rect=[0, 0, 1, 0.90])
    save(fig, 2, "compression_comparison", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 3 — Observer-of-Observer Structural Blindspot (Theorem OoT)
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig03(d, out_dir, dpi):
    oot    = d["fig03_oot_blindspot"]
    traj   = np.array(oot["trajectory"])
    l_alarm = np.array(oot["luenberger_alarm"])
    grammar = oot["dsfb_grammar"]
    k = np.arange(len(traj))

    fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(5.5, 3.8), sharex=True)

    # Top: trajectory + Luenberger threshold
    ax1.plot(k, traj, color=C_DSFB, lw=1.3, label="‖r(k)‖")
    ax1.axhline(0.10, color=C_VIOLATION, lw=1.0, ls="--",
                label="Luenberger threshold δ = 0.10")
    ax1.axhline(0.005, color=C_NEUTRAL, lw=0.7, ls=":",
                label="‖L·r(k)‖ = 0.5·‖r(k)‖")
    # Shade DSFB detection region
    first_boundary = next((i for i, g in enumerate(grammar)
                           if "Boundary" in g or "Violation" in g), len(k))
    ax1.axvspan(first_boundary, len(k), color=C_BOUNDARY, alpha=0.12,
                label=f"DSFB Boundary detected (k={first_boundary})")
    ax1.set_ylabel("Residual Norm")
    ax1.legend(fontsize=7, ncol=2)
    ax1.set_title("Fig. 3 — Observer-of-Observer Structural Blindspot (Theorem OoT)\n"
                  "$\\mathcal{T}_{\\mathrm{blind}}$: small ‖r‖ but persistent $\\dot r>0$")

    # Bottom: Grammar state timeline
    colors = [grammar_color(g) for g in grammar]
    for i, (g, c) in enumerate(zip(grammar, colors)):
        ax2.barh(0, 1, left=i, height=0.7, color=c, edgecolor="none")
    patch_A = mpatches.Patch(color=C_ADMISSIBLE, label="Admissible")
    patch_B = mpatches.Patch(color=C_BOUNDARY,   label="Boundary")
    patch_V = mpatches.Patch(color=C_VIOLATION,  label="Violation")
    ax2.legend(handles=[patch_A, patch_B, patch_V], loc="lower right",
               fontsize=7, ncol=3)
    ax2.set_yticks([])
    ax2.set_xlabel("Observation k")
    ax2.set_ylabel("Grammar")
    ax2.set_xlim(0, len(k))

    fig.tight_layout()
    save(fig, 3, "oot_blindspot", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 4 — DSFB Pipeline DAG
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig04(d, out_dir, dpi):
    stages = d["fig04_pipeline"]["stages"]
    edges  = d["fig04_pipeline"]["edges"]

    fig, ax = plt.subplots(figsize=(6.5, 2.5))
    ax.set_xlim(-0.5, len(stages) - 0.5)
    ax.set_ylim(-1, 1.4)
    ax.axis("off")
    ax.set_title("Fig. 4 — DSFB-RF Deterministic Pipeline\n"
                 "(no_std · no_alloc · zero unsafe — compiler-enforced at every stage)",
                 fontsize=8.5)

    box_colors = ["#aec7e8", "#c5b0d5", "#98df8a", "#ffbb78",
                  "#c49c94", "#f7b6d2", "#dbdb8d"]
    w, h = 0.82, 0.48

    for s in stages:
        x = s["id"]
        col = box_colors[s["id"] % len(box_colors)]
        rect = mpatches.FancyBboxPatch((x - w/2, -h/2), w, h,
            boxstyle="round,pad=0.04", facecolor=col,
            edgecolor="#333333", linewidth=0.8, zorder=3)
        ax.add_patch(rect)
        ax.text(x, 0.06, s["name"], ha="center", va="center",
                fontsize=7.5, fontweight="bold", zorder=4)
        ax.text(x, -0.12, s["module"], ha="center", va="center",
                fontsize=6, color="#444444", zorder=4)
        ax.text(x, -0.28, s["output_type"], ha="center", va="center",
                fontsize=6.5, color="#222222",
                fontstyle="italic", zorder=4)
        ax.text(x, 0.44, s["theorem"], ha="center", va="center",
                fontsize=6, color="#666666", zorder=4)

    for (a, b) in edges:
        x1, x2 = a + w/2 + 0.01, b - w/2 - 0.01
        ax.annotate("", xy=(x2, 0), xytext=(x1, 0),
                    arrowprops=dict(arrowstyle="->", color="#333333",
                                   lw=1.1), zorder=2)

    fig.tight_layout()
    save(fig, 4, "pipeline_dag", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 5 — Finite-Time Envelope Exit (Theorem 1)
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig05(d, out_dir, dpi):
    data  = d["fig05_envelope_exit"]
    rho   = data["rho"]
    curves = data["curves"]

    fig, ax = plt.subplots(figsize=(4.5, 3.0))

    cmap = plt.cm.get_cmap("coolwarm", len(curves))
    n_obs = len(curves[0]["trajectory"])
    k = np.arange(n_obs)

    for i, c in enumerate(curves):
        traj = np.array(c["trajectory"])
        k_star = c["k_star"]
        col = cmap(i / (len(curves) - 1))
        ax.plot(k, traj, color=col, lw=1.2, label=c["label"])
        # Mark theoretical k*
        ks_int = int(min(k_star, n_obs - 1))
        if ks_int < n_obs:
            ax.axvline(ks_int, color=col, lw=0.6, ls=":", alpha=0.7)

    ax.axhline(rho, color=C_VIOLATION, lw=1.4, ls="--",
               label=f"Envelope boundary ρ = {rho}")
    ax.fill_between(k, rho, ax.get_ylim()[1] if ax.get_ylim()[1] > rho else rho * 1.15,
                    alpha=0.08, color=C_VIOLATION)

    ax.set_xlabel("Observation k")
    ax.set_ylabel("Residual norm ‖r(k)‖")
    ax.set_ylim(-0.005, rho * 1.20)
    ax.legend(fontsize=7, loc="upper left", ncol=1)
    ax.set_title("Fig. 5 — Finite-Time Envelope Exit Bound (Theorem 1)\n"
                 "$k^* \\leq \\rho/\\alpha$ — computable without a noise model")
    fig.tight_layout()
    save(fig, 5, "envelope_exit_theorem1", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 6 — Lyapunov Exponent Time Series
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig06(d, out_dir, dpi):
    lp = d["fig06_lyapunov"]
    k       = np.array(lp["k"])
    lam     = np.array(lp["lambda"])
    grammar = lp["grammar"]
    stability = lp["stability"]

    fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(5.5, 3.8), sharex=True)

    # Top: λ(k)
    ax1.plot(k, lam, color=C_DSFB, lw=1.2, label="λ(k) — FTLE")
    ax1.axhline(0.01, color=C_VIOLATION, lw=0.9, ls="--",
                label="λ_crit = 0.01 (ExponentialDivergence)")
    ax1.axhline(0.001, color=C_BOUNDARY, lw=0.7, ls=":",
                label="ε = 0.001 (MarginalDivergence)")
    ax1.axhline(0.0, color=C_NEUTRAL, lw=0.6, ls="-", alpha=0.5)
    ax1.fill_between(k, lam, 0, where=(lam > 0.01),
                     color=C_VIOLATION, alpha=0.18,
                     label="Exponential divergence zone")
    ax1.fill_between(k, lam, 0, where=((lam > 0.001) & (lam <= 0.01)),
                     color=C_BOUNDARY, alpha=0.14,
                     label="Marginal divergence zone")
    ax1.set_ylabel("Lyapunov exp. λ(k)")
    ax1.legend(fontsize=6.5, ncol=2)
    ax1.set_title("Fig. 6 — Finite-Time Lyapunov Exponent vs. Grammar State\n"
                  "Healthy (λ<0) → slow drift (λ>0) → jamming onset (λ spike) → recovery")

    # Bottom: Grammar timeline
    for i, g in enumerate(grammar):
        ax2.barh(0, 1, left=i, height=0.7, color=grammar_color(g), edgecolor="none")
    patch_A = mpatches.Patch(color=C_ADMISSIBLE, label="Admissible (λ<0)")
    patch_B = mpatches.Patch(color=C_BOUNDARY,   label="Boundary (λ>0)")
    patch_V = mpatches.Patch(color=C_VIOLATION,  label="Violation")
    ax2.legend(handles=[patch_A, patch_B, patch_V], loc="lower right",
               fontsize=7, ncol=3)
    ax2.set_yticks([])
    ax2.set_xlabel("Observation k")
    ax2.set_ylabel("Grammar")
    ax2.set_xlim(0, len(k))

    # Regime labels
    for label, start, end in [("Healthy\n(λ < 0)", 0, 30),
                               ("Thermal drift\n(λ > 0)", 30, 70),
                               ("Jamming\nonset", 70, 95),
                               ("Recovery", 95, len(k))]:
        ax1.text((start + end) / 2, ax1.get_ylim()[1] * 0.88, label,
                 ha="center", fontsize=6.5, color="#333333",
                 bbox=dict(boxstyle="round,pad=0.15", facecolor="white", alpha=0.7))

    fig.tight_layout()
    save(fig, 6, "lyapunov_time_series", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 7 — GUM Uncertainty Budget Waterfall
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig07(d, out_dir, dpi):
    g = d["fig07_gum"]
    contribs = g["contributors"]

    fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(7.4, 3.2), gridspec_kw={"width_ratios": [2, 1]})

    # Left: variance budget pie
    names  = [c["name"] for c in contribs]
    values = np.array([c["value"] ** 2 for c in contribs])  # variance
    fracs  = values / values.sum()
    type_colors = [C_DSFB if c["kind"] == "A" else C_BOUNDARY for c in contribs]

    wedges, texts, autotexts = ax1.pie(
        fracs, labels=names, autopct="%1.1f%%",
        colors=type_colors + [C_VIOLATION, C_NEUTRAL, C_HIGHLIGHT],
        startangle=140, textprops={"fontsize": 7}
    )
    for at in autotexts:
        at.set_fontsize(7)
    ax1.set_title("Variance Budget\n(Type A + Type B contributors)")

    # Right: waterfall — u_A, u_B, u_c, U=k·u_c, ρ_GUM
    summaries = [
        ("$u_A$ (Type A)", g["u_a"],         C_DSFB),
        ("$u_B$ (Type B)", g["u_b_combined"], C_BOUNDARY),
        ("$u_c$ combined", g["u_c"],          C_NEUTRAL),
        (f"$U = k{g['coverage_k']:.0f}{chr(183)}u_c$", g["expanded_u"], C_VIOLATION),
        ("$\\rho_{\\mathrm{GUM}} = \\mu + U$",        g["rho_gum"],    C_ADMISSIBLE),
    ]
    labels = [s[0] for s in summaries]
    vals   = [s[1] for s in summaries]
    cols   = [s[2] for s in summaries]

    bars = ax2.barh(range(len(labels)), vals, color=cols,
                    edgecolor="#333333", linewidth=0.6)
    ax2.set_yticks(range(len(labels)))
    ax2.set_yticklabels(labels, fontsize=7.5)
    ax2.set_xlabel("Norm units (‖r‖)")
    ax2.set_title(f"GUM Budget Summary\n"
                  f"ρ_GUM = {g['rho_gum']:.4f}  (k={g['coverage_k']:.0f}, WSS-verified)")
    for i, (bar, val) in enumerate(zip(bars, vals)):
        ax2.text(val + max(vals) * 0.02, i, f"{val:.5f}",
                 va="center", fontsize=7)
    ax2.invert_yaxis()
    ax2.grid(axis="x")

    fig.suptitle("Fig. 7 — GUM/JCGM 100:2008 Uncertainty Budget for Admissibility Envelope\n"
                 "Type A (statistical) + Type B (NF, ADC quantisation, thermal, LO) contributors",
                 fontweight="bold", fontsize=8.5)
    fig.tight_layout(rect=[0, 0, 1, 0.88])
    save(fig, 7, "gum_uncertainty_budget", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 8 — Semiotic Horizon Detection Heatmap
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig08(d, out_dir, dpi):
    hz  = d["fig08_horizon"]
    snr = np.array(hz["snr_levels"])
    alp = np.array(hz["alpha_levels"])
    det = np.array(hz["detection_rate"])    # shape (SNR, alpha)

    fig, ax = plt.subplots(figsize=(5.5, 3.8))

    cmap = LinearSegmentedColormap.from_list(
        "horizon", [(0.0, C_VIOLATION), (0.5, "#ffdd88"), (1.0, C_ADMISSIBLE)])
    im = ax.imshow(det, origin="lower", aspect="auto", cmap=cmap,
                   vmin=0.0, vmax=1.0,
                   extent=[np.log10(alp[0]), np.log10(alp[-1]),
                            snr[0], snr[-1]])

    cbar = fig.colorbar(im, ax=ax, fraction=0.046, pad=0.04)
    cbar.set_label("Detection rate (0=miss, 1=detect)", fontsize=8)

    ax.set_xlabel("log₁₀(drift rate α)")
    ax.set_ylabel("SNR (dB)")
    ax.axhline(-10, color="white", lw=1.2, ls="--", label="SNR floor (−10 dB)")
    ax.legend(fontsize=7, loc="upper left")
    ax.set_title("Fig. 8 — Semiotic Horizon: Detection Surface in (SNR, α) space\n"
                 "Zone of Success (green) / Zone of Failure (red) — maps CRLB-limited boundary")

    # Annotate failure zone
    ax.text(np.log10(alp[0]) * 0.98, -14, "Sub-SNR floor:\nforced Admissible\n(L10 non-claim honored)",
            fontsize=6.5, color="white", ha="left", va="bottom",
            bbox=dict(boxstyle="round", facecolor="#333333", alpha=0.6))

    fig.tight_layout()
    save(fig, 8, "semiotic_horizon_heatmap", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 9 — Physics-of-Failure Mapping
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig09(d, out_dir, dpi):
    phys = d["fig09_physics"]
    nodes = phys["nodes"]
    edges = phys["edges"]

    fig, ax = plt.subplots(figsize=(7.0, 3.6))
    ax.axis("off")
    ax.set_title("Fig. 9 — Physics-of-Failure Map: Grammar State → Candidate Physical Mechanism\n"
                 "(candidate hypotheses — physical attribution requires deployment calibration data)",
                 fontsize=8.5)

    # Grammar nodes: left column; mechanism nodes: right column
    grammar_nodes = [n for n in nodes if n["kind"] == "grammar"]
    mech_nodes    = [n for n in nodes if n["kind"] == "mechanism"]

    gx, mx = 0.14, 0.72
    n_g, n_m = len(grammar_nodes), len(mech_nodes)

    gy = np.linspace(0.85, 0.15, n_g)
    my = np.linspace(0.90, 0.10, n_m)

    pos = {}
    for i, n in enumerate(grammar_nodes):
        pos[n["id"]] = (gx, gy[i])
    for i, n in enumerate(mech_nodes):
        pos[n["id"]] = (mx, my[i])

    # Edges (gradient width by weight)
    for e in edges:
        x0, y0 = pos[e["from"]]
        x1, y1 = pos[e["to"]]
        ax.plot([x0 + 0.09, x1 - 0.12], [y0, y1],
                color="#888888", lw=1.0 + 2.5 * e["weight"],
                alpha=0.35, solid_capstyle="round")
        mx_pos = (x0 + 0.09 + x1 - 0.12) / 2
        my_pos = (y0 + y1) / 2
        ax.text(mx_pos, my_pos, f"{e['weight']:.2f}",
                ha="center", va="center", fontsize=5.5, color="#444444")

    # Grammar node boxes
    GRAMMAR_BOX_COLORS = {
        "grammar":   "#c5b0d5",
        "mechanism": "#aec7e8",
    }
    for n in nodes:
        x, y = pos[n["id"]]
        w, h = (0.20, 0.07) if n["kind"] == "grammar" else (0.22, 0.08)
        rect = mpatches.FancyBboxPatch((x - w/2, y - h/2), w, h,
            boxstyle="round,pad=0.015",
            facecolor=GRAMMAR_BOX_COLORS[n["kind"]],
            edgecolor="#333333", linewidth=0.7, zorder=3,
            transform=ax.transAxes, clip_on=False)
        ax.add_patch(rect)
        ax.text(x, y, n["label"], ha="center", va="center",
                fontsize=6.5, fontweight="bold" if n["kind"] == "grammar" else "normal",
                transform=ax.transAxes, zorder=4, wrap=True)

    # Column labels
    ax.text(gx, 0.97, "Grammar States", ha="center", fontsize=8, fontweight="bold",
            transform=ax.transAxes, color="#333333")
    ax.text(mx, 0.97, "Candidate Physical Mechanisms", ha="center", fontsize=8, fontweight="bold",
            transform=ax.transAxes, color="#333333")

    for x in [gx, mx]:
        ax.plot([x, x], [0.02, 0.94], transform=ax.transAxes,
                color="#cccccc", lw=0.6, ls="--", zorder=0, clip_on=False)

    fig.tight_layout()
    save(fig, 9, "physics_failure_mapping", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 10 — DSA Score Build-Up Time Series
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig10(d, out_dir, dpi):
    ds = d["fig10_dsa"]
    k       = np.array(ds["k"])
    dsa     = np.array(ds["dsa_score"])
    tau     = ds["tau"]
    grammar = ds["grammar"]

    fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(5.5, 3.8), sharex=True,
                                   gridspec_kw={"height_ratios": [3, 1]})

    # DSA composite score
    ax1.plot(k, dsa, color=C_DSFB, lw=1.4, label="DSA(k) composite")
    ax1.axhline(tau, color=C_VIOLATION, lw=1.0, ls="--",
                label=f"Threshold τ = {tau}")
    ax1.fill_between(k, dsa, tau, where=(dsa >= tau),
                     color=C_VIOLATION, alpha=0.20, label="DSA ≥ τ (Policy active)")
    ax1.fill_between(k, 0, dsa, where=(dsa < tau),
                     color=C_DSFB, alpha=0.10)
    ax1.set_ylabel("DSA Score")
    ax1.legend(fontsize=7.5)
    ax1.set_title("Fig. 10 — Deterministic Structural Accumulator (DSA) Score\n"
                  "DSA(k) = w₁·b + w₂·d + w₃·s + w₄·e + w₅·μ  →  Policy escalation")

    # Grammar timeline
    for i, g in enumerate(grammar):
        ax2.barh(0, 1, left=i, height=0.7, color=grammar_color(g), edgecolor="none")
    ax2.set_yticks([])
    ax2.set_xlabel("Observation k")
    ax2.set_ylabel("Grammar")
    ax2.set_xlim(0, len(k))
    patch_A = mpatches.Patch(color=C_ADMISSIBLE, label="Admissible")
    patch_B = mpatches.Patch(color=C_BOUNDARY, label="Boundary")
    patch_V = mpatches.Patch(color=C_VIOLATION, label="Violation")
    ax2.legend(handles=[patch_A, patch_B, patch_V], fontsize=7, ncol=3,
               loc="lower right")

    fig.tight_layout()
    save(fig, 10, "dsa_score_buildup", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 11 — Competitive Differentiation Matrix
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig11(d, out_dir, dpi):
    cm = d["fig11_competitive"]
    methods = cm["methods"]
    caps    = cm["capabilities"]
    mat     = np.array(cm["matrix"], dtype=float)   # rows=capabilities, cols=methods

    fig, ax = plt.subplots(figsize=(7.0, 3.8))

    # Color map: 0=white/red, 1=green, 2=partial/orange
    cell_colors = []
    for row in mat:
        color_row = []
        for v in row:
            if v == 1:
                color_row.append(C_ADMISSIBLE)
            elif v == 2:
                color_row.append(C_BOUNDARY)
            else:
                color_row.append("#eeeeee")
        cell_colors.append(color_row)

    n_cap, n_meth = mat.shape
    for i in range(n_cap):
        for j in range(n_meth):
            rect = mpatches.FancyBboxPatch(
                (j + 0.05, n_cap - 1 - i + 0.05), 0.90, 0.90,
                boxstyle="round,pad=0.02",
                facecolor=cell_colors[i][j], edgecolor="#888888",
                linewidth=0.5, zorder=2)
            ax.add_patch(rect)
            sym = "" if mat[i, j] == 1 else ("" if mat[i, j] == 2 else "")
            ax.text(j + 0.5, n_cap - 1 - i + 0.5, sym,
                    ha="center", va="center",
                    fontsize=10 if mat[i, j] > 0 else 9,
                    color="white" if mat[i, j] == 1 else
                          "#333333" if mat[i, j] == 2 else "#aaaaaa",
                    fontweight="bold" if mat[i, j] == 1 else "normal")

    ax.set_xticks([j + 0.5 for j in range(n_meth)])
    ax.set_xticklabels(methods, fontsize=7.5)
    ax.set_yticks([n_cap - 1 - i + 0.5 for i in range(n_cap)])
    ax.set_yticklabels(caps, fontsize=7.5)
    ax.set_xlim(0, n_meth)
    ax.set_ylim(0, n_cap)
    ax.grid(False)
    ax.set_title("Fig. 11 — Competitive Differentiation Matrix\n"
                 "✓=provided · ○=partial · –=absent   (DSFB=last column, rightmost)",
                 fontsize=8.5)

    # Highlight DSFB column
    rect_h = mpatches.FancyBboxPatch(
        (n_meth - 1, 0), 1, n_cap,
        boxstyle="round,pad=0.0",
        facecolor="none", edgecolor=C_DSFB,
        linewidth=2.0, zorder=5)
    ax.add_patch(rect_h)

    legend_elements = [
        mpatches.Patch(facecolor=C_ADMISSIBLE, label="✓ Provided"),
        mpatches.Patch(facecolor=C_BOUNDARY,   label="○ Partial"),
        mpatches.Patch(facecolor="#eeeeee",     edgecolor="#888888", label="– Absent"),
        Line2D([0],[0], color=C_DSFB, lw=2.0, label="DSFB (this work)"),
    ]
    ax.legend(handles=legend_elements, loc="lower right", fontsize=7)
    ax.axis("on")
    for spine in ax.spines.values():
        spine.set_linewidth(0.6)

    fig.tight_layout()
    save(fig, 11, "competitive_matrix", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 12 — WSS Pre-condition Verification
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig12(d, out_dir, dpi):
    wss = d["fig12_wss"]["scenarios"]
    n_sc = len(wss)

    fig, axes = plt.subplots(2, n_sc, figsize=(8.0, 4.0),
                              gridspec_kw={"height_ratios": [2, 1]})

    for j, sc in enumerate(wss):
        norms = np.array(sc["norms"])
        k     = np.arange(len(norms))
        col   = C_ADMISSIBLE if sc["is_wss"] else C_VIOLATION

        ax_main = axes[0, j] if n_sc > 1 else axes[0]
        ax_bar  = axes[1, j] if n_sc > 1 else axes[1]

        ax_main.plot(k, norms, color=col, lw=1.1)
        ax_main.set_title(sc["name"].replace("(", "\n("), fontsize=7.5)
        ax_main.set_ylabel("‖r(k)‖" if j == 0 else "")
        ax_main.set_xlabel("k")
        verdict = "WSS: PASS" if sc["is_wss"] else "WSS: FAIL"
        ax_main.text(0.97, 0.94, verdict, transform=ax_main.transAxes,
                     ha="right", va="top", fontsize=8, fontweight="bold",
                     color=col,
                     bbox=dict(boxstyle="round,pad=0.2", facecolor="white", alpha=0.8))

        # Metric bar
        metrics = ["Mean dev.", "Var. dev.", "|lag-1|"]
        vals    = [sc["mean_deviation"], sc["variance_deviation"], abs(sc["lag1_autocorr"])]
        thresholds = [0.20, 0.50, 0.70]
        bar_colors = [C_VIOLATION if v > t else C_ADMISSIBLE
                      for v, t in zip(vals, thresholds)]
        ax_bar.bar(metrics, vals, color=bar_colors, edgecolor="#333333", lw=0.5)
        for t_val, metric in zip(thresholds, metrics):
            x_pos = metrics.index(metric)
            ax_bar.hlines(t_val, x_pos - 0.4, x_pos + 0.4,
                          color="#333333", lw=1.0, ls="--")
        ax_bar.set_ylim(0, 1.05)
        ax_bar.set_ylabel("Metric" if j == 0 else "")
        ax_bar.tick_params(labelsize=6.5)

    fig.suptitle("Fig. 12 — WSS Pre-condition Verification (Wiener-Khinchin)\n"
                 "GUM requires stationary calibration window before ρ = μ+kσ is valid",
                 fontsize=8.5, fontweight="bold")
    fig.tight_layout(rect=[0, 0, 1, 0.91])
    save(fig, 12, "wss_verification", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 13 — Episode Precision-Recall Frontier
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig13(d, out_dir, dpi):
    pr = d["fig13_precision_recall"]["methods"]

    fig, ax = plt.subplots(figsize=(4.5, 3.5))

    for m in pr:
        p, r = m["precision"], m["recall"]
        ep   = m["episodes"]
        col  = m["color"]
        ms   = 90 if m["is_dsfb"] else 55
        mk   = "D" if m["is_dsfb"] else "o"
        ax.scatter(r, p, s=ms, c=col, marker=mk, zorder=4,
                   edgecolors="#333333" if m["is_dsfb"] else "none",
                   linewidths=0.8 if m["is_dsfb"] else 0)
        ax.annotate(f" {m['name']}\n({ep:,} ep.)",
                    (r, p), fontsize=6.5,
                    xytext=(5 if p < 0.5 else -5, 1),
                    textcoords="offset points",
                    va="bottom", ha="left" if p < 0.5 else "right")

    # Iso-F1 contours
    ps = np.linspace(0.01, 1.0, 200)
    for f1 in [0.01, 0.05, 0.10, 0.30, 0.50, 0.70]:
        rs = f1 * ps / (2 * ps - f1 + 1e-12)
        valid = (rs >= 0) & (rs <= 1.0)
        ax.plot(rs[valid], ps[valid], "--", color="#aaaaaa", lw=0.7, alpha=0.6)
        if np.any(valid):
            idx = np.argmin(np.abs(rs[valid] - 0.78))
            ax.text(rs[valid][idx], ps[valid][idx], f"F₁={f1:.2f}",
                    fontsize=5.5, color="#888888")

    ax.set_xlim(0.88, 1.005)
    ax.set_ylim(0.0, 1.05)
    ax.set_xlabel("Recall")
    ax.set_ylabel("Episode Precision")
    ax.legend(handles=[
        Line2D([0],[0], marker="D", color="white", markeredgecolor="#333333", ms=8, label="DSFB"),
        Line2D([0],[0], marker="o", color=C_NEUTRAL, ms=7, label="Comparators"),
    ], fontsize=7.5, loc="lower left")
    ax.set_title("Fig. 13 — Episode Precision–Recall Frontier\n"
                 "DSFB: 102.2× precision gain on RadioML, 76.8× on ORACLE")
    fig.tight_layout()
    save(fig, 13, "precision_recall_frontier", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 14 — Multi-Channel Corroboration (Lemma 6)
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig14(d, out_dir, dpi):
    corr  = d["fig14_corroboration"]
    m_v   = np.array(corr["m_values"])
    f_ep  = np.array(corr["false_ep_rate"])

    fig, ax = plt.subplots(figsize=(4.5, 3.0))

    ax.semilogy(m_v, f_ep, "o-", color=C_DSFB, lw=1.6, ms=7, label="False ep. rate")
    for x, y in zip(m_v, f_ep):
        ax.annotate(f"{y:.3f}", (x, y), textcoords="offset points",
                    xytext=(4, 4), fontsize=7)

    ax.axhline(0.05, color=C_NEUTRAL, ls="--", lw=0.8, alpha=0.7,
               label="5% reference")
    ax.set_xlabel("Required corroboration count m")
    ax.set_ylabel("False episode rate P(X ≥ m)")
    ax.legend(fontsize=7.5)
    ax.set_title("Fig. 14 — Multi-Channel Corroboration (Lemma 6)\n"
                 "$P(X\\geq m) = \\binom{M}{m}p_f^m(1-p_f)^{M-m}$, M=6, $p_f$=0.046")
    ax.set_xticks(m_v)
    fig.tight_layout()
    save(fig, 14, "corroboration_lemma6", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 15 — Memory Footprint (no_std Bare-Metal)
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig15(d, out_dir, dpi):
    mem = d["fig15_memory"]["modules"]
    names  = [m["module"] for m in mem]
    sizes  = np.array([m["bytes"] for m in mem])
    colors = [C_DSFB if "total" in m["module"].lower() else C_ADMISSIBLE for m in mem]

    fig, ax = plt.subplots(figsize=(5.5, 3.2))
    bars = ax.barh(range(len(names)), sizes, color=colors,
                   edgecolor="#333333", linewidth=0.7)
    ax.set_yticks(range(len(names)))
    ax.set_yticklabels(names, fontsize=8)
    ax.set_xlabel("Stack bytes (W=10, K=4, M=8)")
    ax.set_title("Fig. 15 — Memory Footprint: no_std / no_alloc / zero-unsafe\n"
                 "All structures stack-allocated — suitable for Cortex-M4F / RISC-V FPGA softcore")
    ax.invert_yaxis()

    for bar, sz in zip(bars, sizes):
        ax.text(sz + 5, bar.get_y() + bar.get_height() / 2,
                f"{sz} B", va="center", fontsize=8)

    # Reference line: 4 KB stack budget
    ax.axvline(4096, color=C_VIOLATION, ls="--", lw=1.0, label="4 KB stack budget")
    ax.axvline(1024, color=C_BOUNDARY,  ls=":", lw=0.8, label="1 KB reference")
    ax.legend(fontsize=7.5)
    fig.tight_layout()
    save(fig, 15, "memory_footprint", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 16 — Complexity Entropy Regime Transition
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig16(d, out_dir, dpi):
    cs = d["fig16_complexity"]
    k      = np.array(cs["k"])
    ent    = np.array(cs["entropy"])
    comp   = np.array(cs["complexity"])
    regime = cs["regime"]

    fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(5.5, 3.8), sharex=True)

    ax1.plot(k, comp, color=C_DSFB, lw=1.3, label="Normalised complexity H/H_max")
    ax1.axhline(0.30, color=C_BOUNDARY, ls="--", lw=0.8, label="Low → Transitional (0.30)")
    ax1.axhline(0.70, color=C_VIOLATION, ls="--", lw=0.8, label="Transitional → High (0.70)")
    ax1.fill_between(k, 0.30, 0.70, alpha=0.08, color=C_BOUNDARY)
    ax1.fill_between(k, 0.70, 1.05, alpha=0.08, color=C_VIOLATION)
    ax1.set_ylim(0, 1.05)
    ax1.set_ylabel("Complexity score")
    ax1.legend(fontsize=7, loc="upper left")
    ax1.set_title("Fig. 16 — MDL/Kolmogorov Complexity (Shannon Entropy Estimator)\n"
                  "Rising complexity = nominal model losing descriptive power = Boundary approaching")

    # Regime timeline
    regime_map = {"LowComplexity": C_ADMISSIBLE,
                  "TransitionalComplexity": C_BOUNDARY,
                  "HighComplexity": C_VIOLATION}
    reg_colors = [regime_map.get(r, C_NEUTRAL) for r in regime]
    for i, c in enumerate(reg_colors):
        ax2.barh(0, 1, left=i, height=0.7, color=c, edgecolor="none")
    ax2.set_yticks([])
    ax2.set_xlabel("Observation k")
    ax2.set_ylabel("Regime")
    ax2.set_xlim(0, len(k))
    patch_L = mpatches.Patch(color=C_ADMISSIBLE, label="LowComplexity")
    patch_T = mpatches.Patch(color=C_BOUNDARY,   label="TransitionalComplexity")
    patch_H = mpatches.Patch(color=C_VIOLATION,  label="HighComplexity")
    ax2.legend(handles=[patch_L, patch_T, patch_H], fontsize=7, ncol=3,
               loc="lower right")

    # Regime labels on ax1
    for label, start, end in [("Nominal\n(low H)", 0, 30),
                               ("Drift onset\n(rising H)", 30, 60),
                               ("High\ncomplexity", 60, 80),
                               ("Recovery", 80, len(k))]:
        ax1.text((start + end) / 2, 0.94, label,
                 ha="center", fontsize=6.5, color="#333333",
                 bbox=dict(boxstyle="round,pad=0.1", facecolor="white", alpha=0.7))

    fig.tight_layout()
    save(fig, 16, "complexity_entropy_regime", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 17 — Grammar FSM Hysteresis
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig17(d, out_dir, dpi):
    fsm    = d["fig17_fsm"]
    k      = np.array(fsm["k"])
    confs  = np.array(fsm["confirmations"])
    states = fsm["committed_state"]

    fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(5.5, 3.4), sharex=True,
                                   gridspec_kw={"height_ratios": [1, 2]})

    # Severity bar
    ax1.step(k, confs, where="mid", color=C_DSFB, lw=1.3, label="Grammar severity (0-2)")
    ax1.set_ylim(-0.1, 2.4)
    ax1.set_yticks([0, 1, 2])
    ax1.set_yticklabels(["Admiss.", "Boundary", "Violation"], fontsize=7.5)
    ax1.set_ylabel("Severity")
    ax1.set_title("Fig. 17 — Grammar FSM Hysteresis Gate\n"
                  "2 consecutive confirmations required before state transition is committed")

    # State timeline with hysteresis annotations
    colors = [grammar_color(s) for s in states]
    for i, c in enumerate(colors):
        ax2.barh(0, 1, left=i, height=0.7, color=c, edgecolor="none")

    # Annotate the transient spike dismissal and the legitimate confirmation
    ax2.annotate("Transient spike\n(dismissed by\nhysteresis)",
                 xy=(4.5, 0.35), xytext=(7.5, 0.52),
                 arrowprops=dict(arrowstyle="->", color="#333333", lw=0.8),
                 fontsize=7, ha="center",
                 bbox=dict(boxstyle="round,pad=0.2", facecolor="white", alpha=0.8))
    ax2.annotate("Confirmed after\n2+ observations",
                 xy=(10.5, 0.35), xytext=(13.5, 0.52),
                 arrowprops=dict(arrowstyle="->", color=C_BOUNDARY, lw=0.8),
                 fontsize=7, ha="center",
                 bbox=dict(boxstyle="round,pad=0.2", facecolor="white", alpha=0.8))

    ax2.set_yticks([])
    ax2.set_xlabel("Observation k")
    ax2.set_ylabel("Committed state")
    ax2.set_xlim(0, len(k))
    patch_A = mpatches.Patch(color=C_ADMISSIBLE, label="Admissible")
    patch_B = mpatches.Patch(color=C_BOUNDARY, label="Boundary")
    patch_V = mpatches.Patch(color=C_VIOLATION, label="Violation")
    ax2.legend(handles=[patch_A, patch_B, patch_V], fontsize=7, ncol=3)

    fig.tight_layout()
    save(fig, 17, "grammar_fsm_hysteresis", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 18 — Standards Alignment Matrix
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig18(d, out_dir, dpi):
    sm = d["fig18_standards"]
    stds    = sm["standards"]
    aspects = sm["aspects"]
    cov     = np.array(sm["coverage"])

    fig, ax = plt.subplots(figsize=(7.0, 3.8))

    n_std, n_asp = cov.shape
    for i in range(n_std):
        for j in range(n_asp):
            v = cov[i, j]
            col = C_ADMISSIBLE if v == 1 else "#f0f0f0"
            rect = mpatches.FancyBboxPatch(
                (j + 0.05, n_std - 1 - i + 0.05), 0.90, 0.90,
                boxstyle="round,pad=0.03",
                facecolor=col, edgecolor="#cccccc",
                linewidth=0.5, zorder=2)
            ax.add_patch(rect)
            sym = "" if v == 1 else "·"
            ax.text(j + 0.5, n_std - 1 - i + 0.5, sym,
                    ha="center", va="center",
                    fontsize=13 if v == 1 else 8,
                    color="white" if v == 1 else "#aaaaaa",
                    fontweight="bold" if v == 1 else "normal")

    ax.set_xticks([j + 0.5 for j in range(n_asp)])
    ax.set_xticklabels(aspects, fontsize=8)
    ax.set_yticks([n_std - 1 - i + 0.5 for i in range(n_std)])
    ax.set_yticklabels(stds, fontsize=8)
    ax.set_xlim(0, n_asp)
    ax.set_ylim(0, n_std)
    ax.grid(False)
    ax.set_title("Fig. 18 — Standards Alignment Coverage Matrix\n"
                 "ITU-R SM.1048-5 · MIL-STD-461G · 3GPP TS 36.141 · VITA 49.2 · SigMF · GUM · SOSA/MORA",
                 fontsize=8.5)
    for spine in ax.spines.values():
        spine.set_linewidth(0.6)

    patch_c = mpatches.Patch(color=C_ADMISSIBLE, label="✓ Covered (clause-depth)")
    patch_n = mpatches.Patch(color="#f0f0f0", edgecolor="#cccccc", label="· Not applicable")
    ax.legend(handles=[patch_c, patch_n], loc="lower right", fontsize=7.5)

    fig.tight_layout()
    save(fig, 18, "standards_alignment", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 19 — Architectural Integration (Read-Only Tap)
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig19(d, out_dir, dpi):
    arch = d["fig19_architecture"]
    layers = arch["layers"]

    fig, ax = plt.subplots(figsize=(7.0, 3.6))
    ax.axis("off")
    ax.set_title("Fig. 19 — DSFB-RF Non-Intrusive Architecture\n"
                 "Read-only tap via immutable  &[f32]  — upstream receiver UNCHANGED",
                 fontsize=9)

    n_layers = len(layers)
    layer_h  = 0.18
    layer_gap = 0.04
    total_h  = n_layers * (layer_h + layer_gap)
    start_y  = (1.0 - total_h) / 2

    for i, layer in enumerate(layers):
        y = 1.0 - start_y - (i + 1) * (layer_h + layer_gap)
        x = 0.04
        w = 0.92
        rect = mpatches.FancyBboxPatch((x, y), w, layer_h,
            boxstyle="round,pad=0.01",
            facecolor=layer["color"], edgecolor="#444444",
            linewidth=1.1 + (0.5 if "DSFB" in layer["name"] else 0),
            transform=ax.transAxes, clip_on=False)
        ax.add_patch(rect)

        # Layer name
        ax.text(x + 0.01, y + layer_h / 2, layer["name"],
                transform=ax.transAxes, va="center", ha="left",
                fontsize=8, fontweight="bold", color="#111111")

        # Modules on the right
        mod_str = "  ·  ".join(layer["modules"])
        ax.text(x + w - 0.01, y + layer_h / 2, mod_str,
                transform=ax.transAxes, va="center", ha="right",
                fontsize=6.5, color="#333333")

        # Arrow between layers
        if i < n_layers - 1:
            arrow_x = 0.5
            arrow_y = y - layer_gap / 2
            if "DSFB" in layer["name"]:
                label = "immutable &[f32] tap\n(read-only · zero write path)"
                col   = C_ADMISSIBLE
            else:
                label = ""
                col   = "#888888"
            ax.annotate("", xy=(arrow_x, arrow_y - layer_gap * 0.1),
                        xytext=(arrow_x, arrow_y + layer_gap * 0.7),
                        xycoords="axes fraction",
                        arrowprops=dict(arrowstyle="->,head_width=0.20",
                                        color=col, lw=1.3),
                        annotation_clip=False)
            if label:
                ax.text(arrow_x + 0.02, arrow_y + layer_gap * 0.3, label,
                        transform=ax.transAxes, fontsize=7, color=col,
                        ha="left", va="center",
                        bbox=dict(boxstyle="round,pad=0.15",
                                  facecolor="white", alpha=0.85))

    # "Remove DSFB → receiver unaffected" callout
    ax.text(0.50, 0.02,
            "If DSFB is removed: upstream receiver behavior is UNCHANGED · "
            "Zero calibration · Zero restart · Zero risk",
            transform=ax.transAxes, ha="center", va="bottom", fontsize=7,
            fontstyle="italic", color="#333333",
            bbox=dict(boxstyle="round,pad=0.2", facecolor="#fffde7", alpha=0.9))

    fig.tight_layout()
    save(fig, 19, "architecture_integration", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Fig 20 — Policy Escalation Logic
# ═══════════════════════════════════════════════════════════════════════════
def plot_fig20(d, out_dir, dpi):
    pl = d["fig20_policy"]
    k       = np.array(pl["k"])
    dsa     = np.array(pl["dsa"])
    grammar = pl["grammar"]
    policy  = pl["policy"]
    tau     = pl["tau"]

    POLICY_COLOR = {
        "Silent":   C_ADMISSIBLE,
        "Watch":    "#aec7e8",
        "Review":   C_BOUNDARY,
        "Escalate": C_VIOLATION,
    }
    POLICY_LEVEL = {"Silent": 0, "Watch": 1, "Review": 2, "Escalate": 3}

    fig, (ax1, ax2, ax3) = plt.subplots(3, 1, figsize=(5.5, 4.5), sharex=True,
                                         gridspec_kw={"height_ratios": [3, 1, 1]})

    # DSA trace
    ax1.plot(k, dsa, color=C_DSFB, lw=1.4, label="DSA(k)")
    ax1.axhline(tau, color=C_VIOLATION, ls="--", lw=1.0, label=f"τ = {tau}")
    ax1.fill_between(k, dsa, tau, where=(dsa >= tau),
                     color=C_VIOLATION, alpha=0.15)
    ax1.set_ylabel("DSA Score")
    ax1.legend(fontsize=7.5)
    ax1.set_title("Fig. 20 — Policy Escalation Logic\n"
                  "Silent → Watch → Review → Escalate (persistence K=4 + DSA ≥ τ + corroboration m=1)")

    # Grammar timeline
    for i, g in enumerate(grammar):
        ax2.barh(0, 1, left=i, height=0.7, color=grammar_color(g), edgecolor="none")
    ax2.set_yticks([])
    ax2.set_ylabel("Grammar")
    ax2.set_xlim(0, len(k))

    # Policy timeline
    pol_levels = [POLICY_LEVEL.get(p, 0) for p in policy]
    pol_colors = [POLICY_COLOR.get(p, C_NEUTRAL) for p in policy]
    for i, (lv, pc) in enumerate(zip(pol_levels, pol_colors)):
        ax3.barh(0, 1, left=i, height=0.7, color=pc, edgecolor="none")

    # Annotate policy transitions
    prev = None
    for i, p in enumerate(policy):
        if p != prev and p != "Silent":
            ax3.text(i + 0.5, 0.5, p[0], ha="center", va="center",
                     fontsize=6, color="white", fontweight="bold")
        prev = p

    ax3.set_yticks([])
    ax3.set_xlabel("Observation k")
    ax3.set_ylabel("Policy")
    ax3.set_xlim(0, len(k))

    legend_elems = [
        mpatches.Patch(color=C_ADMISSIBLE, label="Silent"),
        mpatches.Patch(color="#aec7e8",    label="Watch"),
        mpatches.Patch(color=C_BOUNDARY,   label="Review"),
        mpatches.Patch(color=C_VIOLATION,  label="Escalate"),
    ]
    ax3.legend(handles=legend_elems, fontsize=7, ncol=4, loc="lower right")

    fig.tight_layout()
    save(fig, 20, "policy_escalation_logic", out_dir, dpi)


# ═══════════════════════════════════════════════════════════════════════════
# Entry point
# ═══════════════════════════════════════════════════════════════════════════
FIGURES = {
    1:  plot_fig01,   2:  plot_fig02,   3:  plot_fig03,
    4:  plot_fig04,   5:  plot_fig05,   6:  plot_fig06,
    7:  plot_fig07,   8:  plot_fig08,   9:  plot_fig09,
    10: plot_fig10,   11: plot_fig11,   12: plot_fig12,
    13: plot_fig13,   14: plot_fig14,   15: plot_fig15,
    16: plot_fig16,   17: plot_fig17,   18: plot_fig18,
    19: plot_fig19,   20: plot_fig20,
}


def main():
    parser = argparse.ArgumentParser(description="DSFB-RF figure generator")
    parser.add_argument("--data",  default="figure_data.json",
                        help="Input JSON data file (default: figure_data.json)")
    parser.add_argument("--out",   default="figs",
                        help="Output directory (default: figs/)")
    parser.add_argument("--dpi",   type=int, default=150,
                        help="Figure DPI (default: 150; use 300 for print)")
    parser.add_argument("--fig",   type=int, nargs="*",
                        help="Specific figure numbers (default: all)")
    args = parser.parse_args()

    data_path = Path(args.data)
    if not data_path.exists():
        print(f"ERROR: {data_path} not found.")
        print("Run:  cargo run --features std,serde --example generate_figures")
        sys.exit(1)

    with open(data_path, "r") as fh:
        data = json.load(fh)

    out_dir = Path(args.out)
    selected = args.fig if args.fig else sorted(FIGURES.keys())

    print(f"Rendering {len(selected)} figures → {out_dir}/")
    print(f"DPI = {args.dpi}")
    print()

    for idx in selected:
        if idx not in FIGURES:
            print(f"  WARNING: fig {idx} not defined, skipping.")
            continue
        try:
            FIGURES[idx](data, out_dir, args.dpi)
        except Exception as e:
            print(f"  ERROR in fig {idx}: {e}")
            import traceback
            traceback.print_exc()

    print()
    print(f"Done. {len(selected)} figure(s) written to {out_dir}/")


if __name__ == "__main__":
    main()