
==============================================================================
LAYER 0: exact PFFV arithmetic (fractions only)
==============================================================================
N          = [[Fraction(6, 1), Fraction(2, 1)], [Fraction(2, 1), Fraction(-16, 1)]]
p          = [Fraction(2, 5), Fraction(3, 10)]  (K.p = 0 exact; p > 0)
a.M        = [Fraction(3, 1), Fraction(-24, 1)]   b.M = 7   (integrality OK)
N_flux     = 124 <= Q_D3 = 138
A          = -5/288    c = sqrt(2/pi)*8640/(2 pi i)^2  [derived]
[L] PRINTED c = -sqrt(2/pi)*8640/(2 pi i)^3 differs from derived c by
    -1/(2 pi i): |c_printed|/|c_derived| = 1/(2 pi).  Forensics in Layer 3.
GV multicover check vs [L] F_1, F_2: exact match {(1,0):-540, (0,1):-3, (2,0):-1215/2, (1,1):+1080, (0,2):+45/8}
chi(X) = -540  (X = P[1,1,1,6,9][18]; fiber GV 540 = -chi cross-check OK)

==============================================================================
LAYER 2: variant grid at dps = 50 (the adjudication)
==============================================================================
--- V1  xi(chi_A=-540) [DKMM frame], GV deg<=4  -> expect 2.037e-8, tau=6.856i
    tau  = (7.48553447246e-45 + 6.85545725632j)
    U1   = (2.99431361975e-45 + 2.74217698021j)
    U2   = (2.24572872069e-45 + 2.0566327731j)
    e^-K = 6644.94353599  (CS block 484.646267021)
    |W|  = 1.66057805884e-6
    |W0| = 2.03710609331e-8   [max residual 1.76e-48]
--- V1deg2  same, GV total degree <= 2 (letter's own truncation)
    tau  = (-1.05869647621e-46 + 6.85545725771j)
    U1   = (-4.23351086794e-47 + 2.74217698077j)
    U2   = (-3.17554529551e-47 + 2.05663277352j)
    e^-K = 6644.94354135  (CS block 484.646267313)
    |W|  = 1.66057805727e-6
    |W0| = 2.03710609057e-8   [max residual 3.92e-49]
--- V2  xi = 0 [Broeckel/CMS], GV deg<=2  -> expect 2.048e-8, tau=6.85505i
    tau  = (-1.79725774648e-44 + 6.85506719668j)
    U1   = (-7.18926574175e-45 + 2.74202095175j)
    U2   = (-5.39193228617e-45 + 2.05651575176j)
    e^-K = 6571.68876081  (CS block 479.330732454)
    |W|  = 1.66038985684e-6
    |W0| = 2.04819630466e-8   [max residual 2.35e-48]
--- V2full  xi = 0, GV deg<=4
    tau  = (-6.30855149707e-45 + 6.85506719528j)
    U1   = (-2.52348902273e-45 + 2.74202095119j)
    U2   = (-1.89261628386e-45 + 2.05651575134j)
    e^-K = 6571.68875546  (CS block 479.330732161)
    |W|  = 1.6603898584e-6
    |W0| = 2.04819630742e-8   [max residual 6.56e-49]
--- V3  xi(chi_A=+540) [wrong-sign trap], GV deg<=4  -> matches neither
    tau  = (1.52227646952e-44 + 6.85466912629j)
    U1   = (6.08931193799e-45 + 2.74186171889j)
    U2   = (4.56696902316e-45 + 2.05639632711j)
    e^-K = 6498.4117768  (CS block 474.013526917)
    |W|  = 1.66019622592e-6
    |W0| = 2.05947160558e-8   [max residual 6.77e-49]

==============================================================================
Flux sign-flip test: (M,K) -> (-M,-K) [Broeckel Table 1 vectors]
==============================================================================
--- V2 with sign-flipped fluxes
    tau  = (-1.79725774648e-44 + 6.85506719668j)
    U1   = (-7.18926574175e-45 + 2.74202095175j)
    U2   = (-5.39193228617e-45 + 2.05651575176j)
    e^-K = 6571.68876081  (CS block 479.330732454)
    |W|  = 1.66038985684e-6
    |W0| = 2.04819630466e-8   [max residual 2.35e-48]
| |W0|_flip - |W0|_V2 | = 0.0  -> flux sign flip is NOT the source of the discrepancy

==============================================================================
LAYER 3: racetrack-constant forensics
==============================================================================
2-term racetrack at the V1 vacuum point (e^{K/2}-dressed):
  with c = sqrt(2/pi) 8640/(2pi i)^2  [derived; = Broeckel/CMS printed]:
     |W0| = 2.0372717e-8   (matches the letter's 2.037e-8 scale)
  with c = -sqrt(2/pi) 8640/(2pi i)^3 [as printed in 1912.10047 eq.(19)]:
     |W0| = 3.2424187e-9   = derived/(2 pi) -> the printed c is a typo
  ratio printed/derived = 0.1591549431 = 1/(2 pi) = 0.1591549431

==============================================================================
LAYER 4: headline |W0|, two-precision control (dps 50 vs 90)
==============================================================================
dps=50: |W0| = 0.0000000203710609331118341820831178623429774461
dps=90: |W0| = 0.0000000203710609331118341820831178623429774461
agreement: |diff|/|W0| = 4.0953e-43

HEADLINE (model value, GV deg<=4, DKMM frame, 35 digits):
  |W0| = 0.000000020371060933111834182083117862342977
  Im tau = 6.8554572563187259046361432042141669
  Im U1  = 2.7421769802085209924953942571371399
  Im U2  = 2.056632773104733929945590747451151
  g_s = 1/Im tau = 0.145869190429
  Re parts: -4.31e-94 -1.72e-94 -1.29e-94
  flat-dir displacement: U - tau*p = (-4.7255435e-99 - 5.922319e-6j) (-3.548889e-99 - 4.4037909e-6j)

==============================================================================
LAYER 5: instanton-shell audit and truncation tail bound
==============================================================================
shell d = sum over classes+multicovers at total instanton degree d of
          A_q,k (M.q) e^{-2 pi k q.u}   (units: W_eff = sqrt(2/pi)/(2pi i)^2 * S)
  d= 1:     -8.216975e-5   |shell|/|S| = 1.0
  d= 2:     6.3308643e-9   |shell|/|S| = 7.705e-5
  d= 3:   -1.1068282e-13   |shell|/|S| = 1.347e-9
  d= 4:    2.7015272e-18   |shell|/|S| = 3.288e-14
  d= 5:   -4.2687162e-23   |shell|/|S| = 5.195e-19
  d= 6:    4.3595429e-28   |shell|/|S| = 5.306e-24
  d= 7:   -2.9528686e-33   |shell|/|S| = 3.594e-29
  d= 8:    1.6795793e-38   |shell|/|S| = 2.044e-34
  d= 9:   -1.4017643e-48   |shell|/|S| = 1.706e-44
  d=10:   -9.6330814e-54   |shell|/|S| = 1.172e-49
  d=11:    -2.295309e-62   |shell|/|S| = 2.794e-58
  d=12:    1.8592047e-64   |shell|/|S| = 2.263e-60
  d=13:    -9.809691e-74   |shell|/|S| = 1.194e-69
  d=14:    2.2657725e-78   |shell|/|S| = 2.758e-74
  d=15:   -1.1240567e-82   |shell|/|S| = 1.368e-78
  d=16:    3.8827843e-87   |shell|/|S| = 4.726e-83
  d=17:   -2.0439732e-96   |shell|/|S| = 2.488e-92
  d=18:   -1.0155331e-99   |shell|/|S| = 1.236e-95
  d=19:  -9.7674532e-108   |shell|/|S| = 1.189e-103
  d=20:   8.8535476e-110   |shell|/|S| = 1.078e-105
  d=21:  -1.1645424e-116   |shell|/|S| = 1.417e-112
  d=22:   1.5937041e-123   |shell|/|S| = 1.94e-119
  d=23:  -2.3750014e-130   |shell|/|S| = 2.891e-126
  d=24:   2.0613004e-132   |shell|/|S| = 2.509e-128
  d=25:  -1.2024463e-141   |shell|/|S| = 1.463e-137
  d=26:   4.0657191e-146   |shell|/|S| = 4.948e-142
  d=27:  -1.4983486e-150   |shell|/|S| = 1.824e-146
  d=28:   5.7350832e-155   |shell|/|S| = 6.98e-151
  d=29:  -3.1773866e-164   |shell|/|S| = 3.867e-160
  d=30:  -1.6539265e-167   |shell|/|S| = 2.013e-163
  d=31:  -1.6598092e-175   |shell|/|S| = 2.02e-171
  d=32:   1.5645406e-177   |shell|/|S| = 1.904e-173
  d=33:  -2.1333348e-184   |shell|/|S| = 2.596e-180
  d=34:   3.0184796e-191   |shell|/|S| = 3.674e-187
  d=35:  -4.6395584e-198   |shell|/|S| = 5.647e-194
  d=36:   4.1443057e-200   |shell|/|S| = 5.044e-196
  d=37:  -2.4781331e-209   |shell|/|S| = 3.016e-205
  d=38:   8.6101336e-214   |shell|/|S| = 1.048e-209
  d=39:  -3.2486594e-218   |shell|/|S| = 3.954e-214
  d=40:   1.2712437e-222   |shell|/|S| = 1.547e-218
  d=41:  -7.1910344e-232   |shell|/|S| = 8.752e-228
  d=42:  -3.8172719e-235   |shell|/|S| = 4.646e-231
  d=43:  -3.9024529e-243   |shell|/|S| = 4.75e-239
  d=44:   3.7434709e-245   |shell|/|S| = 4.556e-241
  d=45:  -5.1898509e-252   |shell|/|S| = 6.316e-248
  d=46:   7.4597356e-259   |shell|/|S| = 9.079e-255
  d=47:  -1.1638845e-265   |shell|/|S| = 1.417e-261
  d=48:   1.0545504e-267   |shell|/|S| = 1.283e-263
  d=49:  -6.3918847e-277   |shell|/|S| = 7.779e-273
  d=50:   2.2497284e-281   |shell|/|S| = 2.738e-277
  d=51:  -8.5938164e-286   |shell|/|S| = 1.046e-281
  d=52:   3.4027899e-290   |shell|/|S| = 4.141e-286
  d=53:   -1.946706e-299   |shell|/|S| = 2.369e-295
  d=54:  -1.0446172e-302   |shell|/|S| = 1.271e-298
  d=55:  -1.0790516e-310   |shell|/|S| = 1.313e-306
  d=56:   1.0454344e-312   |shell|/|S| = 1.272e-308
  d=57:  -1.4632658e-319   |shell|/|S| = 1.781e-315
  d=58:   2.1226389e-326   |shell|/|S| = 2.583e-322
  d=59:  -3.3411393e-333   |shell|/|S| = 4.066e-329
  d=60:    3.053095e-335   |shell|/|S| = 3.716e-331
S = -8.21634191997e-5
shell ratios: 2/1 = 7.705e-5  3/2 = 1.748e-5  4/3 = 2.441e-5  (letter's claim: both O(1e-5))

Tail bound (classes with d1>=5 or d2>=5), assumption |n| <= 540 * (1e4)^d1 * (1e2)^d2:
  |S_tail| <= 2.872e-13    relative to |S|: 3.496e-9
  => true-CY |W0| = model |W0| * (1 + delta), |delta| <= 3.496e-9
  (K-side instanton tail is O(1e-5) x this in e^{-K}, subleading)

==============================================================================
LAYER 6: check evaluation
==============================================================================
g2 (DKMM): |W0|_V1 = 2.037e-8  vs printed 2.037e-8 ;  Im tau = 6.855457  vs printed 6.856
g3 (Broeckel/CMS): |W0|_V2 = 2.0482e-8  vs printed 2.048e-8 / 2.0482e-8 ;  Im tau = 6.855067  vs CMS 6.85505
g4 (vevs, DKMM frame): Im U1 = 2.742177  vs 2.742 ;  Im U2 = 2.056633  vs 2.057  ;  CMS U = (2.74202, 2.05652) vs V2: 2.742021 2.056516
g1: headline above (Layer 4), two-precision agreement + stated tail bound

Ratio structure: |W0|_V2/|W0|_V1 = 1.0054441  ;  printed 2.048/2.037 = 1.0054001  ;  2.0482/2.037 = 1.0054983
Done.
