proyecto-estrella-sobol/genz_benchmark_COMBO_vs_JK_output.log at main · tretoef-estrella/proyecto-estrella-sobol · GitHub

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
Last login: Fri May 29 09:43:25 on ttys001
rafa@MacBook-Air-de-RAFAEL Downloads % cd ~/Downloads && python3 compare_COMBO_rank_vs_boxcount.py
=== COMBO cross-check: rank vs box-counting, m in [5,14] ===

[Route A] rank-based partial audit ...
   = 1333   [1.0s]

[Route B] box-counting partial audit (qmcpy points) ...
   = 1333   [47.3s]

====================================================
AGREE: both routes return 1333 over m in [5,14].
The rank-based and equidistribution-based t-values agree on the
COMBO dump. Independent cross-check PASS.
rafa@MacBook-Air-de-RAFAEL Downloads % cd ~/Downloads && python3 genz_benchmark_COMBO_vs_JK.py TOGORDO_COMBO_v1_RECORD.txt new-joe-kuo-6.21201
==============================================================================
  Genz Benchmark — COMBO_3027 vs JK_2008
  Functions: 6 Genz standard
  Dimensions: [10, 20, 30]
  N points per rep: 65536
  Reps per (func, dim, sequence): 30
  Reference truth: MC N=4194304
==============================================================================

[Parse] Loading dumps...
  COMBO_3027: 37 dims loaded
  JK_2008:    21201 dims loaded

[Build] Sobol generators...
  Built in 0.0s

[Generate] Sobol sequences N=65536...
  Generated in 1.0s

[Run] dim=10
  Oscillatory        d=10: COMBO=7.184e-03  JK=7.385e-03  ratio=0.973  ✓ COMBO
  Product peak       d=10: COMBO=4.080e+05  JK=4.068e+05  ratio=1.003  ✗ JK
  Corner peak        d=10: COMBO=1.526e-05  JK=1.526e-05  ratio=1.000  ✓ COMBO
  Gaussian           d=10: COMBO=6.664e-04  JK=6.651e-04  ratio=1.002  ✗ JK
  Continuous         d=10: COMBO=2.143e-04  JK=2.140e-04  ratio=1.002  ✗ JK
  Discontinuous      d=10: COMBO=2.512e+06  JK=2.541e+06  ratio=0.988  ✓ COMBO

[Run] dim=20
  Oscillatory        d=20: COMBO=1.925e-03  JK=2.400e-03  ratio=0.802  ✓ COMBO
  Product peak       d=20: COMBO=1.549e+11  JK=1.681e+11  ratio=0.921  ✓ COMBO
  Corner peak        d=20: COMBO=1.526e-05  JK=1.526e-05  ratio=1.000  ✗ JK
  Gaussian           d=20: COMBO=1.738e-06  JK=1.636e-06  ratio=1.062  ✗ JK
  Continuous         d=20: COMBO=2.730e-07  JK=2.536e-07  ratio=1.076  ✗ JK
  Discontinuous      d=20: COMBO=2.597e+15  JK=6.663e+14  ratio=3.898  ✗ JK

[Run] dim=30
  Oscillatory        d=30: COMBO=2.038e-03  JK=2.504e-03  ratio=0.814  ✓ COMBO
  Product peak       d=30: COMBO=1.147e+16  JK=2.613e+16  ratio=0.439  ✓ COMBO
  Corner peak        d=30: COMBO=1.526e-05  JK=1.526e-05  ratio=1.000  ✗ JK
  Gaussian           d=30: COMBO=1.947e-08  JK=1.500e-08  ratio=1.298  ✗ JK
  Continuous         d=30: COMBO=7.296e-10  JK=5.215e-10  ratio=1.399  ✗ JK
  Discontinuous      d=30: COMBO=4.914e+23  JK=2.894e+23  ratio=1.698  ✗ JK

==============================================================================
  SUMMARY
==============================================================================

  | Function          | dim |  COMBO error  |  JK error    | Ratio  | Winner |
  |-------------------|-----|---------------|--------------|--------|--------|
  | Oscillatory       |  10 | 7.184e-03    | 7.385e-03   | 0.973  | COMBO  |
  | Oscillatory       |  20 | 1.925e-03    | 2.400e-03   | 0.802  | COMBO  |
  | Oscillatory       |  30 | 2.038e-03    | 2.504e-03   | 0.814  | COMBO  |
  | Product peak      |  10 | 4.080e+05    | 4.068e+05   | 1.003  | JK     |
  | Product peak      |  20 | 1.549e+11    | 1.681e+11   | 0.921  | COMBO  |
  | Product peak      |  30 | 1.147e+16    | 2.613e+16   | 0.439  | COMBO  |
  | Corner peak       |  10 | 1.526e-05    | 1.526e-05   | 1.000  | COMBO  |
  | Corner peak       |  20 | 1.526e-05    | 1.526e-05   | 1.000  | JK     |
  | Corner peak       |  30 | 1.526e-05    | 1.526e-05   | 1.000  | JK     |
  | Gaussian          |  10 | 6.664e-04    | 6.651e-04   | 1.002  | JK     |
  | Gaussian          |  20 | 1.738e-06    | 1.636e-06   | 1.062  | JK     |
  | Gaussian          |  30 | 1.947e-08    | 1.500e-08   | 1.298  | JK     |
  | Continuous        |  10 | 2.143e-04    | 2.140e-04   | 1.002  | JK     |
  | Continuous        |  20 | 2.730e-07    | 2.536e-07   | 1.076  | JK     |
  | Continuous        |  30 | 7.296e-10    | 5.215e-10   | 1.399  | JK     |
  | Discontinuous     |  10 | 2.512e+06    | 2.541e+06   | 0.988  | COMBO  |
  | Discontinuous     |  20 | 2.597e+15    | 6.663e+14   | 3.898  | JK     |
  | Discontinuous     |  30 | 4.914e+23    | 2.894e+23   | 1.698  | JK     |

  COMBO_3027 wins: 7 / 18 (38.9%)

==============================================================================
  F75 DUAL-FORM TEST
==============================================================================
  COMBO_3027 wins only 7/18 (<50%).
  → F_2 audit improvement DOES NOT translate to Genz error reduction.
  → F75 v52 dual-form RATIFIED: F_2 audit and F_2 Genz error NOT monotonically related.
  → Practitioner-relevance NOT RATIFIED. Record audit-metric only.
==============================================================================

Total runtime: 604.5s
rafa@MacBook-Air-de-RAFAEL Downloads %