matCercola18 commited on
Commit
20ec019
·
verified ·
1 Parent(s): 929e74d

Add files using upload-large-folder tool

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. results/ablations/logs/anomaly_models.log +28 -0
  2. results/ablations/logs/anomaly_report.log +26 -0
  3. results/ablations/logs/anomaly_targets.log +3 -0
  4. results/ablations/logs/anomaly_train_p25_e3200_s42.log +0 -0
  5. results/ablations/logs/anomaly_train_p25_e3200_s43.log +0 -0
  6. results/ablations/logs/anomaly_train_ref_e200_s42.log +403 -0
  7. results/ablations/logs/anomaly_train_ref_e200_s43.log +403 -0
  8. results/ablations/logs/eval_generative.log +45 -0
  9. results/ablations/logs/eval_generative_siso_seed200.log +24 -0
  10. results/ablations/logs/eval_generative_siso_seed300.log +24 -0
  11. results/ablations/logs/eval_generative_siso_seed400.log +24 -0
  12. results/ablations/logs/eval_generative_siso_seed500.log +24 -0
  13. results/ablations/logs/eval_size_data.log +64 -0
  14. results/ablations/logs/eval_steps.log +28 -0
  15. results/ablations/logs/eval_steps_mimo.log +16 -0
  16. results/ablations/logs/train_cvae_pendulum.log +23 -0
  17. results/ablations/logs/train_cvae_siso.log +23 -0
  18. results/ablations/logs/train_data_p25.log +0 -0
  19. results/ablations/logs/train_data_p25_s43.log +0 -0
  20. results/ablations/logs/train_data_p25_s44.log +0 -0
  21. results/ablations/logs/train_data_p50.log +0 -0
  22. results/ablations/logs/train_data_p50_s43.log +0 -0
  23. results/ablations/logs/train_data_p50_s44.log +0 -0
  24. results/ablations/logs/train_size_2x.log +0 -0
  25. results/ablations/logs/train_size_2x_s43.log +0 -0
  26. results/ablations/logs/train_size_2x_s44.log +0 -0
  27. results/ablations/logs/train_size_quarter.log +0 -0
  28. results/ablations/logs/train_size_quarter_s43.log +0 -0
  29. results/ablations/logs/train_size_quarter_s44.log +0 -0
  30. results/ablations/logs/train_size_ref_s42.log +0 -0
  31. results/ablations/logs/train_size_ref_s43.log +0 -0
  32. results/ablations/logs/train_size_ref_s44.log +0 -0
  33. results/furuta/catch_audit.txt +23 -0
  34. results/furuta/collocation_ceiling.txt +38 -0
  35. results/furuta/eval.txt +5 -0
  36. results/furuta/eval_full.txt +12 -0
  37. results/furuta/friction_drift.txt +17 -0
  38. results/furuta/friction_drift_analysis.txt +35 -0
  39. results/furuta/gen_variance.txt +17 -0
  40. results/furuta/handoff_eval.txt +45 -0
  41. results/furuta/handoff_gate.txt +29 -0
  42. results/furuta/handoff_voltage.txt +14 -0
  43. results/furuta/logs/bank_v2_build.log +234 -0
  44. results/furuta/logs/deploy_extras.log +138 -0
  45. results/furuta/logs/deploy_extras_prev.log +35 -0
  46. results/furuta/logs/deploy_main.log +0 -0
  47. results/furuta/logs/gen_eval_v1.log +471 -0
  48. results/furuta/logs/gen_eval_v2.log +471 -0
  49. results/furuta/logs/ksweep.log +571 -0
  50. results/furuta/logs/margin_check.log +53 -0
results/ablations/logs/anomaly_models.log ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ target seed=2026 draws=(0, 1, 2)
2
+ [models] device=cuda draws=(0, 1, 2) K=8
3
+ f0.25_e800 seed42 seed123: K1 med +1.30% (fail 0.000) K8 med +0.52% dmin 0.010 dnn 0.048
4
+ f0.25_e800 seed42 seed321: K1 med +1.49% (fail 0.000) K8 med +0.50% dmin 0.011 dnn 0.067
5
+ f0.25_e800 seed43 seed123: K1 med +0.93% (fail 0.000) K8 med +0.16% dmin 0.012 dnn 0.066
6
+ f0.25_e800 seed43 seed321: K1 med +0.99% (fail 0.000) K8 med +0.20% dmin 0.013 dnn 0.058
7
+ f0.25_e800 seed44 seed123: K1 med +0.89% (fail 0.000) K8 med +0.07% dmin 0.025 dnn 0.078
8
+ f0.25_e800 seed44 seed321: K1 med +0.89% (fail 0.000) K8 med +0.17% dmin 0.025 dnn 0.069
9
+ f0.50_e800 seed42 seed123: K1 med +1.66% (fail 0.000) K8 med +1.16% dmin 0.004 dnn 0.044
10
+ f0.50_e800 seed42 seed321: K1 med +1.59% (fail 0.000) K8 med +1.10% dmin 0.004 dnn 0.047
11
+ f0.50_e800 seed43 seed123: K1 med +1.19% (fail 0.000) K8 med +0.72% dmin 0.006 dnn 0.051
12
+ f0.50_e800 seed43 seed321: K1 med +1.11% (fail 0.000) K8 med +0.65% dmin 0.005 dnn 0.049
13
+ f0.50_e800 seed44 seed123: K1 med +1.52% (fail 0.000) K8 med +0.92% dmin 0.005 dnn 0.041
14
+ f0.50_e800 seed44 seed321: K1 med +1.59% (fail 0.000) K8 med +0.84% dmin 0.005 dnn 0.047
15
+ f1.00_e800 seed42 seed123: K1 med +1.45% (fail 0.000) K8 med +1.07% dmin 0.004 dnn 0.034
16
+ f1.00_e800 seed42 seed321: K1 med +1.49% (fail 0.000) K8 med +1.02% dmin 0.004 dnn 0.033
17
+ f1.00_e800 seed43 seed123: K1 med +1.00% (fail 0.000) K8 med +0.67% dmin 0.005 dnn 0.040
18
+ f1.00_e800 seed43 seed321: K1 med +1.05% (fail 0.000) K8 med +0.60% dmin 0.005 dnn 0.037
19
+ f1.00_e800 seed44 seed123: K1 med +1.38% (fail 0.000) K8 med +1.00% dmin 0.003 dnn 0.034
20
+ f1.00_e800 seed44 seed321: K1 med +1.34% (fail 0.000) K8 med +0.93% dmin 0.003 dnn 0.033
21
+ f0.25_e3200 seed42 seed123: K1 med +1.46% (fail 0.000) K8 med +1.15% dmin 0.004 dnn 0.040
22
+ f0.25_e3200 seed42 seed321: K1 med +1.48% (fail 0.000) K8 med +1.05% dmin 0.004 dnn 0.060
23
+ f0.25_e3200 seed43 seed123: K1 med +1.40% (fail 0.000) K8 med +1.10% dmin 0.005 dnn 0.054
24
+ f0.25_e3200 seed43 seed321: K1 med +1.40% (fail 0.000) K8 med +0.97% dmin 0.004 dnn 0.046
25
+ f1.00_e200 seed42 seed123: K1 med +1.53% (fail 0.000) K8 med +0.72% dmin 0.012 dnn 0.041
26
+ f1.00_e200 seed42 seed321: K1 med +1.63% (fail 0.000) K8 med +0.78% dmin 0.012 dnn 0.050
27
+ f1.00_e200 seed43 seed123: K1 med +1.25% (fail 0.000) K8 med +0.27% dmin 0.012 dnn 0.049
28
+ f1.00_e200 seed43 seed321: K1 med +1.25% (fail 0.000) K8 med +0.30% dmin 0.012 dnn 0.051
results/ablations/logs/anomaly_report.log ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ target seed=2026 draws=(0, 1, 2)
2
+ === PENDULUM data-size anomaly diagnostics ===
3
+ held-out: seed123 (N=40) + seed321 (N=40), pooled N=80 plants; sampler draws (0, 1, 2); K=1 = first sample of each draw; K=8 = best of 8.
4
+ regret % vs strong per-plant oracle; failures (unstable) counted as +inf in the median, never dropped.
5
+ CI = 95% bootstrap over plants (5000) of the median pooled over (training seed x draw).
6
+
7
+ TARGET native regret (random 40 training targets, seed 2026, on own plants, K=1): median +1.41 [+1.27,+1.51] p90 +3.59 max +4.09 fail 0.000
8
+
9
+ config seeds K1 median [CI] K1 p90 K1 fail K8 median [CI] sd_seed K1 sd_seed K8 d_min d_nnplant
10
+ f0.25_e800 3 +1.16 [+0.99,+1.40] 4.18 0.000 +0.28 [+0.21,+0.37] 0.25 0.22 0.015 0.063
11
+ f0.50_e800 3 +1.48 [+1.33,+1.63] 3.06 0.000 +0.90 [+0.78,+1.04] 0.27 0.22 0.005 0.044
12
+ f1.00_e800 3 +1.32 [+1.15,+1.47] 3.09 0.000 +0.91 [+0.79,+1.03] 0.23 0.22 0.004 0.035
13
+ f0.25_e3200 2 +1.43 [+1.33,+1.58] 3.14 0.000 +1.07 [+0.95,+1.18] 0.03 0.08 0.004 0.049
14
+ f1.00_e200 2 +1.43 [+1.28,+1.63] 3.42 0.000 +0.52 [+0.45,+0.61] 0.28 0.31 0.012 0.046
15
+
16
+ per-plant paired Wilcoxon (plant = mean over seed x draw of regret; same 80 plants), two-sided:
17
+ f0.25_e800 vs f1.00_e800: K1 med diff +0.05pp p=0.146 (n=80) | K8 med diff -0.49pp p=8.46e-14 (n=80)
18
+ f0.25_e3200 vs f0.25_e800: K1 med diff +0.12pp p=0.77 (n=80) | K8 med diff +0.71pp p=3.38e-14 (n=80)
19
+ f0.25_e3200 vs f1.00_e800: K1 med diff +0.12pp p=1.59e-07 (n=80) | K8 med diff +0.12pp p=5.96e-09 (n=80)
20
+ f1.00_e200 vs f1.00_e800: K1 med diff +0.15pp p=0.00533 (n=80) | K8 med diff -0.34pp p=3.5e-11 (n=80)
21
+ f1.00_e200 vs f0.25_e800: K1 med diff +0.15pp p=0.181 (n=80) | K8 med diff +0.18pp p=1.7e-06 (n=80)
22
+ f0.25_e800 vs f0.50_e800: K1 med diff -0.11pp p=0.996 (n=80) | K8 med diff -0.48pp p=1.57e-13 (n=80)
23
+
24
+ Model K=1 vs native target regret (unpaired, different plants): compare K1 median column to the TARGET line above.
25
+ d_min = distance from a generated (clipped, normalized) controller to the nearest vector in that model's own training subset; d_nnplant = distance to the target of the nearest-condition training plant.
26
+
results/ablations/logs/anomaly_targets.log ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ target seed=2026 draws=(0, 1, 2)
2
+ [targets] seed=2026 idx=[67, 204, 253, 305, 379, 408, 427, 456, 458, 588, 721, 730, 769, 879, 912, 937, 951, 1163, 1194, 1199, 1334, 1562, 1640, 1644, 1653, 1670, 1682, 1701, 1812, 1857, 1880, 1948, 2033, 2136, 2140, 2181, 2212, 2329, 2376, 2496]
3
+ [targets] oracle 242s; stable 40/40; median +1.41%
results/ablations/logs/anomaly_train_p25_e3200_s42.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/anomaly_train_p25_e3200_s43.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/anomaly_train_ref_e200_s42.log ADDED
@@ -0,0 +1,403 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [ablation tag=steps_ref_e200] global seed=42
2
+ [ablation tag=steps_ref_e200] device=cuda | N=2600/2600 (frac=1.0) cond=3 out=11 | hidden=256 layers=5 epochs=200 seed=42
3
+
4
+ Epoch 1 | Avg Loss: 1.495968
5
+
6
+ Epoch 2 | Avg Loss: 1.426379
7
+
8
+ Epoch 3 | Avg Loss: 1.310095
9
+
10
+ Epoch 4 | Avg Loss: 1.199908
11
+
12
+ Epoch 5 | Avg Loss: 1.103204
13
+
14
+ Epoch 6 | Avg Loss: 1.058355
15
+
16
+ Epoch 7 | Avg Loss: 1.024843
17
+
18
+ Epoch 8 | Avg Loss: 1.003450
19
+
20
+ Epoch 9 | Avg Loss: 1.020821
21
+
22
+ Epoch 10 | Avg Loss: 0.980699
23
+
24
+ Epoch 11 | Avg Loss: 0.977503
25
+
26
+ Epoch 12 | Avg Loss: 0.960497
27
+
28
+ Epoch 13 | Avg Loss: 0.944223
29
+
30
+ Epoch 14 | Avg Loss: 0.910548
31
+
32
+ Epoch 15 | Avg Loss: 0.889922
33
+
34
+ Epoch 16 | Avg Loss: 0.845595
35
+
36
+ Epoch 17 | Avg Loss: 0.801918
37
+
38
+ Epoch 18 | Avg Loss: 0.718640
39
+
40
+ Epoch 19 | Avg Loss: 0.597195
41
+
42
+ Epoch 20 | Avg Loss: 0.464371
43
+
44
+ Epoch 21 | Avg Loss: 0.352728
45
+
46
+ Epoch 22 | Avg Loss: 0.261917
47
+
48
+ Epoch 23 | Avg Loss: 0.207426
49
+
50
+ Epoch 24 | Avg Loss: 0.179476
51
+
52
+ Epoch 25 | Avg Loss: 0.168767
53
+
54
+ Epoch 26 | Avg Loss: 0.151483
55
+
56
+ Epoch 27 | Avg Loss: 0.152414
57
+
58
+ Epoch 28 | Avg Loss: 0.138554
59
+
60
+ Epoch 29 | Avg Loss: 0.129946
61
+
62
+ Epoch 30 | Avg Loss: 0.129459
63
+
64
+ Epoch 31 | Avg Loss: 0.111340
65
+
66
+ Epoch 32 | Avg Loss: 0.102340
67
+
68
+ Epoch 33 | Avg Loss: 0.106668
69
+
70
+ Epoch 34 | Avg Loss: 0.091032
71
+
72
+ Epoch 35 | Avg Loss: 0.093791
73
+
74
+ Epoch 36 | Avg Loss: 0.090540
75
+
76
+ Epoch 37 | Avg Loss: 0.091575
77
+
78
+ Epoch 38 | Avg Loss: 0.072878
79
+
80
+ Epoch 39 | Avg Loss: 0.068464
81
+
82
+ Epoch 40 | Avg Loss: 0.070691
83
+
84
+ Epoch 41 | Avg Loss: 0.058929
85
+
86
+ Epoch 42 | Avg Loss: 0.067834
87
+
88
+ Epoch 43 | Avg Loss: 0.055984
89
+
90
+ Epoch 44 | Avg Loss: 0.068046
91
+
92
+ Epoch 45 | Avg Loss: 0.063330
93
+
94
+ Epoch 46 | Avg Loss: 0.056505
95
+
96
+ Epoch 47 | Avg Loss: 0.054347
97
+
98
+ Epoch 48 | Avg Loss: 0.050024
99
+
100
+ Epoch 49 | Avg Loss: 0.062993
101
+
102
+ Epoch 50 | Avg Loss: 0.046316
103
+
104
+ Epoch 51 | Avg Loss: 0.047549
105
+
106
+ Epoch 52 | Avg Loss: 0.044382
107
+
108
+ Epoch 53 | Avg Loss: 0.046137
109
+
110
+ Epoch 54 | Avg Loss: 0.048195
111
+
112
+ Epoch 55 | Avg Loss: 0.037185
113
+
114
+ Epoch 56 | Avg Loss: 0.043911
115
+
116
+ Epoch 57 | Avg Loss: 0.043157
117
+
118
+ Epoch 58 | Avg Loss: 0.041899
119
+
120
+ Epoch 59 | Avg Loss: 0.040101
121
+
122
+ Epoch 60 | Avg Loss: 0.038361
123
+
124
+ Epoch 61 | Avg Loss: 0.044358
125
+
126
+ Epoch 62 | Avg Loss: 0.045886
127
+
128
+ Epoch 63 | Avg Loss: 0.037714
129
+
130
+ Epoch 64 | Avg Loss: 0.037354
131
+
132
+ Epoch 65 | Avg Loss: 0.036634
133
+
134
+ Epoch 66 | Avg Loss: 0.042830
135
+
136
+ Epoch 67 | Avg Loss: 0.043315
137
+
138
+ Epoch 68 | Avg Loss: 0.031366
139
+
140
+ Epoch 69 | Avg Loss: 0.039762
141
+
142
+ Epoch 70 | Avg Loss: 0.027531
143
+
144
+ Epoch 71 | Avg Loss: 0.030802
145
+
146
+ Epoch 72 | Avg Loss: 0.031945
147
+
148
+ Epoch 73 | Avg Loss: 0.035084
149
+
150
+ Epoch 74 | Avg Loss: 0.029170
151
+
152
+ Epoch 75 | Avg Loss: 0.030183
153
+
154
+ Epoch 76 | Avg Loss: 0.029024
155
+
156
+ Epoch 77 | Avg Loss: 0.026033
157
+
158
+ Epoch 78 | Avg Loss: 0.030455
159
+
160
+ Epoch 79 | Avg Loss: 0.030781
161
+
162
+ Epoch 80 | Avg Loss: 0.025350
163
+
164
+ Epoch 81 | Avg Loss: 0.023338
165
+
166
+ Epoch 82 | Avg Loss: 0.031032
167
+
168
+ Epoch 83 | Avg Loss: 0.035429
169
+
170
+ Epoch 84 | Avg Loss: 0.032497
171
+
172
+ Epoch 85 | Avg Loss: 0.031067
173
+
174
+ Epoch 86 | Avg Loss: 0.029154
175
+
176
+ Epoch 87 | Avg Loss: 0.031285
177
+
178
+ Epoch 88 | Avg Loss: 0.029843
179
+
180
+ Epoch 89 | Avg Loss: 0.031510
181
+
182
+ Epoch 90 | Avg Loss: 0.027217
183
+
184
+ Epoch 91 | Avg Loss: 0.023474
185
+
186
+ Epoch 92 | Avg Loss: 0.022712
187
+
188
+ Epoch 93 | Avg Loss: 0.021010
189
+
190
+ Epoch 94 | Avg Loss: 0.021846
191
+
192
+ Epoch 95 | Avg Loss: 0.028999
193
+
194
+ Epoch 96 | Avg Loss: 0.030081
195
+
196
+ Epoch 97 | Avg Loss: 0.024542
197
+
198
+ Epoch 98 | Avg Loss: 0.024251
199
+
200
+ Epoch 99 | Avg Loss: 0.028572
201
+
202
+ Epoch 100 | Avg Loss: 0.021054
203
+
204
+ Epoch 101 | Avg Loss: 0.020133
205
+
206
+ Epoch 102 | Avg Loss: 0.026298
207
+
208
+ Epoch 103 | Avg Loss: 0.029485
209
+
210
+ Epoch 104 | Avg Loss: 0.024231
211
+
212
+ Epoch 105 | Avg Loss: 0.030632
213
+
214
+ Epoch 106 | Avg Loss: 0.028009
215
+
216
+ Epoch 107 | Avg Loss: 0.027977
217
+
218
+ Epoch 108 | Avg Loss: 0.029817
219
+
220
+ Epoch 109 | Avg Loss: 0.025838
221
+
222
+ Epoch 110 | Avg Loss: 0.026592
223
+
224
+ Epoch 111 | Avg Loss: 0.020609
225
+
226
+ Epoch 112 | Avg Loss: 0.016672
227
+
228
+ Epoch 113 | Avg Loss: 0.018964
229
+
230
+ Epoch 114 | Avg Loss: 0.018131
231
+
232
+ Epoch 115 | Avg Loss: 0.021116
233
+
234
+ Epoch 116 | Avg Loss: 0.020863
235
+
236
+ Epoch 117 | Avg Loss: 0.022576
237
+
238
+ Epoch 118 | Avg Loss: 0.018232
239
+
240
+ Epoch 119 | Avg Loss: 0.024216
241
+
242
+ Epoch 120 | Avg Loss: 0.025986
243
+
244
+ Epoch 121 | Avg Loss: 0.021779
245
+
246
+ Epoch 122 | Avg Loss: 0.022947
247
+
248
+ Epoch 123 | Avg Loss: 0.021138
249
+
250
+ Epoch 124 | Avg Loss: 0.028706
251
+
252
+ Epoch 125 | Avg Loss: 0.022871
253
+
254
+ Epoch 126 | Avg Loss: 0.017821
255
+
256
+ Epoch 127 | Avg Loss: 0.017886
257
+
258
+ Epoch 128 | Avg Loss: 0.025772
259
+
260
+ Epoch 129 | Avg Loss: 0.022348
261
+
262
+ Epoch 130 | Avg Loss: 0.023009
263
+
264
+ Epoch 131 | Avg Loss: 0.023511
265
+
266
+ Epoch 132 | Avg Loss: 0.025675
267
+
268
+ Epoch 133 | Avg Loss: 0.017721
269
+
270
+ Epoch 134 | Avg Loss: 0.019631
271
+
272
+ Epoch 135 | Avg Loss: 0.020990
273
+
274
+ Epoch 136 | Avg Loss: 0.016071
275
+
276
+ Epoch 137 | Avg Loss: 0.022021
277
+
278
+ Epoch 138 | Avg Loss: 0.018104
279
+
280
+ Epoch 139 | Avg Loss: 0.015225
281
+
282
+ Epoch 140 | Avg Loss: 0.024748
283
+
284
+ Epoch 141 | Avg Loss: 0.022209
285
+
286
+ Epoch 142 | Avg Loss: 0.020707
287
+
288
+ Epoch 143 | Avg Loss: 0.023914
289
+
290
+ Epoch 144 | Avg Loss: 0.018059
291
+
292
+ Epoch 145 | Avg Loss: 0.023437
293
+
294
+ Epoch 146 | Avg Loss: 0.027865
295
+
296
+ Epoch 147 | Avg Loss: 0.017218
297
+
298
+ Epoch 148 | Avg Loss: 0.016681
299
+
300
+ Epoch 149 | Avg Loss: 0.023856
301
+
302
+ Epoch 150 | Avg Loss: 0.017538
303
+
304
+ Epoch 151 | Avg Loss: 0.018230
305
+
306
+ Epoch 152 | Avg Loss: 0.013995
307
+
308
+ Epoch 153 | Avg Loss: 0.014388
309
+
310
+ Epoch 154 | Avg Loss: 0.020273
311
+
312
+ Epoch 155 | Avg Loss: 0.014749
313
+
314
+ Epoch 156 | Avg Loss: 0.017323
315
+
316
+ Epoch 157 | Avg Loss: 0.019504
317
+
318
+ Epoch 158 | Avg Loss: 0.014275
319
+
320
+ Epoch 159 | Avg Loss: 0.019660
321
+
322
+ Epoch 160 | Avg Loss: 0.015002
323
+
324
+ Epoch 161 | Avg Loss: 0.016291
325
+
326
+ Epoch 162 | Avg Loss: 0.019757
327
+
328
+ Epoch 163 | Avg Loss: 0.014090
329
+
330
+ Epoch 164 | Avg Loss: 0.014152
331
+
332
+ Epoch 165 | Avg Loss: 0.012632
333
+
334
+ Epoch 166 | Avg Loss: 0.021982
335
+
336
+ Epoch 167 | Avg Loss: 0.017552
337
+
338
+ Epoch 168 | Avg Loss: 0.020805
339
+
340
+ Epoch 169 | Avg Loss: 0.020367
341
+
342
+ Epoch 170 | Avg Loss: 0.014294
343
+
344
+ Epoch 171 | Avg Loss: 0.011229
345
+
346
+ Epoch 172 | Avg Loss: 0.014651
347
+
348
+ Epoch 173 | Avg Loss: 0.014460
349
+
350
+ Epoch 174 | Avg Loss: 0.020206
351
+
352
+ Epoch 175 | Avg Loss: 0.011293
353
+
354
+ Epoch 176 | Avg Loss: 0.013378
355
+
356
+ Epoch 177 | Avg Loss: 0.016422
357
+
358
+ Epoch 178 | Avg Loss: 0.012518
359
+
360
+ Epoch 179 | Avg Loss: 0.009513
361
+
362
+ Epoch 180 | Avg Loss: 0.013024
363
+
364
+ Epoch 181 | Avg Loss: 0.012743
365
+
366
+ Epoch 182 | Avg Loss: 0.010412
367
+
368
+ Epoch 183 | Avg Loss: 0.014172
369
+
370
+ Epoch 184 | Avg Loss: 0.009194
371
+
372
+ Epoch 185 | Avg Loss: 0.015252
373
+
374
+ Epoch 186 | Avg Loss: 0.009587
375
+
376
+ Epoch 187 | Avg Loss: 0.009375
377
+
378
+ Epoch 188 | Avg Loss: 0.009186
379
+
380
+ Epoch 189 | Avg Loss: 0.008580
381
+
382
+ Epoch 190 | Avg Loss: 0.008260
383
+
384
+ Epoch 191 | Avg Loss: 0.009128
385
+
386
+ Epoch 192 | Avg Loss: 0.016167
387
+
388
+ Epoch 193 | Avg Loss: 0.015715
389
+
390
+ Epoch 194 | Avg Loss: 0.010834
391
+
392
+ Epoch 195 | Avg Loss: 0.009286
393
+
394
+ Epoch 196 | Avg Loss: 0.008440
395
+
396
+ Epoch 197 | Avg Loss: 0.008857
397
+
398
+ Epoch 198 | Avg Loss: 0.008547
399
+
400
+ Epoch 199 | Avg Loss: 0.013890
401
+
402
+ Epoch 200 | Avg Loss: 0.010709
403
+ [ablation tag=steps_ref_e200] DONE train_time=62.5s -> models/Best_pendulum_abl_steps_ref_e200.pth
results/ablations/logs/anomaly_train_ref_e200_s43.log ADDED
@@ -0,0 +1,403 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [ablation tag=steps_ref_e200_s43] global seed=43
2
+ [ablation tag=steps_ref_e200_s43] device=cuda | N=2600/2600 (frac=1.0) cond=3 out=11 | hidden=256 layers=5 epochs=200 seed=43
3
+
4
+ Epoch 1 | Avg Loss: 1.331845
5
+
6
+ Epoch 2 | Avg Loss: 1.280549
7
+
8
+ Epoch 3 | Avg Loss: 1.219884
9
+
10
+ Epoch 4 | Avg Loss: 1.128403
11
+
12
+ Epoch 5 | Avg Loss: 1.083489
13
+
14
+ Epoch 6 | Avg Loss: 1.038621
15
+
16
+ Epoch 7 | Avg Loss: 1.041930
17
+
18
+ Epoch 8 | Avg Loss: 1.017390
19
+
20
+ Epoch 9 | Avg Loss: 0.982101
21
+
22
+ Epoch 10 | Avg Loss: 1.008808
23
+
24
+ Epoch 11 | Avg Loss: 0.988734
25
+
26
+ Epoch 12 | Avg Loss: 0.968774
27
+
28
+ Epoch 13 | Avg Loss: 0.943812
29
+
30
+ Epoch 14 | Avg Loss: 0.929535
31
+
32
+ Epoch 15 | Avg Loss: 0.918892
33
+
34
+ Epoch 16 | Avg Loss: 0.878545
35
+
36
+ Epoch 17 | Avg Loss: 0.818198
37
+
38
+ Epoch 18 | Avg Loss: 0.758015
39
+
40
+ Epoch 19 | Avg Loss: 0.617699
41
+
42
+ Epoch 20 | Avg Loss: 0.487430
43
+
44
+ Epoch 21 | Avg Loss: 0.354470
45
+
46
+ Epoch 22 | Avg Loss: 0.265449
47
+
48
+ Epoch 23 | Avg Loss: 0.216014
49
+
50
+ Epoch 24 | Avg Loss: 0.195541
51
+
52
+ Epoch 25 | Avg Loss: 0.180689
53
+
54
+ Epoch 26 | Avg Loss: 0.166731
55
+
56
+ Epoch 27 | Avg Loss: 0.155794
57
+
58
+ Epoch 28 | Avg Loss: 0.138344
59
+
60
+ Epoch 29 | Avg Loss: 0.132002
61
+
62
+ Epoch 30 | Avg Loss: 0.119396
63
+
64
+ Epoch 31 | Avg Loss: 0.120296
65
+
66
+ Epoch 32 | Avg Loss: 0.114527
67
+
68
+ Epoch 33 | Avg Loss: 0.111389
69
+
70
+ Epoch 34 | Avg Loss: 0.093437
71
+
72
+ Epoch 35 | Avg Loss: 0.093610
73
+
74
+ Epoch 36 | Avg Loss: 0.084453
75
+
76
+ Epoch 37 | Avg Loss: 0.076261
77
+
78
+ Epoch 38 | Avg Loss: 0.074230
79
+
80
+ Epoch 39 | Avg Loss: 0.068763
81
+
82
+ Epoch 40 | Avg Loss: 0.069243
83
+
84
+ Epoch 41 | Avg Loss: 0.060469
85
+
86
+ Epoch 42 | Avg Loss: 0.078794
87
+
88
+ Epoch 43 | Avg Loss: 0.056800
89
+
90
+ Epoch 44 | Avg Loss: 0.057668
91
+
92
+ Epoch 45 | Avg Loss: 0.063404
93
+
94
+ Epoch 46 | Avg Loss: 0.053596
95
+
96
+ Epoch 47 | Avg Loss: 0.051334
97
+
98
+ Epoch 48 | Avg Loss: 0.069718
99
+
100
+ Epoch 49 | Avg Loss: 0.058237
101
+
102
+ Epoch 50 | Avg Loss: 0.048498
103
+
104
+ Epoch 51 | Avg Loss: 0.050047
105
+
106
+ Epoch 52 | Avg Loss: 0.047092
107
+
108
+ Epoch 53 | Avg Loss: 0.046386
109
+
110
+ Epoch 54 | Avg Loss: 0.047934
111
+
112
+ Epoch 55 | Avg Loss: 0.044083
113
+
114
+ Epoch 56 | Avg Loss: 0.054595
115
+
116
+ Epoch 57 | Avg Loss: 0.044890
117
+
118
+ Epoch 58 | Avg Loss: 0.039290
119
+
120
+ Epoch 59 | Avg Loss: 0.043621
121
+
122
+ Epoch 60 | Avg Loss: 0.048209
123
+
124
+ Epoch 61 | Avg Loss: 0.046011
125
+
126
+ Epoch 62 | Avg Loss: 0.035286
127
+
128
+ Epoch 63 | Avg Loss: 0.035557
129
+
130
+ Epoch 64 | Avg Loss: 0.040820
131
+
132
+ Epoch 65 | Avg Loss: 0.047022
133
+
134
+ Epoch 66 | Avg Loss: 0.032354
135
+
136
+ Epoch 67 | Avg Loss: 0.033626
137
+
138
+ Epoch 68 | Avg Loss: 0.029426
139
+
140
+ Epoch 69 | Avg Loss: 0.033933
141
+
142
+ Epoch 70 | Avg Loss: 0.032379
143
+
144
+ Epoch 71 | Avg Loss: 0.029177
145
+
146
+ Epoch 72 | Avg Loss: 0.036466
147
+
148
+ Epoch 73 | Avg Loss: 0.031081
149
+
150
+ Epoch 74 | Avg Loss: 0.030817
151
+
152
+ Epoch 75 | Avg Loss: 0.033649
153
+
154
+ Epoch 76 | Avg Loss: 0.035819
155
+
156
+ Epoch 77 | Avg Loss: 0.030703
157
+
158
+ Epoch 78 | Avg Loss: 0.033565
159
+
160
+ Epoch 79 | Avg Loss: 0.031406
161
+
162
+ Epoch 80 | Avg Loss: 0.026447
163
+
164
+ Epoch 81 | Avg Loss: 0.028661
165
+
166
+ Epoch 82 | Avg Loss: 0.029487
167
+
168
+ Epoch 83 | Avg Loss: 0.024826
169
+
170
+ Epoch 84 | Avg Loss: 0.031393
171
+
172
+ Epoch 85 | Avg Loss: 0.036400
173
+
174
+ Epoch 86 | Avg Loss: 0.027908
175
+
176
+ Epoch 87 | Avg Loss: 0.024144
177
+
178
+ Epoch 88 | Avg Loss: 0.029012
179
+
180
+ Epoch 89 | Avg Loss: 0.023343
181
+
182
+ Epoch 90 | Avg Loss: 0.031519
183
+
184
+ Epoch 91 | Avg Loss: 0.032722
185
+
186
+ Epoch 92 | Avg Loss: 0.024847
187
+
188
+ Epoch 93 | Avg Loss: 0.030518
189
+
190
+ Epoch 94 | Avg Loss: 0.029010
191
+
192
+ Epoch 95 | Avg Loss: 0.023754
193
+
194
+ Epoch 96 | Avg Loss: 0.028472
195
+
196
+ Epoch 97 | Avg Loss: 0.030047
197
+
198
+ Epoch 98 | Avg Loss: 0.027516
199
+
200
+ Epoch 99 | Avg Loss: 0.023539
201
+
202
+ Epoch 100 | Avg Loss: 0.030678
203
+
204
+ Epoch 101 | Avg Loss: 0.030782
205
+
206
+ Epoch 102 | Avg Loss: 0.026083
207
+
208
+ Epoch 103 | Avg Loss: 0.027894
209
+
210
+ Epoch 104 | Avg Loss: 0.025313
211
+
212
+ Epoch 105 | Avg Loss: 0.029632
213
+
214
+ Epoch 106 | Avg Loss: 0.025393
215
+
216
+ Epoch 107 | Avg Loss: 0.021946
217
+
218
+ Epoch 108 | Avg Loss: 0.024098
219
+
220
+ Epoch 109 | Avg Loss: 0.018446
221
+
222
+ Epoch 110 | Avg Loss: 0.030313
223
+
224
+ Epoch 111 | Avg Loss: 0.024365
225
+
226
+ Epoch 112 | Avg Loss: 0.027607
227
+
228
+ Epoch 113 | Avg Loss: 0.025421
229
+
230
+ Epoch 114 | Avg Loss: 0.023045
231
+
232
+ Epoch 115 | Avg Loss: 0.020398
233
+
234
+ Epoch 116 | Avg Loss: 0.019592
235
+
236
+ Epoch 117 | Avg Loss: 0.020794
237
+
238
+ Epoch 118 | Avg Loss: 0.019916
239
+
240
+ Epoch 119 | Avg Loss: 0.026700
241
+
242
+ Epoch 120 | Avg Loss: 0.023696
243
+
244
+ Epoch 121 | Avg Loss: 0.021271
245
+
246
+ Epoch 122 | Avg Loss: 0.019292
247
+
248
+ Epoch 123 | Avg Loss: 0.017333
249
+
250
+ Epoch 124 | Avg Loss: 0.024239
251
+
252
+ Epoch 125 | Avg Loss: 0.029444
253
+
254
+ Epoch 126 | Avg Loss: 0.022161
255
+
256
+ Epoch 127 | Avg Loss: 0.020286
257
+
258
+ Epoch 128 | Avg Loss: 0.020092
259
+
260
+ Epoch 129 | Avg Loss: 0.022430
261
+
262
+ Epoch 130 | Avg Loss: 0.023691
263
+
264
+ Epoch 131 | Avg Loss: 0.022987
265
+
266
+ Epoch 132 | Avg Loss: 0.018329
267
+
268
+ Epoch 133 | Avg Loss: 0.017003
269
+
270
+ Epoch 134 | Avg Loss: 0.016946
271
+
272
+ Epoch 135 | Avg Loss: 0.019491
273
+
274
+ Epoch 136 | Avg Loss: 0.016549
275
+
276
+ Epoch 137 | Avg Loss: 0.016862
277
+
278
+ Epoch 138 | Avg Loss: 0.020227
279
+
280
+ Epoch 139 | Avg Loss: 0.026544
281
+
282
+ Epoch 140 | Avg Loss: 0.022958
283
+
284
+ Epoch 141 | Avg Loss: 0.033135
285
+
286
+ Epoch 142 | Avg Loss: 0.021594
287
+
288
+ Epoch 143 | Avg Loss: 0.022072
289
+
290
+ Epoch 144 | Avg Loss: 0.020449
291
+
292
+ Epoch 145 | Avg Loss: 0.018700
293
+
294
+ Epoch 146 | Avg Loss: 0.020733
295
+
296
+ Epoch 147 | Avg Loss: 0.020617
297
+
298
+ Epoch 148 | Avg Loss: 0.016742
299
+
300
+ Epoch 149 | Avg Loss: 0.015184
301
+
302
+ Epoch 150 | Avg Loss: 0.019570
303
+
304
+ Epoch 151 | Avg Loss: 0.018965
305
+
306
+ Epoch 152 | Avg Loss: 0.018154
307
+
308
+ Epoch 153 | Avg Loss: 0.017541
309
+
310
+ Epoch 154 | Avg Loss: 0.026421
311
+
312
+ Epoch 155 | Avg Loss: 0.018892
313
+
314
+ Epoch 156 | Avg Loss: 0.019163
315
+
316
+ Epoch 157 | Avg Loss: 0.014673
317
+
318
+ Epoch 158 | Avg Loss: 0.018613
319
+
320
+ Epoch 159 | Avg Loss: 0.018231
321
+
322
+ Epoch 160 | Avg Loss: 0.014686
323
+
324
+ Epoch 161 | Avg Loss: 0.020296
325
+
326
+ Epoch 162 | Avg Loss: 0.012193
327
+
328
+ Epoch 163 | Avg Loss: 0.015851
329
+
330
+ Epoch 164 | Avg Loss: 0.016620
331
+
332
+ Epoch 165 | Avg Loss: 0.019167
333
+
334
+ Epoch 166 | Avg Loss: 0.015478
335
+
336
+ Epoch 167 | Avg Loss: 0.020442
337
+
338
+ Epoch 168 | Avg Loss: 0.018287
339
+
340
+ Epoch 169 | Avg Loss: 0.015543
341
+
342
+ Epoch 170 | Avg Loss: 0.016698
343
+
344
+ Epoch 171 | Avg Loss: 0.015689
345
+
346
+ Epoch 172 | Avg Loss: 0.012983
347
+
348
+ Epoch 173 | Avg Loss: 0.017482
349
+
350
+ Epoch 174 | Avg Loss: 0.013971
351
+
352
+ Epoch 175 | Avg Loss: 0.013678
353
+
354
+ Epoch 176 | Avg Loss: 0.016297
355
+
356
+ Epoch 177 | Avg Loss: 0.010570
357
+
358
+ Epoch 178 | Avg Loss: 0.014016
359
+
360
+ Epoch 179 | Avg Loss: 0.015208
361
+
362
+ Epoch 180 | Avg Loss: 0.016265
363
+
364
+ Epoch 181 | Avg Loss: 0.012155
365
+
366
+ Epoch 182 | Avg Loss: 0.012626
367
+
368
+ Epoch 183 | Avg Loss: 0.009315
369
+
370
+ Epoch 184 | Avg Loss: 0.010399
371
+
372
+ Epoch 185 | Avg Loss: 0.013034
373
+
374
+ Epoch 186 | Avg Loss: 0.008753
375
+
376
+ Epoch 187 | Avg Loss: 0.009084
377
+
378
+ Epoch 188 | Avg Loss: 0.013807
379
+
380
+ Epoch 189 | Avg Loss: 0.008235
381
+
382
+ Epoch 190 | Avg Loss: 0.009443
383
+
384
+ Epoch 191 | Avg Loss: 0.014782
385
+
386
+ Epoch 192 | Avg Loss: 0.016910
387
+
388
+ Epoch 193 | Avg Loss: 0.011737
389
+
390
+ Epoch 194 | Avg Loss: 0.012276
391
+
392
+ Epoch 195 | Avg Loss: 0.009440
393
+
394
+ Epoch 196 | Avg Loss: 0.008987
395
+
396
+ Epoch 197 | Avg Loss: 0.013303
397
+
398
+ Epoch 198 | Avg Loss: 0.011389
399
+
400
+ Epoch 199 | Avg Loss: 0.015639
401
+
402
+ Epoch 200 | Avg Loss: 0.016009
403
+ [ablation tag=steps_ref_e200_s43] DONE train_time=62.4s -> models/Best_pendulum_abl_steps_ref_e200_s43.pth
results/ablations/logs/eval_generative.log ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ device=cuda
2
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/pendulum_lmi.py:149: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
3
+ prob.solve(solver=solver, verbose=False)
4
+ [gen-baseline/pendulum] draw seed=0 done
5
+ [gen-baseline/pendulum] draw seed=1 done
6
+ [gen-baseline/pendulum] draw seed=2 done
7
+
8
+ === A3 GENERATIVE BASELINE — PENDULUM (N=40, shared cached oracle, 3 draws/K, seeds=[0, 1, 2]) ===
9
+
10
+ budget: diffusion params=3,823,883 epochs=800 time_to_best=232.0s; cvae params=277,019 epochs=800 time_to_best=278.9s
11
+
12
+ -- K=1 --
13
+ diffusion : stable>=1 1.000+-0.000 cert>=1 0.992+-0.014 persample 1.000+-0.000 median_regret +1.52%+-0.06pp diversity 0.0000+-0.0000
14
+ cvae : stable>=1 1.000+-0.000 cert>=1 1.000+-0.000 persample 1.000+-0.000 median_regret +1.37%+-0.05pp diversity 0.0000+-0.0000
15
+ paired McNemar (diffusion vs cvae, pooled 120 plant x draw trials): cvae wins 0/0 discordant, p=1
16
+
17
+ -- K=8 --
18
+ diffusion : stable>=1 1.000+-0.000 cert>=1 1.000+-0.000 persample 1.000+-0.000 median_regret +1.11%+-0.09pp diversity 0.0142+-0.0007
19
+ cvae : stable>=1 1.000+-0.000 cert>=1 1.000+-0.000 persample 1.000+-0.000 median_regret +1.20%+-0.02pp diversity 0.0027+-0.0001
20
+ paired McNemar (diffusion vs cvae, pooled 120 plant x draw trials): cvae wins 0/0 discordant, p=1
21
+
22
+ [timing] pendulum arm: 131.6s
23
+ [gen-baseline/siso] draw seed=0 done
24
+ [gen-baseline/siso] draw seed=1 done
25
+ [gen-baseline/siso] draw seed=2 done
26
+
27
+ === A3 GENERATIVE BASELINE — SISO (N=60 plants, plant_seed=100, 3 draws/K, seeds=[0, 1, 2]) ===
28
+
29
+ budget: diffusion params=21,445,653 epochs=1200 time_to_best=7940.6s; cvae params=1,108,021 epochs=60 time_to_best=33.9s; regressor params=804,373 epochs=40 time=not recorded
30
+
31
+ -- K=1 --
32
+ diffusion : plant_stabilisable 0.667+-0.033 persample 0.667+-0.033 median_best_J 6.713+-3.633 diversity 0.0000+-0.0000
33
+ cvae : plant_stabilisable 0.750+-0.050 persample 0.750+-0.050 median_best_J 5.167+-2.512 diversity 0.0000+-0.0000
34
+ regressor : plant_stabilisable 0.750+-0.044 persample 0.750+-0.044 median_best_J 2.403+-0.141 diversity 0.0000+-0.0000
35
+ paired McNemar (diffusion vs cvae, pooled 180 plant x draw trials): cvae wins 18/21 discordant, p=0.00149
36
+ paired McNemar (diffusion vs regressor, pooled 180 plant x draw trials): regressor wins 19/23 discordant, p=0.0026
37
+
38
+ -- K=8 --
39
+ diffusion : plant_stabilisable 0.800+-0.050 persample 0.640+-0.024 median_best_J 1.363+-0.120 diversity 0.1844+-0.0076
40
+ cvae : plant_stabilisable 0.767+-0.033 persample 0.751+-0.045 median_best_J 0.748+-0.113 diversity 0.0034+-0.0001
41
+ regressor : plant_stabilisable 0.750+-0.044 persample 0.750+-0.044 median_best_J 2.403+-0.141 diversity 0.0000+-0.0000
42
+ paired McNemar (diffusion vs cvae, pooled 180 plant x draw trials): cvae wins 0/6 discordant, p=0.0312
43
+ paired McNemar (diffusion vs regressor, pooled 180 plant x draw trials): regressor wins 0/9 discordant, p=0.00391
44
+
45
+ [timing] siso arm: 488.6s
results/ablations/logs/eval_generative_siso_seed200.log ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ device=cuda mode=siso plant_seed=200
2
+ [gen-baseline/siso] draw seed=0 done
3
+ [gen-baseline/siso] draw seed=1 done
4
+ [gen-baseline/siso] draw seed=2 done
5
+
6
+ === A3 GENERATIVE BASELINE — SISO (N=60 plants, plant_seed=200, 3 draws/K, seeds=[0, 1, 2]) ===
7
+
8
+ budget: diffusion params=21,445,653 epochs=1200 time_to_best=7940.6s; cvae params=1,108,021 epochs=60 time_to_best=33.9s; regressor params=804,373 epochs=40 time=not recorded
9
+
10
+ -- K=1 --
11
+ diffusion : plant_stabilisable 0.644+-0.067 persample 0.644+-0.067 median_best_J 11.448+-5.757 diversity 0.0000+-0.0000
12
+ cvae : plant_stabilisable 0.739+-0.025 persample 0.739+-0.025 median_best_J 5.993+-2.409 diversity 0.0000+-0.0000
13
+ regressor : plant_stabilisable 0.750+-0.029 persample 0.750+-0.029 median_best_J 2.380+-0.277 diversity 0.0000+-0.0000
14
+ paired McNemar (diffusion vs cvae, pooled 180 plant x draw trials): cvae wins 18/19 discordant, p=7.63e-05
15
+ paired McNemar (diffusion vs regressor, pooled 180 plant x draw trials): regressor wins 19/19 discordant, p=3.81e-06
16
+
17
+ -- K=8 --
18
+ diffusion : plant_stabilisable 0.756+-0.069 persample 0.615+-0.017 median_best_J 1.796+-0.349 diversity 0.1773+-0.0099
19
+ cvae : plant_stabilisable 0.756+-0.038 persample 0.733+-0.030 median_best_J 0.770+-0.091 diversity 0.0034+-0.0001
20
+ regressor : plant_stabilisable 0.750+-0.029 persample 0.750+-0.029 median_best_J 2.380+-0.277 diversity 0.0000+-0.0000
21
+ paired McNemar (diffusion vs cvae, pooled 180 plant x draw trials): cvae wins 2/4 discordant, p=1
22
+ paired McNemar (diffusion vs regressor, pooled 180 plant x draw trials): regressor wins 2/5 discordant, p=1
23
+
24
+ [timing] siso arm: 604.6s
results/ablations/logs/eval_generative_siso_seed300.log ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ device=cuda mode=siso plant_seed=300
2
+ [gen-baseline/siso] draw seed=0 done
3
+ [gen-baseline/siso] draw seed=1 done
4
+ [gen-baseline/siso] draw seed=2 done
5
+
6
+ === A3 GENERATIVE BASELINE — SISO (N=60 plants, plant_seed=300, 3 draws/K, seeds=[0, 1, 2]) ===
7
+
8
+ budget: diffusion params=21,445,653 epochs=1200 time_to_best=7940.6s; cvae params=1,108,021 epochs=60 time_to_best=33.9s; regressor params=804,373 epochs=40 time=not recorded
9
+
10
+ -- K=1 --
11
+ diffusion : plant_stabilisable 0.672+-0.054 persample 0.672+-0.054 median_best_J 7.642+-3.111 diversity 0.0000+-0.0000
12
+ cvae : plant_stabilisable 0.811+-0.035 persample 0.811+-0.035 median_best_J 5.722+-3.110 diversity 0.0000+-0.0000
13
+ regressor : plant_stabilisable 0.806+-0.025 persample 0.806+-0.025 median_best_J 2.596+-0.253 diversity 0.0000+-0.0000
14
+ paired McNemar (diffusion vs cvae, pooled 180 plant x draw trials): cvae wins 26/27 discordant, p=4.17e-07
15
+ paired McNemar (diffusion vs regressor, pooled 180 plant x draw trials): regressor wins 25/26 discordant, p=8.05e-07
16
+
17
+ -- K=8 --
18
+ diffusion : plant_stabilisable 0.839+-0.025 persample 0.681+-0.016 median_best_J 1.486+-0.162 diversity 0.1781+-0.0086
19
+ cvae : plant_stabilisable 0.817+-0.033 persample 0.806+-0.023 median_best_J 0.913+-0.223 diversity 0.0034+-0.0001
20
+ regressor : plant_stabilisable 0.806+-0.025 persample 0.806+-0.025 median_best_J 2.596+-0.253 diversity 0.0000+-0.0000
21
+ paired McNemar (diffusion vs cvae, pooled 180 plant x draw trials): cvae wins 0/4 discordant, p=0.125
22
+ paired McNemar (diffusion vs regressor, pooled 180 plant x draw trials): regressor wins 0/6 discordant, p=0.0312
23
+
24
+ [timing] siso arm: 550.7s
results/ablations/logs/eval_generative_siso_seed400.log ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ device=cuda mode=siso plant_seed=400
2
+ [gen-baseline/siso] draw seed=0 done
3
+ [gen-baseline/siso] draw seed=1 done
4
+ [gen-baseline/siso] draw seed=2 done
5
+
6
+ === A3 GENERATIVE BASELINE — SISO (N=60 plants, plant_seed=400, 3 draws/K, seeds=[0, 1, 2]) ===
7
+
8
+ budget: diffusion params=21,445,653 epochs=1200 time_to_best=7940.6s; cvae params=1,108,021 epochs=60 time_to_best=33.9s; regressor params=804,373 epochs=40 time=not recorded
9
+
10
+ -- K=1 --
11
+ diffusion : plant_stabilisable 0.622+-0.042 persample 0.622+-0.042 median_best_J 9.790+-5.143 diversity 0.0000+-0.0000
12
+ cvae : plant_stabilisable 0.694+-0.025 persample 0.694+-0.025 median_best_J 4.571+-1.419 diversity 0.0000+-0.0000
13
+ regressor : plant_stabilisable 0.700+-0.044 persample 0.700+-0.044 median_best_J 2.706+-0.383 diversity 0.0000+-0.0000
14
+ paired McNemar (diffusion vs cvae, pooled 180 plant x draw trials): cvae wins 20/27 discordant, p=0.0192
15
+ paired McNemar (diffusion vs regressor, pooled 180 plant x draw trials): regressor wins 20/26 discordant, p=0.00936
16
+
17
+ -- K=8 --
18
+ diffusion : plant_stabilisable 0.761+-0.035 persample 0.608+-0.017 median_best_J 1.968+-0.313 diversity 0.1875+-0.0109
19
+ cvae : plant_stabilisable 0.739+-0.042 persample 0.697+-0.031 median_best_J 0.908+-0.171 diversity 0.0034+-0.0000
20
+ regressor : plant_stabilisable 0.700+-0.044 persample 0.700+-0.044 median_best_J 2.706+-0.383 diversity 0.0000+-0.0000
21
+ paired McNemar (diffusion vs cvae, pooled 180 plant x draw trials): cvae wins 1/6 discordant, p=0.219
22
+ paired McNemar (diffusion vs regressor, pooled 180 plant x draw trials): regressor wins 1/13 discordant, p=0.00342
23
+
24
+ [timing] siso arm: 544.8s
results/ablations/logs/eval_generative_siso_seed500.log ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ device=cuda mode=siso plant_seed=500
2
+ [gen-baseline/siso] draw seed=0 done
3
+ [gen-baseline/siso] draw seed=1 done
4
+ [gen-baseline/siso] draw seed=2 done
5
+
6
+ === A3 GENERATIVE BASELINE — SISO (N=60 plants, plant_seed=500, 3 draws/K, seeds=[0, 1, 2]) ===
7
+
8
+ budget: diffusion params=21,445,653 epochs=1200 time_to_best=7940.6s; cvae params=1,108,021 epochs=60 time_to_best=33.9s; regressor params=804,373 epochs=40 time=not recorded
9
+
10
+ -- K=1 --
11
+ diffusion : plant_stabilisable 0.533+-0.017 persample 0.533+-0.017 median_best_J 5.136+-2.005 diversity 0.0000+-0.0000
12
+ cvae : plant_stabilisable 0.633+-0.017 persample 0.633+-0.017 median_best_J 5.881+-2.407 diversity 0.0000+-0.0000
13
+ regressor : plant_stabilisable 0.639+-0.025 persample 0.639+-0.025 median_best_J 2.357+-0.177 diversity 0.0000+-0.0000
14
+ paired McNemar (diffusion vs cvae, pooled 180 plant x draw trials): cvae wins 21/24 discordant, p=0.000277
15
+ paired McNemar (diffusion vs regressor, pooled 180 plant x draw trials): regressor wins 20/21 discordant, p=2.1e-05
16
+
17
+ -- K=8 --
18
+ diffusion : plant_stabilisable 0.711+-0.048 persample 0.537+-0.028 median_best_J 1.432+-0.271 diversity 0.1897+-0.0077
19
+ cvae : plant_stabilisable 0.661+-0.019 persample 0.635+-0.018 median_best_J 0.774+-0.171 diversity 0.0034+-0.0001
20
+ regressor : plant_stabilisable 0.639+-0.025 persample 0.639+-0.025 median_best_J 2.357+-0.177 diversity 0.0000+-0.0000
21
+ paired McNemar (diffusion vs cvae, pooled 180 plant x draw trials): cvae wins 2/13 discordant, p=0.0225
22
+ paired McNemar (diffusion vs regressor, pooled 180 plant x draw trials): regressor wins 1/15 discordant, p=0.000977
23
+
24
+ [timing] siso arm: 540.3s
results/ablations/logs/eval_size_data.log ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ device=cuda training seeds=[42, 43, 44] eval noise seed=0
2
+ [size] hidden=64 (~1/4 width) (base_tag=size_quarter, seeds=[42, 43, 44])
3
+ evaluating size_quarter seed=42 <- models/Best_pendulum_abl_size_quarter.pth (eval noise seed=0)
4
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/pendulum_lmi.py:149: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
5
+ prob.solve(solver=solver, verbose=False)
6
+ evaluating size_quarter seed=43 <- models/Best_pendulum_abl_size_quarter_s43.pth (eval noise seed=0)
7
+ evaluating size_quarter seed=44 <- models/Best_pendulum_abl_size_quarter_s44.pth (eval noise seed=0)
8
+ [size] hidden=256 (fair reference, ablation trainer) (base_tag=size_ref, seeds=[42, 43, 44])
9
+ evaluating size_ref seed=42 <- models/Best_pendulum_abl_size_ref.pth (eval noise seed=0)
10
+ evaluating size_ref seed=43 <- models/Best_pendulum_abl_size_ref_s43.pth (eval noise seed=0)
11
+ evaluating size_ref seed=44 <- models/Best_pendulum_abl_size_ref_s44.pth (eval noise seed=0)
12
+ [size] hidden=512 (~2x width) (base_tag=size_2x, seeds=[42, 43, 44])
13
+ evaluating size_2x seed=42 <- models/Best_pendulum_abl_size_2x.pth (eval noise seed=0)
14
+ evaluating size_2x seed=43 <- models/Best_pendulum_abl_size_2x_s43.pth (eval noise seed=0)
15
+ evaluating size_2x seed=44 <- models/Best_pendulum_abl_size_2x_s44.pth (eval noise seed=0)
16
+ [size] hidden=256 (DEPLOYED checkpoint -- pre-existing train_pendulum.py run, not part of the 3-seed sweep) <- models/Best_pendulum_cond.pth
17
+
18
+ === SIZE ABLATION (pendulum, N=40, K=8, shared cached oracle, eval noise seed=0, 3 training seeds=[42, 43, 44]) ===
19
+
20
+ - hidden=64 (~1/4 width)
21
+ params=255,563 time_to_best=405.9s
22
+ median regret vs oracle: +0.68% +/- 0.16pp over seeds [s42=+0.58%, s43=+0.87%, s44=+0.60%]
23
+ coverage (mean over seeds): certified>=1/plant 0.992 stable>=1/plant 1.000 per-sample stable 0.999
24
+ - hidden=256 (fair reference, ablation trainer)
25
+ params=3,823,883 time_to_best=421.4s
26
+ median regret vs oracle: +0.92% +/- 0.22pp over seeds [s42=+1.06%, s43=+0.67%, s44=+1.03%]
27
+ coverage (mean over seeds): certified>=1/plant 1.000 stable>=1/plant 1.000 per-sample stable 1.000
28
+ - hidden=512 (~2x width)
29
+ params=15,118,859 time_to_best=441.1s
30
+ median regret vs oracle: +0.86% +/- 0.38pp over seeds [s42=+0.98%, s43=+0.43%, s44=+1.16%]
31
+ coverage (mean over seeds): certified>=1/plant 1.000 stable>=1/plant 1.000 per-sample stable 1.000
32
+ - hidden=256 (DEPLOYED checkpoint -- pre-existing train_pendulum.py run, not part of the 3-seed sweep)
33
+ params=3,823,883 time_to_best=232.0s
34
+ median regret vs oracle: +1.15% +/- 0.00pp over seeds [ssingle=+1.15%]
35
+ coverage (mean over seeds): certified>=1/plant 1.000 stable>=1/plant 1.000 per-sample stable 1.000
36
+ [data] 25% data (650/2600 examples) (base_tag=data_p25, seeds=[42, 43, 44])
37
+ evaluating data_p25 seed=42 <- models/Best_pendulum_abl_data_p25.pth (eval noise seed=0)
38
+ evaluating data_p25 seed=43 <- models/Best_pendulum_abl_data_p25_s43.pth (eval noise seed=0)
39
+ evaluating data_p25 seed=44 <- models/Best_pendulum_abl_data_p25_s44.pth (eval noise seed=0)
40
+ [data] 50% data (1300/2600 examples) (base_tag=data_p50, seeds=[42, 43, 44])
41
+ evaluating data_p50 seed=42 <- models/Best_pendulum_abl_data_p50.pth (eval noise seed=0)
42
+ evaluating data_p50 seed=43 <- models/Best_pendulum_abl_data_p50_s43.pth (eval noise seed=0)
43
+ evaluating data_p50 seed=44 <- models/Best_pendulum_abl_data_p50_s44.pth (eval noise seed=0)
44
+ [data] 100% data (= size_ref, same 3 seeds, reused) (base_tag=size_ref, seeds=[42, 43, 44])
45
+ [data] hidden=256 (DEPLOYED checkpoint -- pre-existing train_pendulum.py run, not part of the 3-seed sweep) <- models/Best_pendulum_cond.pth
46
+
47
+ === DATA ABLATION (pendulum, N=40, K=8, shared cached oracle, eval noise seed=0, 3 training seeds=[42, 43, 44]) ===
48
+
49
+ - 25% data (650/2600 examples)
50
+ params=3,823,883 time_to_best=261.8s
51
+ median regret vs oracle: +0.29% +/- 0.27pp over seeds [s42=+0.61%, s43=+0.16%, s44=+0.12%]
52
+ coverage (mean over seeds): certified>=1/plant 0.992 stable>=1/plant 1.000 per-sample stable 1.000
53
+ - 50% data (1300/2600 examples)
54
+ params=3,823,883 time_to_best=355.6s
55
+ median regret vs oracle: +1.00% +/- 0.23pp over seeds [s42=+1.22%, s43=+0.76%, s44=+1.02%]
56
+ coverage (mean over seeds): certified>=1/plant 1.000 stable>=1/plant 1.000 per-sample stable 1.000
57
+ - 100% data (= size_ref, same 3 seeds, reused)
58
+ params=3,823,883 time_to_best=421.4s
59
+ median regret vs oracle: +0.92% +/- 0.22pp over seeds [s42=+1.06%, s43=+0.67%, s44=+1.03%]
60
+ coverage (mean over seeds): certified>=1/plant 1.000 stable>=1/plant 1.000 per-sample stable 1.000
61
+ - hidden=256 (DEPLOYED checkpoint -- pre-existing train_pendulum.py run, not part of the 3-seed sweep)
62
+ params=3,823,883 time_to_best=232.0s
63
+ median regret vs oracle: +1.15% +/- 0.00pp over seeds [ssingle=+1.15%]
64
+ coverage (mean over seeds): certified>=1/plant 1.000 stable>=1/plant 1.000 per-sample stable 1.000
results/ablations/logs/eval_steps.log ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Tue Sep 29 06:26:57 AM PDT 2026
2
+ device=cuda draw seeds=[0, 1, 2] checkpoint=models/Best_pendulum_cond.pth
3
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/pendulum_lmi.py:149: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
4
+ prob.solve(solver=solver, verbose=False)
5
+ (quality-check block took 104.4s)
6
+ steps=5 done in 57.2s
7
+ steps=10 done in 61.2s
8
+ steps=20 done in 64.1s
9
+ steps=50 done in 62.8s
10
+ steps=100 done in 50.9s
11
+ steps=500 done in 56.6s
12
+
13
+ === A4 STEP-0: DDIM@500 vs DDPM@500 quality check (3 draws each, seeds=[0, 1, 2]) ===
14
+
15
+ DDPM@500 : stable 1.000+-0.000 cert 1.000+-0.000 regret +1.11%+-0.09pp gen 1.747+-0.400 ms/ctrl
16
+ DDIM@500 : stable 1.000+-0.000 cert 1.000+-0.000 regret +1.05%+-0.05pp gen 2.218+-0.743 ms/ctrl
17
+ VERDICT: DDIM@500 MATCHES DDPM@500 within noise (|stable diff|=0.000, |regret diff|=0.06pp)
18
+
19
+ === A4 STEPS SWEEP (DDIM, eta=0, N=40, K=8, 3 draws/step-count, seeds=[0, 1, 2]) ===
20
+
21
+ steps= 5 stable>=1 1.000+-0.000 cert>=1 1.000+-0.000 persample 0.995+-0.004 median_regret +0.96%+-0.14pp wall 0.0304+-0.0069 ms/ctrl
22
+ steps= 10 stable>=1 1.000+-0.000 cert>=1 1.000+-0.000 persample 0.999+-0.002 median_regret +1.06%+-0.05pp wall 0.0362+-0.0095 ms/ctrl
23
+ steps= 20 stable>=1 1.000+-0.000 cert>=1 1.000+-0.000 persample 0.999+-0.002 median_regret +1.06%+-0.05pp wall 0.0982+-0.0376 ms/ctrl
24
+ steps= 50 stable>=1 1.000+-0.000 cert>=1 1.000+-0.000 persample 0.999+-0.002 median_regret +1.08%+-0.04pp wall 0.2618+-0.0955 ms/ctrl
25
+ steps= 100 stable>=1 1.000+-0.000 cert>=1 1.000+-0.000 persample 1.000+-0.000 median_regret +1.08%+-0.05pp wall 0.4116+-0.0402 ms/ctrl
26
+ steps= 500 stable>=1 1.000+-0.000 cert>=1 1.000+-0.000 persample 0.999+-0.002 median_regret +1.05%+-0.05pp wall 1.8847+-0.3124 ms/ctrl
27
+
28
+ [exited with code 0]
results/ablations/logs/eval_steps_mimo.log ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ device=cuda ckpt=models/Best_mimo_cond_2x2_opt.pth draw seeds=[0, 1, 2]
2
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/mimo_lmi.py:29: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
3
+ prob.solve(solver=cp.SCS, verbose=False)
4
+ steps=5 done in 187.7s
5
+ steps=20 done in 183.9s
6
+ steps=100 done in 211.3s
7
+ steps=500 done in 200.8s
8
+
9
+ === A4 MIMO bonus: DDPM@500 anchor + DDIM steps sweep (N=60, K=6, 3 draws, seeds=[0, 1, 2]) ===
10
+
11
+ DDPM@500 anchor: per-sample-stable 0.975+-0.006 median_regret_vs_phi0 +0.5%+-0.1 gen 3.952+-1.367 ms/ctrl
12
+
13
+ steps= 5 per-sample-stable 0.975+-0.007 has_stable>=1 1.000+-0.000 within_sector 0.890+-0.018 median_regret_vs_phi0 +1.2%+-0.5 gen 0.0357+-0.0205 ms/ctrl
14
+ steps= 20 per-sample-stable 0.972+-0.007 has_stable>=1 0.994+-0.010 within_sector 0.840+-0.013 median_regret_vs_phi0 +0.6%+-0.5 gen 0.2460+-0.2445 ms/ctrl
15
+ steps= 100 per-sample-stable 0.964+-0.007 has_stable>=1 0.994+-0.010 within_sector 0.789+-0.005 median_regret_vs_phi0 +0.4%+-0.2 gen 0.4678+-0.0237 ms/ctrl
16
+ steps= 500 per-sample-stable 0.968+-0.018 has_stable>=1 0.994+-0.010 within_sector 0.778+-0.007 median_regret_vs_phi0 +0.3%+-0.2 gen 2.6111+-0.1449 ms/ctrl
results/ablations/logs/train_cvae_pendulum.log ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [cvae pendulum] device=cuda N=2600 cond=3 out=11 hidden=256 latent=8 layers=3 params=277019 epochs=800 batch=256 seed=42
2
+ epoch 0 | val MSE 0.04926 (best 0.04926) beta=0.011
3
+ epoch 40 | val MSE 0.00003 (best 0.00003) beta=0.511
4
+ epoch 80 | val MSE 0.00002 (best 0.00002) beta=1.000
5
+ epoch 120 | val MSE 0.00002 (best 0.00001) beta=1.000
6
+ epoch 160 | val MSE 0.00001 (best 0.00001) beta=1.000
7
+ epoch 200 | val MSE 0.00001 (best 0.00000) beta=1.000
8
+ epoch 240 | val MSE 0.00001 (best 0.00000) beta=1.000
9
+ epoch 280 | val MSE 0.00002 (best 0.00000) beta=1.000
10
+ epoch 320 | val MSE 0.00001 (best 0.00000) beta=1.000
11
+ epoch 360 | val MSE 0.00000 (best 0.00000) beta=1.000
12
+ epoch 400 | val MSE 0.00001 (best 0.00000) beta=1.000
13
+ epoch 440 | val MSE 0.00001 (best 0.00000) beta=1.000
14
+ epoch 480 | val MSE 0.00000 (best 0.00000) beta=1.000
15
+ epoch 520 | val MSE 0.00000 (best 0.00000) beta=1.000
16
+ epoch 560 | val MSE 0.00000 (best 0.00000) beta=1.000
17
+ epoch 600 | val MSE 0.00000 (best 0.00000) beta=1.000
18
+ epoch 640 | val MSE 0.00000 (best 0.00000) beta=1.000
19
+ epoch 680 | val MSE 0.00000 (best 0.00000) beta=1.000
20
+ epoch 720 | val MSE 0.00000 (best 0.00000) beta=1.000
21
+ epoch 760 | val MSE 0.00000 (best 0.00000) beta=1.000
22
+ epoch 799 | val MSE 0.00000 (best 0.00000) beta=1.000
23
+ [cvae pendulum] DONE best_val_mse=0.00000 train_time=315.8s params=277019 -> models/Best_cvae_pendulum.pth
results/ablations/logs/train_cvae_siso.log ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [cvae siso] device=cuda N=180160 cond=10 out=21 hidden=512 latent=16 layers=3 params=1108021 epochs=60 batch=2048 seed=42
2
+ epoch 0 | val MSE 0.00177 (best 0.00177) beta=0.165
3
+ epoch 3 | val MSE 0.00176 (best 0.00176) beta=0.665
4
+ epoch 6 | val MSE 0.00176 (best 0.00176) beta=1.000
5
+ epoch 9 | val MSE 0.00176 (best 0.00176) beta=1.000
6
+ epoch 12 | val MSE 0.00176 (best 0.00176) beta=1.000
7
+ epoch 15 | val MSE 0.00176 (best 0.00176) beta=1.000
8
+ epoch 18 | val MSE 0.00176 (best 0.00176) beta=1.000
9
+ epoch 21 | val MSE 0.00176 (best 0.00176) beta=1.000
10
+ epoch 24 | val MSE 0.00176 (best 0.00176) beta=1.000
11
+ epoch 27 | val MSE 0.00176 (best 0.00176) beta=1.000
12
+ epoch 30 | val MSE 0.00176 (best 0.00176) beta=1.000
13
+ epoch 33 | val MSE 0.00176 (best 0.00176) beta=1.000
14
+ epoch 36 | val MSE 0.00176 (best 0.00176) beta=1.000
15
+ epoch 39 | val MSE 0.00176 (best 0.00176) beta=1.000
16
+ epoch 42 | val MSE 0.00176 (best 0.00176) beta=1.000
17
+ epoch 45 | val MSE 0.00176 (best 0.00175) beta=1.000
18
+ epoch 48 | val MSE 0.00175 (best 0.00175) beta=1.000
19
+ epoch 51 | val MSE 0.00175 (best 0.00175) beta=1.000
20
+ epoch 54 | val MSE 0.00175 (best 0.00175) beta=1.000
21
+ epoch 57 | val MSE 0.00176 (best 0.00175) beta=1.000
22
+ epoch 59 | val MSE 0.00176 (best 0.00175) beta=1.000
23
+ [cvae siso] DONE best_val_mse=0.00175 train_time=41.3s params=1108021 -> models/Best_cvae_siso.pth
results/ablations/logs/train_data_p25.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_data_p25_s43.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_data_p25_s44.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_data_p50.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_data_p50_s43.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_data_p50_s44.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_size_2x.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_size_2x_s43.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_size_2x_s44.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_size_quarter.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_size_quarter_s43.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_size_quarter_s44.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_size_ref_s42.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_size_ref_s43.log ADDED
The diff for this file is too large to render. See raw diff
 
results/ablations/logs/train_size_ref_s44.log ADDED
The diff for this file is too large to render. See raw diff
 
results/furuta/catch_audit.txt ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === FURUTA HAND-OFF AUDIT (held-out N=8) ===
2
+ certified = the balance gain engages at a state inside a saturation-free INVARIANT
3
+ ellipsoid (furuta_roa). The gain being certifiable is necessary but not sufficient.
4
+
5
+ --- generated ---
6
+ full-cycle success : 100.0%
7
+ balance GAIN certifiable (§4.12) : 100.0%
8
+ hand-off STATE inside certified set: 0.0%
9
+ x'Px at the hand-off (<=1 certified): med 1.82 p95 3.34 max 3.73
10
+ |Kx| demanded at the hand-off : med 7.00 p95 8.98 max 9.05 V (bound 10.0 V)
11
+
12
+ --- classical ---
13
+ full-cycle success : 100.0%
14
+ balance GAIN certifiable (§4.12) : 100.0%
15
+ hand-off STATE inside certified set: 0.0%
16
+ x'Px at the hand-off (<=1 certified): med 2.15 p95 5.20 max 5.97
17
+ |Kx| demanded at the hand-off : med 7.20 p95 9.00 max 9.01 V (bound 10.0 V)
18
+
19
+ --- the gate as a SET (worst corner of {|a|<a_catch, |ad|<w_catch} at the observed theta) ---
20
+ x'Px worst corner: med 10.94 p95 12.72 max 12.73
21
+ >1 means the gate admits states the certificate does not cover, even when the
22
+ particular state reached happens to be fine.
23
+
results/furuta/collocation_ceiling.txt ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === architecture-free ceiling: direct collocation (IPOPT) ===
2
+ Swing-up optimised freely into the SAME certified ellipsoid the two-stage arm uses,
3
+ then the SAME certified gain balances. Identical cost weights, IC and 4 s horizon, so
4
+ the only difference is the swing-up law: energy shaping vs an optimised trajectory.
5
+
6
+ plant J_collo J_2stage ratio T_su t_ipopt
7
+ 0 3.960 5.804 1.47x 0.50 0.9s
8
+ 2 3.809 4.961 1.30x 0.48 0.6s
9
+ 3 4.496 5.846 1.30x 0.55 1.0s
10
+ 4 4.183 5.475 1.31x 0.52 0.6s
11
+ 5 -- 6.668 n/a -- 2.6s (IPOPT did not converge)
12
+ 6 3.754 5.900 1.57x 0.47 0.8s
13
+ 7 4.382 5.412 1.24x 0.56 0.6s
14
+ 8 4.473 5.850 1.31x 0.56 0.6s
15
+ 9 -- 10.633 n/a 2.17 1.8s (replay failed: caught=False, fe=15.208)
16
+ 10 5.082 5.596 1.10x 0.66 0.7s
17
+ 11 4.055 5.511 1.36x 0.51 0.6s
18
+ 12 4.279 12.403 2.90x 0.54 1.6s
19
+ 13 4.361 13.187 3.02x 0.58 2.0s
20
+ 14 4.501 18.574 4.13x 1.43 1.6s
21
+ 16 4.280 15.157 3.54x 1.61 1.2s
22
+ 17 4.156 5.692 1.37x 0.53 0.9s
23
+ 18 3.683 5.748 1.56x 0.46 0.9s
24
+ 19 4.345 12.268 2.82x 1.01 1.5s
25
+ 20 3.824 10.833 2.83x 0.86 1.4s
26
+ 21 -- 8.833 n/a 3.70 1.5s (replay failed: caught=False, fe=209.667)
27
+
28
+ median J_2stage / J_collocation = 1.47x (n=17)
29
+ range 1.10x .. 4.13x
30
+
31
+ Reading: ~1.0 means energy shaping is already about as good as any swing-up into
32
+ the same certified set, so the model inherits nothing bad from the architecture.
33
+ Well above 1.0 means the ARCHITECTURE, not the synthesis method, is the bottleneck —
34
+ as value iteration showed for the 2-state pendulum (1.16x median, 1.47x worst) —
35
+ and the certified-hand-off comparison would need reframing.
36
+
37
+ Collocation is a local method on a nonconvex problem, so the ratio is a LOWER bound
38
+ on the architecture's suboptimality: a better solution could only widen the gap.
results/furuta/eval.txt ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ held-out Furuta N=60, K=8/plant (region |alpha|<=0.35, Vmax=10.0 V)
2
+ regional certification: per-sample 1.00, >=1 certified 1.00
3
+ empirical balance (>=1 balances region under bound): 0.75
4
+ regret vs per-plant LQR oracle: median +0.0% mean +0.9% (n=45)
5
+ generation: 5.18s for 480 gains (10.8 ms each); per-plant LQR 2.3 ms
results/furuta/eval_full.txt ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ held-out Furuta N=60, K=4/plant (region |alpha|<=0.35, Vmax=10.0 V, Dr=0.0005, Dp=5e-05 pending ID)
2
+ GENERATED FULL controller (swing-up + certified balance, one diffusion pass):
3
+ full-cycle success (>=1 sample): 1.00 per-sample: 1.00
4
+ balance-stage regional certification (per-sample): 1.00
5
+ swing-up hand-off time: median 0.65s (n=60)
6
+ generation: single-plant 391 ms; batched 13.8 ms/controller (240 in 3.31s)
7
+ CLASSICAL baseline (energy-swing-up + LQR, per-plant):
8
+ full-cycle success: 1.00
9
+ synthesis: Riccati 2.3 ms + swing-up forward-sim search 1120 ms = 1122 ms/plant
10
+ NOTE: LQR ALONE cannot do this task (local balance only); the composite generated controller
11
+ is nonlinear. Win = amortisation (one batched pass for the whole family) + certified balance
12
+ stage; the per-plant classical swing-up tuning is a forward-sim search with no closed form.
results/furuta/friction_drift.txt ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === friction drift: is re-synthesis worth anything? (N=40, 3 draws) ===
2
+ Both Dr and Dp scaled by f. 'stale' was synthesised at f=1; 'regenerated' is told the
3
+ true drifted friction; 're-solved' is the per-plant search at the true friction.
4
+
5
+ f stale regenerated re-solved resolve s
6
+ 0.4 15.8% 29.2% 27.5% 5.1s
7
+ 0.6 41.7% 48.3% 47.5% 4.9s
8
+ 1.0 51.7% 55.8% 57.5% 4.8s
9
+ 1.6 48.3% 59.2% 60.0% 6.0s
10
+ 2.5 23.3% 35.8% 40.0% 6.9s
11
+
12
+ Reading: 'regenerated' above 'stale' means telling the model the true friction helps,
13
+ so synthesis is on the critical path and its speed is worth something. Equal columns
14
+ mean re-synthesis buys nothing and the 10x speedup has no use here.
15
+
16
+ mean over drifts: stale 36.2% regenerated 45.7% re-solved 46.5%
17
+ largest regenerated-minus-stale gap at any drift: +13.3 points (at f=0.4)
results/furuta/friction_drift_analysis.txt ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === friction drift, analysed against its own noise floor (N=40, 3 draws, complete) ===
2
+
3
+ f stale regenerated re-solved regen-stale vs the +4.1 null
4
+ 0.4 15.8% 29.2% 27.5% +13.4 ~3.2 se real
5
+ 0.6 41.7% 48.3% 47.5% +6.6 ~1.6 se within noise
6
+ 1.0 51.7% 55.8% 57.5% +4.1 the null itself
7
+ 1.6 48.3% 59.2% 60.0% +10.9 ~2.6 se real
8
+ 2.5 23.3% 35.8% 40.0% +12.5 ~3.0 se real
9
+ mean 36.2% 45.7% 46.5% +9.5
10
+
11
+ THE f=1.0 ROW IS THE CONTROL. There 'stale' and 'regenerated' are the SAME conditioning
12
+ -- both told the true friction -- so its +4.1 points is pure sampler noise, not an
13
+ effect. sample_diffusion is unseeded and this arm has sd 4.5 points per draw
14
+ (results/furuta/gen_variance.txt). Every other row must be read against that floor.
15
+
16
+ CONCLUSION. Re-synthesis is worth nothing while friction is within ~30-60% of what was
17
+ assumed, and worth 11 to 13 points once it has drifted by a factor of 1.6 or more, in
18
+ either direction. The shape is a clean U with its minimum at the control point, which is
19
+ what a real effect should look like and what pure noise would not.
20
+
21
+ This supersedes the +-30% arm in handoff_eval.txt (stale 90.0% vs regenerated 90.0%,
22
+ paired 4-4). That range sits inside the noise floor, so the null there was uninformative
23
+ rather than negative.
24
+
25
+ WHERE THE MODEL FALLS BEHIND. At f=2.5 the per-plant solve reaches 40.0% against the
26
+ model's 35.8%, the only level where the gap favours the solver by more than the noise.
27
+ f=2.5 puts Dp at 0.75x the datasheet value, well outside the 0.6x cap the model was
28
+ trained under, so the model degrades where it is extrapolating furthest while the solver
29
+ does not. That is the expected failure direction and it bounds the claim: conditioning on
30
+ friction pays across the trained range and starts giving way beyond it.
31
+
32
+ COST. 'regenerated' tracks 're-solved' within noise at f <= 1.6 while costing 0.8 s
33
+ against 4.8-6.9 s, roughly a sixth. That is the one setting on a bench rig where
34
+ synthesis speed buys anything: friction drifting within a session, not the
35
+ once-per-swap synthesis nobody waits on.
results/furuta/gen_variance.txt ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === generated arm: independent draws (N=60 plants, K=4, 5 draws) ===
2
+ Same checkpoint, same plants, same verifier. Only the sampler's noise differs.
3
+ The sampler is not seeded anywhere in the evaluation path.
4
+
5
+ draw float64 float32 gain certifies gen ms
6
+ 0 63.3% 63.3% 100.0% 139.4
7
+ 1 75.0% 70.0% 100.0% 138.7
8
+ 2 71.7% 70.0% 100.0% 137.4
9
+ 3 66.7% 61.7% 100.0% 137.8
10
+ 4 68.3% 66.7% 100.0% 150.7
11
+
12
+ float64 scorer: mean 69.0% sd 4.5 range 63.3-75.0
13
+ float32 scorer: mean 66.3% sd 3.8 range 61.7-70.0
14
+ gain certifies: mean 100.0%
15
+
16
+ results/furuta/handoff_eval.txt reported 83.3% from a single unseeded draw.
17
+ Compare that against the range above before treating it as the arm's success rate.
results/furuta/handoff_eval.txt ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === CERTIFIED HAND-OFF EVAL (held-out N=60) ===
2
+ 'certified' = the balance gain engages at a state inside a saturation-free invariant
3
+ ellipsoid for that same gain. Success without certification is NOT counted.
4
+
5
+ arm certified hand-off synthesis / plant
6
+ generated (K=4) 83.3% 46.6 ms
7
+ generated (single shot) 70.0% 11.7 ms
8
+ per-plant search 90.0% 2.8 s
9
+ interpolation table 61.7% ~0
10
+ shipped classical 21.7% (90% success)
11
+
12
+ generated: gain certifies a region 100.0%, speedup vs per-plant 60x
13
+ peak |Kx| at the hand-off: median 3.64 V of 10.0 V
14
+
15
+ --- paired vs generated (McNemar, exact) ---
16
+ generated beats per-plant search on 1, loses on 5 -> p = 0.2188
17
+ generated beats interpolation on 13, loses on 0 -> p = 0.0002 *
18
+ generated beats classical on 37, loses on 0 -> p = 0.0000 *
19
+
20
+ --- friction robustness (the sim->rig question) ---
21
+ the controller is generated for the friction we THINK the rig has;
22
+ the rig's true friction differs. Conditioning should make this recoverable.
23
+ true friction 0.7x reported: stale 90.0% regenerated 90.0% (paired 4-4, p=1.000)
24
+ true friction 1.3x reported: stale 76.7% regenerated 76.7% (paired 2-2, p=1.000)
25
+
26
+ --- RETRACTION (2026-08-26) ---
27
+ The 83.3% above does not reproduce. Five independent sampler draws with the same checkpoint,
28
+ the same 60 plants and this same float32 scorer give 61.7 / 63.3 / 66.7 / 70.0 / 70.0 percent
29
+ (mean 66.3, sd 3.8), so 83.3 is roughly 4.5 sd out. sample_diffusion is unseeded, so every run
30
+ proposes different controllers and this arm's rate is a random variable that was reported as a
31
+ point estimate. Everything checkable is identical between runs: checkpoint, conditioning,
32
+ decode, k_e conversion, catch gate, dt, horizon, candidate ordering, and the gain certifies on
33
+ 100% of plants either way. Cause not established.
34
+ Consequence: the "generated beats interpolation 13-0, p=0.0002" line above is computed from one
35
+ draw of an arm with sd 3.8, and separately dies at matched candidate budget (8-5, p=0.58).
36
+ See results/furuta/gen_variance.txt and docs/furuta_strong_baselines.md.
37
+
38
+ --- FRICTION-ROBUSTNESS ARM SUPERSEDED (2026-08-26) ---
39
+ The "+-30%: stale 90.0% vs regenerated 90.0%, paired 4-4" result above is uninformative,
40
+ not negative. +-30% sits inside this arm's own sampler noise floor: at f=1.0, where stale
41
+ and regenerated are by construction the SAME conditioning, two independent draws differ by
42
+ 4.1 points (results/furuta/friction_drift_partial.txt). Widening the sweep shows a real
43
+ effect once friction drifts by ~1.6x: +13.4 points at f=0.4 and +10.9 at f=1.6, both about
44
+ 3 sd above that null, while f=0.6 (+6.6) stays inside it. Regenerating also tracks the
45
+ per-plant re-solve at every level, at 0.8 s against ~5 s.
results/furuta/handoff_gate.txt ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === FURUTA CERTIFIED HAND-OFF GATE (held-out N=80) ===
2
+ family: mp(0.018, 0.03) Lp(0.1, 0.16) Dr(0.0003, 0.0025) Dp(2e-05, 0.0003) (Dp capped at 0.6x nominal)
3
+
4
+ --- A. per-plant search (the thing a model would amortize) ---
5
+ certified hand-off found: 93.8% (2.7 s/plant)
6
+ winning q_alpha histogram: 1:18 3:8 10:21 30:11 100:17
7
+ feasible fraction of the (gain,k_e) grid: median 88.6% min 1.0% (small = a sharp target a table would miss)
8
+
9
+ --- B. interpolation from a table of per-plant solutions ---
10
+ table: 4^4 = 256 solves, 203 feasible
11
+ interpolated gain CERTIFIES a region: 90.0%
12
+ and its swing-up REACHES that region: 71.2% (vs 93.8% for per-plant search)
13
+ no usable table corner: 10.0%
14
+
15
+ --- C. best FIXED recipe (per-plant Riccati, no search) ---
16
+ best (q_alpha=100, R=1, ke_mult=960)
17
+ gain certifies a region: 100.0%
18
+ certified hand-off achieved: 75.0% (vs 93.8% for per-plant search)
19
+
20
+ --- D. shipped classical (Riccati + box-gate swing-up search, i.e. 4.12) ---
21
+ full-cycle success: 90.0%
22
+ hand-off inside a certified set: 22.5%
23
+
24
+ --- paired comparison vs A (McNemar, exact) ---
25
+ A beats B interpolation on 18 plants, loses on 0 -> exact p = 0.0000 *
26
+ A beats C fixed recipe on 15 plants, loses on 0 -> exact p = 0.0001 *
27
+
28
+ GATE: pass only if B and C lose a large share of CERTIFIED hand-offs that A keeps.
29
+ If a table or a fixed recipe gets them nearly for free, stop and write up the audit.
results/furuta/handoff_voltage.txt ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === |Kx| demanded at the hand-off (bound 10 V, N=60) ===
2
+ Applied voltage is clipped at 10 V for every arm, so this is the pre-clip demand:
3
+ what the gain would need at the state where it takes over.
4
+
5
+ arm median p95 max plants
6
+ classical 8.56 16.90 19.16 54
7
+ fixed recipe 2.25 5.95 6.69 32
8
+ interp K=1 3.08 5.28 5.84 31
9
+ interpolation 3.46 5.27 5.84 38
10
+ generated 3.43 5.71 6.44 42
11
+ per-plant 3.79 5.68 6.34 46
12
+ NMPC n/a (never hands off to a static gain)
13
+ NMPC+cert 3.78 5.13 5.88 57
14
+ collo+TVLQR 5.38 6.65 6.90 42
results/furuta/logs/bank_v2_build.log ADDED
@@ -0,0 +1,234 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ solving 3000 plants on 16 workers (NMPC swing-up into Omega, margin=0.3, seed=13) -> data/furuta_nmpc_bank3000.npz
2
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
3
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
4
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
5
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
6
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
7
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
8
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
9
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
10
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
11
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
12
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
13
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
14
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
15
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
16
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
17
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
18
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
19
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
20
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
21
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
22
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
23
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
24
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
25
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
26
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
27
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
28
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
29
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
30
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
31
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
32
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
33
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
34
+
35
+ ******************************************************************************
36
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
37
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
38
+ For more information visit https://github.com/coin-or/Ipopt
39
+ ******************************************************************************
40
+
41
+
42
+ ******************************************************************************
43
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
44
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
45
+ For more information visit https://github.com/coin-or/Ipopt
46
+ ******************************************************************************
47
+
48
+
49
+ ******************************************************************************
50
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
51
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
52
+ For more information visit https://github.com/coin-or/Ipopt
53
+ ******************************************************************************
54
+
55
+
56
+ ******************************************************************************
57
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
58
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
59
+ For more information visit https://github.com/coin-or/Ipopt
60
+ ******************************************************************************
61
+
62
+
63
+ ******************************************************************************
64
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
65
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
66
+ For more information visit https://github.com/coin-or/Ipopt
67
+ ******************************************************************************
68
+
69
+
70
+ ******************************************************************************
71
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
72
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
73
+ For more information visit https://github.com/coin-or/Ipopt
74
+ ******************************************************************************
75
+
76
+
77
+ ******************************************************************************
78
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
79
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
80
+ For more information visit https://github.com/coin-or/Ipopt
81
+ ******************************************************************************
82
+
83
+
84
+ ******************************************************************************
85
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
86
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
87
+ For more information visit https://github.com/coin-or/Ipopt
88
+ ******************************************************************************
89
+
90
+
91
+ ******************************************************************************
92
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
93
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
94
+ For more information visit https://github.com/coin-or/Ipopt
95
+ ******************************************************************************
96
+
97
+
98
+ ******************************************************************************
99
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
100
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
101
+ For more information visit https://github.com/coin-or/Ipopt
102
+ ******************************************************************************
103
+
104
+
105
+ ******************************************************************************
106
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
107
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
108
+ For more information visit https://github.com/coin-or/Ipopt
109
+ ******************************************************************************
110
+
111
+
112
+ ******************************************************************************
113
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
114
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
115
+ For more information visit https://github.com/coin-or/Ipopt
116
+ ******************************************************************************
117
+
118
+
119
+ ******************************************************************************
120
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
121
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
122
+ For more information visit https://github.com/coin-or/Ipopt
123
+ ******************************************************************************
124
+
125
+
126
+ ******************************************************************************
127
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
128
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
129
+ For more information visit https://github.com/coin-or/Ipopt
130
+ ******************************************************************************
131
+
132
+
133
+ ******************************************************************************
134
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
135
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
136
+ For more information visit https://github.com/coin-or/Ipopt
137
+ ******************************************************************************
138
+
139
+
140
+ ******************************************************************************
141
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
142
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
143
+ For more information visit https://github.com/coin-or/Ipopt
144
+ ******************************************************************************
145
+
146
+ 40/3000 solved, 35 kept, 1.3 s/plant, eta 62.7 min
147
+ 80/3000 solved, 72 kept, 1.2 s/plant, eta 56.8 min
148
+ CasADi - 2026-09-29 06:38:10 WARNING("solver:nlp_jac_g failed: NaN detected for output jac_g_x, at nonzero index 412 (row 101, col 97).") [.../casadi/core/oracle_function.cpp:408]
149
+ 120/3000 solved, 109 kept, 1.1 s/plant, eta 52.7 min
150
+ 160/3000 solved, 145 kept, 1.1 s/plant, eta 50.4 min
151
+ 200/3000 solved, 181 kept, 1.0 s/plant, eta 48.1 min
152
+ 240/3000 solved, 219 kept, 1.0 s/plant, eta 47.8 min
153
+ 280/3000 solved, 257 kept, 1.0 s/plant, eta 47.2 min
154
+ 320/3000 solved, 294 kept, 1.1 s/plant, eta 47.2 min
155
+ CasADi - 2026-09-29 06:42:18 WARNING("solver:nlp_g failed: NaN detected for output g, at (row 101, col 0).") [.../casadi/core/oracle_function.cpp:408]
156
+ CasADi - 2026-09-29 06:42:18 WARNING("solver:nlp_g failed: NaN detected for output g, at (row 101, col 0).") [.../casadi/core/oracle_function.cpp:408]
157
+ CasADi - 2026-09-29 06:42:18 WARNING("solver:nlp_g failed: NaN detected for output g, at (row 101, col 0).") [.../casadi/core/oracle_function.cpp:408]
158
+ 360/3000 solved, 332 kept, 1.1 s/plant, eta 48.0 min
159
+ 400/3000 solved, 372 kept, 1.1 s/plant, eta 48.2 min
160
+ 440/3000 solved, 411 kept, 1.1 s/plant, eta 47.2 min
161
+ 480/3000 solved, 451 kept, 1.1 s/plant, eta 46.5 min
162
+ 520/3000 solved, 488 kept, 1.1 s/plant, eta 46.0 min
163
+ 560/3000 solved, 526 kept, 1.1 s/plant, eta 46.1 min
164
+ 600/3000 solved, 561 kept, 1.1 s/plant, eta 45.7 min
165
+ 640/3000 solved, 597 kept, 1.1 s/plant, eta 44.7 min
166
+ 680/3000 solved, 635 kept, 1.1 s/plant, eta 44.0 min
167
+ 720/3000 solved, 674 kept, 1.1 s/plant, eta 43.3 min
168
+ 760/3000 solved, 713 kept, 1.1 s/plant, eta 42.6 min
169
+ 800/3000 solved, 749 kept, 1.1 s/plant, eta 41.7 min
170
+ 840/3000 solved, 784 kept, 1.1 s/plant, eta 41.2 min
171
+ 880/3000 solved, 822 kept, 1.1 s/plant, eta 40.6 min
172
+ 920/3000 solved, 858 kept, 1.2 s/plant, eta 40.3 min
173
+ 960/3000 solved, 896 kept, 1.2 s/plant, eta 40.0 min
174
+ 1000/3000 solved, 935 kept, 1.2 s/plant, eta 39.3 min
175
+ 1040/3000 solved, 972 kept, 1.2 s/plant, eta 38.7 min
176
+ 1080/3000 solved, 1009 kept, 1.2 s/plant, eta 38.2 min
177
+ 1120/3000 solved, 1048 kept, 1.2 s/plant, eta 37.4 min
178
+ 1160/3000 solved, 1084 kept, 1.2 s/plant, eta 36.8 min
179
+ 1200/3000 solved, 1122 kept, 1.2 s/plant, eta 35.8 min
180
+ CasADi - 2026-09-29 07:00:05 WARNING("solver:nlp_g failed: NaN detected for output g, at (row 101, col 0).") [.../casadi/core/oracle_function.cpp:408]
181
+ CasADi - 2026-09-29 07:00:05 WARNING("solver:nlp_jac_g failed: NaN detected for output jac_g_x, at nonzero index 412 (row 101, col 97).") [.../casadi/core/oracle_function.cpp:408]
182
+ CasADi - 2026-09-29 07:00:06 WARNING("solver:nlp_jac_g failed: NaN detected for output jac_g_x, at nonzero index 412 (row 101, col 97).") [.../casadi/core/oracle_function.cpp:408]
183
+ 1240/3000 solved, 1158 kept, 1.2 s/plant, eta 35.0 min
184
+ 1280/3000 solved, 1192 kept, 1.2 s/plant, eta 34.1 min
185
+ 1320/3000 solved, 1230 kept, 1.2 s/plant, eta 33.1 min
186
+ 1360/3000 solved, 1267 kept, 1.2 s/plant, eta 32.3 min
187
+ 1400/3000 solved, 1305 kept, 1.2 s/plant, eta 31.4 min
188
+ 1440/3000 solved, 1345 kept, 1.2 s/plant, eta 30.5 min
189
+ 1480/3000 solved, 1382 kept, 1.2 s/plant, eta 29.6 min
190
+ 1520/3000 solved, 1418 kept, 1.2 s/plant, eta 28.8 min
191
+ 1560/3000 solved, 1456 kept, 1.2 s/plant, eta 28.0 min
192
+ 1600/3000 solved, 1494 kept, 1.2 s/plant, eta 27.2 min
193
+ 1640/3000 solved, 1534 kept, 1.2 s/plant, eta 26.5 min
194
+ 1680/3000 solved, 1570 kept, 1.2 s/plant, eta 25.7 min
195
+ 1720/3000 solved, 1607 kept, 1.2 s/plant, eta 24.9 min
196
+ 1760/3000 solved, 1645 kept, 1.2 s/plant, eta 24.1 min
197
+ 1800/3000 solved, 1685 kept, 1.2 s/plant, eta 23.2 min
198
+ 1840/3000 solved, 1720 kept, 1.2 s/plant, eta 22.4 min
199
+ 1880/3000 solved, 1757 kept, 1.2 s/plant, eta 21.7 min
200
+ 1920/3000 solved, 1795 kept, 1.2 s/plant, eta 20.9 min
201
+ 1960/3000 solved, 1834 kept, 1.2 s/plant, eta 20.3 min
202
+ 2000/3000 solved, 1874 kept, 1.2 s/plant, eta 19.5 min
203
+ 2040/3000 solved, 1909 kept, 1.2 s/plant, eta 18.7 min
204
+ 2080/3000 solved, 1946 kept, 1.2 s/plant, eta 17.8 min
205
+ 2120/3000 solved, 1980 kept, 1.2 s/plant, eta 17.0 min
206
+ 2160/3000 solved, 2017 kept, 1.2 s/plant, eta 16.3 min
207
+ 2200/3000 solved, 2054 kept, 1.2 s/plant, eta 15.5 min
208
+ 2240/3000 solved, 2092 kept, 1.2 s/plant, eta 14.7 min
209
+ 2280/3000 solved, 2130 kept, 1.2 s/plant, eta 13.9 min
210
+ 2320/3000 solved, 2168 kept, 1.2 s/plant, eta 13.2 min
211
+ 2360/3000 solved, 2204 kept, 1.2 s/plant, eta 12.4 min
212
+ 2400/3000 solved, 2241 kept, 1.2 s/plant, eta 11.6 min
213
+ 2440/3000 solved, 2278 kept, 1.2 s/plant, eta 10.8 min
214
+ 2480/3000 solved, 2318 kept, 1.2 s/plant, eta 10.0 min
215
+ 2520/3000 solved, 2352 kept, 1.2 s/plant, eta 9.2 min
216
+ 2560/3000 solved, 2388 kept, 1.2 s/plant, eta 8.5 min
217
+ 2600/3000 solved, 2426 kept, 1.2 s/plant, eta 7.7 min
218
+ 2640/3000 solved, 2465 kept, 1.2 s/plant, eta 6.9 min
219
+ 2680/3000 solved, 2503 kept, 1.1 s/plant, eta 6.1 min
220
+ 2720/3000 solved, 2542 kept, 1.1 s/plant, eta 5.4 min
221
+ 2760/3000 solved, 2578 kept, 1.1 s/plant, eta 4.6 min
222
+ 2800/3000 solved, 2616 kept, 1.1 s/plant, eta 3.8 min
223
+ 2840/3000 solved, 2654 kept, 1.1 s/plant, eta 3.0 min
224
+ 2880/3000 solved, 2693 kept, 1.1 s/plant, eta 2.3 min
225
+ 2920/3000 solved, 2729 kept, 1.1 s/plant, eta 1.5 min
226
+ 2960/3000 solved, 2765 kept, 1.1 s/plant, eta 0.8 min
227
+ 3000/3000 solved, 2801 kept, 1.1 s/plant, eta 0.0 min
228
+
229
+ bank: 2801/3000 kept (93.4%)
230
+ dropped 143: NMPC never reached Omega
231
+ dropped 56: replay failed
232
+ median per-plant solve 17.0 s
233
+ median achieved J 4.051 T range 0.41..1.53 s
234
+ saved -> data/furuta_nmpc_bank3000.npz
results/furuta/logs/deploy_extras.log ADDED
@@ -0,0 +1,138 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ extras: 2160 tasks
2
+ extras: 360 remaining
3
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
4
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
5
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
6
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
7
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
8
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
9
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
10
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
11
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
12
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
13
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
14
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
15
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
16
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
17
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
18
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
19
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
20
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
21
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
22
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
23
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
24
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
25
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
26
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
27
+ extra 200/360 (23s)
28
+
29
+ === retrieval + re-solved certificate as the deployable Furuta controller (n=60 x 2 draws = 120 plants) ===
30
+ bank data/furuta_nmpc_bank3000.npz (2801 entries, margin 0.3); diffusion models/Best_furuta_traj_cond_v2.pth; draws 4242, 9001; sampler seeds (0, 1, 2); deploy K=8
31
+
32
+ --- TASK 1: nominal, retrieval+P at K (screened + executed on the nominal plant) ---
33
+ per-plant NMPC solve ceiling (reused from traj_gen_ksweep.npz): 93.3% (93.3 / 93.3 per draw)
34
+ K succ 95% CI draw4242 draw9001 hand-off cert medJ med regret p90 regret SDPs/plant med(mean,max) synth s med fail: nocert/rollout
35
+ 4 82.5% [ 75.8, 89.2] 78.3% 86.7% 82.5% 4.109 0.67% 2.51% 1(2.02,4) 0.56 0/21
36
+ 8 90.0% [ 84.2, 95.0] 88.3% 91.7% 90.0% 4.121 0.68% 2.60% 1(2.53,8) 0.56 0/12
37
+ 16 93.3% [ 88.3, 97.5] 93.3% 93.3% 93.3% 4.157 0.71% 2.86% 1(3.13,16) 0.56 0/8
38
+ (hand-off certified == success here: P is re-solved and verified on the very plant that runs.)
39
+ reproduces yesterday's traj_gen_ksweep nearest+P per-plant successes exactly at K=4,8,16: True
40
+ mean SDPs actually needed (first-success stopping) at K=8: 2.53
41
+ diffusion generated+P at K=4/8 (mean+-sd over 3 seeded draws, pooled): K=4: 58.9+-1.0, K=8: 65.3+-0.4
42
+
43
+ --- TASK 2/3: mismatch. Controller designed for NOMINAL params, executed on PERTURBED plant, K=8 ---
44
+ retr = retrieval+P (nominal ID, screened on nominal model); certTrue = nominal P verified on the TRUE plant (all deploys);
45
+ hoTrue = success AND certTrue (certified hand-off on the true plant); NMPC = live NMPC, internal model nominal;
46
+ diff = generated+P (nominal cond.), mean+-sd over 3 seeds; trueID = retrieval+P conditioned on the TRUE params (perfect ID).
47
+
48
+ scenario retr [95% CI] certTrue hoTrue NMPC diff mean+-sd trueID retr-NMPC (b-c,p) retr-diff(s0) (b-c,p) trueID-retr (b-c,p)
49
+ nominal 90.0% [ 84.2, 95.0] 100.0% 90.0% 95.8% 65.3+-0.4 90.0% 0-7 p=0.0156 30-0 p=0.0000 0-0 p=1.0000
50
+ fric0.6 86.7% [ 80.0, 92.5] 0.8% 0.0% 95.8% 62.5+-0.0 99.2% 1-12 p=0.0034 34-5 p=0.0000 16-1 p=0.0003
51
+ fric1.6 9.2% [ 4.2, 15.0] 0.0% 0.0% 64.2% 10.6+-0.8 58.3% 0-66 p=0.0000 3-6 p=0.5078 59-0 p=0.0000
52
+ fric2.5 0.0% [ 0.0, 0.0] 0.0% 0.0% 29.2% 0.0+-0.0 25.0% 0-35 p=0.0000 0-0 p=1.0000 30-0 p=0.0000
53
+ ml_mp-_Lp- 68.3% [ 60.0, 75.9] 100.0% 68.3% 94.2% 50.0+-4.5 85.8% 0-31 p=0.0000 31-8 p=0.0003 22-1 p=0.0000
54
+ ml_mp-_Lp+ 81.7% [ 74.2, 88.3] 0.0% 0.0% 95.8% 61.7+-0.0 83.3% 1-18 p=0.0001 34-10 p=0.0004 14-12 p=0.8450
55
+ ml_mp+_Lp- 48.3% [ 39.2, 56.7] 25.8% 13.3% 95.0% 39.7+-2.7 84.2% 0-56 p=0.0000 26-12 p=0.0336 45-2 p=0.0000
56
+ ml_mp+_Lp+ 75.8% [ 67.5, 83.3] 0.0% 0.0% 95.0% 57.5+-1.2 85.0% 1-24 p=0.0000 34-10 p=0.0004 22-11 p=0.0801
57
+ both_f1.6_mp-_Lp- 9.2% [ 4.2, 15.0] 33.3% 6.7% 55.0% 11.7+-0.7 51.7% 0-55 p=0.0000 3-5 p=0.7266 51-0 p=0.0000
58
+ both_f1.6_mp-_Lp+ 19.2% [ 12.5, 25.8] 0.0% 0.0% 75.0% 17.8+-1.0 57.5% 0-67 p=0.0000 6-3 p=0.5078 49-3 p=0.0000
59
+ both_f1.6_mp+_Lp- 0.8% [ 0.0, 2.5] 0.0% 0.0% 52.5% 0.8+-0.0 52.5% 0-62 p=0.0000 1-1 p=1.0000 62-0 p=0.0000
60
+ both_f1.6_mp+_Lp+ 1.7% [ 0.0, 4.2] 0.0% 0.0% 67.5% 2.5+-0.7 55.0% 0-79 p=0.0000 2-3 p=1.0000 64-0 p=0.0000
61
+ both_f2.5_mp-_Lp- 0.0% [ 0.0, 0.0] 0.0% 0.0% 21.7% 0.3+-0.4 25.8% 0-26 p=0.0000 0-0 p=1.0000 31-0 p=0.0000
62
+ both_f2.5_mp-_Lp+ 0.0% [ 0.0, 0.0] 0.0% 0.0% 33.3% 1.1+-0.4 25.8% 0-40 p=0.0000 0-1 p=1.0000 31-0 p=0.0000
63
+ both_f2.5_mp+_Lp- 0.0% [ 0.0, 0.0] 0.0% 0.0% 18.3% 0.0+-0.0 25.0% 0-22 p=0.0000 0-0 p=1.0000 30-0 p=0.0000
64
+ both_f2.5_mp+_Lp+ 0.0% [ 0.0, 0.0] 0.0% 0.0% 33.3% 0.0+-0.0 23.3% 0-40 p=0.0000 0-0 p=1.0000 28-0 p=0.0000
65
+
66
+ --- EXPLORATORY (added after the pre-registration): retrieval+P with the certificate re-solved at margin 0.3 (retrM) ---
67
+ scenario retrM succ certTrue hoTrue vs retr (b-c,p)
68
+ nominal 88.3% 100.0% 88.3% 0-2 p=0.5000
69
+ fric0.6 88.3% 0.0% 0.0% 3-1 p=0.6250
70
+ fric1.6 9.2% 46.7% 6.7% 0-0 p=1.0000
71
+ fric2.5 0.0% 3.3% 0.0% 0-0 p=1.0000
72
+ ml_mp-_Lp- 66.7% 95.0% 66.7% 4-6 p=0.7539
73
+ ml_mp-_Lp+ 82.5% 8.3% 5.0% 7-6 p=1.0000
74
+ ml_mp+_Lp- 51.7% 80.8% 42.5% 6-2 p=0.2891
75
+ ml_mp+_Lp+ 75.8% 0.0% 0.0% 6-6 p=1.0000
76
+ both_f1.6_mp-_Lp- 10.0% 57.5% 9.2% 1-0 p=1.0000
77
+ both_f1.6_mp-_Lp+ 17.5% 10.0% 0.8% 1-3 p=0.6250
78
+ both_f1.6_mp+_Lp- 0.8% 31.7% 0.0% 0-0 p=1.0000
79
+ both_f1.6_mp+_Lp+ 1.7% 10.0% 0.0% 0-0 p=1.0000
80
+ both_f2.5_mp-_Lp- 0.0% 2.5% 0.0% 0-0 p=1.0000
81
+ both_f2.5_mp-_Lp+ 0.0% 1.7% 0.0% 0-0 p=1.0000
82
+ both_f2.5_mp+_Lp- 0.0% 3.3% 0.0% 0-0 p=1.0000
83
+ both_f2.5_mp+_Lp+ 0.0% 1.7% 0.0% 0-0 p=1.0000
84
+
85
+ --- EXPLORATORY (post-hoc): how much identification is needed? ---
86
+ partialC = design on an estimate recovering fraction C of the log-deviation, screened on that estimate, executed on the true plant;
87
+ hybrid = nominal-screened (u*,T,K) with P re-solved on the TRUE plant (no re-retrieval, no re-screening).
88
+ scenario nominal-ID partial0.5 (cert) partial0.75 (cert) trueID hybrid (cert) NMPC
89
+ fric0.6 86.7% 94.2% 0.8% 95.8% 0.0% 99.2% 87.5% 87.5% 95.8%
90
+ fric1.6 9.2% 30.8% 0.0% 55.0% 0.0% 58.3% 7.5% 7.5% 64.2%
91
+ ml_mp-_Lp- 68.3% 82.5% 82.5% 80.8% 80.8% 85.8% 69.2% 69.2% 94.2%
92
+ ml_mp-_Lp+ 81.7% 87.5% 0.0% 87.5% 0.0% 83.3% 74.2% 74.2% 95.8%
93
+ ml_mp+_Lp- 48.3% 75.8% 75.0% 82.5% 82.5% 84.2% 54.2% 54.2% 95.0%
94
+ ml_mp+_Lp+ 75.8% 79.2% 0.0% 83.3% 0.0% 85.0% 61.7% 61.7% 95.0%
95
+ (cert) = success AND nominal-to-estimate P verified on the true plant, i.e. certified hand-off on the true plant.
96
+
97
+ --- retrieval K=8 (nominal ID) failure causes and per-draw success ---
98
+ scenario d4242 d9001 no-cert rollout ref-div succ,cert-invalid-on-true fallback deploys medJ(succ) NMPC medJ
99
+ nominal 88.3% 91.7% 0 12 0 0 12 4.121 4.112
100
+ fric0.6 85.0% 88.3% 0 16 0 104 12 4.023 3.923
101
+ fric1.6 3.3% 15.0% 0 109 0 11 12 4.332 6.202
102
+ fric2.5 0.0% 0.0% 0 120 0 0 12 nan 10.561
103
+ ml_mp-_Lp- 68.3% 68.3% 0 38 0 0 12 4.163 4.151
104
+ ml_mp-_Lp+ 76.7% 86.7% 0 22 0 98 12 4.264 4.210
105
+ ml_mp+_Lp- 45.0% 51.7% 0 62 0 42 12 4.492 4.494
106
+ ml_mp+_Lp+ 70.0% 81.7% 0 29 0 91 12 4.448 4.353
107
+ both_f1.6_mp-_Lp- 3.3% 15.0% 0 109 0 3 12 4.457 6.108
108
+ both_f1.6_mp-_Lp+ 11.7% 26.7% 0 97 0 23 12 4.387 6.312
109
+ both_f1.6_mp+_Lp- 0.0% 1.7% 0 119 0 1 12 9.895 7.362
110
+ both_f1.6_mp+_Lp+ 1.7% 1.7% 0 118 0 2 12 6.091 7.177
111
+ both_f2.5_mp-_Lp- 0.0% 0.0% 0 120 0 0 12 nan 11.425
112
+ both_f2.5_mp-_Lp+ 0.0% 0.0% 0 120 0 0 12 nan 10.437
113
+ both_f2.5_mp+_Lp- 0.0% 0.0% 0 120 0 0 12 nan 13.458
114
+ both_f2.5_mp+_Lp+ 0.0% 0.0% 0 120 0 0 12 nan 13.293
115
+
116
+ NMPC solve-time tail across all mismatch runs: median of per-run p95 12 ms, median of per-run max 262 ms, max 1820 ms; solver failures 145672/768000 solves
117
+
118
+ --- TASK 4: decision table (deployable arms use nominal params; trueID = bound if identification were perfect) ---
119
+ rule D4: better = success higher by >= 5 pts with McNemar p<0.05 (retr vs NMPC); else TIE. D2/D3 flags per scenario.
120
+ scenario best arm (by success) success D4 retr vs NMPC hand-off cert. on true plant D2 survives D3 ID pays
121
+ nominal NMPC 95.8% NMPC better n/a (NMPC has no certificate) yes no
122
+ fric0.6 NMPC 95.8% NMPC better n/a (NMPC has no certificate) no yes
123
+ fric1.6 NMPC 64.2% NMPC better n/a (NMPC has no certificate) no yes
124
+ fric2.5 NMPC 29.2% NMPC better n/a (NMPC has no certificate) no yes
125
+ ml_mp-_Lp- NMPC 94.2% NMPC better n/a (NMPC has no certificate) no yes
126
+ ml_mp-_Lp+ NMPC 95.8% NMPC better n/a (NMPC has no certificate) no no
127
+ ml_mp+_Lp- NMPC 95.0% NMPC better n/a (NMPC has no certificate) no yes
128
+ ml_mp+_Lp+ NMPC 95.0% NMPC better n/a (NMPC has no certificate) no no
129
+ both_f1.6_mp-_Lp- NMPC 55.0% NMPC better n/a (NMPC has no certificate) no yes
130
+ both_f1.6_mp-_Lp+ NMPC 75.0% NMPC better n/a (NMPC has no certificate) no yes
131
+ both_f1.6_mp+_Lp- NMPC 52.5% NMPC better n/a (NMPC has no certificate) no yes
132
+ both_f1.6_mp+_Lp+ NMPC 67.5% NMPC better n/a (NMPC has no certificate) no yes
133
+ both_f2.5_mp-_Lp- NMPC 21.7% NMPC better n/a (NMPC has no certificate) no yes
134
+ both_f2.5_mp-_Lp+ NMPC 33.3% NMPC better n/a (NMPC has no certificate) no yes
135
+ both_f2.5_mp+_Lp- NMPC 18.3% NMPC better n/a (NMPC has no certificate) no yes
136
+ both_f2.5_mp+_Lp+ NMPC 33.3% NMPC better n/a (NMPC has no certificate) no yes
137
+
138
+ saved results/furuta/retrieval_deploy.txt (+.npz)
results/furuta/logs/deploy_extras_prev.log ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ extras: 2160 tasks
2
+ extras: 2160 remaining
3
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
4
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
5
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
6
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
7
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
8
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
9
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
10
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
11
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
12
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
13
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
14
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
15
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
16
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
17
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
18
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
19
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
20
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
21
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
22
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
23
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
24
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
25
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
26
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
27
+ extra 200/2160 (26s)
28
+ extra 400/2160 (43s)
29
+ extra 600/2160 (65s)
30
+ extra 800/2160 (87s)
31
+ extra 1000/2160 (112s)
32
+ extra 1200/2160 (130s)
33
+ extra 1400/2160 (157s)
34
+ extra 1600/2160 (176s)
35
+ extra 1800/2160 (193s)
results/furuta/logs/deploy_main.log ADDED
The diff for this file is too large to render. See raw diff
 
results/furuta/logs/gen_eval_v1.log ADDED
@@ -0,0 +1,471 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ bank data/furuta_nmpc_bank600.npz: 550 entries
2
+ model models/Best_furuta_traj_cond_v1.pth: cond_dim=4 out_dim=116 hidden=256 layers=5 device=cpu
3
+ [seed4242] margin=0.3: resolving P for 190 unique bank source plants
4
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
5
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
6
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
7
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
8
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
9
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
10
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
11
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
12
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
13
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
14
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
15
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
16
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
17
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
18
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
19
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
20
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
21
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
22
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
23
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
24
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
25
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
26
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
27
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
28
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
29
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
30
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
31
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
32
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
33
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
34
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
35
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
36
+ [seed4242] sampler seed 0: 12.4s for 60x4 candidates (0 decode-invalid of 240)
37
+ [seed4242] sampler seed 1: 14.9s for 60x4 candidates (0 decode-invalid of 240)
38
+ [seed4242] sampler seed 2: 13.4s for 60x4 candidates (0 decode-invalid of 240)
39
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
40
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
41
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
42
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
43
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
44
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
45
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
46
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
47
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
48
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
49
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
50
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
51
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
52
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
53
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
54
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
55
+
56
+ ******************************************************************************
57
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
58
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
59
+ For more information visit https://github.com/coin-or/Ipopt
60
+ ******************************************************************************
61
+
62
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
63
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
64
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
65
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
66
+
67
+ ******************************************************************************
68
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
69
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
70
+ For more information visit https://github.com/coin-or/Ipopt
71
+ ******************************************************************************
72
+
73
+
74
+ ******************************************************************************
75
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
76
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
77
+ For more information visit https://github.com/coin-or/Ipopt
78
+ ******************************************************************************
79
+
80
+
81
+ ******************************************************************************
82
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
83
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
84
+ For more information visit https://github.com/coin-or/Ipopt
85
+ ******************************************************************************
86
+
87
+
88
+ ******************************************************************************
89
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
90
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
91
+ For more information visit https://github.com/coin-or/Ipopt
92
+ ******************************************************************************
93
+
94
+
95
+ ******************************************************************************
96
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
97
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
98
+ For more information visit https://github.com/coin-or/Ipopt
99
+ ******************************************************************************
100
+
101
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
102
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
103
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
104
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
105
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
106
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
107
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
108
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
109
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
110
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
111
+
112
+ ******************************************************************************
113
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
114
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
115
+ For more information visit https://github.com/coin-or/Ipopt
116
+ ******************************************************************************
117
+
118
+
119
+ ******************************************************************************
120
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
121
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
122
+ For more information visit https://github.com/coin-or/Ipopt
123
+ ******************************************************************************
124
+
125
+
126
+ ******************************************************************************
127
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
128
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
129
+ For more information visit https://github.com/coin-or/Ipopt
130
+ ******************************************************************************
131
+
132
+
133
+ ******************************************************************************
134
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
135
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
136
+ For more information visit https://github.com/coin-or/Ipopt
137
+ ******************************************************************************
138
+
139
+
140
+ ******************************************************************************
141
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
142
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
143
+ For more information visit https://github.com/coin-or/Ipopt
144
+ ******************************************************************************
145
+
146
+
147
+ ******************************************************************************
148
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
149
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
150
+ For more information visit https://github.com/coin-or/Ipopt
151
+ ******************************************************************************
152
+
153
+
154
+ ******************************************************************************
155
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
156
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
157
+ For more information visit https://github.com/coin-or/Ipopt
158
+ ******************************************************************************
159
+
160
+
161
+ ******************************************************************************
162
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
163
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
164
+ For more information visit https://github.com/coin-or/Ipopt
165
+ ******************************************************************************
166
+
167
+
168
+ ******************************************************************************
169
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
170
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
171
+ For more information visit https://github.com/coin-or/Ipopt
172
+ ******************************************************************************
173
+
174
+ [seed4242] 10/60 (17s)
175
+
176
+ ******************************************************************************
177
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
178
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
179
+ For more information visit https://github.com/coin-or/Ipopt
180
+ ******************************************************************************
181
+
182
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
183
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
184
+ [seed4242] 20/60 (30s)
185
+ [seed4242] 30/60 (40s)
186
+ [seed4242] 40/60 (50s)
187
+ [seed4242] 50/60 (63s)
188
+ [seed4242] 60/60 (71s)
189
+ [seed9001] margin=0.3: resolving P for 191 unique bank source plants
190
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
191
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
192
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
193
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
194
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
195
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
196
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
197
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
198
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
199
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
200
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
201
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
202
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
203
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
204
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
205
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
206
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
207
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
208
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
209
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
210
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
211
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
212
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
213
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
214
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
215
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
216
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
217
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
218
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
219
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
220
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
221
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
222
+ [seed9001] sampler seed 0: 10.3s for 60x4 candidates (0 decode-invalid of 240)
223
+ [seed9001] sampler seed 1: 10.4s for 60x4 candidates (0 decode-invalid of 240)
224
+ [seed9001] sampler seed 2: 10.4s for 60x4 candidates (0 decode-invalid of 240)
225
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
226
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
227
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
228
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
229
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
230
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
231
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
232
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
233
+
234
+ ******************************************************************************
235
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
236
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
237
+ For more information visit https://github.com/coin-or/Ipopt
238
+ ******************************************************************************
239
+
240
+
241
+ ******************************************************************************
242
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
243
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
244
+ For more information visit https://github.com/coin-or/Ipopt
245
+ ******************************************************************************
246
+
247
+
248
+ ******************************************************************************
249
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
250
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
251
+ For more information visit https://github.com/coin-or/Ipopt
252
+ ******************************************************************************
253
+
254
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
255
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
256
+
257
+ ******************************************************************************
258
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
259
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
260
+ For more information visit https://github.com/coin-or/Ipopt
261
+ ******************************************************************************
262
+
263
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
264
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
265
+
266
+ ******************************************************************************
267
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
268
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
269
+ For more information visit https://github.com/coin-or/Ipopt
270
+ ******************************************************************************
271
+
272
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
273
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
274
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
275
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
276
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
277
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
278
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
279
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
280
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
281
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
282
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
283
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
284
+
285
+ ******************************************************************************
286
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
287
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
288
+ For more information visit https://github.com/coin-or/Ipopt
289
+ ******************************************************************************
290
+
291
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
292
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
293
+
294
+ ******************************************************************************
295
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
296
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
297
+ For more information visit https://github.com/coin-or/Ipopt
298
+ ******************************************************************************
299
+
300
+
301
+ ******************************************************************************
302
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
303
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
304
+ For more information visit https://github.com/coin-or/Ipopt
305
+ ******************************************************************************
306
+
307
+
308
+ ******************************************************************************
309
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
310
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
311
+ For more information visit https://github.com/coin-or/Ipopt
312
+ ******************************************************************************
313
+
314
+
315
+ ******************************************************************************
316
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
317
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
318
+ For more information visit https://github.com/coin-or/Ipopt
319
+ ******************************************************************************
320
+
321
+
322
+ ******************************************************************************
323
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
324
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
325
+ For more information visit https://github.com/coin-or/Ipopt
326
+ ******************************************************************************
327
+
328
+
329
+ ******************************************************************************
330
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
331
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
332
+ For more information visit https://github.com/coin-or/Ipopt
333
+ ******************************************************************************
334
+
335
+
336
+ ******************************************************************************
337
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
338
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
339
+ For more information visit https://github.com/coin-or/Ipopt
340
+ ******************************************************************************
341
+
342
+
343
+ ******************************************************************************
344
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
345
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
346
+ For more information visit https://github.com/coin-or/Ipopt
347
+ ******************************************************************************
348
+
349
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
350
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
351
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
352
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
353
+
354
+ ******************************************************************************
355
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
356
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
357
+ For more information visit https://github.com/coin-or/Ipopt
358
+ ******************************************************************************
359
+
360
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
361
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
362
+
363
+ ******************************************************************************
364
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
365
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
366
+ For more information visit https://github.com/coin-or/Ipopt
367
+ ******************************************************************************
368
+
369
+ [seed9001] 10/60 (19s)
370
+ [seed9001] 20/60 (30s)
371
+ [seed9001] 30/60 (43s)
372
+ [seed9001] 40/60 (53s)
373
+ [seed9001] 50/60 (63s)
374
+ [seed9001] 60/60 (88s)
375
+
376
+ bank data/furuta_nmpc_bank600.npz: 550 entries
377
+ model models/Best_furuta_traj_cond_v1.pth: cond_dim=4 out_dim=116 hidden=256 layers=5 device=cpu
378
+ [seed4242] margin=0.3: resolving P for 190 unique bank source plants
379
+ [seed4242] sampler seed 0: 12.4s for 60x4 candidates (0 decode-invalid of 240)
380
+ [seed4242] sampler seed 1: 14.9s for 60x4 candidates (0 decode-invalid of 240)
381
+ [seed4242] sampler seed 2: 13.4s for 60x4 candidates (0 decode-invalid of 240)
382
+ [seed4242] 10/60 (17s)
383
+ [seed4242] 20/60 (30s)
384
+ [seed4242] 30/60 (40s)
385
+ [seed4242] 40/60 (50s)
386
+ [seed4242] 50/60 (63s)
387
+ [seed4242] 60/60 (71s)
388
+ [seed9001] margin=0.3: resolving P for 191 unique bank source plants
389
+ [seed9001] sampler seed 0: 10.3s for 60x4 candidates (0 decode-invalid of 240)
390
+ [seed9001] sampler seed 1: 10.4s for 60x4 candidates (0 decode-invalid of 240)
391
+ [seed9001] sampler seed 2: 10.4s for 60x4 candidates (0 decode-invalid of 240)
392
+ [seed9001] 10/60 (19s)
393
+ [seed9001] 20/60 (30s)
394
+ [seed9001] 30/60 (43s)
395
+ [seed9001] 40/60 (53s)
396
+ [seed9001] 50/60 (63s)
397
+ [seed9001] 60/60 (88s)
398
+
399
+ === traj_gen_eval (v1): generated vs retrieval vs NMPC ===
400
+
401
+ --- plant set seed4242 (seed 4242, n=60) ---
402
+ arm succ 95% CI medJ med regret p90 synth
403
+ per-plant 93% [ 87, 98] 4.08 0.00% 0.00% 5.9s
404
+ nearest 27% [ 15, 38] 4.15 1.43% 2.99% 161ms
405
+ nearest+P 72% [ 60, 83] 4.07 1.27% 2.99% 960ms
406
+ nearest-margin 40% [ 28, 53] 4.07 1.20% 5.45% 360ms
407
+ NMPC 95% [ 88, 100] 4.08 0.14% 0.24% 32ms
408
+ generated 33.3%(sd4.9) [ 23, 47]* 3.78 0.37%(sd0.1) 1.34%(sd0.4) 225ms
409
+ generated+P 63.9%(sd2.1) [ 55, 78]* 3.99 0.67%(sd0.1) 2.78%(sd0.5) 1.5s
410
+ (* generated rows: mean (sd) over sampler seeds [0, 1, 2]; CI shown is for seed 0 only)
411
+ (certified hand-off rate == success rate for every arm except NMPC (a certificate is a precondition for success here); NMPC's own hand-off is n/a, it never switches to a static gain)
412
+
413
+ --- failure breakdown, primary seed 0 ---
414
+ per-plant NMPC never reached Omega:3 replay failed:1
415
+ nearest Lyapunov decrease fails at a vertex:40 rollout failed:4
416
+ nearest+P rollout failed:17
417
+ nearest-margin rollout failed:22 Lyapunov decrease fails at a vertex:14
418
+ NMPC :3
419
+ generated Lyapunov decrease fails at a vertex:35 rollout failed:4
420
+ generated+P rollout failed:20
421
+
422
+ --- paired McNemar on success (exact binomial on discordant pairs) ---
423
+ per-plant vs nearest beats on 40, loses on 0 p=0.0000 * cost ratio 1.014x p=0.0000
424
+ per-plant vs nearest+P beats on 14, loses on 1 p=0.0010 * cost ratio 1.013x p=0.0000
425
+ per-plant vs nearest-margin beats on 32, loses on 0 p=0.0000 * cost ratio 1.012x p=0.0000
426
+ per-plant vs NMPC beats on 0, loses on 1 p=1.0000 cost ratio 1.001x p=0.0002
427
+ per-plant vs generated beats on 35, loses on 0 p=0.0000 * cost ratio 1.004x p=0.0000
428
+ per-plant vs generated+P beats on 16, loses on 0 p=0.0000 * cost ratio 1.007x p=0.0000
429
+ nearest+P vs per-plant beats on 1, loses on 14 p=0.0010 * cost ratio 0.987x p=0.0000
430
+ nearest+P vs nearest beats on 27, loses on 0 p=0.0000 * cost ratio 1.001x p=0.0469
431
+ nearest+P vs nearest-margin beats on 19, loses on 0 p=0.0000 * cost ratio 1.000x p=0.0930
432
+ nearest+P vs NMPC beats on 0, loses on 14 p=0.0001 * cost ratio 0.989x p=0.0000
433
+ nearest+P vs generated beats on 25, loses on 3 p=0.0000 * cost ratio 0.996x p=0.0104
434
+ nearest+P vs generated+P beats on 10, loses on 7 p=0.6291 cost ratio 0.996x p=0.3392
435
+
436
+ --- plant set seed9001 (seed 9001, n=60) ---
437
+ arm succ 95% CI medJ med regret p90 synth
438
+ per-plant 93% [ 87, 98] 4.16 0.00% 0.00% 7.2s
439
+ nearest 20% [ 10, 30] 4.11 0.35% 1.42% 1ms
440
+ nearest+P 77% [ 67, 87] 4.21 0.75% 4.23% 913ms
441
+ nearest-margin 50% [ 38, 63] 4.20 0.82% 4.53% 399ms
442
+ NMPC 97% [ 92, 100] 4.17 0.07% 0.22% 29ms
443
+ generated 19.4%(sd2.1) [ 12, 32]* 3.94 0.62%(sd0.2) 2.31%(sd0.9) 174ms
444
+ generated+P 61.7%(sd1.4) [ 47, 72]* 4.09 0.99%(sd0.2) 2.80%(sd0.7) 1.5s
445
+ (* generated rows: mean (sd) over sampler seeds [0, 1, 2]; CI shown is for seed 0 only)
446
+ (certified hand-off rate == success rate for every arm except NMPC (a certificate is a precondition for success here); NMPC's own hand-off is n/a, it never switches to a static gain)
447
+
448
+ --- failure breakdown, primary seed 0 ---
449
+ per-plant replay failed:2 NMPC never reached Omega:2
450
+ nearest Lyapunov decrease fails at a vertex:41 rollout failed:7
451
+ nearest+P rollout failed:14
452
+ nearest-margin Lyapunov decrease fails at a vertex:15 rollout failed:15
453
+ NMPC :2
454
+ generated Lyapunov decrease fails at a vertex:46 rollout failed:1
455
+ generated+P rollout failed:24
456
+
457
+ --- paired McNemar on success (exact binomial on discordant pairs) ---
458
+ per-plant vs nearest beats on 45, loses on 1 p=0.0000 * cost ratio 1.004x p=0.0010
459
+ per-plant vs nearest+P beats on 11, loses on 1 p=0.0063 * cost ratio 1.007x p=0.0000
460
+ per-plant vs nearest-margin beats on 27, loses on 1 p=0.0000 * cost ratio 1.008x p=0.0000
461
+ per-plant vs NMPC beats on 0, loses on 2 p=0.5000 cost ratio 1.001x p=0.0064
462
+ per-plant vs generated beats on 43, loses on 0 p=0.0000 * cost ratio 1.005x p=0.0005
463
+ per-plant vs generated+P beats on 20, loses on 0 p=0.0000 * cost ratio 1.006x p=0.0000
464
+ nearest+P vs per-plant beats on 1, loses on 11 p=0.0063 * cost ratio 0.993x p=0.0000
465
+ nearest+P vs nearest beats on 34, loses on 0 p=0.0000 * cost ratio 1.000x p=0.1562
466
+ nearest+P vs nearest-margin beats on 16, loses on 0 p=0.0000 * cost ratio 1.000x p=0.0228
467
+ nearest+P vs NMPC beats on 0, loses on 12 p=0.0005 * cost ratio 0.994x p=0.0000
468
+ nearest+P vs generated beats on 36, loses on 3 p=0.0000 * cost ratio 0.994x p=0.1309
469
+ nearest+P vs generated+P beats on 16, loses on 6 p=0.0525 cost ratio 0.995x p=0.2449
470
+
471
+ saved -> results/furuta/traj_gen_eval_v1.txt (+.npz)
results/furuta/logs/gen_eval_v2.log ADDED
@@ -0,0 +1,471 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ bank data/furuta_nmpc_bank3000.npz: 2801 entries
2
+ model models/Best_furuta_traj_cond_v2.pth: cond_dim=4 out_dim=116 hidden=256 layers=5 device=cpu
3
+ [seed4242] margin=0.3: resolving P for 227 unique bank source plants
4
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
5
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
6
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
7
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
8
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
9
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
10
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
11
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
12
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
13
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
14
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
15
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
16
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
17
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
18
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
19
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
20
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
21
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
22
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
23
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
24
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
25
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
26
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
27
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
28
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
29
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
30
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
31
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
32
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
33
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
34
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
35
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
36
+ [seed4242] sampler seed 0: 16.3s for 60x4 candidates (0 decode-invalid of 240)
37
+ [seed4242] sampler seed 1: 12.4s for 60x4 candidates (0 decode-invalid of 240)
38
+ [seed4242] sampler seed 2: 11.8s for 60x4 candidates (0 decode-invalid of 240)
39
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
40
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
41
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
42
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
43
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
44
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
45
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
46
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
47
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
48
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
49
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
50
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
51
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
52
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
53
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
54
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
55
+
56
+ ******************************************************************************
57
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
58
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
59
+ For more information visit https://github.com/coin-or/Ipopt
60
+ ******************************************************************************
61
+
62
+
63
+ ******************************************************************************
64
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
65
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
66
+ For more information visit https://github.com/coin-or/Ipopt
67
+ ******************************************************************************
68
+
69
+
70
+ ******************************************************************************
71
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
72
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
73
+ For more information visit https://github.com/coin-or/Ipopt
74
+ ******************************************************************************
75
+
76
+
77
+ ******************************************************************************
78
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
79
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
80
+ For more information visit https://github.com/coin-or/Ipopt
81
+ ******************************************************************************
82
+
83
+
84
+ ******************************************************************************
85
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
86
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
87
+ For more information visit https://github.com/coin-or/Ipopt
88
+ ******************************************************************************
89
+
90
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
91
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
92
+
93
+ ******************************************************************************
94
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
95
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
96
+ For more information visit https://github.com/coin-or/Ipopt
97
+ ******************************************************************************
98
+
99
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
100
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
101
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
102
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
103
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
104
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
105
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
106
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
107
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
108
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
109
+
110
+ ******************************************************************************
111
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
112
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
113
+ For more information visit https://github.com/coin-or/Ipopt
114
+ ******************************************************************************
115
+
116
+
117
+ ******************************************************************************
118
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
119
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
120
+ For more information visit https://github.com/coin-or/Ipopt
121
+ ******************************************************************************
122
+
123
+
124
+ ******************************************************************************
125
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
126
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
127
+ For more information visit https://github.com/coin-or/Ipopt
128
+ ******************************************************************************
129
+
130
+
131
+ ******************************************************************************
132
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
133
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
134
+ For more information visit https://github.com/coin-or/Ipopt
135
+ ******************************************************************************
136
+
137
+
138
+ ******************************************************************************
139
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
140
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
141
+ For more information visit https://github.com/coin-or/Ipopt
142
+ ******************************************************************************
143
+
144
+
145
+ ******************************************************************************
146
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
147
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
148
+ For more information visit https://github.com/coin-or/Ipopt
149
+ ******************************************************************************
150
+
151
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
152
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
153
+
154
+ ******************************************************************************
155
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
156
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
157
+ For more information visit https://github.com/coin-or/Ipopt
158
+ ******************************************************************************
159
+
160
+
161
+ ******************************************************************************
162
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
163
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
164
+ For more information visit https://github.com/coin-or/Ipopt
165
+ ******************************************************************************
166
+
167
+
168
+ ******************************************************************************
169
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
170
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
171
+ For more information visit https://github.com/coin-or/Ipopt
172
+ ******************************************************************************
173
+
174
+ [seed4242] 10/60 (20s)
175
+
176
+ ******************************************************************************
177
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
178
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
179
+ For more information visit https://github.com/coin-or/Ipopt
180
+ ******************************************************************************
181
+
182
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
183
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
184
+ [seed4242] 20/60 (31s)
185
+ [seed4242] 30/60 (42s)
186
+ [seed4242] 40/60 (53s)
187
+ [seed4242] 50/60 (64s)
188
+ [seed4242] 60/60 (76s)
189
+ [seed9001] margin=0.3: resolving P for 232 unique bank source plants
190
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
191
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
192
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
193
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
194
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
195
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
196
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
197
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
198
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
199
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
200
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
201
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
202
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
203
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
204
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
205
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
206
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
207
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
208
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
209
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
210
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
211
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
212
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
213
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
214
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
215
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
216
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
217
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
218
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
219
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
220
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
221
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
222
+ [seed9001] sampler seed 0: 10.5s for 60x4 candidates (0 decode-invalid of 240)
223
+ [seed9001] sampler seed 1: 10.7s for 60x4 candidates (0 decode-invalid of 240)
224
+ [seed9001] sampler seed 2: 9.7s for 60x4 candidates (0 decode-invalid of 240)
225
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
226
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
227
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
228
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
229
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
230
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
231
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
232
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
233
+
234
+ ******************************************************************************
235
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
236
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
237
+ For more information visit https://github.com/coin-or/Ipopt
238
+ ******************************************************************************
239
+
240
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
241
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
242
+
243
+ ******************************************************************************
244
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
245
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
246
+ For more information visit https://github.com/coin-or/Ipopt
247
+ ******************************************************************************
248
+
249
+
250
+ ******************************************************************************
251
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
252
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
253
+ For more information visit https://github.com/coin-or/Ipopt
254
+ ******************************************************************************
255
+
256
+
257
+ ******************************************************************************
258
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
259
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
260
+ For more information visit https://github.com/coin-or/Ipopt
261
+ ******************************************************************************
262
+
263
+
264
+ ******************************************************************************
265
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
266
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
267
+ For more information visit https://github.com/coin-or/Ipopt
268
+ ******************************************************************************
269
+
270
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
271
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
272
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
273
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
274
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
275
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
276
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
277
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
278
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
279
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
280
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
281
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
282
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
283
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
284
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
285
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
286
+
287
+ ******************************************************************************
288
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
289
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
290
+ For more information visit https://github.com/coin-or/Ipopt
291
+ ******************************************************************************
292
+
293
+
294
+ ******************************************************************************
295
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
296
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
297
+ For more information visit https://github.com/coin-or/Ipopt
298
+ ******************************************************************************
299
+
300
+
301
+ ******************************************************************************
302
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
303
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
304
+ For more information visit https://github.com/coin-or/Ipopt
305
+ ******************************************************************************
306
+
307
+
308
+ ******************************************************************************
309
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
310
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
311
+ For more information visit https://github.com/coin-or/Ipopt
312
+ ******************************************************************************
313
+
314
+
315
+ ******************************************************************************
316
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
317
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
318
+ For more information visit https://github.com/coin-or/Ipopt
319
+ ******************************************************************************
320
+
321
+
322
+ ******************************************************************************
323
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
324
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
325
+ For more information visit https://github.com/coin-or/Ipopt
326
+ ******************************************************************************
327
+
328
+
329
+ ******************************************************************************
330
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
331
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
332
+ For more information visit https://github.com/coin-or/Ipopt
333
+ ******************************************************************************
334
+
335
+
336
+ ******************************************************************************
337
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
338
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
339
+ For more information visit https://github.com/coin-or/Ipopt
340
+ ******************************************************************************
341
+
342
+
343
+ ******************************************************************************
344
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
345
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
346
+ For more information visit https://github.com/coin-or/Ipopt
347
+ ******************************************************************************
348
+
349
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
350
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
351
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
352
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
353
+
354
+ ******************************************************************************
355
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
356
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
357
+ For more information visit https://github.com/coin-or/Ipopt
358
+ ******************************************************************************
359
+
360
+
361
+ ******************************************************************************
362
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
363
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
364
+ For more information visit https://github.com/coin-or/Ipopt
365
+ ******************************************************************************
366
+
367
+ [seed9001] 10/60 (15s)
368
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
369
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
370
+ [seed9001] 20/60 (22s)
371
+ [seed9001] 30/60 (36s)
372
+ [seed9001] 40/60 (43s)
373
+ [seed9001] 50/60 (53s)
374
+ [seed9001] 60/60 (72s)
375
+
376
+ bank data/furuta_nmpc_bank3000.npz: 2801 entries
377
+ model models/Best_furuta_traj_cond_v2.pth: cond_dim=4 out_dim=116 hidden=256 layers=5 device=cpu
378
+ [seed4242] margin=0.3: resolving P for 227 unique bank source plants
379
+ [seed4242] sampler seed 0: 16.3s for 60x4 candidates (0 decode-invalid of 240)
380
+ [seed4242] sampler seed 1: 12.4s for 60x4 candidates (0 decode-invalid of 240)
381
+ [seed4242] sampler seed 2: 11.8s for 60x4 candidates (0 decode-invalid of 240)
382
+ [seed4242] 10/60 (20s)
383
+ [seed4242] 20/60 (31s)
384
+ [seed4242] 30/60 (42s)
385
+ [seed4242] 40/60 (53s)
386
+ [seed4242] 50/60 (64s)
387
+ [seed4242] 60/60 (76s)
388
+ [seed9001] margin=0.3: resolving P for 232 unique bank source plants
389
+ [seed9001] sampler seed 0: 10.5s for 60x4 candidates (0 decode-invalid of 240)
390
+ [seed9001] sampler seed 1: 10.7s for 60x4 candidates (0 decode-invalid of 240)
391
+ [seed9001] sampler seed 2: 9.7s for 60x4 candidates (0 decode-invalid of 240)
392
+ [seed9001] 10/60 (15s)
393
+ [seed9001] 20/60 (22s)
394
+ [seed9001] 30/60 (36s)
395
+ [seed9001] 40/60 (43s)
396
+ [seed9001] 50/60 (53s)
397
+ [seed9001] 60/60 (72s)
398
+
399
+ === traj_gen_eval (v2): generated vs retrieval vs NMPC ===
400
+
401
+ --- plant set seed4242 (seed 4242, n=60) ---
402
+ arm succ 95% CI medJ med regret p90 synth
403
+ per-plant 93% [ 87, 98] 4.08 0.00% 0.00% 5.1s
404
+ nearest 62% [ 50, 73] 4.15 0.36% 1.51% 388ms
405
+ nearest+P 78% [ 67, 88] 4.09 0.69% 2.52% 760ms
406
+ nearest-margin 62% [ 50, 73] 4.15 0.36% 1.51% 387ms
407
+ NMPC 95% [ 88, 100] 4.08 0.14% 0.24% 29ms
408
+ generated 58.9%(sd2.1) [ 45, 70]* 3.94 0.72%(sd0.0) 3.36%(sd0.6) 561ms
409
+ generated+P 63.3%(sd1.4) [ 52, 75]* 3.98 0.79%(sd0.1) 2.69%(sd0.2) 1.5s
410
+ (* generated rows: mean (sd) over sampler seeds [0, 1, 2]; CI shown is for seed 0 only)
411
+ (certified hand-off rate == success rate for every arm except NMPC (a certificate is a precondition for success here); NMPC's own hand-off is n/a, it never switches to a static gain)
412
+
413
+ --- failure breakdown, primary seed 0 ---
414
+ per-plant NMPC never reached Omega:3 replay failed:1
415
+ nearest Lyapunov decrease fails at a vertex:13 rollout failed:9 true dynamics: 0.02% of the boundary has V-dot >= 0, 0.00% saturates:1
416
+ nearest+P rollout failed:13
417
+ nearest-margin Lyapunov decrease fails at a vertex:13 rollout failed:9 true dynamics: 0.02% of the boundary has V-dot >= 0, 0.00% saturates:1
418
+ NMPC :3
419
+ generated rollout failed:14 Lyapunov decrease fails at a vertex:10 true dynamics: 0.01% of the boundary has V-dot >= 0, 0.00% saturates:1
420
+ generated+P rollout failed:22
421
+
422
+ --- paired McNemar on success (exact binomial on discordant pairs) ---
423
+ per-plant vs nearest beats on 20, loses on 1 p=0.0000 * cost ratio 1.004x p=0.0000
424
+ per-plant vs nearest+P beats on 10, loses on 1 p=0.0117 * cost ratio 1.007x p=0.0000
425
+ per-plant vs nearest-margin beats on 20, loses on 1 p=0.0000 * cost ratio 1.004x p=0.0000
426
+ per-plant vs NMPC beats on 0, loses on 1 p=1.0000 cost ratio 1.001x p=0.0002
427
+ per-plant vs generated beats on 21, loses on 0 p=0.0000 * cost ratio 1.007x p=0.0000
428
+ per-plant vs generated+P beats on 18, loses on 0 p=0.0000 * cost ratio 1.007x p=0.0000
429
+ nearest+P vs per-plant beats on 1, loses on 10 p=0.0117 * cost ratio 0.993x p=0.0000
430
+ nearest+P vs nearest beats on 10, loses on 0 p=0.0020 * cost ratio 1.000x p=0.8291
431
+ nearest+P vs nearest-margin beats on 10, loses on 0 p=0.0020 * cost ratio 1.000x p=0.8291
432
+ nearest+P vs NMPC beats on 0, loses on 10 p=0.0020 * cost ratio 0.995x p=0.0000
433
+ nearest+P vs generated beats on 15, loses on 3 p=0.0075 * cost ratio 1.002x p=0.6511
434
+ nearest+P vs generated+P beats on 12, loses on 3 p=0.0352 * cost ratio 1.001x p=0.6680
435
+
436
+ --- plant set seed9001 (seed 9001, n=60) ---
437
+ arm succ 95% CI medJ med regret p90 synth
438
+ per-plant 93% [ 87, 98] 4.16 0.00% 0.00% 6.0s
439
+ nearest 65% [ 52, 77] 4.10 0.71% 2.11% 168ms
440
+ nearest+P 87% [ 78, 95] 4.14 0.59% 2.34% 374ms
441
+ nearest-margin 65% [ 52, 77] 4.10 0.71% 2.11% 177ms
442
+ NMPC 97% [ 92, 100] 4.17 0.07% 0.22% 23ms
443
+ generated 45.6%(sd2.1) [ 37, 62]* 4.05 0.84%(sd0.3) 2.40%(sd0.1) 466ms
444
+ generated+P 60.0%(sd0.0) [ 48, 72]* 4.08 0.86%(sd0.3) 2.25%(sd0.2) 1.3s
445
+ (* generated rows: mean (sd) over sampler seeds [0, 1, 2]; CI shown is for seed 0 only)
446
+ (certified hand-off rate == success rate for every arm except NMPC (a certificate is a precondition for success here); NMPC's own hand-off is n/a, it never switches to a static gain)
447
+
448
+ --- failure breakdown, primary seed 0 ---
449
+ per-plant replay failed:2 NMPC never reached Omega:2
450
+ nearest Lyapunov decrease fails at a vertex:16 rollout failed:5
451
+ nearest+P rollout failed:8
452
+ nearest-margin Lyapunov decrease fails at a vertex:16 rollout failed:5
453
+ NMPC :2
454
+ generated rollout failed:20 Lyapunov decrease fails at a vertex:11
455
+ generated+P rollout failed:24
456
+
457
+ --- paired McNemar on success (exact binomial on discordant pairs) ---
458
+ per-plant vs nearest beats on 18, loses on 1 p=0.0001 * cost ratio 1.007x p=0.0000
459
+ per-plant vs nearest+P beats on 5, loses on 1 p=0.2188 cost ratio 1.006x p=0.0000
460
+ per-plant vs nearest-margin beats on 18, loses on 1 p=0.0001 * cost ratio 1.007x p=0.0000
461
+ per-plant vs NMPC beats on 0, loses on 2 p=0.5000 cost ratio 1.001x p=0.0064
462
+ per-plant vs generated beats on 27, loses on 0 p=0.0000 * cost ratio 1.007x p=0.0000
463
+ per-plant vs generated+P beats on 20, loses on 0 p=0.0000 * cost ratio 1.011x p=0.0000
464
+ nearest+P vs per-plant beats on 1, loses on 5 p=0.2188 cost ratio 0.994x p=0.0000
465
+ nearest+P vs nearest beats on 13, loses on 0 p=0.0002 * cost ratio 1.000x p=0.1154
466
+ nearest+P vs nearest-margin beats on 13, loses on 0 p=0.0002 * cost ratio 1.000x p=0.1154
467
+ nearest+P vs NMPC beats on 0, loses on 6 p=0.0312 * cost ratio 0.995x p=0.0000
468
+ nearest+P vs generated beats on 25, loses on 2 p=0.0000 * cost ratio 1.005x p=0.0299
469
+ nearest+P vs generated+P beats on 18, loses on 2 p=0.0004 * cost ratio 1.006x p=0.0169
470
+
471
+ saved -> results/furuta/traj_gen_eval_v2.txt (+.npz)
results/furuta/logs/ksweep.log ADDED
@@ -0,0 +1,571 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ bank data/furuta_nmpc_bank3000.npz: 2801 entries
2
+ model models/Best_furuta_traj_cond_v2.pth: cond_dim=4 out_dim=116
3
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
4
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
5
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
6
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
7
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
8
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
9
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
10
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
11
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
12
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
13
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
14
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
15
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
16
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
17
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
18
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
19
+
20
+ ******************************************************************************
21
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
22
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
23
+ For more information visit https://github.com/coin-or/Ipopt
24
+ ******************************************************************************
25
+
26
+
27
+ ******************************************************************************
28
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
29
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
30
+ For more information visit https://github.com/coin-or/Ipopt
31
+ ******************************************************************************
32
+
33
+
34
+ ******************************************************************************
35
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
36
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
37
+ For more information visit https://github.com/coin-or/Ipopt
38
+ ******************************************************************************
39
+
40
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
41
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
42
+
43
+ ******************************************************************************
44
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
45
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
46
+ For more information visit https://github.com/coin-or/Ipopt
47
+ ******************************************************************************
48
+
49
+
50
+ ******************************************************************************
51
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
52
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
53
+ For more information visit https://github.com/coin-or/Ipopt
54
+ ******************************************************************************
55
+
56
+
57
+ ******************************************************************************
58
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
59
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
60
+ For more information visit https://github.com/coin-or/Ipopt
61
+ ******************************************************************************
62
+
63
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
64
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
65
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
66
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
67
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
68
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
69
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
70
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
71
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
72
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
73
+
74
+ ******************************************************************************
75
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
76
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
77
+ For more information visit https://github.com/coin-or/Ipopt
78
+ ******************************************************************************
79
+
80
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
81
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
82
+
83
+ ******************************************************************************
84
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
85
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
86
+ For more information visit https://github.com/coin-or/Ipopt
87
+ ******************************************************************************
88
+
89
+
90
+ ******************************************************************************
91
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
92
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
93
+ For more information visit https://github.com/coin-or/Ipopt
94
+ ******************************************************************************
95
+
96
+
97
+ ******************************************************************************
98
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
99
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
100
+ For more information visit https://github.com/coin-or/Ipopt
101
+ ******************************************************************************
102
+
103
+
104
+ ******************************************************************************
105
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
106
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
107
+ For more information visit https://github.com/coin-or/Ipopt
108
+ ******************************************************************************
109
+
110
+
111
+ ******************************************************************************
112
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
113
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
114
+ For more information visit https://github.com/coin-or/Ipopt
115
+ ******************************************************************************
116
+
117
+
118
+ ******************************************************************************
119
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
120
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
121
+ For more information visit https://github.com/coin-or/Ipopt
122
+ ******************************************************************************
123
+
124
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
125
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
126
+
127
+ ******************************************************************************
128
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
129
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
130
+ For more information visit https://github.com/coin-or/Ipopt
131
+ ******************************************************************************
132
+
133
+
134
+ ******************************************************************************
135
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
136
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
137
+ For more information visit https://github.com/coin-or/Ipopt
138
+ ******************************************************************************
139
+
140
+
141
+ ******************************************************************************
142
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
143
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
144
+ For more information visit https://github.com/coin-or/Ipopt
145
+ ******************************************************************************
146
+
147
+ [seed4242] per-plant ceiling: 93%
148
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
149
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
150
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
151
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
152
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
153
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
154
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
155
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
156
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
157
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
158
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
159
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
160
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
161
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
162
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
163
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
164
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
165
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
166
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
167
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
168
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
169
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
170
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
171
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
172
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
173
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
174
+ [seed4242] nearest+P sequences done
175
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
176
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
177
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
178
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
179
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
180
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
181
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
182
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
183
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
184
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
185
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
186
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
187
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
188
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
189
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
190
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
191
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
192
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
193
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
194
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
195
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
196
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
197
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
198
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
199
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
200
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
201
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
202
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
203
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
204
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
205
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
206
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
207
+ [seed4242] generated+P seed 0 sequences done (41.1s sampling for 60x16)
208
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
209
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
210
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
211
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
212
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
213
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
214
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
215
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
216
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
217
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
218
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
219
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
220
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
221
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
222
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
223
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
224
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
225
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
226
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
227
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
228
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
229
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
230
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
231
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
232
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
233
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
234
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
235
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
236
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
237
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
238
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
239
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
240
+ [seed4242] generated+P seed 1 sequences done (39.2s sampling for 60x16)
241
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
242
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
243
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
244
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
245
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
246
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
247
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
248
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
249
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
250
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
251
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
252
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
253
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
254
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
255
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
256
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
257
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
258
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
259
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
260
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
261
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
262
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
263
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
264
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
265
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
266
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
267
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
268
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
269
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
270
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
271
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
272
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
273
+ [seed4242] generated+P seed 2 sequences done (39.4s sampling for 60x16)
274
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
275
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
276
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
277
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
278
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
279
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
280
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
281
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
282
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
283
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
284
+
285
+ ******************************************************************************
286
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
287
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
288
+ For more information visit https://github.com/coin-or/Ipopt
289
+ ******************************************************************************
290
+
291
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
292
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
293
+
294
+ ******************************************************************************
295
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
296
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
297
+ For more information visit https://github.com/coin-or/Ipopt
298
+ ******************************************************************************
299
+
300
+
301
+ ******************************************************************************
302
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
303
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
304
+ For more information visit https://github.com/coin-or/Ipopt
305
+ ******************************************************************************
306
+
307
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
308
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
309
+
310
+ ******************************************************************************
311
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
312
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
313
+ For more information visit https://github.com/coin-or/Ipopt
314
+ ******************************************************************************
315
+
316
+
317
+ ******************************************************************************
318
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
319
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
320
+ For more information visit https://github.com/coin-or/Ipopt
321
+ ******************************************************************************
322
+
323
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
324
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
325
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
326
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
327
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
328
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
329
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
330
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
331
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
332
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
333
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
334
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
335
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
336
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
337
+
338
+ ******************************************************************************
339
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
340
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
341
+ For more information visit https://github.com/coin-or/Ipopt
342
+ ******************************************************************************
343
+
344
+
345
+ ******************************************************************************
346
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
347
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
348
+ For more information visit https://github.com/coin-or/Ipopt
349
+ ******************************************************************************
350
+
351
+
352
+ ******************************************************************************
353
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
354
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
355
+ For more information visit https://github.com/coin-or/Ipopt
356
+ ******************************************************************************
357
+
358
+
359
+ ******************************************************************************
360
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
361
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
362
+ For more information visit https://github.com/coin-or/Ipopt
363
+ ******************************************************************************
364
+
365
+
366
+ ******************************************************************************
367
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
368
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
369
+ For more information visit https://github.com/coin-or/Ipopt
370
+ ******************************************************************************
371
+
372
+
373
+ ******************************************************************************
374
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
375
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
376
+ For more information visit https://github.com/coin-or/Ipopt
377
+ ******************************************************************************
378
+
379
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
380
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
381
+
382
+ ******************************************************************************
383
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
384
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
385
+ For more information visit https://github.com/coin-or/Ipopt
386
+ ******************************************************************************
387
+
388
+
389
+ ******************************************************************************
390
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
391
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
392
+ For more information visit https://github.com/coin-or/Ipopt
393
+ ******************************************************************************
394
+
395
+
396
+ ******************************************************************************
397
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
398
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
399
+ For more information visit https://github.com/coin-or/Ipopt
400
+ ******************************************************************************
401
+
402
+
403
+ ******************************************************************************
404
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
405
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
406
+ For more information visit https://github.com/coin-or/Ipopt
407
+ ******************************************************************************
408
+
409
+
410
+ ******************************************************************************
411
+ This program contains Ipopt, a library for large-scale nonlinear optimization.
412
+ Ipopt is released as open source code under the Eclipse Public License (EPL).
413
+ For more information visit https://github.com/coin-or/Ipopt
414
+ ******************************************************************************
415
+
416
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
417
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
418
+ [seed9001] per-plant ceiling: 93%
419
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
420
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
421
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
422
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
423
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
424
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
425
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
426
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
427
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
428
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
429
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
430
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
431
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
432
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
433
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
434
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
435
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
436
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
437
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
438
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
439
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
440
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
441
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
442
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
443
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
444
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
445
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
446
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
447
+ [seed9001] nearest+P sequences done
448
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
449
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
450
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
451
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
452
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
453
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
454
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
455
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
456
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
457
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
458
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
459
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
460
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
461
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
462
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
463
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
464
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
465
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
466
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
467
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
468
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
469
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
470
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
471
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
472
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
473
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
474
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
475
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
476
+ [seed9001] generated+P seed 0 sequences done (38.0s sampling for 60x16)
477
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
478
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
479
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
480
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
481
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
482
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
483
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
484
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
485
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
486
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
487
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
488
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
489
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
490
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
491
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
492
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
493
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
494
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
495
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
496
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
497
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
498
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
499
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
500
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
501
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
502
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
503
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
504
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
505
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
506
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
507
+ [seed9001] generated+P seed 1 sequences done (40.2s sampling for 60x16)
508
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
509
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
510
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
511
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
512
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
513
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
514
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
515
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
516
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
517
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
518
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
519
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
520
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
521
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
522
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
523
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
524
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
525
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
526
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
527
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
528
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
529
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
530
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
531
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
532
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
533
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
534
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
535
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
536
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
537
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
538
+ [seed9001] generated+P seed 2 sequences done (40.3s sampling for 60x16)
539
+
540
+ bank data/furuta_nmpc_bank3000.npz: 2801 entries
541
+ model models/Best_furuta_traj_cond_v2.pth: cond_dim=4 out_dim=116
542
+ [seed4242] per-plant ceiling: 93%
543
+ [seed4242] nearest+P sequences done
544
+ [seed4242] generated+P seed 0 sequences done (41.1s sampling for 60x16)
545
+ [seed4242] generated+P seed 1 sequences done (39.2s sampling for 60x16)
546
+ [seed4242] generated+P seed 2 sequences done (39.4s sampling for 60x16)
547
+ [seed9001] per-plant ceiling: 93%
548
+ [seed9001] nearest+P sequences done
549
+ [seed9001] generated+P seed 0 sequences done (38.0s sampling for 60x16)
550
+ [seed9001] generated+P seed 1 sequences done (40.2s sampling for 60x16)
551
+ [seed9001] generated+P seed 2 sequences done (40.3s sampling for 60x16)
552
+
553
+ === candidate-budget sweep: generated+P vs nearest+P vs per-plant ceiling ===
554
+
555
+ --- plant set seed4242 (seed 4242, n=60) ---
556
+ K nearest+P generated+P per-plant
557
+ 1 51.7% t= 0.38s 35.0%(sd6.2) t= 1.02s 93.3% [McNemar nearest+P vs generated+P(seed0): 17-6, p=0.035]
558
+ 4 78.3% t= 0.78s 60.6%(sd2.1) t= 1.85s 93.3% [McNemar nearest+P vs generated+P(seed0): 14-3, p=0.013]
559
+ 8 88.3% t= 0.78s 66.1%(sd1.6) t= 1.91s 93.3% [McNemar nearest+P vs generated+P(seed0): 14-0, p=0.000]
560
+ 16 93.3% t= 0.78s 68.3%(sd1.4) t= 1.91s 93.3% [McNemar nearest+P vs generated+P(seed0): 16-0, p=0.000]
561
+ crossover K (generated+P mean >= nearest+P): never within tested range
562
+
563
+ --- plant set seed9001 (seed 9001, n=60) ---
564
+ K nearest+P generated+P per-plant
565
+ 1 56.7% t= 0.34s 32.8%(sd2.8) t= 1.02s 93.3% [McNemar nearest+P vs generated+P(seed0): 21-5, p=0.002]
566
+ 4 86.7% t= 0.55s 57.2%(sd3.1) t= 1.93s 93.3% [McNemar nearest+P vs generated+P(seed0): 21-2, p=0.000]
567
+ 8 91.7% t= 0.55s 64.4%(sd2.1) t= 2.03s 93.3% [McNemar nearest+P vs generated+P(seed0): 16-0, p=0.000]
568
+ 16 93.3% t= 0.55s 67.8%(sd0.8) t= 2.03s 93.3% [McNemar nearest+P vs generated+P(seed0): 15-0, p=0.000]
569
+ crossover K (generated+P mean >= nearest+P): never within tested range
570
+
571
+ saved -> results/furuta/traj_gen_ksweep.txt (+.npz)
results/furuta/logs/margin_check.log ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ bank data/furuta_nmpc_bank600.npz: 550 entries; 30 fresh held-out plants (seed 4242); K=4; margins [0.0, 0.1, 0.3]
2
+ unique bank source plants touched as a neighbour: 110
3
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
4
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
5
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
6
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
7
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
8
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
9
+ resolved 40/330 (1s)
10
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
11
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
12
+ resolved 80/330 (2s)
13
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
14
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
15
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
16
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
17
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
18
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
19
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
20
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
21
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
22
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
23
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
24
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
25
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
26
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
27
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
28
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
29
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
30
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
31
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
32
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
33
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
34
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
35
+ /home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/furuta_roa.py:189: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
36
+ cp.Problem(cp.Maximize(cp.log_det(S)), cons).solve(solver=solver, verbose=False)
37
+ resolved 120/330 (13s)
38
+ resolved 160/330 (38s)
39
+ resolved 200/330 (58s)
40
+ resolved 240/330 (85s)
41
+ resolved 280/330 (114s)
42
+ resolved 320/330 (143s)
43
+ resolved 330/330 (147s)
44
+
45
+ === margin quick-check: transferred-P certificate rejections (N=30, seed 4242) ===
46
+ bank data/furuta_nmpc_bank600.npz (550 entries); K=4; P re-solved fresh at every margin level (0.0 included) via certified_catch_set(margin=.)
47
+ reference: traj_gate_nmpc.txt reports 40 certificate-rejections / 4 rollout-failures of 60 plants for the 'nearest' arm (bank-stored P, no margin)
48
+
49
+ margin=0.0 transfer success 33.3% (Lyapunov decrease fails at a vertex:18, rollout failed:2) | self-consistency (same u*,T,K, own plant): 100.0% of 110
50
+ margin=0.1 transfer success 46.7% (Lyapunov decrease fails at a vertex:13, rollout failed:3) | self-consistency (same u*,T,K, own plant): 100.0% of 110
51
+ margin=0.3 transfer success 66.7% (Lyapunov decrease fails at a vertex:6, rollout failed:4) | self-consistency (same u*,T,K, own plant): 100.0% of 110
52
+
53
+ saved -> results/furuta/margin_transfer.txt