matCercola18 commited on
Commit
059b34b
·
verified ·
1 Parent(s): e664825

X20 unstable SISO pilot: raw results (HPC)

Browse files
Files changed (49) hide show
  1. results/icml_r3/X20/PREREG.md +158 -0
  2. results/icml_r3/X20/logs/pilot.log +465 -0
  3. results/icml_r3/X20/logs/smoke.log +51 -0
  4. results/icml_r3/X20/pilot/cfg.json +10 -0
  5. results/icml_r3/X20/pilot/plant_00.npz +3 -0
  6. results/icml_r3/X20/pilot/plant_01.npz +3 -0
  7. results/icml_r3/X20/pilot/plant_02.npz +3 -0
  8. results/icml_r3/X20/pilot/plant_03.npz +3 -0
  9. results/icml_r3/X20/pilot/plant_04.npz +3 -0
  10. results/icml_r3/X20/pilot/plant_05.npz +3 -0
  11. results/icml_r3/X20/pilot/plant_06.npz +3 -0
  12. results/icml_r3/X20/pilot/plant_07.npz +3 -0
  13. results/icml_r3/X20/pilot/plant_08.npz +3 -0
  14. results/icml_r3/X20/pilot/plant_09.npz +3 -0
  15. results/icml_r3/X20/pilot/plant_10.npz +3 -0
  16. results/icml_r3/X20/pilot/plant_11.npz +3 -0
  17. results/icml_r3/X20/pilot/plant_12.npz +3 -0
  18. results/icml_r3/X20/pilot/plant_13.npz +3 -0
  19. results/icml_r3/X20/pilot/plant_14.npz +3 -0
  20. results/icml_r3/X20/pilot/plant_15.npz +3 -0
  21. results/icml_r3/X20/pilot/plant_16.npz +3 -0
  22. results/icml_r3/X20/pilot/plant_17.npz +3 -0
  23. results/icml_r3/X20/pilot/plant_18.npz +3 -0
  24. results/icml_r3/X20/pilot/plant_19.npz +3 -0
  25. results/icml_r3/X20/pilot/plant_20.npz +3 -0
  26. results/icml_r3/X20/pilot/plant_21.npz +3 -0
  27. results/icml_r3/X20/pilot/plant_22.npz +3 -0
  28. results/icml_r3/X20/pilot/plant_23.npz +3 -0
  29. results/icml_r3/X20/pilot/plant_24.npz +3 -0
  30. results/icml_r3/X20/pilot/plant_25.npz +3 -0
  31. results/icml_r3/X20/pilot/plant_26.npz +3 -0
  32. results/icml_r3/X20/pilot/plant_27.npz +3 -0
  33. results/icml_r3/X20/pilot/plant_28.npz +3 -0
  34. results/icml_r3/X20/pilot/plant_29.npz +3 -0
  35. results/icml_r3/X20/pilot/plant_30.npz +3 -0
  36. results/icml_r3/X20/pilot/plant_31.npz +3 -0
  37. results/icml_r3/X20/pilot/plant_32.npz +3 -0
  38. results/icml_r3/X20/pilot/plant_33.npz +3 -0
  39. results/icml_r3/X20/pilot/plant_34.npz +3 -0
  40. results/icml_r3/X20/pilot/plant_35.npz +3 -0
  41. results/icml_r3/X20/pilot/plant_36.npz +3 -0
  42. results/icml_r3/X20/pilot/plant_37.npz +3 -0
  43. results/icml_r3/X20/pilot/plant_38.npz +3 -0
  44. results/icml_r3/X20/pilot/plant_39.npz +3 -0
  45. results/icml_r3/X20/smoke_lr0.1/cfg.json +10 -0
  46. results/icml_r3/X20/smoke_lr0.1/plant_00.npz +3 -0
  47. results/icml_r3/X20/smoke_lr0.1/plant_01.npz +3 -0
  48. results/icml_r3/X20/smoke_lr0.1/plant_02.npz +3 -0
  49. results/icml_r3/X20/smoke_lr0.1/plant_03.npz +3 -0
results/icml_r3/X20/PREREG.md ADDED
@@ -0,0 +1,158 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # X20 pre-registration: open-loop-unstable SISO pilot (gate for an unstable-plant dataset)
2
+
3
+ Written 2026-10-14, before any X20 code exists or any X20 plant is drawn.
4
+
5
+ ## 0. Question
6
+
7
+ X19 (validation, 40 plants) found that on the shipped SISO family, which has open-loop-stable, minimum-phase plants with relative degree 1, phi = 0 is certified. The best certified static gain also comes within +0.32% of an independent per-plant optimum. So neither stability nor the nonlinear dynamic controller class matters there.
8
+
9
+ This pilot asks whether a SISO family exists in which (G1) certified stabilization is a real problem and (G2) the nonlinear dynamic class (filter F + nonlinearity phi) is clearly better than tuned linear controllers. The pilot has no learned arm. It decides whether an unstable-plant dataset is built and a generator retrained (GO), or SISO stays as the stable-plant certificate result (NO-GO).
10
+
11
+ ## 1. Plants
12
+
13
+ - **Draws.** N = 40, plant seed 7171. Plants 0-19 form stratum A (minimum phase); plants 20-39 form stratum B (one real right-half-plane zero). A 4-plant smoke set (seed 7272) is used only for the crash check in section 6 and is never scored.
14
+ - **Order and relative degree.** Order 4, strictly proper. Relative degree r is 2 or 3, each with probability 1/2, so there are 4 - r zeros.
15
+ - **Unstable poles.** With probability 1/2 there is one real unstable pole p ~ U[0.2, 1.0]. Otherwise there is one complex pair with real part ~ U[0.1, 0.5] and imaginary part ~ U[0.5, 2.0], conjugate-paired. Note that `generate_random_sys(stable=False)` is NOT used: it draws unpaired complex poles.
16
+ - **Remaining poles.** All real, each = -(U[0, 1] + 0.1).
17
+ - **Zeros, stratum A.** All real, each = -U[0.1, 2.0].
18
+ - **Zeros, stratum B.** One zero is z = c * rho, with c ~ U[2, 5] and rho the largest real part among the unstable poles. The other zeros are drawn as in stratum A. Because z > rho, the parity interlacing property holds, so a stable stabilizing controller exists. This is required because F is constrained stable.
19
+ - **Gain and encoding.** The gain is normalized to |G(0)| = 1, keeping its sign. Coefficients use the `tf_to_coeffs_vector` convention (num 5, den 5); the loop is realized with `k_sweep.realize_pad`, with state [x_G; x_F].
20
+
21
+ ## 2. Controller class and actuator
22
+
23
+ - **Controller.** The class is unchanged from the SISO family: u = -phi(clip(y_F, -5, 5)) with y_F = F y_G. F is stable and strictly proper, of order <= 4. phi is piecewise linear on the 11 knots {-5, ..., 5} with phi(0) = 0.
24
+ - **Actuator bound (new).** |phi(x_k)| <= u_max = 1 at every knot, so |u| <= 1 always.
25
+ - **Why this makes stability hard.** With an exponentially unstable plant and a bounded input, no controller is globally stabilizing. Every certificate below is therefore regional.
26
+
27
+ ## 3. Certificate (the verifier, for every arm)
28
+
29
+ The certificate is a regional circle criterion with a local sector:
30
+ - **Loop and nonlinearity.** The loop xdot = A x + B u is closed by u = -psi(C x).
31
+ - Wiener controllers: psi(y) = phi(clip(y, -5, 5)).
32
+ - LQG arm: psi = sat_1.
33
+ - **Local sector.** For each level ybar in the grid {0.25, 0.5, 0.75, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 6, 8, 10, 15, 20}:
34
+ - k1(ybar) = inf psi(y)/y and k2(ybar) = sup psi(y)/y over 0 < |y| <= ybar.
35
+ - Both are exact for piecewise-linear psi: evaluate at the knots inside the interval and at +-ybar.
36
+ - **SDP per ybar.** Variables P = P' (n x n), tau >= 0, t. Minimize t subject to:
37
+ - [[A'P + PA - tau k1 k2 C'C, PB - tau (k1 + k2)/2 C'], [., -tau]] <= -eps I (S-procedure for the sector);
38
+ - P >= eps I, with eps = 1e-6;
39
+ - [[ybar^2, C], [C', P]] >= 0 (the ellipsoid {x'Px <= 1} lies inside |Cx| <= ybar, so the sector holds on it);
40
+ - d_i' P d_i <= t for the 4 initial-condition directions (section 4).
41
+ - **Certified radius and verdict.** r*(ybar) = 1/sqrt(t*), or 0 if infeasible; r* = max over ybar. The controller is "certified at scale s" iff r* >= s. Then every initial condition s * d_i lies in an invariant ellipsoid on which V = x'Px decreases.
42
+ - **Solver.** CLARABEL, with SCS as fallback. A solve counts only if a numpy re-check passes at tolerance 1e-9 (1 + norm): the max eigenvalue of the first LMI is <= 0, and P and the containment block are PSD. A feasibility-only variant (stop at the first ybar with r*(ybar) >= s) may be used where only "certified at s" is needed.
43
+ - **Conservatism.** The circle criterion with a local sector is more conservative than generalized-sector conditions for saturation (Tarbouriech et al. 2011). The same certificate is applied to every arm, so the comparison is fair, but absolute radii are lower bounds.
44
+ - **Self-check (must pass before any pilot run).**
45
+ - Scalar loop xdot = x + u with u = -sat_1(3x): the exact region of attraction is |x| < 1, so r* along d = 1 must lie in [0.85, 1.0], and s = 1.05 must NOT be certified.
46
+ - A Hurwitz linear loop with psi = identity must have r* > 0.
47
+ - Lives in `nonlinear/testing/test_regional_cert.py`.
48
+
49
+ ## 4. Initial conditions, scales, cost
50
+
51
+ - **Directions.** Per plant, 4 directions d_i = z_i/|z_i|, with z_i ~ N(0, I_4) seeded by (7373, i) on the plant block; the controller starts at rest.
52
+ - **Yardstick.** sigma_p = r* of the nominal saturated LQG (arm c at R = 1). A plant with sigma_p = 0 is dropped and counted.
53
+ - **Scales.**
54
+ - s1 = 0.5 sigma_p: the cost comparison, where reasonable arms should certify.
55
+ - s2 = 1.5 sigma_p: beyond the nominal LQG's certified region.
56
+ - **Cost.** J(s) = mean over the 4 initial conditions s * d_i of int_0^60 q(y_G) + u^2 dt. It is the X19 cost: q = closed_loop.q_state, RK4 with dt 0.03, the `siso_x19.simulate` machinery. Also report the horizon check |x(60 s)|.
57
+
58
+ ## 5. Arms
59
+
60
+ a. **phi = 0.** Certified rate, expected 0%; a sanity check.
61
+
62
+ b. **Saturated static output gain.**
63
+ - Controller: F = +-1, phi_k = clip(k x_k, -1, 1), with k in logspace(-2, 2, 81) and both signs.
64
+ - Linear stabilizability: does some k in +-logspace(-3, 3, 2001), unsaturated, make A_G - k B C_G Hurwitz?
65
+ - r*_SG = max over k.
66
+ - J_SG(s) = min cost over the k certified at s, or inf if none is.
67
+
68
+ c. **Saturated LQG.**
69
+ - Controller: LQR gain K with Q = 2 C'C and R in logspace(-2, 2, 9), plus the X19 `lqg()` observer. u = -sat_1(K xhat), and the observer is driven by the applied input.
70
+ - Lur'e form: A_o = [[A, 0], [L C, A - L C]], B_o = [B; B], C_o = [0, K], psi = sat_1.
71
+ - This is a different architecture from the Wiener class: a full-order observer that sees u, possibly an unstable controller, with relative degree 1. It is reported as the strong linear baseline.
72
+ - R = 1 gives sigma_p. Tuned values: r*_LQG = max over R, and J_LQG(s) = min over the R certified at s.
73
+
74
+ d. **Reference over the Wiener class with |phi| <= 1.** One run per scale s in {s1, s2}, 16 starts each.
75
+ - **Start m.** Draw F from the dataset prior (the F part of `siso_x19.prior_start`: stable, strictly proper, order 4, relative degree 2), with a sign of +-1 at probability 1/2 each. Then set phi_k = clip(k x_k, -1, 1) with k in logspace(-2, 2, 41), taking the cheapest k (J at s) that is certified at s.
76
+ - **Redraws.** If no k is certified, redraw F, up to 32 draws per start; the draw count is recorded. If all 32 fail, the start is empty. Seed (2929, i, scale index).
77
+ - **Optimization.** Same as the X19 `oracle_chunk`: Adam on mean J(s) through `realize_torch` + `simulate`, F-stability rejection with f_margin 0.01, a snapshot every 25 steps. phi = clamp(start * exp(ts), -1, 1).
78
+ - **lr and step budget.** The values X19 selected on its validation set (read from results/icml_r3/X19). Nothing is tuned on the pilot plants.
79
+ - **Selection.** Per start, the last snapshot certified at s, walking back; the start is certified by construction. The reference is the cheapest over the starts. Its r* is recorded.
80
+
81
+ e. **Not run: generator, retrieval, regressor, CVAE.** No unstable-plant training data exists; building it is what this gate decides.
82
+
83
+ ## 6. Smoke rule (the only pre-run adjustment allowed)
84
+
85
+ Run the reference on the 4 smoke plants at the X19-selected lr. If it crashes or produces NaN, use the next smaller value in {0.1, 0.05, 0.02}. Record the outcome below, under "Amendments", before the pilot run.
86
+
87
+ ## 7. Outputs
88
+
89
+ - Per plant: linear stabilizability; r*_SG; sigma_p; r*_LQG (tuned); r*_ref(s1); r*_ref(s2); J at s1 and at s2 for every arm.
90
+ - Start difficulty: the share of prior-F draws (with sign) for which some k certifies at s. It is a sampled measure of how rare certified filters are under the dataset prior.
91
+ - Everything is reported per stratum and pooled.
92
+ - CIs are 95% percentile bootstrap over plants (10,000 resamples). Paired differences use the Wilcoxon signed-rank test.
93
+
94
+ ## 8. Gate (decision rule, on the 40 pilot plants)
95
+
96
+ - **G1, stability is a real problem.** Either
97
+ - no static output gain stabilizes, by the linear check, on >= 25% of plants; or
98
+ - the tuned saturated static gain is not certified at s1 on >= 25% of plants.
99
+ - **G2, the nonlinear dynamic class pays.** Either route is enough.
100
+ - **Cost.** Take the plants where the reference is certified at s1. The median over them of (J_lin(s1) - J_ref(s1))/J_ref(s1) must be >= 5%, with the CI lower bound > 0. Here J_lin = min(J_SG, J_LQG tuned). A plant where neither linear baseline is certified at s1 counts as a win, and those plants are also reported separately.
101
+ - **Region.** Take the plants where neither tuned linear baseline is certified at s2; there must be at least 8 of them. The reference must be certified at s2 on >= 50% of them.
102
+ - **Verdict.** GO iff G1 and G2. Also report:
103
+ - the share of plants where the tuned LQG is cheaper than the reference at s1, which signals whether the controller class should see u;
104
+ - per-stratum results (descriptive only).
105
+
106
+ ## 9. Compute and rules
107
+
108
+ - **Where and when.** virgo, CPU only (`CUDA_VISIBLE_DEVICES=`), at most 12 worker processes in total. Start only after X19's processes have exited (`pgrep -f siso_x19` is empty). There is no wall-time cap.
109
+ - **Code.**
110
+ - `nonlinear/experiments/siso_x20_unstable.py` imports from `siso_x19.py`, which must not be modified.
111
+ - The self-check is `nonlinear/testing/test_regional_cert.py`.
112
+ - **Outputs.** Results go in results/icml_r3/X20/; the figure in figures/icml_r3/x20_unstable.{png,pdf}, with no titles and no in-axes text; the write-up in docs/icml_r3_x20.md (research-doc skill).
113
+ - **Not allowed.** No commit, and no edits to paper_icml.
114
+
115
+ ## Amendments (before the pilot run only; each dated, with the reason)
116
+
117
+ ### A1 (2026-10-14, before any smoke or pilot run; design correction by the orchestrator)
118
+ Found while implementing, from single-process diagnostics on the pilot plants; no pilot outputs exist yet.
119
+
120
+ **Problems.**
121
+ - (a) **Gain cap.** The knots sit on integers in y_F. With |phi| <= 1 the small-signal gain of phi is therefore at most 1, and clip(k x_k, -1, 1) equals sat_1 for every k >= 1. So the gain must live in F.
122
+ - (b) **Phase lag from F.** The dataset prior's F is strictly proper with relative degree 2. On unstable plants of relative degree 2-3 that makes the loop relative degree 4-5, and random prior draws rarely stabilize:
123
+ - with gain <= 1, only 7/40 plants had any stabilizing draw among 60;
124
+ - with gain up to 1e4 in F, 22/40.
125
+ With these starts the reference would be empty on most plants, and G2 could not be evaluated.
126
+
127
+ **Changes.**
128
+ 1. **Controller class.** F is stable and proper (relative degree 0, 1 or 2), of order <= 4, with a free gain. phi is as before, with |phi(x_k)| <= 1. The certificate is unchanged: L = G F is strictly proper because G is.
129
+ 2. **Arm b.** F = c (static), with c in +-logspace(-2, 4, 121) and phi_k = clip(x_k, -1, 1), so u = -sat_1(c y). The linear stabilizability check is unchanged.
130
+ 3. **Reference starts.**
131
+ - Draw F with stable zeros and poles as in the dataset prior, but with relative degree drawn from {0, 1, 2}, gain log-uniform on [1e-1, 1e4], and sign +-1. Set phi_k = clip(x_k, -1, 1).
132
+ - Then minimize the spectral abscissa of the linearized loop A - B C (phi'(0) = 1) over F's coefficients: Adam, at most 300 steps, lr 0.05, stopping once the abscissa is <= -0.05, with the same F-stability rejection as the optimizer. This is the fixed-order stabilization approach of HIFOO (Burke, Henrion, Lewis, Overton 2006).
133
+ - A start is accepted if it is certified at s; otherwise F is redrawn, at most 32 times.
134
+ - Reported: the share of raw prior draws that already stabilize, the old relative-degree-2 prior's share, and the draws and steps needed to reach an accepted start.
135
+ - The reference optimizer runs on the proper-F parametrization: num with 5 coefficients and den with 4, with the same exp/(1+theta) scaling.
136
+ 4. **Self-check.** On the PREREG ybar grid, the scalar loop's exact grid-limited radius is 2.5/3 = 0.833, and that is what is asserted there. The [0.85, 1.0] band is asserted on a fine grid (r* = 0.99). s = 1.05 must not be certified on either grid.
137
+ 5. **Single-gain levels (k1 = k2).** These are solved as the Lyapunov LMI of A - kBC, the tau -> infinity limit, which is valid because psi is linear there. The re-check uses a rescaled (P, tau). Levels are pruned only where pruning cannot change r*.
138
+
139
+ Nothing else changes. The gate (section 8) is unchanged.
140
+
141
+ ### A2 (2026-10-14, before any smoke or pilot run; solver details by the orchestrator)
142
+ 1. **SCS fallback.** SCS runs only when CLARABEL fails: an error, an inaccurate status, or a solution that fails the re-check. A clean CLARABEL infeasibility certificate stands. Under the old reading, SCS ran for about 2.8 s per infeasible level and never once made such a level feasible.
143
+ 2. **Sector parameter.** It is symmetrized; it came out asymmetric at round-off with high-gain C. This never produced a wrong number, only solver errors.
144
+ 3. **State rescaling.** Every SDP is solved after an exact diagonal similarity transform x = T z, with T from matrix balancing of A, and B, C and d_i transformed to match. r* is invariant under it. It is applied because about 20% of the single-gain Lyapunov solves failed the strict re-check with high-gain F (5-plant diagnostic). Such failures count as not certified and could bias arms unequally.
145
+ 4. **As implemented (replaces the T in item 3).** Each SDP is solved by CLARABEL in three exact coordinate systems: a Lyapunov-Cholesky transform of the closed loop, diagonal balancing, and raw. The largest verified radius is kept, since every verified solution is a valid certificate. If none verifies and none returns a clean infeasibility certificate, CLARABEL is retried once with P rescaled to norm ~1, and SCS is tried after that.
146
+ - **5-plant diagnostic (1,462 levels):** re-check failures fell from 3.6% to 1.0%. No level certified in raw coordinates is lost, and 39 more levels are certified.
147
+ - **Accuracy:** coordinate invariance of r* holds only to CLARABEL accuracy (median 7e-6 relative, worst 3.5e-3).
148
+
149
+ ### A3 (2026-10-14 08:47 PT, HPC) smoke: lr 0.1 -> SMOKE OK (no crash, no NaN); pilot uses lr 0.1.
150
+
151
+ ### A4 (2026-10-14, during the pilot; code fix by the orchestrator)
152
+ - **Failure.** Pilot job 14040 stopped with 37/40 plants saved. On plant 30, 33 or 38, the reference's walk-back found no certified snapshot: it re-solved snapshot 0, which is exactly the accepted start, and got r* < s. The run died on the assertion "snapshot 0 is the certified start".
153
+ - **Cause.** The start had already been verified certified at s in `ref_starts`. A re-solve of the identical controller can land a hair below s, because cvxpy reuses CLARABEL's solver state across calls; starts accepted with r* ~ s then flip.
154
+ - **Fix.** If the walk-back reaches snapshot 0 without a certificate, the verified start is kept. This is the PREREG's "start certified by construction". The count is stored as ref_opt_n_start_reused.
155
+ - **Scope.**
156
+ - The 37 completed plants never reached this path, so they are unchanged.
157
+ - The 3 missing plants are re-run with the fixed code.
158
+ - Soundness is unaffected: every selected controller passed the numpy re-check of a solved certificate.
results/icml_r3/X20/logs/pilot.log ADDED
@@ -0,0 +1,465 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 40 plants, 0 already saved, cfg {'starts': 16, 'steps': 400, 'ckpt_every': 25, 'lr': 0.1, 'bcap': 10000.0, 'bisect_iters': 10, 'dt': 0.03, 'f_margin': 0.01}, 28 workers, deadline none
2
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
3
+ warnings.warn(
4
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
5
+ warnings.warn(
6
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
7
+ warnings.warn(
8
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
9
+ warnings.warn(
10
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
11
+ warnings.warn(
12
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
13
+ warnings.warn(
14
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
15
+ warnings.warn(
16
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
17
+ warnings.warn(
18
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
19
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
20
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
21
+ warnings.warn(
22
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
23
+ warnings.warn(
24
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
25
+ warnings.warn(
26
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
27
+ warnings.warn(
28
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
29
+ warnings.warn(
30
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
31
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
32
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
33
+ warnings.warn(
34
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
35
+ warnings.warn(
36
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
37
+ warnings.warn(
38
+ [plant 15 A] linear check False, sigma_p 0.4655, LQG r*/sigma max 1.005 (2 s)
39
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
40
+ warnings.warn(
41
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
42
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
43
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
44
+ warnings.warn(
45
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
46
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
47
+ [plant 15 A] static gain r*/sigma max 0.000 (4 s)
48
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
49
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
50
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
51
+ warnings.warn(
52
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
53
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
54
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
55
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
56
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
57
+ warnings.warn(
58
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
59
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
60
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
61
+ warnings.warn(
62
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
63
+ warnings.warn(
64
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
65
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
66
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
67
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
68
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
69
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
70
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
71
+ warnings.warn(
72
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
73
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
74
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
75
+ warnings.warn(
76
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
77
+ warnings.warn(
78
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
79
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
80
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
81
+ warnings.warn(
82
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
83
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
84
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
85
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
86
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
87
+ warnings.warn(
88
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
89
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
90
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
91
+ warnings.warn(
92
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
93
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
94
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
95
+ num, den = normalize(num, den) # Strips zeros, checks arrays
96
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
97
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
98
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
99
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
100
+ [plant 23 B] linear check False, sigma_p 0.1352, LQG r*/sigma max 1.227 (5 s)
101
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
102
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
103
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
104
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
105
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
106
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
107
+ [plant 23 B] static gain r*/sigma max 0.000 (7 s)
108
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
109
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
110
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
111
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
112
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
113
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
114
+ [plant 21 B] linear check False, sigma_p 0.2917, LQG r*/sigma max 1.000 (9 s)
115
+ [plant 13 A] linear check False, sigma_p 1.021, LQG r*/sigma max 1.009 (9 s)
116
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
117
+ num, den = normalize(num, den) # Strips zeros, checks arrays
118
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
119
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
120
+ [plant 21 B] static gain r*/sigma max 0.000 (11 s)
121
+ [plant 13 A] static gain r*/sigma max 0.000 (11 s)
122
+ [plant 16 A] linear check True, sigma_p 0.5384, LQG r*/sigma max 1.002 (11 s)
123
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
124
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
125
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
126
+ num, den = normalize(num, den) # Strips zeros, checks arrays
127
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
128
+ num, den = normalize(num, den) # Strips zeros, checks arrays
129
+ [plant 11 A] linear check False, sigma_p 0.3086, LQG r*/sigma max 1.089 (13 s)
130
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
131
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
132
+ [plant 11 A] static gain r*/sigma max 0.000 (15 s)
133
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
134
+ num, den = normalize(num, den) # Strips zeros, checks arrays
135
+ [plant 19 A] linear check True, sigma_p 0.698, LQG r*/sigma max 1.128 (19 s)
136
+ [plant 21 B] scale 0: 16/16 starts, 41 draws, raw stable 0.00, HIFOO stable 0.63, median steps 103 (20 s)
137
+ [plant 19 A] static gain r*/sigma max 0.994 (22 s)
138
+ [plant 25 B] linear check False, sigma_p 1.446, LQG r*/sigma max 1.272 (23 s)
139
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
140
+ num, den = normalize(num, den) # Strips zeros, checks arrays
141
+ [plant 03 A] linear check True, sigma_p 0.3032, LQG r*/sigma max 1.210 (25 s)
142
+ [plant 25 B] static gain r*/sigma max 0.000 (25 s)
143
+ [plant 00 A] linear check False, sigma_p 0.1062, LQG r*/sigma max 1.037 (26 s)
144
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
145
+ num, den = normalize(num, den) # Strips zeros, checks arrays
146
+ [plant 03 A] static gain r*/sigma max 0.000 (27 s)
147
+ [plant 04 A] linear check True, sigma_p 0.8838, LQG r*/sigma max 1.002 (27 s)
148
+ [plant 00 A] static gain r*/sigma max 0.000 (28 s)
149
+ [plant 23 B] scale 0: 16/16 starts, 77 draws, raw stable 0.00, HIFOO stable 0.27, median steps 300 (28 s)
150
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
151
+ num, den = normalize(num, den) # Strips zeros, checks arrays
152
+ [plant 04 A] static gain r*/sigma max 0.000 (29 s)
153
+ [plant 19 A] scale 0: 16/16 starts, 38 draws, raw stable 0.03, HIFOO stable 0.82, median steps 28 (29 s)
154
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
155
+ num, den = normalize(num, den) # Strips zeros, checks arrays
156
+ [plant 05 A] linear check False, sigma_p 0.1277, LQG r*/sigma max 1.018 (30 s)
157
+ [plant 12 A] linear check False, sigma_p 0.8353, LQG r*/sigma max 1.109 (30 s)
158
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
159
+ num, den = normalize(num, den) # Strips zeros, checks arrays
160
+ [plant 11 A] scale 0: 16/16 starts, 50 draws, raw stable 0.06, HIFOO stable 0.74, median steps 54 (30 s)
161
+ [plant 12 A] static gain r*/sigma max 0.000 (32 s)
162
+ [plant 22 B] linear check True, sigma_p 0.3161, LQG r*/sigma max 1.513 (32 s)
163
+ [plant 05 A] static gain r*/sigma max 0.000 (32 s)
164
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
165
+ num, den = normalize(num, den) # Strips zeros, checks arrays
166
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
167
+ num, den = normalize(num, den) # Strips zeros, checks arrays
168
+ [plant 10 A] linear check False, sigma_p 0.1591, LQG r*/sigma max 1.147 (34 s)
169
+ [plant 18 A] linear check False, sigma_p 0.1865, LQG r*/sigma max 1.371 (34 s)
170
+ [plant 17 A] linear check True, sigma_p 0.9951, LQG r*/sigma max 1.015 (34 s)
171
+ [plant 16 A] static gain r*/sigma max 1.508 (35 s)
172
+ [plant 18 A] static gain r*/sigma max 0.000 (36 s)
173
+ [plant 10 A] static gain r*/sigma max 0.000 (36 s)
174
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
175
+ num, den = normalize(num, den) # Strips zeros, checks arrays
176
+ [plant 07 A] linear check False, sigma_p 0.4043, LQG r*/sigma max 1.182 (37 s)
177
+ [plant 22 B] static gain r*/sigma max 1.202 (37 s)
178
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
179
+ num, den = normalize(num, den) # Strips zeros, checks arrays
180
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
181
+ num, den = normalize(num, den) # Strips zeros, checks arrays
182
+ [plant 14 A] linear check False, sigma_p 0.1996, LQG r*/sigma max 1.029 (38 s)
183
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
184
+ num, den = normalize(num, den) # Strips zeros, checks arrays
185
+ [plant 07 A] static gain r*/sigma max 0.000 (39 s)
186
+ [plant 15 A] scale 0: 16/16 starts, 119 draws, raw stable 0.00, HIFOO stable 0.53, median steps 96 (39 s)
187
+ [plant 14 A] static gain r*/sigma max 0.000 (40 s)
188
+ [plant 13 A] scale 0: 16/16 starts, 51 draws, raw stable 0.00, HIFOO stable 0.65, median steps 83 (40 s)
189
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
190
+ num, den = normalize(num, den) # Strips zeros, checks arrays
191
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
192
+ num, den = normalize(num, den) # Strips zeros, checks arrays
193
+ [plant 09 A] linear check False, sigma_p 0.52, LQG r*/sigma max 1.064 (48 s)
194
+ [plant 09 A] static gain r*/sigma max 0.000 (50 s)
195
+ [plant 03 A] scale 0: 16/16 starts, 44 draws, raw stable 0.05, HIFOO stable 0.80, median steps 32 (52 s)
196
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
197
+ num, den = normalize(num, den) # Strips zeros, checks arrays
198
+ [plant 01 A] linear check True, sigma_p 0.1761, LQG r*/sigma max 2.498 (53 s)
199
+ [plant 25 B] scale 0: 16/16 starts, 35 draws, raw stable 0.03, HIFOO stable 0.89, median steps 33 (53 s)
200
+ [plant 08 A] linear check False, sigma_p 0.1932, LQG r*/sigma max 1.000 (55 s)
201
+ [plant 08 A] static gain r*/sigma max 0.000 (57 s)
202
+ [plant 04 A] scale 0: 16/16 starts, 34 draws, raw stable 0.03, HIFOO stable 0.79, median steps 41 (57 s)
203
+ [plant 06 A] linear check True, sigma_p 0.628, LQG r*/sigma max 2.034 (58 s)
204
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
205
+ num, den = normalize(num, den) # Strips zeros, checks arrays
206
+ [plant 17 A] static gain r*/sigma max 1.151 (59 s)
207
+ [plant 24 B] linear check False, sigma_p 0.04857, LQG r*/sigma max 1.301 (60 s)
208
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
209
+ num, den = normalize(num, den) # Strips zeros, checks arrays
210
+ [plant 06 A] static gain r*/sigma max 1.934 (63 s)
211
+ [plant 24 B] static gain r*/sigma max 0.000 (62 s)
212
+ [plant 10 A] scale 0: 16/16 starts, 109 draws, raw stable 0.00, HIFOO stable 0.32, median steps 300 (63 s)
213
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
214
+ num, den = normalize(num, den) # Strips zeros, checks arrays
215
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
216
+ num, den = normalize(num, den) # Strips zeros, checks arrays
217
+ [plant 02 A] linear check False, sigma_p 0.1661, LQG r*/sigma max 1.072 (64 s)
218
+ [plant 20 B] linear check True, sigma_p 0.6302, LQG r*/sigma max 1.644 (65 s)
219
+ [plant 02 A] static gain r*/sigma max 0.000 (66 s)
220
+ [plant 05 A] scale 0: 16/16 starts, 131 draws, raw stable 0.00, HIFOO stable 0.31, median steps 300 (67 s)
221
+ [plant 27 B] linear check False, sigma_p 0.178, LQG r*/sigma max 1.110 (67 s)
222
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
223
+ num, den = normalize(num, den) # Strips zeros, checks arrays
224
+ [plant 16 A] scale 0: 16/16 starts, 51 draws, raw stable 0.06, HIFOO stable 0.61, median steps 56 (68 s)
225
+ [plant 20 B] static gain r*/sigma max 0.796 (69 s)
226
+ [plant 27 B] static gain r*/sigma max 0.000 (69 s)
227
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
228
+ num, den = normalize(num, den) # Strips zeros, checks arrays
229
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
230
+ num, den = normalize(num, den) # Strips zeros, checks arrays
231
+ [plant 08 A] scale 0: 16/16 starts, 69 draws, raw stable 0.03, HIFOO stable 0.45, median steps 300 (74 s)
232
+ [plant 18 A] scale 0: 16/16 starts, 54 draws, raw stable 0.07, HIFOO stable 0.81, median steps 28 (78 s)
233
+ [plant 14 A] scale 0: 16/16 starts, 97 draws, raw stable 0.02, HIFOO stable 0.47, median steps 300 (79 s)
234
+ [plant 00 A] scale 0: 16/16 starts, 171 draws, raw stable 0.01, HIFOO stable 0.30, median steps 300 (82 s)
235
+ [plant 26 B] linear check False, sigma_p 0.1039, LQG r*/sigma max 1.713 (86 s)
236
+ [plant 26 B] static gain r*/sigma max 0.000 (87 s)
237
+ [plant 01 A] static gain r*/sigma max 2.651 (89 s)
238
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
239
+ num, den = normalize(num, den) # Strips zeros, checks arrays
240
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
241
+ num, den = normalize(num, den) # Strips zeros, checks arrays
242
+ [plant 22 B] scale 0: 16/16 starts, 33 draws, raw stable 0.00, HIFOO stable 0.82, median steps 42 (93 s)
243
+ [plant 27 B] scale 0: 16/16 starts, 90 draws, raw stable 0.00, HIFOO stable 0.34, median steps 300 (95 s)
244
+ [plant 17 A] scale 0: 16/16 starts, 59 draws, raw stable 0.07, HIFOO stable 0.61, median steps 56 (96 s)
245
+ [plant 24 B] scale 0: 16/16 starts, 171 draws, raw stable 0.00, HIFOO stable 0.17, median steps 300 (105 s)
246
+ [plant 07 A] scale 0: 16/16 starts, 78 draws, raw stable 0.00, HIFOO stable 0.45, median steps 300 (107 s)
247
+ [plant 06 A] scale 0: 16/16 starts, 31 draws, raw stable 0.03, HIFOO stable 0.90, median steps 21 (114 s)
248
+ [plant 01 A] scale 0: 16/16 starts, 24 draws, raw stable 0.12, HIFOO stable 0.92, median steps 26 (114 s)
249
+ [plant 20 B] scale 0: 16/16 starts, 32 draws, raw stable 0.06, HIFOO stable 0.88, median steps 46 (118 s)
250
+ [plant 26 B] scale 0: 16/16 starts, 123 draws, raw stable 0.00, HIFOO stable 0.24, median steps 300 (122 s)
251
+ [plant 02 A] scale 0: 16/16 starts, 52 draws, raw stable 0.00, HIFOO stable 0.58, median steps 66 (127 s)
252
+ [plant 12 A] scale 0: 16/16 starts, 73 draws, raw stable 0.03, HIFOO stable 0.48, median steps 300 (135 s)
253
+ [plant 09 A] scale 0: 16/16 starts, 54 draws, raw stable 0.04, HIFOO stable 0.56, median steps 70 (137 s)
254
+ [plant 21 B] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.59, median steps 108 (156 s)
255
+ [plant 23 B] scale 1: 1/16 starts, 495 draws, raw stable 0.00, HIFOO stable 0.41, median steps 300 (183 s)
256
+ [plant 15 A] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.60, median steps 76 (192 s)
257
+ [plant 01 A] scale 1: 16/16 starts, 51 draws, raw stable 0.10, HIFOO stable 0.88, median steps 28 (204 s)
258
+ [plant 05 A] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.30, median steps 300 (207 s)
259
+ [plant 26 B] scale 1: 9/16 starts, 344 draws, raw stable 0.00, HIFOO stable 0.28, median steps 300 (214 s)
260
+ [plant 00 A] scale 1: 1/16 starts, 490 draws, raw stable 0.00, HIFOO stable 0.35, median steps 300 (215 s)
261
+ [plant 03 A] scale 1: 0/16 starts, 512 draws, raw stable 0.02, HIFOO stable 0.82, median steps 33 (220 s)
262
+ [plant 10 A] scale 1: 1/16 starts, 486 draws, raw stable 0.01, HIFOO stable 0.32, median steps 300 (225 s)
263
+ [plant 24 B] scale 1: 1/16 starts, 493 draws, raw stable 0.00, HIFOO stable 0.16, median steps 300 (224 s)
264
+ [plant 27 B] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.37, median steps 300 (235 s)
265
+ [plant 08 A] scale 1: 0/16 starts, 512 draws, raw stable 0.01, HIFOO stable 0.51, median steps 300 (239 s)
266
+ [plant 11 A] scale 1: 0/16 starts, 512 draws, raw stable 0.02, HIFOO stable 0.72, median steps 57 (242 s)
267
+ [plant 06 A] scale 1: 15/16 starts, 117 draws, raw stable 0.05, HIFOO stable 0.78, median steps 32 (261 s)
268
+ [plant 19 A] scale 1: 0/16 starts, 512 draws, raw stable 0.03, HIFOO stable 0.78, median steps 42 (261 s)
269
+ [plant 14 A] scale 1: 0/16 starts, 512 draws, raw stable 0.02, HIFOO stable 0.54, median steps 99 (284 s)
270
+ [plant 18 A] scale 1: 3/16 starts, 473 draws, raw stable 0.05, HIFOO stable 0.84, median steps 31 (337 s)
271
+ [plant 04 A] scale 1: 0/16 starts, 512 draws, raw stable 0.03, HIFOO stable 0.77, median steps 36 (349 s)
272
+ [plant 16 A] scale 1: 0/16 starts, 512 draws, raw stable 0.05, HIFOO stable 0.69, median steps 52 (349 s)
273
+ [plant 02 A] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.49, median steps 300 (358 s)
274
+ [plant 13 A] scale 1: 0/16 starts, 512 draws, raw stable 0.02, HIFOO stable 0.69, median steps 50 (403 s)
275
+ [plant 07 A] scale 1: 0/16 starts, 512 draws, raw stable 0.02, HIFOO stable 0.44, median steps 300 (413 s)
276
+ [plant 12 A] scale 1: 0/16 starts, 512 draws, raw stable 0.02, HIFOO stable 0.50, median steps 300 (469 s)
277
+ [plant 17 A] scale 1: 0/16 starts, 512 draws, raw stable 0.07, HIFOO stable 0.61, median steps 66 (474 s)
278
+ [plant 20 B] scale 1: 13/16 starts, 302 draws, raw stable 0.03, HIFOO stable 0.81, median steps 40 (517 s)
279
+ [plant 09 A] scale 1: 0/16 starts, 512 draws, raw stable 0.02, HIFOO stable 0.58, median steps 70 (770 s)
280
+ [plant 25 B] scale 1: 0/16 starts, 512 draws, raw stable 0.06, HIFOO stable 0.83, median steps 36 (863 s)
281
+ [plant 22 B] scale 1: 0/16 starts, 512 draws, raw stable 0.02, HIFOO stable 0.87, median steps 42 (1021 s)
282
+ [plant 21 B] done: J_ref s1 0.4491, s2 inf (2021 s)
283
+ saved plant 21 in 2021 s (2022 s elapsed)
284
+ [plant 15 A] done: J_ref s1 0.2685, s2 inf (2043 s)
285
+ saved plant 15 in 2043 s (2043 s elapsed)
286
+ [plant 28 B] linear check False, sigma_p 0.2017, LQG r*/sigma max 1.040 (46 s)
287
+ [plant 28 B] static gain r*/sigma max 0.000 (48 s)
288
+ [plant 10 A] done: J_ref s1 0.07844, s2 1.167 (2083 s)
289
+ saved plant 10 in 2083 s (2083 s elapsed)
290
+ [plant 28 B] scale 0: 16/16 starts, 82 draws, raw stable 0.00, HIFOO stable 0.50, median steps 300 (75 s)
291
+ [plant 00 A] done: J_ref s1 0.07, s2 1.777 (2099 s)
292
+ saved plant 00 in 2099 s (2099 s elapsed)
293
+ [plant 11 A] done: J_ref s1 0.1623, s2 inf (2122 s)
294
+ saved plant 11 in 2122 s (2122 s elapsed)
295
+ [plant 29 B] linear check False, sigma_p 0.2988, LQG r*/sigma max 1.483 (80 s)
296
+ [plant 29 B] static gain r*/sigma max 0.000 (82 s)
297
+ [plant 24 B] done: J_ref s1 0.08776, s2 0.5341 (2131 s)
298
+ saved plant 24 in 2131 s (2132 s elapsed)
299
+ [plant 31 B] linear check False, sigma_p 0.0471, LQG r*/sigma max 1.003 (33 s)
300
+ [plant 31 B] static gain r*/sigma max 0.000 (35 s)
301
+ [plant 30 B] linear check False, sigma_p 0.1173, LQG r*/sigma max 2.347 (52 s)
302
+ [plant 03 A] done: J_ref s1 0.2767, s2 inf (2136 s)
303
+ saved plant 03 in 2136 s (2136 s elapsed)
304
+ [plant 30 B] static gain r*/sigma max 0.000 (54 s)
305
+ [plant 32 B] linear check False, sigma_p 0.4131, LQG r*/sigma max 1.009 (16 s)
306
+ [plant 32 B] static gain r*/sigma max 0.000 (18 s)
307
+ [plant 30 B] scale 0: 16/16 starts, 37 draws, raw stable 0.00, HIFOO stable 0.46, median steps 300 (64 s)
308
+ [plant 29 B] scale 0: 16/16 starts, 88 draws, raw stable 0.00, HIFOO stable 0.34, median steps 300 (105 s)
309
+ [plant 32 B] scale 0: 16/16 starts, 51 draws, raw stable 0.00, HIFOO stable 0.57, median steps 196 (32 s)
310
+ [plant 08 A] done: J_ref s1 0.05793, s2 inf (2170 s)
311
+ saved plant 08 in 2170 s (2170 s elapsed)
312
+ [plant 30 B] scale 1: 16/16 starts, 78 draws, raw stable 0.00, HIFOO stable 0.38, median steps 300 (96 s)
313
+ [plant 33 B] linear check False, sigma_p 0.4651, LQG r*/sigma max 1.432 (50 s)
314
+ [plant 33 B] static gain r*/sigma max 0.000 (52 s)
315
+ [plant 19 A] done: J_ref s1 0.7719, s2 inf (2202 s)
316
+ saved plant 19 in 2202 s (2203 s elapsed)
317
+ [plant 05 A] done: J_ref s1 0.0927, s2 inf (2209 s)
318
+ saved plant 05 in 2209 s (2209 s elapsed)
319
+ [plant 36 B] linear check True, sigma_p 0.5485, LQG r*/sigma max 1.258 (19 s)
320
+ [plant 36 B] static gain r*/sigma max 0.740 (22 s)
321
+ [plant 34 B] linear check False, sigma_p 0.6809, LQG r*/sigma max 1.476 (92 s)
322
+ [plant 33 B] scale 0: 16/16 starts, 136 draws, raw stable 0.01, HIFOO stable 0.24, median steps 300 (97 s)
323
+ [plant 35 B] linear check False, sigma_p 0.2569, LQG r*/sigma max 1.015 (60 s)
324
+ [plant 34 B] static gain r*/sigma max 0.000 (94 s)
325
+ [plant 35 B] static gain r*/sigma max 0.000 (62 s)
326
+ [plant 31 B] scale 0: 5/16 starts, 440 draws, raw stable 0.00, HIFOO stable 0.11, median steps 300 (134 s)
327
+ [plant 29 B] scale 1: 8/16 starts, 340 draws, raw stable 0.00, HIFOO stable 0.28, median steps 300 (207 s)
328
+ [plant 23 B] done: J_ref s1 0.06168, s2 1.294 (2251 s)
329
+ saved plant 23 in 2251 s (2251 s elapsed)
330
+ [plant 27 B] done: J_ref s1 0.1381, s2 inf (2259 s)
331
+ saved plant 27 in 2259 s (2260 s elapsed)
332
+ [plant 28 B] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.60, median steps 118 (243 s)
333
+ [plant 39 B] linear check False, sigma_p 0.2921, LQG r*/sigma max 1.000 (9 s)
334
+ [plant 39 B] static gain r*/sigma max 0.000 (11 s)
335
+ [plant 37 B] linear check False, sigma_p 0.4795, LQG r*/sigma max 1.187 (66 s)
336
+ [plant 37 B] static gain r*/sigma max 0.000 (68 s)
337
+ [plant 14 A] done: J_ref s1 0.06632, s2 inf (2278 s)
338
+ saved plant 14 in 2278 s (2278 s elapsed)
339
+ [plant 39 B] scale 0: 16/16 starts, 37 draws, raw stable 0.00, HIFOO stable 0.73, median steps 79 (26 s)
340
+ [plant 34 B] scale 0: 16/16 starts, 138 draws, raw stable 0.00, HIFOO stable 0.24, median steps 300 (150 s)
341
+ [plant 35 B] scale 0: 16/16 starts, 140 draws, raw stable 0.00, HIFOO stable 0.34, median steps 300 (116 s)
342
+ [plant 36 B] scale 0: 16/16 starts, 36 draws, raw stable 0.03, HIFOO stable 0.89, median steps 46 (84 s)
343
+ [plant 32 B] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.48, median steps 300 (175 s)
344
+ [plant 37 B] scale 0: 16/16 starts, 62 draws, raw stable 0.00, HIFOO stable 0.45, median steps 300 (90 s)
345
+ [plant 38 B] linear check False, sigma_p 0.06082, LQG r*/sigma max 1.016 (53 s)
346
+ [plant 38 B] static gain r*/sigma max 0.000 (55 s)
347
+ [plant 13 A] done: J_ref s1 0.4998, s2 inf (2323 s)
348
+ saved plant 13 in 2323 s (2323 s elapsed)
349
+ [plant 16 A] done: J_ref s1 0.1398, s2 inf (2336 s)
350
+ saved plant 16 in 2336 s (2336 s elapsed)
351
+ [plant 26 B] done: J_ref s1 0.02325, s2 0.6452 (2343 s)
352
+ saved plant 26 in 2343 s (2344 s elapsed)
353
+ [plant 38 B] scale 0: 16/16 starts, 105 draws, raw stable 0.00, HIFOO stable 0.35, median steps 300 (94 s)
354
+ [plant 31 B] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.10, median steps 300 (246 s)
355
+ [plant 06 A] done: J_ref s1 1.455, s2 12.14 (2348 s)
356
+ saved plant 06 in 2348 s (2348 s elapsed)
357
+ [plant 04 A] done: J_ref s1 0.7096, s2 inf (2370 s)
358
+ saved plant 04 in 2370 s (2370 s elapsed)
359
+ [plant 33 B] scale 1: 6/16 starts, 448 draws, raw stable 0.01, HIFOO stable 0.32, median steps 300 (264 s)
360
+ [plant 35 B] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.33, median steps 300 (241 s)
361
+ [plant 02 A] done: J_ref s1 0.1312, s2 inf (2436 s)
362
+ saved plant 02 in 2436 s (2436 s elapsed)
363
+ [plant 18 A] done: J_ref s1 0.2385, s2 3.135 (2439 s)
364
+ saved plant 18 in 2439 s (2439 s elapsed)
365
+ [plant 34 B] scale 1: 3/16 starts, 449 draws, raw stable 0.00, HIFOO stable 0.22, median steps 300 (313 s)
366
+ [plant 07 A] done: J_ref s1 0.1158, s2 inf (2469 s)
367
+ saved plant 07 in 2469 s (2469 s elapsed)
368
+ [plant 39 B] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.56, median steps 148 (213 s)
369
+ [plant 12 A] done: J_ref s1 0.1728, s2 inf (2492 s)
370
+ saved plant 12 in 2492 s (2492 s elapsed)
371
+ [plant 17 A] done: J_ref s1 0.1323, s2 inf (2493 s)
372
+ saved plant 17 in 2493 s (2493 s elapsed)
373
+ [plant 38 B] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.34, median steps 300 (248 s)
374
+ [plant 37 B] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.37, median steps 300 (373 s)
375
+ [plant 20 B] done: J_ref s1 2.162, s2 19.88 (2604 s)
376
+ saved plant 20 in 2604 s (2604 s elapsed)
377
+ [plant 01 A] done: J_ref s1 0.4187, s2 5.441 (2615 s)
378
+ saved plant 01 in 2615 s (2616 s elapsed)
379
+ [plant 09 A] done: J_ref s1 0.1021, s2 inf (2644 s)
380
+ saved plant 09 in 2644 s (2644 s elapsed)
381
+ [plant 25 B] done: J_ref s1 2.155, s2 inf (2796 s)
382
+ saved plant 25 in 2796 s (2797 s elapsed)
383
+ [plant 22 B] done: J_ref s1 2.383, s2 inf (2921 s)
384
+ saved plant 22 in 2921 s (2921 s elapsed)
385
+ [plant 36 B] scale 1: 0/16 starts, 512 draws, raw stable 0.01, HIFOO stable 0.88, median steps 35 (992 s)
386
+ [plant 31 B] done: J_ref s1 0.1876, s2 inf (1831 s)
387
+ saved plant 31 in 1831 s (3930 s elapsed)
388
+ [plant 28 B] done: J_ref s1 0.2483, s2 inf (1968 s)
389
+ saved plant 28 in 1968 s (3989 s elapsed)
390
+ [plant 29 B] done: J_ref s1 0.07799, s2 0.8312 (1949 s)
391
+ saved plant 29 in 1949 s (3992 s elapsed)
392
+ [plant 32 B] done: J_ref s1 0.2592, s2 inf (1906 s)
393
+ [plant 35 B] done: J_ref s1 0.1507, s2 inf (1947 s)
394
+ [plant 39 B] done: J_ref s1 0.2919, s2 inf (1893 s)
395
+ [plant 34 B] done: J_ref s1 0.2059, s2 3.198 (2033 s)
396
+ [plant 37 B] done: J_ref s1 0.2859, s2 inf (2047 s)
397
+ [plant 36 B] done: J_ref s1 1.738, s2 inf (2680 s)
398
+ concurrent.futures.process._RemoteTraceback:
399
+ """
400
+ Traceback (most recent call last):
401
+ File "/home/DATA/software/software/python/3.10.0/lib/python3.10/concurrent/futures/process.py", line 243, in _process_worker
402
+ r = call_item.fn(*call_item.args, **call_item.kwargs)
403
+ File "/home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py", line 528, in plant_job
404
+ o = ref_optimize(g, starts, x0s, ss, D, cfg)
405
+ File "/home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py", line 454, in ref_optimize
406
+ assert sel[m] is not None, "snapshot 0 is the certified start"
407
+ AssertionError: snapshot 0 is the certified start
408
+ """
409
+
410
+ The above exception was the direct cause of the following exception:
411
+
412
+ Traceback (most recent call last):
413
+ File "/home/DATA/software/software/python/3.10.0/lib/python3.10/runpy.py", line 196, in _run_module_as_main
414
+ return _run_code(code, main_globals, None,
415
+ File "/home/DATA/software/software/python/3.10.0/lib/python3.10/runpy.py", line 86, in _run_code
416
+ exec(code, run_globals)
417
+ File "/home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py", line 909, in <module>
418
+ {"selfcheck": _selfcheck, "smoke": stage_smoke, "pilot": stage_pilot, "report": main}[sys.argv[1]]()
419
+ File "/home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py", line 603, in stage_pilot
420
+ run(N_PILOT, PLANT_SEED, f"{OUT}/pilot", frozen_cfg())
421
+ File "/home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py", line 581, in run
422
+ f.result(); walls.append(time.time() - ts)
423
+ File "/home/DATA/software/software/python/3.10.0/lib/python3.10/concurrent/futures/_base.py", line 438, in result
424
+ return self.__get_result()
425
+ File "/home/DATA/software/software/python/3.10.0/lib/python3.10/concurrent/futures/_base.py", line 390, in __get_result
426
+ raise self._exception
427
+ AssertionError: snapshot 0 is the certified start
428
+ 40 plants, 37 already saved, cfg {'starts': 16, 'steps': 400, 'ckpt_every': 25, 'lr': 0.1, 'bcap': 10000.0, 'bisect_iters': 10, 'dt': 0.03, 'f_margin': 0.01}, 3 workers, deadline none
429
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
430
+ warnings.warn(
431
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
432
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
433
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
434
+ warnings.warn(
435
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
436
+ warnings.warn(
437
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
438
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
439
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
440
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
441
+ [plant 33 B] linear check False, sigma_p 0.4651, LQG r*/sigma max 1.432 (46 s)
442
+ [plant 33 B] static gain r*/sigma max 0.000 (48 s)
443
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
444
+ num, den = normalize(num, den) # Strips zeros, checks arrays
445
+ [plant 38 B] linear check False, sigma_p 0.06082, LQG r*/sigma max 1.016 (53 s)
446
+ [plant 38 B] static gain r*/sigma max 0.000 (54 s)
447
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
448
+ num, den = normalize(num, den) # Strips zeros, checks arrays
449
+ [plant 30 B] linear check False, sigma_p 0.1162, LQG r*/sigma max 2.368 (67 s)
450
+ [plant 30 B] static gain r*/sigma max 0.000 (69 s)
451
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
452
+ num, den = normalize(num, den) # Strips zeros, checks arrays
453
+ [plant 30 B] scale 0: 16/16 starts, 37 draws, raw stable 0.00, HIFOO stable 0.46, median steps 300 (79 s)
454
+ [plant 38 B] scale 0: 16/16 starts, 105 draws, raw stable 0.00, HIFOO stable 0.35, median steps 300 (90 s)
455
+ [plant 33 B] scale 0: 16/16 starts, 136 draws, raw stable 0.01, HIFOO stable 0.24, median steps 300 (94 s)
456
+ [plant 30 B] scale 1: 16/16 starts, 78 draws, raw stable 0.00, HIFOO stable 0.38, median steps 300 (110 s)
457
+ [plant 38 B] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.34, median steps 300 (243 s)
458
+ [plant 33 B] scale 1: 6/16 starts, 448 draws, raw stable 0.01, HIFOO stable 0.32, median steps 300 (260 s)
459
+ [plant 38 B] done: J_ref s1 0.1558, s2 inf (1907 s)
460
+ saved plant 38 in 1908 s (1908 s elapsed)
461
+ [plant 30 B] done: J_ref s1 0.0198, s2 0.1767 (1937 s)
462
+ saved plant 30 in 1937 s (1937 s elapsed)
463
+ [plant 33 B] done: J_ref s1 0.08704, s2 1.195 (2015 s)
464
+ saved plant 33 in 2015 s (2015 s elapsed)
465
+ total wall 2015 s; saved 3 plants this run
results/icml_r3/X20/logs/smoke.log ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 4 plants, 0 already saved, cfg {'starts': 16, 'steps': 400, 'ckpt_every': 25, 'lr': 0.1, 'bcap': 10000.0, 'bisect_iters': 10, 'dt': 0.03, 'f_margin': 0.01}, 4 workers, deadline none
2
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
3
+ warnings.warn(
4
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
5
+ warnings.warn(
6
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
7
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
8
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
9
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
10
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
11
+ warnings.warn(
12
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/cvxpy/problems/problem.py:1539: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
13
+ warnings.warn(
14
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
15
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
16
+ [plant 01 A] linear check False, sigma_p 0.2855, LQG r*/sigma max 1.141 (9 s)
17
+ [plant 01 A] static gain r*/sigma max 0.000 (11 s)
18
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
19
+ num, den = normalize(num, den) # Strips zeros, checks arrays
20
+ /home/matteo_cercola/scratch/dc_icml/DiffusionControl/nonlinear/experiments/siso_x20_unstable.py:149: DeprecationWarning: Conversion of an array with ndim > 0 to a scalar is deprecated, and will error in future. Ensure you extract a single element from your array before performing this operation. (Deprecated NumPy 1.25.)
21
+ c = min(1.0, ybar ** 2 / float(C1 @ np.linalg.solve(Pv, C1.T))) if np.abs(C1).max() > 0 else 1.0
22
+ [plant 00 A] linear check True, sigma_p 1.018, LQG r*/sigma max 1.000 (21 s)
23
+ [plant 01 A] scale 0: 16/16 starts, 47 draws, raw stable 0.00, HIFOO stable 0.66, median steps 56 (21 s)
24
+ [plant 00 A] static gain r*/sigma max 1.139 (37 s)
25
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
26
+ num, den = normalize(num, den) # Strips zeros, checks arrays
27
+ [plant 03 B] linear check False, sigma_p 0.7948, LQG r*/sigma max 1.027 (38 s)
28
+ [plant 03 B] static gain r*/sigma max 0.000 (40 s)
29
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
30
+ num, den = normalize(num, den) # Strips zeros, checks arrays
31
+ [plant 03 B] scale 0: 16/16 starts, 61 draws, raw stable 0.00, HIFOO stable 0.49, median steps 300 (76 s)
32
+ [plant 02 B] linear check False, sigma_p 0.3974, LQG r*/sigma max 1.128 (85 s)
33
+ [plant 02 B] static gain r*/sigma max 0.000 (87 s)
34
+ /home/matteo_cercola/matte_server-1/DiffusionControl/myenv/lib/python3.10/site-packages/scipy/signal/_lti_conversion.py:74: BadCoefficients: Badly conditioned filter coefficients (numerator): the results may be meaningless
35
+ num, den = normalize(num, den) # Strips zeros, checks arrays
36
+ [plant 00 A] scale 0: 16/16 starts, 71 draws, raw stable 0.03, HIFOO stable 0.69, median steps 61 (108 s)
37
+ [plant 01 A] scale 1: 0/16 starts, 512 draws, raw stable 0.01, HIFOO stable 0.73, median steps 53 (135 s)
38
+ [plant 02 B] scale 0: 16/16 starts, 116 draws, raw stable 0.01, HIFOO stable 0.30, median steps 300 (136 s)
39
+ [plant 02 B] scale 1: 0/16 starts, 512 draws, raw stable 0.00, HIFOO stable 0.30, median steps 300 (340 s)
40
+ [plant 03 B] scale 1: 0/16 starts, 512 draws, raw stable 0.02, HIFOO stable 0.50, median steps 300 (372 s)
41
+ [plant 00 A] scale 1: 0/16 starts, 512 draws, raw stable 0.03, HIFOO stable 0.66, median steps 57 (818 s)
42
+ [plant 01 A] done: J_ref s1 0.3081, s2 inf (1921 s)
43
+ saved plant 01 in 1921 s (1921 s elapsed)
44
+ [plant 03 B] done: J_ref s1 0.2974, s2 inf (2198 s)
45
+ saved plant 03 in 2198 s (2198 s elapsed)
46
+ [plant 02 B] done: J_ref s1 0.2142, s2 inf (2202 s)
47
+ saved plant 02 in 2202 s (2202 s elapsed)
48
+ [plant 00 A] done: J_ref s1 0.4639, s2 inf (2582 s)
49
+ saved plant 00 in 2582 s (2582 s elapsed)
50
+ total wall 2582 s; saved 4 plants this run
51
+ SMOKE OK (no crash, no NaN)
results/icml_r3/X20/pilot/cfg.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "starts": 16,
3
+ "steps": 400,
4
+ "ckpt_every": 25,
5
+ "lr": 0.1,
6
+ "bcap": 10000.0,
7
+ "bisect_iters": 10,
8
+ "dt": 0.03,
9
+ "f_margin": 0.01
10
+ }
results/icml_r3/X20/pilot/plant_00.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e6f215027e1a2e669ccdfa4c25f02102351409c93a1641e9c2bdc4b4d9f8e0b
3
+ size 69155
results/icml_r3/X20/pilot/plant_01.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:37e9eccafef5249d435b794900efa39de88899862a73789ad0fa1f2895cf373c
3
+ size 109563
results/icml_r3/X20/pilot/plant_02.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c4ae527c3a2bf3970b0a139f84f463a5a0fdd66dd3040155401f35d30a7ff903
3
+ size 63915
results/icml_r3/X20/pilot/plant_03.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2156d8e66c2c05a8baf217dfb0067d83b59e6e1da02c9746eef4a9a6cf512c80
3
+ size 68619
results/icml_r3/X20/pilot/plant_04.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:198f3c9c22e37597f5a6285c866c3f8390200088f30b64ddc8669e056b5b9df6
3
+ size 72793
results/icml_r3/X20/pilot/plant_05.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:13e6646adb434b4f329234be153e0fa48e947398a2eeb550da83df2bafd702f4
3
+ size 59655
results/icml_r3/X20/pilot/plant_06.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d10c925261024318f2ce39b6238d8a1601f39d2cd6c0e0c3b60182e34cb52aaf
3
+ size 86046
results/icml_r3/X20/pilot/plant_07.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1fa5f83af10b0b939d20d6acad2f9a7211d82112d8647a578ede3230802fe41c
3
+ size 68296
results/icml_r3/X20/pilot/plant_08.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d068bc1b19a14d551328f99e740af233ee886a9a3a6a2543e5829f9bdc9756f6
3
+ size 72027
results/icml_r3/X20/pilot/plant_09.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:07edd782c2844a3f712aab0de47890e328f9d9d23901a1deb481d793ea5c7923
3
+ size 70384
results/icml_r3/X20/pilot/plant_10.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:22738b38131c688f45b564967a6486c2da32869625bc493ae09b8f22db9c6a12
3
+ size 70551
results/icml_r3/X20/pilot/plant_11.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:13a5175383dcfb2e2e4a2973bf16af05017111cd53d0af38468ab629a98474bd
3
+ size 71180
results/icml_r3/X20/pilot/plant_12.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0631523acf658dd06b9881a8338bafdeede0fb846e71034c45cdfc42489ce716
3
+ size 71132
results/icml_r3/X20/pilot/plant_13.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2368ac0f193d9eba2203b75ff60331607c6f444a01cf2d24ff31bfb2563b318d
3
+ size 67414
results/icml_r3/X20/pilot/plant_14.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52892408ed99e4c0fec461b5796ee5d691f1434f672c725c795844d31c953c22
3
+ size 67672
results/icml_r3/X20/pilot/plant_15.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bb8165b34921c179548c7628daa3f15ffb872b43d8a1dfc6cbf60a6e2666fe67
3
+ size 73052
results/icml_r3/X20/pilot/plant_16.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:45f7aadac19051b6270564d108cec08ee6a3d0a912eb312634586b8aef3089d2
3
+ size 70672
results/icml_r3/X20/pilot/plant_17.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:104deb78ff7a6148f961a847e201a2cc7d0cbc24f386716b1e5035a1f4f75fa1
3
+ size 70725
results/icml_r3/X20/pilot/plant_18.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aaeb7b8dabdebdfc2ff7358ab12eac0c1689433e9de2f2a79329467f26e13894
3
+ size 75800
results/icml_r3/X20/pilot/plant_19.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fede6b645bfc918584200633b056ba5ca3a1bc0c12856505d90fcd4c12981517
3
+ size 68486
results/icml_r3/X20/pilot/plant_20.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a2c7f11afb5ad0bdefc714a03be539497de302173f1da5fa5e55f39b2e070d2
3
+ size 101543
results/icml_r3/X20/pilot/plant_21.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d725f7a760300568344177f42ef503afe24690741393fa414a69d5a59082197
3
+ size 66597
results/icml_r3/X20/pilot/plant_22.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1002706341993c09d05586ce7efd9096cb36567d09b7b3bd36e9c46f01f50119
3
+ size 68549
results/icml_r3/X20/pilot/plant_23.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b90ed73c7dc0729e773a46d83c5986757d7ac7e706c773e20cd7682b137cbe47
3
+ size 67553
results/icml_r3/X20/pilot/plant_24.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1b6fbf922492bc893ff49bf8f931219c62f8f2fd2eabcb79458e596255989d3f
3
+ size 66062
results/icml_r3/X20/pilot/plant_25.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d4aa4509c1ce249cfda075e9dd7009384884dad2e0e1f8d29d45ec9964cb032
3
+ size 67171
results/icml_r3/X20/pilot/plant_26.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6e074fd796167406b5e97ed44eb3008102afa50da0809bafc157d63e10b5853
3
+ size 74394
results/icml_r3/X20/pilot/plant_27.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f690703d61fc6878a9baa66dee6257677726f471efcaab1ca66192f03b6bd9ba
3
+ size 57124
results/icml_r3/X20/pilot/plant_28.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:29144f8090dd2fc5aacd1418d089ce416df39adc003540309aae20b946cf964e
3
+ size 68916
results/icml_r3/X20/pilot/plant_29.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4275df88fec705087ea3a4368c8752075aeff3a7cb44f7e21e9ffed958af3fae
3
+ size 83626
results/icml_r3/X20/pilot/plant_30.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8000ba73dc42ecb13ec9d5f78ab0d0b434783e3742e359f454817f1add04b2ef
3
+ size 106171
results/icml_r3/X20/pilot/plant_31.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe2eebb2f74a2d2d57d5674436a8660b21f7c397f8f5c174f3dd0966824e19f7
3
+ size 35715
results/icml_r3/X20/pilot/plant_32.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:75f6727fd234b9ce66a3b32100cd476e0ab33af69e4501164951e76e627e863d
3
+ size 52789
results/icml_r3/X20/pilot/plant_33.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0441ee0f5094c426c1c093ef7f108ba2f5da1aa2261009e01d89eb55f28c7d85
3
+ size 78393
results/icml_r3/X20/pilot/plant_34.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4a27416c70d7217e3828d666d9d909e96ddd1f48bebb8530dfa29ef821d326f
3
+ size 59875
results/icml_r3/X20/pilot/plant_35.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba335e1411729e38ecc23e7af8ad9def83a72c84c440dba4db279e05d6d076b5
3
+ size 62727
results/icml_r3/X20/pilot/plant_36.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:648f56e9b0653cad79d123b490c135b19c3af1464e4db7ad8190315a331f069d
3
+ size 66590
results/icml_r3/X20/pilot/plant_37.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9b35053f95a891dcf1e2b0253d5699208d21e4b313d9078c9ea1934c68345bac
3
+ size 51841
results/icml_r3/X20/pilot/plant_38.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:44b8ed71b97f21fdb5c76ef00a51c2a2abbbee24e176fe7724883f5beaf36285
3
+ size 57873
results/icml_r3/X20/pilot/plant_39.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c4bdfc99d2296e9a586a036097d966c310daab42a6ee2cdcb54b350700c36330
3
+ size 63477
results/icml_r3/X20/smoke_lr0.1/cfg.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "starts": 16,
3
+ "steps": 400,
4
+ "ckpt_every": 25,
5
+ "lr": 0.1,
6
+ "bcap": 10000.0,
7
+ "bisect_iters": 10,
8
+ "dt": 0.03,
9
+ "f_margin": 0.01
10
+ }
results/icml_r3/X20/smoke_lr0.1/plant_00.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:85d67d6031ca3a1e60c212662fdb3f65e6618ad873e3737200bb11f68b0dae31
3
+ size 73882
results/icml_r3/X20/smoke_lr0.1/plant_01.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81b514f679c5b59b312ea48d6189bfaf446a015dd738ad2f2752b64e4fb8985a
3
+ size 68795
results/icml_r3/X20/smoke_lr0.1/plant_02.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5606fd0a3461eadd93bcc396b9ba447dc342c4629d17d7ea0cd39496c53eabad
3
+ size 54998
results/icml_r3/X20/smoke_lr0.1/plant_03.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:73fe21804ef827611daa2e6a106eab6b2673cda7e06a728841505038463515d4
3
+ size 65267