DiffusionControl / results /pendulum /constrained_smoke.log
matCercola18's picture
Add files using upload-large-folder tool
929e74d verified
Raw History Blame Contribute Delete
1.33 kB
/home/cercola/matte_server/DiffusionControl/nonlinear/data_generation/pendulum_lmi.py:65: UserWarning: Solution may be inaccurate. Try another solver, adjusting the solver settings, or solve with verbose=True for more information.
prob.solve(solver=cp.SCS, verbose=False)
=== CONSTRAINED OC ===
Constrained optimal control on the pendulum (held-out N=10, cap θ≤0.24)
PRIOR (quadratic-trained, constraint-unaware):
satisfies |θ|≤0.24: 0.30 median peak |θ| = 0.272 median effort 18.557
STEERED toward min ∫u² s.t. |θ|≤0.24 (barrier λ=10000.0):
satisfies: 0.30 median peak |θ| = 0.264 median effort 18.827 certified 1.00
effort REGRET vs per-plant constrained oracle: median +0.5% (n=3)
effort-vs-cap constrained-optimal frontier (steered):
θ_cap mean effort (satisfying) satisfied
0.20 nan 0.00
0.24 7.831 0.30
0.28 21.301 1.00
0.32 21.227 1.00
INTERPRETATION: a non-quadratic, hard-state-constrained optimal-control problem solved
through the differentiable non-linear pendulum: the constraint-unaware prior violates the
cap; steering enforces it at minimum effort (near the per-plant constrained optimum),
certified; tighter caps cost more effort (the constrained-optimal frontier).