Convex rate sublevels for finite-feature geometry design
ProvedPGDRevision20261113.finite_feature_rate_sublevelsOn , let have curvatures and , with bounded. Let be fixed continuously differentiable features satisfying . For and , set
Take the twice integrated curvature as the generator, and use the real implicit update for . Every update exists uniquely. Put
For , define and . Then is equivalent to the following constraints for every real state:
Thus each rate sublevel is convex, including its intersection with any convex coefficient domain. These are constraints on the full state domain; a finite grid does not by itself certify them.
Related results. Exact profile performance (Theorem 4.1) · Geometry classification (Theorem 5.1) · Parameter localization (Theorem 7.1)
Further examples and certificates. finite feature nonuniqueness counterexample · loss matched one step
import Definitions.Def_PGDRevision20261113_Profile open PGDRevision20261113
theorem PGDRevision20261113.finite_feature_rate_sublevels : FiniteFeatureDesignClaim := by sorry