Exact performance for prescribed Bregman profiles
ProvedPGDRevision20261113.exact_profile_reductionLet be a nonempty open interval, and let be twice continuously differentiable real functions on . Their scalar curvature profiles satisfy
Fix and assume that maps onto . For each real tilt , let , and define every update by the unique solution of
The profiles, geometry, and inverse stepsize are fixed across updates; the tilt and initial state vary in the performance criterion. Set
and, for each integer ,
Then the exact performance at every finite horizon is
The equality is understood in the extended nonnegative reals when . Consequently, minimizing over any admissible family with fixed parameters minimizes the performance criterion for every finite horizon.
This result reduces performance analysis for prescribed profiles to one scalar uniform norm. It concerns exact implicit updates and unrestricted linear tilts.
Source: Exact Residual Bounds and Optimal Geometry Selection for Proximal Gradient Methods, revised reading edition, 13 November 2026, Theorem 4.1, page 10; assumptions and criterion in Section 4.
Related results. Separable profiles (Corollary 4.2) · Finite-feature design (Proposition 4.3) · Geometry classification (Theorem 5.1) · Parameter localization (Theorem 7.1) · Entropy-mixture specialization
Further examples and certificates. positive measure entropy extension · loss matched one step
import Definitions.Def_PGDRevision20261113_Profile open PGDRevision20261113
theorem PGDRevision20261113.exact_profile_reduction : ExactProfileReductionClaim := by sorry