Lemma 3.3(i), p. 541 — PMD objective values are nonincreasing
ProvedChenTeboullePMD.Rate.lemma_3_3_ibregman-functionconvex-optimizationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1proximal-point
Under Lemma 3.2's standing assumptions, the PMD objective values do not increase: whenever the preceding iterate belongs to the effective domain ,
Monotonicity is used to identify the limiting objective value in Theorem 3.4.
Formalization Note The guard is needed only at the first step: may lie outside , while every produced iterate lies in . This real inequality represents the nontrivial part of the page's extended-valued monotonicity claim. The ambient space is a finite-dimensional real normed space and the run is explicit.
Preamble
import Mathlib import Definitions.Def_BeckTeboulleMD_EMDA_Setting import Definitions.Def_ChenTeboullePMD_Rate_Setting
Formal statement
namespace ChenTeboullePMD.Rate
/-- Lemma 3.3(i), p. 541. -/
theorem lemma_3_3_i {E : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E]
[FiniteDimensional ℝ E]
(S : Set E) (ψ : E → ℝ) (hψ : IsBregmanFunction S ψ)
(C : Set E) (hCne : C.Nonempty) (hCconv : Convex ℝ C)
(f : E → ℝ) (hf : ConvexOn ℝ C f)
(hlsc : LowerSemicontinuous (extendTop C f))
(hri : intrinsicInterior ℝ C ⊆ S)
(lam : ℕ → ℝ) (hlam : ∀ k, 1 ≤ k → 0 < lam k)
(x : ℕ → E) (hrun : IsPMDRun S C ψ f lam x) :
∀ k : ℕ, x k ∈ C → f (x (k + 1)) ≤ f (x k) := by sorry
end ChenTeboullePMD.Rate
Source
Chen & Teboulle, Convergence analysis of a proximal-like minimization algorithm using Bregman functions, SIAM J. Optim. 3 (1993), p. 541, Lemma 3.3(i)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.