Robustness of the limiting subdifferential (Eq. (3))
ProvedNonconvexSplitting.ProxADMM.limitingSubdiff_robustp2o-batch-p200bp2o-gran-per-chapterp2o-plan-paperp2o-v1variational-analysis
Let be proper, and let with . Suppose with , , and for every , where is the limiting subdifferential. Then
Equivalently, : the graph of is closed along -attentive sequences.
This is the property used to pass to the limit in the optimality conditions of the ADMM subproblems.
Preamble
import Mathlib import Definitions.Def_NonconvexSplitting_Shared_LimitingSubdiff import Definitions.Def_NonconvexSplitting_Shared_StandingAssumptions open NonconvexSplitting.Shared open Filter Topology open scoped InnerProductSpace
Formal statement
namespace NonconvexSplitting.ProxADMM
/-- Eq. (3) of Li–Pong (p. 3): robustness of the limiting subdifferential. -/
theorem limitingSubdiff_robust {m : ℕ} (f : EuclideanSpace ℝ (Fin m) → EReal)
(hf : IsProperFn f) (u v : EuclideanSpace ℝ (Fin m)) (hu : f u ≠ ⊤)
(us vs : ℕ → EuclideanSpace ℝ (Fin m))
(hus : Tendsto us atTop (𝓝 u)) (hfus : Tendsto (fun t => f (us t)) atTop (𝓝 (f u)))
(hvs : Tendsto vs atTop (𝓝 v)) (hmem : ∀ t, vs t ∈ LimitingSubdiff f (us t)) :
v ∈ LimitingSubdiff f u := by sorry
end NonconvexSplitting.ProxADMM
Source
Li & Pong, Global Convergence of Splitting Methods for Nonconvex Composite Optimization, arXiv:1407.0753v6, p. 3, Eq. (3)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.