Proposition 5.2(i), p. 18 — if X_β(s) ≠ ∅ for all s and h is proper, the image function (Ch) is proper
ProvedNonconvexDRS.ADMM.prop_5_2_iimage-functionnonconvex-optimizationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1proper-function
Let be proper, let , and suppose that for some the set
is nonempty for every . Then the image function is proper: it never equals , and it is finite somewhere.
A priori an image function can take the value ; this proposition is what makes the proximal mapping of meaningful in the ADMM–DRS equivalence.
Formalization Note The page does not state that is proper; it is added because for every is all of while is not proper. Properness here includes everywhere.
Preamble
import Mathlib import Definitions.Def_NonconvexSplitting_Shared_LimitingSubdiff import Definitions.Def_NonconvexDRS_ADMM_Setting open NonconvexSplitting.Shared open Filter Topology open scoped InnerProductSpace
Formal statement
namespace NonconvexDRS.ADMM
/-- Proposition 5.2(i), p. 18: if `X_β(s)` is nonempty for every `s` (for some `β > 0`) and `h`
is proper, the image function `(Ch)` is proper. -/
theorem prop_5_2_i {n p : ℕ} (h : EuclideanSpace ℝ (Fin n) → EReal)
(C : EuclideanSpace ℝ (Fin n) →L[ℝ] EuclideanSpace ℝ (Fin p))
(β : ℝ) (hβ : 0 < β) (hh : NonconvexDRS.Tight.IsProper h)
(hX : ∀ s, (Xbeta h C β s).Nonempty) :
NonconvexDRS.Tight.IsProper (imageFn C h) := by sorry
end NonconvexDRS.ADMM
Source
Themelis & Patrinos, Douglas-Rachford splitting and ADMM for nonconvex optimization: tight convergence results, arXiv:1709.05747v4, pp. 17–18, Proposition 5.2(i)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.