§3.1, pp. 436–437 — a proper convex ρ has an algebraic subgradient at every point of int(dom ρ)
ProvedConvexRiskFn.Cont.exists_algSubgradientalgebraic-subgradientconvex-analysishahn-banachp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1risk-function
Let be a real normed space and a proper function satisfying (A1). If , then there is a linear functional (not necessarily continuous) with
i.e. is an algebraic subgradient of at in the sense of (3.1).
The paper uses this as the first step of Proposition 3.1; continuity of is what the lattice structure adds afterwards.
Formalization Note The functional is an arbitrary -linear map E →ₗ[ℝ] ℝ; no continuity is asserted.
Preamble
import Mathlib import Definitions.Def_ConvexRiskFn_Cont_Setting open Filter Topology
Formal statement
namespace ConvexRiskFn.Cont
theorem exists_algSubgradient {E : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E]
(ρ : E → EReal) (hρ : IsProper ρ) (h1 : ConvexRiskFn.Dual.A1 ρ) (Xbar : E)
(hXbar : Xbar ∈ interior (ConvexRiskFn.Dual.dom ρ)) :
∃ l : E →ₗ[ℝ] ℝ, IsAlgSubgradient ρ Xbar l := by sorry
end ConvexRiskFn.Cont
Source
Ruszczyński, Shapiro, Optimization of convex risk functions, Math. Oper. Res. 31 (2006), pp. 436–437, §3.1: "ρ always possesses an algebraic subgradient at any point X̄ ∈ int(dom ρ) (cf. Levin [11, Lemma 1.1])" and the Hahn–Banach sentence on p. 437
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.