Eq. (59) — for each , a.s. and converges
ProvedIncProx.Randomized.eq59In the setting of the randomized incremental methods (42)–(44) (problem (5)–(6) with nonempty closed convex and convex; uniform on and independent of the past history ; Assumption 3(b) or 4(b) with constant ; positive stepsizes ), assume
Then for each there is a set of probability 1 such that for every sample path in
and the sequence converges.
The probability-one set depends on : the statement is "for each , almost surely", not "almost surely, for all ". The proof of Proposition 9 passes from the first to the second through a countable dense subset of .
Formalization Note is written as . The series is stated as Summable of a real sequence; its terms are nonnegative from on (every with lies in ), so this is the paper's . The hypotheses and of Proposition 9 are not needed for (59) and are not assumed. Components are indexed .
import Mathlib import Definitions.Def_IncProx_Randomized_Basic
namespace IncProx.Randomized
open Filter Topology MeasureTheory
/-- Eq. (59) (proof of Proposition 9, p. 21): if Σ α_k² < ∞, then for each x* ∈ X*, with
probability 1, Σ_k (2α_k/m)(F(x_k) − F*) < ∞ and ‖x_k − x*‖ converges. -/
theorem eq59 {Ω : Type*} [MeasurableSpace Ω] {n m : ℕ} [NeZero m]
(μ : Measure Ω) [IsProbabilityMeasure μ] (P : Problem n m) (hP : P.Standing) (A : Alg)
(α : ℕ → ℝ) (hαpos : ∀ k, 0 < α k) (ω : ℕ → Ω → Fin m) (x : ℕ → Ω → Vec n)
(zc gFc gHc xc : ℕ → Fin m → Ω → Vec n) (x₀ : Vec n) (c : ℝ)
(hR : RandHyp μ P A α ω x zc gFc gHc xc x₀ c)
(hα2 : Summable fun k => α k ^ 2) (xs : Vec n) (hxs : xs ∈ P.Xstar) :
∀ᵐ s ∂μ, Summable (fun k => 2 * α k / m * (P.F (x k s) - P.F xs)) ∧
∃ L : ℝ, Tendsto (fun k => ‖x k s - xs‖) atTop (𝓝 L) := by sorry
end IncProx.Randomized
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.