Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 2.4 (a), p. 6 — finitely many points z_k carry every subdifferential of f and of f̂_N

Proved
SAARate.Polyhedral.lemma_2_4_a

by mikedeng1 · Oct 7, 2026 · Mathlib 0df444a (Lean v4.33.1)

p2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1polyhedralsample-average-approximationstochastic-programmingsubdifferential

Let Ω\OmegaΩ be finite, let PPP be a probability measure on Ω\OmegaΩ, and suppose every h(⋅,ω)h(\cdot,\omega)h(⋅,ω), ω∈Ω\omega\in\Omegaω∈Ω, is piecewise linear and convex on Rm\mathbb R^mRm. Write f(x)=EPh(x,ω)f(x)=\mathbb E_P h(x,\omega)f(x)=EP​h(x,ω) and, for a sample path s=(ω1,ω2,… )s=(\omega^1,\omega^2,\dots)s=(ω1,ω2,…) in Ω\OmegaΩ, f^N(x)=N−1∑j=1Nh(x,ωj)\hat f_N(x)=N^{-1}\sum_{j=1}^N h(x,\omega^j)f^​N​(x)=N−1∑j=1N​h(x,ωj). Let ∂g(x)\partial g(x)∂g(x) be the subdifferential of a convex function ggg at xxx.

Then there are finitely many points z1,…,zr∈Rmz_1,\dots,z_r\in\mathbb R^mz1​,…,zr​∈Rm, chosen independently of the sample, such that for every x∈Rmx\in\mathbb R^mx∈Rm there is k∈{1,…,r}k\in\{1,\dots,r\}k∈{1,…,r} with

∂f(x)=∂f(zk)and∂f^N(x)=∂f^N(zk)  for every sample path and every N.\partial f(x)=\partial f(z_k)\qquad\text{and}\qquad \partial\hat f_N(x)=\partial\hat f_N(z_k)\ \text{ for every sample path and every } N.∂f(x)=∂f(zk​)and∂f^​N​(x)=∂f^​N​(zk​)  for every sample path and every N.

This reduces questions about all subdifferentials of the true and the sample average functions to finitely many points; it is the source of the uniform convergence in Lemma 2.4 (b).

Formalization Note The statement assumes that every scenario has positive probability, P{ω}>0P\{\omega\}>0P{ω}>0 for all ω∈Ω\omega\in\Omegaω∈Ω, which is how p. 3 takes Ω={ω1,…,ωK}\Omega=\{\omega_1,\dots,\omega_K\}Ω={ω1​,…,ωK​} to be the support of PPP. Without it the claim fails for sample paths visiting a null scenario whose h(⋅,ω)h(\cdot,\omega)h(⋅,ω) has a kink where fff is affine. The subdifferential is the published ShorNonsmooth.Subdiff.subdifferential with domain Rm\mathbb R^mRm.

Preamble
import Mathlib
import Definitions.Def_SAARate_Polyhedral_Setting
import Definitions.Def_ShorNonsmooth_Subdiff_Subdifferential

open MeasureTheory ProbabilityTheory ShorNonsmooth.Subdiff
Formal statement
namespace SAARate.Polyhedral

/-- Lemma 2.4 (a), p. 6: under (i) `Ω` finite and (ii) every `h(·, ω)` piecewise linear and
convex, there are finitely many points `z_1, …, z_r`, independent of the sample, such that every
`x` has a `k` with `∂f(x) = ∂f(z_k)` and `∂f̂_N(x) = ∂f̂_N(z_k)` for every realization of the
sample. `hP` (every scenario has positive mass, p. 3) is a disclosed pin. -/
theorem lemma_2_4_a {m : ℕ} {Ω : Type*} [Fintype Ω] [MeasurableSpace Ω]
    [MeasurableSingletonClass Ω] (P : Measure Ω) [IsProbabilityMeasure P]
    (hP : ∀ ω, P {ω} ≠ 0)
    (h : SAARate.Sharp.E m → Ω → ℝ) (hpl : ∀ ω, IsPLConvex (fun x => h x ω)) :
    ∃ (r : ℕ) (z : Fin r → SAARate.Sharp.E m), ∀ x : SAARate.Sharp.E m, ∃ k : Fin r,
      subdifferential Set.univ (SAARate.Sharp.expectedObj P h) x =
          subdifferential Set.univ (SAARate.Sharp.expectedObj P h) (z k) ∧
        ∀ (s : ℕ → Ω) (N : ℕ),
          subdifferential Set.univ (SAARate.Sharp.saaObj h s N) x =
            subdifferential Set.univ (SAARate.Sharp.saaObj h s N) (z k) := by sorry

end SAARate.Polyhedral
Source
Shapiro & Homem-de-Mello, On Rate of Convergence of Optimal Solutions of Monte Carlo Approximations of Stochastic Programs, preprint (SPEPS copy, edoc.hu-berlin.de), p. 6, Lemma 2.4 (a); p. 3, (2.1) for the finite support
Human review
  • Endorsed by Shuze Chen · Oct 7, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 7, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me