Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 2.5 — the minimax MMSE problem (2) over the Gaussian Wasserstein ball equals program (5); ψ⋆ is affine and N(µ, S⋆) is least favorable

Proved
WassKF.Reform.theorem_2_5

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

distributionally-robust-optimizationmmse-estimationp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1semidefinite-programming

Let z=[x;y]z = [x; y]z=[x;y] with x∈Rnx \in \mathbb R^nx∈Rn, y∈Rmy \in \mathbb R^my∈Rm and d=n+md = n + md=n+m. Let P=Nd(μ,Σ)\mathbb P = \mathcal N_d(\mu, \Sigma)P=Nd​(μ,Σ) with μ=[μx;μy]\mu = [\mu_x; \mu_y]μ=[μx​;μy​] and Σ≻0\Sigma \succ 0Σ≻0, let σ‾=λmin⁡(Σ)>0\underline\sigma = \lambda_{\min}(\Sigma) > 0σ​=λmin​(Σ)>0, let ρ≥0\rho \ge 0ρ≥0, and let P={Q∈Nd:W2(Q,P)≤ρ}\mathcal P = \{\mathbb Q \in \mathcal N_d : W_2(\mathbb Q, \mathbb P) \le \rho\}P={Q∈Nd​:W2​(Q,P)≤ρ} be the Wasserstein ambiguity set (3). Let L\mathcal LL be the family of all measurable functions Rm→Rn\mathbb R^m \to \mathbb R^nRm→Rn. Consider the finite program

sup⁡Tr⁡[Sxx−SxySyy−1Syx]s.t.S=[SxxSxySyxSyy]∈S+d,Sxx∈S+n,Syy∈S+m,Tr⁡[S+Σ−2(Σ1/2SΣ1/2)1/2]≤ρ2,S⪰σ‾Id.(5)\begin{aligned} \sup \quad & \operatorname{Tr}\bigl[S_{xx} - S_{xy} S_{yy}^{-1} S_{yx}\bigr] \\ \text{s.t.} \quad & S = \begin{bmatrix} S_{xx} & S_{xy} \\ S_{yx} & S_{yy}\end{bmatrix} \in \mathbb S^d_+,\quad S_{xx} \in \mathbb S^n_+,\quad S_{yy} \in \mathbb S^m_+, \\ & \operatorname{Tr}\Bigl[S + \Sigma - 2\bigl(\Sigma^{1/2} S \Sigma^{1/2}\bigr)^{1/2}\Bigr] \le \rho^2, \quad S \succeq \underline\sigma I_d . \end{aligned} \tag{5}sups.t.​Tr[Sxx​−Sxy​Syy−1​Syx​]S=[Sxx​Syx​​Sxy​Syy​​]∈S+d​,Sxx​∈S+n​,Syy​∈S+m​,Tr[S+Σ−2(Σ1/2SΣ1/2)1/2]≤ρ2,S⪰σ​Id​.​(5)

Then the minimax problem (2) is equivalent to (5) in the following sense.

  1. The optimal value of (2), inf⁡ψ∈Lsup⁡Q∈PEQ[∥x−ψ(y)∥2]\inf_{\psi \in \mathcal L} \sup_{\mathbb Q \in \mathcal P} \mathbb E^{\mathbb Q}[\|x - \psi(y)\|^2]infψ∈L​supQ∈P​EQ[∥x−ψ(y)∥2], equals the optimal value of (5).
  2. If S⋆S^\starS⋆ is optimal in (5), then the affine function
ψ⋆(y)=Sxy⋆(Syy⋆)−1(y−μy)+μx\psi^\star(y) = S^\star_{xy} (S^\star_{yy})^{-1} (y - \mu_y) + \mu_xψ⋆(y)=Sxy⋆​(Syy⋆​)−1(y−μy​)+μx​

is measurable and attains the infimum in (2): it is a distributionally robust minimum mean square error estimator. 3. For the same S⋆S^\starS⋆, the normal distribution Q⋆=Nd(μ,S⋆)\mathbb Q^\star = \mathcal N_d(\mu, S^\star)Q⋆=Nd​(μ,S⋆) belongs to P\mathcal PP and attains the supremum in nature's problem sup⁡Q∈Pinf⁡ψ∈LEQ[∥x−ψ(y)∥2]\sup_{\mathbb Q \in \mathcal P} \inf_{\psi \in \mathcal L} \mathbb E^{\mathbb Q}[\|x - \psi(y)\|^2]supQ∈P​infψ∈L​EQ[∥x−ψ(y)∥2]: it is the least favorable prior.

The theorem reduces an infinite-dimensional minimax problem over a nonconvex set of distributions to a finite program over covariance matrices, and recovers both the robust estimator and the least favorable prior from its optimal solution.

Formalization Note "Equivalent" is not defined on the page; it is read as items 1–3, which cover both sentences of the theorem. Values of (2) and of nature's problem are in [0,∞][0,\infty][0,∞] (lower Lebesgue integrals); the value of (5) is the published sdpValue in EReal, compared through the coercion [0,∞]→[0,\infty] \to[0,∞]→ EReal. Program (5) is the published sdpFeasibleSet ρ Σ σ with objective sdpObjective; Sxy=Syx⊤S_{xy} = S_{yx}^\topSxy​=Syx⊤​ holds by symmetry of SSS. σ‾\underline\sigmaσ​ is passed as the least eigenvalue of Σ\SigmaΣ. The word "convex" in "finite convex program" is not part of the formal statement. Neither existence nor uniqueness of an optimal S⋆S^\starS⋆ is asserted: items 2 and 3 hold for every optimal S⋆S^\starS⋆. On the feasible set S⪰σ‾IdS \succeq \underline\sigma I_dS⪰σ​Id​ with σ‾>0\underline\sigma > 0σ​>0, so SyyS_{yy}Syy​ is invertible and the matrix inverse has no junk value. ψ⋆\psi^\starψ⋆ is the published affineEstimator.

Preamble
import Mathlib
import Definitions.Def_WassersteinDRO_Shrinkage_sdpFeasibleSet
import Definitions.Def_WassersteinDRO_Shrinkage_sdpObjective
import Definitions.Def_WassersteinDRO_Shrinkage_sdpValue
import Definitions.Def_WassersteinDRO_Shrinkage_affineEstimator
import Definitions.Def_WassKF_Reform_GaussianBall
import Definitions.Def_WassKF_Reform_expectedLoss
import Definitions.Def_WassKF_Reform_minimaxValue
import Definitions.Def_WassKF_Reform_maximinValue

open MeasureTheory ProbabilityTheory
Formal statement
namespace WassKF.Reform

/-- Theorem 2.5 (Tractable reformulation), Shafieezadeh-Abadeh et al., arXiv:1809.08830v3, p. 3.
Let `ℙ = 𝒩_d(μ, Sig)` with `Sig ≻ 0`, `ρ ≥ 0`, and `σ = λ_min(Sig)`. Then
1. the optimal value of the minimax problem (2) over the ambiguity set (3) equals the optimal value
   of program (5), `sup Tr[S_xx − S_xy S_yy⁻¹ S_yx]` over `sdpFeasibleSet ρ Sig σ`;
2. for every optimal solution `S⋆` of (5), the affine estimator
   `ψ⋆(y) = S⋆_xy (S⋆_yy)⁻¹ (y − μ_y) + μ_x` is measurable and attains the outer infimum of (2);
3. for the same `S⋆`, `Q⋆ = 𝒩_d(μ, S⋆)` lies in the ambiguity set and attains the outer supremum
   of nature's problem `sup_{Q ∈ 𝒫} inf_{ψ ∈ ℒ} E^Q[‖x − ψ(y)‖²]` (it is the least favorable prior).
Here `μ_x = (μ_i)_{i ∈ inl}` and `μ_y = (μ_j)_{j ∈ inr}`. -/
theorem theorem_2_5 {n m : ℕ} (μ : EuclideanSpace ℝ (Fin n ⊕ Fin m))
    (Sig : Matrix (Fin n ⊕ Fin m) (Fin n ⊕ Fin m) ℝ) (hSig : Sig.PosDef)
    (ρ : ℝ) (hρ : 0 ≤ ρ) (σ : ℝ) (hσ : IsLeast (Set.range hSig.isHermitian.eigenvalues) σ) :
    ((minimaxValue μ Sig ρ : ENNReal) : EReal) = WassersteinDRO.Shrinkage.sdpValue ρ Sig σ ∧
    ∀ Sstar ∈ WassersteinDRO.Shrinkage.sdpFeasibleSet ρ Sig σ,
      IsMaxOn WassersteinDRO.Shrinkage.sdpObjective
          (WassersteinDRO.Shrinkage.sdpFeasibleSet ρ Sig σ) Sstar →
        (Measurable (WassersteinDRO.Shrinkage.affineEstimator Sstar.toBlocks₁₂ Sstar.toBlocks₂₂
            ((EuclideanSpace.equiv (Fin n) ℝ).symm (fun i => μ (Sum.inl i)))
            ((EuclideanSpace.equiv (Fin m) ℝ).symm (fun j => μ (Sum.inr j)))) ∧
          (⨆ (Q : Measure (EuclideanSpace ℝ (Fin n ⊕ Fin m))) (_ : Q ∈ GaussianBall μ Sig ρ),
              expectedLoss (WassersteinDRO.Shrinkage.affineEstimator Sstar.toBlocks₁₂
                Sstar.toBlocks₂₂
                ((EuclideanSpace.equiv (Fin n) ℝ).symm (fun i => μ (Sum.inl i)))
                ((EuclideanSpace.equiv (Fin m) ℝ).symm (fun j => μ (Sum.inr j)))) Q) =
            minimaxValue μ Sig ρ) ∧
        (multivariateGaussian μ Sstar ∈ GaussianBall μ Sig ρ ∧
          (⨅ (ψ : EuclideanSpace ℝ (Fin m) → EuclideanSpace ℝ (Fin n)) (_ : Measurable ψ),
              expectedLoss ψ (multivariateGaussian μ Sstar)) =
            maximinValue μ Sig ρ) := by sorry

end WassKF.Reform
Source
Shafieezadeh-Abadeh, Nguyen, Kuhn, Mohajerin Esfahani, Wasserstein Distributionally Robust Kalman Filtering, arXiv:1809.08830v3, p. 3, Theorem 2.5, (5)
Human review
  • Endorsed by Shuze Chen · Oct 8, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 8, 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