Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 5′, first display — at x∗x^*x∗ supported on III, the worst cases over U~\tilde{\mathcal U}U~ and U\mathcal UU, for AAA and A~\tilde AA~, coincide

Proved
RobustRegLasso.Sparsity.theorem_5_prime_first_display

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

p2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1regressionrobust-optimizationsparsity

Let a1,…,am,b∈Rna_1,\dots,a_m, b \in \mathbb R^na1​,…,am​,b∈Rn, radii ci≥0c_i \ge 0ci​≥0, numbers ljl_jlj​, and an index set I⊆{1,…,m}I \subseteq \{1,\dots,m\}I⊆{1,…,m}. Let A~\tilde AA~ have columns a~i\tilde a_ia~i​ with a~i=ai\tilde a_i = a_ia~i​=ai​ for i∈Ii \in Ii∈I and ∥a~j−aj∥2≤lj\|\tilde a_j - a_j\|_2 \le l_j∥a~j​−aj​∥2​≤lj​ for j∉Ij \notin Ij∈/I. Let x∗∈Rmx^* \in \mathbb R^mx∗∈Rm satisfy xj∗=0x^*_j = 0xj∗​=0 for all j∉Ij \notin Ij∈/I. With U\mathcal UU the feature-wise set with radii ccc and U~\tilde{\mathcal U}U~ the enlarged set of Theorem 5′,

max⁡ΔA∈U~∥b−(A+ΔA)x∗∥2=max⁡ΔA∈U∥b−(A+ΔA)x∗∥2=max⁡ΔA∈U∥b−(A~+ΔA)x∗∥2,\max_{\Delta A\in\tilde{\mathcal U}} \|b-(A+\Delta A)x^*\|_2 = \max_{\Delta A\in\mathcal U} \|b-(A+\Delta A)x^*\|_2 = \max_{\Delta A\in\mathcal U} \|b-(\tilde A+\Delta A)x^*\|_2,ΔA∈U~max​∥b−(A+ΔA)x∗∥2​=ΔA∈Umax​∥b−(A+ΔA)x∗∥2​=ΔA∈Umax​∥b−(A~+ΔA)x∗∥2​,

and moreover

max⁡ΔA∈U~∥b−(A~+ΔA)x∗∥2=max⁡ΔA∈U∥b−(A+ΔA)x∗∥2.\max_{\Delta A\in\tilde{\mathcal U}} \|b-(\tilde A+\Delta A)x^*\|_2 = \max_{\Delta A\in\mathcal U} \|b-(A+\Delta A)x^*\|_2.ΔA∈U~max​∥b−(A~+ΔA)x∗∥2​=ΔA∈Umax​∥b−(A+ΔA)x∗∥2​.

No optimality of x∗x^*x∗ is assumed.

This is the first step of the proof of Theorem 5′: since x∗x^*x∗ puts no weight on the features outside III, the columns j∉Ij \notin Ij∈/I of both A~\tilde AA~ and ΔA\Delta AΔA have no effect on the residual, so changing them (or the bounds on them) does not change the worst case at x∗x^*x∗.

Formalization Note The page displays the first three quantities; the proof of Theorem 5′ then uses the fourth equality (A~\tilde AA~ and U~\tilde{\mathcal U}U~ together), which holds for the same reason, so it is stated as an additional conjunct. The maxima are suprema in EReal. The hypothesis ∥a~j−aj∥2≤lj\|\tilde a_j - a_j\|_2 \le l_j∥a~j​−aj​∥2​≤lj​ is used only to force cj+lj≥0c_j + l_j \ge 0cj​+lj​≥0 (so U~\tilde{\mathcal U}U~ is not empty); the radii ci≥0c_i \ge 0ci​≥0 are the paper's standing convention.

Preamble
import Mathlib
import Definitions.Def_RobustRegLasso_Sparsity_Basic
Formal statement
namespace RobustRegLasso.Sparsity

theorem theorem_5_prime_first_display {n m : ℕ} (a aTilde : Fin m → EuclideanSpace ℝ (Fin n))
    (b : EuclideanSpace ℝ (Fin n)) (c l : Fin m → ℝ) (hc : ∀ i, 0 ≤ c i)
    (I : Finset (Fin m)) (xStar : Fin m → ℝ) (hx : ∀ j, j ∉ I → xStar j = 0)
    (hl : ∀ j, j ∉ I → ‖aTilde j - a j‖ ≤ l j) (hI : ∀ i, i ∈ I → aTilde i = a i) :
    robustObjective a b (enlargedSet c l I) xStar = robustObjective a b (RobustRegLasso.FeatureWise.uncertaintySet c) xStar ∧
    robustObjective a b (RobustRegLasso.FeatureWise.uncertaintySet c) xStar =
      robustObjective aTilde b (RobustRegLasso.FeatureWise.uncertaintySet c) xStar ∧
    robustObjective aTilde b (enlargedSet c l I) xStar =
      robustObjective a b (RobustRegLasso.FeatureWise.uncertaintySet c) xStar := by sorry

end RobustRegLasso.Sparsity
Source
Xu, Caramanis, Mannor, Robust Regression and Lasso, arXiv:0811.1790v1, p. 9, proof of Theorem 5′, first display
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