Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

(47) — the minimum of DiD_iDi​ when ∣αTXi∣−ηi≥2λ0λ2|\alpha^TX_i|-\eta_i\ge 2\sqrt{\lambda_0\lambda_2}∣αTXi​∣−ηi​≥2λ0​λ2​​

Proved
L0BnB.Duality.eq_47

by mikedeng1 · Oct 8, 2026 · Mathlib 0df444a (Lean v4.33.1)

lagrangian-dualityp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1reverse-huber

Let λ0,λ2>0\lambda_0,\lambda_2>0λ0​,λ2​>0, let a∈Ra\in\mathbb Ra∈R (standing for αTXi\alpha^TX_iαTXi​) and η≥0\eta\ge 0η≥0, and let D(b)=ψ1(b;λ0,λ2)+ab+η∣b∣D(b)=\psi_1(b;\lambda_0,\lambda_2)+ab+\eta|b|D(b)=ψ1​(b;λ0​,λ2​)+ab+η∣b∣ for b∈Rb\in\mathbb Rb∈R. If ∣a∣−η≥2λ0λ2|a|-\eta\ge 2\sqrt{\lambda_0\lambda_2}∣a∣−η≥2λ0​λ2​​, then the minimum of DDD over R\mathbb RR is attained and

min⁡b∈RD(b)=−14λ2(∣a∣−η)2+λ0.\min_{b\in\mathbb R} D(b) = -\frac{1}{4\lambda_2}\big(|a|-\eta\big)^2+\lambda_0 .b∈Rmin​D(b)=−4λ2​1​(∣a∣−η)2+λ0​.

This is the value of the inner minimization of the Lagrangian in the proof of Theorem 2 when the correlation ∣αTXi∣|\alpha^TX_i|∣αTXi​∣ is large; together with (45) it gives the closed form of the dual function.

Formalization Note "min" is stated as IsLeast of the range of DDD, so attainment is part of the statement. The paper's subsequent combined formula −[(∣a∣−η)2/(4λ2)−λ0]+-[(|a|-\eta)^2/(4\lambda_2)-\lambda_0]_+−[(∣a∣−η)2/(4λ2​)−λ0​]+​ is not stated: it fails when η>∣a∣+2λ0λ2\eta>|a|+2\sqrt{\lambda_0\lambda_2}η>∣a∣+2λ0​λ2​​.

Preamble
import Mathlib
import Definitions.Def_L0BnB_Duality_Dual
Formal statement
namespace L0BnB.Duality

/-- (47), proof of Theorem 2, p. 31. For `a = αᵀXᵢ` and `η = ηᵢ ≥ 0` with
`|a| − η ≥ 2√(λ₀λ₂)`, the minimum over `βᵢ ∈ ℝ` of `Dᵢ(βᵢ) = ψ₁(βᵢ; λ₀, λ₂) + a βᵢ + η|βᵢ|`
is attained and equals `−(|a| − η)²/(4λ₂) + λ₀`. -/
theorem eq_47 (lam0 lam2 a η : ℝ) (hlam0 : 0 < lam0) (hlam2 : 0 < lam2) (hη : 0 ≤ η)
    (hcase : |a| - η ≥ 2 * Real.sqrt (lam0 * lam2)) :
    IsLeast (Set.range (D lam0 lam2 a η)) (-(1 / (4 * lam2)) * (|a| - η) ^ 2 + lam0) := by sorry

end L0BnB.Duality
Source
Hazimeh, Mazumder, Saab, Sparse Regression at Scale: Branch-and-Bound rooted in First-Order Optimization, arXiv:2004.06152v2, p. 31, Proof of Theorem 2, Case 2, (47)
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