Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 3 — the pure corrector direction (17) is orthogonal and leaves the complementarity gap unchanged to first order

Proved
ExpConeIPM.Corrector.lemma_3

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

conic-optimizationinterior-point-methodsp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1

Work in the setting of Lemma 2: the homogeneous model with K=K^×R+K=\hat K\times\mathbb R_+K=K^×R+​ and barrier F(x^,τ)=F^(x^)−log⁡τF(\hat x,\tau)=\hat F(\hat x)-\log\tauF(x^,τ)=F^(x^)−logτ, an iterate z=(x,s,y)z=(x,s,y)z=(x,s,y) with x∈int⁡Kx\in\operatorname{int}Kx∈intK, s∈int⁡K∗s\in\operatorname{int}K^*s∈intK∗, a nonsingular scaling WWW satisfying (8), and a solution Δza\Delta z^{a}Δza of the affine system (11). Let

η:=−12F′′′(x)[Δxa,(F′′(x))−1Δsa](16)\eta:=-\tfrac12F'''(x)[\Delta x^{a},(F''(x))^{-1}\Delta s^{a}] \tag{16}η:=−21​F′′′(x)[Δxa,(F′′(x))−1Δsa](16)

be the corrector term. Then every solution Δzc=(Δxc,Δsc,Δyc)\Delta z^{c}=(\Delta x^{c},\Delta s^{c},\Delta y^{c})Δzc=(Δxc,Δsc,Δyc) of the pure corrector system (17),

G(Δzc)=0,WΔxc+W−TΔsc=−W−Tη,G(\Delta z^{c})=0,\qquad W\Delta x^{c}+W^{-T}\Delta s^{c}=-W^{-T}\eta,G(Δzc)=0,WΔxc+W−TΔsc=−W−Tη,

satisfies

⟨s,Δxc⟩+⟨x,Δsc⟩=0,⟨Δxc,Δsc⟩=0.\langle s,\Delta x^{c}\rangle+\langle x,\Delta s^{c}\rangle=0,\qquad\langle\Delta x^{c},\Delta s^{c}\rangle=0 .⟨s,Δxc⟩+⟨x,Δsc⟩=0,⟨Δxc,Δsc⟩=0.

The corrector therefore changes neither the residual nor the complementarity gap; adding it to the affine and centering directions in (18) does not disturb the uniform decrease established in Lemma 4.

Formalization Note F′′(x)F''(x)F′′(x) is the published hess, F′′′(x)[u,w]F'''(x)[u,w]F′′′(x)[u,w] is fderiv ℝ (hess F) x u w, and (F′′(x))−1(F''(x))^{-1}(F′′(x))−1 is ContinuousLinearMap.inverse (the true inverse: the augmented barrier has a positive definite Hessian at interior points). The barrier in (16) is the augmented FFF, so η\etaη has a τ\tauτ-coordinate; inner products include the τκ\tau\kappaτκ terms. Generalizations as in Lemma 2 (one cone K^\hat KK^, WWW not assumed block diagonal).

Preamble
import Mathlib
import Definitions.Def_SelfScaledIPM_ShortStep_Setting
import Definitions.Def_ExpConeIPM_Corrector_HomogeneousModel
import Definitions.Def_ExpConeIPM_Corrector_SearchDirections

open scoped InnerProductSpace
Formal statement
namespace ExpConeIPM.Corrector

/-- **Lemma 3** (Dahl–Andersen, Math. Program. 194 (2022), p. 352). In the setting of Lemma 2, let
`Δzᵃ` solve the affine system (11), let `η := −½ F'''(x)[Δxᵃ, (F''(x))⁻¹Δsᵃ]` be the corrector (16) of
the augmented barrier `F = F̂ − log τ`, and let `Δzᶜ` solve the pure corrector system (17). Then
`⟨s, Δxᶜ⟩ + ⟨x, Δsᶜ⟩ = 0` and `⟨Δxᶜ, Δsᶜ⟩ = 0`. -/
theorem lemma_3 {n m : ℕ}
    (A : EuclideanSpace ℝ (Fin n) →L[ℝ] EuclideanSpace ℝ (Fin m))
    (b : EuclideanSpace ℝ (Fin m)) (c : EuclideanSpace ℝ (Fin n))
    (hA : Function.Surjective A)
    (Khat : Set (EuclideanSpace ℝ (Fin n))) (Fhat : EuclideanSpace ℝ (Fin n) → ℝ) (ϑhat : ℝ)
    (hK : SelfScaledIPM.ShortStep.IsProperCone Khat)
    (hF : SelfScaledIPM.ShortStep.IsLogHomBarrier Khat Fhat ϑhat)
    (x s : EuclideanSpace ℝ (Fin (n + 1))) (y : EuclideanSpace ℝ (Fin m))
    (hz : IsInteriorIterate Khat x s)
    (W : EuclideanSpace ℝ (Fin (n + 1)) ≃L[ℝ] EuclideanSpace ℝ (Fin (n + 1)))
    (hW : IsDoubleSecantScaling (augCone Khat) (augBarrier Fhat) x s W)
    (dxa dsa : EuclideanSpace ℝ (Fin (n + 1))) (dya : EuclideanSpace ℝ (Fin m))
    (ha : IsAffineDirection A b c x s y W dxa dsa dya)
    (dxc dsc : EuclideanSpace ℝ (Fin (n + 1))) (dyc : EuclideanSpace ℝ (Fin m))
    (hc : IsPureCorrectorDirection A b c W (corrector (augBarrier Fhat) x dxa dsa) dxc dsc dyc) :
    ⟪s, dxc⟫_ℝ + ⟪x, dsc⟫_ℝ = 0 ∧ ⟪dxc, dsc⟫_ℝ = 0 := by sorry

end ExpConeIPM.Corrector
Source
Dahl, Andersen, A primal-dual interior-point algorithm for nonsymmetric exponential-cone optimization, Math. Program. 194 (2022), p. 352, Lemma 3 (with (16), (17))
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