Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 3.5 — limits of KKT points of the Kadrani et al. relaxation are M-stationary under MPEC-CPLD

Proved
MPECRelax.KDBConv.theorem_3_5

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

constraint-qualificationsmpecp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1relaxation-methodsstationarity

Consider the MPEC (1) with continuously differentiable data f,gi,hi,Gi,Hi:Rn→Rf,g_i,h_i,G_i,H_i:\mathbb R^n\to\mathbb Rf,gi​,hi​,Gi​,Hi​:Rn→R, and for t>0t>0t>0 the relaxed program of Kadrani, Dussault and Benchakroun

RKDB(t):min⁡f(x)  s.t.  g(x)≤0, h(x)=0, Gi(x)≥−t, Hi(x)≥−t, (Gi(x)−t)(Hi(x)−t)≤0  (i=1,…,l).R^{KDB}(t):\quad \min f(x)\ \text{ s.t. }\ g(x)\le 0,\ h(x)=0,\ G_i(x)\ge -t,\ H_i(x)\ge -t,\ (G_i(x)-t)(H_i(x)-t)\le 0\ \ (i=1,\dots,l).RKDB(t):minf(x)  s.t.  g(x)≤0, h(x)=0, Gi​(x)≥−t, Hi​(x)≥−t, (Gi​(x)−t)(Hi​(x)−t)≤0  (i=1,…,l).

Theorem 3.5. Let {tk}↓0\{t_k\}\downarrow 0{tk​}↓0 and assume that xkx^kxk is a stationary point of RKDB(tk)R^{KDB}(t_k)RKDB(tk​) for all k∈Nk\in\mathbb Nk∈N. Moreover, suppose that xk→x∗x^k\to x^*xk→x∗ such that MPEC-CPLD holds at x∗x^*x∗. Then x∗x^*x∗ is an M-stationary point of (1): x∗x^*x∗ is feasible for (1), and there are multipliers λ∈Rm\lambda\in\mathbb R^mλ∈Rm, μ∈Rp\mu\in\mathbb R^pμ∈Rp, γ,ν∈Rl\gamma,\nu\in\mathbb R^lγ,ν∈Rl with

∇f(x∗)+∑i=1mλi∇gi(x∗)+∑i=1pμi∇hi(x∗)−∑i=1lγi∇Gi(x∗)−∑i=1lνi∇Hi(x∗)=0,\nabla f(x^*)+\sum_{i=1}^m\lambda_i\nabla g_i(x^*)+\sum_{i=1}^p\mu_i\nabla h_i(x^*)-\sum_{i=1}^l\gamma_i\nabla G_i(x^*)-\sum_{i=1}^l\nu_i\nabla H_i(x^*)=0,∇f(x∗)+i=1∑m​λi​∇gi​(x∗)+i=1∑p​μi​∇hi​(x∗)−i=1∑l​γi​∇Gi​(x∗)−i=1∑l​νi​∇Hi​(x∗)=0,

λ≥0\lambda\ge0λ≥0, λigi(x∗)=0\lambda_ig_i(x^*)=0λi​gi​(x∗)=0, γi=0\gamma_i=0γi​=0 (i∈I+0i\in I_{+0}i∈I+0​), νi=0\nu_i=0νi​=0 (i∈I0+i\in I_{0+}i∈I0+​), and for every i∈I00i\in I_{00}i∈I00​ either γi>0\gamma_i>0γi​>0 and νi>0\nu_i>0νi​>0, or γiνi=0\gamma_i\nu_i=0γi​νi​=0.

The result generalizes the convergence theorem of Kadrani et al. by replacing MPEC-LICQ with the much weaker MPEC-CPLD. It says that the relaxation method, run with parameters tending to zero, can only accumulate at M-stationary points whenever MPEC-CPLD holds there.

Formalization Note "Stationary point" of RKDB(tk)R^{KDB}(t_k)RKDB(tk​) means a feasible point with KKT multipliers (p. 5). "{tk}↓0\{t_k\}\downarrow 0{tk​}↓0" is encoded as tk>0t_k>0tk​>0, ttt nonincreasing and tk→0t_k\to 0tk​→0. The C¹ standing assumption of p. 1 is the hypothesis P.IsC1. M-stationarity (Definition 2.3(c), with the misprints of 2.3(a) corrected) requires a single multiplier tuple for both the weak-stationarity equation and the condition on I00I_{00}I00​, and includes feasibility of x∗x^*x∗, which is not assumed. MPEC-CPLD is standard CPLD for TNLP(x∗)(x^*)(x∗) (Definition 2.4), with linear dependence required on a whole neighbourhood of x∗x^*x∗.

Preamble
import Mathlib
import Definitions.Def_MPECRelax_KDBConv_Basic

open Filter Topology
Formal statement
namespace MPECRelax.KDBConv

/-- Theorem 3.5 (p. 15): let `t_k ↓ 0` (positive, nonincreasing, tending to `0`) and let
`x^k` be a stationary (KKT) point of the relaxed program R^KDB(t_k) for every `k`. If
`x^k → xs` and MPEC-CPLD holds at `xs`, then `xs` is an M-stationary point of the
MPEC (1) (feasibility of `xs` is part of the conclusion). -/
theorem theorem_3_5 {n m p l : ℕ} (P : MPEC n m p l) (hP : P.IsC1)
    (t : ℕ → ℝ) (ht_pos : ∀ k, 0 < t k) (ht_anti : Antitone t)
    (ht_lim : Tendsto t atTop (𝓝 0))
    (x : ℕ → MPECRelax.ScholtesConv.E n) (hx : ∀ k, (P.RKDB (t k)).IsKKTPoint (x k))
    (xs : MPECRelax.ScholtesConv.E n) (hxs : Tendsto x atTop (𝓝 xs)) (hCQ : P.MPEC_CPLD xs) :
    P.IsMStationary xs := by sorry

end MPECRelax.KDBConv
Source
Hoheisel, Kanzow, Schwartz, Theoretical and numerical comparison of relaxation methods for mathematical programs with complementarity constraints, Preprint 299, Univ. Würzburg, Sept. 2010, p. 15, Theorem 3.5
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