Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 3 — each iteration of Algorithm 3 decreases ψ\psiψ by at least 11nlog⁡11−β11n\log\frac{1}{1-\beta}11nlog1−β1​

Proved
AdaptiveStepIPM.Potential.potential_decrease_per_iteration

by mikedeng1 · Oct 6, 2026 · Mathlib 0df444a (Lean v4.33.1)

interior-point-methodslinear-programmingp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1potential-reduction

Let n≥1n \ge 1n≥1, β,γ∈(0,1)\beta, \gamma \in (0,1)β,γ∈(0,1) with γ≤2(1−β)\gamma \le 2(1-\beta)γ≤2(1−β), and ρ=n+(3βγ(1−γ)log⁡11−β)n2\rho = n + \bigl(\frac{3}{\beta\gamma(1-\gamma)}\log\frac{1}{1-\beta}\bigr)n^2ρ=n+(βγ(1−γ)3​log1−β1​)n2. Let (x,s)∈N∞−(β)(x, s) \in \mathcal N_\infty^-(\beta)(x,s)∈N∞−​(β), let d=(dx,dy,ds)d = (d_x, d_y, d_s)d=(dx​,dy​,ds​) solve system (2) at (x,s)(x,s)(x,s) with parameter γ\gammaγ, and let θˉ\bar\thetaθˉ be a step of Algorithm 3: (x(θˉ),s(θˉ))∈N∞−(β)(x(\bar\theta), s(\bar\theta)) \in \mathcal N_\infty^-(\beta)(x(θˉ),s(θˉ))∈N∞−​(β) and ψ(x(θˉ),s(θˉ))≤ψ(x(θ),s(θ))\psi(x(\bar\theta), s(\bar\theta)) \le \psi(x(\theta), s(\theta))ψ(x(θˉ),s(θˉ))≤ψ(x(θ),s(θ)) for every real θ\thetaθ with (x(θ),s(θ))∈N∞−(β)(x(\theta), s(\theta)) \in \mathcal N_\infty^-(\beta)(x(θ),s(θ))∈N∞−​(β). Then

ψ(x(θˉ),s(θˉ))−ψ(x,s)≤−11nlog⁡11−β.\psi(x(\bar\theta), s(\bar\theta)) - \psi(x, s) \le -11n\log\frac{1}{1-\beta}.ψ(x(θˉ),s(θˉ))−ψ(x,s)≤−11nlog1−β1​.

Since ρ−n\rho - nρ−n is of order n2n^2n2 and a total decrease of order n2tn^2tn2t suffices, this fixed decrease per iteration is what yields the O(nt)O(nt)O(nt) bound of Theorem 3.

Formalization Note The paper derives this from (16) and θ2−≥4βγ/n\theta_2^- \ge 4\beta\gamma/nθ2−​≥4βγ/n. For n≥2n \ge 2n≥2 the latter holds because 4βγ≤2≤n4\beta\gamma \le 2 \le n4βγ≤2≤n under γ≤2(1−β)\gamma \le 2(1-\beta)γ≤2(1−β). For n=1n = 1n=1 it can fail, but then no step θˉ\bar\thetaθˉ of Algorithm 3 exists (N∞−(β)=F0\mathcal N_\infty^-(\beta) = \mathcal F^0N∞−​(β)=F0, x(θ)s(θ)=(1−(1−γ)θ)μx(\theta)s(\theta) = (1-(1-\gamma)\theta)\mux(θ)s(θ)=(1−(1−γ)θ)μ and ψ=(ρ−1)log⁡(x(θ)s(θ))\psi = (\rho-1)\log(x(\theta)s(\theta))ψ=(ρ−1)log(x(θ)s(θ)) is unbounded below on [0,1/(1−γ))[0, 1/(1-\gamma))[0,1/(1−γ))), so the statement holds there with no extra hypothesis; n≥1n \ge 1n≥1 is kept as on the page.

Preamble
import Mathlib
import Definitions.Def_LinearOptimization_InteriorPointPotential
import Definitions.Def_AdaptiveStepIPM_Potential_Algorithm3

open Matrix
Formal statement
namespace AdaptiveStepIPM.Potential

/-- **Proof of Theorem 3, p. 12.** Under the hypotheses of (16),
one iteration of Algorithm 3 decreases the potential by at least `11n log(1/(1 − β))`:
`ψ(x(θ̄), s(θ̄)) − ψ(x, s) ≤ −11n log(1/(1 − β))`. -/
theorem potential_decrease_per_iteration {m n : ℕ} (hn : 1 ≤ n)
    (A : Matrix (Fin m) (Fin n) ℝ) (b : Fin m → ℝ)
    (c : Fin n → ℝ) (β γ : ℝ) (hβ0 : 0 < β) (hβ1 : β < 1) (hγ0 : 0 < γ) (hγ1 : γ < 1)
    (hγβ : γ ≤ 2 * (1 - β)) (x s : Fin n → ℝ) (hN : InNinfMinus A b c β x s)
    (dx : Fin n → ℝ) (dy : Fin m → ℝ) (ds : Fin n → ℝ)
    (hd : IsNewtonDirection A γ x s dx dy ds) (θbar : ℝ)
    (hθN : InNinfMinus A b c β (lineStep x dx θbar) (lineStep s ds θbar))
    (hθmin : ∀ θ : ℝ, InNinfMinus A b c β (lineStep x dx θ) (lineStep s ds θ) →
      LinearOptimization.interiorPointPotential (rho n β γ)
          (lineStep x dx θbar) (lineStep s ds θbar) ≤
        LinearOptimization.interiorPointPotential (rho n β γ)
          (lineStep x dx θ) (lineStep s ds θ)) :
    LinearOptimization.interiorPointPotential (rho n β γ)
          (lineStep x dx θbar) (lineStep s ds θbar) -
        LinearOptimization.interiorPointPotential (rho n β γ) x s ≤
      -(11 * (n : ℝ) * Real.log (1 / (1 - β))) := by sorry

end AdaptiveStepIPM.Potential
Source
Mizuno, Todd, Ye, On adaptive-step primal-dual interior-point algorithms for linear programming, Cornell ORIE Technical Report No. 944 (1990, rev. 1991), p. 12, proof of Theorem 3 (after (16))
Human review
  • Endorsed by Shuze Chen · Oct 6, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 6, 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