Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 1 — predictor-corrector termination in O(√n·t) iterations

Proved
AdaptiveStepIPM.PredCorr.theorem_1

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

interior-point-methodslinear-programmingp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1

Let n≥1n\ge1n≥1 and let ttt be a real precision parameter. Start Algorithm 1 at (x0,s0)∈N2(1/4)(x^0,s^0)\in N_2(1/4)(x0,s0)∈N2​(1/4) with (x0)Ts0≤2t(x^0)^Ts^0\le2^t(x0)Ts0≤2t. At every iteration whose gap is greater than 2−t2^{-t}2−t, take the predictor step at γ=0\gamma=0γ=0 to the greatest step whose entire prefix lies in N2(1/2)N_2(1/2)N2​(1/2), followed by one full corrector step at γ=1\gamma=1γ=1. Then for some iteration kkk,

k≤⌈2(log⁡2)81/4n t⌉,(xk)Tsk≤2−t.k\le\left\lceil 2(\log 2)8^{1/4}\sqrt n\,t\right\rceil,\qquad (x^k)^Ts^k\le2^{-t}.k≤⌈2(log2)81/4n​t⌉,(xk)Tsk≤2−t.

Thus Algorithm 1 attains precision ttt in the O(n t)O(\sqrt n\,t)O(n​t) iteration scale stated in Theorem 1.

Formalization Note The explicit ceiling is the count yielded by the paper's contraction factor 1−8−1/4n−1/21-8^{-1/4}n^{-1/2}1−8−1/4n−1/2, with natural logarithm. The run relation uses an attained greatest predictor step. At inputs where no greatest step exists, such as n=1n=1n=1, it has no successor; the statement keeps the report's n≥1n\ge1n≥1 range, although those runs cannot continue above the stopping threshold. No sign condition on ttt is imposed: for t≤0t\le0t≤0 the starting pair already satisfies (x0)Ts0≤2t≤2−t(x^0)^Ts^0\le2^t\le2^{-t}(x0)Ts0≤2t≤2−t and the bound is met at k=0k=0k=0.

Preamble
import Mathlib
import Definitions.Def_AdaptiveStepIPM_PredCorr_Algorithm

open Matrix
Formal statement
namespace AdaptiveStepIPM.PredCorr

/-- Theorem 1, printed p. 8, with the explicit count from its proof. -/
theorem theorem_1 {m n : ℕ} (hn : 1 ≤ n)
    (A : Matrix (Fin m) (Fin n) ℝ) (b : Fin m → ℝ) (c : Fin n → ℝ)
    (t : ℝ) (x s : ℕ → Fin n → ℝ)
    (hinit : N2 A b c (1 / 4) (x 0) (s 0))
    (hgap : x 0 ⬝ᵥ s 0 ≤ (2 : ℝ) ^ t)
    (hrun : ∀ k : ℕ, (2 : ℝ) ^ (-t) < x k ⬝ᵥ s k →
      Alg1Step A b c (x k) (s k) (x (k + 1)) (s (k + 1))) :
    ∃ k : ℕ,
      k ≤ ⌈2 * Real.log 2 * (8 : ℝ) ^ (1 / 4 : ℝ) *
        Real.sqrt (n : ℝ) * t⌉₊ ∧
      x k ⬝ᵥ s k ≤ (2 : ℝ) ^ (-t) := by sorry

end AdaptiveStepIPM.PredCorr
Source
Mizuno, Todd, Ye, On adaptive-step primal-dual interior-point algorithms for linear programming, Cornell ORIE Technical Report No. 944 (1990, rev. 1991), pp. 8–9, Theorem 1 and proof
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