Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 12, p. 14 — λ* of the game (6) known to error ε‴ = ε/(6R(r+1)) decides OPT < α or OPT ≥ α − ε (added |α| ≤ R; (6) row 2 corrected)

Proved
QZeroSum19.LPReduce.lemma_12

by mikedeng1 · Oct 15, 2026 · Mathlib 0df444a (Lean v4.33.1)

linear-programminglp-reductionp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1zero-sum-game

Consider the linear program (4), max⁡{bTy:Ay≤c, y≥0}\max\{b^{\mathsf T}y : Ay \le c,\ y \ge 0\}max{bTy:Ay≤c, y≥0} with A∈Rn×mA \in \mathbb R^{n\times m}A∈Rn×m, and its dual (5), min⁡{cTx:ATx≥b, x≥0}\min\{c^{\mathsf T}x : A^{\mathsf T}x \ge b,\ x \ge 0\}min{cTx:ATx≥b, x≥0}, under the standing assumptions of §4: all entries of AAA, bbb, ccc lie in [−1,1][-1,1][−1,1]; R>0R > 0R>0 and ∣ci∣≤R|c_i| \le R∣ci​∣≤R for all iii; the common optimal value OPT\mathrm{OPT}OPT of (4) and (5) is attained by a primal feasible yyy with ∑jyj≤R\sum_j y_j \le R∑j​yj​≤R and by a dual feasible xxx with ∑ixi≤r\sum_i x_i \le r∑i​xi​≤r. Let α∈R\alpha \in \mathbb Rα∈R with ∣α∣≤R|\alpha| \le R∣α∣≤R, let ε>0\varepsilon > 0ε>0, and put

ε′′′=ε6R(r+1).\varepsilon''' = \frac{\varepsilon}{6R(r+1)} .ε′′′=6R(r+1)ε​.

Let λ∗\lambda^*λ∗ be the value of the zero-sum game with matrix

A′′′=(eT1−1−eT−11−bT0α/RA0−c/R),(6)A''' = \begin{pmatrix} e^{\mathsf T} & 1 & -1 \\ -e^{\mathsf T} & -1 & 1 \\ -b^{\mathsf T} & 0 & \alpha/R \\ A & 0 & -c/R \end{pmatrix}, \qquad\qquad (6)A′′′=​eT−eT−bTA​1−100​−11α/R−c/R​​,(6)

that is, λ∗=min⁡y′′′∈Δm+2max⁡k(A′′′y′′′)k\lambda^* = \min_{y''' \in \Delta^{m+2}} \max_k (A'''y''')_kλ∗=miny′′′∈Δm+2​maxk​(A′′′y′′′)k​. If λ~\tilde\lambdaλ~ is any number with ∣λ~−λ∗∣≤ε′′′|\tilde\lambda - \lambda^*| \le \varepsilon'''∣λ~−λ∗∣≤ε′′′, then

λ~>ε′′′  ⟹  OPT<α,λ~≤ε′′′  ⟹  OPT≥α−ε.\tilde\lambda > \varepsilon''' \implies \mathrm{OPT} < \alpha, \qquad\qquad \tilde\lambda \le \varepsilon''' \implies \mathrm{OPT} \ge \alpha - \varepsilon .λ~>ε′′′⟹OPT<α,λ~≤ε′′′⟹OPT≥α−ε.

In the words of the paper: finding the optimal value λ∗\lambda^*λ∗ of the game (6) up to additive error ε′′′\varepsilon'''ε′′′ suffices to correctly conclude either OPT<α\mathrm{OPT} < \alphaOPT<α or OPT≥α−ε\mathrm{OPT} \ge \alpha - \varepsilonOPT≥α−ε. Combined with a binary search over α∈[−R,R]\alpha \in [-R, R]α∈[−R,R] and an approximate game solver, it turns any algorithm for zero-sum games into an algorithm for linear programming; Theorem 13 of the paper is obtained this way.

Formalization Note Two deviations from the page, both disclosed. (i) The second row of (6) is printed as (−eT,1,1)(-e^{\mathsf T}, 1, 1)(−eT,1,1); the derivation on p. 13 gives (−eT,−1,1)(-e^{\mathsf T}, -1, 1)(−eT,−1,1), which is used. With the printed row the lemma is false (for n=m=1n = m = 1n=m=1, A=1A = 1A=1, c=1c = 1c=1, b=−1b = -1b=−1, R=1R = 1R=1, r=0r = 0r=0, α=0\alpha = 0α=0 one has OPT=0≥α\mathrm{OPT} = 0 \ge \alphaOPT=0≥α but λ∗>0\lambda^* > 0λ∗>0). (ii) The hypothesis ∣α∣≤R|\alpha| \le R∣α∣≤R is added; the proof's step (7) needs it, and §4 supplies it through the binary search over −R≤OPT≤R-R \le \mathrm{OPT} \le R−R≤OPT≤R and the requirement that (6) have entries in [−1,1][-1,1][−1,1]. The decision rule (λ~>ε′′′\tilde\lambda > \varepsilon'''λ~>ε′′′ versus λ~≤ε′′′\tilde\lambda \le \varepsilon'''λ~≤ε′′′) is the one of the first paragraph of the proof, which reduces to the cases λ∗>0\lambda^* > 0λ∗>0 and λ∗≤2ε′′′\lambda^* \le 2\varepsilon'''λ∗≤2ε′′′. The value λ∗\lambda^*λ∗ is the infimum of the image of the simplex, not an infimum over a membership condition.

Preamble
import Mathlib
import Definitions.Def_QZeroSum19_LPReduce_Setting

open Matrix
Formal statement
namespace QZeroSum19.LPReduce
theorem lemma_12 {n m : ℕ} (A : Matrix (Fin n) (Fin m) ℝ) (b : Fin m → ℝ) (c : Fin n → ℝ)
    (R r OPT α ε : ℝ) (hLP : LPStanding A b c R r OPT) (hα : |α| ≤ R) (hε : 0 < ε)
    (lam : ℝ) (hlam : |lam - QZeroSum19.GK.gameValue (gameMatrix A b c α R)| ≤ epsGame ε R r) :
    (epsGame ε R r < lam → OPT < α) ∧ (lam ≤ epsGame ε R r → α - ε ≤ OPT) := by sorry
end QZeroSum19.LPReduce
Source
van Apeldoorn and Gilyén, Quantum algorithms for zero-sum games, arXiv:1904.03180v1, p. 14, Lemma 12 (with (6) on p. 14 and the setting of §4, p. 12)
Human review
  • Endorsed by Shuze Chen · Oct 15, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 15, 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