Prove2Me
Navigate
DiscoverFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

Theorem 4.4 — (EXC) iff every argmax⁡(ω[p])\operatorname{argmax}(\omega[p])argmax(ω[p]) is an integral base set

Proved
SteinitzExchange.Extension.exc_iff_argmax_isIntegralBaseSet

by mikedeng1 · Sep 27, 2026 · Mathlib 0df444a (Lean v4.33.1)

integral-base-setm-concavep2o-batch-p100bp2o-gran-per-chapterp2o-plan-paperp2o-v1

Let B⊆ZVB\subseteq\mathbb Z^VB⊆ZV be a finite integral base set and ω:B→R\omega:B\to\mathbb Rω:B→R. Then ω\omegaω satisfies (EXC) if and only if, for every p:V→Rp:V\to\mathbb Rp:V→R,

argmax⁡(ω[p])={x∈B∣ω(x)+⟨p,x⟩≥ω(y)+⟨p,y⟩ ∀y∈B}\operatorname{argmax}(\omega[p])=\{x\in B\mid \omega(x)+\langle p,x\rangle\ge\omega(y)+\langle p,y\rangle\ \forall y\in B\}argmax(ω[p])={x∈B∣ω(x)+⟨p,x⟩≥ω(y)+⟨p,y⟩ ∀y∈B}

is an integral base set (equivalently, argmax⁡(ω[p])‾\overline{\operatorname{argmax}(\omega[p])}argmax(ω[p])​ is an integral base polytope whose integer points are exactly argmax⁡(ω[p])\operatorname{argmax}(\omega[p])argmax(ω[p])).

This characterizes M-concavity by its maximizers, as concavity of a function is characterized by the convexity of the maximizer sets of all its linear perturbations.

Formalization Note. The page states the condition as "argmax⁡(ω[p])‾\overline{\operatorname{argmax}(\omega[p])}argmax(ω[p])​ is an integral base polytope". Read literally (the convex hull of argmax⁡(ω[p])\operatorname{argmax}(\omega[p])argmax(ω[p]) equals the convex hull of some integral base set) the "if" direction is false: on B={(2,0),(1,1),(0,2)}B=\{(2,0),(1,1),(0,2)\}B={(2,0),(1,1),(0,2)} with ω=0,−1,0\omega=0,-1,0ω=0,−1,0 every argmax⁡(ω[p])\operatorname{argmax}(\omega[p])argmax(ω[p]) is {(2,0)}\{(2,0)\}{(2,0)}, {(0,2)}\{(0,2)\}{(0,2)} or {(2,0),(0,2)}\{(2,0),(0,2)\}{(2,0),(0,2)}, each with an integral base polytope as convex hull, yet (EXC) fails. The paper's proof uses that argmax⁡(ω[p])\operatorname{argmax}(\omega[p])argmax(ω[p]) itself is an integral base set, the reading glossed in Lemma 4.3 and used when Theorem 4.4 is applied on p. 292; that reading is stated here.

Preamble
import Mathlib
import Definitions.Def_SteinitzExchange_Extension_IntegralBaseSet
import Definitions.Def_SteinitzExchange_Extension_Exchange
Formal statement
namespace SteinitzExchange.Extension

/-- Murota 1996, p. 286, Theorem 4.4, in the reading of Lemma 4.3 ("argmax(ω) is an integral base
set, that is, conv(argmax(ω)) is an integral base polytope"): on a finite integral base set `B`,
`ω` satisfies (EXC) iff `argmax(ω[p])` is an integral base set for every `p : V → ℝ`. -/
theorem exc_iff_argmax_isIntegralBaseSet {V : Type*} [Fintype V] [DecidableEq V] [Nonempty V]
    (B : Finset (V → ℤ)) (hB : IsIntegralBaseSet B) (ω : (V → ℤ) → ℝ) :
    SatisfiesEXC B ω ↔ ∀ p : V → ℝ, IsIntegralBaseSet (argmaxB B (perturb ω p)) := by sorry

end SteinitzExchange.Extension
Source
Murota, Convexity and Steinitz's Exchange Property, Adv. Math. 124 (1996), p. 286, Theorem 4.4 (read as in Lemma 4.3, p. 285, and as applied on p. 292)
Read-back

What the Lean code literally says, in plain math · claude-opus-5-5

Setting. VVV is a finite, nonempty type with decidable equality. χu\chi_uχu​ is the indicator vector of uuu, and ⟨p,b⟩=∑vp(v)b(v)\langle p, b\rangle = \sum_v p(v)b(v)⟨p,b⟩=∑v​p(v)b(v).

Hypotheses.

  • B⊆ZVB \subseteq \mathbb{Z}^VB⊆ZV is a finite integral base set. That is, BBB is nonempty, and for all x,y∈Bx, y\in Bx,y∈B and every uuu with (x−y)(u)>0(x-y)(u)>0(x−y)(u)>0 there is some vvv with (x−y)(v)<0(x-y)(v)<0(x−y)(v)<0 and x−χu+χv∈Bx-\chi_u+\chi_v\in Bx−χu​+χv​∈B.
  • ω:ZV→R\omega : \mathbb{Z}^V \to \mathbb{R}ω:ZV→R is arbitrary.

Conclusion. The following two statements are equivalent.

  • ω\omegaω satisfies (EXC) on BBB. For all x,y∈Bx, y\in Bx,y∈B and every uuu with (x−y)(u)>0(x-y)(u)>0(x−y)(u)>0, there is some vvv with (x−y)(v)<0(x-y)(v)<0(x−y)(v)<0 such that x−χu+χv∈Bx-\chi_u+\chi_v\in Bx−χu​+χv​∈B, y+χu−χv∈By+\chi_u-\chi_v\in By+χu​−χv​∈B and
ω(x)+ω(y)≤ω(x−χu+χv)+ω(y+χu−χv).\omega(x)+\omega(y)\le\omega(x-\chi_u+\chi_v)+\omega(y+\chi_u-\chi_v).ω(x)+ω(y)≤ω(x−χu​+χv​)+ω(y+χu​−χv​).
  • For every p∈RVp \in \mathbb{R}^Vp∈RV, the argmax of the perturbation is an integral base set. The set in question is
{x∈B:ω[p](y)≤ω[p](x) ∀y∈B},ω[p](x)=ω(x)+⟨p,ι(x)⟩.\{x\in B : \omega[p](y)\le\omega[p](x)\ \forall y\in B\}, \qquad \omega[p](x)=\omega(x)+\langle p,\iota(x)\rangle.{x∈B:ω[p](y)≤ω[p](x) ∀y∈B},ω[p](x)=ω(x)+⟨p,ι(x)⟩.

Being an integral base set means it is nonempty and has the one-sided exchange property within itself.

Degenerate cases.

  • If BBB has one element:
    • (EXC) holds vacuously;
    • every argmax is that one-element set, which is an integral base set;
    • so both sides are true.
  • Nonemptiness of each argmax is automatic, because BBB is finite and nonempty.
Human review
  • Endorsed by Shuze Chen · Sep 27, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Sep 27, 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 paper
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me