Prove2Me
Navigate
DiscoverCollectionsFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

Larman's bound n 2d−3n\,2^{d-3}n2d−3

Proved
Hirsch.larman_bound

by Shuze Chen · Aug 22, 2026 · Mathlib c5ea003 (Lean v4.30.0)

diameter-boundhirsch-conjecturepolytopes

(Larman 1970.) Every nonempty bounded H-polytope in Rd\mathbb{R}^dRd described by nnn inequalities has combinatorial diameter at most n⋅2d−3n \cdot 2^{d-3}n⋅2d−3. The bound is linear in the number of inequalities for each fixed dimension — still the best known bound of that shape. The exponent d−3d - 3d−3 is truncated natural subtraction, so for d≤3d \le 3d≤3 the asserted bound is nnn, which holds; lower-dimensional polytopes are included.

Preamble
import Mathlib
import Definitions.Def_Hirsch_model
Formal statement
namespace Hirsch

theorem larman_bound (d n : ℕ)
    (a : Fin n → EuclideanSpace ℝ (Fin d)) (b : Fin n → ℝ)
    (hne : (Hpoly a b).Nonempty) (hbd : Bornology.IsBounded (Hpoly a b)) :
    DiamLE (Hpoly a b) (n * 2 ^ (d - 3)) := by sorry

end Hirsch
Source
Larman, Paths on polytopes, Proc. London Math. Soc. s3-20 (1970) 161-178, https://doi.org/10.1112/plms/s3-20.2.249
Read-back

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

Read-back of Hirsch.larman_bound.

For every pair of natural numbers d,n≥0d, n \ge 0d,n≥0, every family of vectors a1,…,an∈Rda_1, \dots, a_n \in \mathbb{R}^da1​,…,an​∈Rd (the Euclidean space Rd\mathbb{R}^dRd with its standard inner product), and every family of reals b1,…,bn∈Rb_1, \dots, b_n \in \mathbb{R}b1​,…,bn​∈R, consider the set

P={ x∈Rd∣⟨ai,x⟩≤bi for all i=1,…,n },P = \{\, x \in \mathbb{R}^d \mid \langle a_i, x\rangle \le b_i \text{ for all } i = 1, \dots, n \,\},P={x∈Rd∣⟨ai​,x⟩≤bi​ for all i=1,…,n},

i.e. the solution set of the nnn linear inequalities (with no assumption that the aia_iai​ are nonzero or distinct; when n=0n = 0n=0 this is all of Rd\mathbb{R}^dRd). Under the two hypotheses that PPP is nonempty and that PPP is bounded (as a subset of the metric space Rd\mathbb{R}^dRd), the theorem asserts the following, where the bound is the natural number

B=n⋅2 d−3B = n \cdot 2^{\,d - 3}B=n⋅2d−3

with truncated natural-number subtraction in the exponent: d−3=0d - 3 = 0d−3=0 whenever d≤3d \le 3d≤3, so for d∈{0,1,2,3}d \in \{0,1,2,3\}d∈{0,1,2,3} the bound is B=nB = nB=n, and B=0B = 0B=0 whenever n=0n = 0n=0.

The conclusion is the predicate "DiamLE(P,B)\mathrm{DiamLE}(P, B)DiamLE(P,B)", which unfolds to: for every two extreme points u,vu, vu,v of PPP (an extreme point of PPP is a point of PPP that does not lie in the open segment between any two points of PPP other than itself), there exists a sequence of points w0,w1,w2,⋯∈Rdw_0, w_1, w_2, \dots \in \mathbb{R}^dw0​,w1​,w2​,⋯∈Rd (defined for all indices) such that

  • w0=uw_0 = uw0​=u and wB=vw_B = vwB​=v, and
  • for every index i<Bi < Bi<B, either wi=wi+1w_i = w_{i+1}wi​=wi+1​ (a stationary step), or wiw_iwi​ and wi+1w_{i+1}wi+1​ are "adjacent on PPP", meaning: wi≠wi+1w_i \ne w_{i+1}wi​=wi+1​ and the closed segment [wi,wi+1][w_i, w_{i+1}][wi​,wi+1​] is an extreme subset of PPP — that is, [wi,wi+1]⊆P[w_i, w_{i+1}] \subseteq P[wi​,wi+1​]⊆P, and whenever a point of [wi,wi+1][w_i, w_{i+1}][wi​,wi+1​] lies strictly between two points x,y∈Px, y \in Px,y∈P (in the open segment from xxx to yyy), both xxx and yyy already belong to [wi,wi+1][w_i, w_{i+1}][wi​,wi+1​].

In words: any two extreme points of PPP are joined by a walk of exactly BBB steps, each step either staying put or moving along a segment that is an extreme subset of PPP; because stationary steps are allowed, this expresses that any two extreme points are connected by at most BBB such moves. The intermediate points wiw_iwi​ are quantified only through these step conditions — they are not separately required to be extreme points of PPP (though each adjacency step forces its endpoints to lie in PPP). Note the degenerate instances included by the quantifiers: if n=0n = 0n=0 then B=0B = 0B=0 and the conclusion demands u=vu = vu=v for all extreme points u,vu, vu,v of P=RdP = \mathbb{R}^dP=Rd (the boundedness hypothesis then restricts this case to d=0d = 0d=0); if PPP has no extreme points, or fewer than two, the conclusion is vacuous or trivial. The theorem is universally quantified over all ddd, nnn, aaa, bbb satisfying the two hypotheses, and the stated inequality on the walk length is non-strict (≤B\le B≤B in effect, via the stationary steps), with BBB computed as the natural number n⋅2d−3n \cdot 2^{d-3}n⋅2d−3 exactly as above.

Human review
  • Endorsed by ShouqiaoWang · Aug 25, 2026

  • Endorsed by Shuze Chen · Aug 25, 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