Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

A non-Hirsch 191919-dimensional polytope with 383838 facets (DeepMind 2025)

Open
HirschDim4.deepmind_nineteen

by LG · Nov 16, 2026 · Mathlib 0df444a (Lean v4.33.1)

discrete-geometryhirsch-conjecturelinear-optimizationpolytopes

There is a bounded, full-dimensional 191919-dimensional polytope with 383838 facets whose combinatorial diameter is greater than 191919 (in fact at least 202020): the smallest-dimensional counterexample to the Hirsch conjecture currently known. It was found in 2025 by Google DeepMind's Hopper multi-agent system: a 555-dimensional prismatoid of width 666 with 242424 vertices (121212 in each base facet), converted by Santos's spindle construction to dimension 24−5=1924-5=1924−5=19 with 2⋅19=382\cdot 19=382⋅19=38 facets. Explicit coordinates of the 242424-vertex prismatoid are given in Appendix A.13 of the source.

Formalization Note. IsNonHirsch a b is ¬ DiamLE (Hpoly a b) (38-19), i.e. diameter >19>19>19. Same shape as Santos's Corollary 1.7 with (n,d)=(38,19)(n,d)=(38,19)(n,d)=(38,19).

Preamble
import Mathlib
import Definitions.Def_Hirsch_model
import Definitions.Def_SantosHirsch_Counter_Setting

open scoped RealInnerProductSpace
Formal statement
namespace HirschDim4

theorem deepmind_nineteen :
    ∃ (a : Fin 38 → EuclideanSpace ℝ (Fin 19)) (b : Fin 38 → ℝ),
      (Hirsch.Hpoly a b).Nonempty ∧
      Bornology.IsBounded (Hirsch.Hpoly a b) ∧
      SantosHirsch.Counter.IsFacetPresentation a b ∧
      SantosHirsch.Counter.IsNonHirsch a b := by sorry

end HirschDim4
Source
G. Swirszcz, A. Z. Wagner, G. Williamson et al. (Google DeepMind), Advancing Geometry with AI: Multi-agent Generation of Polytopes, arXiv:2502.05199 (2025), Section 4.1 and Appendix A.13 (24-vertex width-six 5-prismatoid; non-Hirsch polytope of dimension 19 = 24-5, the smallest known).
Human review
  • Endorsed by Shuze Chen · Nov 17, 2026

    Confirmed by the moderator at approval.

  • Endorsed by LG · Nov 17, 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