QUICK REVIEW
[Paper Review] Hilbert geometry of polytopes
Andreas Bernig|arXiv (Cornell University)|Dec 5, 2008
Point processes and geometric inequalities4 citations
TL;DR
This paper proves that the Hilbert metric on the interior of any convex polytope is bilipschitz equivalent to a normed vector space of the same dimension. The authors construct an explicit diffeomorphism Φ(x) = ∑ log fᵢ(x) grad fᵢ mapping the polytope's interior to the dual space, showing both Lipschitz continuity and a lower bound on the differential's norm, which establishes the bilipschitz equivalence.
ABSTRACT
It is shown that the Hilbert metric on the interior of a convex polytope is bilipschitz to a normed vector space of the same dimension.
Motivation & Objective
- To resolve the open question of whether the Hilbert metric on a convex polytope is bilipschitz equivalent to a normed vector space.
- To establish a quantitative geometric equivalence between Hilbert geometry and finite-dimensional normed spaces for polytopal domains.
- To provide a constructive proof using a specific diffeomorphism based on the defining affine functions of the polytope.
- To extend prior results on Hilbert metrics of polygons and C¹¹ bodies to general polytopes, clarifying the bilipschitz classification.
Proposed method
- Define the Hilbert metric d(x,y) via cross-ratio on the line through x and y intersecting the boundary ∂P.
- Construct the map Φ: int(P) → V* by Φ(x) = ∑ᵢ log fᵢ(x) dfᵢ, where fᵢ are the affine functions defining the polytope.
- Prove Lipschitz continuity of Φ using bounds on the log differences |log fᵢ(x) - log fᵢ(y)| and the gradient norms of fᵢ.
- Establish injectivity of Φ via strict positivity of the inner product ⟨Φ(x) - Φ(y), x - y⟩ when x ≠ y.
- Use induction on face dimensions to prove the differential dΦ satisfies a lower bound ‖dΦ(v)‖₂ ≥ C_F ‖v‖ for v in tangent spaces.
- Apply geometric estimates near faces and use quotient spaces to reduce to lower-dimensional polyhedral cones, ensuring uniform lower bounds on the differential.
Experimental results
Research questions
- RQ1Is the Hilbert metric on a convex polytope bilipschitz equivalent to a normed vector space?
- RQ2Can a canonical diffeomorphism be constructed that realizes this bilipschitz equivalence?
- RQ3What is the behavior of the Hilbert metric near the boundary of a polytope, and how does it relate to the geometry of the defining affine functions?
- RQ4Does the bilipschitz property hold uniformly across all tangent directions, even near lower-dimensional faces?
- RQ5How does the differential of the proposed map Φ behave in relation to the Finsler norm and the Euclidean norm on the tangent space?
Key findings
- The Hilbert metric on the interior of any compact convex polytope is bilipschitz equivalent to a normed vector space of the same dimension.
- The map Φ(x) = ∑ᵢ log fᵢ(x) grad fᵢ is a bilipschitz diffeomorphism from int(P) to V*.
- The bilipschitz constant depends only on the number of facets m, the maximum gradient norm R of the defining affine functions, and the diameter D of the polytope.
- Lipschitz continuity of Φ is established via the bound ‖Φ(x) - Φ(y)‖₂ ≤ 2mR d(x,y).
- The differential dΦ satisfies a uniform lower bound ‖dΦ(v)‖₂ ≥ C_F ‖v‖ for all tangent vectors v, with C_F depending on the face dimension and the geometry of P.
- The result holds uniformly across all compact subsets of int(P), with the lower bound on dΦ being uniform near each face of the polytope.
Better researchstarts right now
From reading papers to final review, dramatically reduce your research time.
No credit card · Free plan available
This review was created by AI and reviewed by human editors.