MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $\{ G_n \}_{n \ge 1}$ be a sequence of graphs such that the number of vertices of $G_n$ tends to $\infty$ as $n \to \infty$. We say that $\{ G_n \}_{n \ge 1}$ is an expander family if $\lambda_2( G_n)$ is bounded away from $0$ as $n \to \infty$. Here $\lambda_2(H)$ is the smallest positive eigenvalue of the normalized graph Laplacian of $H$.

If every $G_n$ is the $1$-skeleton of a simple $3$-polytope, you can not have an expander family. This follows from the planar separator theorem.

What if every $G_n$ is the $1$-skeleton of a simple $4$-polytope? If you relax "simple" then there are obvious examples — in particular, cyclic polytopes in dimension $4$ have complete graphs for their $1$-skeletons, so they are expanders.

If it is hard to come up with an example for simple $4$-polytope, does it make it easier if we allow the dual graph of an arbitrary simplicial $3$-sphere?

share|cite|improve this question
up vote 7 down vote accepted

As far as I know there are no examples of graphs of simple 4-polytopes (or simple $d$-polytopes, for fixed $d$) that are expanders, and not even such examples for the dual graphs for triangulations of spheres of fixed dimensions. Some information on expansion properties of graphs of polytopes can be found in my chapter polytope skeletons and paths of the Handbook of Discrete and Computational Geometry (Goodman and O'Rourke, eds.).

There is a conjecture that graphs of polytopes are not very good expanders: Let $d$ be fixed. The conjecture asserts that a graph of every simple $d$-polytope with $n$ vertices can be separated into two parts, each having at least $n/3$ vertices by removing $O(n^{1-1/(d-1)})$ vertices.

This is known to be false for dual graphs to triangulations. It is known that there are dual graphs to triangulations of $S^3$ which cannot be separated even by $O(n / \log n)$ vertices. This is a construction from G. L. Miller, S.-H. Teng, W. Thurston, and S. A. Vavasis. "Separators for sphere packings and nearest neighbor graphs." J. ACM, 44(1):1-29, 1997. (PDF download) (I would be happy to see a nice exposition of this construction.)

On the positive side it is known that dual to neighborly $d$ polytopes with $n$ facets are $\epsilon$-expanders for $\epsilon =n^{-4}$. This is far from the good expansion in the question but good enough to give a polynomial upper bound in $n$ for the diameter. Unfortunately general simple $d$ polytopes can be fairly poor expanders as seen by gluing two large simplicial polytopes along a facet and taking the dual.

share|cite|improve this answer
(Added a link to the Miller et al. paper.) – Joseph O'Rourke Jun 24 '13 at 0:35
I think I must have missed something in [MTTV97], but I didn't find an explicit construction in that paper, would you be more specific? Or maybe you mean a construction is possible using their result? – Hao Chen Nov 7 '14 at 19:36

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.