Tight triangulations of closed 3-manifolds

TL;DR

This paper characterizes tight triangulations of closed 3-manifolds over fields of odd characteristic as neighborly, orientable, and stacked, and explores properties and bounds related to tight triangulations over fields of characteristic two.

Contribution

It provides a complete characterization of tight triangulations in dimension three for odd characteristic fields and establishes bounds and conditions for characteristic two cases.

Findings

Tight triangulations are neighborly, orientable, and stacked in odd characteristic fields.

Converse holds for certain vertex counts and Betti numbers in characteristic two.

An upper bound theorem relates vertices, Betti number, and tightness conditions.

Abstract

It is well known that a triangulation of a closed 2-manifold is tight with respect to a field of characteristic two if and only if it is neighbourly; and it is tight with respect to a field of odd characteristic if and only if it is neighbourly and orientable. No such characterization of tightness was previously known for higher dimensional manifolds. In this paper, we prove that a triangulation of a closed 3-manifold is tight with respect to a field of odd characteristic if and only if it is neighbourly, orientable and stacked. In consequence, the K\"{u}hnel-Lutz conjecture is valid in dimension three for fields of odd characteristic. Next let $\mathbb{F}$ be a field of characteristic two. It is known that, in this case, any neighbourly and stacked triangulation of a closed 3-manifold is $\mathbb{F}$-tight. For triangulated closed 3-manifolds with at most 71 vertices or with first Betti number at most 188, we show that the converse is true. But the possibility of an $\mathbb{F}$-tight non-stacked triangulation on a larger number of vertices remains open. We prove the following upper bound theorem on such triangulations. If an $\mathbb{F}$-tight triangulation of a closed 3-manifold has $n$ vertices and first Betti number $\beta_1$, then $(n-4)(617n- 3861) \leq 15444\beta_1$. Equality holds here if and only if all the vertex links of the triangulation are connected sums of boundary complexes of icosahedra.