New bounds on the maximum number of neighborly boxes in R^d

TL;DR

This paper establishes new upper and lower bounds on the maximum size of $k$-neighborly families of axis-aligned boxes in $ e^d$, advancing understanding of their combinatorial and geometric properties.

Contribution

It introduces improved bounds on $n(k,d)$, linking geometric configurations to graph coverings, and constructs explicit examples using structural properties like total laminations.

Findings

Upper bound: $n(k,d) o (2- ext{delta})^d$ for certain $k,d$

Lower bound: $n(k,d) o rac{d^k}{k!}$ asymptotically

Structural analysis of total laminations for explicit constructions

Abstract

A family of axis-aligned boxes in $\er^d$ is \emph{$k$-neighborly} if the intersection of every two of them has dimension at least $d-k$ and at most $d-1$. Let $n(k,d)$ denote the maximum size of such a family. It is known that $n(k,d)$ can be equivalently defined as the maximum number of vertices in a complete graph whose edges can be covered by $d$ complete bipartite graphs, with each edge covered at most $k$ times. We derive a new upper bound on $n(k,d)$, which implies, in particular, that $n(k,d)\leqslant (2-\delta)^d$ if $k\leqslant (1-\varepsilon)d$, where $\delta>0$ depends on arbitrarily chosen $\varepsilon>0$. The proof applies a classical result of Kleitman, concerning the maximum size of sets with a given diameter in discrete hypercubes. By an explicit construction we obtain also a new lower bound for $n(k,d)$, which implies that $n(k,d)\geqslant (1-o(1))\frac{d^k}{k!}$. We also study $k$-neighborly families of boxes with additional structural properties. Families called \emph{total laminations}, that split in a tree-like fashion, turn out to be particularly useful for explicit constructions. We pose a few conjectures based on these constructions and some computational experiments.