The Complexity of Recognizing Geometric Hypergraphs

As set systems, hypergraphs are omnipresent and have various representations. In a geometric representation of a hypergraph $H=(V,E)$, each vertex $v\in V$ is a associated with a point $p_v\in \mathbb{R}^d$ and each hyperedge $e\in E$ is associated with a connected set $s_e\subset \mathbb{R}^d$ such that $\{p_v\mid v\in V\}\cap s_e=\{p_v\mid v\in e\}$ for all $e\in E$. We say that a given hypergraph $H$ is representable by some (infinite) family $\mathcal{F}$ of sets in $\mathbb{R}^d$, if there exist $P\subset \mathbb{R}^d$ and $S \subseteq \mathcal{F}$ such that $(P,S)$ is a geometric representation of $H$. For a family $\mathcal{F}$, we define RECOGNITION($\mathcal{F}$) as the problem to determine if a given hypergraph is representable by $\mathcal{F}$. It is known that the RECOGNITION problem is ER-hard for halfspaces in $\mathbb{R}^d$. We study the families of balls and ellipsoids in $\mathbb{R}^d$, as well as other convex sets, and show that their RECOGNITION problems are also ER-complete. This means that these recognition problems are equivalent to deciding whether a multivariate system of polynomial equations with integer coefficients has a real solution.