A counterexample to Wegner's conjecture on good covers (Q664352)

The nerve \(N\) of a family \(\mathcal{U}\) of convex open sets in \(\mathbb{R}^d\) has trivial homology in degrees greater than or equal to \(d\). Moreover, \(N\) satisfies a condition called \textit{\(d\)-collapsibility} introduced by \textit{G. Wegner} [Arch. Math. 26, 317--321 (1975; Zbl 0308.52005)]. A simplicial complex is \(d\)-collapsible if we can obtain the empty complex by removing simplices of dimension smaller than or equal to \(d-1\) which are contained in a unique maximal simplex. The notion of \(d\)-collapsibility is related to Whitehead's classical collapses in the following way: If a complex \(K\) is \(d\)-collapsible, then it collapses to a subcomplex of dimension smaller than \(d\). If we replace the family \(\mathcal{U}\) by a family of open sets in which every non-empty intersection is homeomorphic to an open disk, then the nerve \(N\) still has trivial homology in high degrees. Wegner conjectured that \(N\) is \(d\)-collapsible also in this case. In the paper under review the author shows that Wegner's conjecture is false. He gives an explicit example for \(d=2\) with 19 open sets in the plane. A counterexample for dimension \(d+1\) is obtained from one in dimension \(d\) by considering cylinders \(U\times \mathbb{R}\) of the sets and adding two new disjoint sets \(\mathbb{R}^d \times V\), \(\mathbb{R}^d \times W\). The nerve of the new family is then the suspension of the nerve of the previous.