Minors in random and expanding hypergraphs

Uli Wagner
2011 Proceedings of the 27th annual ACM symposium on Computational geometry - SoCG '11  
We introduce a new notion of minors for simplicial complexes (hypergraphs), so-called homological minors. Our motivation is to propose a general approach to attack certain extremal problems for sparse simplicial complexes and the corresponding threshold problems for random complexes. In this paper, we focus on threshold problems. The basic model for random complexes is the Linial-Meshulam model X k (n, p). By definition, such a complex has n vertices, a complete (k − 1)-dimensional skeleton,
more » ... every possible k-dimensional simplex is chosen independently with probability p. We show that for every k, t ≥ 1, there is a constant C = C(k, t) such that for p ≥ C/n, the random complex X k (n, p) asymptotically almost surely contains K k t (the complete k-dimensional complex on t vertices) as a homological minor. As corollary, the threshold for (topological) embeddability of X k (n, p) into R 2k is at p = Θ(1/n). The method can be extended to other models of random complexes (for which the lower skeleta are not necessarily complete) and also to more general Tverberg-type problems, where instead of continuous maps without doubly covered image points (embeddings), we consider maps without qfold covered image points. 1 Even the dependence of the constants on t is now known quite precisely, C(t) ≈ 0.319t √ log t and CTOP(t) = Θ(t 2 ), see [45] and [5, 21], respectively. In the special case t = 5, Mader [31] also showed that 3n−5 edges are already enough to ensure K5 as a (topological) minor.
doi:10.1145/1998196.1998256 dblp:conf/compgeom/Wagner11 fatcat:unq53imjuramndjovjaxwkgdbq