arXiv++ Combinatorics

This is to suggest a new approach to the old and open problem of counting the number f_n of Z-singular n x n matrices with entries from {-1,+1}: Comparison of two measures, none of them the uniform measure, one of them closely related to it, the other asymptotically under control by a recent theorem of Bourgain, Vu and Wood. We will define a measure P_chio on the set {-1,0,+1}^([n-1]^2) of all (n-1)x(n-1)-matrices with entries from {-1,0,+1} which (owing to a determinant identity published by M. F. Chio in 1853) is closely related to the uniform measures on {-1,+1}^([n]^2) and {0,1}^([n-1]^2) and at the same time it intriguingly mimics the so-called lazy coin flip distribution P_lcf on {-1,0,+1}^([n-1]^2), with the resemblance fading more and more as the events get smaller. This is relevant in view of a recent theorem of J. Bourgain, V. H. Vu and P. M. Wood (J. Funct. Anal. 258 (2010), 559--603) which proves that if the entries of an n x n matrix whose {-1,0,+1}-entries are governed by P_lcf and fully independent (they are not when governed by P_chio), then an asymptotically optimal bound on the singularity probability over Z can be proved. We will characterize P_chio graph-theoretically and use the characterization to prove that given a B in {-1,0,+1}^([n-1]^2), deciding whether P_chio[B] = P_lcf[B] is equivalent to deciding an evasive graph property, hence the time complexity of this decision is Omega(n^2). Moreover, we will prove k-wise independence properties of P_chio. Many questions suggest themselves that call for further work. In particular, the present paper will close with more constrained equivalent formulations of the conjecture f_n/2^(n^2) ~ (1/2 + o(1))^n.

The conjecture of Bollobás and Komlós, recently proved by Böttcher, Schacht, and Taraz [Math. Ann. 343(1), 175--205, 2009], implies that for any $γ>0$, every balanced bipartite graph on $2n$ vertices with bounded degree and sublinear bandwidth appears as a subgraph of any $2n$-vertex graph $G$ with minimum degree $(1+γ)n$, provided that $n$ is sufficiently large. We show that this threshold can be cut in half to an essentially best-possible minimum degree of $(\frac12+γ)n$ when we have the additional structural information of the host graph $G$ being balanced bipartite. This complements results of Zhao [to appear in SIAM J. Discrete Math.], as well as Hladký and Schacht [to appear in SIAM J. Discrete Math.], who determined a corresponding minimum degree threshold for $K_{r,s}$-factors, with $r$ and $s$ fixed. Moreover, it implies that the set of Hamilton cycles of $G$ is a generating system for its cycle space.

Let the Andrásfai graph $\mathrm{And}_k$ be defined as the graph with vertex set $\{v_0,v_1,...c, v_{3k-2}\}$ and two vertices $v_i$ and $v_j$ being adjacent iff $|i-j| \equiv 1\mod 3$. The graphs $\mathrm{And}_k$ are maximal triangle-free and play a role in characterizing triangle-free graphs with large minimum degree as homomorphic preimages. A minimal bipartification of a graph $G$ is defined as a set of edges $F\subset E(G)$ having the property that the graph $(V(G), E(G)\backslash F)$ is bipartite and for every $e \in F$ the graph $(V(G), E(G)\backslash (F\backslash e))$ is not bipartite. In this note it is shown that there is a minimal bipartification $F_k$ of $\mathrm{And}_k$ which consists of exactly $\lfloor\frac{k^2}{4}\rfloor$ edges. This equals $\lfloor{1/36}\bigl(|\mathrm{And}_k|+1\bigr)^2\rfloor$, where $|\cdot|$ denotes the number of vertices of a graph. For all $k$ this is consistent with a conjecture of Paul Erdős that every triangle-free graph $G$ can be made bipartite by deleting at most ${1/25} |G|^2$ edges. Bipartifications like $F_k$ may be useful for proving that arbitrary homomorphic preimages of an Andrásfai graph can be made bipartite by deleting at most ${1/25} |G|^2$ edges.