Rainbow matchings in edge-colored simple graphs

There has been much research on the topic of finding a large rainbow matching (with no two edges having the same color) in a properly edge colored graph, where a proper edge coloring is a coloring of the edge set such that no same-colored edges are incident. Barat, Gyarfas, and Sarkozy conjectured that in every proper edge coloring of a multigraph (with parallel edges allowed, but not loops) with $2q$ colors where each color appears at least $q$ times, there is always a rainbow matching of size $q$. Recently, Aharoni, Berger, Chudnovsky, Howard, and Seymour proved a relaxation of the conjecture with $3q-2$ colors. Our main result proves that $2q + o(q)$ colors are enough if the graph is simple, confirming the conjecture asymptotically for simple graphs. This question restricted to simple graphs was considered before by Aharoni and Berger. We also disprove one of their conjectures regarding the lower bound on the number of colors one needs in the conjecture of Barat, Gyarfas, and Sarkozy for the class of simple graphs. Our methods are inspired by the randomized algorithm proposed by Gao, Ramadurai, Wanless, and Wormald to find a rainbow matching of size $q$ in a graph that is properly edge colored with $q$ colors, where each color class contains $q + o(q)$ edges. We consider a modified version of their algorithm, with which we are able to prove a generalization of their statement with a slightly better error term in $o(q)$. As a by-product of our techniques, we obtain a new asymptotic version of the Brualdi-Ryser-Stein Conjecture.