Spanning trees in randomly perturbed graphs

Abstract

A classical result of Komlós, Sárközy, and Szemerédi states that every n‐vertex graph with minimum degree at least (1/2 + o(1))n contains every n‐vertex tree with maximum degree . Krivelevich, Kwan, and Sudakov proved that for every n‐vertex graph Gα with minimum degree at least αn for any fixed α > 0 and every n‐vertex tree T with bounded maximum degree, one can still find a copy of T in Gα with high probability after adding O(n) randomly chosen edges to Gα. We extend the latter results to trees with (essentially) unbounded maximum degree; for a given and α > 0, we determine up to a constant factor the number of random edges that we need to add to an arbitrary n‐vertex graph with minimum degree αn in order to guarantee with high probability a copy of any fixed n‐vertex tree with maximum degree at most Δ.