KAIST Discrete Math Seminar

The chromatic number of a graph is the minimum k such that the graph has a proper k-coloring. It is known that if T is a tree, then every graph with large chromatic number contains T as a subgraph. In this talk, we discuss this phenomena for star-coloring (a proper coloring forbidding a bicolored path on four vertices) and acyclic-coloring (a proper coloring forbidding bicolored cycles). Specifically, we will characterize all graphs T such that every graph with sufficiently large star-chromatic number (acyclic-chromatic number) contains T as a subgraph.

A chordless cycle in a graph G is an induced subgraph of G which is a cycle of length at least four. We prove that the Erdős-Pósa property holds for chordless cycles, which resolves the major open question concerning the Erdős-Pósa property. Our proof for chordless cycles is constructive: in polynomial time, one can find either k+1 vertex-disjoint chordless cycles, or ck² log k vertices hitting every chordless cycle for some constant c. It immediately implies an approximation algorithm of factor O(OPT log OPT) for Chordal Vertex Deletion. We complement our main result by showing that chordless cycles of length at least ℓ for any fixed ℓ≥ 5 do not have the Erdős-Pósa property.

We prove that a class of graphs obtained by gluing complete multipartite graphs in a tree-like way satisfies a conjecture of Kohayakawa, Nagle, Rödl, and Schacht on random-like counts for small graphs in locally dense graphs. This implies an approximate version of the conjecture for graphs with bounded tree-width. We also prove an analogous result for odd cycles instead of complete multipartite graphs.
The proof uses a general information theoretic method to prove graph homomorphism inequalities for tree-like structured graphs, which may be of independent interest.

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 at most O(n/log n) as a subgraph, and the bounds on the degree conditions are sharp.
On the other hand, Krivelevich, Kwan and Sudakov recently proved that for every n-vertex graph G 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 G with high probability after adding O(n) randomly-chosen edges to G.
We extend this result to trees with unbounded maximum degree. More precisely, for a given n^ε ≤ Δ≤ cn/log n and α>0, we determined the precise number (up to a constant factor) of random edges that we need to add to an arbitrary n-vertex graph G with minimum degree αn in order to guarantee with high probability a copy of any fixed T with maximum degree at most Δ. This is joint work with Felix Joos.