Rainbow Colouring of Split Graphs

A rainbow path in an edge coloured graph is a path in which no two edges are coloured the same. A rainbow colouring of a connected graph G is a colouring of the edges of G such that every pair of vertices in G is connected by at least one rainbow path. The minimum number of colours required to rainbow colour G is called its rainbow connection number. Between them, Chakraborty et al. [J. Comb. Optim., 2011] and Ananth et al. [FSTTCS, 2012] have shown that for every integer k, k \geq 2, it is NP-complete to decide whether a given graph can be rainbow coloured using k colours. A split graph is a graph whose vertex set can be partitioned into a clique and an independent set. Chandran and Rajendraprasad have shown that the problem of deciding whether a given split graph G can be rainbow coloured using 3 colours is NP-complete and further have described a linear time algorithm to rainbow colour any split graph using at most one colour more than the optimum [COCOON, 2012]. In this article, we settle the computational complexity of the problem on split graphs and thereby discover an interesting dichotomy. Specifically, we show that the problem of deciding whether a given split graph can be rainbow coloured using k colours is NP-complete for k \in {2,3}, but can be solved in polynomial time for all other values of k.
Comment: This is the full version of a paper to be presented at ICGT 2014. This complements the results in arXiv:1205.1670 (which were presented in COCOON 2013), and both will be merged into a single journal submission