Optimal diameter computation within bounded clique-width graphs

Abstract

Coudert et al. (SODA'18) proved that under the Strong Exponential-Time Hypothesis, for any $\epsilon >0$, there is no ${\cal O}(2^{o(k)}n^{2-\epsilon})$-time algorithm for computing the diameter within the $n$-vertex cubic graphs of clique-width at most $k$. We present an algorithm which given an $n$-vertex $m$-edge graph $G$ and a $k$-expression, computes all the eccentricities in ${\cal O}(2^{{\cal O}(k)}(n+m)^{1+o(1)})$ time, thus matching their conditional lower bound. It can be modified in order to compute the Wiener index and the median set of $G$ within the same amount of time. On our way, we get a distance-labeling scheme for $n$-vertex $m$-edge graphs of clique-width at most $k$, using ${\cal O}(k\log^2{n})$ bits per vertex and constructible in ${\cal O}(k(n+m)\log{n})$ time from a given $k$-expression. Doing so, we match the label size obtained by Courcelle and Vanicat (DAM 2016), while we considerably improve the dependency on $k$ in their scheme. As a corollary, we get an ${\cal O}(kn^2\log{n})$-time algorithm for computing All-Pairs Shortest-Paths on $n$-vertex graphs of clique-width at most $k$. This partially answers an open question of Kratsch and Nelles (STACS'20).