Solving Hamiltonian Cycle by an EPT Algorithm for a Non-sparse Parameter

TL;DR

This paper demonstrates that the Hamiltonian Cycle problem can be solved in fixed-parameter tractable (FPT) and even efficiently in exponential time (EPT) when parameterized by split-matching-width, a graph parameter between treewidth and clique-width.

Contribution

It introduces an EPT algorithm for Hamiltonian Cycle parameterized by split-matching-width and provides a method to approximate this parameter, advancing algorithms for dense graphs.

Findings

Hamiltonian Cycle is in EPT when parameterized by split-matching-width.

An algorithm for approximating split-matching-width within a constant factor.

Optimality of algorithms under the Exponential Time Hypothesis.

Abstract

Many hard graph problems, such as Hamiltonian Cycle, become FPT when parameterized by treewidth, a parameter that is bounded only on sparse graphs. When parameterized by the more general parameter clique-width, Hamiltonian Cycle becomes W[1]-hard, as shown by Fomin et al. [5]. S{\ae}ther and Telle address this problem in their paper [13] by introducing a new parameter, split-matching-width, which lies between treewidth and clique-width in terms of generality. They show that even though graphs of restricted split-matching-width might be dense, solving problems such as Hamiltonian Cycle can be done in FPT time. Recently, it was shown that Hamiltonian Cycle parameterized by treewidth is in EPT [1, 6], meaning it can be solved in $n^{O(1)} 2^{O(k)}$-time. In this paper, using tools from [6], we show that also parameterized by split-matching-width Hamiltonian Cycle is EPT. To the best of our knowledge, this is the first EPT algorithm for any "globally constrained" graph problem parameterized by a non-trivial and non-sparse structural parameter. To accomplish this, we also give an algorithm constructing a branch decomposition approximating the minimum split-matching-width to within a constant factor. Combined, these results show that the algorithms in [13] for Edge Dominating Set, Chromatic Number and Max Cut all can be improved. We also show that for Hamiltonian Cycle and Max Cut the resulting algorithms are asymptotically optimal under the Exponential Time Hypothesis.