Improved Combinatorial Approximation Algorithms for MAX CUT in Sparse Graphs

The Max-Cut problem is a fundamental NP-hard problem, which is attracting attention in the field of quantum computation these days. Regarding the approximation algorithm of the Max-Cut problem, algorithms based on semidefinite programming have achieved much better approximation ratios than combinatorial algorithms. Therefore, filling the gap is an interesting topic as combinatorial algorithms also have some merits. In sparse graphs, there is a linear-time combinatorial algorithm with the approximation ratio $\frac{1}{2}+\frac{n-1}{4m}$ [Ngoc and Tuza, Comb. Probab. Comput. 1993], which is known as the Edwards-Erdős bound. In subcubic graphs, the combinatorial algorithm by Bazgan and Tuza [Discrete Math. 2008] has the best approximation ratio $\frac{5}{6}$ that runs in $O(n^2)$ time. Based on the approach by Bazgan and Tuza, we introduce a new vertex decomposition of graphs, which we call tree-bipartite decomposition. With the decomposition, we present a linear-time $(\frac{1}{2}+\frac{n-1}{2m})$-approximation algorithm for the Max-Cut problem. As a derivative, we also present a linear-time $\frac{5}{6}$-approximation algorithm in subcubic graphs, which solves an open problem in their paper.