Core Concepts
This paper presents efficient approximation and fixed-parameter tractable (FPT) algorithms for finding minimum DM-irreducible spanning subgraphs in bipartite graphs, which generalizes the well-known strongly connected spanning subgraph problem.
Abstract
The paper considers the DM-Irreducible Spanning Subgraph (DMISS) problem, which is a generalization of the Strongly Connected Spanning Subgraph (SCSS) problem. In DMISS, the goal is to find a minimum-weight spanning subgraph of a given DM-irreducible balanced bipartite graph that is also DM-irreducible.
The key results are:
A 2-approximation algorithm for DMISS that runs in O(n^3) time, by extending the 2-approximation algorithm for SCSS by Frederickson and Jája.
An FPT algorithm for the unweighted version of DMISS, parameterized by the difference from the trivial upper bound of the optimal value, by extending the FPT algorithm for the unweighted version of SCSS by Bang-Jensen and Yeo.
The approximation algorithm works by first finding a minimum-weight perfect matching, and then finding a minimum-weight spanning in-arborescence and out-arborescence in the auxiliary graph. The FPT algorithm uses an odd proper ear decomposition of the input graph to identify a small vertex set that determines the structure of any minimum DM-irreducible spanning subgraph.