Perfect Graph Modification Problems: An Integer Programming Approach

TL;DR

This paper introduces integer programming methods and heuristics for solving perfect graph modification problems, including editing, completion, and sandwich problems, with practical algorithms and empirical validation.

Contribution

It develops exact integer programming formulations and cutting plane algorithms for NP-hard perfect graph modification problems, enhancing solution efficiency and practical applicability.

Findings

Effective integer programming models for perfect graph problems.

Cutting plane algorithms efficiently find solutions in practice.

Heuristic methods improve upper bounds and solution quality.

Abstract

Graph modification problems, which aim to find a small set of modifications to a graph so that it satisfies a desired property, have been studied for several special graph classes. The literature is rather rich in NP-completeness results and polynomial time solvable cases. However, to the best of our knowledge, only a few exact algorithms have been suggested to address NP-hard cases. In this work, we propose exact solution methods based on integer programming for three perfect graph modification problems: minimum perfect editing, minimum perfect completion and the perfect sandwich problem. The minimum perfect editing problem inquires the smallest number of edge additions and deletions to make a graph perfect, while the completion problem allows only edge additions. In the perfect sandwich problem, only a given subset of non-edges can be changed to edges, and the problem asks whether a perfect graph can be obtained in this way. The proposed methods are based on the Strong Perfect Graph Theorem. We represent odd holes and odd antiholes as linear inequalities, and formulate an integer programming model to solve minimum perfect editing problem. To address the exponential number of constraints, we propose a cutting plane algorithm which relies on finding odd holes and odd antiholes. To enhance the practical efficiency of the cutting plane algorithm, we address the expected number of odd holes and odd antiholes in random graphs. In addition, we propose a heuristic algorithm to make a given graph perfect, which is used to obtain improved upper bounds for the editing and the completion problems. Finally, we demonstrate empirical effectiveness of the proposed methods through computational experiments.