Bounded Representations of Interval and Proper Interval Graphs

TL;DR

This paper investigates the bounded representation problem for interval and proper interval graphs, providing efficient algorithms for some cases and highlighting complexity differences between related graph classes.

Contribution

It introduces linear and quadratic time algorithms for bounded representations and reveals surprising complexity distinctions between proper and unit interval graphs.

Findings

Linear time algorithm for interval graphs

Quadratic time algorithm for proper interval graphs

NP-completeness for bounded representations of unit interval graphs

Abstract

Klavik et al. [arXiv:1207.6960] recently introduced a generalization of recognition called the bounded representation problem which we study for the classes of interval and proper interval graphs. The input gives a graph G and in addition for each vertex v two intervals L_v and R_v called bounds. We ask whether there exists a bounded representation in which each interval I_v has its left endpoint in L_v and its right endpoint in R_v. We show that the problem can be solved in linear time for interval graphs and in quadratic time for proper interval graphs. Robert's Theorem states that the classes of proper interval graphs and unit interval graphs are equal. Surprisingly the bounded representation problem is polynomially solvable for proper interval graphs and NP-complete for unit interval graphs [Klav\'{\i}k et al., arxiv:1207.6960]. So unless P = NP, the proper and unit interval representations behave very differently. The bounded representation problem belongs to a wider class of restricted representation problems. These problems are generalizations of the well-understood recognition problem, and they ask whether there exists a representation of G satisfying some additional constraints. The bounded representation problems generalize many of these problems.