Article ID Journal Published Year Pages File Type
4651783 Electronic Notes in Discrete Mathematics 2013 8 Pages PDF
Abstract

We address a particular case of the quadratic minimum spanning tree problem in which interaction costs only apply for adjacent edges. Motivated by the fact that Gilmore-Lawler procedures in the literature underestimate the contribution of interaction costs to compute lower bounds, we introduce a reformulation that allows stronger linear programming bounds to be computed. An algorithm based on dynamic column and row generation is presented for evaluating these bounds. Our computational experiments indicate that the reformulation introduced here is indeed much stronger than those in the literature.

Related Topics
Physical Sciences and Engineering Mathematics Discrete Mathematics and Combinatorics