A simulated annealing approach for the capacitated minimum spanning tree problem

Abstract

Presents a novel approach to solve an NP-complete problem that is very important from the theoretical and practical point of view, namely the capacitated minimum spanning tree (CMST) problem. This approach has the following features: (a) it is based on the simulated annealing (SA) algorithm; (b) it represents a tree with N edges using N-2 integers; and (c) it defines a consistent ordering between feasible and infeasible trees. Our SA implementation was tested against the most referenced algorithms for the CMST: the Essau-Williams (1966) algorithm, the Prim (1957) algorithm and the Kruskal (1993) algorithm. The results indicate that our novel approach is very promising for solving CMST problem instances, because it consistently obtains the best results (for the tested cases) but it takes more time.