Producción CyT
Tesis
Autoría
Fecha
01/01/2012
Resumen
Información suministrada por el agente en
SIGEVA
The Time Dependent Travelling Salesman Problem (TDTSP) is a generalization of the well known Travelling Salesman Problem (TSP) in which we look for a minimum cost tour that visits each client exactly once where the travel times (or costs) among them are not assumed to be constant and may vary depending on many different factors. The motivation to consider the time dependence factor is that it enables to have better approximations to many problems from practice, mainly from the industry and the ...
The Time Dependent Travelling Salesman Problem (TDTSP) is a generalization of the well known Travelling Salesman Problem (TSP) in which we look for a minimum cost tour that visits each client exactly once where the travel times (or costs) among them are not assumed to be constant and may vary depending on many different factors. The motivation to consider the time dependence factor is that it enables to have better approximations to many problems from practice, mainly from the industry and the service sector. In the related literature there are different versions of the TDTSP that consider that variations occur for different reasons and over distinct parameters. For example, some of them consider that variations affect travel times (and/or costs) while others assume that variations influence travel speeds. Moreover, in the first case, one version assumes that the travel time depends on the position of the arc in the tour while another one assumes that it depends on the particular starting instant for travelling the arc. As well as the TSP, the TDTSP belongs to the class N P-Hard and therefore it is not known an algorithm that solves it in polynomial time. In this thesis we approach two of the versions mentioned above by means of Integer Linear Programming formulations. For each formulation, we perform a theoretical study focused on deriving families of valid inequalities that exploit the particular characteristics of the problem. In particular, for one of the variants, we prove that some of them are facet defining. From a practical standpoint, for the versions considered we develop exact Branch and Cut algorithms that incorporate these families of valid inequalities and we evaluate them on instances from the related literature, obtaining good computational results that show that both approaches are feasible to be used in practice.
Ver más
Ver menos
Palabras Clave
INTEGER PROGRAMMINGBRANCH AND CUTTIME-DEPENDENT TSP