Bjørn Petersen

Subset-Row Inequalities Applied to the Vehicle-Routing Problem with Time Windows

This paper presents a branch-and-cut-and-price algorithm for the vehicle-routing problem with time windows. The standard Dantzig-Wolfe decomposition of the arc flow formulation leads to a set-partitioning problem as the master problem and an elementary shortest-path problem with resource constraints as the pricing problem. We introduce the subset-row inequalities, which are Chvatal-Gomory rank-1 cuts based on a subset of the constraints in the master problem. Applying a subset-row inequality in the master problem increases the complexity of the label-setting algorithm used to solve the pricing problem because an additional resource is added for each inequality. We propose a modified dominance criterion that makes it possible to dominate more labels by exploiting the step-like structure of the objective function of the pricing problem. Computational experiments have been performed on the Solomon benchmarks where we were able to close several instances. The results show that applying subset-row inequalities in the master problem significantly improves the lower bound and, in many cases, makes it possible to prove optimality in the root node.

Chvátal-Gomory Rank-1 Cuts Used in a Dantzig-Wolfe Decomposition of the Vehicle Routing Problem with Time Windows

This chapter shows how Chvatal-Gomory (CG) rank-1 cuts can be used in a Branch-and-Cut-and-Price algorithm for the Vehicle Routing Problem with Time Windows (VRPTW). Using Dantzig-Wolfe decomposition we split the problem into a Set Partitioning Problem as master problem and an Elementary Shortest Path Problem with Resource Constraints as pricing problem. To strengthen the formulation we derive general CG rank-1 cuts based on the master problem formulation. Adding these cuts to the master problem means that an additional resource is added to the pricing problem for each cut. This increases the complexity of the label algorithm used to solve the pricing problem since normal dominance tests become weak when many resources are present and hence most labels are incomparable. To overcome this problem we present a number of improved dominance tests exploiting the step-like structure of the objective function of the pricing problem. Computational experiments are reported on the Solomon test instances showing that the addition of CG rank-1 cuts improves the lower bounds significantly and makes it possible to solve a majority of the instances in the root node of the branch-and-bound tree. This indicates that CG rank-1 cuts may be essential for solving future large-scale VRPTW problems where we cannot expect that the branching process will close the gap between lower and upper bounds in reasonable time.

Investigating solvability and complexity of linear active networks by means of matroids

The solvability and complexity problems of finear active network are approached from a purely combinatorial point of view, using the concepts of matroid theory. Since the method is purely combinatorial, we take into account the network topology alone. Under this assumption necessary and sufficient conditions are given for the unique solvablity of linear active networks. The complexity and the number of dc-eigenfrequencies are also given. The method enables.you to decide if degeneracies are due to the topology alone, or if they are caused by special relations among network parameter values. If the network parameter values are taken into account, the complexity and number of dc-eigenfrequencies given by the method, are only upper and lower bounds, respectively. The above conditions are fairly easily checked, and the complexity and number of dc-elgenfrequencies are found, using polynomially bounded algorithms (matroid partition and intersection algorithms).

A branch-and-cut algorithm for the capacitated profitable tour problem

This paper considers the Capacitated Profitable Tour Problem (CPTP) which is a special case of the Elementary Shortest Path Problem with Resource Constraints (ESPPRC). The CPTP belongs to the group of problems known as traveling salesman problems with profits. In CPTP each customer is associated with a profit and a demand and the objective is to find a capacitated tour (rooted in a depot node) that minimizes the total travel distance minus the profit of the visited customers. The CPTP can be recognized as the sub-problem in many column generation applications, where it is traditionally solved through dynamic programming. In this paper we present an alternative framework based on a formulation for the undirected CPTP and solved through branch-and-cut. Valid inequalities are presented among which we introduce a new family of inequalities for the CPTP denoted rounded multistar inequalities and we prove their validity. Computational experiments are performed on a set of instances known from the literature and a set of newly generated instances. The results indicate that the presented algorithm is highly competitive with the dynamic programming algorithms. In particular, we are able to solve instances with 800 nodes to optimality where the dynamic programming algorithms cannot solve instances with more than 200 nodes. Moreover dynamic programming and branch-and-cut complement each other well, giving us hope for solving more general problems through hybrid approaches. The paper is intended to serve as a platform for further development of branch-and-cut algorithms for CPTP hence also acting as a survey/tutorial.

A solution approach based on Benders decomposition for the preventive maintenance scheduling problem of a stochastic large-scale energy system

This paper describes a Benders decomposition-based framework for solving the large scale energy management problem that was posed for the ROADEF 2010 challenge. The problem was taken from the power industry and entailed scheduling the outage dates for a set of nuclear power plants, which need to be regularly taken down for refueling and maintenance, in such a way that the expected cost of meeting the power demand in a number of potential scenarios is minimized. We show that the problem structure naturally lends itself to Benders decomposition; however, not all constraints can be included in the mixed integer programming model. We present a two phase approach that first uses Benders decomposition to solve the linear programming relaxation of a relaxed version of the problem. In the second phase, integer solutions are enumerated and a procedure is applied to make them satisfy constraints not included in the relaxed problem. To cope with the size of the formulations arising in our approach we describe efficient preprocessing techniques to reduce the problem size and show how aggregation can be applied to each of the subproblems. Computational results on the test instances show that the procedure competes well on small instances of the problem, but runs into difficulty on larger ones. Unlike heuristic approaches, however, this methodology can be used to provide lower bounds on solution quality.