Filipe Pereira e Alvelos

New mixed integer-programming model for the pickup-and-delivery problem with transshipment

In recent years, many important real-world applications are studied as “rich” vehicle routing problems that are variants and generalizations of the well-known vehicle routing problem. In this paper we address the pickup-and-delivery version of this problem and consider further generalization by allowing transshipment in the network. Moreover, we allow heterogenous vehicles and flexible fleet size. We describe mixed integer-programming formulations for the problem with and without time windows for services. The number of constraints and variables in the models are bounded by polynomial size of the problem. We discuss several problem variants that are either captured by our models or can be easily captured through simple modifications. Computational work gave promising results and confirms that transshipment in network can indeed enhance optimization.

Comparing Dantzig-Wolfe decompositions and branch-and-price algorithms for the multi-item capacitated lot sizing problem

In this article, we consider the multi-item capacitated lot sizing problem with setup times. Starting from an original mixed integer programming model, we apply the standard Dantzig–Wolfe decomposition (DWD) in two different ways: defining the subproblems by items and defining the subproblems by periods. A third decomposition is developed in which the subproblems of both types are integrated in the same model. The linear relaxation of this last approach, which we denote as multiple DWD, provides lower bounds (equal to or) better than the bounds obtained by the other decompositions, which in turn, provide lower bounds (equal to or) better than the ones given by the original model. For solving the three decomposition models, we implemented three branch-and-price algorithms. We describe their main aspects and report on their computational results in instances from the literature.

Traffic Engineering of Multiple Spanning Tree Routing Networks: the Load Balancing Case

This paper deals with optimal load balancing in telecommunication networks based on multiple spanning tree routing. This is the case in switched Ethernet networks where the operator configures different routing spanning trees and assigns each demand VLAN to one of the spanning trees. We consider modeling and solving three load balancing objectives: (i) minimization of the average link load with a guaranteed optimal worst case link load, (ii) minimization of the worst case link load with a guaranteed optimal average link load and (iii) the min-max optimization of link loads. We also propose heuristic techniques to compute both feasible solutions and lower bounds for the addressed optimization problems. Finally, we assess both the efficiency and the efficacy of the different solution techniques and compare the quality of each problem solution taking into account the optimization criteria of the other problems.

Sequence based heuristics for two-dimensional bin packing problems

This article addresses several variants of the two-dimensional bin packing problem. In the most basic version of the problem it is intended to pack a given number of rectangular items with given sizes in rectangular bins in such a way that the number of bins used is minimized. Different heuristic approaches (greedy, local search, and variable neighbourhood descent) are proposed for solving four guillotine two-dimensional bin packing problems. The heuristics are based on the definition of a packing sequence for items and in a set of criteria for packing one item in a current partial solution. Several extensions are introduced to deal with issues pointed out by two furniture companies. Extensive computational results on instances from the literature and from the two furniture companies are reported and compared with optimal solutions, solutions from other five (meta)heuristics and, for a small set of instances, with the ones used in the companies.

A branch-and-price approach for harvest scheduling subject to maximum area restrictions

Recently, research on exact methods has been undertaken to solve forest management problems subject to constraints on the maximum clearcut area by using the area restriction model approach. Three main basic integer programming models for these problems have been discussed in the literature: the so-called cluster, path and bucket formulations. Solving these models via branch-and-bound, where all variables and constraints are used a priori, is adequately suited for real problems of a small to medium size, but is not appropriate for larger problems. In this paper, we describe a branch-and-price approach for the cluster model, and we show that this formulation dominates the bucket model, by completing the results of the dominance relationships between the bounds of the three models. Branch-and-price was tested on real and hypothetical forests ranging from 45 to 2945 stands and temporal horizons ranging from three to twelve periods were employed. Results show that the solutions obtained by the proposed approach stood within 1% of the optimal solution and were achieved in a short computation time. It was found that branch-and-bound was unable to produce solutions for most forests from 850 stands with either eleven or an average number of stands per clearcut greater or equal than eight.

Integrating two-dimensional cutting stock and lot-sizing problems

The two-dimensional cutting stock problem (2DCSP) consists in the minimization of the number of plates used to cut a set of items. In industry, typically, an instance of this problem is considered at the beginning of each planning time period, what may result in solutions of poor quality, that is, excessive waste, when a set of planning periods is considered. To deal with this issue, we consider an integrated problem, in which the 2DCSP is extended from the solution in only a single production planning period to a solution in a set of production planning periods. The main difference of the approach in this work and the ones in the literature is to allow sufficiently large residual plates (leftovers) to be stored and cut in a subsequent period of the planning horizon, which may further help in the minimization of the waste. We propose two integrated integer programming models to optimize the combined two-dimensional cutting stock and lot-sizing problems, minimizing the total cost, which includes material, waste and storage costs. Two heuristics based on the industrial practice to solve the problem were also presented. Computational results for the proposed models and for the heuristics are presented and discussed.

A new branch-and-price approach for the kidney exchange problem

The kidney exchange problem (KEP) is an optimization problem arising in the framework of transplant programs that allow exchange of kidneys between two or more incompatible patient-donor pairs. In this paper an approach based on a new decomposition model and branch-and-price is proposed to solve large KEP instances. The optimization problem considers, hierarchically, the maximization of the number of transplants and the minimization of the size of exchange cycles. Computational comparison of different variants of branch-and-price for the standard and the proposed objective functions are presented. The results show the efficiency of the proposed approach for solving large instances.

Combining column generation and metaheuristics

In this Chapter, we consider the hybridization of column generation (CG) with metaheuristics (MHs) for solving integer programming and combinatorial optimization problems. We describe a general framework entitled ”metaheuristic search by column generation” (for short, SearchCol). CG is a decomposition approach in which one linear programming master problem interacts with subproblems to obtain an optimal solution to a relaxed version of a problem. The subproblems may be solved by problem-specific algorithms. After CG is applied, a set of subproblem’s solutions, optimal primal and dual values of the master problem variables and a lower bound to the optimal value of the problem are available. In contrast with enumerative approaches (e.g, branch-and-price), in SearchCol the information provided by CG is used in a MH search. The search is based on representing a solution (to the overall problem) as being composed by one solution from each subproblem. After a search is conducted, a perturbation for CG is defined and a new iteration begins. The perturbation consists in forcing or forbidding attributes of the subproblem’s solutions and, in general, leads to the generation of new subproblem’s solutions and different optimal primal and dual values of the master problem variables. In this Chapter, we discuss (i) which models are suitable for decomposition approaches as SearchCol, (ii) different alternatives for generating initial solutions for the search (with different degrees of randomization, greediness and influence of CG) (iii) different search approaches based on local search, (iv) different alternatives for perturbing CG (influenced by CG, based on the incumbent, and based on the memory of the search).

A hybrid column generation with GRASP and path relinking for the network load balancing problem

In this paper, a hybrid meta-heuristic is proposed which combines the GRASP with path relinking method and Column Generation. The key idea of this method is to run a GRASP with path relinking search on a restricted search space, defined by Column Generation, instead of running the search on the complete search space of the problem. Moreover, column generation is used not only to compute the initial restricted search space but also to modify it during the whole algorithm. The proposed heuristic is used to solve the network load balancing problem: given a capacitated telecommunications network with single path routing and an estimated traffic demand matrix, the network load balancing problem is the determination of a routing path for each traffic commodity such that the network load balancing is optimized, i.e., the worst link load is minimized, among all such solutions, the second worst link load is minimized, and continuing in this way until all link loads are minimized. The computational results presented in this paper show that, for the network load balancing problem, the proposed heuristic is effective in obtaining better quality solutions in shorter running times.

Optimizing network load balancing: an hybridization approach of metaheuristics with column generation

Given a capacitated telecommunications network with single path routing and an estimated traffic demand matrix, we aim to determine the routing path of each traffic commodity such that the whole set of paths provide an optimal network load balancing. In a recent paper, we have proposed a column generation based heuristic where, in the first step, we use column generation to solve a linear programming relaxation of the original problem (obtaining, in this way, a lower bound and a set of paths for each commodity) and, in the second step, we apply a multi-start local search with path relinking heuristic on the solution space defined by the paths of the first step. Here, we propose a hybridization approach of the metaheuristic with column generation that can be seen as an enhanced version of the previous approach: we run column generation not only at the beginning (to define the initial search space) but also during the search. These additional column generation steps consist in solving a perturbed problem defined by the incumbent solution. In the previous paper, we have shown that the first approach is efficient in obtaining near optimal routing solutions within short running times. With the enhanced version, we show through computational results that the additional paths, introduced by the additional column generation steps, either improve the efficiency of the algorithm or show similar efficiency in the cases where the original algorithm is already very efficient.