Nicos Christofides

Exact algorithms for the vehicle routing problem, based on spanning tree and shortest path relaxations

We consider the problem of routing vehicles stationed at a central facility (depot) to supply customers with known demands, in such a way as to minimize the total distance travelled. The problem is referred to as the vehicle routing problem (VRP) and is a generalization of the multiple travelling salesman problem that has many practical applications.We present tree search algorithms for the exact solution of the VRP incorporating lower bounds computed from (i) shortest spanningk-degree centre tree (k-DCT), and (ii)q-routes. The final algorithms also include problem reduction and dominance tests.Computational results are presented for a number of problems derived from the literature. The results show that the bounds derived from theq-routes are superior to those fromk-DCT and that VRPs of up to about 25 customers can be solved exactly.

Project scheduling with resource constraints: A branch and bound approach

Abstract This paper describes a branch and bound algorithm for project scheduling with resource constraints. The algorithmis based on the idea of using disjunctive arcs for resolving conflicts that are created whenever sets of activities have to be scheduled whose total resource requirements exceed the resource availabilities in some periods. Four lower bounds are examined. The first is a simple lower bound based on longest path computations. The second and third bounds are derived from a relaxed integer programming formulation of the problem. The second bound is based on the Linear Programming relaxation with the addition of cutting planes, and the third bound is based on a Lagrangean relaxation of the formulation. This last relaxation involves a problem which is a generalization of the longest path computation and for which an efficient, though not polynomial, algorithm is given. The fourth bound is based on the disjunctive arcs used to model the problem as a graph. We report computational results on the performance of each bound on randomly generated problems involving up to 25 activities and 3 resources.

The period routing problem

In this paper we present heuristic algorithms for the period vehicle routing problem, the problem of designing vehicle routes to meet required service levels for customers and minimize distribution costs over a given several-day period of time. These heuristic algorithms are based on an initial choice of customer delivery days which meet the service level requirements, followed by an interchange procedure in an attempt to minimize distribution costs. The heuristic algorithms represent distribution costs by replacing the vehicle routing problem for each day of the period by (I) a median problem and (II) a traveling salesman problem. Computational results and comparisons are given for the algorithms, based on test problems derived from the literature with up to 126 customers. The largest of these problems is the one given and solved by Russell and Igo. The solution obtained for this problem by the heuristic algorithms shows an improvement of 13% over the previous best solution. (Author/TRRL)

An algorithm for the resource constrained shortest path problem

In this paper we examine an integer programming formulation of the resource constrained shortest path problem. This is the problem of a traveller with a budget of various resources who has to reach a given destination as quickly as possible within the resource constraints imposed by his budget. A lagrangean relaxation of the integer programming formulation of the problem into a minimum cost network flow problem (which in certain circumstances reduces to an unconstrained shortest path problem) is developed which provides a lower bound for use in a tree search procedure. Problem reduction tests based on both the original problem and this lagrangean relaxation are given. Computational results are presented for the solution of problems involving up to 500 vertices, 5000 arcs, and 10 resources.

State‐space relaxation procedures for the computation of bounds to routing problems

It is well-known that few combinatorial optimization problems can be solved effectively by dynamic programming alone, since the number of vertices of the state space graph is enormous. What we are proposing here is a general relaxation procedure whereby the state-space associated with a given dynamic programming recursion is relaxed in such a way that the solution to the relaxed recursion provides a bound which could be embedded in general branch and bound schemes for the solution of the problem. This state space relaxation method is analogous to Langrangian relaxation in integer programming. This paper gives a survey of this new methodology, and gives, as examples, applications to the traveling salesman problem (TSP), the timeconstrained TSP and the vehicle routing problem (VRP). Valid state space relaxations are discussed for these problems and several bounds are derived in each case. Subgradient optimization and “state space ascent” are discussed as methods of maximizing the resulting lower bounds. More details of the procedures surveyed in this paper can be found in [2 ,3 ,41.

Distribution Management-Mathematical Modelling and Practical Analysis

Despite the extensive advances that have been made over the last twenty years in network analysis, our ability to design optimal networks in practical situations is often extremely limited. For some problems, in particular for transportation and logistics, this capability has, in fact, been almost nonexistent. To be sure, our accumulated experience in transport planning does enable us to avoid the most ineffective kinds of networks. However, this is quite different from knowing what kinds of networks are truly optimal. For many modes of transport, we do not even know which of significantly different networks are better. In air transport, for example,

An exact algorithm for the vehicle routing problem based on the set partitioning formulation with additional cuts

This paper presents a new exact algorithm for the Capacitated Vehicle Routing Problem (CVRP) based on the set partitioning formulation with additional cuts that correspond to capacity and clique inequalities. The exact algorithm uses a bounding procedure that finds a near optimal dual solution of the LP-relaxation of the resulting mathematical formulation by combining three dual ascent heuristics. The first dual heuristic is based on the q-route relaxation of the set partitioning formulation of the CVRP. The second one combines Lagrangean relaxation, pricing and cut generation. The third attempts to close the duality gap left by the first two procedures using a classical pricing and cut generation technique. The final dual solution is used to generate a reduced problem containing only the routes whose reduced costs are smaller than the gap between an upper bound and the lower bound achieved. The resulting problem is solved by an integer programming solver. Computational results over the main instances from the literature show the effectiveness of the proposed algorithm.

An exact algorithm for general, orthogonal, two-dimensional knapsack problems

Abstract We present a new exact tree-search procedure for solving two-dimensional knapsack problems in which a number of small rectangular pieces, each of a given size and value, are required to be cut from a large rectangular stock plate. The objective is to maximise the value of pieces cut or minimise the wastage. We consider the case where there is a maximum number of times that a piece may be used in a cutting pattern. The algorithm limits the size of the tree search by using a bound derived from a Langrangean relaxation of a 0–1 integer programming formulaton of the problem. Subgradient optimisation is used to optimise this bound. Reduction tests derived from both the original problem and the Lagrangean relaxation produce substantial computational gains. The computational performance of the algorithm indicates that it is an effective procedure capable of solving optimally practical two-dimensional cutting problems of medium size.

A restricted Lagrangean approach to the traveling salesman problem

We describe an algorithm for the asymmetric traveling salesman problem (TSP) using a new, restricted Lagrangean relaxation based on the assignment problem (AP). The Lagrange multipliers are constrained so as to guarantee the continued optimality of the initial AP solution, thus eliminating the need for repeatedly solving AP in the process of computing multipliers. We give several polynomially bounded procedures for generating valid inequalities and taking them into the Lagrangean function with a positive multiplier without violating the constraints, so as to strengthen the current lower bound. Upper bounds are generated by a fast tour-building heuristic. When the bound-strengthening techniques are exhausted without matching the upper with the lower bound, we branch by using two different rules, according to the situation: the usual subtour breaking disjunction, and a new disjunction based on conditional bounds. We discuss computational experience on 120 randomly generated asymmetric TSPs with up to 325 cities, the maximum time used for any single problem being 82 seconds. This is a considerable improvement upon earlier methods. Though the algorithm discussed here is for the asymmetric TSP, the approach can be adapted to the symmetric TSP by using the 2-matching problem instead of AP.

Extensions to a Lagrangean relaxation approach for the capacitated warehouse location problem

Abstract In this paper we present a lower bound for the capacitated warehouse location problem based upon the Lagrangean relaxation of a mixed-integer formulation of the problem, where we use subgradient optimisation in an attempt to maximise this lower bound. Problem reduction tests based upon this lower bound and the original problem are given. Incorporating this bound and the reduction tests into a tree search procedure enables us to solve problems involving up to 50 warehouses and 150 customers.