Diptesh Ghosh

Neighborhood search heuristics for the uncapacitated facility location problem

Abstract The uncapacitated facility location problem is one of choosing sites among a set of candidates in which facilities can be located, so that the demands of a given set of clients are satisfied at minimum costs. Applications of neighborhood search methods to this problem have not been reported in the literature. In this paper we first describe and compare several neighborhood structures used by local search to solve this problem. We then describe neighborhood search heuristics based on tabu search and complete local search with memory to solve large instances of the uncapacitated facility location problem. Our computational experiments show that on medium sized problem instances, both these heuristics return solutions with costs within 0.075% of the optimal with execution times that are often several orders of magnitude less than those required by exact algorithms. On large sized instances, the heuristics generate low cost solutions quite fast, and terminate with solutions whose costs are within 0.0345% of each other.

Branch and peg algorithms for the simple plant location problem

The simple plant location problem is a well-studied problem in combinatorial optimization. It is one of deciding where to locate a set of plants so that a set of clients can be supplied by them at the minimum cost. This problem often appears as a subproblem in other combinatorial problems. Several branch and bound techniques have been developed to solve these problems. In this paper we present two techniques that enhance the performance of branch and bound algorithms. The new algorithms thus obtained are called branch and peg algorithms, where pegging refers to fixing values of variables at each subproblem in the branch and bound tree, and is distinct from variable fixing during the branching process. We present exhaustive computational experiments which show that the new algorithms generate less than 60% of the number of subproblems generated by branch and bound algorithms, and in certain cases require less than 10% of the execution times required by branch and bound algorithms.

Tabu search for the single row facility layout problem using exhaustive 2-opt and insertion neighborhoods

The single row facility layout problem (SRFLP) is the problem of arranging facilities with given lengths on a line, while minimizing the weighted sum of the distances between all pairs of facilities. The problem is NP-hard. In this paper, we present two tabu search implementations, one involving an exhaustive search of the 2-opt neighborhood and the other involving an exhaustive search of the insertion neighborhood. We also present techniques to significantly speed up the search of the two neighborhoods. Our computational experiments show that the speed up techniques are effective, and our tabu search implementations are competitive. Our tabu search implementations improved previously known best solutions for 23 out of the 43 large sized SRFLP benchmark instances.

Complete Local Search with Memory

Neighborhood search heuristics like local search and its variants are some of the most popular approaches to solve discrete optimization problems of moderate to large size. Apart from tabu search, most of these heuristics are memoryless. In this paper we introduce a new neighborhood search heuristic that makes effective use of memory structures in a way that is different from that in common implementations of tabu search. We report computational experiments with this heuristic on the traveling salesperson problem and the subset sum problem.

A scatter search algorithm for the single row facility layout problem

The single row facility layout problem (SRFLP) is the problem of arranging facilities with given lengths on a line, with the objective of minimizing the weighted sum of the distances between all pairs of facilities. The problem is NP-hard and research has focused on heuristics to solve large instances of the problem. In this paper we present a scatter search algorithm to solve large size SRFLP instances. Our computational experiments show that the scatter search algorithm is an algorithm of choice when solving large size SRFLP instances within limited time.

Insertion based Lin-Kernighan heuristic for single row facility layout

The single row facility layout problem (SRFLP) is the problem of arranging facilities with given lengths on a line, while minimizing the weighted sum of the distances between all pairs of facilities. The problem is known to be NP-hard. In this paper, we present a neighborhood search heuristic called LK-INSERT which uses a Lin-Kernighan neighborhood structure built on insertion neighborhoods. To the best of our knowledge this is the first such heuristic for the SRFLP. Our computational experiments show that LK-INSERT is competitive for most instances, and it improves the best known solutions for several large sized benchmark SRFLP instances.

An efficient genetic algorithm for single row facility layout

The single row facility layout is the NP-Hard problem of arranging facilities with given lengths on a line, so as to minimize the weighted sum of the distances between all pairs of facilities. Owing to its computational complexity, researchers have developed several heuristics to obtain good quality solutions. In this paper, we present a genetic algorithm called GENALGO to solve large single row facility layout problem instances. Our algorithm uses standard genetic operators and periodically improves the fitness of all individuals. Our computational experiments show that our genetic algorithm yields high quality solutions in spite of starting with an initial population that is randomly generated. Our algorithm improves the previously best known solutions for the 19 instances of 58 benchmark instances and is competitive for most of the remaining ones.

Data aggregation for p-median problems

In this paper, we use a pseudo-Boolean formulation of the p-median problem and using data aggregation, provide a compact representation of p-median problem instances. We provide computational results to demonstrate this compactification in benchmark instances. We then use our representation to explain why some p-median problem instances are more difficult to solve to optimality than other instances of the same size. We also derive a preprocessing rule based on our formulation, and describe equivalent p-median problem instances, which are identical sized instances which are guaranteed to have identical optimal solutions.

Binary knapsack problems with random budgets

The binary knapsack problem is a combinatorial optimization problem in which a subset of a given set of elements needs to be chosen in order to maximize profit, given a budget constraint. In this paper, we study a stochastic version of the problem in which the budget is random. We propose two different formulations of this problem, based on different ways of handling infeasibility, and propose an exact algorithm and a local search-based heuristic to solve the problems represented by these formulations. We also present the results from some computational experiments.

A competitive local search heuristic for the subset sum problem

Subset sum problems are a special class of difficult singly constrained zero–one integer programming problems. Several heuristics for solving these problems have been reported in the literature. In this paper we propose a new heuristic based on local search which improves upon the previous best. Scope and purpose Subset sum problems (SSP) are a widely studied class of integer programming problems. These frequently occur as subproblems in other integer programs. They have been used for calculating bounds for general integer programs and power indices of players in cooperative voting games. In this paper we present a competitive heuristic for approximately solving this problem.