A Parallel Grasp for the Steiner Tree Problem in Graphs Using a Hybrid Local Search Strategy

In this paper, we present a parallel greedy randomized adaptive search procedure (GRASP) for the Steiner problem in graphs. GRASP is a two-phase metaheuristic. In the first phase, solutions are constructed using a greedy randomized procedure. Local search is applied in the second phase, leading to a local minimum with respect to a specified neighborhood. In the Steiner problem in graphs, feasible solutions can be characterized by their non-terminal nodes (Steiner nodes) or by their key-paths. According to this characterization, two GRASP procedures are described using different local search strategies. Both use an identical construction procedure. The first uses a node-based neighborhood for local search, while the second uses a path-based neighborhood. Computational results comparing the two procedures show that while the node-based variant produces better quality solutions, the path-based variant is about twice as fast. A hybrid GRASP procedure combining the two neighborhood search strategies is then proposed. Computational experiments with a parallel implementation of the hybrid procedure are reported, showing that the algorithm found optimal solutions for 45 out of 60 benchmark instances and was never off by more than 4% of the optimal solution value. The average speedup results observed for the test problems show that increasing the number of processors reduces elapsed times with increasing speedups. Moreover, the main contribution of the parallel algorithm concerns the fact that larger speedups of the same order of the number of processors are obtained exactly for the most difficult problems.     

A GRASP and Path Relinking Heuristic for Rural Road Network Development

This paper presents a model for rural road network design that involves two objectives: maximize all season road connectivity among villages in a region and maximize route efficiency, while allocating a fix budget among a number of possible road projects. The problem is modeled as a bicriterion optimization problem and solved heuristically through a greedy randomized adaptive search procedure (GRASP) in conjunction with a path relinking procedure. The implementation of GRASP and path relinking includes two novel modifications, a new form of reactive GRASP and a new form of path relinking. Overall, the heuristic approach is streamlined through the incorporation of advanced network flow reoptimization techniques. Results indicate that this implementation outperforms both GRASP as well as a straightforward form of GRASP with path relinking. For small problem instances, for which optimality could be verified, this new, modified form of GRASP with path relinking solved all but one known instance optimally.     

Piecewise Linear Time Series Estimation with GRASP

This paper describes a heuristic to build piecewise linear statistical models with multivariate thresholds, based on a Greedy Randomized Adaptive Search Procedure (GRASP). GRASP is an iterative randomized sampling technique that has been shown to quickly produce good quality solutions for a wide variety of optimization problems. In this paper we describe a GRASP to sequentially split an n-dimensional space in order to build a piecewise linear time series model.     

Grasp Embedded Scatter Search for the Multicommodity Capacitated Network Design Problem

A GRASP embedded Scatter Search is developed for the multicommodity capacitated network design problem. Difficulty for this problem arises from the fact that selection of the optimal network design is an NP-complete combinatorial problem. There exist no polynomial exact algorithms which can solve this problem in a reasonable period of time for realistically sized instances. In such cases, heuristic procedures are commonly used. Two strategies were designed for GRASP: a traditional approach and a memory based technique. As for Scatter Search, 5 different strategies were used to update the reference set. Computational results on a large set of randomly generated instances show the convenience of the proposed procedures.     

Adaptive and multi-mining versions of the DM-GRASP hybrid metaheuristic

Metaheuristics represent an important class of techniques to solve, approximately, hard combinatorial optimization problems for which the use of exact methods is impractical. Some researches have been combining machine learning techniques with metaheuristics to adaptively guide and improve the search for near optimal solutions. An example of such development is the DM-GRASP, a hybrid version of the Greedy Randomized Adaptative Search Procedures (GRASP) metaheuristic which incorporates a data mining process. In this hybrid proposal, after executing half of the total number of iterations, the data mining process extracts patterns from an elite set of sub-optimal solutions for the optimization problem. These patterns present characteristics of near optimal solutions and can be used to guide the following half GRASP iterations in the search through the solution space. In this work, we explore new versions of the DM-GRASP metaheuristic to experiment, not a single activation, but multiple and adaptive executions of the data mining process during the metaheuristic execution. We also applied the data mining technique into a reactive GRASP to show that a more sophisticated and not memoryless GRASP approach can also benefit from the use of this technique. In order to evaluate these new proposals, we adopted the server replication for reliable multicast problem since the best known results for this problem were obtained by GRASP and DM-GRASP implementations. The computational experiments, comparing traditional GRASP, DM-GRASP, and the new proposals, showed that multiple and adaptive executions of the data mining process can improve the results obtained by the DM-GRASP hybrid metaheuristic—the new proposals were able to find better results in less computational time for the reliable multicast problem.     

A nonmonotone GRASP

A greedy randomized adaptive search procedure (GRASP) is an iterative multistart metaheuristic for difficult combinatorial optimization problems. Each GRASP iteration consists of two phases: a construction phase, in which a feasible solution is produced, and a local search phase, in which a local optimum in the neighborhood of the constructed solution is sought. Repeated applications of the construction procedure yields different starting solutions for the local search and the best overall solution is kept as the result. The GRASP local search applies iterative improvement until a locally optimal solution is found. During this phase, starting from the current solution an improving neighbor solution is accepted and considered as the new current solution. In this paper, we propose a variant of the GRASP framework that uses a new “nonmonotone” strategy to explore the neighborhood of the current solution. We formally state the convergence of the nonmonotone local search to a locally optimal solution and illustrate the effectiveness of the resulting Nonmonotone GRASP on three classical hard combinatorial optimization problems: the maximum cut problem (MAX-CUT), the weighted maximum satisfiability problem (MAX-SAT), and the quadratic assignment problem (QAP).     

Multiobjective GRASP with Path Relinking

In this paper we review and propose different adaptations of the GRASP metaheuristic to solve multiobjective combinatorial optimization problems. In particular, we describe several alternatives to specialize the construction and improvement components of GRASP when two or more objectives are considered. GRASP has been successfully coupled with Path Relinking for single-objective optimization. Moreover, we propose different hybridizations of GRASP and Path Relinking for multiobjective optimization. We apply the proposed GRASP with Path Relinking variants to two combinatorial optimization problems, the biobjective orienteering problem and the biobjective path dissimilarity problem. We report on empirical tests with 70 instances and 30 algorithms, that show that the proposed heuristics are competitive with the state-of-the-art methods for these problems.     

Grasp with memory-based mechanisms for minimizing total tardiness in single machine scheduling with setup times

This paper addresses the problem of scheduling jobs in a single machine with sequence dependent setup times in order to minimize the total tardiness with respect to job due dates. We propose variants of the GRASP metaheuristic that incorporate memory-based mechanisms for solving this problem. There are two mechanisms proposed in the literature that utilize a long-term memory composed of an elite set of high quality and sufficiently distant solutions. The first mechanism consists of extracting attributes from the elite solutions in order to influence the construction of an initial solution. The second one makes use of path relinking to connect a GRASP local minimum with a solution of the elite set, and also to connect solutions from the elite set. Reactive GRASP, which probabilistically determines the degree of randomness in the GRASP construction throughout the iterations, is also investigated. Computational tests for instances involving up to 150 jobs are reported, and the proposed method is compared with heuristic and exact methods from the literature.     

A hybrid GRASP/VND algorithm for two- and three-dimensional bin packing

The three-dimensional bin packing problem consists of packing a set of boxes into the minimum number of bins. In this paper we propose a new GRASP algorithm for solving three-dimensional bin packing problems which can also be directly applied to the two-dimensional case. The constructive phase is based on a maximal-space heuristic developed for the container loading problem. In the improvement phase, several new moves are designed and combined in a VND structure. The resulting hybrid GRASP/VND algorithm is simple and quite fast and the extensive computational results on test instances from the literature show that the quality of the solutions is equal to or better than that obtained by the best existing heuristic procedures.     

Speeding up continuous GRASP

Continuous GRASP (C-GRASP) is a stochastic local search metaheuristic for finding cost-efficient solutions to continuous global optimization problems subject to box constraints (Hirsch et al., 2007). Like a greedy randomized adaptive search procedure (GRASP), a C-GRASP is a multi-start procedure where a starting solution for local improvement is constructed in a greedy randomized fashion. In this paper, we describe several improvements that speed up the original C-GRASP and make it more robust. We compare the new C-GRASP with the original version as well as with other algorithms from the recent literature on a set of benchmark multimodal test functions whose global minima are known. Hart’s sequential stopping rule (1998) is implemented and C-GRASP is shown to converge on all test problems.     

Applying the pilot method to improve VNS and GRASP metaheuristics for the design of SDH/WDM networks

Variable neighborhood search (VNS) and Greedy randomized adaptive search procedure (GRASP) are among the well studied local search based metaheuristics providing good results for many combinatorial optimization problems throughout the last decade. While they are usually explored in different environments one may encounter quite obvious commonalities. Based on previous successful applications of these two types of metaheuristics on various network design problems in telecommunications, we further enhance these approaches by incorporating ideas from the pilot method. The different heuristics are compared among each other as well as against objective function values obtained from a mathematical programming formulation based on a commercial solver. The problem instances cover a large variety of networks and demand patterns.     

Avoiding local optima in thep-hub location problem using tabu search and GRASP

In the discretep-hub location problem, various nodes interact with each other by sending and receiving given levels of traffic (such as telecommunications traffic, data transmissions, airline passengers, packages, etc.). It is necessary to choosep of the given nodes to act as hubs, which are fully interconnected; it is also necessary to connect each other node to one of these hubs so that traffic can be sent between any pair of nodes by using the hubs as switching points. The objective is to minimize the sum of the costs for sending traffic along the links connecting the various nodes. Like many combinatorial problems, thep-hub location problem has many local optima. Heuristics, such as exchange methods, can terminate once such a local optimum is encountered. In this paper, we describe new heuristics for thep-hub location problem, based on tabu search and on a greedy randomized adaptive search procedure (GRASP). These recently developed approaches to combinatorial optimization are capable of examining several local optima, so that, overall, superior solutions are found. Computational experience is reported in which both tabu search and GRASP found optimal hub locations (subject to the assumption that nodes must be assigned to the nearest hub) in over 90% of test problems. For problems for which such optima are not known, tabu search and GRASP generated new best-known solutions.     

A hybrid heuristic for the diameter constrained minimum spanning tree problem

We propose a hybrid GRASP and ILS based heuristic for the diameter constrained minimum spanning tree problem. The latter typically models network design applications where, under a given quality requirement, all vertices must be connected at minimum cost. An adaptation of the one time tree heuristic is used to build feasible diameter constrained spanning trees. Solutions thus obtained are then attempted to be improved through local search. Four different neighborhoods are investigated, in a scheme similar to VND. Upper bounds within 2% of optimality were obtained for problems in two test sets from the literature. Additionally, upper bounds stronger than those previously obtained in the literature are reported for OR-Library instances.     

Hybridization of GRASP Metaheuristic with Data Mining Techniques

In this work, we propose a hybridization of GRASP metaheuristic that incorporates a data mining process. We believe that patterns obtained from a set of sub-optimal solutions, by using data mining techniques, can be used to guide the search for better solutions in metaheuristics procedures. In this hybrid GRASP proposal, after executing a significant number of GRASP iterations, the data mining process extracts patterns from an elite set of solutions which will guide the following iterations. To validate this proposal we have worked on the Set Packing Problem as a case study. Computational experiments, comparing traditional GRASP and different hybrid approaches, show that employing frequent patterns mined from an elite set of solutions conducted to better results. Besides, additional performed experiments evidence that data mining strategies accelerate the process of finding good solutions. ^★★Work sponsored by CNPq research grants 300879/00-8 and 475124/03-0. ^†Work sponsored by CNPq research grant 475124/03-0.     

Greedy Randomized Adaptive Search Procedures

Today, a variety of heuristic approaches are available to the operations research practitioner. One methodology that has a strong intuitive appeal, a prominent empirical track record, and is trivial to efficiently implement on parallel processors is GRASP (Greedy Randomized Adaptive Search Procedures). GRASP is an iterative randomized sampling technique in which each iteration provides a solution to the problem at hand. The incumbent solution over all GRASP iterations is kept as the final result. There are two phases within each GRASP iteration: the first intelligently constructs an initial solution via an adaptive randomized greedy function; the second applies a local search procedure to the constructed solution in hope of finding an improvement. In this paper, we define the various components comprising a GRASP and demonstrate, step by step, how to develop such heuristics for combinatorial optimization problems. Intuitive justifications for the observed empirical behavior of the methodology are discussed. The paper concludes with a brief literature review of GRASP implementations and mentions two industrial applications.     

Single machine scheduling with flow time and earliness penalties

This paper considers the problem of schedulingn jobs on a single machine to minimize the total cost incurred by their respective flow time and earliness penalties. It is assumed that each job has a due date that must be met, and that preemptions are not allowed. The problem is formulated as a dynamic program (DP) and solved with a reaching algorithm that exploits a series of dominance properties and efficiently generated bounds. A major factor underlying the effectiveness of the approach is the use of a greedy randomized adaptive search procedure (GRASP) to construct high quality feasible solutions. These solutions serve as upper bounds on the optimum, and permit a predominant portion of the state space to be fathomed during the DP recursion.To evaluate the performance of the algorithm, an experimental design involving over 240 randomly generated problems was followed. The test results indicate that problems with up to 30 jobs can be readily solved on a microcomputer in less than 12 minutes. This represents a significant improvement over previously reported results for both dynamic programming and mixed integer linear programming approaches.     

GRASP with path relinking for the orienteering problem

In this paper, we address an optimization problem resulting from the combination of the well-known travelling salesman and knapsack problems. In particular, we target the orienteering problem, originated in the context of sport, which consists of maximizing the total score associated with the vertices visited in a path within the available time. The problem, also known as the selective travelling salesman problem, is NP-hard and can be formulated as an integer linear program. Since the 1980s, several solution methods for this problem have been developed and applied to a variety of fields, particularly in routing and tourism. We propose a heuristic method—based on the Greedy Randomized Adaptive Search Procedure (GRASP) and the Path Relinking methodologies—for finding approximate solutions to this optimization problem. We explore different constructive methods and combine two neighbourhoods in the local search of GRASP. Our experimentation with 196 previously reported instances shows that the proposed procedure obtains high-quality solutions employing short computing times.     

A GRASP for simultaneously assigning and sequencing product families on flexible assembly lines

This paper introduces a new model and solution methodology for a real-world production scheduling problem arising in the electronics industry. The production environment is a high volume, just-in-time, make-to-order facility with volatile demand over many product families that are assembled on flexible lines. A distinguishing characteristic of the problem is the presence of non-traditional sequence-dependant setup costs, which complicate our ability to find high-quality solutions. The scheduling problem arose when product variety exceeded the mix that the existing lines could accommodate. A nonlinear integer programming formulation is presented for the problem of minimizing setup costs, and a greedy randomized adaptive search procedure (GRASP) is developed to find solutions. To select the GRASP parameter values, an efficient, space-filling experimental design method is used based on nearly orthogonal Latin hypercubes. The proposed methodology is tested on actual factory data and compared to a prior heuristic presented in the literature; our heuristic provides a cost savings in 7 out of the 10 cases examined, and an average improvement of 17.39 % which is shown to be highly statistically significant. This improvement is due in part to the introduction of a pre-processing step to determine preferential and non-preferential line assignment information.     

GRASP with path-relinking for the non-identical parallel machine scheduling problem with minimising total weighted completion times

In this work, we tackle the problem of scheduling a set of jobs on a set of non-identical parallel machines with the goal of minimising the total weighted completion times. GRASP is a multi-start method that consists of two phases: a solution construction phase, which randomly constructs a greedy solution, and an improvement phase, which uses that solution as an initial starting point. In the last few years, the GRASP methodology has arisen as a prospective metaheuristic approach to find high-quality solutions for several difficult problems in reasonable computational times. With the aim of providing additional results and insights along this line of research, this paper proposes a new GRASP model that combines the basic scheme with two significant elements that have been shown to be very successful in order to improve GRASP performance. These elements are path-relinking and evolutionary path-relinking. The benefits of our proposal in comparison to existing metaheuristics proposed in the literature are experimentally shown.     

GRASP and Path Relinking for the Clustered Prize-collecting Arc Routing Problem

The Clustered Prize-collecting Arc Routing Problem is an arc routing problem where each demand edge is associated with a profit which is collected once if the edge is serviced, independently of the number of times it is traversed. It is further required that if a demand edge is serviced, then all the demand edges of its component are also serviced. This paper presents GRASP and Path Relinking heuristics for the Clustered Prize-collecting Arc Routing Problem. For the constructive phase of the GRASP two different strategies are considered. One of them follows a bottom-up style whereas the other one is a top-down procedure. The best solutions obtained with both strategies are used as elite solutions for the Path Relinking. The results of extensive computational experiments are presented and analyzed.