## Guided Local Search (1995)

Venue: | European Journal of Operational Research |

Citations: | 57 - 6 self |

### BibTeX

@TECHREPORT{Voudouris95guidedlocal,

author = {Chris Voudouris and Edward Tsang},

title = {Guided Local Search},

institution = {European Journal of Operational Research},

year = {1995}

}

### Years of Citing Articles

### OpenURL

### Abstract

Guided Local Search (GLS) is an intelligent search scheme for combinatorial optimization problems. A main feature of the approach is the iterative use of local search. Information is gathered from various sources and exploited to guide local search to promising parts of the search space. The application of the method to the Travelling Salesman Problem and the Quadratic Assignment Problem is examined. Results reported show that the algorithm outperforms or compares very favorably with well-known and established optimization techniques such as simulated annealing and tabu search. Given the novelty of the approach and the very encouraging results, the method could have an important contribution to the development of intelligent search techniques for combinatorial optimization. 1. Introduction Guided Local Search is the outcome of a research project with main aim to extend the GENET neural network [29,26,5] for constraint satisfaction problems to partial constraint satisfaction [6,26] and...

### Citations

2833 |
Adaptation in natural and artificial systems
- Holland
- 1975
(Show Context)
Citation Context ... In another GLS work, aspiration (inspired by TS) is used to favour promising moves [55]. 7.2 GLS and Genetic Algorithms As a meta-heuristic method, GLS can also sit on top of genetic algorithms (GA) =-=[27, 33]-=-. This has been demonstrated in Guided Genetic Algorithm (GGA) [44, 45, 46, 47]. GGA is a hybrid of GA and GLS. It is designed to extend the domain of both GA and GLS. One major objective is to furthe... |

1818 |
Genetic Algorithms in Search Optimization and Machine Learning
- Goldberg
- 1989
(Show Context)
Citation Context ... In another GLS work, aspiration (inspired by TS) is used to favour promising moves [55]. 7.2 GLS and Genetic Algorithms As a meta-heuristic method, GLS can also sit on top of genetic algorithms (GA) =-=[27, 33]-=-. This has been demonstrated in Guided Genetic Algorithm (GGA) [44, 45, 46, 47]. GGA is a hybrid of GA and GLS. It is designed to extend the domain of both GA and GLS. One major objective is to furthe... |

816 | Foundations of Constraint Satisfaction
- Tsang
- 1993
(Show Context)
Citation Context ... development of intelligent search techniques for combinatorial optimization. 1. Introduction Guided Local Search is the outcome of a research project with main aim to extend the GENET neural network =-=[29,26,5]-=- for constraint satisfaction problems to partial constraint satisfaction [6,26] and combinatorial optimization problems. Beginning with GENET, we developed a number of intermediate algorithms such as ... |

679 | A new method for solving hard satisfiability problems
- Selman, Levesque, et al.
- 1992
(Show Context)
Citation Context ...the SAT and weighted MAXSAT problem [54]. On a set of SAT problems from DIMACS, GLSSAT produced more frequently better or comparable solutions than those produced by WalkSAT [64], a variation of GSAT =-=[65]-=-, which was specifically designed for the SAT problem. On a popular set of benchmark weighted MAX-SAT problems, GLSSAT produced better or comparable solutions, more frequently than state-of-the-art al... |

605 | Tabu search
- Glover, Laguna
- 1998
(Show Context)
Citation Context ...is denoted by ImprovementMethod. This method can be a simple local search algorithm or a more sophisticated one such as Variable neighbourhood Search [30], Variable Depth Search [50], Ejection Chains =-=[25]-=- or combinations of local search methods with exact search algorithms [60].6 Christos Voudouris, Edward P. K. Tsang and Abdullah Alsheddy It is not essential for the improvement method to generate hi... |

512 |
Optimization by Simulated Annealing: An Experimental Evaluation
- Johnson, Aragon, et al.
- 1991
(Show Context)
Citation Context ...ium. After reaching equilibrium, the temperature was multiplied by the cooling rate a which was set to a high value (a = 0.9) . To stop the algorithm, we used the scheme with the counter described in =-=[12]-=-. 6.1.5.2 Tabu Search The tabu search variant implemented is using a combination of tabu restrictions and aspiration level criteria that bias the algorithm towards favoring short edges and avoiding lo... |

475 |
Tabu-search Part I
- Glover
- 1989
(Show Context)
Citation Context ... reducing so the number of candidate solutions to be considered. Generally speaking, GLS objectives are similar to those of a wider class of combinatorial optimization algorithms known as tabu search =-=[7,8,9]-=-. In fact, GLS could be classified as a tabu search method though there are many and distinct differences with the other methods developed so far within that framework. To mention a few, GLS is a comp... |

422 | Partial constraint satisfaction
- Freuder, Wallace
- 1992
(Show Context)
Citation Context ...ntroduction Guided Local Search is the outcome of a research project with main aim to extend the GENET neural network [29,26,5] for constraint satisfaction problems to partial constraint satisfaction =-=[6,26]-=- and combinatorial optimization problems. Beginning with GENET, we developed a number of intermediate algorithms such as the Tunneling Algorithm [28] to conclude with Guided Local Search (GLS) present... |

399 | Minimizing conflicts: a heuristic repair method for constraint satisfaction and scheduling problems. Ar-hf. /tall. 58
- Minton, Johnston, et al.
- 1992
(Show Context)
Citation Context ...pe local optimum. This chapter focuses on GLS [81]. GLS can be seen as a generalization of techniques such as GENET [15, 78, 79, 87, 88] and the min-conflicts heuristic repair method by Minton et al. =-=[56]-=- developed for constraint satisfaction problems. GLS also relates to ideas from the area of Search Theory on how to distribute the search effort (e.g. see [41, 68]). The principles of GLS can be summa... |

368 |
Neural Networks
- Haykin
- 1994
(Show Context)
Citation Context ...solutions generated by local search to be in accordance to prior or gathered during search information. The approach taken by GLS is similar to that of regularization methods for `illposed ' problems =-=[25,10]-=-. The idea behind regularization methods and GLS up to an extent is the use of prior information to help us solve an approximation problem. Prior information translates to constraints which further de... |

323 |
An effective heuristic algorithm for the travelling salesman problem
- Lin, Kernighan
- 1973
(Show Context)
Citation Context ... testing of new optimization methods. In this context, we examine how guided local search has been applied to the problem. 6.1.2 Local Search Local search for the TSP is synonymous to the k-opt moves =-=[15]-=-. The simplest and less expensive of the k-opt moves is the famous 2-opt. 2-opt removes two edges from the tour to replace them with two new edges not previously included in the tour. Particular care ... |

272 |
Tabu Search - Part II
- Glover
- 1990
(Show Context)
Citation Context ... reducing so the number of candidate solutions to be considered. Generally speaking, GLS objectives are similar to those of a wider class of combinatorial optimization algorithms known as tabu search =-=[7,8,9]-=-. In fact, GLS could be classified as a tabu search method though there are many and distinct differences with the other methods developed so far within that framework. To mention a few, GLS is a comp... |

259 |
TSPLIB – A Traveling Salesman Problem Library
- REINELT
- 1991
(Show Context)
Citation Context ...ghborhoods corresponding to the cities at the ends of that edge are activated. 6.1.4 Results on Small to Medium Size TSPs In the continue, we report results for GLS on known TSP instances from TSPLIB =-=[19]-=-. TSPLIB is a publicly available library of TSP problems. Most of the instances included in TSPLIB have already been solved to optimality and they have been used in many works in the past. For each TS... |

226 | G.: The Reactive Tabu Search
- Battiti, Tecchiolli
- 1994
(Show Context)
Citation Context ... the QAP has been already described in section 5 and it is the general one for permutation-based representations. Evaluating a swap can be made in constant time for best improvement local search (see =-=[1]-=-). Unfortunately, the scheme requires all possible swaps to be evaluated before a move is performed and therefore can not be used in the case of fast local search which for the QAP requires O(n) opera... |

215 | Domain-Independent Extensions to GSAT: Solving Large Structured Satisfiability Problems
- Selman, Kautz
- 1993
(Show Context)
Citation Context ... GLS, was applied to both the SAT and weighted MAXSAT problem [54]. On a set of SAT problems from DIMACS, GLSSAT produced more frequently better or comparable solutions than those produced by WalkSAT =-=[64]-=-, a variation of GSAT [65], which was specifically designed for the SAT problem. On a popular set of benchmark weighted MAX-SAT problems, GLSSAT produced better or comparable solutions, more frequentl... |

164 | QAPLIB – a quadratic assignment problem library
- Burkard, Karisch, et al.
- 1997
(Show Context)
Citation Context ...uniformly across the feature set. Comparative tests we conducted between GLS and the Tabu Search of [70] indicate that both algorithms are performing equally well when applied to the QAPLIB instances =-=[8]-=- with no clear winner across the instance set. GLS, although not using feature costs in this problem, is still very competitive to state-of-the-art techniques such as Tabu Search. To determine λ in th... |

157 |
The traveling salesman: computational solutions for TSP applications
- Reinelt
- 1994
(Show Context)
Citation Context ...N. The cost of a permutation is defined as: ( ) ( ) g d i i N p p = = �� 1 (1-14) and gives the cost function of the TSP [17]. An up to date and comprehensive survey of TSP methods is that by Rein=-=elt [20]-=-. The reader may also refer to [14] for a classical text on the TSP. The state of the art is that problems up to 1.000.000 cities are within the reach of specialised approximation algorithms [2]. More... |

148 |
Local Optimization and the Traveling Salesman Problem
- Johnson
- 1990
(Show Context)
Citation Context ... we could not find any other approximation method that uses only the simple 2-opt move and consistently finds optimal solutions for problems up to 318 cities. Only the iterated LinKernighan algorithm =-=[11]-=- shares the same consistency in reaching the optimal solutions [20] though it uses k-opt moves of higher order. A meaningful comparison that can be made is between GLS and other general methods that a... |

138 | A user's guide to tabu search
- Glover, Taillard, et al.
- 1993
(Show Context)
Citation Context ... reducing so the number of candidate solutions to be considered. Generally speaking, GLS objectives are similar to those of a wider class of combinatorial optimization algorithms known as tabu search =-=[7,8,9]-=-. In fact, GLS could be classified as a tabu search method though there are many and distinct differences with the other methods developed so far within that framework. To mention a few, GLS is a comp... |

131 |
Fast Algorithms for Geometric Traveling Salesman Problems
- Bentley
- 1992
(Show Context)
Citation Context ...es of problems. We present here fast local search (FLS) which drastically speeds up the neighborhood search process. The method is a generalization of the approximate 2-opt method proposed by Bentley =-=[2]-=- for the TSP. Fast local search works as follows. The problem's neighborhood is broken down to a number of small sub-neighborhoods and an activation bit is attached to each one of them. The idea is to... |

122 | Efficient exploration in reinforcement learning
- Thrun
- 1992
(Show Context)
Citation Context ... from the optimal search theory of Koopman [13,22]. Also, counter based schemes for search diversification like that of GLS are used under the name counter-based exploration in reinforcement learning =-=[24]-=-. Something that has been left out from the so far analysis is the regularization parameter l in the augmented cost function (1-6). This parameter determines the degree up to which constraints on feat... |

115 |
A Traveling Salesman Problem
- Reinelt
- 1991
(Show Context)
Citation Context ...ghborhoods corresponding to the cities at the ends of that edge are activated. 6.1.4 Results on Small to Medium Size TSPs In the continue, we report results for GLS on known TSP instances from TSPLIB =-=[19]-=-. TSPLIB is a publicly available library of TSP problems. Most of the instances included in TSPLIB have already been solved to optimality and they have been used in many works in the past. For each TS... |

107 | Quadratic Assignment Problem
- Burkard
- 1984
(Show Context)
Citation Context ...cations In the applications section, we examine how the method of GLS has been used to tackle two well-known NP-hard problems, the Traveling Salesman Problem [14] and the Quadratic Assignment Problem =-=[3]-=-. All experiments reported in this paper performed on a DEC Alpha 3000/600. Unless otherwise stated, the algorithms were implemented in C++. 6.1 Traveling Salesman Problem(TSP) 6.1.1 The Problem There... |

96 | GENET: A connectionist architecture for solving constraint satisfaction problems by iterative improvement
- Davenport, Tsang, et al.
- 1994
(Show Context)
Citation Context ... development of intelligent search techniques for combinatorial optimization. 1. Introduction Guided Local Search is the outcome of a research project with main aim to extend the GENET neural network =-=[29,26,5]-=- for constraint satisfaction problems to partial constraint satisfaction [6,26] and combinatorial optimization problems. Beginning with GENET, we developed a number of intermediate algorithms such as ... |

91 | The quadratic assignment problem: A survey and recent developments, in: Quadratic Assignment and Related Problems, in
- Pardalos, Rendl, et al.
- 1994
(Show Context)
Citation Context ...ion. The problem can model a variety of applications but it is mainly known for its use in facility location problems. For a recent QAP survey, the reader is referred to Pardalos, Rendl and Wolkowicz =-=[18]. In-=- the following, we describe the QAP in its simple form. Guided Local Search 14 Given a set N = {1, 2, ..., n} and n �� n matrices A= [ a ij ] and B = [b kl ], the QAP can be stated as follows: ( )... |

90 |
Comparison of iterative searches for the quadratic assignment problem
- Taillard
- 1995
(Show Context)
Citation Context ... values of the first neighbourhood search are stored and updated eachGUIDED LOCAL SEARCH 27 time a new neighbourhood search is performed to take into account changes from the move last executed, see =-=[71]-=- for details. Move values do not need to be evaluated from scratch and thus the neighbourhood can be fully searched in roughly O(n 2 ) time instead of O(n 3 ) 1 . To evaluate moves in constant time, w... |

81 | Combining Simulated Annealing with Local Search Heuristics
- Martin, Otto
- 1996
(Show Context)
Citation Context ...bly in their application to the travelling salesman problem (TSP). The Lin-Kernighan algorithm (LK) is a specialised algorithm for the TSP that has long been perceived as the champion of this problem =-=[50, 51]-=-. We tested GLS+FLS+2Opt against LK [85] on a set of benchmark problems from a public TSP library [61]. Given the same amount of time, GLS+FLS+2Opt found better results than LK on average. GLS+FLS+2Op... |

78 | AGenetic Local Search Algorithm for Solving Symmetric and Asymmetric Traveling Salesman Problems
- Freisleben, Merz
(Show Context)
Citation Context ...TSP library [61]. Given the same amount of time, GLS+FLS+2Opt found better results than LK on average. GLS+FLS+2Opt also out-performed Simulated Annealing [36], Tabu Search [40] and Genetic Algorithm =-=[23]-=- implementations for the TSP. One must be cautious when interpreting such empirical results as they could be affected by many factors, including implementation details. But given that the TSP is an ex... |

77 |
Tabu search applied to the quadratic assignment problem
- Skorin-Kapov
- 1990
(Show Context)
Citation Context ...n GLS or robust taboo search did. A second set of runs was performed on large QAP instances. Thirteen instances from QAPLIB were used with sizes from 42 to 100 which have been randomly generated (see =-=[21]-=- for more). A limited amount of time equal to 5 minutes on a DEC Alpha 3000/600 was given to each algorithm. Ten runs performed on each instance from random initial solutions. The objective of this se... |

59 |
Variable neighbourhood search
- Mladenović, Hansen
- 1997
(Show Context)
Citation Context ...lution is also required. In the pseudo-code, this is denoted by ImprovementMethod. This method can be a simple local search algorithm or a more sophisticated one such as Variable neighbourhood Search =-=[30]-=-, Variable Depth Search [50], Ejection Chains [25] or combinations of local search methods with exact search algorithms [60].6 Christos Voudouris, Edward P. K. Tsang and Abdullah Alsheddy It is not e... |

59 | A discrete Lagrangian-based global-search method for solving satisfiability problems
- Wah, Shang
- 1998
(Show Context)
Citation Context ...ly designed for the SAT problem. On a popular set of benchmark weighted MAX-SAT problems, GLSSAT produced better or comparable solutions, more frequently than state-of-the-art algorithms, such as DLM =-=[66]-=-, WalkSAT [64] and GRASP [63]. 8.6 Generalized Assignment Problem The Generalized Assignment Problem is a generic scheduling problem in which the task is to assign agents to jobs. Each job can only be... |

49 | Guided Local Search and its application to the travelling salesman problem
- Voudouris, Tsang
- 1999
(Show Context)
Citation Context ...Voudouris, Edward P. K. Tsang and Abdullah Alsheddy It is not essential for the improvement method to generate high quality local minima. Experiments with GLS and various local heuristics reported in =-=[85]-=- have shown that high quality local minima take time to produce, resulting in less intervention by GLS in the overall allocated search time. This may sometimes lead to inferior results compared to a s... |

47 | Solving vehicle routing problems using constraint programming and metaheuristics
- Backer, Furnon, et al.
- 1997
(Show Context)
Citation Context ...ould improve upon the most frequently employed strategy based on di, j only. Experimental results confirm the effectiveness of GTS, producing new best results for several benchmarks. De Backer et al. =-=[3]-=- also proposed a Guided Tabu Search hybrid in their work on the VRP. GLS has been also successfully hybridized with Ant Colony Optimization (ACO) by Hani et al. [29]. This hybrid algorithm was applied... |

43 | Guided local search for the vehicle routing problem
- Kilby, Prosser, et al.
- 1997
(Show Context)
Citation Context ...fying time and capacity constraints. This is a practical problem which, like many practical problems, is NP-hard. Kilby et al. applied GLS to vehicle routing problems and achieved outstanding results =-=[38, 39]-=-. As a result, their work was incorporated in Dispatcher, a commercial package developed by ILOG [3]. Recently, the application of GLS and its hybrids to the VRP have been considerably extended to sev... |

43 |
Guided Local Search for Combinatorial Optimisation Problems
- Voudouris
- 1997
(Show Context)
Citation Context ...rious techniques have been introduced over the years. Simulated Annealing (SA), Tabu Search (TS) and Guided Local Search (GLS) all attempt to help LS escape local optimum. This chapter focuses on GLS =-=[81]-=-. GLS can be seen as a generalization of techniques such as GENET [15, 78, 79, 87, 88] and the min-conflicts heuristic repair method by Minton et al. [56] developed for constraint satisfaction problem... |

42 | Fast Local Search and Guided Local Search and Their Application to British Telecom’s Workforce Scheduling Problem
- Tsang, Voudouris
- 1995
(Show Context)
Citation Context ...y of problems from the problem libraries TSPLIB and QAPLIB proving the robustness of GLS across combinatorial optimization problems, problem instances and instance sizes. The reader may also refer to =-=[27]-=- where further evidence is provided on the effectiveness of both GLS and fast local search this time in the context of a real-world and difficult scheduling problem. Future research on GLS will invest... |

40 |
A genetic algorithm for the generalized assignment problem
- Chu, Beasley
- 1997
(Show Context)
Citation Context ...ts of resources when doing the same job. In a set of benchmark problems, GGA found results as good as those produced by a state-of-the-art algorithm (which was also a GA algorithm) by Chu and Beasley =-=[11]-=-, with improved robustness [47]. GLS hybrids have been proposed for the related QAP. Zhang et al. [92] proposed the GLS/EDA hybrid metaheuristic. In addition, the hybrid of GLS with ACO (ACO GLS) has ... |

39 |
A Method for Solving Traveling-Salesman Problems
- Croes
- 1958
(Show Context)
Citation Context ...deleting k edges from the current tour and reconnecting the resulting paths using k new edges. The k-Opt moves are the basis of the three most famous local search heuristics for the TSP, namely 2-Opt =-=[13]-=-, 3-Opt [49] and Lin-Kernighan (LK) [50]. The reader can consider using the simple 2-Opt method, which in addition to its simplicity is very effective when combined with GLS. With 2-Opt, a neighbourin... |

37 | Solving constraint satisfaction problems using neuralnetworks
- Wang, Tsang
- 1991
(Show Context)
Citation Context ... development of intelligent search techniques for combinatorial optimization. 1. Introduction Guided Local Search is the outcome of a research project with main aim to extend the GENET neural network =-=[29,26,5]-=- for constraint satisfaction problems to partial constraint satisfaction [6,26] and combinatorial optimization problems. Beginning with GENET, we developed a number of intermediate algorithms such as ... |

33 | Guided local search for solving SAT and weighted MAX-SAT problems
- Mills, Tsang
- 2000
(Show Context)
Citation Context ...SAT problems, hence these problems have received significant attention in recent years, e.g. see Gent et al. [24]. GLSSAT, an extension of GLS, was applied to both the SAT and weighted MAXSAT problem =-=[54]-=-. On a set of SAT problems from DIMACS, GLSSAT produced more frequently better or comparable solutions than those produced by WalkSAT [64], a variation of GSAT [65], which was specifically designed fo... |

31 | Partial constraint satisfaction problems and guided local search, in
- Voudouris, Tsang
- 1996
(Show Context)
Citation Context ...at applied GENET to radio length frequency assignment. For the CALMA set of benchmark problems, which has been widely used, GLS+FLS reported the best results compared to all work published previously =-=[84]-=-. In the NATO Symposium on RLFAP in Denmark, 1998, GGA was shown to improve the robustness of GLS [46]. In the same symposium, new and significantly improved results by GLS were reported [83]. At the ... |

30 | 2009a) “Random Assignment Problems
- Krokhmal, Pardalos
(Show Context)
Citation Context ...sembling14 Christos Voudouris, Edward P. K. Tsang and Abdullah Alsheddy the tabu lists idea, a limited number of penalties are used when GLS is applied to the radio link frequency assignment problem =-=[58]-=-. When the list is full, old penalties are overwritten [83]. In another GLS work, aspiration (inspired by TS) is used to favour promising moves [55]. 7.2 GLS and Genetic Algorithms As a meta-heuristic... |

29 | A generic neural network approach for constraint satisfaction problems. Neural network applications
- Tsang, Wang
- 1992
(Show Context)
Citation Context ...ing (SA), Tabu Search (TS) and Guided Local Search (GLS) all attempt to help LS escape local optimum. This chapter focuses on GLS [81]. GLS can be seen as a generalization of techniques such as GENET =-=[15, 78, 79, 87, 88]-=- and the min-conflicts heuristic repair method by Minton et al. [56] developed for constraint satisfaction problems. GLS also relates to ideas from the area of Search Theory on how to distribute the s... |

25 | A constraint programming framework for local search methods
- Pesant, Gendreau, et al.
- 1999
(Show Context)
Citation Context ...algorithm or a more sophisticated one such as Variable neighbourhood Search [30], Variable Depth Search [50], Ejection Chains [25] or combinations of local search methods with exact search algorithms =-=[60]-=-.6 Christos Voudouris, Edward P. K. Tsang and Abdullah Alsheddy It is not essential for the improvement method to generate high quality local minima. Experiments with GLS and various local heuristics... |

23 |
Shmoys (Eds.), The Travelling Salesman Problem: A guided tour in combinatorial optimization
- Lawler, Lenstra, et al.
- 1985
(Show Context)
Citation Context ...ering all possible modifications. 6. Applications In the applications section, we examine how the method of GLS has been used to tackle two well-known NP-hard problems, the Traveling Salesman Problem =-=[14]-=- and the Quadratic Assignment Problem [3]. All experiments reported in this paper performed on a DEC Alpha 3000/600. Unless otherwise stated, the algorithms were implemented in C++. 6.1 Traveling Sale... |

23 |
The Traveling Salesman: Computational Solutions for TSP
- Reinelt
- 1994
(Show Context)
Citation Context .... ,N. The cost of a permutation is defined as: N = ∑ i= 1 ( π) π() g d i i (1-14) and gives the cost function of the TSP [17]. An up to date and comprehensive survey of TSP methods is that by Reinelt =-=[20]-=-. The reader may also refer to [14] for a classical text on the TSP. The state of the art is that problems up to 1.000.000 cities are within the reach of specialised approximation algorithms [2]. More... |

22 |
Tabu Search Performance on the Symmetric Traveling Salesman Problem
- Knox
- 1994
(Show Context)
Citation Context ...ark problems from a public TSP library [61]. Given the same amount of time, GLS+FLS+2Opt found better results than LK on average. GLS+FLS+2Opt also out-performed Simulated Annealing [36], Tabu Search =-=[40]-=- and Genetic Algorithm [23] implementations for the TSP. One must be cautious when interpreting such empirical results as they could be affected by many factors, including implementation details. But ... |

21 | A comparison of traditional and constraintbased heuristic methods on vehicle routing problems with side constraints
- Kilby, Prosser, et al.
(Show Context)
Citation Context ...fying time and capacity constraints. This is a practical problem which, like many practical problems, is NP-hard. Kilby et al. applied GLS to vehicle routing problems and achieved outstanding results =-=[38, 39]-=-. As a result, their work was incorporated in Dispatcher, a commercial package developed by ILOG [3]. Recently, the application of GLS and its hybrids to the VRP have been considerably extended to sev... |

20 |
Path assignment for Call Routing: An Application of Tabu Search
- Anderson, Fraughnaugh, et al.
- 1993
(Show Context)
Citation Context ...nts which state that certain requests should always be allocated a resource, in which case there is no need to define a feature for them. Problems in this category include the Path Assignment Problem =-=[1]-=-, Maximum Channel Assignment Problem [67], Workforce Scheduling Problem [2] and others. 9.4 Constrained Optimization Problems Constraints are very important in capturing processes and systems in the r... |

20 |
The traveling-salesman problem
- Flood
- 1956
(Show Context)
Citation Context ...OCAL SEARCH As mentioned earlier, GLS augments the given objective function with penalties. To apply GLS, one needs to define features for the problem. For example, in the travelling salesman problem =-=[21]-=-, a feature could be “whether the candidate tour travels immediately from city A to city B”. GLS associates a cost and a penalty with each feature. The costs can often be defined by taking the terms a... |