Glover, F. and M. Laguna, \Tabu search", a chapter in Modern Heuristic Techniques for Combinatorial Optimization, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....very dicult task. Most of such optimization problems are multiextremal NP hard problems. Well known methods for both continuous and discrete optimization problems are simulated annealing [1] 9] 13] 40] initiated by Metropolis [32] and later generalized by Hastings [19] and [26] tabu search [15] and genetic algorithms [16] For a very interesting landmark paper on simulated annealing see Romeijn and Smith [40] Recent works on combinatorial optimization include the nested partitioning (NP) method [45] the stochastic comparison method [17] and the method of Andrad ottir [6] 7] In ....

....optimization problems indicating polynomial (in the size of the problem) running time complexity are given as well. Our algorithm employs quite di erent ideas than the other well known random search algorithms for global optimization including simulated annealing [1] 13] 40] tabu search [15] and genetic algorithms [16] These, in fact, represent local search heuristics and employ the notion of so called local neighborhood structures. Although our method, like many others, reconstructs the initial probability measure to make occurrence of the events more likely in the vicinity of a ....

Glover, F. and M. Laguna, \Tabu search", a chapter in Modern Heuristic Techniques for Combinatorial Optimization, 1992.

....or graph, where each edge has a given length, the longest path and the TSP. For additional references on CE for COP see [14] and [20] 21] and for application of simulated annealing to BAP, see [26] Alternative well known heuristics for cop, capable of handling the BAP problem are tabu search [9], genetic algorithms [11] nested partitioning (NP) 24] 25] and the Ant Colony Optimization (ACO) meta heuristic of Dorigo and colleagues [6] 8] see also Gutjahr [12] for the prove of convergence of the ant based algorithm) Although CE can handle successfully both deterministic and ....

Glover, F. and M. Laguna, \Tabu search", a chapter in Modern Heuristic Techniques for Combinatorial Optimization, 1992.

....(k = ffi f log Gamma 1 Delta ) In this implementation the function k 0:98 is used for practical heat descent. A bad solution is accepted if e Gammaffi f k is greater than a random number uniformly chosen in (0; 1) see figure 1) Tabu search. It is a deterministic meta heuristic [5]. As for the simulated annealing, a lot of parameters have to be tuned or defined [1] The tabu search starts from a given solution and improves it by local pair wise exchanges. Accessible solutions using local moves are called neighbors. At a given step, the best unexplored neighbor is chosen. ....

F. Glover and M. Laguna. Tabu Search, a chapter in Modern Heuristic Techniques for Combinatorial Problems. W.H. Freeman, N-Y, 1992.

....is not to find a hill but to reach the top. Recently several variants of local search have been developed to overcome this defect. These variants include the simulated annealing algorithm [1, 7, 10, 27, 28, 30, 45] the genetic algorithm [23, 25, 35, 36] neural net approach [26] and tabu search [3, 18, 19, 20, 21, 22]. In this paper, we derive a new variant of local search called the Life Span Method (LSM) which is much simpler than many other variants of local search. 2. Local Search and Tabu Search In this section, we briefly review local search and tabu search on which our new algorithm is based. Here, ....

....x 0 2 N(x) we store attribute (x 0 ; x) in the tabu list. Then, if attribute (x; x 0 ) is in the tabu list, move (x; x 0 ) cannot be used, i.e. the move is tabu , for tabulength iterations. Some heuristics to diversify and to intensify the search were proposed. For more details, see [18, 19, 21, 22]. 3. Life Span Method In this section, we describe a new variant of local search named the Life Span Method (LSM) 3.1. Origin Before describing our algorithm, we mention the origin of the LSM. Our first aim was to compare some variants of local search in a fair manner. When we started the ....

[Article contains additional citation context not shown here]

F. Glover and M. Laguna. Tabu search. A chapter in Modern Heuristic Techniques for Combinatorial Problems, C. Reeves, ed., Blackwell Scientific Publishing, pages 70--141, 1993.

....1 for all fi 2 B such that LS(fi) 0 9 return x Figure 3: Life Span Method. Not only are the mathematical definitions between tabu search and the LSM different, but the fundamental philosophy is also different. The philosophy of tabu search is to collect principles of intelligent problem solving [17]; so the parameters to control the algorithm may be very large. Meanwhile, our philosophy is to keep the number of control parameters as small as possible. The details of the LSM can be found in the companion paper [27] 4 The Life Span Method for the Maximum Stable Set Problem To design an ....

F. Glover. Tabu search. A chapter in Modern Heuristic Techniques for Combinatorial Problems, 1992.

....can locally increase the energy of the system. This phenomenon occurs with a given probability decreasing with the temperature. It corresponds mathematically to giving a chance to leave a local minimum of the function to optimize. 2.2. 3 Tabu search Tabu search is an iterative meta heuristic [5]. It tries to find the best neighbor of a given solution. To avoid cycling and local optima, a tabu list is established. This tabu list contains information concerning the last moves. A tabu move is not allowed except if an aspiration criterion is satisfied (i.e. if the proposed neighbor gives a ....

Fred Glover and Manuel Laguna. Tabu Search, a chapter in Modern Heuristic Techniques for Combinatorial Problems. W.H. Freeman, N-Y, 1992.

....Allowing infeasible solutions makes it possible to escape from local optima. Not only are the mathematical definitions between tabu search and the LSM different, but the fundamental philosophy is also different. The philosophy of tabu search is to collect principles of intelligent problem solving [15]; so the parameters to control the algorithm may be very large. Meanwhile, our philosophy is to keep the number of control parameters as small as possible. The details of the LSM can be found in the companion paper [21] 4 Life Span Method for the Quadratic Assignment Problem In this section, we ....

F. Glover. Tabu search. A chapter in Modern Heuristic Techniques for Combinatorial Problems, 1992.

....k follows the curve 1 log(k) It is well known that this algorithm is very hard to tune: finding the starting temperature or the heat decreasing steps are very difficult in practice and have to be done after many experiments. 3.3. 3 Tabu search Tabu search is an iterative meta heuristic [GL92]. Contrary to the simulated annealing, it is deterministic. As for the simulated annealing, a lot of parameters have to be tuned or defined[BBTW95] The tabu search improves iteratively a cost function: the tabu starts from a given solution and tries to improve it by doing some local moves (e.g. ....

F. Glover and M. Laguna. Tabu Search, a chapter in Modern Heuristic Techniques for Combinatorial Problems. W.H. Freeman, N-Y, 1992.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC