S.W. Mahfoud and D.E. Goldberg. Parallel recombinative simulated annealing: A genetic algorithm. Parallel Computing, 21(1):1--28, 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... interacting walks [40] or population approaches [28] Procedures such as coarse grained parallel genetic algorithms (PGA) 17,33,35] and ne grained PGA [16,26,30] Memetic Algorithms [6,27,29] parallel Branchand Bound [13] tabu search algorithms [8,10,11,23] simulating annealing procedures [22,24,25], and special purpose Arti cial Intelligence (AI) cooperative searches [4,19] are examples of parallel search methods based on cooperation born from sharing gathered information among several sequential search programs. The primary rationale behind sharing information among search programs is to ....

....to the search parameters of each individual program which are primarily a ecting the search pattern of the sequential programs as such. The existence of these parameters as part of the design of cooperation among sequential search programs has been discussed by Laursen [22] Mahfoud and Goldberg [25], and Toulouse, Crainic and Gendreau [37] Cooperative search algorithms obtain speed up similar to parallel independent search algorithms. The most noticeable impact of cooperation consists in the modi cation of the search paths of the cooperating search programs. One can measure the extend of ....

S. W. Mahfoud and D. E. Goldberg. Parallel Recombinative Simulated Annealing: a Genetic Algorithm. Parallel Computing, 21:1-28, 1995.

....techniques requiring the calculation of Z . 5.2.3.3. Parallel recombinative simulated annealing (PRSA) A hybrid algorithm of parallel recombinative simulated annealing (PRSA) blending desirable features of genetic algorithms (GA) briefly discussed above, with standard SA has been proposed [24,93]. Crossover and mutation techniques of GA are performed during various stages of SA. Parallelism is a feature typically incorporated with GA, and this is added here as well. If two conditions are strictly satisfied (a) that the system can move to an optimal solution in a finite number of ....

S.W. Mahfoud and D.E. Goldberg, Parallel recombinative simulated annealing: A genetic algorithm, IlliGAL Report No. 92002, University of Illinois, Urbana, IL, (1992).

....analyze and decompose, e.g. problems composed of multiple inter dependant NP hard problems, each of which has huge solution spaces. 4. 4 Parallel recombinative simulated annealing (PRSA) PRSA algorithms have some of the best attributes of both genetic algorithms and simulated annealing algorithms [122]. This class of algorithms is best understood as genetic algorithms that use Boltzmann trials between modified and existing solutions, in order to select the solutions that will exist in the next generation. The greediness of a PRSA algorithm starts low and increases during an optimization run, ....

S. W. Mahfoud and D. E. Goldberg, "Parallel recombinative simulated annealing: A genetic algorithm," Parallel Computing, vol. 21, pp. 1--28, Jan. 1995.

....Davis and Principe emphasized the relationship existing between the mutation probability and a temperature for the simple genetic algorithm [5] Suzuki [26, 27] analyzed Markov chain models of this algorithm using the analogy with simulated annealing. Rudolph [21] and Mahfoud and Goldberg [18] compared massively parallel simulated annealing and genetic algorithms. Van Nimwegen et al. 19] studied meta stability in the royal road genetic algorithm deeply with low mutation probabilities. The connection has been investigated by Cerf in a series of papers [3, 4] for a genetic algorithm ....

S.W. Mahfoud and D.E. Golberg. Parallel recombinative simulated annealing: A genetic algorithm. Parallel Computing, 21 (1995),1,1995.

.... The genetic algorithm (GA) 23, 33, 45] is a population based general purpose combinatorial optimization technique which has also proven to be quite effective for solving various NP hard VLSI DA problems [16, 17, 23, 81] Unlike SA, GAs have the advantage of being readily parallelizable [17, 33, 34, 71]. Unlike SA, however, GAs suffer from a difficulty in convergence 10 control, leading sometimes to prohibitive computational cost. Recently, researchers have been investigating thermodynamic genetic hybrid algorithms that attempt to mix GA and SA techniques to overcome the disadvantages of each ....

....SA, however, GAs suffer from a difficulty in convergence 10 control, leading sometimes to prohibitive computational cost. Recently, researchers have been investigating thermodynamic genetic hybrid algorithms that attempt to mix GA and SA techniques to overcome the disadvantages of each paradigm [10, 11, 66, 71, 92, 98]. Specifically, SA GA hybrids attempt to present a general purpose combinatorial optimization technique that is both readily parallelizable with good convergence control characteristics. One such approach is population oriented simulated annealing (POSA) 34] Chapter 5 of this dissertation ....

[Article contains additional citation context not shown here]

S.W. Mahfoud and D.E. Goldberg, "Parallel Recombinative Simulated Annealing: A Genetic Algorithm," Parallel Computing, vol. 21, 1-28, 1995.

....conditions on the sequence of temperatures fc k g k=0;1; used. The use of a Boltzmann distribution is clearly inspired by the well known results obtained for the SA [17] There have been in the literature other attempts to bring the desirable convergence properties of SA into the GA realm [3,2,11,16]. All of them have failed to prove the convergence to the optimum of their respective instances of the GA, unless they introduce basic changes in the algorithm. Davis and Pr ncipe [3] tried to use the mutation probability as a temperature parameter. They proved that as the mutation probability ....

....Bilbro et al. 2] give an algorithm called genetic annealing algorithm where they consider basically a SA but, instead of maintaining an individual in each step, they have a population. They showed that this algorithm can inherit the convergence properties of SA. In Mahfoud and Goldberg [11] a summary of other attempts made in the same way can be found. In addition they propose a new algorithm called Parallel Recombinative Simulated Annealing (PRSA) PRSA can be interpreted as a GA where to choose the individuals that go to the next population a competition is carried out between the ....

S.W. Mahfoud and D.E. Goldberg, Parallel Recombinative Simulated Annealing: A Genetic Algorithm, Parallel Computing 21 (1995) 1-28. 14

....obvious set of methods to consider here are the genetic algorithms [25] whose population based evolution models distribute over parallel machines more naturally. However, we do not wish to abandon the direct hill climbing of annealing, which has empirically performed well in this task. Goldberg [26] suggests a solution here: parallel recombinative simulated annealing (PRSA) PRSA, which has its roots in genetic algorithms, can be regarded as a strategy for synchronizing a population of annealers as they cooperatively search a cost surface. The idea is conceptually simple. Suppose in a serial ....

....Mellon, Dec. 1997. 24] K. Nakamura and L.R. Carley, A current based positive feedback technique for efficient cascode bootstrapping, Proc. VLSI Circuits Symposium, June 1991. 25] J.H. Holland. Adaptation in Nature and Artificial Systems, University of Michigan Press, Ann Arbor, 1975. [26] S. W. Mahfoud and D.E. Goldberg, Parallel Recombinative Simulated Annealing: A Genetic Algorithm, Parallel Computing, vol. 21, 1995. 27] A. Geist, A. Beguelin, J. Dongarra, W. Jiang, R. Manchek, V. Sunderam. PVM: Parallel Virtual Machine A User s Guide and Tutorial for Network Parallel ....

[Article contains additional citation context not shown here]

S. W. Mahfoud and D.E. Goldberg, "Parallel Recombinative Simulated Annealing: A Genetic Algorithm," Parallel Computing, vol. 21, 1995.

....In this section, we describe the evolutionary algorithm used by CORDS to optimize PE allocations, communication resource allocations, task assignments, and communication resource connectivities. This algorithm shares some properties with parallel recombinative simulated annealing algorithms [11], and multiobjective genetic algorithms [12] CORDS maintains a pool of architectures. A generation is a discrete unit of time. In every generation, architectures reproduce. The new architectures mutate and trade information with each other. The architectures are then ranked, relative to each ....

....which depend on the lost PE such that none of the tasks or communication resources depend on the lost PE. Communication resource allocation mutation is analogous to PE allocation mutation. Information trading: CORDS uses an evolutionary algorithm which is based on two types of genetic algorithm [11], 12] However, each of these algorithms has problems dealing with the optimization of multi dimensional information. Below, we describe these problems and explain how CORDS avoids them. In a genetic algorithm, each architecture is represented by a string, i.e. a linear array, of values. ....

S. W. Mahfoud and D. E. Goldberg, "Parallel recombinative simulated annealing: A genetic algorithm," Parallel Computing, vol. 21, pp. 1--28, Jan. 1995.

....of a genetic algorithm might require substantial computational resources, but a parallel implementation might make optimal use of the available resources. Mahfoud and Goldberg describe an interesting parallel implementation of a combination of simulated annealing with a genetic algorithm in (Mahfoud and Goldberg 1995). A parallel genetic algorithm might require O(1) fitness evaluations given enough processors. However, these fitness evaluations themselves could be performed in parallel since the criteria of equations 3.9 and 3.13 in chapter 3 are parallel iterative. The availability of parallel and hardware ....

S. W. Mahfoud and D. E. Goldberg (1995). Parallel recombinative simulated annealing: A genetic algorithm. Parallel Computing 21, 1--28.

....of the process. Other methods are based on improving fast approximate partitions by exchanging interface nodes or elements, until the partition is optimal; simulated annealing (SA) has for example been used [FMB95] for the element exchange optimisation. Combining the parallel GA with SA, as in [MG95] appears to be an attractive procedure for accelerating such a process, but remains to be tried. Parallel generation and partitioning The above approach to mesh partitioning for parallel CFD computations suffers two serious deficiencies. In the first place, it is assumed that there is sufficient ....

Mahfoud S. W. and Goldberg D. E. (1995) Parallel recombinative simulated annealing: a genetic algorithm. Parallel Computing 21: 1--28.

....same order for both sequences. As a consequence, both sequences either converge or diverge, i.e. either the optimum is not found with probability one or the sequence (F k : k 0) oscillates forever. This observation reveals that the selection mechanism must be time dependent. Mahfoud and Goldberg [18] used time homogeneous mutations fulfilling assumption (A 0 2 ) and adopted the time inhomogeneous selection method as it is known from simulated annealing. Assumption (A 0 2 ) ensures fi k = fi 0 and hence the divergence of the sequence (fi k : k 0) while the simulated annealing like ....

S. W. Mahfoud and D. E. Goldberg. Parallel recombinative simulated annealing: A genetic algorithm. Parallel Computing, 21(1):1--28, 1995. 18 author/ short title

....and the fitness ratio to zero. Although we do not provide optimal parameter schedules, nor any definite functions for changing the three parameters, we do avoid introducing a SA like temperature schedule (and so we can use Theorem 2) This is unlike a procedure used by Mahfoud and Goldberg [17] [18], where the parameters are fixed but temperature changes determine the survival probability between two pairs of parents and children. NOTATION The following notation will be used. At generation t, individuals, each having L genes, are randomly chosen. Subsequent generations at t = 1; 2; 1 1 1 ....

Mahfoud, S. W., and Goldberg, D. E., "Parallel Recombinative Simulated Annealing: a Genetic Algorithm", Parallel Computing 21, pages 1-28, 1995.

No context found.

S.W. Mahfoud and D.E. Goldberg. Parallel recombinative simulated annealing: A genetic algorithm. Parallel Computing, 21(1):1--28, 1995.

No context found.

Goldberg, D. and Mahfoud, S. (1993). Parallel Recombinative Simulated Annealing: A Genetic Algorithm.

No context found.

S. W. Mahfoud and D. E. Goldberg, "Parallel recombinative simulated annealing: A genetic algorithm," Parallel Computing, vol. 21, pp. 1-28, Jan. 1995.

No context found.

S. W. Mahfoud and D. E. Goldberg. Parallel Recombinative Simulated Annealing: a Genetic Algorithm. Parallel Computing, pages 1--28, 1995.

No context found.

S. W. Mahfoud and D. E. Goldberg. Parallel recombinative simulated annealing: A genetic algorithm. Parallel computing, 21:1--28, 1995. 15

No context found.

S. W. Mahfoud and D. E. Goldberg. Parallel Recombinative Simulated Annealing: A Genetic Algorithm. Parallel Computing, 21:1--28, 1995.

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