T. N. Bui and B. R. Moon, "Genetic algorithm and graph partitioning," IEEE Transactions on Computers, vol. 45, pp. 841--855, Jul 1996.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....recursive Kernighan Lin, pairwise Kernighan Lin and Cyclic Partitioning, using many repeated applications of the optimisation algorithm on different, initial, randomly generated partitions. The authors have used the results to establish a set of high quality, benchmark partitions documented in [5]; 8 way 32 way partitioning were performed. Soper et al. have recently constructed a genetic algorithm which uses neither a linear chromosomal representation nor a traditional crossover operator, 16] The crossover is implemented by modifying the graph to record where the parents had cut edges ....

....each subdomain and so the differences vanish altogether. 4. 2 Comparison with the results of Kang Moon In this section we compare our results against a recent benchmark of Kang Moon, 10] Some of these graphs have similar structure to meshes, but some less structured examples are included, [5, 9]. Three types of graph were tested: Un.d, random geometric graphs of n vertices that lie in the unit square and whose co ordinates are chosen uniformly from the unit interval; Gridn.b, a grid graph of n vertices whose optimal bisection size is known to be b and W gridn.b, the same graph with ....

T. N. Bui and B. R. Moon. Genetic Algorithms and Graph Partitioning. IEEE Trans. Comput., 45(7):841-- 855, 1996.

....5. 2 Related Work Perhaps the best known algorithm for the graph bisection problem is the Kernighan Lin (K L) heuristic [15] and its modification by Fiduccia and Mattheyses [8] Several other algorithms for graph bisection and graph l partitioning have been proposed, including genetic algorithms [4]. Johnson et al. 13] experimentally compared the performance of the K L and simulated annealing algorithms on several random graph models. Overall, simulated annealing was found to be superior to K L on the G(n; p) graphs tested (the parameters were chosen so that in effect p = O(1=n) while ....

T. N. Bui and B. R. Moon. "Genetic algorithm and graph partitioning," IEEE Transactions on Computers, 45:7 (1996), 841--855.

....the best choice for # is closely related to a transition from ergodic to non ergodic behavior, with optimal performance of EO obtained near the edge of ergodicity. This will be the subject of future investigation. To evaluate EO, we applied the algorithm to a testbed of graphs 2 discussed in [10,18, 23,27,28]. The first set of graphs, originally introduced in [23] consists of eight geometric and eight random graphs. The geometric graphs in the testbed, labeled UN.C , are of sizes N = 500 and 1000 and connectivities C = 5, 10, 20 and 40. In a random graph, points are not related by a metric. ....

....with probability p, leading to an average connectivity C # pN . The random graphs in the testbed, labeled GNp , are of sizes N = 500 and 1000 and connectivities pN = 2.5, 5, 10 and 20. The best results reported to date on these graphs have been obtained from finely tuned GA implementations [10,27,28]. EO reproduces most of these cutsizes, and often at a fraction of the runtime, using # = 1.4 and 30 runs of 200N update steps each. Comparative results are given in the upper half of Table 1. The next set of graphs in our testbed are of larger size (up to N = 143,437) The lower half of ....

T.N. Bui, B.R. Moon, Genetic algorithm and graph partitioning, IEEE Trans. Comput. 45 (1996) 841--855.

....solution method for graph partition problems based on simple bounding functions was given by Clausen and Traff [7] All these exact solution methods have been limited to instances of general graphs with around 60 vertices. Many heuristic methods have been proposed in the literature, see e.g. [6, 10, 23, 28, 31, 34], which are designed to approximate large scale problems as they appear in real world applications. Nevertheless, it is a great challenge to develop exact methods for general problem graphs with more than 100 vertices. The approach presented in this article is certainly a step in this direction. ....

T.N. BUI and B.R. MOON. Genetic algorithm and graph partitioning. IEEE Trans. Comput., 45:841--855, 1995.

....work. In particular, different aspects of greedy heuristics are described in [20, 21, 55, 38, 5] while [33] presents a detailed empirical study of Simulated Annealing (SA) following the application proposed in [36, 37] The starting point of our investigation was the recent paper by Bui and Moon [14]. It contains a detailed discussion of previous studies and heuristics for the problem, presents a state of the art Genetic Algorithm (GA) for its solution, called BFS GBA, with extensive experimental comparisons with Simulated Annealing approaches [37, 54, 51, 57, 33] and with the classic ....

.... comparisons with Simulated Annealing approaches [37, 54, 51, 57, 33] and with the classic Kernighan Lin (KL) algorithm [35] Different implementation aspects of the KL algorithm are discussed in [23, 31, 17] while a combination of KL with SA is proposed in [41] While the reader is referred to [14] for a detailed discussion of the KL, SA, and GA heuristics, we will briefly summarize some Tabu Search approaches for the problem in Sec. 4. Tabu Search (TS) introduced by Glover [27] and, independently, by Hansen and Jaumard [29] with the term SAMD ( steepest ascent mildest descent ) is based ....

[Article contains additional citation context not shown here]

T. N. Bui and B. R. Moon, "Genetic Algorithm and Graph Partitioning," IEEE Transactions on Computers, vol. 45, no. 7, pp. 841--855, 1996.

....5. 2 Related Work Perhaps the best known algorithm for the graph bisection problem is the Kernighan Lin (K L) heuristic [15] and its modification by Fiduccia and Mattheyses [8] Several other algorithms for graph bisection and graph l partitioning have been proposed, including genetic algorithms [4]. Johnson et al. 13] experimentally compared the performance of the K L and simulated annealing algorithms on several random graph models. Overall, simulated annealing was found to be superior to K L on the G(n; p) graphs tested (the parameters were chosen so that in effect p = O(1=n) while ....

T. N. Bui and B. R. Moon. "Genetic algorithm and graph partitioning," IEEE Transactions on Computers, 45:7 (1996), 841--855.

.... (Simon, 1991) spectral partitioning (Hendrickson and Leland, 1995) and multilevel approaches (Karypis and Kumar, 1995) and (b) general purpose heuristic optimization approaches, such as simulated annealing (Johnson et al. 1989) tabu search (Battiti and Bertossi, 1998b) and genetic algorithms (Bui and Moon, 1996; Steenbeek et al. 1998; Laszewski and Muhlenbein, 1991) In this paper, we first analyze the fitness landscape of the GBP and then present a memetic algorithm that is able to produce near optimum solutions for several GBP instances efficiently. The fitness landscapes of geometric and ....

....and d the expected average vertex degree. The coordinates of the vertices are chosen randomly from within the unit square and only vertex pairs with a squared distance smaller or equal to d n constitute an edge to the graph. The second set of benchmark graphs has been provided by Bui and Moon (Bui and Moon, 1996) and consists of (c) regular graphs denoted Breg.n.b with n vertices and optimum cut size b (Bui et al. 1987) d) caterpillar graphs denoted Cat.n and Rcat.n with a known optimum cut size of 1 (Bui and Moon, 1994) and (e) regular grid graphs (grid.n.b and W grid.n.b) with optimum bisection size ....

[Article contains additional citation context not shown here]

Bui, T. N. and Moon, B. R. (1996). Genetic Algorithm and Graph Partitioning. IEEE Transactions on Computers, 45(7):841--855.

....solution method for graph partition problems based on simple bounding functions was given by Clausen and Traff [7] All these exact solution methods have been limited to instances of general graphs with around 60 vertices. Many heuristic methods have been proposed in the literature, see e.g. [6, 10, 23, 27, 30, 33], which are designed to approximate large scale problems as they appear in real world applications. Nevertheless, it is a great challenge to develop exact methods for general problem graphs with more than 100 vertices. The approach presented in this article is certainly a step in this direction. ....

T.N. BUI and B.R. MOON. Genetic algorithm and graph partitioning. IEEE Trans. Comput., 45:841--855, 1995.

No context found.

T. N. Bui and B. R. Moon, "Genetic algorithm and graph partitioning," IEEE Transactions on Computers, vol. 45, pp. 841--855, Jul 1996.

No context found.

T. N. Bui and B. R. Moon, "Genetic algorithm and graph partitioning," IEEE Trans. Computers, vol. 45, no. 7, pp. 841-855, July 1996.

No context found.

T. N. Bui and B. R. Moon, "Genetic Algorithm and Graph Partitioning," IEEE Transactions on Computers, vol. 45, no. 7, pp. 841--855, 1996.

No context found.

Bui, T.N. and B.R. Moon (1996) Genetic algorithm and graph partitioning. IEEE Trans. on Computers, 45(7), pp. 841-855.

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