Multi-Way Partitioning Via Geometric Embeddings, Orderings, and Dynamic Programming
نویسندگان
چکیده
30 Acknowledgements We thank Pak K. Chan, Martine Schlag and Jason Zien for past research discussions, for the use of the LASO interface written by Martine Schlag, and for the use of their KP and SB codes. Ken D. Boese supplied the 3-Opt optimization code. Lars Hagen and Jen-Hsin Huang developed the ideas behind the partitioning-speciic net model. The anonymous reviewers provided many comments which substantially improved this work, particularly in the experimental design. Part of this work (ABK) was performed in part during a Spring 1993 sabbatical visit to UC Berkeley; support from NSF MIP-9117328 and the hospitality of Professor Ernest S. Kuh and his research group is gratefully acknowledged. Benz ecri, \Construction d'une Classiication Ascendante Hi erarchique par la Rechereche en Chaine des Voisins R eciproques", Les Cahiers de l'Analyse des Donn ees (VII)2, pp. Table 7: Runtimes in seconds on a Sun Sparc 10, for a single run of DP-RP for each of the ten d-dimensional embeddings (constructed from ~ 2 ; : : :; ~ d+1). Each Embedding entry gives the time required to generate d eigenvectors, and a DP-RP entry gives the time needed to generate 2-through 10-way partitioning solutions.
منابع مشابه
Multiway partitioning via geometric embeddings, orderings, and dynamic programming
AbstructThis paper presents effective algorithms for multiway partitioning. Confirming ideas originally due to Hall, we demonstrate that geometric embeddings of the circuit netlist can lead to high-quality k-way partitionings. The netlist embeddings are derived via the computation of d eigenvectors of the Laplacian for a graph representation of the netlist. As Hall did not specify how to partit...
متن کاملMulti-Way Partitioning Via Space lling Curves and Dynamic Programming
Spectral geometric embeddings of a circuit netlist can lead to fast, high quality multi-way partitioning solutions. Furthermore, it has been shown that d-dimensional spectral embeddings (d > 1) are a more powerful tool than single-eigenvector embeddings (d = 1) for multi-way partitioning [2] [4]. However, previous methods cannot fully utilize information from the spectral embedding while optimi...
متن کاملKc 34.6 33.6 34.4 30.7 27.5 16.4 17.4 13.5
Spectral geometric embeddings of a circuit netlist can lead to fast, high quality multi-way partitioning solutions. Furthermore, it has been shown that d-dimensional spectral embeddings (d > 1) are a more powerful tool than single-eigenvector embeddings (d = 1) for multi-way partitioning [2] [4]. However, previous methods cannot fully utilize information from the spectral embedding while optimi...
متن کاملSpectral Partitioning : The More Eigenvectors , The BetterCharles
The graph partitioning problem is to divide the vertices of a graph into disjoint clusters to minimize the total cost of the edges cut by the clusters. A spectral partitioning heuristic uses the graph's eigenvectors to construct a geometric representation of the graph (e.g., linear orderings) which are subsequently partitioned. Our main result shows that when all the eigenvectors are used, grap...
متن کاملSpectral Partitioning with Multiple Eigenvectors
The gvuph partitioning problem is to divide the vertices of a graph into disjoint clusters to minimize the total cost of the edges cut by the clusters. A spectral partitioning heuristic uses the graph’s eigenvectors to construct a geometric representation of the graph (e.g., linear orderings) which are subsequently partitioned. Our main result shows that when all the eigenvectors are used, grap...
متن کامل