Hypergraph partition method based on multi-level framework and hyperedge migration

Abstract

The invention relates to a hypergraph partition method based on a multi-level framework and hyperedge migration. The hypergraph partition method comprises the following steps of: (1), generating an empowerment undirected hypergraph file, and storing the empowerment undirected hypergraph file in a predetermined file format; (2), reading the empowerment undirected hypergraph file, and storing a empowerment undirected hypergraph in a memory; and (3), performing partition of the empowerment undirected hypergraph, starting a step of performing partition of a multi-level hypergraph based on hyperedge migration, performing partition of the empowerment undirected hypergraph stored in the memory, and storing the final partition result in an empowerment undirected hypergraph partition file. By means of the method disclosed by the invention, the problems that the number of cross-partition nodes is too high and the uniformity degree of a hyperedge sub-set is not good in the current hypergraph partition method based on hyperedge partition can be solved; therefore, the partition sub-hypergraph is relatively uniform; and the processing efficiency of an application based on the hypergraph partition method is increased.

Description

technical field [0001] The invention relates to a method for dividing a graph, and more specifically, relates to a method for dividing a hypergraph based on a multi-level framework and hyperedge migration. Background technique [0002] In addition to the definition of general graph theory, hypergraph can also solve some problems that cannot be solved by general graph theory. Real-world applications, such as workflow management, task scheduling, and data mining, are more suitable for modeling with hypergraphs. [0003] Existing graph partitioning systems can be divided into node partitioning based methods and edge partitioning based methods. At present, there are research results of the above two division methods for the division of Putu. For hypergraphs, hypergraph partitioning methods based on node partitioning are mainly used, and hypergraph partitioning methods based on hyperedge partitioning are rare. In the existing hypergraph partitioning methods, the hypergraph par...

AI Technical Summary

Problems solved by technology

[0005] For this reason, the purpose of the present invention is to provide a hypergraph division method based on multi-level framework and hyperedge migration, by designing a hypergraph division method based on multi-level framework and hyperedge migration, to solve the current problem based on hyperedge migration In the partitioned hypergraph partitioning method, the number of cross-partitioned nodes is too high and the uniformity of hyperedge subsets is not good enough to make the partitioned sub-hypergraphs more uniform and improve the processing efficiency of applications based on hypergraph partitioning methods

Images