Copyrights to these papers may be held by the publishers. The download files are preprints. It is understood that all persons copying this information will adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.
Richard Veras, Tze-Meng Low and Franz Franchetti (Proc. High Performance Extreme Computing (HPEC), pp. 1-7, 2016)
A Scale-Free Structure for Power-Law Graphs
Preprint (1.5 MB)
Published paper (link to publisher)
Many real-world graphs, such as those that arise from the web, biology and transportation, appear random and without a structure that can be exploited for performance on modern computer architectures. However, these graphs have a scale-free graph topology that can be leveraged for locality. Existing sparse data formats are not designed to take advantage of this structure. They focus primarily on reducing storage requirements and improving the cost of certain matrix operations for these large data sets. Therefore, we propose a data structure for storing real-world scale-free graphs in a sparse and hierarchical fashion. By maintaining the structure of the graph, we preserve locality in the graph and in the cache. For synthetic scale-free graph data we outperform the state of the art for graphs with up to 10^7 non-zero edges.Keywords: (No keyword)