Most applications with graphs use adjacency lists to represent them in memory. The purpose of this work is to put edge lists as a viable alternative. In order to achieve that, the algorithm of interpolation search is used. Assuming that the values in the array are uniformly distributed (i.e. there is no prior knowledge about the array), it is proven that the number of iterations has a mean and variance bounded to log2 log2 E (where E is number of directed edges). Its performance is measured and compared for edge lists of graphs with different properties. Generally, higher average degree gives better results and ErdosRenyi graphs outperform power law graphs. Additionally, the algorithm is evaluated with timing random walks on real and generated graphs.
[1]
PerlYehoshua,et al.
Interpolation searcha log logN search
,
1978
.
[2]
Gaston H. Gonnet,et al.
An algorithmic and complexity analysis of interpolation search
,
2004,
Acta Informatica.
[3]
Hosung Park,et al.
What is Twitter, a social network or a news media?
,
2010,
WWW '10.
[4]
Willy Zwaenepoel,et al.
X-Stream: edge-centric graph processing using streaming partitions
,
2013,
SOSP.
[5]
José Fernando Rodrigues,et al.
M-Flash: Fast Billion-Scale Graph Computation Using a Bimodal Block Processing Model
,
2015,
ECML/PKDD.