A Parallel Algorithm for Generating a Random Graph with a Prescribed Degree Sequence

TL;DR

This paper introduces a parallel algorithm for generating random graphs with a specified degree sequence, significantly improving efficiency and including the first parallel method for Erdős-Gallai characterization verification.

Contribution

It presents the first non-trivial parallel algorithm for Erdős-Gallai characterization and demonstrates a highly efficient shared-memory parallel graph generation method.

Findings

Achieves a speedup of 20.5 with 32 cores for graph generation.

Achieves a speedup of 23 with 32 cores for Erdős-Gallai check.

Enables efficient sampling and analysis of complex networks.

Abstract

Random graphs (or networks) have gained a significant increase of interest due to its popularity in modeling and simulating many complex real-world systems. Degree sequence is one of the most important aspects of these systems. Random graphs with a given degree sequence can capture many characteristics like dependent edges and non-binomial degree distribution that are absent in many classical random graph models such as the Erd\H{o}s-R\'{e}nyi graph model. In addition, they have important applications in the uniform sampling of random graphs, counting the number of graphs having the same degree sequence, as well as in string theory, random matrix theory, and matching theory. In this paper, we present an OpenMP-based shared-memory parallel algorithm for generating a random graph with a prescribed degree sequence, which achieves a speedup of 20.5 with 32 cores. One of the steps in our parallel algorithm requires checking the Erd\H{o}s-Gallai characterization, i.e., whether there exists a graph obeying the given degree sequence, in parallel. This paper presents the first non-trivial parallel algorithm for checking the Erd\H{o}s-Gallai characterization, which achieves a speedup of 23 using 32 cores.