Title:A Scalable Acyclic Graph Generator

Abstract:Acyclic graphs, much like scale-free graphs, appear in many areas of computation and engineering. These include, but are not limited to, knowledge representation, binary decision diagrams, dependency graphs, semantic web, and, binaries of computer programs. In the field of life-sciences and bio-informatics, in particular, such structures are used to create ontologies that represent the compendium of factual information and are central towards organizing and cataloging scientific data. I show why most naive generators might produce degenerate (bipartite graphs) and propose a generator that can produce a stochastic acyclic graph satisfying constraints on the in-degree, the out-degree, the depth, and the path-length (distribution of random walks from source to sink) attributes of the resulting dag. I provide suit of distribution functions that yield non-degenerate real-world acyclic graphs.