On Ranking Components in Scientific Software

DSM 2015: Modeling and managing complex systems - Proceedings of the 17th International DSM Conference Fort Worth (Texas, USA), 4-6 November 2015

Year: 2015
Editor: Browning, T. R.; Eppinger, S. D.; Schmidt, D. M.; Lindemann, U.
Author: Hossain, S.; Khan, S. F.; Quashem, R.
Series: DSM
Section: Systems' Architecture and Modularities
Page(s): 245-254

Abstract

We explore centrality rankings of functions in call graphs of scientific software. Dependency Structure Matrix (DSM) is used as a modeling tool to capture and analyze pattern of interdependencies among functions. The non-symmetric nature of the dependency relations is addressed by computing dual scores for each function: as a caller and as a callee. The centrality schemes we utilize here are firmly based on spectral properties of the underlying call graph DSM.

Keywords: Scientific Software, Complex Network, Topological Property, Software Architecture, Spectral Analysis