Providing High Performance Computing based Models as a Service: Architecture and Services for Modeling Contagions on Large Networked Populations

Abstract

Network science emerged as an interdisciplinary field over the last 20 years, and played a central role to address fundamental problems in other fields, e.g., epidemiology, public health, and transportation, and is now part of most university curriculums. Network dynamics is a major area within network science where researchers study different forms of processes in networked populations, such as the spread of emotions, influence, opinions, flu, ebola, and mass movements. These processes often referred to individually and collectively as contagions. Contagions are increasingly studied because of their economic, social, and political impacts. Yet, resources for studying network dynamics are largely dispersed and stand-alone. Furthermore, many researchers interested in the study of networks are not computer scientists. As a result, they do not have easy access to computing and data resources. Even with the presence of software or tools, it is challenging to install, build, and maintain software. These challenges create a barrier for researchers and domain scientists. The goal of this work is the design and implementation of a research framework for modeling contagions on large networked populations. The framework consists of various systems and services that provide support for researchers and domain scientists at different stages of their research workflow.