PHILADELPHIA –- The University of Pennsylvania School of Engineering and Applied Science has received a $7.5 million, five-year grant to improve basic understanding of network science, an emerging field of research seeking the unifying principles that govern the diverse networks that make up the economic, political and social cores of the 21st century. From Facebook and the Internet to gene regulatory networks and financial markets, there is rapidly growing demand for new knowledge to analyze, design and operate next-generation networks.
Funded by the Office of Naval Research, Penn will lead a five-school consortium of mathematicians, engineers and computer scientists to develop paradigms for study of the structure, dynamics and behaviors of large-scale physical, social and information networks and to understand and reliably predict their behavior under stress.
Advances in network science could lead to new insight into how large-scale power outages propagate from small faults in the power grid and why we are all connected by the famous six degrees of separation.
“Network science is something that is still in its infancy,” Michael Kearns, professor in the Department of Computer and Information Science at Penn, said. “But the idea is to use the mathematical knowledge of graph theory and other discrete objects, as well as domain-specific, to understand, predict and design the behavior of networks.”
A core challenge for researchers and a central issue for the government is the ability to forward and reverse engineer “local” rules and protocols that produce global behaviors, particularly when the nodes are not directly linked to a centralized coordinator, and to be able to identify network fragilities from its structure.
With Facebook, for example, people are nodes, and the connections between them are their relationships. Millions of people are connected to this network, yet not everyone is connected to everyone else. Similar structure is exhibited in the complex network of traders and investors on Wall Street. Network science will enable researchers to predict how opinions propagate in such networks and how suddenly special patterns and social norms emerge. Other questions addressed in this research include social aggregation phenomena in networks of species such as flocking in birds and schooling in fish.
“By the pattern of connections in a social network, can you identify the people involved?” Ali Jadbabaie, a professor of electrical and systems engineering at Penn, said. “Can you extract information from a network by observing only a few nodes? Are there universal rules that can help us predict the behavior of social, informational and physical networks? Unfortunately, our current mathematical understanding of this topic is not mature enough to answer these questions.”
Kearns and Jadbabaie will be joined by Shawndra Hill, assistant professor in the Department of Operations and Information Management of Penn’s Wharton School. The Penn team will lead top faculty from the California Institute of Technology; University of California, San Diego; University of California, Santa Barbara; and Naval Postgraduate School.