Sep 10 2013
Understanding complex networked systems, from nanomachines to nation-states, and learning how to control them are the aims of two new projects led by researchers at the University of California, Davis, Complexity Sciences Center. Each project is funded with $6.25 million over five years from the U.S. Department of Defense's Multidisciplinary University Research Initiative.
One project, led by Raissa D'Souza, professor of mechanical and aerospace engineering and computer science at UC Davis, aims to find how collections of networks interact to create new, stable functions — or not. Ultimately, D'Souza hopes to find ways to intervene in networks to better control them or to prevent a catastrophic cascading breakdown.
The other grant, headed by Jim Crutchfield, professor of physics and director of the Complexity Sciences Center, will study how networks can manipulate both information and energy, and develop what Crutchfield calls a new “physics of information." Future applications could range from microscopic computers to directing molecule-sized machines.
Complex systems are more than "the sum of their parts," D'Souza said. While the individual parts may be simple, you cannot predict the system's final collective behavior from the basic parts. For example, you cannot predict the behavior of an ant colony from studying a single ant.
But once you understand the collective behavior of the whole system, you can work back to the individual parts to understand how these behaviors emerge, she said.
Scientists typically consider systems in isolation. But in the real world, any networked system is linked to many other networked systems. An anthill is linked to networks of predators, changes in climate and food availability. Your local electrical power grid relies on communication networks and new, long-distance networks of renewable sources; it is affected by networks of traders in energy markets, by natural disasters, and by policy and public sentiment.
The U.S. military, which is funding the research through the Army Research Office, depends on complex, dynamic networks that are mostly built and maintained by civilians for communications, supplies and transportation, D'Souza noted. The Army is also affected by shifts in social and political networks.
D'Souza's project will take a highly interdisciplinary approach. It will draw on both theoretical work and specific examples such as disaster recovery in critical infrastructure.
"Engineers have a lot of practical experience with how networks behave, and we want to translate that real world experience into theory," D'Souza said.
One sub-project will look at behavioral studies of monkey colonies at the University of Wisconsin’s Center for Complexity and Collective Computation, while another, based at the California Institute of Technology, will study networks of nano-electrical mechanical devices in the laboratory. Zeev Maoz and Kyle Joyce, both professors of political science at UC Davis, will collaborate on studies of trade and political alliances between nation-states.
The aim is to find general principles for intervening in networks to prevent a disastrous failure, such as a stock market crash, and to prevent a failure in one layer -- for example, an earthquake taking down the power grid -- from affecting other layers of networks.
Crutchfield's project aims to understand information processing and storage at the scale of atoms.
"Traditional physics asks questions about energy — we're asking questions about information," he said. "How much of a molecule's history is stored in the present? How is it stored? And how does it produce future behavior?"
For example, some proteins in living cells use energy to move items around, allowing cells to move, grow and carry out essential functions. These protein motors are using energy, and creating structure.
Classical thermodynamics says that energy always runs downhill, and disorder, or entropy, always increases. If there is a "secret of life," it is in using energy to create order and structure.
"Most natural systems are open; energy enters and cascades through, and some of it is shunted aside to do useful work," Crutchfield said. "Just around the corner are new laws of organization that explain how nanoscale systems circumvent increasing disorder to support life."
Additional key collaborators on the grant led by D'Souza are: at UC Davis, Crutchfield, and Vladimir Filkov and Prem Devanbu, both professors of computer science; Leonardo Dueñas-Osorio, Rice University; Jessica Flack and David Krakauer, University of Wisconsin-Madison; and Michael Roukes, Caltech. Participating on Crutchfield's project are: Gavin Crooks and Michael DeWeese, UC Berkeley; Henry Hess, Columbia University; Christopher Jarzynski and P.S. Krishnaprasad, University of Maryland.