For Immediate Release
November 29, 1995

CREATING ORDER WITH DISORDER: COMPLEX NATURAL & ARTIFICIAL SYSTEMS USE DIVERSITY, VARIABILITY TO ORGANIZE

Bringing order out of chaos can require a little disorder.

That's the conclusion drawn by a team of physicists who report in the November 30 issue of the journal Nature that adding variability and disorder to certain complex systems can help tame their chaotic behavior.

This unexpected conclusion could require scientists and engineers to take a new look at the operation and interaction of both natural and artificial nonlinear systems. It could ultimately lead to methods for improving the performance of electronic systems by exploiting variations in their components, and to new techniques for controlling disease processes such as epilepsy -- by restoring proper amounts of disorder.

"We have found that nature utilizes disorder to create organization, and that there are situations where the lack of disorder will create disorganization," said William Ditto, assistant professor of physics at the Georgia Institute of Technology. "We think many patterns we see in nature are aided by randomness and disorder. This will lead us to think about systems in dramatically different ways."

Ditto and colleagues John Lindner of The College of Wooster and Yuri Braiman of Emory University used computer simulations to study a variety of coupled nonlinear systems, including a series of chaotic pendula and a system with a hundred identical oscillators. The systems exhibited chaotic behavior over both time and space (spatiotemporal chaos), and the activity of each individual element could affect the behavior of others.

To see what would happen if they increased the disorder and variability of the chaotic systems, the researchers made each pendulum a different length, and programmed each oscillator to respond in a slightly different way.

"We expected that we would get even more disorder and even more turbulent behavior, but what we got was organized behavior patterns coming out of the systems," explained Ditto, director of Georgia Tech's Applied Chaos Laboratory. "The diversity or disorder provided a mechanism by which the systems could organize themselves."

How the process works to control chaos isn't fully understood yet, but Ditto believes the disorder may help move groups of chaotic elements into similar modes of behavior. Neighboring elements then begin to lock into the same mode, and "a local domino effect" spreads that behavior. The result is an organized system of individual elements that repeats its behavior in a complex but regular way.

But not just any amount of disorder will do. The researchers found that a 30 percent variation in the length of pendula or behavior of oscillators produced the most regular behavior patterns. Small amounts of disorder could not prompt changes in the system, while more disorder simply "overwhelmed" it.

The study demonstrates the importance of considering how natural and artificial systems interact with other systems in real-world conditions that include noise and variability, said Lindner, an associate professor at The College of Wooster.

"Real systems are never completely homogeneous and you can never work in an environment without noise," he said. "It is important to move beyond the study of completely homogeneous systems. Scientists can be misled in important qualitative ways if they simply study ideal examples."

While knowing the laws governing individual systems is important, that won't necessarily help understand the activity of complex systems made up of many individual systems, Ditto noted.

"In Monopoly, you can understand the rules of the game, but the way that everybody interacts is quite different from game to game," he explained. "You must be able to understand the consequences of the rules for a variety of players. We have to understand how systems obeying the physical rules interact."

The work described in Nature may be related to stochastic resonance, a phenomenon in which adding noise to a system actually improves its ability to receive weak signals. Stochastic resonance is already finding applications in electronic systems, and Lindner believes engineers may one day use disorder to enhance performance of electronic systems.

"For certain nonlinear systems, maybe you can not only get away with greater variability in your components, but maybe that's what you want," he explained. "A clever engineer may be able to exploit this basic phenomenon to lead to better devices. Surprising as that may sound, having a little inhomogeneity in a system may provide better performance if the elements are nonlinear."

The importance of variability and diversity is well known in the biological sciences, where systems built of identical elements are normally weaker than more diverse systems. But physical scientists and engineers tend to see diversity and variability as harmful, Ditto noted.

The next step in the research will be to determine the specific mechanisms by which disorder helps organize nonlinear systems, and which types of systems could benefit from the introduction of diversity and variability. The phenomenon must also be tested experimentally in real systems.

Besides putting a positive spin on noise and disorder, the findings also illustrate how much territory needs to be explored in the strange new world of chaotic systems that interact with each other. Just as scientists have learned to use chaos techniques to control systems like irregular heartbeat, Lindner believes they will also learn to use the non-intuitive behavior of disorder in nonlinear systems.

"When you look at nonlinear systems, you can find very non-intuitive behaviors," Lindner concluded. "We are not exactly sure where this will be ten years from now, but it is certainly very exciting. It's a new field, and there is a lot to be discovered."

RESEARCH NEWS AND PUBLICATIONS OFFICE
Georgia Institute of Technology
75 Fifth Street, N.W., Suite 100
Atlanta, Georgia 30308

MEDIA RELATIONS CONTACTS:
John Toon (404-894-6986);
Internet: john.toon@edi.gatech.edu;
FAX: (404-894-4545)

WRITER: John Toon