Oliver Guenther, Tad Hogg and Bernardo A. Huberman
Dynamics of Computation Group
Xerox Palo Alto Research Center
Palo Alto, CA 94304
guenther@parc.xerox.com,
hogg@parc.xerox.com,
huberman@parc.xerox.com
We present a market solution to the problem of maintaining a spatially
extended physical system near an unstable configuration. Computational
agents representing individual controllers bid to buy or sell power that
is used to push the system to a desired unstable configuration.
Simulation results show that the auction leads to a stable control of an
otherwise unstable system by focussing the power in those parts of the
system where is most needed. We also studied the effectiveness of
different market organizations in achieving control, and we noticed that
the nearest neighbor structure uses the least amount of power to control
the unstable system while also capable of reducing the average
displacement of the chain.
postcript