Thursday, February 9, 2012, 3pm
***Note special location: Johnson Center, Room B
Juan G. Restrepo
Department of Mathematics
University of Colorado, Boulder
Criticality
and Dynamic Range in Network Cascading Processes
I will present recent work on the effect of network topology on
cascading processes. The motivation for our work is a series of recent
experiments on cascades of excitation in rat cortical tissue cultures,
where it is found that these neural networks maximize their dynamic
range (the range of stimulus intensities resulting in distinguishable
network responses). We develop a theoretical framework to study the
effect of network topology on the response to a stochastic stimulus,
and find that the dynamic range is maximized when the largest
eigenvalue of the transmission probability matrix is one. In the
critical regime with maximum dynamic range, the response of the network
is characterized by excitation avalanches with power-law distributions
of size and duration. We find that these experimental signatures of
criticality are robust to the underlying network structure. Using our
theory, we characterize the network topologies that can achieve the
largest dynamic range. I will discuss potential applications of these
results to other networked systems.