 Quote by Pythagorean
There's lots of ways to characterize it; Ichiro Tsuda has published a couple papers on it (I think he might have coined the term) and they all generally contain a definition or a list of characteristics... but basically we have a system that contains several transient attractors (i.e repellors) intersecting, possibly even with a true chaotic attractor.
Trajectories can get caught in one repellor for a while. It's associated with a particular lyapunov exponent, but given a perturbation (either internal or external, depending on how your system is defined), it can be knocked into another repellor/attractor where the (short time) lyapunov exponent now shifts to another value.
It's fairly common to have this kind of behavior in network of coupled spiking neurons.
|
Thanks, that's interesting. Google gave
http://chaos.c.u-tokyo.ac.jp/papers/...os_chaotic.pdf . Do you think the heteroclinic channel
http://www.ploscompbiol.org/article/...l.pcbi.1000072 is a simple (maybe non-chaotic version) of chaotic itinerancy? Actually, now reading the second article more carefully, in the section "Mathenatical Image and Models", they do explicitly discuss chaotic itinerancy as being similar, topologically, but they weren't able to find parameters that made it work for their application.