CSSS 2009 Santa Fe-Blog: Difference between revisions
From Santa Fe Institute Events Wiki
No edit summary |
No edit summary |
||
| Line 1: | Line 1: | ||
{{CSSS 2009 Santa Fe}} | {{CSSS 2009 Santa Fe}} | ||
Post your reactions to lecture material, observations about the school and projects, and general conversation here. | Post your reactions to lecture material, observations about the school and projects, after hours events, and general conversation here. | ||
==Monday 6/8== | ==Monday 6/8== | ||
Revision as of 23:27, 9 June 2009
| CSSS Santa Fe 2009 |
Post your reactions to lecture material, observations about the school and projects, after hours events, and general conversation here.
Monday 6/8
Lectures:
Geoffery West & Dan Rockmore: Intro to Complex Systems
Liz Bradley: Nonlinear Dynamics I
Peter Dodds: Networks I: Introduction & Overview
jp: We're getting a few suggestions from people: -Wifi on during break/after lectures. -No NASA microphone headsets.
Lucas Lacasa: Regarding Liz talk, we had an interesting debate on lunch time regarding the choice of discrete vs continuous dynamical system when modeling a physical system that I would like to reproduce here. In discrete time systems you don't impose continuity on the variable x: x(n+1) can in principle be arbitrarily far from x(n) (depending on the specific map). Is that a problem? Within population dynamics (as the case of the logistic map Liz talked about), you can say 'ok, I'm coarse graining, between n and n+1 I allow population to vary as much as wanted, so eventually it can vary in the whole range [0,1] between n and n+1'. But if x characterizes a physical observable (thermodynamical quantity for instance such as temperature or pressure), then continuity is mandatory: small changes in time are related to small changes in the variable. Now, since in discrete systems n and n+1 are arbitrary time discretizations, you can in principle define the time unity as small as wanted. So we conclude that, in the limit, discrete dynamical systems do not fulfill the hypothesis of continuously varying variables. So, is it eventually a useless technique for physical modeling purposes? One possible answer could be the existence of different time scales: if the time scale in which a physical observable varies is much smaller than the scale where the modeling is addressed, you could use discrete time dynamical systems, couldn't you? Enough for today, maybe tomorrow I talk about randomness in discrete versus continuous time series!
Tuesday 6/9
Liz Bradley: Nonlinear Dynamics II: Flows
Peter Dodds: Networks II: Scale-free networks, power laws, history
Nathan Collins: Adaptive Modeling: Aspiration-based Models
Nathan Collins: Reinforcement Learning Methods