User:Carl Boettiger/Notebook/Stochastic Population Dynamics/2010/02/09

{| width="800"
 * style="background-color: #EEE"|[[Image:owwnotebook_icon.png|128px]] Stochastic Population Dynamics
 * style="background-color: #F2F2F2" align="center"|  |Main project page
 * style="background-color: #F2F2F2" align="center"|  |Main project page


 * colspan="2"|
 * colspan="2"|

Code Development for Exploring Regimes Shifts

 * Just committed changes to the google project.   now provides a fully functional, ensemble enabled simulation written with functional API for general Gillespie simulations (at last! I've come quite a way in my C coding from those early Gillespie simulations for adaptive dynamics two years ago).  Still need to pull the API out of the single file.  Doxygen-compatible literate programming included, but needs significant expansion before it can really be used effectively as an API for other users.  Getting there... Meanwhile, on the science side:
 * Simulation represents demographic stochasticity in a population following dynamics that are capable of the saddle node bifurcation. Can add environmental noise, remove demographic stochasticity, and allow a slow parameter change that will force the saddle node bifurcation either deterministically or stochastically.

Next steps

 * add reporter functions to monitor the classic warning signals, based a single replicate (using time-averaging) and on the ensemble.
 * Variance is the easiest warning signal to implement for both cases.
 * Autocorrelation over a specified window-size
 * Power spectrum over time window (chance to become familiar with GSL's fft, though I have this analysis implemented in some old Python all ready.
 * Observer function will need to be able to pass its data up to a higher-level language like R. Might be time to learn .Call over .C interface...
 * Would be nice to have an R (command-line) and a Python based GUI, slider-bar, real-time graphing interface to the simulation to make it easy for users to explore how slow changes needed to be, how big the autocorrelation window, etc, to detect a warning signal.

Notebook Reflections
Ok, so that's part vague list of accomplishments, part dense list of goals. Code development stage doesn't really have simulations to chronicle with all there parameters and results to plot and show, but I'll get there. Still figuring out what kind of notes are most useful for this stage, will be more obvious once things are coded and I can start exploring preliminary results. [mailto:cboettig@gmail.com Comments] welcome; they can also be added to the talk page or directly to this notebook. Hope to figure out how to get a comment form bubble to appear so viewers can leave in-notebook comments without wiki-editing (possibly without logging in to OWW?). Thoughts on how to do that also appreciated.


 * }