Coin3D 64-bit dmg for Mac

Needed for AUTO’s PLAUT07 extension (SoQt needs 64-bit Coin to compile.)

http://grey.colorado.edu/emergent/index.php/Build_from_source_(Snow_Leopard)

High dimensional basins of attraction

Some ideas about N-dim basin detection:

• Using relatively small number of points, identify possible trapping region(s). Case net to find more points, expand boundary of basin entering trapping region(s).  (via Sarah) A conley index-type approach, since we are finding an isolating neighborhood with the index of a fixed point.
• Also, view saddle points as “scattering points”. This would potentially leave “holes” in space around which there is a neighborhood that no trajectories enter.
• Can also use above to hone trapping region and study any chaotic behavior within.
• Local Lyapunov exponent detection to detect points/regions with low/high exponents–indicating stable/unstable behavior.

Github convert push to ssh (from https) to avoid username/password prompt.

http://stackoverflow.com/questions/6565357/git-push-requires-username-and-password

An alternative that seemed to work even better is here. Quick demo:

$ git config -l
user.name=Jesse Berwald
user.email=jberwald@gmail.com
github.user=caja-matematica
github.token=39acf791af96e5f89d369ff3aa3cfc4d
core.excludesfile=/Users/jberwald/.gitignore_global
core.repositoryformatversion=0
core.filemode=true
core.bare=false
core.logallrefupdates=true
core.ignorecase=true
remote.origin.fetch=+refs/heads/*:refs/remotes/origin/*
remote.origin.url=https://github.com/caja-matematica/climate_attractors.git
branch.master.remote=origin
branch.master.merge=refs/heads/master
$ git config remote.origin.url git@github.com:caja-matematica/climate_attractors    <–
$ git push origin master
Everything up-to-date

June 29

RBC: we have demonstrated a partial separation between old and new cells. By looking at those persistent generators with a long lifespan we can distinguish cells. There is still some overlap, but ary and madalena are happy with the results. The is probably a further project using zig zag persistence to hone in on the temporal dynamics, too. In the end, I’ve learned a lot about persistence and how to use it on real world data. It provides far more challenges than the nice data that the Duke camp writes papers on.

MCRN: the project on climate attractors is taking shape. We have decided to consider measuring the basins of attraction. We’ll start with Lorenz 96. Looking into using the database framework, if it would make sense to consider a point are map of Lorenz 96.

ML: the above project could double as an application to the ml work. In particular, studying the bifurcation in more detail with a system whose spatial and parameter spaces can be changed might provide a nice experiment.

June 22, 2012

• Working towards reproducing Warwick’s Lorenz data. In reading http://www2.math.uu.se/~warwick/main/rodes/JFoCM.pdf, it turns out that the flow is re-initialized at various planes. So it may be necessary to create a few instances of PoincareMap with different sections.
• Closer to compiling RADS with CAPD. With proper 64-bit Python, it now recognizes the shared objects. But, still having trouble further along:

ImportError: dlopen(/Users/jberwald/github/local/caja-matematica/rads/build/lib/rads/enclosure/cymapper.so, 2): Symbol not found: __ZTV11HenonMapper
Referenced from: /Users/jberwald/github/local/caja-matematica/rads/build/lib/rads/enclosure/cymapper.so
Expected in: flat namespace
in /Users/jberwald/github/local/caja-matematica/rads/build/lib/rads/enclosure/cymapper.so

• RBC project: Many new ideas from the SOCG conference. Zig zag persistence could provide (a) a way to connect fixed thresholds across time; (b) a way to consider interleaved interlevel/sublevel/level sets.
• It will probably be necessary to split the RBC work into two parts. The plain homology and the persistence. The results from the former (homology, plus FFT thresholding) may be strong enough of a so-so paper. It’s definitely interesting. The persistence would be stronger.
• The machine learning database work: The holy grail would be a method to detect when one is near a bifurcation, similar to Scheffer’s work. Worried the Morse sets’ robustness to noise will cause problems there (the invariant sets are by nature also robust to noise).

June 17, 2012

Modified CAPD for Lorenz poincare map. Maps intervals forward, but has division by zero error if diam( interval) >;~ 0.4.

RADS: integration of capd is part way there. make file finds limbs, etc., but g++ fails with no such file error. Something is out of order.

Other recent news:

test

EPD “lost ipython” issues

A painful couple of hours was spent searching for ways to reintroduce ipython to my new mac. Encountered the following error:

jberwald@hyperion:~$ ipython
Traceback (most recent call last):
  File "/Library/Frameworks/Python.framework/Versions/Current/bin/ipython", line 8, in <module>
    from IPython.frontend.terminal.ipapp import launch_new_instance
ImportError: No module named IPython.frontend.terminal.ipapp

Turns out, somehow, the path:

/Library/Frameworks/Python.framework/Versions/Current/lib/python2.7/site-packages

was erased from the .bashrc file recently. This path does not get appended automatically to PATH or PYTHONPATH by EPD. So even a fresh install did not fix it. Finally tracked down location of the of ipapp in EPD’s structure. Added the above to the .bashrc file. Now we’re back in business.