A few computing details ...

Most of our large-scale runs are done on the NSF Teragrid resource. Many thanks to Jeff Gardiner at UW and others for their extremely effective assistance here.

Links to our submission scripts, including an excellent walltime-estimator written by Nick Cain, will be available and updated on the group wiki.


For local runs on AMATH machines, a couple of very basic tips for getting started:

To run a code -- say in matlab, "mycode.m" in the background on a linux machine ...

ssh in
go to directory w/ code
make sure you have deleted any file in that folder with name scr.log (see below)

nohup nice -19 matlab -nojvm -nosplash -nodisplay < mycode.m > scr.log &

If you'd like, you can then type

top (and then control-q to exit)

to make sure that this is up and running. Please make sure to type this top command as well before you start things up, to make sure that the machine you're running on is sufficiently free. Most have 4 cores, so can run 4 codes at once at more or less full-speed, unless they need lots of memory.

What the above command does is it gets a code running in the background, and it will keep it running even if you log out / switch your mac off / whatever. The nice means it runs at low priority, so nobody logging in on the lab machine and trying to do something themselves will be affected significantly (unless your code eats lots of memory -- as you can see in top). Finally, the > scr.log means that whatever matlab would have displayed on the screen is dumped to this text file. Useful to periodically view it using

more scr.log

to see how things are coming.



On various platforms -- we use EPD python (free for academic use) from Enthought. Double-click install. Comes with cython, numpy, scipy, etc. You also get ipython (which is a matlab-like interface to python, with searchable command history, tab-completion, etc.) ...


Miscellaneous Mac software used by our group ...

SKIM for pdf markup: skim-app.sourceforge.net/

Papers.app, Things.app, and Evernote to stay above water!