Re: [yt-users] Iteration Problems
Brian,
I think you might be running into a known & fixed memory leak in the halo finding stuff. It is fixed in the svn trunk version of yt, and you might consider switching to that. There will be a point release of yt very soon (today or in a few days) if you want to wait for that. There are a whole slew of changes between 1.5 and what's coming, which will be outlined in the announcement email. If you feel up to switching now, you can do that inside the yt-1.5 directory like this: svn sw http://svn.enzotools.org/yt/trunk python setup.py develop There are other possibilities, but this is the best one I can think of right now. Good luck! _______________________________________________________ sskory@physics.ucsd.edu o__ Stephen Skory http://physics.ucsd.edu/~sskory/ _.>/ _Graduate Student ________________________________(_)_\(_)_______________
As a quick note, I have recently been notified that the installation instructions in the documentation instruct users to grab yt-1.5's install script. This was my error, and I apologize -- it seems to have affected both BrianC and another user as well. I'll be updating the docs as soon as I have a new build, hopefully today or tomorrow. -Matt On Fri, Jun 25, 2010 at 1:35 PM, Stephen Skory <stephenskory@yahoo.com> wrote:
Hi all, so it looks like Brian Crosby's problem is that in order to run a multiple-cycle halo finding script, the halos object needs to be explicitly deleted. This is probably my fault with all the various levels of referencing done in halo finding, but for the time being, a script like this will *not* run away in memory. I've confirmed this on both Ranger & Triton for yt-1.7. ---- from yt.mods import * for n in range(0,28): pf = load("RD%04i/RedshiftOutput%04i" % (n,n)) halos = HaloFinder(pf,padding=0.1) halos.write_out("RD%04i/MergerHalos%04i.txt" % (n,n)) halos.write_particle_lists("RD%04i/MergerHalos%04i" % (n,n)) halos.write_particle_lists_txt("RD%04i/MergerHalos%04i" % (n,n)) del halos --- (p.s. Brian Crosby - please note that this is based on your halos.py script on Ranger, but you missed the dir names in the output commands. Without them you're just over-writing your old work!) _______________________________________________________ sskory@physics.ucsd.edu o__ Stephen Skory http://physics.ucsd.edu/~sskory/ _.>/ _Graduate Student ________________________________(_)_\(_)_______________
As a quick note, I have recently been notified that the installation instructions in the documentation instruct users to grab yt-1.5's install script. This was my error, and I apologize -- it seems to have affected both BrianC and another user as well. I'll be updating the docs as soon as I have a new build, hopefully today or tomorrow. -Matt On Fri, Jun 25, 2010 at 1:35 PM, Stephen Skory <stephenskory@yahoo.com> wrote:
Hi all, so it looks like Brian Crosby's problem is that in order to run a multiple-cycle halo finding script, the halos object needs to be explicitly deleted. This is probably my fault with all the various levels of referencing done in halo finding, but for the time being, a script like this will *not* run away in memory. I've confirmed this on both Ranger & Triton for yt-1.7. ---- from yt.mods import * for n in range(0,28): pf = load("RD%04i/RedshiftOutput%04i" % (n,n)) halos = HaloFinder(pf,padding=0.1) halos.write_out("RD%04i/MergerHalos%04i.txt" % (n,n)) halos.write_particle_lists("RD%04i/MergerHalos%04i" % (n,n)) halos.write_particle_lists_txt("RD%04i/MergerHalos%04i" % (n,n)) del halos --- (p.s. Brian Crosby - please note that this is based on your halos.py script on Ranger, but you missed the dir names in the output commands. Without them you're just over-writing your old work!) _______________________________________________________ sskory@physics.ucsd.edu o__ Stephen Skory http://physics.ucsd.edu/~sskory/ _.>/ _Graduate Student ________________________________(_)_\(_)_______________
participants (2)
-
Matthew Turk -
Stephen Skory