I was trying to run the halo finder in parallel this morning after updating YT and the particle IO stopped working. It was working last friday with YT version 1578. I have since reverted back to 1578 and made sure that it works. So the update made sometime in between caused this error. Can someone else please confirm the error?
From G.S.
Hi G.S.,
I've been using the halo finder all day without problem on the trunk. Unfortunately, you haven't really given us enough information to figure out what the issue is, either. Any chance you could supply a traceback? Or, if it segfaults, a traceback from the core dump? I've skimmed the changesets and it all looks good to me.
Thanks!
-Matt
On Mon, Jan 25, 2010 at 5:07 PM, gso@physics.ucsd.edu wrote:
I was trying to run the halo finder in parallel this morning after updating YT and the particle IO stopped working. It was working last friday with YT version 1578. I have since reverted back to 1578 and made sure that it works. So the update made sometime in between caused this error. Can someone else please confirm the error?
From G.S.
yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
Stephen traced the problem to where the array of particles are generated, instead of returning 1 or 0 when it checking for the particle age to see if it's less than 0, it was returning false, so when running the halo finder it will say Initializing hop ..... 0 nan. He was unable to reproduce this error on his Triton account, but he verified that it was indeed in mine. So I was hoping maybe to find some other people with the same problem, but maybe it was just something screwed up with my YT, and reverting back to the older version gotten rid of the problem for me.
From G.S.
Hi G.S.,
I've been using the halo finder all day without problem on the trunk. Unfortunately, you haven't really given us enough information to figure out what the issue is, either. Any chance you could supply a traceback? Or, if it segfaults, a traceback from the core dump? I've skimmed the changesets and it all looks good to me.
Thanks!
-Matt
On Mon, Jan 25, 2010 at 5:07 PM, gso@physics.ucsd.edu wrote:
I was trying to run the halo finder in parallel this morning after updating YT and the particle IO stopped working. It was working last friday with YT version 1578. I have since reverted back to 1578 and made sure that it works. So the update made sometime in between caused this error. Can someone else please confirm the error?
From G.S.
yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
G.S.,
I am able to run HaloFinder in parallel on Verne on one of my smaller datasets. Could you point me to the location of one of your problem datasets? And make sure it's public?
Thanks! _______________________________________________________ sskory@physics.ucsd.edu o__ Stephen Skory http://physics.ucsd.edu/~sskory/ _.>/ _Graduate Student ________________________________(_)_\(_)_______________
I am able to run HaloFinder in parallel on Verne on one of my smaller datasets. Could you point me to the location of one of your problem datasets? And make sure it's public?
Scratch that. I am seeing a problem... more to come. _______________________________________________________ sskory@physics.ucsd.edu o__ Stephen Skory http://physics.ucsd.edu/~sskory/ _.>/ _Graduate Student ________________________________(_)_\(_)_______________
Hi all,
We hammered it down to be, indeed, an issue with a null array being passed into the HDF5 reader. I think the issue was that on some platforms, for some particles, it was showing up when generating conversion factors before feeding those conversion factors into the new Particle IO. It should be fixed in r1595 in trunk and r1596.
Thanks for the tip, GSo!
-Matt
On Mon, Jan 25, 2010 at 7:36 PM, Stephen Skory stephenskory@yahoo.com wrote:
I am able to run HaloFinder in parallel on Verne on one of my smaller datasets. Could you point me to the location of one of your problem datasets? And make sure it's public?
Scratch that. I am seeing a problem... more to come. _______________________________________________________ sskory@physics.ucsd.edu o__ Stephen Skory http://physics.ucsd.edu/~sskory/ _.>/ _Graduate Student ________________________________(_)_\(_)_______________
yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
participants (3)
-
gso＠physics.ucsd.edu -
Matthew Turk -
Stephen Skory