On Fri, Jul 1, 2016 at 4:37 PM, Stephanie Tonnesen <stonnes@gmail.com> wrote:
Hi Nathan,

Yahoo!  doing the conda install instead of pip has fixed it! 

The 4th of July weekend is saved!
 
For completion, though, here are the answers to the other two questions:

Yes, it all runs nicely (until scipy is needed) if I don't have scipy installed. 

If I start python and type

import scipy.optimize

I get a Segmentation fault --when I used the pip install version.  with conda install, import scipy.optimize seems to work fine.

This must be a bug in scipy then, triggered by pip installing in whatever environment you have set up on this cluster. 
 
Thanks!!!
Stephanie


--
Dr. Stephanie Tonnesen
Alvin E. Nashman Postdoctoral Fellow
Carnegie Observatories, Pasadena, CA
stonnes@gmail.com

On Fri, Jul 1, 2016 at 2:15 PM, Nathan Goldbaum <nathan12343@gmail.com> wrote:

On Fri, Jul 1, 2016 at 4:00 PM, Stephanie Tonnesen <stonnes@gmail.com> wrote:
Hi Nathan,

I am not sure how to get a core dump or how to make gdb give me useful things.  But maybe this will help...

By running my code via a command line, I find that
import h5py
import scipy
import yt
from yt.analysis_modules.cosmological_observation.api import LightRay

all are fine.  But,

from yt.analysis.modules.absorption_spectrum.api import AbsorptionSpectrum
from yt.analysis.modules.absorption_spectrum.api import generate_total_fit

And both of these succeed when scipy isn't installed?

What happens if you do:

import scipy.optimize

Also you're using the conda-based python that the latest version of the install script builds, right? Maybe you should try the conda binary builds instead:

$ pip uninstall scipy
$ conda install scipy
 

both give me a Segmentation fault and kick me out of python

I can keep trying to figure out how to have gdb give me something useful, but does this give any clues in the meantime?

Thanks,
Stephanie

--
Dr. Stephanie Tonnesen
Alvin E. Nashman Postdoctoral Fellow
Carnegie Observatories, Pasadena, CA
stonnes@gmail.com

On Fri, Jul 1, 2016 at 12:28 PM, Nathan Goldbaum <nathan12343@gmail.com> wrote:
Hi Stephanie,

It's very hard to say what's going wrong without more information.

Can you get a core dump or run python under gdb to get a stack trace at the place where the segmentation fault happens?

Can you run this on your laptop? If so, do you get a segmentation fault there?

-Nathan

On Fri, Jul 1, 2016 at 2:25 PM, Stephanie Tonnesen <stonnes@gmail.com> wrote:
Hi all,

I have the yt-conda installation of yt.  I am trying to create an absorption spectrum, so need the command "generate_total_fit".

yt told me I needed scipy for this to work.  yt does not have scipy.  So I typed:
>pip install scipy

scipy was installed.

However, then my submission script no longer worked at all and I was given the error:

/var/spool/torque/mom_priv/jobs/50052.mies.stanford.edu.SC: line 26:  3750 Segmentation fault      python HI_absorbers.py

When I pip uninstall scipy, my code runs again! 
The only commands in my script are
source /home/stonnes/yt-conda/bin/activate
python HI_absorbers.py

Can anyone help me out with this?

Thanks!
Stephanie

--
Dr. Stephanie Tonnesen
Alvin E. Nashman Postdoctoral Fellow
Carnegie Observatories, Pasadena, CA
stonnes@gmail.com

_______________________________________________
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



_______________________________________________
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



_______________________________________________
yt-users mailing list
yt-users@lists.spacepope.org
http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org