I would like to add that this feature would be *really* nice to have. Flash checkpoint files are huge, and I'm doing large (and going to be much larger) runs. The plot and particle files are nice and small, easy to store and move around. I cringe at the thought of analyzing a 5 M(cpu-hrs) run of checkpoint files for time series data. Let me know if I can help by providing test data files or anything! My OOP and python skills are probably not at the level you'd want for actual code development, but I'll pitch in where I can.

Joshua Wall

Drexel University

On Mon, May 11, 2015 at 4:40 PM John ZuHone <jzuhone@gmail.com> wrote:

I don't think we should expect this to work, actually. We never specified that it should, to my knowledge. So we need to implement it.

John ZuHone

Kavli Center for Astrophysics and Space Research
Massachusetts Institute of Technology
77 Massachusetts Ave., 37-582G
Cambridge, MA 02139
(w) 617-253-2354
(m) 781-708-5004
jzuhone@space.mit.edu
jzuhone@gmail.com
http://www.jzuhone.com

On May 11, 2015, at 4:31 PM, Nathan Goldbaum <nathan12343@gmail.com> wrote:

On Mon, May 11, 2015 at 1:24 PM, Nathan Goldbaum <nathan12343@gmail.com> wrote:

On Mon, May 11, 2015 at 1:19 PM, Joshua Wall <joshua.e.wall@gmail.com> wrote:

Dear users,

    Since I'm already 0 for 1 today, I thought I'd also ask about this error I've seen. I'm trying to get the time series of momentum density in the x-dir from a simulation in FLASH where I have both the plot and particle files with:

############################################################

ts = yt.load("/home/josh/amuse-svn/src/amuse/community/flash/SinkMomTest2/SMT_hdf5_plt_cnt_0??[!0]",
             particle_filename='/home/josh/amuse-svn/src/amuse/community/flash/SinkMomTest2/SMT_hdf5_part_0??[!0]')

times = []
Mom_x = []
Mom_y = []

for ds in ts:
   
   
    try:
        ds.add_field("total_px", function=_px, units="kg/s/m**2")
        ds.add_field("total_py", function=_py, units="kg/s/m**2")
    except:
        continue

    dd = ds.all_data()

    Mom_x.append(dd["total_px"])
    Mom_y.append(dd["total_py"])
    times.append(ds.current_time.in_units("Myr"))

###########################################################

But this gives the error:

---------------------------------------------------------------------------
IOError                                   Traceback (most recent call last)
<ipython-input-3-4be7349ef199> in <module>()
      9 Mom_y = []
     10
---> 11 for ds in ts:
     12
     13

/usr/local/lib/python2.7/dist-packages/yt/data_objects/time_series.pyc in __iter__(self)
    158         for o in self._pre_outputs:
    159             if isinstance(o, types.StringTypes):
--> 160                 ds = load(o, **self.kwargs)
    161                 self._setup_function(ds)
    162                 yield ds

/usr/local/lib/python2.7/dist-packages/yt/convenience.pyc in load(*args, **kwargs)
     78         if c._is_valid(*args, **kwargs): candidates.append(n)
     79     if len(candidates) == 1:
---> 80         return output_type_registry[candidates[0]](*args, **kwargs)
     81     if len(candidates) == 0:
     82         if ytcfg.get("yt", "enzo_db") != '' \

/usr/local/lib/python2.7/dist-packages/yt/frontends/flash/data_structures.pyc in __init__(self, filename, dataset_type, storage_filename, particle_filename, units_override)
    206                 self._particle_handle = HDF5FileHandler(self.particle_filename)
    207             except:
--> 208                 raise IOError(self.particle_filename)
    209         # These should be explicitly obtained from the file, but for now that
    210         # will wait until a reorganization of the source tree and better

IOError: /home/josh/amuse-svn/src/amuse/community/flash/SinkMomTest2/SMT_hdf5_part_0??[!0]

Unfortunately the FLASH frontend is gobbling the error h5py is raising here.

A quick way to get the real error from h5py would be to replace "raise IOError..." on line 208 with just "raise" and then rerun the script.

Ah, but now that I think about it, I don't think the glob pattern should have made it all the way down to the FLASH I/O layer.  Maybe there's a bug in the way we set up time series for FLASH datasets that include particle files?

 

 

#############################################################

Any ideas?

Joshua Wall

Drexel University

_______________________________________________
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