yt-dev July 2014

yt-dev@python.org

16 participants
70 discussions

by Nathan Goldbaum

Hi all, There has been some lag with the bitbucket API lately. Unfortunately when Fido tries to pull changes via the API it shows no changes even if there are new changes via the web interface. Long story short, if fido says your tests fail, take a look at the console log. If it says near the bottom that there was nothing to merge in, you've hit the bitbucket API issue. Nathan

6 years

TimeSeries Objects

by Cameron Hummels

Hey everyone, I just discovered by accident that one can now generate a TimeSeries DatasetSeries object with the load() command in yt-3.0. import yt time_series = yt.load('DD????/DD????') for ds in time_series: print ds.current_time I don't know when this functionality was created or by whom, but it's great and simplifies the timeseries analysis steps. Does it work for all frontends? I'm happy to update the timeseries documentation with this information, but I wanted to make sure it worked with all frontends, and see if there were any other new tricks that were added when this functionality was added .so we don't have to iterate too much on the docs update. Since I didn't write this functionality, I don't know what else was done and it doesn't look like it was written down in the docs. Cameron -- Cameron Hummels Postdoctoral Researcher Steward Observatory University of Arizona http://chummels.org

6 years

adding datasets

by Michael Zingale

Hi, I'd like to add some BoxLib datasets to the suite of public datasets available for testing and experiementation. Can someone point me to the procedure of how to do that? and any guidelines regarding size. Also, are both 2- and 3-d (and 1-d?) datasets acceptable there? Thanks, Mike -- Michael Zingale Associate Professor Dept. of Physics & Astronomy • Stony Brook University • Stony Brook, NY 11794-3800 *phone*: 631-632-8225 *e-mail*: Michael.Zingale(a)stonybrook.edu *web*: http://www.astro.sunysb.edu/mzingale

6 years

Removing some analysis modules, removing yt.analysis_modules.api

by John ZuHone

Hi all, I've issued a PR with some potentially controversial changes: https://bitbucket.org/yt_analysis/yt/pull-request/1009/removing-analysis-mo… which removes a few analysis modules as per this Trello card: https://trello.com/c/tJuM4kkk/23-remove-analysis-modules-as-needed Britton has also suggested that we do away with yt.analysis_modules.api entirely, as importing it imports a whole bunch of stuff you don't need, especially when you only need one module at a time (which is usually the case). I'm +1 on this, but wanted to get others' takes. Best, John

6 years

Scipy sprint today and tomorrow morning

by Nathan Goldbaum

Hi all, A group of us are here in Austin for the scipy conference. Today we're going to be sprinting on yt, focusing mostly on closing tickets for the yt-3.0 release. If you're in Austin, please feel free to come on down to the AT&T center to meet with us. My cell phone number is 720-201-2231 - give me a call if you need help finding us. We'd love to have remote participants. I will be watching the IRC channel - we could probably also set up a google hangout if there is interest. Hope you all are having a great Friday :) -Nathan

6 years

SDF pull request

by Matthew Turk

Hi all, Can we get some eyes on the SDF pull request? Since the that uses it went out yesterday, I was hoping to get it in, as long as we can identify any blockers that can and will be addressed before 3.0 goes out. https://bitbucket.org/yt_analysis/yt/pull-request/955/wip-sdf-frontend-and-… I think it now depends on thingking when actually being used for http SDF, but that's pure python and easy to install. -Matt

6 years

Issue #855: [flash] undefined 'unitsystems' (yt_analysis/yt)

by Kacper Kowalik

New issue 855: [flash] undefined 'unitsystems' https://bitbucket.org/yt_analysis/yt/issue/855/flash-undefined-unitsystems Kacper Kowalik: Pull request #1003 causes: ``` #!python import yt yt.load("Orbit/orbit_hdf5_chk_0000") ``` to fail with: ``` #!text Traceback (most recent call last): File "<stdin>", line 1, in <module> File "yt/src/yt-my/yt/convenience.py", line 80, in load return output_type_registry[candidates[0]](*args, **kwargs) File "yt/src/yt-my/yt/frontends/flash/data_structures.py", line 206, in __init__ Dataset.__init__(self, filename, dataset_type) File "yt/src/yt-my/yt/data_objects/static_output.py", line 186, in __init__ self.set_units() File "yt/src/yt-my/yt/data_objects/static_output.py", line 603, in set_units self.set_code_units() File "yt/src/yt-my/yt/frontends/flash/data_structures.py", line 241, in set_code_units super(FLASHDataset, self).set_code_units() File "yt/src/yt-my/yt/data_objects/static_output.py", line 631, in set_code_units self._set_code_unit_attributes() File "yt/src/yt-my/yt/frontends/flash/data_structures.py", line 214, in _set_code_unit_attributes if self.parameters['unitsystem'].lower() == "cgs": KeyError: 'unitsystem' ``` Responsible: jzuhone

6 years

yt with python3

by Jill Naiman

Hi Folks! So, I'm trying to build yt with python3. My python/c is still noobish, but so far I've converted everything with 2to3, however I'm getting some errors in regards to the kdtree libraries. I was wondering if anybody has tried a python3 build and/or if anybody has an idea on how to properly install new kdtree libraries for python3? Thanks for any insights! -Jill

6 years

Forthon 0.8.13

by Douglas Harvey Rudd

Hi all, Has anyone tried to build YT using a recent version of Forthon? I've been building using 0.8.10 on an older Python installation but when I use any of 3 installations using Forthon 0.8.13 (pip installed, 0.8.15 is current) it fails due to misunderstanding how to construct the module fortran file from the module name: Traceback (most recent call last): File "/software/python-2.7-2014q2-el6-x86_64/bin/Forthon", line 2, in <module> import Forthon.Forthon_builder File "/software/python-2.7-2014q2-el6-x86_64/lib/python2.7/site-packages/Forthon/Forthon_builder.py", line 87, in <module> raise Exception('Main fortran file can not be found, please specify using the fortranfile option') Exception: Main fortran file can not be found, please specify using the fortranfile option .... subprocess.CalledProcessError: Command '['/software/python-2.7-2014q2-el6-x86_64/bin/Forthon', '-F', 'gfortran', '--compile_first', 'fKD_source', '--no2underscores', '--fopt', "'-O3'", 'fKD', 'fKD_source.f90']' returned non-zero exit status 1 Looking through the differences in Forthon_builder.py between those two versions looks like fortranfile is set to a default in the older version: 0.8.10:34 fortranfile = options.fortranfile or (pkg + '.' + options.fixed_suffix) 0.8.13:34 fortranfile = options.fortranfile Adding a --fortranfile option for fKD.f90 seems to work, but I'm not certain this is the correct solution. I'll submit a PR for the following change unless anyone else has a better idea (or reason it's a bad idea): setup.py:63 cmd = [Forthon_exe, "-F", "gfortran", "--compile_first", "fKD_source", "--no2underscores", "--fopt", "'-O3'", "fKD", "fKD_source.f90"] -> cmd = [Forthon_exe, "-F", "gfortran", "--compile_first", "fKD_source", "--no2underscores", "--fopt", "'-O3'", "-f", "fKD.f90", "fKD", "fKD_source.f90"] Douglas Rudd Scientific Computing Consultant Research Computing Center drudd(a)uchicago.edu

6 years

Issue #854: Different dimensions for coordinates and fields in `smoothed_covering_grid` (yt_analysis/yt)

by John ZuHone

New issue 854: Different dimensions for coordinates and fields in `smoothed_covering_grid` https://bitbucket.org/yt_analysis/yt/issue/854/different-dimensions-for-coo… John ZuHone: This script: ``` #!python import yt import numpy as np pf = yt.load("GasSloshing/sloshing_nomag2_hdf5_plt_cnt_0150") L = 617. # width in kpc dx = 2.*pf.index.get_smallest_dx().in_units("kpc").v nx = int(L/dx) le = pf.arr([-0.5*nx*dx]*3, "kpc").in_units("code_length").v dims = np.array([nx]*3) cube = pf.smoothed_covering_grid(pf.index.max_level-1, le, dims) print cube["x"].shape print cube["density"].shape ``` Gives this output: ``` #!python (131, 131, 131) (132, 132, 132) ``` So the shapes of the coordinates and the fields have gotten out of sync somehow. This only happens for certain values of `L` and `nx`, setting `L = 600.` results in both dimensions being the same.

6 years

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

yt-dev July 2014