[yt-users] Accumulation in 3.0

Nathan Goldbaum nathan12343 at gmail.com
Tue Nov 4 11:42:25 PST 2014 

Dug into this a little bit.

The issue is that querying the particle_mass field returns CGS values, but
the field_info for particle_mass says it should be in code_mass. This is
illustrated by the following script: https://bpaste.net/show/6ee56ae61fa6

Does anyone remember why this is the case?

The reason this is coming up here is because the profiling code passes raw
numpy arrays to the cython parallel profiling machinery, so the units that
are attached to the data when we query the data source are stripped off.
We're relying on the field_info units entry for all fields being identical
to the units that get appended to the results of a field access on a data
object.

Unfortunately I don't have time right now to figure out where this
inconsistency is coming from...

-Nathan

On Tue, Nov 4, 2014 at 10:42 AM, John Regan <johnanthonyregan at gmail.com>
wrote:

> Hi Nathan,
>
> Sure. Using the enzo_tiny_cosmology I see the same problem. Script
> attached.
>
> On Tue, Nov 4, 2014 at 6:29 PM, Nathan Goldbaum <nathan12343 at gmail.com>
> wrote:
>
>> Hi John,
>>
>> Do you have a short script using one of the test datasets on
>> yt-project.org/data that reproduces this behavior?
>>
>> -Nathan
>>
>> On Tue, Nov 4, 2014 at 10:26 AM, John Regan <johnanthonyregan at gmail.com>
>> wrote:
>>
>>> Hi Matt,
>>>
>>> It looks like the particle mass is in grams but the unit assignment sets
>>> it to code_mass. Then when the units is set to Msun the numbers get crazy.
>>> I can't see easily where the units for the particle mass get set in the
>>> create_profile routine (in field/particle_fields the mass is set to g in
>>> particle_mass()). Any idea where this might be getting all mixed up or how
>>> to force the mass units to be grams in create_profile()?
>>>
>>> Cheers,
>>> John
>>>
>>> On Mon, Nov 3, 2014 at 5:18 PM, John Regan <johnanthonyregan at gmail.com>
>>> wrote:
>>>
>>>> Yeah that's what I thought first but 1 Msun = 10^33 g. So that would
>>>> put the inner DM mass at ~ 10^14 right?
>>>>
>>>> The total mass inside the sphere of 1kpc is only about 10^7 though.
>>>> Something is not quite correct somewhere in the profiling calculation.
>>>>
>>>> On Mon, Nov 3, 2014 at 5:16 PM, Matthew Turk <matthewturk at gmail.com>
>>>> wrote:
>>>>
>>>>> Yup, those sure are weird!  Looks to me like a units issue -- any
>>>>> chance those could be in grams but reported in Msun?
>>>>>
>>>>>
>>>>> On Mon Nov 03 2014 at 11:15:23 AM John Regan <
>>>>> johnanthonyregan at gmail.com> wrote:
>>>>>
>>>>>> Hi Matt,
>>>>>>
>>>>>> Ah I see what you mean now. Changing 'radius' to 'particle_radius'
>>>>>> worked nicely. However, the results are bit weird.
>>>>>>
>>>>>> rpm = yt.create_profile(sp, ['particle_radius'],  ['particle_mass'],
>>>>>>                         units = {'particle_radius': 'pc',
>>>>>> 'particle_mass' : 'Msun'},
>>>>>>                         n_bins=20, weight_field=None,
>>>>>> accumulation=True, fractional=False)
>>>>>> print "Particle Mass = ", rpm["particle_mass"]
>>>>>>
>>>>>> Particle Mass =  [  0.00000000e+00   0.00000000e+00
>>>>>> 0.00000000e+00   2.17730865e+47
>>>>>>    2.39503952e+48   5.00780990e+48   9.36242721e+48   2.87404742e+49
>>>>>>    8.03426893e+49   1.91385431e+50   4.44824158e+50   1.02681876e+51
>>>>>>    2.28900459e+51   4.74174278e+51   1.05066029e+52   2.07602025e+52
>>>>>>    3.49614805e+52   4.89060538e+52   6.82621100e+52   1.08907237e+53]
>>>>>> Msun
>>>>>>
>>>>>> Any idea what is causing the crazy values?
>>>>>>
>>>>>> On Mon, Nov 3, 2014 at 4:23 PM, Matthew Turk <matthewturk at gmail.com>
>>>>>> wrote:
>>>>>>
>>>>>>> Hi John,
>>>>>>>
>>>>>>> It was exactly the same error, when you used particle_radius in both
>>>>>>> places that "radius" was used before?
>>>>>>>
>>>>>>> -Matt
>>>>>>>
>>>>>>>
>>>>>>> On Mon Nov 03 2014 at 7:06:44 AM John Regan <
>>>>>>> johnanthonyregan at gmail.com> wrote:
>>>>>>>
>>>>>>>> Hi Kacper,
>>>>>>>>
>>>>>>>> Of course - I should have noticed that. Thanks for finding that.
>>>>>>>> However, I notice that when I try to create a profile for the Dark
>>>>>>>> (particle) matter I run into some trouble as well.
>>>>>>>>
>>>>>>>> rpm = yt.create_profile(sp, 'radius',  ['particle_mass'],
>>>>>>>>                         units = {'radius': 'pc', 'particle_mass' :
>>>>>>>> 'Msun'},
>>>>>>>>                         n_bins=20, weight_field=None,
>>>>>>>> accumulation=True, fractional=False)
>>>>>>>>
>>>>>>>> This gives the following error:
>>>>>>>>
>>>>>>>> Traceback (most recent call last):
>>>>>>>>   File "EnclosedMass.py", line 35, in <module>
>>>>>>>>     n_bins=20, weight_field=None, accumulation=True,
>>>>>>>> fractional=False)
>>>>>>>>   File
>>>>>>>> "/homeappl/home/regan/appl_taito/YT/Dev-3.0/yt/yt/data_objects/profiles.py",
>>>>>>>> line 1361, in create_profile
>>>>>>>>     obj.add_fields([field for field in fields])
>>>>>>>>   File
>>>>>>>> "/homeappl/home/regan/appl_taito/YT/Dev-3.0/yt/yt/data_objects/profiles.py",
>>>>>>>> line 782, in add_fields
>>>>>>>>     self._bin_chunk(chunk, fields, temp_storage)
>>>>>>>>   File
>>>>>>>> "/homeappl/home/regan/appl_taito/YT/Dev-3.0/yt/yt/data_objects/profiles.py",
>>>>>>>> line 979, in _bin_chunk
>>>>>>>>     rv = self._get_data(chunk, fields)
>>>>>>>>   File
>>>>>>>> "/homeappl/home/regan/appl_taito/YT/Dev-3.0/yt/yt/data_objects/profiles.py",
>>>>>>>> line 910, in _get_data
>>>>>>>>     arr[:,i] = chunk[field][filter]
>>>>>>>>   File
>>>>>>>> "/homeappl/home/regan/appl_taito/YT/Dev-3.0/yt/yt/units/yt_array.py", line
>>>>>>>> 963, in __getitem__
>>>>>>>>     ret = super(YTArray, self).__getitem__(item)
>>>>>>>> ValueError: too many boolean indices
>>>>>>>>
>>>>>>>>
>>>>>>>> Has anyone seen this before? For the 'cell_mass' everything works
>>>>>>>> fine.
>>>>>>>>
>>>>>>>> Cheers,
>>>>>>>> John
>>>>>>>>
>>>>>>>> On Sat, Nov 1, 2014 at 6:49 PM, Kacper Kowalik <
>>>>>>>> xarthisius.kk at gmail.com> wrote:
>>>>>>>>
>>>>>>>>> Hi John,
>>>>>>>>> I think that if you want to compute "total" instead of mean you
>>>>>>>>> need to set weight_field to None in create_profile.
>>>>>>>>> Cheers,
>>>>>>>>> Kacper
>>>>>>>>>
>>>>>>>>> On Sat Nov 01 2014 at 6:10:50 AM John Regan <
>>>>>>>>> johnanthonyregan at gmail.com> wrote:
>>>>>>>>>
>>>>>>>>>> Hi All,
>>>>>>>>>>
>>>>>>>>>> Is the accumulation flag working in 3.0?
>>>>>>>>>> I tried to plot the enclosed mass in a sphere and I got some
>>>>>>>>>> funny results.
>>>>>>>>>>
>>>>>>>>>> rpm = yt.create_profile(sp, 'radius',  'cell_mass',
>>>>>>>>>>                         units = {'radius': 'pc', 'cell_mass' :
>>>>>>>>>> 'Msun'},
>>>>>>>>>>                         weight_field='density',
>>>>>>>>>> accumulation=True, fractional=False)
>>>>>>>>>>
>>>>>>>>>> print "Mass = ", rpm["gas", "cell_mass"]
>>>>>>>>>>
>>>>>>>>>> In this case bin[n-1] gives a mass of something like 0.5 Msun but
>>>>>>>>>> when I print the totals quantity I get a value of several orders of
>>>>>>>>>> magnitude higher and closer to what I would expect.
>>>>>>>>>>
>>>>>>>>>> sp.quantities.total_quantity(["cell_mass", "particle_mass"])
>>>>>>>>>>
>>>>>>>>>> Cheers,
>>>>>>>>>> John
>>>>>>>>>> _______________________________________________
>>>>>>>>>> yt-users mailing list
>>>>>>>>>> yt-users at lists.spacepope.org
>>>>>>>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> yt-users mailing list
>>>>>>>>> yt-users at lists.spacepope.org
>>>>>>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>>>>>>>
>>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> yt-users mailing list
>>>>>>>> yt-users at lists.spacepope.org
>>>>>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> yt-users mailing list
>>>>>>> yt-users at lists.spacepope.org
>>>>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>>>>>
>>>>>>>
>>>>>> _______________________________________________
>>>>>> yt-users mailing list
>>>>>> yt-users at lists.spacepope.org
>>>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> yt-users mailing list
>>>>> yt-users at lists.spacepope.org
>>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>>>
>>>>>
>>>>
>>>
>>> _______________________________________________
>>> yt-users mailing list
>>> yt-users at lists.spacepope.org
>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>
>>>
>>
>> _______________________________________________
>> yt-users mailing list
>> yt-users at lists.spacepope.org
>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>
>>
>
> _______________________________________________
> yt-users mailing list
> yt-users at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.spacepope.org/pipermail/yt-users-spacepope.org/attachments/20141104/df0611ac/attachment.htm>

More information about the yt-users mailing list