[yt-users] Error in making light ray for a gadget dataset

Matthew Turk matthewturk at gmail.com
Tue Mar 1 06:57:35 PST 2016


Hi Cameron,

No, I think it's unrelated -- probably clowser to the way selectors work.

On Tue, Mar 1, 2016 at 8:46 AM, Cameron Hummels <chummels at gmail.com> wrote:
> Do you guys think this is related to the problem we encountered a few months
> ago with SPH datasets where regions of "zero" density were just filtered out
> of the LightRay to avoid NaNs and such?  It seemed like our fix was too
> kludgy, but perhaps this is unrelated.
>
> On Thu, Feb 25, 2016 at 6:39 AM, Britton Smith <brittonsmith at gmail.com>
> wrote:
>>
>> Hi Bili,
>>
>> I have looked into this and it seems there are some problems with
>> generating rays with particle datasets.  I cannot tell whether yours is a
>> special case, but what seems to be happening is that certain fields are
>> being returned with different sizes.  The error happens in the LightRay code
>> because it first tries to access the "t" field, which comes back with size
>> 40.  After that, it tries to access one of the coordinate fields ("x", "y",
>> or "z") and that is being returned with size 81, which doesn't fit into the
>> arrays made to hold it that already have size 40.
>>
>> Anyway, you're not seeing this when you make the ray yourself if you first
>> try to access a field other than the "t" field.  This results in the field
>> arrays being made with the larger size.  If you do this and then try to
>> access the "t" array, you will find that the first 40 values have real
>> numbers but that the rest are just uninitialized values.
>>
>> Unfortunately, I haven't gotten much past this as I'm having trouble
>> seeing where the fcoords for this particular object are getting generated.
>> I have created an issue, which you can follow here:
>>
>> https://bitbucket.org/yt_analysis/yt/issues/1175/fcoords-and-tcoords-not-the-same-shape-for
>>
>> That's about all I can do for now, but if anyone has any thoughts, help
>> would be appreciated.
>>
>> Britton
>>
>> On Wed, Feb 24, 2016 at 7:21 PM, Bili Dong - Gmail <qobilidop at gmail.com>
>> wrote:
>>>
>>> Hi Britton,
>>>
>>> I was using the yt sample dataset 'GadgetDiskGalaxy/snapshot_200.hdf5' to
>>> test the functionality.
>>>
>>> Thanks,
>>> Bili
>>>
>>> On Wed, Feb 24, 2016 at 2:39 AM, Britton Smith <brittonsmith at gmail.com>
>>> wrote:
>>>>
>>>> Hi Bili,
>>>>
>>>> My apologies for the long delay in replying to this thread.  By now, I
>>>> think you may be working with Trident.  If things are working, that's great.
>>>> However, Trident relies on much of the machinery that you were trying to
>>>> use, so it's still possible that you see this error.  The problem is likely
>>>> related to how the LightRay generator is calculating the trajectory for the
>>>> ray and not with the underlying ray machinery which it relies on.  If it's
>>>> possible for you to give me access to this dataset, I would be happy to take
>>>> a look at this.
>>>>
>>>> Britton
>>>>
>>>> On Mon, Feb 15, 2016 at 9:30 PM, Nathan Goldbaum <nathan12343 at gmail.com>
>>>> wrote:
>>>>>
>>>>> I think Cameron, Britton, and Devin have done this for their Trident
>>>>> project - you might want to contact them.
>>>>>
>>>>> On Mon, Feb 15, 2016 at 3:23 PM, Bili Dong - Gmail
>>>>> <qobilidop at gmail.com> wrote:
>>>>>>
>>>>>> Hi Matt,
>>>>>>
>>>>>> I've tried to construct the ray and ortho_ray objects, neither of
>>>>>> which can reproduce the error. They work just fine. I haven't succeeded in
>>>>>> tracing down the error yet, but I'll keep working on that.
>>>>>>
>>>>>> Besides, I'm wondering if there are examples of successfully making
>>>>>> light rays in a particle dataset? That may help figuring out what might go
>>>>>> wrong.
>>>>>>
>>>>>> Thanks,
>>>>>> Bili
>>>>>>
>>>>>> On Fri, Feb 12, 2016 at 12:42 PM, Matthew Turk <matthewturk at gmail.com>
>>>>>> wrote:
>>>>>>>
>>>>>>> Hi Bili,
>>>>>>>
>>>>>>> Hm, I'm honestly not sure.  This looks like an oddity in the way the
>>>>>>> octree is being constructed in spatial dimensions.  Any chance you can
>>>>>>> reproduce this with a ray or ortho_ray object, which should simplify the
>>>>>>> number of moving parts?
>>>>>>>
>>>>>>> On Tue, Feb 9, 2016 at 2:04 PM, Bili Dong - Gmail
>>>>>>> <qobilidop at gmail.com> wrote:
>>>>>>>>
>>>>>>>> Hi yt users,
>>>>>>>>
>>>>>>>> I'm trying to make a light ray for a gadget dataset using this
>>>>>>>> script: http://paste.yt-project.org/show/6234/
>>>>>>>>
>>>>>>>> But I get the following error:
>>>>>>>>
>>>>>>>> ```
>>>>>>>>
>>>>>>>> /Users/qobilidop/repos/yt/yt/geometry/geometry_handler.py in
>>>>>>>> fcoords(self)
>>>>>>>>     320             c = obj.select_fcoords(self.dobj)
>>>>>>>>     321             if c.shape[0] == 0: continue
>>>>>>>> --> 322             ci[ind:ind+c.shape[0], :] = c
>>>>>>>>     323             ind += c.shape[0]
>>>>>>>>     324         return ci
>>>>>>>>
>>>>>>>> ValueError: could not broadcast input array from shape (81,3) into
>>>>>>>> shape (40,3)
>>>>>>>>
>>>>>>>> ```
>>>>>>>>
>>>>>>>>
>>>>>>>> My yt is on the changeset d47150dfbde6.
>>>>>>>>
>>>>>>>>
>>>>>>>> I'm wondering if I'm doing it the correct way? And how could I solve
>>>>>>>> the problem?
>>>>>>>>
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>>
>>>>>>>> Bili
>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> 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
>>
>
>
>
> --
> Cameron Hummels
> NSF Postdoctoral Fellow
> Department of Astronomy
> California Institute of Technology
> http://chummels.org
>
> _______________________________________________
> yt-users mailing list
> yt-users at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>



More information about the yt-users mailing list