[yt-users] error in light_ray periodic ray creation

Joshua Eric Moloney Joshua.Moloney at Colorado.EDU
Mon Jul 30 11:39:43 PDT 2012 

Britton,

I updated to the yt-dev branch with your fix in it, but it still doesn't
seem to have fixed the problem. I reran my script (modified for the update
version of yt) with the same seed values, and it fails in the same place
with an identical error.

The light ray segment that it fails on is

[ 0.16103374  0.78830872  0.24846128]    to     [ 0.13268221  0.49455532
1.2770327 ]

which is still getting split into subsegments:

[[array([ 0.16103374,  0.78830872,  0.24846128]), array([ 0.14031834,
0.57367409,  1.        ])], [array([ 0.14031834,  0.57367409,  1.
]), array([ 0.14031834,  0.57367409,  1.        ])], [array([ 0.14031834,
0.57367409,  0.        ]), array([ 0.13268221,  0.49455532,  0.2770327 ])]]

the subsegment that it actually fails on is

Getting subsegment: [0.14031833857011128, 0.5736740852620702,
0.99999999999999989] to [0.14031833857011128, 0.5736740852620702, 1.0]

which I think confirms that the issue is a rounding error leading to the
segment not getting split properly. (It looks like the start and endpoints
in the third dimension are exactly one bit off.) One thing I didn't notice
before is this error will probably only occur if you're using your
ion_balance module to get ion fractions, because it fails on a call for
interpolation. This wouldn't happen with any of the standard fields (I
think) so all it would do is attach a subsegment with nothing in it and
continue. I'll send the exact script I'm using to you off list in case
there's something I'm doing wrong in there.
         - Josh

On Sat, Jul 28, 2012 at 2:23 PM, Britton Smith <brittonsmith at gmail.com>wrote:

> Hi Josh,
>
> I think I was able to find the source of the error.  There was a case in
> which rays start on the edge of the domain and are oriented outward that I
> hadn't accounted for correctly.  I've just pushed a fix for this to the
> development branch of yt.  If you update to the yt branch within the repo,
> it should be fine.  However, there was also a minor change in the LightRay
> api, so you'll need to adjust your script a little bit.  If you change your
> script to resemble something like the following:
> http://paste.yt-project.org/show/2579/
> everything should be fine.
>
> Let me know if you encounter any more issues.
>
> Britton
>
>
> On Fri, Jul 27, 2012 at 7:18 PM, Britton Smith <brittonsmith at gmail.com>wrote:
>
>> Hi Josh,
>>
>> I'm not sure what's causing that as it's never happened to me before. I'm
>> away from my computer at the moment but I will take a look at this later.
>> Is the script you're using unaltered?  If not, please send me the one
>> you're using off-list.
>>
>> Also, please make sure to send emails to the list from the address that
>> you signed up with. Otherwise, it has to be approved by the moderator each
>> time.
>>
>> Britton
>> On Jul 27, 2012 7:05 PM, "Joshua Eric Moloney" <
>> Joshua.Moloney at colorado.edu> wrote:
>>
>>> The install I'm running should be the latest stable version of the code:
>>>
>>> The current version of the code is:
>>>
>>> ---
>>> c23cb264f04c (stable)
>>> ---
>>>
>>> -Josh
>>>
>>>
>>> On Fri, Jul 27, 2012 at 4:54 PM, Nathan Goldbaum <nathan12343 at gmail.com>wrote:
>>>
>>>> Hi Joshua,
>>>>
>>>> I can't speak to the problem you're having specifically.  However,
>>>> since there have been some recent updates to the light ray, it would help
>>>> us to figure out what's happening if you could reply with the output of the
>>>> "yt instinfo" command, run on the command line.
>>>>
>>>> Cheers,
>>>>
>>>> Nathan
>>>>
>>>> On Jul 27, 2012, at 3:49 PM, Joshua Eric Moloney wrote:
>>>>
>>>> Hi all,
>>>>
>>>> I've been trying to get a script (provided to me by Britton) working to
>>>> create a bunch of light rays from an Enzo simulation. Depending on the
>>>> initial seed value, it makes some of them correctly but eventually fails
>>>> with the following error:
>>>>
>>>> Traceback (most recent call last):
>>>>   File "make_light_rays.py", line 32, in <module>
>>>>     get_los_velocity=True, get_nearest_galaxy=False)
>>>>   File "/home/student/jomo9722/Enzo/
>>>> yt-x86_64/src/yt-hg/yt/analysis_modules/light_ray/light_ray.py", line
>>>> 275, in make_light_ray
>>>>     (sub_ray[field])])
>>>>   File
>>>> "/home/student/jomo9722/Enzo/yt-x86_64/src/yt-hg/yt/data_objects/data_containers.py",
>>>> line 319, in __getitem__
>>>>     self.get_data(key)
>>>>   File
>>>> "/home/student/jomo9722/Enzo/yt-x86_64/src/yt-hg/yt/data_objects/data_containers.py",
>>>> line 508, in get_data
>>>>     if field not in ("dts", "t") and self._generate_field(field):
>>>>   File
>>>> "/home/student/jomo9722/Enzo/yt-x86_64/src/yt-hg/yt/data_objects/data_containers.py",
>>>> line 349, in _generate_field
>>>>     self[field] = self.pf.field_info[field](self)
>>>>   File
>>>> "/home/student/jomo9722/Enzo/yt-x86_64/src/yt-hg/yt/data_objects/field_info_container.py",
>>>> line 316, in __call__
>>>>     dd = self._function(self, data)
>>>>   File
>>>> "/data/moloney/cosmology_test/test_new/ion_balance/SD93IonBalance.py", line
>>>> 178, in _ion_number_density
>>>>     return solarAbundance[atom] * data[fractionField] *
>>>> data['Metallicity'] * \
>>>>   File
>>>> "/home/student/jomo9722/Enzo/yt-x86_64/src/yt-hg/yt/data_objects/data_containers.py",
>>>> line 319, in __getitem__
>>>>     self.get_data(key)
>>>>   File
>>>> "/home/student/jomo9722/Enzo/yt-x86_64/src/yt-hg/yt/data_objects/data_containers.py",
>>>> line 508, in get_data
>>>>     if field not in ("dts", "t") and self._generate_field(field):
>>>>   File
>>>> "/home/student/jomo9722/Enzo/yt-x86_64/src/yt-hg/yt/data_objects/data_containers.py",
>>>> line 349, in _generate_field
>>>>     self[field] = self.pf.field_info[field](self)
>>>>   File
>>>> "/home/student/jomo9722/Enzo/yt-x86_64/src/yt-hg/yt/data_objects/field_info_container.py",
>>>> line 316, in __call__
>>>>     dd = self._function(self, data)
>>>>   File
>>>> "/data/moloney/cosmology_test/test_new/ion_balance/SD93IonBalance.py", line
>>>> 193, in _ion_fraction_field
>>>>     fraction = na.power(10, interp(data))
>>>>   File
>>>> "/home/student/jomo9722/Enzo/yt-x86_64/src/yt-hg/yt/utilities/linear_interpolators.py",
>>>> line 44, in __call__
>>>>     x_i = (na.digitize(x_vals, self.x_bins) - 1).astype('int32')
>>>> ValueError: Both x and bins must have non-zero length
>>>>
>>>> Working through the debugger, I found that for the segment where it
>>>> failed the periodic_ray function had produced the following list of
>>>> segments.
>>>>
>>>> [[array([ 0.16103374,  0.78830872,  0.24846128]), array([ 0.14031834,
>>>> 0.57367409,  1.        ])], [array([ 0.14031834,  0.57367409,  1.
>>>> ]), array([ 0.14031834,  0.57367409,  1.        ])], [array([ 0.14031834,
>>>> 0.57367409,  0.        ]), array([ 0.13268221,  0.49455532,  0.2770327 ])]]
>>>>
>>>> The second segment here has identical start and end points, which is
>>>> leading yt to try to interpolate a field with nothing in it (and therefore
>>>> throwing the above error). It seems that what's happening is that due to
>>>> rounding errors the first segment being produced actually has a value < 1
>>>> for the third dimension so that when it checks to see if the segment hit
>>>> any of the box walls it thinks that it didn't. It then tries to start a new
>>>> segment that's already at the boundary of the box, producing a ray segment
>>>> containing 0 grid cells.
>>>>
>>>> Has anyone run into this problem before, and is there a fix for it
>>>> floating around out there somewhere? If not I could probably work something
>>>> up that checks to see if the distance from the segment endpoint to the wall
>>>> is < SmallNumber and set it equal to the value at the wall if it is.
>>>>
>>>> Thanks.
>>>> _______________________________________________
>>>> 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/20120730/3d6962a0/attachment.html>

More information about the yt-users mailing list