[yt-users] GDF format issue

[Andrea Negri]

Hi Nathan,

thanks for clarifying this. Actually what I have is an octree, where
every level is a division of the parent level. My mindset was on this
kind of grid, while apparently yt works with patches. It seems that
using a dimension of [4,4,4] for every grid in the example above, all
the structure works.

What was confising me was that in maps of my script, a 2x2 grid (in
the projection) is divided in 4, so I was assuming that the next level
was in the right place. So it seems that the left_edge and right_edge
of a grid MUST land on a grid division of the  parent level, if I
understand it correctly.


Andrea

On 4 November 2016 at 21:47, Nathan Goldbaum <nathan12343 at gmail.com> wrote:
> Hi Andrea,
>
> In your script, you have that the dimensions of the root grid are 2x2x2, but
> you have a grid on AMR level 1 that goes from 0.25 to 0.75 along z. This
> means that the grid on AMR level 1 partially overlaps with some of the root
> grid zones. This means that there are zones in your root grid that are
> "half" refined. In yt's AMR implementation this is not allowed, grids must
> be strictly nested (i.e. the boundary of a child grid *must* fall on a zone
> boundary of the enclosing parent grid).
>
> I think that's causing your issue.
>
> In principle load_amr_grids should be able to detect this case and raise an
> error message that the AMR grid structure is not well-formed.
>
> Was that a formatting error in your test script, or is this what your "real"
> data look like?
>
> -Nathan
>
> On Fri, Nov 4, 2016 at 12:58 PM, Nathan Goldbaum <nathan12343 at gmail.com>
> wrote:
>>
>> We've opened an issue to track this:
>>
>>
>> https://bitbucket.org/yt_analysis/yt/issues/1295/gdf-writing-and-loading-dataset-yields
>>
>> -Nathan
>>
>> On Fri, Nov 4, 2016 at 12:39 PM, Nathan Goldbaum <nathan12343 at gmail.com>
>> wrote:
>>>
>>>
>>>
>>> On Fri, Nov 4, 2016 at 12:36 PM, Andrea Negri <negri.andre at gmail.com>
>>> wrote:
>>>>
>>>> Hi Nathan,
>>>>
>>>> yt version returns
>>>>
>>>> ---
>>>> Version = 3.4-dev
>>>> Changeset = 334cceb2c6e1 (yt) tip
>>>> ---
>>>>
>>>>
>>>> Now I'll try to switch to the stable version.
>>>
>>>
>>> If it's broken there it's broken on stable too.
>>>
>>> Kacper is taking a look at this right now.
>>>
>>>>
>>>>
>>>> On 4 November 2016 at 18:32, Nathan Goldbaum <nathan12343 at gmail.com>
>>>> wrote:
>>>> >
>>>> > On Fri, Nov 4, 2016 at 12:28 PM, Nathan Goldbaum
>>>> > <nathan12343 at gmail.com>
>>>> > wrote:
>>>> >>
>>>> >> Hi Andrea
>>>> >>
>>>> >> Different how?
>>>> >
>>>> >
>>>> > Ah I see it's in the script you attached, sorry for the confusion.
>>>> >
>>>> >>
>>>> >> I recently fixed a bug in the GDF writer that would cause the field
>>>> >> data
>>>> >> for some grids not to get written out properly That fix should be in
>>>> >> yt
>>>> >> 3.3.2 so if you're using an older version you might want to update.
>>>> >> Here's
>>>> >> the PR with the fix:
>>>> >>
>>>> >>
>>>> >>
>>>> >> https://bitbucket.org/yt_analysis/yt/pull-requests/2296/iterate-over-all-grids-in-the-gdf-writer/diff
>>>> >>
>>>> >> Nathan
>>>> >>
>>>> >> On Friday, November 4, 2016, Andrea Negri <negri.andre at gmail.com>
>>>> >> wrote:
>>>> >>>
>>>> >>> Hi all,
>>>> >>>
>>>> >>> As you have already noted, I'm trying to use the GDF format. Just as
>>>> >>> an example I created a simple AMR grid, loaded into yt with load_amr
>>>> >>> and plotted a slice. I then write down the dataset to a gdf file
>>>> >>> using
>>>> >>> yt.utilities.grid_data_format.writer.write_to_gdf,  and after
>>>> >>> loading
>>>> >>> the very same dataset and doint the slice again, the result is
>>>> >>> different (verything is in the attached script).
>>>> >>>
>>>> >>> Curiously, if I load the sedov dataset downloaded from
>>>> >>> http://yt-project.org/data/, the slice seems to be OK, I think (but
>>>> >>> I
>>>> >>> could be wrong) that there is a problem in the gdf writer.
>>>> >>>
>>>> >>>
>>>> >>> Cheers,
>>>> >>> Andrea
>>>> >
>>>> >
>>>> >
>>>> > _______________________________________________
>>>> > 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
>