[yt-users] load spherical polar coordinate data with mesh refinement in HDF5

Matthew Turk matthewturk at gmail.com
Mon Apr 4 14:35:17 PDT 2016 

Hi Mike and Sam,

That makes sense; one thing we need to be slightly careful about is if
the shape function needs to be modified considerably, or if it's
multi-variate.  Sam, I'm sorry I missed your message or didn't put
things together, but I think this is worth looking into further.  If
you'd like to correspond about this, yt-dev might be a better choice,
or we can chat offline and I'll send a summary.

On Mon, Apr 4, 2016 at 4:31 PM, Michael Zingale
<michael.zingale at stonybrook.edu> wrote:
> This mapped grid capability is similar to what Sam Jones e-mailed the list
> about a week or two ago.  Their grid is logically Cartesian, but a mapping
> function transforms it to a different shape.
>
> On Mon, Apr 4, 2016 at 5:13 PM, Matthew Turk <matthewturk at gmail.com> wrote:
>>
>> Hi Yan-Fei,
>>
>> I've thought about this a bunch.  I *think* that it's feasible, with
>> only very minor modifications to the existing system, as long as
>> there's a mapping we can construct for the spacing, and that mapping
>> is unique across the grids at fixed levels.
>>
>> What I mean by that is, if we can have a function that operates such
>> that it's given a set of integer values (integer coordinate [xyz],
>> resolution level) and it can return a position and a width (in linear
>> space), we can implement this without too much trouble.  I can see
>> some tricky things relating to how to switch between log/lin
>> visualization, but they should be surmountable. (And presumably, if
>> you have a log spaced dataset, you won't want linear viz terribly
>> frequently anyway.)  Generating the integer coordinate might be the
>> funniest part; in linear space it's just
>> (pos-domain_left_edge)/local_dx, but I don't know the specifics of
>> your data format.
>>
>> -Matt
>>
>> On Thu, Mar 31, 2016 at 9:55 AM, Jonah Miller
>> <jonah.maxwell.miller at gmail.com> wrote:
>> > Hi Yan-Fei,
>> >
>> > I don't think this will work out of the box. Matt may have more insight
>> > here.
>> >
>> > Best,
>> > Jonah
>> >
>> >
>> > On 16-03-31 01:03 AM, Jiang, Yanfei wrote:
>> >
>> > Hi Matt and Jonah,
>> >
>> >     I think I can use load_hexahedral_mesh to load data from each mesh
>> > block. But it cannot use three arrays (xgrid, ygrid, zgrid) to define
>> > the
>> > whole mesh due to mesh refinement. I guess I can create a dict to
>> > collect
>> > the data from each mesh block, as in load_amr_grids (but not the same).
>> > The
>> > question is will such a data structure (a dict of hexahedral_mesh) be
>> > recognized by yt?
>> >
>> > On Wed, Mar 30, 2016 at 6:01 PM, Jonah Miller
>> > <jonah.maxwell.miller at gmail.com> wrote:
>> >>
>> >> Hi Yanfei,
>> >>
>> >> To add to Matt's comment. Here is the documentation for using
>> >> load_hexahedral_mesh.
>> >>
>> >>
>> >> http://yt-project.org/doc/examining/loading_data.html#semi-structured-grid-data
>> >>
>> >>
>> >> http://yt-project.org/doc/reference/api/generated/yt.frontends.stream.data_structures.load_hexahedral_mesh.html
>> >>
>> >> Best,
>> >> Jonah
>> >>
>> >>
>> >> On 16-03-30 05:15 PM, Matthew Turk wrote:
>> >>>
>> >>> Hi Yanfei,
>> >>>
>> >>> This is possible -- but it will also slow some things down.  You can
>> >>> use the load_hexahedral_mesh function, and there're examples of this.
>> >>> This takes away some of the optimizations that regular grids can do,
>> >>> but we're hoping to add them back in shortly.  Like the other
>> >>> functions it too takes a geometry argument.
>> >>>
>> >>> -Matt
>> >>>
>> >>> On Wed, Mar 30, 2016 at 2:49 PM, Jiang, Yanfei
>> >>> <yanfei.jiang at cfa.harvard.edu> wrote:
>> >>>>
>> >>>> Hi Jonah,
>> >>>>
>> >>>>     I am able to load the data to yt now using load_amr_grids(). I
>> >>>> notice
>> >>>> that it can also set geometry to spherical. This is great. But to
>> >>>> make
>> >>>> life
>> >>>> more complicate. Is it possible to let yt know that the grid is
>> >>>> logarithmic?
>> >>>> Currently, load_amr_grids() just take left and right edges of each
>> >>>> block. I
>> >>>> guess yt assumes the grid is uniform between left and right edges. Is
>> >>>> it
>> >>>> possible to provide arrays of coordinate values for each cell in each
>> >>>> block?
>> >>>>
>> >>>> Thanks.
>> >>>>
>> >>>> On Tue, Mar 29, 2016 at 6:58 PM, Jonah Miller
>> >>>> <jonah.maxwell.miller at gmail.com> wrote:
>> >>>>>
>> >>>>> Hi Yan-Feng,
>> >>>>>
>> >>>>> I haven't tried it, but I think there is functionality for this. You
>> >>>>> can
>> >>>>> use the load_amr_grids() function which can load amr data in
>> >>>>> Cartesian,
>> >>>>> spherical, or polar coordinates. Here's the documentation:
>> >>>>>
>> >>>>>
>> >>>>>
>> >>>>>
>> >>>>> http://yt-project.org/doc/examining/generic_array_data.html#loading-numpy-array
>> >>>>>
>> >>>>>
>> >>>>>
>> >>>>>
>> >>>>> http://yt-project.org/doc/reference/api/generated/yt.frontends.stream.data_structures.load_amr_grids.html#yt.frontends.stream.data_structures.load_amr_grids
>> >>>>>
>> >>>>> Best,
>> >>>>> Jonah
>> >>>>>
>> >>>>> On Tue, Mar 29, 2016, 6:35 PM Jiang, Yanfei
>> >>>>> <yanfei.jiang at cfa.harvard.edu>
>> >>>>> wrote:
>> >>>>>>
>> >>>>>> Hi,
>> >>>>>>           I have some data from a new version of Athena
>> >>>>>> simulations.
>> >>>>>> It is
>> >>>>>> in spherical polar coordinate with mesh refinement written in HDF5.
>> >>>>>> The data
>> >>>>>> structure is similar to flash data but not the same. Does anyone
>> >>>>>> have
>> >>>>>> suggestions to load such data to yt directly?
>> >>>>>>
>> >>>>>> Thank you.
>> >>>>>> --
>> >>>>>> Yan-Fei Jiang
>> >>>>>>
>> >>>>>> Einstein Fellow
>> >>>>>> Harvard-Smithsonian Center for Astrophysics
>> >>>>>> 60 Garden Street, Cambridge MA USA 02138
>> >>>>>> _______________________________________________
>> >>>>>> 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
>> >>>>>
>> >>>>
>> >>>>
>> >>>> --
>> >>>> Yan-Fei Jiang
>> >>>>
>> >>>> Einstein Fellow
>> >>>> Harvard-Smithsonian Center for Astrophysics
>> >>>> 60 Garden Street, Cambridge MA USA 02138
>> >>>>
>> >>>> _______________________________________________
>> >>>> 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
>> >
>> >
>> >
>> >
>> > --
>> > Yan-Fei Jiang
>> >
>> > Einstein Fellow
>> > Harvard-Smithsonian Center for Astrophysics
>> > 60 Garden Street, Cambridge MA USA 02138
>> >
>> >
>> > _______________________________________________
>> > 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
>
>
>
>
> --
> Michael Zingale
> Associate Professor
>
> Dept. of Physics & Astronomy • Stony Brook University • Stony Brook, NY
> 11794-3800
> phone:  631-632-8225
> e-mail: Michael.Zingale at stonybrook.edu
> web: http://www.astro.sunysb.edu/mzingale
> github: http://github.com/zingale
>
>
> _______________________________________________
> 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