[petsc-dev] hdf5 self-describing output?

[Matthew Knepley]

On Sat, Oct 15, 2011 at 4:35 PM, Jed Brown <jedbrown at mcs.anl.gov> wrote:

> On Sat, Oct 15, 2011 at 16:24, Matthew Knepley <knepley at gmail.com> wrote:
>
>> Jed just added some VTK output that has appended binary data. I plan to
>> add the strategy we use in
>> PyLith which is to write an HDF5 file, and then an xdmf file to describe
>> that layout to VTK.
>>
>
>  You can -snes_view_solution_vtk foo.vts or -ts_monitor_solution_vtk
> foo-%03d.vts to get binary output for anything that was based on a DA.
> Notes:
>
> 1. If you set field names with DMDASetFieldName(), those will be present in
> the VTK file (hence your GUI of choice).
>
> 2. If you name Vecs differently, you can view multiple Vecs to the same
> file provided they are associated with a DMDA. The variable name will
> include the Vec name.
>
> 3. Time series can't go in the same file because VTK is a lame format.
>

You can however point the xdmf to an HDF5 file with multiple timesteps.

   Matt


> 4. You can't view multiple Vecs on _different_ DAs because VTK is a lame
> format.
>
> 5. DMComposite can't go in the same file because VTK is a lame format. (It
> could go to different files, but I haven't written support for this.)
>
>


-- 
What most experimenters take for granted before they begin their experiments
is infinitely more interesting than any results to which their experiments
lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20111016/299c3079/attachment.html>