[Users] CarpetIOHDF5 to S5 converter for yt

Jonah Miller jonah.maxwell.miller at gmail.com
Thu Oct 6 20:45:15 CDT 2016

Hi Roland,

Here are build and use instructions for SimulationIO and the 
SimulationIO converter. You can find much of the same information in the 
slides for my presentation at the ET workshop:


Let me know if you have any trouble.



To build the release version of SimulationIO, use my fork of Erik's repo:


And the instructions are here:


This installs the SimulationIO reader and the python API. It also adds 
the utilities sio-convert-carpet-output and sio-list, which you can use 
to generate and look at SimulationIO files.

The version of yt with a working SimulationIO frontend is:


To use it:

hg clone ssh://hg@bitbucket.org/Yurlungur/yt-simulationio

cd yt-simulationio

hg update simulationio

python setup.py install

(add --user or --prefix=path or whatever you prefer)

Then you can load simulationio files by

import yt

ds = yt.load(filename,configuration=configuration)

where configuration is a name for the time level and timestep in the 
SimulationIO file. You may also need to specify that there are no ghost 
zones, etc.

On 16-10-05 06:46 PM, Roland Haas wrote:
> Hello Erik, Jonah,
> yes, Llama seems involved. If I use only the data for the Cartesian
> grid it aborts with a different error:
>    opening dataset WEYLSCAL4::Psi4r it=357120 tl=0 m=0 rl=5 c=6...
>      field name: WEYLSCAL4::Psi4r
>      tensor rank: 0
>      tensor type: Scalar3D
>      tensor indices: []
>      iteration: 357120
>      time level: 0
>      map: 0
>      refinement level: 5
>      component: 6
> sio-convert-carpet-output: convert-carpet-output.cpp:612: main(int, char**)::<lambda(const H5::Group&, const string&, const H5L_info_t*)>: Assertion `loc == val' failed.
> Aborted (core dumped)
> Yours,
> Roland
>> Roland
>> Yes, such a converter exists; it is part of the SimulationIO package <
>> https://github.com/eschnett/SimulationIO>, and is called
>> "sio-convert-carpet-output". It should be built by the default makefile
>> target.
>> You call it as "./sio-convert-carpet-output [--copy|--extlink]? <output
>> file name> {<input file name>}*". There is one output file, and there can
>> be multiple input files (e.g. one for each process). The "copy" option
>> copies datasets, the "extlink" option creates external links to the
>> original HDF5 files. The latter is faster and creates smaller files, but
>> you obviously need to keep the originals around, and need to preserve their
>> path (either their relative or absolute path, depending on what you
>> specified on the command line).
>> I believe one of these two options is currently broken in the development
>> version, sorry about this. I don't recall which.
>> -erik
>> On Fri, Sep 30, 2016 at 5:01 PM, Roland Haas <rhaas at illinois.edu> wrote:
>>> Hello Jonah, Einstein Toolkit users,
>>> I think you were working on a plugin to read S5 files into yt, right?
>>> Do you also have a converter tool to convert CarpetIOHDF5 3d output
>>> files (vertex centered, mesh refined) to S5 files? If only cell
>>> centering is supported this is also fine as or this project a slight
>>> inaccuracy does not matter.
>>> I would be working with Nathan Goldbaum locally to read them into yt so
>>> once I have them in that format I will hopefully be home free.
>>> Yours,
>>> Roland
>>> --
>>> My email is as private as my paper mail. I therefore support encrypting
>>> and signing email messages. Get my PGP key from http://keys.gnupg.net.
>>> _______________________________________________
>>> Users mailing list
>>> Users at einsteintoolkit.org
>>> http://lists.einsteintoolkit.org/mailman/listinfo/users
> _______________________________________________
> Users mailing list
> Users at einsteintoolkit.org
> http://lists.einsteintoolkit.org/mailman/listinfo/users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.einsteintoolkit.org/pipermail/users/attachments/20161006/eb4b56a6/attachment.html 

More information about the Users mailing list