Re: [h5md-user] VMD reader plugin for H5MD

[Peter Colberg]

Hi Olaf,

On Mon, Apr 15, 2013 at 03:05:35PM +0200, Olaf Lenz wrote:
> I think one of the things that the h5md format is lacking at the moment
> is that there is no simple way to visualize h5md files.

I briefly started writing a VMD plugin a while ago. The idea was to
integrate the LuaJIT engine using a molfile plugin [1], and then run
a Lua script to load the particle data from an H5MD file.
I scratched the idea when confronted with the plugin callback
structure. It contains function pointers with up to 10 (or 20?)
arguments, which is not supported by LuaJIT callbacks.

Then, the availability of the source code only after registration,
the non-free license [2], the missing source code repository, and
the GUI, were simply not worth the trouble to write a full-fledged
molfile plugin in pure C/C++.

[1] http://www-s.ks.uiuc.edu/Research/vmd/plugins/doxygen/molfileplugins.html
[2] http://www-s.ks.uiuc.edu/Research/vmd/current/LICENSE.html


So I went the manual route, and now use the VTK toolkit [3] for
visualisation of trajectories. I wrote a partial LuaJIT wrapper
for the VTK C++ library. While the code to display a trajectory
is quite verbose, it is also rock solid and very fast. LuaJIT
compiles the script code to machine code, so the display script
can perform numerical preprocessing if needed.

For a production-ready solution, I would recommend either mayavi [4],
or the Python bindings for VTK [5]. If your distribution does not
ship these packages, a binary Python bundle [6] is provided by the
same company that develops mayavi.

[3] http://www.vtk.org/VTK/resources/software.html
[4] http://docs.enthought.com/mayavi/mayavi/mlab.html
[5] http://www.vtk.org/Wiki/VTK/Examples/Python
[6] https://www.enthought.com/products/epd/

> As we want to use h5md in our software ESPResSo, one of our students
> (Sascha Ehrhardt) has started to write a file reader plugin for the
> VMD software
> 
>   http://www.ks.uiuc.edu/Research/vmd/
> 
> If anybody has any material that might help him, we would be happy to
> hear about that. Of course, we will write to the list as soon as we have
> something that works.

I will be happy to help with any HDF5 questions that arise.

Peter

trajectory.lua
Description: Text document