opmd2VTK: Converter of openPMD files to VTK

This library provides an easy way to export 3D and quasi-3D simulation data from openPMD format to VTK files. Present API for OpenPMD is based on openPMD-viewer and accepts only HDF5 input format. For VTK, opmd2VTK can use either PyVTK or tvtk. PyVTK is a more lightweight dependancy, while tvtk can be much faster as it works with a newer VTK format.

Installation

The code was tested only under Linux and MacOS. The recommended installation is through the Anaconda distribution. If Anaconda is not your default Python installation, download and install it from here.

First install the dependencies:

• VTK can be installed from anaconda:

conda install vtk

• openPMD-viewer can be installed from PyPi:

pip install openPMD-viewer

• pyvtk can be installed from PyPi:

pip install pyvtk

• tvtk is a part of Mayavi and can be installed with it from PyPi:

pip install mayavi

After dependencies are satisfied, clone opmd2VTK, cd into the folder and install it:

git clone https://github.com/hightower8083/opmd2VTK.git
cd opmd2VTK
python setup.py install

Basic usage

(for pyvtk API replace opmd2VTK.tvtk with opmd2VTK.pyvtk):

• serial

from opmd_viewer import OpenPMDTimeSeries
from opmd2VTK.tvtk import Opmd2VTK

ts = OpenPMDTimeSeries('./diags/hdf5/')
conv = Opmd2VTK(ts)

conv.write_fields_vtk(iteration=ts.iterations[-1])
conv.write_species_vtk(iteration=ts.iterations[-1])

• parallel script (needs mpi4py)

from opmd_viewer import OpenPMDTimeSeries
from opmd2VTK.tvtk import Opmd2VTK

from mpi4py.MPI import COMM_WORLD  as comm

ts = OpenPMDTimeSeries('./diags/hdf5/')
conv = Opmd2VTK(ts)

for it in ts.iterations[comm.rank::comm.size]:
    conv.write_fields_vtk(iteration=it, zmin_fixed=0)
    conv.write_species_vtk(iteration=it)

Useful features :

• Angular resolution of cylindric grid is defined by the argument Nth of write_fields_vtk;
• zmin_fixed allows to fix the origin of visualization domain both for fields and particles (useful for animations);
• argument CommonMesh=True groups all fields to the same mesh, while otherwise each component is written to a separate file (resolves the issue of the staggered fields);
• sampling of field arrays allows in particual reducing the longitudinal resolution, which may often be excess for visualization needs;

Work in progress

TBD:

• error messages
• debugging