Visualizing outputs

Converting output to VTK format

To convert the binary output files to VTU files, run

2vtk.py modelname

modelname should be the one defined in the config file.

To see full usage information (including options for marker data VTP files), run

2vtk.py -h

The processed VTU (node and cell data) and VTP (marker data) files can be visualized with ParaView or VisIt.

Some outputs can be disabled by editing 2vtk.py and output.cxx.

Restarted models

2vtk.py follows a restarted model's .manifest back to its parent, so the whole series converts as one:

2vtk.py modelname-r

Without -g, frames from the parent that were already converted are linked in under the restarted name (modelname-r.000000.vtu -> modelname.000000.vtu, and so on); any parent frame with no .vtu yet prints a warning naming it, instead of being silently skipped. 2vtk.py -g modelname-r converts the parent's frames first, so the series runs from frame 0.

Each frame also carries its own origin — which model and machine wrote it, and, on a restart, what it restarted from — as described in Provenance.

VTKHDF output format

DES3D can write output in the VTKHDF format (.vtkhdf), a self-contained HDF5-based format supported by ParaView 5.11+. VTKHDF files bundle all time steps into a single file, making them easier to manage for long runs than the equivalent per-step VTU collection.

Updating an existing VTKHDF file

The -u / --update-vtkhdf flag lets you add or recompute derived fields (e.g., strain-rate invariants, second invariant of stress) in an existing .vtkhdf file without regenerating it from scratch. This is useful when you want to add a newly supported output quantity to a completed run.

2vtk.py --update-vtkhdf modelname.vtkhdf

Combined with -g on a restarted model, 2vtk.py -u -g modelname-r.vtkhdf writes a .vtkhdf.series file in place alongside it, one entry per frame across the whole restart chain, so ParaView opens the run as a single series instead of one file per restart leg.

Requirements: h5py must be installed (pip install h5py or conda install h5py).

Safety: The update is performed in-place. Back up your file before running if it is large or irreplaceable. The script prints a warning and exits if the target file does not look like a valid VTKHDF file.

HDF5 file locking

On some parallel or network filesystems, HDF5 file locking can cause a BlockingIOError. DES3D disables HDF5 locking automatically when writing VTKHDF output, so ParaView can open a frame while a run is still writing it. If you encounter this error when running 2vtk.py externally, set HDF5_USE_FILE_LOCKING=FALSE in your environment.

The trade-off: with locking off, two runs that share a modelname are no longer refused, and will corrupt each other's file. Give every run its own modelname.