Visualization and outputting formats

Visualization and outputting formats#

Visualization#

In this workshop, we will often need to inspect the meshes, markers and functions that we create, beyond what is possible with the print function in Python.

As seen for the most basic interactions on an element level, or things such as line plots, we can use the matplotlib library to create plots within our scripts.

However, we will often deal with unstructured grids, and sometimes with curved elements, which is not easy to visualize with matplotlib.

We therefore recommend using pyvista for visualization within your codes. Pyvista can either be used for interactive plotting, or to create screen-shots/pngs. In this tutorial the primary focus is interactive plotting.

The easiest way to get going with pyvista is to use the following commands

import os, sys, pyvista

if sys.platform == "linux" and (os.getenv("CI") or pyvista.OFF_SCREEN):
    pyvista.start_xvfb(0.05)

plotter = pyvista.Plotter()
plotter.show()
plotter.screenshot("test.png")

pyvista_ndarray([[[255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255],
                  ...,
                  [255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255]],

                 [[255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255],
                  ...,
                  [255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255]],

                 [[255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255],
                  ...,
                  [255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255]],

                 ...,

                 [[255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255],
                  ...,
                  [255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255]],

                 [[255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255],
                  ...,
                  [255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255]],

                 [[255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255],
                  ...,
                  [255, 255, 255],
                  [255, 255, 255],
                  [255, 255, 255]]], dtype=uint8)

The environment variables that has been set for DOLFINx in the provided conda environment is listed below

 PYVISTA_TRAME_SERVER_PROXY_PREFIX="/proxy/"
 PYVISTA_TRAME_SERVER_PROXY_ENABLED="True"
 PYVISTA_OFF_SCREEN=false
 PYVISTA_JUPYTER_BACKEND="html"

Outputting#

Sometimes one doesn’t want to inspect or visualize the result right away. Then it is convenient to save the solution to a file that can be opened in an external visualization software, such as Paraview.

DOLFINx supports a variety of output formats compatible formats, each with its own benefits and drawbacks.

XDMFFile#

Pros#

  • Preferred format for reading in meshes and mesh tags

  • Can store multiple meshes in a single file

  • Binary format (.h5) allows efficient parallel reading

  • h5py from Python can be used to inspect the h5-files

  • h5dump can be used to inspect the files in the terminal

Cons#

  • Only supports first and second order geometries

  • Functions can only be outputted as continuous Lagrange functions of the same degree as the mesh coordinate element

  • No longer maintained by Kitware

VTXWriter#

Pros#

  • Flexible format that can output arbitrary order continuous and discontinuous Lagrange functions (similar to Pyvista)

  • Uses a binary format (binary pack) from ADIOS2 as backend, which is easy to adapt to your own purposes.

  • bpls can be used to inspect the files from terminal

Cons#

  • Some limited support for DG-0 functions (time-dependent functions currently not working)

  • Storage of a single mesh

FidesWriter#

Pros#

  • Flexible format that can output arbitrary order continuous and discontinuous Lagrange functions (similar to Pyvista)

  • Uses a binary format (binary pack) from ADIOS2 as backend, which is easy to adapt to your own purposes.

  • bpls can be used to inspect the files from terminal

Cons#

  • Can only store linear (first order) meshes

VTKFile#

Pros#

  • Flexible format that can output arbitrary order continuous and discontinuous Lagrange functions (similar to Pyvista)

Cons#

  • ASCII (text) based output that creates many files when used with multiple MPI processes

Checkpointing#

Checkpointing in DOLFINx is supported with the extension ADIOS4DOLFINx. See the online documentation for illustrative use-cases.