Some general questions about PyFR


I have been testing and analyzing the PyFR For some time, and thus have some questions about PyFR. They might not be related with each other, but for convenience, I list them all here:

  1. Is there any convenient tool to visualize the VTU file with the post-processed data? Some tools seems too complex or improper for a simple visualization usage.

  2. Is there a rule of thumb to estimate the memory space required by the software, since it seems having a high demand for memory with bigger test cases?

  3. Is there any way of verifying the correctness of solutions? Or, how to define the “correctness” in such kind of computation?

  4. What is the best practice in using MPI with OpenMP? How many partitions should I have?

  5. Where will this software be going? It seems to be focusing on solving a limited set of problems, so I have this question.

If the above questions seem to ambiguous, please point it out :slight_smile:

Thanks for any help!

Zhen Zhang

Hi Zhen,

Regarding 1:


Regarding 3:

Some test problems have analytical solutions, such as the 2D ‘Euler Vortex’ and ‘Couette Flow’ test cases supplied with PyFR. For these you can assess accuracy by comparing with the analytical solutions (see for example). Other problems (the majority!) do not have nice analytical solutions, and hence you may need to compare with experimental data, or other peoples CFD results.

Regarding 4:

See this recent thread, and this post in particular:!msg/pyfrmailinglist/osp16U_0UCE/cZSVSjcZaI8J

Regarding 5:

The current focus is on solving compressible Navier-Stokes problems. We are starting a project to look at incompressible problems soon. The framework could also be extended to solve other governing systems.



The _v t u file can also be opened by tecpolt.