Incorrect results for HDG Stokes examples

[Christoph]

I wanted to use the HDG method for my stokes problem. To get started I analyzed the examples from your 2024 Usermeeting. Unfortunately, I cannot reproduce the results you show on your website. This is your result:

I get this:

I then ran the 11.5 normal cont. HDG-tricks tutorial and added the visualization for all results for all spaces:
hdivhdg_tricks_debug.ipynb (12.8 KB)
I only get correct results for the FE spaces “n.+t. red.”, i.e. with highest_order_dc=True. All other pairs give incorrect results like above.
My NGSolve version is 6.2.2406.
Can you reproduce this?

[joachim]

the webpage
https://docu.ngsolve.org/ngs24/CFD/hdivhdg_tricks.html

shows that result:

and this I get also 8 times from you notebook, using the latest version ngsolve 2501. I am using MacOS, where the default inverse is “umfpack”.

Joachim