From both the documentation and my tests, it appears that neknek assumes field symmetry across sessions. In particular, scalar arrays are expected to exist (and potentially be exchanged) in both sessions.
Am I correct in understanding that there is no straightforward way to:
More specifically, has anyone successfully implemented a setup where:
As an alternative, I also explored a “nested mesh” approach (slave mesh as a strict subset, with scalar-only solve using interpolated velocity). However, this led to robustness issues, particularly related to gs_setup, which is why I am revisiting neknek.
Any insights, or confirmation that this is fundamentally unsupported, would be very helpful.
Chinthaka.