Closed jlippuner closed 3 years ago
Toby has been looking into this kind of functionality, and I'd like to offer it in the upcoming version 3. However, for neither there is a schedule. If you could work with the current API that would be best. If your coarsening indicators are conservative enough, then coarsening should not add significantly to the overall error.
Thanks for making this library available, it's been very useful for my work!
I've found that
p8est_coarsen_ext
often coarses families of quadrants that should not be coarsened and they get refined again when I callp8est_balance_ext
. Would it be possible to combine the refine, coarsen, and balance steps into a single function so that it won't coarsen and then immediately refine again but rather be aware of the balance constraint and not coarsen in the first place?