Various bugs related to BCs #1255
Conversation
|
Yes this is what we had talked about! I can see if I have a small mesh to test this on. |
| call this%gs_Xh%op(this%u,GS_OP_MIN) | ||
| call this%gs_Xh%op(this%v,GS_OP_MIN) | ||
| call this%gs_Xh%op(this%w,GS_OP_MIN) | ||
| call bc_list_apply_vector(this%bclst_vel,& | ||
| this%u%x, this%v%x, this%w%x, this%dm_Xh%size(), t, tstep) | ||
| call this%gs_Xh%op(this%u,GS_OP_MAX) | ||
| call this%gs_Xh%op(this%v,GS_OP_MAX) | ||
| call this%gs_Xh%op(this%w,GS_OP_MAX) |
There was a problem hiding this comment.
Would it be possible to only add a subset of the bclst? to reduce the work, and similarly for the gs, are these additional ops always needed, if not, can we possibly detect when they are, and have a if block around them?
There was a problem hiding this comment.
I think this is the only bulletproof way of doing this, but we can give it a thought I guess. For now I would keep it this way.
MartinKarp
changed the title
|
I have updated this PR. Please have a look when you have time! |
MartinKarp
changed the title
|
@njansson can we merge this? We should probably try to get it in as soon as possible, if device-aware mpi doesn't work maybe we can let that be an exception or something. |
I suggest we wait a bit, I'll have a go at the min/max kernels next week |
|
How is this going? |
Fixes the issue of applying bcs that are not necessarily on a facet. Enables the use of geometries where only the edge of an element is touching the boundary.
This is a major bugfix as currently we are constrained to rather well-posed geometries.
I also fixes Non-normal ('on') and Symmetry ('sym') bcs as they currently do not seem to work as intended as their init resets the marked facets.