Adding support for pml in domain for multiple patches #4632
Conversation
RevathiJambunathan
added
component: mesh refinement
| } | ||
| if (do_pml_Hi[idim]){ | ||
| domain0.growHi(idim, -ncell); | ||
| BoxList bl = grid_ba.boxList(); |
There was a problem hiding this comment.
We could also use simplified_list() instead of boxList here. That way, it's less likely to hit the assert failure below.
There was a problem hiding this comment.
Thank you Weiqun for the suggestions!
Done
| // ncell is divided by refinement ratio to ensure that the | ||
| // physical width of the PML region is equal in fine and coarse patch | ||
| cdomain.growHi(idim, -ncell/ref_ratio[idim]); | ||
| BoxList bl = grid_cba.boxList(); |
There was a problem hiding this comment.
Here we can use simplified_list as well.
There was a problem hiding this comment.
discussed offline, switched back to boxList due to failing CI
Source/BoundaryConditions/PML.cpp
Outdated
| @@ -565,35 +565,43 @@ PML::PML (const int lev, const BoxArray& grid_ba, const DistributionMapping& gri | |||
| // In order to implement this, a reduced domain is created here (decreased by ncells in all direction) | |||
There was a problem hiding this comment.
Maybe the term "reduced domain" is not appropriate anymore in this comment, since it is not a single contiguous domain anymore.
| for (auto& b : bl) { | ||
| for (int idim = 0; idim < AMREX_SPACEDIM; ++idim) { | ||
| if (do_pml_Lo[idim]) { | ||
| Box const& bb = amrex::adjCellLo(b, idim); |
There was a problem hiding this comment.
Could you add a comment to explains what this code (the call to adjCellLo and the subsequent intersects) does?
There was a problem hiding this comment.
added comments to clarify
| if (do_pml_Hi[idim]){ | ||
| domain0.growHi(idim, -ncell); | ||
| BoxList bl = grid_ba.boxList(); | ||
| for (auto& b : bl) { |
There was a problem hiding this comment.
Could you add a comment to explain what this for loop does?
(In my understanding, it loops over the box at this level, and whenever a box touches a PML, it reduces the box by ncell, since this is where the surrounding "in-domain" PML will be created.)
There was a problem hiding this comment.
added comments to clarify
| @@ -760,24 +768,28 @@ PML::PML (const int lev, const BoxArray& grid_ba, const DistributionMapping& gri | |||
| BoxArray grid_cba = grid_ba; | |||
| grid_cba.coarsen(ref_ratio); | |||
|
|
|||
| // assuming that the bounding box around grid_cba is a single patch, and not disjoint patches, similar to fine patch. | |||
| amrex::Box cdomain = grid_cba.minimalBox(); | |||
| BoxArray grid_cba_reduced = grid_cba; | |||
There was a problem hiding this comment.
The code below duplicates the above code.
Maybe it would be good, in a separate follow-up PR, to create a separate function for this.
There was a problem hiding this comment.
yes. I ll do this in a separate PR
for more information, see https://pre-commit.ci
|
Added a CI test by modifying existing 2D mr particles_in_pml test to include two patches |
Thanks to @WeiqunZhang for the discussion and prototype!!!
The current implementation for
pml_in_domainonly works for a single patchThis PR adds capability to support 'pml_in_domain' for multiple distinct mesh-refinement patches
without the support for this feature : (PML in domain only on the left most edge of the left patch and right edge of the right patch) - i.e., the interiors edges dont apply PML correctly
With this PR :
