Symmetrize #99
Symmetrize #99
Conversation
Add routine to see if a curve is a straight line by checking the second derivative at a point. Add tests for that, too. See comments for limitations.
Start adding code to mirror across a symmetry line.
Add code to compute the reflection of a point about a line, plus tests. Put in stubs for mirroring a mesh.
Add changes to flip a mesh about a symmetry line and add the new elements to the mesh. Only does one symmetry boundary in the model ATM. Haven't checked the mesh file yet, and edges need to be added still to handle ISM-V2.
Add reflection of the edges in the mesh by removing the old ones and re-computing them. Also reflect the xPatch array, which is used in the SEM plotting procedure. Finally update the makefile to include the new reflection code.
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #99 +/- ##
==========================================
+ Coverage 74.19% 74.88% +0.69%
==========================================
Files 67 68 +1
Lines 10155 10429 +274
Branches 2 2
==========================================
+ Hits 7534 7810 +276
+ Misses 2621 2619 -2
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
Fix ifort error regarding class/type of a variable. Fix typos.
Add routines to test if all curves marked symmetry are co-linear straight lines. If not, posts a warning and the mesh is not reflected.
Fixes a longstanding error in assigning the boundary curve to an element boundary edge. Instead of using ROW_END as a discriminant to choose which curve the two nodes fall on (a corner node can be on two curves simultaneously), it checks to see if the node is at the endpoints of its curve and if so, uses the curve attached to the other node.
Make boundary curves int the boundaryInfo consistent with the defined directions.
Added PacMan example doing symmetry reflection to test suite. Change test on SYMMETRY_CURVE_NAME to be case insensitive.
Update the documentation to include the PacMan example.
Update to the previous to include one more file.
Append "_R" to boundary names that are on the reflected portion of the model domain, to allow for different BCs if desired.
|
I've updated everything I can think of. One thing especially open for discussion is that I append "R" to a boundary name on the reflected portion of the mesh. I guess if one does not want that, it's a quick search/replace (maybe) in the file. Maybe something else truly unique would be better ? "#R? |
There was a problem hiding this comment.
Looks good overall; just a small suggestion with the literals.
One follow-up question. If one sets multiple symmetry boundary it only reflects the "base" mesh, right? It would not be the case that the base mesh gets reflected over one boundary and the entire mesh is then reflected over the second boundary?
Add literal type where it was missing. Co-authored-by: Andrew Winters <[email protected]>
|
No, there's nothing that you have to do. Just call at least one of the boundary curves "symmetry" when you are generating a model and off it will go. (Hey, this might be a way to define an interior boundary! If one of the curves along a line is called symmetry and the others aren't, then the nodes will be duplicated, along those others and the curve names are not deleted. Hmm.) This is what one user was asking for at one point. This does work, but messages saying that there are duplicate nodes are posted. That's OK, because one expects that to happen. Huh. |
|
Add capability to take a mesh and reflect it about a (straight line) boundary with the name "symmetry". This should answer the request of some users for a symmetric mesh. TODO: Add example, testing, documentation, and test the mesh files.