[bug correction] trirefine is now independant of triangulation numbering #2363
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Correction in cubic interpolation function:
the interpolated values on refined triangular grids should be the same regardless of the triangles ordering.
As of today, this was not the case for the cubic interpolation, as discovered by @ianthomas23 (refer to the new test enclosed in this PR, which was failing and should now pass). The differences were small but I consider this was a bug (at least highly unexpected).
The solution is a new choice for the Gauss quadrature points.
I added a new (non-image) test and updated the 2 baseline images involving this kind of cubic interpolation (one was failing as a result of this PR ; I also updated the 2nd to get a fresh start).
Note:
As changing the Gauss points was not my first guess (unfortunately...) I also made a double-check of the various shape functions used in
_ReducedHCT_Element
For this I found that it was more convenient to arrange them by columns, even if it this is not strictly pep8. If this is considered bad practise, please tell me ; I would be happy to revert.