Hello, I have been working with FETools::extrapolate, and I think there is a slight issue with its documentation (at https://www.dealii.org/current/doxygen/deal.II/namespaceFETools.html#ab2c5f72f3b99909fdf7d7bb215792400 ).
Specifically, the third bullet point appears to be false. It states that: “Since we traverse non-active cells from the coarsest to the finest levels, we may find patches that correspond to child cells of previously treated patches if the mesh had been refined adaptively (this cannot happen if the mesh has been refined globally because there the children of a patch are all active). We also perform the operation described above on these patches, but it is easy to see that on patches that are children of previously treated patches, the operation is now the identity operation (since it interpolates from the children of the current patch a function that had previously been interpolated to these children from an even coarser patch). Consequently, this does not alter the solution vector any more.” >From looking at the implementation in extrapolate_serial starting on line 1712 of fe/fe_tools_extrapolate.templates.h, it seems that because dof values are always read from the const input vector `u3` (line 1755) and written to the output vector `u2` (line 1756), any previously extrapolated values on grandchild cells will not be read when the child is revisited. I’m not entirely sure about the parallel case, but it looks as though the logic is similar. I don't think this is a problem with the algorithm (users may well want the extrapolation to respect their adapted mesh!), but it seems to be a disagreement with the documentation. I felt it important to flag it up as I know the importance of high quality documentation to the deal.II community. All the best, Oliver -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/6f376f1a-9319-46de-9d2f-1083e6aa0ca4o%40googlegroups.com.
