Fix elements orderings considering element volumes #40
Conversation
|
@alexbenedicto I don't understand the fix neither the initial program. If you follow the right-hand rule their is only one possible permutation for vertices: https://examples.vtk.org/site/Cxx/GeometricObjects/LinearCellsDemo/ I don't understand why this function takes as arguments another permutations. You have to check that every tuple of indices follows this arrangement. I am not sure someone wants to use another geometric convention. |
|
@untereiner This program is a tool that allows user to correct the nodes ordering of specific cells of a mesh. Therefore in the original version, if the user knows the correct convention for the tetrahedrons, he can change the nodes ordering of ALL tetrahedrons at once. The correction I added was to answer a demand from users where their grids still contained invalid volumes like in the case with the Tetrahedron grid. But this time, not all tetrahedrons had an invalid volume, only half of them for example (and yes I received a grid with this issue). So if you were to use the original code, you would reorder all of your tetrahedrons but still end up with half of them with an invalid volume (this time the other half). This new version allows for more control to only operate the nodes ordering on cells that have an invalid volume, or still on all cells. |
|
Ok but the ordering should be deduced for each tetrahedron depending on the sign of the volume (if you look inside or outside the volume). What I don’t understand is why the ordering as a parameter. |
|
@untereiner So for each cell found to be problematic in the mesh, I should have an algorithm that can find the right ordering ? |
|
If you have negative volume then take a face and invert two vertices to turn right hand. |
…e node ordering himself. Just needs to specify the name of elements to check.
…e node ordering himself. Just needs to specify the name of elements to check.
…ttps://github.com/GEOS-DEV/geosPythonPackages into origin/feature/fix_elements_orderings/useCellVolume
…erated cells cannot be treated because the problem lies more in the generation of the mesh itself than with a convention error.
Linked with issue #35
This is the second attempt at making a PR after sustaining merge errors in previous PR, now closed at: #39
Like stated previously, fix_elements_orderings feature would operate nodes ordering modification on all cells of a mesh that have a specific element type.
But if only half of the cells being of that specific cell type were invalid, all of them would still be reordered.
This PR aims at considering the volume of the cells to be taken into account when considering to reorder or not the nodes of a cell.