Update Laplacian jump indicator tests #10288
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
* Drop redundant periodic, FunctionNeumannBCs, and associated Functions. * We have to use FunctionPenaltyDirichletBC with Hermites because MOOSE does not correctly handle the setting of DirichletBCs for HERMITE elements. * Switch test to use new [Adaptivity] system. Note that this changes this output of the 'dofs' postprocessor since there are now elemental dofs for the marker variable in the system. This required regolding the tests, but I checked to make sure that the solutions were still the same. * Drop gold files that were not used for anything. * If you use periodic BCs, you should not apply any other kind of BC on that boundary. * Reduce the order of the custom quadrature rule used. This seemed to make no difference in the results and might speed up the test just a bit. Refs idaholab#2190.
These Kernels/BCs are needed for properly testing the LaplacianJumpIndicator, but they are probably general enough that they could be added to MOOSE proper, if desired. Refs idaholab#2190.
These new tests are based on the Biharmonic equation (and its transient variant) so the Laplacian jump indicator is appropriate for estimating the error. In the process of developing these tests, it was discovered that MOOSE is not currently imposing Dirichlet BCs for Hermite elements correctly. Specifically, MOOSE constrains only the *value* degrees of freedom when Dirichlet BCs are imposed, but setting BCs for Hermite elements involves the gradient degrees of freedom as well, according to @roystgnr. The current workaround is to use a penalty approach for enforcing the boundary conditions, but this comes with the obvious drawbacks of dramatically increasing the initial residual and changing the conditioning of the system. Refs idaholab#2190.
|
I also fixed the tests in |
permcody
requested changes
Nov 17, 2017
test/include/bcs/BiharmonicLapBC.h
Outdated
| { | ||
| public: | ||
| BiharmonicLapBC(const InputParameters & parameters); | ||
| virtual ~BiharmonicLapBC() {} |
There was a problem hiding this comment.
I usually don't bother putting these in on derived classes anymore.
test/include/bcs/BiharmonicLapBC.h
Outdated
| virtual ~BiharmonicLapBC() {} | ||
|
|
||
| protected: | ||
| virtual Real computeQpResidual(); |
|
Job Documentation on 07cdc2c wanted to post the following: View the site here This comment will be updated on new commits. |
Shhh don't tell Andrew... |
|
BTW, here's what the error indicator looks like as we refine the grid for the new problem (true solution is a Gaussian centered at the origin).
|
permcody
approved these changes
Nov 17, 2017
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
These tests were previously solving an incorrect weak form of the advection-diffusion equation that somehow "worked" although @roystgnr and myself remain stumped as to why. This PR adds test Kernels/BCs for solving the biharmonic equation (which requires C1 elements) and tests the Laplacian jump indicator with this PDE instead.
In the course of investigating this issue, it was discovered that the way MOOSE imposes Dirichlet boundary conditions is not correct for Hermite elements, since the constraints must involve not only the value dofs but the gradient dofs as well. The current workaround is to use penalty BCs for these tests, but there are known drawbacks to this. A longer term fix is to look into the
DirichletBoundaryobjects in libMesh to see if they can be used in MOOSE, or to employ a non-penalty approach like Nitsche's method for imposing BCs (I know this works for H1 problems and Lagrange elements, not sure about the Biharmonic problem.)Refs #2190.