Mohamad/author fix #7
Open
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
Communicate refinement flags in p:s:T
Use pre-increment operator.
Clarify documentation of p::s::T.
Ensure to look at square of tolerances
Fix a reference.
Add some assertions about hypercube reference cells.
Reduce the volume of communication somewhat.
…ia_fix p:d:GridRefinement: actually accept other types of triangulations
Expand a discussion in step-58.
Fix type in FPE
Non-affine piola transformation for mf RaviartThomas
Specifically, we don't need it before 'std::enable_if_t' because we are not referencing a nested type.
While there, also use it in fe_system.h.
…ation DoFHandlerPolicy: Avoid one write access to an array
Enable SolverIDR for block vectors
…etricSum_precompile Precompile TensorProductMatrixSymmetricSum kernels
Add MatrixFreeTools::ElementBirthAndDeathMatrixFree
Fuse some loops in Triangulation::execute_coarsening
…tion MatrixFree: Fix assertion in presence of FE_Nothing
FEEvaluation: Create temporary copy of Jacobian matrix in submit_XXX
…precond_specialization Add specializations of PreconditionRelaxation functions
Fix a warning
* Ensure to look at square of tolerances * copy boundary ids to remaining boundary faces Co-authored-by: Peter Munch <[email protected]> * parallel intermediate out: compression levels - support different compression levels - add a test * Fix reference entry * Provide implementations of Utilities::MPI::isend/irecv(). * Add a test. * Add a changelog. * Simplify a function by using range-based for loops. * Fix label fo bibtex entry * compute_affine_transformation in codim1 and codim2 case * Communicate refinement flags in p:s:T * deprecate VtkFalgs::ZlibCompressionLevel * enum class * Fix a reference. * Clarify documentation of p::s::T. * Use only enough bits in an enum as necessary. * Use pre-increment operator. * Add FieldTransfer for the transfer of field with FE_Nothing * Add some assertions about hypercube reference cells. * Reduce the volume of communication somewhat. * p:d:GridRefinement: actually accept other types of triangulations * Fix type in FPE * Expand a discussion in step-58. * Add some functions to convert active cell iterators * avoid deprecated names * Added quicktests to Intel workflow. * Do not use 'typename' where not necessary. Specifically, we don't need it before 'std::enable_if_t' because we are not referencing a nested type. * Replace std::enable_if by std::enable_if_t throughout. * Declare 'enable_if_all_t' in analogy to std::enable_if_t. While there, also use it in fe_system.h. * Alternative implementation to convert iterators * Show a formula. * Update tutorials to use cell->as_dof_handler_iterator() * FEEvaluation: Fix two places where the address sanitizer complained * MatrixFree: Use separate data arrays for two separate evaluators * Add test for FEEvaluationShift with FE_Q and FE_DGQ * matrix-free FEEvaluation: initialize with nan Initialize the internal scratch_data of FEEvaluation with signaling_nans in debug mode. * qualify namspace for MSVC otherwise MSVC tries to use numbers::internal from signaling_nan.h * Make all template specialization of FPE compile for float * epic doxygen fixes It turns out that grouping using @{ and @} only works if it is inside a doxygen comment like /** */ and not inside a normal comment /* */. This meant that most of our groupings were incorrect. - fix @{ and @} - avoid using /*! which is equivalent to /** - make some formatting prettier * Implement get_affine_constraints() * more fixes * CMake: remove find_package override * CMake: use proper FIND_PACKAGE call in python bindings setup * Failing test. * Added a fix. * Fixed clang-tidy. * normal vector doc * Clean cgal utilities * Require PETSc 3.7.0 or newer. This is six years old at this point. * Remove some old PETSc workarounds. * Fix a typo in deal.II/base/numbers.h The refactoring done in b1883d9 introduced a typo that wasn't spotted by our CI. Fix this. * Fix generation of the tutorial graph. A new code gallery program has quotes which we need to escape in its tooltip. * Fix various typos Fix typos found in source/, include/, and doc/ Found via `codespell -q 3 -S ./bundled,./doc/news -L ba,commend,nd,pres,ro,rotat,stoer,workd` * Revert fix in previous commit per request * Clean up of VectorTools::project functions * Fix typos in tests/ subdirectory * Fixed indentation * Implement TensorProductMatrixSymmetricSumBase::memory_consumption() * Add some missing headers. Other changes in this PR necessitate these additions. * Add FEFaceEvaluation::at_boundary() and boundary_id() * Don't include Sacado.hpp in base/numbers.h. Sacado.hpp, when preprocessed, is over 200,000 lines. We can get things working in an equivalent way by using argument dependent lookup instead. * Get things working with adol-c too * Add second version to initialize MFF::ConstraintInfo * Remove some template specializations for QAnisotropic. Clang now complains about adding undeclared specializations in the source file - besides, we can combine them pretty easily. * Move Threads::Mutex to a dedicated header. Most files don't need anything in Threads other than Threads::Mutex so we can decouple them from TBB et. al. by moving this class to its own header. * Fix some theory in step-44 introduction The deviator tensor involves the 4th-order symmetric identity tensor, not the "normal" 4th-order identity tensor. * Consistently update notation in SymmetricTensor docs * Fix typo in step-52. * Move 'using MPI_Comm = int' et al to a separate header. Most of the time we only need declarations of MPI classes (like MPI_Comm) and not our own utility functions in headers. To this end, we can just put mpi.h (or our equivalent stub declarations) in a separate header and use that most of the time. * Pass ReferenceCell by reference in QProjector. This lets us use a forward declaration. * Clean up some header inclusions. * Remove now redundant namespace identifier. * Added missing header. * FEEvaluationData: add virtual destructor * fix missing header * Introduce inverse of MF::get_cell_iterator() * Assert for released handle on Transfer classes. * no normal flux on levels * combine into one function * add more tests * stokes test * fix a bug * address comments * Fix header in Zoltan tests. * Remove an unused forward declaration. * Move Trilinos utilities into a separate header. * Fix more headers in Zoltan tests. * Fix MatrixFree + FE_Q_iso_Q1(2) * Add second version of TensorProductMatrixSymmetricSum::apply_inverse() * Use ReferenceCell instead of StaticMappingQ1. * Add a missing header to vector_relations.h. * Clean up inclusions of mapping_q1.h * Add some more missing Trilinos headers. * Fix Kokkos flags * CMake: Avoid --date=iso-strict for git versions older than 2.2 * Change get_active_fe_indices() to return the result. * Added get/set functions for future FE indices. * Move active_fe_index_type to types. * address comments * Fix Tensor assignment of nested Tensors and provide a function overload of apply_transformation to get rid of casts in FEPointEvaluation * Clarify comments about get_generalized_support_points(). * Move invalid_fe_index to numbers. * Add tests for FieldTransfer * address comments * Fix find_package for boost python * Generalized VT::point_gradients() for multiple components Co-authored-by: Peter Munch <[email protected]> Co-authored-by: Martin Kronbichler <[email protected]> rename file * Optimize PreconditionRelaxation * Avoid warning about unused variable * MatrixFree: Fix bug in setup of new cell_index for MG * Fix test case with clang compiler * Update output * Fix bug in matrix-free operators regarding diagonal * Set another OMP environment variable. Both of these environment variables control the number of threads - if we set one we should set the other. In my shell I always set OMP_THREAD_LIMIT=1 since I don't want to use OpenMP. For whatever reason stopped working recently, resulting in OpenBLAS trying to yield execution to nonexistent other threads since some functions expect one thread to exist and others expect two. * Move up the check with ExcGridHasInvalidCell. We can do better than an internal error in the GridTools function. * Add some more output to the ExodusII test. * Add pyramids and wedges to invert_cells_with_negative_measure(). * Refactor the way we apply grid fixup functions. This exposes a few issues with the ExodusII test grids. * Merge two loops in setup of triangulation * Add specialization of PreconditionChebyshev functions * Fuse some loops in Triangulation::execute_coarsening * Fix PETSc version detection. New versions of PETSc now define PETSC_VERSION_LT etc. in one line, so our detection of PETSC_VERSION_MAJOR will now also pick up that usage. Get around this by making the regex more correct. * Clean up side effect * Update checks for AD-types. At some point these stopped working and something like #include <deal.II/lac/vector.h> #include <deal.II/lac/vector.templates.h> #include <adolc/adouble.h> int main() { using namespace dealii; Vector<adouble> vec; } managed to compile (but compiling Vector<adouble>::l2_norm() fails). In practice we don't support any type here which we cannot call std::abs() on, so lets just switch to a simpler check for these types. * Handle a test which fails under GCC 5.4 * Restructure computation of diagonal * Add test case * Extend comment * Remove an outdated comment. Link to step-19. * Clarify a comment for one of the particle generators. * Doc update in GridTools. * fix: step-50 PETSc add a missing include * Add MatrixFreeTools::ElementActivationAndDeactivationMatrixFree * Enable SolverIDR for block vectors * Precompile TensorProductMatrixSymmetricSum kernels * Add specializations of PreconditionRelaxation functions * MatrixFree: Fix assertion in presence of FE_Nothing * FEEvaluation: Create temporary copy of Jacobian matrix in submit_gradient * Fix a warning Co-authored-by: Wolfgang Bangerth <[email protected]> Co-authored-by: Martin Kronbichler <[email protected]> Co-authored-by: Johannes Heinz <[email protected]> Co-authored-by: Peter Munch <[email protected]> Co-authored-by: Timo Heister <[email protected]> Co-authored-by: Fabian Castelli <[email protected]> Co-authored-by: Marco Feder <[email protected]> Co-authored-by: David Wells <[email protected]> Co-authored-by: Daniel Arndt <[email protected]> Co-authored-by: Bruno Turcksin <[email protected]> Co-authored-by: Jean-Paul Pelteret <[email protected]> Co-authored-by: Matthias Maier <[email protected]> Co-authored-by: Marc Fehling <[email protected]> Co-authored-by: Maximilian Bergbauer <[email protected]> Co-authored-by: Luca Heltai <[email protected]> Co-authored-by: Jiaqi Zhang <[email protected]> Co-authored-by: luz paz <[email protected]> Co-authored-by: Magdalena Schreter <[email protected]> Co-authored-by: Maximilian Bergbauer <[email protected]>
PreconditionChebyshev: reduce number of writes by one
Specialize EvaluatorTensorProduct<evaluate_evenodd> for non-templated execution
DiagonalMatrix: specialize for LA:d:V
PreconditionRelaxation: add specialization
Fix warning regarding unused variables
Fix another warning about an unused variable
Function to restrict matrices
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.
Author name is fixed on all past commits.