deal.II Newsletter #359
Rene Gassmoeller
Hello everyone!
This is deal.II newsletter #359.
It automatically reports recently merged features and discussions about the deal.II finite element library.
Below you find a list of recently proposed or merged features:
#19902: Fix GridIn::read() for abaqus input files (proposed by mschreter)
#19901: Ensure internal::expand is instantiated (proposed by masterleinad)
#19900: Improve error messages in FiniteElement::get_sub_fe(). (proposed by bangerth)
#19899: Better document FiniteElement::get_sub_fe(). (proposed by bangerth)
#19898: Fix initialization of hessians_quad_submitted in FEEvaluationData (proposed by mschreter; merged)
#19897: Revert "Use set of pairs for hp_quad_dof_identities" (proposed by bangerth; merged)
#19896: Fix 1D path of interpolate_boundary_values() for hp::DoFHandler (proposed by mschreter; merged)
#19895: Additive Schwarz from PETSc (proposed by fdrmrc)
#19894: Deprecate other now-unused parts of MeshWorker. (proposed by bangerth)
#19893: Deprecate those classes that were only used in step-39. (proposed by bangerth)
#19892: Better document DoFTools::extract_dofs_with_support_contained_within(). (proposed by bangerth)
#19891: Support for agglomeration-based DG discretizations (proposed by fdrmrc)
#19890: Document 'quad' in FiniteElement::hp_quad_dof_identities(). (proposed by bangerth; merged)
#19889: Better document the Gauss-Jacobi polynomials. (proposed by bangerth; merged)
#19888: Arkode: fix callback documentation for ARKode and steppers (proposed by vovannikov; merged)
#19887: Update Trilinos build instructions (proposed by masterleinad; merged)
#19886: MatrixFree: if there is only one partitioner, then return it. (proposed by drwells; merged)
#19885: MatrixFree: make two scratch vectors small vectors. (proposed by drwells; merged)
#19884: enable affine constraints to use PETScWrappers::MPI full matrices (proposed by SanojKnalp; merged)
#19883: Add an assertion. (proposed by bangerth)
#19882: Only work-around HDF5 bug for sufficiently old versions. (proposed by bangerth)
#19881: Add `adjust_quad_dof_index_for_face_orientation_table` for `FE_SimplexP` (proposed by dominiktassilostill)
#19880: Fix vectorized_load_and_transpose for AVX/AVX-512 (proposed by kronbichler; merged)
#19879: TriangulationDescription::CellData: use an inplace_vector. (proposed by drwells; merged)
#19878: MappingFE::compute_mapping_support_points(): re-use an array. (proposed by drwells)
#19877: Arkode: improve robustness of tests arkode_01 and arkode_02 (proposed by vovannikov; merged)
#19876: Be more consistent about the use of types::fe_index. (proposed by bangerth)
#19875: KellyErrorEstimator: only compare unsigned ints with unsigned ints. (proposed by drwells; merged)
#19874: News changes minor entry for PR #18659. (proposed by gsegon; merged)
#19873: Add log entry for diagonal operator. (proposed by quangx; merged)
#19871: PreconditionChebyshev: Add ability to compute residual norm (proposed by kronbichler)
#19870: Fix calling TpetrWrappers::vmult with dealii::Vector arguments (proposed by masterleinad; merged)
#19869: Utilities::MPI: don't compress data before sending it. (proposed by drwells; merged)
#19868: Try setting CMAKE_MSVC_DEBUG_INFORMATION_FORMAT=ProgramDatabase to minimize library size with MSVC (proposed by masterleinad)
#19867: Make Tpetra vector element operations thread-safe (proposed by gassmoeller; merged)
#19866: [WIP] Remove IC preconditioner from tests if unavailable (proposed by gassmoeller)
#19865: Arkode: tighten tolerances for tests 07, 08 and erkstep (proposed by vovannikov; merged)
#19864: Simplify TimeStepping with std::unique_ptr and initializer lists (proposed by JR-Khan; merged)
#19863: Remove dependence of step-39 on old MeshWorker classes. (proposed by bangerth)
#19862: Arkode: rethrow recoverable error after SundialsOperator::vmult() and SundialsPreconditioner::vmult() (proposed by vovannikov; merged)
#19861: Rename a few variables into n_lanes. Remove unneeded ones (proposed by kronbichler; merged)
#19860: KellyErrorEstimator: avoid allocating a scratch vector. (proposed by drwells; merged)
#19859: make_hanging_node_constraints(): avoid setting up a std::set. (proposed by drwells; merged)
#19858: Replace PreconditionIC in step-43 if unavailable (proposed by gassmoeller)
#19857: Replace PreconditionerIC in step 31 for Tpetra (proposed by gassmoeller; merged)
#19856: Guard AMG parameter only available for Epetra ML (proposed by gassmoeller; merged)
#19855: Arkode: introduce the default eigenvalue estimator for LSRKStepperSTS (proposed by vovannikov; merged)
#19854: Add missing empty lines in header (proposed by gassmoeller; merged)
#19853: Add missing compress() calls in example programs (proposed by gassmoeller; merged)
#19852: Provide MueLu if ML is not available (proposed by gassmoeller; merged)
#19851: Simplify MeshWorkers::Assemblers::CellsAndFaces. (proposed by bangerth)
#19850: Deprecate MeshWorker::Functional. (proposed by bangerth; merged)
#19849: Link namespace TimeSteppers to other resources. (proposed by bangerth; merged)
#19848: Deprecate the classes in namespace Algorithms. (proposed by bangerth; merged)
#19847: Remove incompatibilities between mpc and mpich (proposed by luca-heltai; merged)
#19846: SolutionTransfer: add a scratch array for packing data. (proposed by drwells; merged)
#19844: Only keep Tpetra vector view as long as necessary (proposed by gassmoeller; merged)
#19843: fix latex formula issues (proposed by tjhei; merged)
#19839: Avoid some static variables. (proposed by drwells; merged)
#19837: Specialize vectorized_load_and_transpose for remainder cases 1,2,3 (proposed by kronbichler; merged)
#19836: Arkode: Improve robustness of ARKode tests 06-10 and compat_interfaces (proposed by vovannikov; merged)
#19827: Evaluation kernels hanging nodes docu (proposed by peterrum; merged)
#19826: Use a single implementation for vectorized transpose functions (proposed by kronbichler; merged)
#19784: Base & Tests: Implement and verify fixed-step and embedded Tsitouras 5(4) (proposed by JR-Khan; merged)
#19706: Add a diagonal operator. (proposed by quangx; merged)
#19662: Nodal basis for simplices (proposed by dominiktassilostill; merged)
#19291: Keep order of fe indices when determining hp dof identities (proposed by dominiktassilostill; merged)
#19016: write CF/UGRID output (proposed by dabele; merged)
#17578: Arbitrary order quadrature on pyramids (proposed by dominiktassilostill; merged)
And this is a list of recently opened or closed discussions:
#19872 get_sub_fe() throws with FESystems whose sub-FE were not explicitly created as FESystem (opened and closed)
#19842 Deprecate everything in namespace Algorithms. (closed)
#19818 Improve code structure of time integrators: Use array assignment. (closed)
#19817 Improve code structure of time integrators: Use std::unique_ptr over plain pointer. (closed)
#19814 SUNDIALS test failures (closed)
#19799 Compilation issues with locally built deal.II and PETSc (closed)
#19781 matrix_free/evaluation_kernels_hanging_nodes.h: Improve documentation (closed)
#19700 Regression tester regressed 3770e0 (closed)
#19650 General status of simplex support (closed)
#19610 sundials/invocation_context.h is missing an include. (closed)
#18903 Netcdf/CF/Ugrid output (closed)
#18834 Change auto-generated labels in tutorials. (closed)
#18676 xref validation (closed)
#17836 tutorial step-92: "parallel handling of complicated meshes" (closed)
#17719 Make DoFRenumbering::component_wise() work with FEValuesExtractors. (closed)
#15624 LinearAlgebra::{E,T}petraWrappers::Vector::compress() doesn't do anything. (closed)
#15519 Testsuite: improve test timings (closed)
#15476 Testsuite: don't print exc.what() in tests (closed)
#15395 Testsuite: tests should not output and compare initial and final residuals (closed)
#13949 Optimize compilation units and object libraries: investigate sizes, minimize includes, minimize compilation times, enforce DAG (closed)
#6045 Link the TimeStepping classes to the tutorials. (closed)
A list of all major changes since the last release can be found at this website.
Thanks for being part of the community!
Let us know about questions, problems, bugs or just share your experience by writing to this mailing list, or by opening issues or pull requests on Github.
Additional information can be found at our official website.