1. 29 Dec, 2020 2 commits
  2. 14 Oct, 2020 1 commit
  3. 09 Feb, 2020 1 commit
  4. 20 Jan, 2020 1 commit
  5. 03 Jan, 2020 3 commits
  6. 27 Sep, 2019 1 commit
    • Praetorius, Simon's avatar
      add PETSc find module and configuration to cmake · 3a2c0797
      Praetorius, Simon authored
      changed petsc cmake configuration
      
      added computation of global DOF ids for pre-bases
      
      Global basis ids for dune 2.6 and dune 2.7
      
      corrected the unbind function
      
      wrong unbind function call corrected
      
      Added partition type query to GlobalIdSet
      
      added documentation and extended the test
      
      add general implementation and specialization for TaylorHoodBasis
      
      corrected treepath of velocity component in taylorhoodbasis
      
      add PETSc find module and configuration to cmake
      
      add DofMapping for the association of local and global indices
      
      changed petsc cmake configuration
      
      added DOFMapping and test, to be used with petsc or pmtl backend
      
      extension of istl preconditioner and amg precon
      
      added block jacobi preconditioner with sequential sub precon
      
      update ISTLCommTest for new Communication Creators
      
      removed some smoothers due to bugs in dune-istl
      
      added krylov AMG preconditioner
      
      restructuring of istl solvers and preconditioners
      
      Corrected errors in uniqueBorderPartitionTest and ProblemStat
      
      changed default solver and removed solver category argument from precon creator
      
      restructured stolver and precon creators in istl. Added precompiled explicit template instantiations.
      
      cholmod for dune-2.7 only
      
      removed explicit instantiation of AMG precons due to dune-istl bugs
      
      Cleanup of AMG and more documentation, removed some precompiled preconditioner and solvers
      
      redesign of linear algebra backend for inclusion of PETSc
      
      petsc example added for study of the implementation
      
      Matrix pattern communicates non-owning rows
      
      petsc example extended for debugging
      
      output of vectors does not use std::vector directly anymore
      
      dirichletBC redesigned
      
      restructuring of vectors and matrices, added BiLinearForm and LinearForm
      
      cleanup of VectorBackend design. Needs to be applied to all backends!
      
      adopt istl backend to changes in VectorBase and MatrixBase
      
      Update CMakeLists and implemented twist of globalBasisIds
      
      Communication in MatrixNnzStructure and DofMapping corrected.
      
      not HAVE_MPI caseimplemented
      
      do not use lv.maxSize() for mat and vec sizes
      
      meshcreator with overlap and periodicity for YaspGrid
      
      Twist indexing corrected and new UniqueBorderPartition strategy implemented, In MeshCreated the MAcroGridFactory removed, since it does not work in parallel
      
      PETScCommTest added
      
      MatrixNNZStructure extended to include remote dnnz numbers
      
      made all backends conforming to the new structure
      
      MeshCreator now reads AlbertaGrid only on rank 0
      
      make vector copyable
      
      export BACKEND cmake flag to amdis-config file
      
      cmake-pkg config file added
      
      pkg config file updated and PETSc direct solver added
      
      update MTL and PETSc backend after complex test
      
      make petsc constraints work on DirichletNodes that are not on boundary intersections
      
      moved AMDIS_INFO_LEVEL to Output
      
      updated the backup-restore test and corrected some small errors
      
      make friend use struct instead of class for Constraints
      
      change solver parameter in generated initfile of amdisproject and add PETSc to README
      
      update choldmod to recent dune version
      3a2c0797