1. 19 Mar, 2020 1 commit
  2. 16 Mar, 2020 1 commit
  3. 15 Mar, 2020 1 commit
  4. 04 Mar, 2020 1 commit
  5. 26 Feb, 2020 1 commit
  6. 13 Feb, 2020 5 commits
  7. 30 Jan, 2020 2 commits
  8. 17 Dec, 2018 2 commits
  9. 27 Sep, 2018 1 commit
  10. 24 Jun, 2018 1 commit
  11. 18 Jan, 2018 1 commit
    • Timo Koch's avatar
      [solver] Remove TypeTag dependency of linear solver · ac426f91
      Timo Koch authored
      This patch reduces dependencies and makes usage of linear solvers more flexible, i.e. for two problems with different typetag
      
      * Linear solvers don't need type tag anymore, in fact they don't need any template params
      * the preconditioner block size is now provided for the solve method (see e.g. staggered newton controller)
      * the amg has a convencience alias using a TypeTag but is "independent" through indirection via the traits
      ac426f91
  12. 21 Dec, 2017 1 commit
  13. 20 Dec, 2017 1 commit
  14. 19 Dec, 2017 1 commit
  15. 18 Dec, 2017 1 commit
  16. 16 Dec, 2017 1 commit
  17. 11 Dec, 2017 1 commit
  18. 08 Dec, 2017 1 commit
  19. 06 Dec, 2017 1 commit
    • Timo Koch's avatar
      [amg] Decouple AmgTraits from property system · a794364e
      Timo Koch authored
      The Amg properties where defined in an amg header for other discretization TypeTags.
      This leads to a lot of dependencies because files from all model types where included.
      Now the includes are reduced to property forward declarations and disc methods.
      a794364e
  20. 29 Nov, 2017 1 commit
  21. 25 Oct, 2017 1 commit
  22. 06 Apr, 2017 1 commit
  23. 28 Mar, 2017 2 commits
  24. 09 Feb, 2017 1 commit
  25. 23 Feb, 2016 1 commit
  26. 02 Feb, 2016 1 commit
  27. 22 Jan, 2016 3 commits
  28. 04 Nov, 2015 1 commit
  29. 03 Nov, 2015 1 commit
    • Bernd Flemisch's avatar
      [AMGBackend] fix AMGBackend for sequential grids and HAVE_MPI == 1 · 21c2b3d8
      Bernd Flemisch authored
      So far, the AMGBackend only worked for sequential grids if HAVE_MPI
      was 0. This was due to the fact that AmgTraits distinguished between
      parallel and sequential by evaluating HAVE_MPI. If HAVE_MPI was 1
      but the Grid was sequential, this lead to compiler errors for the
      constructor calls of the Communicator, LinearOperator and
      ScalarProduct.
      
      Do this properly now by using Dune::Capabilities for distinction
      between the sequential and parallel case. For Dune 2.4, the
      capability isParallel has to be used, while for Dune 3.0, it is
      canCommunicate. The correct AmgTraits and constructor calls are
      chosen by template specialization.
      
      This solves FS#300.
      21c2b3d8
  30. 14 Oct, 2015 1 commit
  31. 13 Oct, 2015 1 commit