1. 03 Feb, 2020 1 commit
  2. 20 Nov, 2019 1 commit
  3. 30 Jul, 2019 1 commit
    • Benoit LE's avatar
      Added a toExport function to PostProcessing · b769d03c
      Benoit LE authored
      May be useful if exporting a certain quantity requires some computations, but
      we want to do these computations only in the case where this quantity is
      supposed to be exported.
  4. 14 Jun, 2019 1 commit
  5. 25 Feb, 2019 1 commit
  6. 22 Jul, 2016 1 commit
    • Kévin Moreau's avatar
      This commit bring the following changes · 0d20941f
      Kévin Moreau authored
      * clean up unused feature.
      it affects:
        - Export.cc
        - Import.cc
      * improvement of transfer of displacement field after remeshing
      it affects:
        - Formulation.cc
      * correct a small bug that happens sometimes in restart, "mean" vectors
      had wrong size after restart. It was always too long and filled with NaN.
      it affects:
        - FormulationQS.cc
      * had several "guards" that end computation and give an error message when
      something goes wrong.
      it affects:
        - TLSSolver.cc
        - FormulationQS.cc
      * correct an error in the getGrad function of CohesiveLinearHardeningFunction
      it affects:
        - MaterialFunction.cc
      * swap orientation in comp mesh generation. It seems that it is test case dependent.
      TO BE INVESTIGATED if your comp_mesh has two colours when plotting in GMSH.
      it affects:
        - MeshGeneration.h
      * correct a bug resulting from \Gamma_in recent commit. FastMarchingModeExtension
      was updating phi based on \Gamma_in. The obtained phi was not anymore corresponding
      to the iso-lc discretisation, it leads to damage equal to 1 inside the mechanical domain.
      It leads to singular stiffness matrix.
      it affects:
        - TLSSolver.cc
  7. 20 Jun, 2016 1 commit
  8. 30 May, 2016 1 commit
  9. 03 May, 2016 1 commit
  10. 12 Feb, 2016 1 commit
    • Kevin Moreau's avatar
      New stable release. · 7e38f088
      Kevin Moreau authored
      Major changes:
        - integ_mesh is computed only on fully damaged support,
        - comp_mesh adds new vertices and edges inside the fully damaged zone and then erase useless entities,
        - the code is almost 3D,
        - algorithm part of the code moved to TLSAlgorithm, in order to share it with DamageBand, DamageBandDyn...
      Minor changes:
        - small improvement of Export,
        - get rid of every xIter except in TLSGeom so that if someone wants
        to use other kind of iterators (that deref in mEntity*),
        - dispatch code in several different files.
        - remove the TODO spread in the code (boring),
        - finish the cut of tetrahedra in Cutter,
        - clean deletion of mEntity* after adpating comp_mesh (I forgot to do this => memory leaks),
        - get rid of mEntity* (is it even possible?).
      git-svn-id: https://svn.ec-nantes.fr/eXlibris/Applis/TLSDuctile@2261 fbbead7c-fb4d-4173-aa67-51132c73c120
  11. 26 Jan, 2016 1 commit
    • Kevin Moreau's avatar
      Among other things: · 319163af
      Kevin Moreau authored
          1. pass superlu to xGenericSparseMatrix,
          2. improve integration rule smart, go back to integ rule partition,
          3. fixed a non initialized variable (nb_group)
          4. transfer export specific to FormulationQS to it
          5. add a tls submesh on which we restrain phi field. With this improvement we can decrease computational costs and treat inclusions (still experimental this zone is currently all the mesh).
          6. add export dedicated function in TLSSolver.
          7. add several keywords for superposition feature.
          8. restrain save of field to a specific region
          9. possibility to restrain integration mesh generation to a specific region only (experimental, to be continued)
          10. add fit to vertices or repulse from vertices feature to improve robustness, add export to debug mesh generation
          11. add a fit_to_vertices keyword (actually it is right now an repulse_from_vertices feature, to be continued...)
          12. handle anticipated exit properly, use characteristic length of the body as a maximum displacement limit.
      git-svn-id: https://svn.ec-nantes.fr/eXlibris/Applis/TLSDuctile@2253 fbbead7c-fb4d-4173-aa67-51132c73c120
  12. 05 Jan, 2016 1 commit
  13. 06 Nov, 2015 1 commit
  14. 30 Oct, 2015 1 commit
  15. 29 Oct, 2015 1 commit
    • Kevin Moreau's avatar
      This commit bring many modifications into the code. · a1bf0674
      Kevin Moreau authored
      - classical enrichment using xSpaceXFEM is removed and replaced by a mesh modification with superimposed elements,
      - cutting procedure using xPhysSurfByTagging is removed and is now closely related to mesh modification using an integration mesh,
      => see MeshGeneration.{h,cc}
      - xSubMesh are limited to TLSGeom.cc so that it is orthogonal to the rest of the application and so that anyone annoyed by such
       an implementation can change it,
      => see TLSGeom.cc
      - work of Benoit Le on solver interface is added (it has been refactored to treat plasticity case),
      => see LinearSystem.{h,cc}
      - the whole application is simplified and several member function names changed, mostly inside TLSSolver and Formulation (QS and Dyn).
      => see TLSSolver.{h,cc} and Formulation.{h,cc}
      - numerical integration changed and is now based on an specific integration with fixed integration point positions and evolving weights
      to integrate a polynom exactly on a part of an element.
      => see MeshGeneration.{h,cc}
      - Space.{h,cc} is removed since implementation was slow and replaced by linear algebra using a sparse matrix.
      => see TLSSolver.cc
      git-svn-id: https://svn.ec-nantes.fr/eXlibris/Applis/TLSDuctile@2178 fbbead7c-fb4d-4173-aa67-51132c73c120
  16. 03 Sep, 2015 1 commit
  17. 08 Jul, 2015 1 commit
  18. 30 Jun, 2015 1 commit
    • Benoît LÉ's avatar
      Added the possibility to save state during computation, and restart calculations · d5bfe2ed
      Benoît LÉ authored
      from this state. 
      To that end, a Preprocessing class was added to Import.*.
      To save state, the user just has to add the keyword "save" in the info.dat file,
      followed by the frequency of save. An archive (containing two binary files
      containing displacement and level-set field) is generated at each required step
      To do a restart, the user must instantiate and object of the PreProcessing
      class, and then call the restartFromFile function, AFTER the constructor of
      Algorithm. The name of the restart file can be parsed using ParseRestartFileName
      (it still has to be improved).
      Then when excecuting the program, the user needs to give the main data file, and 
      the name of the restart file, for exemple :
      >> ./executable data/main.dat save_step_0010.tar
      Normally it should work.
      In Export.*, a new exportOnSpace function has been added, to save field when
      needed (it can save both scalar field like level-set of vector field like
      The level-set field is saved in TLSSolver, after reinitiating the level-set
      In Formulation.* a saveStep function has been added to save displacement, and
      the create an archive file.
      In Util.* some functions have been added (toString to convert char* into string,
      and fexists to test file existence).
      git-svn-id: https://svn.ec-nantes.fr/eXlibris/Applis/TLSDuctile@2136 fbbead7c-fb4d-4173-aa67-51132c73c120
  19. 16 Jun, 2015 1 commit
  20. 10 Jun, 2015 1 commit
  21. 29 May, 2015 1 commit
  22. 28 May, 2015 1 commit
  23. 15 Apr, 2015 1 commit
  24. 13 Apr, 2015 1 commit
  25. 10 Apr, 2015 1 commit
  26. 09 Apr, 2015 1 commit
  27. 08 Apr, 2015 1 commit
  28. 07 Apr, 2015 1 commit
  29. 03 Apr, 2015 2 commits
  30. 01 Apr, 2015 1 commit