* Added build directly to gitignore
* First cut at mfem windows build
* Adding windows build files
* Fixed relative path issue
* Remove cmake rebuild from mfem project, fix filters issues with ms
* Response to Review, making sure it builds for different windows build types
* Windows build fixes
* Cleaning up the build fixes
* Fixing .gitignore for some of the newer packages
* Fixing code for MSVC Compiler in preperation for NVCC build
* Upgrade to latest VS2017
* commit after merge
* Tweeks
* Fix ordering of help options, protect agains not having SNLS in main code
* Fixed windows build
* more project files
* Move Eng_Matmodel under class form, remove some errors/warnings in RDG
* Response to review 5
* Fix accidentally deleted constructor
* Windows build fixes; further refactoring of .vcxproj files
* Added FindHDF5.cmake file to mfem
* Added time, time step, and cycle to conduit data collection
* Further compatibility fixes, Allowing non-chemical materials to use region inflow, updating documentation
* Fix merge
* Adding a preprocessor definition for hypre.
* Update mfem project
* Update project
* Fix typos in mfem project
* Update MFEM HYPRE version
* Split mfem projects
* Update projects
* Update projects
* Update projects
* More tweaks for windows build
* Fix tribol and mfem build on windows
* Fix hypre and mfem projects
* Fix tribol project; add opac gitignore
* Update project for new mfem version
* Update mfem build to import umpire target so that camp can be found; add empty lapack include directory that is required if mfem config files are used
* Fix to mfem for new umpire
* Update Windows project files
* Undo changes to windows project files
* Update windows project files
* Fix all windows configurations for several libraries
* Fix CMake error in MFEM
* Adding functions to set another preconditioner, instead of BoomerAMG.
Adding EliminationGMRESSolver and PenaltyGMRESSolver to use GMRES
instead of CG.
* Adding some functions to get the number of iterations and residual norm
from Hypre solvers.
* Adding two lines to return the right residual and convergence information.
* Adding some functionality to choose the preconditioner for the SchurConstrainedHypreSolver.
* Vectorizing the penalty function, as we can have different penalty values
for each constraint equation.
* We should initialize a diagonal HypreParMatrix here, instead of trying
to do row scalings.
* Adding some functions to the Hypre solver classes to set parameters.
* Whoops, this function should be for logging.
* Make this a multiplication by a diagonal SparseMatrix.
* Memory fix for SuperLU solver. This passes the address sanitizer now.
* Update to VS2022 and Intel oneAPI (with some Intel Classic)
* Fixing the Windows build for MFEM.
* Applying the change to include directories to all builds.
* Fixing undefined references for MFEM.
* make style
* Fix copyright
* Fix hidden overload
* Fixing the hidden overload here.
* Some fixes to the Windows build, after adding the SuperLU_Dist option.
* Fix MFEM build
* Update .gitignore
* fix windows build issues
* Removing dead code, adding comments.
* Consistend FindHDF5.cmake (and don't mention ALE3D)
* Rename property sheets to more consistent pattern
* Moving some things around gitignore, moving the Win32 directory to config.
Changing the relative directory structure in the project files after moving
the directory.
* Adding a function to allow for parallel symbolic factorization.
* Editing the CHANGELOG and INSTALL files.
* Editorial updates
* Remove config/Win32 (will be maintained externally)
Co-authored-by: Jamie A. Bramwell <bramwell1@llnl.gov>
Co-authored-by: Albert L. Nichols, III <nichols5@llnl.gov>
Co-authored-by: Arlie Capps <capps2@llnl.gov>
Co-authored-by: Patrick Greene <greene30@llnl.gov>
Co-authored-by: Benjamin T. Liu <liu15@llnl.gov>
Co-authored-by: Alan Dayton <dayton8@llnl.gov>
Co-authored-by: Tzanio Kolev <tzanio@llnl.gov>
GatherBlockOffsetData().
Add overflow check in GatherBlockOffsetData().
In the SuperLURowLocMatrix constructor from HypreParMatrix,
remove the check for the number of columns and add a note why it is
not necessary at the moment.
Add a compile-time check when building with SuperLU_DIST support
that HYPRE_Int is int which is required by the current implementation.
Nuno Nobre