Dear Professor Bangerth, The problem that I'm interested in solving is multi-physical and thus I'm handling with block matrices. These are treated in a particular way inside my library which is different than the BlockSparseMatrix type defined in deal.II. So the idea was to create interfaces for calling my solver and also for building the block matrices according to the data structure that I need for running my solver. Digging into the library I saw that a similar thing is done for the integration of PETSc and Trilinos to deal.II. Then I tried to follow how their integration is done and apply it for my solver. Please, let me know if this is not the best approach :)
Best! On Tuesday, July 4, 2017 at 11:14:22 PM UTC+2, Wolfgang Bangerth wrote: > > On 07/04/2017 08:01 AM, Victor Magri wrote: > > Hello Jean-Paul, sorry for not being clear about this... Answer is the > first > > option. > > > > I'm using the development version of libdeal.II and I added an interface > for > > calling my solver under include/lac/mysolver.h (In the solve method I > have a > > call to a function defined in my library). > > That seems like the more complicated version, compared to putting the new > solver into your own project -- any reason to put it there? > > Cheers > W. > > -- > ------------------------------------------------------------------------ > Wolfgang Bangerth email: bang...@colostate.edu > <javascript:> > www: http://www.math.colostate.edu/~bangerth/ > > -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to dealii+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
