Wolfgang,
Thanks yet again.
> No, it sees 4 matrices. But if you don't need the blocks, just use a
> SparseMatrix instead of a BlockSparseMatrix.
Ok I see now... I just understood how that works. I don't need a
BlockSparseMatrix unless I really actually do need to access blocks; I did
not see it like that before now. The documentation is actually rather
clear about this, so I don't know what I missed or how I got confused...
I wanted to do it like that to compare the result with my own hand-made
SchurComplement solver (which I do realise I can do *likewise* to the
deal.II tutorials by adding something like a "D-SchurComplement" before
solving) I need a comparator to that for debugging (hence, GMRES).
> > Any guesses which is the way to go?
>
> This doesn't depend on the structure of the matrix but the
> problem this structure results from. The implementation will follow what you
> want to do, not the other way around ;-)
You are absolutely right of course. That is such clear and yet valuable
advice, that I am going to write it down and read it every day so I
never again will forget! :-)
Best,
Toby
-----
Toby D. Young
Assistant Professor
Institute of Fundamental Technological Problems
Polish Academy of Sciences
ul Adolfa Pawinskiego 5b
02-106 Warsaw
Poland
www: http://www.ippt.gov.pl/~tyoung
skype: stenografia
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
