21. mai 2015 kl. 10:25 skrev Tobias Ritschel <[email protected]>:

> Thanks for the great feedback. I will proceed with the DUNE solvers and I 
> might compare to the Eigen solvers for a simple setup.
> 
> I also considered PETSc as I use TAO (part of PETSc) to carry out 
> optimization, though I think it is easier to use DUNE together with the Eigen 
> SparseMatrix format.
> 
> Regarding the DuneMatrix object, it seems that it takes a SparseMatrix<double 
> Eigen::RowMajor> object as input, whereas the M object from the AutoDiffBlock 
> class is of type SparseMatrix<double>, i.e. using the default 
> Eigen::ColumnMajor as the _Options input.
> 
> Could this be a problem?

You will need to convert the matrix to row format first, but that is very easy, 
you just construct
a new matrix using the copy constructor:

SparseMatrix<double Eigen::RowMajor> row_matrix(col_matrix);

Atgeirr


_______________________________________________
Opm mailing list
[email protected]
http://www.opm-project.org/mailman/listinfo/opm

Reply via email to