Mariia
Anyway, I rewrote several script parts to get rid of block matrices and
vectors. The compiler does not complain anymore, while building CXX object,
but it returns an error while linking CXX executable. It concerns the
affine_constraints (as provided below). I wonder, whether there is
Dear Wolfgang,
Thank you for your response. There is indeed not much of a need to keep the
block structure in the code with a direct solver, but I was wondering
still, as the script looks more compact this way.
Anyway, I rewrote several script parts to get rid of block matrices and
vectors. Th
On 1/19/22 03:41, Мария Бронзова wrote:
I checked the member function documentation for the solve() function of
MUMPS and found only the solve() function that depends on the *const
MatrixBase, *but I guess in my case *const* *BlockMatrixBase * would
make more sense.
Could you please tell me,
Dear Wolfgang,
Thank you a lot for the explanation, I got PETSc preinstalled already and
didn't have a chance to see there were different compiling options. Now
it's clear!
Anyway, I ran into a further issue with my script that I wanted to
parallelize. I am dealing with non spd matrices in ter
Mariia,
I was trying to make adjustments in my script to enable parallel computing (as
it takes unbearably long to calculate it even for small problems due to big
sparsity matrix, I guess).
In my script I am using block matrices and vectors with complex elements
(i.e. BlockSparseMatrix>).
Dear colleagues,
I was trying to make adjustments in my script to enable parallel computing
(as it takes unbearably long to calculate it even for small problems due to
big sparsity matrix, I guess).
In my script I am using block matrices and vectors with complex elements
(i.e. BlockSparseMatr
