Hi all,

I have used* constraints.condense* function in a serial code to apply 
periodic constraints on system_matrix and system_rhs after their assembly 
in which I do not use *constraints.distribute_local_to_global* for 
cell_matrix and cell_rhs but I first assemble mass_matrix , laplace_matrix, 
nl_matrix and nl_term (the same approach as in step-25) and then 
syetem_matrix and system_rhs are obtained after some manipulations on the 
mentioned assembled matrices and vectors.

The problem is that* constraints.condense* doesn't work with 
*TrilinosWrappers::SparseMatrix 
*so the question is when we can not use *constraints.distribute_local_to_global 
*how we can apply constraints in the assembly for parallel codes. In the 
https://www.dealii.org/8.4.1/doxygen/deal.II/group__constraints.html 
<https://www.google.com/url?q=https%3A%2F%2Fwww.dealii.org%2F8.4.1%2Fdoxygen%2Fdeal.II%2Fgroup__constraints.html&sa=D&sntz=1&usg=AFQjCNHazlbNq8-QxRUDRTTpEWHT5IXCdg>
 the 
following is mentioned which I can't truly understand. 

"The condensation functions exist for different argument types: 
SparsityPattern 
<https://www.dealii.org/8.4.1/doxygen/deal.II/classSparsityPattern.html>, 
SparseMatrix 
<https://www.dealii.org/8.4.1/doxygen/deal.II/classSparseMatrix.html> and 
BlockSparseMatrix 
<https://www.dealii.org/8.4.1/doxygen/deal.II/classBlockSparseMatrix.html>. 
Note that there are no versions for arguments of type 
PETScWrappers::SparseMatrix() 
<https://www.dealii.org/8.4.1/doxygen/deal.II/classPETScWrappers_1_1SparseMatrix.html>
 or 
any of the other PETSc or Trilinos matrix wrapper classes. This is due to 
the fact that it is relatively hard to get a representation of the sparsity 
structure of PETSc matrices, and to modify them efficiently; this holds in 
particular, if the matrix is actually distributed across a cluster of 
computers. If you want to use PETSc/Trilinos matrices, you can either copy 
an already condensed deal.II matrix, or assemble the PETSc/Trilinos matrix 
in the already condensed form, see the discussion below."

Thanks and regards,
Hamed

-- 
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 [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to