Dear deal.II developers,

Thomas C. and myself are facing the following problem:
we work with the FESystem<dim>
                           (FE_Q<dim>(degree+1), dim,
                            FE_Nothing<dim>(), dim,
                            FE_Q<dim>(degree+1), dim,
                            FE_DGQ<dim>(degree), 1
                            )

and BlockCompressedSimpleSparsityPattern csp (4,4);

After initializing the different
blocks and calling  sparsity_pattern.copy_from (csp);
we get the error message

An error occurred in line <139> of file </home/people/twick/deal.II/source/lac/block_sparsity_pattern.cc> in function void dealii::BlockSparsityPatternBase<SparsityType>::collect_sizes() [with SparsityPatternBase = dealii::SparsityPattern]
The violated condition was:
    row_sizes[r] == sub_objects[r][c]->n_rows()
The name and call sequence of the exception was:
    ExcIncompatibleRowNumbers (r,0,r,c)
Additional Information:
The blocks [0,0] and [0,1] have differing row numbers.


The reason is that the blocks which couple with
the FENothing component have
n_rows() != 0 but n_cols() = 0!!!



In the function BlockSparsityPattern::copy_from (const BlockCompressedSimpleSparsityPattern &csp) we have checked that
  std::cout <<  sub_objects[0][1]->n_rows() << std::endl;
  std::cout <<  csp.block(0,1).n_rows()  << std::endl;

give different n_rows values:
sub_objects[0][1]->n_rows()  = 0
csp.block(0,1).n_rows() != 0

In our opinion there should be the same numbers.
It seems that during some operations (we do not know  where, therefore,
we write this email), for the object sub_objects[0][1]
n_rows() is set to zero because n_cols is zero.

Do you have any ideas for us?

Thanks,

Thomas & Thomas





--
++----------------------------------------++
Thomas Wick
University of Heidelberg
AG Numerik, Room 213
Im Neuenheimer Feld 293
69120 Heidelberg, Germany

Phone:  +49(0)6221 / 54-5449
Email:  [email protected]
www:    http://numerik.uni-hd.de/~twick/
++----------------------------------------++
--
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii

Reply via email to