It seems to me that what happens is that all values in the vectors
associated with variable 0 are copied to variable 1 and 2.
Also, when I run in devel mode, I get the error:
Assertion `var < this>n_vars(s)' failed.
Stack frames: 11
0: libMesh::print_trace(std::ostream&)
1: libMesh::DofObject::dof_number(unsigned int, unsigned int, unsigned
int) const
2: libMesh::DofMap::old_dof_indices(libMesh::Elem const*,
std::vector<unsigned int, std::allocator<unsigned int> >&, unsigned int)
const
3:
libMesh::BuildProjectionList::operator()(libMesh::StoredRange<libMesh::MeshBase::const_element_iterator,
libMesh::Elem const*> const&)
4: libMesh::System::project_vector(libMesh::NumericVector<double>
const&, libMesh::NumericVector<double>&) const
5: libMesh::System::project_vector(libMesh::NumericVector<double>&) const
6: libMesh::System::restrict_vectors()
7: libMesh::EquationSystems::reinit()
On 08/27/2012 01:48 PM, Jens Lohne Eftang wrote:
> Hi,
>
> I have an EquationSystems object with one System. This system has many
> additional vectors that I have added using System::add_vector(). I
> write everything out using EquationSystems::write() with the flags
> specified as (WRITE_DATA  WRITE_ADDITIONAL DATA).
>
> My System has three variables, and each added vector has nonzero
> entries only for one of the variables. If I print each of the vectors
> to screen prior to writing them to disk, everything looks fine.
> However, if I open the .xda file and look at the additional vectors,
> they are messed up. Now, some vectors have nonzero entries for all
> variables (all entries in the vector), and some vectors have all zero
> variables. Interestingly, it looks as if only every third vector is
> nonzero.
>
> Any ideas on what could be the issue here?
>
> Best,
> Jens
