recalculateHierarchy() assumed that there is information avaiable
that is only there in parallel runs. This failed miserably.

This commit fixes this be implementing dummy versions for
redistributing matrices in  a sequential run and does not
query any parallel stuff any more

This closes #11.

[amg][bugfix] Fix dirichlet processing in FastAMG

The comment and surroundig code indicate that FastAMG should
solve during pre() for dirichlet entries. This is determined
by checking if a row does contain a diagonal entry while
all other entries are zero.

The check for this was buggy because hasDiagonal was always false.
This merge commits sets it to true if a diagonal entry is found and nonzero.

See merge request !32

This allows other code to access field_type and real_type for a MultiTypeBlockVector
using the FieldTraits class.  Caveat: in principle, individual entries of a
MultiTypeBlockVector could use different field_types.  The implementation always
returns the first one.

Adding a new template perameter before the existing ones is
a non-compatible interface change, because code that explicitly
instantiates the template will fail to compile afterwards.

Putting the new template parameter to the end fixes the issue

I missed a few occurrences of 'nblocks' when renaming that to 'rows_',
because it was hidden behind the DUNE_ISTL_WITH_CHECKING macro.

[amg] Silence FastAMG if debugLevel=0

There was an unconditional debug output in FastAMG.
With this commit this is will only be printed if
debugLevel>0.

See merge request !31

Carsten Gräser authored 9 years ago and Christoph Grüninger committed 9 years ago

There was an unconditional debug output in FastAMG.
With this commit this is will only be printed if
debugLevel>0.

This additional check currently fails.

[fix] Remove coumns_, probably a copy-n-paste artifact



See merge request !33

The comment and surroundig code indicate that FastAMG should
solve during pre() for dirichlet entries. This is determined
by checking if a row does contain a diagonal entry while
all other entries are zero.

The check for this was buggy because hasDiagonal was always false.
Now it's set to true if a diagonal entry is found and nonzero.

The dune-common module now contains short-cuts for static indices of type
integral_constant<size_t,...>, and we should try to standardize on those.

In particular, this means that static indices are integral_constant<std::size_t,...> now,
rather than integral_constant<int,...>.  We decided to standardize on std::size_t.

This array was a left-over from the VariableBlockVector class.  In short,
it stores the length of each matrix row.  Of course that is redundant,
because all rows in a dense matrix have the same length.

This patch makes the class allocate less memory, and use the known row
size to compute the offsets into the data array.  In principle, the code
becomes faster by this, but I am not convinced that the effect will be
measurable.

Still, even without measurable speedup: this patch is an improvement,
because it makes the code simpler.

doc: fix typo: writee -> write



See merge request !30

[bugfix] add missing header

Header `dune/common/exceptions.hh` is not included but exceptions are used.

See merge request !29

Matrix loadbalance to more

There was a problem when having the global matrix and using load balance
to distribute it. This is fixed by the MR.

These changes should be cherrypicked to the release, too.

See merge request !28

One of the use cases for load balancing a matrix is reading it globally
on one process and then calling the load balancer to distribute it among
the processes. Unfortunately this did not work due to a bug that prevented
the correct remapping of the domain numbers from the load balancers. This
lead to negative rank numbers.

With this commit we use yet unassigned process numbers for domains that
are not yet mapped (due to having no dofs before the load balancing).

[cmake][bugfix] Correct handling of the EXCLUDE_FROM_ALL property:

The signature of `dune_add_test`, that takes the `TARGET` from the user
produced excluded targets if added in a directory that has the directory
property `EXCLUDE_FROM_ALL` set. The correct solution IMO is to not use
that property on test subdirectories anymore (it is a relic from the old
testing magic), because we rely on `dune_add_test` to handle exclusion.

See dune-common#21 for details.

See merge request !27

The signature of `dune_add_test`, that takes the `TARGET` from the user
produced excluded targets if added in a directory that has the directory
property `EXCLUDE_FROM_ALL` set. The correct solution IMO is to not use
that property on test subdirectories anymore (it is a relic from the old
testing magic), because we rely on `dune_add_test` to handle exclusion.

See core/dune-common#21 for details.

I had removed it by accident in one of the previous commits.

All blocks have the same size, therefore 'block' is not actually needed.
This is the first step towards getting rid of it.

The recent changes to VariableBlockVector broke the Matrix class -- shame on me
for testing thoroughly enough before pushing.  However, using VariableBlockVector
internally in Matrix was a hack anyway.  So, as a first step towards a cleaner
solution, this patch copies the (old) VariableBlockVector class into the file,
renames VariableBlockVector to DenseMatrixBase, and does some obvious cleanup.
This is still not particularly pretty, but better than before.

This fixes a valgrind warning in the test for VariableBlockVector, where the reference type
is an internal proxy class.

Just as containers from the standard library do.  These types are needed for
proper interface testing of the dune-istl vector interface.  For VariableBlockVector,
in particular, the type used for references is *not* value_type&, and therefore
the more general approach using a separate 'reference' type is needed.

Yes, strictly speaking this is an API change.  But the VariableBlockVector code
has been so buggy until the last few patches that I doubt anyone actually used it.

Previously, a method of the same name was inherited from the base class, which
however returned bogus results.

- much shorter code
- makes the iterators pass the test suite

The FieldVector class resides in dune-common, and is tested there.  However, we also
want to check it against the tests specifying the dune-istl vector interface in vectortest.hh,
therefore we add a second test for FieldVector to the dune-istl module.

Previously, they were inherited from the base class, resulting in the wrong types.

So far, each dune-istl vector class had its own unit test.  They all tested, among
other things, for compliance with the dune-istl vector interface.  This lead, or had
the risk to lead, to subtle differences between the individual tests.  Also, since
there is no formal specification of the interface, it was somewhat unclear what the
interface actually requires.

This patch tries to improve the situation by providing a generic set of tests that
should be passed by anything claiming to be a dune-istl vector.  This will save work,
because it allows to centralize tests that are now distributed across the different
unit tests for the different dune-istl vectors.  Also, it is my hope that these
centralized tests can serve as some form of formalized specification of the dune-istl
vector interface.