Currently RestartedGMRes and MINRes don't work with multiple rhs, as
it is not straight forward to rewrite the Givens-rotation in a
vectorized way.

- create dedicated "eigenvalue solvers" group
- move implementation details into the "Impl" namespace
- minor cleanups of doxygen documentation

This header is included by the installed headers superlu.hh and
supermatrix.hh.

Markus Blatt authored 8 years ago and Christoph Grüninger committed 8 years ago

Previously, we only supported one numeric type (float, double, complex<float>,
or complex<double>) that the user could select via a cryptic preprocessor
define. This was done this way because when including SuperLU headers we
would otherwise have multiply defined structs in SuperLU. At least in 4.3
GlobalLU_t is defined in slu_sdefs.h, slu_ddefs.h, slu_cdefs.h, and
slu_zdefs.h

With this commit we do not include the problematic SuperLU headers anymore.
Instead why define the functions called by us with extern C within DUNE.
In addition we now provide all the numeric types for which SuperLU drivers are
found at the same time.

Hopefully this will also fix problems with multiply defined BLAS routines experienced
with arpack and SuperLU.

Oliver Sander authored 8 years ago and Ansgar Burchardt committed 8 years ago

There are various classes in dune-istl that are effectively implementation
classes internal to dune-istl. A few of them are base_array_unmanaged,
base_array_window, compressed_base_array_unmanaged,block_vector_unmanaged,
BlockVectorWindow, compressed_block_vector_unmanaged, CompressedBlockVectorWindow

See #22 for a brief
discussion.

This patch moves the classes into the namespace 'Imp'.  That way it is
clear that they are internal.

All code outside of dune-istl using those classes will break, but does
such code really exist?

is selected. Since UMFPack does not support this, it cannot be used in this
case.

By using the doxygen \internal command.

The implementation for this was hidden in the base classes
before and relied on undefined behaviour by using a downcast.
Now it is explicitly implemented for the derived classes
which also makes it more visible to users.

These constructors and assignments from base class are
all using a downcast which may result in undefined behaviour.
Since this is potentially dangerous and not used anywhere
in the core modules, this patch removes those methods.

These constructors and assignments are all implemented using
a downcast to derived class and may lead to undefined behaviour.
This commit removes the checks for these methods in preparation
for their removal.

The doc now correctly describes
* that the overflow area is used during compress and not during assembly
* the meaning of avg and overflowsize
* the numbering dependence of this build mode
* the magic number 4
* the late failure during compress()

Stephan Hilb authored 8 years ago and Christoph Grüninger committed 8 years ago

This fixes the clang compiler warnings about tautological compare
when idxtype is unsigned. For clang behaviour see [1].

[1]: https://llvm.org/bugs/show_bug.cgi?id=8682

This reverts ddf49c23 and removes the check if
the average parameter behaves as expected.

similar to Matlab matrix writer

Remove stamp-vc file.

Since the istl view is, that a matrix is a container of rows,
it should have a size method.

This is in fact an _unsafe_ downcast to a derived
class. Accessing members via this pointer is undefined
if it is not guaranteed that the passed pointer
points to an object of the _derived_ class - even
if the memory layout is the same. In this case
this static_cast acts like a reinterpret_cast.

If the correct average is provided, compress() may still
fail if more than the average numebr of entries are added
to an early row (e.g. the first one)

Due to a confusion between the number of lines N and the number of
columns M, Matrix::transpose() gave wrong results for matrices with
N != M.

Due to a bug in Matrix::transpose() it will give wrong results for
non-quadratic matrices. Thus the new test case fails for the moment.

Remove output that printed the matrix for parallel runs for small grids n<20.

According to git blame the last time these constructors were changed was
in 2011.

Simplifies the memory management.