when LAPACK is not found.

Merge branch 'fix/typo' into 'master'

See merge request [!789]

  [!789]: Nonecore/dune-common/merge_requests/789

Merge branch 'fix/typo' into 'master'

See merge request [!788]

  [!788]: Nonecore/dune-common/merge_requests/788

Merge branch 'eigenvalues_for_float' into 'master'

ref:core/dune-common\> This MR adds support for computing the eigenvalues and
vectors for FieldMatrices of field_type float

See merge request [!787]

  [!787]: gitlab.dune-project.org/core/dune-common/merge_requests/787

is not double or float

with field_type float

Merge branch 'feature/adjust-c++-checks' into 'master'

See merge request [!785]

  [!785]: Nonecore/dune-common/merge_requests/785

C++17 provides std::aligned_alloc.

Set the following variables always to 1:
 DUNE_HAVE_CXX_CLASS_TEMPLATE_ARGUMENT_DEDUCTION
 DUNE_HAVE_CXX_OPTIONAL
 DUNE_HAVE_CXX_VARIANT
 DUNE_SUPPORTS_CXX_THROW_IN_CONSTEXPR
 DUNE_HAVE_C_ALIGNED_ALLOC

Merge branch 'update-CI' into 'master'

See merge request [!786]

  [!786]: Nonecore/dune-common/merge_requests/786

Merge branch 'features/no-fortran' into 'master'

ref:core/dune-common\> FindLAPACK and FindBLAS check for the name mangling
with added underline as a fallback, anyway. Hence there is no need for using a
Fortran compiler. With this commit we now simply add another check for a
LAPACK function to see whether it has an underline added and adapt the
existing macro in config.h accordingly.

Closes [#184]

See merge request [!767]

  [#184]: gitlab.dune-project.org/NoneNone/issues/184
  [!767]: gitlab.dune-project.org/core/dune-common/merge_requests/767


Closes #184

Markus Blatt authored 5 years ago and Christoph Grüninger committed 5 years ago

FindLAPACK and FindBLAS check for the name mangling with added
underline as a fallback, anyway. Hence there is no need for using
a Fortran compiler. With this commit we now simply add another check
for a LAPACK function to see whether it has an underline added and
adapt the existing macro in config.h accordingly.

Closes #184

Merge branch 'feature/3d-eigenvalues+vectors' into 'master'

ref:core/dune-common\> When dealing with small matrices calling LAPACK to
calculate Eigenvalues and Eigenvectors can be quite slow. 1d- and
2d-specializations are quite straightforward, for the 3d-specialization I´m
using the approach presented in
[https://www.geometrictools.com/Documentation/RobustEigenSymmetric3x3.pdf].

See merge request [!776]

  [https://www.geometrictools.com/Documentation/RobustEigenSymmetric3x3.pdf]: https://www.geometrictools.com/Documentation/RobustEigenSymmetric3x3.pdf
  [!776]: gitlab.dune-project.org/core/dune-common/merge_requests/776

Merge branch 'loopsimd_alignment' into 'master'

ref:core/dune-common\> This MR remerges [!670], which was reverted due to to
much time and memory requirement for looptest. This MR merges the branch again
but with two changes:

-   It adds explicit instantiations to the looptest
-   It changes the default template argument of the alignment to 0, which
    means that it has no impact. In particular, if it is set to 0, rebound
    types have the template argument 0 as well. That reduces the number of
    types for that the tests are compiled and executed.

See merge request [!784]

  [!670]: gitlab.dune-project.org/NoneNone/merge_requests/670
  [!784]: gitlab.dune-project.org/core/dune-common/merge_requests/784

This reverts commit ebae92f0.

Merge branch 'revert-3c5922d7' into 'master'

ref:core/dune-common\> This reverts merge request [!670]

As explained by @carsten.graeser the compiler costs increased dramatically.
Let's revert this MR until we sorted out the actual problem and keep the
compile time in the usual bounds.

See merge request [!782]

  [!670]: gitlab.dune-project.org/NoneNone/merge_requests/670
  [!782]: gitlab.dune-project.org/core/dune-common/merge_requests/782

This reverts merge request !670

Merge branch 'loopsimd_alignment' into 'master'

ref:core/dune-common\> This MR adds a additional template parameter to
LoopSIMD to specify the alignment.

The default alignment is the alignment of the underlying type.

I'm not sure how to determine the proper alignment for some return types or
RebindType. To mitigate that problem I define conversion constructors and
assignment operators from LoopSIMD types with different alignment.

This MR depends on [!669].

Feel free to push at the source branch!

See merge request [!670]

  [!669]: gitlab.dune-project.org/NoneNone/merge_requests/669
  [!670]: gitlab.dune-project.org/core/dune-common/merge_requests/670

Merge branch 'loopsimd_for_vector_types' into 'master'

ref:core/dune-common\> This MR adapts LoopSIMD to take another vectorized type
as template argument. If Tis a vectorized type, then LoopSIMD<T,k> has
lanes<T>()*k lanes and the same scalar as T (Scalar<LoopSIMD<T,k>> ==
Scalar<T>).

This is useful with dune-vectorclass, to create arbitrary large vector types
with explicit SIMD support, for example.

Feel free to push any changes to the source branch!

See merge request [!669]

  [!669]: gitlab.dune-project.org/core/dune-common/merge_requests/669