]>
https://gitweb.dealii.org/ - dealii.git/log
Matthias Maier [Mon, 7 Sep 2015 22:16:09 +0000 (17:16 -0500)]
Testsuite: Fix invalid constructor calls
This should have never been syntactically correct code...
Timo Heister [Mon, 7 Sep 2015 21:16:55 +0000 (17:16 -0400)]
Merge pull request #1545 from tamiko/fix_cxx11_thread_check
CMake: Bugfix: Export -pthread when trying to link
Matthias Maier [Mon, 7 Sep 2015 21:04:01 +0000 (16:04 -0500)]
CMake: Bugfix: Export -pthread when trying to link
It turns out that clang is a bit more picky and wants to have -pthread
on the command line.
Matthias Maier [Mon, 7 Sep 2015 19:15:44 +0000 (14:15 -0500)]
Merge pull request #1517 from tamiko/complexify
Tensor<rank,dim,Number> - major cleanup
Matthias Maier [Mon, 7 Sep 2015 19:13:59 +0000 (14:13 -0500)]
Bugfix: Use generic initializer in Tensor::norm()
Matthias Maier [Mon, 7 Sep 2015 19:03:55 +0000 (14:03 -0500)]
Do not use Tensor<..>(bool) contstructor in headers and testsuite
Matthias Maier [Mon, 7 Sep 2015 18:49:05 +0000 (13:49 -0500)]
Cosmetic change: Attach operator to conversion type
Matthias Maier [Mon, 7 Sep 2015 18:46:20 +0000 (13:46 -0500)]
Tensor, Point: Remove constructor taking a boolean argument
This boolean has no effect any more. Simply remove it.
Matthias Maier [Mon, 7 Sep 2015 18:38:50 +0000 (13:38 -0500)]
Cosmetic change: Use member initializer list
Matthias Maier [Mon, 7 Sep 2015 16:05:27 +0000 (11:05 -0500)]
Bugfix: Put tensor_base.h back in place.
Matthias Maier [Mon, 7 Sep 2015 16:02:34 +0000 (11:02 -0500)]
Bugfix: Remove an unnecessary second pass for value initialization
Matthias Maier [Mon, 7 Sep 2015 04:42:48 +0000 (23:42 -0500)]
Bugfix: Avoid C++11 only syntax
Matthias Maier [Mon, 7 Sep 2015 04:37:27 +0000 (23:37 -0500)]
Workaround for a bug in gcc-4.8 wrt to operator* and mixed std::complex types
Matthias Maier [Mon, 7 Sep 2015 02:35:26 +0000 (21:35 -0500)]
Bugfix: Dissallow access to an object of type Tensor<rank,0,Number>
Well, there is no sane way of accessing such an object. Unfortunately, the
"degeneration" operator[](void) is not part of the C++ language...
So, dissallow any access to the value but allow creation and comparison of
this object...
Matthias Maier [Mon, 7 Sep 2015 02:20:09 +0000 (21:20 -0500)]
Remove unnecessary specializations and a confusing comment
Matthias Maier [Mon, 7 Sep 2015 02:15:41 +0000 (21:15 -0500)]
Obsolete tensor_base.h, refactor everything to tensor.h
Matthias Maier [Mon, 7 Sep 2015 02:02:51 +0000 (21:02 -0500)]
Bugfix: Avoid temporaries
Matthias Maier [Mon, 7 Sep 2015 01:39:32 +0000 (20:39 -0500)]
Bugfix: Only run this test in C++11 mode
Matthias Maier [Mon, 7 Sep 2015 01:28:14 +0000 (20:28 -0500)]
Bugfix: Remove ambiguity.
With the rest of Tensor<0,dim,Number>'s interface this operator= is
redundant.
Matthias Maier [Mon, 7 Sep 2015 01:05:16 +0000 (20:05 -0500)]
Bugfix: Do not use C++11 features
Matthias Maier [Mon, 7 Sep 2015 01:03:36 +0000 (20:03 -0500)]
Bugfix: Add more typetraits to ProductType for non C++11
Matthias Maier [Fri, 4 Sep 2015 07:24:08 +0000 (02:24 -0500)]
Bugfix: Allow creation and handling of Tensor<rank, dim ==0>
Such objects only store a zero (which doesn't make much sense to work with
them). But for certain situations in the library we need to be able to
create such objects.
Matthias Maier [Fri, 4 Sep 2015 05:43:46 +0000 (00:43 -0500)]
Bugfix: Remove superfluous operator<< and fix ambiguity
Matthias Maier [Fri, 4 Sep 2015 04:01:11 +0000 (23:01 -0500)]
Bugfix: Tensor<rank,dim>: Always store a tensor type internally
Matthias Maier [Fri, 4 Sep 2015 04:57:08 +0000 (23:57 -0500)]
Bugfixes: Update wrong descriptions, fix typos
Matthias Maier [Fri, 4 Sep 2015 04:26:41 +0000 (23:26 -0500)]
Bugfix: Update Point<dim,Number> documentation, fix norm, norm_square
Both are supposed to return a real valued scalar product. Update
documentation and implementation accordingly.
Matthias Maier [Thu, 3 Sep 2015 05:54:41 +0000 (07:54 +0200)]
Bugfix: Avoid C++11 features
Do not use template default arguments for function templates.
Also provide fully templated interface for scalar multiplication with
point, too.
Matthias Maier [Thu, 3 Sep 2015 04:15:06 +0000 (23:15 -0500)]
Bugfix: Promote VectorizedArray<Number> to a scalar type
Now, every operation concerning Tensors are guarded with the EnableIfScalar
type trait. Consequently, a VectorizedArray has also to be made a scalar
type.
Matthias Maier [Thu, 3 Sep 2015 00:07:08 +0000 (19:07 -0500)]
run astyle
Matthias Maier [Wed, 2 Sep 2015 23:38:21 +0000 (18:38 -0500)]
add a news/changes.h entry
Matthias Maier [Wed, 2 Sep 2015 22:44:17 +0000 (17:44 -0500)]
Minor Bugfixes on Point<dim>
- use numbers::NumberTraits<Number>::abs_square to compute the squared
euclidian norm
- allow scaling with arbitrary, compatible number type
- implement some functions directly to be independent of the Tensor<>
implementation
Matthias Maier [Wed, 2 Sep 2015 22:17:06 +0000 (17:17 -0500)]
Restructure declarations in tensor_base.h and tensor.h
Move everything that isn't necessary from tensor_base.h to tensor.h
Matthias Maier [Wed, 2 Sep 2015 03:51:05 +0000 (22:51 -0500)]
Eliminate specialization for Tensor<1,dim>
This is more or less an exact code copy of Tensor<rank_,dim>.
tensor_base.h now contains the full definition of Tensor<rank_,dim>. All
external functions operating on tensors that are not strictly needed reside
in tensor.h.
Matthias Maier [Wed, 2 Sep 2015 03:18:22 +0000 (22:18 -0500)]
Tensor<rank, dim>: Finish interface cleanup
Matthias Maier [Tue, 1 Sep 2015 00:40:51 +0000 (19:40 -0500)]
Testsuite: Add a test for mixed type operations
Matthias Maier [Wed, 2 Sep 2015 00:43:12 +0000 (19:43 -0500)]
Tensor<1, dim>: Finish interface cleanup
Matthias Maier [Tue, 1 Sep 2015 23:37:21 +0000 (18:37 -0500)]
Tensor<0,dim>: Use cast instead of friend declaration to access value
Matthias Maier [Tue, 1 Sep 2015 23:29:00 +0000 (18:29 -0500)]
Tensor<0,dim>: Also provide operators for scalar multiplication
Matthias Maier [Tue, 1 Sep 2015 22:43:56 +0000 (17:43 -0500)]
Tensor<k,dim>: Remove superfluous operator* variants
It is completely sufficient to provide 1 ("one") templated variant.
I guess, over time more and more variants were added. This increased the
grade of ambiguity for the compiler to chose the one intended
implementation.
Matthias Maier [Mon, 31 Aug 2015 23:21:45 +0000 (18:21 -0500)]
Tensor<0,dim,Number>: Implement mixed (tensor) operations
This allows arbitrary arithmetic operations of mixed value_type operands
such as real and complex valued tensors.
Matthias Maier [Tue, 1 Sep 2015 00:23:06 +0000 (19:23 -0500)]
Bugfix: Fix type ambiguity for two calls to std::min
Wolfgang Bangerth [Mon, 7 Sep 2015 17:56:45 +0000 (12:56 -0500)]
Merge pull request #1515 from tcclevenger/write_svg
Add SparsityPattern::write_svg
Martin Kronbichler [Mon, 7 Sep 2015 16:30:37 +0000 (18:30 +0200)]
Merge pull request #1544 from maieneuro/remove_compute_2nd
removed compute_2nd as hessians are analytically computed
Maien Hamed [Mon, 7 Sep 2015 16:03:48 +0000 (18:03 +0200)]
removed compute_2nd as hessians are analytically computed
tcclevenger [Wed, 2 Sep 2015 20:38:03 +0000 (16:38 -0400)]
Add SparsityPattern::print_svg
Martin Kronbichler [Mon, 7 Sep 2015 14:39:21 +0000 (16:39 +0200)]
Merge pull request #1541 from tjhei/enable_muelu_tests
enable muelu tests with newer Trilinos
Timo Heister [Mon, 7 Sep 2015 14:38:16 +0000 (10:38 -0400)]
enable muelu tests with newer Trilinos
Timo Heister [Mon, 7 Sep 2015 14:08:26 +0000 (10:08 -0400)]
Merge pull request #1516 from tjhei/new_inst_file_splitting
rework fe_values.cc inst splitting
Timo Heister [Mon, 7 Sep 2015 13:23:08 +0000 (09:23 -0400)]
address comments
- rename files
Timo Heister [Mon, 7 Sep 2015 12:58:35 +0000 (08:58 -0400)]
Merge pull request #1534 from drwells/refactor-q1-mapping-with-exceptions
Use exceptions to mark Q1 transform failures.
Timo Heister [Mon, 7 Sep 2015 01:41:06 +0000 (21:41 -0400)]
Merge pull request #1536 from tamiko/fix_cxx11_thread_detection
CMake: Avoid running CXX11 Thread support test
Wolfgang Bangerth [Sun, 6 Sep 2015 22:42:13 +0000 (17:42 -0500)]
Merge pull request #1539 from tamiko/gold
Use gold instead of the traditional linker if possible.
David Wells [Tue, 1 Sep 2015 21:51:37 +0000 (17:51 -0400)]
Use exceptions to mark Q1 transform failures.
The previous implementation returned magical numbers in some instances.
Matthias Maier [Sun, 6 Sep 2015 21:46:50 +0000 (16:46 -0500)]
CMake: Refactor check for -fuse-ld=gold into check_01_compiler_features.cmake
Wolfgang Bangerth [Mon, 31 Aug 2015 16:42:20 +0000 (11:42 -0500)]
Use gold instead of the traditional linker if possible.
Wolfgang Bangerth [Sun, 6 Sep 2015 19:47:45 +0000 (14:47 -0500)]
Merge pull request #1522 from bangerth/further-FEValues-cleanups-32
Make MappingQ1::InternalData more generic
Wolfgang Bangerth [Sun, 6 Sep 2015 18:46:39 +0000 (13:46 -0500)]
Reapply a patch that would otherwise have been lost in a rebase.
Wolfgang Bangerth [Sat, 5 Sep 2015 13:07:33 +0000 (08:07 -0500)]
Store the correct polynomial degree in MappingQGeneric.
Currently, MappingQ derives from MappingQ1 which passes 1 down as polynomial degree.
This worked 'by accident' because the polynomial degree is only used by MappingQGeneric
in get_data() and friends, which are overloaded by MappingQ. However, the correct
place to store the polynomial degree is clearly in MappingQGeneric.
Wolfgang Bangerth [Sat, 5 Sep 2015 00:05:32 +0000 (19:05 -0500)]
Remove MappingQ copy constructor.
It is apparently not needed.
Wolfgang Bangerth [Fri, 4 Sep 2015 23:13:00 +0000 (18:13 -0500)]
Simplify code.
Wolfgang Bangerth [Fri, 4 Sep 2015 16:18:13 +0000 (11:18 -0500)]
Some more preparatory work for separating MappingQ/Q1.
Some documentation updates.
Also make the Q1::InternalData object in MappingQ a pointer since that
will fit better into the structure we will need later on.
Wolfgang Bangerth [Fri, 4 Sep 2015 01:57:34 +0000 (20:57 -0500)]
Refactor MappingQ/Q1 into a common base class.
This patch separates out the common functionality for generic polynomial
mappings into a base class. It contains most of what used to be in MappingQ1
but really was not specific to p=1 mappings.
Wolfgang Bangerth [Thu, 3 Sep 2015 15:59:45 +0000 (10:59 -0500)]
Make MappingQ::InternalData protected.
Wolfgang Bangerth [Thu, 3 Sep 2015 15:46:52 +0000 (10:46 -0500)]
Remove a whole bunch of code.
None of this is needed any more now that the InternalData
class has been largely separated.
Wolfgang Bangerth [Thu, 3 Sep 2015 15:00:04 +0000 (10:00 -0500)]
Move the computation of the Mapping values into initialize().
Rather than having to do it yourself right after initialize(), let initialize()
call the function that does that. This greatly simplifies a good deal of code,
and also leads to better parallelization.
Wolfgang Bangerth [Thu, 3 Sep 2015 13:15:26 +0000 (08:15 -0500)]
Move the code that initializes the Mapping{Q,Q1}::InternalData into that structure.
This allows us to get rid of the two implementations of compute_shapes(). It also
significantly simplifies the logic which of the two functions actually need to
be called, since the InternalData object actually knows that itself.
Wolfgang Bangerth [Wed, 2 Sep 2015 22:44:59 +0000 (17:44 -0500)]
Get rid of the awkward dispatch mechanism in MappingQ*::compute_shapes().
In particular, remove the manual dispatch to either the Q1 or Qp functions.
Wolfgang Bangerth [Wed, 2 Sep 2015 19:46:45 +0000 (14:46 -0500)]
Attempt to disentangle MappingQ1 and MappingQ.
Store the polynomial degree in MappingQ1::InternalData, rather than
which kind of mapping created it.
Wolfgang Bangerth [Sun, 6 Sep 2015 17:06:30 +0000 (12:06 -0500)]
Merge pull request #1533 from drwells/tria-iterator-greater-than
Add operator< to TriaRawIterator.
Matthias Maier [Sun, 6 Sep 2015 08:10:42 +0000 (03:10 -0500)]
Documentation: Increase minimal compiler versions to reflect reality
Matthias Maier [Sun, 6 Sep 2015 06:04:52 +0000 (01:04 -0500)]
CMake: Avoid running CXX11 Thread support test
We have had multiple problems with running this test so far. The latests is
pull request #1524, the attempted fix opens another problem #1535.
Given the fact that the test tries to detect a problem with gcc-4.5, just
check for successful compilation.
David Wells [Sat, 5 Sep 2015 21:26:20 +0000 (17:26 -0400)]
Add operator< to TriaRawIterator.
This just switches the arguments to operator>.
Wolfgang Bangerth [Sat, 5 Sep 2015 20:12:24 +0000 (15:12 -0500)]
Merge pull request #1532 from QiaoLei-88/step15Eqn
correct equation in intro of step-15
Lei Qiao [Sat, 5 Sep 2015 19:45:29 +0000 (14:45 -0500)]
correct equation in intro of step-15
Luca Heltai [Sat, 5 Sep 2015 16:41:49 +0000 (18:41 +0200)]
Merge pull request #1530 from asartori86/fix_multiple_parameter_loop
fix multiple parameter loop tests
Timo Heister [Sat, 5 Sep 2015 16:20:52 +0000 (12:20 -0400)]
new .inst splitting insolution_transfer.cc
alberto sartori [Sat, 5 Sep 2015 16:18:02 +0000 (18:18 +0200)]
fix multiple parameter loop tests
Timo Heister [Sat, 5 Sep 2015 15:28:49 +0000 (11:28 -0400)]
add changelog
Timo Heister [Fri, 4 Sep 2015 17:44:10 +0000 (13:44 -0400)]
move error_estimator code into templates.h
Timo Heister [Sat, 5 Sep 2015 15:36:25 +0000 (11:36 -0400)]
split compilation units
Timo Heister [Wed, 2 Sep 2015 22:39:37 +0000 (18:39 -0400)]
Implement .inst splitting in make_instantiations
Wolfgang Bangerth [Sat, 5 Sep 2015 12:49:13 +0000 (07:49 -0500)]
Merge pull request #1377 from maieneuro/analytical_hessian_for_fe_poly_tensor
Analytical hessian for fe poly tensor
Maien Hamed [Thu, 13 Aug 2015 10:02:29 +0000 (12:02 +0200)]
changed shape hessians in FE_PolyTensor to use analytical instead of numerical differentiation
Wolfgang Bangerth [Fri, 4 Sep 2015 19:38:54 +0000 (14:38 -0500)]
Merge pull request #1520 from asartori86/prm_preprocess
Allow to use "\" in the file of parameters
alberto sartori [Thu, 3 Sep 2015 11:09:21 +0000 (13:09 +0200)]
allows continuation lines in ParameterHandler
Matthias Maier [Fri, 4 Sep 2015 03:21:05 +0000 (22:21 -0500)]
Merge pull request #1524 from Rombur/cxx11_thread_bug
Workaround a bug with Ubuntu 14.04
Wolfgang Bangerth [Fri, 4 Sep 2015 01:48:44 +0000 (20:48 -0500)]
Merge pull request #1525 from tjhei/sparsity_pattern_docs
update SparsityPattern documentation
Timo Heister [Fri, 4 Sep 2015 01:32:31 +0000 (21:32 -0400)]
update documentation
Bruno Turcksin [Thu, 3 Sep 2015 23:33:32 +0000 (18:33 -0500)]
Workaround a bug with Ubuntu 14.04
Wolfgang Bangerth [Thu, 3 Sep 2015 14:18:51 +0000 (09:18 -0500)]
Merge pull request #1521 from kronbichler/master
Attempt to fix non-C++11 code path
Martin Kronbichler [Thu, 3 Sep 2015 13:49:31 +0000 (15:49 +0200)]
Attempt to fix non-C++11 code path
Wolfgang Bangerth [Thu, 3 Sep 2015 11:25:02 +0000 (06:25 -0500)]
Merge pull request #1518 from tamiko/fix_openmp_simd_configuration
CMake: Fix openmp-simd configuration
Matthias Maier [Thu, 3 Sep 2015 05:10:18 +0000 (00:10 -0500)]
add news/changes.h entry
Matthias Maier [Thu, 3 Sep 2015 05:05:26 +0000 (00:05 -0500)]
CMake: Fix openmp-simd configuration
This enables openmp simd for Clang version 3.6 or newer (and the
corresponding version for XCode). Further, it unconditionally disables
the openmp check for any older clang version - checking for openmp support
with the flag only is unreliable
Wolfgang Bangerth [Wed, 2 Sep 2015 21:54:59 +0000 (16:54 -0500)]
Merge pull request #1424 from davydden/fe_nothing_dominate
let make_hp_hanging_node_constraints() search for a common space in FECollection
Denis Davydov [Wed, 2 Sep 2015 20:58:54 +0000 (22:58 +0200)]
fixed compiler warning in dof_tools_constraints
Denis Davydov [Wed, 2 Sep 2015 20:51:22 +0000 (22:51 +0200)]
test make_hanging_node_constraints() with hp-ref and neither_element_dominates
Denis Davydov [Wed, 2 Sep 2015 20:47:09 +0000 (22:47 +0200)]
simplified Grid generation in hp_constraints_neither_dominate_01
Also fixed indentation.
Denis Davydov [Wed, 2 Sep 2015 20:44:04 +0000 (22:44 +0200)]
extend make_hp_hanging_node_constraints to hp-ref with neither_element_dominates
Remove no more used get_most_dominating_subface_fe_index().
Added an item to changes.h.
Denis Davydov [Wed, 2 Sep 2015 11:33:29 +0000 (13:33 +0200)]
fix find_least_face_dominating_fe() for self-domination
In the beginning the Universe was created. This has made a lot of
people very angry and has been widely regarded as a bad move.
Douglas Adams
Typeset in
Trocchi and Trocchi Bold Sans Serif.