]>
https://gitweb.dealii.org/ - dealii.git/log
Wolfgang Bangerth [Mon, 11 Jan 2016 00:54:40 +0000 (18:54 -0600)]
Mark all functions that are always_inline also as inline.
Wolfgang Bangerth [Thu, 7 Jan 2016 22:32:28 +0000 (16:32 -0600)]
Add a changelog entry.
Wolfgang Bangerth [Thu, 7 Jan 2016 22:30:53 +0000 (16:30 -0600)]
Use a portable syntax to always inline functions.
Wolfgang Bangerth [Thu, 7 Jan 2016 22:27:32 +0000 (16:27 -0600)]
Check for the availability of the always_inline attribute.
If not available, simply set DEAL_II_ALWAYS_INLINE to the empty string.
Matthias Maier [Thu, 7 Jan 2016 22:14:39 +0000 (16:14 -0600)]
Always inline some tensor functions.
This drastically reduces the time in assembly in ASPECT:
DEBUG:
| Assemble Stokes system | 8 | 38.3s | 27% |
| Assemble temperature system | 8 | 51.4s | 36% |
| Build Stokes preconditioner | 3 | 13.8s | 9.7% |
| Build temperature preconditioner| 8 | 0.192s | 0.13% |
| Solve Stokes system | 8 | 12.3s | 8.6% |
DEBUG WITH PATCH:
| Assemble Stokes system | 8 | 30.8s | 25% |
| Assemble temperature system | 8 | 41.7s | 34% |
| Build Stokes preconditioner | 3 | 12.1s | 9.9% |
| Build temperature preconditioner| 8 | 0.192s | 0.16% |
| Solve Stokes system | 8 | 12.2s | 10% |
(Run with 16 processors, using the testcase discussed in https://github.com/geodynamics/aspect/pull/654 .)
There are no significant differences in release mode with this patch,
as probably expected.
Martin Kronbichler [Mon, 11 Jan 2016 07:07:09 +0000 (08:07 +0100)]
Merge pull request #2041 from bangerth/update-step-17-2
Rewrite the step-17 in-code documentation.
Martin Kronbichler [Mon, 11 Jan 2016 07:06:09 +0000 (08:06 +0100)]
Merge pull request #2050 from jppelteret/fix_docs_schur_complement
Fixed formula typesetting in schur_complement linear operator doc.
Jean-Paul Pelteret [Mon, 11 Jan 2016 06:46:47 +0000 (07:46 +0100)]
Fixed formula typesetting in schur_complement linear operator doc.
Fixes #2049 .
Wolfgang Bangerth [Mon, 11 Jan 2016 02:30:47 +0000 (20:30 -0600)]
Merge pull request #2048 from kronbichler/master
Fix AlignedVector for classes that do not implement copy assignment
Wolfgang Bangerth [Mon, 11 Jan 2016 01:00:25 +0000 (19:00 -0600)]
Address comments in the second round of reviews.
Martin Kronbichler [Sun, 10 Jan 2016 08:02:00 +0000 (09:02 +0100)]
Update output due to change
Martin Kronbichler [Sat, 9 Jan 2016 19:57:39 +0000 (20:57 +0100)]
Avoid initializing memory twice for non-trivial classes in table constructor.
Martin Kronbichler [Sat, 9 Jan 2016 19:56:54 +0000 (20:56 +0100)]
Fix AlignedVector for classes that do not implement copy assignment operator.
Wolfgang Bangerth [Sat, 9 Jan 2016 02:49:04 +0000 (20:49 -0600)]
Update authorship.
Wolfgang Bangerth [Fri, 8 Jan 2016 23:50:07 +0000 (17:50 -0600)]
Reorder the declaration of member variables.
This allows a significant simplification of the code flow in the
constructor.
Wolfgang Bangerth [Fri, 8 Jan 2016 23:37:00 +0000 (17:37 -0600)]
Address all but one point in the review of the previous commits.
Wolfgang Bangerth [Thu, 7 Jan 2016 20:07:03 +0000 (14:07 -0600)]
More work on the documentation.
Wolfgang Bangerth [Thu, 7 Jan 2016 14:01:16 +0000 (08:01 -0600)]
Go through the documentation of the first part of step-17.
Wolfgang Bangerth [Fri, 8 Jan 2016 23:57:17 +0000 (17:57 -0600)]
Merge pull request #2047 from jppelteret/debug_tensor_1
Minor changes to Tensor class
Jean-Paul Pelteret [Fri, 8 Jan 2016 23:26:28 +0000 (00:26 +0100)]
Tensor::operator<< for dim==0 now accesses values by reference.
This is done instead of making a copy, and is useful when non-trivial
number types (e.g. auto-differentiable) are stored.
Jean-Paul Pelteret [Fri, 8 Jan 2016 22:36:09 +0000 (23:36 +0100)]
Tensor::operator[] should always return by reference.
One of the access operators returned by value, instead of by reference.
Martin Kronbichler [Fri, 8 Jan 2016 22:17:46 +0000 (23:17 +0100)]
Merge pull request #2044 from kronbichler/master
Fix memory leak in AlignedVector
Martin Kronbichler [Fri, 8 Jan 2016 22:16:40 +0000 (23:16 +0100)]
Update documentation
Wolfgang Bangerth [Fri, 8 Jan 2016 17:23:36 +0000 (11:23 -0600)]
Merge pull request #2043 from bangerth/update-step-18
Add some more material to the introduction of step-18.
Wolfgang Bangerth [Fri, 8 Jan 2016 17:22:40 +0000 (11:22 -0600)]
Comment on a couple of things brought up in review.
Martin Kronbichler [Fri, 8 Jan 2016 15:56:09 +0000 (16:56 +0100)]
Mention change
Martin Kronbichler [Fri, 8 Jan 2016 15:51:37 +0000 (16:51 +0100)]
Add another test
Martin Kronbichler [Fri, 8 Jan 2016 15:39:41 +0000 (16:39 +0100)]
Fix memory leak in AlignedVector
Martin Kronbichler [Fri, 8 Jan 2016 07:31:44 +0000 (08:31 +0100)]
Merge pull request #1956 from jppelteret/schur_operator
Schur operator
Wolfgang Bangerth [Thu, 7 Jan 2016 23:10:55 +0000 (17:10 -0600)]
Minor stylistic changes.
Wolfgang Bangerth [Thu, 7 Jan 2016 23:10:02 +0000 (17:10 -0600)]
Add some more material to the introduction of step-18.
Jean-Paul Pelteret [Mon, 7 Dec 2015 08:01:21 +0000 (09:01 +0100)]
Implementation of Schur complement linear operator and auxillary functions
A new linear operator representing the Schur complement,
namely schur_complement(), has been implemented. Some auxiliary functions
that are often used in conjunction with the Schur complement
(condense_schur_rhs() and postprocess_schur_solution()) are also provided
as a PackagedOperation. Addresses #1439 .
Martin Kronbichler [Thu, 7 Jan 2016 17:31:41 +0000 (18:31 +0100)]
Merge pull request #2040 from bangerth/fix-changelog
Add a missing name to an entry in changes.h.
Wolfgang Bangerth [Thu, 7 Jan 2016 17:28:29 +0000 (11:28 -0600)]
Add a missing name to an entry in changes.h.
Timo Heister [Thu, 7 Jan 2016 15:22:56 +0000 (10:22 -0500)]
Merge pull request #2017 from kronbichler/master
Parallel adaptive multigrid for parallel::distributed::Vector
Martin Kronbichler [Thu, 7 Jan 2016 12:28:16 +0000 (13:28 +0100)]
Add instantiations
Martin Kronbichler [Wed, 6 Jan 2016 20:18:35 +0000 (21:18 +0100)]
Fix 64 bit bug
Martin Kronbichler [Wed, 6 Jan 2016 11:04:56 +0000 (12:04 +0100)]
Fix typo
Martin Kronbichler [Wed, 6 Jan 2016 11:00:09 +0000 (12:00 +0100)]
Implement optimized copy for generic vectors
Martin Kronbichler [Tue, 5 Jan 2016 17:44:41 +0000 (18:44 +0100)]
Augment documentation
Martin Kronbichler [Thu, 24 Dec 2015 21:51:21 +0000 (22:51 +0100)]
Mention change
Martin Kronbichler [Thu, 24 Dec 2015 21:46:36 +0000 (22:46 +0100)]
Test for parallel adaptive multigrid with matrix-free.
Martin Kronbichler [Wed, 23 Dec 2015 22:08:59 +0000 (23:08 +0100)]
Implement parallel MGTransferPrebuilt for parallel::distributed::Vector
This commit includes a refactoring of MGTransferPrebuilt into a base class
MGLevelGlobalTransfer that contains the copy operation between the multigrid
hierarchy and the global degrees of freedom, including a specialization for
parallel::distributed::Vector that takes care of the particular behavior in
this vector type. It also allows us to use optimized access in terms of local
MPI indices.
Matthias Maier [Wed, 6 Jan 2016 17:38:18 +0000 (11:38 -0600)]
Merge pull request #2037 from bangerth/hook-in-code-gallery-3
Hook in code gallery 3
Wolfgang Bangerth [Wed, 6 Jan 2016 17:07:12 +0000 (11:07 -0600)]
Build and track a page that lists all code gallery programs.
Wolfgang Bangerth [Wed, 6 Jan 2016 16:00:56 +0000 (10:00 -0600)]
Deal with ```...``` markdown commands.
doxygen doesn't appear to understand them, so replace them by the
doxygen equivalent.
Wolfgang Bangerth [Wed, 6 Jan 2016 15:19:56 +0000 (09:19 -0600)]
Sort markdown files first.
This allows bringing any readme.md or similar files to the top.
Wolfgang Bangerth [Wed, 6 Jan 2016 14:31:30 +0000 (08:31 -0600)]
Shorten text.
Wolfgang Bangerth [Wed, 6 Jan 2016 14:29:58 +0000 (08:29 -0600)]
Skip header lines only for tutorial programs.
This leaves copyright notices and author information visible for
code gallery programs.
Wolfgang Bangerth [Wed, 6 Jan 2016 14:25:46 +0000 (08:25 -0600)]
More accurately describe dependencies.
Wolfgang Bangerth [Wed, 6 Jan 2016 13:40:03 +0000 (07:40 -0600)]
Copy annotated versions of .md and .cc files into the page.
Also provide links to them at the top.
Wolfgang Bangerth [Wed, 6 Jan 2016 13:17:59 +0000 (07:17 -0600)]
Do not create the heading in program2doxygen.
Instead create it where we call it.
Matthias Maier [Wed, 6 Jan 2016 12:54:55 +0000 (06:54 -0600)]
Merge pull request #2034 from jppelteret/adjust_lin_ops_tests
Minor adjustments to some linear operator tests
Wolfgang Bangerth [Fri, 1 Jan 2016 22:12:20 +0000 (16:12 -0600)]
Create a gallery of pictures at the top of the page.
Timo Heister [Tue, 5 Jan 2016 21:27:06 +0000 (22:27 +0100)]
Merge pull request #2036 from guidokanschat/arpack_warning
Change the exception type to avoid the compiler warning
Guido Kanschat [Tue, 5 Jan 2016 17:51:43 +0000 (18:51 +0100)]
Change the exception type to avoid the compiler warning
Matthias Maier [Tue, 5 Jan 2016 10:17:58 +0000 (04:17 -0600)]
Merge pull request #2035 from davydden/slepc_rework_fix
add missing DEAL_II_WITH_SLEPC in petsc_solver.h
Denis Davydov [Tue, 5 Jan 2016 06:33:43 +0000 (07:33 +0100)]
add missing DEAL_II_WITH_SLEPC in petsc_solver.h
Wolfgang Bangerth [Thu, 31 Dec 2015 22:50:47 +0000 (16:50 -0600)]
Properly track dependencies, and install copies of the code gallery.
Wolfgang Bangerth [Thu, 31 Dec 2015 22:50:19 +0000 (16:50 -0600)]
Extend the stub to generate a page for each code gallery.
Wolfgang Bangerth [Thu, 31 Dec 2015 04:21:08 +0000 (22:21 -0600)]
Provide the basic building blocks to hook in the code gallery.
In particular, track dependencies and set up a stub for generating doxygen input.
Jean-Paul Pelteret [Mon, 4 Jan 2016 15:16:47 +0000 (16:16 +0100)]
Renamed test to help "categorise" it correctly.
Jean-Paul Pelteret [Mon, 4 Jan 2016 17:10:05 +0000 (18:10 +0100)]
Fixed some tests.
linear_operator_08 failed when deal.II is not built with UMFPACK.
There's now a switch for that.
Martin Kronbichler [Mon, 4 Jan 2016 17:12:58 +0000 (18:12 +0100)]
Merge pull request #1804 from davydden/slepc_rework
Allow usage of PETSc solvers/preconditioners in SLEPc spectral transformation
David Wells [Mon, 4 Jan 2016 14:09:37 +0000 (09:09 -0500)]
Merge pull request #2028 from bangerth/hook-in-code-gallery-2
Generate the legend for the tutorial connection graph automatically.
Wolfgang Bangerth [Fri, 1 Jan 2016 19:30:48 +0000 (13:30 -0600)]
Merge pull request #2031 from kronbichler/cleanup_tests
Only run tests if MPI (and p4est for mg_transfer) is available.
Denis Davydov [Fri, 1 Jan 2016 14:38:57 +0000 (15:38 +0100)]
further initialise()->initialize() and alike
Denis Davydov [Fri, 1 Jan 2016 11:50:19 +0000 (12:50 +0100)]
make sure that inside solve() PETSc preconditioner has a matrix set
Martin Kronbichler [Fri, 1 Jan 2016 09:00:05 +0000 (10:00 +0100)]
Merge pull request #2029 from tamiko/bugfix
Populate Tvmult and Tvmult_add of an inverse_operator correctly
Martin Kronbichler [Fri, 1 Jan 2016 08:53:55 +0000 (09:53 +0100)]
Only run tests if MPI (and p4est for mg_transfer) is available.
Timo Heister [Thu, 31 Dec 2015 22:46:49 +0000 (23:46 +0100)]
Merge pull request #2030 from davydden/step54_input
step-54: move input vtk to a separate folder to avoid its removal by …
Denis Davydov [Thu, 31 Dec 2015 05:53:56 +0000 (06:53 +0100)]
step-54: move input vtk to a separate folder to avoid its removal by make distclean
Matthias Maier [Wed, 30 Dec 2015 17:33:30 +0000 (18:33 +0100)]
break indenting to make astyle happy
Matthias Maier [Wed, 30 Dec 2015 17:30:27 +0000 (18:30 +0100)]
add a changes.h entry
Matthias Maier [Wed, 30 Dec 2015 17:26:48 +0000 (18:26 +0100)]
Add a regression test
Martin Kronbichler [Wed, 30 Dec 2015 17:20:51 +0000 (18:20 +0100)]
Merge pull request #2001 from tjhei/mg_remove_deprecated
remove deprecated stuff in MGConstrainedDoFs
Matthias Maier [Wed, 30 Dec 2015 17:08:37 +0000 (18:08 +0100)]
Bugfix: Populate Tvmult and Tvmult_add of an inverse_operator correctly
Fix a typo in linear_operator.h: The transpose operation function-object
Tvmult was accidentally overwritten by Tvmult_add (and the latter left
uninitialized).
Wolfgang Bangerth [Wed, 30 Dec 2015 16:26:36 +0000 (10:26 -0600)]
Merge pull request #2026 from kronbichler/vector_parallel_layout
Vectorize one loop
Martin Kronbichler [Wed, 30 Dec 2015 13:45:14 +0000 (14:45 +0100)]
Add instantiations
Wolfgang Bangerth [Tue, 29 Dec 2015 23:50:36 +0000 (17:50 -0600)]
Also generate the legend for the tutorial connection graph automatically.
This allows us to show the actual symbols for each kind, rather than having to
describe them in words.
Wolfgang Bangerth [Tue, 29 Dec 2015 21:37:42 +0000 (15:37 -0600)]
De-emphasize connections to the code gallery.
Wolfgang Bangerth [Tue, 29 Dec 2015 21:23:29 +0000 (15:23 -0600)]
Merge pull request #2025 from kronbichler/minor_multigrid_cleanup
Minor multigrid cleanup
Wolfgang Bangerth [Tue, 29 Dec 2015 21:22:02 +0000 (15:22 -0600)]
Merge pull request #2027 from kronbichler/change_minimum_parallel_grainsize
Change minimum parallel grainsize of vector to power of 2
Martin Kronbichler [Tue, 29 Dec 2015 20:16:12 +0000 (21:16 +0100)]
Change minimum parallel grainsize of vector to power of 2
Martin Kronbichler [Tue, 29 Dec 2015 20:12:40 +0000 (21:12 +0100)]
Vectorize one loop
Martin Kronbichler [Tue, 29 Dec 2015 18:59:15 +0000 (19:59 +0100)]
Correct header name
Martin Kronbichler [Tue, 29 Dec 2015 18:58:50 +0000 (19:58 +0100)]
Re-enable deallog output in multigrid tutorials.
Wolfgang Bangerth [Tue, 29 Dec 2015 16:37:24 +0000 (10:37 -0600)]
Merge pull request #2024 from drwells/update-relaxationtype-example
Update the documentation of RelaxationType.
David Wells [Tue, 29 Dec 2015 15:55:11 +0000 (10:55 -0500)]
Update the documentation of RelaxationType.
This commit consistently labels classes that satisfy this concept and
clarifies the description of the concept.
Wolfgang Bangerth [Tue, 29 Dec 2015 15:36:19 +0000 (09:36 -0600)]
Merge pull request #2022 from bangerth/fix-markup
Fix markup of references.
Wolfgang Bangerth [Mon, 28 Dec 2015 22:40:53 +0000 (16:40 -0600)]
Fix markup of references.
doxygen interpreted our bibtex-references as C++ code which led
to funny HTML markup. Fortunately, there is a way to tell
doxygen to interpret the code block as just some piece of text
it will not understand regardless of how hard it tries.
Denis Davydov [Tue, 29 Dec 2015 09:44:19 +0000 (10:44 +0100)]
rename initialise() to initialize() and remove an empty line
Denis Davydov [Thu, 17 Dec 2015 15:58:36 +0000 (16:58 +0100)]
slepc: add an entry in changes.h
Denis Davydov [Thu, 17 Dec 2015 15:50:54 +0000 (16:50 +0100)]
SLEPcWrappers: adjust documentation
Denis Davydov [Tue, 15 Dec 2015 10:33:14 +0000 (11:33 +0100)]
extend SLEPc unit test to check different preconditioners in KrylovSchur solver
Denis Davydov [Tue, 15 Dec 2015 10:32:18 +0000 (11:32 +0100)]
allow creation of PETSc preconditioners without a matrix (Jacobi/BoomerAMG)
Denis Davydov [Tue, 15 Dec 2015 10:20:11 +0000 (11:20 +0100)]
impliment missing set_initial_vector() for SLEPc solvers
Denis Davydov [Sat, 12 Dec 2015 20:08:37 +0000 (21:08 +0100)]
SLEPc unit test which uses PETSc preconditioner and solver inside SLEPc
Denis Davydov [Sat, 12 Dec 2015 20:03:03 +0000 (21:03 +0100)]
rework SLEPc solver classes
(i) initialise underlying SLEPc objects in constructors. This allows to...
(ii) apply all the solver settings immediately without using a cache for parameters
Denis Davydov [Sat, 12 Dec 2015 20:00:08 +0000 (21:00 +0100)]
rework SLEPc spectral transformation
(i) create unerlying SLEPc object in constructor; require MPI_Comm.
(ii) add a function to set PETSc solver
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.