dune-gdt issues
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues
2022-10-20T14:35:03+02:00
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/40
ensure working examples
2022-10-20T14:35:03+02:00
Dr. Felix Tobias Schindler
ensure working examples
* [ ] ensure `adaptive_elliptic_swipdg.cc` works
* [ ] ensure `adaptive_interpolation.cc` works
* [x] ensure `elliptic_lrbms_fom.cc` works
* [ ] ensure `mpi_2019_02_talk_on_hyperbolic_equations.cc` works
* [ ] ensure `stationary-heat-equ...
* [ ] ensure `adaptive_elliptic_swipdg.cc` works
* [ ] ensure `adaptive_interpolation.cc` works
* [x] ensure `elliptic_lrbms_fom.cc` works
* [ ] ensure `mpi_2019_02_talk_on_hyperbolic_equations.cc` works
* [ ] ensure `stationary-heat-equation.cc` works
* [ ] ensure `gamm_2019_talk_on_conservative*.py` works
* [ ] ensure examples are compiled and executed in the CI
2022.1
Dr. Felix Tobias Schindler
Dr. Felix Tobias Schindler
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/39
check linear solvers in newton
2022-05-09T17:11:42+02:00
Dr. Felix Tobias Schindler
check linear solvers in newton
Make sure we currently do what we want to do. In particular
* balance linear solver accuracy vs. newton accuracy (currently: magic number 0.1)
* trying multiple linear solvers might be unexpected
Make sure we currently do what we want to do. In particular
* balance linear solver accuracy vs. newton accuracy (currently: magic number 0.1)
* trying multiple linear solvers might be unexpected
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/38
Fix old style cast in bindings and retry sanitizer test
2022-05-02T16:18:57+02:00
Dr. Tim Keil
Fix old style cast in bindings and retry sanitizer test
Currently, there is an issue with the sanitizer test, for instance in `dune/gdt/operators/bilinear-form.hh` which is why [it is added to the blacklist](https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/config-opts/-/blob/d735...
Currently, there is an issue with the sanitizer test, for instance in `dune/gdt/operators/bilinear-form.hh` which is why [it is added to the blacklist](https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/config-opts/-/blob/d735cd576f5e004f359038d0c9af8c3eec373314/.clang-sanitizer-blacklist#L35). Locally I have seen an "use of old-style cast" warning which @r_milk01 claimed to be related to the test fail. Once this is fixed, we should retry to remove it from the blacklist.
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/37
CMake Support for X-sanitizers
2022-04-12T14:44:35+02:00
René Fritze
CMake Support for X-sanitizers
Currently the load/start logic lives partially in `travis` opts and the test launch scripts for CI.
That makes it hard to roll out locally. Instead move all the settings to cmake, and expose only on/off toggles to the user.
Currently the load/start logic lives partially in `travis` opts and the test launch scripts for CI.
That makes it hard to roll out locally. Instead move all the settings to cmake, and expose only on/off toggles to the user.
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/36
keep copy of grid view in operators and spaces
2022-04-05T13:43:20+02:00
Dr. Felix Tobias Schindler
keep copy of grid view in operators and spaces
As established [here](https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-xt/-/merge_requests/52#note_2170603), we can copy grid views even in the adaptive setting without harm. We should do so in all places, in particular...
As established [here](https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-xt/-/merge_requests/52#note_2170603), we can copy grid views even in the adaptive setting without harm. We should do so in all places, in particular:
- operators
- spaces
2022.2
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/35
regression: newton requires more iterations after refactoring
2022-04-01T13:06:46+02:00
Dr. Felix Tobias Schindler
regression: newton requires more iterations after refactoring
After the refactoring in !10, the newton in `test_linear_transport__1d__implicit__fv_LinearTransport1dImplicitFvTest` required more iterations (248) than before (<= 100).
This hints at potential bugs, not necessarily in the newton but in...
After the refactoring in !10, the newton in `test_linear_transport__1d__implicit__fv_LinearTransport1dImplicitFvTest` required more iterations (248) than before (<= 100).
This hints at potential bugs, not necessarily in the newton but in some of the operators.
We should investigate this by comparing to a previously working stage, e.g. [fabc279a on github.com/dune-community/dune-gdt](https://github.com/dune-community/dune-gdt/commit/fabc279a79e7362181701866ce26133ec40a05e0) with passing CI.
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/34
computing jacobi matrix takes more time in later newton iterations
2022-04-01T13:05:03+02:00
Dr. Felix Tobias Schindler
computing jacobi matrix takes more time in later newton iterations
Where I would have thought that the complexity should stay constant over all newton steps. Investigate!
Where I would have thought that the complexity should stay constant over all newton steps. Investigate!
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/33
make logging more consistent
2022-04-01T11:41:35+02:00
Dr. Felix Tobias Schindler
make logging more consistent
E.g., in `newton_solve` the logger is initialized with default state, rendering it useless. Here, and in many other places, we need to think about how to derive the logger state more cleverly...
E.g., in `newton_solve` the logger is initialized with default state, rendering it useless. Here, and in many other places, we need to think about how to derive the logger state more cleverly...
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/32
add apply_adjoint to ForwardOperatorInterface
2022-03-24T16:47:48+01:00
Dr. Felix Tobias Schindler
add apply_adjoint to ForwardOperatorInterface
as required in `test_stokes`, for instance
as required in `test_stokes`, for instance
next release
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/31
refactor ordering of dim template arguments
2022-03-24T16:20:54+01:00
Dr. Felix Tobias Schindler
refactor ordering of dim template arguments
Since we rarely have matrix-valued functions, it makes more sense to have
```c++
template <class AssemblyGridView,
size_t s_r = 1,
size_t r_r = s_r,
size_t s_rC = 1,
size_t r_rC = s_rC,
c...
Since we rarely have matrix-valued functions, it makes more sense to have
```c++
template <class AssemblyGridView,
size_t s_r = 1,
size_t r_r = s_r,
size_t s_rC = 1,
size_t r_rC = s_rC,
class F = double,
class SGV = AssemblyGridView,
class RGV = AssemblyGridView>
class BilinearForm
```
as template arguments than (as we currently have)
```c++
template <class AssemblyGridView,
size_t s_r = 1,
size_t s_rC = 1,
size_t r_r = s_r,
size_t r_rC = s_rC,
class F = double,
class SGV = AssemblyGridView,
class RGV = AssemblyGridView>
class BilinearForm
```
This applies to all places with similar arguments.
next release
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/30
merge dune-gdt-super into dune-gdt
2022-03-15T12:58:29+01:00
Dr. Felix Tobias Schindler
merge dune-gdt-super into dune-gdt
The idea is to store the information of required dependencies directly within dune-gdt.
Possibly achieved by git submodules residing in some folder.
This
- should make dune-gdt-super obsolete,
- simplify CI and
- gives you everything wit...
The idea is to store the information of required dependencies directly within dune-gdt.
Possibly achieved by git submodules residing in some folder.
This
- should make dune-gdt-super obsolete,
- simplify CI and
- gives you everything with all dependencies after cloning one repo
next release
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/29
merge dune-xt into dune-gdt
2022-04-01T15:27:15+02:00
Dr. Felix Tobias Schindler
merge dune-xt into dune-gdt
As discussed offline with @r_milk01, we consider to merge dune-xt into dune-gdt.
Pros:
* less projects, less repos, less super repos
* (CI) dependency simplification (no need to commit dune-xt in gdt-super, wait for base images update, ...
As discussed offline with @r_milk01, we consider to merge dune-xt into dune-gdt.
Pros:
* less projects, less repos, less super repos
* (CI) dependency simplification (no need to commit dune-xt in gdt-super, wait for base images update, commit dune-gdt, ...)
* you clone one repo and get everything
Cons:
* wheels would get too big, need to be split into several ones
* no standalone dune-xt without dune-gdt:
- add a `bindings_dune_xt` only target if someone wants only those bindings?
Please voice your opinions, @all!
next release
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/28
refine CI stages
2022-10-19T15:54:57+02:00
Dr. Felix Tobias Schindler
refine CI stages
Currently in !10, we have `images` -> `cpp` -> `headercheck` -> `python`. I would propose something like
```
images -> cpp-make-all
| |
| +--------> cpp-make-test -> cpp-run-tests
| |
...
Currently in !10, we have `images` -> `cpp` -> `headercheck` -> `python`. I would propose something like
```
images -> cpp-make-all
| |
| +--------> cpp-make-test -> cpp-run-tests
| |
| +--------> cpp-headercheck
| |
| +--------> python-make-bindings -> python-run-tests
|
+--------------------> wheels-build -> wheels-test
```
to allow for as much in parallel as possible. I am putting `wheels` independently of `bindings` here, since the first is in an `manylinux` image, while the latter is with all our usual c++ compilers. One could even think about branching `wheels-build` off directly after `images` already, but I guess thats a bit much.
What do you think, @r_milk01?
next release
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/27
Refactor functionals according to operators
2022-02-24T21:42:16+01:00
Dr. Felix Tobias Schindler
Refactor functionals according to operators
After merging !10
After merging !10
2022.2
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/26
make use of pytest-cpp
2022-02-24T16:36:11+01:00
Dr. Felix Tobias Schindler
make use of pytest-cpp
https://github.com/pytest-dev/pytest-cpp
https://github.com/pytest-dev/pytest-cpp
next release
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/25
Release 2023.1
2022-12-27T14:47:44+01:00
Dr. Felix Tobias Schindler
Release 2023.1
Collection of what has to be done before the next release:
- [ ] merge the big refactoring branch !10, therefore
* [ ] consider #23
- [ ] apply clang tidy #24
- [ ] rename and clean up #14
- [ ] ~~fix lifetime issues #3 and #13~~
N...
Collection of what has to be done before the next release:
- [ ] merge the big refactoring branch !10, therefore
* [ ] consider #23
- [ ] apply clang tidy #24
- [ ] rename and clean up #14
- [ ] ~~fix lifetime issues #3 and #13~~
Not sure if having [the milestone](https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/milestones/4) is enough, or if we want this issue to tick things off.
2022.1
Dr. Felix Tobias Schindler
Dr. Felix Tobias Schindler
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/24
Apply clang tidy
2022-02-24T16:22:28+01:00
Tobias Leibner
Apply clang tidy
After !10 is merged, we should apply clang-tidy as I did in https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-xt/-/merge_requests/92 and https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-xt/-/merge_reque...
After !10 is merged, we should apply clang-tidy as I did in https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-xt/-/merge_requests/92 and https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-xt/-/merge_requests/94. The infrastructure is already in place (assuming we are using the recent `dune-xt` master branch), we only have to update `.vcsetup` and can then just run `ninja -j8 fix_tidy_parallel` in the build directory (or maybe just `ninja fix_tidy` if the parallel fixes cause lots of errors). Maybe we should apply the `-modernize-use-override`, `modernize-concat-nested-namespaces` and `readability-else-after-return` checks first in separate tidy runs (by replacing the `Checks:` list in `.vcsetup/.clang-tidy` by the respective check) and commit the results after each run as these checks introduce a lot of changes.
2022.1
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/23
Roadmap to common main and paper
2022-02-24T16:23:13+01:00
Dr. Felix Tobias Schindler
Roadmap to common main and paper
Requirements for merging to main:
- [ ] address [issues raised in the original PR](https://github.com/dune-community/dune-gdt/pull/198)
- [ ] make sure C++ tests pass in !5
- (for Tim & Felix:) as soon as this works, Felix starts on L...
Requirements for merging to main:
- [ ] address [issues raised in the original PR](https://github.com/dune-community/dune-gdt/pull/198)
- [ ] make sure C++ tests pass in !5
- (for Tim & Felix:) as soon as this works, Felix starts on LRBMS C++ example
- [ ] make sure Python bindings compile in !5
- [ ] make sure Python Tutorials work with !5
- (for Tim & Felix:) as soon as this works, Tim starts on LRBMS Python tutorial
- [ ] merge !5 into main
Afterwards (new issues to be created yet), roadmap to paper:
- [ ] ensure wheels can be built and are importable
- [ ] update C++ functionals to match operator refactoring
- [ ] update Python bindings to support new syntax after refatoring
- [ ] add Python tests
- [ ] pyMOR LRBMS discretizer
Dr. Felix Tobias Schindler
Dr. Felix Tobias Schindler
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/16
Allow for larger stencil in finite difference jacobians of element operators
2022-02-24T16:26:56+01:00
Tobias Leibner
Allow for larger stencil in finite difference jacobians of element operators
Copied over from [Github](https://github.com/dune-community/dune-gdt/issues/154).
While being only element operators, they might depend on values of the source from neighbors (by internally doing a walk over the intersections). Thus, wh...
Copied over from [Github](https://github.com/dune-community/dune-gdt/issues/154).
While being only element operators, they might depend on values of the source from neighbors (by internally doing a walk over the intersections). Thus, when appending those to the matrix jacobian we should also specify the stencil used in the FD approximation...
This is required for higher order implicit DG to work for hyperbolic poblems...
Related to #15, should be implemented alongside.
Future
Dr. Felix Tobias Schindler
Dr. Felix Tobias Schindler
https://zivgitlab.uni-muenster.de/ag-ohlberger/dune-community/dune-gdt/-/issues/15
Derive local operators from local functionals
2022-02-24T16:27:11+01:00
Tobias Leibner
Derive local operators from local functionals
Copied over from [Github](https://github.com/dune-community/dune-gdt/issues/174).
Once [#170](https://github.com/dune-community/dune-gdt/pull/170) is merged, the only difference between local operators and functionals is `with_source`, ...
Copied over from [Github](https://github.com/dune-community/dune-gdt/issues/174).
Once [#170](https://github.com/dune-community/dune-gdt/pull/170) is merged, the only difference between local operators and functionals is `with_source`, so derive one from the other. Also check here whether we really need `with_source` (the jacobian FD assemblers seem to work fine without).
Future