Closed amartinhuertas closed 3 years ago
Here are the outstanding tasks that I'm aware of (there may be others)
shallow_water_rosenbrock_time_stepper
for the mean depth, H_0
as used int eh approximate Jacobian (2998.1154702758267 for the Williamson2 test case, 10000.0 or more correctly the mean h globally integrated for the Galewsky test case) [u,h]
coupled system with (B
) and without (A
) the mass matricesdu, dh
in shallow_water_rosenbrock_time_step!
MultiFieldFESpace
implementationMerging #18 (1b8cee6) into master (e1439c5) will increase coverage by
6.32%
. The diff coverage is91.54%
.
@@ Coverage Diff @@
## master #18 +/- ##
==========================================
+ Coverage 62.72% 69.05% +6.32%
==========================================
Files 11 13 +2
Lines 440 559 +119
==========================================
+ Hits 276 386 +110
- Misses 164 173 +9
Impacted Files | Coverage Δ | |
---|---|---|
src/DiagnosticTools.jl | 58.33% <0.00%> (ø) |
|
src/Helpers.jl | 90.90% <90.90%> (ø) |
|
src/ShallowWaterRosenbrock.jl | 91.66% <91.66%> (ø) |
|
src/ShallowWaterExplicit.jl | 97.19% <100.00%> (+1.43%) |
:arrow_up: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update e1439c5...1b8cee6. Read the comment docs.
Hey @cbladwell , you've done some nice work here, looks like we're almost finished, well done! I'm just setting up to run your code now, but here are a few (minor) and one major comment in the interim:
Amat*get_free_dof_values(duh1)
may create a new vector (not sure), much be safer to create a work vector at the beginning of the simulation, and pass this and then do as mul(my_work_vec, Amat, get_free_dof_values(duh1))
: https://github.com/gridapapps/GridapGeosciences.jl/blob/shallow_water_rosenbrock/src/ShallowWaterRosenbrock.jl#L84Amat
as in https://github.com/gridapapps/GridapGeosciences.jl/blob/shallow_water_rosenbrock/src/ShallowWaterRosenbrock.jl#L179 needs to be multiplied by -\lambda*dt
before it is applied to the RHS of the solveI have update the multiplication by `Amat.
looks like this has been defined in a couple of places, might want to consolidate... https://github.com/gridapapps/GridapGeosciences.jl/blob/shallow_water_rosenbrock/src/ShallowWaterRosenbrock.jl#L10
Agreed. I haven't moved it yet but this could be an internal function _clone_fe_function
in Operators.jl
Thanks @cbladwell , I just fixed up a bug and made a minor optimisation and pushed the changes...
Galewsky test case results look good (instability & conservation errors). Ready for revie w
@davelee2804 @cbladwell thanks for your work. I am happy with the current version. Once the tests pass we can accept the PR.
Creating PR so that we can start discussing this work as needed ... WIP ...