Remove `normal_direction_ll` for nonconservative terms #2062

patrickersing · 2024-09-04T08:16:32Z

This PR simplifies the nonconservative terms to only depend on a single averaged normal_direction, instead of both normal_direction_average and normal_direction_ll (see #2049).
This should also enable setting boundary conditions for nonconservative terms #1445

github-actions · 2024-09-04T08:16:47Z

Review checklist

This checklist is meant to assist creators of PRs (to let them know what reviewers will typically look for) and reviewers (to guide them in a structured review process). Items do not need to be checked explicitly for a PR to be eligible for merging.

Purpose and scope

The PR has a single goal that is clear from the PR title and/or description.
All code changes represent a single set of modifications that logically belong together.
No more than 500 lines of code are changed or there is no obvious way to split the PR into multiple PRs.

Code quality

The code can be understood easily.
Newly introduced names for variables etc. are self-descriptive and consistent with existing naming conventions.
There are no redundancies that can be removed by simple modularization/refactoring.
There are no leftover debug statements or commented code sections.
The code adheres to our conventions and style guide, and to the Julia guidelines.

Documentation

New functions and types are documented with a docstring or top-level comment.
Relevant publications are referenced in docstrings (see example for formatting).
Inline comments are used to document longer or unusual code sections.
Comments describe intent ("why?") and not just functionality ("what?").
If the PR introduces a significant change or new feature, it is documented in NEWS.md with its PR number.

Testing

The PR passes all tests.
New or modified lines of code are covered by tests.
New or modified tests run in less then 10 seconds.

Performance

There are no type instabilities or memory allocations in performance-critical parts.
If the PR intent is to improve performance, before/after time measurements are posted in the PR.

Verification

The correctness of the code was verified using appropriate tests.
If new equations/methods are added, a convergence test has been run and the results
are posted in the PR.

Created with ❤️ by the Trixi.jl community.

patrickersing · 2024-09-05T12:57:31Z

As expected changing the normal directions caused slightly different results and with that some failing mhd tests.
The p4est and t8code test results are almost the same and only the psi value fails testing with differences in L2 between 1e-8 and 1e-10.
For UnstructuredMesh and StructuredMesh there are more failing tests, but the differences in the L2-Error remain relatively small between 1e-5 and 1e-8. The reason is probably that we use a very coarse mesh for the testing, with only four elements per dimension, which leads to larger differences in the normal direction.
Entropy is still conserved in all relevant tests.

For validation I additionally ran convergence tests using structured_2d/elixir_mhd_alfen_wave.jl and compared the result with the previous implementation using normal_direction_ll. The results show almost identical convergence rates.

polydeg=3, normal_direction_average:

polydeg=3, normal_direction_ll:

polydeg=4, normal_direction_average:

polydeg=4, normal_direction_ll:

Looking at these results, I think it would be fine to introduce the averaged normal direction and update the test values.
It would be great if someone working with MHD (maybe @amrueda @SimonCan @andrewwinters5000) can have a look at this and let me know if they think this is fine or if more validation is needed.

Another open problem is that the test dgmulti_2d/elixir_mhd_reflective_wall.jl is failing. @jlchan do you have any idea why this test could be failing?

jlchan · 2024-09-05T14:25:40Z

Another open problem is that the test dgmulti_2d/elixir_mhd_reflective_wall.jl is failing. @jlchan do you have any idea why this test could be failing?

Thanks for testing it out in DGMulti too! I'll take a look this afternoon.

Since the test is failing due to domain error (negative density or pressure) I think it's a real failure. My guess is that it's related to the boundary condition imposition

Trixi.jl/src/solvers/dgmulti/dg.jl

Lines 555 to 560 in aaf0166

    
           noncons_flux_at_face_node = boundary_condition(u_face_values[i, f], 
        
                                                          face_normal, 
        
                                                          face_coordinates, 
        
                                                          t, 
        
                                                          nonconservative_flux, 
        
                                                          equations)

since all the other MHD tests still pass.

patrickersing · 2024-09-05T15:08:06Z

Since the test is failing due to domain error (negative density or pressure) I think it's a real failure. My guess is that it's related to the boundary condition imposition

That makes sense. I just looked at the boundary conditions that are used in the elixir and noticed that BoundaryConditionDoNothing needs to be adapted. Currently this always calls the flux function (even if we call it with the nonconservative term) so this definitely imposes a wrong boundary condition.

I will try to replace this with

@inline function (::BoundaryConditionDoNothing)(u_inner,
                                                outward_direction::AbstractVector,
                                                x, t, surface_flux, equations)
    #return flux(u_inner, outward_direction, equations)
    return surface_flux(u_inner, u_inner, outward_direction, equations)
end```

jlchan · 2024-09-06T00:05:59Z

@patrickersing looks like your new BoundaryConditionDoNothing fixed the issue; the failing tests for unstructured_dgmulti now just seem to be due to floating point differences.

DanielDoehring · 2024-09-10T14:41:57Z

Once this is ready to go we can schedule the next breaking release 0.9, cf. #1997

andrewwinters5000

LGTM! I just left one question for Andrés. It would be good to get this PR ship-shape and ready for the next breaking release. Can you have a look @amrueda or @SimonCan as you both more actively work with the MHD system.

src/equations/ideal_glm_mhd_2d.jl

amrueda

It looks good. I leave here only a couple of comments.

amrueda · 2024-09-20T05:52:29Z

src/basic_types.jl

@@ -71,14 +71,14 @@ struct BoundaryConditionDoNothing end
                                                orientation_or_normal_direction,
                                                direction::Integer, x, t, surface_flux,
                                                equations)
-    return flux(u_inner, orientation_or_normal_direction, equations)
+    return surface_flux(u_inner, u_inner, orientation_or_normal_direction, equations)


Why do we need to use surface_flux here? isn't that adding computational complexity for the same output?

This became necessary, since we now also apply the boundary condition for nonconservative terms.
When you call BoundaryConditionDoNothing with nonconservative terms, the old implementation still evaluates the inner flux instead of the inner nonconservative term. Therefore, we erroneously impose the conservative flux twice.
For conservative fluxes this increases the computational complexity, but the modification is necessary to return a consistent boundary condition for both conservative and nonconservative terms.

amrueda · 2024-09-20T05:52:38Z

src/basic_types.jl

 end

 # This version can be called by hyperbolic solvers on unstructured, curved meshes
 @inline function (::BoundaryConditionDoNothing)(u_inner,
                                                outward_direction::AbstractVector,
                                                x, t, surface_flux, equations)
-    return flux(u_inner, outward_direction, equations)
+    return surface_flux(u_inner, u_inner, outward_direction, equations)


Same as above

amrueda · 2024-09-20T05:54:36Z

src/equations/ideal_glm_mhd_2d.jl

-contravariant vector (normal direction) at the current node and the averaged
-one. This is different from numerical fluxes used to discretize conservative
-terms.
-
 ## References


It should be mentioned here that this term is not exactly equivalent to the one proposed by Bohm et al., as the "metrics dealiasing" is done in a different manner.

I added an additional comment about this to the docstring in 95d846a

amrueda · 2024-09-20T05:55:03Z

src/equations/ideal_glm_mhd_3d.jl

-contravariant vector (normal direction) at the current node and the averaged
-one. This is different from numerical fluxes used to discretize conservative
-terms.
-
 ## References


Same as above

ranocha · 2024-09-30T18:35:14Z

A lot of tests fail. Could you please check what's going on?

patrickersing · 2024-10-01T07:50:22Z

A lot of tests fail. Could you please check what's going on?

Most tests are failing, because we get slightly different results with the averaged normal direction. Since there haven't been any objections, I will proceed and update the reference values to fix the tests.

To fix the downstream test for TrixiShallowWater.jl I prepared a corresponding PR trixi-framework/TrixiShallowWater.jl#58, that needs to be merged after this one.

codecov · 2024-10-01T13:27:07Z

Codecov Report

Attention: Patch coverage is 96.29630% with 1 line in your changes missing coverage. Please review.

Project coverage is 96.34%. Comparing base (fa43fb2) to head (be1f9e2).

Files with missing lines	Patch %	Lines
src/solvers/dgmulti/flux_differencing.jl	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2062      +/-   ##
==========================================
+ Coverage   89.86%   96.34%   +6.48%     
==========================================
  Files         470      470              
  Lines       37486    37495       +9     
==========================================
+ Hits        33685    36123    +2438     
+ Misses       3801     1372    -2429

Flag	Coverage Δ
unittests	`96.34% <96.30%> (+6.48%)`	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

ranocha · 2024-10-02T06:12:33Z

@amrueda Could you please review this PR again?

ranocha

Could you please add some NEWS.md?

remove_normal_direction_ll

bf728f7

patrickersing added the breaking label Sep 4, 2024

Merge branch 'main' into remove_normal_direction_ll

aaf0166

patrickersing mentioned this pull request Sep 5, 2024

Remove normal_direction_ll trixi-framework/TrixiShallowWater.jl#58

Draft

patrickersing and others added 3 commits September 5, 2024 17:10

update testset name StructuredMesh3D

10dc9cf

fix BoundaryConditionDoNothing

cc40791

Merge branch 'main' into remove_normal_direction_ll

974b3ba

adjust docstrings

47dab1a

patrickersing changed the title ~~[WIP] Remove normal_direction_ll for nonconservative terms~~ Remove normal_direction_ll for nonconservative terms Sep 10, 2024

Merge branch 'main' into remove_normal_direction_ll

c69719b

patrickersing marked this pull request as ready for review September 12, 2024 09:58

Merge branch 'main' into remove_normal_direction_ll

66c5bd8

andrewwinters5000 previously approved these changes Sep 13, 2024

View reviewed changes

src/equations/ideal_glm_mhd_2d.jl Show resolved Hide resolved

DanielDoehring mentioned this pull request Sep 18, 2024

Release v0.9 #1997

Open

10 tasks

Merge branch 'main' into remove_normal_direction_ll

d25e179

amrueda reviewed Sep 20, 2024

View reviewed changes

add comment to docstring

95d846a

patrickersing dismissed andrewwinters5000’s stale review via 95d846a September 20, 2024 06:49

Merge branch 'main' into remove_normal_direction_ll

4e13e6d

ranocha requested review from amrueda, andrewwinters5000 and SimonCan September 30, 2024 18:34

patrickersing and others added 2 commits October 1, 2024 14:37

Merge branch 'main' into remove_normal_direction_ll

681430d

update test values

2684172

Merge branch 'main' into remove_normal_direction_ll

038e1e3

ranocha requested changes Oct 2, 2024

View reviewed changes

This was linked to issues Oct 2, 2024

Remove normal_direction_ll in nonconservative terms #2049

Open

Boundary conditions on nonconservative terms not supported #1445

Open

patrickersing and others added 2 commits October 2, 2024 09:18

add news item

72cf490

Merge branch 'main' into remove_normal_direction_ll

be1f9e2

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Remove `normal_direction_ll` for nonconservative terms #2062

Remove `normal_direction_ll` for nonconservative terms #2062

patrickersing commented Sep 4, 2024 •

edited

Loading

github-actions bot commented Sep 4, 2024

patrickersing commented Sep 5, 2024

jlchan commented Sep 5, 2024

patrickersing commented Sep 5, 2024

jlchan commented Sep 6, 2024

DanielDoehring commented Sep 10, 2024

andrewwinters5000 left a comment

amrueda left a comment

amrueda Sep 20, 2024

patrickersing Sep 20, 2024

amrueda Sep 20, 2024

amrueda Sep 20, 2024

patrickersing Sep 20, 2024

amrueda Sep 20, 2024

ranocha commented Sep 30, 2024

patrickersing commented Oct 1, 2024

codecov bot commented Oct 1, 2024 •

edited

Loading

ranocha commented Oct 2, 2024

ranocha left a comment

Remove normal_direction_ll for nonconservative terms #2062

Are you sure you want to change the base?

Remove normal_direction_ll for nonconservative terms #2062

Conversation

patrickersing commented Sep 4, 2024 • edited Loading

github-actions bot commented Sep 4, 2024

Review checklist

Purpose and scope

Code quality

Documentation

Testing

Performance

Verification

patrickersing commented Sep 5, 2024

jlchan commented Sep 5, 2024

patrickersing commented Sep 5, 2024

jlchan commented Sep 6, 2024

DanielDoehring commented Sep 10, 2024

andrewwinters5000 left a comment

Choose a reason for hiding this comment

amrueda left a comment

Choose a reason for hiding this comment

amrueda Sep 20, 2024

Choose a reason for hiding this comment

patrickersing Sep 20, 2024

Choose a reason for hiding this comment

amrueda Sep 20, 2024

Choose a reason for hiding this comment

amrueda Sep 20, 2024

Choose a reason for hiding this comment

patrickersing Sep 20, 2024

Choose a reason for hiding this comment

amrueda Sep 20, 2024

Choose a reason for hiding this comment

ranocha commented Sep 30, 2024

patrickersing commented Oct 1, 2024

codecov bot commented Oct 1, 2024 • edited Loading

Codecov Report

ranocha commented Oct 2, 2024

ranocha left a comment

Choose a reason for hiding this comment

Remove `normal_direction_ll` for nonconservative terms #2062

Remove `normal_direction_ll` for nonconservative terms #2062

patrickersing commented Sep 4, 2024 •

edited

Loading

codecov bot commented Oct 1, 2024 •

edited

Loading