WIP: Add FV method for T8codeMesh #1844

bennibolm · 2024-02-13T11:55:59Z

This PR adds the implementation of a first-order and second order finite volume scheme.
The mesh is organized with t8code. Therefore, it supports MPI parallelization.

Visualization is implemented using a t8code routine directly and therefore without .h5 files and Trixi2Vtk.

Validation is done in comments.

TODOs:

Check if convergence_test works parallel. If I remember correctly, it doesn't. Update: Seems to work now.
Reconstruction stencil on periodic boundaries
Second order extended reconstruction stencil. Adapt smaller reconstruction stencils
MPI parallel with extended reconstruction stencil (Problem: Using data from other ranks)

…aries)

github-actions · 2024-02-13T11:56:12Z

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.

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.

codecov · 2024-02-13T12:49:17Z

Codecov Report

Attention: Patch coverage is 0.29326% with 1020 lines in your changes missing coverage. Please review.

Project coverage is 86.80%. Comparing base (cde00a8) to head (8d7c2e1).
Report is 3 commits behind head on main.

Files with missing lines	Patch %	Lines
src/solvers/fv_t8code/fv.jl	0.00%	316 Missing ⚠️
src/meshes/t8code_mesh.jl	1.02%	290 Missing ⚠️
src/solvers/fv_t8code/containers.jl	0.00%	276 Missing ⚠️
src/callbacks_step/analysis_dg2d.jl	0.00%	56 Missing ⚠️
examples/t8code_2d_fv/elixir_advection_basic.jl	0.00%	23 Missing ⚠️
src/callbacks_step/stepsize_dg2d.jl	0.00%	22 Missing ⚠️
examples/t8code_2d_fv/elixir_euler_blast_wave.jl	0.00%	12 Missing ⚠️
src/callbacks_step/save_solution.jl	0.00%	10 Missing ⚠️
...2d_fv/elixir_euler_kelvin_helmholtz_instability.jl	0.00%	9 Missing ⚠️
examples/t8code_2d_fv/elixir_euler_free_stream.jl	0.00%	6 Missing ⚠️

❗ There is a different number of reports uploaded between BASE (cde00a8) and HEAD (8d7c2e1). Click for more details.

HEAD has 4 uploads less than BASE

Flag BASE (cde00a8) HEAD (8d7c2e1)

unittests 26 22

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1844      +/-   ##
==========================================
- Coverage   96.32%   86.80%   -9.52%     
==========================================
  Files         470      476       +6     
  Lines       37485    38502    +1017     
==========================================
- Hits        36106    33420    -2686     
- Misses       1379     5082    +3703

Flag	Coverage Δ
unittests	`86.80% <0.29%> (-9.52%)`	⬇️

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.

bennibolm · 2024-02-20T11:17:20Z

src/meshes/t8code_mesh.jl

+# Simple meshes
+# Temporary routines to create simple `cmesh`s by hand
+
+# Directly ported from: `src/t8_cmesh/t8_cmesh_examples.c: t8_cmesh_new_periodic_hybrid`.
+function cmesh_new_periodic_hybrid(comm)::t8_cmesh_t
+    n_dims = 2
+    vertices = [ # Just all vertices of all trees. partly duplicated
+        -1.0, -1.0, 0, # tree 0, triangle
+        0, -1.0, 0,
+        0, 0, 0,
+        -1.0, -1.0, 0, # tree 1, triangle


The following 300 lines are just by-hand building routines for simple cmeshs. This is only temporary and will be deleted in the future.

src/meshes/mesh_io.jl

src/meshes/t8code_mesh.jl

src/solvers/fv_t8code/fv.jl

src/callbacks_step/analysis.jl

src/callbacks_step/analysis_dg2d.jl

bennibolm · 2024-02-20T14:27:04Z

src/callbacks_step/analysis_dg2d.jl

+function integrate_via_indices(func::Func, u,
+                               mesh::T8codeMesh, equations,
+                               solver::FV, cache, args...;
+                               normalize = true) where {Func}


TODO: Relocate this routine, calc_error_norms and analyze to analysis.jl since they are dimension independent.

…h_info!`

bennibolm · 2024-07-19T12:47:51Z

After c4a4813, doing some validation:

Convergence test
First order with elixir_advection_basic.jl (with a periodic quad mesh and time integration method CarpenterKennedy2N54 and T=1.0):

convergence_test("examples/t8code_2d_fv/elixir_advection_basic.jl", 7, order=1, initial_refinement_level=2)

l2
scalar
error     EOC
2.33e-01  -
1.50e-01  0.64
8.55e-02  0.81
4.58e-02  0.90
2.37e-02  0.95
1.21e-02  0.98
6.08e-03  0.99

linf
scalar
error     EOC
3.28e-01  -
2.11e-01  0.63
1.21e-01  0.81
6.47e-02  0.90
3.35e-02  0.95
1.71e-02  0.97
8.60e-03  0.99

non-periodic domain:

convergence_test("../examples/t8code_2d_fv/elixir_advection_nonperiodic.jl", 7, initial_refinement_level=3, order=1)

l2
scalar
error     EOC
1.94e-01  -
1.24e-01  0.64
7.22e-02  0.78
3.92e-02  0.88
2.05e-02  0.93
1.05e-02  0.97
5.32e-03  0.98

linf
scalar
error     EOC
3.30e-01  -
2.11e-01  0.64
1.21e-01  0.81
6.47e-02  0.90
3.35e-02  0.95
1.71e-02  0.97
8.60e-03  0.99

Second order with elixir_advection_basic.jl (with a periodic quad mesh and time integration method CarpenterKennedy2N54 and T=1.0):
Using parameter r = 2, epsilon = 1.0e-13

convergence_test("examples/t8code_2d_fv/elixir_advection_basic.jl", 6, order=2, initial_refinement_level=1)

l2
scalar
error     EOC
2.54e-01  -
7.64e-02  1.73
1.89e-02  2.02
5.81e-03  1.70
1.74e-03  1.74
5.15e-04  1.75

linf
scalar
error     EOC
3.56e-01  -
1.19e-01  1.58
3.77e-02  1.66
1.42e-02  1.41
5.41e-03  1.39
2.07e-03  1.39

When I remove the limiting completely by lambda = [0.0; 1.0], I get

l2
scalar
error     EOC
2.09e-01  -
5.31e-02  1.97
9.68e-03  2.46
1.97e-03  2.29
4.59e-04  2.11
1.12e-04  2.03

linf
scalar
error     EOC
2.95e-01  -
7.20e-02  2.03
1.36e-02  2.40
2.78e-03  2.30
6.48e-04  2.10
1.59e-04  2.03

Free stream preservation:
Note: There is still a problem with using the mapping for FV with t8codemesh. Therefore, it's just the basic mesh used in this test.
Simulation of compressible Euler equation with a constant initial condition

trixi_include("examples/t8code_2d_fv/elixir_euler_free_stream.jl", tspan=(0.0, 100.0))

order=1: cfl=0.9, ~7000 time steps: Errors <1.0e-15
order=2: cfl=0.9, ~7000 time steps: Errors <1.0e-10 (smaller CFL doesn't make a difference) :/
order=2: No Limiting activated, cfl=0.9, ~7000 time steps: Errors <1.0e-15

bennibolm · 2024-07-24T09:15:58Z

src/meshes/t8code_mesh.jl

        mesh.tree_node_coordinates = tree_node_coordinates
+        mesh.max_number_faces = 2 * NDIMS # TODO: How to automatically adapt for other element types, e.g. triangles?


Is there a t8code routine that gives the maximum number of faces of all elements in a forest?

bennibolm · 2024-08-26T16:15:04Z

As in the dg elixirs with T8codeMesh, I added !isinteractive() && finalize(mesh) to the end of each fv elixir in 0994aab. This caused segmentation fault errors in the mpi tests (see here).
Also, somehow the nonperiodic test case failed.
So, I remove it again in a46520f.

sloede · 2024-08-28T04:25:10Z

As in the dg elixirs with T8codeMesh, I added !isinteractive() && finalize(mesh) to the end of each fv elixir in 0994aab. This caused segmentation fault errors in the mpi tests (see here). Also, somehow the nonperiodic test case failed. So, I remove it again in a46520f.

@jmark Could these issues be fixed by your PR we discussed yesterday? Or are they likely to be caused by the same (or similar) causes?

bennibolm · 2024-08-28T07:52:14Z

As in the dg elixirs with T8codeMesh, I added !isinteractive() && finalize(mesh) to the end of each fv elixir in 0994aab. This caused segmentation fault errors in the mpi tests (see here). Also, somehow the nonperiodic test case failed. So, I remove it again in a46520f.

@jmark Could these issues be fixed by your PR we discussed yesterday? Or are they likely to be caused by the same (or similar) causes?

Actually, the PR we talked about was the reason I added the lines in the first place. After failing, I was not sure, it should work in my examples at all, so I removed it again.
But @jmark and I already talked about this yesterday, and I will add it again and try to figure out what's going on here.

Long story short, I'm aware of the merged PR, and I am in contact with @jmark.

bennibolm added 2 commits February 13, 2024 12:53

Add first working first order FV method on T8codeMesh (periodic bound…

38ebc81

…aries)

Add simple routine to create cmeshs by hand

3c75fb9

bennibolm mentioned this pull request Feb 13, 2024

First-order Finite Volume Method organized with t8code #1627

Closed

4 tasks

fmt

2a7f417

bennibolm added 7 commits February 13, 2024 15:13

Fix bug in visualization

f59dc75

Relocate function to adapt to "normal" structure

398ed0d

Add tests

1c5a941

Rename Interface Container

67a25da

Activate tests; clean up

2d51511

Merge branch 'main' into t8codemesh-fv

173e4b9

Remove comments in elixirs [skip ci]

3f09f80