Weak form modal DG methods on simplicial (and quad/hex meshes) #647
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Codecov Report
@@ Coverage Diff @@
## dev #647 +/- ##
========================================
Coverage 93.63% 93.63%
========================================
Files 171 177 +6
Lines 16535 16763 +228
========================================
+ Hits 15482 15696 +214
- Misses 1053 1067 +14
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
todo: remove later
introduces a wrapper around MeshData and boundary tags
jlchan
changed the title
|
@ranocha should be ready for merging. Went ahead and added tets too 😁 |
ranocha
reviewed
Jun 30, 2021
|
I started reviewing this. Not sure when it will be done, but it would be great if only necessary bug fixes could be pushed forth etime being, to reduce the rsik of one of my comments being gobbled by the changes. |
sloede
reviewed
Jun 30, 2021
Comment on lines
13
to
16
| @test_trixi_include(joinpath(EXAMPLES_DIR, "elixir_euler_triangular_mesh.jl"), | ||
| l2 = [7.687514677136661e-5, 8.844681835345058e-5, 8.844681835367038e-5, 0.0002667915678724591], | ||
| linf = [0.00023033713406972467, 0.0001967281286732181, 0.00019672812868742895, 0.0004570579510136952] | ||
| ) |
There was a problem hiding this comment.
Finally, do these test adhere to the ten second rule and if not, would it be possible to make them?
There was a problem hiding this comment.
Does this mean the entire testset should run in under 10 seconds?
There was a problem hiding this comment.
Each test should run in under 10 seconds (roughly)
| @@ -0,0 +1,47 @@ | |||
| # !!! warning "Experimental features" | |||
There was a problem hiding this comment.
Also, is there another name than triangulate that is more descriptive? Without having dug into the details, I don't understand why there is a triangular_mesh and a triangulate_mesh. But maybe it's ok, if there is no better distinguishing characteristic!
There was a problem hiding this comment.
Hm...the Triangulate_mesh is an unstructured mesh created by the Triangulate.jl package, which is also why it was capitalized previously.
Maybe eilxir_euler_triangulate_pkg_mesh.jl?
|
I've started to review this. Since changes pushed after I add a comment sometimes gobbles these comments, I'd prefer if only
I just submitted a first round of comments, so you can go ahead and change away :-) |
Co-authored-by: Michael Schlottke-Lakemper <michael@sloede.com> Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com>
Co-authored-by: Michael Schlottke-Lakemper <michael@sloede.com>
new folders: `simplicial_2d_dg` and `simplicial_3d_dg`
- defaults to `mul!` for `A::UniformScaling`
|
@ranocha I think this should be ready to merge into |
ranocha
approved these changes
Jul 1, 2021
ranocha
added a commit
that referenced
this pull request
Jul 27, 2021
* Weak form modal DG methods on simplicial (and quad/hex meshes) (#647) * first draft of weak form modal DG methods * fix naming of new folder * adding structarrays and Trixi interface stuff * adding calc_source! * adding sandbox folder for experiments todo: remove later * new VertexMappedMesh type introduces a wrapper around MeshData and boundary tags * adding timers * removing "has_nonconservative_terms" for now * updating scratch elixir * adding draft of BoundaryStateDirichlet/Wall * updating elixir * updating elixir for testing * adding specialization for nodal SBP methods * minor minor formatting and docstring edits, removing an unnecessary type annotation, renaming variables * formatting + fixed BC bug * working elixir with weak form DG on triangles * calc_error_norms returns component-wise errors * removing unused boundary condition code * adding `func` to calc_error_norms * making AbstractMeshData !<: AbstractMesh * minor formatting * adding periodic/non-periodic elixirs * remove unused type alias / annotations * adding integrate * adding analysis callback code adding analysis to Trixi.jl, moving analysis out of dg.jl * removing extra ::AbstractVector annotation * adding tests * adding `2d_tri` tests to CI * force add Manifest.toml * Revert "force add Manifest.toml" This reverts commit 74a9a46. * adding ape triangular example * modified @requires statements make it so that Trixi @requires both StructArrays/StartUpDG for triangular mesh * switch to L2 projection for compute_coefficients * removing cruft "sandbox" testing folder * formatting (2-space tabs, alignment) * renaming folder dg_tri -> dg_triangles * Update src/solvers/dg_triangles/mesh.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update .github/workflows/ci.yml Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update examples/triangular_mesh_2D/elixir_ape_triangular_mesh.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update examples/triangular_mesh_2D/elixir_ape_triangular_mesh.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update src/solvers/dg_triangles/mesh.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update examples/triangular_mesh_2D/elixir_euler_triangular_mesh.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update examples/triangular_mesh_2D/elixir_euler_periodic_triangular_mesh.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update src/solvers/dg_triangles/dg.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update test/test_examples_2d_triangular_mesh.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update test/runtests.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update src/solvers/dg_triangles/dg.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update src/solvers/dg_triangles/dg.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update src/solvers/dg_triangles/dg.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update src/solvers/dg_triangles/dg.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update src/solvers/dg_triangles/mesh.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update src/solvers/dg_triangles/mesh.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * adding new elixirs + tagging as experimental * using `each___` for iteration * minor doc changes * update NEWS.md * compat bounds * fix eltype in analysis * update path dg_tri -> dg_triangles * Apply suggestions from code review Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * caught mistake in L2 norm computation * fixed elixir name: euler -> ape * adding convergence elixir * adding compat with StartUpDG@0.9.1 new SBP type selector API. minor version bump was to avoid conflict in exporting DGSEM... * remove SBP() from convergence + bump test project.toml versions * renaming folder dg_triangles - > dg_simplices * Apply suggestions from code review Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * fixing bug in digest_boundary_conditions for triangles * specializing source_terms in calc_sources! * change dg_triangles -> dg_simplices in Trixi.jl * adding Triangulate VertexMappedMesh constructor * adding example using Triangulate * specializing digest_boundary_conditions to avoid ambiguity * adding docs for simplicial meshes * update gitignore for MacOS ignore .DS_Store * updating docs [skip ci] * Apply suggestions from code review Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * updating docs - capping lines at 100 columns - replacing N -> polydeg - replacing md.sJ -> md.Jf - adding more descriptions of quadrature rules and SBP nodes - finishing descriptions of plotting-related fields in RefElemData * replacing N -> polydeg in elixirs * replacing md.sJ -> md.Jf in solver code * bump StartUpDG compat * specializing existing `digest_boundary_conditions` routines - restricting x_neg, x_pos, ... unpacking routines to mesh::TreeMesh * Apply suggestions from code review Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * removing elixir which duplicates coverage updating elixirs for CI tests set up parameters to reduce runtimes * updating # todos - # todo -> # Todo: simplices, removing old todos. - fixing todo for calc_error_norms updating todos in dg.jl * updating `each___` iteration function names * adding @threaded * adding Base.show for VertexMappedMesh * adding Octavian - adding Octavian to Project.toml - adding `using Octavian` to `Trixi.jl` * updating l2, linf errors for test for triangular elixirs * updating test/Project.toml - temporarily removing Plots.jl to test CI on the PR * trying Octavian: matmul! * Update test/test_examples_2d_triangular_mesh.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * fixing linf::MVector issue - `similar(::SVector)` returns an `MVector`, so I replaced it * selectively using Octavian `matmul!` revert to `mul!` for 5-argument multiply with vector arguments. Octavian hangs (see JuliaLinearAlgebra/Octavian.jl#103) * compat bound for Octavian, reverting OrdinaryDiffEq compat * fixing bugs for SBP-type discretizations - specialize `mul_by!` for UniformScaling - using LinearAlgebra: UniformScaling - dispatch on subtype of SBP * fixing Octavian compat had put Octavian compat in test/Project.toml instead of Project.toml * fixing thread-related bug - adding threading and preallocated thread arrays other minor change: removed unnecessary StructArray conversion in `compute_coefficients` * update `todo` label for Octavian.matmul! * update NEWS.md with simplicial mesh updates * renaming elixirs and test: triangular_mesh -> simplicial_mesh * adding support for 3D tet meshes - made most simplicial routines dimension-agnostic * added `min_max_speed_naive` for CompressibleEulerEquations3D * updating NEWS.md with tet meshes * fixing test naming: 2d_triangles -> simplices * updating docs with news about simplices * adding minor comments * VertexMappedMesh: fix order of type parameters * Apply suggestions from code review Co-authored-by: Michael Schlottke-Lakemper <michael@sloede.com> Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update docs/make.jl Co-authored-by: Michael Schlottke-Lakemper <michael@sloede.com> * reorganizing simplicial examples new folders: `simplicial_2d_dg` and `simplicial_3d_dg` * resolving issue with `mul_by!` dispatch for UniformScaling - defaults to `mul!` for `A::UniformScaling` * renaming/reorganizing simplicial tests in ci * updating docs with tentative name for Triangulate.jl elixir * adding Plots.jl v1.16 to test compat * adding convergence test for triangular mesh * updating `eachdim(mesh)`, also `@inline`ing `each___` functions Co-authored-by: Jesse Chan <jesse.chan@rice.edu> Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> Co-authored-by: Michael Schlottke-Lakemper <michael@sloede.com> * renaming cache variables in preparation for flux differencing * adding StructArrays and LazyArrays in prep for flux differencing * Replacing .1 -> 0.1, .5 -> 0.5 * fixing spacing in `mesh_data_meshes.md` docs * fixing elixir comments * adding `TODO` comment to `nelements(solver, cache)` * removing fluxdiff.jl from Trixi to be added in #695 * updating VertexMappedMesh - swapping Dim -> NDIMS - removing Tv, Ti from VertexMappedMesh type parameters * replacing @trixi_testset with outer @testset + inner @trixi_testset * renaming variables - VX, VY, VZ -> vertex_coordinates_x, ... - Dim -> NDIMS - invMQrstTrW -> weak_differentiation matrices * renaming DG{<:RefElemData,...} -> MultiDG{...} * Update src/Trixi.jl Co-authored-by: Michael Schlottke-Lakemper <michael@sloede.com> * adding name to contributors in AUTHORS.md [skip ci] * fix Plots compat in test/Project.toml * genenralize print_boundary_conditions * Apply suggestions from code review Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * adding Markdown indentation to docs * add pretty printing for boundary conditions * add Base.summary printing for MultiDG types * implementing `nelements(mesh, dg, cache)` for analysis * moving files to appropriate folders * reducing allocations in `calc_boundary_flux!` * Flux differencing on simplicial elements (#695) * implemented flux differencing volume kernel for SBP triangles * renaming flux differencing elixir * removing StructArrays from elixirs * factoring type aliases into separate file * update type aliases more restrictive subtyping * implemented entropy stable flux differencing on triangular meshes tet meshes still have some bugs * removing StructArrays from tet elixir * fixing bug in tet flux differencing * clean up comments * adding inlines and mul-adds * changing tet mesh size in elixir (to satisfy 10 second test rule) * adding `using Tet` from StartUpDG * specializing `build_lazy_physical_derivative` restricting to `VertexMappedMesh{NDIMS, <:Union{Tri, Tet}}`, which have constant geometric terms. Will specialize later to include affine quad/hex meshes. Todo: specialize for curved meshes. * minor rearrangement of `using ...` and trimming whitespace * reducing number of elements in tests (make them slightly cheaper) * adding simplicial flux differencing tests * renaming type_aliases.jl -> types_and_traits.jl - also updating PolyDGFluxDiff and SBPDGFluxDiff to use MultiDG type * adding flux diff elixirs to docs * renaming type_aliases.jl -> types_and_traits.jl in Trixi.jl * adding MultiDG convenience constructor * Apply suggestions from code review Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * Update examples/simplicial_2d_dg/elixir_euler_sbp_triangular_mesh_flux_diff.jl Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * renaming folder to `dg_multi` * removing extra l2, linf error printouts * fixing names of elixirs in docs * trim whitespace * renaming flux_diff -> ec * increasing tspan for simplicial tests * removing cruft * Apply suggestions from code review Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * fixing _ec name inconsistency in elixirs * Apply suggestions from code review Co-authored-by: Michael Schlottke-Lakemper <michael@sloede.com> * Update src/solvers/dg_multi/fluxdiff.jl Co-authored-by: Michael Schlottke-Lakemper <michael@sloede.com> * adding hex elixir * removing cruft `solvers/dg_simplices` folder * making StartUpDG an explicit dependency * update elixirs - use `DGMulti` constructor - updating function signatures which involve `RefElemData` - remove `using StartUpDG` * MultiDG -> DGMulti MultiDG -> DGMulti * name clarifications + function specializiations * combining dg_multi tests into a single `@testset` * adding back commented out tests * removing unnecessary annotations * renaming type aliases for DGMulti{...} - removed PolyDGFluxDiff, replaced with DGMultiFluxDiff{Polynomial} - removed SBPDGFluxDiff, replaced with DGMultiFluxDiff{SBP} * elem_type -> element_type * minor formatting * consolidating elixirs and updating docs - switching from multiple elixirs to just using `trixi_include` to change parameters in a few elixirs - minor formatting minor formatting * renaming VX, ... -> vertex_coordinates_x, ... * adding periodic 3d dgmulti elixir + tests * renaming CI tests * Apply suggestions from code review Co-authored-by: Michael Schlottke-Lakemper <michael@sloede.com> Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * renaming dg_multi -> dgmulti * more renaming - dg_multi -> dgmulti - estimate_dt(dg, mesh) -> estimate_dt(mesh, dg) * fix indentation in docstring * renaming dg_multi -> dgmulti in tests * separating dgmulti elixirs into 2d/3d * selectively importing/exporting from StartUpDG * adding compat bounds for StartUpDG * adding kwarg option for DGMulti passed onto RefElemData Co-authored-by: Jesse Chan <jesse.chan@rice.edu> Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> Co-authored-by: Michael Schlottke-Lakemper <michael@sloede.com> * moving function definitions to after DGMulti defined * moving "includes" to appropriate locations * update .codecov.yml * add VertexMappedMesh to overview table * fix docstring * fix hadamard_sum_A_transposed * update test compat bounds * use at-muladd * Update test/Project.toml Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> * removing let blocks * add solver/discretization type to overview table * updating docs with new `DGMulti` solver info * minor changes to docs * add some cross references to the docs Co-authored-by: Jesse Chan <jesse.chan@rice.edu> Co-authored-by: Hendrik Ranocha <ranocha@users.noreply.github.com> Co-authored-by: Michael Schlottke-Lakemper <michael@sloede.com> Co-authored-by: Hendrik Ranocha <mail@ranocha.de>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
After reading through @ranocha and @andrewwinters5000 new solvers, I think I have a better idea of how to structure mine. Closed #557 in favor of this PR, which I think will be better organized and more readable (hopefully).
Currently passes the "preserve constant state" test for
CompressibleEulerEquations2Dusing a triangular DG discretization. Periodic entropy sine wave seems to work now too.Todo:
I think I may save flux differencing for another PR so this one is more readable.