bump version

See merge request !69

* bump version

Cleanup

See merge request !68

* rm deps BenchmarkTools, GLob, InteractiveUitls, LaTeXStrings, REPL; update dg1dTests' dependencies

* remove Plots, PlotThemes, Interpolations

improve a few things in the `make_movie.jl` script

See merge request !63

* improve a few things

- avoid axis jittering by using the Inside() alignmode for axes
- allow to plot cell data as constant functions
- update manifest for v1.9

Improve `HeatEq`

See merge request !65

* add heateq_sine reftest

* rename reftest from heateq to heateq_gaussian

* rename numflux methods

* change time step heuristic to align with what nodal DG book suggests (p. 250)

* add sine initial data for heateq

Stepper for time evolution

See merge request !67

* fix evolve! by wrapping stepper result also in callback

previously we copied the stepper's result into the initial data buffer
but now we just swap the vectors
we now push a wrapper for the dynamic state vector (like we also use it
for the rhs_fn, timestep_fn) to the fullstep callback so that we don't
have to do the wrapping in projects

* teach pushfirst to CallbackSet

* avoid adding cfl twice

* disable the timestep-retaking tests as they are now deprecated

* grab cfl parameter from mesh parameters

* fix while condition to be aware of symbol return values

* fix up evolve test

* fix time stepping with Evolution type

- make a copy of initial data to initialize up1, because
  we need to allocate memory anyways and the stepper will also swap it before
  stepping
- apply cfl to timestep

* fix construction of Evolution type

- fix sanity checks for callbacks
- add sanity check for cfl
- introduce TimeStepAlgorithm super type and move all algorithm
  types into types.jl
- add docstring to Evolution type
- update tests

* update unit tests

* introduce a state full type Evolution so that we can evolve step by step

Fix CI

See merge request !66

* fix undef var

* fix vector signature

* try to capture and log full backtrace in ref tests

* revert some changes

* ignore v2.2.3 in PrettyTables in IntegrationTests

* what is the real error we see?

* wrap parameter printing into try-finally, because PrettyTables.jl causes the CI to crash here

* also upload errors.log

* log the captured error to a file

* also resolve before running ref tests

convert error to warning when unknown parameters are found

See merge request !64

* convert error to warning when unknown parameters are found

`@with_signature`: Add debug options

See merge request !62

* add tests

* allow to selectively insert for loop macros

* add inbounds,simd,debug to @with_signature options

Add `HeatEq`

See merge request !61

* add ref test

* add HeatEq project

add helpers to run test parfiles with main()

See merge request !60

* add helpers to run test parfiles with main()

Migrate all of EulerEq project to use new boradcasting methods

See merge request !59

* remove t=0.6 output step from euler_sod_shock_tube_entropyav reftest

* rename euler_sod_shock_tube test to euler_sod_shock_tube_bernstein
also set `every_dt = 0` for ref test parfile

* migrate rest of Euler 1d + AV to use new broadcasting

* fixup new_broadcast_bdry and test it with 1d + AV method

* ref test euler_sod_shock_tube_entropyav: crank up CFL

* enable output checking for euler_sod_shock_tube test

* add ref test euler_sod_shock_tube_entropyav

* migrate 1d + AV method

* fixup migration of 1d eqs

* strip new_ from new 2d equations

* implement interpolate_face_data! for 1d mesh; migrate llf for 1d mesh

* migrate 1d equations to use new broadcasting, pt 1

* clean up more equations

* convert entropy variables computation and maxspeed
to use new broadcast interface

needed to update 2d ref tests, because somehow changing
speed computation showed massive differences in ref output

* use new broadcasts for other 2d rhs

* ref test: use aligned_ts for euler2d isentropic

* use new broadcast for 2d + AV method

* add klevin helmholtz as ref test

use aligned_ts for output times

New broadcasting methods based on face-buffer abstractions

See merge request !58

* wrap new_broadcast_* methods into TimerOutput

* Add LV to optimize `dg_rhs` loops v2 (!57)

* prefer @turbo over mul! in dg_rhs.jl

* test mul! instead of @turbo for mass matrix applicationg

* don't use mapreduce for 2 arg absmax

* allow using Cart[1d,2d,3d].[Directions,Axis] as tuple indices

* wrap broadcast_[volume,faces]_2! into global timer

* split up turbo loops due to limitation of single inner loops in LV; see discussion LV.jl/issues/506

* accelerate dg rhs evaluation using LoopVectorization.jl

* put new_broadcast_[volume,faces] to use in EulerEq2d

* introduce bulkfaceindices, bulkbdryindices to work with faceindices, bdryindices; implement face interpolation method

* make has_signature also dispatch on Type

* add print_idxs option to @with_signature

* add faceindices, bdryindices to Mesh; implement broadcasting for volume, faces, bdry

Add LV to optimize `dg_rhs` loops v2

See merge request !57

* prefer @turbo over mul! in dg_rhs.jl

* test mul! instead of @turbo for mass matrix applicationg

* don't use mapreduce for 2 arg absmax

* allow using Cart[1d,2d,3d].[Directions,Axis] as tuple indices

* wrap broadcast_[volume,faces]_2! into global timer

* split up turbo loops due to limitation of single inner loops in LV; see discussion LV.jl/issues/506

* accelerate dg rhs evaluation using LoopVectorization.jl

Refactor `@with_signature` to allow iterations over vectors and custom ranges

See merge request !56

* extend docs

* move new with_signature tests into test suite

* refactor keyword ex parsing; add more safeguards for options parsing; add more tests

* implemented tests for refactored version

* fixup to redesign

* rework with_signature to insert a loop into function body

Make EulerEq2d run with Kelvin Helmholtz test

See merge request !54

* add TODO on bdryconds not being implemented for AV 2d

* fixup dg_rhs in special case when one flux component is nonzero

* cleanup initialdata

* make EulerEq2d run with Kelvin Helmholtz test (not stable yet)
also substitute all max with absmax in equations.jl

* export broadcast_bdry_2!

* clean up EulerEq/equation.jl; remove local LLF definitions

* move out numerical utils in separate file

remove old precompile tools; bump julia version to 1.9

See merge request !53

* separate out movie.jl script

* bump julia version to 1.9

* remove precompile tools

improvements to ScalarEq, EulerEq, SRHD

See merge request !52

* make EulerEq sod shock tube work with entropy AV

* add Mesh.periodic parameters and update Mesh constructors and reftests

* fixup av parameters and edge case in mda viscosity

* hack in bdryconds for EulerEq

* get sod shock tube with EV going

* switch mesh1d methods from EulerEq over to use new broadcast; update integrationtests to include ID output

* fixup EulerEq 1d with reconstructors

* filter tests to run by regex

fixup regex dispatch for runtests

* filter tests to run by regex

* switch ScalarEq 1d versions over to use new
broadcasting methods; updated all ScalarEq tests

* fixup cons2prim call in RHS; some formatting

* fix find_cell_index edge case

* wip

* use vtk as default 2d output format

* add smoothed_bump initial data

* wip: hack in shocktube problem with non-periodic domain

* can't smooth boundary cells in non-periodic problems

* add cell index helper for mesh

* add bdrycell array to tree

* add absolute maximum with index helper

* add logging parameter section

* hack in entropy av method

Enable entropy av for 2d Euler equation

See merge request !50

* add kelvin helmholtz test which is not yet working

* fixup EulerEq reftests

* enable initialdata setup for hrsc modules

* add initial data for Kelvin Helmholtz test

* quick shot on EulerEquation with entropy viscosity

* add overloads of dg rhs weak form for more efficiency in LDG

convert some TCI and HRSC for 2d runs

See merge request !49

* disable advection2d_bump_entropyav test

* only log error type when ref test fails to run

* update ref tests

* initialize hrsc and tci variables

* add todos

* fix entropy av to account for jacobian

* tciav: remove unnecessary dispatch

* remove unused code

* fix typos in EV computation

* integration tests: copy and upload if two output files differ

* also update registry in ci

CI: update registry before activating env?

CI: debug registry status

CI: readd general registry

CI: rm registry update

CI: somehow need registry update now ...

* commit Manifest and let CI run on 1.9

remove Manifest again

readd manifest

* add missing output.h5 for ref test

* remove xdmf output from refs

* fix duplicated index in av smoother

* rework AV smoother for mesh1d

* fix cell offsets and iterator for mesh1d

remove debug statement

* use 1d output for ref tests of 2d runs

update ref test vars for 2d runs

* test entropy AV with burgers2d

* fix HRSC smoother

* no need to add up results for cell integration

* remove branch when integrating

* enforce type stability for evolve method

* fix entropy AV computation

* fix type instability in glgl

* also import dependencies ...

* add integration test for advection2d + tci av + threshold tci

* change default format2d to xdmf

* add integration for advection2d + entropy av

* fix vtk

* update dependencies

* add unittests for 2d output formats

* cleanup vtk

* enable vtk, xdmf+bin, xdmf+h5 2d output methods

* add vtk output method for 2d

* refactor xdmf output to allow hdf5 and bin format for heavy data

* quick&dirty enabling of entropy viscosity for advection2d

* add integrate_cell for meshes

* inline integrate

* export cellindices

* extend entropy viscosity to 2d

* adjust dispatch

* remove irrelevant size check

* enable av smoothing in 2d

* implement C0 smoother in 2d

* make SmoothedArtificialViscosity type stable

* cure type stability in Environment

* quick shot on ScalarEq with AV

* extend TCIViscosity to 2d

* merge HRSC/interface.jl into HRSC/HRSC.jl

* extend TCI/threshold to Mesh2d

* remove TCI/interface.jl

Streamline DG rhs computation interface

See merge request !48

* remove dg strong form for now

* streamline dg weak form interface

* add ref tests for EulerEq2d and SRHD2d