Release 2022.09.02

Upgrade instructions

From #4130 (Unify iterable action return type):

For an iterable action, the following changes should be applied:

• #include <optional>
• #include "Parallel/AlgorithmExecution.hpp"
• remove #include "Parallel/AlgorithmMetafunctions.hpp" if present
• Declare the return type of the apply function of an iterable action to be Parallel::iterable_action_return_t
• Replace the return statements of the iterable action in the following way:
  • If returning a one argument tuple of a DataBox, instead return {Parallel::AlgorithmExecution::Continue, std::nullopt};
  • If returning a two argument tuple of a DataBox and bool, instead return {Parallel::AlgorithmExecution::Continue, std::nullopt}; if the bool is false and return {Parallel::AlgorithmExecution::Pause, std::nullopt}; if the bool is true
  • If returning a two argument tuple of a DataBox and Parallel::AlgorithmExecution, instead return the Parallel::AlgorithmExecution follwed by std::nullopt
  • If returning a three argument tuple of a DataBox, Parallel::AlgorithmExecution, and size_t, just remove the first argument (i.e. the DataBox)

From #4108 (Cce data in separate file):

Cce output will now be written to the reductions file instead of to a volume data file.

From #4153 (Remove support for Charm v6.10.2):

upgrade the version of charm++ you are using to v7.0.0

From #4151 (Dynamical creation and destruction of array element chares):

If you create an array element on a specified chare, you may have to add the type of to the constructor call to disambiguate from the newly added constructor.

From #4164 (Replace spin_vector function with new version.):

• spin_vector now takes different arguments than previously, and it has an additional template parameter.
• DimensionfulSpinVectorCompute now takes different Tags, and has an additional template parameter.

Neither one of these functions were actually being used in any of the executables.

From #4165 (AH finder computes inertial AH coords, not inertial Strahlkorper.):

The full Inertial-frame Strahlkorper is no longer available after horizon finds. This is ok, because except for spin direction (see #4164) all observables are scalars should be computed in the frame that the Strahlkorper was found, which is more efficient and more straightforward.

From #4183 (Add new stages and images to the Dockerfile):

Once this is merged, rebase on the latest develop branch to use the new container for CI on your PR.

Merged pull-requests (54)

General changes (47):

• Unify iterable action return type (#4130)
• Add predicted_zero_crossing function generalized for Datavectors (#4004)
• Add spectre h5 table reader (#4127)
• UniformCylindricalEndcap: make clang-tidy happy. (#4142)
• Fix new compilation errors on gcc 12 (#4140)
• Add absolute cutoff to ScalarAdvection subcell TCI (#4141)
• Cce data in separate file (#4108)
• Add LinearRegression. (#4136)
• Add size control logic and test. (#4097)
• Add Beginners tutorial to SpECTRE (#4146)
• Add actions for evaluating AMR criteria and updating AMR decisions based on neighbor decisions (#4134)
• Add momentum density to scalar wave (#4122)
• Change PredictedZeroCrossing algorithm. (#4137)
• Add CombineH5 executable (#4129)
• Remove support for Charm v6.10.2 (#4153)
• Add python code to visualize Dat files (#4062)
• Add ZeroCrossingPredictor (#4147)
• Add Christoffel and ExtrinsicCurvatureFunctions to KerrSchild, Minkowski and SphericalKerrschild (#4148)
• Replace CSW compute tags (#4149)
• Assume frustum not invertible below lower bound (#4161)
• Add clone methods to MathFunction (#4156)
• Dynamical creation and destruction of array element chares (#4151)
• Replace spin_vector function with new version. (#4164)
• Add StrahlkorperCoordsInDifferentFrame. (#4158)
• Add zero spin optimizations to KerrSchild (#4168)
• Split dense trigger triggering and rescheduling (#4077)
• Remove ScalarWaveGr (#4157)
• Add clang build analysis tools to build optimization docs (#4160)
• Remove quotes from python test names (#4169)
• Add filtering to BBH exec for local or global time stepping (#4144)
• Set SpherepackIterator by compact index (#4072)
• Add CompOSE EOS table reader (#4087)
• Add make_with_value implementation for Rational (#4175)
• Add MP5 reconstruction to finite difference (#4064)
• Add MP5 reconstructor to ValenciaDivClean system (#4066)
• Add Andrea Ceja to Metadata (#4180)
• Add new stages and images to the Dockerfile (#4183)
• Correct name of subcell option struct in docs (#4190)
• Add time dependence to CSW system (#4152)
• Remove functions/tags for taking square roots (#4191)
• Low-level performance improvements (#4187)
• Disable HDF5 default initialize datasets (#4194)
• Push new demo Docker image on PostRelease workflow (#4193)
• Make RunEventsAndDenseTriggers more generic (#4177)
• Upgrade docs stylesheets (#4188)
• Add executable to convert CompOSE EOS table to SpECTRE table (#4176)
• Skip TciAndSwitchToDg action for elements that just switched from DG to FD (#4198)

Bugfixes (7):

• Fix formatting in post-release workflow (#4143)
• Fix bug in test_copy_semantics where constructor instead of assignment operator is called (#4154)
• Fix finding Python on macOS CI (#4170)
• Fix a check in CubicScale. (#4173)
• Conditionally resize LHS Tensors before RHS TensorExpression evaluation (#4159)
• AH finder computes inertial AH coords, not inertial Strahlkorper. (#4165)
• Don't die on non-real quadratic equation roots (#4189)

Contributors (13): @kidder, @MarloMo, @nilsdeppe, @markscheel, @wthrowe, @yoonso0-0, @nilsvu, @geoffrey4444, @knelli2, @acifajkya, @asaju7, @nikwit, @macedo22