1# Changes: 3.24 2 3% STYLE GUIDELINES: 4% * Capitalize sentences 5% * Use imperative, e.g., Add, Improve, Change, etc. 6% * Don't use a period (.) at the end of entries 7% * If multiple sentences are needed, use a period or semicolon to divide sentences, but not at the end of the final sentence 8 9```{rubric} General: 10``` 11 12- Add `PETSCPYTHONPATH` to the generated `$PETSC_DIR/$PETSC_ARCH/lib/petsc/conf/petscvariables` file so it is available to users through the makefile system 13- Add `PETSCPYTHONPATH` to the generated `$PETSC_DIR/$PETSC_ARCH/lib/pkgconfig/PETSc.pc` file so it is available to users with 14 `PKG_CONFIG_PATH=$PETSC_DIR/$PETSC_ARCH/lib/pkgconfig pkg-config --variable=PETSCPYTHONPATH PETSc.pc` 15- Add `EXTRA_OPTIONS_INITIAL` to the test system, which prepends options 16 17```{rubric} Configure/Build: 18``` 19 20- Change `make sphinxhtml` in the `doc` directory to be `make docs` 21- Change `make docs` to put all its artifacts in `${PETSC_ARCH}-doc` instead of `doc` 22- Add support for `--download-xxx`, `--with-xxx`, and `-with-xxx-dir` for Python packages that install with pip. See `packages.py` 23- Change `make alletags` to `make etags` 24- Deactivate Fortran bindings of SuperLU_DIST and ExodusII by default, but they can still be built with the configure options `--with-superlu_dist-fortran-bindings` and `--with-exodusii-fortran-bindings`, respectively 25- Add `--with-python-exec-from-env=python` to allow providing the Python to be used by mpi4py and petsc4py without hard-wiring the full path at ./configure time 26 27```{rubric} Sys: 28``` 29 30- Add `PetscOptionsBool3()` 31- Add `PETSC_E` 32- Deprecate `PetscSSEIsEnabled()` 33- Add `PetscBTCopy()` 34- Change `PetscBool` to be a C bool. It now has a size of one byte, when previously it had a size of four bytes 35- Deprecate `MPIU_BOOL` in favor of `MPI_C_BOOL`. This is now possible since `PetscBool` is now a C bool 36- Add `PetscStackView()` to the public API 37- Change the default file pointer used in `PetscStackView()` if `NULL` is used to `PETSC_STDERR` 38- Add `-petsc_viewer_stdout_format formatname` to allow initializing the format of `PETSC_VIEWER_STDOUT_()` 39 40```{rubric} Event Logging: 41``` 42 43```{rubric} PetscViewer: 44``` 45 46- Add `PetscViewerHDF5SetCompress()` and `PetscViewerHDF5GetCompress()` 47 48```{rubric} PetscDraw: 49``` 50 51- Add `PetscDrawLGGetData()` 52 53```{rubric} AO: 54``` 55 56```{rubric} IS: 57``` 58 59```{rubric} VecScatter / PetscSF: 60``` 61 62- Change `VecSetOperation()` and `VecGetOperation()` so that their final argument must be casted with `PetscErrorCodeFn *` and `PetscErrorCodeFn **` 63 64```{rubric} PF: 65``` 66 67```{rubric} Vec: 68``` 69 70```{rubric} PetscSection: 71``` 72 73- Add `PetscSectionArrayView()` 74 75```{rubric} PetscPartitioner: 76``` 77 78- Add `PETSCPARTITIONERMULTISTAGE` for improved load balance, edge cut, and performances of large scale unstructured mesh partitioning. 79 80```{rubric} Mat: 81``` 82 83- Add `MatConstantDiagonalGetConstant()` 84- Add `MatNullSpaceRemoveFn` type definition 85- Add `MatMFFDFn`, `MatMFFDiFn`, `MatMFFDiBaseFn`, and `MatMFFDCheckhFn` type definitions 86- Add `MatFDColoringFn` type definition 87- Add support for `-mat_mumps_icntl_15 1` with the companion function `MatMumpsSetBlk()` 88- Change `MatSetOperation()` and `MatGetOperation()` so that their final argument must be casted with `PetscErrorCodeFn *` and `PetscErrorCodeFn **` 89- Change `MatShellSetOperation()` and `MatShellGetOperation()` so that their final argument must be casted with `PetscErrorCodeFn *` and `PetscErrorCodeFn **` 90 91```{rubric} MatCoarsen: 92``` 93 94```{rubric} PC: 95``` 96 97- Add `PCMatApplyTranspose()` and `PCShellSetMatApplyTranspose()` 98- Remove `PC_ApplyMultiple` 99- Add `PCShellPSolveFn` 100- Add `PCModifySubMatricesFn` 101- Remove `PCSetPreSolve()` and `PCSetPostSolve()`. See `PCSetPostSetUp()` 102 103```{rubric} KSP: 104``` 105 106- Add `MatLMVMGetLastUpdate()` 107- Add `MatLMVMMultAlgorithm`, `MatLMVMSetMultAlgorithm()`, and `MatLMVMGetMultAlgorithm()` 108- Add `MatLMVMSymBroydenGetPhi()` and `MatLMVMSymBroydenSetPhi()` 109- Add `MatLMVMSymBadBroydenGetPsi()` and `MatLMVMSymBadBroydenSetPsi()` 110- Deprecate `KSP_CONVERGED_RTOL_NORMAL` in favor of `KSP_CONVERGED_RTOL_NORMAL_EQUATIONS` and `KSP_CONVERGED_ATOL_NORMAL` in favor of `KSP_CONVERGED_ATOL_NORMAL_EQUATIONS` 111- Add `KSPFlexibleSetModifyPC()` to provide a common API for setting the modification function for all flexible `KSP` methods 112- Add `KSPFlexibleModifyPCFn` function prototype 113- Change the function signature of the `destroy()` argument to `KSPSetConvergenceTest()` to `PetscCtxDestroyFn*`. If you provide custom destroy 114 functions to `KSPSetConvergenceTest()` you must change them to expect a `void **` argument and immediately dereference the input 115- Add `KSPPSolveFn` 116- Change `KSPMonitorResidualDraw()` to `KSPMonitorResidualView()` 117- Change `KSPMonitorTrueResidualDraw()` to `KSPMonitorTrueResidualView()` 118 119```{rubric} SNES: 120``` 121 122- Change `SNESTestJacobian()` to report the norms 123- Add `SNESNormSchedule` support to `SNESKSPONLY` 124 125```{rubric} SNESLineSearch: 126``` 127 128- Rename option `snes_linesearch_maxstep` to `snes_linesearch_maxlambda` to better coincide with its purpose in the various `SNESLineSearch`es 129- Rename `SNESLineSearchL2` to `SNESLineSearchSecant` (and hence its option entry `snes_linesearch_type` form `l2` to `secant`) to better represent the underlying approach 130 131```{rubric} TS: 132``` 133 134- Add `TSSetRunSteps()` and `-ts_run_steps` for better control of restarted jobs 135- Add `-ts_monitor_solution_skip_initial` to skip first call to the solution monitor 136- Add `-ts_monitor_wall_clock_time` to display the elapsed wall-clock time for every step 137- Change `TSDiscGradIsGonzalez()`, `TSDiscGradUseGonzalez()` to `TSDiscGradSetType()`,`TSDiscGradGetType()` 138 139```{rubric} TAO: 140``` 141 142- Add `TaoBRGNSetRegularizationType()`, `TaoBRGNGetRegularizationType()` 143- Add `TaoGetInequalityConstraintsRoutine()`, `TaoGetEqualityConstraintsRoutine()`, `TaoGetJacobianInequalityRoutine()` and `TaoGetJacobianEqualityRoutine()` 144 145```{rubric} PetscRegressor: 146``` 147 148- Add new component to support regression and classification machine learning tasks: [](ch_regressor) 149- Add `PetscRegressor` type `PETSCREGRESSORLINEAR` for solving linear regression problems with optional regularization 150 151```{rubric} DM/DA: 152``` 153 154- Add `DMHasBound()`, `DM_BC_LOWER_BOUND` and `DM_BC_LOWER_BOUND` 155- Add `DMSetCellCoordinateField()` 156- Add ``localized`` argument to `DMSetCoordinateDisc()` and `DMCreateAffineCoordinates_Internal()` 157- Add `DMCreateGradientMatrix()` 158 159```{rubric} DMSwarm: 160``` 161 162- Add `DMSwarmProjectFields()` and `DMSwarmProjectGradientFields()` 163 164```{rubric} DMPlex: 165``` 166 167- Add `DMPlexGetTransform()`, `DMPlexSetTransform()`, `DMPlexGetSaveTransform()`, and `DMPlexSetSaveTransform()` 168- Add `DMPlexGetCoordinateMap()` and `DMPlexSetCoordinateMap()` 169- Add `DMPlexTransformCohesiveExtrudeGetUnsplit()` 170- Add `DMFieldCreateDefaultFaceQuadrature()` 171- Rename `DMPlexComputeResidual_Internal()` to `DMPlexComputeResidualForKey()` 172- Rename `DMPlexComputeJacobian_Internal()` to `DMPlexComputeJacobianByKey()` 173- Rename `DMPlexComputeJacobian_Action_Internal()` to `DMPlexComputeJacobianActionByKey()` 174- Rename `DMPlexComputeResidual_Hybrid_Internal()` to `DMPlexComputeResidualHybridByKey()` 175- Rename `DMPlexComputeJacobian_Hybrid_Internal()` to `DMPlexComputeJacobianHybridByKey()` 176- Add `DMPlexInsertBounds()` 177- Change argument order for `DMPlexComputeBdResidualSingle()` and `DMPlexComputeBdJacobianSingle()` to match domain functions 178- Add `DMPlexComputeBdResidualSingleByKey()` and `DMPlexComputeBdJacobianSingleByLabel()` 179- Add ``localized`` argument to `DMPlexCreateCoordinateSpace()` 180- Remove ``coordFunc`` argument from `DMPlexCreateCoordinateSpace()` 181- Change `DMPlexExtrude()` to take a label argument 182- Rename `DMPlexVecGetOrientedClosure_Internal()` to `DMPlexVecGetOrientedClosure()` 183- Correctly handle `Mat` preallocation for isoperiodic boundary conditions 184 185```{rubric} FE/FV: 186``` 187 188- Add `PetscFEExpandFaceQuadrature()` 189- Add `PetscFECreateBrokenElement()` 190- Change `PetscFEIntegrateJacobian()` signature to allow rectangular operators 191 192```{rubric} DMNetwork: 193``` 194 195```{rubric} DMStag: 196``` 197 198```{rubric} DT: 199``` 200 201- Deprecate `PetscSimplePointFunc` in favor of `PetscSimplePointFn *` 202- Deprecate `PetscPointFunc` in favor of `PetscPointFn *` 203- Deprecate `PetscPointJac` in favor of `PetscPointJacFn *` 204- Deprecate `PetscBdPointFunc` in favor of `PetscBdPointFn *` 205- Deprecate `PetscBdPointJac` in favor of `PetscBdPointJacFn *` 206- Deprecate `PetscRiemannFunc` in favor of `PetscRiemannFn *` 207- Deprecate `PetscProbFunc` in favor of `PetscProbFn *` 208- Add `PetscDTCreateQuadratureByCell()` 209 210```{rubric} Fortran: 211``` 212 213- Add `PetscObjectNullify()` 214- Require Fortran compiler to have `.true.=b00000001` and `.false.=b00000000` for `logical(C_BOOL)`. Thus require the compiler flags `-fpscomp logicals` for Intel and `-Munixlogical` for NVIDIA compilers 215- Change `PetscBool` to be `logical(C_BOOL)` (equivalent to a `logical(kind=1)`). It now has a size of one byte, previously it was a `logical(kind=4)` and had a size of four bytes 216- Remove the `./configure` option `-with-fortran-type-initialize=0`. Hence, it is now not possible to include PETSc objects in common blocks 217