1# Changes: 3.22 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 `-mpiuni-allow-multiprocess-launch` to allow mpiexec to launch multiple indendent MPI-Uni jobs. Previously a sanity check 13 error message was always produced 14 15```{rubric} Configure/Build: 16``` 17 18- Add `--with-openmp-kernels` 19- Update to work with python-3.13 20- Change `MPIU_Allreduce()` to always returns an MPI error code that should be checked with `PetscCallMPI()` 21- Add `PetscCallReturnMPI()`, `PetscCallMPIReturnMPI()`, and `PetscCheckReturnMPI()` to check for in PETSc provided MPI callbacks 22 23```{rubric} Sys: 24``` 25 26- Add `PetscPragmaUseOMPKernels` 27- Deprecate `PetscOptionsRestoreViewer()` in favor of `PetscViewerDestroy()` 28- Deprecate `PetscOptionsGetViewer()` and `PetscOptionsGetViewers()` in favor of `PetscOptionsCreateViewer()` and `PetscOptionsCreateViewers()` 29- Deprecate `PetscOptionsPushGetViewerOff()`, `PetscOptionsPopGetViewerOff()`, and `PetscOptionsGetViewerOff()` in favor of 30 `PetscOptionsPushCreateViewerOff()`, `PetscOptionsPopCreateViewerOff()`, and `PetscOptionsGetCreateViewerOff()` 31- Add `PetscObjectContainerCompose()` and `PetscObjectContainerQuery()` 32- Add `size_t` argument to `PetscMPIErrorString()` 33- Add `PetscCallExternalAbort()` for calling external library functions from functions not returning `PetscErrorCode` 34- Add `PetscRandomSetOptionsPrefix()` 35- Add `PetscOptionsGetBool3()` to get a ternary logical value from a command line option 36 37```{rubric} Event Logging: 38``` 39 40```{rubric} PetscViewer: 41``` 42 43- Add `PetscViewerASCIIStdoutSetFileUnit()` 44- Add `PetscShmgetAllocateArrayScalar()`, `PetscShmgetDeallocateArrayScalar()`, `PetscShmgetAllocateArrayInt()`, and `PetscShmgetDeallocateArrayInt()` for Fortran 45 46```{rubric} PetscDraw: 47``` 48 49```{rubric} AO: 50``` 51 52```{rubric} IS: 53``` 54 55```{rubric} VecScatter / PetscSF: 56``` 57 58- Add `PetscSFRegisterPersistent()` and `PetscSFDeregisterPersistent()` for safe optimization of persistent communication 59 60```{rubric} PF: 61``` 62 63```{rubric} Vec: 64``` 65 66- The `IS` passed to `VecISAXPY()`, `VecISCopy()`, `VecISSet()`, and `VecISShift()` must have the same communicator of the vectors used 67- Make `VecLock` API active in optimized mode 68- `VecNestSetSubVec()` and `VecNestSetSubVecs()` now take references to input vectors rather than creating duplicates 69- Deprecate `VecSetInf()` with `VecFlag()` 70 71```{rubric} PetscSection: 72``` 73 74```{rubric} PetscPartitioner: 75``` 76 77```{rubric} Mat: 78``` 79 80- Change `MatProductSetFill()` to support `PETSC_DETERMINE` and `PETSC_CURRENT`. `MatMatMult()` and its friends and relations now accept 81 `PETSC_DETERMINE` and `PETSC_CURRENT` in the `fill` argument. `PETSC_DEFAULT` is deprecated for those functions 82- Change the default `MatType` of the output `Mat` of `MatSchurComplementComputeExplicitOperator()` to be `MATDENSE`. It may be changed from the command line, e.g., `-fieldsplit_1_explicit_operator_mat_type aij` 83- Add `MATOP_GET_BLOCK_DIAGONAL` and `MATOP_GET_VBLOCK_DIAGONAL` to set methods returning the diagonal point blocks of a matrix 84 85```{rubric} MatCoarsen: 86``` 87 88```{rubric} PC: 89``` 90 91- Add support in `PCFieldSplitSetFields()` including with `-pc_fieldsplit_%d_fields fields` for `MATNEST`, making it possible to 92 utilize multiple levels of `PCFIELDSPLIT` with `MATNEST` from the command line 93- Add `PCCompositeSpecialSetAlphaMat()` API to use a matrix other than the identity in 94 preconditioners based on an alternating direction iteration, e.g., setting $M$ for 95 $P = (A + alpha M) M^{-1} (alpha M + B)$ 96- Reuse the result of $T = A_{00}^-1 A_{01}$ in `PCApply_FieldSplit_Schur` with `-pc_fieldsplit_schur_fact_type full` 97- Change the option database keys for coarsening for `PCGAMG` to use the prefix `-pc_gamg_`, for example `-pc_gamg_mat_coarsen_type` 98- Add `PCGAMGSetGraphSymmetrize()` and `-pc_gamg_graph_symmetrize` to control symmetrization when coarsening the graph 99- Add `-pc_hypre_type ilu` with `-pc_hypre_ilu_type`, `-pc_hypre_ilu_iterative_setup_type`, `-pc_hypre_ilu_iterative_setup_maxiter`, 100 `-pc_hypre_ilu_iterative_setup_tolerance`, `-pc_hypre_ilu_print_level`, `-pc_hypre_ilu_logging`, `-pc_hypre_ilu_level`, 101 `-pc_hypre_ilu_max_nnz_per_row`, `-pc_hypre_ilu_tol`, `-pc_hypre_ilu_maxiter`, `-pc_hypre_ilu_drop_threshold`, 102 `-pc_hypre_ilu_tri_solve`, `-pc_hypre_ilu_lower_jacobi_iters`, `-pc_hypre_ilu_upper_jacobi_iters`, and `-pc_hypre_ilu_local_reordering` 103- Improve `-pc_type boomeramg` with `-pc_hypre_boomeramg_smooth_num_sweeps`, `-pc_hypre_boomeramg_ilu_type`, `-pc_hypre_boomeramg_ilu_iterative_setup_type`, 104 `-pc_hypre_boomeramg_ilu_iterative_setup_option`, `-pc_hypre_boomeramg_ilu_iterative_setup_maxiter`, `-pc_hypre_boomeramg_ilu_iterative_setup_tolerance`, 105 `-pc_hypre_boomeramg_ilu_print_level`, `-pc_hypre_boomeramg_ilu_logging`, `-pc_hypre_boomeramg_ilu_level`, `-pc_hypre_boomeramg_ilu_max_nnz_per_row`, 106 `-pc_hypre_boomeramg_ilu_maxiter`, `-pc_hypre_boomeramg_ilu_drop_tol`, `-pc_hypre_boomeramg_ilu_tri_solve`, `-pc_hypre_boomeramg_ilu_lower_jacobi_iters`, 107 `-pc_hypre_boomeramg_ilu_upper_jacobi_iters`, and `-pc_hypre_boomeramg_ilu_local_reordering` 108 109```{rubric} KSP: 110``` 111 112- Add support for `PETSC_DETERMINE` as an argument to `KSPSetTolerances()` to set the parameter back to its initial value when the object's type was set 113- Deprecate `PETSC_DEFAULT` in favor of `PETSC_CURRENT` for `KSPSetTolerances()` 114 115```{rubric} SNES: 116``` 117 118- Add support for `PETSC_DETERMINE` as an argument to `SNESSetTolerances()` to set the parameter back to its initial value when the object's type was set 119- Deprecate `PETSC_DEFAULT` in favor of `PETSC_CURRENT` for `SNESSetTolerances()` 120- Add `DMAdaptorMonitor()`, `DMAdaptorMonitorSet()`, `DMAdaptorMonitorCancel()`, and `DMAdaptorMonitorSetFromOptions()` 121- Add `DMAdaptorMonitorSize()`, `DMAdaptorMonitorError()`, `DMAdaptorMonitorErrorDraw()`, `DMAdaptorMonitorErrorDrawLGCreate()`, and `DMAdaptorMonitorErrorDrawLG()` 122- Add `DMAdaptorMonitorRegister()`, `DMAdaptorMonitorRegisterAll()`, and `DMAdaptorMonitorRegisterDestroy()` 123- Add `DMAdaptorGetCriterion()` and `DMAdaptorSetCriterion()` 124- Add `DMAdaptorSetOptionsPrefix()` 125- Add Newton's method with arc length continuation: `SNESNEWTONAL` with `SNESNewtonALSetFunction()`, `SNESNewtonALGetFunction()`, `SNESNewtonALComputeFunction()`, `SNESNewtonALGetLoadParameter()`, and `SNESNewtonALSetCorrectionType()` 126- Add `SNESNewtonTRSetTolerances()` and `SNESNewtonTRSetUpdateParameters()` to programmatically set trust region parameters 127- Deprecate `SNESSetTrustRegionTolerance()` in favor of `SNESNewtonTRSetTolerances()` 128- Add `SNESResetCounters()` to reset counters for linear iterations and function evaluations 129 130```{rubric} SNESLineSearch: 131``` 132 133```{rubric} TS: 134``` 135 136- Add Rosenbrock-W methods from {cite}`rang2015improved` with $B_{PR}$ stability: `TSROSWR34PRW`, `TSROSWR3PRL2`, `TSROSWRODASPR`, and `TSROSWRODASPR2` 137- Add support for `PETSC_DETERMINE` as an argument to `TSSetTolerances()` to set the parameter back to its initial value when the object's type was set 138- Deprecate `PETSC_DEFAULT` in favor of `PETSC_CURRENT` for `TSSetTolerances()` 139- Add support for `PETSC_DETERMINE` as an argument to `TSSetMaxSteps()` and `TSSetMaxTime()` 140- Deprecate `PETSC_DEFAULT` in favor of `PETSC_CURRENT` for `TSAdaptSetSafety()` 141- Deprecate `PETSC_DEFAULT` in favor of `PETSC_CURRENT` for `TSAdaptSetClip()` 142- Deprecate `PETSC_DEFAULT` in favor of `PETSC_CURRENT` for `TSAdaptSetStepLimits()` 143- Add `TSGetStepResize()` 144- Add `-ts_monitor_solution_vtk_interval` to control the interval for dumping files 145- Add a new ARKIMEX solver for fast-slow systems that are partitioned component-wise and additively at the same time 146- Add `TSRHSSplitSetIFunction()`, `TSRHSSplitSetIJacobian()`, `TSRHSSplitSetSNES()`, `TSRHSSplitGetSNES()`, `TSARKIMEXSetFastSlowSplit()`, and `TSARKIMEXGetFastSlowSplit()` to support the new solver 147 148```{rubric} TAO: 149``` 150 151- Add support for `PETSC_DETERMINE` as an argument to `TaoSetTolerances()` and `TaoSetConstraintTolerances()` to set the parameter back to its initial value when the object's type was set 152- Deprecate `PETSC_DEFAULT` in favor of `PETSC_CURRENT` for `TaoSetTolerances()` and `TaoSetConstraintTolerances()` 153 154```{rubric} DM/DA: 155``` 156 157- Add `DMGetSparseLocalize()` and `DMSetSparseLocalize()` 158- Add `DMGeomModelRegister()`, `DMGeomModelRegisterAll()`, `DMGeomModelRegisterDestroy()`, `DMSnapToGeomModel()`, and `DMSetSnapToGeomModel()` to support registering geometric models 159- Add `DMGetOutputSequenceLength()` 160- Add an additional return vector to `DMCreateMassMatrixLumped()` to retrieve the local mass lumping 161- Add `DMPlexMigrateGlobalToNaturalSF()` modifies the NaturalSF to map from the SF's old global section to the new global section 162 163```{rubric} DMSwarm: 164``` 165 166```{rubric} DMPlex: 167``` 168 169- Add `DMLabelGetValueBounds()` 170- Add `DMPlexOrientLabel()` 171- Add an argument to `DMPlexLabelCohesiveComplete()` in order to change behavior at surface boundary 172- Remove `DMPlexSnapToGeomModel()` 173- Add refinement argument to `DMPlexCreateHexCylinderMesh()` 174- Now `DMPlexComputeBdIntegral()` takes one function per field 175- Add `DMPlexCreateEdgeNumbering()` 176- Add `DMPlexComputeL2FluxDiffVec()` and `DMPlexComputeL2FluxDiffVecLocal()` 177- Add `DMAdaptorSetType()`, `DMAdaptorGetType()`, `DMAdaptorRegister()`, `DMAdaptorRegisterAll()`, and `DMAdaptorRegisterDestroy()` 178- Add `DMAdaptorGetMixedSetupFunction()` and `DMAdaptorSetMixedSetupFunction()` 179- Add `DMPlexCreateCellNumbering()` 180- Add `DMPlexBuildFromCellSectionParallel()` and `DMPlexCreateFromCellSectionParallel()` 181- Add `-dm_plex_box_label` to add "Face Sets" label with current "box" conventions 182- Add "Face Sets" label to simplex meshes using current "box" conventions 183- Update `PetscViewerExodusII` object to cache variable names for faster and more robust look-ups 184- Add `PetscViewerExodusIISetNodalVariable()`, `PetscViewerExodusIISetZonalVariable()`, `PetscViewerExodusIIGetNodalVariable()`, and `PetscViewerExodusIIGetZonalVariable()` to set or get the number of variables in a file 185- Add `PetscViewerExodusIISetNodalVariableName()`, `PetscViewerExodusIISetZonalVariableName()`, `PetscViewerExodusIIGetNodalVariableName()`, and `PetscViewerExodusIIGetZonalVariableName()` to set or get a single variable name 186- Add `PetscViewerExodusIISetNodalVariablesNames()`, `PetscViewerExodusIISetZonalVariablesNames()`, `PetscViewerExodusIIGetNodalVariablesNames()`, and `PetscViewerExodusIIGetZonalVariablesNames()` to set or get all variable names at the same time (not available in Fortran) 187- Add degree bounds to `DMCopyFields()`, `DMCopyDS()`, `PetscDSCopy()`, and `PetscDSSelectDiscretizations()` 188- Add `PetscFELimitDegree()` 189- Add localizationHeight and sparseLocalize arguments to `DMPlexCreateBoxMesh()` for coordinate localization on periodic meshes 190- Add parallel CGNS reader, enabled by `-dm_plex_cgns_parallel` 191- Add CGNS function for `VecLoad()` of solutions in parallel (must be run with `-dm_plex_cgns_parallel`) 192- Add `PetscViewerCGNSOpen()` convenience function 193- Add `PetscViewerCGNSGetSolutionTime()`, `PetscViewerCGNSGetSolutionName()`, `PetscViewerCGNSSetSolutionIndex()`, and `PetscViewerCGNSGetSolutionIndex()` 194- Add `DMPlexGetDepthStratumGlobalSize()` 195 196```{rubric} FE/FV: 197``` 198 199```{rubric} DMNetwork: 200``` 201 202```{rubric} DMStag: 203``` 204 205```{rubric} DT: 206``` 207 208- Add `PetscDSSetIntegrationParameters()` and `PetscDSSetCellParameters()` 209 210```{rubric} Fortran: 211``` 212 213- Add `PETSC_NULL_ENUM` to be used instead of `PETSC_NULL_INTEGER` when a pointer to an `enum` is expected in a PETSc function call 214- Add `PETSC_NULL_INTEGER_ARRAY`, `PETSC_NULL_SCALAR_ARRAY`, and `PETSC_NULL_REAL_ARRAY` for use instead of 215 `PETSC_NULL_INTEGER`, `PETSC_NULL_SCALAR`, and `PETSC_NULL_REAL` when an array is expected in a PETSc function call 216- Add automatically generated interface definitions for most PETSc functions to detect illegal usage at compile time 217- Add `PetscObjectIsNull()` for users to check if a PETSc object is `NULL` 218- Change the PETSc Fortran API so that non-array values, `v`, passed to PETSc routines expecting arrays must be cast with `[v]` in the calling sequence 219