AMReX 26.03

What's Changed

• Fix access specifier with HDF5 enabled by @gardner48 in #4921
• ParmParse: Add support for AMREX_USE_GPU macro by @WeiqunZhang in #4927
• Added multi-array version of EBData by @ankithadas in #4929
• CI: Enable MPI in CUDA test on Windows by @WeiqunZhang in #4869
• Reducer: New wrapper class for ReduceOps and ReduceData by @WeiqunZhang in #4933
• Add more Doxygen comments to Array4 by @WeiqunZhang in #4935
• Generalize SIMD Single Source Design by @ax3l in #4924
• Add more doxygen comments to Array by @WeiqunZhang in #4937
• Use Clang-Tidy 21 in CIs by @WeiqunZhang in #4939
• FindAMReXFFTW.cmake Support Native Threading by @ax3l in #4934
• FFTW CMake: Hint More Lib Paths (Windows) by @ax3l in #4940
• Test: SIMD by @ax3l in #4938
• Protect from using MarchingCubes if not 3D by @cgilet in #4946
• ParticleContainerToBlueprint: Allocator (2) by @ax3l in #4948
• Fix: AMReX w/ OpenMP and Ignore FFTW OMP by @ax3l in #4941
• SENSEI: Allocator by @ax3l in #4949
• SYCL: Add a new path for big kernels by @WeiqunZhang in #4952
• Guard againt potential OOB error in ParticleContainer OK() by @cgilet in #4951
• Respect NoSync region in FabArray::Copy and setBndry by @WeiqunZhang in #4955
• CI: Fix clang-tidy cache by @WeiqunZhang in #4944
• Doc: Details on AMREX_DEFAULT_INIT by @ax3l in #4974
• Doxygen FFT by @WeiqunZhang in #4975
• prevent auto-converting const char * to bool in ParmParse::add by @charlesgwaldman in #4969
• Fix (currently latent) bug with Particles + tiling + GPU by @atmyers in #4973
• Fix (currently latent) bug with pure SoA particles if periodic_shift … by @atmyers in #4972
• Fix particle communication bug when calling RedistributeCPU but with … by @atmyers in #4970
• Follow-on to #4973 by @atmyers in #4978
• Fix host_idcpu to use finest_level_in_file when restarting [Particles] [IO] by @atmyers in #4971
• [LinearSolvers] Fix bug in setBCoeffs(Vector) by @WeiqunZhang in #4968
• [LinearSolvers][1D Overset] Fix missing component index by @WeiqunZhang in #4967
• Add Gpu::SyncAtExit and Gpu::streamSyncActive by @WeiqunZhang in #4956
• NonLocalBC: Fix Boolean condition by @WeiqunZhang in #5065
• Fix SYCL device property init by @WeiqunZhang in #5066
• BaseFab::indexFromValue: Fix index typo by @WeiqunZhang in #5067
• EB VTK writer: Fix attribute name by @WeiqunZhang in #5069
• AmrMesh: Fix invalid iterator range by @WeiqunZhang in #5068
• add ReduceToPlaneMF2Patchy by @BenWibking in #4958
• Improve clarity of AMR assign grid function by @atmyers in #5072
• Use Long instead of int for maxnextid in pre/post particle IO by @atmyers in #5073
• Fix sub-communicator in CPU fillNeighbors() by @atmyers in #5077
• assert that pureSoA containers have at least SpaceDim real components by @atmyers in #5076
• Use size_t instead of int for buffer offset in CPU fillNeighbors by @atmyers in #5078
• Print/PrintToFile: Fix potential MPI_Comm issue by @WeiqunZhang in #5079
• Doxygen HYPRE by @WeiqunZhang in #5070
• AllGatherBoxes: Fix MPI type by @WeiqunZhang in #5081
• SUNDIALS alloc-failure-output-pointer by @ajnonaka in #5094
• SUNDIALS Stale Flag by @ajnonaka in #5092
• Fix impicit MRI configuration calls by @gardner48 in #5093
• PoissonHybrid: Validate length of dz by @WeiqunZhang in #5085
• R2X: Assert zero-based domain indexing by @WeiqunZhang in #5084
• FFT: Early abort for unsupported path by @WeiqunZhang in #5083
• 1D CurlCurl: Update variable beta smoother for consistence by @WeiqunZhang in #5082
• Fix GPU memory issue in the ReduceToPlanePatchy test by @WeiqunZhang in #5086
• Update CHANGES for 26.03 by @WeiqunZhang in #5102

Full Changelog: 26.02...26.03