Modified update-release.sh script to improve portability (for LINUX, UNIX and macOS).
* The use of the 'date' command has been modified to use GNUs 'date' command (if installed).
* In addition, single quotes for sed commands have been replaced by double quotes to allow the use of single quotes
around internal variables. This appears to be more portable than the use of '\x27'. Note that this means shell meta-
characters need to be escaped if they need to be treated as string literals. Other said lines are also modified
accordingly for consistency.
This sets the default convergence tolerance to 1.0e-6 uniformly in hypre. Only three solvers had to be updated (AMG, ILU, and MGR), along with the corresponding documentation.
This PR fixes some memory leaks of BoomerAMGDD. Additional topics covered are:
* Perform code refactoring to meet the style used in hypre.
* Add GetIterations and FinalResidual interfaces for BoomerAMGDD.
* Add new regression tests.
* Make FAC Relaxation dependent on the execution policy.
* added extended+e interpolation, CUDA implementation
* added two regression tests for extended+e interpolation
* renamed some routines to better reflect type of interpolation
Co-authored-by: Ulrike M. Yang <ulrikey@ray53.coralea.llnl.gov>
Co-authored-by: Ruipeng Li <li50@llnl.gov>
This includes the implementation of the AMG-DD algorithm, a variant of BoomerAMG designed to limit communication.
AMG-DD may be used as a standalone solver or a preconditioner for Krylov methods (note that AMG-DD is a non-symmetric preconditioner). For an example of how to set up and use AMG-DD, see the IJ driver (src/test/ij.c).
A list with the parameters of AMG-DD is given below:
Padding (recommended default 1): HYPRE_BoomerAMGDDSetPadding(...)
Number of ghost layers (recommended default 1): HYPRE_BoomerAMGDDSetNumGhostLayers(...)
Number of inner FAC cycles per AMG-DD iteration (default 2): HYPRE_BoomerAMGDDSetFACNumCycles(...)
FAC cycle type: HYPRE_BoomerAMGDDSetFACCycleType(...)
1 = V-cycle (default)
2 = W-cycle
3 = F-cycle
Number of relaxations on each level during FAC cycle: HYPRE_BoomerAMGDDSetFACNumRelax(...)
Type of local relaxation during FAC cycle: HYPRE_BoomerAMGDDSetFACRelaxType(...)
0 = Jacobi
1 = Gauss-Seidel
2 = ordered Gauss-Seidel
3 = C/F L1-scaled Jacobi (default)
For more details of the algorithm, see Mitchell W.B., R. Strzodka, and R.D. Falgout (2020), Parallel Performance of Algebraic Multigrid Domain Decomposition (AMG-DD).
