Symbol "bHYPRE
Symbol "bHYPRE.BoomerAMG" (version 1.0.0)Algebraic multigrid solver, based on classical Ruge-Stueben.
BoomerAMG requires an IJParCSR matrix
The following optional parameters are available and may be set using the appropriate Parameter function (as indicated in parentheses):
- MaxLevels
- (Int) - maximum number of multigrid levels.
- StrongThreshold
- (Double) - AMG strength threshold.
- MaxRowSum
- (Double) -
- CoarsenType
- (Int) - type of parallel coarsening algorithm used.
- MeasureType
- (Int) - type of measure used; local or global.
- CycleType
- (Int) - type of cycle used; a V-cycle (default) or a W-cycle.
- NumGridSweeps
- (IntArray 1D) - number of sweeps for fine and coarse grid, up and down cycle. DEPRECATED: Use NumSweeps or Cycle?NumSweeps instead.
- NumSweeps
- (Int) - number of sweeps for fine grid, up and down cycle.
- Cycle1NumSweeps
- (Int) - number of sweeps for down cycle
- Cycle2NumSweeps
- (Int) - number of sweeps for up cycle
- Cycle3NumSweeps
- (Int) - number of sweeps for coarse grid
- GridRelaxType
- (IntArray 1D) - type of smoother used on fine and coarse grid, up and down cycle. DEPRECATED: Use RelaxType or Cycle?RelaxType instead.
- RelaxType
- (Int) - type of smoother for fine grid, up and down cycle.
- Cycle1RelaxType
- (Int) - type of smoother for down cycle
- Cycle2RelaxType
- (Int) - type of smoother for up cycle
- Cycle3RelaxType
- (Int) - type of smoother for coarse grid
- GridRelaxPoints
- (IntArray 2D) - point ordering used in relaxation. DEPRECATED.
- RelaxWeight
- (DoubleArray 1D) - relaxation weight for smoothed Jacobi and hybrid SOR. DEPRECATED: Instead, use the RelaxWt parameter and the SetLevelRelaxWt function.
- RelaxWt
- (Int) - relaxation weight for all levels for smoothed Jacobi and hybrid SOR.
- TruncFactor
- (Double) - truncation factor for interpolation.
- JacobiTruncThreshold
- (Double) - threshold for truncation of Jacobi interpolation.
- SmoothType
- (Int) - more complex smoothers.
- SmoothNumLevels
- (Int) - number of levels for more complex smoothers.
- SmoothNumSweeps
- (Int) - number of sweeps for more complex smoothers.
- PrintFileName
- (String) - name of file printed to in association with SetPrintLevel.
- NumFunctions
- (Int) - size of the system of PDEs (when using the systems version).
- DOFFunc
- (IntArray 1D) - mapping that assigns the function to each variable (when using the systems version).
- Variant
- (Int) - variant of Schwarz used.
- Overlap
- (Int) - overlap for Schwarz.
- DomainType
- (Int) - type of domain used for Schwarz.
- SchwarzRlxWeight
- (Double) - the smoothing parameter for additive Schwarz.
- Tolerance
- (Double) - convergence tolerance, if this is used as a solver; ignored if this is used as a preconditioner
- DebugFlag
- (Int) -
- InterpType
- (Int) - Defines which parallel interpolation operator is used. There are the following options for interp_type:
0 classical modified interpolation 1 LS interpolation (for use with GSMG) 2 classical modified interpolation for hyperbolic PDEs 3 direct interpolation (with separation of weights) 4 multipass interpolation 5 multipass interpolation (with separation of weights) 6 extended classical modified interpolation 7 extended (if no common C neighbor) classical modified interpolation 8 standard interpolation 9 standard interpolation (with separation of weights) 10 classical block interpolation (for use with nodal systems version only) 11 classical block interpolation (for use with nodal systems version only) with diagonalized diagonal blocks 12 FF interpolation 13 FF1 interpolation The default is 0.
- NumSamples
- (Int) - Defines the number of sample vectors used in GSMG or LS interpolation.
- MaxIterations
- (Int) - maximum number of iterations
- Logging
- (Int) - Set the
logging level , specifying the degree of additional informational data to be accumulated. Does nothing by default (level = 0). Other levels (if any) are implementation-specific. Must be called before Setup and Apply.- PrintLevel
- (Int) - Set the
print level , specifying the degree of informational data to be printed either to the screen or to a file. Does nothing by default (level=0). Other levels (if any) are implementation-specific. Must be called before Setup and Apply.The following function is specific to this class:
- SetLevelRelxWeight
- (Double , \tt Int) - relaxation weight for one specified level of smoothed Jacobi and hybrid SOR.
Objects of this type can be cast to Solver objects using the __cast methods.
alphabetic index hierarchy of classes
generated by doc++