Package no.uib.cipr.matrix.distributed
Class BlockDiagonalPreconditioner
java.lang.Object
no.uib.cipr.matrix.distributed.BlockDiagonalPreconditioner
- All Implemented Interfaces:
Preconditioner
- Direct Known Subclasses:
TwoLevelPreconditioner
Deprecated.
Block diagonal preconditioner
-
Constructor Summary
ConstructorsConstructorDescriptionDeprecated.Constructor for BlockDiagonalPreconditioner -
Method Summary
Modifier and TypeMethodDescriptionDeprecated.Solves the approximate problem with the given right hand side.void
Deprecated.Sets the operator matrix for the preconditioner.transApply
(Vector b, Vector x) Deprecated.Solves the approximate transpose problem with the given right hand side.
-
Constructor Details
-
BlockDiagonalPreconditioner
Deprecated.Constructor for BlockDiagonalPreconditioner- Parameters:
prec
- Preconditioner to apply on the blocks. As this preconditioner is meant to be used for distributed memory matrices, the preconditioner should be constructed onA.getBlock()
-
-
Method Details
-
apply
Deprecated.Description copied from interface:Preconditioner
Solves the approximate problem with the given right hand side. Result is stored in given solution vector- Specified by:
apply
in interfacePreconditioner
- Parameters:
b
- Right hand side of problemx
- Result is stored here- Returns:
- x
-
transApply
Deprecated.Description copied from interface:Preconditioner
Solves the approximate transpose problem with the given right hand side. Result is stored in given solution vector- Specified by:
transApply
in interfacePreconditioner
- Parameters:
b
- Right hand side of problemx
- Result is stored here- Returns:
- x
-
setMatrix
Deprecated.Description copied from interface:Preconditioner
Sets the operator matrix for the preconditioner. This method must be called before a preconditioner is used by an iterative solver- Specified by:
setMatrix
in interfacePreconditioner
- Parameters:
A
- Matrix to setup the preconditioner for. Not modified
-
no.uib.cipr.matrix.distributed
package has been deprecated because of a number of hard to fix concurrency bugs. It is distributed only for backwards compatibility, but is not recommended. The utility of this package is questionable, as it does not allow distribution of computation between JVMs or across a network. For many people, distributed computing of multiple matrices can be achieved at a user-level through the JPPF Framework. Users who need to deal with few very large matrices may wish to implement their own storage classes and solvers using JPPF, but this will not be supported directly in matrix-toolkits-java.