SPARSKIT
Name SPARSKITAbsTolerance
Section Math::SPARSKIT
Type float
Default 1e-10
Some SPARSKIT solvers use an absolute tolerance as a stopping criterion
for the iterative solution process. This variable can be used to
specify the tolerance.
Name SPARSKITIterOut
Section Math::SPARSKIT
Type integer
Default -1
Determines how often status info of the solver is printed.
If <= 0, will never be printed.
Name SPARSKITKrylovSubspaceSize
Section Math::SPARSKIT
Type integer
Default 15
Some of the SPARSKIT solvers are Krylov subspace methods.
This variable determines what size the solver will use
for the subspace.
Name SPARSKITMaxIter
Section Math::SPARSKIT
Type integer
Default 50000
This variable controls the maximum number of iteration steps that
will be performed by the (iterative) linear solver.
Name SPARSKITRelTolerance
Section Math::SPARSKIT
Type float
Default 1e-8
Some SPARSKIT solvers use a relative tolerance as a stopping criterion
for the iterative solution process. This variable can be used to
specify the tolerance.
Name SPARSKITSolver
Section Math::SPARSKIT
Type integer
Default sk_bcg
Specifies what kind of linear solver will be used.
Options:
- sk_cg:
Conjugate Gradient Method
- sk_cgnr:
Conjugate Gradient Method (Normal Residual equation)
- sk_bcg:
Bi-Conjugate Gradient Method
- sk_dbcg:
BCG with partial pivoting
- sk_bcgstab:
BCG stabilized
- sk_tfqmr:
Transpose-Free Quasi-Minimum Residual method
- sk_fom:
Full Orthogonalization Method
- sk_gmres:
Generalized Minimum Residual method
- sk_fgmres:
Flexible version of Generalized Minimum Residual method
- sk_dqgmres:
Direct versions of the Quasi-Generalized Minimum Residual method
Name SPARSKITVerboseSolver
Section Math::SPARSKIT
Type logical
Default no
When set to yes, the SPARSKIT solver will write more detailed output.