petsc4py.PETSc.KSP.Type#
- class petsc4py.PETSc.KSP.Type#
Bases:
objectKSP Type.
The available types are:
RICHARDSONThe preconditioned Richardson iterative method
KSPRICHARDSON.CHEBYSHEVThe preconditioned Chebyshev iterative method.
KSPCHEBYSHEV.CGThe Preconditioned Conjugate Gradient (PCG) iterative method.
KSPCGGROPPCGA pipelined conjugate gradient method (Gropp).
KSPGROPPCGPIPECGA pipelined conjugate gradient method.
KSPPIPECGPIPECGRRPipelined Conjugate Gradients with Residual Replacement.
KSPPIPECGRRPIPELCGDeep pipelined (length l) Conjugate Gradient method.
KSPPIPELCGPIPEPRCGPipelined predict-and-recompute conjugate gradient method.
KSPPIPEPRCGPIPECG2Pipelined conjugate gradient method with a single non-blocking reduction per two iterations.
KSPPIPECG2CGNEApplies the preconditioned conjugate gradient method to the normal equations without explicitly forming AᵀA.
KSPCGNENASHConjugate gradient method subject to a constraint on the solution norm.
KSPNASHSTCGConjugate gradient method subject to a constraint on the solution norm.
KSPSTCGGLTRConjugate gradient method subject to a constraint on the solution norm.
KSPGLTRFCGFlexible Conjugate Gradient method (FCG). Unlike most KSP methods this allows the preconditioner to be nonlinear.
KSPFCGPIPEFCGPipelined, Flexible Conjugate Gradient method.
KSPPIPEFCGGMRESGeneralized Minimal Residual method with restart.
KSPGMRESPIPEFGMRESPipelined (1-stage) Flexible Generalized Minimal Residual method.
KSPPIPEFGMRESFGMRESImplements the Flexible Generalized Minimal Residual method.
KSPFGMRESLGMRESAugments the standard Generalized Minimal Residual method approximation space with approximations to the error from previous restart cycles.
KSPLGMRESDGMRESDeflated Generalized Minimal Residual method. In this implementation, the adaptive strategy allows to switch to the deflated GMRES when the stagnation occurs.
KSPDGMRESPGMRESPipelined Generalized Minimal Residual method.
KSPPGMRESTCQMRA variant of Quasi Minimal Residual (QMR).
KSPTCQMRBCGSStabilized version of Biconjugate Gradient (BiCGStab) method.
KSPBCGSIBCGSImproved Stabilized version of BiConjugate Gradient (IBiCGStab) method in an alternative form to have only a single global reduction operation instead of the usual 3 (or 4).
KSPIBCGSQMRCGSQuasi- Minimal Residual variant of the Bi-CGStab algorithm (QMRCGStab) method.
KSPQMRCGSFBCGSFlexible Stabilized version of BiConjugate Gradient (BiCGStab) method.
KSPFBCGSFBCGSRA mathematically equivalent variant of flexible stabilized BiConjugate Gradient (BiCGStab).
KSPFBCGSRBCGSLVariant of the L-step stabilized BiConjugate Gradient (BiCGStab(L)) algorithm. Uses “L-step” Minimal Residual (MR) polynomials. The variation concerns cases when some parameters are negative due to round-off.
KSPBCGSLPIPEBCGSPipelined stabilized BiConjugate Gradient (BiCGStab) method.
KSPPIPEBCGSCGSConjugate Gradient Squared method.
KSPCGSTFQMRA Transpose Tree Quasi- Minimal Residual (QMR).
KSPCRCR(Preconditioned) Conjugate Residuals (CR) method.
KSPCRPIPECRPipelined Conjugate Residual (CR) method.
KSPPIPECRLSQRLeast squares solver.
KSPLSQRPREONLYApplies ONLY the preconditioner exactly once. This may be used in inner iterations, where it is desired to allow multiple iterations as well as the “0-iteration” case. It is commonly used with the direct solver preconditioners like PCLU and PCCHOLESKY. There is an alias of KSPNONE.
KSPPREONLYNONENo solver
KSPNONEQCGConjugate Gradient (CG) method subject to a constraint on the solution norm.
KSPQCGBICGImplements the Biconjugate gradient method (BiCG). Similar to running the conjugate gradient on the normal equations.
KSPBICGMINRESMinimum Residual (MINRES) method.
KSPMINRESSYMMLQSymmetric LQ method (SymmLQ). Uses LQ decomposition (lower trapezoidal).
KSPSYMMLQLCDLeft Conjugate Direction (LCD) method.
KSPLCDPYTHONPython shell solver. Call Python function to implement solver.
KSPPYTHONGCRPreconditioned flexible Generalized Conjugate Residual (GCR) method.
KSPGCRPIPEGCRPipelined Generalized Conjugate Residual method.
KSPPIPEGCRTSIRMTwo-Stage Iteration with least-squares Residual Minimization method.
KSPTSIRMCGLSConjugate Gradient method for Least-Squares problems. Supports non-square (rectangular) matrices.
KSPCGLSFETIDPDual-Primal (DP) Finite Element Tearing and Interconnect (FETI) method.
KSPFETIDPHPDDMInterface with the HPDDM library. This KSP may be used to further select methods that are currently not implemented natively in PETSc, e.g., GCRODR, a recycled Krylov method which is similar to KSPLGMRES.
KSPHPDDM
See also
Attributes Summary
Object
BCGSof typestrObject
BCGSLof typestrObject
BICGof typestrObject
CGof typestrObject
CGLSof typestrObject
CGNEof typestrObject
CGSof typestrObject
CHEBYSHEVof typestrObject
CRof typestrObject
DGMRESof typestrObject
FBCGSof typestrObject
FBCGSRof typestrObject
FCGof typestrObject
FETIDPof typestrObject
FGMRESof typestrObject
GCRof typestrObject
GLTRof typestrObject
GMRESof typestrObject
GROPPCGof typestrObject
HPDDMof typestrObject
IBCGSof typestrObject
LCDof typestrObject
LGMRESof typestrObject
LSQRof typestrObject
MINRESof typestrObject
NASHof typestrObject
NONEof typestrObject
PGMRESof typestrObject
PIPEBCGSof typestrObject
PIPECGof typestrObject
PIPECG2of typestrObject
PIPECGRRof typestrObject
PIPECRof typestrObject
PIPEFCGof typestrObject
PIPEFGMRESof typestrObject
PIPEGCRof typestrObject
PIPELCGof typestrObject
PIPEPRCGof typestrObject
PREONLYof typestrObject
PYTHONof typestrObject
QCGof typestrObject
QMRCGSof typestrObject
RICHARDSONof typestrObject
STCGof typestrObject
SYMMLQof typestrObject
TCQMRof typestrObject
TFQMRof typestrObject
TSIRMof typestrAttributes Documentation