MFNConvergedReason#

Reason a matrix function iteration was determined to have converged or diverged.

Synopsis#

typedef enum {/* converged */
              MFN_CONVERGED_TOL                =  1,
              MFN_CONVERGED_ITS                =  2,
              /* diverged */
              MFN_DIVERGED_ITS                 = -1,
              MFN_DIVERGED_BREAKDOWN           = -2,
              MFN_CONVERGED_ITERATING          =  0} MFNConvergedReason;

Values#

MFN_CONVERGED_TOL - converged up to tolerance
MFN_CONVERGED_ITS - solver completed the requested number of steps
MFN_DIVERGED_ITS - exceeded the maximum number of allowed iterations
MFN_DIVERGED_BREAKDOWN - generic breakdown in method

Note#

Basic solvers (e.g., unrestarted Krylov iterations) cannot determine if the computation is accurate up to the requested tolerance. In that case, the converged reason is set to MFN_CONVERGED_ITS if the requested number of steps (for instance, the ncv value in unrestarted Krylov methods) have been completed successfully.

Level#

intermediate

Location#

include/slepcmfn.h

Examples#

src/mfn/tutorials/ex37.c
src/mfn/tutorials/ex23.c
src/mfn/tutorials/ex23f.F90

Index of all MFN routines Table of Contents for all manual pages Index of all manual pages