| 724f369d | 07-Mar-2014 |
Peter Brune <brune@mcs.anl.gov> |
SNESASPIN: Fix for the case of zero iterations
In the case where the absolute tolerance is satisfied in the first subSNES step, the operators are never set on the subKSP, leading to an error in the
SNESASPIN: Fix for the case of zero iterations
In the case where the absolute tolerance is satisfied in the first subSNES step, the operators are never set on the subKSP, leading to an error in the outer KSP iteration. For the sake of generality, have the ASPIN Jacobian call KSPSolve() instead of PCApply() when multiplying.
show more ...
|