[petsc-users] help needed for -snes_mf_operator

Dominik Szczerba dominik at itis.ethz.ch
Wed Apr 25 13:19:23 CDT 2012


> Notice that there is no increase in the residual as is required. Notice also
> that
> there is no decrease. This must mean that the preconditioner is rank
> deficient,
> and I think it probably means that it is 0 since I find it hard to believe
> that P x
> produces no component in the b direction.

Should this be the case, how would I get correct results for several
benchmarks without using the option?

Dominik


More information about the petsc-users mailing list