I apparently was explicitly changing the PC in my code after PCSetFromOptions, so the options weren't getting set properly (which I found out by using -snes_view). Sorry to bother.
Andrew ---- Original message ---- >Date: Fri, 8 Feb 2008 12:02:45 -0600 >From: "Matthew Knepley" <knepley at gmail.com> >Subject: Re: sub_pc lu zero pivot >To: petsc-users at mcs.anl.gov > >We would like to see: > > 1) The complete error message > > 2) The output of -snes_view so we can see exactly what solver setup you have > > Thanks, > > Matt > >On Feb 8, 2008 11:46 AM, Andrew T Barker <Andrew.Barker at colorado.edu> wrote: >> >> For my code on a single processor, using the options >> >> -ksp_type gmres -pc_type lu >> >> works fine, but >> >> -ksp_type gmres -pc_type asm -sub_pc_type lu >> >> produces a zero pivot. On one processor I would expect these two to be >> identical, and in fact when I print out the asm submatrices (there is only >> one) in the second case and the original matrix in the first case, they are >> identical. Even stranger is that >> >> -ksp_type gmres -pc_type lu -sub_pc_type lu >> >> also produces a zero pivot. I would expect it to ignore the sub_pc in this >> case. >> >> Thanks for any help, >> >> Andrew >> >> --- >> Andrew T. Barker >> andrew.barker at colorado.edu >> Applied Math Department >> University of Colorado, Boulder >> 526 UCB, Boulder, CO 80309-0526 >> >> > > > >-- >What most experimenters take for granted before they begin their >experiments is infinitely more interesting than any results to which >their experiments lead. >-- Norbert Wiener >
