<div class="gmail_quote">On Fri, Jun 1, 2012 at 9:27 AM, Jozsef Bakosi <span dir="ltr"><<a href="mailto:jbakosi@lanl.gov" target="_blank">jbakosi@lanl.gov</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div id=":3sh">I mean I grab one node with a Dirichlet condition and apply a Neumann on<br>
all boundaries, i.e. the problem is well-posed: the additive constant,<br>
up to which the solution is determined, is accounted for.<br></div></blockquote><div><br></div><div>This is fragile because smoothed aggregation can "coarsen out" this constraint. We usually recommend informing the KSP of the null space. See the section of the user's manual on solving singular systems.</div>
<div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div id=":3sh">
<div class="im"><br>
> > I did another test and set the PETSC_OPTIONS environment variable to<br>
> > '-options_table -options_left', then I get:<br>
> ><br>
> > ...<br>
> ><br>
> > [49]PETSC ERROR: Zero pivot row 537 value 9.36276e-13 tolerance 1e-12!<br>
> ><br>
><br>
> This is an error. How do you get to the output shown below?<br>
<br>
</div>I'm not sure if I understand your question. I set PETSC_OPTIONS<br>
environment variable to '-options_table -options_left'.<br></div></blockquote><div><br></div><div>You must not have been checking errors or something because the code evidently continued past the error, getting all the way to PetscFinalize().</div>
<div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div id=":3sh"><div><div class="h5">
</div></div>It seems like your suggestion<br>
<div class="im"><br>
-mg_coarse_redundant_pc_factor_zeropivot <2.22045e-14>: Pivot is considered zero if less than (PCFactorSetZeroPivot)<br>
<br>
</div>remedies the problem as I'm not getting the zeropivot error anymore. I<br>
guess the problem was at the coarsest MG level. I set<br>
mg_coarse_redundent_pc_factor to 1.0e-20 and that helped.<br></div></blockquote></div><br><div>Yes, but is the convergence good now?</div>