<div dir="ltr">This error seems to be coming from the computation of the extreme eigenvalues of the matrix for smoothing in smoothed aggregation.<div><br></div><div>Are you getting good solutions with hypre? This error looks like it might just be the first place where a messed up matrix fails in GAMG.</div></div><div class="gmail_extra"><br><div class="gmail_quote">On Sat, Jan 7, 2017 at 10:23 PM, Matthew Knepley <span dir="ltr"><<a href="mailto:knepley@gmail.com" target="_blank">knepley@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div class="gmail_extra"><div class="gmail_quote"><span class="">On Sat, Jan 7, 2017 at 7:38 PM, Manuel Valera <span dir="ltr"><<a href="mailto:mvalera@mail.sdsu.edu" target="_blank">mvalera@mail.sdsu.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Ok great, i tried those command line args and this is the result:<div><br></div><div>when i use -pc_type gamg:</div><div><br></div><div><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: --------------------- Error Message ------------------------------<wbr>------------------------------<wbr>--</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: Petsc has generated inconsistent data</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: Have un-symmetric graph (apparently). Use '-pc_gamg_sym_graph true' to symetrize the graph or '-pc_gamg_threshold -1.0' if the matrix is structurally symmetric.</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: See <a href="http://www.mcs.anl.gov/petsc/documentation/faq.html" target="_blank">http://www.mcs.anl.gov/petsc/d<wbr>ocumentation/faq.html</a> for trouble shooting.</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: Petsc Release Version 3.7.4, unknown </span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: ./ucmsMR on a arch-linux2-c-debug named ocean by valera Sat Jan  7 17:35:05 2017</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: Configure options --with-cc=gcc --with-cxx=g++ --with-fc=gfortran --download-fblaslapack --download-mpich --download-hdf5 --download-netcdf --download-hypre --download-metis --download-parmetis --download-trillinos</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: #1 smoothAggs() line 462 in /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/pc/impls/gamg/agg.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: #2 PCGAMGCoarsen_AGG() line 998 in /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/pc/impls/gamg/agg.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: #3 PCSetUp_GAMG() line 571 in /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/pc/impls/gamg/gamg.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: #4 PCSetUp() line 968 in /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/pc/interface/precon.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: #5 KSPSetUp() line 390 in /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/ksp/interface/itfunc.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">application called MPI_Abort(comm=0x84000002, 77) - process 1</span></p></div><div><span style="font-variant-ligatures:no-common-ligatures"><br></span></div><div><span style="font-variant-ligatures:no-common-ligatures"><br></span></div><div><span style="font-variant-ligatures:no-common-ligatures">when i use </span>-pc_type gamg and -<span style="font-family:menlo;font-size:11px">pc_gamg_sym_graph true:</span></div></div></blockquote><div><br></div></span><div>Do everything Barry said.</div><div><br></div><div>However, I would like to track down this error. It could be a bug in our code. However, it appears to happen in the call</div><div>to LAPACK, so it could also be a problem with that library on your machine. Could you run this case in the debugger</div><div>and give the stack trace?</div><div><br></div><div>  Thanks,</div><div><br></div><div>     Matt</div><div><div class="h5"><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures"> -----------------------------<wbr>------------------------------<wbr>-------------</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[0]PETSC ERROR: Caught signal number 8 FPE: Floating Point Exception,probably divide by zero</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[0]PETSC ERROR: Try option -start_in_debugger or -on_error_attach_debugger</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[0]PETSC ERROR: or see <a href="http://www.mcs.anl.gov/petsc/documentation/faq.html#valgrind" target="_blank">http://www.mcs.anl.gov/petsc/d<wbr>ocumentation/faq.html#valgrind</a></span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[0]PETSC ERROR: or try <a href="http://valgrind.org" target="_blank">http://valgrind.org</a> on GNU/linux and Apple Mac OS X to find memory corruption errors</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: ------------------------------<wbr>------------------------------<wbr>------------</span></p></div><div><span style="font-variant-ligatures:no-common-ligatures"><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: ---------------------  Stack Frames ------------------------------<wbr>------</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: Note: The EXACT line numbers in the stack are not available,</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR:       INSTEAD the line number of the start of the function</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR:       is given.</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: [1] LAPACKgesvd line 42 /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/ksp/impls/gmres/gmreig.<wbr>c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: [1] KSPComputeExtremeSingularValue<wbr>s_GMRES line 24 /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/ksp/impls/gmres/gmreig.<wbr>c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: [1] KSPComputeExtremeSingularValue<wbr>s line 51 /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/ksp/interface/itfunc.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: [1] PCGAMGOptProlongator_AGG line 1187 /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/pc/impls/gamg/agg.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: [1] PCSetUp_GAMG line 472 /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/pc/impls/gamg/gamg.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: [1] PCSetUp line 930 /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/pc/interface/precon.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[1]PETSC ERROR: [1] KSPSetUp line 305 /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/ksp/interface/itfunc.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[0] PCGAMGOptProlongator_AGG line 1187 /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/pc/impls/gamg/agg.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[0]PETSC ERROR: [0] PCSetUp_GAMG line 472 /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/pc/impls/gamg/gamg.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[0]PETSC ERROR: [0] PCSetUp line 930 /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/pc/interface/precon.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[0]PETSC ERROR: [0] KSPSetUp line 305 /usr/dataC/home/valera/petsc/s<wbr>rc/ksp/ksp/interface/itfunc.c</span></p>
<p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">[0]PETSC ERROR: --------------------- Error Message ------------------------------<wbr>------------------------------<wbr>--</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures"><br></span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">when i use -pc_type hypre it actually shows something different on -ksp_view :</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures"><br></span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">KSP Object: 2 MPI processes</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">  type: gcr</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    GCR: restart = 30 </span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    GCR: restarts performed = 37 </span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">  maximum iterations=10000, initial guess is zero</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">  tolerances:  relative=1e-14, absolute=1e-50, divergence=10000.</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">  right preconditioning</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">  using UNPRECONDITIONED norm type for convergence test</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">PC Object: 2 MPI processes</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">  type: hypre</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG preconditioning</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Cycle type V</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Maximum number of levels 25</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Maximum number of iterations PER hypre call 1</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Convergence tolerance PER hypre call 0.</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Threshold for strong coupling 0.25</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Interpolation truncation factor 0.</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Interpolation: max elements per row 0</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Number of levels of aggressive coarsening 0</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Number of paths for aggressive coarsening 1</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Maximum row sums 0.9</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Sweeps down         1</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Sweeps up           1</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Sweeps on coarse    1</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Relax down          symmetric-SOR/Jacobi</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Relax up            symmetric-SOR/Jacobi</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Relax on coarse     Gaussian-elimination</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Relax weight  (all)      1.</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Outer relax weight (all) 1.</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Using CF-relaxation</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Not using more complex smoothers.</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Measure type        local</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Coarsen type        Falgout</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Interpolation type  classical</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: Using nodal coarsening (with HYPRE_BOOMERAMGSetNodal() 1</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    HYPRE BoomerAMG: HYPRE_BoomerAMGSetInterpVecVar<wbr>iant() 1</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">  linear system matrix = precond matrix:</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">  Mat Object:   2 MPI processes</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    type: mpiaij</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    rows=200000, cols=200000</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    total: nonzeros=3373340, allocated nonzeros=3373340</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">    total number of mallocs used during MatSetValues calls =0</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo">









































</p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">      not using I-node (on process 0) routines</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures"><br></span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures"><br></span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures">but still the timing is terrible.</span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures"><br></span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures"><br></span></p><p style="margin:0px;font-size:11px;line-height:normal;font-family:menlo"><span style="font-variant-ligatures:no-common-ligatures"><br></span></p></span></div></div><div class="m_2658651402405135346HOEnZb"><div class="m_2658651402405135346h5"><div class="gmail_extra"><br><div class="gmail_quote">On Sat, Jan 7, 2017 at 5:28 PM, Jed Brown <span dir="ltr"><<a href="mailto:jed@jedbrown.org" target="_blank">jed@jedbrown.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span>Manuel Valera <<a href="mailto:mvalera@mail.sdsu.edu" target="_blank">mvalera@mail.sdsu.edu</a>> writes:<br>
<br>
> Awesome Matt and Jed,<br>
><br>
> The GCR is used because the matrix is not invertible and because this was<br>
> the algorithm that the previous library used,<br>
><br>
> The Preconditioned im aiming to use is multigrid, i thought i configured<br>
> the hypre-boomerAmg solver for this, but i agree in that it doesn't show in<br>
> the log anywhere, how can i be sure is being used ? i sent -ksp_view log<br>
> before in this thread<br>
<br>
</span>Did you run with -pc_type hypre?<br>
<span><br>
> I had a problem with the matrix block sizes so i couldn't make the petsc<br>
> native multigrid solver to work,<br>
<br>
</span>What block sizes?  If the only variable is pressure, the block size<br>
would be 1 (default).<br>
<span><br>
> This is a nonhidrostatic pressure solver, it is an elliptic problem so<br>
> multigrid is a must,<br>
<br>
</span>Yes, multigrid should work well.<br>
</blockquote></div><br></div>
</div></div></blockquote></div></div></div><br><br clear="all"><span class=""><div><br></div>-- <br><div class="m_2658651402405135346gmail_signature" data-smartmail="gmail_signature">What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.<br>-- Norbert Wiener</div>
</span></div></div>
</blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><div class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><i>The secret to doing good research is always be a little underemployed. You waste years by not being able to waste hours</i> -- Amos Tversky<br></div></div></div></div>
</div>