<html>
<head>
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">Thanks a lot for your answers.<br>
<br>
I ran it with<tt><br>
</tt>
<blockquote><tt>-ksp_type gmres -pc_type hypre -pc_hypre_type
euclid</tt><br>
</blockquote>
and it worked very well. Thanks.<br>
<br>
I then tried to use boomeramg as a preconditioner coupled with
Richardson but I was not successful, it failed to solve the system
and returned nans.<br>
<blockquote><tt>-ksp_type richardson -pc_type hypre -pc_hypre_type
boomeramg -pc_hypre_boomeramg_relax_type_all SOR/Jacobi
-pc_hypre_boomeramg_print_debug -ksp_view
-ksp_monitor_true_residual</tt><br>
</blockquote>
and i got the following<br>
<tt><br>
</tt>
<blockquote><tt>�===== Proc = 0���� Level = 0� =====</tt><br>
<tt>Proc = 0��� Coarsen 1st pass = 0.000000</tt><br>
<tt>Proc = 0��� Coarsen 2nd pass = 0.000000</tt><br>
<tt>Proc = 0��� Initialize CLJP phase = 0.000000</tt><br>
<tt>Proc = 0� iter 1� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 18308 nc_offd
= 0</tt><br>
<tt>Proc = 0� iter 2� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 7 nc_offd = 0</tt><br>
<br>
<tt>�===== Proc = 0���� Level = 1� =====</tt><br>
<tt>Proc = 0��� Coarsen 1st pass = 0.010000</tt><br>
<tt>Proc = 0��� Coarsen 2nd pass = 0.000000</tt><br>
<tt>Proc = 0��� Initialize CLJP phase = 0.000000</tt><br>
<tt>Proc = 0� iter 1� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 8725 nc_offd
= 0</tt><br>
<tt>Proc = 0� iter 2� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 16 nc_offd =
0</tt><br>
<br>
<tt>�===== Proc = 0���� Level = 2� =====</tt><br>
<tt>Proc = 0��� Coarsen 1st pass = 0.000000</tt><br>
<tt>Proc = 0��� Coarsen 2nd pass = 0.000000</tt><br>
<tt>Proc = 0��� Initialize CLJP phase = 0.000000</tt><br>
<tt>Proc = 0� iter 1� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 4721 nc_offd
= 0</tt><br>
<tt>Proc = 0� iter 2� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 13 nc_offd =
0</tt><br>
<tt>Proc = 0� iter 3� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 4 nc_offd = 0</tt><br>
<br>
<tt>�===== Proc = 0���� Level = 3� =====</tt><br>
<tt>Proc = 0��� Coarsen 1st pass = 0.000000</tt><br>
<tt>Proc = 0��� Coarsen 2nd pass = 0.000000</tt><br>
<tt>Proc = 0��� Initialize CLJP phase = 0.000000</tt><br>
<tt>Proc = 0� iter 1� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 2495 nc_offd
= 0</tt><br>
<tt>Proc = 0� iter 2� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 22 nc_offd =
0</tt><br>
<tt>Proc = 0� iter 3� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 4 nc_offd = 0</tt><br>
<br>
<tt>�===== Proc = 0���� Level = 4� =====</tt><br>
<tt>Proc = 0��� Coarsen 1st pass = 0.000000</tt><br>
<tt>Proc = 0��� Coarsen 2nd pass = 0.000000</tt><br>
<tt>Proc = 0��� Initialize CLJP phase = 0.000000</tt><br>
<tt>Proc = 0� iter 1� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 1337 nc_offd
= 0</tt><br>
<tt>Proc = 0� iter 2� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 13 nc_offd =
0</tt><br>
<tt>Proc = 0� iter 3� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 2 nc_offd = 0</tt><br>
<br>
<tt>�===== Proc = 0���� Level = 5� =====</tt><br>
<tt>Proc = 0��� Coarsen 1st pass = 0.000000</tt><br>
<tt>Proc = 0��� Coarsen 2nd pass = 0.000000</tt><br>
<tt>Proc = 0��� Initialize CLJP phase = 0.000000</tt><br>
<tt>Proc = 0� iter 1� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 695 nc_offd =
0</tt><br>
<tt>Proc = 0� iter 2� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 3 nc_offd = 0</tt><br>
<br>
<tt>�===== Proc = 0���� Level = 6� =====</tt><br>
<tt>Proc = 0��� Coarsen 1st pass = 0.000000</tt><br>
<tt>Proc = 0��� Coarsen 2nd pass = 0.000000</tt><br>
<tt>Proc = 0��� Initialize CLJP phase = 0.000000</tt><br>
<tt>Proc = 0� iter 1� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 343 nc_offd =
0</tt><br>
<tt>Proc = 0� iter 2� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 21 nc_offd =
0</tt><br>
<tt>Proc = 0� iter 3� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 2 nc_offd = 0</tt><br>
<br>
<tt>�===== Proc = 0���� Level = 7� =====</tt><br>
<tt>Proc = 0��� Coarsen 1st pass = 0.000000</tt><br>
<tt>Proc = 0��� Coarsen 2nd pass = 0.000000</tt><br>
<tt>Proc = 0��� Initialize CLJP phase = 0.000000</tt><br>
<tt>Proc = 0� iter 1� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 174 nc_offd =
0</tt><br>
<tt>Proc = 0� iter 2� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 15 nc_offd =
0</tt><br>
<tt>Proc = 0� iter 3� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 2 nc_offd = 0</tt><br>
<br>
<tt>�===== Proc = 0���� Level = 8� =====</tt><br>
<tt>Proc = 0��� Coarsen 1st pass = 0.000000</tt><br>
<tt>Proc = 0��� Coarsen 2nd pass = 0.000000</tt><br>
<tt>Proc = 0��� Initialize CLJP phase = 0.000000</tt><br>
<tt>Proc = 0� iter 1� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 81 nc_offd =
0</tt><br>
<tt>Proc = 0� iter 2� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 13 nc_offd =
0</tt><br>
<br>
<tt>�===== Proc = 0���� Level = 9� =====</tt><br>
<tt>Proc = 0��� Coarsen 1st pass = 0.000000</tt><br>
<tt>Proc = 0��� Coarsen 2nd pass = 0.000000</tt><br>
<tt>Proc = 0��� Initialize CLJP phase = 0.000000</tt><br>
<tt>Proc = 0� iter 1� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 37 nc_offd =
0</tt><br>
<tt>Proc = 0� iter 2� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 6 nc_offd = 0</tt><br>
<br>
<tt>�===== Proc = 0���� Level = 10� =====</tt><br>
<tt>Proc = 0��� Coarsen 1st pass = 0.000000</tt><br>
<tt>Proc = 0��� Coarsen 2nd pass = 0.000000</tt><br>
<tt>Proc = 0��� Initialize CLJP phase = 0.000000</tt><br>
<tt>Proc = 0� iter 1� comm. and subgraph update = 0.000000</tt><br>
<tt>Proc = 0��� CLJP phase = 0.000000 graph_size = 11 nc_offd =
0</tt><br>
<br>
<br>
<tt>� 0 KSP preconditioned resid norm 7.299769365830e+14 true
resid norm 8.197927963033e-03 ||r(i)||/||b||
1.000000000000e+00</tt><br>
<tt>� 1 KSP preconditioned resid norm 2.319459389445e+28 true
resid norm 6.152576199945e+12 ||r(i)||/||b||
7.505038136086e+14</tt><br>
<tt>KSP Object: 1 MPI processes</tt><br>
<tt>� type: richardson</tt><br>
<tt>��� Richardson: damping factor=1</tt><br>
<tt>� maximum iterations=90, initial guess is zero</tt><br>
<tt>� tolerances:� relative=0.1, absolute=1e-50,
divergence=100000</tt><br>
<tt>� left preconditioning</tt><br>
<tt>� using PRECONDITIONED norm type for convergence test</tt><br>
<tt>PC Object: 1 MPI processes</tt><br>
<tt>� type: hypre</tt><br>
<tt>��� HYPRE BoomerAMG preconditioning</tt><br>
<tt>��� HYPRE BoomerAMG: Cycle type V</tt><br>
<tt>��� HYPRE BoomerAMG: Maximum number of levels 25</tt><br>
<tt>��� HYPRE BoomerAMG: Maximum number of iterations PER hypre
call 1</tt><br>
<tt>��� HYPRE BoomerAMG: Convergence tolerance PER hypre call 0</tt><br>
<tt>��� HYPRE BoomerAMG: Threshold for strong coupling 0.25</tt><br>
<tt>��� HYPRE BoomerAMG: Interpolation truncation factor 0</tt><br>
<tt>��� HYPRE BoomerAMG: Interpolation: max elements per row 0</tt><br>
<tt>��� HYPRE BoomerAMG: Number of levels of aggressive
coarsening 0</tt><br>
<tt>��� HYPRE BoomerAMG: Number of paths for aggressive
coarsening 1</tt><br>
<tt>��� HYPRE BoomerAMG: Maximum row sums 0.9</tt><br>
<tt>��� HYPRE BoomerAMG: Sweeps down�������� 1</tt><br>
<tt>��� HYPRE BoomerAMG: Sweeps up���������� 1</tt><br>
<tt>��� HYPRE BoomerAMG: Sweeps on coarse��� 1</tt><br>
<tt>��� HYPRE BoomerAMG: Relax down��������� SOR/Jacobi</tt><br>
<tt>��� HYPRE BoomerAMG: Relax up����������� SOR/Jacobi</tt><br>
<tt>��� HYPRE BoomerAMG: Relax on coarse����
Gaussian-elimination</tt><br>
<tt>��� HYPRE BoomerAMG: Relax weight� (all)����� 1</tt><br>
<tt>��� HYPRE BoomerAMG: Outer relax weight (all) 1</tt><br>
<tt>��� HYPRE BoomerAMG: Using CF-relaxation</tt><br>
<tt>��� HYPRE BoomerAMG: Measure type������� local</tt><br>
<tt>��� HYPRE BoomerAMG: Coarsen type������� Falgout</tt><br>
<tt>��� HYPRE BoomerAMG: Interpolation type� classical</tt><br>
<tt>� linear system matrix = precond matrix:</tt><br>
<tt>� Matrix Object:�� 1 MPI processes</tt><br>
<tt>��� type: seqbaij</tt><br>
<tt>��� rows=22905, cols=22905, bs=5</tt><br>
<tt>��� total: nonzeros=785525, allocated nonzeros=785525</tt><br>
<tt>��� total number of mallocs used during MatSetValues calls
=0</tt><br>
<tt>������� block size is 5</tt></blockquote>
Do you guys have any suggestion?� Is it possible that I am haven't
initialized boomeramg properly? Or it is just my system equations
that can not be solved by AMG?<br>
<br>
Sincerely,<br>
Dario<br>
<br>
<br>
<br>
<br>
On 05/16/2014 11:54 AM, Barry Smith wrote:<br>
</div>
<blockquote
cite="mid:B37A27EA-6E44-4036-AF2F-8E8EC34E3E68@mcs.anl.gov"
type="cite">
<pre wrap="">
On May 16, 2014, at 10:46 AM, Dario Isola <a class="moz-txt-link-rfc2396E" href="mailto:dario.isola@newmerical.com"><dario.isola@newmerical.com></a> wrote:
</pre>
<blockquote type="cite">
<pre wrap="">Dear all,
I am investigating the use of hypre+petsc. I was able to successfully configure, install, compile petsc 3.3 with the external package for hypre.
I tried to run it with the following options
-pc_type hypre -pc_type_hypre pilut -ksp_type richardson
and, although he did not complain, it does not solve the system either.
</pre>
</blockquote>
<pre wrap="">
Do you meaning it did not converge? At first always run with -ksp_view (or -snes_view if using snes or -ts_view if using ts) and -ksp_monitor_true_residual to see what is going on.
</pre>
<blockquote type="cite">
<pre wrap="">-pc_type_hypre pilut
</pre>
</blockquote>
<pre wrap="">
is wrong it is -pc_hypre_type pilut
Note that pilut will generally not work with Richardson you need a �real� Krylov method like GMRES.
Also the ilu type preconditioners don�t scale particularly well though occasionally they can be fine.
</pre>
<blockquote type="cite">
<pre wrap="">
To what extent is hypre supported by petsc? More specifically, what kind of matrices?
</pre>
</blockquote>
<pre wrap="">
If it cannot handle the matrix type it would give an error message. Hypre uses a format like AIJ so you should use AIJ. Note that you can make the matrix type a runtime option so you don�t have to compile in that it is BAIJ.
</pre>
<blockquote type="cite">
<pre wrap="">I am using a baij matrix.
Thanks in advance,
D
</pre>
</blockquote>
<pre wrap="">
</pre>
</blockquote>
<br>
</body>
</html>