[petsc-dev] Problem with conjugate gradient solver

[Keita Teranishi]

The default solver is GMRES, which can solve indefinite systems.

================================
 Keita Teranishi
 Scientific Library Group
 Cray, Inc.
 keita at cray.com
================================

From: petsc-dev-bounces at mcs.anl.gov [mailto:petsc-dev-bounces at mcs.anl.gov] On Behalf Of Anush Krishnan
Sent: Thursday, August 26, 2010 12:08 PM
To: petsc-dev at mcs.anl.gov
Subject: Re: [petsc-dev] Problem with conjugate gradient solver


Message: 2
Date: Tue, 24 Aug 2010 10:56:53 -0500
From: Barry Smith <bsmith at mcs.anl.gov<mailto:bsmith at mcs.anl.gov>>
Subject: Re: [petsc-dev] Problem with conjugate gradient solver
To: For users of the development version of PETSc
       <petsc-dev at mcs.anl.gov<mailto:petsc-dev at mcs.anl.gov>>,        Victor Eijkhout <eijkhout at tacc.utexas.edu<mailto:eijkhout at tacc.utexas.edu>>
Message-ID: <A4C6A1ED-CDE7-4653-8046-C2797AD5AF31 at mcs.anl.gov<mailto:A4C6A1ED-CDE7-4653-8046-C2797AD5AF31 at mcs.anl.gov>>
Content-Type: text/plain; charset="us-ascii"


  Well you are doing nothing wrong. When I run the code I get the same results as you. I dump the matrix into Matlab and it has a negative eigenvalue so the example is wrong. The code is handling it correctly.

>> eig(full(Mat_0))

ans =

  -0.8042
   0.1716
   0.6489
   1.0000
   1.0000
   1.0000
   1.0000

So are you saying that trying to solve the system of equations in the example will always result in divergence? I tried the same program with a different system that I know has a solution, and even that diverges. I get a proper solution if I use the default KSP solver.

Anush


  I am cc:iing Victor who supposedly wrote the original example.

  Barry

Attached is the matrix saved with -mat_view_matlab
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20100826/0e6594f5/attachment.html>