[petsc-users] Use of PCApplyBAorAB with PC_SYMMETRIC side

Sylvain Mercier sylvainmercier85 at gmail.com
Mon Feb 17 04:42:47 CST 2014

Hello everyone,

I want to use PCApplyBAorAB symmetrically with a preconditioner from an ICC
However, I don't understand why the result is not the same if I use a
natural or a RCM ordering during the factorization.
This problem does not occur if I use this routine with PC_LEFT or PC_RIGHT

I have put in attachment a code to illustrate the problem.
I compute a symmetric matix A in R^{n*n} and a full rank matrix B in R^{m*n}
I compute an ICC factorization of  the SPD matrix A+B^TB to precondition A.

Thank you!

Sylvain Mercier
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140217/fb010eb6/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: testPCApplyBAorAB.F
Type: text/x-fortran
Size: 4981 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140217/fb010eb6/attachment.bin>

More information about the petsc-users mailing list