[Trilinos-Users] Big Big Error on generating Tridiag matrix, am I wrong?

Heroux, Michael A maherou at sandia.gov
Fri Oct 30 10:30:11 MDT 2009


John,

Thanks for reporting this.  You are right.  I have fixed this bug and the fix will go out in the next minor release of Trilinos 10.

Thanks again.

Mike


On 10/29/09 10:02 PM, "john" <ax.equal.b at gmail.com> wrote:

Dear all,

I updated to Trililnos 10.0.2 yesterday.
But I found that the generated Tridiag matrix in Both triutils and galeri packages are wrong.

In source Trilinos_Util_CrsMatrixGallery.cpp reads following lines
...
    } else {
      // off-diagonal for internal row
      Indices[0] = MyGlobalElements_[i]-1;
      Values[1] = b_;// Should the index be Values[0] = b_
      Indices[1] = MyGlobalElements_[i]+1;
      Values[0] = c_;// Should the index be Values[1] = c_
      NumEntries = 2;
    }
...
Here b_ is the subdiagonal and c_ is the superdiagonal.

The same code appears in Galeri_Tridiag.h.

Does my observation is wrong?

John.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://software.sandia.gov/pipermail/trilinos-users/attachments/20091030/ebcf15ce/attachment.html 


More information about the Trilinos-Users mailing list