[Trilinos-Users] result of belos iterative solver depends on number of processors

[Joachim Wuttke]

> ... for certain input data the number of processes determines
> whether or not the solver converges. Is this normal behavior, or
> indicative of yet another bug?

using CrsMatrix2MATLAB from
/trilinos/11.12.1/didasko/examples/epetra/ex15.cpp,
I find that the number of processors affects the
internal order of the sparse matrix:

with one processor
...
A(5,2) = -4;
A(5,5) = 5.41421;
A(5,6) = 2.82843;
A(5,7) = -1.41421;
A(5,8) = 0.585786;
...

with two processors
...
A(5,5) = 5.41421;
A(5,6) = 2.82843;
A(5,7) = -1.41421;
A(5,8) = 0.585786;
A(5,2) = -4;
...

Why does this happen?
And how does it cause Belos::BlockGmresSolMgr to fail in case 2?

- Joachim

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5110 bytes
Desc: S/MIME Cryptographic Signature
URL: <https://trilinos.org/pipermail/trilinos-users/attachments/20170627/9cb78b7e/attachment.p7s>