[petsc-users] Solve equivalent real forms of complex linear systems.

[Jed Brown]

Ce Qin <qince168 at gmail.com> writes:

> Dear all,
>
> Now I want to solve a complex linear system arise in EM problems. And I
> also want to use TAO to solve a optimization problem simultaneously. So I
> must solve this complex linear system via its equivalent real form. To the
> best of my knowledge, the nested matrix and vector and fieldsplit
> preconditioner can do this. 

Don't use VecNest/MatNest/FieldSplit for this.  It will be worse for
this purpose and more complicated to implement.  Just use a block size
of 2.

We are aware of the problem that scalar=complex is not supported by TAO,
but we don't know what is the best approach to fix it.

> But I don't know how to setup the fieldsplit preconditioner. Does
> anyone have such experiences?
>
> BTW, I fount that the BCGS solver does not support nested vector because it
> does not have an axpbypcz operator.

Thanks for the report.  Will add this.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140218/a8435446/attachment-0001.pgp>