[petsc-users] Curiosity about MatSetOptionsPrefix on a_11 in PCSetUp_FieldSplit

[Eric Chamberland]

On 09/11/2014 12:26 PM, Matthew Knepley wrote:
> I am trying to understand the problem. This matrix is created by by the
> MatGetSubMatrix() call.

If this is right, then I misunderstood the effect of MatGetSubMatrix on 
a MatNest.  We created a MatNest which holds the a_11 matrix...

>
>    1) What options are you trying to pass to it?

mat_type sbaij
mat_block_size  3 (not in this example however)

>
>    2) Why would they interfere with the KSP options?

I didn't mentioned that they could interfere, just that the name was 
changed and that it was "curious" from my point of view.

But if the sub-matrix extracted from the matnest is not the a_11 matrix 
we gave to the matnest, then maybe we did something wrong?  Or you can't 
verify that the extracted matrix did exists before the extraction?

Again, the options we gave are treated correctly, since we create the 
matrix and give it to the nest.  The "curious" thing is when petsc 
retreive (the same?) matrix and changes its name...

Thanks again!

Eric