Thanks, Matt and Umut. I will look into these possibilities.

Shao-Ching

On Wed, Oct 27, 2010 at 4:29 PM, Umut Tabak <u.tabak at tudelft.nl> wrote:
> Matthew Knepley wrote:
>>
>>
>> I think the right thing to do is build M from scratch. Its unlikely that
>> you actually
>> want the parallel matrix to be partitioned this way.
>>
>> ? Matt
>>
>
> Or you can take a look at PetscExt, thanks to Dave May, which provides
> functions to build block matrices. I also learned about that from here.
> Best,
> Umut
>

Reply via email to