Hello,

I'm facing the same problem and since this post is 4 years old, I wonder if 
Arpack and SparseDirectUMFPACK still do not support 
TrilinosWrappers::BlockSparseMatrix. 

In my case, I need to calculate the minimum 10 eigenvalues of a large 
TrilinosWrappers::BlockSparseMatrix and since Arpack has the shifted 
eigenvalue method it would be very efficient to use it. 

If you have any other suggestions to solve this problem using other 
packages, I'd be glad to hear them.

Thank you for your time.

Best,
Mert Can



On Wednesday, July 1, 2015 at 9:42:02 AM UTC-7, Wolfgang Bangerth wrote:
>
> On 05/18/2015 06:50 AM, Shiraz Farouq wrote: 
> > 
> > 
> > I am not sure, but it looks that Arpack and SparseDirectUMFPACK do not 
> support 
> > Trilinos::BlockVector and  TrilinosWrappers::BlockSparseMatrix data 
> structures 
> > ! So, how am I supposed to solve for the eigenvalues for a block sparse 
> system 
> > in deal.ii ! 
> > 
> > Any hints to get around it in deal.ii ! 
>
> Shiraz, you may have long found a different solution (this question sat 
> unanswered for a long time), but I think the answer is that Arpack and 
> UMFPACK 
> simply don't go together with Trilinos data structures. 
>
> There are, however, corresponding classes in Trilinos that you could use. 
> I 
> think we have interfaces to the direct solvers in Trilinos (I'm not 
> entirely 
> sure, though), and there are probably also eigenvalue solvers (though we 
> don't 
> have wrappers for those). 
>
> In the end, your problem illustrates the curse of interfacing with 
> external 
> packages: it starts to get difficult when you want to mix and match them, 
> because supporting N such packages requires, in the worst case, writing 
> N^2 
> wrappers if you want to allow all combinations to work with each other. 
> That's 
> not practically feasible. 
>
> Best 
>   W. 
>
>
> -- 
> ------------------------------------------------------------------------ 
> Wolfgang Bangerth               email:            [email protected] 
> <javascript:> 
>                                  www: http://www.math.tamu.edu/~bangerth/ 
>
>

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to