[petsc-users] Large rectangular Dense Transpose multiplication with sparse

[Hong]

Ghosh:
>
>
>    Thanks for the suggestion of Elemental. However I need to have the
> dense matrix have the same parallel communicator as the sparse matrix.
> If I use elemental, then the numbering changes, will I be able to multiply
> with the sparse matrix which has a different numbering scheme and
> communicator?
>
> In any case I would want the resultant matrices A and M to be elemental
> type since I need to solve a dense eigenvalue problem.
>

If your final goal is solving a  dense eigenvalue problem, and the only
sparse matrix is a banded matrix H, Elemental is a suitable package.

I just added HermitianGenDefiniteEig() into petsc(master)-elemental(v0.84)
interface and tested on some large matrices - seems quite efficient. You
may take a look at ~petsc(master)/src/mat/examples/tests/ex174.cxx

I'm not sure if the latest Elemental supports banded dense matrix - if not,
ask Jack to added :-)

Hong

>
> ------------------------------
> *From: *"Hong" <hzhang at mcs.anl.gov>
> *To: *"Swarnava Ghosh" <sghosh2012 at gatech.edu>
> *Cc: *"PETSc users list" <petsc-users at mcs.anl.gov>
> *Sent: *Thursday, February 5, 2015 8:22:13 PM
> *Subject: *Re: [petsc-users] Large rectangular Dense Transpose
> multiplication with sparse
>
>
> Swarnava:
> The matrix product A will be a dense matrix. You may consider using
> Elemental package for such matrix product.
>
> Hong
>
>>
>> Dear all,
>>
>>   I am trying to compute matrices A = transpose(R)*H*R and M =
>> transpose(R)*R where
>>   H is a sparse (banded) matrix in MATMPIAIJ format (5 million x 5
>> million total size)
>>   R is a MPI dense matrix of size 5 million x 2000.
>>
>>   I tried 1) MatPtAP - Failed, realized this only works for pairs of AIJ
>> matrices
>>           2) First multiplying H*R and storing in 5 million x 2000 MPI
>> dense. Then MatTranspose of R and multiplying the transposed R with 5
>> million x 2000 dense. This multiplication fails.
>>
>>    Could someone please suggest a way of doing this.
>>
>> Regards,
>> Swarnava
>> --
>> Swarnava Ghosh
>>
>
>
>
>
> --
> Swarnava Ghosh
> PhD Candidate,
> Structural Engineering, Mechanics and Materials
> School of Civil and Environmental Engineering
> Georgia Institute of Technology
> Atlanta, GA 30332
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20150206/da5a8259/attachment.html>