[Trilinos-Users] [EXTERNAL] Anasazi solver with ML preconditioner won't converge

[Thornquist, Heidi K]

Hi Ray and Andris,

I believe that the test matrix is the BCSSTK01/BCSSTM01 matrix pair from
the BCSSTRUC1 set in Matrix Market.  That was just a guess from the
statistics.  The convergence of this matrix pair can be affected by many
things, including which eigenvalues you are looking for, the size of
subspace, and the number of allowed iterations.  I think that the
inclusion of the ML preconditioner may be obfuscating the issue.  It's
easiest to reduce the complexity of the possibilities (i.e. remove the
preconditioner) and see if you can something to converge for the simplest
case, especially when the problem is so small.  Then start adding
preconditioning, spectral transformations, etc. to speed up the solve and
build more intuition about the solver.

Since you are looking for the smallest eigenvalues and the system is SPD,
have you also considered using LOBPCG?  Is this exercise with the Matrix
Market test matrices in anticipation of a larger generalized eigenproblem
with SPD matrices?  We can give guidance on how to solve this
eigenproblem, but knowing what your final target will help us out.

Thanks,
Heidi

-- 

Heidi K. Thornquist
Electrical Models & Simulation
Sandia National Laboratories
Albuquerque, NM  87185-1323





On 2/1/17, 6:46 PM, "Trilinos-Users on behalf of Raymond Tuminaro"
<trilinos-users-bounces at trilinos.org on behalf of rstumin at sandia.gov>
wrote:

>Andris,
>    I can't tell if it is an ML problem or an Anasazi problem. I've
>actually never combined ML with Anasazi ... I'm not
>sure who would be best to address this?  Could you add the following
>after the SetDefaults() line in your code:
>
>precOptions.set("ML output",10);
>
>
>This should at least get ML to spit out some output. Could you tell us
>more about where your matrix comes from. Is it a
>discretization of a PDE system? If so, what type of PDE?
>
>-Ray
>
>
>On 01/27/17 02:08, Andris Freimanis wrote:
>> Hello,
>>
>> I'm writing a small program that will compute few smallest eigenpairs.
>>I stiched it together from anasazi and ML
>> examples, tested it with 48x48 mass and stiffness matrices from matrix
>>market website, but can't get the solver to
>> converge.
>>
>> I attached a log file with debug info and source code.
>>
>> Can you help me?
>>
>>
>>
>> This body part will be downloaded on demand.
>>
>
>-- 
>Ray Tuminaro                      phone: (925) 294-2564
>MS 9159                           fax:   (925) 294-2234
>Sandia National Laboratories      email: rstumin at sandia.gov
>PO Box 969                        http://www.cs.sandia.gov/~rstumin
>Livermore, CA 94551
>
>
>_______________________________________________
>Trilinos-Users mailing list
>Trilinos-Users at trilinos.org
>https://trilinos.org/mailman/listinfo/trilinos-users