[Trilinos-Users] AztecOO and Problem Size

[Ammar T. Al-Sayegh]

----- Original Message ----- 
From: "Mike Heroux" <maherou at sandia.gov>
To: "'Ammar T. Al-Sayegh'" <alsayegh at purdue.edu>; <trilinos-users at software.sandia.gov>
Sent: Sunday, November 20, 2005 12:49 AM
Subject: RE: [Trilinos-Users] AztecOO and Problem Size


> [...] The error you are seeing comes from
> AztecOO seeing the difference between the implicit residual norm and the
> true norm, which is computed at the end of the solve.

I can see that now. So even though it's displayed in the
output, the actual residual is not known when convergence
creterion is tested--it's tested against recursive redsidual.
When convergence is confirmed based on recursive residual,
the actual residual is calculated, from which the norm is
computed and compared to the requested norm. If it turns
out to be greater than the requested norm, it gives out
this error message. Am I right?
 
> There is a variety of convergence tests available for AztecOO.  The default
> is the implicit 2-norm of the residual, divided by the 2-norm of the initial
> residual.  I don't know which one you are using, or if you have a non-zero
> initial guess, so I can't give you precise advice, but it is possible that
> setting AZ_conv to AZ_noscaled might help. [...]

I tried, it helped having more elements before hitting
this error again. I see that the error message suggests
using the solution is an initial guess. Instead of
exiting with an error and suggesting to restart with
the solution as an initial guess manually, is there
an option to let AztecOO do that automatically? i.e.,
whenver it converges but the calculated norm turns
out to be greater than requested norm, then let AztecOO
automatically recompute the same problem using the
last result as an initial guess?

Thank you for your prompt followup.


-ammar