Dear all,

Just a few comments on the solvers, as I was searching for the one
that would match my problem best. In PySparse it is LU, the default
one, as was predictable cause I didn't see anything suspicious in the
solution. In Trilinos it turned out to be LinearBicgstabSolver,
although there are still some residuals left which I can push lower
decreasing tolerance and residual limit while sweeping.
LinearBicgstabSolve also turned out to be ~30% faster than Trilinos
Deffault one. However, all this may just be specific to my problem.

> Well, as we say there, you *might* wish to make these changes. There is no 
> requirement not to use LU, just that GMRES is >generally better in Trilinos. 
> GMRES is broken in PySparse, so we can't use it there (and PySparse LU is 
> quite fast).
I found GMRES working in PySparse. LU is excessively slow in Trilinos
and does not solve anything (solution variable is 0 at each time
step). I decided to try it first cause PySparse version is very good
for my problem. I didn't try the ones for symmetric matrix, cause mine
is asymmetric.

> When comparing different solvers, you can be affected by different 
> preconditioners (and preconditioner settings), different solution 
> >tolerances, and different metrics for judging solution tolerance. For 
> instance, the default metrics used by PySparse PCG and >Trilinos PCG are 
> different. We probably need to decide whether FiPy's default settings should 
> force them to be identical or not. >Ultimately, that may not be possible; 
> these packages offer different capabilities.
I have found that without Preconditioners all Solvers are noticeably
slower. This may reflect what James said in one of the last posts.

Regards,
Igor.


Reply via email to