Wolfgang,

> So I vote that the command line wins over what we> specify in the code. This 
> is what your patch did, right?

This is what I was going for. Yes and that's the PETSc philosophy.
Coming from mostly a PETSc background, I found it incomprehensible to
see the command line options ignored until I saw the code. But I
figured the deal.II philosophy is different in this sense but
nevertheless a little confusing. All the Mat, Vec, PC options are
being set from command line while the KSP options are being ignored. I
find this inconsistent in the deal.II call sequence.

> Can I ask you to write a paragraph in the documentation of the
> PETScWrappers::SolverBase class on how one could use this feature in
> applications?

I can definitely send you something to clarify these aspects so that
you can add them to the documentation.

Thanks,
Vijay

On Wed, Dec 14, 2011 at 11:05 PM, Wolfgang Bangerth
<[email protected]> wrote:
>
> Vijay and Toby,
> as I said, the deal.II way has traditionally been to specify everything in
> the code and nothing on the command line. That has a lot to do with the fact
> that I personally think that it's clearer and more reproducible, and I have
> never quite understood how the PETSc behavior can reliably be controlled if
> one has several nested solvers. But I accept (i) that that conflicts with
> the PETSc viewpoint, and (ii) that there are useful things one can do with
> PETSc by explicitly giving command line flags. So I agree that what you try
> to do is useful.
>
> There remains the question what should happen if what we specify in the code
> does not match what is passed on the command line. I don't think it is very
> useful if someone goes to lengths specifying something on the command line
> just to find that it is ignored (that's not a question of right or wrong;
> simply of utility). So I vote that the command line wins over what we
> specify in the code. This is what your patch did, right?
>
> Can I ask you to write a paragraph in the documentation of the
> PETScWrappers::SolverBase class on how one could use this feature in
> applications?
>
> Best & thanks to both of you!
>  Wolfgang
>
> ------------------------------------------------------------------------
> Wolfgang Bangerth               email:            [email protected]
>                                www: http://www.math.tamu.edu/~bangerth/
>
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii

Reply via email to