> Hmm... we have multiple scalars with the same unknown_name, right?
Yep, the variable order determines the number of scalars.
> So we do need some way to disambiguate them. We could pass (0,0,0)
> for the
> first, (1,0,0) for the second, etc... or we could append their indices
> to the unknow
On Wed, 21 Oct 2009, David Knezevic wrote:
> OK, it's all checked in now. SCALARs should now work in parallel and with
> project_vector.
>
> I haven't done the line 255 (now line 280) fptr thing yet though since the
> fptr depends on a spatial coordinate. Shall I just set the SCALAR to be fptr(
On Oct 21, 2009, at 1:19 PM, David Knezevic wrote:
> OK, it's all checked in now. SCALARs should now work in parallel and
> with project_vector.
Great! I'll get this tested on our end and let you know if I find
anything wrong...
Thanks!
Derek
> Unfortunately, there are no project_vector gurus; Ben and I wrote all
> that code, and I wouldn't say either of us did a great job at it. ;-)
>
> You can stay out of System::ProjectVector::operator(), surprisingly -
> that takes an element range, to enable threading, but copying SCALAR
> values
On Tue, 20 Oct 2009, David Knezevic wrote:
> I had a look at getting SCALAR variables working with n_processors() > 1. The
> attached patches for dof_map take care of things, but I wanted to run them by
> the list before checking in since 0.6.4 is just around the corner.
>
> The change I made i
Dave... just wanted to let you know that the work on SCALARs is really
appreciated! We're using it here to solve some constraint equations
simultaneously with our non-linear equations and it's working great!
Being able to do parallel with SCALARs would be a really nice addition...
Derek
On Tue,
