Thanks Roy,
Solved. I was trying to do just one time step and where I was calling the
sequence wasn't the right place so the constraints weren't imposed
properly. Now it works.
Best,
Ata
On Fri, Nov 15, 2013 at 4:50 PM, Roy Stogner wrote:
>
> On Fri, 15 Nov 2013, Ataollah Mesgarnejad wrote:
>
On Fri, 15 Nov 2013, Ataollah Mesgarnejad wrote:
> It still doesn't work. Where exactly would you call that command sequence?
Anywhere after your solve for the previous time step but before your
assembly for the next time step should have (in theory - in practice
we don't have this in test cover
Roy,
It still doesn't work. Where exactly would you call that command sequence?
Thanks,
Ata
On Fri, Nov 15, 2013 at 12:45 PM, Ataollah Mesgarnejad <
ames...@tigers.lsu.edu> wrote:
> Thanks Roy,
>
> I will try it and get back to you.
>
> As for the example I have bunch of derived classes from
Thanks Roy,
I will try it and get back to you.
As for the example I have bunch of derived classes from AnalyticFunction
that compute BCs for me. I can try add one of these to Elasticity examples
you already have and generate a new example for you.
Best,
Ata
On Fri, Nov 15, 2013 at 12:37 PM, Ro
On Fri, 15 Nov 2013, Ataollah Mesgarnejad wrote:
> I just noticed that. But isn't that rather a high price to pay to reinit
> the whole EquationSystems to get the updated constraints? Can't there be a
> specific function that instead of doing a reinit just updates the
> constraints?
Not one spec
Thanks Paul,
I just noticed that. But isn't that rather a high price to pay to reinit
the whole EquationSystems to get the updated constraints? Can't there be a
specific function that instead of doing a reinit just updates the
constraints?
Thanks,
Ata
On Fri, Nov 15, 2013 at 12:22 PM, Paul T.
I wouldn't call that a bug :)
You could consider using a "lifting function" approach, where you impose
zero Dirichlet boundary conditions, and then solve for u_hat = u - g_BC,
where g_BC is a (possibly time-dependent) lifting function that satifies
the non-zero boundary condition that you want.
Dear all,
I noticed that the Dirichlet BCs through DirichletBoundary API are only
applied before the the first system assemble!! This does not make sense
for time variable Dirichlet BCs which require time to be passed through the
FunctionBase to the boundaries.
Best,
Ata
I've never tried this, but at a minimum something like dof_map.reinit() would
be needed I think, as it is here the object->dof index mapping is constructed.
The map stating which variables are active where is still valid, but the
changes in the mesh are not automatically propagated.
-Ben
On
