On Wed, Jan 23, 2013 at 12:43:17PM -0600, Roy Stogner wrote:
> On Thu, 10 Jan 2013, Robert wrote:
> > On Thu, Jan 10, 2013 at 07:27:43AM -0600, Roy Stogner wrote:
> >> On Thu, 10 Jan 2013, Robert wrote:
> >>> The compression feature depends on a somewhat recent version of vtk,
> >>> I can do some research about the version, when it was introduced, if
> >>> needed.
> >> That actually would be pretty helpful; I'd like to avoid breaking
> >> things for users with old VTK versions but I'm not sure if I've got
> >> any handy to test with myself.
> > I looked through the git of vtk. The call used in the patch was
> > introduced in vtk 5.6.0 (Released in April 2010). The feature itself is
> > in vtk somewhat longer, but has to be accessed differently. I suggest to
> > surround this call by version-check-macros. So user of this old vtk
> > installation just don't have this addional feature.
> Thanks again!
> I think I've now got your vtk patch and your nonlinear elasticity
> example working with the newest libMesh - just need to add the
> copyright boilerplate before I commit. You're the author and
> copyright holder of those contributions and you license them for
> public use under the LGPL, correct?
Yes, I'll licence them under LGPL.
> While I'm waiting for your reply I'll be committing the library fixes
> that were necessary to get your elasticity example working again.
> Sorry for regressing the ALE code on you recently - since that's
> always been half-broken (it was work-in-progress for a project that
> got replaced by a different formulation), I didn't expect that anyone
> had started to depend on the non-broken half!
This is not specific to ALE, this is usefull for every problem that
needs to distinguish between undeformed (reference) and deformed
Some time ago I switched my primary variable from the node position back
to the node displacement, to be able to use the (at this point
homogenious only) periodic boundary conditions, so I don't use this part
of the code anymore.
As I still need quantities on the reference configuration (mapping
jacobian), on the current configuration (shape functions,
their derivations, qp positions, normals) and the mapping between
undeformed and deformed configuration I currently deform each
element manually and reinit the fe myself.
Does anyone know a better way to do this with libmesh? Is the FEMap
class of any use here?