On Sun, 15 Jul 2012, Derek Gaston wrote:
Our motivation is that we're using the mapping stuff independent of
finite-element assembly quite a bit. Specifically in mechanical contact
simulations we do a _ton_ of stuff with mapping... and it's kinda weird that
we have to go through the FE object to do that stuff. It would be great to
be able to interact with the FEMap object directly.
That's reasonable. Plus, "we'd like to see an aesthetically better
design" always seems to result in better long term refactoring than
"we've got this new feature we need to add by last week".
But I'm not sure whether to feel disappointed or relieved that
"someone's interested in adding NURBS support, you could join in..."
is not dangling tantalizingly in front of me.
For instance... for a while there I was thinking that for doing contact
point mapping on second order lagrange geometry we would need a second order
lagrange FE object. That turns out not to be the case though because
underneath the mapping was automatically doing second order lagrange mapping
regardless of what kind of FE object we had. It just feels a little odd to
have to instantiate an FE object in order to do mapping...
It's also probably a little inefficient when users need second-order
element topology for the higher p support but only use first-order
geometry. The affine element code paths were based on real profiling
results that suggested shaving ~90% off mapping calculation time would
be worth the effort, but there are non-affine first-order meshes where
that code path is inapplicable yet we could at least shave 50%-70% off
mapping calculations by letting the user specify in advance that
they're all first-order.
---
Roy
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Libmesh-devel mailing list
Libmesh-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/libmesh-devel
