On Jan 16, 2012, at 11:59 AM, Roy Stogner wrote:

> 
> If you want a per-subdomain tensor as a property that gets saved in
> your output file...  a separate ExplicitSystem with a SCALAR variable
> for each independent tensor element, maybe?
This one seems to be excessive since I'll need to create corresponding FEbase 
and at least update the shape function. 

> If you just want it as a property for use in your assembly, I'd just
> do a switch test on the subdomain id to determine which tensor to use
> in which subdomain.

I actually started doing it this way. My problem is that I wanted to have 
flexibility for number of subdomains and different tensors corresponding to 
them. Ultimately I want the program to get the different tensors and their 
corresponding subdomain(s) as an input and do the switching automatically. The 
difficulty implementing this is storing these tensors. only way I can think of 
to store these values was in EquationSystems::Parameters, however this will get 
very messy when you want to add all the components from several tensors!

So my questions at this point are:

1- If there is any way around adding tensor components as scalars to 
EquationSystems::Parameters? I'm not sure but can I add a custom type as a 
parameter?
2- Is there a way to work with Parameters without using their string names? 
(some sort of enumeration in parameters)
3- Or if there is a better place to store these tensors (components) so I can 
access them anywhere that I access the LinearImplicitSystem that uses them?

Best,
Ata
------------------------------------------------------------------------------
RSA(R) Conference 2012
Mar 27 - Feb 2
Save $400 by Jan. 27
Register now!
http://p.sf.net/sfu/rsa-sfdev2dev2
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to