Re: [Libmesh-users] Augmenting sparsity pattern and sendlist

[Kirk, B. (JSC-EG311) <ben...@na...>]

We do this in a thermal analysis code. I can send you some snippets if you like. 

I plan to add an example demonstrating this using nitsches method for nonmatched boundaries, but I'd be happy to have you beat me!

-Ben



On May 29, 2013, at 3:45 PM, "David Knezevic" <dkn...@se...> wrote:

> In order to model a "film resistance" thermal interface condition, I'm 
> using a mesh with a "crack", and I need to assemble an interface term on 
> the crack.
> 
> More precisely, I have two mesh subdomains, \Omega_1 and \Omega_2, with 
> coincident boundaries, but which are not connected in the FE sense. Let 
> \Gamma denote the coincident boundary. The term on \Gamma takes the form:
> 
> \int_\Gamma (u_1 - u_2) (v_1 - v_2) ds,
> 
> where u_1, u_2 (resp. v_1, v_2) are the trial (resp. test) functions on 
> either side of \Gamma.
> 
> This term looks much like an interface term in DG, and I've assembled it 
> using an approach similar to miscellaneous_ex5 (note: I'm using C0 basis 
> functions, but of course the solution is discontinuous across \Gamma).
> 
> However, when I call solve, I get:
> 
> [0]PETSC ERROR: --------------------- Error Message 
> ------------------------------------
> [0]PETSC ERROR: Argument out of range!
> [0]PETSC ERROR: New nonzero at (203,206) caused a malloc!
> 
> This isn't surprising since the interface term introduces new non-zeros 
> in the sparsity pattern. In order to make this work, it seems I just 
> need to:
> - Augment the sparsity pattern
> - Augment the send_list (for it to work in parallel)
> 
> I see that dof_map has functionality for doing this "augmentation", so 
> I'm going to try those functions out. But also I was wondering if anyone 
> has any example code that demonstrates this, or any 
> comments/suggestions, in case there are any "gotchas" with this type of 
> thing?
> 
> Also, if there's interest, I'd be happy to use this model problem to 
> make a new libMesh example, since it exercises some functionality that 
> may be worthwhile having in the example suite.
> 
> David
> 
> 
> ------------------------------------------------------------------------------
> Introducing AppDynamics Lite, a free troubleshooting tool for Java/.NET
> Get 100% visibility into your production application - at no cost.
> Code-level diagnostics for performance bottlenecks with <2% overhead
> Download for free and get started troubleshooting in minutes.
> http://p.sf.net/sfu/appdyn_d2d_ap1
> _______________________________________________
> Libmesh-users mailing list
> Lib...@li...
> https://lists.sourceforge.net/lists/listinfo/libmesh-users