On Tue, 2008-08-12 at 17:31 -0500, Kirk, Benjamin (JSC-EG) wrote:
> I will see about sending you a patch to ex9 demonstrating SUPG for a
> scalar problem. John is right, for systems it gets a little too
> complicated for an example...
Thanks, I'd appreciate it!
> ----- Original Message -----
> From: libmesh-devel-bounces@...
> To: Adam C Powell IV <hazelsct@...>
> Cc: libmesh-devel@...
> Sent: Tue Aug 12 17:19:46 2008
> Subject: Re: [Libmesh-devel] Example code with upwinding
> On Tue, Aug 12, 2008 at 5:15 PM, Adam C Powell IV
> <hazelsct@...> wrote:
> > [Sorry to post a -users message to -devel; I can't post to -users.]
> Interesting. I can see about adding you to the list, if you want? Or
> is it just an issue with your domain?
Strangest thing: I'm a member of the list, and get all of its posts, and
had no trouble posting through 7/15. Then 7/21 my posts started
vanishing without showing up on the list. And I can still post to
-devel just fine.
But I think you (John) were a part of that discussion a week or two ago.
> > I have a very high-Peclet number advection-diffusion problem, and
> > like to use upwind differencing for the advection terms to reduce
> > numerical diffusion. I don't see this in any of the libmesh
> > but the online gallery shows some nice examples which might have
> > it. (Also the deal.II FEA library has an example or two which uses
> AFAIK, upwind differencing *adds* numerical diffusion, to improve
Right, uh, that's what I meant. :-)
> > Is any libMesh code available with upwinding?
> I have some code which uses SUPG but it's not simple enough to be an
> example. Nothing using simple upwinding though. Was this in 1D, 2D,
> 3D? Maybe the dealII example could be co-opted to work in libmesh?
This is example 12, in arbitrary dimensions; intro.dox ends with:
Finally, we note that the program is dimension independent, i.e.
after replacing <code>DGMethod<2></code> by
<code>DGMethod<3></code> the code runs in 3d.
The API is somewhat different, and this example's documentation makes it
sound more FV-ish than FEA-ish. So I think it would be hard to
Frankly, I find the libMesh API, examples, and documentation easier to
learn and more intuitive. But then, perhaps that's why you don't have
an example with upwinding!
Thanks again for the replies.
GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6
Engineering consulting with open source tools