Hi,

I CCed to our hpfem list, so that people in our group can have a look
at this too.

On Mon, Mar 2, 2009 at 5:35 PM, Roy Stogner <[email protected]> wrote:
>
>
> On Mon, 2 Mar 2009, Ondrej Certik wrote:
>
>> On Mon, Mar 2, 2009 at 12:16 PM, Roy Stogner <[email protected]>
>> wrote:
>>>
>>>
>>> On Mon, 2 Mar 2009, Ondrej Certik wrote:
>>>
>>>> If I find time, I'd like to create Python interface to libmesh,
>>>> because I used it in the past and I would like to compare libmesh
>>>> h-adaptivity with our hp-adaptivity.
>>>
>>> If you're already doing that, you might want to go one step farther
>>> and compare libmesh hp-adaptivity with yours.  The former is
>>> theoretically working but I've rarely gotten the convergence rates I
>>> expected out of it, so I suspect there's some subtle bugs left.
>>
>> Ah, yes, absolutely! I didn't know that libmesh also has
>> hp-adaptivity.
>
> Well, that might be overstating the case.  ;-)  libMesh has the
> ability to adaptively refine p-hierarchic element types in h and/or p,
> which technically might qualify as "has (isotropic) hp-adaptivity"...
> but libMesh lacks a good heuristic for deciding which type of
> adaptation to do.  IIRC the only strategy classes I wrote were "h
> adapt if you touch a prespecified singularity, p adapt otherwise" and
> "try to see whether an h-coarsening or a p-coarsening would be worse,
> then h-adapt or p-adapt respectively instead".  The former seemed to
> give exponential but probably non-optimal convergence in a couple
> benchmark problems; the latter didn't even go exponential.
>
>> Then I will definitely give it a shot.
>
> Thanks!  I would have liked to spend more time on this myself, but
> eventually dissertation work took precedence.
>
> For that matter, I'd love to have an excuse to play with it now, but I
> still haven't figured out how to make Ben's hypersonics formulations
> work perfectly with hierarchic elements.


That's awesome that you are interested in this. At this moment, I am
quite busy with other things, but I am very interested in this
comparison. Does the isotropic hp-adaptivity work in 3d? The difficult
part is in choosing the right candidate for hp-adaptivity, and I know
that the guys in my group spent months in getting it right, so I am
curious in comparing it with your version to see if it's different. I
want to see the graph, then I will believe. :)

On Sun, Mar 1, 2009 at 3:30 PM, Kirk, Benjamin (JSC-EG311)
<[email protected]> wrote:
> is that these tend to be more "industrial-type" applications, where
> higher-order elements are often not used for various reasons (non-smooth
> solutions, sharp complex geometry, etc...)

In fact, hp-fem performs the best exactly with solutions that are both
non-smooth and sharp somewhere (it uses a low polynomial order there)
and very smooth somewhere else in the domain (it uses a high
polynomial order there).

But as I said, it's tough to get everything into a production ready
state, but it's exciting when we get there finally both in 2D and in
3D.

Ondrej

------------------------------------------------------------------------------
Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
-Strategies to boost innovation and cut costs with open source participation
-Receive a $600 discount off the registration fee with the source code: SFAD
http://p.sf.net/sfu/XcvMzF8H
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel

Reply via email to