dolfin team mailing list archive

[Johan Hake <hake@xxxxxxxxx>]

On Thursday 12 March 2009 22:55:08 Anders Logg wrote:
> On Thu, Mar 12, 2009 at 09:45:41PM +0000, Garth N. Wells wrote:
> > Anders Logg wrote:
> > > On Thu, Mar 12, 2009 at 09:12:48PM +0000, Garth N. Wells wrote:
> > >> DOLFIN wrote:
> > >>> One or more new changesets pushed to the primary dolfin repository.
> > >>> A short summary of the last three changesets is included below.
> > >>>
> > >>> changeset:   5860:54b6f356ba99be68a5caa08fa66031fc1572ac0e
> > >>> tag:         tip
> > >>> user:        "Johan Hake <hake@xxxxxxxxx>"
> > >>> date:        Thu Mar 12 22:00:22 2009 +0100
> > >>> files:       dolfin/function/SpecialFunctions.cpp
> > >>> dolfin/function/SpecialFunctions.h
> > >>> site-packages/dolfin/specialfunctions.py description:
> > >>> Added a SUPGStablizer built in function
> > >>>   sigma - diffusion coeffisient
> > >>>   a     - advective field
> > >>>   s     - SUPGStabilizer(a,sigma)
> > >>>   v     - TestFunction
> > >>>
> > >>>   # SUPG stabilized TestFunction
> > >>>   v = v + dot(s,grad(v))
> > >>>
> > >>>   (Will add a demo later)
> > >>
> > >> Some of this is pretty specific, e.g. tau = 1/std::tanh(PE)-1/PE. I'm
> > >> not convinced that it should be in SpecialFunctions (very worthwhile
> > >> in a demo though which users can mimic).
> > >>
> > >> Garth
> > >
> > > I think it could be useful to have. We could also add other optional
> > > stabilizers. It's in SpecialFunctions after all. It's a matter of
> > > taste how special the functions can be.
> >
> > Part of my motivation is that I have a number of SUPG stabilised solvers
> > and I haven't felt the need to move parts into SpecialFunctions, in
> > contrast to upwinding at facets which I like in SpecialFunctions.

This is your field Garth, so it's fine with me that you deside. 

I know for sure though that if _any_ SUPGStabilizer were available in DOLFIN 
or any demo when I started working on this I would have saved a lot of time. 
I was aware of the demo included in DOLFIN 0.6.4 which I had some help from.

Is it possible to categorize such functions into some few general one that 
covers some common user cases? These could then be included either as 
SpecialFunctions or in the demos. 

I have to admit that I have more the engineering approach to this, a toolbox 
with usefull Functions, with for example a reference to the literature, is 
better than no toolbox at all :)

Johan

> I don't mind either way.