ufl team mailing list archive

[Kristian Oelgaard <k.b.oelgaard@xxxxxxxxxx>]

Quoting "Garth N. Wells" <gnw20@xxxxxxxxx>:

>
>
> Kristian Oelgaard wrote:
> > Quoting Martin Sandve Alnæs <martinal@xxxxxxxxx>:
> >
> >> On Thu, May 14, 2009 at 5:59 PM, Garth N. Wells <gnw20@xxxxxxxxx> wrote:
> >>>
> >>> Martin Sandve Alnæs wrote:
> >>>> On Thu, May 14, 2009 at 9:43 AM, Anders Logg <logg@xxxxxxxxx> wrote:
> >>>>> On Thu, May 14, 2009 at 08:40:41AM +0100, Garth N. Wells wrote:
> >>>>>> Martin Sandve Alnæs wrote:
> >>>>>>> Is there a mathematical difference?
> >>>>>>>
> >>>>>> Splitting hairs, I would say yes and that grad(u('+')) is correct and
> >>>>>> grad(u)('+') is not because grad(u) is not defined in a classical
> sense.
> >>>>>> I prefer grad(u('+')), but FFC used to use grad(u)('+'). My preference
> >>>>>> is however not strong.
> >>>>>>
> >>>>>> My immediate question is will the generated code differ for
> grad(u('+'))
> >>>>>> and grad(u)('+')?
> >>>>>>
> >>>>>> Garth
> >>>>> No, since the restriction will be propagated to the leaves... :-)
> >>>> The leaves are terminal expressions, that is literal constants,
> >>>> geometric quantities, functions and basis functions.
> >>>> "Propagating to the leaves" means modifying for example
> >>>>
> >>>>  (grad(f) + grad(g))('+')
> >>> So the above works now? It would be super useful for me.
> >> I don't know what FFC does.
> >>
> >> But I implemented the propagation right now, only tested
> >> for a simple case but it's a simple algorithm:
> >>
> >> # test_restrictions.py:
> >> from ufl import *
> >>
> >> e = FiniteElement("DG", triangle, 1)
> >> f = Function(e)
> >> g = Function(e)
> >> a = (grad(f) + grad(g))('-')
> >> print a
> >>
> >> from ufl.algorithms import propagate_restrictions
> >>
> >> b = propagate_restrictions(a)
> >> print b
> >>
> >>
> >> martinal@martinal-znote:~/dev/fenics/ufl/sandbox/algorithmtests$
> >> python test_restrictions.py
> >> ((grad(w_0)) + (grad(w_1)))('-')
> >> (grad((w_0)('-'))) + (grad((w_1)('-')))
> >>
> >>
> >> FFC just needs to apply propagate_restrictions to the integrand of
> >> interior facet integrals.
> >
> > I just tried it and it appears to be working. However, I got the error:
> >
> > *** FFC: Not expecting twice restricted expression.
> > /home/oelgaard/fenics/clone/ffc/scripts/ffc:131: DeprecationWarning:
> > BaseException.message has been deprecated as of Python 2.6
> >   error(exception.message)
> > *** FFC: Not expecting twice restricted expression.
> > *** FFC: To get more information about this error, rerun FFC with --debug.
> >
> > for the following form (from Garth)
> >
> > sigma_a = -b*u - gamma*b*(uhat - u)
> > sigma_d = kappa*grad(u) + (alpha/h)*kappa*(uhat -u)*n('+')
> > sigma_i = dot(sigma_a + sigma_d, n)
> >
> > sigma_plus = sigma_i('+')
> >
> > a = v('+')*sigma_plus*dS
> >
> > The reason is that sigma_d already contains n('+'). What should happen in a
> case
> > like this? For complex forms it is sometimes easier to declare temporary
> > expressions and in these cases one might accidentally (or intentionally?)
> > restrict a function multiple times. Should a restriction just be
> overwritten or
> > what would you expect Garth for this form?
> >
>
> I would like to get an error. n('+') shouldn't be in sigma_d (my mistake).

OK, fine by me. I have updated FFC to use propagate_restrictions() it should be
working now.

Kristian

> Garth
>
> > Kristian
> >
> >> Martin
> >> _______________________________________________
> >> UFL-dev mailing list
> >> UFL-dev@xxxxxxxxxx
> >> http://fenics.org/mailman/listinfo/ufl-dev
> >>
> >
> >
>