fiat team mailing list archive

["Garth N. Wells" <g.n.wells@xxxxxxxxxx>]

This is my first foray into Python, so I don't really understand the
finer points. Drawing on the function existing tabulate_jet in the class
VectorPolynomialSet, should a function trace_tabulate_jet in
VectorPolynomialSet look like (?)

def trace_tabulate_jet( self , d , e , order , xs ):
    myshape = self.domain_shape()
    xs_dim = map( shapes.pt_maps[ myshape ][ d ]( e ) , xs )        
    return [ self.select_vector_component( i ).tabulate_jet( order ,
xs_dim ) \
             for i in range(self.tensor_shape()[0]) ]

I'd really like to have something that does this because I've added
functionality to FFC to handle boundary integrals and it relies on
having such a function to work with both scalar and vector polynomials.

Garth


On Mon, 2006-05-01 at 10:12 -0500, Robert C. Kirby wrote:
> It works because you applied it to an instance of a subclass that  
> supports tabulate_jet.  Dynamic binding in Python hides some of these  
> issues (it wouldn't compile in C++).  To do the vector-valued case,  
> you need to do it to every component of the vectors.  It's a similar  
> function, but trickier.
> 
> 
> On May 1, 2006, at 10:01 AM, Garth N. Wells wrote:
> 
> > Rob,
> >
> > Is it just a question of adding a similar function to the class
> > VectorPolynomialSet as below?  When I do this, the output looks
> > reasonable.
> >
> > Garth
> >
> > --- polynomial.py       1 May 2006 14:28:43 -0000       1.8
> > +++ polynomial.py       1 May 2006 14:56:33 -0000
> > @@ -283,6 +283,12 @@
> >                   for i in range(self.tensor_shape()[0]) ]
> >
> >
> > +    def trace_tabulate_jet( self , d , e , order , xs ):
> > +        myshape = self.domain_shape()
> > +        xs_dim = map( shapes.pt_maps[ myshape ][ d ]( e ) , xs )
> > +        return self.tabulate_jet( order, xs_dim )
> > +
> > +
> >  # The code for TensorPolynomialSet will look just like
> > VectorPolynomialSet
> >  # except that we will flatten all of the coefficients to make them
> >  # look like vectors instead of tensors.
> >
> >
> > On Mon, 2006-05-01 at 16:28 +0200, fiat@xxxxxxxxxx wrote:
> >> Commit from kirby (2006-05-01 16:28 CEST)
> >> -----------------
> >>
> >> Added trace_tabulate_jet to ScalarPolynomialSet.
> >> Patch thanks to Garth.
> >> Only problem is that tabulate_jet is undefined on the
> >> AbstractPolynomialSet (in the patch), so I
> >> moved it to ScalarPolynomialSet.  It's not hard to generalize the
> >> jet to vector-valued spaces, but gives for a large number of
> >> indices.  Perhaps in the future...
> >>
> >>   fiat  FIAT/polynomial.py  1.8
> >>
> >> _______________________________________________
> >> FIAT-dev mailing list
> >> FIAT-dev@xxxxxxxxxx
> >> http://www.fenics.org/cgi-bin/mailman/listinfo/fiat-dev
> >
> >
> > _______________________________________________
> > FIAT-dev mailing list
> > FIAT-dev@xxxxxxxxxx
> > http://www.fenics.org/cgi-bin/mailman/listinfo/fiat-dev