fiat team mailing list archive

["Robert C. Kirby" <kirby@xxxxxxxxxxxx>]

Garth,
I'm not sure why you get an unsubscriptable object, but
there are at least two problems in how you are using/understanding  
the function.

1.) First, it's not necessarily a quadrature rule that you're  
tabulating at, but any set of points.  This has no impact on your  
problem.
2.) Second, you have to provide the one-dimensional points at which  
to tabulate since you said "d=1".  So if you wanted the Gauss points  
along edge number 1, you would compute the Gauss points on the  
reference line (q=quadrature.make_quadrature(shapes.LINE,degree)),  
and then call trace_tabulate( d , 1 , q.get_points() ).   
Trace_tabulate maps the points on the reference line to the  
appropriate edge.

Let me know if there is still a problem.

Rob

On Apr 27, 2006, at 10:14 AM, Garth N. Wells wrote:

> Hi Rob,
>
> I'm playing around with calling trace_tabulate from FFC. Before I
> elaborate my problem, let me check that I'm on the right path with
> trace_tabulate.
>
> In a nutshell, does trace_tabulate return the basis functions on
> edge/face e of dimension d (d=2->face, d=1->edge, etc) given a
> quadrature scheme of dimension d for an element of dimension n? (e  
> and d
> being the same as used in the trace_tabulate argument list).
>
> When I try in FFC
>
>   junk =  self.element.function_space().trace_tabulate(1, 1,  points)
>
> where
>   self.element.function_space() is of type
> FIAT.polynomial.ScalarPolynomialSet
> and
>   points = [(0.0)]
>
> I get the error
>
>   *** unsubscriptable object
>
> Garth