ufl team mailing list archive

["Martin Sandve Alnæs" <martinal@xxxxxxxxx>]

On Sat, Jan 10, 2009 at 3:19 PM, Anders Logg <logg@xxxxxxxxx> wrote:
> On Sat, Jan 10, 2009 at 03:14:02PM +0100, UFL wrote:
>> One or more new changesets pushed to the primary ufl repository.
>> A short summary of the last three changesets is included below.
>>
>> changeset:   492:982082a9d9df0fad6ceab20e57f7a734d7aae9ea
>> tag:         tip
>> user:        "Martin Sandve Alnæs <martinal@xxxxxxxxx>"
>> date:        Sat Jan 10 15:14:29 2009 +0100
>> files:       ufl/algorithms/ufl2latex.py ufl/classes.py ufl/differentiation.py ufl/exproperators.py ufl/indexing.py ufl/tensors.py
>> description:
>> Removed AxisType, simplifying MultiIndex and related code a bit.
>> Indexing with an axis is now represented differently:
>>
>>   A = <rank 2 tensor expression>
>>   b = A[i,:] # rank 1 expression with one free index
>>
>> is the same as
>>
>>   b = as_tensor(A[i,j], (j,))
>>
>> or
>>
>>   b = A[i,j](j)
>
> The first one looks good to have (move from free-indexed expression to
> tensor) but is it something that it should be mapped to __call__?
>
> __call__ can be saved for other things like for example
>
>  A[i,j](x)
>
> which would be the value at a specific point x.

The __call__ mapping was just an experiment.
I agree it doesn't feel right, so I'll remove it.

Evaluating an expression at a specific point doesn't make much
sense without values for basis functions and functions though.
But it would be good to have for testing!

Martin