[Dune] Question about the GenericReferenceElements

[Andreas Dedner]

 > actually, this is the internal structure of the 
GenericReferenceElement. In my
I know. That why I made the suggestion and it would not interfere with 
the old
methods... The problem with confusion i and c seems impossible to solve.
But mostly one iterates over i=0...I-1 and keeps c fix. Asserting c<dim 
will therefore
in most cases lead to a runtime error when i and c are confused since 
I-1>=d in general.

Best
Andreas

> opinion, this is a great idea (though it is still possible to confuse i and c --
> and I don't mean variable names, here).
>
> Best,
>
> Martin
>
> On 10/22/2010 11:24 PM, Andreas Dedner wrote:
>    
>> Hi Martin,
>>
>> how about a method
>> GenericSubEntity subEntity(c,i)
>> return a ''proxy'' or reference with an index method
>> index(cc,ii) or something like that, then there should be
>> less possibilities to do it wrong and since everything is Cached
>> anyway it should be possible to do something like that without
>> loss in efficiency?
>>
>> Best
>> Andreas
>>
>> On 10/22/2010 08:23 PM, Martin Nolte wrote:
>>      
>>> Hi Oliver,
>>>
>>> when implementing the generic reference element, I implemented all methods of
>>> the original ReferenceElement class. Especially, this included the method
>>>
>>> global ( const FieldVector<   ctype, dim-codim>   &local, int i, int c ) const.
>>>
>>> When I saw the redundancy, I simply added another method, dropping the redundant
>>> argument (for convenience).
>>>
>>> Back then, I somehow assumed that the interface for the ReferenceElement has
>>> been thoroughly discussed. Over time, this impression has changed. Since the old
>>> ReferenceElement is now gone, I would be very happy to discuss what methods
>>> should be present on the GenericReferenceElement. In particular, I would very
>>> much like to get rid of the (i,c,ii,cc) methods. In my mind it is (c,i,cc,ii)
>>> and, since the types don't differ, I frequently get weird errors from this.
>>>
>>> I would very much appreciate, if an "interface" for the reference element could
>>> be discussed on the developer meeting in Münster.
>>>
>>> Best,
>>>
>>> Martin
>>>
>>>
>>> On 10/22/2010 02:43 PM, Oliver Sander wrote:
>>>
>>>        
>>>> Dear Dune!
>>>> I just had a look at the GenericReferenceElement class.
>>>> I learned that said class has methods
>>>>
>>>> template<   int codim>
>>>> FieldVector<   ctype, dim>
>>>> global( const FieldVector<   ctype, dim-codim>   &local, int i )
>>>>
>>>> and
>>>>
>>>> template<   int codim>
>>>> FieldVector<   ctype, dim>
>>>> global( const FieldVector<   ctype, dim-codim>   &local, int i, int c ) const
>>>>
>>>> Their documentations appear to be identical, with the exception of a
>>>> note saying "The runtime argument c is redundant and must equal codim"
>>>> for the second method.  And indeed the method throws an exception
>>>> if c!=codim.
>>>>
>>>>
>>>> Why do we have this seemingly pointless method as part
>>>> of our grid interface?
>>>>
>>>> Best,
>>>> Oliver
>>>>
>>>> -- 
>>>>
>>>> ************************************************************************
>>>> * Oliver Sander                ** email: sander at mi.fu-berlin.de        *
>>>> * Freie Universität Berlin     ** phone: + 49 (30) 838 75348           *
>>>> * Institut für Mathematik      ** URL  : page.mi.fu-berlin.de/~sander  *
>>>> * Arnimallee 6                 ** -------------------------------------*
>>>> * 14195 Berlin, Germany        ** Member of MATHEON (www.matheon.de)   *
>>>> ************************************************************************
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Dune mailing list
>>>> Dune at dune-project.org
>>>> http://lists.dune-project.org/mailman/listinfo/dune
>>>>
>>>>          
>>>
>>>        
>>      
>