fenics team mailing list archive

[Harish Narayanan <harish.mlists@xxxxxxxxx>]

On 18/12/2009 14:19, Marie Rognes wrote:
> Marie Rognes wrote:
>> Harish Narayanan wrote:
>>> Dear fellow PDE solvers,
>>>
>>> The sort of question I am about the phrase might seem strange to some,
>>> but perhaps it will seem natural to others.
>>>
>>> Last evening, I was curious about meshing quirky geometries like the
>>> Möbius strip in DOLFIN. This morning, I was curious about solving a PDE
>>> on such a domain, and so I came up with the attached python script.
>>>
>>> The script purportedly sets up a Poisson equation (with homogeneous
>>> boundary conditions and a constant source) on a Möbius strip (with four
>>> twists) and "solves it." I see a pretty looking plot with some
>>> psychedelic colours, but can someone who knows more about this stuff be
>>> kind enough to clarify what I've ended up doing?
>>>
>>>   
>>
>> I haven't got the foggiest, 
> 
> Modifying this statement a bit:
> 
> In the assembling of the matrices etc, the third coordinate is probably
> just
> ignored. (At least the FFC generated code probably just ignores the third
> coordinate). Try removing the last coordinate from your  mesh --
> you'll get a very similar picture.

This is probably what is going on. Without the third coordinate, the
domain (obviously) looks different, but the "solution" numbers look the
same. But most importantly, the essential prettiness of the plots is
retained.

Harish