Hello Daniel,

I made a big mistake. I was somehow under the impression that Fipy is a
finite element library. No wonder that I couldn't find information on
the base functions of the elements that are used. Sorry for the
confusion. getfem++ seems to solve my problem.

Thanks alot!
Benjamin


Am 17.01.2014 00:26, schrieb Daniel Wheeler:
> On Wed, Jan 15, 2014 at 5:15 PM, Benjamin Hepp
> <[email protected]> wrote:
>> Hello,
>>
>> my question from a few days ago might have been a bit imprecise. I
>> reformulate it and hope this helps:
>>
>> I could not find any details in the FiPy documentation so I assume it is
>> using first order Lagrange elements psi_i, where i is the element index.
> 
> I guess that is correct. FiPy just uses linear interpolation between
> cell centers.
> 
>> I would like to compute the integrals
>>   <psi_i, psi_j>
>> and
>>   <grad psi_i, grad psi_j>
>> for each i, j. Is this possible with FiPy?
> 
> This is confusing. In the FVM, $\psi_i$ only really means anything in
> cell $i$. By definition,
> 
>    \psi_i = \int \psi dV_i / \int dV_i
> 
> over a given cell. The integrals might not make sense in the context
> of the FVM, while making sense in terms of the FEM. This may be to do
> with the notion of local support in FEM, where the $phi_i$ is defined
> on the whole domain, but is only non-zero in the neighbourhood of
> element $i$ so the integral makes sense. If one was to do this naively
> with the output from fipy, then you would just have a diagonal matrix,
> which doesn't show you anything. What you need is a reconstruction of
> the non-averaged discretized $\psi$'s. I don't believe that FiPy has
> any inbuilt functionality to do this.
> 
> Hope that helps or at least clarifies things a bit.
> 


Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
fipy mailing list
[email protected]
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Reply via email to