Re: [Geopdes-users] Some difficulties with problems, which are not difficult

[Carlo de F. <car...@po...>]

On 21 Sep 2011, at 14:52, Dominik Mokriš wrote:

> Hello all,

> One more question, if you don't mind: Is there a reasonably simple way to
> define my own operators? With my mentor, we are trying to solve the
> convection -- diffusion problem (Bazilevs et.al. (2005) were showing
> advantages of IGA on this topic) but we're stuck in making the operator (b .
> grad(u) )v. You can have a look at our attempts in the attachement

The attachment is missing so we cannot have a look at the code,
anyway implementing the advection operator, without stabilization, should be quite easy.
If you want to implement the SUPG stabilization mentioned in the paper you cite, it is 
a bit more challenging but it would also be a very welcome addition to geopdes.

> but I
> hope, there exists a simpler way, than trying to glue it from op_u_v and
> op_gradu_gradv.

actually that is exactly the same approach I would follow, why do you find it so complicated?
If the difficulties you have are due to all the extra gymnics used to optimize memory and speed
in version 2.0 of the operators, you can start working on 1.1
which is much simpler (though A LOT more memory consuming).  
Moving to 2.0 once you have a working function is very simple.

> Thanks a lot.
> 
> Dominik.

Carlo