Local Raviart-Thomas space

[Giselle Sosa Jones]

Hello,

I am trying to implement a projection of a DG solution to a Raviart-Thomas space. For this, I need a local space Q_{p-1,p}(K) \times Q_{p,p-1}(K). I have been using FE_DGRaviartThomas of degree p-1, and then I use shape_value_component(i, q, 1) when I want to use the x component of the test function, and shape_value_component(i,q,0) when I want to use the y component of the test function. I have a bug in my code and I am wondering if it is coming from the way I am creating this local space. Does this construction of the space make sense? Is there maybe a better way of doing it?

Many thanks.

Giselle

[Wolfgang Bangerth]

On 6/16/21 2:00 PM, Giselle Sosa Jones wrote:
>
> I am trying to implement a projection of a DG solution to a Raviart-Thomas
> space. For this, I need a local space Q_{p-1,p}(K) \times Q_{p,p-1}(K). I have
> been using FE_DGRaviartThomas of degree p-1, and then I
> use shape_value_component(i, q, 1) when I want to use the x component of the
> test function, and shape_value_component(i,q,0) when I want to use the y
> component of the test function.

I suspect that you made a mistake when you say
component=1 => x-component
component=0 => y-component
It should be the other way around.

> I have a bug in my code and I am wondering if
> it is coming from the way I am creating this local space. Does this
> construction of the space make sense? Is there maybe a better way of doing it?

It's hard to tell without actually seeing what you do. But I would suggest you
take a look at step-61, which does the kind of thing you're looking for, I think.

Best
W.


--
------------------------------------------------------------------------
Wolfgang Bangerth email: bang...@colostate.edu
www: http://www.math.colostate.edu/~bangerth/