Hello Lukas

You may want to look up this thread

https://groups.google.com/d/msg/dealii/RbcPNELLHZA/eZxRYjzlVXIJ

I started implementing Taylor basis as FE_DGT class some years back but did
not complete it, see code here

https://bitbucket.org/cpraveen/deal_ii/src/master/fe_dgt/?at=master
https://bitbucket.org/cpraveen/deal_ii/src/master/
dg/2d_scalar_unsteady_taylor/?at=master

FE_DGT is basically monomials, see
https://bitbucket.org/cpraveen/deal_ii/src/497a4c7b2aa0d07f92e7c9f58a2b4c
00ca1c1206/fe_dgt/fe_dgt.cc?at=master&fileviewer=file-
view-default#fe_dgt.cc-135

These shape functions are defined in physical space. They are not
orthogonal.

In my implementation, I transform as

(x-xc)/h, (y-yc)/h     where h = cell diameter

You probably want to transform as

(x-xc)/dx, (y-yc)/dy

then calculate shape_value, subtract integral of shape_function also. Then
all shape functions have zero integral and only first one has non-zero
integral. This should be possible to implement for Cartesian grids.


Best
praveen

On Wed, Apr 19, 2017 at 1:50 AM, Lukas Korous <[email protected]>
wrote:

> Hello,
>
> for solving the (ideal) MHD equations, I would like to implement two
> custom shapesets:
>
>
>    - one for the flow part, completely discontinuous, scalar FE space,
>    analogy to FE_DGQ, but based on Taylor expansion in the cell center
>
>
>    - this should be the easy part, but I would like to ask how to go
>       about it in deal.ii - I would like to implement at least linear and
>       quadratic functions
>
>
>    - one for the magnetic field, a vector Hdiv space, analogy to
>    FE_RaviartThomas, but in this case a space of divergence-free functions
>    that satisfy div F = 0 within the reference cell
>       - this I assume will be more difficult, but here I am only after
>       linear functions
>
>
> Reasoning behind usage of these is in https://github.com/l-korous/mh
> deal/blob/master/papers/Compumag2017_MHD.pdf, section IV.A (divergence
> free space), IV.B (Vertex-based limiting which needs the Taylor-basis FE
> space). Currently the code in https://github.com/l-korous/mh
> deal/blob/master/code/ is failing for DG order > 0 because of undershoots
> and overshoots present in the unlimited flow solution, mag field works, but
> there is no divergence cleaning employed.
>
>
> Could you please point me in the right direction where to start with this
> and what all needs to be done for these two new shapesets / spaces to be
> employed?
>
>
> Many thanks
>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/fo
> rum/dealii?hl=en
> ---
> You received this message because you are subscribed to the Google Groups
> "deal.II User Group" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/d/optout.
>

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to