[Carpet] problems applying RotatingSymmetry90 when both CCTK_INT and CCTK_REAL gridfns are present

[Erik Schnetter]

On Jun 14, 2006, at 12:10:00, Jonathan Thornburg wrote:

> Hi, Erik,
>
> I have a wierd error appearing which looks like it may be a problem
> with RotatingSymmetry90.

You are correct, Slab expects that all variables have the same type.   
A quick workaround is obvious; either apply the boundary conditions  
sequentially, or change your grid function to real.  Obviously, this  
will be inconvenient to you.

The real solution would be to enhance thorn Slab.  Slab also contains  
special cases for CCTK_REAL (which improve efficiency); there should  
probably also be special cases for CCTK_INT and CCTK_COMPLEX.

Most of the code in Slab handles the variables one after the other,  
so that different variable types could easily be supported.  The  
exception is the big MPI_Alltoallv in the middle which shuffles the  
data between the processors.  Making this more generic would also  
require introducing separate buffers for the different data types.

Maybe the easiest thing would be to have a wrapper which splits the  
input data into the different data types, and then calls the original  
function in a loop.  In fact, that shouldn't be difficult to do.

-erik

-- 
Erik Schnetter <schnetter at cct.lsu.edu>

My email is as private as my paper mail.  I therefore support encrypting
and signing email messages.  Get my PGP key from www.keyserver.net.



-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 186 bytes
Desc: This is a digitally signed message part
Url : /archives/developers/attachments/20060614/532ddcd1/attachment.pgp