[Carpet] Regridding problem

[Erik Schnetter]

Hi Uli,

I believe that bitant mode does not have this problem.  We have seen  
the problem mostly when two refined regions merge.

I checked the code, and I found that Carpet does in fact initialise  
the past time levels.  It may have instead been an error in the way  
the boundary conditions are scheduled to be applied.  After modifying  
MoL to apply the MoL_PostStep group automatically in the postregrid  
and postrestrict bins, the problem may already be solved.  I haven't  
tested this carefully yet.

Depending how you regrid and how you apply the boundary conditions,  
you may not have seen the problem at all.  One good way to test for  
the problem is 3D output of the constraints a few time steps after  
regridding; e.g. an isosurface with a value slightly above the  
"background noise" generated by the mesh refinement boundaries should  
show whether there are problems.

-erik

On Jun 16, 2007, at 09:38:38, Ulrich Sperhake wrote:

> Hi Erik,
>
> thank you for that information. I have one question: does this mean  
> that
> all simulations using symmetry will be subject to this problem? For
> example in a standard BH binary evolution in the xy-plane, the boxes
> moving with the holes will always have refined regions touching the
> equatorial plane, i.e. refinement boundary and symmetry boundary have
> non-zero overlap.
>
> Cheers,
>
> Uli
>
>
> On Wed, 13 Jun 2007, Erik Schnetter wrote:
>
>> Dear Carpet users,
>>
>> we have found a problem in Carpet's regridding algorithm.  After  
>> regridding,
>> the boundary conditions need to be re-applied if the refined  
>> regions are near
>> a boundary (e.g. a symmetry boundary).  Carpet does not apply the  
>> boundary
>> conditions to the past time levels, so that the next time  
>> interpolation leads
>> to wrong results at the mesh refinement boundaries.  This error is  
>> somewhat
>> difficult to see, since it is present mostly on the past time  
>> levels, which
>> are usually not output.
>>
>> I am working on a correction.  In the mean time, please be careful  
>> about
>> refinement boundaries near symmetry boundaries.
>>
>> -erik
>>
>>
>
> -- 
> ====================================================================== 
> ======
> Ulrich Sperhake                        Tel.  : ++49 3641 947146
> Institute of Theoretical Physics       Fax   : ++49 3641 947102
> Friedrich-Schiller University          Email : Ulrich.Sperhake at uni- 
> jena.de
> Max-Wien-Platz 1                       http://www.uni-jena.de/~p5spul2
> D-07743 Jena
> ====================================================================== 
> ======
> _______________________________________________
> developers mailing list
> developers at lists.carpetcode.org
> http://lists.carpetcode.org/listinfo/developers
>


-- 
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/20070616/52f5cf2e/attachment.pgp