[Carpet] missing data for grid component

[Erik Schnetter]

On Mar 6, 2007, at 05:59:31, Giovanni Corvino wrote:

> Hi, I have a problem with a compnent of my grid.
>
> I'm running with 2 refinement levels like
>
> [ [ ([-40.0,-40.0,0.0]:[40.0,40.0,18.0]:[1,1,1]) ],
>  [ ([-30.0,-30.0,0.0]:[30,30,8]:[0.5,0.5,0.5]) ] ]"
>
> the outer boundary is at 50 (bitant symmetry).
>
> First of all I get the following warning while generating the initial
> conditions:
>
> INFO (Carpet): Recomposing the grid hierarchy for map 0...
> Not properly nested: rl=1, required distance=4
> Received from level 0:
> bboxset<CCTK_INT4,3>:size=113049,setsize=5,set={([14,14,6]: 
> [30,186,54]:[2,2,2]),([32,14,6]:[186,30,54]:[2,2,2]),([170,32,6]: 
> [186,186,54]:[2,2,2]),([32,170,6]:[168,186,54]:[2,2,2]),([32,32,38]: 
> [168,168,54]:[2,2,2])}
> Taboo region on level 1:
> bboxset<CCTK_INT4,3>:size=238289,setsize=5,set={([6,6,-2]: 
> [38,194,62]:[2,2,2]),([40,6,-2]:[194,38,62]:[2,2,2]),([162,40,-2]: 
> [194,194,62]:[2,2,2]),([40,162,-2]:[160,194,62]:[2,2,2]), 
> ([40,40,30]:[160,160,62]:[2,2,2])}
> Sent to level 2:
> bboxset<CCTK_INT4,3>:size=35262,setsize=5,set={([36,36,8]: 
> [46,164,32]:[2,2,2]),([48,36,8]:[164,46,32]:[2,2,2]),([154,48,8]: 
> [164,164,32]:[2,2,2]),([48,154,8]:[152,164,32]:[2,2,2]),([48,48,22]: 
> [152,152,32]:[2,2,2])}
> Overlap on level 1:
> bboxset<CCTK_INT4,3>:size=13994,setsize=5,set={([36,36,8]: 
> [38,164,32]:[2,2,2]),([40,36,8]:[164,38,32]:[2,2,2]),([162,40,8]: 
> [164,164,32]:[2,2,2]),([40,162,8]:[160,164,32]:[2,2,2]),([40,40,30]: 
> [160,160,32]:[2,2,2])}
>
>
> and then wheh I look at the data I see that one or more of the
> components of the grid has not been computed, on all refinement  
> levels.
> Sometimes is only one component, sometimes two. What can it depend on?
> Caould it be something tied to the boudary conditions of the boxes?

Béla's answer is correct, there is some space necessary between  
different refinement level.  The warning you see is a warning that  
there was not enough space.  (Unfortunately, this warning is  
currently in terms of grid points and not coordinates, so that it is  
difficult to read.)

The problematic region -- the "overlap" between the two levels -- is  
described on the last line.  The overlap extends mostly from grid  
points 8 to 32 in the z direction.  Depending on your number of ghost  
and buffer zones, this could be near the lower z boundary, i.e., near  
the symmetry boundary.  Maybe, as Béla, says, there is not enough  
space on the coarse grid to interpolate to the fine grid.

However, I cannot fully tell without knowing the details.  If the  
problem persists, can you send the parts of the parameter file that  
set up the grid hierarchy?  How many processors did you use?  What  
did you mean by "component" in your description above?

You can enable debug output by setting CarpetLib::output_bboxes=yes.   
This will produce a scary amount of output, but it often helps to  
pinpoint the problem.

-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/20070306/52b74ad7/attachment.pgp