[ET Trac] [Einstein Toolkit] #1795: PITTNullCode results depend on number of processors

[Einstein Toolkit]

#1795: PITTNullCode results depend on number of processors
------------------------------------+---------------------------------------
  Reporter:  rhaas                  |       Owner:                     
      Type:  defect                 |      Status:  closed             
  Priority:  major                  |   Milestone:                     
 Component:  EinsteinToolkit thorn  |     Version:  development version
Resolution:  fixed                  |    Keywords:  PITTNullCode       
------------------------------------+---------------------------------------

Comment (by hinder):

 AEILocalInterp (is that the one being used?) by default adjusts the
 interpolation stencil if there are insufficient ghost zones, and the
 interpolation point is next to an interprocessor boundary.  To avoid this,
 you should use
 {{{
   boundary_off_centering_tolerance={0.0 0.0 0.0 0.0 0.0 0.0}
   boundary_extrapolation_tolerance={0.0 0.0 0.0 0.0 0.0 0.0}
 }}}
 in the interpolator parameter string.

 See
 https://bitbucket.org/einsteintoolkit/einsteinanalysis/src/69f49fffa270/Multipole/param.ccl.

 See also https://trac.einsteintoolkit.org/ticket/181, in which it was
 decided to make AEILocalInterp abort if this condition was detected,
 rather than changing the interpolation operator, but as far as I know this
 was not implemented.

-- 
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1795#comment:13>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit