[FLASH-BUGS] Bug in Unsplit Hydro_computeDt

[Klaus Weide]

On Fri, 12 Feb 2016, Jason Galyardt wrote:

> ...                                                      There are two new
> bugs in source/physics/Hydro/HydroMain/unsplit/Hydro_computeDt.F90 on lines
> 188 and 190, also having to do with array indices:
> 
> !~~~~
> jmS = max(blkLimits(LOW,IAXIS), jmS)
> !...
> jpS = min(blkLimits(HIGH,IAXIS), jpS)
> !~~~~
> 
> The 'IAXIS' on both lines should be 'JAXIS', I believe. Here are the
> corrected lines:
> 
> !~~~~
> jmS = max(blkLimits(LOW,JAXIS), jmS)
> !...
> jpS = min(blkLimits(HIGH,JAXIS), jpS)
> !~~~~
> 
> Sorry about the confusion. Again, these bugs are also in
> source/physics/Hydro/HydroMain/unsplit_rad/Hydro_computeDt.F90.

Jason, 

you are right about this additional bug in Hydro_computeDt.F90.

Similar to the bug reported by Mark Richardson, this one has the 
potential to cause problems only under certain unusual conditions.
Specifically, all of the following must be true to cause problems:

 (1) 2D or 3D setup with unsplit Hydro or MHD;
 (2) non-square / non-cubic (logical) block sizes - 
     specifically, NXB .ne. NYB.
 (3) hy_cflStencil defined to be greater than 0 in Hydro_data.
 (4) A domain with non-periodic boundaries is used.
 (5) CFL_VAR is defined.


Klaus