[FLASH-USERS] A minor bug in FieldLoop initialization

[Suoqing Ji]

Hi Dongwook,

Thanks for your info! I misread the code previously and I do agree with the cell corner vs. cell center scheme.

However, now I revise my concern into the following:

The i index goes up to (blkLimitsGC(HIGH,IAXIS)+1), thus xCoord(i) will go beyond index boundary when i=blkLimitsGC(HIGH,IAXIS)+1 (the same thing happens on zCoord(k)). Although cell center value should be used here, the boundary needs to be taken care of slightly. Say,

if (i <= blkLimitsGC(HIGH,IAXIS)) then
    x1 = (cos_ang * xCoord(i) + sin_ang * zz)
else
    x1 = (cos_ang * (xCoord(i-1) + dx) + sin_ang * zz)
endif

Thank you!

Best wishes,
—
Suoqing Ji
Ph.D Candidate
Department of Physics
University of California, Santa Barbara
http://physics.ucsb.edu/~suoqing

On May 22, 2017, 1:18 PM -0700, Dongwook Lee <dongwook at flash.uchicago.edu>, wrote:
> Hi Souqing,
>
> I think the code lines are correct. Please first note that the filed loop is rotated about the y-axis, following the transformation rule:
>
>  x1  = x*cos(ang) - z*sin(ang)
>  x2  = y
>  x3  = x*sin(ang) + z*cos(ang)
>
> Now, you want to define vector potentials to define divergence-free face magnetic fields using the following rules:
>
>  Bx = dAz/dy - dAy/dz
>  By = dAx/dz - dAz/dx
>  Bz = dAy/dx - dAx/dy
>
> When considering Ax in this case, you want to stagger (i.e., use coordinates at faces, rather than centers) y and z coordinates for the straightforward differencing. Since y is not rotated, you just need to stagger z-coordinate, leaving x-coordinate at cell-centers.
>
> Likewise, you would need to stagger x-y for Az, for which you just need x-staggering only, keeping z-coordinates at cell-centers.
>
> I double checked this with the code implementations in the ATHENA code setup, and they look the same as the FLASH implementation.
>
> Hope this helps.
>
> Best,
> Dongwook
>
> =========================================
> Dongwook Lee, Ph.D., Assistant Professor
> Applied Mathematics and Statistics
> University of California, Santa Cruz
> Baskin Engineering, Room 353C
> 1156 High Street, Santa Cruz, CA 95064
> https://users.soe.ucsc.edu/~dongwook/
>
>
>
> > On Fri, Apr 21, 2017 at 12:57 AM, Suoqing Ji <suoqing at physics.ucsb.edu> wrote:
> > > Hi,
> > >
> > > I would like to report a very minor bug in magnetoHD/FieldLoop/Simulation_initBlock.F90: in line 253 and 273, the variable xCoord(i) and zCoord(k) should be replaced by xx and zz.
> > >
> > > With current code, depending on platforms and ways of setup, the simulation has a certain possibility of hanging forever at initialization stage without any error output (sometimes it can run successfully). I encounter this issue when I use FieldLoop for performance tests on Stampede.
> > >
> > > Thanks!
> > >
> > > Best wishes,
> > > —
> > > Suoqing Ji
> > > Ph.D Candidate
> > > Department of Physics
> > > University of California, Santa Barbara
> > > http://physics.ucsb.edu/~suoqing
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20170522/7e161a37/attachment-0001.htm>