[FLASH-USERS] potential issue on map particle to grid at periodic boundary

[Suoqing Ji]

Hi Klaus,

This patch exactly resolves the issue. Thank you!

Best wishes,
—
Suoqing JI
Ph.D Candidate
Department of Physics
University of California, Santa Barbara
http://physics.ucsb.edu/~suoqing <http://physics.ucsb.edu/~suoqing>
> On Mar 29, 2017, at 1:40 PM, Klaus Weide <klaus at flash.uchicago.edu> wrote:
> 
> On Mon, 27 Mar 2017, Suoqing Ji wrote:
> 
>> I’m mapping a few particle properties to grid variables, say, mapping 
>> from particle MASS_PART_PROP to grid PDEN with CIC method. I initialized 
>> uniformly distributed particles with sufficient large number, and set 
>> all the boundary conditions to be periodic. I tested three setups with 
>> UG and AMR:
>> 
>> Test 1: nxb=nyb=32, iProcs=2, jProcs=1
>> Test 2: nxb=nyb=16, iProcs=4, jProcs=2
>> Test 3: nxb=nyb=32, nblockx=2, nblocky=1, lrefine_max=1
>> 
>> All the tests have the same effective resolution, and should generate 
>> the same mapping to PDEN. However, both Test 2 and Test 3 produce single 
>> value PDEN as expected, while Test 1 produces artifacts at y-low and 
>> y-high boundaries (see attached figure).
>> 
>> I think this issue is related with the boundaries along which there’s 
>> only one block/processor with UG. Say, if I change set iProcs=1 in Test 
>> 1, the same artifacts also begin to appear at x-low and x-high 
>> boundaries. In addition, I did some additional tests with randomly 
>> distributed particles, and can still see the artifacts (but less 
>> obvious).
> 
> Please try the attached short patch and let me know whether it solves the 
> problem.
> 
> Klaus<getblkbc.diff>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20170329/2417079e/attachment.htm>