[MITgcm-support] Process tracking

Martin Losch martin.losch at awi.de
Mon Nov 21 08:41:23 EST 2016


Ryan’s solution is probably the safest one, but you can also use “global indices”. 

Here’s an excerpt of pkg/cost/cost_test.F that shows you how to define the global indices:
          do j=1,sNy
          jg = myYGlobalLo-1+(bj-1)*sNy+j
            do i=1,sNx
            ig = myXGlobalLo-1+(bi-1)*sNx+i
you can then use ig,jg to find out where you are on your global domain (1:nx,1:ny), see pkg/cost/cost_test.F for more details.

Martin

> On 19 Nov 2016, at 23:39, Ryan Abernathey <ryan.abernathey at gmail.com> wrote:
> 
> Define your forcing in terms of physical space coordinates (e.g. XC, YC) rather than grid indices and you will be fine. 
> 
> -Ryan
> 
> Sent from my iPhone
> 
> On Nov 19, 2016, at 5:24 PM, Bertrand Louis Rene Delorme <bdelorme at stanford.edu> wrote:
> 
>> Hi,
>> 
>> I added a body force to gV in external_forcing.f, which I apply only in a very specific region of my domain, and I ran my simulation by using only one process. Now, I would like to use several processes (by increasing nPx and nPy) but I do not find a way to know where I am in the domain to apply the body force at the same precise location as before. Is there a way of finding out from inside of external_forcing.f what the location of the particular tile is? If not, would you have any ideas on how to run my experiment in parallel?
>> 
>> Thanks,
>> 
>> Bertrand Delorme
>> 
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-support
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support




More information about the MITgcm-support mailing list