[MITgcm-devel] stripes in adjoint along tile edges

[Martin Losch]

Hi Torge and others,

just avoid any misunderstanding: this does not have anything to do with the sea ice model or any parts of it, the seaice code is turned off in the simulation (there might be another similar problem in the sea ice code, but that will not show up here).

In my example, these stripes appear only in the adjoint variables (visibly).

Martin


On Nov 20, 2014, at 9:54 PM, Torge Martin <torge.martin at gmail.com> wrote:

> Hi Martin, 
> 
> That looks a lot like the issue I had when I last tested the ITD code on a linux cluster using 24 cpus. Now I remember talking to Patrick about the possibility that the overlap regions might not be handled well in the ITD code. Have to check this. Would be great to hear how you solve your issue in the end.
> 
> Torge
> 
> On Thu, Nov 20, 2014 at 4:28 PM, Martin Losch <Martin.Losch at awi.de> wrote:
> Hi adjoint people,
> 
> while debugging, I stumbled over interesting stripes in the adjoint fields along tile edges. See attached files for an example (based on arctic_cs36 with 24 cpus, sNx = 35, sNy = 48), log10(sqrt(adU**2)) for layers 1 and 10 after 1 adjoint timestep (after evaluating the cost function). As a cost function I evaluate theta(80,30), at the spot with the highest sensitivity. The stripes have low magnitudes (1e-8), but still large compared to the surrounding values. Does that look familiar in any way? Should I be worried about this? (I am because the model later explodes along these stripes). 
> 
> Martin
> 
> PS. I am using cdTargetResidual=1e-13 and cg2dMaxIters=1000, and these stripes appear when the seaice package is turned off totally (forwards and backwards), with advection scheme 30 and 33 for T/S.
> 
> 
> <adU_k1_t1.png><adU_k10_t1.png>
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel