[MITgcm-devel] gmredi taper_scheme stableGmAdjTap

[gael forget]

Hi Jean-Michel,

I do not have a clear recollection of the details (this code dates back to 2008…) but based on 
the associated comment in the code this "special choice for adjoint/optimization of parameters
(~ strong clipping, reducing non linearity of psi=f(K))” must have helped stabilize the adjoint.
That being said I agree that the factor 5 is odd and the hard-coding of slopeMax would
definitely be worth revisiting. Thanks for pointing this out.

Cheers,
Gael

On Oct 18, 2017, at 3:52 PM, Jean-Michel Campin <jmc at mit.edu> wrote:

> Hi,
> 
> I just had a quick look at the implementation of 'stableGmAdjTap' tapering
> scheme in pkg/gmredi/gmredi_slope_psi.F
> and I don't understand why the 2 components of the slope are multiplied by 5.
> 
> No very clear neither why run-time parameter "GM_maxSlope" is ignored when
> using stableGmAdjTap and replaced with hard-coded value (slopeMax= 2. _d -3 
> in gmredi_slope_limit.F and slopeMaxSpec=i 1. _d -4 in gmredi_slope_psi.F)
> 
> Cheers,
> Jean-Michel
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-devel

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 1839 bytes
Desc: not available
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-devel/attachments/20171018/2e2b9501/attachment.p7s>