[MITgcm-devel] Adjoint and Latest pkg/seaice modifs

[Martin Losch]

Thanks Patrick,

when I made these changes I had though that TAF would recognize that I  
had just moved some code, but I guess the problem is that I put it  
into seaice_lsr, that is called more than once and that  
stressDivergence is also used in the EVP code ... sorry.

Martin

On Jun 2, 2009, at 6:13 PM, Patrick Heimbach wrote:

> Hi there,
>
> I think I've fixed this for now
> (additional storing of stressDivergenceX/Y ).
>
> In the long run, once pkg/seaice is more stable (will it ever be? ;o)
> we might revise/overhaul the connection between all those IF's and  
> STORE's.
>
> Cheers
> -p.
>
> On Jun 2, 2009, at 6:07 AM, Martin Losch wrote:
>
>> Patrick,
>>
>> probably the culprit: my recent check-in when I moved the  
>> computation of hibler-bryan-stress in part into seaice_lsr.F (and  
>> put within if (ilcall.EQ.NPSEUDOSTEP)) then/endif).
>> We can either put all  IF ( useHB87StressCoupling ) THEN into ndef  
>> ALLOW_AUTODIFF, or do something about this if-statement.
>>
>> The rest of my modifications are only about diagnostics (and in  
>> fact, I had to modify the hb87 code to be able to do consistent  
>> diagnostics of yield curves, something, nobody will ever use,  
>> except for myself).
>>
>> Martin
>>
>>
>> On May 30, 2009, at 4:57 PM, Jean-Michel Campin wrote:
>>
>>> Hi Martin and Patrick,
>>>
>>> The 3 lab_sea AD tests are taking for ever to run:
>>> last night (early May 30), 70 to 75 minutes for each of them
>>> (there are 3) compare to 25 to 30 second before.
>>> I checked both on faulks and dickens.
>>>
>>> The only cvs check-in I've seeen (apart from Patrick's bug fix)
>>> during this period are those in pkg/seaice, so I suspect it's
>>> coming from there.
>>>
>>> I might disable this lab_sea AD test on few platforms (until it
>>> get fixed) since it's realy taking too long (~5.h on faulks).
>>>
>>> Cheers,
>>> Jean-Michel
>>>
>>> _______________________________________________
>>> 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
>
> ---
> Patrick Heimbach | heimbach at mit.edu | http://www.mit.edu/~heimbach
> MIT | EAPS 54-1518 | 77 Massachusetts Ave | Cambridge MA 02139 USA
> FON +1-617-253-5259 | FAX +1-617-253-4464 | SKYPE patrick.heimbach
>
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel