[MITgcm-devel] reference for adjoint

[Martin Losch]

Thanks, I wasn't sure about the adjoint.

I decided to make only changes to growth that will not affect the adjoint solutions (so I am sticking with divisions like this one: 1./integer)

Martin

On Nov 11, 2010, at 5:29 PM, Jean-Michel Campin wrote:

> Hi Martin,
> 
> We switch to baudelaire.csail.mit.edu with gfortran 
> for reference output, both forward and adjoint.
> 
> Not yet decided regarding the 2 fizhi experiments
> (still faulks + pgf77).
> 
> Cheers,
> Jean-Michel
> 
> On Thu, Nov 11, 2010 at 10:52:54AM +0100, Martin Losch wrote:
>> Hi there,
>> 
>> I am making a small change in seaice_growth.F: something/MULTDIM -> something*rnDim where rnDim = 1/MULTIM and that changes the lab_sea adjoint results.
>> 
>> Just so that I do the right thing: which machine is now the reference for the adjoint tests?
>> 
>> Martin
>> 
>> 
>> _______________________________________________
>> 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