[MITgcm-devel] [MITgcm-cvs] MITgcm/pkg/salt_plume CVS Commit

[Jean-Michel Campin]

Hi Martin,

I am going to check-in your fix. 
Can wait for more than 4 months to get this fixed by S/R author.

Thanks,
Jean-Michel

On Thu, Feb 02, 2012 at 01:42:29PM +0100, Martin Losch wrote:
> what I forgot to say, is of course that the problem goes away, when I remove the masks for taux/y in seaice_get_dynforcing.F but that does not seem to be a good solution.
> 
> What also works is replacing
> if (tmpscal4.GT.0) with
> if (tmpscal3.LT.0. _d 0)
> because then the discrimintant tmpscal4 is not only larger 0, but also larger than tmpscal2^2 so that the outer sqrt is also defined
> 
> >From this and from explicitly trying, tt appears to me that is is the case of tmpscal3=0. that causes the problem.
> 
> M.
> 
> On Feb 2, 2012, at 11:37 AM, Martin Losch wrote:
> 
> > Hi Jean-Michel,
> > the floating point exception occurs in s/r seaice_freedrift, l122:
> > sol_n=sqrt(HALF*(sqrt(tmpscal4)-tmpscal2))
> > here sqrt(tmpscal4)-tmpscal2<0 (-1e-20) and the sqrt(-1e-20) fails.
> > 
> > One could catch this by changing the if-statement to
> > if (tmpscal4.GT.0. _d 0 .AND. sqrt(tmpscal4)-tmpscal2.GT.0. _d 0) then
> > 
> > But maybe there is something better? Like avoid the multiplication and division by SEAICE_waterDrag, which I do not see necessary in the first place, because taux/y is the air-ice stress already (this is a problem that you reported earlier, didn't you?)
> > 
> > Martin
> > 
> > 
> > On Feb 1, 2012, at 4:36 PM, Jean-Michel Campin wrote:
> > 
> >> Hi Martin,
> >> 
> >> The latest test on solasrv is better today, but there is still
> >> lab_sea.hb87 which stop with "Arithmetic Exception".
> >> It could be due to the changes I made on Monday, less likely
> >> due the changes I made yesterday (since this one is using
> >> VARIABLE_FREEZING_POINT). Do you have an idea ?
> >> 
> >> Otherwise, offline_exf_seaice.seaice is now passing (12)
> >> but was failing (9 digits) until yesterday, but tests on other
> >> platforms (as far as I can tell) did not change for this experiment.
> >> curious.
> >> 
> >> Cheers,
> >> Jean-Michel
> >> 
> >> On Tue, Jan 31, 2012 at 06:38:23PM +0100, Martin Losch wrote:
> >>> Hi Jean-Michel,
> >>> thanks for spotting this. Now fixed.
> >>> M.
> >>> On Jan 31, 2012, at 4:55 PM, Jean-Michel Campin wrote:
> >>> 
> >>>> Sorry, I did reply to mitgcm-cvs list !
> >>>> now fwd to mitgcm-devel list.
> >>>> Jean-Michel
> >>>> 
> >>>> On Tue, Jan 31, 2012 at 10:42:15AM -0500, Jean-Michel Campin wrote:
> >>>>> Hi Martin,
> >>>>> 
> >>>>> I am not sure if there is a problem here, but on latest solasrv
> >>>>> testreport result, lab_sea & seaice_obcs are not compiling with
> >>>>> this error in salt_plume_frac.f :
> >>>>>>      plumek(i) = 0._d 0
> >>>>>>                  ^
> >>>>>> "salt_plume_frac.f", Line = 786, Column = 21: ERROR: The kind type parameter must be a scalar integer constant.
> >>>>> Is it related to your changes ?
> >>>>> 
> >>>>> Thanks,
> >>>>> Jean-Michel
> >>>>> 
> >>>>> On Mon, Jan 30, 2012 at 06:18:15AM -0500, Martin Losch wrote:
> >>>>>> Update of /u/gcmpack/MITgcm/pkg/salt_plume
> >>>>>> In directory forge:/tmp/cvs-serv22267/pkg/salt_plume
> >>>>>> 
> >>>>>> Modified Files:
> >>>>>> 	salt_plume_frac.F 
> >>>>>> Log Message:
> >>>>>> o vectorize salt_plume package: 
> >>>>>> - add extra code to salt_plume_frac, so that i-loop vectorizes
> >>>>>> - add code to avoid divisions by zero
> >>>>>> - efficiency of salt_plume_frac is still sub-optimal, but acceptable
> >>>>>> 
> >>>>>> 
> >>>>>> _______________________________________________
> >>>>>> MITgcm-cvs mailing list
> >>>>>> MITgcm-cvs at mitgcm.org
> >>>>>> http://mitgcm.org/mailman/listinfo/mitgcm-cvs
> >>>>> 
> >>>>> _______________________________________________
> >>>>> MITgcm-cvs mailing list
> >>>>> MITgcm-cvs at mitgcm.org
> >>>>> http://mitgcm.org/mailman/listinfo/mitgcm-cvs
> >>>> 
> >>>> _______________________________________________
> >>>> 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
> >> 
> >> _______________________________________________
> >> 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
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel