[MITgcm-devel] seaice_calc_viscosities
Martin Losch
Martin.Losch at awi.de
Tue May 5 06:53:13 EDT 2015
Here's what I get, so there are differences and they are a little
ambiguous. I'd have to modify 3 experiments plus the forward
experiments. Shall I go ahead with this?
M.
bkli04l006:verification> cat tr_out.txt
Tue May 5 12:23:35 CEST 2015
run: ./testreport -adm -t 'global_ocean.cs32x15 lab_sea
offline_exf_seaice seaice_itd seaice_obcs' -j 8
on : Linux bkli04l006 3.13.0-49-generic #83-Ubuntu SMP Fri Apr 10
20:11:33 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux
No "OPTFILE" was specified ; genmake2 found and uses:
OPTFILE=../../../tools/build_options/linux_amd64_gfortran
Adjoint generated by TAF
default 10
G D M C A F
e p a R o d D
n n k u s G G
2 d e n t r r
Y Y Y Y 16>16<16 pass global_ocean.cs32x15 (e=0, w=50)
Y Y Y Y 16>14< 9 pass global_ocean.cs32x15.seaice
Y Y Y Y 8>10< 3 pass global_ocean.cs32x15.seaice_dynmix
Y Y Y Y 16>16<16 pass global_ocean.cs32x15.thsice
Y Y Y Y 16>13<16 pass lab_sea (e=0, w=7)
Y Y Y Y 16>16<16 pass lab_sea.noseaice
Y Y Y Y 16>16<16 pass lab_sea.noseaicedyn
Y Y Y Y 16>16<16 pass offline_exf_seaice (e=0, w=38)
Y Y Y Y 16>16<16 pass offline_exf_seaice.thsice
Start time: Tue May 5 12:23:35 CEST 2015
End time: Tue May 5 12:44:50 CEST 2015
On 05/01/2015 12:43 AM, Jean-Michel Campin wrote:
> Hi Martin,
>
> I think we can agree that the expression that ensure a positive value
> (even with some machine trucations) is better.
>
> The changes in the FWD verification experiments are not large (keep 11
> digits or more), so I would go for this modification and may be leave
> the previous expression commented out (but no new CPP options)
> We could always add this CPP option later on if, for backward compatibility,
> we need to bring back the old expression.
>
> But it would be good, before checking-in this modif, to check what the AD tests
> are doing.
>
> Cheers,
> Jean-Michel
>
> On Thu, Apr 30, 2015 at 06:48:27PM +0200, Martin Losch wrote:
>> Hi Jean-Michel,
>>
>> in seaice_calc_viscosities, there???s a piece of code (the computation of deltaC) that causes problems.
>>
>> this is how it should look, delta=sqrt( (eps11+eps22)**2 + (eps11-eps22)**2/e**2 + 4 eps12**2 /e**2 )
>> for some reason I expanded the squares and then I get this formulation:
>> deltaC(I,J,bi,bj) =
>> & (e11(i,j,bi,bj)**2+e22(i,j,bi,bj)**2)*(ONE+ecm2)
>> & + 4. _d 0*ecm2*e12Csqr(i,j)
>> & + 2. _d 0*e11(i,j,bi,bj)*e22(i,j,bi,bj)*(ONE-ecm2)
>>
>> Unfortunately, this can be less that zero in some rare cases (we have one at very high resolution) and then the sqrt(deltaC) produces NaN. I would like to change the code, but the many tests suffer from the truncation error. by up to 7digits (see below, I didn???t try the adjont yet). Should I include the modification within ifdefs for now, so that we can reproduce old results, or should I just update the tests in this case? I have it in the back of my head that I preferred the current version because it avoids evaluating e11-e22. Was that a good reason?
>>
>> Martin
>>
>> Thu Apr 30 18:21:36 CEST 2015
>> run: ./testreport -t 'global_ocean.cs32x15 lab_sea offline_exf_seaice seaice_itd seaice_obcs' -j 8 -q
>> on : Linux bkli04l006 3.13.0-49-generic #83-Ubuntu SMP Fri Apr 10 20:11:33 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux
>>
>> No "OPTFILE" was specified ; genmake2 found and uses:
>> OPTFILE=../../../tools/build_options/linux_amd64_gfortran
>>
>> default 10 ----T----- ----S----- ----U----- ----V----- --PTR 01-- --PTR 02-- --PTR 03-- --PTR 04-- --PTR 05--
>> G D M c m s m s m s m s m s m s m s m s m s
>> e p a R g m m e . m m e . m m e . m m e . m m e . m m e . m m e . m m e . m m e .
>> n n k u 2 i a a d i a a d i a a d i a a d i a a d i a a d i a a d i a a d i a a d
>> 2 d e n d n x n . n x n . n x n . n x n . n x n . n x n . n x n . n x n . n x n .
>>
>> Y Y Y Y>16<16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 pass global_ocean.cs32x15
>> Y Y Y Y>11<16 16 16 16 16 16 16 14 16 16 14 16 16 16 14 16 16 16 16 16 16 16 pass global_ocean.cs32x15.icedyn
>> Y Y Y Y>11<16 16 16 16 16 16 16 16 16 16 14 13 16 16 14 16 22 16 16 16 22 16 16 16 pass global_ocean.cs32x15.seaice
>> Y Y Y Y>16<16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 13 16 13 16 16 16 pass global_ocean.cs32x15.thsice
>> Y Y Y Y>16<16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 pass global_ocean.cs32x15.viscA4
>> Y Y Y Y>12<16 16 16 16 16 16 16 16 16 16 14 16 16 16 13 14 22 16 16 16 22 16 16 16 pass lab_sea
>> Y Y Y Y>16<16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 22 16 16 16 22 16 16 16 pass lab_sea.fd
>> Y Y Y Y>16<16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 22 16 16 16 22 16 16 16 pass lab_sea.hb87
>> Y Y Y Y>12<16 16 16 16 16 16 16 16 16 16 13 16 16 14 13 16 22 16 16 16 22 16 16 16 pass lab_sea.salt_plume
>> Y Y Y Y -- 16 16 16 13 16 16 16 16 12 13 13>12<13 13 12 13 pass offline_exf_seaice
>> Y Y Y Y -- 13 14 14 12 13 12 13 16 12 12 12>12<12 11 11 11 pass offline_exf_seaice.dyn_jfnk
>> Y Y Y Y -- 13 16 16 13 13 14 16 13 12 13 13>13<13 13 11 13 pass offline_exf_seaice.dyn_lsr
>> Y Y Y Y -- 16 16 16 16 16 16 16 22 22 16 16 16 22 16 16>16<pass offline_exf_seaice.thermo
>> Y Y Y Y -- 16 16 16 16 16 16 16 22 16 16 16 16 16 16 16>16<pass offline_exf_seaice.thsice
>> Y Y Y Y -- 16 16 16 16 16 16 16 16 12 14 12>12<13 16 12 13 pass seaice_itd
>> Y Y Y Y -- 16 16 16 13 16 16 16 16 12 14 13>12<13 16 13 13 pass seaice_itd.lipscomb07
>> Y Y Y Y -- 16 11 8 8 16 11 12 9 7 7 8> 7< 6 9 7 7 FAIL seaice_itd.thermo
>> Y Y Y Y>14<16 16 16 16 16 16 16 16 16 16 14 16 16 16 14 16 22 16 16 16 22 16 16 16 pass seaice_obcs
>> Y Y Y Y>13<16 16 16 16 16 16 16 16 16 16 13 14 16 16 16 16 22 16 16 16 22 16 16 16 pass seaice_obcs.seaiceSponge
>> Y Y Y Y>14<16 16 16 16 16 16 16 16 13 16 16 16 16 16 16 16 22 16 16 16 22 16 16 16 pass seaice_obcs.tides
>> Start time: Thu Apr 30 18:21:36 CEST 2015
>> End time: Thu Apr 30 18:27:26 CEST 2015
>>
>>
>> _______________________________________________
>> 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
>
--
Martin Losch
Alfred Wegener Institute for Polar and Marine Research
Postfach 120161, 27515 Bremerhaven, Germany;
Tel./Fax: ++49(0471)4831-1872/1797
More information about the MITgcm-devel
mailing list