Parent Directory
|
Revision Log
|
Revision Graph
Links to HEAD: | (view) (download) (annotate) |
Links to checkpoint65b: | (view) (download) (annotate) |
Sticky Tag: |
update experiment after (hopefully) fixing the adjoint of sea ice dynamics
update results after changes made in pkg/seaice/seaice_lsr.F
update results after changing STORE directives and (more importantly) CPP flags in SEAICE_VECTORIZE_LSR and SEAICE_LSR_ADJOINT_ITER
put back previous output (from before yesterday update)
Modified output fullowing An Nguyen's changes to pkg/salt_plume (carry corresponding heat flux along with salt redistribution)
update AD output after removal of TICE from SEAICE.h
update output after changing global_sum call in ecco_cost_final.F: affects Fwd grad: TLM: 5 , ADM: 5 , 8 (noseaice), 5 (noseaicedyn) but all @ 16 before
- update results after change of experiment seaice.
* #undef SEAICE_EXCLUDE_FOR_EXACT_AD_TESTING * #define ALLOW_SST_CONTROL & ALLOW_SST_CONTROL * nTimeSteps 3 -> 4
- update results at truncation error level. Y Y Y Y 16>12< 5 pass lab_sea Y Y Y Y 13> 9< 3 FAIL lab_sea.evp Y Y Y Y 16>12< 4 pass lab_sea.noseaicedyn Y Y Y Y>10<16 16 16 16 16 16 16 16 16 16 13 16 16 16 12 16 22 16 16 16 22 13 16 16 pass lab_sea Y Y Y Y>12<16 16 16 16 16 16 16 16 16 16 13 16 16 16 13 16 22 16 16 16 22 13 16 16 pass lab_sea.fd Y Y Y Y> 3< 8 9 9 8 13 11 10 8 5 7 2 4 6 6 3 5 22 16 6 6 22 5 6 6 FAIL lab_sea.hb87 Y Y Y Y>14<16 16 16 16 16 16 16 16 16 16 13 16 16 16 14 16 22 16 16 16 22 16 16 16 pass lab_sea.salt_plume
un-do previous modif after fixing calculation of recip_nDim (precision issue)
update output after replacing few divisions by multiplication by reciprocal; this affects results at the machine truncation level: Y Y Y Y 12> 7< 4 FAIL lab_sea Y Y Y Y 9> 5< 2 FAIL lab_sea.evp Y Y Y Y 12> 7< 4 FAIL lab_sea.noseaicedyn
- update results that changed at truncation level (except evp) fwd Y Y Y Y>11<13 16 16 16 16 16 16 16 16 16 14 16 16 16 12 16 22 16 16 16 22 16 16 16 . . . . . . . . . . . . pass lab_sea Y Y Y Y>12<16 16 16 16 16 16 16 16 16 16 13 16 16 16 13 16 22 16 16 16 22 16 16 14 . . . . . . . . . . . . pass lab_sea.fd Y Y Y Y> 2< 8 9 8 7 13 11 9 6 5 7 2 4 6 6 3 4 22 16 6 6 22 4 5 5 . . . . . . . . . . . . FAIL lab_sea.hb87 Y Y Y Y>14<16 16 16 16 16 16 16 16 16 16 13 16 16 16 14 16 22 16 16 16 22 16 16 16 . . . . . . . . . . . . pass lab_sea.salt_plume ad Y Y Y Y 16>12< 4 pass lab_sea Y Y Y Y 12> 8< 2 FAIL lab_sea.evp Y Y Y Y 16>16<16 pass lab_sea.noseaice Y Y Y Y 16>13< 4 pass lab_sea.noseaicedyn
changes in seaice_solve4temp.F (SOLVE4TEMP_LEGACY) affects results at the truncation level (cg2d: 11 digits match except if using evp): update output.
seawater freezing temp changed from hard coded 271.2 (=previous value and current setting in "data" file of celsius2K + default value of SEAICE_freeze) to celsius2K+SEAICE_freeze : this introduce machine truncation differences: Y Y Y Y 15>12< 5 FAIL lab_sea Y Y Y Y 4> 7< 3 FAIL lab_sea.evp Y Y Y Y 4> 7< 4 FAIL lab_sea.noseaicedyn
update results after a) changing EXF_LWDOWN_WITH_EMISSIVITY (now #define in ECCO_CPPOPTIONS.h) and using pkg/exf default emissivities. and b) using default SEAICEuseFluxForm with simple harmonic diffusivity
Reduced from 4 to 3 timesteps (avoids td. gradient problems as temporary solution), and reflect changes in seaice_lsr adjoint
update output after changing seaice_lsr.F (add parenthesis in Water-Ice drag calculations; simplify and fix AA3 setting)
update output after adding Finite-difference gradient report (+ changing description of cost-function & adjoint gradient output)
Update results which now includes adjoint monitor for exf, seaice
identical results but changes in printed debug messages
Updating of verification experiments now that SEAICE_SALINITY has been changed to SEAICE_VARIABLE_SALINITY and SEAICE_salinity has been changed to SIsalFRAC. Noticed that ICE2WATR was still being used to initialize the HSALT in the lab_sea verification experiments. Removed it from the code but that broke testreport. Hence, some lab_sea/results/*txt files had to be replaced. Now pretty sure that ICE2WATR has been complete neutralized. : Modified Files: : lab_sea/results/output_adm.evp.txt : lab_sea/results/output_adm.noseaicedyn.txt : lab_sea/results/output_adm.txt : lab_sea/input_ad/data.seaice lab_sea/input_ad.evp/data.seaice : lab_sea/input_ad.noseaicedyn/data.seaice : lab_sea/input.salt_plume/data.seaice : lab_sea/code_ad/SEAICE_OPTIONS.h lab_sea/code/SEAICE_OPTIONS.h : 1D_ocean_ice_column/input_ad/data.seaice : 1D_ocean_ice_column/input/data.seaice : 1D_ocean_ice_column/code_ad/SEAICE_OPTIONS.h : 1D_ocean_ice_column/code/SEAICE_OPTIONS.h : seaice_obcs/input/data.seaice : seaice_obcs/code/SEAICE_OPTIONS.h : global_ocean.cs32x15/code_ad/SEAICE_OPTIONS.h : global_ocean.cs32x15/code/SEAICE_OPTIONS.h : : offline_exf_seaice/code_ad/SEAICE_OPTIONS.h : offline_exf_seaice/code/SEAICE_OPTIONS.h : ----------------------------------------------------------------------
generate reference output using reference code (which does not produce any lines starting with 'sfs ' or 'efs ' as in previous output)
Verification experiment update 1) Remove "backward compatability only" lines from lab_sea, offline_exf_seaice, and seaice_obcs data.seaice files < # for backward compatibility only: < SEAICE_cpAir = 1.0039560439560439e+03, < SEAICE_lhSublim = 2.8340219780219775e+06, < SEAICE_rhoAir = 1.3E0, 2) Added rhoConstFresh = 999.8 in input*/data where no rhoConstFresh was previously defined 3) Update output*.txt where necessary Modified Files: 1D_ocean_ice_column/input/data.seaice 1D_ocean_ice_column/results/output.txt lab_sea/input_ad/data lab_sea/input_ad/data.seaice lab_sea/input_ad.evp/data lab_sea/input_ad.evp/data.seaice lab_sea/input_ad.noseaicedyn/data.seaice lab_sea/input_ad_seaice_ocean/data lab_sea/input_ad_seaice_ocean/data.seaice lab_sea/input_ad_seaice_only/data lab_sea/input_ad_seaice_only/data.seaice lab_sea/input/data lab_sea/input/data.seaice lab_sea/input.fd/data.seaice lab_sea/input.hb87/data lab_sea/input.hb87/data.seaice lab_sea/input.salt_plume/data lab_sea/input.salt_plume/data.seaice lab_sea/results/output_adm.evp.txt lab_sea/results/output_adm.noseaicedyn.txt lab_sea/results/output_adm.noseaice.txt lab_sea/results/output_adm.txt lab_sea/results/output.fd.txt lab_sea/results/output.hb87.txt lab_sea/results/output.salt_plume.txt lab_sea/results/output.txt offline_exf_seaice/input_ad/data offline_exf_seaice/input_ad/data.seaice offline_exf_seaice/input/data offline_exf_seaice/input.seaicetd/data offline_exf_seaice/input.seaicetd/data.seaice offline_exf_seaice/results/output_adm.txt offline_exf_seaice/results/output.seaicetd.txt seaice_obcs/input/data seaice_obcs/input/data.seaice seaice_obcs/results/output.txt
Update lab_sea and seaice_obcs verification experiments #1) Removed ICE2WATR from input/ and input_ad/data.seaice #2) Updated results/output.txt and output_adm.txt in those verif. experiments that used the incorrect ICE2WATR values : Modified Files: : lab_sea/input_ad/data.seaice lab_sea/input_ad.evp/data.seaice : lab_sea/input_ad.noseaicedyn/data.seaice : lab_sea/input_ad_seaice_ocean/data.seaice : lab_sea/input_ad_seaice_only/data.seaice : lab_sea/input/data.seaice lab_sea/input.fd/data.seaice : lab_sea/input.hb87/data.seaice : lab_sea/input.salt_plume/data.seaice : lab_sea/results/output_adm.evp.txt : lab_sea/results/output_adm.noseaicedyn.txt : lab_sea/results/output_adm.noseaice.txt : lab_sea/results/output_adm.txt : lab_sea/results/output.salt_plume.txt : seaice_obcs/input/data.seaice : seaice_obcs/results/output.txt : ----------------------------------------------------------------------
- update results after seaice_growth.F rev. 1.110 - the changed order of mult. factors lead to Y Y Y Y>11<16 16 16 16 16 16 16 16 16 16 13 16 13 13 12 14 . . . . . . . . . . . . . . . . . . . . pass lab_sea Y Y Y Y>12<16 16 16 16 16 16 16 16 16 13 13 16 16 16 13 14 . . . . . . . . . . . . . . . . . . . . pass lab_sea.fd Y Y Y Y> 2< 8 10 7 6 13 12 9 6 6 7 2 4 6 7 3 4 . . . . . . . . . . . . . . . . . . . . FAIL lab_sea.hb87 Y Y Y Y>14<16 16 16 16 16 16 16 16 16 16 13 16 16 16 14 16 . . . . . . . . . . . . . . . . . . . . pass lab_sea.salt_plume Y Y Y Y 16>13<pass lab_sea Y Y Y Y 16>12<pass lab_sea.evp Y Y Y Y 16>13<pass lab_sea.noseaicedyn
update results following revision 105 of seaice_growth.F that, on baudelaire with gfortran, had lead to Y Y Y Y 16>10<pass lab_sea Y Y Y Y 16>10<pass lab_sea.evp Y Y Y Y 16>15<pass lab_sea.noseaice Y Y Y Y 16>10<pass lab_sea.noseaicedyn
update results after seaice_growth.F revision 1.89
update results after seaice_growth.F revision 1.87 for ad experiments only
update results after seaice_growth.F revision 1.87
update results after seaice_growth.F revision 1.85
update results after seaice_growth.F revision 1.84
update results after seaice_growth.F revision 1.83
update results after seaice_growth.F revision 1.80
update results after seaice_growth.F revision 1.78
update results after seaice_growth.F revision 1.77
update output after changes made in exf_bulkformulae.F
Change gradient check to common position between 1-tile and 4-tile setup for better comparison.
Update results after correcting store dir.
Change results following switch to multi-tile
update output after adding missing "_d 0" in cost_readers.F, cost_readtopex.F and cost_readtopexmean.F (in pkg/ecco).
update output after gael fix in seaice_model.F
update output after changing CD-scheme Adams-Bashforth.
Adapt reference output to changed setup: o #define SEAICE_MULTICATEGORY o staggered time-stepping
new output after adding missing "_d 0" in HSNOW initialisation
new output after adding missing "_d 0" in swfrac.F
Update once again following bug(fix) in re-initialisation
Updating results following change of 3D to 2D arrays; Gradient checks have improved by an order of magnitude!
Modified output following changes in depth restrictions of cost functions
Update adjoint output following Martin's changes.
Make isbyte=8 default, update output.
Extend testing.
Revert to original ones (as of before rhoInSitu changes) for now.
lab_sea adjoint results change slighly following new variable rhoInSitu
Forgot to check these in yesterday.
Update (or add new) reference output * timestep shortended to 4 steps * extended cost (data.ecco) * additional verif. experiments
Shorten integration to 4 timesteps.
Modified ADM and TLM output after switch to multiDimAdvection and DST3 (=30)
update after adding "_d 0" in kpp routines (only get 8 matching digits) (but without recent params changes of Patrick to turn on Muti-Dim)
update output after changing EmPmR units: affects (@ truncation level) results (only 12 digits match).
Change reference output for adjoint yet again (after checking that recent fixes reproduce old reference): o SEAICE_EXCLUDE_FOR_EXACT_AD_TESTING o SEAICE_salinity = 0.3 o SEAICE_availHeatFrac = 0.8
Update lab_sea adjoint output which now reverts to LSR solver and SEAICE_SALT.
Update adjoint output after numerous changes (and prior to further numerous changes).
add new AD output (from today test on faulks) with new format for Cost & Gradient output (in preparation for changing AD testing)
This form allows you to request diffs between any two revisions of this file. For each of the two "sides" of the diff, select a symbolic revision name using the selection box, or choose 'Use Text Field' and enter a numeric revision.
ViewVC Help | |
Powered by ViewVC 1.1.22 |