Parent Directory
| 
Revision Log
| 
Revision Graph
| Links to HEAD: | (view) (download) (annotate) |
| Links to checkpoint59r: | (view) (download) (annotate) |
| Sticky Tag: |
Clean
More hidden recomputations, this time more thoroughly: avoid seaice_model and load_fields_driver
Avoid hidden recomp. of do_oceanic_phys (needs clean-up)
implementing TAF directives for gchem,cfc,dic
One more store for NLFS
Put in place a i-j-k comlev storing for messy Leith scheme loop. Allowing recomputation produces wrong TAF adjoint so we have to store here (if we want to include this code).
Modify few stores.
split PTRACERS.h in 2 header files: PTRACERS_FIELDS.h & PTRACERS_PARAMS.h
Put back some AD stuff (diva-testing)
Toward fixing multi-dim. adv. adjoint?
Put back IF(useSEAICE) for adjoint (seems benign).
allow user defined cost function contributions CPP: ALLOW_GENCOST_CONTRIBUTION USAGE: compute contributions in cost_gencost_all.F
remove THSICE_2DYN.h
internal parameter controls: completements of implementations
Fix TAF problems when including correction code for source/sink of tracer due to the linear free surface (S/R CALC_WSURF_TR) (I had removed the ifndef ALLOW_AUTODIFF_TAMC, thought it was trivial)
Further cleanup of top-level routines.
Re-organized adjoint checkpointing according to Matt Mazloff (but with modifs/completions/cleanups) Storing is now bundled into large arrays to optimized I/O (and second step in cleanup of the_main_loop)
Removing exf_clim code.
2nd set of modifs for thsice adjoint.
move EXF header files from lower_case.h to UPPER_CASE.h
Getting rid of mytime in adthe_main_loop for all.
move calls to MONITOR & DO_THE_MODEL_IO from the_main_loop.F to the end of initialise_varia.F
Change common block size
seaice adjoint, part 2 (SEAICE_ALLOW_EVP).
Updating seaice adjoint, step 1 (everything, except SEAICE_EVP).
Port ptracers comlev1 from model/src/
Add storing for ptracer
And more changes, adapting from forward_step
Add missing end-of-loop call to diagnostics in ECCO-specific main loop
o Yet another change to suppress recomp. loop for DIVA while keeping monitor and diagnostics (but suppress checkpointing) o Catch a few divisions by zero for non-used cost terms (weights=0)
improve pkg/profiles namelist structure & fix for divided adjoint
Attempt to resolve some DIVA-related issues (mixed success).
Added CPU time numbers for the performance statistics
consistent with addition of ptracer and ssh in pkg/profiles
o Add hooks for rbcs checkpointing o Add rbcs_init_varia
o Ported new load_fields_driver structure over to ECCO o adjusted exf and seaice store directives accordingly
Fix stupid typos
Make pkg/profile fully independent of ecco,cost, etc. stuff to be able to use it in pure forward.
Updates to support PCL performance counters, fix real*4 bug for PAPIS and enhance PAPI counter support (including IPC per timestep output)
o SEAICE_CGRID adjoint, part 2. (all stores seem to be sorted out, but NANs in adjoint).
More storing for NLFS
Merging NLFS stores from forward_step.F to the_main_loop.F
Change of EmPmR in external_forcing_surf necessitates re-arranging adjoint STORE (what a pain).
Fix GAD keys that are now also used by seaice.
Move the mean flux terms to cost_forcing
Updated code to provide timing information per timestep to include cases where there is no pressure solve by moving the code to the main loop. Modified the timing in the pressure solver for the case where TAF needs to be fooled. Added support for PAPI summary MFlop/s information per timestep along side the simple (user/system/wallclock) timing. To use PAPI one currently needs to have declared PAPIINC and PAPILIB in the optfile for the include and library linking stings.
Updating calls from forward_step
Fix call to DO_STATEVARS_TAVE.
Adding DIAGNOSTICS hooks to ecco-specific the_main_loop.
o replace various cost routines by single generic routine cost_generic.F o add weights for SST, SSS control o bracket GAD.h for #undef ALLOW_GENERIC_ADVDIFF version
o Adding cost term for seaice obs (daily SMR vs. AREA) new seaice_cost init/weight/final routines o Modularized cost_averages routines (new cost_averagesgeneric.F)
Moving seaice-related cost to pkg/seaice/
Implement 4-level checkpointing. Needed for high-res. runs with large global files to be able to limit tapelev size (otherwise a potential show-stopper).
Adding precip control
use baseTime as time origin ; DIFF_BASE_MULTIPLE replaces DIFFERENT_MULTIPLE
Move call to dummy_in_stepping
Adding eddy stress controls a la Ferreira et al.
Add hooks for inAdExact
Re-instate useSeaice flag. May come in handy also for the adjoint.
2-level checkpoint switch ported to pkg/ecco Noww tested for Exp. 1 of natl_box_adjoint/
Restoring adjoint.
o Fixing directives for non-ECCO forward_step, after ECCO directives have been fixed to avoid recomp. of do_oceanic_phys o Works well for global_ocean.90x40x15, but recomp. issue remains for carbon
o some delicate re-shuffle of store directives to avoid one extra call of do_oceanic_physics o NB: this may break global_ocean adjoint temporarily, but it is clear how to fix it. Will do later, need this now.
updating global1x1_tot
o ECCO specific cost function terms (up-to-date with 1x1 runs) o ecco_cost_weights is modified to 1x1 runs o modifs to allow observations to be read in as single file or yearly files
o bringing adjoint up to date for sheduled c55
max number of passive tracers is now defined in PTRACERS_SIZE.h
o Additional storing in an attempt to fix TAMC problem with adjoint of KPP code. (no success...) Identical code works fine with TAF.
packages_write_pickup added to timestepping loop.
o paramter list of CALL ADTHE_MAIN_LOOP o initial do_the_model_io
fixes for DIVA
added missing cost_averages_bar_directives.h (plus name change)
o merging from ecco-branch o pkg/ecco now containes ecco-specific part of cost function o top level routines the_main_loop, forward_step supersede those in model/src/ previous input data.cost now in data.ecco (new namelist ecco_cost_nml)
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 |