| 3 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 4 |
============================== |
============================== |
| 5 |
|
|
| 6 |
|
o modifications to make FREEZE flux visible to pkg/kpp |
| 7 |
|
- moved surfaceTendencyTice from pkg/seaice to main code |
| 8 |
|
- FREEZE moved to FORWARD_STEP |
| 9 |
|
- subroutine FREEZE now limits only surface temperature |
| 10 |
|
this means new output.txt for global_ocean.90x40x15, |
| 11 |
|
global_ocean.cs32x15, and global_with_exf, but note |
| 12 |
|
that results for these three experiments remain |
| 13 |
|
bit-identical to before if allowFreezing=.FALSE. |
| 14 |
|
o added surface flux output variables to TIMEAVE_STATVARS |
| 15 |
|
o time-averaged output for pkg/ptracers |
| 16 |
|
|
| 17 |
|
checkpoint52a_pre |
| 18 |
|
o some bug fixes for #undef REAL4_IS_SLOW |
| 19 |
|
|
| 20 |
|
checkpoint52 = ecco_c52_e35 |
| 21 |
|
o after testing adjoints of checkpoint51u_post |
| 22 |
|
|
| 23 |
|
checkpoint51u_post |
| 24 |
|
o this is the post-merge tag of the ecco-branch |
| 25 |
|
(needs another clean checkout and re-testing) |
| 26 |
|
o various natl_box_adjoint configs and results |
| 27 |
|
but no input fields |
| 28 |
|
o obcs adjoint not yet merged! |
| 29 |
|
o adding ECCO global 1x1 and 2x2 degree configs |
| 30 |
|
o updated/extended some store directives for seaice |
| 31 |
|
o pkg/ecco now containes ecco-specific part of cost function |
| 32 |
|
o top level routines the_main_loop, forward_step |
| 33 |
|
supersede those in model/src/ |
| 34 |
|
previous input data.cost now in data.ecco |
| 35 |
|
(new namelist ecco_cost_nml) |
| 36 |
|
o in ctrl/ cleaned some cross-dependencies and updated CPP options |
| 37 |
|
|
| 38 |
|
checkpoint51t_post |
| 39 |
|
o this is the pre-merge tag of the ecco-branch |
| 40 |
|
|
| 41 |
|
checkpoint51s_post |
| 42 |
|
o cleanup: removed various obsolete #define-s |
| 43 |
|
o pkg/grdchk |
| 44 |
|
- renamed 'ph-grd 3' to 'grad-res' |
| 45 |
|
- merged from ecco-branch |
| 46 |
|
(remaining bug fixes for obcs gradient checks) |
| 47 |
|
- additional high-precision output for testreport |
| 48 |
|
(grep for 'precision_grdchk_result') |
| 49 |
|
|
| 50 |
|
checkpoint51r_post |
| 51 |
|
o cleaned up or removed CPP_OPTIONS.h files |
| 52 |
|
o updated results for some of the tests to reflect the new |
| 53 |
|
optfiles (-ieee: -O0 instead of -O3) |
| 54 |
|
|
| 55 |
|
checkpoint51q_post |
| 56 |
|
o cd_code: |
| 57 |
|
- moved cd_scheme.F -> cd_code_scheme.F |
| 58 |
|
- separate read_checkpoint from cd_code_ini_vars.F |
| 59 |
|
- separated cd_code part from write_checkpoint |
| 60 |
|
- updated AD_SOURCE, generated .flow |
| 61 |
|
- added CD_CODE_VARS.h to the_main_loop |
| 62 |
|
- Moving cd_code specific store dir.s to pkg/cd_code |
| 63 |
|
o ctrl package totally restructured |
| 64 |
|
- pack/unpack now optional and decoupled from |
| 65 |
|
xx_/adxx_ I/O |
| 66 |
|
- ctrl_pack/unpack cleaned |
| 67 |
|
(new routines ctrl_init_ctrlvar.F, pkg/ctrl/ctrl_init_wet.F) |
| 68 |
|
- confined inclusion of AD_CONFIG.h to where necessary. |
| 69 |
|
o genmake2: changed AD dependencies from SRCFILES to AD_FILES |
| 70 |
|
|
| 71 |
|
checkpoint51p_post |
| 72 |
|
o create initial cd_code package |
| 73 |
|
- WARNING: the initialization of fields within packages |
| 74 |
|
is broken because INI_FIELDS (where READ_CHECKPOINT is |
| 75 |
|
called) is called *before* PACKAGES_INIT_VARIABLES. |
| 76 |
|
This seems to be backwards since |
| 77 |
|
1) its important to initially zero the entire field including |
| 78 |
|
the "halo" region |
| 79 |
|
2) zeroing needs to be done *before* reading pickup info |
| 80 |
|
and the pickup call might or might not happen in any |
| 81 |
|
particular run |
| 82 |
|
- this needs an overhaul for R2 |
| 83 |
|
o cleanup verification/*/code/GMREDI_OPTIONS.h |
| 84 |
|
o cleanup testreport |
| 85 |
|
|
| 86 |
|
checkpoint51o_post |
| 87 |
|
o bringing AD-related verif's up to date with genmake2 |
| 88 |
|
- carbon/ |
| 89 |
|
- global_ocean.90x40x15/ |
| 90 |
|
- hs.94.1x64x5/ |
| 91 |
|
- (front_relax/ ) |
| 92 |
|
o (more) consistent directory structure for AD-related verif.'s |
| 93 |
|
o cleaning ALLOW_GRADIENT_CHECK -> ALLOW_GRDCHK |
| 94 |
|
o cleaning some ALLOW_TANGENTLINEAR_RUN -> ALLOW_AUTODIFF |
| 95 |
|
o bug fix in find_alpha.F for MDJWF: |
| 96 |
|
- modif. to alpha = 1/D*( dN/dT - rho*dD/Dt) to account for |
| 97 |
|
change rho -> rho-rhoConst |
| 98 |
|
- replace call find_rho to find_rhonum |
| 99 |
|
o initialisation of rFlx extended to full array (required by TAF) |
| 100 |
|
and shifted to thermodynamics |
| 101 |
|
o removed PTRACERS.h in ptracers routine |
| 102 |
|
o added surfacetendencyPtr to S/R parameter list pracers_forcing |
| 103 |
|
o Correcting initialisations for fVerU, fVerV. |
| 104 |
|
o helping TAF to recognize dependency on kUp for |
| 105 |
|
fVerU, fVerV |
| 106 |
|
o Replacing delZ by delR in pkg/ctrl/ |
| 107 |
|
checkpoint51o_pre |
| 108 |
|
|
| 109 |
|
o undid all of cp51 pending some cleanups and discussion |
| 110 |
|
o added the [#include "AD_CONFIG.h"] statement to all files that need |
| 111 |
|
it for adjoint/tl #defines |
| 112 |
|
o re-worked the build logic in genmake2 to support AD_CONFIG.h |
| 113 |
|
o removed tools/genmake since it no longer works |
| 114 |
|
|
| 115 |
|
checkpoint51n_post |
| 116 |
|
o modifications to make FREEZE flux visible to pkg/kpp |
| 117 |
|
- moved surfaceTendencyTice from pkg/seaice to main code |
| 118 |
|
- FREEZE & EXTERNAL_FORCING_SURF moved to FORWARD_STEP |
| 119 |
|
- subroutine FREEZE now limits only surface temperature |
| 120 |
|
(this means new output.txt for global_ocean.90x40x15, |
| 121 |
|
global_ocean.cs32x15, and global_with_exf) |
| 122 |
|
o added surface flux output variables to TIMEAVE_STATVARS |
| 123 |
|
|
| 124 |
|
checkpoint51n_pre |
| 125 |
|
o added "#undef ALLOW_USE_MPI" support to eesupp/src/gather_2d.F |
| 126 |
|
o added [#include "AD_CONFIG.h"] to all files that use the following |
| 127 |
|
defines: |
| 128 |
|
- ALLOW_ADJOINT_RUN |
| 129 |
|
- ALLOW_TANGENTLINEAR_RUN |
| 130 |
|
- ALLOW_ECCO_OPTIMIZATION |
| 131 |
|
|
| 132 |
|
checkpoint51m_post |
| 133 |
|
o incorporating changes to get TAMC/TAF working in genmake2 |
| 134 |
|
- added new TAMC/TAF targets to the Makefile |
| 135 |
|
- various cleanups related to package handling |
| 136 |
|
- these changes should have no impact on the forward model |
| 137 |
|
|
| 138 |
|
checkpoint51l_post |
| 139 |
|
o changes to pkg/cal and pkg/exf that allow and speed-up long integrations |
| 140 |
|
- faster algorithm for cal_addtime.F |
| 141 |
|
- higher precision arithmetic, where needed |
| 142 |
|
- pkg/exf *startdate variables changed from calendar date to |
| 143 |
|
model integration time |
| 144 |
|
|
| 145 |
|
checkpoint51l_pre |
| 146 |
|
o add salt flux to forcing fields (used when salty sea-ice forms or melts) |
| 147 |
|
o enable to apply restoring on S & T only in a latitude band. |
| 148 |
|
o update output (external_forcing_surf modif. change truncation errors) |
| 149 |
|
global_ocean.cs32x15, global_ocean_pressure, lab_sea, natl_box |
| 150 |
|
|
| 151 |
|
checkpoint51k_post |
| 152 |
|
o change default to: useNHMTerms=.FALSE. |
| 153 |
|
- new output.txt for atmos. exp (NHMTerms were wrong but now turned off): |
| 154 |
|
adjustment.128x64x1, aim.5l_Eq_Ch, aim.5l_LatLon, hs94.128x64x5, hs94.1x64x5 |
| 155 |
|
- new output.txt for idealized oceanic exp (NHMTerms are now turned off): |
| 156 |
|
exp1, ideal_2D_oce |
| 157 |
|
- change data file of oceanic exp (leave the results unchanged for now) |
| 158 |
|
exp2, global_ocean[90x40x15,pressure,with_exf], lab_sea, natl_box |
| 159 |
|
o set gravitySign automatically according to the vertical coordinate used |
| 160 |
|
(and change sign: back to "logical" value: +1 when R increases downward) |
| 161 |
|
o added an include barrier to the default CPP_OPTIONS.h and |
| 162 |
|
removed the #include "PACKAGES_CONFIG.h" from it |
| 163 |
|
o fix a bug in mom_fluxform (from chk51j_post) |
| 164 |
|
|
| 165 |
|
checkpoint51j_post |
| 166 |
|
o Added some AD-related initialisations in mom_vecinv/ mom_fluxform/ |
| 167 |
|
o adjusted some flow directives (dynamics, thermodynamics) |
| 168 |
|
o added cubed-sphere case to test cost function |
| 169 |
|
o heckpoint_lev?_directives.h now include |
| 170 |
|
pkg-specific directives, whenever those are enabled. |
| 171 |
|
o Moving pkg-specific flow directives to corresponding pkg |
| 172 |
|
Moving pkg-specific differentiation list to corresponding pkg |
| 173 |
|
Moving pkg-specific checkpoint-lev storing to corresponding pkg |
| 174 |
|
---> conventions are |
| 175 |
|
- pkg_ad.flow |
| 176 |
|
- pkg_ad_diff.list |
| 177 |
|
- pkg_ad_check_lev?_dir.h |
| 178 |
|
|
| 179 |
|
checkpoint51i_post |
| 180 |
|
o merge of the branch-genmake2 |
| 181 |
|
- knowingly breaks the older genmake |
| 182 |
|
- automatic generation of all ALLOW_${PKG_NAME} defines |
| 183 |
|
- new testreport supersedes older testscript |
| 184 |
|
o large number of package inclusion and header cleanups |
| 185 |
|
checkpoint51i_pre |
| 186 |
|
|
| 187 |
|
o new output for adjoin of global ocean vector invariant |
| 188 |
|
o fix for checkpoint of thermodynamic seaice and bulk_force |
| 189 |
|
o Steph gchem various cleanups |
| 190 |
|
|
| 191 |
|
checkpoint 51h_pre |
| 192 |
|
o fixed some single quotes |
| 193 |
|
o modify gchem pkg to be more versatile |
| 194 |
|
o modifications to clean up dic pkg |
| 195 |
|
o add pkg/cfc |
| 196 |
|
|
| 197 |
|
checkpoint51g_post |
| 198 |
|
o provide links for ARPACK and ADM*TLM |
| 199 |
|
o start removing verification/vero/ |
| 200 |
|
o update AD-related stuff |
| 201 |
|
(in particular some IF-statements in model/src/ routines) |
| 202 |
|
o adding to verification/global_ocean.90x40x15/ adjoint of |
| 203 |
|
vector invariant setup |
| 204 |
|
|
| 205 |
|
checkpoint51f_post |
| 206 |
|
o update advect_xz/output.txt (left from chekpoint50f_pre, ini_vel modif) |
| 207 |
|
o Mods and bug fixes to pkg/cal, pkg/exf, etc., needed for computation |
| 208 |
|
of tracer Green's fucntions for ocean inversion project. |
| 209 |
|
|
| 210 |
|
checkpoint51f_pre |
| 211 |
|
o Added on-the-fly spatial interpolation capability |
| 212 |
|
"USE_EXF_INTERPOLATION" to pkg/exf. |
| 213 |
|
- This is a temporary Cartesian-grid hack until |
| 214 |
|
the super-duper ESMF coupler becomes available. |
| 215 |
|
- See verification/global_with_exf/README for usage example. |
| 216 |
|
- Removed obsolete EXFwindOnBgrid and SEAICEwindOnCgrid |
| 217 |
|
flags and modified pkg/seaice accordingly. |
| 218 |
|
o Bug fix to pkg/ptracers, pkg/generic_advdiff/gad_calc_rhs.F, |
| 219 |
|
and pkg/kpp/kpp_transport_ptr.F for dealing with tracer |
| 220 |
|
non-local transport term. |
| 221 |
|
|
| 222 |
|
checkpoint51e_post |
| 223 |
|
o pkg/mom_vecinv: |
| 224 |
|
- changes related to hFac: use S/R from aim.5l_cs/code (coriolis & vort.advec) |
| 225 |
|
- add time-ave diagnostic of vorticity advection. |
| 226 |
|
o add diagnostic (instantaneous) of Shapiro Filter effect for T,S & UV. |
| 227 |
|
o atmospheric geopotential: - include water vapor effect; |
| 228 |
|
- use p^kappa in p* scaling. |
| 229 |
|
o land pkg: read initial state from files. |
| 230 |
|
o AIM : add stratospheric drag in the upper level |
| 231 |
|
|
| 232 |
|
checkpoint51d_post |
| 233 |
|
o change function ILNBLNK (=> last non-blank char.) to return zero |
| 234 |
|
(instead of the string length) when the string. char is empty |
| 235 |
|
o fix bug in dxV,dyU definition of cubed-sphere grid (ini_curvilinear_grid) |
| 236 |
|
o modif's to restore TAMC compatibility |
| 237 |
|
- avoid "_d" constructs in PARAMETER declarations |
| 238 |
|
- avoid interrupting lines by comment lines |
| 239 |
|
o Use cluster local disks for purely local I/O |
| 240 |
|
vs. globally visible disks needed for ctrl stuff |
| 241 |
|
(merged from ecco-branch) |
| 242 |
|
- new active_file_loc... and mdsio_..._loc routines |
| 243 |
|
- affected packages: autodiff, ctrl, grdchk, mdsio |
| 244 |
|
|
| 245 |
|
checkpoint51c_post |
| 246 |
|
o introducing integer flag debugLevel |
| 247 |
|
- debLevZero: suppress all output |
| 248 |
|
- debLevA: standard |
| 249 |
|
- debLevB: equivalent to current debugMode=.TRUE. |
| 250 |
|
o introducing pathname variable mdsioLocalDir for mdsio |
| 251 |
|
that can be specified at runtime |
| 252 |
|
(relevant for cluster I/O to local disk) |
| 253 |
|
` |
| 254 |
|
>>>checkpoint51b_post |
| 255 |
|
adjoint of ptracers |
| 256 |
|
o disentangled ALLOW_PTRACERS using new ALLOW_GCHEM |
| 257 |
|
o extended state in checkpoint_lev?, etc. |
| 258 |
|
o alternative ctrl/cost with tr1 or ptracer |
| 259 |
|
o new comlev1_bibj_ptracers |
| 260 |
|
o new ptracers.flow |
| 261 |
|
o made ptracers_forcing same as external_forcing_s |
| 262 |
|
o added verification for adjoint with ptracers |
| 263 |
|
- adm. grdchk fail so far |
| 264 |
|
- tlm. grdchk OK |
| 265 |
|
- checked tapelev/comlev recomputations: OK |
| 266 |
|
<<<checkpoint51b_pre |
| 267 |
|
|
| 268 |
|
checkpoint51a_post |
| 269 |
|
o added code for biogeochemistry package |
| 270 |
|
- new pkgs: dic, gchem |
| 271 |
|
- new subroutine in ptracers |
| 272 |
|
o modify code for biogeo packages |
| 273 |
|
- in model/src and pkg/ptracers |
| 274 |
|
|
| 275 |
|
checkpoint51 |
| 276 |
|
o merged improved packages from ecco-branch (ecco_c51_e34) to MAIN |
| 277 |
|
- autodiff, ctrl, exf, grdchk, seaice |
| 278 |
|
kept separate packages |
| 279 |
|
- cost, ecco, obcs |
| 280 |
|
o updating verification/internal_wave, verification/lab_sea |
| 281 |
|
(all verif's pass on faulks.lcs.mit.edu RH7.3, g77) |
| 282 |
|
|
| 283 |
|
checkpoint50i_post |
| 284 |
|
o Preparing next differentiable checkpoint and sync |
| 285 |
|
of MAIN vs. ecco-branch |
| 286 |
|
- 1st step: this one is differentiable |
| 287 |
|
- 2nd step: next(! not this) checkpoint will have updated packages |
| 288 |
|
from ecco-branch) |
| 289 |
|
|
| 290 |
|
checkpoint50h_post |
| 291 |
|
o new package "land": provide surface BC for AIM. |
| 292 |
|
o pkg/ptracers: apply exchange to all tracers after reading a pickup. |
| 293 |
|
o AIM: use true log-P extrapolation for near surface temp (= as in |
| 294 |
|
F.Molteni paper) ; update aim.5l_cs output. |
| 295 |
|
o New variable in PARM03 for more flexible pickups |
| 296 |
|
o somewhat cleaned package initialisation sequence for |
| 297 |
|
ctrl/ cost/ ecco/ in packages_... routines |
| 298 |
|
|
| 299 |
|
checkpoint50g_post |
| 300 |
|
o merged with release1_p17 (pkg/seaice and verification/lab_sea) |
| 301 |
|
- added SEAICE_MULTILEVEL for 8-category sea-ice thermodynamics |
| 302 |
|
- LSR sea-ice dynamic solver moved to SouthWest B-grid location and |
| 303 |
|
made the default because of faster convergence than ADI |
| 304 |
|
|
| 305 |
|
checkpoint50f_post |
| 306 |
|
o exf totally restructured |
| 307 |
|
- replaced all exf_set_ routines by one generic exf_set_gen.F |
| 308 |
|
and made sure it's still differentiable |
| 309 |
|
- split exf_getffields.F into new exf_bulkformulae.F |
| 310 |
|
- exf_inscal_/exf_outscal_ suffixes consistent with field names |
| 311 |
|
- updated adjoint/makefile and relevant verif. accordingly |
| 312 |
|
|
| 313 |
|
checkpoint50f_pre |
| 314 |
|
o AIM: near surface temp evaluated from Pot.Temp: fixed for Part Cell ; |
| 315 |
|
+ change Evap over land. (update aim.5l_cs output) |
| 316 |
|
o added missing .cvsignores and made uniform across expts |
| 317 |
|
o changed configuration of tidal_basin_2d expt |
| 318 |
|
o added pkg/debug functions for tracing code in fatal crashes |
| 319 |
|
o tidied pkg/monitor to use MONITOR_OPTIONS.h |
| 320 |
|
o new fn in pkg/monitor to stop model if solution grows too big |
| 321 |
|
o removed lots of single quotes from comments |
| 322 |
|
|
| 323 |
|
checkpoint50e_post |
| 324 |
|
o make KPP work with PTRACERS |
| 325 |
|
- fix gad_calc_rhs to call new routine kpp_transport_ptr, which is |
| 326 |
|
nearly a copy of kpp_transport_s |
| 327 |
|
- there is no analogue to SurfaceTendencyS, so I have to use |
| 328 |
|
gPtr(of the surface layer) instead |
| 329 |
|
o add a new platform SunFire+mpi (SunFire 15000) to genmake |
| 330 |
|
checkpoint50e_pre |
| 331 |
|
|
| 332 |
|
checkpoint50d_post |
| 333 |
|
o change kpp output from multiple-record state files to single-record state |
| 334 |
|
files analogous to write_state.F |
| 335 |
|
o reduce the output frequency of cg3d-related stuff to the monitor frequency, |
| 336 |
|
analogous to the cg2d-related output. |
| 337 |
|
o fix small problem with in ptracers_write_checkpoint.F: len(suff)=512, |
| 338 |
|
so that writing to internal file fn (with length 512) fails. |
| 339 |
|
checkpoint50d_pre |
| 340 |
|
|
| 341 |
|
checkpoint50c_post |
| 342 |
|
o fix problem in ini_procs.F when using MPI (was there since checkpoint48f) |
| 343 |
|
o add monitor diagnostic of vorticity (mon_vort3.F) and potential energy. |
| 344 |
|
o add time-average diagnostic of U*V and vertical diffusive flux for theta. |
| 345 |
|
o bug fixed in convective adjustment diagnostic (multi-tile set-up). |
| 346 |
|
o 4th Order Adv scheme: return to previous (c40pre7) masking near boundaries. |
| 347 |
|
|
| 348 |
|
checkpoint50c_pre |
| 349 |
|
Merging from release1_p13: |
| 350 |
|
o bug fix for pkg/seaice dynamic solver |
| 351 |
|
o Added SEAICE_initialHEFF to pkg/seaice |
| 352 |
|
|
| 353 |
|
checkpoint50b_post |
| 354 |
|
o store u*,v* in gU,V instead of in guNm1,gvNm1 |
| 355 |
|
o in order to put the momForcing out of the Adams-Bashforth: |
| 356 |
|
move forcing & CD-scheme calls from mom_fluxform & mom_vecinv |
| 357 |
|
to timestep.F |
| 358 |
|
o new flag "useCDscheme" (default=F); replace guCD,gvCD by local arrays |
| 359 |
|
o re-generate output for highly sensitive test-exp: |
| 360 |
|
global_ocean_pressure, ideal_2D_oce, lab_sea, natl_box |
| 361 |
|
|
| 362 |
|
checkpoint50b_pre |
| 363 |
|
o restore the use of momAdvection & useCoriolis flags (mom_vecinv) |
| 364 |
|
o corrected CPP option mismatch in exf_Bulk... functions for case |
| 365 |
|
#define ALLOW_ATM_WIND && #undef ALLOW_ATM_TEMP |
| 366 |
|
|
| 367 |
|
checkpoint50a_post |
| 368 |
|
o Bug fix for merging between c50 and KPP. |
| 369 |
|
ikey was passed from thermodynamics to kpp_calc via |
| 370 |
|
common block rather than being recomputed in kpp_calc, |
| 371 |
|
in contradiction with new key itdkey. |
| 372 |
|
New key ikppkey created, and tamc.h headers updated. |
| 373 |
|
|
| 374 |
|
checkpoint50 |
| 375 |
|
- 2nd step of merge with ecco-branch: |
| 376 |
|
merge changes in ecco-branch back into main trunk |
| 377 |
|
- bug fix in find_alpha |
| 378 |
|
- more storing in gmredi_calc_tensor.F, kpp_routines.F |
| 379 |
|
- avoid iterative loop in ini_pressure for ALLOW_AUTODIFF_TAMC |
| 380 |
|
- merged some sliced AD support routines |
| 381 |
|
|
| 382 |
checkpoint49 |
checkpoint49 |
| 383 |
|
merged from ecco-branch: |
| 384 |
|
o exf: |
| 385 |
|
- Enable initialisation of forcing fields to constant |
| 386 |
|
(runtime) values. |
| 387 |
|
- in exf_getffields.F |
| 388 |
|
Reduce i-/j-loop to interior domain, discarding overlaps. |
| 389 |
|
That also fixes wrong TAF-key computations for key_1, key_2 |
| 390 |
|
with bulf formulae. |
| 391 |
|
- exf_init.F modify #ifdef for exf_init_evap |
| 392 |
|
- exf_getffieldrec.F, ctrl_getrec.F |
| 393 |
|
The following INT-usages are not safe: |
| 394 |
|
fldsecs = int(fldsecs/fldperiod)*fldperiod |
| 395 |
|
fldcount = int(fldsecs/fldperiod) + 1 |
| 396 |
|
and were modified. |
| 397 |
|
o autodiff: checkpoint_lev... |
| 398 |
|
- updated storing for bulk formulae (some arrays were missing) |
| 399 |
|
- added state for seaice package |
| 400 |
|
- synchronized with main branch |
| 401 |
|
o DIVA: |
| 402 |
|
Modif.s in |
| 403 |
|
- the_main_loop.F |
| 404 |
|
- the_model_main.F |
| 405 |
|
and added cost_final_restore.F cost_final_store.F |
| 406 |
|
o With these changes, it is possible to fully separate |
| 407 |
|
the first forward run to compute the cost function (mdthe_main_loop) |
| 408 |
|
and adjoint runs over individual DIVA intervals (adthe_main_loop) |
| 409 |
|
into separate model executions. |
| 410 |
|
This enables better tuning of DIVA to given queue sizes. |
| 411 |
|
o Loss of cost function value in consecutive DIVA runs is |
| 412 |
|
fixed by storing to 'costfinal' file. |
| 413 |
|
o Appropriate call of ctrl_unpack/_pack,grdchk_main at |
| 414 |
|
beginning/end of adjoint integration is accounted for. |
| 415 |
|
The current changes apply to TAF-1.4.26. |
| 416 |
|
TAF has one remaining bug, requiring one manual change |
| 417 |
|
(see MITgcm manual). |
| 418 |
|
o cal: synchronized between MAIN and ecco-branch |
| 419 |
|
|
| 420 |
|
checkpoint48i_post |
| 421 |
o Changes to restore differentiability of code w.r.t. previous tag |
o Changes to restore differentiability of code w.r.t. previous tag |
| 422 |
(mostly adding new routines to make list and replacing |
(mostly adding new routines to make list and replacing |
| 423 |
pressure by totPhiHyd). |
pressure by totPhiHyd). |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch