| 4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 5 |
============================== |
============================== |
| 6 |
|
|
| 7 |
|
o pkg/layers (adjust changes from Sept 19): |
| 8 |
|
- do snap-shot output and fill diagnostics from inside iLa loop (layers_calc.F) |
| 9 |
|
this allows to remove "layers_maxNum" dimension from all layers arrays. |
| 10 |
|
o pkg/layers (adjust changes from Sept 19): |
| 11 |
|
- remove ALLOW_LAYERS_OUTPUT ; fix snap-shot and time-ave output ; |
| 12 |
|
- snap-shot extended to all type of layers (not just the 1rst type); |
| 13 |
|
- remove layers variables when corresponding CPP option is undef. |
| 14 |
|
o pkg/layers: |
| 15 |
|
- rename pkg/layers variable and add new output (layer exist & un-weighted |
| 16 |
|
velocity in layer) |
| 17 |
|
o pkg/seaice: |
| 18 |
|
- simplify (?) ocean-drag coefficent computations by using |
| 19 |
|
the same routine in all three solvers (saves some code lines) |
| 20 |
|
- remove global fields DRAGA and DRAGS from C-grid code and |
| 21 |
|
declare them locally in lsr.F (the only place where they are |
| 22 |
|
still used) |
| 23 |
|
o pkg/seaice: add Jacobian-free Newton Krylov solver for ice dynamics |
| 24 |
|
(Lemieux et al. 2010, 2012). Turn on with the usual SEAICE_ALLOW_JFNK |
| 25 |
|
and SEAICEuseJFNK=.true. A few things to note: |
| 26 |
|
- involves a FGMRES, that is parallel but very inefficient, |
| 27 |
|
because of many scalar products (global sums) in the Gram-Schmidt |
| 28 |
|
orthogonalization method. Needs to be replaced. Original code |
| 29 |
|
from J.-F. Lemieux |
| 30 |
|
- preconditioner for FGMRES is a stripped down version of LSOR |
| 31 |
|
- SOLV_MAX_ITERS is now used for LSOR as a solver and as a |
| 32 |
|
preconditioner. For the preconditioner, the default is set to 10. |
| 33 |
|
- Jacobian times vector is computed with finite differences |
| 34 |
|
- marginally related to JFNK-solver: add a smooth regularization for |
| 35 |
|
zeta with tanh that will improve convergence (Lemieux and Tremblay, |
| 36 |
|
2009) for any implicit solver (not implemented for EVP because EVP |
| 37 |
|
does not need this regularization in the first place). Define |
| 38 |
|
SEAICE_ZETA_SMOOTHREG in SEAICE_OPTIONS.h to use. |
| 39 |
|
o eesupp/src: |
| 40 |
|
- check in genmake2 that intrinsic S/R flush is available. |
| 41 |
|
- add S/R to flush IO unit (if intrinsic S/R flush is available) |
| 42 |
|
- flush standard & error msg unit before MPI termination |
| 43 |
|
- if debugMode, flush IO unit after printing msg |
| 44 |
|
o pkg/ctrl: |
| 45 |
|
- ctrl_unpack.F, ALLOW_GENARR3D_CONTROL code: fix missing argument in call |
| 46 |
|
to ctrl_set_unpack_xyz. |
| 47 |
|
|
| 48 |
|
checkpoint64 (2012/10/09) |
| 49 |
|
o pkg/shelfice |
| 50 |
|
- in mom_fluxform.F, mom_vecinv.F: |
| 51 |
|
CALL SHELFICE_V_DRAG(bi,bj,k,vFld,KE,KappaRU,vF,myThid) |
| 52 |
|
Bug fix KappaRU -> KappaRV |
| 53 |
|
does not change verification/isomip but may change other setups |
| 54 |
|
(presumably constant KappaRU=KappaRV) |
| 55 |
|
- fix diagnostics SHIU/Vdrag (thanks to Tom Millgate) |
| 56 |
|
o pkg/obcs: |
| 57 |
|
- fixed some indexing bugs in obcs_seaice_sponge.F |
| 58 |
|
- added obcs relaxation subroutines for HSALT and HSNOW |
| 59 |
|
- changes verification/seaice_obcs/results/obcs_seaice_sponge.F |
| 60 |
|
o pkg/seaice: |
| 61 |
|
- add "multicategory" parameterization for snow in analogy to ice, |
| 62 |
|
default is off for now. |
| 63 |
|
o pkg/obcs: |
| 64 |
|
- added obcs_seaice_sponge.F for obcs relaxation of seaice variables; |
| 65 |
|
see verification/seaice_obcs/input.seaiceSponge for example application. |
| 66 |
|
- remove local commom bloc "OBCS_GLOBAL" in obcs_readparms.F and do all the OB |
| 67 |
|
index setting by master-thread. This fix multi-threads Pb in previous modifs. |
| 68 |
|
o pkg/layers : |
| 69 |
|
- add diagnostics to pkg/layers |
| 70 |
|
- move the meat of pkg/layers from layers_calc.F to layers_fluxcalc.F |
| 71 |
|
(from R.Tulloch). Layers_calc.F now calls layers_fluxcalc.F accordingly. |
| 72 |
|
o pkg/generic_advdiff/gad_implicit_r.F : |
| 73 |
|
- fix ADVr diagnostics in case of both implicit vertical advection and diffusion. |
| 74 |
|
o pkg/seaice/seaice_growth.F: |
| 75 |
|
- fix SItflux diagnostics in virtual salt flux + balanceEmpmr case. |
| 76 |
|
o pkg/obcs: |
| 77 |
|
- change default setting of OBCS_UVICE_OLD from defined to undef; |
| 78 |
|
- use new parameter OB_indexNone for null index value (instead of hard-coded 0) |
| 79 |
|
in all pkg/obcs code and in other pkgs; remove k loop in obcs_check_depths.F; |
| 80 |
|
- pass OB_indexNone as (new) argument to S/R CTRL_MASK_SET_[X,Y]Z; |
| 81 |
|
- change OB null index value (OB_indexNone) to -99 ; this fix the problem |
| 82 |
|
with Southern & Western OB at a tile edge (at least in a simple set-up). |
| 83 |
|
o pkg/obcs: |
| 84 |
|
- change settings of OB index in case OB coincides with tile edge (shared by |
| 85 |
|
2 tiles), so that common OB location is registered by both tiles (instead of |
| 86 |
|
only 1). However, for Southern and Western OB, index value = 0 is not |
| 87 |
|
recognized as valid OB index (=default null value) and needs to be fixed. |
| 88 |
|
- modify accordingly OBCS_BALANCE_FLOW and OBCS_MONITOR (avoid double counting) |
| 89 |
|
- new parameter "OB_indexNone" for null index value (for now, still zero) |
| 90 |
|
to replace hard-coded null index value of zero. |
| 91 |
|
o pkg/exch2: |
| 92 |
|
- improve W2_USE_E2_SAFEMODE (faster) and W2_E2_DEBUG_ON: |
| 93 |
|
always check buffer-size (but outside the loop); move checking for valid |
| 94 |
|
index from W2_USE_E2_SAFEMODE to W2_E2_DEBUG_ON; in W2_E2_DEBUG_ON, |
| 95 |
|
print each tile and point connection only if |W2_printMsg|>= 2 and 3. |
| 96 |
|
o pkg/ctrl (pack & unpack): |
| 97 |
|
- in ctrl_pack.F & ctrl_unpack.F, use the same local file-name variable |
| 98 |
|
(locally defined) for different variables (+ remove the variable-specific |
| 99 |
|
file name in CTRL_GENARR.h). For now, only implemented for GENARR2/3D_CONTROL |
| 100 |
|
(to be extended to other local file names in ctrl.h). |
| 101 |
|
o pkg/flt: |
| 102 |
|
- add a FLT_Iter0 option (similar to PTRACERS_Iter0). |
| 103 |
|
o pkg/mdsio: |
| 104 |
|
- fix missing STOP in mdsio_readvec_loc.F (case ioUnit=0) when file is missing |
| 105 |
|
o tools/genmake2: |
| 106 |
|
- add similar dependences for target EXE_AD & EXE_FTL as we have for EXECUTABLE |
| 107 |
|
(added by Alistair in version 1.51 for building without make depend); |
| 108 |
|
- remove Header files list (H_SRC_FILES) from the dependence of targets |
| 109 |
|
ad_input_code.f & ftl_input_code.f (not clear why it was added in the first |
| 110 |
|
place in version 1.23): this save going through staf when unrelated header |
| 111 |
|
file has been modified. |
| 112 |
|
Note: for an intermediate target (e.g., ad_input_code.f or adtaf) without |
| 113 |
|
"make depend" step, still need to do "make links" before "make adtaf". |
| 114 |
|
o eesupp/src (+ few pkg src files) & genmake2: |
| 115 |
|
- finish to remove ALWAYS_USE_MPI in source code that TAF does not see; |
| 116 |
|
- no longer set ALWAYS_USE_MPI in Makefile (was done by genmake2). |
| 117 |
|
- improve genmake2 checking for ALLOW_${PKG} in CPP_OPTIONS.h files (distinguish |
| 118 |
|
upper/lower case ; account for space between "#" and undef/define) |
| 119 |
|
o pkg/flt (with exch2): |
| 120 |
|
- cleaning: fix so that it compiles; move remaining MPI calls out of flt_exch2.F; |
| 121 |
|
- simplify threads synchronisation; fix (many issues) multi-threaded with MPI. |
| 122 |
|
|
| 123 |
|
checkpoint63s (2012/09/05) |
| 124 |
|
o pkg/ecco: |
| 125 |
|
- remove #ifdef ALLOW_SMOOTH_CORREL3D brackets. |
| 126 |
|
- add more relevant #ifdef ALLOW_SMOOTH ones. |
| 127 |
|
- sort out useAtmWind, useSMOOTH, ctrlSmoothCorrel2D. |
| 128 |
|
o pkg/ctrl: |
| 129 |
|
- CTRL_OPTIONS.h : document CPPs related to smoothing; ALLOW_SMOOTH_CORREL2D/3D, |
| 130 |
|
ALLOW_CTRL_SMOOTH, ALLOW_SMOOTH_CTRL2D, ALLOW_SMOOTH_CTRL3D. |
| 131 |
|
- ctrl.h : document run time params doInitXX, doMainPack, doMainUnpack, doPackDiag, |
| 132 |
|
doSinglePrecTapelev (audodiff param!), ctrlSmoothCorrel2D (new) and |
| 133 |
|
ctrlSmoothCorrel3D (new). |
| 134 |
|
The added ctrlSmoothCorrel2D/3D replace the old ALLOW_SMOOTH_CORREL2D/3D CPP. |
| 135 |
|
- ctrl_readparms.F : initialise ctrlSmoothCorrel2D/3D and read from namelist. |
| 136 |
|
- ctrl_init.F ctrl_pack.F ctrl_unpack.F : switch off ustress/vstress |
| 137 |
|
controls when useAtmWind, and vice versa. |
| 138 |
|
- ctrl_map_ini_ecco.F : UVEL0, VVEL0, edtaux, edtauy cannot be smoothed as tracers. |
| 139 |
|
- ctrl_get_gen.F : replace #ifdef ALLOW_SMOOTH_CORREL2D with |
| 140 |
|
#ifdef ALLOW_SMOOTH, IF ctrlSmoothCorrel2D, IF useSMOOTH. |
| 141 |
|
- ctrl_map_ini_gen.F : replace #ifdef ALLOW_SMOOTH_CORREL3D with |
| 142 |
|
#ifdef ALLOW_SMOOTH, IF ctrlSmoothCorrel3D, IF useSMOOTH. |
| 143 |
|
- ctrl_set_pack_xy[z].F, ctrl_set_unpack_xy[z].F: |
| 144 |
|
replace ALLOW_SMOOTH_CORREL2D/3D brackets with ctrlSmoothCorrel2D/3D. |
| 145 |
|
o pkg/smooth: |
| 146 |
|
- remove un-necessary includes. |
| 147 |
|
- remove ALLOW_SMOOTH*D and ALLOW_SMOOTH_CORREL*D brakets. |
| 148 |
|
Those CPP options were never defined, and not necessary. |
| 149 |
|
o model/src: |
| 150 |
|
- fix CPP brackets in ini_mixing.F (GM_background_K belongs to gmredi). |
| 151 |
|
- standard treatment of useSMOOTH. |
| 152 |
|
o pkg/autodiff: |
| 153 |
|
- add LOGICAL useGGL90inAdMode, useGGL90inFwdMode, useSALT_PLUMEinAdMode, |
| 154 |
|
useSALT_PLUMEInFwdMode allowing to switch off ggl90 and salt_plume adjoint. |
| 155 |
|
o pkg/exch2: |
| 156 |
|
- add "if usingMPI" where it was missing (in exch2_rx?_cube*.template) |
| 157 |
|
+ start to remove ALWAYS_USE_MPI in source code that TAF does not see. |
| 158 |
|
- make new S/R W2_MAP_PROCS out of w2_eeboot.F to setup exch2 CommonFlag |
| 159 |
|
and tiles to process mapping. |
| 160 |
|
- re-arrange 3 header files (W2_EXCH2_PARAMS.h, _BUFFER.h & _TOPOLOGY.h) |
| 161 |
|
moving /EXCH2_FILLVAL_RX/ & /W2_EXCH2_COMMFLAG/ to W2_EXCH2_TOPOLOGY.h; |
| 162 |
|
rename "exch2_tProc" to "W2_tileProc" and move seting of e2FillValue_RX |
| 163 |
|
from w2_eeboot.F to w2_readparms.F. |
| 164 |
|
o eesupp/src: |
| 165 |
|
- set exchNeedsMemsync & exchUsesBarrier according to "usingMPI" (instead of |
| 166 |
|
ALWAYS_USE_MPI) + moved from ini_communication_patterns.F to eeset_parms.F; |
| 167 |
|
also move setting of exchCollectStatistics from exch_init.F to eeset_parms.F |
| 168 |
|
- in eedie.F and all_proc_die.F, always call MPI_FINALIZE when ALLOW_USE_MPI |
| 169 |
|
is defined (remove condition on usingMPI) to match cases where MPI_INIT |
| 170 |
|
has been called. |
| 171 |
|
- fix bug I put in previous modif of exch_sm_3d_rx.template (not much used); |
| 172 |
|
- change default (non-MPI) value of myPid (from 1 to 0) to match MPI settings |
| 173 |
|
(so that on 1 proc, get same myPid independently of usingMPI) |
| 174 |
|
- add a stop if multi-procs but usingMPI=F ; stop also if usingMPI=T and |
| 175 |
|
useCubedSphereExchange with EXCH-1 (safer). |
| 176 |
|
- add "if usingMPI" where it was missing (+ start to remove ALWAYS_USE_MPI). |
| 177 |
|
o pkg/ggl90: |
| 178 |
|
- fix bug in filling corner mask array "mskCor" (FILL_CS_CORNER_TR_RL call) |
| 179 |
|
used in ALLOW_GGL90_SMOOTH code. |
| 180 |
|
o pkg/flt: |
| 181 |
|
- fix typo in FLT_WITHOUT_Y_PERIODICITY (the effect was to ignore stopping |
| 182 |
|
floats at the Southern end of the domain). |
| 183 |
|
o eesupp/inc: |
| 184 |
|
- add a list of CPP-options which can be set by Makefile (DEF_IN_MAKEFILE.h); |
| 185 |
|
but not used (and not to be used) anywhere, just for documentation. |
| 186 |
|
o pkg/autodiff, exf, ptracers & seaice (AD & TL output): |
| 187 |
|
- comment out the very nasty "writeBinaryPrec = writeStatePrec" line, since: |
| 188 |
|
a) can damage previously written records in any other multi-record files. |
| 189 |
|
b) if needed, can easily do it cleanly, by calling S/R WRITE_REC_3D_RL/S. |
| 190 |
|
o tools/genmake2: |
| 191 |
|
- define variable MAKE within Makefile and use it in place of 'make'; |
| 192 |
|
- update AD_CONFIG.h (if needed) before building any of the executable |
| 193 |
|
(forward, adjoint or tangent-lin); this involves a 2nd invocation of make; |
| 194 |
|
and add the usual #ifndef/#define AD_CONFIG_H in AD_CONFIG.h; |
| 195 |
|
- always print usual warning comment 'this file is automatically generated' |
| 196 |
|
from script convert_cpp_cmd2defines (+ remove this warning from genmake2). |
| 197 |
|
o pkg/cheapaml: |
| 198 |
|
- updated to correspond to the code used in accepted paper (Nico Wienders). |
| 199 |
|
o pkg/exf: |
| 200 |
|
added run time switch useAtmWind to replace ALLOW_ATM_WIND cpp switch. |
| 201 |
|
ALLOW_ATM_WIND now just sets the useAtmWind default (see exf_readparms.F) |
| 202 |
|
and force defines ALLOW_BULKFORMULAE (EXF_OPTIONS.h). |
| 203 |
|
o pkg/exf, autodiff, ctrl, ecco and seaice: |
| 204 |
|
remove ALLOW_ATM_WIND brackets, or replace them with useAtmWind ones. |
| 205 |
|
o pkg/ctrl, ecco: |
| 206 |
|
allow to compile both ALLOW_U/VSTRESS_CONTROL and ALLOW_U/VWIND_CONTROL. |
| 207 |
|
Depending on useAtmWind, one is inactive, and the other is active (see |
| 208 |
|
exf_getffields.F/exf_getsurfacefluxes.F). |
| 209 |
|
o tools/genmake2: |
| 210 |
|
- improve error msg when failing in 2nd test of check_for_broken_Ff |
| 211 |
|
- fix option '-makefile' short-name (-mf according to genmake2 -help) |
| 212 |
|
- source local config file "genmake_local" just after processing the genmake2 |
| 213 |
|
command arguments (previously done before processing arg). |
| 214 |
|
o pkg/grdchk: |
| 215 |
|
- add a call to TURNOFF_MODEL_IO for Tangent-Linear run (in this case, call |
| 216 |
|
to this routine from cost_final has been dropped in g_cost_final) |
| 217 |
|
- improve printed information (more explicit msg, no longer using "print *," |
| 218 |
|
fix fcpertminus printed value). |
| 219 |
|
o model/src: |
| 220 |
|
- move call to COST_INIT_VARIA out of packages_init_variables.F directly |
| 221 |
|
into initialise_varia.F (just after calling PACKAGES_INIT_VARIABLES): |
| 222 |
|
this prevents getting cost-funct reset in adinitialise_varia recomputations. |
| 223 |
|
o pkg/dic: |
| 224 |
|
- move initialisation of "freefe" array from from dic_biotic_forcing.F to |
| 225 |
|
fe_chem.F: this prevents TAF from dropping this initialisation in TLM code. |
| 226 |
|
o pkg/grdchk: |
| 227 |
|
- grdchk_main.F : added CALL GLOBAL_SUM_INT( ierr , myThid ) |
| 228 |
|
- grdchk_getadxx.F etc. : omit I/O if ierr.EQ.0 |
| 229 |
|
o pkg/autodiff/ctrl/grdchk : |
| 230 |
|
- attempt to sort out what works and what does not, |
| 231 |
|
with regard to I/O, and to warn/stop accordingly. |
| 232 |
|
o pkg/grdchk: |
| 233 |
|
- attempt to fix grdchk in the case of useSingleCpuIO. |
| 234 |
|
- "if ( myProcId .EQ. grdchkwhichproc .AND. ierr .EQ. 0 )" |
| 235 |
|
statements were moved from grdchk_main.F and into |
| 236 |
|
grdchk_getadxx.F grdchk_getxx.F grdchk_setxx.F. |
| 237 |
|
- a global sum is applied to record numbers icvrec/layer |
| 238 |
|
out of grdchk_loc so that they be passed to I/O routines. |
| 239 |
|
o pkg/smooth: |
| 240 |
|
- fix recomputation that appeared with nlfs : use smooth_recip_hFacC, |
| 241 |
|
smooth_hFacW, smooth_hFacS (copied at init_fixed) in 3d case. |
| 242 |
|
o verification/testreport: |
| 243 |
|
- add option '-tlm' for Tangent-Linear test (-adm,-oad,-tlm are all exclusive); |
| 244 |
|
- can also use compressed (gzip) reference output file; |
| 245 |
|
- change format (in pkg/grdchk) of TLM output (+ add forward gradient) |
| 246 |
|
to be easier to collect by testreport. |
| 247 |
|
- update existing TLM output (half out-of-date by more than 3 yrs). |
| 248 |
|
- report NETCDF & LAPACK settings (from genmake2) in output dir. |
| 249 |
|
- report run-time in each individual summary (previously only for AD test); |
| 250 |
|
for TLM test, add also TAF version. |
| 251 |
|
- add daily TLM test on (old) aces cluster + weekly test on baudelaire. |
| 252 |
|
|
| 253 |
|
checkpoint63r (2012/08/15) |
| 254 |
|
o tools/genmake2: |
| 255 |
|
- fix Makefile for target ftlall (+ update files to clean after make ftlall) |
| 256 |
|
o pkg/ecco/ecco_cost_final.F : |
| 257 |
|
- fix fc print statements (to stdout and costfunction0000) so |
| 258 |
|
that the printed value is always after the global sum. |
| 259 |
|
o pkg/seaice/seaice_dynsolver.F : |
| 260 |
|
- fix ice strength assymptote at SEAICE_area_max<1 |
| 261 |
|
(no change for the standard case of SEAICE_area_max=1) |
| 262 |
|
o some cleaning: |
| 263 |
|
- pkg source files: ensure that all *.F files from each pkg included the |
| 264 |
|
corresponding ${PKG}_OPTIONS.h file (except stand-alone src file which |
| 265 |
|
only includes CPP_EEOPTIONS.h); fix S/R CTRL_SMOOTH ; update AD experiment |
| 266 |
|
code_ad/*.F src files; add new option-file in pkg/admtlm + corresponding |
| 267 |
|
includes; |
| 268 |
|
- model/src: add PACKAGES_CONFIG.h wherever ALLOW_AUTODIFF[_TAMC] is used. |
| 269 |
|
- ALLOW_DIFFKR_CONTROL: add a stop in S/R CTRL_CHECK in case ALLOW_3D_DIFFKR |
| 270 |
|
is undef; also define /DYNVARS_DIFFKR/ common block in DYNVARS.h when |
| 271 |
|
ALLOW_DIFFKR_CONTROL is defined, to allow to compile and reach this stop. |
| 272 |
|
- update option-file CPP_OPTIONS.h of all AD experiments: |
| 273 |
|
* no longer include PACKAGES_CONFIG.h ; |
| 274 |
|
* includes ECCO_CPPOPTIONS.h if PACKAGES_CONFIG.h has been included ; |
| 275 |
|
* def/undef ALLOW_3D_DIFFKR independently of ALLOW_DIFFKR_CONTROL (before |
| 276 |
|
including ECCO_CPPOPTIONS.h). |
| 277 |
|
o pkg/cost, pkg/ctrl, pkg/ecco option-file: |
| 278 |
|
- clean-up and split content of old pkg/ecco/ECCO_CPPOPTIONS.h file into |
| 279 |
|
separated pkg option-file (done according to old doc from ECCO_CPPOPTIONS.h, |
| 280 |
|
not necessary accurate). |
| 281 |
|
- add a template for option-file ECCO_CPPOPTIONS.h, mostly empty, in |
| 282 |
|
pkg/autodiff, i.e., the 1rst pkg to compile for adjoint built, with lines to |
| 283 |
|
deliberately break compilation (a remainder that it needs to be customized) |
| 284 |
|
- in pkg/ecco, includes ECCO_OPTIONS.h instead of COST_OPTIONS.h |
| 285 |
|
- rename COST_CPPOPTIONS.h to COST_OPTIONS.h |
| 286 |
|
- rename CTRL_CPPOPTIONS.h to CTRL_OPTIONS.h |
| 287 |
|
o tools/genamke2: |
| 288 |
|
- check for lapack. |
| 289 |
|
o pkg/cost and pkg/ecco : attempt to sort out some of the limiting intricacies. |
| 290 |
|
- pkg/cost.cost_final.F : allow for use of pkg/ecco and pkg/cost cost functions |
| 291 |
|
jointly. The exclusive ifdefs were removed, to add costs terms (when defined) |
| 292 |
|
from both packages. |
| 293 |
|
- the fc global sum in pkg/ecco/ecco_cost_final.F was removed accordingly. |
| 294 |
|
- for objf_atl we use either the pkg/ecco or the pkg/cost term depending on CPPs. |
| 295 |
|
- pkg/ecco and model/src : add CPP brackets around pkg/cost and pkg/ctrl includes |
| 296 |
|
of "CTRL_SIZE.h" "ctrl.h" "ctrl_dummy.h" "optim.h" / "cost.h" accordingly. |
| 297 |
|
o pkg/cost : add comments to explain special and/or untested setups. |
| 298 |
|
o pkg/exf: |
| 299 |
|
- use EXF_OPTIONS.h to set pkg/exf options except if done in ECCO_CPPOPTIONS.h |
| 300 |
|
(previously, ALLOW_AUTODIFF_TAMC was controling this switch). |
| 301 |
|
o verification/*/code_ad: |
| 302 |
|
mimimal cleaning of AD experiments ECCO_CPPOPTIONS.h option-file: |
| 303 |
|
- reccord when this file is included (define ECCO_CPPOPTIONS_H) |
| 304 |
|
- remove the most obvious obsolete features (e.g., #define ALLOW_COST not |
| 305 |
|
allowed since Oct 2003 in any option-file except in PACKAGES_CONFIG.h) |
| 306 |
|
o pkg/seaice : |
| 307 |
|
- added possibility to change ice strength/PRESS0 relationship to |
| 308 |
|
HEFF, according to the following run time parameters |
| 309 |
|
SEAICEpresPow0 :: HEFF exponent for ice strength below SEAICEpresH0 (integer, >0) |
| 310 |
|
SEAICEpresPow1 :: HEFF exponent for ice strength above SEAICEpresH0 (integer, >0) |
| 311 |
|
SEAICEpresH0 :: HEFF threshold for ice strength (m) |
| 312 |
|
o pkg/ggl90 : |
| 313 |
|
- added run time flag mxlSurfFlag to include the code that ensure |
| 314 |
|
mixing between first and second level (previously included as a comment) |
| 315 |
|
o verification/1D_ocean_ice_column/input_ad/data.autodiff : |
| 316 |
|
- add test of newly added dumpAdByRec=T option where |
| 317 |
|
we write all records to one file per variable. |
| 318 |
|
o verification/global_ocean.cs32x15/code_ad/CPP_OPTIONS.h : |
| 319 |
|
- switch to #define ALLOW_BALANCE_RELAX and ALLOW_BALANCE_FLUXES |
| 320 |
|
to check that this code is ok with taf and compiles in adjoint. |
| 321 |
|
o pkg/autodiff : |
| 322 |
|
- added run time param dumpAdByRec : ad dump format switch. |
| 323 |
|
If true then write all records to one file per variable. |
| 324 |
|
Else write one file per record (remains the default). |
| 325 |
|
- revise seaice_ad_dump.F and exf_adjoint_snapshots_ad.F accordingly. |
| 326 |
|
o pkg/seaice: |
| 327 |
|
- Added SEAICEuseTILT run time switch : if set to |
| 328 |
|
false then omit the tilt term in seaice dynamics. |
| 329 |
|
- Fix free drift calculation. |
| 330 |
|
o pkg/shelfice: |
| 331 |
|
When initializing SHELFICEDragQuadratic, set default to be |
| 332 |
|
consistent with shiCdrag in case SHELFICEuseGammaFrict is used, |
| 333 |
|
but retain option to set it independently |
| 334 |
|
(inconsistenty spotted by Veronique Dansereau) |
| 335 |
|
o eesupp/inc: |
| 336 |
|
- in EEPARAMS.h, add real-type constant for some frequently used simple number |
| 337 |
|
(0,1,2,1/2), with name explicitly referring to RS/RL type. |
| 338 |
|
o model/src: |
| 339 |
|
- move balancing of relaxation terms to new S/R balance_relax.F. |
| 340 |
|
o pkg/autodiff: |
| 341 |
|
re-write active I/O high-level (control) S/R: |
| 342 |
|
- rename S/R ACTIVE_READ/WRITE_RL to ACTIVE_READ/WRITE_3D_RL |
| 343 |
|
- add 1 more argument (useCurrentDir) to replace the "_LOC" 2nd set of S/R. |
| 344 |
|
- use standard pkg/mdsio I/O routines (MDS_READ/WRITE_SEC_XZ/YZ and |
| 345 |
|
MDS_READ/WRITE_FIELD); change loops to multi-threads. |
| 346 |
|
- update and merge active_file_control_slice.flow into active_file_control.flow |
| 347 |
|
- update the calls in lower level S/R (active_file.F, active_file_g.F and |
| 348 |
|
active_file_ad.F) keeping the argument list and function unchanged. |
| 349 |
|
- also update lower-level "_LOC" version (call same S/R with "useCurrentDir |
| 350 |
|
=.TRUE.") and remove high-level (control) _loc version (not called anymore). |
| 351 |
|
- new low-level active I/O routines, one for each type (RL/RS), for active var |
| 352 |
|
of various (XY,XZ,YZ) shape and level number (arguments: arrShape, myNr); |
| 353 |
|
fix ctrl_init_wet.F & ecco_cost_weights.F by calling the new S/R with the |
| 354 |
|
right type. |
| 355 |
|
o autodiff and RS=real*4: |
| 356 |
|
- call the right type (_RS instead of _RL) of S/R in few output S/R (pkg/exf |
| 357 |
|
ad-monitor and snapshots); disable arrType='RS' in mdsio_gl*.F ; |
| 358 |
|
- change argument type of 3-D mask argument to _RS to match type in the call |
| 359 |
|
(in seaice_cost*.F, ctrl_*_gen.F and *ctrl_bound.F); to be continued. |
| 360 |
|
o pkg/thsice: |
| 361 |
|
- re-anable balanceEmPmR when useSEAICE=T with useThSIce=T. |
| 362 |
|
- option to correct ocean fresh-water flux for imbalance of Atmos+Land |
| 363 |
|
fresh-water flux: uniform correction or proportional to local precip. |
| 364 |
|
o pkg/ctrl: |
| 365 |
|
- new header file CTRL_SIZE.h for parameters maxCtrlArr2D, maxCtrlArr3D |
| 366 |
|
- redefine CPP option ALLOW_BOTTOMDRAG_CONTROL to |
| 367 |
|
ALLOW_BOTTOMDRAG_CONTROL_NONGENERIC |
| 368 |
|
- verification/hs94.1x64x5/ tests ALLOW_GENARR{2D,3D}_CONTROL |
| 369 |
|
|
| 370 |
|
checkpoint63q (2012/07/30) |
| 371 |
|
o pkg/autodiff: |
| 372 |
|
note about monitor and snap-shot of ad-variables: |
| 373 |
|
a) need to apply ADEXCH before monitor or output of ad-vars to include |
| 374 |
|
overlap contributions (was done for adjDump but missing for monitor). |
| 375 |
|
b) since ADEXCH implies adding overlap values to interior array, machine |
| 376 |
|
truncation error is expected when adding an ADEXCH call in the middle of |
| 377 |
|
the time-stepping (although mathematically identical). |
| 378 |
|
modifs: |
| 379 |
|
- fix some incorrect type (RL/RS, UV/scalar) in ADEXCH calls and ad-vars |
| 380 |
|
declaration (adEddyPsiX,adEddyPsiY in adcommon.h). |
| 381 |
|
- add new run-time params (dumpAdVarExch & mon_AdVarExch) to select how to |
| 382 |
|
apply ADEXCH before monitor or snap-shot of ad-variables: |
| 383 |
|
=0 : no ADEXCH ; =1 : do ADEXCH ; =2 : apply ADEXCH on a local copy. |
| 384 |
|
- implement new method (copy+adexch) for snap-shot and monitor of main model |
| 385 |
|
ad-vars (initially not used by default -> output unchanged). |
| 386 |
|
- change default (set to 2) to use new method and update few output_adm.txt |
| 387 |
|
with accurate ad-monitor (global_ocean.90x40x15, tutorial_global_oce_biogeo |
| 388 |
|
and both tutorial_tracer_adjsens tests). |
| 389 |
|
- switch 2 global_ocean.cs32x15 AD-tests to AB-2 (seaice_dynmix & thsice) |
| 390 |
|
and turn off momDissip_In_AB in the 2 other (needed for stable AB-3). |
| 391 |
|
o pkg/ctrl: |
| 392 |
|
enable the use of a variable number (maxCtrlArr2D, maxCtrlArr3D) of |
| 393 |
|
generic control variables via |
| 394 |
|
#define ALLOW_GENARR2D_CONTROL and/or #define ALLOW_GENARR3D_CONTROL |
| 395 |
|
Only required customization should be S/R ctrl_map_ini_genarr.F |
| 396 |
|
(routine in repository contains example) |
| 397 |
|
o pkg/exf: |
| 398 |
|
- fix and refine checking of surface-relaxation setting when using EXF |
| 399 |
|
o verification (adjoint exp): |
| 400 |
|
- reduce length of simulation by half for experiment global_ocean.cs32x15 |
| 401 |
|
(std, seaice & seaice_dynmix) and exp tutorial_dic_adjoffline; update output. |
| 402 |
|
o pkg/mom_common (mom_u/v_bottomdrag), pkg/shelfice (shelfice_u/v_drag) |
| 403 |
|
- replace if statements KE.ne.0 with KE.gt.0 to avoid sqrt of negative numbers |
| 404 |
|
of KE that can occur with aggressive optimization |
| 405 |
|
+ put quadratic bottom drag calculation in separated loops (only done if |
| 406 |
|
using bottomDragQuadratic) |
| 407 |
|
o model/src: CG2D_NSA |
| 408 |
|
- clean-up: stop if cg2dNormaliseRHS=T and AUTODIFF |
| 409 |
|
and replace GLOBAL_SUM calls with GLOBAL_SUM_TILE. |
| 410 |
|
- in test-exp that uses CG2D_NSA (bottom_ctrl & tutorial_tracer_adjsens), |
| 411 |
|
change cg2d residual target to be expressed in Wunits (=> do not normalise |
| 412 |
|
RHS ; this fix global_max Pb with TAF); update output. |
| 413 |
|
o verification/tutorial_tracer_adjsens: |
| 414 |
|
improve parameter consistency (and get rid of warnings): |
| 415 |
|
- use deltaTFreeSurf=deltaTtracer (better with NonLin FreeSurf); |
| 416 |
|
- switch Temp Adv Scheme from 81 to 80 (no limiter since theta < 0 is physical) |
| 417 |
|
- use staggerTimeStep (for stability) in som81; use default rotation period. |
| 418 |
|
o pkg/exch2: |
| 419 |
|
- with empty facet: fix compact-format (W2_mapIO=1) definition (was previously |
| 420 |
|
reset to 0 = 1 long line in X) |
| 421 |
|
o pkg/mdsio: |
| 422 |
|
- fix reading of dimensions from meta file for large-size domain |
| 423 |
|
o model/src: |
| 424 |
|
- fix saltIsActiveTr setting for IdealGas eosType (eosType is 6 character long); |
| 425 |
|
- to refine CS-grid check, add logical flag "hasWetCSCorners": true if using |
| 426 |
|
Cubed-Sphere Exch with CS-corners inside the domain; |
| 427 |
|
make new S/R INI_GLOBAL_DOMAIN from code in ini_linear_phisurf.F to calculate |
| 428 |
|
globalArea and to set hasWetCSCorners. |
| 429 |
|
- add warning if usingCurvilinearGrid and momAdvection in flux-form (missing |
| 430 |
|
metric terms); stop if hasWetCSCorners=T and momAdvection in flux-form. |
| 431 |
|
- re-activate the stop if useCDscheme and hasWetCSCorners=T (previously if |
| 432 |
|
useCDscheme and useCubedSphereExchange). |
| 433 |
|
o pkg/seaice/seaice_growth.F : |
| 434 |
|
- compute the SEAICE_DISABLE_HEATCONSFIX term accordingly |
| 435 |
|
for the (.NOT.useRealFreshWaterFlux).OR.(nonlinFreeSurf.EQ.0) |
| 436 |
|
when the previous code was not correct. |
| 437 |
|
- in all cases diagnoze the boundary condition mismatch to SIaaflux |
| 438 |
|
- in the case of useRealFreshWaterFlux.AND.(nonlinFreeSurf.GT.0) add |
| 439 |
|
that term to QNET (essentially to set the melt/freeze water to 0degC). |
| 440 |
|
when SEAICEheatConsFix is true, so that the ocean+ice budget heat is closed. |
| 441 |
|
o verification/global_ocean.cs32x15/input.seaice : |
| 442 |
|
- add SEAICEheatConsFix in data.seaice, ensuring closed ocean+ice heat budget. |
| 443 |
|
- add diagnostics (commented out) for budget test using my matlab codes (gcmfaces). |
| 444 |
|
- update results (output.seaice.txt) accordingly. |
| 445 |
|
o verification/lab_sea/input/data.seaice : |
| 446 |
|
- remove SEAICEheatConsFix, which is now be tested |
| 447 |
|
in global_ocean.cs32x15. The presently coded fix |
| 448 |
|
is only consistent with real fresh water + nlfs. |
| 449 |
|
- update results (output.txt) accordingly. |
| 450 |
|
o pkg/salt_plume: added diagnotic for salt tendency (oceSPtnd). |
| 451 |
|
o AD_CONFIG.h file: |
| 452 |
|
- in genmake2: remove "#undef ALLOW_ECCO_OPTIMIZATION" from AD_CONFIG.h; |
| 453 |
|
- remove #include "AD_CONFIG.h" from all included OPTIONS files and |
| 454 |
|
add it explicitly in each fortran src file where it is needed. |
| 455 |
|
motivation: header file AD_CONFIG.h is generated/modified after "make depend" |
| 456 |
|
thus introducing potentially wrong dependency; therefore, better to include |
| 457 |
|
it only where it's really needed, and after any other included OPTIONS file |
| 458 |
|
(less tempting to bring conditional includes according to ALLOW_ADJOINT_RUN |
| 459 |
|
or ALLOW_TANGENTLINEAR_RUN). |
| 460 |
|
|
| 461 |
|
checkpoint63p (2012/07/06) |
| 462 |
|
o pkg/autodiff: |
| 463 |
|
- rename S/R INI_AUTODIFF to AUTODIFF_INIT_VARIA ; |
| 464 |
|
- create new header file "AUTODIFF_PARAMS.h" for pkg/autodiff parameters |
| 465 |
|
which are now read from new parameter file "data.autodiff" (inAdExact |
| 466 |
|
previously in "data", use{KPP/GMRedi/SEAICE}inAdMode previously in data.pkg) |
| 467 |
|
- add file "data.autodiff" in all verification/*/input_ad/ directories. |
| 468 |
|
- add run-time parameter to turn On/Off SEAICEuseFREEDRIFT in adjoint/forward |
| 469 |
|
mode (used in global_ocean.cs32x15.seaice_dynmix test exp.) |
| 470 |
|
o verification/tutorial_tracer_adjsens/ |
| 471 |
|
- add extra sub-verification which uses "SOM" advection scheme |
| 472 |
|
(2nd Order-Moment Advection Scheme, Prather Limiter) with |
| 473 |
|
#define GAD_ALLOW_TS_SOM_ADV in GAD_OPTIONS.h |
| 474 |
|
and in data: |
| 475 |
|
multiDimAdvection=.TRUE., |
| 476 |
|
tempAdvScheme=81, |
| 477 |
|
saltAdvScheme=81, |
| 478 |
|
o model/src: |
| 479 |
|
- introduce ALLOW_BALANCE_RELAX which allow the removal |
| 480 |
|
of the global mean of relaxation terms by setting |
| 481 |
|
balanceThetaClimRelax and balanceSaltClimRelax |
| 482 |
|
- disable balanceEmPmR and balanceQnet in the case when useSeaice. |
| 483 |
|
This case is now treated appropriately in seaice_growth.F |
| 484 |
|
o pkg/seaice: |
| 485 |
|
- do balanceEmPmR and balanceQnet based on the net fluxes |
| 486 |
|
(SItflux) that enter the ocean+ice system rather than QNET |
| 487 |
|
and EMPMR. Works best in real fresh water. |
| 488 |
|
- SIatmFW, SIatmQnet, and SItflux computations had to be moved |
| 489 |
|
out of the diagnotics bloc, and stored in global arrays. |
| 490 |
|
|
| 491 |
|
checkpoint63o (2012/06/29) |
| 492 |
|
o vertical mixing schemes: |
| 493 |
|
- remove unused S/R CALC_BUOYANCY (left from branch-atmos merging) |
| 494 |
|
- replace computation of density (in Function STATE1, opps_calc.F) by a call |
| 495 |
|
to S/R FIND_RHO_SCALAR |
| 496 |
|
- pass sigmaR to S/R PP81_CALC, MY82_CALC (but not yet used) and to GGL90_CALC |
| 497 |
|
and use it for N^2 calculation in ggl90_calc.F (save 2 density computations); |
| 498 |
|
affects output (vermix.ggl90) at truncation level. |
| 499 |
|
o testreport: |
| 500 |
|
- change adjustment of SIZE.h for mpi+mth testing: preserve enough tiles |
| 501 |
|
(nSx,nSy) for the number of threads (from eedata.mth) |
| 502 |
|
o model/src: |
| 503 |
|
- remove 2nd set of calls at the top of initialise_varia.F (Adjoint part) |
| 504 |
|
related to Non_Lin Free-Surf or Bottom-Control (but not longer necessary). |
| 505 |
o verification/tutorial_tracer_adjsens/ |
o verification/tutorial_tracer_adjsens/ |
| 506 |
- compile with ALLOW_CG2D_NSA |
- compile with ALLOW_CG2D_NSA |
| 507 |
- run with nonlinFreeSurf=3 |
- run with nonlinFreeSurf=3 |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch