3 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
4 |
============================== |
============================== |
5 |
|
|
6 |
|
Revision history for branch c24_e25_ice |
7 |
|
(branch for adding sea-ice code to ecco-branch) |
8 |
|
|
9 |
|
02/12/03 cvs tag ecco_ice2 |
10 |
|
Added pkg/seaice and modified pkg/exf as per release1_p12_pre |
11 |
|
|
12 |
|
02/11/03 cvs tag ecco_ice1 |
13 |
|
Updated verification/natl_box_forward and natl_box_adjoint |
14 |
|
for SGI platform |
15 |
|
|
16 |
|
12/06/02 |
17 |
|
Branched off tag ecco_c44_e25 |
18 |
|
|
19 |
|
================================================================= |
20 |
|
|
21 |
|
ecco_c44_e25 |
22 |
|
o Merging some GM changes from main trunk (c47) to ecco-branch. |
23 |
|
o Updating verif. natl_box_adjoint with GMRedi. |
24 |
|
o Adding setup with explicit differentiaton of pressure solver. |
25 |
|
o Bug fix: missing OBCS_OPTIONS.h for sponge layer code. |
26 |
|
o (Re-?) Commiting old bug fixes in kpp to ecco-branch |
27 |
|
(kpp_calc.F, kpp_routines.F) |
28 |
|
o exf: |
29 |
|
updated external forcing package |
30 |
|
- cleaned old exfa stuff |
31 |
|
- bug fixes |
32 |
|
(missing OBCS_OPTIONS.h in two routines) |
33 |
|
- enable easy to use "no forcing". |
34 |
|
- added exf I/O for atmospheric loading |
35 |
|
- added exf I/O for runoff data |
36 |
|
- transfered scaling between exf <-> MITgcm to exf namelist |
37 |
|
|
38 |
|
ecco_c44_e24 |
39 |
|
o modifications to gradient check package (Martin Losch) |
40 |
|
- enable centered differences |
41 |
|
- modified format of standard output |
42 |
|
o bug fix for nwettile contents |
43 |
|
so far assumed global sum in first entry (1,1,k), |
44 |
|
needed for global control vector |
45 |
|
-> introduced new variable nwet[c/w/s]global(nr) |
46 |
|
o shifts of calls to cal/exf package to be accessible |
47 |
|
in MITgcm stand-alone |
48 |
|
o shifted packages_init_variables before convective_adjustment_ini |
49 |
|
(corresponding update of TAF tape initialisation). |
50 |
|
o removed auxiliary field in impldiff |
51 |
|
(caused numerical differences for GMRedi cases) |
52 |
|
o updated obcs control part (G. Gebbie) |
53 |
|
o removed some f90 comments ("!") |
54 |
|
o Bug fixes and updates in cost package |
55 |
|
* cost_final: correct multiplier mult_tauv, mult_sflux |
56 |
|
* cost_hyd: added call cost_ssh |
57 |
|
* cost_init_barfiles.F: replaced arrays by tmpfld's |
58 |
|
* cost_readtopexmean.F: ersmask -> tpmeanmask |
59 |
|
* cost_weights.F: corrected wtheta[2], wsalt[2] |
60 |
|
o cleaned exchanges in cg2d.F |
61 |
|
|
62 |
|
ecco_c44_e23 |
63 |
|
merged patches between release1_p1 and release1_p2: |
64 |
|
o fixed array boundary problem for adjoint |
65 |
|
(related to partial recomputation within subroutine |
66 |
|
related to dynamics -> calc_viscosity -> mom_... |
67 |
|
o fixed recomputation in thermodynamics |
68 |
|
o prepare (but not complete) recomputation problem in impldiff |
69 |
|
o fixed problem for adjoint of global_max |
70 |
|
o switched off sponge layer code in external_forcing |
71 |
|
Bug fix in ctrl_mask_set_yz (spotted by G. Gebbie) |
72 |
|
|
73 |
|
ecco_c44_e22 |
74 |
|
o Finally, the merges with the Scripps cost function terms: |
75 |
|
Various additions, modifications, quality checks. |
76 |
|
o Plus: Included switch ALLOW_AUTODIFF_MONITOR to handle |
77 |
|
monitoring of adjoint state. |
78 |
|
o Plus: removed comm_stats, ini_communication_patterns from diff. list |
79 |
|
(no more need to "keep" MPI headers for adjoint generation) |
80 |
|
o Plus: exf changes: |
81 |
|
o added run-off routine |
82 |
|
o additional storing needed for ustress,vstess |
83 |
|
because of state-dependent quality check |
84 |
|
o added missing initialisations |
85 |
|
o Plus: genmake with MAKEDEPEND variable was still missing in ecco-branch |
86 |
|
NB: - verification/natl_box_adjoint/ 240 steps run |
87 |
|
* looks ok under Linux (twain.lcs.mit.edu) |
88 |
|
* looks ok on MIT SUN's (fjord.mit.edu) |
89 |
|
* NaN in reverse mode on Scripps SUN's at iter=214 |
90 |
|
* NaN in reverse mode on IBM SP3 (BlueHorizon) at iter=204 |
91 |
|
- 1 year forward int. global 2x2 deg. produces similar cost functions |
92 |
|
* all cost terms of new code are slightly smaller (0.5 to 3%). |
93 |
|
|
94 |
|
ecco_c44_e21 |
95 |
|
o reversed order of ctrl/cost initialisation |
96 |
|
o gmredi_slope_psi.F: Inconsistent types in MIN/MAX expressions |
97 |
|
o modified to enable ctrl field swapping |
98 |
|
o bug fix for global sum of ncvarlength in ctrl_init |
99 |
|
o no more call to ctrl_unpack from ctrl_init |
100 |
|
o no more write of ctrl vector in ctrl_pack |
101 |
|
however, init. of control vector from ctrl_unpack |
102 |
|
o various bug fixes & consistent nobcs handling in |
103 |
|
ctrl_pack/unpack routines (by G. Gebbie) |
104 |
|
o optim package: Added obcs control part for lsopt I/O |
105 |
|
(by G. Gebbie) |
106 |
|
o missing call to obcs_diags |
107 |
|
o moved calls in packages_init_fixed to packages_readparms |
108 |
|
because some quantities are needed earlier. |
109 |
|
|
110 |
|
ecco_c44_e20 |
111 |
|
o before major merges from the Scripps setup |
112 |
|
here are two small updates: |
113 |
|
- field swapping for surface flux controls enabled |
114 |
|
(remains t.b.d. for atmos. state plus obcs) |
115 |
|
- bug fix in active_file.F for xz-/yz- routines |
116 |
|
- added missing xz-/yz- routines to active_file_ad.F |
117 |
|
|
118 |
|
ecco_c44_e19 |
119 |
|
o merged necessary patches: |
120 |
|
- AD-related changes for GMRedi |
121 |
|
- fixes i KPP (delZ -> drF) |
122 |
|
- hook to OBCS songe layer code in external_forcing |
123 |
|
- errorMessageUnit non-zero in eeboot.F |
124 |
|
- included GMREDI, EXACT_CONSERV to natl_box_adjoint verif. |
125 |
|
|
126 |
|
ecco_c44_e18 |
127 |
|
o sponge layer code for obcs was missing |
128 |
|
- added hook in external_forcing.F |
129 |
|
- added storing for TAF |
130 |
|
o new routine cost_init_barfiles to initialise |
131 |
|
adjoint fields adtbar, adsbar, adpsbar |
132 |
|
(needed in conjunction with corrected ladinit for active I/O) |
133 |
|
o cleaned up and corrected order in packages_readparms, ecco_readparms |
134 |
|
o corrected and simplified weights for obcs controls |
135 |
|
- weight fields simplified |
136 |
|
- bug fix in ctrl_pack |
137 |
|
- no more write of control vector in ctrl_pack |
138 |
|
o i/j index bug fixes in cost_ssh_mean |
139 |
|
|
140 |
|
ecco_c44_e17 |
141 |
|
o bug fixes in cost package |
142 |
|
* wrong identifier for active I/O of monthly mean psbar |
143 |
|
(cost_averagesfields, cost_ssh) xx_psbar_dummy |
144 |
|
* cleaned all identifier for active I/O of monthly fields |
145 |
|
* wrong ladinit in various cost routines |
146 |
|
o merge with interval (chkpt44c_post - chkpt44d_post) |
147 |
|
* added missing EXCLUDE_MONITOR flags |
148 |
|
* changed "e" to "_d" in gmredi_slope_limit, gmredi_slope_psi |
149 |
|
(incompatible typ in MIN/MAX expressions caused problems |
150 |
|
on IBM SP3) |
151 |
|
* in genmake added variable MAKEDEPEND |
152 |
|
plus resetting for case SunOS |
153 |
|
added timer_stats.c routine for IBM SP3 |
154 |
|
* removed variables in dynamics |
155 |
|
|
156 |
|
ecco_c44_e16 |
157 |
|
o new ECCO checkpoint, the first since ecco_c43_e15 |
158 |
|
which is branched from MITgcm. |
159 |
|
All changes were merged in between (see ecco-branch-mod? tags). |
160 |
|
o last changes: |
161 |
|
* assuming path for TAF and TAMC are known (e.g. in $path) |
162 |
|
* genmake: |
163 |
|
- added tools/xmakedepend to circumvene shortcoming of |
164 |
|
SUN's makedepend (size too small) |
165 |
|
- invoking ../tools/xmakedepend instead of makedepend |
166 |
|
- changed compiler flags for case SunOS |
167 |
|
|
168 |
|
ecco-branch-mod5 |
169 |
|
o mod's to reproduce open boundary run |
170 |
|
o enable field swapping for OB's and climatology |
171 |
|
o update store directives for OB control |
172 |
|
o added option ALLOW_OBCS_PRESCRIBE |
173 |
|
o call of getobcs now out of obcs_calc (param. list modified) |
174 |
|
|
175 |
|
ecco-branch-mod4 |
176 |
|
o this is the second part of changes necessary to reconcile |
177 |
|
the natl_box_adjoint verification (with and without grdchk). |
178 |
|
>>>!!! CHANGES BETWEEN THE TAGS ecco-branch-mod3 !!!<<< |
179 |
|
>>>!!! AND ecco-branch-mod4 !!!<<< |
180 |
|
>>>!!! SHOULD BE MERGED INTO THE MAIN TRUNK !!!<<< |
181 |
|
- genmake: removed $S64 overwrite for case SunOS |
182 |
|
- pkg/exf: update and corrections for field swapping and obcs |
183 |
|
- pkg/ecco: parameter lists for the_model_main, the_main_loop |
184 |
|
harmonized between ECCO and MITgcm |
185 |
|
- pkg/autodiff: added flow directives for obcs, mdsio_gl_slice |
186 |
|
updated checkpointing_lev... lists for obcs |
187 |
|
- model/src: minor changes in forward_step, plot_field |
188 |
|
- pkg/mdsio: added mdsio_gl_slice |
189 |
|
|
190 |
|
ecco-branch-mod3 |
191 |
|
o this is the first part of changes necessary to reconcile |
192 |
|
natl_box_adjoint verification (with and without grdchk). |
193 |
|
This set of changes is relevant for ECCO only. |
194 |
|
The next set/tag of changes (ecco-branch-mod4) |
195 |
|
should be merged into the main trunk. |
196 |
|
- update some options |
197 |
|
- parameter lists of ecco_the_model_main and ecco_the_main_loop |
198 |
|
differed from those of the MITgcm counterparts. |
199 |
|
|
200 |
|
ecco-branch-mod2 |
201 |
|
o update ecco-branch-mod1 to checkpoint44 |
202 |
|
o exclude from updates: |
203 |
|
- verification/ |
204 |
|
- adjoint/ |
205 |
|
- pkg/cost |
206 |
|
- pkg/ctrl |
207 |
|
- pkg/ecco |
208 |
|
- pkg/obcs |
209 |
|
- model/src/the_model_main.F |
210 |
|
- model/src/the_main_loop.F |
211 |
|
|
212 |
|
checkpoint44 |
213 |
|
o apply Shapiro filter on T&S before calc_PhiHyd (stagger time stepping) |
214 |
|
o fix bug in mom_vi_del2uv |
215 |
|
o added some stuff for AD |
216 |
|
|
217 |
|
ecco-branch-mod1 |
218 |
|
o based on c43 |
219 |
|
o replaced packages cost, ctrl, ecco, obcs |
220 |
|
o removed MITgcm verif.'s, added ECCO verif.'s |
221 |
|
o added lsopt, optim for line search optimization. |
222 |
|
|
223 |
|
checkpoint43 |
224 |
|
o added two new advection tests |
225 |
|
o added global 4x4 degree ocean test |
226 |
|
o added test of H&S94 with free surface on cube |
227 |
|
o some final mods for gradient check, adjoint,... |
228 |
|
|
229 |
|
checkpoint42 |
230 |
|
Fixed AD-related problems: |
231 |
|
o Store directives up-to-date with re-arranged Adams-Bashforth |
232 |
|
(mainly thermodynamics.F) |
233 |
|
o New store directives for multi-dim. advection schemes |
234 |
|
* new CPP flag ALLOW_MULTI_DIM_ADVECTION |
235 |
|
* new common block and key passkey |
236 |
|
(mainly gad_advection.F) |
237 |
|
o Modified store directives for split of dynamics/thermodynamics |
238 |
|
for the case ALLOW_KPP |
239 |
|
o Cleaned argument list for timestep_tracer.F |
240 |
|
|
241 |
|
checkpoint41 |
242 |
|
o re-formatted/added comments in prologues of key subroutines |
243 |
|
to work with protex |
244 |
|
|
245 |
|
checkpoint40 |
246 |
|
o moved calc_gs, calc_gt and calc_tr1 back to model/src |
247 |
|
o added FLT (floats) package from Arne along with demo. experiment |
248 |
|
o re-arranged Adams-Bashforth scheme for tracers to allow easier |
249 |
|
implementation of forward-in-time forcing terms |
250 |
|
o more mods for non-linear free-surface including new variable |
251 |
|
o modified multi-dim method to work on cube |
252 |
|
o bug fix in KPP (from DM) |
253 |
|
|
254 |
|
checkpoint40pre9 |
255 |
|
o added DST advection schemes |
256 |
|
o fix MPI bug on SGI's |
257 |
|
o modifications for non-linear free-surface |
258 |
|
o fixed conditionals for fixed flow/passive tracer only mode |
259 |
|
o added tension/strain form of stress tensor |
260 |
|
o changes to commenting of arguments for TAF/TAMC |
261 |
|
o added/updated many matlab scripts |
262 |
|
o added multi-dimensional advection method for tracers |
263 |
|
|
264 |
|
checkpoint40pre8 |
265 |
|
o new package rw/ |
266 |
|
Added method for dumping intermediate local arrays: |
267 |
|
mdsio_writetile - same as mdsio_writefield except works from inside |
268 |
|
bi,bj loop |
269 |
|
mdsio_writelocal - same as mdsio_writetile except works for local |
270 |
|
arrays |
271 |
|
write_local_r? - higher-level wrapper for mdsio_writelocal |
272 |
|
Controlled by diagFreq. Defaults to zero (ie. no dumps) |
273 |
|
Example given at end of mom_vecinv.F that dumps some local arrays. |
274 |
|
o replaced calc_mom_rhs.F by mom_fluxform.F/mom_vecinc.F to |
275 |
|
control flux-form versus vector invariant code. |
276 |
|
Switch is runtime flag vectorInvariantMomentum (defaults to false) |
277 |
|
or genmake -DISABLE ... |
278 |
|
o TAMC-related update to cope with mom_fluxform.F/mom_vecinc.F |
279 |
|
(additional storing is now needed for case vecinv; |
280 |
|
currently nothing implemented to switch this storing off |
281 |
|
in case fluxform is used since no CPP flag available). |
282 |
|
|
283 |
|
checkpoint40pre7 |
284 |
|
o bug fix in dynamics.F |
285 |
|
calc_viscosity called before k=1,Nr loop. |
286 |
|
o added some #ifdef's ALLOW_PASSIVE_TRACER |
287 |
|
o cleaned initialisations in ctrl_map_ini |
288 |
|
related to TAF |
289 |
|
|
290 |
|
checkpoint40pre6 |
291 |
|
o split dynamics into thermodynamics & dynamics |
292 |
|
o corresponding split of calc_diffusivity into |
293 |
|
calc_diffusivity, calc_viscosity |
294 |
|
(kpp_calc_diff, kpp_calc_visc). |
295 |
|
o Added diffkr, kapgm to control variables |
296 |
|
o bug fix in gmredi_slope_limit for ldd97 |
297 |
|
|
298 |
checkpoint40pre5 |
checkpoint40pre5 |
299 |
o proper initialization of Adams-Bashforth time stepping |
o proper initialization of Adams-Bashforth time stepping |
300 |
+ uses forward step for very first step of model (iter=0) |
+ uses forward step for very first step of model (iter=0) |