3 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
4 |
============================== |
============================== |
5 |
|
|
6 |
|
ecco_c44_e23 |
7 |
|
merged patches between release1_p1 and release1_p2: |
8 |
|
o fixed array boundary problem for adjoint |
9 |
|
(related to partial recomputation within subroutine |
10 |
|
related to dynamics -> calc_viscosity -> mom_... |
11 |
|
o fixed recomputation in thermodynamics |
12 |
|
o prepare (but not complete) recomputation problem in impldiff |
13 |
|
o fixed problem for adjoint of global_max |
14 |
|
o switched off sponge layer code in external_forcing |
15 |
|
Bug fix in ctrl_mask_set_yz (spotted by G. Gebbie) |
16 |
|
|
17 |
|
ecco_c44_e22 |
18 |
|
o Finally, the merges with the Scripps cost function terms: |
19 |
|
Various additions, modifications, quality checks. |
20 |
|
o Plus: Included switch ALLOW_AUTODIFF_MONITOR to handle |
21 |
|
monitoring of adjoint state. |
22 |
|
o Plus: removed comm_stats, ini_communication_patterns from diff. list |
23 |
|
(no more need to "keep" MPI headers for adjoint generation) |
24 |
|
o Plus: exf changes: |
25 |
|
o added run-off routine |
26 |
|
o additional storing needed for ustress,vstess |
27 |
|
because of state-dependent quality check |
28 |
|
o added missing initialisations |
29 |
|
o Plus: genmake with MAKEDEPEND variable was still missing in ecco-branch |
30 |
|
NB: - verification/natl_box_adjoint/ 240 steps run |
31 |
|
* looks ok under Linux (twain.lcs.mit.edu) |
32 |
|
* looks ok on MIT SUN's (fjord.mit.edu) |
33 |
|
* NaN in reverse mode on Scripps SUN's at iter=214 |
34 |
|
* NaN in reverse mode on IBM SP3 (BlueHorizon) at iter=204 |
35 |
|
- 1 year forward int. global 2x2 deg. produces similar cost functions |
36 |
|
* all cost terms of new code are slightly smaller (0.5 to 3%). |
37 |
|
|
38 |
|
ecco_c44_e21 |
39 |
|
o reversed order of ctrl/cost initialisation |
40 |
|
o gmredi_slope_psi.F: Inconsistent types in MIN/MAX expressions |
41 |
|
o modified to enable ctrl field swapping |
42 |
|
o bug fix for global sum of ncvarlength in ctrl_init |
43 |
|
o no more call to ctrl_unpack from ctrl_init |
44 |
|
o no more write of ctrl vector in ctrl_pack |
45 |
|
however, init. of control vector from ctrl_unpack |
46 |
|
o various bug fixes & consistent nobcs handling in |
47 |
|
ctrl_pack/unpack routines (by G. Gebbie) |
48 |
|
o optim package: Added obcs control part for lsopt I/O |
49 |
|
(by G. Gebbie) |
50 |
|
o missing call to obcs_diags |
51 |
|
o moved calls in packages_init_fixed to packages_readparms |
52 |
|
because some quantities are needed earlier. |
53 |
|
|
54 |
|
ecco_c44_e20 |
55 |
|
o before major merges from the Scripps setup |
56 |
|
here are two small updates: |
57 |
|
- field swapping for surface flux controls enabled |
58 |
|
(remains t.b.d. for atmos. state plus obcs) |
59 |
|
- bug fix in active_file.F for xz-/yz- routines |
60 |
|
- added missing xz-/yz- routines to active_file_ad.F |
61 |
|
|
62 |
|
ecco_c44_e19 |
63 |
|
o merged necessary patches: |
64 |
|
- AD-related changes for GMRedi |
65 |
|
- fixes i KPP (delZ -> drF) |
66 |
|
- hook to OBCS songe layer code in external_forcing |
67 |
|
- errorMessageUnit non-zero in eeboot.F |
68 |
|
- included GMREDI, EXACT_CONSERV to natl_box_adjoint verif. |
69 |
|
|
70 |
|
ecco_c44_e18 |
71 |
|
o sponge layer code for obcs was missing |
72 |
|
- added hook in external_forcing.F |
73 |
|
- added storing for TAF |
74 |
|
o new routine cost_init_barfiles to initialise |
75 |
|
adjoint fields adtbar, adsbar, adpsbar |
76 |
|
(needed in conjunction with corrected ladinit for active I/O) |
77 |
|
o cleaned up and corrected order in packages_readparms, ecco_readparms |
78 |
|
o corrected and simplified weights for obcs controls |
79 |
|
- weight fields simplified |
80 |
|
- bug fix in ctrl_pack |
81 |
|
- no more write of control vector in ctrl_pack |
82 |
|
o i/j index bug fixes in cost_ssh_mean |
83 |
|
|
84 |
|
ecco_c44_e17 |
85 |
|
o bug fixes in cost package |
86 |
|
* wrong identifier for active I/O of monthly mean psbar |
87 |
|
(cost_averagesfields, cost_ssh) xx_psbar_dummy |
88 |
|
* cleaned all identifier for active I/O of monthly fields |
89 |
|
* wrong ladinit in various cost routines |
90 |
|
o merge with interval (chkpt44c_post - chkpt44d_post) |
91 |
|
* added missing EXCLUDE_MONITOR flags |
92 |
|
* changed "e" to "_d" in gmredi_slope_limit, gmredi_slope_psi |
93 |
|
(incompatible typ in MIN/MAX expressions caused problems |
94 |
|
on IBM SP3) |
95 |
|
* in genmake added variable MAKEDEPEND |
96 |
|
plus resetting for case SunOS |
97 |
|
added timer_stats.c routine for IBM SP3 |
98 |
|
* removed variables in dynamics |
99 |
|
|
100 |
|
ecco_c44_e16 |
101 |
|
o new ECCO checkpoint, the first since ecco_c43_e15 |
102 |
|
which is branched from MITgcm. |
103 |
|
All changes were merged in between (see ecco-branch-mod? tags). |
104 |
|
o last changes: |
105 |
|
* assuming path for TAF and TAMC are known (e.g. in $path) |
106 |
|
* genmake: |
107 |
|
- added tools/xmakedepend to circumvene shortcoming of |
108 |
|
SUN's makedepend (size too small) |
109 |
|
- invoking ../tools/xmakedepend instead of makedepend |
110 |
|
- changed compiler flags for case SunOS |
111 |
|
|
112 |
|
ecco-branch-mod5 |
113 |
|
o mod's to reproduce open boundary run |
114 |
|
o enable field swapping for OB's and climatology |
115 |
|
o update store directives for OB control |
116 |
|
o added option ALLOW_OBCS_PRESCRIBE |
117 |
|
o call of getobcs now out of obcs_calc (param. list modified) |
118 |
|
|
119 |
|
ecco-branch-mod4 |
120 |
|
o this is the second part of changes necessary to reconcile |
121 |
|
the natl_box_adjoint verification (with and without grdchk). |
122 |
|
>>>!!! CHANGES BETWEEN THE TAGS ecco-branch-mod3 !!!<<< |
123 |
|
>>>!!! AND ecco-branch-mod4 !!!<<< |
124 |
|
>>>!!! SHOULD BE MERGED INTO THE MAIN TRUNK !!!<<< |
125 |
|
- genmake: removed $S64 overwrite for case SunOS |
126 |
|
- pkg/exf: update and corrections for field swapping and obcs |
127 |
|
- pkg/ecco: parameter lists for the_model_main, the_main_loop |
128 |
|
harmonized between ECCO and MITgcm |
129 |
|
- pkg/autodiff: added flow directives for obcs, mdsio_gl_slice |
130 |
|
updated checkpointing_lev... lists for obcs |
131 |
|
- model/src: minor changes in forward_step, plot_field |
132 |
|
- pkg/mdsio: added mdsio_gl_slice |
133 |
|
|
134 |
|
ecco-branch-mod3 |
135 |
|
o this is the first part of changes necessary to reconcile |
136 |
|
natl_box_adjoint verification (with and without grdchk). |
137 |
|
This set of changes is relevant for ECCO only. |
138 |
|
The next set/tag of changes (ecco-branch-mod4) |
139 |
|
should be merged into the main trunk. |
140 |
|
- update some options |
141 |
|
- parameter lists of ecco_the_model_main and ecco_the_main_loop |
142 |
|
differed from those of the MITgcm counterparts. |
143 |
|
|
144 |
|
ecco-branch-mod2 |
145 |
|
o update ecco-branch-mod1 to checkpoint44 |
146 |
|
o exclude from updates: |
147 |
|
- verification/ |
148 |
|
- adjoint/ |
149 |
|
- pkg/cost |
150 |
|
- pkg/ctrl |
151 |
|
- pkg/ecco |
152 |
|
- pkg/obcs |
153 |
|
- model/src/the_model_main.F |
154 |
|
- model/src/the_main_loop.F |
155 |
|
|
156 |
|
checkpoint44 |
157 |
|
o apply Shapiro filter on T&S before calc_PhiHyd (stagger time stepping) |
158 |
|
o fix bug in mom_vi_del2uv |
159 |
|
o added some stuff for AD |
160 |
|
|
161 |
|
ecco-branch-mod1 |
162 |
|
o based on c43 |
163 |
|
o replaced packages cost, ctrl, ecco, obcs |
164 |
|
o removed MITgcm verif.'s, added ECCO verif.'s |
165 |
|
o added lsopt, optim for line search optimization. |
166 |
|
|
167 |
checkpoint43 |
checkpoint43 |
168 |
o added two new advection tests |
o added two new advection tests |
169 |
o added global 4x4 degree ocean test |
o added global 4x4 degree ocean test |