4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
5 |
============================== |
============================== |
6 |
|
|
7 |
|
checkpoint65h (2014/12/17) |
8 |
|
o verification/global_ocean.90x40x15, global_ocean_ebm: |
9 |
|
- update to generic controls. |
10 |
|
o pkg/ctrl: |
11 |
|
- ctrl_map_ini_genarr.F: test for igen>0 |
12 |
|
- ctrl_init_variables.F: add call to CTRL_MAP_INI_GENTIM2D. |
13 |
|
- move gentim2d part of ctrl_map_ini_genarr.F to |
14 |
|
new separate routine : ctrl_map_ini_gentim2d.F |
15 |
|
- ctrl_cost_driver.F,ctrl_init.F ctrl_map_ini_genarr.F: |
16 |
|
add time variable weights and cyclic controls (time mean, |
17 |
|
seasonal cycle, etc) capabilities to gentim2d |
18 |
|
o pkg/ecco: |
19 |
|
- move reading of error, conversion to weights from |
20 |
|
ecco_cost_init_fixed.F to ecco_toolbox.F (ecco_readwei) |
21 |
|
o global_oce_llc90, cs32: |
22 |
|
- activte sbo. |
23 |
|
- CTRL_SIZE.h : add maxCtrlProc. |
24 |
|
- activate the new genctrl capabilities. |
25 |
|
- update following reorganization of global_oce_input_fields.tar.gz |
26 |
|
o pkg/ctrl/ctrl_ad_diff.list: |
27 |
|
- add missing ctrl_init_rec.f |
28 |
|
o tools/OAD_support/keepOriginal: |
29 |
|
- add ctrl_init_rec.F |
30 |
|
o model/src/the_main_loop.F, pkg/ctrl/ctrl_map_ini_genarr.F: |
31 |
|
- avoid recomputations of xx_gentim2d_dummy. |
32 |
|
o momentum diagnostics: |
33 |
|
- move filling of diagnostics Um_Diss & Vm_Diss from mom_fluxform.F and |
34 |
|
mom_vecinv.F to timestep.F (to include Smag-3D contribution). |
35 |
|
o pkg/seaice/seaice_lsr.F: |
36 |
|
- fix computation of BU/BV, make sure they are non-zero, only necessary |
37 |
|
for cubed sphere exchanges and SEAICE_OLx/y>0. |
38 |
|
This makes calling s/r fill_cs_corner_rl obsolete (removed). |
39 |
|
o verification/global_ocean.cs32x15/input.seaice |
40 |
|
- add test for strong implicit coupling and restricted addtive Schwarz |
41 |
|
methods for LSR |
42 |
|
- update results/output.seaice.txt |
43 |
|
o verification/tutorial_global_oce_biogeo: |
44 |
|
- AD & OAD set-up: switch off ALLOW_OLD_VIRTUALFLUX (like in fwd set-up) |
45 |
|
and set PTRACERS_EvPrRn(1,2) to zero (for DIC & ALK); |
46 |
|
update output_adm,_oadm,_tlm.txt |
47 |
|
o pkg/dic: |
48 |
|
- change default to "#undef ALLOW_OLD_VIRTUALFLUX" in DIC_OPTIONS.h |
49 |
|
- add a stop when ALLOW_OLD_VIRTUALFLUX is defined and PTRACERS_EvPrRn |
50 |
|
for tracer 1 & 2 is set (to avoid double counting of dilution effect). |
51 |
|
o verification/tutorial_global_oce_biogeo: |
52 |
|
- AD & OAD set-up: do not set PTRACERS_EvPrRn for tracer 1 & 2 since |
53 |
|
dilution is already added by pkg/dic ALLOW_OLD_VIRTUALFLUX ; |
54 |
|
update output_adm,_oadm,_tlm.txt |
55 |
|
o eesupp/src: |
56 |
|
- add fill_cs_corner_rl.F (_RL-copy of fill_cs_corner_rs.F) |
57 |
|
and call it in seaice_lsr when using nonzero SEAICE_OLx/y |
58 |
|
o pkg/seaice: introduce strongly implicit coupling, intended for stabilizing |
59 |
|
LSR (following Hutchings et al. 2004) |
60 |
|
- introduce zetaZ as a global variable (requires adjustments in |
61 |
|
seaice_evp.F and seaice_jfnk.F) and compute analoguously to etaZ |
62 |
|
in seaice_calc_viscosities.F |
63 |
|
- new flag SEAICEuseStrImpCpl (default off) |
64 |
|
- add new terms zetaZ*du/dy and zetaZ*dv/dx on both sides of the |
65 |
|
momentum equations |
66 |
|
- JFNK-solver (preconditioner) adjusted to allow using this method |
67 |
|
(where it does not have any positive effect) |
68 |
|
o pkg/seaice |
69 |
|
- add overlap for restrictive additive Schwartz method also to LSR, but |
70 |
|
turn it off by default (because it only seems to work for non-cubed |
71 |
|
sphere/llc topologies) |
72 |
|
o pkg/autodiff |
73 |
|
- add extra exchanges in addummy_in_steppging also for mnc output |
74 |
|
o pkg/seaice: |
75 |
|
- add extra exchanges before dumping adjoint fields in seaice_ad_dump.F |
76 |
|
to avoid stripes in the adjoint output (similar to addummy_in_stepping.F) |
77 |
|
o pkg/obcs: |
78 |
|
- add option to specify domain-connected piece Id along OB though files |
79 |
|
(for now, only used to diagnose PhiVel; default =1 -> global correction) |
80 |
|
- modify obcs_diag_balance.F to process each connected-part separately; |
81 |
|
exclude OB where Id = 0 but keep the full divergence there (-> constant |
82 |
|
Psi along this portion of OB and all inflow derived from grad(Phi)) |
83 |
|
- add example of OB-conect files in verification/so_box_biogeo/input. |
84 |
|
o pkg/diagnostics: |
85 |
|
- improve treatment of Open-Boundaries in Velocity-Potential calculation: |
86 |
|
a) use non zero matrix element across OB and solve for Phi 1 grid point |
87 |
|
outside OB (i.e, where tracer OBCS are applied). |
88 |
|
b) global imbalance of inFlow/outFlow across OB: instead of modifying |
89 |
|
velocity across OB, spread it uniformly along the OB in the RHS. |
90 |
|
|
91 |
|
checkpoint65g (2014/11/19) |
92 |
|
o pkg/diagnostcis: |
93 |
|
- add preconditioner off-diagonal factor (diagCG_pcOffDFac) as run-time param |
94 |
|
o pkg/my82,pp81,kl10: |
95 |
|
- fix ALLOW_3D_DIFFKR implementation (previously had no effect): use 3-D |
96 |
|
field diffKr as background diffusivity to set minimum diffusivity value. |
97 |
|
o pkg/seaice: |
98 |
|
- add new parameter SEAICEuseEVPrev (for revisited EVP by |
99 |
|
Bouillon et al 2013) |
100 |
|
- add SEAICE_evpAlpha/Beta to namelist |
101 |
|
- modify seaice_evp code to able to use SEAICEuseEVPrev. These changes |
102 |
|
involve trivial re-ordering of divisions/multiplications that should not |
103 |
|
have any effects, but ... they do change the results of lab_sea.hb87 so |
104 |
|
that only 2 digits of agreement remain, in line with the general |
105 |
|
instability/fragility of the evp-code. For now, I do not care and just |
106 |
|
update the verification experiment |
107 |
o pkg/ctrl : add generic pre-processing (xx_genarr2d_preproc etc) |
o pkg/ctrl : add generic pre-processing (xx_genarr2d_preproc etc) |
108 |
- remove xx_genarr2d_numsmooth etc |
- remove xx_genarr2d_numsmooth etc |
109 |
- use xx_genarr2d_preproc etc instead |
- use xx_genarr2d_preproc etc instead |
110 |
- carry out smooth_correl2D at initialization |
- carry out smooth_correl2D at initialization stage rather than during |
111 |
stage rather than during time stepping. |
time stepping. |
112 |
- get smoothOpNb from xx_gentim2d_preproc |
- get smoothOpNb from xx_gentim2d_preproc |
113 |
o global_oce_cs32 : |
o global_oce_cs32 : |
114 |
- add test of generic ecco capabilities. |
- add test of generic ecco capabilities. |
115 |
o pkg/exf/exf_getsurfacefluxes.F : |
o pkg/exf/exf_getsurfacefluxes.F : |
116 |
- add to tmpUE/tmpVN rather than re-initializing them (conventional behavior). |
- add to tmpUE/tmpVN rather than re-initializing them (conventional behavior). |
117 |
o pkg/ctrl,ecco : |
o pkg/ctrl,ecco : |
118 |
- omit un-necessary arguments to ctrl_cost_gen.F |
- omit un-necessary arguments to ctrl_cost_gen.F unless ECCO_CTRL_DEPRECATED |
119 |
unless ECCO_CTRL_DEPRECATED is defined. |
is defined. |
120 |
o pkg/seaice: add diagnostics for Delta (deformation parameter) |
o pkg/seaice: add diagnostics for Delta (deformation parameter) |
121 |
o model/src: |
o model/src: |
122 |
- when ALLOW_3D_DIFFKR is defined, add a STOP if any of the vertical |
- when ALLOW_3D_DIFFKR is defined, add a Warning if any of the vertical |
123 |
diffusivity (background) is specified but ignored. |
diffusivity (background) is specified but ignored. |
124 |
o verification: |
o verification: |
125 |
- in testreport, convert hostname to lower case for default output dir name |
- in testreport, convert hostname to lower case for default output dir name |
519 |
nearly a factor of 10 (but keep in mind there is mostly seaice code in |
nearly a factor of 10 (but keep in mind there is mostly seaice code in |
520 |
this experiment): |
this experiment): |
521 |
> size mitgcmuv_ad |
> size mitgcmuv_ad |
522 |
text data bss dec hex |
text data bss dec hex |
523 |
5100659 1216 220458952 225560827 d71c8fb |
5100659 1216 220458952 225560827 d71c8fb |
524 |
5339425 1200 1868783272 1874123897 6fb4dc79 (SEAICE_LSR_ADJOINT_ITER) |
5339425 1200 1868783272 1874123897 6fb4dc79 (SEAICE_LSR_ADJOINT_ITER) |
525 |
but only a moderate increase of 36% for global_ocean.cs32x15: |
but only a moderate increase of 36% for global_ocean.cs32x15: |
526 |
text data bss dec hex |
text data bss dec hex |