4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
5 |
============================== |
============================== |
6 |
|
|
7 |
|
o utils/matlab/rdmnc.m: adapt to be able to use the generic matlab netcdf API |
8 |
|
o pkg/ctrl: require a specific set of control variables and exf-flags for |
9 |
|
adjoint of seaice (stop in ctrl_check.F if conditions are not satisfied) |
10 |
|
o model/src: |
11 |
|
a) 1rst version of a new driver S/R to turn-off output freq and flags after the |
12 |
|
1rst forward sweep, to avoid over-writing output files during recomputations |
13 |
|
recomputations (& grdchk); called from cost_final.F (& ctrl_init_ctrlvar.F ?) |
14 |
|
b) 2nd version: |
15 |
|
- add specific pkg S/R for seaice & ptracers (called from S/R TURNOFF_MODEL_IO); |
16 |
|
- switch-off more output flags (exf_monFreq, chkPtFreq, dumpInitAndLast, |
17 |
|
writePickupAtEnd) + seaice & ptracers output; |
18 |
|
- do not switch off MNC flags (otherwise, won't get admonitor netcdf file) |
19 |
|
o model/src: |
20 |
|
- new S/R rotate_uv2en_rl (..._rs) that rotates vector fields. |
21 |
|
o pkg/exch2: |
22 |
|
- remove "simulationMode" from arg. list of S/R EXCH2_RX1,2_CUBE ; |
23 |
|
- add argument "signOption" to EXCH2_RX1_CUBE (will be needed for SM exch). |
24 |
|
o pkg/autodiff & exch2: |
25 |
|
process exch_uv_agrid_3d_rx.f & exch2_uv_agrid_3d_rx.f through TAF |
26 |
|
(for now, only if ALLOW_EXCH2 is defined). |
27 |
|
o pkg/autodiff: |
28 |
|
- include PACKAGES_CONFIG.h in exch*ad.flow & exch*ad.F |
29 |
|
(not supposed to be in CPP_OPTIONS.h) to get ALLOW_EXCH2 defined correctly. |
30 |
|
+ clean-up comments (_RS is not always real*4) |
31 |
|
- fix bug in ADEXCH_UV_3D_RX (without EXCH2, was calling the forward exch S/R) |
32 |
|
- add a stop in ADEXCH_UV_AGRID_3D_RX if trying to use it with EXCH2: |
33 |
|
was calling a forward exch2, which is wrong ; used in SEAICE & EXF pkgs. |
34 |
|
o testreport (AD-tests): |
35 |
|
no longer link files from dir "../input" (1rst link files from |
36 |
|
"../input_ad.[xx]" if secondary test "xx" and then from "../input_ad") |
37 |
|
|
38 |
|
checkpoint62f (2010/04/29) |
39 |
|
o model/src/the_main_loop.F: |
40 |
|
adjust store directives for seaice evp and multicategory according to |
41 |
|
pkg/ecco/the_main_loop.F |
42 |
|
o adjoint: |
43 |
|
global_sum_ad was incomplete (noted many times by many people ;) |
44 |
|
nice description by Jean U. as follows: |
45 |
|
* allreduce(var) in global_sum.F can be interpreted as |
46 |
|
reduce(sum) followed by broadcast of sum. |
47 |
|
* The adjoint of that is |
48 |
|
sum(adjoint) followed by broadcast(summed adjoint value) |
49 |
|
or in other words, allreduce(sum, adjoint vars) |
50 |
|
We were missing the sum(adjoint)! |
51 |
|
Benign, except(!) for "global" scalars appearing in tile loops (bi,bj) |
52 |
|
(noted by J.M. Campin, e.g., in cost_ssh_mean, variable offset). |
53 |
|
The fix should produce 16-digit accuracy between |
54 |
|
single-CPU multi-tile (e.g. nSx=2, nPx=1) and |
55 |
|
multi-CPU single-tile (e.g. nSx=1, nPx-2) |
56 |
|
We'll probably change this yet again shortly |
57 |
|
o eesupp/src: |
58 |
|
- fix global_sum_r4 (wrong from previous modif in Jun 2009; ever used ?) |
59 |
|
o pkg/icefront |
60 |
|
- ALLOW_SUBGLACIAL_RUNOFF option adds capability for time-evolving |
61 |
|
specification of addmass array based on a 2D field, which is added |
62 |
|
at bottommost wet level at each horizontal location. |
63 |
|
o model/src: |
64 |
|
horizontal grid initialisation (other than curvilinear): add code for Exch2 IO |
65 |
|
layout; (not always compatible with delX,delY setting; commented out for now). |
66 |
o pkg/exch2: |
o pkg/exch2: |
67 |
change tile to processor setting (case nSy*nPy > 1): allows default model mapping |
change tile to processor setting (case nSy*nPy > 1): allows default model mapping |
68 |
(myX and myYGlobalLow) to be used for simple (1 facet, no blank tile) config. |
(myX and myYGlobalLow) to be used for simple (1 facet, no blank tile) config. |