| 4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 5 |
============================== |
============================== |
| 6 |
|
|
| 7 |
|
o OpenAD |
| 8 |
|
- Adding preliminary ggl90 test to OpenAD verification. |
| 9 |
|
Config. with ggl90 currently diverges after 64 time steps |
| 10 |
|
so will need more work to become more stable |
| 11 |
|
- Adding preliminary kpp test to OpenAD verification. |
| 12 |
|
required small tweakes in kpp_routines.F |
| 13 |
|
- Merge and update from Krishna Narayanan's contrib area: |
| 14 |
|
* Add one variable for preliminary OpenAD DIVA |
| 15 |
|
* Add replacement time-stepping for OpenAD DIVA |
| 16 |
|
* genmake2 flag -diva (but only for OpenAD; preliminary) |
| 17 |
|
* required modifs for OAD_support |
| 18 |
|
successful preliminary test for tutorial_global_oce_biogeo/ |
| 19 |
|
o pkg/seaice: |
| 20 |
|
- initialise deltaC in seaice_lsr, cosmetic changes in seaice_lsr |
| 21 |
|
- add cpp-flag SEAICE_DELTA_SMOOTHREG for option of regularising |
| 22 |
|
delta with a smooth function in s/r seaice_calc_viscosity |
| 23 |
|
(no effect on EVP) |
| 24 |
|
- rename local suffix "sqr" into "sq" for more consistent variable names |
| 25 |
|
o pkg/seaice: add new parameter SEAICE_deltaMin that is used ***only*** |
| 26 |
|
for regularizing Delta (and nothing else, like the parameter |
| 27 |
|
which is also used for all sorts of things SEAICE_EPS). Defaults to |
| 28 |
|
SEAICE_EPS for backward compatibility only. |
| 29 |
|
o pkg/seaice: replace erroneously storing e12Csqr on the wrong tape with |
| 30 |
|
the wrong key by inialising it before use -> fixes a recomputation |
| 31 |
|
problem but does not fix the gradient |
| 32 |
|
o pkg/layers: |
| 33 |
|
- move resetting of flux arrays to outside the iLa loop |
| 34 |
|
(this caused layers_diapycnal to not work with more than one layer coord.) |
| 35 |
|
- fix bug in calculation of layers_recip_delta in layers_init_fixed.F |
| 36 |
|
o pkg/layers: |
| 37 |
|
- remove from common block 3D layers fields (used to diagnose transport |
| 38 |
|
in tracer coords) and declare them as local arrays in layers_calc.F |
| 39 |
|
o pkg/seaice: |
| 40 |
|
- Rename SEAICE_VECTORIZE_LSR_ZEBRA to SEAICE_LSR_ZEBRA and define it |
| 41 |
|
in lab_sea forward experiment (changes results -> updated) |
| 42 |
|
- update adjoint experiment lab_sea after cleaning up seaice_lsr_tridiagu/v |
| 43 |
|
o pkg/seaice: modify seaice_lsr.F in order to improve the gradient |
| 44 |
|
computations; for SEAICE_VECTORIZE_LSR |
| 45 |
|
- move the loops over which the tridiagonal solvers (seaice_lsr_tridiagu/v) |
| 46 |
|
sweep to the outside, |
| 47 |
|
- remove store directives and add "CADJ loop sequential" directives |
| 48 |
|
in analogy to model/src/solve_tridiagonal.F |
| 49 |
|
- replace many "#ifdef SEAICE_VECTORIZE_LSR_ZEBRA" by variable loop |
| 50 |
|
boundaries to yield more readable code. This has also the charming side |
| 51 |
|
effect that your can use SEAICE_VECTORIZE_LSR_ZEBRA without |
| 52 |
|
SEAICE_VECTORIZE_LSR (i.e. adjoint without recomputations in |
| 53 |
|
seaice_lsr_tridiagu/v now requires either of these flags, vectorization |
| 54 |
|
still requires SEAICE_VECTORIZE_LSR with SEAICE_VECTORIZE_LSR_ZEBRA as |
| 55 |
|
an additional option) |
| 56 |
|
The tridiagonal solvers are now completely analoguous to solve_tridiagonal.F |
| 57 |
|
but the adjoint gradients (which are affected by this change) still explode. |
| 58 |
|
o coupler (atm & ocn _compon_interf): |
| 59 |
|
- avoid using Exch2 IO mapping if not suitable. |
| 60 |
|
|
| 61 |
|
checkpoint65m (2015/06/15) |
| 62 |
|
o pkg/layers: |
| 63 |
|
- clean-up old parameters setting (from Sep 2012): |
| 64 |
|
+ remove old parameters setting from common block (declared locally); |
| 65 |
|
+ stop if any old parameter is mixed with new parameters; |
| 66 |
|
+ check for valid or consistent parameter setting (e.g., layers_name) |
| 67 |
|
- check that layers_bounds for Pot.Density are in the right range; |
| 68 |
|
this is supposed to catch un-updated setting after recently shifting |
| 69 |
|
potential density field "prho" by -1000 (now in "sigma" scale). |
| 70 |
|
o pkg/layers: |
| 71 |
|
- Added missing call to layers_fill for keeping track of tottend in |
| 72 |
|
pkg/diagnostics/diagnostics_fill_state.F |
| 73 |
|
o pkg/layers: |
| 74 |
|
- added CPP flux LAYERS_FINEGRID_DIAPYCNAL to disable fine-grid binning |
| 75 |
|
of diapycnal fluxes |
| 76 |
|
o pkg/layers: |
| 77 |
|
- reverted changes to interpolation to previous treatment; there |
| 78 |
|
was never actually a problem |
| 79 |
|
o pkg/layers: |
| 80 |
|
- fix interpolation of layers tracers in the presence of partial cells |
| 81 |
|
- add lots of new functionality to compute water mass transformation |
| 82 |
|
(enabled by CPP flag LAYERS_THERMODYNAMICS) |
| 83 |
|
- add new calls from main code (diags_oceanic_surf_flux.F, |
| 84 |
|
thermodynamics.F, impldiff.F) and gad to save the necessary fields |
| 85 |
|
for layers_thermodynamics calculations |
| 86 |
|
o pkg/ptracers: |
| 87 |
|
- add capability of stepping a passive tracer forward without advection |
| 88 |
|
by setting PTRACERS_advScheme to 0. |
| 89 |
|
o pkg/diagnostics: |
| 90 |
|
- add few (threads) _BARRIER around any update of "diag_pkgStatus". |
| 91 |
|
o pkg/layers: |
| 92 |
|
- shift potential density field "prho" by -1000 for density layer averaging so |
| 93 |
|
that it corresponds to usual "sigma" definition (i.e., rho(p=const)-1000); |
| 94 |
|
Note that density bin boundaries (layers_bounds in data.layers) have |
| 95 |
|
to be set accordingly (i.e., shifted, in sigma scale). |
| 96 |
|
o pkg/layers: |
| 97 |
|
- fix snap-shot output of layer thickness @ V.pt and output of pot.density |
| 98 |
|
- fix calculation of layer non-weighted velocity and layer probability |
| 99 |
|
- add diagnostics for pot.density used for layer averaging |
| 100 |
|
o pkg/seaice: |
| 101 |
|
- seaice_calc_viscosity.F: change computation of deltaC**2 to ensure |
| 102 |
|
positiveness, modify a few comments and improve variable names, |
| 103 |
|
- the changed computation of deltaC affects most seaice-related |
| 104 |
|
experiments at the truncation level. |
| 105 |
|
seaice_itd.thermo is affected by far the most, with |
| 106 |
|
only 7 digits of agreement remaining; update experiments. |
| 107 |
|
|
| 108 |
|
checkpoint65l (2015/05/04) |
| 109 |
|
o pkg/shelfice: |
| 110 |
|
- in shelfice_thermodynamics.F, if SHELFICEBoundaryLayer=.true., then set |
| 111 |
|
factor drKp1 to zero where it is negative (ie. do not use |
| 112 |
|
SHELFICEBoundaryLayer if hFacC>1) |
| 113 |
|
- the same for factor drLoc in shelfice_forcing.F |
| 114 |
|
o tools/genmake2: move FEXTRAFLAGS from FFLAGS to FOPTIM to avoid a problem |
| 115 |
|
with modules in ptracers, when using testreport option "-small_f" |
| 116 |
|
o pkg/seaice/seaice_growth.F (in case of SEAICE_ITD) |
| 117 |
|
- replace tmpscal1**1.36 by faster exp(1.36*log(tmpscal1)) |
| 118 |
|
o verification/testreport |
| 119 |
|
- add new option -small_f |
| 120 |
|
with this option testreport first makes the target small_f (*.f files) |
| 121 |
|
before running make (all). This is useful for explicit inlining when |
| 122 |
|
all sources need to be full available. |
| 123 |
|
o pkg/exf/exf_interp.F: |
| 124 |
|
remove previous changes related to reverse (decreasing) input grids latitude |
| 125 |
|
|
| 126 |
|
checkpoint65k (2015/04/01) |
| 127 |
|
o pkg/ecco: |
| 128 |
|
fix missing cosphi initialization; use eccoiter in place of optimcycle; |
| 129 |
|
if autodiff is not compiled then use READ_REC_XY_RL/READ_REC_XYZ_RL |
| 130 |
|
instead of active read/write |
| 131 |
|
o pkg/ctrl: |
| 132 |
|
if autodiff is not compiled then use READ_REC_XY_RL/READ_REC_XYZ_RL |
| 133 |
|
instead of active read/write. |
| 134 |
|
o pkg/autodiff: make autodiff_readparms.F and autodiff_whtapeio_sync.F |
| 135 |
|
return if .NOT.useAUTODIFF |
| 136 |
|
o pkg/seaice: add CPP brackets (cost function related). |
| 137 |
|
o model/src: |
| 138 |
|
add useAUTODIFF run time switch. |
| 139 |
|
o pkg/exf/exf_interp.F: |
| 140 |
|
add code to allow input grids with latitude starting in the north (i.e. when |
| 141 |
|
j=1 corresponds to northern edge of field); modify global_with_exf to test |
| 142 |
|
this feature. |
| 143 |
|
o verification/testreport: |
| 144 |
|
- replace "grep -A3" with POSIX conformal sed command, because there are |
| 145 |
|
implementations of grep that cannot do contextual searches |
| 146 |
|
- replace "cat -n" by "nl" (two occurrences), because not all implementations |
| 147 |
|
of "cat" have the option "-n" |
| 148 |
|
o model/src: |
| 149 |
|
add new file gsw_teos10.F that contains 3 functions, more or less a copy |
| 150 |
|
of the gsw_toolbox.f90 of the TEOS-10 fortran interface, that are used to |
| 151 |
|
convert from conservative to potential temperature. For now they are not |
| 152 |
|
called anywhere. |
| 153 |
|
o pkg/ggl90: make "useIDEMIX" useful |
| 154 |
|
- re-organize code in s/r ggl90_calc a little to move computations that |
| 155 |
|
are different for idemix into separate loops (changes verification |
| 156 |
|
experiment idemix at truncation level, 12 digits remain -> updated) |
| 157 |
|
- requires 2 new 2D fields KappaM, verticalShear |
| 158 |
|
- make hFacI and recip_hFacI local 3D variables that are passed to |
| 159 |
|
s/r ggl90_idemix |
| 160 |
|
- fix and improve some comments |
| 161 |
|
|
| 162 |
|
checkpoint65j (2015/02/25) |
| 163 |
|
o pkg/openad |
| 164 |
|
- synchronize seawater.F with version in model/src: replace routine |
| 165 |
|
body of sw_temp by calling sw_ptmp and fix/add a few comments. |
| 166 |
|
o pkg/mnc: |
| 167 |
|
- change default value of pickup_read/write_mnc params to FALSE |
| 168 |
|
(since few features are missing in reading MNC pickup files). |
| 169 |
|
o model/src: |
| 170 |
|
- change default (background) vertical diffusivity for salt to be the |
| 171 |
|
temperature (background) vertical diffusivity. |
| 172 |
|
- change background vertical diffusivity in vertical mixing pkgs ggl90, |
| 173 |
|
kl10, my82 and pp81 from temperature diffusivity to salinity diffusivity. |
| 174 |
|
This makes ptracers default diffusivity (that uses salt diffKr) more |
| 175 |
|
consistent with vertical mixing schemes. |
| 176 |
|
o verification/testreport: |
| 177 |
|
- fix typo in "-pcls" option ; |
| 178 |
|
- put temp files in /tmp (instead of in local dir) |
| 179 |
|
- add option "-ncad" to use genmake2 new option "-nocat4ad" |
| 180 |
|
- allow to use "-j" option (for multi-procs make) for AD, OAD and TLM. |
| 181 |
|
o pkg/ggl90 (idemix code): |
| 182 |
|
- apply few fixes ; add some missing "_d 0" ; update output of exp. |
| 183 |
|
global_ocean.90x40x15.idemix |
| 184 |
|
o OpenAD |
| 185 |
|
- enable active file I/O and begin removal of work-around code |
| 186 |
|
for OpenAD for handling control variables (memory vs. active files) |
| 187 |
|
- new CPP options file OPENAD_OPTIONS.h |
| 188 |
|
- revert to genarr2d gradient check for hs94.1x64x5 |
| 189 |
|
- (genarr3d tested by tutorial_global_oce_biogeo, tutorial_tracer_adjsens) |
| 190 |
|
- gentim2d works for halfpipe_streamice but not yet hs94.1x64x5 |
| 191 |
|
o tools/genmake2: |
| 192 |
|
- add option for an alternative Makefile (from Martin) where multiple |
| 193 |
|
source files are sent to TAF (instead of all in one file). |
| 194 |
|
o pkg/ggl90: add IDEMIX (Olbers and Eden, 2013) |
| 195 |
|
- code provided by Carsten Eden as an extension of ggl90 |
| 196 |
|
- so far the code is turned on within ggl90 by setting a CPP-flag at |
| 197 |
|
compile time; a runtime flag implementation is not yet complete |
| 198 |
|
- tested in new verification experiment global_ocean.90x40x15.idemix |
| 199 |
|
o verification: |
| 200 |
|
- reduce length of the AD test run by half for experiments |
| 201 |
|
global_ocean.cs32x15 (std + thsice) and isomip |
| 202 |
|
- enable active file I/O for OpenAD, adjust verifications accordingly |
| 203 |
|
- halfpipe_streamice: change gradient check from genarr2d to gentim2d |
| 204 |
|
o verification/testreport: |
| 205 |
|
- fix two non-POSIX sed statements so that there are no error messages |
| 206 |
|
on Mac OS X |
| 207 |
|
- replace hostname -s with hostname | sed 's/\..*$//' |
| 208 |
|
o model/src/seawater.F: |
| 209 |
|
- simplify SW_TEMP to just calling SW_PTMP with P and PR exchanged |
| 210 |
|
(following the matlab code), which would have avoided the bug caught |
| 211 |
|
by Matt (see below) in the first place |
| 212 |
|
o pkg/shelfice: |
| 213 |
|
- implement alternative discretisation of quadratic drag (selectBotDragQuadr) |
| 214 |
|
and true distance in no-slip BC (bottomVisc_pCell) similar to bottom-drag. |
| 215 |
|
- fix sign of friction tendency (no-slip BC) and fix u,v drag for thin |
| 216 |
|
ice-shelf case (i.e., within surface level). |
| 217 |
|
- add config summary in shelfice_check.F |
| 218 |
|
- remove global variable recip_SHELFICElatentHeat and replace by a local |
| 219 |
|
variable recip_latentHeat |
| 220 |
|
- introduce diagnostics for friction velocity uStar (in case of |
| 221 |
|
SHELFICEuseGammaFrict) |
| 222 |
|
- change verification experiment isomip.htd: |
| 223 |
|
+ start from pickup (as input_ad at nIter0=8640) |
| 224 |
|
+ start from SHELFICEmassFile instead of SHELFICEloadAnomaly file |
| 225 |
|
+ test SHELFICEuseGammaFrict = .TRUE. |
| 226 |
|
+ test SHELFICEadvDiffHeatFlux = .TRUE. |
| 227 |
o model/src: |
o model/src: |
| 228 |
- fix bug (Thanks to Matt) in function SW_TEMP (to convert Pot.Temp to |
- fix bug (Thanks to Matt) in function SW_TEMP (to convert Pot.Temp to |
| 229 |
in-situ Temp) and in routine SW_TEMP (in pkg/openad); |
in-situ Temp) and in routine SW_TEMP (in pkg/openad); |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch