| 4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 5 |
============================== |
============================== |
| 6 |
|
|
| 7 |
|
o model: add Christopher Wolfe's implemetation of the single reduction cg2d |
| 8 |
|
(d'Azevedo, Eijkhout, and Romine, 1999) with CPP-flag ALLOW_SRCG and |
| 9 |
|
runtime flag useSRCGSolver. Modify verification experiment global_with_exf |
| 10 |
|
to test this code |
| 11 |
|
o pkg/cd_code: |
| 12 |
|
Change defaut D-grid time-stepping to get the same for D-grid momentum as for |
| 13 |
|
C-grid momentum. D-grid velocity used to be stepped forward in time with |
| 14 |
|
Adams-Bashforth only on surface pressure term. Tests show that using AB on |
| 15 |
|
D-grid coriolis term improves stability (as expected from CD-scheme paper). |
| 16 |
|
To reproduce old results, add 2 options (in CD_CODE_OPTIONS.h). |
| 17 |
|
Update results of exp: global_with_exf (x2), ideal_2D_oce, isomip (x2), |
| 18 |
|
lab_sea (3 Fwd +3 AD), natl_box (x2), tutorial_global_oce_biogeo (Fwd+AD), |
| 19 |
|
tutorial_global_oce_latlon, tutorial_global_oce_optim & tutorial_tracer_adjsens. |
| 20 |
|
Keep old results by adding CD_CODE_OPTIONS.h in exp: cfc_example, exp2, |
| 21 |
|
lobal_ocean.90x40x15(+AD), global_ocean_ebm(+AD) and OpenAD. |
| 22 |
|
o pkg/obcs: |
| 23 |
|
- fix bug in initialisation (broken since Oct 1rst modification) |
| 24 |
|
for useOBCSprescribe without useEXF. |
| 25 |
|
|
| 26 |
|
checkpoint61y |
| 27 |
|
o pkg/seaice |
| 28 |
|
- change (once again) the discretization for the evp solver: rather than |
| 29 |
|
squaring the averages of strain rates, average the squares to give |
| 30 |
|
a better approximation to VP behavior (in simple test); this breaks |
| 31 |
|
verification experiments that use EVP (lab_sea and lab_sea.hb87) |
| 32 |
|
--> update outputs |
| 33 |
|
o add CPP-flag (in mom_calc_visc.F) and TAF directive (in gmredi_slope_limit) |
| 34 |
|
to improve vectorization of adjoint code on TARGET_NEC_SX |
| 35 |
|
o verification/vermix: |
| 36 |
|
- change vermix to a more "classical" 1d column with high vertical |
| 37 |
|
resolution, seasonal heat flux forcing and constant wind forcing |
| 38 |
|
--> update outputs |
| 39 |
|
o pkg/seaice |
| 40 |
|
- fix diagnostics of principal stress components for EVP |
| 41 |
|
- add optional diagnostics of non-linear convergence for LSOR solver |
| 42 |
|
- make maximum zeta a runtime parameter |
| 43 |
|
o model/src (3-d solver): |
| 44 |
|
Undo modif from Dec 12, 2005 (solve_for_pressure.F cvs version v.49), |
| 45 |
|
which consisted in removing the Surf. Non-Hydrostatic pressure from the |
| 46 |
|
whole column and adding it to Eta (inspired from Casualli, 1999): |
| 47 |
|
Only affects 3-D solver initial guess and only if using ExactConserv, |
| 48 |
|
but tests did show a degradation of solver convergence. For this reason, |
| 49 |
|
decide to disable this piece of code. |
| 50 |
|
to recover previous version, just set zeroPsNH = exactConserv (line 78). |
| 51 |
|
Update output of exp. global_ocean.cs32x15.viscA4 & hs94.cs-32x32x5.impIGW |
| 52 |
|
o pkg/ggl90: |
| 53 |
|
- fix previous modif (unfortunatly has not been documented here). |
| 54 |
|
o pkg/seaice/autodiff: |
| 55 |
|
- add CPP flag to include clipping of zeta in EVP code; |
| 56 |
|
although clipping of zeta is not necessary for EVP and not recommended |
| 57 |
|
it was done by default, now clipping is disabled by default |
| 58 |
|
- modify computation and store zetaC/Z in seaice_evp for TAF |
| 59 |
|
- add CPP-flag SEAICE_ALLOW_CLIPVELS (undefined by default) |
| 60 |
|
- change seaice_calc_viscosities/strainrates for better |
| 61 |
|
vectorization after AD by TAF |
| 62 |
|
- rearrange common blocks in SEAICE.h once again for easier dumps of |
| 63 |
|
adjoint output, adjust adcommon.h and g_common.h |
| 64 |
|
- change flux form of default advection scheme for better vectorization |
| 65 |
|
of adjoint code |
| 66 |
|
- modify lab_sea/input.lsr to test the flux form of default advection |
| 67 |
|
scheme |
| 68 |
|
o autodiff: fix seaice fields in g_common.h (left over from 3 to 1 level |
| 69 |
|
transition in checkpoint61r) |
| 70 |
|
|
| 71 |
|
checkpoint61x |
| 72 |
|
o adjoint: adding two new verification experiments |
| 73 |
|
* tutorial_global_biogeo: an online dic sensitivity experiment |
| 74 |
|
(work with G. Forget) |
| 75 |
|
* tutorial_dic_adjoffline: an offline dic sensitivity experiment |
| 76 |
|
(an update of the Dutkiewicz et al., 2006 paper) |
| 77 |
|
Both use the same CO2FLUX cost function, but different controls |
| 78 |
|
Gradient checks look good, but some recomp. could still be traightened out |
| 79 |
|
Should add xx_dic control variable to test at some point |
| 80 |
|
o model/src: |
| 81 |
|
add run-time parameters to check and/or mask initial Temp & Salt |
| 82 |
|
o pkg/ptracers: |
| 83 |
|
apply zonal filter (if used) to all passive tracers |
| 84 |
|
|
| 85 |
|
checkpoint61w |
| 86 |
|
o adjoint: |
| 87 |
|
* update and finish work on adjoint of dic package (G. Forget) |
| 88 |
|
* start adjoint of offline package |
| 89 |
|
* bug fix for cube-sphere TLM code (active_file_g.F) |
| 90 |
|
o tutorial_global_oce_optim: |
| 91 |
|
now tested with 4 tiles (change data.grdchk and update output_adm.txt) |
| 92 |
|
and add SIZE.h_mpi for MPI testing |
| 93 |
|
o pkg autodiff |
| 94 |
|
- fix bug in autodiff_restore |
| 95 |
|
- make genmake2 remove comments and empty lines from ad_input_code.f in |
| 96 |
|
order to reduce file size (new script file remove_comments_sed) |
| 97 |
|
o allows to specify vertical profile of vertical viscosity (viscArNr) |
| 98 |
|
o pkg autodiff |
| 99 |
|
autodiff_ini_model_io: comment out writing of grid information to netcdf |
| 100 |
|
files, because at this time the grid fields are a bunch of zeros |
| 101 |
|
(this is analogous to ini_mnc_vars.F) |
| 102 |
|
o pkg seaice/autodiff: |
| 103 |
|
left-over from removing the 3-time-level fields: |
| 104 |
|
- small rearrangement of common blocks (SEAICE.h) |
| 105 |
|
- fix field declaration in order to get meaningful output in |
| 106 |
|
adseaice.*.nc (adcommon.h) |
| 107 |
o pkg obcs: |
o pkg obcs: |
| 108 |
- go through NH code only if nonHydrostatic=T |
- go through NH code only if nonHydrostatic=T |
| 109 |
- fix restart for Orlanski (Hydrostatic; still to fix NH+Orlanski restart) |
- fix restart for Orlanski (Hydrostatic; still to fix NH+Orlanski restart) |
| 110 |
- move call to OBCS_CALC before SEAICE_MODEL which needs seaice-obcields |
- move call to OBCS_CALC before SEAICE_MODEL which needs seaice-obc fields |
| 111 |
=> Update output of experiment seaice_obcs |
=> Update output of experiment seaice_obcs |
| 112 |
- useOBCSbalance: code is wrong with multi-tile / proc: add a stop. |
- useOBCSbalance: code is wrong with multi-tile / proc: add a stop. |
| 113 |
- hack obcs_calc.F to fix restart in experiment seaice_obcs. |
- fix NH+Orlanski restart (add wvel OBC value in pickup file) |
| 114 |
|
- fix restart for useOBCSprescribe + useEXF |
| 115 |
o pkg ctrl |
o pkg ctrl |
| 116 |
- make obcs as control parameter work also with useSingleCPUio |
- make obcs as control parameter work also with useSingleCPUio |
| 117 |
- replace a few sny and snx by Ny and Nx to be consistent with |
- replace a few sny and snx by Ny and Nx to be consistent with |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch