| 4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 5 |
============================== |
============================== |
| 6 |
|
|
| 7 |
|
o adjoint: |
| 8 |
|
global_sum_ad was incomplete (noted many times by many people ;) |
| 9 |
|
nice description by Jean U. as follows: |
| 10 |
|
* allreduce(var) in global_sum.F can be interpreted as |
| 11 |
|
reduce(sum) followed by broadcast of sum. |
| 12 |
|
* The adjoint of that is |
| 13 |
|
sum(adjoint) followed by broadcast(summed adjoint value) |
| 14 |
|
or in other words, allreduce(sum, adjoint vars) |
| 15 |
|
We were missing the sum(adjoint)! |
| 16 |
|
Benign, except(!) for "global" scalars appearing in tile loops (bi,bj) |
| 17 |
|
(noted by J.M. Campin, e.g., in cost_ssh_mean, variable offset). |
| 18 |
|
The fix should produce 16-digit accuracy between |
| 19 |
|
single-CPU multi-tile (e.g. nSx=2, nPx=1) and |
| 20 |
|
multi-CPU single-tile (e.g. nSx=1, nPx-2) |
| 21 |
|
We'll probably change this yet again shortly |
| 22 |
|
o pkg/icefront |
| 23 |
|
- ALLOW_SUBGLACIAL_RUNOFF option adds capability for time-evolving |
| 24 |
|
specification of addmass array based on a 2D field, which is added |
| 25 |
|
at bottommost wet level at each horizontal location. |
| 26 |
|
o model/src: |
| 27 |
|
horizontal grid initialisation (other than curvilinear): add code for Exch2 IO |
| 28 |
|
layout; (not always compatible with delX,delY setting; commented out for now). |
| 29 |
|
o pkg/exch2: |
| 30 |
|
change tile to processor setting (case nSy*nPy > 1): allows default model mapping |
| 31 |
|
(myX and myYGlobalLow) to be used for simple (1 facet, no blank tile) config. |
| 32 |
|
|
| 33 |
|
checkpoint62e (2010/04/15) |
| 34 |
|
o pkg/obcs: |
| 35 |
|
- fix wrong index in OBCS_SPONGE_V for Eastern-OB (reported by David Hebert). |
| 36 |
|
o pkg/dic: |
| 37 |
|
-fix bug in fe_chem.F for multi-tile (nSx*nSy > 1) set-up; |
| 38 |
|
(note: not presently tested in any forward verification exp.) |
| 39 |
|
update tutorial_global_oce_biogeo AD output file. |
| 40 |
|
o pkg/exf: |
| 41 |
|
- account for the variation of albedo as a function of zenith angle |
| 42 |
|
by defining ALLOW_ZENITHANGLE in EXF_OPTIONS.h |
| 43 |
|
o build_options: |
| 44 |
|
- rename "linux_ia64_ifort+mpi_ice_nas" to "linux_amd64_ifort+mpi_ice_nas" |
| 45 |
|
o pkg/dic: |
| 46 |
|
change DIC atmos-CO2 box (method 3, dic_int1=3) initialisation & restart: |
| 47 |
|
- use binary pickup file for restart; |
| 48 |
|
- initialise atmos-CO2 box from dic_pCO2 parameter; |
| 49 |
|
- fix multi-threaded. |
| 50 |
|
o pkg/rbcs: |
| 51 |
|
- clean-up RBCS code: add RBCS_OPTIONS.h file ; fix multi-threaded ; |
| 52 |
|
change (+ fix IO calls) type of array RBC_mask from _RL to _RS. |
| 53 |
|
- add RBCS to exp4 passive tracer 01 (rbcs was previously untested) |
| 54 |
|
and update output.txt |
| 55 |
|
o pkg/kpp: |
| 56 |
|
- add diagnostics for KPP non-local flux of Temp, Salinity and pTracers; |
| 57 |
|
- change diagnostic of KPP non-local term: output average of KPPghat*KPPdiffKzS |
| 58 |
|
(= ratio of KPP non-local (salt) flux relative to surface-flux) when |
| 59 |
|
using pkg/timeave & pkg/diagnostics (diag 'KPPghatK' replaces 'KPPghat '); |
| 60 |
|
- change pkg/offline to read-in average KPPghat*KPPdiffKzS from file |
| 61 |
|
KPP_ghatKFile (instead of the average of KPPghat from file KPP_ghatFile). |
| 62 |
|
o pkg/offline: |
| 63 |
|
fix bug in S/R OFFLINE_FIELDS_LOAD regarding surfaceForcing conversion factor |
| 64 |
|
(bug was added in version 1.10 on 2007-08-23). |
| 65 |
|
o pkg/offline: |
| 66 |
|
- no need to specify periodicExternalForcing, etc ... to use offline pack |
| 67 |
|
- offline input files no longer just single prec --> parameter offlineLoadPrec |
| 68 |
|
o verification/lab_sea: |
| 69 |
|
change AD test to multi-tile setup and update all 4 AD output. |
| 70 |
|
NOTE: while cost function values remain very similar, |
| 71 |
|
gradients are very different, although gradient checks are self-consistent |
| 72 |
|
Should try to get to the bottom of this (cost function summation order?) |
| 73 |
|
o pkg/dic: |
| 74 |
|
in carbon_chem.F, replace log10 (AUTODIFF) with explicit form to avoid |
| 75 |
|
a weakness of TAF that drops part of double precision in log10 derivative. |
| 76 |
|
<= changes removed since problem has been fixed in current TAF version 1.9.71 |
| 77 |
|
update AD output of exp. tutorial_dic_adjoffline & tutorial_global_oce_biogeo. |
| 78 |
|
o pkg/ecco: |
| 79 |
|
add missing "_d 0" in cost_readers.F, cost_readtopex.F & cost_readtopexmean.F |
| 80 |
|
update all 4 lab_sea AD output. |
| 81 |
|
|
| 82 |
|
checkpoint62d (2010/03/22) |
| 83 |
|
o build_options: |
| 84 |
|
remove '-r8' (and '-i4') from linux_*pgf and linux_*ifort optfiles. |
| 85 |
|
o remove unbalanced quote (single or double) in commented line of source code |
| 86 |
|
(may cause some warning with some preprocessor/makedepend versions). |
| 87 |
|
o pkg/seaice: |
| 88 |
|
fix seaice_model.F for adjoint (gael) and update AD results of experiments |
| 89 |
|
lab_sea (std + evp + noseaicedyn) and offline_exf_seaice. |
| 90 |
|
o testreport: |
| 91 |
|
add option '-runonly' (short: '-ro') to just do the run part (without make). |
| 92 |
|
o eesupp: remove JAM communication code (obsolete). |
| 93 |
|
o add option to genmake2 & testreport to switch "_RS" type to "real*4" |
| 94 |
|
(if allowed in CPP_EEOPTIONS.h); update default CPP_EEOPTIONS.h and return |
| 95 |
|
to default CPP_EEOPTIONS.h for exp. inverted_barometer & solid-body.cs-32x32x1 |
| 96 |
|
(update also the output). |
| 97 |
|
|
| 98 |
checkpoint62c (2010/03/02) |
checkpoint62c (2010/03/02) |
| 99 |
o tools/build_options : |
o tools/build_options : |
| 100 |
move (some) old optfiles from build_options/ to build_options/unsupported/ |
move (some) old optfiles from build_options/ to build_options/unsupported/ |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch