--- MITgcm/doc/tag-index 2013/07/01 16:55:57 1.1889 +++ MITgcm/doc/tag-index 2013/08/27 21:34:21 1.1919 @@ -1,12 +1,176 @@ -$Header: /home/ubuntu/mnt/e9_copy/MITgcm/doc/tag-index,v 1.1889 2013/07/01 16:55:57 dimitri Exp $ +$Header: /home/ubuntu/mnt/e9_copy/MITgcm/doc/tag-index,v 1.1919 2013/08/27 21:34:21 dimitri Exp $ $Name: $ Notes on tags used in MITgcmUV ============================== +checkpoint64n (2013/08/26) +o pkg/bbl: + - update formulation to reduce dispersion of tracer properties in bbl + - replaced horizontal and vertical transport parameters with velocities + - changes results for verification/global_with_exf.yearly +o verification/halfpipe_streamice: + - change to matrix construction; update both Fwd & AD output. +o verification/tutorial_tracer_adjsens: + - OpenAD test: switch to vector-invariant momentum since fluxform with + rStar causes problems with OpenAD; update output. +o tools: + - genmake2: define a common prefix for all temp-file names ; move them + to local /tmp disk (but would be easy to move them back); + - xmakedepend: apply the same switch for temp-files. +o pkg/offline: + - changes for use with rStar (NLFS): assume that we continue to provide + UVELMASS/hFacW_ini & VVELMASS/hFacS_ini (hFacWi,S_ini as written in grid + output files) and apply r* scaling at each time-step (function of the time + varying Eta field); use normal pickup files (needed for restart with NLFS). +o adjoint with OpenAD: + - move OpenAD-specific files from verification/OpenAD/code_oad_all to a new + package openad. This package has to be compiled for all runs with OpenAD + instead of specifying code_oad_all as a modification directory. + - link the fortran templates from tools/OAD_support to the build directory + for preprocessing to avoid conflicts between different setups compiled + simultaneously and simplify cleanup. +o pkg/exf: + - in exf_getforcing.F: comment out most of the EXCH calls (fields used by + main model or other pkgs are exchanged in EXF_MAPFIELDS). +o model/src: + - do_stagger_fields_exchanges.F: skip EXCH calls when using pkg/offline + (similar to do_fields_blocking_exchanges.F). +o tools/genmake2: + - remove TAF output src code just before calling staf (when not doing a fresh + built, will stop if things go wrong with staf). + - fix target "adobj" (left from TAF version-2 transition) +o pkg/diagnostics: + - add a parameter to track the status of pkg/diagnostics activation; + this replace/extend the use of logical param "settingDiags" (to check + when adding diag to the list) to also check any DIAGNOSTICS_[]FILL* call. + New S/R to print error message and stop when pkgStatus is not right. + - to try: add a warning when "data.diagnostics" exist but useDiagnostics=F. + +checkpoint64m (2013/08/13) +o model/inc: + - move FCORI_MACROS.h from PARAMS(_MACROS).h to GRID_MACROS.h (left from + 8 yrs ago modif when Coriolis arrays were moved from PARAMS.h to GRID.h) +o pkg/kpp & aim_v23: + - only call DIAGNOSTICS_FILL when useDiagnostics=TRUE. +o pkg/seaice: + - restore (last time on 2011/12/24) ability to compile pkg/seaice with + undefined pkg/exf ALLOW_ATM_TEMP or ALLOW_RUNOFF CPP-options. + - only call DIAGNOSTICS_FILL when useDiagnostics=TRUE. +o eesupp/src: + - fix check for overlap-size argument (myOLn.GT.MAX_OLY_EXCH) + in exch1_rx[_ad].template (thanks to Francois Lefeuvre report). +o switch to TAF version-2: + - add flow directives to "freeze" naming convention for ad-routines that + are used in I/O routines of adjoint state and associated common blocks + that are used in adcommon.h (backward compatible with version-1) + - update all ad-optfiles (in tools/adjoint_options + few local to some + verification exp) to switch to latest TAF version (currently 2.5.5); + merge "*_mpi" adoptfiles into the corresponding non-mpi (but commented + out except for diva). + - the major change with TAF version-2 is that both the differentiated + routines and the forward routines (new) are returned as part of the TAF + output source code. Update genmake2 accordingly. + - new output for AD experiment tutorial_global_oce_biogeo (better AD gradient + with version-2 when compared to TLM or FWD gradient); other experiments + output show minor changes (machine truncation level); update the few AD & + TLM output with 12 matching digits only (some prior to version-2 switch). + +checkpoint64l (2013/08/05) +o model/src: + - in dynamics.F, skip the call to CALC_VISCOSITY if momViscosity=F + - main_do_loop.F: no longer includes additional (AUTODIFF) header files + when using TAF but only if using OpenAD. +o pkg/mom_common+fluxform+vecinv: + - always set horiz. viscosity arrays to background value before calling + MOM_CALC_VISC (in MOM_FLUXFORM & MOM_VECINV) and call S/R MOM_CALC_VISC + only when using variable horiz. viscosity (useVariableVisc=T); + - simplify mom_vecinv.F (only 1 block for momViscosity). +o pkg/mom_common: variable horiz. viscosity in NH code + - move horiz-viscosity 3-D arrays viscAh_W & viscA4_W (used in calc_gw.F) + from model/inc/NH_VARS.h to pkg/mom_common/MOM_VISC.h + - store in common block (in MOM_VISC.h) logical parameters: useHarmonicVisc, + useBiharmonicVisc & useVariableVisc. + - fix old bug (already in 1rst implementation of variable horizontal viscosity + in calc_gw.F, version 1.27, 2006/06/20): + horizontal bi-harmonic viscosity in gW was missing when wiscA4W=0 + even if variable bi-harmonic viscosity (e.g., viscA4Grid) was non-zero. + - change background horiz viscosity in vertical momentum (gW) to always be + viscAhW,viscA4W (instead of switching to viscAhD, viscA4D when using + any type of variable viscosity, e.g., viscAh/4Grid) + - change default value of viscAhW,viscA4W to viscAhD,viscA4D (instead of + viscAh,viscA4) which default to viscAh,viscA4 : this avoid missing + horiz. viscosity in W eq when just setting viscAhD/Z,viscA4D/Z (since + default viscAh,viscA4 is zero). +o model/src: + - extend length of drC & recip_drC from Nr to Nr+1. +o Divided Adjoint (DIVA) with MPI: + - genmake2: do not create local ./mpi_headers dir when no file to copy into. + - remove option ALLOW_DIVIDED_ADJOINT_MPI (replaced by ALLOW_DIVIDED_ADJOINT + and ALLOW_USE_MPI) from model/src code and from ECCO_CPPOPTIONS.h. + +checkpoint64k (2013/07/23) +o tools/OAD_support: + - avoid the need for file "oadtempflile" in insertTemplateDir.bash + and remove it from genmake2. +o tools/build_options: + - remove MPI_HEADER_FILES_INC setting (no longer used) and comment out + MPI_HEADER_FILES list (let genmake2 find the missing headers) +o verification/halfpipe_streamice: + - change to driving stress formulation; update both Fwd & AD output. +o tools/genmake2 and Divided Adjoint with MPI: + - remove default setting of MPIINCLUDEDIR from $MPIHOME (always been broken); + - only create local ./mpi_headers dir if using MPI and MPIINCLUDEDIR is set; + - add new variable (LOCAL_MPI_HEADERS, replace MPI_HEADER_FILES_INC) in + Makefile, which is set by genmake2, and check which mpi-headers are needed. + Note: need to update adjoint_options/adjoint_diva_mpi accordingly. + - update target "CLEAN" (to also remove ./mpi_headers/ dir). +o testreport: + - changes for Divided Adjoint Run: when file "run_ADM_DIVA" is present, + perform additional runs (number of additional runs is "add_DIVA_runs" taken + from file run_ADM_DIVA ; should match "nchklev_3" in file "code_ad/tamc.h"). +o verification/lab_sea (AD test): + - change to test Divided Adjoint Run in this set-up (both standard AD test + and 2 secondary AD test: noseaice + noseaicedyn); drop the evp AD test. +o pkg atm_ & ocn_compon_interf: + - remove cpl_earlyExpImpCall=F option (was mainly a hack added for the old + staggerTimeStep implementation) and disable (retired) corresponding + parameter. +o pkg/offline: + - comment out unused field-array: ICEM, ConvectCount, hflx[01], icem[01]; + - add S/R OFFLINE_CHECK to print offline parameter summary + and stop if loading a file that is not used; + - enable to calculate GMRedi tensor, KPP mixing and convective index + when the corresponding file(s) is/are not specified (but pkg is used). + - comment out resetting of monitorFreq (to endTime) since it still useful + for monitoring CFL numbers and checking offline fields. +o OpenAD: + - do not define OAD_DEBUG (in tools/adjoint_options/adjoint_oad) to get + smaller standard output) and generate new (smaller) output (on baudelaire + using gfortran with -ieee) for exp. tutorial_tracer_adjsens +o model/src: + - call EXTERNAL_FORCING_SURF outside bi,bj loops (in do_oceanic_phys.F), + (where external_forcing_adjust was called) and change list of arguments; + - merge external_forcing_adjust.F into external_forcing_surf.F + - put surface relaxation of SST & SSS in new S/R (forcing_surf_relax.F) + which also contains balancing surface relaxation (ALLOW_BALANCE_RELAX, + previously in file "balance_relax.F", now removed); remove corresponding + arrays in FFIELDS.h. + - EXTERNAL_FORCING_SURF is now called when using pkg/offline; implement + a hack for getting back surfaceForcingS (loaded from file) when using + ALLOW_OLD_VIRTUALFLUX (pkg/dic, in DIC_OPTIONS.h) in offline set-up. + - update output of AD test experiment global_ocean.cs32x15.seaice_dynmix + (was getting only 11 digits for AD-grad, due to different truncation + error in AD code). + +checkpoint64j (2013/07/04) +o pkg/offline: + - to enable the use of other features with pkg offline (e.g., obcs), + also call S/R DO_OCEANIC_PHYS in offline run - more modifs to come. + - comment out "allowFreezing" (now active) in 2 offline testing exp. o pkg/exf ALLOW_RUNOFTEMP - fix sign and make compatible with temp_EvPrRn .NE. UNSET_RL - - added RUNOFTEMP test to verification/global_ocean.cs32x15/input.seaice + - ahdded RUNOFTEMP test to verification/global_ocean.cs32x15/input.seaice - changes verification/global_ocean.cs32x15/results/output.seaice.txt o pkg/atm_phys: - allows to add some damping on uVel,vVel (strength only function of k).