/[MITgcm]/MITgcm/doc/tag-index

Diff of /MITgcm/doc/tag-index

Parent Directory | Revision Log | View Revision Graph Revision Graph | View Patch Patch

-revision 1.53.4.33 by heimbach,
Sat Mar  8 00:46:48 2003 UTC
+revision 1.71 by jmc,
Wed May  1 01:35:30 2002 UTC
 Line 3 
 $Header$
      Notes on tags used in MITgcmUV
      ==============================
- ecco_c50_e28
+ checkpoint45a_post
- o major merge:
+ o compute Rho over the whole domain (extend iMin,jMin indices)
-   ecco-branch merged with checkpoint50
+ o specific exchange for X-slice domain (case Ny=1)
-   gmredi now available, grdchk, DIVA up-to-date for MPI
+ o argument futureIter of S/R obcs_calc missing in exp4 & internal_wave /code
-   However, two outstanding problems:
+ o solve_for_pressure : set cg2d_x to zero where OBCS are applied
-   - kpp adjoint is broken, needs fixing!
-   - a few minor modif's in handling of CPP option
+ checkpoint44h_post = checkpoint45
-     ALLOW_TANGENTLINEAR_RUN in conjunction with
+ o Fix to restore differentiability:
-     ALLOW_ECCO_OPTIMIZATION are necessary to enable
+   place at which packages_init_variables is called in
-     TLM runs with ECCO setup
+   initialise_varia is crucial for reverse control flow.
+ o modifications in GMRedi to enable stable adjoint
- ecco_c44_e27
+   (corresponding modif's in thermodynamics, dynamics)
- o exf:
+ o added missing hook for sponge layer code in external_forcing
-   - Enable initialisation of forcing fields to constant
+ o modified test cost functions
-     (runtime) values.
+ o storing in gad_advection (plus removal of write(0,*)!
-   - in exf_getffields.F
+ o missing headers TR1.h added in convective_adjustment routines
-     Reduce i-/j-loop to interior domain, discarding overlaps.
+ o errorMessageUnit set non-equal zero in eeboot
-     That also fixes wrong TAF-key computations for key_1, key_2
+   (conflict on some platforms, e.g. IBM SP3)
-     with bulf formulae.
+ o modified carbon verif.
-   - exf_init.F modify #ifdef for exf_init_evap
-   - exf_getffieldrec.F, ctrl_getrec.F
+ checkpoint44h_pre
-     The following INT-usages are not safe:
+ o vertical grid option: allow to put the Interface at the middle between
-       fldsecs  = int(fldsecs/fldperiod)*fldperiod
+cell-centers ; replace delZ in pkg/kpp by drF.
-       fldcount = int(fldsecs/fldperiod) + 1
+ o GM Advective form: Tracers are advected using the residual transport (=
-     and were modified.
+   Euler+GM-bolus); set param GM_AdvSeparate=T to return to previous form.
-   - added repeatPeriod (from MAIN branch)
- o grdchk: updated and merged
+ checkpoint44g_post
-   - has both ADM and TLM checks
+ o fix surface correction term in multiDimAdvection (affects plume_on_slope
-   - works for single- and multi-proc.
+   results) ; ifdef missing in thermodynamics.F
-   - output cleaned
-   - worked successfully for parallel DIVA
+ checkpoint44f_pre,checkpoint44f_post
-   Needed one update in EEPARAMS.h and updates in ctrl package
+ o added PTRACERS package
-   More modif's may be necessary for ECCO code to support TLM.
+   This allows an arbitrary number of passive tracers to be integrated
- o autodiff: checkpoint_lev...
+   forward simultaneously with the dynamicaly model.
-   - updated storing for bulk formulae (some arrays were missing)
+   + Implemented so far:
-   - added state for seaice package
+     - basic forward algorithm (time-stepping, advection, diffusion, convection)
-   - synchronized with main branch
+     - I/O and checkpointing
- o DIVA:
+     - GM/Redi  *but*  using the GM/Redi coefficient of Salt
-   Modif.s in
+   + Not implemented so far:
-   - cost_init.F
+     - KPP
-   - cost_averagesfields.F
+     - OBCS
-   - ecco_the_main_loop.F
+   + No specific example supplied (yet) but global_ocean.90x40x15 has the
-   - ecco_the_model_main.F
+     necessary data.ptracer file. Simply use -enable=ptracers and uncomment
-   and added cost_final_restore.F cost_final_store.F
+     line in data.pkg.
-   o With these changes, it is possible to fully separate
+   + This package is disabled by default since it increases storage.
-     the first forward run to compute the cost function (mdthe_main_loop)
-     and adjoint runs over individual DIVA intervals (adthe_main_loop)
-     into separate model executions.
+ o testscript: added -cleanup option
-     This enables better tuning of DIVA to given queue sizes.
+ o Shap_filt: a) new shap_filt S/R to use no-slip BC with S2 filter
-   o Loss of cost function value in consecutive DIVA runs is
+              b) enable to filter 2D fields.
-     fixed by storing to 'costfinal' file.
-   o Appropriate call of ctrl_unpack/_pack,grdchk_main at
+ checkpoint44e_pre,post
-     beginning/end of adjoint integration is accounted for.
+ o re-wrote convective adjustment to use pre-calculated weights
-   The current changes apply to TAF-1.4.26.
+   Affects: model/src/convective_adjustment.F, convective_adjustment_ini.F
-   TAF has one remaining bug, requiring one manual change
+   Adds:    model/src/convective_weights.F, convectively_mixtracer.F
-   (see MITgcm manual).
+   Original calls to convect() still in place.
- o bug fix plus modif for ubar, vbar in cost_averagesfields
- o cal: synchronized between MAIN and ecco-branch
+ chkpt44d_pre,post
+ o added missing EXCLUDE_MONITOR flags
- icebear4
+ o changed "e" to "_d" in gmredi_slope_limit, gmredi_slope_psi
- o Added cost_ice for sea-ice volume sensitivity experiment.
+   (incompatible typ in MIN/MAX expressions caused problems
- o ecco_the_model_main.F: SEAICE_WRITE_PICKUP must precede WRITE_CHECKPOINT
+   on IBM SP3)
- o Reduced derivative singularities in pkg/seaice.
+ o in genmake added variable MAKEDEPEND
- o Updated verification/natl_box_adjoint
+   plus resetting for case SunOS
+ o added timer_stats.c routine for IBM SP3
- icebear3
+ o removed variables in dynamics
- o Removed spurious iikey and idkey
+ !!! I forgot to create tag chkpt44d_pre !!!
- o pkg/exf
+ !!! But chkpt44d_pre = chkpt44c_post    !!!
-   - bug fix in treatment of downward radiation
- o pkg/seaice
+ o real fresh water flux implemented with non-linear free-surface.
-   - added initial sea-ice thickness parameter SEAICE_initialHEFF
- o updated verification/natl_box_adjoint
+ chkpt44c_pre,post
+ o few fix (mask in shap_s2, EmPmR in external_field_load,
- icebear2
+   USE_NATURAL_BCS in solve_for_P);
- o Added missing /* */ to CPP comments in pkg/seaice, pkg/exf,
+ o add arguments myIter & myTime to S/R obcs_calc & solve_for_P
-   kpp_transport_t.F,and ecco_the_main_loop.F
- o adjoint-friendly modifications to pkg/seaice
+ checkpoint44b_pre/_post
- o added verification/lab_sea as per release1_p12_pre
+ o merge of relevant stuff from the ecco-branch:
- ecco_c44_e26
- o Merging pkg/exf and pkg/seaice from cvs tag ecco_ice2
- o exf: Modifications to correctly handle evap field for adjoint
- o grdchk:
-   Added code to perform gradient checks for bulk formulae/atmos. state
-   (atemp, aqh, uwind, vwind).
-   NOTE:
-   This package is out of synch with c48 package.
-   The latter also has the tangent linear gradient checks.
-   Need to be merged.
- o verification/natl_box_adjoint:
-   * added new setup: code_bulk/, input_bulk, results_bulk
-     (Experiment 2, described in README)
-     uses bulk formulae and atmos. state as control instead of
-     air-sea fluxes.
-     Performs gradient checks for first 4 elements of atemp control.
-   * changed standard verification to perform
-     gradient checks on first 4 elements on hflux control.
-   * gmredi switched off (will come after merge with c48).
- ecco_c44_e25
- o Merging some GM changes from main trunk (c47) to ecco-branch.
- o Updating verif. natl_box_adjoint with GMRedi.
- o Adding setup with explicit differentiaton of pressure solver.
- o Bug fix: missing OBCS_OPTIONS.h for sponge layer code.
- o (Re-?) Commiting old bug fixes in kpp to ecco-branch
-   (kpp_calc.F, kpp_routines.F)
- o exf:
-   updated external forcing package
-   - cleaned old exfa stuff
-   - bug fixes
-     (missing OBCS_OPTIONS.h in two routines)
-   - enable easy to use "no forcing".
-   - added exf I/O for atmospheric loading
-   - added exf I/O for runoff data
-   - transfered scaling between exf <-> MITgcm to exf namelist
- ecco_c44_e24
- o modifications to gradient check package (Martin Losch)
-   - enable centered differences
-   - modified format of standard output
- o bug fix for nwettile contents
-   so far assumed global sum  in first entry (1,1,k),
-   needed for global control vector
-   -> introduced new variable nwet[c/w/s]global(nr)
- o shifts of calls to cal/exf package to be accessible
-   in MITgcm stand-alone
- o shifted packages_init_variables before convective_adjustment_ini
-   (corresponding update of TAF tape initialisation).
- o removed auxiliary field in impldiff
-   (caused numerical differences for GMRedi cases)
- o updated obcs control part (G. Gebbie)
- o removed some f90 comments ("!")
- o Bug fixes and updates in cost package
-  * cost_final: correct multiplier mult_tauv, mult_sflux
-  * cost_hyd: added call cost_ssh
-  * cost_init_barfiles.F: replaced arrays by tmpfld's
-  * cost_readtopexmean.F: ersmask -> tpmeanmask
-  * cost_weights.F: corrected wtheta[2], wsalt[2]
- o cleaned exchanges in cg2d.F
- ecco_c44_e23
- merged patches between release1_p1 and release1_p2:
- o fixed array boundary problem for adjoint
-   (related to partial recomputation within subroutine
-   related to dynamics -> calc_viscosity -> mom_...
- o fixed recomputation in thermodynamics
- o prepare (but not complete) recomputation problem in impldiff
- o fixed problem for adjoint of global_max
- o switched off sponge layer code in external_forcing
- Bug fix in ctrl_mask_set_yz (spotted by G. Gebbie)
- ecco_c44_e22
- o Finally, the merges with the Scripps cost function terms:
-   Various additions, modifications, quality checks.
- o Plus: Included switch ALLOW_AUTODIFF_MONITOR to handle
-   monitoring of adjoint state.
- o Plus: removed comm_stats, ini_communication_patterns from diff. list
-   (no more need to "keep" MPI headers for adjoint generation)
- o Plus: exf changes:
-   o added run-off routine
-   o additional storing needed for ustress,vstess
-     because of state-dependent quality check
-   o added missing initialisations
- o Plus: genmake with MAKEDEPEND variable was still missing in ecco-branch
- NB: - verification/natl_box_adjoint/ 240 steps run
-       * looks ok under Linux (twain.lcs.mit.edu)
-       * looks ok on MIT SUN's (fjord.mit.edu)
-       * NaN in reverse mode on Scripps SUN's at iter=214
-       * NaN in reverse mode on IBM SP3 (BlueHorizon) at iter=204
-     - 1 year forward int. global 2x2 deg. produces similar cost functions
-       * all cost terms of new code are slightly smaller (0.5 to 3%).
- ecco_c44_e21
- o reversed order of ctrl/cost initialisation
- o gmredi_slope_psi.F: Inconsistent types in MIN/MAX expressions
- o modified to enable ctrl field swapping
- o bug fix for global sum of ncvarlength in ctrl_init
- o no more call to ctrl_unpack from ctrl_init
- o no more write of ctrl vector in ctrl_pack
-   however, init. of control vector from ctrl_unpack
- o various bug fixes & consistent nobcs handling in
-   ctrl_pack/unpack routines (by G. Gebbie)
- o optim package: Added obcs control part for lsopt I/O
-   (by G. Gebbie)
- o missing call to obcs_diags
- o moved calls in packages_init_fixed to packages_readparms
-   because some quantities are needed earlier.
- ecco_c44_e20
- o before major merges from the Scripps setup
-   here are two small updates:
-   - field swapping for surface flux controls enabled
-     (remains t.b.d. for atmos. state plus obcs)
-   - bug fix in active_file.F for xz-/yz- routines
-   - added missing xz-/yz- routines to active_file_ad.F
- ecco_c44_e19
- o merged necessary patches:
-   - AD-related changes for GMRedi
-   - fixes i KPP (delZ -> drF)
-   - hook to OBCS songe layer code in external_forcing
-   - errorMessageUnit non-zero in eeboot.F
-   - included GMREDI, EXACT_CONSERV to natl_box_adjoint verif.
- ecco_c44_e18
- o sponge layer code for obcs was missing
-   - added hook in external_forcing.F
-   - added storing for TAF
- o new routine cost_init_barfiles to initialise
-   adjoint fields adtbar, adsbar, adpsbar
-   (needed in conjunction with corrected ladinit for active I/O)
- o cleaned up and corrected order in packages_readparms, ecco_readparms
- o corrected and simplified weights for obcs controls
-   - weight fields simplified
-   - bug fix in ctrl_pack
-   - no more write of control vector in ctrl_pack
- o i/j index bug fixes in cost_ssh_mean
- ecco_c44_e17
- o bug fixes in cost package
-   * wrong identifier for active I/O of monthly mean psbar
-     (cost_averagesfields, cost_ssh) xx_psbar_dummy
-   * cleaned all identifier for active I/O of monthly fields
-   * wrong ladinit in various cost routines
- o merge with interval (chkpt44c_post - chkpt44d_post)
-   * added missing EXCLUDE_MONITOR flags
-   * changed "e" to "_d" in gmredi_slope_limit, gmredi_slope_psi
-     (incompatible typ in MIN/MAX expressions caused problems
-     on IBM SP3)
-   * in genmake added variable MAKEDEPEND
-     plus resetting for case SunOS
-     added timer_stats.c routine for IBM SP3
-   * removed variables in dynamics
- ecco_c44_e16
- o new ECCO checkpoint, the first since ecco_c43_e15
-   which is branched from MITgcm.
-   All changes were merged in between (see ecco-branch-mod? tags).
- o last changes:
-   * assuming path for TAF and TAMC are known (e.g. in $path)
-   * genmake:
-     - added tools/xmakedepend to circumvene shortcoming of
-       SUN's makedepend (size too small)
-     - invoking ../tools/xmakedepend instead of makedepend
-     - changed compiler flags for case SunOS
- ecco-branch-mod5
- o mod's to reproduce open boundary run
- o enable field swapping for OB's and climatology
- o update store directives for OB control
- o added option ALLOW_OBCS_PRESCRIBE
- o call of getobcs now out of obcs_calc (param. list modified)
- ecco-branch-mod4
- o this is the second part of changes necessary to reconcile
-   the natl_box_adjoint verification (with and without grdchk).
-   >>>!!! CHANGES BETWEEN THE TAGS ecco-branch-mod3 !!!<<<
-   >>>!!!                      AND ecco-branch-mod4 !!!<<<
-   >>>!!! SHOULD BE MERGED INTO THE MAIN TRUNK      !!!<<<
    - genmake: removed $S64 overwrite for case SunOS
    - pkg/exf: update and corrections for field swapping and obcs
    - pkg/ecco: parameter lists for the_model_main, the_main_loop
-Line 269 
 o this is the second part of changes nec
+Line 88 
 o this is the second part of changes nec
    - pkg/autodiff: added flow directives for obcs, mdsio_gl_slice
                    updated checkpointing_lev... lists for obcs
    - model/src: minor changes in forward_step, plot_field
+                added directive for divided adjoint in the_main_loop
    - pkg/mdsio: added mdsio_gl_slice
- ecco-branch-mod3
+ o check parameters & config (chkpt44a_pre,post)
- o this is the first part of changes necessary to reconcile
+ o OBC and NonLin_FrSurf.
-   natl_box_adjoint verification (with and without grdchk).
-   This set of changes is relevant for ECCO only.
-   The next set/tag of changes (ecco-branch-mod4)
-   should be merged into the main trunk.
-   - update some options
-   - parameter lists of ecco_the_model_main and ecco_the_main_loop
-     differed from those of the MITgcm counterparts.
- ecco-branch-mod2
- o update ecco-branch-mod1 to checkpoint44
- o exclude from updates:
-   - verification/
-   - adjoint/
-   - pkg/cost
-   - pkg/ctrl
-   - pkg/ecco
-   - pkg/obcs
-   - model/src/the_model_main.F
-   - model/src/the_main_loop.F
  checkpoint44
- o apply Shapiro filter on T&S before calc_PhiHyd (stagger time stepping)
  o fix bug in mom_vi_del2uv
+ o select when filters are applied ; add options to zonal_filter (data.zonfilt)
+ o gmredi: fix Pb in the adiabatic form ; add options (.e.g. Bolus advection)
+ o update AIM experiments (NCEP input files)
+ o improve and extend diagnostics (Monitor, TimeAve with NonLin-FrSurf)
  o added some stuff for AD
- ecco-branch-mod1
- o based on c43
- o replaced packages cost, ctrl, ecco, obcs
- o removed MITgcm verif.'s, added ECCO verif.'s
- o added lsopt, optim for line search optimization.
  checkpoint43
  o added two new advection tests
  o added global 4x4 degree ocean test

 Legend:



Removed from v.1.53.4.33
 


changed lines


 
Added in v.1.71
 Legend:



Removed from v.1.53.4.33
 


changed lines


 
Added in v.1.71
-Removed from v.1.53.4.33
+Added in v.1.71

	ViewVC Help
Powered by ViewVC 1.1.22