C $Header: /u/gcmpack/MITgcm/pkg/ctrl/optim.h,v 1.2 2007/10/09 00:00:01 jmc Exp $ C $Name: $ c ================================================================== c HEADER OPTIMIZATION c ================================================================== c c o Header for the large scale optimization. c c This header file contains those parameters that are monitoring c the state of the off-line optimization procedure. One c optimization cycle is considered to consist of a forward/adjoint c model run followed by a cost function minimization that then c yields a new guess for the control vector: c c optimcycle i: xx(i) --> fc(i) adxx(i) --> xx(i+1) c c c started: Christian Eckert eckert@mit.edu 30-Jun-1999 c c changed: Christian Eckert eckert@mit.edu 24-Feb-2000 c c c ================================================================== c HEADER OPTIMIZATION c ================================================================== c Optimization version: c ===================== c c lsoptversion - Version of the large scale optimization. character*(5) lsoptversion parameter ( lsoptversion = '2.1.0' ) character*(5) offlineversion parameter ( offlineversion = '0.1.1' ) c The off-line optimization part: c =============================== c c Starting an optimization experiment implies setting optimcycle to c zero. During cycle number I the model is called first, then the c offline optimization is run. The latter provides a new guess for c the vector of control variables for the next cycle I+1. c c optimcycle - cycle number of the off-line optimization. common /optiparm_i/ & optimcycle, & nvars, & numiter, & nfunc, & iprint, & nupdate integer optimcycle integer nvars integer numiter integer nfunc integer iprint integer nupdate common /optiparm_r/ & fmin, & epsf, & epsx, & epsg, & eps _RL epsf _RL epsx _RL fmin _RL epsg _RL eps common /optiparm_l/ & nondimcontrol logical nondimcontrol common /optiparm_c/ & copt character*( 3) copt c ================================================================== c END OF HEADER OPTIMIZATION c ==================================================================