1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
|
//
// Scilab ( http://www.scilab.org/ ) - This file is part of Scilab
// Copyright (C) ????-2008 - INRIA
// Copyright (C) 2010 - DIGITEO - Yann COLLETTE
//
// This file is distributed under the same license as the Scilab package.
//
function [co,u,g,itv,dtv]=icsu(u,simu,nap,imp)
// Computation of the optimal control with weighting of the initial control.
// A scaling brings all the componant of the initial control to 1.
// The componant initially null will stay null.
// Only works if the lower bound is positive.
// Syntax
// [co,u,g,itv,dtv] = icsu(u,simu,nap,imp)
// input variables :
// u(nu) : initial parameters
// simu : string which contains the name of the sub program
// which describes the problem (second member, criterion
// and initial state)
// nap : maximum number of call to the simulator
// imp : debug value during optimization
// output variables :
// co : final cost
// u(nu) : final parameters
// g(nu) : final gradient
// itv(nitv) : work area (fortran integers)
// dtv(ndtv) : work area (fortran double precision)
// Use the macros icot and icob to extract the total state
// or the state at measure time instants of dtv.
df0 = 1;
if min(binf) <=0 then
error("appel de icsu avec binf non strictement positif");
end
for i=1:nu
u(1,i) = max( [binf(1,i),min([u(1,i),bsup(1,i)])] );
end
ech = u;
binf = binf./u;
bsup = bsup./u;
u = ones(1,nu);
cof = ones(1,ntob*nob);
[co,u,g,itv,dtv] = icsegen(u,simu,nap,imp);
u = ech.*u;
endfunction
|
