[Scilab-users] Optimization question

[Stéphane Mottelet]

Le 13/08/2014 03:11, ebmb.sci a écrit :
>   Hello All,
>   please, could someone guide me on how to proceed to minimize the following
> polynomial using functions optim and numdiff. Since the parameters A and B
> have different constraints (i.e.: A = [0 1] and B = [1000 2000]), how must I
> sign the values and pass them for optim? I implemented a basic script to try
> to do it, but it always returns with wrong results.
>
> function f = myFc(x)
>     a = x(1);
>     b = x(2);
>     f = 0.000791 + 0.000027*a + -0.000037*b + 0.000180*a*b + -0.000126*a^2 +
> -0.000162*b^2;
> endfunction
>
> function [f, g, ind] = myFcCost(x, ind)
>          f = myFc(x);
>          g = numdiff(myFc, x);
> endfunction
>
> x0 = [0 1];
>
> [fopt, xopt] = optim(myFcCost, 'b', [0 1], [1000 2000], x0);
>
>   I am so thanks for any help in advance!
>
>   Best regards,
>   Eduardo.

Hello,

your bounds should be given this way (first vector for lower bounds and 
second for upper ones) :

[fopt, xopt] = optim(myFcCost, 'b', [0 1000], [1 2000], x0);


S.


>
>
>
>
> --
> View this message in context: http://mailinglists.scilab.org/Optimization-question-tp4031028.html
> Sent from the Scilab users - Mailing Lists Archives mailing list archive at Nabble.com.
> _______________________________________________
> users mailing list
> users at lists.scilab.org
> http://lists.scilab.org/mailman/listinfo/users