Dear sir, I have questions on fmincon.
I hope many experts might help me.
Would you mind giving me some answers? :)
Question 1.
It's fmincon cetering problem. In matlab R2013a optimization toolbox user' guide(http://www.mathworks.co.kr/kr/help/optim/ug/whenthesolverfails.html?searchHighlight=example%3A+centering+and+scaling.+consider+minimizing#br53c391)
There is a word 'centering'. In the problem sentence "fcc = @(w)fc([w(1)+1e6;w(2)1e6]); % centered ", I can not understand the meaning of centering. The function handle fcc seems to fix the error caused by fc in the text, then what is centered?
Question 2.
In the fmincon scaling, I think that the key point is making the weight of variables become equal. So, if we consider an exemplary case that x(1), x(2), x(3) are variables of input power at each second and y(1), y(2), y(3) are variables of charged battery state in each second, then the initial point of fmincon starts with the form [x(1) x(2) x(3) y(1) y(2) y(3)] with the constraints 0<=x<=1000 (unit of W) and 0<=y<=1 (no unit, full charge 1 and no charge 0). In that case, the user gudie recommend to use input [(x(1)/1000 x(2)/1000 x(3)/1000 y(1) y(2) y(3)] and rescale x values 1000 times in the calculation again. Is my though right?
Question 3.
This is question about inverse function estimation of fmincon. We assume f(m)=u and g(u)=n (then n=g(f(m))). Is fmincon find soutions well, when the object is minimizing inverse function of f (that is m) with input u with constrains g(u). The reason of I ask this is as follows. Should I make u the function of m and n the function of m and proceed with the bottom variable of m. (in fact m is fuel usage, u is generation power and g(f(m)) is the constraing of the battery). Or just give u as the input and finding m as a object function with inverse of f O.K.?
