fmincon start point

hi, I use fmincon function to resolve my problem, but the difficulty is : I have many variables in my objective function and it is very difficult to find a start point? if I use a random x0, the solution isn't optimal !

thinx

Answer 1

[Mouloud]

The length (x0) more than 1000 and I used the random start point but solutions are not interest ( there are bad locals optimums) (I tried a multi start point) how to choose an intelligent start point to find an interesting local optimum ? I know where I can put a zero in x0 to find the local optimum but I don't have an explication for that :-(

cordially

Answer 2

[Walter Roberson]

If you can, it would be better for you to switch to the Global Optimizer Toolbox and use one of the routines from that; see the function guide

[Mouloud]

hi;

I trying to use multistart point (rmsearch) but it is difficult to find a global minimum. In fact , I need an intelligent multistart.