|
|
Re: Lower bound of a non-convex program through the dual problem in Matlab
Posted:
Jan 22, 2013 12:33 PM
|
|
Dear Bruno,
Thanks for your reply. I tried the 'interior-point-method' to solve the following (trial) optimization problem:
min_x exp(-x) * cos(2*pi*x)
s.t. x >= 0
I used the following options:
options = optimset('Display','off','Algorithm','interior-point','MaxFunEvals', 5000,'TolFun', 1e-6,'TolX',1e-6,'TolCon',1e-6, 'MaxIter', 500,'Display','off');
[x,netval,exitflag] = fmincon(@NonConvexFunction,x0,[],[],[],[],0,[]); %% 0 is the %% lower bound of x
Unfortunately, my solution is getting stuck at a local minimum. Please let me know if I can get the lower bounds of this problem through the interior point method.
Please also inform me if you know of any other software that provides bounds for non-convex problems.
Thanks,
Nazmul
"Bruno Luong" <b.luong@fogale.findmycountry> wrote in message <kddrp8$p10$1@newscl01ah.mathworks.com>...
> Use fmincon with interior-point method. I believe it minimizes the dual-gap.
>
> Bruno
|
|
