Non-negativity for unconstrained nonlinear multivariable function

I don't have any constraint for my nonlinear problem, so I am using either fminsearch or fminunc. If I want to get only positive solutions, am I still able to use fminsearch or fminunc? If so, how can I do that?

Thanks,

Answer 1

[Sean de Wolski]

You probably could, but it is not ideal. fmincon() will be your friend, though.

More per follow up Here are two ways to do it:

f = @(x)x(1)+x(2); %function with a minimum @ [-inf -inf]

A = [-1 0; 0 -1]; %-x(1);-x(2);
b = [0;0]; %less than zero, i.e. negative x<0
fmincon(f,[pi pi],A,b) %go!

%or
fmincon(f,[pi pi],[],[],[],[],zeros(1,2),inf(1,2)) %lower bound at zero, upper at infinity

Welcome to MATLAB Answers!

Answer 2

[Jalali]

Thanks! How do you define non-negativity for constraints in fmincon? Would you give me a simple example?

Answer 3

[Sargondjani]

if you want your inputs to be positive, just add them as a lower bound, see documention. For example if your x is a vector of three then just add: lb=zeros(3,1);

if the objective has to be positive you have to put in an nonlinear constraint, see also documention. Basically you repeat the objective and set Cineq = -function_objective (the constraint is Cineq<=0)