Error using the fmincon function

This is my code:

nonlcon = @(x) deal(g_fun_vec(x), []);
options = optimoptions('fmincon','Algorithm','sqp','Display','iter');
[x_sol,~,exitflag, output] = fmincon(@(x) 0, x0, [], [], [], [], lb, ub, nonlcon, options);

And I get this error message:

The strange things is that, as you can see on the screenshoot of the Command Window, the constraint fuction is defined correctly at x0

[Walter Roberson]

Although nonlcon is not specifically documented as having to return real values,

https://www.mathworks.com/help/optim/ug/fmincon.html#busog7r-nonlcon

Nonlinear constraints, specified as a function handle or function name. nonlcon is a function that accepts a vector or array x and returns two arrays, c(x) and ceq(x).

• c(x) is the array of nonlinear inequality constraints at x. fmincon attempts to satisfyc(x) <= 0 for all entries of c.
• ceq(x) is the array of nonlinear equality constraints at x. fmincon attempts to satisfyceq(x) = 0 for all entries of ceq.

If c(x) were permitted to be complex, then c(x) <= 0 is not well defined. If c(x) were permitted to be complex, then is the implied definition real(c(x))<=0 or would it be real(c(x))<=0 & imag(c(x))<=0 -- leaving open the possibility that c(x) were (for example) -1-1i as satisfying constraints ?

Answer 1

[Matt J]

No, the constraint and objective functions have to return real values to be considered "defined". Possibly you have log() or sqrt() operations in your constraints that produce complex values when their arguments go negative.

[Giacomo]

Yes, thanks I didn't put the expression in abs()