What is wrong with my objective function, called by bintprog?

Hello!

I am solving an optimization problem with 8 variables (x1,...,x8) and each variable has specific benefit and cost. I have to maximize the benefit with the condition that the cost do not exceed the certain value. There is also a condition that if x2 is 1 then x8 also has to be 1, otherwise they are both 0.

My objective function is:

function f = benefit(x)

if ~xor(x(2),x(8)) == 1
f = [-950 -780 -440 -215 -630 -490 -560 -600];
else
  f = [100 100 100 100 100 100 100 100]; %extremely non-beneficial
end

This produces the error:

Error using benefit (line 3). Not enough input arguments.

I have no idea what is wrong.

Regards, Anze

Answer 1

[Matt J]

Error using benefit (line 3). Not enough input arguments.

You haven't shown your call to BINTPROG, so we cannot diagnose the error.

Independently of that, though, bintprog does not maximize arbitrary objective functions. The objective function has to be a linear function dot(f,x). In other words, f has to be a vector independent of x.

[Matt J]

Anze Commented:

My call to bintprog is

bintprog(benefit,[400 350 200 100 300 250 300 350],1000)

the second argument is the vector of costs and the third argument is the cost limit.

[Matt J]

Again, the first argument, "benefit", isn't allowed to be the name of a function. It has to be a constant vector of weights.

[Anze]

So it is not possible to take the second condition into account with bintprog?

[Matt J]

I think the following might be what you want. The main point is that the "f" input must be a fixed vector.

 f=[-950 -780 -440 -215 -630 -490 -560 -600];

 A=[400 350 200 100 300 250 300 350];
 b=1000;

 Aeq=[0 1 0 0 0 0 0 -1]; %equivalent to xor(x(2),x(8)) == 0
 beq=0;

 bintprog(f,A,b,Aeq,beq);

Answer 2

[Alan Weiss]

Please take a look at the documentation and some examples for bintprog. bintprog takes a single vector as an objective function, not a program.

Alan Weiss

MATLAB mathematical toolbox documentation