How to formulate a fitness function cosists of sum of different variables?

In this fitness function I couldn't run the code. It don't know the appropriate way to formulate it.

function  y= fitness(b,a,c,d,N,S)
y=2*sum (a1+b1+c1-d1+(N1/S)) + sum (a2+b2+c2-d2+(N2/S));

a1= [0 23];
a2= [0 38];
c1= [0 4];
c2= [0 4];
d1= [5 4];
d2= [4 6];
N1= [4 5];
N2= [6 3];

fun=@(b)fitness(b,[a1,a2],[c1,c2],[d1,d2],[N1,N2],S);

lb= [0 0]; 
ub= [10 10]
b= fmincon(fun,b0)

end

[michio]

I do not follow that you have tried. What is the exact step that we need to follow to reproduce your issues? It would be helpful if you could clearly mentioned which part of your code corresponds to the function and script at least.

When you "couldn't run the code" there should be some error message or some indication of something is going wrong. Any additional information would also be helpful.

[Sherif Shokry]

Thanks for the quick response. This is the error I got: "Undefined function or variable 'b'." Although I edited the code a bit, I got the same error.

function  y= fitness(b,a,c,d,N,S)
 b= [1,2]
a1= [0,23];
a2= [0,38];
c1= [0,4];
c2= [0,4];
d1= [5,4];
d2= [4,6];
N1= [4,5];
N2= [6,3];
y=2*sum (a1+b1+c1-d1+(N1/S)) + sum (a2+b2+c2-d2+(N2/S));
fun=@(b)fitness(b,[a1,a2],[c1,c2],[d1,d2],[N1,N2],S);
Aeq=[];
beq=[];
lb= [0,0]; 
ub= [10,10]
b= fmincon(fun,b0,,Aeq,beq,lb,ub)
end

[michio]

Thanks for the comments. What is the reproduction step? Do you put all of the code into the command line? Do you define a function?

[Sherif Shokry]

I type the code in the (editor) window and then try to run. I got this error in the command window.

Answer 1

[michio]

Hmm, I'm still struggling to understand your situation. This is pure speculation, but I'd try the followings.

First, define a objective function. For example, save the following as fitness.m

function  y= fitness(b,a,c,d,N,S)
y=2*sum (a+b+c-d+(N/S)) + sum(a+b+c-d+(N/S));
end

Then copy & paste the following as a script, say as samplefmincon.m.

a1= [0 23];
a2= [0 38];
c1= [0 4];
c2= [0 4];
d1= [5 4];
d2= [4 6];
N1= [4 5];
N2= [6 3];
fun=@(b)fitness(b,[a1,a2],[c1,c2],[d1,d2],[N1,N2],S);
lb= [0 0]; 
ub= [10 10]
b= fmincon(fun,b0)

Then run the samplefmincon.m script. In this case I get the error saying ""Undefined function or variable 'b0'". This is because I didn't define b0.

If you are not sure the difference between script and function, then please see https://jp.mathworks.com/help/matlab/learn_matlab/scripts-and-functions.html

Also there are several working example that uses fmincon. So it should be a good starting point. https://jp.mathworks.com/help/optim/ug/fmincon.html

[Sherif Shokry]

When I run it after giving an initial value for b0= [1 2], I got this error: "Error in samplefmincon (line 13) b= fmincon(fun,b0)

Caused by: Failure in initial objective function evaluation. FMINCON cannot continue."

[michio]

On the previous comment, I just wanted to give you an example that you'd better follow in posting questions, be specific so that others can reproduce your issues. I didn't meant it as a solution to your issues, because I still don't understand your problem.

Now, you, or we, need to review your objective function/script and check if it does what you want them to do. Obvious problem is this one.

 fun=@(b)fitness(b,[a1,a2],[c1,c2],[d1,d2],[N1,N2],S);

In the script, a variable S is not defined. Also 4 input values are essentially 1x4 array, ie. [a1,a2], [c1,c2], [d1,d2],[N1,N2]), but not sure if b is 1x4 array or not. Whatever the size of b would be, you need to be very careful not to have dimension mismatch on the fitness.

I believe this is all I can say based on my understanding of your problem statement. Please refer to the documentation page of fmincon https://jp.mathworks.com/help/optim/ug/fmincon.html very carefully. Again you can find some "working" examples on the page.

[Sherif Shokry]

Since the fmincon Fun is used to find a minimizer x of the function described

x = fmincon(fun,x0,A,b) starts at x0

So if I have to optimization variables, how can I modify fmincon fun to minimize both variables