How to use an optimization expression if in a function?

Hi everyone!

I'm new to Matlab and i'm trying to write a function that is used in a problem based optimization.

This is the code that i wrote:

function [array] = getmax (matrix)
    row = size(matrix,1);
    column = size(matrix,2);
    array = optimexpr;
    for i=1:1:row
        maxr = 0;
        for j=1:1:column
           if matrix(i,j) >= maxr
               maxr = matrix(i,j)
           end
        end
        array(i,1) = maxr;
    end
end

Matrix is an optimization expression (es. 4 row and 5 column)

Array is the optimization expression that i want to get as the result of the function. (in this example it should be 4 row 1 column with tha max value of each matrix's row).

If i run this code i get "Conversion to logical from optim.problemdef.OptimizationInequality is not possible" in the if line.

I read about the big-M method, but i can't apply that method to my problem. (There's a lot about the big-M method for constrain but not a lot for function).

Can someone help me? I think it's a very simple function but i'm a newbie :(

Thanks

Answer 1

[Matt J]

You cannot express such an operation through the problem-based framework. If you are trying to solve a minimax problem, then you should use fminimax()

https://www.mathworks.com/help/optim/ug/fminimax.html

[Diego Cecchinato]

Thank you Matt J.

You made me save lot of time.

I try to explain my problem, maybe if you want to read you can give me some good advices.

I'm trying to solve a Warehouse allocation problem.

I have n components, n locations and m kits (n>m).

A kit is a group of code.

The binary array "Kitconfig" (m*n) tells which components is part of the kit (see the image to undestand).

I also have two other array:

• Distance (1*n) with the distance of every location to a specific point
• Qnt (1*m) with the quantity request every year for each kit

In reality workers will receive the list of components that make up the kit; then they will reach the location associated with the components and they will pick them up.

The objective function i want to minimize is the total distance travelled in a year.

The decision variable x is a binary array (n*n) with 1 if the cod is associated with the location. (something like this:)

(as you can see a location is associated with only one component, and every component is associated with only one location).

Total Distance = qnt * Dkit (for every kit)

Dkit = // distance of most far component that is part of the kit// (<- this is what i was trying to get with the function i describe previously).

Dcod = x*Distance (for every components)

I hope you understand my problem, i tried to explain it as simple as possible.

Do you think i can solve this with Matlab? It's a problem based optimization or a minmax problem? Every advice will be appreciate :)

Thanks in advance for your help. (I will certainly mention you in the thanks of my degree thesis).