Hello everyone, Can I use predefined scattered values for the variables used in Genetic algorithms instead of ranges?

Can I use predefined scattered values for the variables used in Genetic algorithms instead of ranges?

EX.

I have 2 variables to be obtimized using an objective function; I have 5 scaterred values for each variable and want to use them instead of using ranges(upper and lower bounds).

Answer 1

[John D'Errico]

Well, you can, IF that is correct for your problem.

For example, suppose you knew that you wanted variable x to take on 5 different values, and ONLY those 5 values. What is more, the values of interest are scattered. (Why? Hey, that is your problem to know.) For example, suppose x can take only the values [1, 2.5, 3.14, 5.653, 12.2].

Just tell GA that x is integer valued, taking on the values only [1 2 3 4 5]. Then inside your objective function, remap the integers [1:5] into [1, 2.5, 3.14, 5.653, 12.2]. This is just a table lookup.

Sorry, I don't have the Global optimization toolbox, so I don't have the syntax to give you an example.