Function evaluations in GA

What is function evaluation in GA? How to relate the values of function evaluations with global minimum? If function evaluation is 170 and global minimum is -0.5, what does it mean? Is there a formula for function evaluation? How to calculate the numerical values of function evaluations in GA by using either command line or GUI?

Answer 1

[Walter Roberson]

If the function evaluation is 170 and the global minimum is -0.5, then all it means is that that location that was evaluated was not the global minimum.

The actual value of the global minimum and the actual value of the function evaluation do not matter (provided you work within the precision limits of MATLAB); all that matters to ga() is whether one value is less than another.

There is no formula for function evaluation. Just have the function return some measure of how good the location is, with lower values to be interpreted as better than higher values.

The function handle you pass to ga() is evaluated with a vector of parameter values. You can call the function yourself with a vector of parameter values.

For example,

ga(@std, 5)

would call std() passing in a vector of 5 values. You could call std yourself:

std([8 .2 -4 7 51])

[Dua]

Basically I understand your explanation. I need to calculate the average number of function evaluations for the performance evaluation criteria. For instance, when comparing the objective function values obtained through different crossover operators.

[Walter Roberson]

Ah, number of function evaluations was not clear in your question. I thought you were asking about the value returned by the function.

I do not know how ga() decides the number of function evaluations to use.