The Individual Comparators Package
This package specifies the comparators used by the Search Engines to determine which individual is fitter.
All selection and search is comparison based. Since we support multi-dimensional searches, we cannot simple
reduce the fitness of an element to one single number and compare them without lost of information. When
performing a multi-dimensional search, the fitness is always a vector. To compare vectors, one might use
arbitrary methods, some of them are specified here:
- Staggared Comparator
- The fitness functions are as important as their index is small. This means, that this comparator first
compares the fitness functions at index 0. If one of them is bigger, we're done. Otherwise we need to
go on at the second fitness function and so on.
- Majority Comparator
- The individual which dominates the other in the most fitness functions wins.
- Weightes Majority Comparator
- The same as
Majority Comparator, but a weight value is assigned to each fitness function.
- Sum Comparator
- This comparator adds up all fitness functions and compares the results. This roughly resembles to
a transformation from a multi-dimensional search to a single-dimensional one.
- Weighted Sum Comparator
- The same as
Sum Comparator, but a weight value is assigned to each fitness function.