Genetic Algorithm Definition

The genetic algorithm is top-most entity in any Genetic Algorithm implementation. There can be only one genetic algorithm construct in any DARWIN program and this construct should contain at least one population as a member. It is responsible for creating its population and evolving it until a termination condition is reached. The genetic algorithm also specifies the replacement strategy. The genetic algorithm has terminator, replacement, evolver, initializer and printer as its moderators.

The syntax of the genetic algorithm is as follows :

alg_typedef	::=	"algorithm" id
		"{" memb_decls "}"
		alg_moderators ";"
alg_moderator	::=	initializer ":" prototype
	$\vert$	printer ":" prototype
	$\vert$	evolver ":" prototype
	$\vert$	terminator ":" prototype
	$\vert$	replacement ":" prototype

For example, an algorithm implementing Simple GA can be declared by just specifying a population :

algorithm GA
{
   Pop population;
};

The genetic algorithm initializer

is responsible for initializing its population and parameters. The default is to call the population initializer to initializer its population.

The genetic algorithm terminator

is a boolean function deciding when a termination condition is reached. The default stops if number of generations is 100.

The genetic algorithm replacement

moderator is used to decide if which of the chromosomes in the parent and offspring populations should survive. The generated replacement moderator is dummy, so the offspring replaces its parent population.

The genetic algorithm evolver moderator

is the hart of any DARWIN program. This moderator is responsible for first creating and then evolving a population until the termination moderator decides stopping this process. The generated genetic algorithm evolver implements Simple GA.

The genetic algorithm printer

provides formatted output of the genetic algorithm construct. By default, all the algorithm members, including the population, are printed. This moderator is called when the termination condition of the genetic algorithm is reached.