Chapter 3 . Applications of Genetic Aigorithms
41
uninitiated. And finally, many of the combinatorial, parametrie and pattern
recognition applications of GAs require large datasets.
With these limitations in mind, it seems probable that new ground will be
broken by GAs that do things for wh ich GAs are uniquely suited. For example,
GAs that incorporate evolution into ecological systems can allow us to truly
explore the science of evolution, to understand ecosystems and open the doors to a
multitude of practical applications. These GAs are distinct from earlier process
models in that they can evolve both the equations governing them and the
equation rate parameters. Through the GA, the modeler produces adynamie
"movie" of the evolution of mechanisms producing the patterns. Questions can be
posed regarding whether and how a particular mechanism might have arisen to
produce a natural phenomenon. Typically few assumptions and constraints, other
than Darwinian selection, are built into a genetic algorithm model. As a resuIt,
GAs can substantiate ecological theory by recreating it based on fundamental rules
(i.e., dynamic induction).
There are some fascinating examples of explorations into evolutionary theory
and the emergence of ecological systems and properties. The fundamentals of
how to build these models came largely from computer game playing strategies
(e.g., Bouskila et al. 1998). Koza (1992) in his book on genetic programming
presents a thorough discussion of a genetic program designed to find the optimal
foraging strategies for an Anolis lizard and emergent properties in ants. In the
lizard example, the optimal foraging strategy as described by Roughgarden (1992)
was discovered based simply on the location of the prey, the abundance of the
prey, and the velocity of the lizard. Over time, the lizard can adapt his strategy if
the environmental conditions change. This provides both an example of controleost strategy and a demonstration that evolutionary programs can evolve the
theories governing compIex behavioral strategies using only simple governing
rules. In the ant emergent behavior program (Deneubourg et al. 1986; 1991), a set
of rules is used that governs the actions of individual ants. When these rules are
simultaneously executed, a complex pattern of behavior emerges that causes the
ants to collect and consolidate food pellets into a single pile.
Researchers have long hypothesized that collective behavior could arise from
colonial organisms operating according to very simple rules. Traditional
engineering style models would require tremendous computer power to simulate
individuals and explicitly code the possible sets of interactions. Genetic
algorithms, on the other hand, are especially weIl suited for this type of
applications and demonstrated that the emergence of complex behavior was
indeed possible from simple ruIes. Others, including K vasnicka and Pospichal
(1999) and Reuter and Breckling (1999) have similarly demonstrated emergent
behavior among artificial agents and organisms. GAs, such as Giske et al.'s (1998)
model of spatial dynamics in fish, that simulate coordinated movement of
organisms based on simple rules have greatly improved our understanding of how
groups of organisms travel, sometimes great distances, together in a coordinated
fashion.
The transition from population and community ecology to ecosystem science
occurs when organisms are placed in a physical and chemical environment.
Ecosystem science deals with the interaction between organisms and their non-
41
uninitiated. And finally, many of the combinatorial, parametrie and pattern
recognition applications of GAs require large datasets.
With these limitations in mind, it seems probable that new ground will be
broken by GAs that do things for wh ich GAs are uniquely suited. For example,
GAs that incorporate evolution into ecological systems can allow us to truly
explore the science of evolution, to understand ecosystems and open the doors to a
multitude of practical applications. These GAs are distinct from earlier process
models in that they can evolve both the equations governing them and the
equation rate parameters. Through the GA, the modeler produces adynamie
"movie" of the evolution of mechanisms producing the patterns. Questions can be
posed regarding whether and how a particular mechanism might have arisen to
produce a natural phenomenon. Typically few assumptions and constraints, other
than Darwinian selection, are built into a genetic algorithm model. As a resuIt,
GAs can substantiate ecological theory by recreating it based on fundamental rules
(i.e., dynamic induction).
There are some fascinating examples of explorations into evolutionary theory
and the emergence of ecological systems and properties. The fundamentals of
how to build these models came largely from computer game playing strategies
(e.g., Bouskila et al. 1998). Koza (1992) in his book on genetic programming
presents a thorough discussion of a genetic program designed to find the optimal
foraging strategies for an Anolis lizard and emergent properties in ants. In the
lizard example, the optimal foraging strategy as described by Roughgarden (1992)
was discovered based simply on the location of the prey, the abundance of the
prey, and the velocity of the lizard. Over time, the lizard can adapt his strategy if
the environmental conditions change. This provides both an example of controleost strategy and a demonstration that evolutionary programs can evolve the
theories governing compIex behavioral strategies using only simple governing
rules. In the ant emergent behavior program (Deneubourg et al. 1986; 1991), a set
of rules is used that governs the actions of individual ants. When these rules are
simultaneously executed, a complex pattern of behavior emerges that causes the
ants to collect and consolidate food pellets into a single pile.
Researchers have long hypothesized that collective behavior could arise from
colonial organisms operating according to very simple rules. Traditional
engineering style models would require tremendous computer power to simulate
individuals and explicitly code the possible sets of interactions. Genetic
algorithms, on the other hand, are especially weIl suited for this type of
applications and demonstrated that the emergence of complex behavior was
indeed possible from simple ruIes. Others, including K vasnicka and Pospichal
(1999) and Reuter and Breckling (1999) have similarly demonstrated emergent
behavior among artificial agents and organisms. GAs, such as Giske et al.'s (1998)
model of spatial dynamics in fish, that simulate coordinated movement of
organisms based on simple rules have greatly improved our understanding of how
groups of organisms travel, sometimes great distances, together in a coordinated
fashion.
The transition from population and community ecology to ecosystem science
occurs when organisms are placed in a physical and chemical environment.
Ecosystem science deals with the interaction between organisms and their non-