74
F. Recknagel
series data. Even though the range and extent of data available may limit ANN,
EC can explore both causal and empirical information by means of hybrid
frameworks to induce and evolve models (Bobbin and Recknagel2001; Whigham
and Recknagel 2001). However predictive capacity of resulting models still relies
on underlying causal and empirical knowledge. The application of adaptive agents
(AA) (Holland 1992; Holland 1998) is an attempt to go one step further: to evolve
ecosystem structures and behaviours by emerging, submerging, interacting and
evolving ecological entities simulated by adaptive agents.
The present paper reviews current developments of individual-based AA for
microbial and terrestrial ecosystems, and designs a concept how state variablebased AA can be applied in order to simulate evolving species abundance and
succession in aquatic ecosystems. The proposed concept is currently developed
and tested towards adaptive lake ecosystem simulation. It is expected to overcome
constraints by the rigidity of traditional dynamic ecosystem models and enable to
evolve ecosystem structures and behaviours.
5.2
Adaptive Agents Framework
Holland (1992) introduced Echo (Fig. 5.1.) as a generic simulator designed to
explore interactions among large numbers of different adaptive agents (AA). It
provides for the study of populations of evolving, reproducing agents distributed
over a geography with different inputs of renewable resources at various sites.
Each agent has simple capabilities - offence, defense, trading, mate selection -
determined by a set of "chromosomes". Chromosomes in each agent are
differentiated into two classes:
Tag chromosomes determine the agent's external phenotypic characteristics
and distinguish: offence tag, defence tag and mating tag. Tags are displayed on the
exterior of an agent and are analogous to signature groups of an antigen or the
logo of an organisation. Condition chromosomes determine what kinds of
interactions take place when agents encounter one another and distinguish: combat
(competition), trading (mutualism) or mating (reproduction).
The fact that an agent's structure is completely defined by its chromosomes,
which are just strings over the resource alphabet {a,b,c,d}, plays a critical role in
its reproduction. An agent reproduces when it "collects" enough letters to make
copies of its chromosomes. An agent can collect these letters through its
interactions: combat, trade, or uptake from the environment. Each agent has a
reservoir in which it stores collected letters until there are enough of them for
reproduction to take place. Interactions between agents, when they come into
contact are determined by a simple sequence of tests based on their tags and
conditions. In the simplest model they first test for combat, then they test for
trading and fInally they test for mating as follows:
F. Recknagel
series data. Even though the range and extent of data available may limit ANN,
EC can explore both causal and empirical information by means of hybrid
frameworks to induce and evolve models (Bobbin and Recknagel2001; Whigham
and Recknagel 2001). However predictive capacity of resulting models still relies
on underlying causal and empirical knowledge. The application of adaptive agents
(AA) (Holland 1992; Holland 1998) is an attempt to go one step further: to evolve
ecosystem structures and behaviours by emerging, submerging, interacting and
evolving ecological entities simulated by adaptive agents.
The present paper reviews current developments of individual-based AA for
microbial and terrestrial ecosystems, and designs a concept how state variablebased AA can be applied in order to simulate evolving species abundance and
succession in aquatic ecosystems. The proposed concept is currently developed
and tested towards adaptive lake ecosystem simulation. It is expected to overcome
constraints by the rigidity of traditional dynamic ecosystem models and enable to
evolve ecosystem structures and behaviours.
5.2
Adaptive Agents Framework
Holland (1992) introduced Echo (Fig. 5.1.) as a generic simulator designed to
explore interactions among large numbers of different adaptive agents (AA). It
provides for the study of populations of evolving, reproducing agents distributed
over a geography with different inputs of renewable resources at various sites.
Each agent has simple capabilities - offence, defense, trading, mate selection -
determined by a set of "chromosomes". Chromosomes in each agent are
differentiated into two classes:
Tag chromosomes determine the agent's external phenotypic characteristics
and distinguish: offence tag, defence tag and mating tag. Tags are displayed on the
exterior of an agent and are analogous to signature groups of an antigen or the
logo of an organisation. Condition chromosomes determine what kinds of
interactions take place when agents encounter one another and distinguish: combat
(competition), trading (mutualism) or mating (reproduction).
The fact that an agent's structure is completely defined by its chromosomes,
which are just strings over the resource alphabet {a,b,c,d}, plays a critical role in
its reproduction. An agent reproduces when it "collects" enough letters to make
copies of its chromosomes. An agent can collect these letters through its
interactions: combat, trade, or uptake from the environment. Each agent has a
reservoir in which it stores collected letters until there are enough of them for
reproduction to take place. Interactions between agents, when they come into
contact are determined by a simple sequence of tests based on their tags and
conditions. In the simplest model they first test for combat, then they test for
trading and fInally they test for mating as follows: