Foreword
Simulation is a powerful tool, that allows domain experts to test their theories as safe virtual experiments. But as the systems being modeled grow and
become complex, with many interacting elements, the code also becomes extremely complex. Whether it be modeling an ant colony, or human interactions
in economic systems, these problems not only help the domain experts, but
also require immense effort from computer scientists. A multitude of computer
science techniques are involved such as how to design models, build code, simulate and analyze data. Agent-based modeling is an example of simulation
technique, which can help researchers deviate from stochastic and differential
equations, to more granular approaches of building models based on interactions.
Agent-based models have shown applications in various fields such as biology, economics and social sciences. Over the years, multiple agent-based modeling frameworks have been produced, allowing experts with non-computing
background to easily write and simulate their models. However, most of these
models are limited by the capability of the framework, time it takes for a simulation to finish, or handling the massive amounts of data produced. FLAME
(Flexible Large-scale Agent-based Modeling Environment) was produced at
the University of Sheffield, and developed through the years, with multiple
grants and projects from biology, sociology and economics. As a challenge, it
was able to produce an economic agent-based model, EURACE, consisting of
three markets integrated together, which had never been done before.
This book contains a comprehensive summary of the field and how concepts of X-machines can be stretched across multiple fields to produce agent
models. It has been written with several audiences in mind. First, it is organized as a collection of models, with detail descriptions of how models can be
designed, especially for beginners in agent-based models. A number of theoretical aspects of software engineering and how they relate to agent-based
models have been discussed for students interested in software engineering
and parallel computing. Finally, it is intended as a guide to developers from
biology, economics and sociologists, who want to explore how to write agentbased models for their research area. By working through model examples
provided, anyone should be able to design and build their agent-based models
and deploy them on their machines. With FLAME, they can easily increase
the agent number and run models on parallel computers, in order to save on
simulation complexity and waiting time for results.
xiii
Simulation is a powerful tool, that allows domain experts to test their theories as safe virtual experiments. But as the systems being modeled grow and
become complex, with many interacting elements, the code also becomes extremely complex. Whether it be modeling an ant colony, or human interactions
in economic systems, these problems not only help the domain experts, but
also require immense effort from computer scientists. A multitude of computer
science techniques are involved such as how to design models, build code, simulate and analyze data. Agent-based modeling is an example of simulation
technique, which can help researchers deviate from stochastic and differential
equations, to more granular approaches of building models based on interactions.
Agent-based models have shown applications in various fields such as biology, economics and social sciences. Over the years, multiple agent-based modeling frameworks have been produced, allowing experts with non-computing
background to easily write and simulate their models. However, most of these
models are limited by the capability of the framework, time it takes for a simulation to finish, or handling the massive amounts of data produced. FLAME
(Flexible Large-scale Agent-based Modeling Environment) was produced at
the University of Sheffield, and developed through the years, with multiple
grants and projects from biology, sociology and economics. As a challenge, it
was able to produce an economic agent-based model, EURACE, consisting of
three markets integrated together, which had never been done before.
This book contains a comprehensive summary of the field and how concepts of X-machines can be stretched across multiple fields to produce agent
models. It has been written with several audiences in mind. First, it is organized as a collection of models, with detail descriptions of how models can be
designed, especially for beginners in agent-based models. A number of theoretical aspects of software engineering and how they relate to agent-based
models have been discussed for students interested in software engineering
and parallel computing. Finally, it is intended as a guide to developers from
biology, economics and sociologists, who want to explore how to write agentbased models for their research area. By working through model examples
provided, anyone should be able to design and build their agent-based models
and deploy them on their machines. With FLAME, they can easily increase
the agent number and run models on parallel computers, in order to save on
simulation complexity and waiting time for results.
xiii