top banner

Japanese | English

Multi-agent Simulation


Chapter9 Slide(1)

Chapter9 Slide(2)

Chapter9 Slide(3)

Chapter9 Slide(4)

Chapter9 Slide(5)

Chapter9 Slide(6)

Chapter9 Slide(7)

Chapter9 Slide(8)

Chapter9 Slide(9)

Chapter9 Slide(10)

Chapter9 Slide(11)

Chapter9 Slide(12)

Chapter9 Slide(13)

Chapter9 Slide(14)

Chapter9 Slide(15)

Chapter9 Slide(16)

Chapter9 Slide(17)

Chapter9 Slide(18)

Chapter9 Slide(19)

Chapter9 Slide(20)

Chapter9 Slide(21)

Chapter9 Slide(22)

Chapter9 Slide(23)

Chapter9 Slide(24)

Chapter9 Slide(25)

Chapter9 Slide(26)

Chapter9 Slide(27)

Chapter9 Slide(28)

Chapter9 Slide(29)

Chapter9 Slide(30)

Chapter9 Slide(31)

Chapter9 Slide(32)

Chapter9 Slide(33)

Chapter9 Slide(34)

go to 10 pages prev   prev   next   go to 10 pages next

Slide data PDF, PPTX
Movie (P.20) movie1(21.5MB)
Movie (P.20) movie2(16.0MB)
Movie (P.23) movie3(9.96MB)
Movie (P.23) movie4(14.6MB)
Movie (P.26) movie5(6.84MB)
Movie (P.26) movie6(8.86MB)

Chapter9: Multi-agent Simulation (Textbook, pdf, 1,613KB)

  • 9.1 Why Multi-agent Simulations?
  • 9.2 Creating and Analyzing Social Systems
  • 9.3 Participatory Simulations
  • 9.4 Examples of Multi-agent Simulations
    • 9.4.1 Evacuation Simulation
    • 9.4.2 Traffic Simulations
    • 9.4.3 Economic Simulations
  • 9.5 Using Multi-agent Simulations

What is the relevance of Multi-agent Simulation in Field Informatics

Innovations emerge by appearing new technologies as well as by the elicitation of needs in society and daily-life. ICT technologies which can facilitate fragmentation and the recombination of social systems boost such innovations and inspire unlimited possibilities of society. The question is how to predict ICT-driven innovations emerge in society and daily-life? The required simulations are not traditional ones such as physical or chemical simulations, but novel simulations that can calculate a sequence of interactions between humans/organizations based on individual models of autonomous decision-making entities. Multiagent-based simulations introduced in this chapter can be used to predict possible changes in society and life caused by human behaviors with new technologies and needs.


Essentially, simulations are used to virtually reproduce complex phenomena that are difficult to observe in the real world, on computers. Simulations can be divided into two classes; macroscopic and microscopic simulations according to the abstraction level of models of simulation targets. System dynamics is a typical example of macroscopic simulations. Macro simulation reproduces a phenomenon based on macroscopic viewpoint so that the whole of a simulation target is represented as a single model and its behavior is defined by governing equations. Consequently, macro simulations allow observation of behaviors or changes in the overall system, but the local properties of individual elements or interactions among elements cannot be reproduced. On the other hand, micro simulation reproduces a complex social phenomenon by accumulating microscopic behaviors of models of social entities (e.g., humans or organizations) including interactions among them. Assuming that human society consists of a lot of decision-making entities, it seems natural to predict behaviors of society with a micro simulation. In particular, micro simulations manifest their ability to clearly present a variety of individual core behaviors in the reproduction and analysis of complex collective behaviors.

Multiagent-based simulations have become increasingly popular as a type of micro simulation to represent diversity and heterogeneity of behaviors of entities and enable us to observe micro and macro phenomena. In multiagent-based simulations, humans and organizations are modeled as agents. An agent can autonomously determine his/her behavior and interact with the environment and other agents based on surrounding environment and information from others. In contrast with traditional micro-simulations, which tend to be the same in their entity behavior mechanisms, multi-agent simulations can represent individual decision-making in detail according to an agent's circumstances, so they can reproduce the complex phenomena that arise from the results of interactions between different agents.

Case Study

  1. Evacuation simulation
  2. In a disaster evacuation simulation, the evacuees and rescue workers have different attributes under disaster conditions, such as collapsed houses, fire, etc., and it simulates the conditions under which they perform evacuation behavior and rescue activities, respectively. At such time the behaviors of the evacuees and the rescue workers are not the same, so it is necessary to analyze their behaviors individually. Furthermore, the analysis of micro-macro relationship is needed because panic may fueled by a chain-reaction of local behaviors.

  3. Traffic simulation
  4. Traffic is a social phenomenon that is represented as an accumulation of interactions among vehicles caused by drivers' behavior. Drivers will act in a variety of ways even under identical traffic conditions due to their individuality, cognitive ability, as well as the performance of the vehicles they drive. This results in a wide variety of traffic phenomena. Therefore, to efficiently control traffic systems for ensuring smooth traffic flow, it is desirable to comprehend emerged changes in the entire traffic resulting of each driver's behavior caused by enforcement of traffic controls such as speed limits. In other words, if we could run a simulation in advance that applied a variety of traffic policies and then compared the effects and influence of each policy, we would be able to effectively promote designs of mechanisms, such as the traffic rules, etc.

  5. Economic simulation
  6. Market economy analysis using simulations has a long history. Even today researchers are very active in looking at the construction of artificial markets and the reproduction of economic phenomena. The market is a social system where participants have various and complicated preferences and strategies. Traditional economic theories usually assume rigid (sometimes unrealistic) rationality of players in the market, but irrational behaviors can be observed in the real market. Thus, not only theoretical analysis with an assumption of rational players, but behavior analysis of real-life players based on practical viewpoint are required. With economic simulations it is desirable to model and analyze the economic behavior of each player under free market mechanisms.