Abstract Agent-based Models Research Articles

When Does Simulated Data Match Real Data? Comparing Model Calibration Functions Using Genetic Algorithms

Agent-based models can be manipulated to replicate real- world datasets, but choosing the best set of parameters to achieve this result can be difficult. To validate a model, the real-world dataset is often divided into a training and test set. The training set is used to calibrate the parameters and the test set is used to determine if the calibrated model represents the real-world data. The difference between the real-world data and the simulated data is determined using an error measure. When using an evolutionary computation technique to choose the parameters, this error measure becomes the fitness function, and choosing the appropriate measure becomes even more crucial for a successful calibration process. We survey the effect of five different error measures in the context of a toy problem and a real world problem (simulating on-line news consumption). We use each error measure in turn to calibrate on the training dataset, and then examine the results of all five error measures on both the training and testing datasets. For the toy problem, one measure was the Pareto-dominant choice for calibration, but no error measure dominated all the others for the real-world problem.

Agentenbasierte Modellierung von Märkten

Agent-based modeling of markets The paper gives an overview of agent-based models of markets as they have been published during the past years. Starting from concept-based models which describe the behaviour of market participants in an abstract manner, recent approaches model real markets in order to find out possible future developments and to evaluate strategies to influence real markets. After a short description of two recent applications of agent-based modeling, a housing market in a city in Brazil and the German market for pharmaceutical products, the paper discusses how this technique could be applied to wood markets.

Verifying agent-based models with steady-state analysis

Agent-based modeling has been well received in the simulation community. Complex systems are simulated by many autonomous agents whose behavior is defined by a conceptual model. However, the model can be improperly implemented or misinterpreted resulting in an implementation that does not reflect the conceptual rules. It is imperative that the implementation’s function be tested against the model’s expected outcome. In this paper, we present certain steady-state techniques that can be used to verify the operation of agent-based simulations. These methods are introduced and then applied to an ecological model which simulates reproductive dynamics of mosquitoes.

Spatial Dynamic Factors Affecting Population-Level Risk Assessment for a Terrestrial Arthropod: An Agent-Based Modeling Approach

ABSTRACT Agent-based models (ABMs) explicit handling of space and time and integration of nonlinear interactions between system components to create a system response could facilitate realistic risk assessment modeling. We used this approach to evaluate the impact of spatial dynamic factors on a theoretical risk assessment of an insecticide on a carabid beetle population (based on Bembidion lampros). Results indicated that both impacts and recovery were dependent on period of application, area treated, spatial distribution of stressor, beetle dispersal, and underlying habitat suitability. The impact of the stressor was detected far outside the area to which it was applied, and the extent of this impact was affected by beetle dispersal and landscape structure. The results call into question the validity of the recovery endpoint as assessed by field trials because model recovery was primarily influenced by reinvasion, depleting surrounding areas. Despite the simplicity of the acute stressor/Bembidion system modeled, the results obtained were highly variable dependent on the precise inputs used. Therefore, we suggest that application of modeling to risk assessment in the future will require a much more complete description of the risk assessment problem than has hitherto been the case.

Application of Chaotic Number Generators in Econophysics

Agent-based models have demonstrated their power and flexibility in Econophysics. However their major challenge is still to devise more realistic simulation scenarios. The complexity of Economy makes appealing the idea of introducing chaotic number generators as simulation engines in these models. Chaos based number generators are easy to use and highly configurable. This makes them just perfect for this application.

Safety assessment of a future taxi into position and hold operation by agent-based dynamic risk modelling

Agent-based dynamic risk modelling supports the design of future air traffic operations by risk analysis methods that account for the performance variability of the interacting operators and systems, and the resulting emergence of safety occurrences. The paper shows the application of this modelling approach in a risk assessment cycle of a future A-SMGCS level 3 supported taxiing into position and hold operation. Accident risk results have been obtained by Monte Carlo simulations of a multi-agent dynamic risk model. The uncertainty in the risk has been evaluated using sensitivity analysis and feedback of operational experts.

A FRAMEWORK FOR PROCESSING AGENT-BASED PEDESTRIAN ACTIVITY SIMULATIONS IN SHOPPING ENVIRONMENTS

Agent-based modeling is a computational methodology that allows the analyst to create, analyze, and experiment with artificial worlds populated by agents. A specific research area is microscale agent-based modeling, which can be used for the simulation of pedestrian movement for low- and high-density scenarios and for the effect of changes in an environment. Such models can also be used for pedestrian dynamics in city centers to show the design effects in the shopping environment. The main contribution of this article is an agent-based model that provides an activity agenda for pedestrian agents that guides their shopping behavior in terms of destination and time spent in shopping areas. This model involves choice mechanisms including where to stop, in what order, and which route to take. The article describes a framework for processing agent-based pedestrian activity simulations within a shopping environment. The main achievement of this research is a validation of the approach leading to a working system. Preliminary findings are reported here.

Agent-Based Simulation Models in Organization Science

Agent-based simulation models allow to reproduce the structure of interactions between the members of an organization, deriving organizational decision-making from individual behavior. This article explains the philosophy of agent-based models, expounds exemplary applications to crucial issues in Organization Science and discusses their validation and acceptance by the scientific community.

Re-thinking simulation: a methodological approach for the application of data mining in agent-based modelling

Agent-based models informed by empirical data are growing in popularity. Many models make extensive use of collected data for the development, initialisation or validation. In parallel, models are growing in size and complexity, generating large amounts of output data. On the other hand, Data Mining is used to extract hidden patterns from large collections of data using different techniques. This work proposes the intense use of Data Mining techniques for the improvement and development of agent-based models. It presents a methodological approach explaining why and when to use Data Mining, with a formal description of each stage of the corresponding process. This is illustrated with a case study, showing the application of the proposed approach step by step.

Statistical Agent-Based Models for Discrete Spatio-Temporal Systems

Agent-based models have been used to mimic natural processes in a variety of fields, from biology to social science. By specifying mechanistic models that describe how small-scale processes function and then scaling them up, agent-based approaches can result in very complicated large-scale behavior while often relying on only a small set of initial conditions and intuitive rules. Although many agent-based models are used strictly in a simulation context, statistical implementations are less common. To characterize complex dynamic processes, such as the spread of epidemics, we present a hierarchical Bayesian framework for formal statistical agent-based modeling using spatiotemporal binary data. Our approach is based on an intuitive parameterization of the system dynamics and can explicitly accommodate directionally varying dispersal, long distance dispersal, and spatial heterogeneity.

Social Dynamics: Theory and Applications

Agent-based models typically involve large numbers of interacting individuals with widely differing characteristics, rules of behavior, and sources of information. The dynamics of such systems can be extremely complex due to their high dimensionality. This chapter discusses a general method for rigorously analyzing the long-run behavior of such systems using the theory of large deviations in Markov chains. The theory highlights certain qualitative features that distinguish agent-based models from more conventional types of equilibrium analysis. Among these distinguishing features are: local conformity versus global diversity, punctuated ideas equilibrium, and the persistence of particular states in the presence of random shocks. These ideas are illustrated through a variety of examples, including competition between technologies, models of sorting and segregation, and the evolution of contractual customs.

COMPUTATIONAL FEATURES OF AGENT-BASED MODELS

Agent-based modeling allows the description of very complex systems. To run large scale simulations of agent-based models in a reasonable time, it is crucial to carefully design data structures and algorithms. We describe the main computational features of agent-based models and report about the solutions we adopted in two applications: The simulation of the immune system response and the simulation of the stock market dynamics.

Swarms of Particle Agents with Harmonic Interactions

Agent-based modeling is a powerful methodology to describe the occurence of complex behavior in biological systems. The interaction of a large number of individuals (agents) may for example lead to the emergence of new forms of collective motion. In this paper, we investigate a particle-based approach to the coherent motion of a swarm with parabolic (i. e. harmonic) interactions between the agents. It is based on generalized Langevin equations for the particle agents, which take into account (i) energetic conditions for active motion, (ii) linear attractive forces between each two agents. The complex collective motion observed can be explained as the result of these different influences: the active motion of the agents, which is driven by the energy-take up, would eventually lead to a spatial dispersion of the swarm, while the mutual interaction of the agents results in a tendency of spatial concentration. In addition to particle-based computer simulations, we also provide a mathematical framework for investigating the collective dynamics.