Jan Drchal, Michal  Čertický and Michal Jakob.
Data Driven Validation Framework for Multi-agent Activity-based Models.

Abstract: Activity-based models, as a specific instance of agent-based models, deal with afebts that structure their activity in terms of (daily) activity schedules. An activity schedule consists of a sequence of activity instances, each with its assigned start time, duration and location, together with transport modes used for travel between subsequent activity locations. A critical step in the development of simulation models is validation. Despite the growing mportance of activity-based models in modelling transport and mobility, there has been so far no work focusing specifically on the validation of such models. In this paper, we therefore propose a six-step Validation Framework for Activity-based Models (VALFRAM) that allows exploiting historical real-world data to assess the validity of activity-based models. The framework compares temporal and spatial properties and the structure of activity schedules against real-world travel diaries and origin-destination matrices. We confirm the usefulness of the framework on several real-world activity-based transport models.

 

Emmanuel Hermellin and Fabien Michel. 
GPU Environmental Delegation of Agent Perceptions: Application to Reynolds's Boids.

Abstract: Using Multi-Agent Based Simulation (MABS), computing resources requirements often limit the extent to which a model could be experimented with. Regarding this issue, some research works propose to use the General-Purpose Computing on Graphics Processing Units (GPGPU) technology. GPGPU indeed allows to use the massively parallel architectures of graphic cards to perform general-purpose computing with huge speedups. Still, GPGPU requires the underlying program to be compliant with the specific architecture of GPU devices, which is very constraining. Especially, it turns out that doing MABS using GPGPU is very challenging because converting Agent Based Models (ABM) accordingly is a very difficult task. In this context, the \emph{GPU Environmental Delegation of Agent Perceptions} principle has been proposed to ease the use of GPGPU for MABS. This principle consists in making a clear separation between the agent behaviors, managed by the CPU, and environmental dynamics, handled by the GPU. For now, this principle has shown good results, but only on one single case study. In this paper, we further trial this principle by testing its feasibility and genericness on a classic ABM, namely Reynolds's boids. To this end, we first review existing boids implementations and propose our own benchmark model. The paper then shows that applying GPU delegation not only speeds up boids simulations but also produces an ABM which is easy to understand, thanks to a clear separation of concerns.

 

Sara Montagna, Andrea Omicini and Danilo Pianini. 
Extending the Gillespie's Stochastic Simulation Algorithm for Integrating Discrete-Event and Multi-Agent Based Simulation.

Abstract: Whereas Multi-Agent Based Simulation (MABS) is emerging as a reference approach for complex system simulation, the event-driven approach of Discrete-Event Simulation (DES) is the most used approach in the simulation mainstream. In this paper, we elaborate on two intuitions: (i) event-based systems and multi-agent systems are amenable of a coherent interpretation within a unique conceptual framework; (ii) integrating MABS and DES can lead to a more expressive and powerful simulation framework. Accordingly, we propose a computational model integrating DES and MABS based on an extension of the Gillespie’s stochastic simulation algorithm. Then we discuss a case of a simulation platform (ALCHEMIST) specifically targeted at such a kind of complex models, and show an example of urban crowd steering simulation.

Chi Quang Truong, Patrick Taillandier, Benoit Gaudou, Quang Minh Vo, Hieu Trung Nguyen and Alexis Drogoul. 
Exploring agent architectures for farmer behavior in land-use change.

Abstract: Farmers are the key actors of land-use change processes. It is thus essential to choose a suitable architecture for farmer behaviors to model such processes. In this paper, we compared three models with different architectures to model the farmer behaviors in the coastal areas of the Ben Tre province: (i) The first one is a probabilistic model that allows farmer to select the land-use pattern based on land change probability; (ii) The second model is based on multi-criteria decision making and takes into account the land suitability of the parcel and the farmer benefit; (iii) The third model used a BDI (Beliefs - Desires - Intentions) architecture. For each of these models, we have compared the difference between simulated data and real data by using the Fuzzy Kappa coefficient. The results show that the suitability of the BDI architecture to build land-use change model and to support decision making in land-use planning.

James Archbold and Nathan Griffiths. 
Maximising Influence in Non-blocking Cascades of Interacting.

Abstract: In large populations of autonomous individuals, the propagation of ideas, strategies or infections is determined by the composite effect of interactions between individuals. The propagation of concepts in a population is a form of influence spread and can be modelled as a cascade from a set of initial individuals through the population. Understanding influence spread and information cascades has many applications, from informing epidemic control and viral marketing strategies to understanding the emergence of conventions in multi-agent systems. Existing work on influence spread has mainly considered single concepts, or small numbers of blocking (exclusive) concepts. In this paper we focus on non-blocking cascades, and propose a new model for characterising concept interaction in an independent cascade. Furthermore, we propose two heuristics, Concept Aware Single Discount and Expected Infected, for identifying the individuals that will maximise the spread of a particular concept, and show that in the non-blocking multi-concept setting our heuristics out-perform existing methods.

 

Concepts Benjamin Herd, Simon Miles, Peter McBurney and Michael Luck. 
MC2MABS: A Monte Carlo Model Checker for Multiagent-based Simulations.

Abstract: Agent-based simulation has shown great success for the study of complex adaptive systems and could in many areas show advantages over traditional analytical methods. Due to their internal complexity, however, agent-based simulations are notoriously difficult to verify and validate. 
This paper presents MC2MABS, a Monte Carlo Model Checker for Multiagent-Based Simulations. It incorporates the idea of statistical runtime verification, a combination of statistical model checking and runtime verification, and is tailored to the approximate verification of large-scale agent-based simulations. We provide a description of the underlying theory together with design decisions, an architectural overview, and implementation details. The performance of MC2MABS in terms of both runtime consumption and memory allocation is evaluated against a set of example properties.

 

Diego Queiroz and Jaime Sichman. 
Experiments of von Neumann Neighborhoods in Parallel Multi-Agent Simulations.

Abstract: Parallel computation of simulations involving multiple agents is a hard task to accomplish due the high dependence that is hold by the agents’ interactions. These interactions usually are determined by environmental constraints which specify rules defining how agents interacts with each other. One of these predetermined rules is called von Neumann neighborhood, which is applied, specifically, to two-dimensional lattices. This work intends to analyze how simulations of multiple agents that interacts based on von Neumann neighborhoods can be parallelized in a way that reduces overhead, focusing to enhance the performance of the system.

 

Arnaud Grignard, Guillaume Fantino, J. Wesley Lauer, Alexandre Verpeaux and Alexis Drogoul. 
Agent-Based Visualization: A simulation tool for the analysis of river morphosedimentary adjustments.

Abstract: Spatially explicit agent-based models and simulations are playing an increasing role in the modelling of complex natural and social systems. The ARCHEM project belongs to this new research area. It proposes a new methodology to visualize the fine-scale sediment transport of a river. In this paper, we present the first implementation of ARCHEM on the case study of the Rhone river. Even though visualization cannot replace the analysis of simulation results, it often constitutes the more accessible medium to orientate a more specific and accurate analysis. It offers immediate feedback as well as a way to interact with and analyze results. We show how to support multiple viewpoints and different levels of abstraction using an agent-based visualization approach. We present a specific application focusing on dynamical 3D rendering of a GIS file and the analysis of morphosedimentary adjustments.

 

Banafsheh Hajinasab, Paul Davidsson, Jan A. Persson and Johan Holmgren. 
Towards an agent-based model of passenger transportation.

Abstract: In this paper, we present a multi-agent based simulation model for supporting the decision making in urban transport planning. The model can be used to investigate how different transport infrastructure investments and policy instruments will affect the travel choices of passengers. We have identified four main categories of factors influencing the choice of travel: cost, time, conven-ience, and social norm. However, travelers value these factors differently de-pending on their individual preferences, something which can be modeled in an agent-based model. Moreover, instead of modeling the transport system explic-itly, on-line web services are used to generate travel options. The model can support transport planners by providing modal share, as well as economical and environmental consequences. As a first validation of the model, we have made a simple case study of three scenarios where we analyze the effects of changes to the public transport fees on commuter’s travel choices in the Malmö-Lund region in Sweden.

 

Itsuki Noda. 
Optimality and Equilibrium of Exploration Ratio for Multiagent Learning in Nonstationary Environments --- Case Study of Resource Sharing Problems ---.

Abstract: I investigate relations among  exploration ratio,  total performance of multiagent learning,  and relative performance of individual agents. Exploration ratio is a key parameter to determine features of multiagent learning in two aspects: as a speed controller of learning in individual agents, and as a reciprocal noise factor for other agents. The investigation figures out trade-off of the two aspects and shows existence of single optimal value of the ratio to minimize the learning errors. I also carried out simulation experiments to compare the performances of agents who use different exploration ratios. The results of the experiments tells existence of equilibrium points to choose the ratio by individual agents. Finally, we discuss the relationship between optimal and equilibrium values of the exploration ratio, which might bring dilemma of selection of the exploration ratio. 

 

Arnaud Banos, Nathalie Corson, Benoit Gaudou, Vincent Laperrière and Sébastien Rey Coyrehourcq. 
Coupling micro and macro dynamics models on networks: Application to disease spread.

Abstract: A hybrid model coupling an aggregated equation-based model and an agent-based model is presented in this article. It is applied to the simulation of a disease spread in a city network. We focus here on the evaluation of our hybrid model by comparing it with a simple aggregated model. We progressively introduce heterogeneities in the model and measure their impact on three indicators: the maximum intensity of the epidemic, its duration and the time of the epidemic peak. Finally we present how to integrate mitigation strategies in the model and the benefits we can get from our hybrid approach over single paradigm models.

 

Zulkuf Genc, Michel Oey, Hendrik van Antwerpen and Frances Brazier. 
Dynamic Data-Driven Experiments in the Smart Grid Domain with a Multi-Agent Platform.

Abstract: Pervasive information and communication technologies and large-scale complex systems, are strongly influencing today’s networked society. Understanding the behaviour and impact of such distributed, often emergent systems on society is of vital importance. This paper proposes a new approach to understanding the complexity of large-scale complex systems in the context of smart grids. Multi-agent based distributed simulations of realistic multi-actor scenarios incorporating real-time dynamic data and active participation of actors is the means to this purpose. The Symphony experiment platform, developed to study complex emergent behaviours and to facilitate the analysis of the system dynamics and actor interactions, is the enabler.