Data Provenance for Agent-Based Models in a Distributed Memory

Multi-agent simulations are useful for exploring collective patterns of individual behavior in social, biological, economic, network, and physical systems. However, there is no provenance support for multi-agent models (MAMs) in a distributed setting. To this end, we introduce ProvMASS, a novel approach to capture provenance of MAMs in a distributed memory by combining inter-process identification, lightweight coordination of in-memory provenance storage, and adaptive provenance capture. ProvMASS is built on top of the Multi-Agent Spatial Simulation (MASS) library, a framework that combines multi-agent systems with large-scale fine-grained agent-based models, or MAMs. Unlike other environments supporting MAMs, MASS parallelizes simulations with distributed memory, where agents and spatial data are shared application resources. We evaluate our approach with provenance queries to support three use cases and performance measures. Initial results indicate that our approach can support various provenance queries for MAMs at reasonable performance overhead.

From the viewpoint of strong spatial cognition in graph problems, the shortest path can be identified in one physical action using strings and pins that respectively represent graph edges and vertices. By pulling a start and an end pin, we can construct a series of stretched strings as the shortest path. We use agent-based models (ABMs) to translate this action into computational representations. Assuming that a set of strings and pins are hung on a wall with a start pin, agents are disseminated downward to a destination as gravity forces. We implemented three models: a discrete-event, an asynchronous, and an aggregated agent dissemination on top of the MASS (multi-agent spatial simulation) library. To address large-scale network environments, we blended HDFS into MASS so that a graph data set is read over a cluster system in parallel. This paper presents these ABM implementations and performance measurements over a cluster system.

Autonomous vehicles (AVs) may have a transformative influence on transportation systems. AV use is expected to reduce crash, transport costs, and land use. However, the net effects of the autonomous vehicle are unknown because there are no fully-autonomous vehicles in reality. The simulation systems are indispensable to be built up not only for cost but also for obtaining massive data. An agent-based simulation is capable of simulating the entities in a complex system, such as the AV system. Considering the potential of agent-based models to complement existing models and the difficulty to understand, to duplicate and to compare different agent-based models, this paper review existing agent-based modeling of autonomous vehicles with the help of ODD protocol, which has been applied in the review of agent-based land-use change models and agricultural policy evaluation. The significant variables that researchers have taken into consideration in sensitivity analysis and in the different scenarios are explored in the paper. Fleet size is the primary variable researchers considered, and we also discussed the probable causes of different AV replacement rates. We have assessed the quality of reviewed articles to gain more acceptance from the agent-based modeling community. Most of the reviewed papers follow the ODD protocol of agent decision making. However, in order to improve the level of modeling transparency, the executable and related data shall be available, at least the data source and the programming language/simulation platform should be provided. Several recommendations for further research are also presented.

Urban land-use modeling methods have experienced substantial improvements in the last several decades. With the advancement of urban land-use change theories and modeling techniques, a considerable number of models have been developed. The relatively young approach, agent-based modeling, provides urban land-use models with some new features and can help address the challenges faced by traditional models. Applications of agent-based models to study urban dynamics have increased steadily over the last twenty years. To offer a retrospective on the developments in agent-based models (ABMs) of urban residential choices, we review fifty-one relevant models that fall into three general categories: (i) urban land-use models based on classical theories; (ii) different stages of the urbanization process; and (iii) integrated agent-based and microsimulation models. We summarize and compare the main features of these fifty-one models within each category. This review focuses on three fundamental new features introduced byABMs. The first is agent heterogeneity with particular attention to the method of introducing heterogeneity in agents' attributes and behaviors. The second is the representation of land-market processes, namely preferences, resources constraints, competitive bidding, and endogenous relocation. The third is the method of measuring the extensive model outputs. In addition, we outline accompanying challenges to, and open questions for, incorporating these new features. We conclude that, by modeling agent heterogeneity and land markets, and by exploiting a much broader dimension of output, we will enhance our understanding of urban land-use change and are hopefully able to improve model fitness and robustness.

The ‘keep it simple, stupid’ slogan, or the KISS principle has been the basic guideline in agent-based modeling (ABM). While the KISS principle or parsimony is vital in modeling attempts, conventional agent-based models remain abstract and are rarely incorporated or validated with empirical data, leaving the links between theoretical models and empirical phenomena rather loose. This article reexamines the KISS principle and discusses the recent modeling attempts that incorporate and validate agent-based models with spatial (geo-referenced) data, moving beyond the KISS principle. This article also provides a working example of such time and space specified (TASS) agent-based models that incorporates Schelling's (1971) classic model of residential segregation with detailed geo-referenced demographic data on the city of Chicago derived from the US Census 2010.

Flood risk management in Sint Maarten – A coupled agent-based and flood modelling method

One of the major trends in traffic simulations is to take into account microscopic aspects of traffic flows at the street level. Multi-agent models such as MATSim (multi-agent transport simulation) have been highlighted for recent years as a solution to address these complex and microscopic simulation requirements. They are viewed as an emergent and collective behavior of agents, (i.e., vehicles). However, as the simulations scale up, their computational requirements could get increased beyond the capability of a single CPU and thus should be fulfilled with parallelization. Multithreading can partially contribute to parallelization by utilizing multi-cores, but cannot give full scalability of both CPU power and memory space. To support distributed-memory parallelization for multi-agent models, we have developed the MASS (multi-agent spatial simulation) library. This paper presents how to parallelize MATSim using the MASS library and demonstrates the library's portability and execution performance in practical transport simulations.

Agent-Based Modeling can contribute to the understanding of international trade processes. Models for the effects of culture and cultural differences on agent behavior are required for realistic agent-based simulation of international trade. This paper makes a step toward modeling of culture in agents. The focus is one of the five dimensions of culture according to Hofstede: individualism versus collectivism. The paper presents an analysis based on social science literature about national cultures. For cultural differentiation of agent behavior, rules are formulated for individualist versus collectivist agent behavior with respect to negotiations, cooperation or defection in the delivery phase of transactions, trade partner selection, and trust. Example computations demonstrate the feasibility in multi-agent simulations.

The aim of the paper is to discuss the possibility of using complex software agents in a simulation model in order to represent and analyse the dynamics of certain types of criminal systems via Agent Based Modelling (ABM), in particular Extortion Racket Systems (ERSs). It presents a simulation model and the results of a large number of runs to find out under which parameter constellations governing the normative behaviour of the software agents the model replicates the macro observations in a number of provinces in Southern Italy. Finally the simulation model is applied to analysing strategies and their effect on the behaviour of the agents and the system as a whole.

Some graph analyses, such as social network and biological network, need large-scale graph construction and maintenance over distributed memory space. Distributed data-streaming tools, including MapReduce and Spark, restrict some computational freedom of incremental graph modification and run-time graph visualization. Instead, we take an agent-based approach. We construct a graph from a scientific dataset in CSV, tab, and XML formats; dispatch many reactive agents on it; and analyze the graph in the form of their collective group behavior: propagation, flocking, and collision. The key to success is how to automate the run-time construction and visualization of agent-navigable graphs mapped over distributed memory. We implemented this distributed graph-computing support in the multi-agent spatial simulation (MASS) library, coupled with the Cytoscape graph visualization software. This paper presents the MASS implementation techniques and demonstrates its execution performance in comparison to MapReduce and Spark, using two benchmark programs: (1) an incremental construction of a complete graph and (2) a KD tree construction.

Mathematical models in systems biology can use quantitative techniques to study the integrated behaviors of biological systems in macro level. Agent-based modeling provides a qualitative framework, focusing on how each individual molecule behaves in micro level. We are motivated to combine their features together to describe the gene regulatory networks, in particular the emergence of macro phenomena from micro interactions. An agent-based model will be well defined as general as possible, including of many agents of various molecular species interacting in it. The mathematical premises, which are grounded in the micro behaviors of agents, can derive the chemical master equations and the stochastic differential equations for modeling the integrated behaviors of systems in macro level. Combing agent-based models with stochastic differential equations affords a more complete perspective on gent regulatory networks. In addition, the sources and magnitudes of deterministic dynamics and random noises are explicitly characterized.

Integrating sensor networks in cloud computing gives new opportunities of using as many cloud-compute nodes as necessary to analyze real-time sensor data on the fly. However, most cloud services for parallelization such as OpenMP, MPI, and MapReduce are not always fitted to on-the-fly sensor-data analyses that are implemented as model-based entity-based, and multi-agent simulations. To address this semantic gap between analyzing algorithms and their actual implementations, we are developing MASS: a library for multi-agent spatial simulation that composes a user application of distributed array elements and multi-agents, each representing an individual simulation place or an active entity. All computation is enclosed in each of elements and agents that are automatically distributed over different computing nodes. Their communication is then scheduled as periodical data exchanges among those entities using their logical indices. We are currently implementing a multi-process and a multi-threaded version of the MASS library, both to be combined in a single version in the near future. This paper focuses on an implementation and preliminary performance of the multi-process version.

Agent-based modeling (ABM) is a computational technique employed to simulate the behavior of autonomous units called agents. An agent is a virtual entity which interacts with its environment and other agents and is capable of making decisions and taking actions based on these interactions. Relatively simple rules are utilized to govern individual interactions; however the cumulative behavior of all agents may be complex and may sometimes reveal interesting patterns. Agent-based modeling, though first introduced in the field of artificial intelligence, has gained acceptance in a many diverse fields such as finance engineering, bio-film modeling, social computation, and ecology. The systems studied in these fields are inherently complex involving many interacting components. These systems are modeled by representing the various system components as agents and having simple rules governing the behavior of agents. These rules can be changed to match the emergent behavior of the model with the behavior of the actual system. However, validating such a model is not straight forward as it is difficult to objectively determine similarity between the patterns in the model output and data. We propose a universal paradigm to facilitate such comparisons by extracting features from patterns in the data and the model output and use these features to train and validate the model. The merit of this methodology will be adopted to train an ABM simulating the movement of cattle grazing in a pasture using data from GPS units installed on individual cows.

Supply chain simulation is a fundamental approach for supply chain prediction, management, evaluation, and improvement. In order to simulate member behavior, organizational strategy, and management strategy of the supply chain, an agent-based modeling and simulation approach on multi-stage supply chain operation was proposed in this paper. First, the model structure of multi-stage supply chain operation was introduced. Then, the individual agent behavior model was emphatically studied, and hierarchical colored Petri-nets were used to describe the agent behavior process and cooperative behavior process. Finally, the model was validated using a series of simulation which focused on the agent inventory strategy.

Teamwork has become an important research field and its contribution to organizational performance has attracted attention of researchers from several disciplines. The development and application of newly emerged technologies such as Industry 4.0, Internet of Things, and cyber physical systems create additional concerns for teamwork which claim to be integrated into existing models. The objective of this chapter is to advance research on teamwork, by facilitating researchers with a review which identifies the key factors that affect teamwork behavior both in human and in agent-based teamwork models, while indicating if and how they are inter-related. A review of related studies was conducted, and, as a result, a range of factors that affect teamwork behavior to both human and agent-based models was identified and analyzed. From the analysis, stand out factors that gain attention while newly-appeared factors are determined from recent studies about models shift towards dynamic and realistic environments. These discoveries point to new aspects of teamwork behavior.