Catálogo de publicaciones - libros

This paper addresses the problem regarding the parameter exploration of Multi-Agent Based Simulation for social systems. We focus on the principles of and . In conventional artificial society models, the simulation is executed straightforwardly: Initially, many micro-level parameters and initial conditions are set, then, the simulation steps are executed, and finally the macro-level results are observed. Unlike this, executes these steps in the reverse order: set a macro-level objective function, evolve the worlds to fit to the objectives, then observe the micro-level agent characteristics. Another unique point of our approach is that, using Genetic Algorithms with the functionalities of multi-modal and multi-objective function optimization, we are able to validate the sensitivity of the solutions. This means that, from the same initial conditions and the same objective function, we can evolve different results, which we often observe in real world phenomena. This is the principle of .

This paper provides a survey and analysis of applications of Agent Based Simulation (ABS). A framework for describing and assessing the applications is presented and systematically applied. A general conclusion from the study is that even if ABS seems a promising approach to many problems involving simulation of complex systems of interacting entities, it seems as the full potential of the agent concept and previous research and development within ABS often is not utilized. We illustrate this by providing some concrete examples. Another conclusion is that important information of the applications, in particular concerning the implementation of the simulator, was missing in many papers. As an attempt to encourage improvements we provide some guidelines for writing ABS application papers.

This paper discusses dynamic properties of peer-to-peer communication networks, which emerge from information exchanges among people. First, we gather activity data of communication among people through questionnaires in order to categorize both information (contents) and people, then we develop agent-based simulation models to examine implicit mechanisms behind the dynamics. The agent-based models enable us to discover the quality of information exchanged and the preferences of specific communication groups. The simulation results have suggested that 1) peer-to-peer communication networks have scale-free and small world properties, 2) the characteristics of contents and users are observed in word-of-mouth communications, and 3) the combination of real survey data and agent-based simulation is effective.

Agent-based simulation is increasingly used to analyze the performance of complex systems. In this paper we describe results of our work on one specific agent-based model, showing how it can be validated against the equation-based model from which it was derived, and demonstrating the extent to which it can be used to derive additional results over and above those that the equation-based model can provide.

The agent-based model that we build deals with , the number of years of formal schooling that an individual chooses to undertake. For verification, we show that our agent-based model makes similar predictions about the growth in inequality — that is the growth of the variance in human capital across the population — as th equation-based model from which it is derived. In addition, we show that our model can make predictions about the change in human capital from generation to generation that are beyond the equation-based model.

An agent-based model of food web evolution is presented and contrasted with a particular system dynamics model. Both models examine the effects of speciation and species invasion of food webs, but the agent-based approach focuses on the interactions between individuals in the food web, whereas the system dynamics approach focuses on the overall system dynamics. The system dynamics model is an abstract model of species co-evolution that shows similar characteristics to many natural food webs. The agent-based model attempts to model a similarly abstract food web (in which species are characterised by abstract features that determine how they will fare against any other species). The ultimate aim of this exercise is to explore the many of the assumptions inherent in the system dynamics model; the current challenge is to simply replicate the system dynamics results using agent-based modelling. Preliminary studies have revealed some underlying assumptions in the system dynamics model, as well as some intrinsic difficulties in linking the two different approaches. The paper discusses the key difficulties in linking these different types of models, and presents some discussion of the limits and benefits benefits that each approach may bring to the analysis of the problem.

In this paper we analyse the roosting effect among artificial vampires as a way to preserve altruism from cheaters exploitation. We simulate the formation and maintenance of new social structures (roosts) from initial populations as a consequence of both demographic growth and social organisation. Food-sharing among vampire bats (Desmodus Rotundus) is a well-known form of altruism, necessary for the survival of this species, supported by wide ethological evidence. By means of simulation, we study the performance of the system under varying mutation rate (giving rise to cheaters that exploit the altruistic mechanism) and roost size. Results show that the roosting effect can cope with sensible mutation levels. Moreover, the most robust size of roosts indicated by our simulations is shown to be comparable with the size actually found in nature.

Tax compliance is a field that crosses over several research areas, from economics to machine learning, from sociology to artificial intelligence and multi-agent systems. The core of the problem is that the standing general theories cannot even explain why people comply as much as they do, much less make predictions or support prescriptions for the public entities. The compliance decision is a challenge posed to rational choice theory, and one that defies the current choice mechanisms in multi-agent systems. The key idea of this project is that by considering rationally-heterogeneous agents immersed in a highly social environment we can get hold of a better grasp of what is really involved in the individual decisions. Moreover, we aim at understanding how those decisions determine tendencies for the behaviour of the whole society, and how in turn those tendencies influence individual behaviour. This paper presents the results of some exploratory simulations carried out to uncover regularities, correlations and trends in the models that represent first and then expand the standard theories on the field. We conclude that forces like social imitation and local neighbourhood enforcement and reputation are far more important than individual perception of expected utility maximising, in what respects compliance decisions.

In this paper, a variant of Kiyotaki and Wright’s model of emergence of money is investigated. In the model, each good has different durability rather than storage cost as in Kiyotaki and Wright’s model. Two goods are infinitely durable but one is not durable. With certain conditions, non-durable good can be money as a medium of exchange. But equilibrium condition may be sensitive to the time evolution of the distribution of goods that each agent holds in its inventory. We test, with several learning models using different level of information, whether or not the steady state in this economy can be attainable if the distribution of goods is far from the steady state distribution. Belief learning with full information outperforms the other models. The steady state equilibrium is never attained by belief learning with partial information. A few agents learn to use non-durable good as money by reinforcement learning which does not use information about distribution of goods. It is surprising that providing partial information is rather detrimental for attaining emergence of a non-durable good money.

Multi-Agent Based Simulations (MABS) for real-world problems may require a large number of agents. A possible solution is to distribute the simulation in multiple machines. Thus, we are forced to consider how Large Scale MABS can be deployed in order to have an efficient system. Even more, we need to consider how to cluster those agents in the different execution servers. In this paper we propose an approach based on a holonic model for the construction and update of clusters of agents. We also present two modules to facilitate the deployment and control of distributed simulations.

Agent-based modeling is a powerful tool for systems modeling. Instantiating each domain entity with an agent captures many aspects of system dynamics and interactions that other modeling techniques do not. However, an entity’s agent can execute only one trajectory per run, and does not sample the alternative trajectories accessible to the entity in the evolution of a realistic system. Averaging over multiple runs does not capture the range of individual interactions involved. We address these problems with a new modeling entity, the polyagent, which represents each entity with a single persistent avatar supported by a swarm of transient ghosts. Each ghost interacts with the ghosts of other avatars through digital pheromone fields, capturing a wide range of alternative trajectories in a single run that can proceed faster than real time.