• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 2
  • 1
  • Tagged with
  • 3
  • 3
  • 3
  • 3
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 1
  • 1
  • 1
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

Un enfoque de simulación basada en agentes en procesos estratégicos de formación de redes

Larrosa, Juan Manuel C. 20 May 2010 (has links)
La modelación individual de la interacción entre agentes en Economía ha recibido particular atención en los últimos años. Estos aportes han recaído inicialmente, y siguiendo la tradición académica, en modelos analíticos que contemplan la aparición de interacciones locales entre los individuos. Esta ineracción permite, en algún punto, determinar resultados de conducta diferentes comparados con un análisis en el que los agentes se comporten aisladamente. La moderación de dicha interacción en términos formales ha utilizado herramitentas matemáticas que permiten la traslación de efectos por parte de un individuo hacia otro. La teoría de juegos, dado que analiza en el comportamiento óptimo de un agente que se ve restringido en su accionar por las estrategias de otros agentes, resulta un modelo analítico clásico de ciertos casos de interacción. La tesis utiliza el enfoque basado en agentes para modelar las interacciones entre agentes y comprobar si los resultados teóricos pueden ser alcanzados a través de un modelo computacionalmente evolutivol. Sus resultados no son formales sino numéricos. Su sustento no es básicamente matemático sino computacional. Los agentes resultan ser bajo este marco de investigaciòn objetos de programas. Los objetos agente tienen estados y reglas de comportamiento. Correr un modelo de agentes comprende la inicialización de la población de agentes, el proceso subsiguientede interacción entre los mismos y el monitoreo del resultado final. En definitiva, correr el modelo en el tiempo la suficiente cantidad de veces, en caso de ser estocástico, es lo único que se necesita para "resolverlo". La tesos comprende 5 capítulos. El primero es una introducción general mientras que el segundo se aboca a explayar los modelos teóricos que serán testeados por modelos computacionales. El tercer capítulo presenta la metodología computacional de algoritmos evolutivos y el capítulo 4 analiza el resultado de los experimentos computacionales. Finalmente, la tesis concluye con el capítulo 5. Como síntesis del trabajo, los algoritmos evolutivos fueron exitosos en reproducir los resultados extremos de formación de redes (la red vacía y la red completa) pero fallaron en replicar los resultados de arquitecturas intermedias propuestas por la teoría. / The modeling of interaction between agents in economics has received particular attention in recent years. These contribu-tions have focused intially, and following the acaademic tradition, in analytically tracktable models that showed emer-gence of local interactions among individuals. This interaction can, at some point, determining different behavioral outcomes compared with an analysis in which agents behave in isolation. The modeling of this interaction in formal terms used mathematical tools that allow the translation of effects from one individual to another. Game theory, modelling the optimal behavior of an agent constrained in its actions by the strategies of other agents, is a classical analytic model for certain cases of interaction. The thesis uses agent-based approach to model interactions between agents and contrast if theoretical results can be achieved through an evolutionary computational model.Their results are not formal but numeri-cal. Their background is not mathematical but basically computing. The agents appear under this framework of re-search as program objects. Agent as objects have sates and rules of behavior.Running an agent model includes te initia-lization of the population of agents, the subsequent process of interaction among them and monitoring the outcome. In short, running the model a sufficient number of times, if it is stochastic, is all that is needed to "solve" it. The thesis consists of 5 chapters. The first is a general intoduction while in the second es summarized the theoretical models that are going to be tested by computer models. The third chapter presents the computational methodology of evolutionary algorithms and chapter 4 discusses the results of computa-tional experiments. Finally, the thesis concludes with Chapter 5. As a summary of the work, evolutionary algorithms were successful in reproducing the extreme results (the empty network and the complete network) of network formation games but failed to replicate the results of inter-mediate architectures proposed by the theory.
2

Jogos de Steiner / Steiner Games

Machado, César Gamboa 11 May 2012 (has links)
Neste projeto analisamos jogos de formação de redes que são variantes do problema da floresta de Steiner, nos quais indivíduos desejam conectar conjuntos de vértices terminais em um grafo de forma a minimizar seus custos, podendo dividir o custo das arestas com os demais participantes. Estudamos como o método de divisão de custos influencia na existência e na qualidade dos equilíbrios desses jogos em comparação com o valor da solução ótima centralizada. / In this project we analyze network formation games that are variants of the Steiner forest problem, in which individuals wish to connect sets of terminal vertices of a graph in a way that minimizes their costs, being able to divide the cost of an edge with the other participants. We study how the method used to divide the costs influences the existence and quality of the equilibria of these games in relation to the centralized optimal solution.
3

Jogos de Steiner / Steiner Games

César Gamboa Machado 11 May 2012 (has links)
Neste projeto analisamos jogos de formação de redes que são variantes do problema da floresta de Steiner, nos quais indivíduos desejam conectar conjuntos de vértices terminais em um grafo de forma a minimizar seus custos, podendo dividir o custo das arestas com os demais participantes. Estudamos como o método de divisão de custos influencia na existência e na qualidade dos equilíbrios desses jogos em comparação com o valor da solução ótima centralizada. / In this project we analyze network formation games that are variants of the Steiner forest problem, in which individuals wish to connect sets of terminal vertices of a graph in a way that minimizes their costs, being able to divide the cost of an edge with the other participants. We study how the method used to divide the costs influences the existence and quality of the equilibria of these games in relation to the centralized optimal solution.

Page generated in 0.1126 seconds