No description available.

Systemic risk is a complex topic, with a large number of variables and constraints. In this thesis we introduce an agent-based network to study the effects of financial shocks on the financial network. The model takes into consideration the repurchase agreement (repo) market and rehypothecation. We introduce a financial network consisting of financial agents who are connected through direct channels (bilateral contracts) and indirect channels (markets). Each fi- nancial agent has a balance sheet with liquid assets (cash), collateral (bonds, shares), reverse repo assets, fixed assets (loans and mortgages) on the asset side and repo loans, deposits and equities on the liability side. Agents (i.e., banks) need to satisfy constraints on (i) liquidity, which deals with financial shocks, (ii) collateral, related to repo liabil- ities, rehypothecation, and (iii) solvency constraints, ensuring that equity is positive. Liquidity constrain can be broken by a financial shock (e.g., a bank run), while the collateral constraint can be broken by hoarding credit and collateral price reduction. When liquidity and collateral constraints are broken the financial agent will try to fix them through recalling reverse repos and firesale of fixed assets. Banks that fail to fix their constraints by the end of the day will be considered defaulted. We introduce netting and novation techniques to deal with defaulted banks and lower the stress on the financial markets. In the netting step we lower the exposure of financial agents by removing cycles in the repo liabilities between banks, while in the novation we redistribute the ownership of bilateral contracts and settle any residuals that are left. We also establish that, under certain conditions on the set of defaulted banks, that the novation step is order indifferent. Different network topologies and balance sheet compositions are tested under several financial shocks to check the robustness of the financial network under our framework. / Thesis / Doctor of Philosophy (PhD)

The process of enrolling and completing the courses needed to earn an undergraduate degree involves complex interactions between individual students and institutional policies and procedures, especially because student and institutional priorities do not always align. Traditional social and behavioral statistical methods are ineffective for modeling these interactions. Simulation and algorithm-based modeling approaches have been underutilized in higher education, but their adaptability can accommodate the complexity of the degree attainment process. The purpose of this research was to design, develop, validate, and apply a multi-method Course Enrollment Simulation Model (CESM), which mirrored the process of college students enrolling in courses required for a specific undergraduate degree program. Simulated output from the model included graduation outcomes, like six-year graduation rates and average terms for students to obtain the degree, which are metrics commonly tracked by institutions of higher education. As proof of concept, data from a Fall 2015 student cohort and graduation requirements for an undergraduate computer science program at a large public university were used to create and test the CESM. The model integrated elements of discrete event simulation, agent-based modeling, and microsimulation methods into one architecture. Monte Carlo experiments were used to assess the validity of the model, which was more accurate than comparable inferential statistics. Finally, the CESM was used to evaluate summer enrollment policy options intended to improve graduation outcomes in a computer science program, finding that the effectiveness of the proposed policies depended on student factors as well as course requirements of the degree program. This dissertation is formatted as a collection of three studies, each organized into a publishable manuscript.

As democracias estão fundamentadas na conexão entre atitudes dos cidadãos e políticas. Numa arena política onde agentes têm racionalidade limitada, a competição dos políticos e partidos pelo apoio e voto da população depende de como está distribuída a opinião pública. Contudo, modelos explícitos que gerem essas distribuições, ao invés de pressupô-las, ainda são escassos. O presente trabalho gera estas distribuições por meio de um modelo computacional baseado em agentes de influência social. Nele os agentes interagem em díades e tornam-se mais próximos dos outros agentes em alguma questão a depender da distância de seus posicionamentos ideológicos, da incerteza dos agentes e do quanto confiam na opinião do outro. Por meio da combinação de análise de sensibilidade, histogramas e séries temporais de medidas do sistema e gráficos de dispersão, encontramos que o quão incertos são os agentes, o número de questões subjacentes ao seu posicionamento e o quanto estão sujeitos a mudanças aleatórias de atitude (ruído) são determinantes importantes do formato da distribuição de atitudes populacional. Nossos resultados nos levam a questionamentos empíricos quanto a interação dos parâmetros do modelo e a plausibilidade das distribuições geradas por ele a serem respondidos por trabalhos futuros / Democracies are based on the connection between citizens\' attitudes and policies. In a political arena where agents are boundedly rational, the competition of politicians and parties for the support and vote of the population depends on how the public opinion is distributed. However, explicit models that generate, instead of presupposing, them are still rare. This work generates those distributions through a computational agent-based model of social influence. In it the agents interact in pairs and become more similar to the others in some issue conditional on their ideological positioning, their uncertainty and on how much they trust another agents opinion. Through a combination of sensitivity analysis, histograms and time series of the systems measures and scatterplots, we find that how uncertain the agents are, the number of issues underneath their positioning and how much theyre subject to random attitude change (noise) are important determinants of the shape of the population attitudinal distribution. Our results lead us to empirical questions regarding the model parameters interaction and the plausibility of the distributions generated by it which should be answered by future works

Este trabalho adota a técnica de simulação baseada em agentes para analisar o processo de precificação de energia comercializada no ambiente de contratação regulada (leilões) do mercado elétrico do Brasil. Nesse contexto, são simulados leilões de energia nova - produto fonte eólica. O simulador dos leilões de energia do Brasil foi construído através de programação realizada em matlab. Nesse programa, é possível comparar a escolha de lances de vendedores participantes nos pleitos. Os agentes (vendedores) participantes dos leilões aprendem com o decorrer dos leilões simulados. A aprendizagem é determinada através da utilização de variação do algoritmo Q-learning. Os resultados claramente demonstram que as técnicas de aprendizagem consideradas têm resultados mais favoráveis do que escolhas aleatórias (sem aprendizagem). Considerando outro ponto de vista, é possível verificar diferença de média de preços nos leilões entre os perfis de geradores públicos e privados. Além disso, é possível afirmar que o preço da energia se altera dada alteração na participação relativa de vendedores públicos ou privados nos pleitos. / This thesis adopts an agent based simulation in order to analyses the pricing process of energy that is negotiate by auctions in Brazil. In this work, wind energy power auctions are simulate. The model was developed in Matlab platform, and so, it was possible to compare the bidding process of the players in those auctions. The players learn during the auctions, and the process of learning is defined by a variation of the Q-learning algorithm. The results of the research show that when Q-learning is considered by generators there are more benefits than it is not. From another point of view, it is possible to say that there is difference between the prices of public and private players (enterprises that sell Wind energy). Besides it is possible to say that when the number of public and private players in an auction change the energy price.

As democracias estão fundamentadas na conexão entre atitudes dos cidadãos e políticas. Numa arena política onde agentes têm racionalidade limitada, a competição dos políticos e partidos pelo apoio e voto da população depende de como está distribuída a opinião pública. Contudo, modelos explícitos que gerem essas distribuições, ao invés de pressupô-las, ainda são escassos. O presente trabalho gera estas distribuições por meio de um modelo computacional baseado em agentes de influência social. Nele os agentes interagem em díades e tornam-se mais próximos dos outros agentes em alguma questão a depender da distância de seus posicionamentos ideológicos, da incerteza dos agentes e do quanto confiam na opinião do outro. Por meio da combinação de análise de sensibilidade, histogramas e séries temporais de medidas do sistema e gráficos de dispersão, encontramos que o quão incertos são os agentes, o número de questões subjacentes ao seu posicionamento e o quanto estão sujeitos a mudanças aleatórias de atitude (ruído) são determinantes importantes do formato da distribuição de atitudes populacional. Nossos resultados nos levam a questionamentos empíricos quanto a interação dos parâmetros do modelo e a plausibilidade das distribuições geradas por ele a serem respondidos por trabalhos futuros / Democracies are based on the connection between citizens\' attitudes and policies. In a political arena where agents are boundedly rational, the competition of politicians and parties for the support and vote of the population depends on how the public opinion is distributed. However, explicit models that generate, instead of presupposing, them are still rare. This work generates those distributions through a computational agent-based model of social influence. In it the agents interact in pairs and become more similar to the others in some issue conditional on their ideological positioning, their uncertainty and on how much they trust another agents opinion. Through a combination of sensitivity analysis, histograms and time series of the systems measures and scatterplots, we find that how uncertain the agents are, the number of issues underneath their positioning and how much theyre subject to random attitude change (noise) are important determinants of the shape of the population attitudinal distribution. Our results lead us to empirical questions regarding the model parameters interaction and the plausibility of the distributions generated by it which should be answered by future works

The ant colony, known as a self-organized system, can adapt to the environment by a series of negative and positive feedbacks. There is still a lack of mechanistic understanding of how the factors, such as temperature and food, coordinate the labor of ants. According to the Metabolic Theory of Ecology (MTE), the metabolic rate could control ecological process at all levels. To analyze self-organized process of ant colony, we constructed an agent-based model to simulate the energy and population dynamics of ant colony. After parameterizing the model, we ran 20 parallel simulations for each experiment and parameter sweeps to find patterns and dependencies in the food and energy flow of the colony. Ultimately this model predicted that ant colonies can respond to changes of temperature and food availability and perform differently. We hope this study can improve our understanding on the self-organized process of ant colony.

No description available.

Sex ratio theory suggests that the strength of intersexual selection will increase as a population more male-biased; reflecting increased selectivity in mate choice. Populations of pond turtle have varying adult sex ratios, in painted turtles (Chrysemys picta), reported sex ratios range from female biased (1:3) to male biased (3:1). I investigated the effect of sex ratio on male reproductive success (quantified as “fertilization success”) in painted turtles. I examined the mating system of painted turtles in a female-biased population using microsatellite paternity analysis, relating variation in male fertilization success to male phenotype and offspring survival, employed ex situ behavioural observation to clarify the mechanism behind the variation in male fertilization success and used agent-based modeling to simulate the effects of changing sex ratio, population density and proportions of male phenotype on male fertilization success. Small males contributed sperm to a greater number of clutches than did larger males, but were not more likely to reproduce in a season than larger males. There was no offspring fitness advantage related to male body size and no relationship between male claw length and fertilization success. Large male painted turtles courted at a higher frequency than small males. I found no relationship between male courtship behavior and claw length. Females showed no preference for males of any phenotype. Agent-based simulations were based on the distribution of best fit from the observed data; an amalgam of two Poisson distributions, each with its own probability of success and proportional representation in the final distribution. Increased female sex ratio bias, increased population density and increased proportions of “more successful” males all increased the mean and variance of male fertilization success, based on increased encounter rate among turtles. Small and large male painted turtles enjoy different fertilization success. It is uncertain whether this difference is based on active female choice, cryptic female choice, sperm competition or a combination of factors. Sex ratio simulations predict the opposite result as that predicted by sex ratio theory. These contrary results should be compared to simulations manipulating choosiness and field data from painted turtle populations to clarify mechanisms influencing male reproductive success.

This D.Phil. thesis develops a new agent-based simulation model to improve the results of analysis, which solely uses discrete choice modelling, as well as to analyse the effects of a road user charging scheme for the Upper Derwent Valley in the Peak District National Park. The advantages of discrete choice analysis are well known. However, results with these conventional conventional approaches, which conduct analysis solely with discrete choice models, can be biased if interaction and learning effects are significant. The Minority Game, in which agents try to choose the option of the minority side, is an appropriate tool to deal with these problems. The situation in the Upper Derwent Valley can be explained with economic game theories and the Minority Game. The two approaches mutually help to analyse the situation in the Upper Derwent Valley leading to the development of a stochastic Minority Game. The stochastic Minority Game was tested with an online game (questionnaire), which was played 3,886 times by response in all around the world. The practical part of this thesis examines the components of the stochastic Minority Game with the data collected around the Upper Derwent Valley. The main data was collected using a stated preference survey. Overall, 700 questionnaires were distributed and 323 of them were returned (i.e. a return rate of 46.1 %). In the practical part, the agent-based model has four sub modules: 1) Multinomial mixed logit model for mode choice, 2) Binary logit model for parking location choice, 3) Markov queue model for parking network, and 4) the Minority Game for parking congestion and learning. This simulation model produces comprehensive outputs including mode choices, congestion levels, and user utilities. The results show that the road user charging scheme reduces car demand in the Upper Derwent Valley and ensures a reduction in congestion at the parking areas. The model also shows that an exemption will increase the utilities of elderly visitors without substantially sacrificing those of younger visitors. In conclusion, the simulation model demonstrated that oversimplification in conventional approaches solely using discrete choice models gave significant biases when real world problems were analysed.