Dynamic Programming Approaches for Estimating and Applying Large-scale Discrete Choice Models

People go through their life making all kinds of decisions, and some of these decisions affect their demand for transportation, for example, their choices of where to live and where to work, how and when to travel and which route to take. Transport related choices are typically time dependent and characterized by large number of alternatives that can be spatially correlated. This thesis deals with models that can be used to analyze and predict discrete choices in large-scale networks. The proposed models and methods are highly relevant for, but not limited to, transport applications. We model decisions as sequences of choices within the dynamic discrete choice framework, also known as parametric Markov decision processes. Such models are known to be difficult to estimate and to apply to make predictions because dynamic programming problems need to be solved in order to compute choice probabilities. In this thesis we show that it is possible to explore the network structure and the flexibility of dynamic programming so that the dynamic discrete choice modeling approach is not only useful to model time dependent choices, but also makes it easier to model large-scale static choices. The thesis consists of seven articles containing a number of models and methods for estimating, applying and testing large-scale discrete choice models. In the following we group the contributions under three themes: route choice modeling, large-scale multivariate extreme value (MEV) model estimation and nonlinear optimization algorithms. Five articles are related to route choice modeling. We propose different dynamic discrete choice models that allow paths to be correlated based on the MEV and mixed logit models. The resulting route choice models become expensive to estimate and we deal with this challenge by proposing innovative methods that allow to reduce the estimation cost. For example, we propose a decomposition method that not only opens up for possibility of mixing, but also speeds up the estimation for simple logit models, which has implications also for traffic simulation. Moreover, we compare the utility maximization and regret minimization decision rules, and we propose a misspecification test for logit-based route choice models. The second theme is related to the estimation of static discrete choice models with large choice sets. We establish that a class of MEV models can be reformulated as dynamic discrete choice models on the networks of correlation structures. These dynamic models can then be estimated quickly using dynamic programming techniques and an efficient nonlinear optimization algorithm. Finally, the third theme focuses on structured quasi-Newton techniques for estimating discrete choice models by maximum likelihood. We examine and adapt switching methods that can be easily integrated into usual optimization algorithms (line search and trust region) to accelerate the estimation process. The proposed dynamic discrete choice models and estimation methods can be used in various discrete choice applications. In the area of big data analytics, models that can deal with large choice sets and sequential choices are important. Our research can therefore be of interest in various demand analysis applications (predictive analytics) or can be integrated with optimization models (prescriptive analytics). Furthermore, our studies indicate the potential of dynamic programming techniques in this context, even for static models, which opens up a variety of future research directions.

Modelos autorregressivos com valores inteiros não negativos : aplicação em séries económicas de Cabo Verde

No estudo de séries temporais, os processos estocásticos usuais assumem que as distribuições marginais são contínuas e, em geral, não são adequados para modelar séries de contagem, pois as suas características não lineares colocam alguns problemas estatísticos, principalmente na estimação dos parâmetros. Assim, investigou-se metodologias apropriadas de análise e modelação de séries com distribuições marginais discretas. Neste contexto, Al-Osh and Alzaid (1987) e McKenzie (1988) introduziram na literatura a classe dos modelos autorregressivos com valores inteiros não negativos, os processos INAR. Estes modelos têm sido frequentemente tratados em artigos científicos ao longo das últimas décadas, pois a sua importância nas aplicações em diversas áreas do conhecimento tem despertado um grande interesse no seu estudo. Neste trabalho, após uma breve revisão sobre séries temporais e os métodos clássicos para a sua análise, apresentamos os modelos autorregressivos de valores inteiros não negativos de primeira ordem INAR (1) e a sua extensão para uma ordem p, as suas propriedades e alguns métodos de estimação dos parâmetros nomeadamente, o método de Yule-Walker, o método de Mínimos Quadrados Condicionais (MQC), o método de Máxima Verosimilhança Condicional (MVC) e o método de Quase Máxima Verosimilhança (QMV). Apresentamos também um critério automático de seleção de ordem para modelos INAR, baseado no Critério de Informação de Akaike Corrigido, AICC, um dos critérios usados para determinar a ordem em modelos autorregressivos, AR. Finalmente, apresenta-se uma aplicação da metodologia dos modelos INAR em dados reais de contagem relativos aos setores dos transportes marítimos e atividades de seguros de Cabo Verde.

The use of mathematical modelling in fire safety engineering

Fire is a form of uncontrolled combustion which generates heat, smoke, toxic and irritant gases. All of these products are harmful to man and account for the heavy annual cost of 800 lives and £1,000,000,000 worth of property damage in Britain alone. The new discipline of Fire Safety Engineering has developed as a means of reducing these unacceptable losses. One of the main tools of Fire Safety Engineering is the mathematical model and over the past 15 years a number of mathematical models have emerged to cater for the needs of this discipline. Part of the difficulty faced by the Fire Safety Engineer is the selection of the most appropriate modelling tool to use for the job. To make an informed choice it is essential to have a good understanding of the various modelling approaches, their capabilities and limitations. In this paper some of the fundamental modelling tools used to predict fire and evacuation are investigated as are the issues associated with their use and recent developments in modelling technology.

Coupled diusion and mechanics in battery electrodes

Thesis (Ph.D.)--University of Washington, 2016-08

Vitamines B, éléments clés de l'efficacité métabolique : Effets de la nature de la diète sur les apports

Les travaux sur la nutrition en vitamines B des ruminants montrent des résultats très variés sur les quantités de ces nutriments disponibles pour l’animal selon la nature de la ration. Ces divergences sont dues à des changements des populations microbiennes dans le rumen, causées par les facteurs physico-chimiques de la ration. Une amélioration de la compréhension des effets de la nature de la diète sur la synthèse et l’utilisation des vitamines B dans le rumen pourrait aider à identifier les conditions sous lesquelles une supplémentation en ces vitamines serait bénéfique pour la vache. Le but de ce travail de thèse est donc d’améliorer la compréhension des effets de l’espèce fourragère, de la maturité et de la longueur des particules de fourrage sur les apports en vitamines B chez la vache laitière. Pour évaluer chacune de ces variables, les concentrations de thiamine, riboflavine, niacine, vitamine B6, folates et vitamine B12 ont été mesurées dans les échantillons d’aliments et de digesta duodénal recueillis lors de trois projets réalisés à l’Université du Michigan par l’équipe du Dr. M. Allen. Dans la première étude, l’effet de l’espèce fourragère des ensilages a été évalué au cours de deux expériences similaires durant lesquelles les vaches recevaient une diète à base d’ensilage de luzerne ou de dactyle. Les diètes à base de luzerne ont été associées à une augmentation de la dégradation de la thiamine et de la vitamine B6 dans le rumen par rapport aux diètes à base d’ensilage de dactyle. La deuxième étude visait à évaluer les effets de la maturité des plantes lors de la mise en silo sur les quantités de vitamines B disponibles pour la vache; les deux expériences se différenciaient par l’espèce fourragère étudiée, soit la luzerne ou le dactyle. Une récolte à un stade de maturité plus élevé a augmenté les flux duodénaux de thiamine, de niacine et de folates lorsque les vaches recevaient des diètes à base d’ensilage de luzerne mais n’a diminué que le flux duodénal de riboflavine chez les animaux recevant des diètes à base d’ensilage de dactyle. La troisième étude a comparé les effets de la longueur de coupe (10 vs. 19 mm) d’ensilages de luzerne et de dactyle sur le devenir des vitamines B dans le système digestif de la vache laitière. Cette étude a permis de constater qu’une augmentation du temps de séchage au champ diminuait les concentrations de vitamines B dans les ensilages. Cependant, la taille des particules des ensilages de luzerne et de dactyle n’a pas affecté les quantités des vitamines B arrivant au duodénum des vaches. En général, les résultats de ces études montrent qu’il existe une corrélation négative entre la synthèse de riboflavine, de niacine et de vitamine B6 et leur ingestion, suggérant une possible régulation de la quantité de ces vitamines B par les microorganismes du rumen. De plus, l’ingestion d’amidon et d’azote a été corrélée positivement avec la synthèse de thiamine, de folates et de vitamine B12, et négativement avec la synthèse de niacine. Ces corrélations suggèrent que les microorganismes qui utilisent préférentiellement l’amidon jouent un rôle majeur pour la synthèse ou la dégradation de ces vitamines. De plus, la présence d’une quantité suffisante d’azote semble avoir un impact majeur sur ces processus. La suite de ces travaux devrait viser la modélisation de ces données afin de mieux appréhender la physiologie de la digestion de ces vitamines et permettre la création de modèles mathématiques capables de prédire les quantités de vitamines disponibles pour les vaches. Ces modèles permettront, lorsqu’intégrés aux logiciels de formulation de ration, d’élaborer une diète plus précise, ce qui améliorera la santé du troupeau et la performance laitière et augmentera les profits du producteur.

Asymptotic and structural properties of special cases of the Wright function arising in probability theory

This analysis paper presents previously unknown properties of some special cases of the Wright function whose consideration is necessitated by our work on probability theory and the theory of stochastic processes. Specifically, we establish new asymptotic properties of the particular Wright function 1Ψ1(ρ, k; ρ, 0; x) = X∞ n=0 Γ(k + ρn) Γ(ρn) x n n! (|x| < ∞) when the parameter ρ ∈ (−1, 0)∪(0, ∞) and the argument x is real. In the probability theory applications, which are focused on studies of the Poisson-Tweedie mixtures, the parameter k is a non-negative integer. Several representations involving well-known special functions are given for certain particular values of ρ. The asymptotics of 1Ψ1(ρ, k; ρ, 0; x) are obtained under numerous assumptions on the behavior of the arguments k and x when the parameter ρ is both positive and negative. We also provide some integral representations and structural properties involving the ‘reduced’ Wright function 0Ψ1(−−; ρ, 0; x) with ρ ∈ (−1, 0) ∪ (0, ∞), which might be useful for the derivation of new properties of members of the power-variance family of distributions. Some of these imply a reflection principle that connects the functions 0Ψ1(−−;±ρ, 0; ·) and certain Bessel functions. Several asymptotic relationships for both particular cases of this function are also given. A few of these follow under additional constraints from probability theory results which, although previously available, were unknown to analysts.

Modeling of burden distribution in the blast furnace

The blast furnace is the main ironmaking production unit in the world which converts iron ore with coke and hot blast into liquid iron, hot metal, which is used for steelmaking. The furnace acts as a counter-current reactor charged with layers of raw material of very different gas permeability. The arrangement of these layers, or burden distribution, is the most important factor influencing the gas flow conditions inside the furnace, which dictate the efficiency of the heat transfer and reduction processes. For proper control the furnace operators should know the overall conditions in the furnace and be able to predict how control actions affect the state of the furnace. However, due to high temperatures and pressure, hostile atmosphere and mechanical wear it is very difficult to measure internal variables. Instead, the operators have to rely extensively on measurements obtained at the boundaries of the furnace and make their decisions on the basis of heuristic rules and results from mathematical models. It is particularly difficult to understand the distribution of the burden materials because of the complex behavior of the particulate materials during charging. The aim of this doctoral thesis is to clarify some aspects of burden distribution and to develop tools that can aid the decision-making process in the control of the burden and gas distribution in the blast furnace. A relatively simple mathematical model was created for simulation of the distribution of the burden material with a bell-less top charging system. The model developed is fast and it can therefore be used by the operators to gain understanding of the formation of layers for different charging programs. The results were verified by findings from charging experiments using a small-scale charging rig at the laboratory. A basic gas flow model was developed which utilized the results of the burden distribution model to estimate the gas permeability of the upper part of the blast furnace. This combined formulation for gas and burden distribution made it possible to implement a search for the best combination of charging parameters to achieve a target gas temperature distribution. As this mathematical task is discontinuous and non-differentiable, a genetic algorithm was applied to solve the optimization problem. It was demonstrated that the method was able to evolve optimal charging programs that fulfilled the target conditions. Even though the burden distribution model provides information about the layer structure, it neglects some effects which influence the results, such as mixed layer formation and coke collapse. A more accurate numerical method for studying particle mechanics, the Discrete Element Method (DEM), was used to study some aspects of the charging process more closely. Model charging programs were simulated using DEM and compared with the results from small-scale experiments. The mixed layer was defined and the voidage of mixed layers was estimated. The mixed layer was found to have about 12% less voidage than layers of the individual burden components. Finally, a model for predicting the extent of coke collapse when heavier pellets are charged over a layer of lighter coke particles was formulated based on slope stability theory, and was used to update the coke layer distribution after charging in the mathematical model. In designing this revision, results from DEM simulations and charging experiments for some charging programs were used. The findings from the coke collapse analysis can be used to design charging programs with more stable coke layers.

An approach to emotion recognition in single-channel EEG signals: a mother child interaction

In this work, we perform a first approach to emotion recognition from EEG single channel signals extracted in four (4) mother-child dyads experiment in developmental psychology -- Single channel EEG signals are analyzed and processed using several window sizes by performing a statistical analysis over features in the time and frequency domains -- Finally, a neural network obtained an average accuracy rate of 99% of classification in two emotional states such as happiness and sadness

Ecological and evolutionary effects of dispersal on freshwater zooplankton

A recent focus on contemporary evolution and the connections between communities has sought to more closely integrate the fields of ecology and evolutionary biology. Studies of coevolutionary dynamics, life history evolution, and rapid local adaptation demonstrate that ecological circumstances can dictate evolutionary trajectories. Thus, variation in species identity, trait distributions, and genetic composition may be maintained among ecologically divergent habitats. New theories and hypotheses (e.g., metacommunity theory and the Monopolization hypothesis) have been developed to understand better the processes occurring in spatially structured environments and how the movement of individuals among habitats contributes to ecology and evolution at broader scales. As few empirical studies of these theories exist, this work seeks to further test these concepts. Spatial and temporal dispersal are the mechanisms that connect habitats to one another. Both processes allow organisms to leave conditions that are suboptimal or unfavorable, and enable colonization and invasion, species range expansion, and gene flow among populations. Freshwater zooplankton are aquatic crustaceans that typically develop resting stages as part of their life cycle. Their dormant propagules allow organisms to disperse both temporally and among habitats. Additionally, because a number of species are cyclically parthenogenetic, they make excellent model organisms for studying evolutionary questions in a controlled environment. Here, I use freshwater zooplankton communities as model systems to explore the mechanisms and consequences of dispersal and to test these nascent theories on the influence of spatial structure in natural systems. In Chapter one, I use field experiments and mathematical models to determine the range of adult zooplankton dispersal over land and what vectors are moving zooplankton. Chapter two focuses on prolonged dormancy of one aquatic zooplankter, Daphnia pulex. Using statistical models with field and mesocosm experiments, I show that variation in Daphnia dormant egg hatching is substantial among populations in nature, and some of that variation can be attributed to genetic differences among the populations. Chapters three and four explore the consequences of dispersal at multiple levels of biological organization. Chapter three seeks to understand the population level consequences of dispersal over evolutionary time on current patterns of population genetic differentiation. Nearby populations of D. pulex often exhibit high population genetic differentiation characteristic of very low dispersal. I explore two alternative hypotheses that seek to explain this pattern. Finally, chapter four is a case study of how dispersal has influenced patterns of variation at the community, trait and genetic levels of biodiversity in a lake metacommunity.

Spectral graph theory with applications to quantum adiabatic optimization

In this dissertation I draw a connection between quantum adiabatic optimization, spectral graph theory, heat-diffusion, and sub-stochastic processes through the operators that govern these processes and their associated spectra. In particular, we study Hamiltonians which have recently become known as ``stoquastic'' or, equivalently, the generators of sub-stochastic processes. The operators corresponding to these Hamiltonians are of interest in all of the settings mentioned above. I predominantly explore the connection between the spectral gap of an operator, or the difference between the two lowest energies of that operator, and certain equilibrium behavior. In the context of adiabatic optimization, this corresponds to the likelihood of solving the optimization problem of interest. I will provide an instance of an optimization problem that is easy to solve classically, but leaves open the possibility to being difficult adiabatically. Aside from this concrete example, the work in this dissertation is predominantly mathematical and we focus on bounding the spectral gap. Our primary tool for doing this is spectral graph theory, which provides the most natural approach to this task by simply considering Dirichlet eigenvalues of subgraphs of host graphs. I will derive tight bounds for the gap of one-dimensional, hypercube, and general convex subgraphs. The techniques used will also adapt methods recently used by Andrews and Clutterbuck to prove the long-standing ``Fundamental Gap Conjecture''.

Dynamic Programming Approaches for Estimating and Applying Large-scale Discrete Choice Models

People go through their life making all kinds of decisions, and some of these decisions affect their demand for transportation, for example, their choices of where to live and where to work, how and when to travel and which route to take. Transport related choices are typically time dependent and characterized by large number of alternatives that can be spatially correlated. This thesis deals with models that can be used to analyze and predict discrete choices in large-scale networks. The proposed models and methods are highly relevant for, but not limited to, transport applications. We model decisions as sequences of choices within the dynamic discrete choice framework, also known as parametric Markov decision processes. Such models are known to be difficult to estimate and to apply to make predictions because dynamic programming problems need to be solved in order to compute choice probabilities. In this thesis we show that it is possible to explore the network structure and the flexibility of dynamic programming so that the dynamic discrete choice modeling approach is not only useful to model time dependent choices, but also makes it easier to model large-scale static choices. The thesis consists of seven articles containing a number of models and methods for estimating, applying and testing large-scale discrete choice models. In the following we group the contributions under three themes: route choice modeling, large-scale multivariate extreme value (MEV) model estimation and nonlinear optimization algorithms. Five articles are related to route choice modeling. We propose different dynamic discrete choice models that allow paths to be correlated based on the MEV and mixed logit models. The resulting route choice models become expensive to estimate and we deal with this challenge by proposing innovative methods that allow to reduce the estimation cost. For example, we propose a decomposition method that not only opens up for possibility of mixing, but also speeds up the estimation for simple logit models, which has implications also for traffic simulation. Moreover, we compare the utility maximization and regret minimization decision rules, and we propose a misspecification test for logit-based route choice models. The second theme is related to the estimation of static discrete choice models with large choice sets. We establish that a class of MEV models can be reformulated as dynamic discrete choice models on the networks of correlation structures. These dynamic models can then be estimated quickly using dynamic programming techniques and an efficient nonlinear optimization algorithm. Finally, the third theme focuses on structured quasi-Newton techniques for estimating discrete choice models by maximum likelihood. We examine and adapt switching methods that can be easily integrated into usual optimization algorithms (line search and trust region) to accelerate the estimation process. The proposed dynamic discrete choice models and estimation methods can be used in various discrete choice applications. In the area of big data analytics, models that can deal with large choice sets and sequential choices are important. Our research can therefore be of interest in various demand analysis applications (predictive analytics) or can be integrated with optimization models (prescriptive analytics). Furthermore, our studies indicate the potential of dynamic programming techniques in this context, even for static models, which opens up a variety of future research directions.

Co-evolution of friendships and antipathies: A longitudinal study of preschool peer groups

Material suplementar está disponível em: http://journal.frontiersin.org/article/10.3389/fpsyg. 2016.01509

Modelos autorregressivos com valores inteiros não negativos : aplicação em séries económicas de Cabo Verde

No estudo de séries temporais, os processos estocásticos usuais assumem que as distribuições marginais são contínuas e, em geral, não são adequados para modelar séries de contagem, pois as suas características não lineares colocam alguns problemas estatísticos, principalmente na estimação dos parâmetros. Assim, investigou-se metodologias apropriadas de análise e modelação de séries com distribuições marginais discretas. Neste contexto, Al-Osh and Alzaid (1987) e McKenzie (1988) introduziram na literatura a classe dos modelos autorregressivos com valores inteiros não negativos, os processos INAR. Estes modelos têm sido frequentemente tratados em artigos científicos ao longo das últimas décadas, pois a sua importância nas aplicações em diversas áreas do conhecimento tem despertado um grande interesse no seu estudo. Neste trabalho, após uma breve revisão sobre séries temporais e os métodos clássicos para a sua análise, apresentamos os modelos autorregressivos de valores inteiros não negativos de primeira ordem INAR (1) e a sua extensão para uma ordem p, as suas propriedades e alguns métodos de estimação dos parâmetros nomeadamente, o método de Yule-Walker, o método de Mínimos Quadrados Condicionais (MQC), o método de Máxima Verosimilhança Condicional (MVC) e o método de Quase Máxima Verosimilhança (QMV). Apresentamos também um critério automático de seleção de ordem para modelos INAR, baseado no Critério de Informação de Akaike Corrigido, AICC, um dos critérios usados para determinar a ordem em modelos autorregressivos, AR. Finalmente, apresenta-se uma aplicação da metodologia dos modelos INAR em dados reais de contagem relativos aos setores dos transportes marítimos e atividades de seguros de Cabo Verde.

Phase transitions in gene networks evolved under different selection rules

Mathematical models of gene regulation are a powerful tool for understanding the complex features of genetic control. While various modeling efforts have been successful at explaining gene expression dynamics, much less is known about how evolution shapes the structure of these networks. An important feature of gene regulatory networks is their stability in response to environmental perturbations. Regulatory systems are thought to have evolved to exist near the transition between stability and instability, in order to have the required stability to environmental fluctuations while also being able to achieve a wide variety of functions (corresponding to different dynamical patterns). We study a simplified model of gene network evolution in which links are added via different selection rules. These growth models are inspired by recent work on `explosive' percolation which shows that when network links are added through competitive rather than random processes, the connectivity phase transition can be significantly delayed, and when it is reached, it appears to be first order (discontinuous, e.g., going from no failure at all to large expected failure) instead of second order (continuous, e.g., going from no failure at all to very small expected failure). We find that by modifying the traditional framework for networks grown via competitive link addition to capture how gene networks evolve to avoid damage propagation, we also see significant delays in the transition that depend on the selection rules, but the transitions always appear continuous rather than `explosive'.

Synthetic Circuits for Feedback and Detection in Bacteria

Synthetic biology, by co-opting molecular machinery from existing organisms, can be used as a tool for building new genetic systems from scratch, for understanding natural networks through perturbation, or for hybrid circuits that piggy-back on existing cellular infrastructure. Although the toolbox for genetic circuits has greatly expanded in recent years, it is still difficult to separate the circuit function from its specific molecular implementation. In this thesis, we discuss the function-driven design of two synthetic circuit modules, and use mathematical models to understand the fundamental limits of circuit topology versus operating regimes as determined by the specific molecular implementation. First, we describe a protein concentration tracker circuit that sets the concentration of an output protein relative to the concentration of a reference protein. The functionality of this circuit relies on a single negative feedback loop that is implemented via small programmable protein scaffold domains. We build a mass-action model to understand the relevant timescales of the tracking behavior and how the input/output ratios and circuit gain might be tuned with circuit components. Second, we design an event detector circuit with permanent genetic memory that can record order and timing between two chemical events. This circuit was implemented using bacteriophage integrases that recombine specific segments of DNA in response to chemical inputs. We simulate expected population-level outcomes using a stochastic Markov-chain model, and investigate how inferences on past events can be made from differences between single-cell and population-level responses. Additionally, we present some preliminary investigations on spatial patterning using the event detector circuit as well as the design of stationary phase promoters for growth-phase dependent activation. These results advance our understanding of synthetic gene circuits, and contribute towards the use of circuit modules as building blocks for larger and more complex synthetic networks.