Assembling spatial units into meaningful clusters is a challenging task, as it must cope with a consequential computational complexity while controlling for the modifiable areal unit problem (MAUP), spatial autocorrelation and attribute multicollinearity. Nevertheless, we sustain that these effects can reveal significant interactions among diverse spatial phenomena, such as segregation and economic specialization, but most methods treat this apparent disorder as noise.

In order to address this issue, we have developed a hierarchical regionalization algorithm that is sensitive to scalar variations of multivariate spatial correlations, recalculating PCA scores at all aggregation steps in order to account for differences in the span of autocorrelation effects for diverse variables. In such a way, we intend to provide a method that minimizes the information loss associated with both MAUP zoning and scale effects, while providing results that allow studying the self-organization of spatial patterns avoiding arbitrary zoning decisions. This algorithm produces a hierarchical cartography, which has multiple applications, where two particular cases were studied in Santiago de Chile.

With these settings, the scalar evolution of several social distress measures is compared between empirical and 120 random datasets. Remarkably, adjusting several indicators with real and simulated data allows for a clear definition of a stopping rule for spatial hierarchical clustering. Indeed, increasing correlations with scale in random datasets are spurious MAUP effects, so they can be discounted from real data results in order to identify an optimal clustering level, as defined by the maximum of authentic spatial self-organization. This allows to single out the most socially distressed areas in Greater Santiago, thus providing relevant socio-spatial insights from their cartographic and statistical analysis, which agrees to independent diagnostics.

On the other hand, despite the abundance of works in hedonic mass appraisal, the potential of implementing hierarchical structures to market segmentation has not been fully explored. The purpose of this research is to fill this gap in the literature by studying the impact of incorporating complex architectures to predictive models, such as: econometrics models, artificial neural networks and hybrid models of combined forecasts. Our results confirm that all models exceed their predictive capability when applied in a hierarchical framework.

In sum, a useful methodology is developed to systematically explore the black box of spatial interdependence and multiscalar self-organizing phenomena, while linking these questions to relevant real world issues.

Several models have been developed to study different emergent phenomena in society. Many of them have been proposed to explain various aspects of spatial pattern formations; some others have been used to understand phenomena related to timing. All of them suppose a consequence of a particular interaction among individuals of the system.

In this thesis work, the study of spatio-temporal component of social interacting systems is analyzed. The spatial component is studied through an exhaustive computational simulation work, where individuals of at least two different groups are interacting in a two dimensional grid. Two different models were explored to study the basic formations of a bunch of social systems: the Schelling model of social segregation and the Sakoda model of social interactions. Both models explain the global formation of social structures through local interactions of individuals. Variations of different parameters of interaction ranges and movements produce many different structures. Despite of this, commonalities appear allowing to understand different aspects of these dynamics. The events timing in systems composed by two populations in conflict is studied. This study is based in the escalation and de-escalation patterns of those events.

A version of the original Schelling model was developed. The principal characteristic of this version is that there is not considered vacancy spaces where unhappy individuals may move. Instead, different from original Schelling model, two unhappy individuals of opposite states exchange its positions. The happiness of an individual is determined by the presence of less than certain number of individuals in its neighborhood of its opposite state. This number is namely the tolerance parameter and its fixed during a simulation. In addition, because of similarities with the Ising model in physics, an energy function was proposed. This energy characterizes pertinently different levels of segregation. This model was studied deeply such that, despite inherent randomness of the evolution protocols, robust behaviors and commonalities are characterized.

The Sakoda model of social interactions was deeply characterized and complemented. This model 6 Abstract 7 allows to study all possible social structures formed by the repulsive, attractive or neutral interactions among individuals of two populations. The segregation case turns out to be one special case of this model. In fact, in this thesis work, the 45 possible social structures were characterized in terms of the parameters of the interaction matrix. This matrix mandates the evolution of dynamics. A Potts like energy function is proposed to characterize different interactions. This function is a Lyapunov function in the case of symmetric interactions that quantifies different attractors in terms of the parameters of the interaction matrix and the fraction of initial individuals.

Events timing in two interacting populations were studied through the progress curves. Data across several domains were analyzed to support our approach. All of theme exhibit progress curves that escalate or de-escalate following a power-law trend. This yields two important consequences: the characterization of different conflicts, and the predictions of future events.

La investigación propuesta en esta tesis describe la formación de un sistema de comunicación compartido, en particular, una asociación entre forma y significado, en una población artificial sujeta a restricciones cognitivas. La formación de asociaciones se estudia en un entorno computacional basado en un modelo de agentes: Naming Game. En tal modelo, pares de agentes, un oyente (hearer) y un hablante (speaker), participan en interacciones cuya finalidad es alcanzar acuerdos colectivos. En esta tesis, se propone un modelo de Redes de Autómatas como un marco alternativo para estudios computacionales de la evolución del lenguaje. De esta forma, se estudian dos aspectos fundamentales mediante simulaciones por computador y herramientas de complejidad algorítmica.

En primer lugar, se describe el tiempo necesario para que la comunidad alcance acuerdos (globales o locales). En segundo lugar, se introduce parámetros que restringen las capacidades de memoria a corto plazo de los agentes. En valores críticos de estos parámetros, la población exhibe una transición drástica entre la presencia y la ausencia de consenso global. Este hecho sugiere nuevos argumentos acerca de la ausencia de etapas en la formación del lenguaje humano.

The main goal of this thesis, relies the dynamics of a reversible and conservative cellular automaton Q2R model. Q2R is a automaton that runs on a two-dimensional grid of finite size and is reversible in a physical sense, that is, not only is the automaton rule invertible, but the backward rule reads exactly the same as the forward one. This model is a dynamical variation of the Ising model for ferromagnetism that possesses quite a rich and complex dynamics.

As expected, the Q2R automaton only possesses fixed points and periodic orbits and it has been shown that possesses an energy like quantity, and, at least an extra conserved quantity. Although, the dynamics includes only fixed points and periodic orbits, numerical simulations show that the system exhibits a ferromagnetic phase transition in the large system size limit for a well defined critical energy.

In the present work, we characterize the configuration space, that is composed of a huge number of cycles with exponentially long periods. More precisely, we quantify the probability distribution functions of states in terms of the aforementioned invariants. We show that the dynamics of the system in the phase space appears to be, depending on the energy, a random walk or a Levy flight.

The main contribution of the present thesis is the application of a coarse-graining approach that allows to write a coarse-grained master equation, which characterizes equilibrium and non equilibrium statistical properties, for the Q2R model. Following Nicolis and collaborators, a coarse-graining approach is applied to the time series of the total magnetization, leading to a consistent master equation that governs the macroscopic irreversible dynamics of the Q2R automata. The methodology is replicated for various lattice sizes. In the case of small systems, we show that the master equation leads to a tractable probability transfer matrix of moderate size, which provides a master equation for a coarse-grained probability distribution. The method is validated and some explicit examples are discussed.

Societies, earth’s global changes, the human brain, ecosystems and social or economic organizations may be considered as complex systems, which present emergent properties, uncertainty dynamics, non-linearities and macro interactions; as a consequence of the interdependency of several and diverse elements. The terms “complex systems” can be used to state systems with complex characteristics, and also, the set of integrated methodologies used to address problems arisen in these types of systems. In this work, we review both, the characteristics of the complex systems and the methodologies to address them, through an interdisciplinary and top-down approach.

Mainly, we focus in the analysis of the segregation, as a natural consequence of an homophilic and spatial process. The segregation is presented as an emergent phenomenon of complex systems that may produce negative particular consequences in the heterogeneous societies. At the basis of simulations, mathematical and mechanical statistic models, we shall contribute with another point of view to understand the quantitative segregation measures and the evolutionary dynamics of this important complex problem. In this work, we characterize quantitatively the temporal evolution of the Schelling’s model, applied in spatial arrangement and networks; according to the tolerance levels of the individuals of a simulated society. Finally, we provide probabilistic arguments to explain the observed behaviors, and also, we reflect about some topics that allow addressing the segregation through an organic manner.

Since the current sequencing technologies cannot read complete genome sequences, it is necessary to perform an intermediate step to reconstruct the genome. Such step is called genome assembly. Large genomes being highly heterozygous, highly repetitive or polyploids are considered complex genomes, because they resist the genome assembly algorithms. The present PhD thesis contributes to the assembly of complex genomes by introducing new algorithms and by addressing the de novo genome assembly of two complex genomes.

Chapter 2 is dedicated to an early result of the present PhD thesis, namely the complete de novo assembly reconstruction of the highly heterozygous ‘Sultanina’ genome. Our approach produced a draft genome of 466 Mb, recovering 82% of the genes present in the grapevine reference genome; in addition, we identified 240 novel genes.

Chapter 3 introduces what we consider is the major contribution of this PhD thesis, namely the FAST-SG algorithm which uses a new ultra-fast alignment-free algorithm specifically designed for constructing a scaffolding graph using light-weight data structures. Using comprehensive standard datasets and benchmarks, we show how FAST-SG outperforms the state-of-the-art short read aligners when building the scaffolding graph, and can be used to extract linking information from either raw or error-corrected long reads.

Finally, Chapter 4 presents the de novo genome reconstruction of the Chinook salmon, which is a pseudo-tetraploid genome of 2.4Gb size. The genome was assembled using the algorithms developed in Chapter 3. The Chinook salmon (Oncorhynchus tshawytscha) is an anadromous fish species with considerable ecological, economic and social value. Overall, we consider that the algorithmic and genome assembly results described in the present PhD thesis are a significant contribution to the cracking of complex genomes.

In this thesis three works on complexity in economics and finance has been developed.

In the first essay, we deal with the modern retail competition. In particular, a dynamic multi-market Cournot–Nash equilibrium with global economies and diseconomies of scale model is developed. We confirm the non-intuitive hypothesis that retail multi-store competition is more unstable than traditional small business that cover the same demand. The main sources of stability are the scale parameter, the number of markets, and the number of firms.

In the second work, we consider a trader who wants to buy or sell a European option. Nevertheless, there is not a market for this security. In this paper, firstly, we elaborate further the definition of proxy, and introduce the concept of twin assets, based on two basic criteria of similarity. Besides, an option pricing model of twin assets is developed considering risk-neutral valuation. The proposed model allows us to price an option of one non-traded derivative asset using its twin asset, but this time knowing explicitly (analytically) what levels of errors we are facing.
Finally, the last paper deals with the Constant Elasticity of Variance (CEV) model, which has been empirically validated as a valuable improvement over the Black Scholes model. In this essay, a new numerical method for the CEV model is developed, based on a path integral approximation. This new computational method is quite efficient and accurate compare to the well-established non-central chi-squared approach for the pricing of European call options.

Cellular metabolism is a biological complex system made of many metabolites that interact through a large network of biochemical reactions catalyzed by cellular enzymes. To model these systems, the first step is to have a good notion of which are its components and how they interact. The process of putting together a comprehensive representation of a cellular metabolic network is called genome-scale metabolic network reconstruction. This is commonly accompanied with constraint-based analysis which allows to model metabolic networks at steady-state. This dissertation presents contributions to genome-scale metabolic network reconstruction and modeling.

The first contribution is a workflow designed to assist genome-scale reconstructions which allows to use and connect different tools, ensuring each step is documented and reproducible. The rest of the contributions are applications of genome-scale network reconstruction and analysis on different microorganisms. One is Nannochlorpsis salina, a microalga that has a potential as a cell factory for lipid production. A first metabolic network reconstruction and model of this alga are presented. Another application is on Piscirickettsia salmonis, a fish pathogenic bacteria that threatens the salmon industry. A comprehensive reconstruction of its metabolic network is introduced revealing particular metabolic features. The analysis of its network also allowed to identify metabolites required for P. salmonis growth used to develop defined media for its culture.

As a final application the metabolic networks of two bioleaching bacteria are reconstructed and used to model their co-culture. Different growth scenarios are considered and trade-offs between the growth of both organisms are analyzed.

Modeling of plant responses to the environment through gene regulatory networks contributes to the understanding of plants as the complex systems that they are. This was the aim of this thesis. New knowledge in this field can help agriculture to face environmental challenges and improve productivity.

The human microbiome is the internal ecosystem of microorganisms that live in the human body. The gut microbiome represents the volume/part of the human body with the greatest abundance and diversity of microorganisms. Several works have established the importance of the gut microbiome for human health, in association with metabolic functions, immune and nervous systems and even with behaviour. The Fragile X syndrome (FXS) is a genetic condition that causes a range of nervous system and behavioural problems and is the most common known cause of inherited autism. For the study of FXS, the fmr1 gene knockout (fmr1 KO) mouse reproduces most of this phenotype and represents a preclinical model for the identification of new biomarkers and potential drug treatments for FXS.

This PhD thesis research was focused in the application of different bioinformatics methods to identify potential interactions between the gut microbiome and the brain in the FXS, paying attention to its influence with behaviour and gene expression. It was found a distinctive profile of the fmr1 KO2 gut microbiome in comparison with wild type controls, characterized by different bacterial groups including an increased level of Akkermansia and Sutterella genera. The interactions of these microorganisms with the host are largely unknown and its influence on health still remains unclear. A potential harmful mechanism of interaction may be related to the excessive mucin degradation produced by these intestinal bacterial groups. All together the findings of this PhD thesis research contributes to the better understanding of the FXS pathology and to propose novel possible interactions between the brain and gut microbiome in FXS.

Network Virtualization is the creation of multiple virtual networks on the same physical substrate. In the case of an optical network, the physical substrate is composed of physical nodes (switches or routers) interconnected by optical links. The communication between each pair of nodes is one by encoding the information as a beam of light traveling through the fiber from one node to another.

Each component (nodes and links) of the physical substrate has a finite amount of resources, which will be occupied by virtual elements to be hosting there. The 3 elements that make possible the implementation of a network virtualization are: physical networking with virtualization capability on nodes and links, requests for virtual networks and a control plane responsible for assigning virtual network to the physical network.

The flexible grid allows the center frequency and the spectral width are variable according to the needs of communication to establish. This is achieved by dividing the spectrum into small units, typically 12.5[GHz], known as FSU (Frequency Slot Unit) [12]. Each FSU can be coupled with other channels to achieve a desired spectrum utilization.

In dynamic conditions, due to the release of funds by the virtual networks that fail to operate, yet generate gaps between occupied slots. This is called spectrum fragmentation. The main problems associated with this have to do with the spectral efficiency, increasing the blocking probability. To resolve this problem appears as a reasonable alternative to defragment the spectrum.

Defragmenting the spectrum is actually a reconfiguration of this. The goal of the reconfiguration is to rearrange the resources used by the virtual networks in the physical network.

Thus appears the general objective of this thesis, which is to reduce the blocking probability in the allocation of resources to virtual networks on a flexible optical grid substrate using reconfiguration techniques

Bifurcations in continuous dynamical systems, i.e., those described by ordinary differential equations, are found in a multitude of models such as those used to study phenomena related to natural and social sciences. Using this concept as a unifying idea, we apply it to model both Social as well as Physical systems.

In the first part, we apply tools of statistical physics and bifurcation theory to model a problem of binary decision in Social Sciences. We find a scheme to predict the appearance of extreme jumps based on the notion of precursors which act as a kind of warning signal for the upcoming appearance of these catastrophic events. We also solve a mathematical model of social collapse based on a logistic re-growing equation used to model population growth and how limited resources change growth patterns in combination with the so-called sunk-cost effect, the which has been used as a feasible explanation for understanding the disintegration of some ancient societies.

In the second part, we study systems described by stochastic differential equations or equivalently by their deterministic counterparts, the Fokker-Planck equation. A new and alternative scheme of computation based on Path Integrals, is introduced in order to calculate the Probability Distribution Function. The computations based on this Path Integral scheme are performed on systems in one and two dimensions and contrasted to soluble models validating this method. We also extended this scheme to the case of computation of Mean Exit Time which resulted in a new expression for systems in arbitrary dimensions

Los sistemas complejos están representados por una serie de sistemas, incluidos, entre otros, los ecosistemas, el cerebro, la economía, las organizaciones y la sociedad. Estos sistemas presentan propiedades emergentes, dinámicas estocásticas y caóticas de difícil predicción y una estructura de red entre los agentes que los componen, todo lo cual es consecuencia de la interdependencia entre los elementos que componen los sistemas complejos. La ingeniería de sistemas complejos proporciona un conjunto de metodologías integradas para la descripción de sistemas complejos y para abordar problemas originados en este tipo de sistemas y la Microbiología no ha dudado en incorporar el enfoque de sistemas complejos para caracterizar la dinámica de los ensamblajes microbianos.

En esta tesis, estudiamos la dinámica de la comunidad bacteriana a una alta resolución temporal temprana para reconstruir las redes de interacción ecológica y para describir cómo las redes bacterianas reflejan la dinámica de la comunidad bacteriana. Ofreceremos una descripción detallada del análisis de redes de comunidades bacterianas en dos ecosistemas diferentes diseñados (microcosmos del suelo y superficies artificiales sumergidas en agua de mar) y demostraremos cómo se puede aplicar para evaluar la estabilidad del ecosistema bacteriano, contribuyendo tanto a la Ecología Microbiana básica como la aplicada.

La histéresis es la tendencia de un material a conservar una de sus propiedades, en ausencia del estímulo que la ha generado. Los materiales que tienen una respuesta de histéresis a un campo externo, como el magnético, son esenciales en el almacenamiento de información y en tecnologías de procesamiento. Una gran cantidad de curvas de magnetización, de materiales artificiales y naturales, han sido previamente observadas y en cada una de ellas se han encontrado mecanismos ad-hoc que explican el fenómeno de histéresis. Sin embargo, un modelo fenomenológico simple que reproduzca cualitativamente distintos ciclos de histéresis y que al mismo tiempo provea de un modelo para diseñar la respuesta magnética de un material, no existe.

En este trabajo proponemos y demostramos experimentalmente un método elemental para diseñar ciclos de histéresis en un metamaterial construido con cadenas dipolares. Demostramos que controlando las interacciones y geometría del sistema, podemos configurar el ciclo de histéresis, permitiendo el diseño de un material magnético a voluntad. Adicionalmente, este mecanismo permite el control del número de ciclos con el fin de realizar tecnologías de valor lógico múltiple. Nuestros hallazgos allanan el camino para el diseño de respuestas de histéresis en una variedad de sistemas físicos, como dipolos fríos, materiales ferroeléctricos o redes reticulares magnéticas, entre otros.

A microbiome is a community of microorganisms that inhabit an environment like soils or human body and are made of different species that interacts among them. The microbiome behaves as a complex adaptive system and fluctuates in response to changes in environmental factors. The recovery of the genetic material, the classification and study of these microbiomes is called Metagenomics. The digitalization of the genetic material is done by sequencing technologies which shreds the DNA into fragments called reads.

In the first part, we propose an alignment-free method capable of assign a taxa to a read by analyzing its statistical properties. A database is constructed and used to build a robust machine learning procedure based on multiple CART trees.

In the second part we study the dynamics of a gut microbiome under antibiotic perturbation and Clostridium difficile infection (CDI). We demonstrate that a previous Boolean model is a threshold Boolean network (TBN). We generate an evolutionary algorithm to identify alternative TBNs and construct a neutral space conformed by a set of models that differ in their interactions and dynamics but share the observed final states. Relevant interactions are identified and we discuss how can affect or prevent CDI.

In the third part we examined microbiomes of a transect from Atacama Desert. We studied the compositional structure and generate two co-occurrence networks. We examined changes in interactions and the associations to physicochemical variables. Finally, we examined the nature of ecological rearrangements that take place in the microbial community when facing contrasting environments.

Nowadays, due to advances in technology, data is generated at an incredible rate, resulting in large data sets of ever-increasing size and dimensionality. Therefore, it is important to have efficient computational methods and machine learning algorithms that can handle large data sets, such that they may be analyzed in reasonable time. One particular approach that has gained popularity in recent years is Extreme Learning Machine (ELM), which is the name given to neural networks that employ randomization in their hidden layer, and that can be trained efficiently. This Ph.D. thesis introduces several machine learning methods based on neural networks with random weights aimed at dealing with the challenges that modern data sets pose.

This thesis presents new algorithms and applications of randomized neural networks, a kind of artificial neural network that runs very fast on a computer. Five contributions are presented. First, a new parallel architecture is proposed. Second, a real problem with noise data is analyzed with this type of neural networks. Another direction of research was to develop an algorithm for pruning irrelevant neurons in the hidden layer. Fourth, an empirical study of the hidden matrix rank is presented. Finally, a novel non-iterative deep neural network is used for sentiment classification in Twitter.

Biological communities are conventionally described as assemblages of species with well-defined ecological roles. In microbial ecology, however, such a taxonomic approach is hindered by our limited knowledge of the functions of most microorganisms, which often alter their genetic material through diverse mechanisms. Thus, some microbial ecologists have made use of culture-independent genetic approaches to study the whole pool of functional genes instead. However, this approach requires dealing with gene categories not necessarily related to the ecology of the organisms.

In this study, it is demonstrated that the genes encoding oxidoreductases characterize the microbial communities better than other categories of genes, including those associated with the taxonomy. Additionally, with this approach, the role of the microbial communities from different ecosystems in the biogeochemical cycles becomes readily apparent. The importance of this result is, however, limited by the coverage of known genetic functions in the total pool of metagenomic genes. To increase this coverage a methodology for the recognition of the protein-coding potential of DNA sequences is proposed. The results of this study should pave the way for a better assessment and evaluation of microbial ecosystem services from different environments of our planet.

In this thesis, we tackle the study of the dynamics of ecological models, using different techniques such as: blow up, Poincaré compactification, normalization, bifurcation theory and averaging theory of first order. Initially, we analyze the dynamics of a planar system of ordinary differential equations, which is a modified Leslie-Gower model, considering a Beddington-DeAngelis functional response. It generates a rich dynamics of the predator-prey interactions according to the asso- ciated parameters. From the system obtained we characterize all the equilibria and its local behaviour and the existence of a attraction basin is proved. We describe different types of bifurcations (such as, Hopf, Bogdanov- Takens and homoclinic bifurcation) and the existence of limit cycles is shown. Analytic proofs are pro- vided for all results. Ecological implications and a set of numerical simulations supporting the mathematical results are also presented.

Next, we study general dynamic features of an Intraguild Predation (IGP) model where the top predator feeds only on the mesopredator and also affects its consumption rate. Important dynamical aspects of the model are described. Specifically, we prove that the trajectories of the associated system are bounded and defined for all positive time; there is a trapping domain; there are open subsets of parameters, such that the system in the first octant has at most five equilibrium solutions and at to three of them are of co-existence. Here we characterize the exis- tence of Hopf bifurcations and we prove that this model exhibits either one, two or three small amplitude periodic solutions which arise from a zero-Hopf bifurcation. We prove the existence of multi stability. Finally, some numerical computations have been given in order to support our analytical results. The importance of some parameters of the model is discussed, in particular the role of the interference rate. The numerical exploration suggests that the equilibrium biomass of each of the three species grow as the level of interference grows.

In this thesis, our main interest is focused specifically on the dynamics of cellular automata. Dynamics will be addressed by the invariance of the automaton, and in some cases invariance will be addressed by its reversibility. Our concept of invariance will be considered under the set of attractors of the automaton, i.e. its periodic configurations. This concept suggests to question the robustness of the automaton, this is, the stability of the behavior regardless external disturbances, such as different update schedules. This is important in dynamical systems, in order to determine and prove strong properties that are invariant under structural modifications.

More precisely, we have studied the block invariance and attractor invariance of the elementary cellular automata. On the one hand, we have studied 11 conjectures about block invariance that were previously established. We managed to prove 9 of them and refute the other 2 left. Also, for all 256 elementary cellular automata we stablished equivalences in between them by means of the configurations of their set of attractors. In the case of attractor invariance, we managed to characterize the set of attractor of the elementary cellular automata rules under sequential update schedules, so to establish equivalences (classes) in between these rules by means of the configurations of their set of attractors. We have proven 2 of these classes, leaving the rest of them as future work. On the other hand, we were able to characterize the update schedules for which the linear rules 90 and 150 are invariant, and the same was done for linear rules with radius 2. The key tool to prove invariance for linear rules was the study of the reversibility of each rule.

In intertidal marine crustaceans, phenotypic variation in physiological and lifehistory traits is pervasive along latitudinal clines. However, organisms have complex phenotypes, and their traits do not vary independently but rather interact differentially between them, effect that is caused by genetic and/or environmental forces. One of the abiotic factors that most influences organisms is the temperature that affects practically all the biochemical and physiological processes of organisms, which has been linked to growth, survival and reproduction patterns. In this sense, these ectothermal organisms have had to adapt to different environmental conditions, modifying or adjusting their physiology and / or behavior to tolerate the extreme environmental conditions of the intertidal. This allows us to ask some questions such as: 1) Are there compromises between physiological components and life-history traits in marine ectothermal organisms?. 2) What is the role in acclimatization to different geographical gradients in physiological responses (thermal tolerances) in ectothermic marine organisms? 3) Does the integration of physiological traits responses vary between populations of different species?.

To answer these questions, we evaluate the phenotypic integration of life-history and physiological traits based on geographic variations found at latitudinal and intertidal levels. This study covered a latitudinal gradient of approximately 3,000 km from the coast around Iquique to the coast of Chiloé; we study different populations of decapod crabs that are distributed in vertical bands in the rocky intertidal zone exposing themselves to different thermal variations. Our main results reflect (1) Variations in physiological traits with latitude appear to be a plastic response controlled primarily by temperature, rather than a fixed traits of site-specific populations, therefore, population differences in physiological rates result in a differential mean in SFG of ovigerous and non-ovigerous females, which could be related to the frequency of reproductive events in those species that are distributed across wide geographical gradients. (2) The species that inhabit the intertidal-high maintain a greater integration between their physiological traits and greater thermal tolerances and present less plasticity than those that inhabit the zones of low intertidal. (3) The physiological behavior of the studied species differs along a clinal gradient, and the differentiation of the P-Matrix depends mainly on the microhabitat in which these species are distributed vertically in addition to the environment barriers such as biogeographic breaks.

Automata Networks are discrete dynamical systems initially introduced by von Neumann, Mc-Culloch and Ulam and they have been used to model diverse complex systems such as the study of the evolution and self-organization of physical and biological systems. Automata Networks are composed of a graph, where each node acquires different states (from a finite set) and evolves in units of discrete time, according to a certain function -known as the local transition rule- that depends on the neighboring states of the network.

Cellular Automata (CA) and Boolean Networks (BN) are particular cases of Automata Networks. In the CA, the neighborhood structure and the transition rules are the same for all nodes. On the other hand, Boolean Networks are non-uniform, binary systems, meaning that each node can take only two possible states and evolves according to its own local Boolean transition rule, on an arbitrary finite graph (orientated or not oriented).

The thesis consists of three parts. The first two study problems related to Cellular Automata: a class of decision problems with binary, one-dimensional CAs, and the complexity analysis of a specific decision problem for elementary CA, the prediction of the so-called stability problem. The third part is focused on the dynamics of Boolean Networks with Memory (RBM) and their applications.

Abstract Jumping locomotion is present in many organisms, such as plants, fungi, insects and vertebrates that describe a wide variety of jumping mechanisms. Surprisingly, jumping organisms are limited to a very narrow range of take-off velocity (below 6 m/s). To explain the jumping limit observed in nature, we propose a definition of jumping that encompasses a diversity of organisms. We present a physics model that provides an understanding of the limit. The model hand over kinematic and dynamic constraints added to biological aspects. The main results are that general mechanical considerations provide an upper bound for the take-off velocity of any jumper, animate or inanimate, rigid or soft body, animal or vegetal. The take-off velocity is driven by the ratio of released energy to body mass. Further, the mean reaction force on a rigid platform during push-off is inversely proportional to the characteristic size of the jumper. These general considerations are illustrated in the context of Alexander’s jumper model which can be solved exactly and which shows an excellent agreement with the mechanical results. These aspects of the jump can be applied to understand the limit of nature and the construction of artificial devices to jump’s way locomotion.

Finally, we complement the jumping research with a case of study of a very good jumper, the jerboa(rodent). This is a highly maneuverable bipedal rodent, with multiple tail behaviors such, the tail is elevated and actively controlled during locomotion, provides support during standing, and acts as a fifth limb for juvenile jerboas learning to walk. We collect experimental data obtained specifically from the tail of Jerboa and we proposed models representing its through an elastic structure. The first approach contributes to developing a model of the particular S shape of the animate tail; the second one allowed to calculate bending stiffness of the tail with experimental mechanics and elastic model (inanimate tail).

En el presente trabajo se discute la importancia del espacio geográfico en el contexto de la generación de marcos muestrales de encuestas, poniendo en tensión la premisa estadística tradicional de la aleatoriedad e independencia de las observaciones. Para esto se analiza el aporte de la geografía cuantitativa en la generación de metodologías de regionalización que permitan, de manera efectiva, mejorar el error muestral de las encuestas, enfocados principalmente en las áreas urbanas, en presencia de variables de estratificación con autocorrelación espacial. Se testean de forma empírica algoritmos de regionalización con y sin procesos de optimización heurística, utilizando datos censales, para posteriormente definir el nivel de error y establecer comparaciones contra muestreos tradicionales de corte aleatorio y aleatorio bi-etápico, por medio de un procedimiento Montecarlo. Los resultados obtenidos dan cuenta de una disminución de hasta un 20% en el error contra metodologías tradicionales o en su defecto la disminución de hasta 100 casos con el mismo nivel de error. Se concluye que las metodologías de muestreo espacializado con optimización heurística ofrecen ventajas evidentes en áreas urbanas, en presencia de autocorrelación espacial.

Finalmente se analiza la incidencia de la compacidad cartográfica, primero como un resultado directo de los algoritmos y su importancia en el contexto de la optimización de la varianza y la operatividad de aplicar la regionalización a encuestas, o bien, como unidad mínima territorial para instrumentos exhaustivos. Se establece que la prioridad está dada por mantener los preceptos de optimización de los algoritmos de regionalización por sobre la consistencia de la forma. Asimismo, antes que la coherencia geométrica, es clave mantener estables los pisos de población. En tal sentido se proponen cambios en zonas extremas de compacidad crítica como procedimiento plausible para no bajar la efectividad del proceso del muestreo espacializado.

Public and Private enrollment in Chile: School choice, enrollment trends, and school characteristics. Public education is crucial to satisfy the right to education and provide nonexclusive educative projects (Bellei, González, and Valenzuela 2010). Most OECD countries have a large public school sector. In contrast, only 34% of students attend this sector in Chile and, since the 90s, public enrollment has decreased (OECD 2020; Bellei 2009; Hsieh and Urquiola 2006). We delve into the different factors affecting the enrollment distribution between public and private school sectors.

To do so, we follow three broad steps. First, we delve into the demand side. We study family preferences and the choice-set formation process, using both a latent class logit model and a computational model to represent how families become aware of different school options. Second, we develop an Agent-Based Model (ABM) to represent the Chilean school system. This ABM extends the previous models and goes further by modeling schools’ behavior and temporal trends. We used this ABM to run several experiments to contrast different hypotheses about the mechanisms behind the decline in public enrollment. Finally, we use the ABM to focus on how the different educational levels offered by the schools (i.e., primary and secondary education) affect the enrollment distribution between the public and private sectors. We show that (i) the educational levels offered by the schools are important for parents’ decisions and for explaining enrollment trends, (ii) social networks are a key element in choosing schools and school enrollment, and (ii) there is inertia in the Chilean educational system.

In this thesis, I present three original articles on the economics of crime using elements of chaos theory. Firstly, we model the economics of crime and punishment from an evolutionary perspective. We first model the interaction between agents and controllers as an asymmetric game and then, we explore the dynamics of the static model using a replicator dynamics equation, given exogenous levels of monitoring and criminal sanctions. The results point out that an equilibrium of honest agents is never stable, however when the penalty is high enough, the system will neutrally tend to an equilibrium of honest agents and a monitoring firm. On the other hand, when the probability of detecting wrongdoing is rather small, the system, in some cases, will remain in a transient state, where it will be impossible to predict the proportion of honest agents. Secondly, we model corruption as a repeated stochastic asymmetric game with discounting.

Finally, we investigate the role and potential impact of whistleblowing on the persistence of corruption in organisations. We focus specifically on the role of incentives aimed at encouraging whistleblowing behavior. We use the concept of replicator dynamics to explore formally the local asymptotical stability of honest behavior, corrupt behavior, and whistleblowing (honest enforcer) behavior. Assuming that the population shares of an organization pursuing these three behavior types, we analyze which of these population shares will become stable in the long run.

Por décadas, los out-of-stock (OOS) han sido un problema para los retailers y elaboradores y comercializadores de bienes de consumo de alta rotación -Fast-moving consumer goods (FMCG)-. Un evento de OOS se refiere a una condición en la cual los consumidores no encuentran un cierto producto, cuándo tienen la intención de adquirirlo. Esta tesis tiene como propósito abordar dicha problemática desde la perspectiva del manufacturador, proponiendo un caso de estudio, en una empresa elaboradora de productos alimenticios empacados para retail, ubicada en Latino América. Se desarrollaron e implementaron dos sistemas basados en Machine Learning para detectar automáticamente OOS.

El primer sistema se basó en un algoritmo de clasificación de Random Forest con data balanceada, y el segundo, en un ensamble de 6 diferentes algoritmos de clasificación. Se utilizaron datos transaccionales obtenidos desde el sistema de información del manufacturador y a través de auditorías físicas a los puntos de venta. La novedad de este trabajo es que las nuevas variables predictoras de OOS que han sido definidas e incorporadas, no han sido utilizadas anteriormente. Estas variables fueron determinadas manualmente a partir del conocimiento del dominio, pudiendo ser posible medir y cuantificar su impacto en el mundo real. Este sistema fue implementado exitosamente en una empresa elaboradora de productos alimenticios para retail. La incorporación de las nuevas variables predictivas nos permitió mejorar el poder predictivo del sistema, logrando que Random Forest alcanzara un 72% de precisión en la detección, identificando un 68% del total de los OOS en un escenario de la vida real.

Engineering rhizospheric environments is a crucialstep towardssustainable agriculture processes. These complex inter-kingdom environments are characterized by the inherent interactions between the microbial members of the system and their relationship with the associated host plant. This dynamism represents a significant factor of complexity within the rhizospheric system. It is influenced by the plant genotype, development, and environmental responses, mainly throughout the exudation of organic components from the root system to the soil. These factors fine-tune the rhizospheric-related microbial communities to better adapt to the plants' conditions. Positive interactions are mainly abstracted from the ecologic relationship on the rhizosphere, and plants could benefit from growth and health and are the primary subject for engineering principles.

These beneficial plant-related microbial characteristics are often present on low abundant microbiome members, and little is known about their interactions and growth characteristics. This thesis explores how members of a low abundant rhizospheric community, composed of four Burkholderiales, grow and interact within the co-culture and with each other. Additionally, it studies how they adapt towards changes in plant root exudate patterns in short and long term situations; trying to elucidate possible predictive attributes from their ecologic behavior, genetic characteristics, and adaptation capabilities that could explain the microbial shifts within these microbial communities, always thinking towards the application of these attributes to the engineering of rhizospheric environments.

Multi-Attribute Vehicle Routing Problems (MAVRP) are variants of Vehicle Routing Problems (VRP) in which, besides the original constraint on vehicle capacity present in Capacitated Vehicle Routing Problem (CVRP), other restrictions model diverse real-life system attributes. Among the most common attributes studied in the literature are vehicle capacity and route duration constraints. The impact of these restrictions on the overall structure of the problem and on the performance of local search algorithms used to solve it is not well known. This work aims to explain how constraints impact different variants of VRP by altering the structure of the underlying search space.

We work on two research articles: The first proposes a new scheme for Fitness Landscape Analysis (FLA) and provides valu able information to characterize the fitness landscape’s structure of CVRP and its relation with the algorithmic performance. The second focuses on Local Optima Network Analysis (LONA) for multiple Traveling Salesman Problem (m-TSP) and VRP with vehicle capacity (CVRP), route duration (DVRP), and both (DCVRP) constraints. We present results that indicate that measures obtained for a sample of local optima provide valuable information on the behavior of the landscape under modifications in the problem’s constraints.