Introdução: as infecções são responsáveis por mais de um terço das mortes neonatais, sendo a sepse a de maior expressão. A sepse neonatal precoce (SNP) aumenta em cinco vezes a taxa de mortalidade entre recém-nascidos (RN) de termo e pré-termos tardios e em vinte e cinco vezes a taxa de mortalidade entre pré-termos extremos. A Variabilidade da Frequência Cardíaca (VFC) vem se mostrando capaz de antecipar o diagnóstico, reduzindo em 22% o risco relativo de mortalidade entre os pré-termos de muito baixo peso com sepse. Objetivo: verificar a VFC de RN com suspeita de SNP; identificar medidas significantes e verificar se, associadas a variáveis clínico-laboratoriais, aumentam a chance de identificar pacientes passíveis de desenvolver a doença. Método: utilizou-se um frequencímetro portátil para aferir as medidas da VFC nos domínios do tempo, da frequência e do Caos. Dentre elas, elegeu-se aquelas com curva ROC > 0,75 para, com seus valores de cut-off, aplicar regressão logística e identificar as que melhor se adequassem ao modelo. As variáveis clínico-laboratoriais significantes que restaram da análise univariada foram submetidas à análise multivariada a fim de identificar, também, as que melhor se adequassem ao modelo. Resultados: foram identificadas quatro variáveis clínico-laboratoriais (sexo masculino, idade gestacional < 34 semanas, Score de Rodwell e proteína C-reativa), e nove medidas da VFC (RRmean, SDNN, RR tri index, LF, HF, SD2 do plot de Poincaré, ShaEnt, CorrDim D2 e SamAsy). Cinco apresentaram área sob a curva ROC > 0,75: SDNN, RR tri index, LF, SD2 e CorrDim D2. Depois de aplicada a regressão logística, restou apenas a CorrDim D2 com cut-off = 0,1164. Identificadas, as variáveis foram submetidas ao cálculo do logit para, com isso, quantificar a chance do paciente suspeito de evoluir para a doença, na dependência das três variáveis eleitas: CorrDim D2, idade gestacional < 34 semanas e score de Rodwell. Conclusão: o estudo mostrou que tanto o modelo preditivo clinico como a CorrDim D2 são capazes, de forma independente, de estimar a chance que um futuro paciente com sepse suspeitada tem de efetivamente confirmar o diagnóstico de sepse. E que, além disso, se a CorrDim D2 for agregada ao modelo preditivo clínico, esta chance aumenta consideravelmente. / Introduction: infectious diseases are responsible for more than one-third of neonatal deaths, and sepsis is of great importance. Early Neonatal Sepsis (ENS) increases five times the mortality rate among term and late preterm newborns. Among extreme preterm babies the mortality rate increases by twenty-five times. Heart Rate Variability (HRV) has been showing to be capable of anticipating the diagnosis, thus reducing by 22 percent the relative risk of mortality of very low preterm babies with sepsis. Objective: to verify HRV of newborn babies suspects of having ENS; to identify significant measures of HRV and to verify if, associated to clinical-laboratorial variables, if they improve the chance of identifying patients exposed to developing ENS. Method: a portable frequencimeter was used to registering the measurements of HRV in the domains of time, of frequency and of chaos. Among them, only those with a ROC curve greater than 0.75 were chosen, with their cut-off values to enter a logistic regression with the aim of identifying those more suitable to the model. The clinical-laboratorial variables which were considered significant in a univariete analysis were also submitted to the multivariate analysis with the purpose of identifying those better fitted to the model. Results: four clinical-laboratorial variables (male sex, gestational age less than 34 weeks, the Rodwell Score and C-Reactive Protein); similarly, nine measures of HRV (RRmean, SDNN, RR tri index, LF, HF, SD2 of Poincaré plot, ShaEnt, CorrDimD2 and SamAsy). Five of them presented an area under the ROC curve greater than 0.75: SDNN, RR tri index, LF, SD2 and CorrDim D2. Once applied the logistic regression, only CorrDim D2 remained, with a cut-off value of 0.1164. The identified variables were, then, submitted to the calculation of the logit, with the purpose of quantifying the chance of a patient suspect of evolving to a sepsis, in the dependence of the three elected variables: CorrDim D2, gestational age under 34 weeks and the Score of Rodwell. Conclusion: the study showed that both, the predictive clinical model as CorrDim D2 are able, independently, to estimate the chance that a patient with suspected sepsis future has to effectively confirm the diagnosis of sepsis. And that, moreover, if the CorrDim D2 is aggregated to clinical predictive model, this chance increases considerably.

Introdução: as infecções são responsáveis por mais de um terço das mortes neonatais, sendo a sepse a de maior expressão. A sepse neonatal precoce (SNP) aumenta em cinco vezes a taxa de mortalidade entre recém-nascidos (RN) de termo e pré-termos tardios e em vinte e cinco vezes a taxa de mortalidade entre pré-termos extremos. A Variabilidade da Frequência Cardíaca (VFC) vem se mostrando capaz de antecipar o diagnóstico, reduzindo em 22% o risco relativo de mortalidade entre os pré-termos de muito baixo peso com sepse. Objetivo: verificar a VFC de RN com suspeita de SNP; identificar medidas significantes e verificar se, associadas a variáveis clínico-laboratoriais, aumentam a chance de identificar pacientes passíveis de desenvolver a doença. Método: utilizou-se um frequencímetro portátil para aferir as medidas da VFC nos domínios do tempo, da frequência e do Caos. Dentre elas, elegeu-se aquelas com curva ROC > 0,75 para, com seus valores de cut-off, aplicar regressão logística e identificar as que melhor se adequassem ao modelo. As variáveis clínico-laboratoriais significantes que restaram da análise univariada foram submetidas à análise multivariada a fim de identificar, também, as que melhor se adequassem ao modelo. Resultados: foram identificadas quatro variáveis clínico-laboratoriais (sexo masculino, idade gestacional < 34 semanas, Score de Rodwell e proteína C-reativa), e nove medidas da VFC (RRmean, SDNN, RR tri index, LF, HF, SD2 do plot de Poincaré, ShaEnt, CorrDim D2 e SamAsy). Cinco apresentaram área sob a curva ROC > 0,75: SDNN, RR tri index, LF, SD2 e CorrDim D2. Depois de aplicada a regressão logística, restou apenas a CorrDim D2 com cut-off = 0,1164. Identificadas, as variáveis foram submetidas ao cálculo do logit para, com isso, quantificar a chance do paciente suspeito de evoluir para a doença, na dependência das três variáveis eleitas: CorrDim D2, idade gestacional < 34 semanas e score de Rodwell. Conclusão: o estudo mostrou que tanto o modelo preditivo clinico como a CorrDim D2 são capazes, de forma independente, de estimar a chance que um futuro paciente com sepse suspeitada tem de efetivamente confirmar o diagnóstico de sepse. E que, além disso, se a CorrDim D2 for agregada ao modelo preditivo clínico, esta chance aumenta consideravelmente. / Introduction: infectious diseases are responsible for more than one-third of neonatal deaths, and sepsis is of great importance. Early Neonatal Sepsis (ENS) increases five times the mortality rate among term and late preterm newborns. Among extreme preterm babies the mortality rate increases by twenty-five times. Heart Rate Variability (HRV) has been showing to be capable of anticipating the diagnosis, thus reducing by 22 percent the relative risk of mortality of very low preterm babies with sepsis. Objective: to verify HRV of newborn babies suspects of having ENS; to identify significant measures of HRV and to verify if, associated to clinical-laboratorial variables, if they improve the chance of identifying patients exposed to developing ENS. Method: a portable frequencimeter was used to registering the measurements of HRV in the domains of time, of frequency and of chaos. Among them, only those with a ROC curve greater than 0.75 were chosen, with their cut-off values to enter a logistic regression with the aim of identifying those more suitable to the model. The clinical-laboratorial variables which were considered significant in a univariete analysis were also submitted to the multivariate analysis with the purpose of identifying those better fitted to the model. Results: four clinical-laboratorial variables (male sex, gestational age less than 34 weeks, the Rodwell Score and C-Reactive Protein); similarly, nine measures of HRV (RRmean, SDNN, RR tri index, LF, HF, SD2 of Poincaré plot, ShaEnt, CorrDimD2 and SamAsy). Five of them presented an area under the ROC curve greater than 0.75: SDNN, RR tri index, LF, SD2 and CorrDim D2. Once applied the logistic regression, only CorrDim D2 remained, with a cut-off value of 0.1164. The identified variables were, then, submitted to the calculation of the logit, with the purpose of quantifying the chance of a patient suspect of evolving to a sepsis, in the dependence of the three elected variables: CorrDim D2, gestational age under 34 weeks and the Score of Rodwell. Conclusion: the study showed that both, the predictive clinical model as CorrDim D2 are able, independently, to estimate the chance that a patient with suspected sepsis future has to effectively confirm the diagnosis of sepsis. And that, moreover, if the CorrDim D2 is aggregated to clinical predictive model, this chance increases considerably.

Avec la hausse des niveaux de dimensionnement sismique il est devenu nécessaire de limiter les chargements internes dans les structures, notamment en utilisant des dispositifs glissants. Ces dispositifs plafonnent les efforts internes en déclenchant un glissement de la structure. Il devient cependant nécessaire d'estimer l'amplitude des déplacements de corps rigide, notamment pour les structures stockées dans des réservoirs. Dans ce cas, il est nécessaire de prévenir les impacts entre la structure glissante et les bords du réservoir pour contrôler les risques de fuite. Parmi les structures glissantes immergées, on citera les ponts, les structures côtières en maçonnerie, les râteliers de stockage de combustible nucléaire, etc...Les équations de dynamique associées au comportement de ces structures sont non-linéaires et nécessitent l'utilisation de simulations numériques coûteuses en temps de calcul et ne permettant pas de faire des études de sensibilité rapides. On propose donc une méthode de résolution quasi-analytique de ces équations en traitant dans un premier temps, l'évaluation analytique des matrices de masses ajoutées du couplage fluide-structure, dans un second temps, une méthode de résolution quasi-analytique du glissement d'une structure quelconque immergée dans un fluide avec une actualisation de la géométrie de lames d'eau. Les résultats obtenus présentent une bonne adéquation avec des simulations numériques et offrent un temps de calcul quasiment instantané compatible avec une étude paramétrique ou stochastique de ces structures / As the seismic loadings are increasing in accordance to the recent regulations regarding Earthquake design, the use of sliding devices in structures is becoming more common. These devices limitate the internal forces by creating a rigid body sliding. It is then necessary to estimate the global displacement of the structure, especially concerning structures that are immersed in a reservoir. In this case, the displacement must be well estimated in order to prevent impacts between the sliding structure and the boundaries of the reservoir. We can find such structures in : bridges, costal structures in brick and masonry, or in the nuclear industry with the underwater fuel storage racks, ...The governing equations for the behaviour of these structures are non linear and must be solved using time-consuming computer simulations which are not fit for a stochastic study. Our method consists in, firstly, evaluating analytically the added masses of the fluid-structure interaction, secondly, a semi-analytical solving of the governing equations including the updating of the dimensions of the fluid layers surrounding the sliding structure. The results of this new method are in accordance with the numerical simulations and can be obtained in a short time (1 or 2 seconds) which offers the possibility to make a stochastic analysis of the non linear behaviour

Molecular Dynamics (MD) is the numerical simulation of a large system of interacting molecules, and one of the key components of a MD simulation is the numerical estimation of the solutions to a system of nonlinear differential equations. Such systems are very sensitive to discretization and round-off error, and correspondingly, standard techniques such as Runge-Kutta methods can lead to poor results. However, MD systems are conservative, which means that we can use Hamiltonian mechanics and symplectic transformations (also known as canonical transformations) in analyzing and approximating solutions. This is standard in MD applications, leading to numerical techniques known as symplectic integrators, and often, these techniques are developed for well-understood Hamiltonian systems such as Hill’s lunar equation. In this presentation, we explore how well symplectic techniques developed for well-understood systems (specifically, Hill’s Lunar equation) address discretization errors in MD systems which fail for one or more reasons.

For this project, we use a modified version of a previously developed mathematical model, which describes the relationships among matrix metalloproteinases (MMPs), their tissue inhibitors (TIMPs), and extracellular matrix (ECM). Our ultimate goal is to quantify and understand differences in parameter estimates between patients in order to predict future responses and individualize treatment for each patient. By analyzing parameter confidence intervals and confidence and prediction intervals for the state variables, we develop a parameter space reduction algorithm that results in better future response predictions for each individual patient. Moreover, use of another subset selection method, namely Structured Covariance Analysis, that considers identifiability of parameters, has been included in this work. Furthermore, to estimate parameters more efficiently and accurately, the standard error (SE- )optimal design method is employed, which calculates optimal observation times for clinical data to be collected. Finally, by combining different parameter subset selection methods and an optimal design problem, different cases for both finding optimal time points and intervals have been investigated.

The piston-driven oscillator is traditionally modeled by directly applying boundary conditions to the acoustic wave equations; with better models re-deriving the wave equations but retaining nonlinear and viscous effects. These better models are required as the acoustic solution exhibits singularity near the natural frequencies of the cavity, with an unbounded (and therefore unphysical) solution. Recently, a technique has been developed to model general pressure oscillations in propulsion systems and combustion devices. Here, it is shown that this technique applies equally well to the piston-driven gas-column oscillator; and that the piston experiment provides strong evidence for the validity of the general theory. Using a modified piston-tube apparatus, agreement between predicted and observed limit-cycle amplitudes is observed to be on the order of 1%.

The piston-driven oscillator is traditionally modeled by directly applying boundary conditions to the acoustic wave equations; with better models re-deriving the wave equations but retaining nonlinear and viscous effects. These better models are required as the acoustic solution exhibits singularity near the natural frequencies of the cavity, with an unbounded (and therefore unphysical) solution. Recently, a technique has been developed to model general pressure oscillations in propulsion systems and combustion devices. Here, it is shown that this technique applies equally well to the piston-driven gas-column oscillator; and that the piston experiment provides strong evidence for the validity of the general theory. Using a modified piston-tube apparatus, agreement between predicted and observed limit-cycle amplitudes is observed to be on the order of 1%.

Submitted by Reinaldo Kagi (reinaldo.kagi@gmail.com) on 2014-03-07T16:23:26Z No. of bitstreams: 1 KAGI, Reinaldo Kenji. (2014) - Fragmentos da Complexidade Aplicados ao Mercado Financeiro.pdf: 976757 bytes, checksum: 053f1bfa52560a1fc533f8c2860a3caa (MD5) / Rejected by JOANA MARTORINI (joana.martorini@fgv.br), reason: prezado Reinaldo, boa tarde A numerações das paginas não podem ficar visíveis antes da introdução, existem paginas em branco que devem ser retiradas e o titulo não está de acordo com o titulo apresentado na Ata de Dissertação conforme segue abaixo: Fragmentos DE complexidade aplicados ao mercado financeiro Fragmentos DA complexidade aplicados ao mercado financeiro Por gentileza fazer os ajustes listados acima. on 2014-03-07T18:20:57Z (GMT) / Submitted by Reinaldo Kagi (reinaldo.kagi@gmail.com) on 2014-03-07T19:05:43Z No. of bitstreams: 1 KAGI, Reinaldo Kenji. (2014) - Fragmentos de Complexidade Aplicados ao Mercado Financeiro.pdf: 953573 bytes, checksum: d484fb79cf4de8d48c79d7b72dd959a0 (MD5) / Rejected by JOANA MARTORINI (joana.martorini@fgv.br), reason: Prezado Reinaldo, boa tarde. Tem um único ponto a ser ajustado, na folha de assinatura a data está errada, o correto seria 07/02/2014 (data de sua apresentação de dissertação. por gentileza ajustar. on 2014-03-07T19:25:51Z (GMT) / Submitted by Reinaldo Kagi (reinaldo.kagi@gmail.com) on 2014-03-07T19:31:16Z No. of bitstreams: 1 KAGI, Reinaldo Kenji. (2014) - Fragmentos de Complexidade Aplicados ao Mercado Financeiro.pdf: 975473 bytes, checksum: 955bb2b02ccec9d2b6a97da997eb9af3 (MD5) / Approved for entry into archive by JOANA MARTORINI (joana.martorini@fgv.br) on 2014-03-07T19:32:30Z (GMT) No. of bitstreams: 1 KAGI, Reinaldo Kenji. (2014) - Fragmentos de Complexidade Aplicados ao Mercado Financeiro.pdf: 975473 bytes, checksum: 955bb2b02ccec9d2b6a97da997eb9af3 (MD5) / Made available in DSpace on 2014-03-07T19:50:04Z (GMT). No. of bitstreams: 1 KAGI, Reinaldo Kenji. (2014) - Fragmentos de Complexidade Aplicados ao Mercado Financeiro.pdf: 975473 bytes, checksum: 955bb2b02ccec9d2b6a97da997eb9af3 (MD5) Previous issue date: 2014-03-10 / The rise of complexity in financial markets has been reported by Rajan (2005), Gorton (2008) and e Haldane & May (2011) as one of the main features that led to the increase of systemic risk, which climaxed in the financial crisis of 2007/08. The Bank for International Settlements (2013) covers the matters of complexity in the context of banking regulation and discusses the comparability of capital adequacy among banks and jurisdictions. Nonetheless, definitions for concepts such as complexity and complex adaptive systems are omitted from the major discussions. This paper elucidates some concepts related to the Theories of Complexity, how this phenomenon arises, how they may be applied to financial markets. We discuss the use of two tools in the context of complex adaptive systems: Agent Based Models (ABMs) and entropy. We come to the conclusion that although the complexity research agenda still leaves us some gaps, it most definitely contributes to the economic research in understanding the mechanisms that trigger systemic risks, as well as adding tools that allows us model interacting heterogeneous agents, which leads to the rise of emergent phenomena in the system. Some research hypotheses are suggested for later development. / O aumento da complexidade do mercado financeiro tem sido relatado por Rajan (2005), Gorton (2008) e Haldane e May (2011) como um dos principais fatores responsáveis pelo incremento do risco sistêmico que culminou na crise financeira de 2007/08. O Bank for International Settlements (2013) aborda a questão da complexidade no contexto da regulação bancária e discute a comparabilidade da adequação de capital entre os bancos e entre jurisdições. No entanto, as definições dos conceitos de complexidade e de sistemas adaptativos complexos são suprimidas das principais discussões. Este artigo esclarece alguns conceitos relacionados às teorias da Complexidade, como se dá a emergência deste fenômeno, como os conceitos podem ser aplicados ao mercado financeiro. São discutidas duas ferramentas que podem ser utilizadas no contexto de sistemas adaptativos complexos: Agent Based Models (ABMs) e entropia e comparadas com ferramentas tradicionais. Concluímos que ainda que a linha de pesquisa da complexidade deixe lacunas, certamente esta contribui com a agenda de pesquisa econômica para se compreender os mecanismos que desencadeiam riscos sistêmicos, bem como adiciona ferramentas que possibilitam modelar agentes heterogêneos que interagem, de forma a permitir o surgimento de fenômenos emergentes no sistema. Hipóteses de pesquisa são sugeridas para aprofundamento posterior.

Dans un contexte industriel où l’utilisation de matériaux composites s’est généralisée jusqu’à atteindre les structures primaires, la menace foudre se révèle être une problématique majeure.Avec un coup de foudre en moyenne par an et par avion en service, les nouvelles structures composites, moins bonnes conductrices que leurs prédécesseurs en métal, doivent être protégées. Les protections mises en œuvre par les fabricants et les équipementiers sont des couches minces ajoutées à l’empilement composite, initialement choisi pour le compromis optimal qu’il offre entre résistance et légèreté. L’optimisation et le conseil concernant les protections foudre deviennent alors un enjeu industriel d’importance. Dans ce cadre, le travail de thèse a porté sur l’étude et la compréhension des dommages issus des chocs foudre sur des structures protégées dans le but ultérieur d’optimiser ou de créer des protections adaptées.Nous proposons une méthode qui permet de déterminer les caractéristiques d’un impact mécanique sur une plaque composite nue, équivalent à un choc foudre sur une structure protégée. Une campagne d’essais d’impacts avec un canon du laboratoire couplé à une campagne numérique ont permis de conclure que la stratégie et la méthode d’équivalence sont fondées, et permettent de prendre en compte les paramètres constitutifs de la protection de surface. Les modèles proposés permettront d’aborder les questions de conception des protections. / In an industrial context where more and more composite materials are integrated into primary structures, the lightning threat has become a major issue for aircraft manufacturers. As lightning strikes in service airplane about once a year, the new composite structures, with a lower electrical conductivity than their metallic predecessors, must be protected. The protections already integrated by manufacturers are mainly made of expanded metallic foil layered above the composite lay-up, thus adding weight on the low density structures and reducing the gain of weight. The optimization of such structures and counsel concerning lightning protection become a major industrial stake. In the scope of the PhD work, the proposed work focused on the study of the damage mechanisms due to lightning strikes on protected composite panels in order to optimize or offer adequate protections against this threat. A methodology is proposed to determine a mechanical impact on a bare composite plate equivalent to a protected and even painted structure submitted to a lightning impact. An experimental campaign of mechanical impacts using a canon gas gun coupled to a numerical plan is led and allows concluding on the strategy and its validity by taking into account the state surface of the lightning samples.

Cette recherche porte sur le développement d’un modèle de structure de géométrie complexe équipée de composants à comportement non linéaire viscoélastique dans le but de simuler sa réponse à des excitations définies par des densités spectrales de puissance (DSP). L’application industrielle concerne l’isolation vibratoire d’un module de refroidissement automobile monté sur plot de suspension en élastomère. Une revue du comportement des élastomères en fonction de leurs conditions d’environnement et de sollicitations identifie les paramètres des différents modèles analysés. Des essais préliminaires ont été menés pour définir les intervalles des niveaux de sollicitations et quantifier l’échauffement des plots. La caractérisation expérimentale de la suspension porte sur des plots en élastomère munis de leurs interfaces afin d’agréger dans un seul modèle les non linéarités du comportement viscoélastique, des glissements et des frottements. Les boucles effort-déflexion axiales et radiales mesurées sont traitées avec un système expert développé spécialement pour caler les paramètres du modèle retenu. Il s’agit du modèle de Dahl généralisé qu’il a fallu étendre aux aspects viscoélastiques. Ce processus de calage automatique a été codé avec un logiciel développé sous Octave/Matlab. Des méthodes d’interpolations et extrapolations rendent opérationnel ce modèle sur toute la gamme fonctionnement de l’application définie dans l’espace fréquence-déflexion. Ce processus a été codé dans le module UserSubroutine pour Abaqus. Soumettre le système mécanique non linéaire à des excitations définies par une DSP nécessite de s’appesantir sur le traitement des vibrations aléatoires. En effet il faut, pour calculer les réponses, considérer le passage fréquence-temps et inversement pour les comparer éventuellement aux exigences des normes. De plus, la taille et la complexité du modèle EF de la structure industrielle rendent impossible une résolution temporelle sur l’ensemble de ses degrés de liberté. Il s’agit alors de faire appel à des techniques d’homogénéisation et de condensation dynamique afin de prévoir la réponse aux excitations à large bande fréquentielle dans le but d’analyser les performances de l’isolation vibratoire. / This research work regards the development of a complex structure model with non-linear viscoelastic components. The purpose of this study is to simulate the response of this structure submitted to a random vibration excitation based on a power spectral density definition (PSD). The industrial applicative case is the vibratory insulation of a automotive engine cooling module supported by elastomer mounts. A brief review of elastomers behavior depending on solicitations types enables to identify the parameters of the different investigated models. Preliminary tests have been conducted to define the range of amplitudes of excitations and evaluate the internal warming of rubbers during the full structure validation test. The experimental characterization of the suspension is based on rubbers mounts and their interfaces with the cooling module, in order to take into account in a unique model all nonlinearities due to the viscoelastic behavior, the slidings, and the friction. Measured force-deflection hysteretic cycles in axial and radial direction are post-processed with an expert system developed to obtain the parameters of the retained model: the modified Dahl’s model, generalized to viscoleastic aspect. This process has been developed with Octave/Matlab code. Interpolation and extrapolation methods enable to obtain a good model response on the global operating range. These methods have been coded in an Abaqus UserSubroutine. Imposing random vibration excitation of a non linear mechanical system based on PSD imposes to take into account signal processing aspects. To evaluate response levels versus norms requirements, it’s mandatory to consider the time-frequency transfer. In addition, the size and the complexity of the total finite element model of the industrial structure don’t allow a global resolution in the time domain for all the degrees of freedom. Homogenization and dynamic reduction techniques are used to evaluate the response of the system submitted to large frequency range excitations, and to analyse the behavior of the suspension.