Thèse numérisée par la Direction des bibliothèques de l'Université de Montréal.

Les recherches décrites dans ces pages portent sur l'étude de l'aurore à protons par télédétection spatiale en ultraviolet lointain. Ce travail s'inscrit dans le cadre de la mission IMAGE, dont l'instrument FUV offre la possibilité d'observer les émissions aurorales dues à la précipitation d'électrons et de protons. L'imageur spectral SI12, construit au Centre Spatial de Liège, donne une image de la raie Lyman-alpha déplacée par effet Doppler produite par la désexcitation de l'hydrogène. Il permet l'observation globale de l'ovale auroral avec une résolution temporelle de 2 minutes. Les données couplées à un modèle de transport et de dégradation d'énergie des particules qui précipitent ont été utilisées pour étudier l'aurore à protons tant d'un point de vue morphologique que d'un point de vue quantitatif. La première étape du travail consistait à valider les observations et à calibrer la méthode développée pour obtenir les flux énergétiques des particules accélérées vers l'atmosphère à partir des taux de comptage SI12. Des comparaisons des observations SI12 et des mesures in situ de particules des satellites NOAA ont montré un désaccord de 50%, qui a été pris en considération dans la suite du travail. J'ai ensuite entrepris une étude statistique des caractéristiques morphologiques de l'ovale protonique. Cette statistique présente un très bon accord avec les études précédentes, bien que celles-ci soient basées sur des mesures locales de la précipitation de protons. Les résultats quantitatifs quant à l'intensité de la précipitation des protons ont montré une variation saisonnière, avec un maximum dans l'hémisphère estival. J'ai également mis en évidence l'influence de l'orientation de la composante nord/sud du champ magnétique interplanétaire et de la pression dynamique du vent solaire sur la précipitation de protons tant au point de vue de l'énergie accumulée durant la phase de croissance, que des processus de déclenchement de leur précipitation et des processus de précipitation eux-mêmes. L'analyse de la dynamique de l'ovale auroral protonique durant les phases de croissance des sous-tempêtes magnétiques a montré que le mouvement vers l'équateur des frontières de l'ovale pouvait être local, global ou une superposition des deux. Ce résultat est interprété comme la combinaison d'un étirement global de la magnétosphère dû à l'influence du vent solaire et d'un étirement local trouvant son origine dans le feuillet de plasma interne à la magnétosphère. Dans la dernière étape du travail, j'ai analysé les conséquences de la précipitation aurorale sur l'ionosphère. J'ai montré que les conductivités de Hall et de Pedersen et leur évolution temporelle pouvaient être déduites de la contribution des trois imageurs ultraviolets à une échelle spatiale globale. Le résultat principal de cette partie montre que la contribution des protons à la puissance et aux conductances aurorales ne peut localement pas être négligée par rapport à celle des électrons.

Pour étudier le rôle de la mitochondrie dans la tolérance à la sécheresse, la réponse à la contrainte hydrique a été comparée entre une lignée sauvage (WT) et un mutant CMSII (Cytoplasmic Male Sterile) de Nicotiana sylvestris. Chez le mutant CMSII, le complexe I mitochondrial est absent et la respiration est assurée par les NAD(P)H déshydrogénases alternes et elle est maintenue à un niveau supérieur de l’ordre de 20 à 30% à celui du WT. La différence observée entre les plantes WT et CMSII met en jeu non seulement le fonctionnement mitochondrial, mais également le fonctionnement des chloroplastes. En effet, l’activité photosynthétique du mutant est plus faible que celui du WT et elle est corrélée avec une plus faible conductance stomatique (gs) et mésophyllienne (gm).Après l’arrêt de l’arrosage, on observe que le contenu relatif en eau (RWC) diminue plus lentement chez les feuilles du CMSII. Ceci n’était pas le résultat d’une plus petite surface de transpiration ou d’une masse racinaire d’absorption plus élevé puisque le rapport partie aérienne/racine et la surface foliaire totale ont été similaires au début de l’expérience chez les deux génotypes. De plus la mutation n’a pas induit des changements au niveau des paramètres hydriques (P0, PTLP, RWCTLP) ni au niveau de la densité stomatique. La tolérance des plantes CMSII a été le plus probablement la conséquence de sa plus faible transpiration en conditions bien hydratées et aux premiers jours de déshydratation et non pas d’une meilleure efficacité d’absorption de l’eau puisque le contenu en eau du sol reste plus élevé chez CMSII après l’arrêt de l’arrosage. La plus faible conductance stomatique chez le CMSII bien hydraté a été expliquée par sa plus faible conductance hydraulique. De plus, contrairement au WT, le niveau des acides aminés totaux diminue au cours de la déshydratation lorsque le contenu en protéines solubles augmente chez les feuilles du CMSII, suggérant une accélération de la remobilisation des acides aminés. D’autre part, il a été aussi montré que le mutant CMSII est capable de s’acclimater mieux à la sécheresse que le WT lorsqu’ils ont été maintenus à un RWC de 80 % sur plusieurs jours. Sous ces conditions, la photosynthèse reste plus élevée chez le mutant que chez le WT. Cette meilleure acclimatation corrèle avec une plus forte photorespiration du CMSII sous conditions bien hydratées et sous conditions d’acclimatation. La photorespiration chez CMSII et le WT a été estimée par le transport électronique dévolu à l’oxygénation de RuBP et en plus par l’accumulation des métabolites impliqués dans la photorespiration. D’une part, l’acclimatation à la sécheresse diminue gm plus fortement chez le WT que chez le CMSII. D’autre part, le WT accumule la glycine ce qui laisse supposer que le glycine décarboxylase mitochondrial est plus affectée chez le WT que chez le CMS et inhibe ainsi la photorespiration. En effet, cette plus faible photorespiration chez le WT affecte les réactions primaires de la photosynthèse par une accumulation d’un gradient de protons estimé par le quenching non-photochimique (NPQ) de la fluorescence chlorophyllienne ce qui induit une diminution du transport électronique des réactions primaires de la photosynthèse. / To investigate the role of mitochondria in drought stress, the response to water deprivation was compared between Nicotiana sylvestris wild type (WT) plants and the CMSII respiratory complex I mutant. In CMSII, alternative NAD(P)H-dehydrogenases bypassing complex 1 allow respiration.. The difference of mitochondrial function between WT and CMSII plants affect also photosynthesis. The CMSII has lower photosynthetic actitvity than the WT and lower stomatal (gs) and internal (gm) conductances to CO2. When watering of plants with similar leaf surface and similar shoot/root ratio was stopped the relative water content (RWC) declined faster in WT as compared to CMSII leaves. Furthermore, CMSII and WT leaves had the same osmotic potential at leaf saturation (P0) and at leaf turgor lost pressure (PTLP) and the same stomatal density. The slower decline of RWC in CMSII, compared to WT leaves, was most likely the consequence of the lower stomatal conductance (gs) under well-watered conditions and during the first days after withholding watering, The lower stomatal conductance of well-watered CMSII leaves correlated with a lower hydraulic conductance of leaves. Remarkably, total free amino acid levels declined and total soluble protein content increased in CMSII leaves, while the opposite was observed in WT leaves. This suggests protein synthesis in CMSII but protein degradation in WT leaves during drought stress. We also show that CMSII leaves better acclimate to drought stress than the WT leaves. After several days at 80 % RWC , photosynthesis is higher in the mutant than in WT. As compared to the WT, the mutant shows higher rates of photorespiration before and after acclimation to drought.The strong accumulation of glycine in the WT suggests that photorespiration may be limited at the level of glycine decarboxylase. In addition, after acclimation to drought gm declined markedly in WT but not in CMSII leaves, thus further limiting CO2 supply for photosynthesis in the WT. The resulting lower photosynthesis and photorespiration in WT leaves affect also the primary reaction of photosynthesis by increasing the non-photochemical fluorescence quenching (NPQ) and decreasing linear electron transport.

L'objet de cette thèse est l'étude d'une classe importante de marches aléatoires en milieu aléatoire, appelée marches aléatoires avec conductances aléatoires. Nous présentons trois principaux résultats montrant des comportements opposés, irrégulier et standard du noyau de la chaleur des marches aléatoires avec conductances aléatoires à queue polynômiale. Les deux premiers (cf. Chapitre 2) portent sur les marches aléatoires simples dans $\Z^d, d>1$, gouvernées par une famille de conductances aléatoires i.i.d. à valeurs dans l'intervalle $[0,1]$, avec une queue polynomiale d'exposant $\gamma$ au voisinage de $0$. Nous montrons en premier lieu pour toute dimension supérieure à $4$ que la probabilité de retour après $2n$ sauts décroit de façon irrégulière en ce sens qu'elle admet une borne inférieure que l'on peut rendre, à un terme sous-polynomial près, aussi proche que l'on veut de $1/n^{2}$ en laissant le paramètre $\gamma$ tendre vers $0$. En considérant le même modèle et à l'opposé du premier résultat, nous montrons en second lieu pour toute dimension $d$ supérieure à $2$ que le noyau de la chaleur de la marche aléatoire admet une borne supérieure que l'on peut rendre, à un terme sous-polynomial près, aussi proche que l'on veut de la borne standard $1/n^{d/2}$ en laissant le paramètre $\gamma$ tendre vers l'infini. Nous considérons dans le troisième résultat (cf. Chapitre 3) les mêmes chaînes de Markov mais en temps continu et étudions la décroissance de la probabilité de retour asymptotique. Nous prouvons pour tout $\gamma> d/2$ que la dimension spectrale est standard, i.e. égale à $d$. Une conséquence prévisible de ce résultat est que ceci reste tout aussi vrai en temps discret.

O método de Lyapunov, também conhecido como método direto, é eficiente para análise de estabilidade transitória em sistemas de potência. Tal método possibilita a análise de estabilidade sem requerer o conhecimento das soluções das equações diferenciais que modelam o problema. A maior desvantagem da utilização dos métodos diretos, é sem dúvida encontrar uma função (V) que satisfaça as condições do Teorema de Lyapunov, ou seja, V > 0 e V \'< ou =\' 0. Durante muitos anos a inclusão das condutâncias de transferência na modelagem do sistema de potência, com a rede reduzida aos nós dos geradores, foi um assunto que despertou interesse em vários pesquisadores. Em 1989, Chiang provou a não existência de uma Função de Lyapunov para sistemas de potência quando as condutâncias de transferência são consideradas. Essas condutâncias de transferência são responsáveis por gerar regiões no espaço de estados onde tem-se V > 0, não satisfazendo as condições do Teorema de Lyapunov. Recentemente, Rodrigues, Alberto e Bretas (2000) apresentaram a Extensão do Princípio de Invariância de LaSalle, onde é permitido que a Função de Lyapunov possua, em algumas regiões limitadas do espaço de estados, a derivada positiva. Neste caso, estas funções passam a ser denominadas Funções de Lyapunov Estendidas (FLE). Neste trabalho, são utilizadas a Extensão do Princípio de Invariância de LaSalle e as Funções de Lyapunov Estendidas para a análise de estabilidade transitória, considerando o efeito das condutâncias de transferência na modelagem do problema. Para isto, são propostas Funções de Lyapunov Estendidas para modelos de sistemas de potência que não apresentam uma Função de Lyapunov no sentido usual. Essas FLE\'s são propostas tanto para sistemas de 1-máquina versus barramento infinito quanto para sistemas multimáquinas. Para a obtenção de boas estimativas do tempo de abertura, nos estudos de estabilidade transitória, é proposto um algoritmo iterativo. Este algoritmo fornece uma boa estimativa local da área de atração do ponto de equilíbrio estável de interesse. / The method of Lyapunov, one of the direct method, is efficient for transient stability analysis of power systems. The direct methods are well-suited for stability analysis of power systems, since they do not require the solution of the set of differential equations of the system model. The great difficulty of the direct methods is to find an auxiliary function (V) which satisfies the conditions of Lyapunov\'s Theorem V > 0 and V \'< or =\' 0. For many years the inclusion of the transfer conductances in the power system model, with the reduced network, is a issue of interest for several researchers. In 1989, Chiang studied the existence of energy functions for power systems with losses and he proved the non existence of a Lyapunov Function for power systems when the transfer conductance is taken into account. The transfer conductances are responsible for generating regions in the state space where the derivative of V is positive. Therefore, the function V is nor a Lyapunov Function, because its derivative is not semi negative definite. Recently, an Extension of the LaSalle\'s Invariance Principle has been proposed by Rodrigues, Alberto and Bretas (2000). This extension relaxes some of the requirements on the auxiliary function which is commonly called Lyapunov Function. In this extension, the derivative of the auxiliary function can be positive in some bounded regions of the state space and, for distinction purposes, it is called, as Extended Lyapunov Function. Inthis work, the Extension of the LaSalle\'s Invariance Principle and the Extended Lyapunov Function are used for the transient stability analysis of power systems with the model taking transfer conductances in consideration. For at purpose in this research, Extended Lyapunov Functions for power system models which do not have Lyapunov Functions in the usual sense are proposed. Extended Lyapunov Functions are proposed for a single-machine-infinite- bus-system and multimachine systems. For obtaining good estimates of the critical clearing time in transient stability analysis, an iterative algorithm is proposed. This algorithm supplies a good local estimate of the attraction area for the post fault stable equilibrium point.

Une vision externe du caloduc permet de le considérer comme un système passif de transfert de chaleur assimilable à un matériaux de conductivité thermique 100 à 1000 fois plus importante que le cuivre. Les multiples applications sont évoquées au travers d'exemples. Leur principe de fonctionnement est basé sur l'exploitation du changement de phase liquide vapeur et la circulation d'un fluide à l'état de saturation. Ainsi, nous proposons une classification fine permettant de réduire, aux caloducs circulaires avec ou sans structure capillaire et soumis ou non à des forces de volume, notre analyse détaillée de la physique régissant le fonctionnement des caloducs.<br />Les différentes limites de fonctionnement sont abordées en détails à partir d'une étude bibliographique importante. Des hypothèses nouvelles sont avancées pour le calcul de la limite sonique.<br />Au travers d'une vaste étude bibliographique, nous abordons les notions de conductances thermiques dans la zone évaporateur et condenseur du caloduc, que ce soit pour les caloducs ayant une structure capillaire ou pour les thermosiphons à tube lisse. Cette observation minutieuse de la physique interne nous a conduit à examiner, à l'évaporateur, les échanges thermiques à travers des films liquides minces en introduisant la notion de pression de disjonction et de résistance d'évaporation.<br />D'autre part, un montage expérimental complexe et unique au monde à notre connaissance nous a permis de tester trois caloducs cuivre-méthanol sur une plage de température de [0, 100°C] pour des puissances jusqu'à 1400 W et soumis à des champs d'accélération de 1 à 9 g.<br />Si une partie des résultats expérimentaux obtenus est conforme à nos attentes, de nombreux comportements restent inexpliqués mais permettent d'orienter les recherches futures pour augmenter la prédictibilité des modèles ou des corrélations sur de vastes gammes d'évolution des paramètres (température, puissance, accélération, fluide, matériau, dimensions, structure capillaire, ...)

O método de Lyapunov, também conhecido como método direto, é eficiente para análise de estabilidade transitória em sistemas de potência. Tal método possibilita a análise de estabilidade sem requerer o conhecimento das soluções das equações diferenciais que modelam o problema. A maior desvantagem da utilização dos métodos diretos, é sem dúvida encontrar uma função (V) que satisfaça as condições do Teorema de Lyapunov, ou seja, V > 0 e V \'< ou =\' 0. Durante muitos anos a inclusão das condutâncias de transferência na modelagem do sistema de potência, com a rede reduzida aos nós dos geradores, foi um assunto que despertou interesse em vários pesquisadores. Em 1989, Chiang provou a não existência de uma Função de Lyapunov para sistemas de potência quando as condutâncias de transferência são consideradas. Essas condutâncias de transferência são responsáveis por gerar regiões no espaço de estados onde tem-se V > 0, não satisfazendo as condições do Teorema de Lyapunov. Recentemente, Rodrigues, Alberto e Bretas (2000) apresentaram a Extensão do Princípio de Invariância de LaSalle, onde é permitido que a Função de Lyapunov possua, em algumas regiões limitadas do espaço de estados, a derivada positiva. Neste caso, estas funções passam a ser denominadas Funções de Lyapunov Estendidas (FLE). Neste trabalho, são utilizadas a Extensão do Princípio de Invariância de LaSalle e as Funções de Lyapunov Estendidas para a análise de estabilidade transitória, considerando o efeito das condutâncias de transferência na modelagem do problema. Para isto, são propostas Funções de Lyapunov Estendidas para modelos de sistemas de potência que não apresentam uma Função de Lyapunov no sentido usual. Essas FLE\'s são propostas tanto para sistemas de 1-máquina versus barramento infinito quanto para sistemas multimáquinas. Para a obtenção de boas estimativas do tempo de abertura, nos estudos de estabilidade transitória, é proposto um algoritmo iterativo. Este algoritmo fornece uma boa estimativa local da área de atração do ponto de equilíbrio estável de interesse. / The method of Lyapunov, one of the direct method, is efficient for transient stability analysis of power systems. The direct methods are well-suited for stability analysis of power systems, since they do not require the solution of the set of differential equations of the system model. The great difficulty of the direct methods is to find an auxiliary function (V) which satisfies the conditions of Lyapunov\'s Theorem V > 0 and V \'< or =\' 0. For many years the inclusion of the transfer conductances in the power system model, with the reduced network, is a issue of interest for several researchers. In 1989, Chiang studied the existence of energy functions for power systems with losses and he proved the non existence of a Lyapunov Function for power systems when the transfer conductance is taken into account. The transfer conductances are responsible for generating regions in the state space where the derivative of V is positive. Therefore, the function V is nor a Lyapunov Function, because its derivative is not semi negative definite. Recently, an Extension of the LaSalle\'s Invariance Principle has been proposed by Rodrigues, Alberto and Bretas (2000). This extension relaxes some of the requirements on the auxiliary function which is commonly called Lyapunov Function. In this extension, the derivative of the auxiliary function can be positive in some bounded regions of the state space and, for distinction purposes, it is called, as Extended Lyapunov Function. Inthis work, the Extension of the LaSalle\'s Invariance Principle and the Extended Lyapunov Function are used for the transient stability analysis of power systems with the model taking transfer conductances in consideration. For at purpose in this research, Extended Lyapunov Functions for power system models which do not have Lyapunov Functions in the usual sense are proposed. Extended Lyapunov Functions are proposed for a single-machine-infinite- bus-system and multimachine systems. For obtaining good estimates of the critical clearing time in transient stability analysis, an iterative algorithm is proposed. This algorithm supplies a good local estimate of the attraction area for the post fault stable equilibrium point.

L’oxaliplatine (OXA) est un anticancéreux couramment employé en oncologie, son efficacité étant reconnue en première ligne dans le traitement de nombreux cancers. Cependant, son utilisation est limitée par l’apparition de neuropathies impactant la qualité de vie du patient. Pour identifier et étudier des marqueurs cliniques, plasmatiques et électrochimiques de neuropathie chronique induite par l’oxaliplatine (NPIO), deux études pilotes (LIPIDOXA et CANALOXA) ont été réalisées. Les patients de l’étude LIPIDOXA ont été inclus avant traitement par OXA et ont été évalués avant traitement, tout au long du traitement et six mois après l’arrêt. Entre mai 2014 et juin 2016, 35 patients ont été inclus. L’apparition d’une hypoesthésie thermique a été mise en évidence six mois après arrêt de l’OXA, révélateur d’un processus neurotoxique qui se poursuit même après l’arrêt du traitement. L’étude des marqueurs plasmatiques a révélé, en fin de traitement, une augmentation de la production de prostaglandines E2, un niveau moins élevé de stress oxydant chez les patients souffrant d’une neuropathie de grade 2 ainsi qu’une diminution de la concentration des triglycérides polyinsaturés associée à une tendance à l’augmentation des acides gras libres polyinsaturés. Les patients de l’étude CANALOXA, tous neuropathiques, ont été inclus en cours de traitement par OXA et évalués une seule fois. Entre avril 2016 et mars 2017, 36 patients ont été inclus. Les valeurs des conductances électrochimiques de la peau (ESC) étaient pathologiques pour un tiers des patients. Les valeurs d’ESC étaient bien corrélées avec le score de douleur neuropathique.Ces marqueurs sont cependant peu spécifiques et semblent d’apparition tardive. Aussi, ils sont difficilement utilisables en pratique clinique pour un éventuel suivi de la tolérance neurologique au traitement. Au-delà de l’apport clinique et thérapeutique modéré, ce travail renforce la compréhension de la physiopathologie de la NPIO dans le domaine de la neuropathie à petites fibres, les processus inflammatoires associés et la perturbation du métabolisme lipidique chez le patient traité par oxaliplatine.La complémentarité et les similarités de ces travaux rappellent que la prise en charge du patient doit être globale et conjuguer des éléments variés au sein desquels le patient doit rester au cœur de la gestion de la NPIO. Ce projet de recherche s’inscrit dans une dynamique d’amélioration continue et les résultats des études pilotes constituent une base pour approfondir les travaux. / Oxaliplatin (OXA) is an anti-cancer drug widely used in oncology, its effectiveness is recognized in first-line chemotherapy regimen in many cancers. However, its use is limited by an onset of a disabling peripheral neuropathy with a negative association with quality of life. To identify and study clinical, plasma and electrochemical markers of oxaliplatin-induced chronic neuropathy (OIPN), two pilot studies (LIPIDOXA and CANALOXA) were conducted.Patients in the LIPIDOXA study were included prior to OXA treatment and were evaluated before treatment, during treatment, and six months after discontinuation. Between May 2014 and June 2016, 35 patients were included. The onset of thermal hypoaesthesia was highlighted six months after OXA completion, revealing a neurotoxic process that extends beyond treatment. The study of plasma markers revealed, at the end of treatment, an increase of prostaglandins E2 release, a lower level of oxidative stress in patients suffering from grade-2 neuropathy as well as a decrease in the concentration of polyunsaturated triglycerides associated with a tendency in polyunsaturated free fatty acids increase. The patients in the CANALOXA study, all neuropathic, were included during treatment with OXA and evaluated once. Between April 2016 and March 2017, 36 patients were included. The values of the electrochemical conductances of the skin (ESC) were pathological for one third of patients. ESC values were well correlated with the neuropathic pain score.However, these markers are sparsely specific and seem to be of late onset. Thus, they are are not suitable for a possible monitoring of the neurological tolerance of the treatment in the current care practice. Beyond the moderate clinical and therapeutic contribution, this work strengthens the understanding of the pathophysiology of OIPN in the field of small-fiber neuropathy, the associated inflammatory processes and the disruption of lipid metabolism in the patient treated with oxaliplatin.The complementarity and similarity of this work remind us that the patient care must be global and combine various elements in which the patient must remain at the heart of the management of the OIPN. This project takes part in a dynamic of continuous improvement and the results of these pilot studies constitute the basis for further research in this field.

Ce mémoire de thèse constitue une contribution à l'investigation des génératrices destinées à une application éolienne en mer. Le but est de rechercher les solutions optimales dans un domaine de recherche vaste contenant entre huit et onze variables d'optimisation, tout en respectant un cahier de charge bien précis. Afin d'y parvenir, un modèle multi-physique a été développé permettant la détermination des distributions du champ électromagnétique et de température dans les topologies de génératrices choisies avec prise en considération des non-linéarités des matériaux. La méthode des constantes localisées couplée aux fonctions d'interpolations a été choisie comme solution offrant un bon rapport temps de calculs/précision, prenant ainsi en considération les caractéristiques des matériaux (thermiques et magnétiques). Le modèle développé a été couplé à un algorithme d'optimisation génétique, NSGAII, permettant dans un premier temps ; d'investiguer le poids nécessaire des aimants permanents et des parties actives de deux topologies de génératrices synchrones à aimant permanent où dans la première les aimants sont montés en surfaces et dans la seconde insérés en concentration de flux, trois puissances différentes ont été investiguées 5, 8 et 15 [MW]. Dans un second temps, l'intégration d'un multiplicateur de vitesse magnétique dans la chaîne de conversion éolienne a été étudiée à travers la même approche optimale utilisée précédemment tout en comparant le poids nécessaires des aimants permanents pour ce type d'entrainement avec celle des topologies à attaque directe (sans multiplicateur de vitesse). / The aim of this PhD report is the investigation of electrical generators dedicated to an offshore wind turbine application. The main goal is to find optimal solutions in a vast research domain containing between eight and eleven optimization variables, while respecting a the imposed constraints. In order to achieve this goal, a multi-physics model was developed allowing the determination of the electromagnetic and temperature fields distributions in the selected topologies. Lumped models coupled to the interpolation functions were chosen as a solution offering a good computation time / precision ratio, thus taking into consideration the characteristics of the materials (thermal and magnetic). The developed model was coupled to a genetic optimization algorithm, NSGAII, allowing at first; the investigation of the necessary permanent magnets weight and the active parts one of two permanent magnet synchronous generator topologies where in the first the magnets are mounted in surfaces and in the second inserted in order to have flux concentration, three different powers have been investigated 5, 8 and 15 [MW]. Second, the integration of a magnetic gear in the wind energy conversion chain was studied using the same optimal approach previously used, while comparing the weight required of permanent magnets for this type of training with that of direct drive topologies (without gearbox).

Local field potentials (LFPs), extracellular potentials that reflect localized electrical activity, have long been used as a window to understand the behavioural dependence and mechanistic aspects of brain physiology. A principal premise that has driven the interpretation of LFPs is that they largely reflect the synaptic drive that impinges on neurons located in the vicinity of the recording microelectrode. An implicit, yet critical, assumption that led to the emergence of this premise is that dendrites, the structures onto which most synaptic inputs project, are purely passive compartments. However, there is a growing body of evidence demonstrating that dendrites express a plethora of active conductance, like voltage-gated ion channels, several of which are active in the subthreshold regime. These subthreshold-activated ion channels and their intra-neuronal localization profiles play widely acknowledged regulatory roles in the physiology, plasticity and pathophysiology of synapses and neurons. Despite this, the implications for the existence of these subthreshold conductances on constituent oscillatory patterns in LFPs and on the phase of neuronal spiking with reference to oscillating LFPs have surprisingly remained unexplored. The aim of this thesis is to examine if there exists a role of subthreshold conductances in regulating LFPs and the phase of spikes with reference to these LFPs. To address this, we chose to study LFPs and spikes from the CA1 region of the rat hippocampus, with hyperpolarization-activated cyclic-nucleotide-gated (HCN) channels forming the specific subthreshold conductance of focus. The reasons behind these choices were manifold. First, CA1 pyramidal neurons are arranged in a laminar open-field configuration, making the interpretation of the source-sink formation in this region relatively tractable. Second, the dendrites of these neurons are endowed with a multitude of subthreshold conductances whose expression profiles, physiology and plasticity have been characterized in great detail. Third, this brain region has been implicated in coding for episodic and spatial memories. The phase of the spikes of the CA1 pyramidal neurons, with reference to the LFP, is believed to serve as a code that can be used to decode the location of the animal. Given that the most dominant LFP pattern seen in the CA1 region during such active exploration (and possibly encoding of spatial memories) consists of oscillations in the 4–10 Hz theta frequency band, we decided to focus our study on theta-frequency LFPs. Finally, consistent with the choice of the specific band of LFP frequencies, we focused on HCN channels because of their predominantly dendritic expression and their ability to bestow resonance and impedance phase lead, both in the theta-frequency range, on CA1 pyramidal neurons. In exploring the role of HCN channels on LFPs, we used a multi-compartmental morphologically realistic CA1 pyramidal neuron model and introduced an HCN channel conductance gradient that was constrained with several experimental measurements. This neuron was driven by dendritic excitatory synapses and perisomatic inhibitory synapses, both theta-modulated with a phase difference of +60º between their arrivals timings. We increased the excitatory synaptic conductance with distance from the soma to account for the fact that irrespective of the location of the synapse in the dendrites, the unitary excitatory post-synaptic potential remains the same at the soma. Employing these model configurations, we generated 25 different synaptic distributions on the same neuronal morphology to account for the input variability and for each of these models, we recorded transmembrane currents from all the compartments, for 8–10 cycles of the theta-modulated inputs. To model LFPs using the forward modelling scheme of line source approximation, we designed a cylindrical neuropil of 40 µm height and 100 µm radius and inserted a virtual linear electrode with 7 contact points distributed on the probe at the canter of the neuropil such that we could compute the LFP at all the strata of the CA1 region. Accounting for the volume of the neuropil and the density of neurons in this region, we took 440 instances of the morphology, rotated them at uniformly distributed angles, and distributed the somata of these model neurons within the neuropil. Each of these 440 neurons received transmembrane currents from one of the 25 models picked uniformly. With a passive model, where we did not introduce HCN channels, we expectedly observed the formation of a source-sink structure that expressed as a progressive phase shift spanning different strata, owing to the perisomatic inhibitory currents coupled with the dendritic excitatory currents. On introducing a somatodendritic gradient of HCN conductance with identical input conditions, we observed a phase lead in the LFPs across all the layers, with the magnitude of the lead increasing with distance from the soma in a manner that was correlated with the increase in HCN conductance. Next, we computed spike phases, for each of the 25 neuron models, with reference to the stratum pyramidale (SP) LFP for model configurations with and without HCN channels. We found that the spikes showed a phase lag in the presence of a gradient of HCN channels when compared to the spike phases measured from the passive neuron models. Finally, we computed the coherence of spikes across all the 25 passive or 25 active (with HCN channels) neuron models and found that the presence of HCN channels greatly enhanced spike phase coherence across neurons. Together, these results demonstrate that the presence of HCN channels introduces a lead in the theta-frequency LFP phase, a lag in the associated spike phase, and a significant enhancement of spike phase coherence. Exploring the robustness of these findings to the model configuration, we first found these conclusions to be robust to increases in neuropil size (400-µm diameter neuropil with 1797 neurons, and 1-mm diameter neuropil with 11297 neurons). Next, we introduced heterogeneities in the population of neurons (in terms of morphology as well as passive and active properties) that formed the neuropil, and found our conclusions to be invariant to such degeneracy in the underlying neuronal population. It has been observed that under certain pathological conditions like epilepsy, an entire population of CA1 neurons can undergo intrinsic plasticity, such as global (i.e., across the entire neuronal topograph) downregulation of HCN channels. To assess the impact of such up/downregulation on LFPs, we respectively increased/decreased HCN channel conductance globally in our model neurons, and found the magnitude of the lead in the LFP phase to progressively increase with HCN-channel conductance. Similarly, the magnitude of the spike-phase lag and the spike phase coherence also progressively increased as functions of HCN-channel conductance. Although such population-level global intrinsic plasticity is observed under pathological conditions, a more physiological scenario would be when a single neuron, in the process of encoding new inputs (such as encoding spatial or episodic memories), undergoes intrinsic plasticity. To assess this, we increased or decreased HCN-channel conductance specifically in a single neuron placed closest to the electrode, while leaving the HCN expression in other neurons of the neuropil at the baseline level. Expectedly, we did not find significant changes in LFP amplitude or phase, but we did find a significant lag in the spike phase preference of the neuron that underwent an upregulation of HCN conductance. Another physiological scenario is when the rat experiences a reward or exhibits anxiety-like behaviour, which can lead to changes in hormonal or neuromodulator concentrations. These changes, functioning through the activation of G-protein coupled receptors and the consequent elevation of cytosolic cyclic adenosine monophosphate (cAMP) concentrations, could shift the half-maximal activation voltage ( V1/2 ) of HCN channels to a more depolarized potential. Would such a shift in V1/2 impact LFPs and spike phases in a manner similar to that observed with increasing the conductance of HCN channels? Assessing this within our modeling framework, we found that shifting the V1/2 by +5 mV resulted in an increased lead in the LFP phase, an increased lag in the spike phase and an enhanced spike phase coherence compared to the case with a hyperpolarized V1/2 . What are the biophysical mechanisms that underlie these robust changes observed in LFPs and spike phases observed as a consequence of these several ways of increasing the current through HCN channels? We reasoned that our observations could be explained by one of the two distinct changes conferred on CA1 pyramidal neuron physiology by the presence of HCN channels. First, in the presence of HCN channels, the voltage response of CA1 pyramidal neurons shows a phase lead with reference to a sinusoidal current input (inductive phase lead) in the theta frequency range. Second, HCN channels regulate the excitability of these cells by decreasing the input resistance and impedance amplitude. To delineate the differential role of the inductive changes vs. changes in excitability, we replaced HCN channels by a faster variant (HCNFast) such that neuronal excitability remained the same while abolishing the inductive phase lead in the theta band. On doing so, we found that the lead in the LFP phase and the lag in the spike phase brought about by HCN channels was partially reversed when HCN conductance values were low. However the reversal was not substantial when HCN conductance values were high, suggesting that the inductive phase component dominates at lower HCN channel conductances, whereas the excitability component plays a critical role at higher HCN conductances. Akin to intrinsic plasticity mentioned above, under certain pathological conditions, an entire population of neurons can undergo scaling of their excitatory or inhibitory synapses. In assessing the implications for such synaptic plasticity, we first found that our conclusions on the roles of HCN channels in introducing a lead in the LFP phase, a lag in the spike phase and an enhancement of spike phase coherence were invariant to the specific values of synaptic conductances, or the phase difference between excitatory and inhibitory theta-modulated inputs. While these observations further established the robustness of the changes brought about by HCN channels to LFPs and associated spikes, we next asked whether synaptic plasticity, mediated by changes in subthreshold synaptic conductances, could itself bring about changes in the LFP and spike phase. Expectedly, we found that scaling up of excitatory synapses introduced a mild lag in the LFP phase and a lead in the spike phase, whereas scaling up of inhibitory synapses introduced a lead in the LFP phase and a lag in the spike phase. Finally, we observed a critical role of the arrival phase of inhibition with reference to excitation in altering both, the stratum pyramidale LFP and associated spike phases, with the magnitude of change in both the LFP and the spike phase roughly following the magnitude of the shift in the excitatory-inhibitory phase difference. However, in contrast to changes observed with HCN-channel plasticity, there was no significant change in spike phase coherence with any of the three forms of synaptic changes explored. Together, our results identify definite roles for HCN channels and synaptic receptors in phase-coding schemas and in the formation and dynamic reconfiguration of neuronal cell assemblies and present a clear case for the incorporation of subthreshold-activated ion channels, their gradients, and their plasticity into the computation of LFPs. Given the rich expression of several subthreshold ion channels — including HCN, A-type potassium and T-type calcium — in neuronal dendrites, future work could focus on the impact of subthreshold channels on LFPs recorded in different brain regions under different behavioral states. This thesis is organized into seven chapters. Chapter 1 provides the motivations for the study, introduces the aim of the study and poses the specific questions asked in our endeavor to understand the role of subthreshold conductances in regulating LFPs and spike phases. Chapter 2 discusses the physiological foundations and relevant literature that places the questions posed in the first chapter in the context of the aim of the thesis, with an emphasis on the literature on HCN channels. In chapter 3, we introduce the computational and theoretical foundations required to model neurons and to compute LFPs. In chapter 4, we look at the consequences of the presence of a non-uniform density of somatodendritic HCN channels on LFPs and spike phase and test the robustness of the effects observed. In chapter 5, we present our assessment of the impact of intrinsic plasticity/modulation of HCN channels on LFPs and spike phases, also exploring the biophysical mechanisms underlying such an impact. In chapter 6, we test if the observed effects still hold under synaptic plasticity, and assess the regulation of LFPs and spike phases by synaptic changes. In chapter 7, we summarize and conclude the results presented in the preceding chapters and provide some potential directions for future studies.