<p dir="ltr">This thesis presents advances in numerical methods for partial differential equations (PDEs) and optimization problems, with a focus on improving efficiency, stability, and accuracy across various applications. We begin by addressing 3D Poisson-type equations, developing a GPU-accelerated spectral-element method that utilizes the tensor product structure to achieve extremely fast performance. This approach enables solving problems with over one billion degrees of freedom in less than one second on modern GPUs, with applications to Schrödinger and Cahn<i>–</i>Hilliard equations demonstrated. Next, we focus on parabolic PDEs, specifically the Cahn<i>–</i>Hilliard equation with dynamical boundary conditions. We propose an efficient energy-stable numerical scheme using a unified framework to handle both Allen<i>–</i>Cahn and Cahn<i>–</i>Hilliard type boundary conditions. The scheme employs a scalar auxiliary variable (SAV) approach to achieve linear, second-order, and unconditionally energy stable properties. Shifting to a machine learning perspective for PDEs, we introduce an unsupervised learning-based numerical method for solving elliptic PDEs. This approach uses deep neural networks to approximate PDE solutions and employs least-squares functionals as loss functions, with a focus on first-order system least-squares formulations. In the realm of optimization, we present an efficient and robust SAV based algorithm for discrete gradient systems. This method modifies the standard SAV approach and incorporates relaxation and adaptive strategies to achieve fast convergence for minimization problems while maintaining unconditional energy stability. Finally, we address optimization in the context of machine learning by developing a structure-guided Gauss<i>–</i>Newton method for shallow ReLU neural network optimization. This approach exploits both the least-squares and neural network structures to create an efficient iterative solver, demonstrating superior performance on challenging function approximation problems. Throughout the thesis, we provide theoretical analysis, efficient numerical implementations, and extensive computational experiments to validate the proposed methods. </p>

Cette thèse de doctorat composée de trois chapitres contribue au développement de tests statistiques et à analyser la transmission financière dans un cadre multivarié hétéroscédastique. Le premier chapitre propose deux tests du multiplicateur de Lagrange de constance des corrélations conditionnelles dans les modèles GARCH multivariés. Si l'hypothèse nulle repose sur des corrélations conditionnelles constantes, l'hypothèse alternative propose une première spécification basée sur des réseaux de neurones artificiels et une seconde représentée par une forme fonctionnelle inconnue qui est linéarisée à l'aide d'un développement de Taylor.Dans le deuxième chapitre, un nouveau modèle est introduit dans le but de tester la non-linéarité des (co)variances conditionnelles. Si l'hypothèse nulle repose sur une fonction linéaire des innovations retardées au carré et des (co)variances conditionnelles, l'hypothèse alternative se caractérise quant à elle par une fonction de transition non-linéaire : exponentielle ou logistique ; une configuration avec effets de levier est également proposée. Dans les deux premiers chapitres, les expériences de simulations et les illustrations empiriques montrent les bonnes performances de nos tests de mauvaise spécification.Le dernier chapitre étudie la transmission d'information en séance et hors séance de cotation en termes de rendements et de volatilités entre la Chine, l'Amérique et l'Europe. Le problème d'asynchronicité est considéré avec soin dans la modélisation bivariée avec la Chine comme référence. / This Ph.D. thesis composed by three chapters contributes to the development of test statistics and to analyse financial transmission in a multivariate heteroskedastic framework.The first chapter proposes two Lagrange multiplier tests of constancy of conditional correlations in multivariate GARCH models. Whether the null hypothesis is based on constant conditional correlations, the alternative hypothesis proposes a first specification based on artificial neural networks, and a second specification based on an unknown functional form linearised by a Taylor expansion.In the second chapter, a new model is introduced in order to test for nonlinearity in conditional (co)variances. Whether the null hypothesis is based on a linear function of the lagged squared innovations and the conditional (co)variances, the alternative hypothesis is characterised by a nonlinear exponential or logistic transition function; a configuration with leverage effects is also proposed.In the two first chapters, simulation experiments and empirical illustrations show the good performances of our misspecification tests.The last chapter studies daytime and overnight information transmission in terms of returns and volatilities between China, America and Europe. The asynchronicity issue is carefully considered in the bivariate modelling with China as benchmark.

Wards occupy significant proportions of hospital floor areas and due to their constant use, represent a worthwhile focus of study. Single-bed wards are specifically of interest owing to the isolation aspect they bring to infection control, including airborne pathogens, but threats posed by airborne pandemics and family-involvement in hospital care means cross-infection is still a potential problem. In its natural mode, ventilation driven by combined wind and buoyancy forces can lead to energy savings and achieve thermal comfort and high air change rates through secure openings. These are advantageous for controlling indoor airborne pathogens and external air and noise pollution. However, there is lack of detailed evidence and guidance is needed to gain optimum performance from available natural ventilation systems. This research is a proof of concept investigation into the feasibility and impact of natural ventilation systems targeting airflow rates, thermal comfort, heating energy and control of pathogenic bio-aerosols in hospital wards. In particular, it provides insights into the optimal areas of vent openings which could satisfy the complex three-pronged criteria of contaminant dilution, low heating energy and acceptable thermal comfort for occupants in a naturally ventilated single bed ward. The main aim of this thesis is the structured study of four systems categorised into three groups: Simple Natural Ventilation (SNV) in which single and dual-openings are used on the same external wall; Advanced Natural Ventilation (ANV) which is an emerging concept; and finally Natural Personalised Ventilation (NPV) which is an entirely new concept borne out of the limitations of previous systems and gaps in literature. The focus of this research is in the exploratory study of the weaknesses and potentials of the four systems, based on multi-criteria performances metrics within three architecturally distinct single-bed ward designs. In contributing to the body of existing knowledge, this thesis provides a better understanding of the performances of three existing systems while presenting the new NPV system. The analysis is based on dynamic thermal modelling and computational fluid dynamics and in the case of the NPV system, salt-bath experiments for validation and visualisation of transient flows. In all cases, wards were assumed to be free of mechanical ventilation systems that might influence the natural flow of air. The thesis meets three major objectives which have resulted in the following contributions to current knowledge: An understanding of the limitations and potentials of same-side openings, especially why and how dual-openings can be useful when retrofitted into existing wards. Detailed analysis of bulk airflow, thermal comfort, heating energy and room air distribution achievable from existing SNV and ANV systems, including insights to acceptable trickle ventilation rates, which will be particular useful in meeting minimum dilution and energy requirements in winter. This also includes qualitative predictions of the airflow pattern and direction obtainable from both systems. The innovation and study of a new natural ventilation system called Natural Personalised Ventilation (NPV) which provides fresh air directly over a patient s bed, creating a mixing regime in the space and evaluation of its comfort and energy performances. A low-energy solution for airborne infection control in clinical spaces is demonstrated by achieving buoyancy-driven mixing ventilation via the NPV system, and a derivative called ceiling-based natural ventilation (CBNV) is shown. A comparative analysis of four unique natural ventilation strategies including their performance rankings for airflow rates, thermal comfort, energy consumption and contaminant dilution or removal using an existing single-bed ward design as case study. Development of design and operational recommendations for future guidelines on utilising natural ventilation in single-bed wards either for refurbishment or for proposed designs. These contributions can be extended to other clinical and non-clinical spaces which are suitable to be naturally ventilated including treatment rooms, office spaces and waiting areas. The findings signify that natural ventilation is not only feasible for ward spaces but that there is opportunity for innovation in its application through further research. Future work could focus on related aspects like: impacts of fan-assisted ventilation for a hybrid flow regime; pre-heating of supply air; integration with passive heat recovery systems as well the use of full-scale experiments to fine-tune and validate findings.

We discuss a full-waveform based material profile reconstruction in two-dimensional heterogeneous semi-infinite domains. In particular, we try to image the spatial variation of shear moduli/wave velocities, directly in the time-domain, from scant surficial measurements of the domain's response to prescribed dynamic excitation. In addition, in one-dimensional media, we try to image the spatial variability of elastic and attenuation properties simultaneously. To deal with the semi-infinite extent of the physical domains, we introduce truncation boundaries, and adopt perfectly-matched-layers (PMLs) as the boundary wave absorbers. Within this framework we develop a new mixed displacement-stress (or stress memory) finite element formulation based on unsplit-field PMLs for transient scalar wave simulations in heterogeneous semi-infinite domains. We use, as is typically done, complex-coordinate stretching transformations in the frequency-domain, and recover the governing PDEs in the time-domain through the inverse Fourier transform. Upon spatial discretization, the resulting equations lead to a mixed semi-discrete form, where both displacements and stresses (or stress histories/memories) are treated as independent unknowns. We propose approximant pairs, which numerically, are shown to be stable. The resulting mixed finite element scheme is relatively simple and straightforward to implement, when compared against split-field PML techniques. It also bypasses the need for complicated time integration schemes that arise when recent displacement-based formulations are used. We report numerical results for 1D and 2D scalar wave propagation in semi-infinite domains truncated by PMLs. We also conduct parametric studies and report on the effect the various PML parameter choices have on the simulation error. To tackle the inversion, we adopt a PDE-constrained optimization approach, that formally leads to a classic KKT (Karush-Kuhn-Tucker) system comprising an initial-value state, a final-value adjoint, and a time-invariant control problem. We iteratively update the velocity profile by solving the KKT system via a reduced space approach. To narrow the feasibility space and alleviate the inherent solution multiplicity of the inverse problem, Tikhonov and Total Variation (TV) regularization schemes are used, endowed with a regularization factor continuation algorithm. We use a source frequency continuation scheme to make successive iterates remain within the basin of attraction of the global minimum. We also limit the total observation time to optimally account for the domain's heterogeneity during inversion iterations. We report on both one- and two-dimensional examples, including the Marmousi benchmark problem, that lead efficiently to the reconstruction of heterogeneous profiles involving both horizontal and inclined layers, as well as of inclusions within layered systems. / text

為了探討注意力與年齡因素對計時行為的影響，本實驗選用事前知道計時派典（prospective paradigm）與不涉及人為時間單位之知識的計時複製方式。針對注意力的因素，本研究採用雙重作業以操弄注意力資源集中或分散，檢測其對計時作業的影響。另針對年齡的因素，本研究比較低年級小學生、高年級小學生、及大學生成人三種不同年齡組的受試。 本研究利用四種計時作業，針對三個不同年齡層的受試進行對兩種不等的時距計時資料收集。計時作業及年齡因子的實驗設計採受試者間設計，而時距因子則採受試者內設計。研究結果發現（1）低年級組的計時準確性較高年級組或成人組差；（2）受試在雙重作業的計時表現的確較單一作業差，這個計時受干擾的結果，在複雜的雙重作業又較在簡單雙重作業明顯，且年紀愈小的受試所受干擾影響愈大；（3）三年齡層受試在雙重作業中皆有低估目標時距的現象，其中以低年級在複雜雙重作業中的低估程度最明顯；（4）低年級組的計時能力之穩定性較高年級或成人組低；（5）在無需對不斷出現之無意義刺激作偵測反應時，三年齡受試的計時準確性未受影響。 這些結果顯示計時的誤差的確隨著注意力需分派於非時間作業的增加，受試所複製的時距也隨之縮短。這項由於注意力分散所致的計時誤差在國小低年級組最明顯，但隨著年齡的成長及腦發展愈臻成熟，這項影響效果愈趨式微，致使計時的表現愈正確。 關鍵詞：時間複製、年齡與發展、雙重作業、注意力、小學生、純量計時 / The purpose of this study was to investigate the effects of the attention and age on time estimation based on the prospective timing paradigm. Four timing tasks, with different degrees of attentional requirement, were utilized to test timing of reproducing the target intervals in three groups of different ages including the lower-grade (M= 7.5 years), higher-grade (M= 11.2 years) elementary school students and adult (M= 20.1 years). The subjects in this work were asked to estimate a stimulus duration lasting for 7 or 14 s, during which they are required to either do or do not perform a concurrent non-temporal task. Thus, the experimental design for the present study was mixed with between-subject factors (age and task) and a within-subject factor (target interval). The results showed (1) the lower-grade children performed less accurate than the higher-grade children; (2) all subjects had worse timing performance in the dual-task condition than the single-task or control condition, which effect was most evident in the lower-grade children; (3) all subjects reproduced shorter interval as compared to the target interval, which effect was most apparently observed from the dual-task for the lower-grade children; (4) the lower-grade children had less reliability in timing; and (5) the effect of age on the single-task and control condition is not significant. Together, these data indicate that concurrent non-temporal task cause temporal reproduction shorter, especially under high attention demand in lower-grade children. In addition to the influence by the task complexity, the accuracy of timing estimation depends on the age. Keywords: time reproduction、age and development、dual task、attention、elementary school student、Scalar expectancy theory(SET)

Cette étude s'intéresse aux risques associés à la formation d'une ATmosphère EXplosive (ATEX) née d'une fuite d'hydrogène et de sa dispersion dans l'air ambiant. La fuite a été modélisée par un jet turbulent à densité variable libre, impactant sur une sphère de diamètre 20mm ou sur une plaque plane. Dans un premier temps, les champs de vitesses et de concentration ont été obtenus expérimentalement en proche sortie grâce à des mesures de Vélocimétrie par Images de Particules (PIV) et de Fluorescence Induite par Plan Laser sur l'acétone (PLIF). La turbulence et le mélange ont été caractérisés pour le cas d'un jet libre ou en présence d'un obstacle. A partir de ces mesures, la structure générale de l'écoulement a été étudiée à partir des champs moyens et fluctuants par comparaison avec les données de la bibliographie. Puis, les données issues des fluctuations ont été analysées statistiquement par l'étude des fonctions de densité de probabilité du scalaire. Ces travaux se sont poursuivis avec la mise en relation des résultats expérimentaux avec ceux obtenues par des simulations numériques DNS (Direct Numerical Simulation) utilisant la méthode Boltzmann sur Réseau (LBM) d'un scalaire passif dans un jet d'air. Cette étude a permis de recueillir et d'analyser des données supplémentaires sur le mélange d'un jet à masse volumique variable libre ou impactant. Ces données sont directement applicables à la maitrise des risques liés aux fuites d'hydrogène. / This study examines the risks associated with the formation of an explosive atmosphere from a hydrogen leak and its dispersion into the air. We considered the leak as a turbulent jet with density variable, free and impinging a $20,mm$ diameter sphere or a flat plate. Firstly, velocity and scalar fields have been measured experimentally in the near field through Particle Image Velocimetry (PIV) and acetone Planar Laser Induced Fluorescence (LIF). Turbulence and mixing have been defined in the case of free jet and impinging jet. From this measurements, the flow structure has been presented from the mean and fluctuating flow measurements by comparison with literature data. Next, the fluctuation scalar fields are studied with the probability density function method. Finally, a comparison has been conducted between the experiments and direct numerical simulation (DNS) of turbulence based on the lattice Boltzmann method (LBM) for passive scalar in air jet. This study is gathering and analyzing data on the mixing of jet with density variable, free and impinging jet. Such data is directly useful to identify and control risks incurred due to hydrogen leak.

Cette thèse est consacrée aux écoulements homogènes de bulles, ainsi qu'à leur couplage avec le transport d'un scalaire et la turbulence. Elle s'intéresse plus spécifiquement aux effets de taille finie, des interactions hydrodynamiques et de la microstructure de la suspension qui sont étudiés à l'aide de simulations numériques directes à l'échelle d'une seule bulle. La dynamique d'une suspension laminaire de bulles induite par la seule gravité est d'abord revisitée. L'influence de la fraction volumique sur la vitesse de dérive des bulles est établie analytiquement et numériquement pour une suspension parfaitement ordonnée, puis des ressemblances entre suspensions ordonnées et suspensions désordonnées sont mises en évidence. Ces résultats sont ensuite mis à profit pour la modélisation du transport d'un scalaire passif au sein d'une suspension laminaire, tel que décrit par une diffusivité effective tensorielle, et des différences essentielles entre systèmes ordonnés et systèmes désordonnés concernant le transport de scalaire sont mises en exergue. Enfin, la turbulence est prise en compte dans les simulations et son interaction avec une bulle de taille finie est caractérisée. Il est montré que le comportement dynamique d'une bulle de taille comparable à la microéchelle de Taylor ressemble qualitativement à celui d'une microbulle, avec, notamment, une préférence pour certaines régions caractéristiques de l'écoulement. Une définition de l'écoulement vu par la bulle compatible avec les modèles standards de masse ajoutée et de portance est finalement proposée / This thesis is devoted to the study of homogeneous bubbly flows and their coupling with scalar transport and turbulence. It focuses on the effects of finite size, hydrodynamic interactions, and suspension microstructure, which are investigated using direct numerical simulations at the bubble scale. The dynamics of laminar buoyancy-driven bubbly suspensions is first revisited. More specifically, the effect of volume fraction on the bubble drift velocity is clarified by connecting numerical results to theory for dilute ordered systems, and similarities between perfectly ordered and free disordered suspensions are evidenced. These results are then used for the modeling of passive scalar transport in laminar suspensions as described by an effective diffusivity tensor, and crucial differences between ordered and disordered systems with respect to scalar transport are highlighted. Lastly, turbulence is included in the simulations, and its interaction with a finite-size bubble is characterized. The behavior of a bubble as large as Taylor microscale is shown to share a number of common features with that of a microbubble, most notably, the flow sampled by the bubble is biased. A definition of the liquid flow seen by the bubble, as it enters in usual models for the added mass and the lift forces, is finally proposed

In dieser Arbeit berechnen wir mit Hilfe der Gitter-QCD Observablen, die in Beziehung zur Struktur des Nukleons stehen. Ein Teil dieser Arbeit beschäftigt sich mit Momenten von Parton-Verteilungsfunktionen. Solche Momente sind wichtig für das Verständnis der Nukleon-Struktur und werden durch globale Analysen von tief-inelastischen Streuexperimenten bestimmt. Eine theoretische, nicht-perturbative Rechnung der Momente in der Gitter-QCD ist möglich. Allerdings existiert, seit solche Gitter-QCD Rechnungen vorliegen, eine Diskrepanz zwischen diesen Rechnungen und den Ergebnissen globaler Analysen experimenteller Daten. Wir untersuchen, ob systematische Effekte für diese Diskrepanz verantwortlich sind, dabei studieren wir insbesondere die Effekte angeregter Zustände. Zudem führen wir eine erste Rechnung mit vier dynamischen Quark-Flavors durch. Ein weiterer Aspekt dieser Arbeit ist eine Machbarkeitsstudie zur Berechnung des skalaren Quark-Inhalts des Nukleons in der Gitter-QCD. Dieser bestimmt den Wirkungsquerschnitt der durch ein skalares Teilchen (z.B. ein Higgs-Teilchen) vermittelten Wechselwirkung eines schweren Teilchens mit einem Nukleon und kann somit einen Einfluss bei der Suche nach Dunkler Materie haben. Bisherige Gitter-Rechnungen dieser Größe besitzen große Unsicherheiten und sind daher von geringer Signifikanz für phenomenologische Anwendungen. Wir benutzen eine Varianz-Reduktions-Methode zur Auswertung von unverbundenen Diagrammen um ein präzises Ergebnis zu erhalten. Des Weiteren stellen wir eine neue stochastische Methode zur Berechnung von Nukleon-Dreipunkt-Korrelationsfunktionen vor, die für die Berechnung von Observablen der Nukleon-Struktur benötigt werden. Wir testen die Konkurrenzfähigkeit dieser neuen Methode gegenüber der Standard-Methode. In allen Rechnungen benutzen wir Wilson twisted-Mass Fermionen mit maximalem Twist, so dass die hier berechneten Observablen nur O(a^2) Diskretisierungsfehler aufweisen. / In this thesis we compute within lattice QCD observables related to the structure of the nucleon. One part of this thesis is concerned with moments of parton distribution functions (PDFs). Those moments are essential elements for the understanding of nucleon structure and can be extracted from a global analysis of deep inelastic scattering experiments. On the theoretical side they can be computed non-perturbatively by means of lattice QCD. However, since the time lattice calculations of moments of PDFs are available, there is a tension between these lattice calculations and the results from a global analysis of experimental data. We examine whether systematic effects are responsible for this tension, and study particularly intensively the effects of excited states by a dedicated high precision computation. Moreover, we carry out a first computation with four dynamical flavors. Another aspect of this thesis is a feasibility study of a lattice QCD computation of the scalar quark content of the nucleon, which is an important element in the cross-section of a heavy particle with the nucleon mediated by a scalar particle (e.g. Higgs particle) and can therefore have an impact on Dark Matter searches. Existing lattice QCD calculations of this quantity usually have a large error and thus a low significance for phenomenological applications. We use a variance-reduction technique for quark-disconnected diagrams to obtain a precise result. Furthermore, we introduce a new stochastic method for the calculation of connected 3-point correlation functions, which are needed to compute nucleon structure observables, as an alternative to the usual sequential propagator method. In an explorative study we check whether this new method is competitive to the standard one. We use Wilson twisted mass fermions at maximal twist in all our calculations, such that all observables considered here have only O(a^2) discretization effects.

Dans ce Manuscrit, nous étudions l'influence de la géométrie sur les équations de Hardy-Sobolev perturbées ou non sur toute variété Riemannienne compacte sans bord de dimension supérieure ou égale à 3. Plus précisément, dans le cas non perturbé nous démontrons que pour toute dimension de la variété strictement supérieure à, l'existence d'une solution (ou plutôt une condition suffisante d'existence) dépendra de la géométrie locale autour de la singularité. En revanche, dans le cas où la dimension est égale à 3, c'est la géométrie globale (particulièrement, la masse de la fonction de Green) de la variété qui comptera. Dans le cas d'une équation à terme perturbatif sous-critique, nous démontrons que l'existence d'une solution dépendra uniquement de la perturbation pour les grandes dimensions et qu'une interaction entre la géométrie globale de la variété et la perturbation apparaîtra en dimension 3. Enfin, nous établissons une inégalité optimale de Hardy-Sobolev Riemannienne, la variété étant avec ou sans bord, où nous démontrons que la première meilleure constante est celle des inégalités Euclidiennes et est atteinte / In this Manuscript, we investigate the influence of geometry on the Hardy-Sobolev equations on the compact Riemannian manifolds without boundary of dimension greateror equal to 3. More precisely, we prove in the non perturbative case that the existence of solutions depends only on the local geometry around the singularity when the dimension is greater or equal to 4 while it is the global geometry of the manifold when the dimension is equal to 3 that matters. In the presence of a perturbative subcritical term, we prove that the existence of solutions depends only on the perturbation when the dimension is greater or equal to 4 while an interaction between the perturbation and the global geometry appears in dimension 3. Finally, we establish an Optimal Hardy-Sobolev inequality for all compact Riemannian manifolds, with or without boundary, where we prove that the Riemannian sharp constant is the one for the Euclidean inequality and is achieved

In dieser Doktorarbeit werden wir das Renormierungsproblem von Gravitationstheorien im Kontext der Renormierungsgruppe (RG) unter Anwendung von perturbativen und nicht-perturbativen Methoden untersuchen. Insbesondere werden wir uns auf verschiedene Gravitationsmodelle und Näherungen konzentrieren, in welchen die zentrale Rolle von einem skalaren Freiheitsgrad eingenommen wird. Wir konzentrieren uns besonders auf zwei Ansätze für Quantengravitation, die in letzter Zeit viel Aufmerksamkeit erhalten haben, nämlich den asymptotisch sicheren Fall der Gravitation und die Hořava-Lifshitz Quantengravitation. Das Prinzip der Asymptotischen Sicherheit beruht auf der Annahme, dass das hochenergetische Gravitationsregime von einem nicht-Gaußschen Fixpunkt bestimmt wird, der nicht-perturbative Renormierung und Endlichkeit der Korrelationsfunktionen sicherstellt. Wir werden die Existenz eines solchen nicht-trivialen Fixpunktes mit Hilfe der funktionalen Renormierungsgruppe untersuchen. Insbesondere werden wir den einzigen konformen Freiheitsgrad quantisieren. Die Frage nach der Existenz eines nicht-Gaußschen Fixpunktes in einem unendlich- dimensionalen Parameterraum, das heißt für eine generische f(R)-Theorie, kann jedoch nicht mit einem solchen konform reduzierten Model analysiert werden. Deshalb werden wir es untersuchen, indem wir eine skalare dynamische Äquivalentstheorie, das heißt eine generische Brans-Dicke Theorie in der lokal-Potential Näherung mit ω = 0, quantisieren. Schließlich werden wir mittels einer perturbativen RG Methode die asymptotische Freiheit der Hořava-Lifshitz Gravitationstheorie analysieren. Diese Gravitationstheorie beruht auf der Entstehung einer Anisotropie zwischen Raum und Zeit, die Newtons Konstante zu einer marginalen Koppelung werden lässt und explizit die Unitarität bewahrt. Insbesondere werden wir die Einschleifenkorrektur in 2+1 Dimensionen berechnen, indem wir nur den konformen Freiheitsgrad quantisieren. / In this Ph.D. thesis we will study the issue of renormalizability of gravitation in the context of the renormalization group (RG), employing both perturbative and non-perturbative techniques. In particular, we will focus on different gravitational models and approximations in which a central role is played by a scalar degree of freedom, since their RG flow is easier to analyze. We restrict our interest in particular to two quantum gravity approaches that have gained a lot of attention recently, namely the asymptotic safety scenario for gravity and the Hořava-Lifshitz quantum gravity. In the so-called asymptotic safety conjecture the high energy regime of gravity is controlled by a non-Gaussian fixed point which ensures non-perturbative renormalizability and finiteness of the correlation functions. We will then investigate the existence of such a non trivial fixed point using the functional renormalization group, a continuum version of the non-perturbative Wilson’s renormalization group. In particular we will quantize the sole conformal degree of freedom, which is an approximation that has been shown to lead to a qualitatively correct picture. The question of the existence of a non-Gaussian fixed point in an infinite-dimensional parameter space, that is for a generic f(R) theory, cannot however be studied using such a conformally reduced model. Hence we will study it by quantizing a dynamically equivalent scalar-tensor theory, i.e. a generic Brans-Dicke theory with ω = 0 in the local potential approximation. Finally, we will investigate, using a perturbative RG scheme, the asymptotic freedom of the Hořava-Lifshitz gravity, that is an approach based on the emergence of an anisotropy between space and time which lifts the Newton’s constant to a marginal coupling and explicitly preserves unitarity. In particular we will evaluate the one-loop correction in 2+1 dimensions quantizing only the conformal degree of freedom.