Graph Neural Networks and Transformers are very powerful frameworks for learning machine learning tasks. While they were evolved separately in diverse fields, current research has revealed some similarities and links between them. This work focuses on bridging the gap between GNNs and Transformers by offering a uniform framework that highlights their similarities and distinctions. We perform positional encodings and identify key properties that make the positional encodings node embeddings. We found that the properties of expressiveness, efficiency and interpretability were achieved in the process. We saw that it is possible to use positional encodings as node embeddings, which can be used for machine learning tasks such as node classification, graph classification, and link prediction. We discuss some challenges and provide future directions.

Salamander limb regeneration is an accurate process which gives rise exclusively to the missing structures, irrespective of the amputation level. This suggests that cells in the stump have an awareness of their spatial location, a property termed positional identity. Little is known about how positional identity is encoded, in salamanders or other biological systems. Through single-cell RNAseq analysis, we identified Tig1/Rarres1 as a potential determinant of proximal identity. Tig1 encodes a conserved cell surface molecule, is regulated by retinoic acid and exhibits a graded expression along the proximo-distal axis of the limb. Its overexpression leads to regeneration defects in the distal elements and elicits proximal displacement of blastema cells, while its neutralisation blocks proximo-distal cell surface interactions. Critically, Tig1 reprogrammes distal cells to a proximal identity, upregulating Prod1 and inhibiting Hoxa13 and distal transcriptional networks. Thus, Tig1 is a central cell surface determinant of proximal identity in the salamander limb.

本篇論文檢視台灣華語為母語之正常幼童以及構音異常幼童之滑音[j],[w], [ɥ]發展與產製表現，採長期觀察之方式，詳細描述正常幼兒滑音產製之出現及穩定年齡、頻率、正確率和發展順序，並且比較此三滑音於各音節位置之產製表現，進而比較兩組幼童之台灣華語滑音產製及音韻歷程之使用行為。本研究以標記理論及位置層級理論來檢驗幼兒滑音之發展與表現。 本研究總共觀察了四位幼童，將其分為兩組，第一組為兩位正常幼童，年齡在九個月至二十八個月和十個月至二十九個月，為期十九個月的觀察，另一組為兩位構音異常幼童，年齡在三歲十個月至四歲三個月和四歲三個月至四歲九個月，為期六個月的觀察。每兩週收錄一次長達一小時之語料，並利用錄製之高規格影音檔做譯寫及分析。 研究結果顯示，正常幼兒之滑音出現順序和穩定順序皆符合標記理論之預測，無標音早於有標音，其滑音產製之穩定度與音節位置相關，音節首位之滑音表現較音節中與音節末之滑音表現來的穩定，這與位置層級理論之推測相符合。構音異常幼童組中發現了有別於正常幼童之特殊取代模式，由較晚習得語音取代較早習得語音，其所產製的語音中會違反華語的音法限制。音韻歷程之表現在兩組幼童中皆偵測到刪除、換位、取代三種模式，其中皆以刪除為主要策略。 / The purpose of the present study is to report the developmental process of three Mandarin glides [w, j, ɥ] in terms of three word positions by examining the age of emergence and stabilization, the order of stabilization, and accuracy rate of thechildren’s production, and further compare the normative data with phonological disorder data in order to explore the possible phonological processes. This study alsoaccounts for the developmental process of glides on the basis of markedness theory and positional prominence hierarchy. A longitudinal study was carried out for the investigation of two normally-developing children, aged between 0;9-2;4 and 0;10-2;4, and two phonologically-disordered children, one of whom is between 4;3 and 4;9, and the other between 3;10 and 4;3. The data were collected at two-week intervals. The results showed that the order of glide emergence and stabilization of the normally-developing group is in accordance with the markedness theory. The unmarked [j], [w] precede marked [ɥ]. Moreover, the stabilization order of the three glides in terms of the three syllable positions was found to reflect the interaction between markedness constraint and positional prominence hierarchy. The unmarked glide in the initial position is the first to stabilize and the marked glide in the non-initial position is the last to stabilize. On the other hand, the children in the phonologically-disordered group were found to consistently replace the presumably earlier-developing glide with a presumably later-developing one, which differs from the process used in the normally-developing group. Furthermore, this group of children produced the combination that violates the phonotactic constraints of Taiwan Mandarin. In addition, there are three phonological processes, including deletion, metathesis, and substitution detected in children’s data. The most commonly used process is deletion in both groups of children.

This research both develops new methods and expands upon existing methodologies in order to improve our understanding of voting patterns and party competition in Britain. The thesis comprises five sections, each of which relates to a particular research focus. The first and principal section describes the process of determining a new method for decomposing electoral bias for three-party competition under simple plurality rules of voting. The study of electoral bias is important for voting systems that requires periodic boundary reviews intended to equalise electorate and to remove malapportionment. These papers describe both the process for developing the threeparty bias method and later its application to UK general elections from 1983 onwards. The second section uses aggregate data gathered for the elections to the Greater London Authority in order to understand the patterns of electoral support across the capital, particularly support for minor parties. A considerable amount of research effort has been expended upon providing reliable models for electoral forecasting both in the UK and elsewhere. The third section includes a paper that develops a forecast model that utilises aggregate local election data to estimate national vote shares for the three main parties in the UK. A fourth section brings together a series of papers that are linked by the themes of voter behaviour, either in terms of geographical or ballot context. A study of voter turnout in a London borough describes the relationship between proximity to polling station and electoral turnout at different types of election. A 8 number of papers included in this section also detail the effects of candidate ballot order on electoral support. The fifth and final section groups together two papers that using individual-level survey data to describe the pattern of candidate recruitment for local elections in Britain and, specifically, the under-recruitment of both women and Black, Asian and other minority ethnic candidates.

Title: Marketing of selected Czech football league clubs Objective: Analysis and comparison of the marketing activities of selected Czech football league clubs and proposition of measures that would lead to the improvement of marketing activities, with an emphasis on marketing communications. Methods: PEST analysis SWOT analysis Porters five forces analysis Marketing research - qualitative method - in-depth interview - quantitative method - questionnaires Results: The results of evaluation of the marketing of selected Czech football league clubs showed significant differences in the quality of marketing activities between clubs. In some areas, such as websites, printed materials, clubs perform well. In others, such as, social networks or programs during breaks, they do not. Despite the lack of financial and human resources some elements are made with precision. I also tried to propose some measures that might lead to improvement and streamlining of marketing activities of clubs and that would be realistic even at third league level. These recommendations are mainly in the field of internet communication, accompanying programs, competitions and events that may increase the interest of viewers and partners of the clubs. Key words: Sports marketing, marketing mix, marketing communications, marketing...

Ce travail a porté sur l'identification et la caractérisation de nouveaux facteurs de l'hôte nécessaires au cycle infectieux des virus à ARN (genre Potyvirus) chez Arabidopsis thaliana, ces facteurs pouvant constituer des cibles pour la lutte génétique contre les phytovirus. L'exploration de la diversité génétique naturelle d'Arabidopsis a conduit à l'identification d'un gène récessif de résistance au Watermelon mosaic virus (WMV) chez l'accession Cvi-0. Il s'agit du gène rwm1 qui agit à un stade précoce de l'interaction en empêchant l'accumulation du virus au niveau des premières cellules infectées. Les résultats de clonage positionnel de rwm1, de séquençage d'allèles de résistance et de sensibilité et de validation fonctionnelle argumentent en faveur d'un rôle de la phosphoglycérate kinase chloroplastique (cPGK), une enzyme clé de la photosynthèse, dans la résistance conférée par rwm1. En parallèle, une stratégie gène candidat a été mise en œuvre pour étudier le rôle de la protéine kinase Target Of Rapamycin (TOR) dans l'interaction plante-potyvirus. La caractérisation de lignées RNAi pour le gène TOR d'Arabidopsis montre une diminution voire une suppression de l'accumulation du WMV, ainsi qu'une réduction de l'accumulation du Turnip mosaic virus. L'étude comparative de la composition des complexes de liaison à la coiffe entre une lignée sauvage et une lignée RNAi TOR en conditions saines et infectées par le WMV suggère un rôle de TOR dans l'interaction avec les potyvirus par le biais d'une régulation de l'activité du facteur d'initiation de la traduction eIF3. / This work describes the use of forward and reverse genetics in Arabidopsis thaliana to identify and characterize new host factors required for the infectious cycle of plant RNA viruses (genus Potyvirus). The exploration of the Arabidopsis natural genetic diversity led to the identification of the recessive rwm1 gene for resistance to Watermelon mosaic virus (WMV). Rwm1-mediated resistance was identified in the Cvi-0 accession where it acts at an early stage of the infection process by impairing virus accumulation in initially infected tissues. Map-based cloning results, allelic sequence analysis and functional validation experiments, strongly suggest the involvement of chloroplast phosphoglycerate kinase (cPGK), a key photosynthetic enzyme, in rwm1-mediated resistance. In parallel, a candidate gene approach was developed to investigate the potential role of the TOR kinase (Target Of Rapamycin) in the control of susceptibility to potyviruses. Arabidopsis RNAi lines displaying partial TOR inactivation showed a significant decrease of susceptibility to the infection by two potyviruses, namely WMV and Turnip mosaic virus. The results from a comparative proteomic analysis of cap-binding complexes between wild type and TOR RNAi infected and healthy lines suggest that TOR could play a role in plant-potyvirus interactions by regulating the activity of the translation initiation factor eIF3. Overall, this work opens a challenging research area to provide novel insights on the molecular principles underlying viral infection processes and opportunities to improve and optimize the way we tackle plant resistance to viruses through a diversification of genetic targets.

Le premier chapitre propose une revue des différents moyens de quantification du bien-être subjectif. Il s'agit des auto-évaluations globales de satisfaction, des mesures en temps réel (DRM et ESM) et de l'indice U. Dans le second chapitre, nous exploitons la base IFLS pour tester l'hypothèse de statut relatif en Indonésie. Le bien-être des indonésiens est positivement influencé par le revenu des agents dans leur entourage. Cette influence positive est clairement plus importante pour les pauvres que pour les riches. Le troisième chapitre examine l'effet de l'obésité sur la santé mentale des femmes mexicaines. Nos résultats montrent que la grosseur des femmes exerce des effets hétérogènes sur leur bien-être émotionnel, avec notamment un effet négatif observé pour les riches contrairement aux groupes de revenu inférieurs et moyens qui semblent tirer satisfaction de leur excès de poids. Les différences également observées selon d'autres dimensions (éducation et ruralité) supportent l'hypothèse liée aux différences de normes sociales. Ce constat est robuste à des estimations alternatives considérant les biais de causalité inverse et de variables omises. Le dernier chapitre étudie la présence d'une prime de la taille dans la vie adulte d'un échantillon d'hommes actifs dans un pays relativement pauvre, l'Indonésie. Les résultats montrent que chaque accroissement d'un centimètre entraîne une augmentation du revenu de 2%. Les conditions socio-économiques des individus durant leur enfance sont à l'origine du quart de cet effet. Les capacités cognitives en expliquent seulement une petite partie. Nos résultats montrent également un effet positif de la taille sur le bonheur. / Recent years have witnessed an upsurge of interest among economists in the measurement and analysis of subjective well-being. The present thesis consists of a survey and three empirical studies relate to the determinants of subjective well-being in developing countries. The first chapter is a summary of approaches used to measure subjective well-being. It analyses the advantages and disadvantages of each approach as well as the theoretical background that lies behind them.The second investigates the presence and the sign of positional concerns in Indonesia. Estimates reveal that relative income positively affects individual well-being. This positive influence is unambiguously more important for the rich than for the poor, and is consistent to various specifications. The third chapter analyses the causal effect of obesity on the emotional well-being of Mexican women, with a special focus on the role of norms. Results point to a negative effect of obesity on emotional well-being among the rich Mexican but positive or insignificant effects among the poor. Heterogeneous effects are also observed along other dimensions such as education and rurality. This heterogeneity is attributed to contrasted body-type norms that characterize different groups. The fourth chapter tests the presence of a beneficial effect of height on different concepts of objective and subjective well-being among Indonesian men. Our analysis concludes to several height premiums. In particular, taller men receive higher wages (because they sort into more lucrative occupations), and are on average happier than their shorter peers.

No presente trabalho, desenvolve-se um programa computacional para análise de instabilidade de perfis de parede fina por meio do método dos elementos finitos (MEF), com discretização em elementos de casca. Para tal finalidade, utiliza-se uma formulação não-linear geométrica do MEF, com descrição lagrangeana total do equilíbrio, tendo posições nodais e vetores generalizados como variáveis fundamentais da formulação, possibilitando a adoção de lei constitutiva tridimensional completa. Dada a adoção de vetores generalizados ao invés de giros, surge o problema de não unicidade desses vetores nas regiões de encontro entre elementos não coplanares. Para contornar esse problema, desenvolvem-se algumas estratégias de acoplamento que são eficientes e que não comprometem o condicionamento do sistema resultante. Em seguida, introduz-se no programa uma estratégia, baseada na análise linear de instabilidade, que consiste na obtenção de autovalores e autovetores correspondendo, respectivamente, a cargas críticas e modos de instabilidade associados. É realizada uma extensão dessa estratégia para a incorporação da análise não-linear de instabilidade, possibilitando a determinação de pontos críticos ao longo da trajetória de equilíbrio de um ponto da estrutura. Desenvolve-se, também, uma interface gráfica para o programa, para a qual se implementam algoritmos para geração de malha de elementos finitos triangulares e quadrilaterais e se possibilita a aplicação de condições de contorno de forma simples. Por fim, apresentam-se exemplos para validar o código computacional desenvolvido e para explorar as potencialidades do mesmo. A partir desses exemplos, conclui-se que a estratégia proposta e a ferramenta computacional desenvolvida funcionam adequadamente, oferecendo como principal vantagem respostas em geral livres de travamento volumétrico quando comparadas aos resultados provenientes da formulação convencional do MEF, encontrados na literatura. / In the present work, a computational program is developed to perform instability analysis in thin-walled profiles employing the finite element method (FEM), using shell elements. For this purpose, a non-linear geometric formulation of FEM is adopted, with Total Lagrangean description of the equilibrium, having nodal positions and unconstrained vectors as fundamental variables of the formulation, instead of displacements and rotations, making possible the adoption of complete three-dimensional constitutive law. Given the adoption of generalized vectors instead of rotations, the problem arises of the vectors\' non-uniqueness in the regions of connection between non-coplanar shell elements. To overcome this problem, some coupling strategies are developed that are efficient and do not result in ill conditioning of the resulting system of equilibrium equations. Then, a strategy based on buckling analysis is considered in the program, which consists of obtaining eigenvalues and eigenvectors related, respectively, to critical loads and instability modes. An extension of this strategy is developed to consider the nonlinear analysis of instability, making possible to determine critical points along the equilibrium path of a point in the structure. A graphical interface is also developed for the program, for which algorithms are implemented for triangular and quadrilateral finite elements mesh generations and easy boundary conditions assignments. Finally, some examples are presented to validate the developments and to explore the potentialities of the computational tool obtained in the work. From the results, it is possible to conclude that the program works properly, offering as main advantage volumetric responses, in general, free of locking when compared to results using the conventional FEM formulation, as found in the literature.

A análise de compósitos laminados apresenta grandes desafios, pois, diferentemente dos materiais isotrópicos homogêneos, os compósitos laminados são constituídos de materiais heterogêneos e anisotrópicos. Além disso, as distribuições de tensões interlaminares obtidas com as formulações convencionais são descontínuas e imprecisas. Sua melhoria, portanto, é imprescindível para buscar e modelar critérios de falha relacionados às estruturas formadas por compósitos laminados. Diante disso, este trabalho se concentrou no desenvolvimento e implementação computacional de um elemento finito posicional de pórtico plano laminado cuja cinemática é descrita ao longo da espessura do laminado de acordo com a teoria Layerwise. A formulação do elemento considera a não linearidade geométrica, originada pela ocorrência de grandes deslocamentos e rotações, e admite deformações moderadas, em função da lei constitutiva de Saint-Venant-Kirchhoff. O desenvolvimento deste trabalho se iniciou com uma preparação teórica sobre mecânica dos sólidos deformáveis e métodos numéricos para que fossem adquiridos os subsídios teóricos necessários ao desenvolvimento de códigos computacionais, à interpretação dos resultados e à tomada de decisões quando das análises numéricas. A formulação desenvolvida é Lagrangiana total com emprego do método dos elementos finitos baseado em posições. Inicialmente o elemento finito posicional de pórtico plano homogêneo é proposto, uma vez que sua cinemática possibilita uma expansão natural para o caso laminado. Os graus de liberdade são compostos por posições nodais e por vetores generalizados que representam o giro e a variação na altura da seção transversal. A eficiência do elemento é constatada através de análises realizadas em problemas de pórtico sujeitos a grandes deslocamentos e rotações. Os resultados obtidos apresentaram excelente concordância com soluções numéricas e analíticas disponíveis na literatura. Uma expansão natural da cinemática é empregada na formulação do elemento laminado. Os graus de liberdade do elemento são as posições nodais e as componentes de vetores generalizados associados às seções transversais de cada lâmina. Dessa forma, as lâminas têm liberdade para variação de espessura e giro independente das demais, mas com as posições compatibilizadas nas interfaces. Os resultados de análises numéricas realizadas em vários exemplos demonstram a eficiência da formulação proposta, pois as distribuições de deslocamentos e tensões ao longo da espessura do laminado apresentaram excelente concordância com as obtidas a partir de análises numéricas utilizando um elemento finito bidimensional em uma discretização bastante refinada. Os exemplos analisados contemplam problemas com seção laminada fina ou espessa. / The analysis of laminated composites presents challenges because, unlike homogeneous isotropic materials, the laminated composites are made up of heterogeneous and anisotropic materials. Moreover, the distribution of interlaminar stresses obtained with conventional formulations are discontinuous and inaccurate. His improvement is therefore essential to check and modeling failure criteria related to structures formed by laminates. Thus, this work focused on developing and computational implementation of a positional finite element of laminated plane frame whose kinematics is described throughout the thickness of the laminate according to Layerwise theory. The formulation element considers the geometric nonlinearity, caused by the occurrence of large displacements and rotations, and admits moderate deformation, in the constitutive law function of Saint-Venant-Kirchhoff. The development of this work began with a theoretical preparation on mechanics of deformable solids and numerical methods for the acquired of the theoretical support needed for the development of computational codes, interpretation of results and decision-making when of the numerical analyzes. The developed formulation is total Lagrangian with use of the finite element method based on positions. Initially the positional finite element of homogeneous plane frame is proposed, since their kinematic enables a natural expansion for the laminate case. The degrees of freedom are composed of nodal positions and generalized vectors representing the spin and the variation in the height of the cross section. The efficiency of the element is verified through analyzes performed in frame problems subject to large displacements and rotations. The results showed excellent agreement with numerical and analytical solutions available in the literature. A natural expansion of the kinematics is used in the formulation of the laminate element. The degrees of freedom of the element are the nodal positions and components of the generalized vectors associated to cross-sections of each lamina. Thus, the laminas are free for the thickness variation and for independent spin, but with the positions matched in the interfaces. The results of numerical analysis performed in various examples show the effectiveness of the proposed formulation, since the distributions of displacements and stresses through the thickness of the laminate agreed well with those obtained from numerical analysis using a discretization with two-dimensional finite elements in a very refined. The examples discussed include problems with thin or thick laminated section.

As deformações derivadas da retração e da fluência estão entre os principais fatores que afetam a integridade e facilidade de manutenção das estruturas de concreto a longo prazo. Outro fator que corrobora para a deterioração do concreto é a Reação Álcali-Agregado (RAA), que devido à sua característica expansiva é considerada uma manifestação patológica de difícil tratamento, sendo necessário medidas mitigadoras para evitar sua ocorrência. A retração, a fluência e a RAA são deformações expressivas em estruturas com grandes volumes de concreto, em que as deformações devido à retração e à fluência potencializam o efeito da RAA, criando um ciclo de degradação do concreto de difícil controle. Deste modo, este trabalho tem como objetivo a análise numérica que possibilite a previsão do comportamento da retração e da fluência quando associadas à RAA. O código computacional é baseado no Método de Elementos Finitos Posicional considerando o elemento de chapa com elementos triangulares de ordem cúbica e utilizando a lei constitutiva de Saint-Venant-Kirchhoff. Para a modelagem da RAA utiliza-se o estudo desenvolvido por Carrazedo & Lacerda (2008). A fluência e a retração são modeladas implementando-se os modelos FIB, JSCE e B4. Os resultados deste trabalho demonstram a eficiência da modelagem e da análise numérica para a previsão do comportamento das deformações do concreto sujeito à RAA, considerando os efeitos de retração e fluência. / Shrinkage and creep strains are among the main factors that affect the integrity and serviceability of concrete structures in long term. Another corroborating factor for the concrete deterioration is the Alkali-Aggregate Reaction (AAR), which due to its expansive characteristic is considered a pathological manifestation of difficult treatment, and mitigating measures are necessary to avoid its occurrence. Shrinkage, creep and AAR are expressive strains in structures with large volumes of concrete, in which strains due shrinkage and creep enhance the effect of AAR, creating a concrete degradation cycle that is difficult to control. Thus, this work has the aim of numerical analysis that allows the prediction of the shrinkage and creep behavior when associated with AAR. The computational code is based on the Positional Finite Elements Method considering the element of plate with triangular elements of cubic order and using the constitutive law of Saint-Venant-Kirchhoff. For the modeling of AAR is used the study developed by Carrazedo & Lacerda (2008). The creep and shrinkage are modeled by implementing the FIB, JSCE and B4 models. The results of this work demonstrate the efficiency of modeling and numerical analysis for the prediction of the strains behavior of concrete subject to AAR, considering the effects of shrinkage and creep.