L'acide ribonucléique (ARN) est une molécule impliquée dans de nombreuses fonctions cellulaires comme la traduction génétique et la régulation de l’expression des gènes. Les récents succès des vaccins à ARN témoignent du rôle que ce dernier peut jouer dans le développement de traitements thérapeutiques. La connaissance de la fonction d’un ARN passe par sa séquence et sa structure lesquelles déterminent quels groupes chimiques (et de quelles manières ces groupes chimiques) peuvent interagir avec d’autres molécules. Or, les structures connues sont rares en raison du coût et de l’inefficacité des méthodes expérimentales comme la résonnance magnétique nucléaire et la cristallographie aux rayons X. Par conséquent, les méthodes calculatoires ne cessent d’être raffinées afin de déterminer adéquatement la structure d’un ARN à partir de sa séquence. Compte tenu de la croissance des jeux de données et des progrès incessants de l’apprentissage profond, de nombreuses architectures de réseaux neuronaux ont été proposées afin de résoudre le problème du repliement de l’ARN. Toutefois, les jeux de données actuels et la nature des mécanismes de repliement de l’ARN dressent des obstacles importants à l’application de l’apprentissage statistique en prédiction de structures d’ARN. Ce mémoire de maîtrise se veut une couverture des principaux défis inhérents à la résolution du problème du repliement de l’ARN par apprentissage automatique. On y formule une tâche fondamentale afin d’étudier le comportement d’une multitude d’algorithmes lorsque confrontés à divers contextes statistiques, le tout dans le but d’éviter le surapprentissage, problème dont souffre une trop grande proportion des méthodes publiées jusqu’à présent. / Ribonucleic acid (RNA) is a molecule involved in many cellular functions like translation and regulation of gene expression. The recent success of RNA vaccines demonstrates the role RNA can play in the development of therapeutic treatments. The function of an RNA depends on its sequence and structure, which determine which chemical groups (and in what ways these chemical groups) can interact with other molecules. However, only a few RNA structures are known due to the high cost and low throughput of experimental methods such as nuclear magnetic resonance and X-ray crystallography. As a result, computational methods are constantly being refined to accurately determine the structure of an RNA from its sequence. Given the growth of datasets and the constant progress of deep learning, many neural network architectures have been proposed to solve the RNA folding problem. However, the nature of current datasets and RNA folding mechanisms hurdles the application of statistical learning to RNA structure prediction. Here, we cover the main challenges one can encounter when solving the RNA folding problem by machine learning. With an emphasis on overfitting, a problem that affects too many of the methods published so far, we formulate a fundamental RNA problem to study the behaviour of a variety of algorithms when confronted with various statistical contexts.

This article combines Aoki’s institutional complementarity concept with actor-centered institutional analysis of action situations to study herder behavior and institutional change in a complex pastoral social–ecological system. Transformation of the Mongolian Steppe Ecosystem in the face of climate and social change has led to a decline in pastoral mobility, which in turn is making the ecosystem less sustainable. Responding to this concern, Mongolian policymakers have designed pasture use and conservation policies. We evaluate whether the enacted policies are complementary to herders’ strategic choices. First, we reconstruct institutional choices herders make in the commons domain, where herders interact to use common pastures. Second, we track this process in the political economy domain, where pasture users support or resist government policies. Finally, we evaluate the complementarity of the strategic choices and resulting institutions in the interdependent action situations of both domains. In combination with game-theoretic model building, we have employed the process tracing method during field research in Mongolia. We have not identified any evident, stable institutional complementarity between high pastoral mobility and support for a policy of leasing and certification of land for winter and spring camps. Conversely, our findings do suggest that policies for establishing pasture user groups and pasture use planning can be effective. A critical mass of herders choosing to comply with these policies and engage in pastoral mobility will be crucial for sustaining the ecosystem. This will strengthen conditions for institutional complementarity and create a new institutional arrangement overall.

In dieser Arbeit präsentieren wir eine neue Art von Newton-Verfahren mit Liniensuche, basierend auf Interpolation im Bildbereich nach Wedin et al. [LW84]. Von dem resultierenden stabilisierten Newton-Algorithmus wird theoretisch und praktisch gezeigt, dass er effizient ist im Falle von nichtsingulären Lösungen. Darüber hinaus wird beobachtet, dass er eine superlineare Rate von Konvergenz bei einfachen Singularitäten erhält. Hingegen ist vom Newton-Verfahren ohne Liniensuche bekannt, dass es nur linear von fast allen Punkten in der Nähe einer singulären Lösung konvergiert. In Hinsicht auf Anwendungen auf Komplementaritätsprobleme betrachten wir auch Systeme, deren Jacobimatrix nicht differenzierbar sondern nur semismooth ist. Auch hier erreicht unser stabilisiertes und beschleunigtes Newton- Verfahren Superlinearität bei einfachen Singularitäten. / In this thesis we present a new type of line-search for Newton’s method, based on range space interpolation as suggested by Wedin et al. [LW84]. The resulting stabilized Newton algorithm is theoretically and practically shown to be efficient in the case of nonsingular roots. Moreover it is observed that it maintains a superlinear rate of convergence at simple singularities. Whereas Newton’s method without line-search is known to converge only linearly from almost all points near the singular root. In view of applications to complementarity problems we also consider systems, whose Jacobian is not differentiable but only semismooth. Again, our stabilized and accelerated Newton’s method achieves superlinearity at simple singularities.

This work is concerned with conditions ensuring stability of a given solution of a system of nonlinear equations with respect to large (not asymptotically thin) classes of right-hand side perturbations. Our main focus is on those solutions that are in a sense singular, and hence, their stability properties are not guaranteed by “standard” inverse function-type theorems. In the twice differentiable case, these issues have received some attention in the existing literature. Moreover, a few results in this direction are known in the case when the first derivative is merely B-differentiable. Here, we further elaborate on a similar setting, but the main attention is paid to the case of piecewise smooth equations. Specifically, we study the effect of singularity of a solution for some active smooth selection on the overall stability properties, and we provide sufficient conditions ensuring the needed stability properties in the cases when such smooth selections may exist. Finally, an application to a piecewise smooth reformulation of complementarity problems is given.

本論文係研究和發展平行疊代法（PARALLEL ITERATIVE METHOD ）以解決數學規劃（ MATHEMATICAL PROGRAMMING）中之互補問題（COMPLEMENTA-RITY PROBLEM）。互補問 題源自解決國防軍事、工程經濟及管理科學等領域之應用，而由於近年來各種超級或 平行電腦不斷地創新，使得發展平行演算法以充分並有效地應用超級或平行電腦來解 決大型科學計算的問題日趨重要。 在本篇論文中，我們分別探討線性互補問題以及非線性互補問題。首先我們發展出一 半非同步（SEMI-ASYNCHRONOUS ）法來解決線性互補問題，此法之特性在於其能大幅 地減低因同步法所造成處理機閒置（IDLING）之冗額成本（OVERHEAD）；同時，也放 寬了非同步法對問題所加諸之限制，因而擴大了半非同步法所能應用之範圍。我們也 建立了有關該法收斂性（CONVERGENCE ）之理論根據。此外，線性互補問題之探討， 實為進一步研究非線性互補問題之基礎。 其次，我們提出一個整體性之架構，探討平行牛頓法（NEWTON METHOD ）及其各種變 型（VARIATIONS）來解決各種非線性互補問題，比較並研究各種方法的特性、限制及 執行效率。 然後，針對上述各種演算法，我們在教育部電算中心之IBM ３０９０上發展並模擬各 該法之平行運算，經由廣泛地實驗測試，以獲得具體之數值結果，來檢驗其效率，並 比較研究各法之適用性與優劣。最後，我們也提出一些相關之問題，以供未來後續研 究之參考。

La lentille (Lens culinaris Med.) est une composante importante des régimes alimentaires de nombreuses populations à travers le monde mais sa consommation en Europe est relativement faible. L’Europe produit seulement 26% de sa consommation de lentille et ce déficit est en partie causé par d’importants verrous agronomiques comme la verse, les bruches et la compétition des adventices qui réduisent ses rendements, notamment en agriculture biologique. Les associations de cultures, définies comme la culture simultanée d’au moins deux espèces différentes sur une même surface pendant une durée significative, sont considérées comme une option pour lever ces verrous agronomiques et ainsi développer la production de lentille en agriculture biologique. Les objectifs de cette thèse étaient de (1) évaluer le potentiel des associations de lentille et de blé de printemps pour produire de la lentille en conditions d’agriculture biologique et (2) comprendre les principaux mécanismes sous-jacents à la performance des associations. Des essais agronomiques ont été mis en place en 2015 et 2016 en conditions d’agriculture biologique. Quatre variétés de lentille et de blé de printemps ont été conduites en culture pures et en plusieurs associations de type substitutif et additif. Nos résultats montrent que le rendement moyen des associations avant récolte mécanique était plus élevé que le rendement moyen des cultures pures. Néanmoins, le rendement de lentille en association était inférieur à celui de la lentille en culture pure en raison d’une compétition forte et précoce du blé pour les ressources qui a causé la diminution nombre de ramifications par plante de la lentille. Le prix de la lentille étant environ quatre fois plus élevé que celui du blé, la marge brute des associations avant récolte était inférieure à celle de la lentille en culture pure. Cependant, la verse de la lentille a été fortement réduite en association, entrainant une augmentation de l’efficacité de sa récolte mécanique. En conséquence les rendements de lentille issus de la récolte mécanique se sont avérés similaires en association et en culture pure. Enfin, après tri et nettoyage des graines, la marge brute des associations sur le rendement commercialisable était supérieure à celle des cultures pures. Nos résultats montrent que (1) les associations n’ont pas eu d’effet sur le taux de bruchage des lentilles, (2) l’association la plus performante est constituée de lentille à densité équivalente à la culture pure dans laquelle on ajoute 15-20% de blé, (3) la performance des associations est due à une utilisation complémentaire de l’azote rendue possible par la fixation symbiotique de l’azote par la lentille et (4) l’intensité des compétitions entre espèces dépendent de l’année, de la densité de blé et des génotypes. En conclusion, nos travaux indiquent que les associations de lentille et de blé de printemps peuvent permettre de développer la production de lentille en agriculture biologique mais qu’une meilleure compréhension des interactions de type génotype x environnement x conduite pourrait permettre de mettre au point des couverts encore plus performants. / Lentil (Lens culinaris Med.) is an important component of the human diet in the world, but in the meantime, Europe produces only 26% of the lentils it consumes. This is partly due to strong agronomic weaknesses that reduce yield such as lodging, bruchid beetles and weeds, especially in organic farming. Intercropping, the simultaneous growing of two or more species in the same field is tested here as an option to reduce these drawbacks and develop organic lentil production. The aims of this thesis were to (1) assess the potential of lentil-spring wheat intercrops to produce organic lentil, (2) understand the mechanisms that explain their performances, and (3) evaluate the profitability of such intercrops. A two-year field experiment was carried out in southwestern France in 2015 and 2016 under organic farming rules. Four lentil and two wheat cultivars were grown as sole crops and intercrops in multiple additive and substitutive designs. Our results showed that the total intercrop attainable grain yield was higher than the mean of sole crops. Yet, lentil yield in intercrop was lower than in sole crop as the result of a strong competition for resources from wheat in early lentil growth stages reducing the number of branches per plant of lentil. This led to lower gross margins of intercrops. However, lentil lodging was strongly reduced in intercrops thus its mechanical harvest efficiency increased. This led to similar mechanically harvested yields of lentil in intercrop and sole crop. Consequently, after mechanical harvest and grain cleaning, the marketable gross margin of intercrops was higher than that of sole crops. Our results suggest that (1) intercrop had no effect on bruchids, (2) the most effective intercrop is when lentil is at sole crop density and wheat at 15-20%, (3) intercrop performance is due to complementary use of N pools through legume N2 fixation and (4) the intensity of interspecific interactions depends on year, wheat density and genotypes. Our work indicates that lentil-spring wheat intercrop can develop organic lentil production but a better understanding of Genotype x Environment x Cropping system interactions may be useful to design optimized managements.

L’ouverture de la fenêtre des Combrailles apporte des enseignements nouveaux sur la puissance et la richesse de la cité des Arvernes.L’occupation du territoire est continue, de la Protohistoire au Moyen Âge, avec un très grand développement au moment de l’Empire romain.Le tracé des deux voies romaines majeures est établi : la densité de la population le long de leur passage prouve le rôle qu’elles ont joué.L’installation des habitats (avec la présence des élites) sur la totalité du territoire montre la complémentarité économique entre la capitale et la moyenne montagne.La densité des espaces miniers et l’exploitation de l’or à la période gauloise ont certainement contribué à la richesse des Arvernes. / The opening of the window of study of Combrailles provides new insights into the power and the wealth of the city of Arvernes.The occupacy of the territory is continuous from the Protohistory to the Middle Ages, with a great development at the time of the Roman Empire.The layout of two major ways is established : the density of population along their path proves the role they played.The settlement of the habitat (including those of the elites) on the whole territory shows the economic complementarity between rural areas and the cities.The density of mining spaces and the exploitation of gold during the Gallic period certainly contribute to the richness of Arvernes.

<p>A time-discrete formulation of the variational principle of mechanics is used to provide a consistent theoretical framework for the construction and analysis of low order integration methods. These are applied to mechanical systems subject to mixed constraints and dry frictional contacts and impacts---machines. The framework includes physics motivated constraint regularization and stabilization schemes. This is done by adding potential energy and Rayleigh dissipation terms in the Lagrangian formulation used throughout. These terms explicitly depend on the value of the Lagrange multipliers enforcing constraints. Having finite energy, the multipliers are thus massless ghost particles. The main numerical stepping method produced with the framework is called SPOOK.</p><p>Variational integrators preserve physical invariants globally, exactly in some cases, approximately but within fixed global bounds for others. This allows to product realistic physical trajectories even with the low order methods. These are needed in the solution of nonsmooth problems such as dry frictional contacts and in addition, they are computationally inexpensive. The combination of strong stability, low order, and the global preservation of invariants allows for large integration time steps, but without loosing accuracy on the important and visible physical quantities. SPOOK is thus well-suited for interactive simulations, such as those commonly used in virtual environment applications, because it is fast, stable, and faithful to the physics.</p><p>New results include a stable discretization of highly oscillatory terms of constraint regularization; a linearly stable constraint stabilization scheme based on ghost potential and Rayleigh dissipation terms; a single-step, strictly dissipative, approximate impact model; a quasi-linear complementarity formulation of dry friction that is isotropic and solvable for any nonnegative value of friction coefficients; an analysis of a splitting scheme to solve frictional contact complementarity problems; a stable, quaternion-based rigid body stepping scheme and a stable linear approximation thereof. SPOOK includes all these elements. It is linearly implicit and linearly stable, it requires the solution of either one linear system of equations of one mixed linear complementarity problem per regular time step, and two of the same when an impact condition is detected. The changes in energy caused by constraints, impacts, and dry friction, are all shown to be strictly dissipative in comparison with the free system. Since all regularization and stabilization parameters are introduced in the physics, they map directly onto physical properties and thus allow modeling of a variety of phenomena, such as constraint compliance, for instance.</p><p>Tutorial material is included for continuous and discrete-time analytic mechanics, quaternion algebra, complementarity problems, rigid body dynamics, constraint kinematics, and special topics in numerical linear algebra needed in the solution of the stepping equations of SPOOK.</p><p>The qualitative and quantitative aspects of SPOOK are demonstrated by comparison with a variety of standard techniques on well known test cases which are analyzed in details. SPOOK compares favorably for all these examples. In particular, it handles ill-posed and degenerate problems seamlessly and systematically. An implementation suitable for large scale performance and accuracy testing is left for future work.</p>

A time-discrete formulation of the variational principle of mechanics is used to provide a consistent theoretical framework for the construction and analysis of low order integration methods. These are applied to mechanical systems subject to mixed constraints and dry frictional contacts and impacts---machines. The framework includes physics motivated constraint regularization and stabilization schemes. This is done by adding potential energy and Rayleigh dissipation terms in the Lagrangian formulation used throughout. These terms explicitly depend on the value of the Lagrange multipliers enforcing constraints. Having finite energy, the multipliers are thus massless ghost particles. The main numerical stepping method produced with the framework is called SPOOK. Variational integrators preserve physical invariants globally, exactly in some cases, approximately but within fixed global bounds for others. This allows to product realistic physical trajectories even with the low order methods. These are needed in the solution of nonsmooth problems such as dry frictional contacts and in addition, they are computationally inexpensive. The combination of strong stability, low order, and the global preservation of invariants allows for large integration time steps, but without loosing accuracy on the important and visible physical quantities. SPOOK is thus well-suited for interactive simulations, such as those commonly used in virtual environment applications, because it is fast, stable, and faithful to the physics. New results include a stable discretization of highly oscillatory terms of constraint regularization; a linearly stable constraint stabilization scheme based on ghost potential and Rayleigh dissipation terms; a single-step, strictly dissipative, approximate impact model; a quasi-linear complementarity formulation of dry friction that is isotropic and solvable for any nonnegative value of friction coefficients; an analysis of a splitting scheme to solve frictional contact complementarity problems; a stable, quaternion-based rigid body stepping scheme and a stable linear approximation thereof. SPOOK includes all these elements. It is linearly implicit and linearly stable, it requires the solution of either one linear system of equations of one mixed linear complementarity problem per regular time step, and two of the same when an impact condition is detected. The changes in energy caused by constraints, impacts, and dry friction, are all shown to be strictly dissipative in comparison with the free system. Since all regularization and stabilization parameters are introduced in the physics, they map directly onto physical properties and thus allow modeling of a variety of phenomena, such as constraint compliance, for instance. Tutorial material is included for continuous and discrete-time analytic mechanics, quaternion algebra, complementarity problems, rigid body dynamics, constraint kinematics, and special topics in numerical linear algebra needed in the solution of the stepping equations of SPOOK. The qualitative and quantitative aspects of SPOOK are demonstrated by comparison with a variety of standard techniques on well known test cases which are analyzed in details. SPOOK compares favorably for all these examples. In particular, it handles ill-posed and degenerate problems seamlessly and systematically. An implementation suitable for large scale performance and accuracy testing is left for future work.

本論文主要由兩篇文章所組成，探討有關跨國購併活動中目標公司的選擇，以及聯貸市場中參貸銀行的選擇。第一部份旨在分析市場集中程度與跨國購併在垂直相關產業的議題。Beladi, Chakrabarti and Marjit (2013)建立一般均衡寡佔模型，連結當地國的市場競爭力和跨國購併在垂直相關產業的論點。他們模型認為當地國家的垂直整合程度會改變國外主併者策略優勢。我們使用1990年至2012年涵蓋86個國家，之全球購併活動案件，我們衡量當地國家的市場競爭與垂直整合程度，呈現當地國家前期的市場競爭力將會誘發國外市場主併者進入，以垂直購併的方式購併當地國家之目標公司。本研究結果提供了實證性的結果來支持過去理論之發現，認為產業的集中程度會影響跨國購併。 本論文的第二部份，在研究知識技能互補和銀行商譽在參貸銀行的選擇，觀察主貸銀行該如何選擇參貸銀行之決策分析。延伸Diamond (1991)的商譽建立假說，透過主貸銀行本身的特性因子與工作經驗，來探討主貸銀行選擇參貸銀行的誘因動機。研究結果發現，當主貸銀行具有較高的自身商譽佳、經營及投資具效率性、內部監理機制較佳、且市場經驗較為豐富時，會誘使主貸銀行減少對高商譽參貸銀行的需求。呈現知識技能互補的現象於主貸銀行選擇參貸銀行的決策。本研究結果可提供我們對聯貸銀行團商譽互補現象及分析。 / Two essays are comprised in this dissertation to study on choices of target firms in cross-border M&As and choices of participant banks in the syndicated loan market. In the first essay, cross-border mergers and market concentration in a vertically related industry, we examine the relationship between market concentration and cross-border M&A. Beladi, Chakrabarti and Marjit (2013) present an oligopoly in general equilibrium model to identify the linkages between local market competition and cross-border mergers in a vertically related industry. Their model predicts that a vertical integration at home changes the strategic advantage for foreign acquirers. Using firm-level data from 86 countries between 1990 and 2012, we calculate proxies for local market competition and show that lower (higher) pre-merger local competition at home country will increase (decrease) mergers between a foreign firm and a vertically integrated home firm. These findings provide empirical supports for the significant impact of industry concentration on the decisions of cross-border M&A. In the second essay, the effects of knowledge complementarities and bank reputation on participant banks choices, we focus on the decision of lead arrangers on participant bank choices in the syndicated loan market. We extend reputation building theory (Diamond, 1991) and model the lead arranger’s partner choice problem through the effect of self-related and task-related factors. Our paper show that when lead arrangers have higher reputation, operating efficiency, and market experience, lead arrangers tend to choose less reputable partners. These results help to explain how lead arrangers, through their partner selection decisions, manage the reputation pool among banks in the syndicated loan market.