Les travaux présentés dans cette thèse ciblent la recherche de nouvelle physique au-delà du Modèle Standard dans le cadre des modèles composites, lesquels font intervenir des secteurs fortement couplés à l'échelle du TeV comprenant des particules exotiques. Leur existence peut être testée auprès du collisionneur proton-proton LHC situé au CERN, grâce au détecteur ATLAS. La production d'un quark lourd, comme le T5/3, prédit dans ces modèles, peut être déduite d'une section efficace anormalement élevée d'une signature spécifique. Certains états finaux, comme celui étudié dans cette thèse impliquant 2 leptons de même signe, étant très rares, possèdent une excellente sensibilité. Les recherches des processus de création du T5/3 au LHC en utilisant les données d'ATLAS à 8 et 13 TeV sont présentées. Un léger excès a été observé à 8 TeV. Des résultats préliminaires de l'analyse à 13 TeV sont présentés et ne montrent pas d'excès. Les limites inférieures en masse sur cette particule hypothétique sont de 745 GeV à 8 TeV et de 990 GeV à 13 TeV. Les modèles composites prévoient aussi l'existence de bosons lourds comme par exemple un Z' topophile qui ne se couplerait qu'aux quarks top et dont la production en association d'une paire tt donnerait un état final similaire aux événements à 4 quarks top. Une étude phénoménologique sur le potentiel de découverte de ces processus exotiques (Z' et théorie effective) ou standard (tttt) à 13 et 14 TeV est effectuée et donne les luminosités minimales nécessaires pour pouvoir observer des déviations au Modèle Standard. Une étude instrumentale d'une nouvelle carte de déclenchement de niveau 1 du calorimètre électromagnétique est aussi décrite en prévision de la phase d'Upgrade du détecteur. / The work of this thesis focuses on the search for new physics beyond the Standard Model in the compositeness framework, which postulates strongly coupled sectors at the TeV-scale consisting of exotic particles. Their existence can be tested at the CERN proton-proton collider, the LHC, using the ATLAS detector. The production of an heavy quark, such as the T5/3 predicted in those models, can be inferred from an enhancement of the cross-section of a given signature. Some final states, such as the one studied in this thesis involving 2 same-sign leptons, are very rare and have an excellent sensitivity. Searches for T5/3 production at the LHC using 8 and 13 TeV ATLAS data are presented. A small excess is seen at 8 TeV. Preliminary results of the 13 TeV analysis are presented and do not show any excess. The inferior limits put on the T5/3 mass is 745 GeV at 8 TeV and 990 GeV at 13 TeV. Composite models include also heavy bosons such as top-philic Z' coupled only to the top quark whose production in association to a top quark pair gives a final state similar to those from 4 tops events. A phenomenological study on the discovery potential of those exotic (Z' and effective theory) or standard (tttt) processes at 13 and 14 TeV is done and gives the minimal luminosities needed to observe deviations from the Standard Model. An instrumental study for the conception of the new LTDB card, part of the electromagnetic calorimeter's level 1 trigger system is described in the context of the ATLAS upgrade program.

Auto-completion is a task that requires an algorithm to give suggestions for completing sentences. Specifically, the history of live chat and the words already typed by the agents are provided to the algorithm for outputting the suggestions to finish the sentences. This study aimed to investigate if the above task can be handled by fine-tuning a pre-trained T5 model on the target dataset. In this thesis, both an English and a Portuguese dataset were selected. Then, T5 and its multilingual version mT5were fine-tuned on the target datasets. The models were evaluated with different metrics (log perplexity, token level accuracy, and multi-word level accuracy), and the results are compared to those of the baseline methods. The results on these different metrics show that a method based on pre-trained T5 is a promising approach to handle the target task. / Automatisk komplettering är en uppgift som kräver en algoritm för att ge förslag på hur man kan slutföra meningar. Specifikt levereras historien om livechatt och de ord som redan har skrivits av agenterna till algoritmen för att mata ut förslagen för att avsluta meningarna. Denna studie syftade till att undersöka om ovanstående uppgift kan hanteras genom att finjustera en förtränad T5-modell på måldatamängden. I denna avhandling valdes både en engelsk och en portugisisk datamängd. Därefter finjusterades T5 och dess flerspråkiga version mT5 på måldatauppsättningarna. Modellerna utvärderades med olika mätvärden (log-perplexitet, precision på ordnivå och flerordsnivå), och resultaten jämförs med baslinjemetoderna. Resultaten på dessa olika mätvärden visar att en metod baserad på en förtränad T5 är ett lovande tillvägagångssätt för att hantera uppgiften.

В современном мире текстовая обработка и искусственный интеллект активно используются для автоматизации различных процессов, включая создание аннотаций для художественных произведений. Автоматическая генерация аннотаций помогает читателям быстро понять содержание книги и принять решение о её прочтении. В этой диссертации проводится исследование различных моделей генерации аннотаций, таких как Bag-of-Words (BoW), TF-IDF, Latent Dirichlet Allocation (LDA), Recurrent Neural Networks (RNNs), BERT (Bidirectional Encoder Representations from Transformers), T5 и PEGASUS. Эффективность этих моделей оценивается с помощью метрик BLEU Score, ROUGE Score, METEOR Score, F1 Score и CIDEr Score. Для тестирования моделей используется датасет, состоящий из книг в формате .docx. Результаты работы позволяют выявить наиболее эффективные методы автоматической генерации аннотаций и предлагают направления для дальнейшего совершенствования этих моделей. / In today's world, text processing and artificial intelligence are actively used to automate various processes, including the creation of annotations for fiction works. Automatic annotation generation helps readers quickly grasp the content of a book and decide whether to read it. This dissertation investigates various models for generating annotations, such as Bag-of-Words (BoW), TF-IDF, Latent Dirichlet Allocation (LDA), Recurrent Neural Networks (RNNs), BERT (Bidirectional Encoder Representations from Transformers), T5, and PEGASUS. The effectiveness of these models is evaluated using metrics such as BLEU Score, ROUGE Score, METEOR Score, F1 Score, and CIDEr Score. A dataset of books in .docx format is used to test the models. The results of the study identify the most effective methods for automatic annotation generation and suggest directions for further improvement of these models.

This thesis describes the experimental implementation of novel intersecting cavity terahertz optical parametric oscillators based on bulk and periodically-poled magnesium oxide-doped lithium niobate. Both collinear and non-collinear phase-matching geometries have been demonstrated and injection-seeding has been implemented in devices using periodically-poled material to reduce threshold and increase the down-conversion efficiency. A comprehensive characterisation of the original intracavity terahertz OPO was undertaken, which revealed the parameters having the greatest impact on OPO efficiency (idler mirror reflectivity and cavity length) and led to a better understanding of the losses in the system. During the characterisation process, generation of further terahertz radiation at the same frequency as that generated by the parametric process was observed and identified as being a result of difference frequency generation (DFG) between the parametrically-generated idler and terahertz waves. This phenomenon had previously only been observed when periodically-poled materials were employed in the system. The effect of this additional DFG process has been analysed in terms of the enhancement of the terahertz field on the basis of the coupled wave equations and physically measured quantities. The use of periodically-poled lithium niobate has been a major part of the research presented in this thesis. A comprehensive study of the modified phase-matching conditions was carried out and both collinear and novel hybrid non-collinear phase-matching geometries were identified. Several computer models were developed to assess the performance of any given grating design in these different geometries and the effects of temperature tuning and pump wavelength variation were also investigated using the models. Experimental studies confirmed the viability of the modelling approach but material limitations (particularly the early onset of crystal damage) limited the outcomes of the experiments. A detailed comparison of the poled and bulk materials was made to highlight the present drawbacks of the poled material. Finally, injection seeding was used to improve the efficiency of the collinear phase-matched PPLN OPOs. When seeding was used the depletion of the pump pulse was increased to the point of being measurable, reaching an upper level of 10%. Coupling constraints placed on the seed laser limited the amount of depletion attained. The potential for injection seeding to be used in the hybrid non-collinear phase-matching scheme was also identified but not realised during the course of this work. Were this technique successful, the tuning range of the intersecting cavity terahertz OPO could be extended to encompass the sub-1THz region, something that has previously been limited by the available idler cavity angles.

Extractive question answering is a popular domain in the field of natural language processing, where machine learning models are tasked with answering questions given a context. Historically the field has been centered on monolingual models, but recently more and more multilingual models have been developed, such as Google’s MT5 [1]. Because of this, machine translations of English have been used when training and evaluating these models, but machine translations can be degraded and do not always reflect their target language fairly. This report investigates if complementary information stored in other languages can improve monolingual QA performance for languages where only machine translations are available. It also investigates if exposure to more languages can improve zero-shot cross-lingual QA performance (i.e. when the question and answer do not have matching languages) by providing complementary information. We fine-tune 3 different MT5 models on QA datasets consisting of machine translations, as well as one model on the datasets together in combination with 3 other datasets that are not translations. We then evaluate the different models on the MLQA and XQuAD datasets. The results show that for 2 out of the 3 languages evaluated, complementary information stored in other languages had a positive effect on the QA performance of the MT5. For zero-shot cross-lingual QA, the complementary information offered by the fused model lead to improved performance compared to 2/3 of the MT5 models trained only on translated data, indicating that complementary information from other languages do not offer any improvement in this regard. / Frågebesvaring (QA) är en populär domän inom naturlig språkbehandling, där maskininlärningsmodeller har till uppgift att svara på frågor. Historiskt har fältet varit inriktat på enspråkiga modeller, men nyligen har fler och fler flerspråkiga modeller utvecklats, till exempel Googles MT5 [1]. På grund av detta har maskinöversättningar av engelska använts vid träning och utvärdering av dessa modeller, men maskinöversättningar kan vara försämrade och speglar inte alltid deras målspråk rättvist. Denna rapport undersöker om kompletterande information som lagras i andra språk kan förbättra enspråkig QA-prestanda för språk där endast maskinöversättningar är tillgängliga. Den undersöker också om exponering för fler språk kan förbättra QA-prestanda på zero-shot cross-lingual QA (dvs. där frågan och svaret inte har matchande språk) genom att tillhandahålla kompletterande information. Vi finjusterar 3 olika modeller på QA-datamängder som består av maskinöversättningar, samt en modell på datamängderna tillsammans i kombination med 3 andra datamängder som inte är översättningar. Vi utvärderar sedan de olika modellerna på MLQA- och XQuAD-datauppsättningarna. Resultaten visar att för 2 av de 3 utvärderade språken hade kompletterande information som lagrats i andra språk en positiv effekt på QA-prestanda. För zero-shot cross-lingual QA leder den kompletterande informationen som erbjuds av den sammansmälta modellen till förbättrad prestanda jämfört med 2/3 av modellerna som tränats endast på översättningar, vilket indikerar att kompletterande information från andra språk inte ger någon förbättring i detta avseende.

That kings throughout the entire Middle Ages used the marriages of themselves and their children to further their political agendas has never been in question. What this thesis examines is the significance these marriage alliances truly had to domestic and foreign politics in England from the accession of Henry II in 1154 until the death of his grandson Henry III in 1272. Chronicle and record sources shed valuable light upon the various aspects of royal marriage at this time: firstly, they show that the marriages of the royal family at this time were geographically diverse, ranging from Scotland and England to as far abroad as the Empire, Spain, and Sicily, Most of these marriages were based around one primary principle, that being control over Angevin land-holdings on the continent. Further examination of the ages at which children were married demonstrates a practicality to the policy, in that often at least the bride was young, certainly young enough to bear children and assimilate into whatever land she may travel to. Sons were also married to secure their future, either as heir to the throne or the husband of a wealthy heiress. Henry II and his sons were almost always closely involved in the negotiations for the marriages, and were often the initiators of marriage alliances, showing a strong interest in the promotion of marriage as a political tool. Dowries were often the centre of alliances, demonstrating how much the bride, or the alliance, was worth, in land, money, or a combination of the two. One of the most important aspects for consideration though, was the outcome of the alliances. Though a number were never confirmed, and most royal children had at least one broken proposal or betrothal before their marriage, many of the marriages made were indeed successful in terms of gaining from the alliance what had originally been desired.

Organic solar cells offer the possibility for lightweight, flexible, and inexpensive photovoltaic devices. This thesis studies the physics of a wide range of materials designed for use in organic solar cells. The materials investigated include conjugated polymers, conjugated dendrimers, and inorganic nanocrystals. The materials studied in this thesis fall into five categories: conjugated polymers blended with a buckminsterfullerene derivative PCBM, nanocrystals synthesised in a conjugated polymer matrix, conjugated polymers designed for intramolecular charge separation, conjugated dendrimers blended with PCBM, and nanocrystals synthesised in a matrix of conjugated small molecules or dendrimers. Conjugated polymers blended with PCBM have been extensively studied for photovoltaic applications, and hence form an ideal test bed for new experiments. In this thesis this blend was used to achieve the first pulsed electrically detected magnetic resonance experiments on organic solar cells. Nanocrystals are attractive for photovoltaics because it is possible to tune their band gap across the solar spectrum. In this thesis a one-pot synthesis is used to grow PbS and CdS nanocrystals in conjugated polymers, soluble small molecules, and dendrimers, and characterisation is performed on these composites. Previous work on dendrimer: nanocrystal composites has been limited to non-conjugated molecules, and the synthesis developed in this thesis extends this work to a conjugated oligomer and a conjugated dendrimer. This synthesis can potentially be extended to a variety of conjugated soluble small molecule: nanocrystal and dendrimer: nanocrystal systems. Conjugated dendrimers have been successfully employed in organic light emitting diodes, and in this thesis they are applied to organic solar cells. Materials based on fluorene and cyanine dye cores show excellent absorption tunability across the solar spectrum. A set of electronically asymetric polymers designed for intramolecular charge separation were investigated. Quenching of the luminescence was observed, and light induced electron paramagnetic resonance measurements revealed that photoexcitation led to approximately equal numbers of positive polarons and nitro centred radical anions. This indicates that charge separation is occurring in these molecules.

This thesis proposes two distinct but connected ideas: that John Keats’s idiom of friendship was haunted by “sequestered” longings and that he ultimately valued specific, one-on-one partnerships as a basis for his poetical character. The Introduction places the thesis within its critical context and outlines “paradoxical solitude,” a concept the poet expressed by joining a “kindred spirit” in a wilderness retreat in “O, Solitude.” I begin by examining the evolving role of solitude in Keats’s literary predecessors (Chapter I). I then trace the development of ideas of creativity and solitude from his 1814-1815 verse, including his first association with a coterie and the influence of Wordsworth (Chapter II). Building on these findings, I explore the poet’s introduction to the Hunt circle in 1816, assessing his relationships with its members and their overstated roles in the production of Poems (Chapter III). I then discuss how Keats regarded the composition of Endymion in 1817 as a poetic “test,” specifically tailored to reinforce his identity as a solitary poet (Chapter IV). I contend that Keats engaged in a dialogue of independence with Reynolds, adapted the theories of Hazlitt, and restlessly travelled throughout England as a means of rejecting the highly social periods of 1818 (Chapter V). I then consider the creative gains of his northern expedition with Brown in the summer of 1818. I argue that Keats exaggerated his development into a “post-Wordsworthian” poet, positioning himself outside both the coterie’s sphere and the reach of Blackwood’s criticism, and inspiring the theme of Hyperion (Chapter VI). In closing, I analyze Keats’s advice to Shelley to be a selfish creator of his poetic identity. Only through paradoxical solitude, I argue, was Keats able to construct the poetic identity that led him to compose the poems on which his fame rests in the 1820 volume.

Ce travail porte sur l'étude expérimentale de deux mécanismes de translocations d'acides nucléiques au travers d'une membrane lipidique : la translocation, forcée électrophorétiquement, d'oligomères au travers d'un pore d'alpha-hémolysine et la translocation passive d'un ADN génomique hors de la capside du bactériophage T5. La première partie de la thèse porte sur l'ouverture de molécules d'ADN double brin à travers le nanopore d'alpha hémolysine. Les temps de passage individuels de molécules d'ADN à travers le pore sont mesurés expérimentalement en fonction de la séquence, de la longueur et de la force appliquée sur l'ADN. Les distributions obtenues sont confrontées à un modèle décrivant le passage de l'ADN par la diffusion d'une fourche d'ouverture dans un paysage énergétique unidimensionnel, déterminé par la séquence de la molécule. La deuxième partie porte sur un système in vitro reconstituant les étapes initiales d'infection du bactériophage T5. L'interaction de T5 avec son récepteur membranaire FhuA purifié en détergent, génère une séquence d'événements qui conduit à l'éjection du génome viral hors de la capside : (i) fixation du récepteur ; (ii) activation conduisant à l'ouverture d'un canal d'ADN ; (iii) éjection de l'ADN. La dynamique des trois étapes est mesurée à l'aide d'expériences en population et en virus unique. La dernière étape est comparée à un modèle physique qui révèle une dynamique fortement hors d'équilibre à l'initiation de l'éjection. Enfin, FhuA est reconstitué dans des vésicules lipidiques géantes afin de suivre l'éjection par microscopie de fluorescence et par électrophysiologie à travers une membrane lipidique.

This thesis explores various aspects of long memory behaviour in African stock markets (ASMs). First, we examine long memory in both equity returns and volatility using the weak-form version of the efficient market hypothesis (EMH) as a criterion. The results show that these markets (largely) display a predictable component in returns; while evidence of long memory in volatility is mixed. In general, these findings contradict the precepts of the EMH and a variety of remedial policies are suggested. Next, we re-examine evidence of volatility persistence and long memory in light of the potential existence of neglected breaks in the stock return volatility data. Our results indicate that a failure to account for time-variation in the unconditional mean variance can lead to spurious conclusions. Furthermore, a modification of the GARCH model to allow for mean variation is introduced, which, generates improved volatility forecasts for a selection of ASMs. To further evaluate the quality of volatility forecasts we compare the performance of a number of long memory models against a variety of alternatives. The results generally suggest that over short horizons simple statistical models and the short memory GARCH models provide superior forecasts of volatility; while, at longer horizons, we find some evidence in favour of long memory models. However, the various model rankings are shown to be sensitive to the choice of error statistic used to assess the accuracy of the forecasts. Finally, a wide range of volatility forecasting models are evaluated in order to ascertain which method delivers the most accurate value-at-risk (VaR) estimates in the context of Basle risk framework. The results show that both asymmetric and long memory attributes are important considerations in delivering accurate VaR measures.