The introduction of the Euro in 1999 has led to significant economic divergences among European countries, particularly between Northern and Central European nations like Germany and Austria, and Mediterranean countries such as France and Italy. These divergences are prominently reflected in trade imbalances. This dissertation investigates the primary drivers of these economic disparities, focusing on the role of international competitiveness. Competitiveness, defined as a nation's ability to market its products globally, is analyzed through both price and non-price factors. The research highlights several methodological challenges in assessing trade flows and competitiveness. Key findings indicate that price competitiveness, measured through unit labour costs, significantly impacts export performance. Empirical evidence reveals a negative export elasticity to prices in the European context, emphasizing the importance of cost competitiveness even for high-tech sectors. Non-price competitiveness factors, although less robust in empirical models, contribute to a comprehensive narrative when assessed descriptively. The dissertation’s contributions enrich the debate on intra-European trade imbalances, particularly through comparative analyses of Germany and Italy. The findings suggest that Germany’s export success is strongly linked to price competitiveness, while Italy’s weaker performance is due to deteriorations in both price and non-price competitiveness. These insights underline the necessity for economic policies to prioritize cost competitiveness as a determinant of industrial success. In conclusion, this dissertation provides a clearer understanding of the complex factors influencing international trade dynamics within the Eurozone, offering a foundation for future research to explore causal relationships and further investigate non-price competitiveness.:List of contents Preface of the Author i Overview of Research Papers iii List of Tables iv List of Figures v I. Executive Summary 1 1.1 Introduction 1 1.2 Background and Literature Review 2 1.2.1 Cumulative causation and export-led growth 2 1.2.2 The Kaldor paradox and the discovery of non-price competitiveness 3 1.2.3 The case of Germany and Italy 5 1.3 Presentation of Research Articles 7 1.3.1 Idea 7 1.3.2 Shared methodological framework 8 1.3.3 Summary of Research Article I 8 1.3.4 Summary of Research article II 10 1.3.5 Summary of Research article III 11 1.4 Conclusion 12 1.4.1 Main results 12 1.4.2 Contribution to the European debate 13 1.4.3 Closing remarks 14 II. Research Articles 15 2.1 Article I: The Challenging Estimation of Trade Elasticities: Tackling the Inconclusive Eurozone Evidence 15 2.1.1 Introduction 15 2.1.2 The empirical export equation 16 2.1.3 The unreliable price coefficient 21 2.1.4 Inferential Evidence 26 2.1.5 Conclusion 33   2.2 Article II: Competing for Manufacturing Value Added: How Strong is Competitive Cost Pressure on Sectoral Level? 35 2.2.1 Introduction 35 2.2.2 Background 36 2.2.3 Empirical approach 39 2.3.4 Results 43 2.3.5 Conclusion 51 2.3. Article III: Assessing the Role of Non-Price Factors: Shedding New Light on the European Competitiveness Puzzle 52 2.3.1 Introduction 52 2.3.2 Non-price competitiveness: Meaning, proxies and procedural Issues 53 2.3.3 Empirical Evidence 57 2.3.4 Discussion 70 2.3.5 Conclusion 74 References 76 Appendix I i Appendix II viii Appendix III xvii

Las aplicaciones Web son consideradas actualmente un elemento esencial e indispensable en toda actividad empresarial, intercambio de información y motor de redes sociales. La usabilidad, en este tipo de aplicaciones, es reconocida como uno de los factores clave más importantes, puesto que la facilidad o dificultad que los usuarios experimentan con estas aplicaciones determinan en gran medida su éxito o fracaso. Sin embargo, existen varias limitaciones en las propuestas actuales de evaluación de usabilidad Web, tales como: el concepto de usabilidad sólo se soporta parcialmente, las evaluaciones de usabilidad se realizan principalmente cuando la aplicación Web se ha desarrollado, hay una carencia de guías sobre cómo integrar adecuadamente la usabilidad en el desarrollo Web, y también existe una carencia de métodos de evaluación de la usabilidad Web que hayan sido validados empíricamente. Además, la mayoría de los procesos de desarrollo Web no aprovechan los artefactos producidos en las fases de diseño. Estos artefactos software intermedios se utilizan principalmente para guiar a los desarrolladores y para documentar la aplicación Web, pero no para realizar evaluaciones de usabilidad. Dado que la trazabilidad entre estos artefactos y la aplicación Web final no está bien definida, la realización de evaluaciones de usabilidad de estos artefactos resulta difícil. Este problema se mitiga en el desarrollo Web dirigido por modelos (DWDM), donde los artefactos intermedios (modelos) que representan diferentes perspectivas de una aplicación Web, se utilizan en todas las etapas del proceso de desarrollo, y el código fuente final se genera automáticamente a partir estos modelos. Al tener en cuenta la trazabilidad entre estos modelos, la evaluación de estos modelos permite detectar problemas de usabilidad que experimentaran los usuarios finales de la aplicación Web final, y proveer recomendaciones para corregir estos problemas de usabilidad durante fases tempranas del proceso de desarrollo Web. Esta tesis tiene como objetivo, tratando las anteriores limitaciones detectadas, el proponer un método de inspección de usabilidad que se puede integrar en diferentes procesos de desarrollo Web dirigido por modelos. El método se compone de un modelo de usabilidad Web que descompone el concepto de usabilidad en sub-características, atributos y métricas genéricas, y un proceso de evaluación de usabilidad Web (WUEP), que proporciona directrices sobre cómo el modelo de usabilidad se puede utilizar para llevar a cabo evaluaciones específicas. Las métricas genéricas del modelo de usabilidad deben operacionalizarse con el fin de ser aplicables a los artefactos software de diferentes métodos de desarrollo Web y en diferentes niveles de abstracción, lo que permite evaluar la usabilidad en varias etapas del proceso de desarrollo Web, especialmente en las etapas tempranas. Tanto el modelo de usabilidad como el proceso de evaluación están alineados con la última norma ISO/IEC 25000 estándar para la evaluación de la calidad de productos de software (SQuaRE). El método de inspección de usabilidad propuesto (WUEP) se ha instanciado en dos procesos de desarrollo Web dirigido por modelos diferentes (OO-H y WebML) a fin de demostrar la factibilidad de nuestra propuesta. Además, WUEP fue validado empíricamente mediante la realización de una familia de experimentos en OO-H y un experimento controlado en WebML. El objetivo de nuestros estudios empíricos fue evaluar la efectividad, la eficiencia, facilidad de uso percibida y la satisfacción percibida de los participantes; cuando utilizaron WUEP en comparación con un método de inspección industrial ampliamente utilizado: La Evaluación Heurística (HE). El análisis estadístico y meta-análisis de los datos obtenidos por separado de cada experimento indicaron que WUEP es más eficaz y eficiente que HE en la detección de problemas de usabilidad. Los evaluadores también percibieron más satisfacción cuando se aplicaron WUEP, y les / Fernández Martínez, A. (2012). A Usability Inspection Method for Model-driven Web Development Processes [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/17845

Despite much scepticism and problems for its adoption, the Model-Driven Development (MDD) is being used and improved to provide many inherent benefits for industry. One of its greatest benefits is the ability to handle the complexity of software development by raising the abstraction level. Models are expressed using concepts that are not related to a specific implementation technology (e.g. Unified Modelling Language -UML, Object Constraint Language -OCL, Action Language for Foundational UML -ALF), which means that the models can be easier to specify, maintain and document. As in Model-Driven Engineering (MDE), the primary artefacts are the conceptual models, efforts are focused on their creation, testing and evolution at different levels of abstraction through transformations because if a conceptual schema has defects, these are passed on to the following stages, including coding. Thus, one of the challenges for researchers and developers in Model-Driven Development is being able to identify defects early on, at the conceptual schema level, as this helps reduce development costs and improve software quality. Over the last decade, little research work has been performed in this area. Some of the causes of this are the high theoretical complexity of testing conceptual schemas and the lack of adequate software support. This research area thus admits new methods and techniques, facing challenges such as generation of test cases using information external to the conceptual schemas (i.e. requirements), the measurement of possible automation, selection and prioritization of test cases, the need for an efficient support tool using standard semantics, the opportune feedback to support the software quality assurance process and facilitate making decisions based on the analysis and interpretation of the results. The aim of this thesis is to mitigate some of the problems that affect conceptual schema validation by providing a novel testing-based validation framework based on Model-Driven Development. The use of MDD improves abstraction, automation and reuse, which allows us to alleviate the complexity of our validation framework. Furthermore, by leveraging MDD techniques (such as metamodeling, model transformations, and models at runtime), our framework supports four phases of the testing process: test design, test case generation, test case execution and the evaluation of the results. In order to provide software support for our proposal, we developed the CoSTest ALF-based testing environment. To ensure that CoSTest offers the necessary functionality, we first identified a set of functional requirements. Then, after these requirements were identified, we defined the architecture and testing environment of the validation framework, and finally we implemented the architecture in the Eclipse context. CoSTest has been developed to test several properties on the executable model, such as syntactic correctness (i.e. all the elements in the model conform to the syntax of the language in which it is described), consistency between the structural and behavioural parts (its integrity constraints) and completeness (i.e. all possible changes on the system state can be performed through the execution of the operations defined in the executable model). For defective models, the CoSTest report returns a meaningful feedback that helps locate and repair any defects detected. / A pesar del escepticismo y dificultades en su adopción, el Desarrollo Orientado por Modelos (MDD, por sus siglas en inglés) está siendo usado y mejorado para proveer muchos beneficios inherentes a la industria. Uno de sus mayores beneficios es la capacidad de manejar la complejidad del desarrollo de software elevando el nivel de abstracción. Los modelos se expresan utilizando conceptos que no están relacionados con una tecnología de implementación específica (por ejemplo, Lenguaje de Modelado Unificado -UML, Lenguaje de Restricción de Objetos -OCL, Lenguaje de Acción para el Foundational UML - ALF), lo que significa que los modelos pueden ser más fáciles de especificar, mantener y documentar. Debido a que en una Ingeniería dirigida por modelos (MDE), los artefactos primarios son los modelos conceptuales, los esfuerzos se centran en su creación, prueba y evolución a diferentes niveles de abstracción a través de transformaciones, porque si un esquema conceptual tiene defectos, éstos se pasan a las siguientes etapas, incluida la codificación. Por lo tanto, uno de los retos para los investigadores y desarrolladores in MDD es poder identificar los defectos temprano, a nivel de esquemas conceptuales, ya que esto ayudaría a reducir los costos de desarrollo y mejorar la calidad del software. Durante la última década, pocos trabajos de investigación se han realizado en esta área. Algunas de las causas de esta realidad son la alta complejidad teórica de probar esquemas conceptuales y la falta de soporte de software adecuado. Por lo tanto, este área de investigación admite nuevos métodos y técnicas, enfrentando retos como la generación de casos de prueba utilizando información externa a los esquemas conceptuales (es decir, los requisitos), la medición de una posible automatización, selección y priorización de casos de prueba, la necesidad de una herramienta de soporte eficiente que utilice una semántica estándar, la retroalimentación oportuna para apoyar el proceso de aseguramiento de la calidad del software y facilitar la toma de decisiones basadas en el análisis y la interpretación de los resultados. El objetivo de esta tesis es mitigar algunos de los problemas que afectan la validación de los esquemas conceptuales, proporcionando un nuevo marco de validación basado en pruebas que fue construido usando un desarrollo dirigido por modelos. El uso de MDD permite un aumento en la abstracción, automatización y reutilización que nos permite aliviar la complejidad de nuestro marco de validación. Además, al aprovechar las técnicas MDD (como el metamodelado, las transformaciones de modelos y los modelos en tiempo de ejecución), nuestro marco soporta cuatro fases del proceso de prueba: diseño de pruebas, generación de casos de prueba, ejecución de casos de prueba y la evaluación de los resultados. Con el fin de proporcionar soporte de software para nuestra propuesta, hemos desarrollado CoSTest, un entorno de pruebas basado en el lenguaje ALF. Para asegurar que CoSTest ofrece la funcionalidad necesaria, primero identificamos un conjunto de requisitos funcionales. Luego, después de identificar estos requisitos, definimos la arquitectura y el ambiente de pruebas de nuestro marco de validación y, finalmente, implementamos la arquitectura en el contexto de Eclipse. CoSTest ha sido desarrollado para probar varias propiedades sobre el modelo ejecutable como la corrección sintáctica (es decir, todos los elementos del modelo se ajustan a la sintaxis del lenguaje en el que se describe), consistencia entre la parte estructural y el comportamiento (sus restricciones de integridad) y completitud (es decir, todos los cambios posibles en el estado del sistema se pueden realizar a través de la ejecución de las operaciones definidas en el modelo ejecutable). Para los modelos defectuosos, el informe de CoSTest devuelve una retroalimentación significativa que ayuda a localizar y reparar los defectos detec / A pesar de l'escepticisme i les dificultats en la seua adopció, el Desenvolupament Orientat per Models (MDD, segons les sigles en anglès) està sent usat i millorat per tal de proveir molts beneficis potencials inherents a l' indústria. Un dels majors beneficis és la capacitat de manejar la complexitat del desenvolupament del programari elevant el nivell d'abstracció. Els models s'expressen mitjançant conceptes que no estan relacionats amb una tecnologia d'implementació específica (per exemple, el Llenguatge de Modelat Unificat - UML, Llenguatge de Restricció d'Objectes -OCL, Llenguatge d'Acció per al Foundational UML - ALF), el que significa que els models poder ser més fàcils d'especificar, mantindre i documentar. A causa de que en una Enginyeria dirigida per models (MDE), els artefactes primaris són els models conceptuals, els esforços es centren en la seua creació, prova i evolució a diferents nivells d'abstracció mitjançant transformacions, perquè si un esquema conceptual té defectes, aquestos es passen a les següents etapes, inclosa la codificació. Per tant, un del reptes per als investigadors i desenvolupadors en MDD és poder identificar els defectes des del principi, a nivell de esquemes conceptuals, perquè açò ajudaria a reduir els costos de desenvolupament i millora de la qualitat del programari. Durant l'última dècada, pocs treballs d'investigació s'han fet en aquesta àrea. Algunes de les causes d'aquesta realitat són l'alta complexitat teòrica de provar esquemes conceptuals i la falta de suport de programari adequat. Per tant, aquesta àrea d'investigació admet nous mètodes i tècniques, enfrontant reptes com la generació de casos de prova mitjançant informació externa als esquemes conceptuals (es a dir, requisits), la medició de una possible automatització, selecció i priorització de casos de prova, la necessitat de una ferramenta de suport rentable que utilitze una semàntica estàndard, la retroalimentació oportuna per suportar el procés d'assegurament de la qualitat del programari i la facilitat per a prendre decisions basades en l'anàlisi i la interpretació dels resultats. En aquesta tesi intentem mitigar alguns dels problemes que afecten a la validació dels esquemes conceptuals, proporcionant un nou marc de validació basat en proves que va ser construït mitjançant un desenvolupament dirigit per models. L'ús de MDD permet un augment en l'abstracció, automatització i reutilització que ens permet alleujar la complexitat del nostre marc de validació. A més a més, al aprofitar les tècniques MDD (com el metamodelat, les transformacions de models i els models en temps d'execució), el nostre marc suporta quatre fases del procés de prova: disseny, generació i execució de casos de prova, així com l'avaluació de resultats del procés de prova. Amb la finalitat de proporcionar suport de programari per a la nostra proposta, hem desenvolupat un entorn de proves basat en el llenguatge ALF que s'anomena CoSTest. Per tal d'assegurar que CoSTest ofereix la funcionalitat necessària, identifiquem un conjunt de requisits funcionals abans de desenvolupar la ferramenta. Després d'identificar aquestos requisits, definim l'arquitectura i l'ambient de proves del nostre marc de validació, i finalment, implementem l'arquitectura en el context Eclipse. CoSTest ha sigut desenvolupat per provar diverses propietats sobre el model executable com la correcció sintàctica (és a dir, tots els elements del model s'ajusten a la sintaxi del llenguatge en el que es descriu), consistència antre la part estructural i el comportament (les seues restriccions d'integritat) i completitud (és a dir, tots els canvis possibles en l'estat del sistema es poden realitzar mitjançant l'execució de les operacions definides en el model executable). Per als models defectuosos, l'informe de CoSTest retorna una retroalimentació significativa que ajuda a localitzar i reparar els defectes dete / Granda Juca, MF. (2017). Testing-Based Conceptual Schema Validation in a Model-Driven Environment [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/89091

La reutilización es un factor clave para reducir los costos y mejorar la calidad de las propiedades de productos software como la seguridad, fiabilidad o rendimiento. Siguiendo este factor surge la aproximación para el desarrollo de software de Líneas de Productos Software; esta aproximación promete, entre otras cosas, acortar el tiempo del desarrollo de los sistemas software y reducir significativamente los costes de desarrollo y mantenimiento. Por otro lado, el Desarrollo Dirigido por Modelos es un enfoque para el desarrollo de software que propone el uso de modelos en varios niveles de abstracción y transformaciones de modelo como artefactos principales. El uso de modelos como los principales artefactos en el desarrollo de software ofrece muchas ventajas a los desarrolladores, por ejemplo, las transformaciones de modelo permiten la conversión de un modelo fuente en otro modelo objetivo, el aumento del nivel de abstracción permite a los desarrolladores centrarse en el problema a resolver y restar importancia a los detalles de implementación. Ambos paradigmas, en la búsqueda de optimizar el tiempo de producción y calidad en el software generado, pueden reunir importantes ventajas en el proceso de producción de software. La combinación de Desarrollo Dirigido por Modelos y Líneas de Producto Software para producir productos software requiere la identificación de nuevos retos y necesidades de los stakeholders involucrados. La investigación presentada en esta tesis tiene el objetivo, apoyada en varios estudios empíricos realizados en entornos industriales, de aumentar el conocimiento y realizar una serie de propuestas de mejora del proceso de desarrollo software fundamentado en la combinación de Desarrollo Dirigido por Modelos y Líneas de Producto Software. Para alcanzar este objetivo se han estudiado cuatro dimensiones: procesado de requisitos, usabilidad, comprensión (en la configuración de productos) y gestión de errores. Cada una de estas dimensiones ha sido abordada en un estudio empírico, estudios presentados en trabajos de investigación y que forman la parte nuclear de esta tesis. Como resultado del trabajo realizado en esta tesis se han elaborado una serie de propuestas para mejorar el proceso de desarrollo software basado en la combinación de Desarrollo Dirigido por Modelos y Líneas de Producto Software y se han generado siete trabajos de investigación. Cinco de estos trabajos han sido presentados en conferencias de relevancia en el ámbito de la Ingeniería del Software: CAiSEForum'15, CAiSE'16, ESEM'16, ISD'17 y ESEM'17. Estos resultados de investigación han sido aplicados en el proceso de desarrollo de software de placas de inducción de la división de electrodomésticos de BSH (Bosch, Siemens, Gaggenau, Neff y Balay) y están siendo utilizados para su aplicación en la actual implantación para el desarrollo de software del PLC que controla los trenes en la empresa Construcciones y Auxiliar de Ferrocarriles. / Software reuse is a key factor in reducing costs and improving the quality of software product properties such as security, reliability, or performance. Taking this factor in account, the Software Product Line approach appears for software development. This approach promises to decrease the time spent in developing software systems and to significantly reduce the costs for development and maintenance of software systems, among other things. In addition, Model Driven Development is an approach for software development that proposes the use of models at various levels of abstraction and model transformations as main artifacts. The use of models as the main artifacts in software development offers many advantages for developers. For instance, model transformations allow the conversion of a source model into another target model. The increase in the level of abstraction allows the developers to focus on the problem to solve, subtracting importance to the implementation details. Both paradigms search to optimize the production time and quality of the generated software, and can result in important advantages in the software production process. However, the combination of Model Driven Development and Software Product Lines to develop software products requires the identification of the challenges and needs of the involved stakeholders. The research presented in this dissertation, supported by several empirical studies carried out in industrial environments, aims to increase the knowledge in the field and to do a set of proposals to improve the software development process based on the combination of Model Driven Development and Software Product Lines. To achieve this objective, four dimensions have been studied: requirements processing, usability, comprehension (configuring software products), and error management. The dimensions have been addressed through empirical studies, presented in research papers. These papers conform the core of this dissertation. As result of the work carried out for this dissertation, a set of proposals to improve the software development process based on the combination of Model Driven Development and Software Product Lines have been generated. Furthermore, seven research papers have been published. Five of these works have been presented at relevant conferences in the Software Engineering field: CAiSE Forum'15, CAiSE'16, ESEM'16, ISD'17 and ESEM'17. These research results have been applied in the software development process of the induction hobs from the electrical appliances division of BSH (under the brands Bosch, Siemens, Gaggenau, Neff, and Balay), and are also being used in the current implementation of the PLC that controls the trains manufactured by the Construcciones y Auxiliar de Ferrocarriles company. / La reutilització és un factor clau per a reduir els costos i millorar la qualitat de les propietats de productes programari com la seguretat, fiabilitat o rendiment. Seguint aquest factor sorgeix l'aproximació per al desenvolupament de programari utilitzant Línies de Productes Programari; aquesta aproximació promet, entre altres coses, escurçar el temps del desenvolupament dels sistemes programari i reduir significativament els costos de desenvolupament i manteniment. D'altra banda, el Desenvolupament Dirigit per Models és un enfocament per al desenvolupament de programari que proposa l'ús de models en diversos nivells d'abstracció i transformacions de model com artefactes principals. L'ús de models com els principals artefactes en el desenvolupament de programari ofereix molts avantatges als desenvolupadors, per exemple, les transformacions de model permeten la conversió d'un model font en un altre model objectiu, l'augment del nivell d'abstracció permet als desenvolupadors centrar-se en el problema a resoldre i restar importància als detalls d'implementació. Tots dos paradigmes, en la cerca d'optimitzar el temps de producció i qualitat en el programari generat, poden reunir importants avantatges en el procés de producció de programari.La combinació de Desenvolupament Dirigit per Models i Línies de Producte Programari per a produir productes programari requereix la identificació de nous reptes i necessitats dels \ingles{stakeholders} involucrats. La recerca presentada en aquesta tesi té l'objectiu, recolzada en diversos estudis empírics realitzats en entorns industrials, d'augmentar el coneixement i realitzar una sèrie de propostes de millora del procés de desenvolupament de programari fonamentat en la combinació de Desenvolupament Dirigit per Models i Línies de Producte Programari. Per a aconseguir aquest objectiu s'han estudiat quatre dimensions: processament de requisits, usabilitat, comprensió (en la configuració de productes) i gestió d'errors. Cadascuna d'aquestes dimensions ha sigut abordada en un estudi empíric, estudis presentats en treballs de recerca i que formen la part nuclear d'aquesta tesi. Com a resultat del treball realitzat en aquesta tesi s'han elaborat una sèrie de propostes per a millorar el procés de desenvolupament programari basat en la combinació de Desenvolupament Dirigit per Models i Línies de Producte Programari i s'han generat set treballs de recerca. Cinc d'aquests treballs han sigut presentats en conferències de rellevància en l'àmbit de l'Enginyeria del Programari: CAiSEForum'15, CAiSE'16, ESEM'16, ISD'17 i ESEM'17. Aquests resultats de recerca han sigut aplicats en el procés de desenvolupament de programari de plaques d'inducció de la divisió d'electrodomèstics de BSH (Bosch, Siemens, Gaggenau, Neff i Balay) i estan sent utilitzats per a la seua aplicació en l'actual implantació per al desenvolupament de programari del PLC que controla els trens en l'empresa Construcciones y Auxiliar de Ferrocarriles. / Echeverría Ochoa, J. (2018). Analysis and Improvement of a Software Production Process based on the Combination of Model Driven Development and Software Product Lines [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/107734

<p dir="ltr">Deep neural networks (DNNs) have demonstrated outstanding performance in numerous tasks such as image recognition and natural language processing. However, in dynamic systems modeling, the tasks of estimating and uncovering the potentially nonlinear structure of systems represented by ordinary differential equations (ODEs) pose a significant challenge. In this dissertation, we employ DNNs to enable precise and efficient parameter estimation of dynamic systems. In addition, we introduce a highly flexible neural ODE model to capture both nonlinear and sparse dependent relations among multiple functional processes. Nonetheless, DNNs are susceptible to overfitting and often struggle to accurately assess predictive uncertainty despite their widespread success across various AI domains. The challenge of defining meaningful priors for DNN weights and characterizing predictive uncertainty persists. In this dissertation, we present a novel neural adaptive empirical Bayes framework with a new class of prior distributions to address weight uncertainty.</p><p dir="ltr">In the first part, we propose a precise and efficient approach utilizing DNNs for estimation and inference of ODEs given noisy data. The DNNs are employed directly as a nonparametric proxy for the true solution of the ODEs, eliminating the need for numerical integration and resulting in significant computational time savings. We develop a gradient descent algorithm to estimate both the DNNs solution and the parameters of the ODEs by optimizing a fidelity-penalized likelihood loss function. This ensures that the derivatives of the DNNs estimator conform to the system of ODEs. Our method is particularly effective in scenarios where only a set of variables transformed from the system components by a given function are observed. We establish the convergence rate of the DNNs estimator and demonstrate that the derivatives of the DNNs solution asymptotically satisfy the ODEs determined by the inferred parameters. Simulations and real data analysis of COVID-19 daily cases are conducted to show the superior performance of our method in terms of accuracy of parameter estimates and system recovery, and computational speed.</p><p dir="ltr">In the second part, we present a novel sparse neural ODE model to characterize flexible relations among multiple functional processes. This model represents the latent states of the functions using a set of ODEs and models the dynamic changes of these states utilizing a DNN with a specially designed architecture and sparsity-inducing regularization. Our new model is able to capture both nonlinear and sparse dependent relations among multivariate functions. We develop an efficient optimization algorithm to estimate the unknown weights for the DNN under the sparsity constraint. Furthermore, we establish both algorithmic convergence and selection consistency, providing theoretical guarantees for the proposed method. We illustrate the efficacy of the method through simulation studies and a gene regulatory network example.</p><p dir="ltr">In the third part, we introduce a class of implicit generative priors to facilitate Bayesian modeling and inference. These priors are derived through a nonlinear transformation of a known low-dimensional distribution, allowing us to handle complex data distributions and capture the underlying manifold structure effectively. Our framework combines variational inference with a gradient ascent algorithm, which serves to select the hyperparameters and approximate the posterior distribution. Theoretical justification is established through both the posterior and classification consistency. We demonstrate the practical applications of our framework through extensive simulation examples and real-world datasets. Our experimental results highlight the superiority of our proposed framework over existing methods, such as sparse variational Bayesian and generative models, in terms of prediction accuracy and uncertainty quantification.</p>

Accurate modeling of structural ensembles is essential for understanding molecular function, predicting molecular interactions, refining molecular potentials, protein engineering, drug discovery, and more. Here, we enhance molecular modeling through Bayesian Inference of Conformational Populations (BICePs), a highly versatile algorithm for reweighting simulated ensembles with experimental data. By incorporating replica-averaging, improved likelihood functions to better address systematic errors, and adopting variational optimization schemes, the utility of this algorithm in the refinement and validation of both structural ensembles and empirical models is unmatched. Utilizing a set of diverse experimental measurements, including NOE distances, chemical shifts, and vicinal J-coupling constants, we evaluated nine force fields for simulating the mini-protein chignolin, highlighting BICePs’ capability to correctly identify folded conformations and perform objective model selection. Additionally, we demonstrate how BICePs automates the parameterization of molecular potentials and forward models—computational frameworks that generate observable quantities—while properly accounting for all sources of random and systematic error. By reconciling prior knowledge of structural ensembles with solution-based experimental observations, BICePs not only offers a robust approach for evaluating the predictive accuracy of molecular models but also shows significant promise for future applications in computational chemistry and biophysics. / Chemistry

This thesis investigates the circumstances under which parties change their positions, with Swedish nuclear power policy as a case to adress the purpose. The thesis employed qualitative text analysis and the theoretical framework Advocacy Coalition Framework (ACF) to examine and adress the the thesis's research questions and purpose. ACF was used to formulate the research questions, focusing on wether there had been changes in the Swedish parliamentary parties' policy core belifes regardning nuclear power policy and, if so, wether external subsystem events and or\changes in coalition formation could explain these changes. The study found that three out of eight Swedish parliamentary parties had changed their policy core belifes regarding the nuclear power issue:the Social Democrats, the Green Party, and the Center party.  At least one of the examined external subsystem events could either explain the changes or reinforce previous changes. Furthermore, the sudy showed that changes in coalition formation could potentially explain the Social Democrats' and the Green Party's altered policy core belifes. Other potential explanations included the influence of the Janaury Agreement, where the Liberals, as a strongly pro-nuclear party, may have influenced the Green party and the Social Democrats to become more positive towards nuclear power. In summary, the sudy demonstrated that changes in coalition formation and\or external subsystem evenets are circumstances that can explain why parties change their positions. Additionally, the thesis suggested that ACF should include two additional external subsystem events, which this sudy has shown to have an impact on parties' changed policy core belifes:war and economic changes.

Spin-orbit coupling (SOC) is a relativistic effect whose origin lies in the Dirac’s equation – a relativistic analogue of Schrödinger’s equation. SOC corrects the electronic states of a quantum mechanical system up to ~1 eV in case of semiconductors and ~ 2 – 3.6 eV in case of actinides and heavy elements by considering not only the coordinates but also the spin of the electrons in the system. Most of the applications of the present day technology are based on manipulating the electronic structure of a system with very high accuracy and precision. This demands availability of correct electronic structure of a material or molecule within a feasible computational time. Some direct consequences of SOC in materials can be noticed in analyzing the charge-transport properties of a semiconductor, evaluating the candidature of transition metal dichalcogenides (TMDCs) for spintronic, twistronic and valleytronic applications, and in the origin of topological properties of a material. Not only in materials but also in molecules the SOC effects can be observed. Fine-structure of atomic spectra was explained on the account of SOC. Several additional peaks and wavelength shift in UV-vis spectroscopy of Gold Superatoms can only be explained by correctly considering the energy level splittings caused by SOC. SOC allows intersystem and reverse intersystem crossing by mixing the spin states, ultimately opening various chemical reaction pathways which were spin forbidden before. Current advancements in computational power enrich us to work shoulder to shoulder with experiments where one can simulate the synthesized structures containing thousands of atoms using semi-empirical methods as in DFTB, GFN-XTB. These methods so far considered SOC effects but only as case studies in testing the implementation of SOC Hamiltonian rather than a systemic extension of SOC parameters to most part of the periodic table and studying SOC effects for different categories of materials and molecules. This motivated us to implement the SOC either in the form of highly accurate parameters throughout the periodic table or as addition in hamiltonian in such methods. Twisted van der Waals 2D materials as in twisted TMDC bilayers shows exciting electronic and optoelectronic properties and depending on the twist angle and chemical composition they can have thousands of atoms in their superlattices. A correct electronic analysis of such structures with SOC corrected DFT is computationally very expensive but is feasible at semi-empirical level. Here, we have applied our implementation on TMDC homo and heterobilayer twisted superlattices and studied the effect of SOC on the excitonic properties of the system. Therefore, this work opens the way for realizing various exotic applications of present day materials as well as molecules.:Table of Contents Abstract 4 1 Introduction 8 1.1 Quantum Chemistry: 8 1.2 HF based Semi-Empirical Methods 9 1.3 DFT based Semi-Empirical Methods 11 1.3.1 Density Functional based Tight-Binding Method (DFTB) 11 1.3.2 Geometry, Frequency, Non-Covalent, extended Tight Binding (GFN-xTB) 12 1.4 Spin-Orbit Coupling (SOC) 14 1.4.1 SOC in Materials 18 1.4.2 SOC in Molecular Structures 22 1.5 Theoretical Models for Accounting SOC 24 1.6 Motivation, Objective and Outline of thesis 26 2 Methodology 29 2.1 Quantum Chemistry 30 2.1.1 Schrödinger equation 30 2.2 Density Functional Theory 33 2.2.1 Generalized Gradient Approximations 39 2.3 Spin-orbit Coupling (SOC) 41 2.3.1 Classical Picture of SOC in LS model 42 2.3.2 Quantum Picture of SOC in LS model: 43 2.3.3 Calculation of SOC Paramentes 45 2.4 Density Functional Based Semi-empirical Quantum Mechanical Methods 48 2.4.1 Self-Consistent Charge Density Functional Based Tight Binding Method (SCC-DFTB) 48 2.4.2 Extended Tight-Binding (GFN1-xTB) 51 2.4.3 Addition of Spin-Orbit Coupling Hamiltonian in DFTB and GFN-xTB 54 3 Benchmarking Spin-Orbit Coupling Parameters for DFTB 56 3.1 Introduction 58 3.2 Computational Details of the DFT benchmark calculations 60 3.3 Benchmarking Spin-Orbit Coupling Parameters 60 3.3.1 III-V Bulk Semiconductor 61 3.3.2 Transition Metal Dichalcogenide 2D Crystals 65 3.3.3 Topological Insulators 68 3.4 Conclusions 70 4 Spin-Orbit Coupling Corrections for the GFN-xTB method 71 4.1.1 Introduction 73 4.2 Computational Details of The Benchmark Calculations 75 4.3 Results & Discussion 76 4.3.1 Geometries 76 4.3.2 Effect of SOC on Charge Transport Properties of Chromophores in MOFs 77 4.3.3 Superatoms 82 4.3.4 Effect of SOC on Binding of O2 on Ferrous Deoxyheme 85 4.4 Conclusions 86 5 Spin Orbit Coupling Effects on The Excitonic Properties of Twisted Moiré Transition Metal Dichalcogenides 88 5.1 Introduction 90 5.2 Computational Details 92 5.3 Results & Discussions 93 5.4 Excitons in Twisted Moiré Homobilayers 93 5.5 Excitons in Twisted Moiré Heterobilayers 102 5.6 Conclusions 109 6 Summary 112 A. Acronym 116 B. Appendices 120 SOC Parameters 120 7 References 147 C. Acknowledgement 173

The main focus of this study is to determine if systems development methodologies are being utilised in the development of electronic learning systems in South Africa and if these methodologies are being applied effectively. Essentially this study can be viewed as exploratory research, utilising a conceptual research model to investigate the relationships between the constructs and measurements. Electronic learning, or e-learning, is being employed to educate millions of learners, students and employees around the world and it is a critical component of modern educational systems. E-learning systems, or learning management systems, as it is known in the field, sit at the heart of these educational systems and are used to systematically deliver on-line content and facilitate the learning experience around that content. There is still much confusion and misconceptions surrounding e-learning and learning management systems abound. This study will try and clarify some of these misconceptions. In e-learning systems, the effective use of information systems is especially relevant as it is used to educate the minds of the future. To ensure that e-learning systems of outstanding quality are being developed, it is therefore crucial that systems development methodologies are being used as they can have a significant impact on the development process. There is a dearth of empirical research available on the use and effectiveness of systems development methodologies in South Africa. This study aims, amongst other things to make a contribution to the availability of empirical results. By empirically evaluating the conceptual research model, utilising a survey as the main research method and statistically analysing the dataset, meaningful results were obtained. This study gave some insights into how learning management system procurement and development is being done in South Africa and revealed that the use of open-source systems currently exceeds the use of proprietary systems. The results of the research showed that systems development methodologies (e.g. Object-Oriented Analysis and Rapid Application Development) are being used effectively in the development of e-learning systems. Strong relationships exist between many of the systems development methodology factors identified (e.g. performance expectancy and the perceived support of the methodology) and the quality and productivity of the development process. This in turn has a strong influence on the impact systems development methodologies have on the quality of learning management systems. / MCom (Computer Science & Information Systems), North-West University, Potchefstroom Campus, 2014

The main focus of this study is to determine if systems development methodologies are being utilised in the development of electronic learning systems in South Africa and if these methodologies are being applied effectively. Essentially this study can be viewed as exploratory research, utilising a conceptual research model to investigate the relationships between the constructs and measurements. Electronic learning, or e-learning, is being employed to educate millions of learners, students and employees around the world and it is a critical component of modern educational systems. E-learning systems, or learning management systems, as it is known in the field, sit at the heart of these educational systems and are used to systematically deliver on-line content and facilitate the learning experience around that content. There is still much confusion and misconceptions surrounding e-learning and learning management systems abound. This study will try and clarify some of these misconceptions. In e-learning systems, the effective use of information systems is especially relevant as it is used to educate the minds of the future. To ensure that e-learning systems of outstanding quality are being developed, it is therefore crucial that systems development methodologies are being used as they can have a significant impact on the development process. There is a dearth of empirical research available on the use and effectiveness of systems development methodologies in South Africa. This study aims, amongst other things to make a contribution to the availability of empirical results. By empirically evaluating the conceptual research model, utilising a survey as the main research method and statistically analysing the dataset, meaningful results were obtained. This study gave some insights into how learning management system procurement and development is being done in South Africa and revealed that the use of open-source systems currently exceeds the use of proprietary systems. The results of the research showed that systems development methodologies (e.g. Object-Oriented Analysis and Rapid Application Development) are being used effectively in the development of e-learning systems. Strong relationships exist between many of the systems development methodology factors identified (e.g. performance expectancy and the perceived support of the methodology) and the quality and productivity of the development process. This in turn has a strong influence on the impact systems development methodologies have on the quality of learning management systems. / MCom (Computer Science & Information Systems), North-West University, Potchefstroom Campus, 2014