In this thesis, we study the problem of protein side chain assignment (SCA) given multiple sources of experimental and modeling data. In particular, the mechanism of X-ray crystallography (X-ray) is re-examined using Fourier analysis, and a novel probabilistic model of X-ray is proposed for SCA's decision making. The relationship between the measurements in X-ray and the desired structure is reformulated in terms of Discrete Fourier Transform (DFT). The decision making is performed by developing a new resolution-dependent electron density map (EDM) model and applying Maximum Likelihood (ML) estimation, which simply reduces to the Least Squares (LS) solution. Calculation of the condence probability associated with this decision making is also given. One possible extension of this novel model is the real-space refinement when the continuous conformational space is used. Furthermore, we present a data fusion scheme combining multi-sources of data to solve SCA problem. The merit of our framework is the capability of exploiting multi-sources of information to make decisions in a probabilistic perspective based on Bayesian inference. Although our approach aims at SCA problem, it can be easily transplanted to solving for the entire protein structure.

No description available.

Background: The influence of typical manufacturing regimes for producing fixed dental prostheses (FDPs) from yttria partly-stabilized zirconia polycrystals (3Y/4Y/5Y-TZP) on the phase composition is quantified. Methods: Fixed dental prostheses (FDPs) were designed using a CAD process and machined from different Y-TZP blanks from two manufacturers differing in yttria contents. Subsequent to sintering, the FDPs were glaze fired and air-blasted using alumina particles. Phase composition was determined with X-ray diffraction and quantified with Rietveld refinement. Results: The blanks from VITA Zahnfabrik (VITA YZ HT, VITA YZ ST, VITA YZ XT) and Dental Direct (DD Bio ZX2, DD cube ONE, DD cube X2) featured a rhombohedral portion with rather small crystallites and a small monoclinic portion for 3Y/4Y-TZPs, which increased after machining and disappeared after sintering. Glaze firing and air-blasting with alumina particles had no significant influence on the phase composition. Conclusion: The phase history of dental zirconia is revealed, which may have implications on further processing and aging of the FDP (e.g. low temperature degradation) in mouth.

Numerical methods for solving partial differential equations are commonplace in the engineering community and their popularity can be attributed to the rapid performance improvement of modern workstations and desktop computers. The ubiquity of computer technology has allowed all areas of engineering to have access to detailed thermal, stress, and fluid flow analysis packages capable of performing complex studies of current and future designs. The rapid pace of computer development, however, has begun to outstrip efforts to reduce analysis overhead. As such, most commercially available software packages are now limited by the human effort required to prepare, develop, and initialize the necessary computational models. Primarily due to the mesh-based analysis methods utilized in these software packages, the dependence on model preparation greatly limits the accessibility of these analysis tools. In response, the so-called meshless or mesh-free methods have seen considerable interest as they promise to greatly reduce the necessary human interaction during model setup. However, despite the success of these methods in areas demanding high degrees of model adaptability (such as crack growth, multi-phase flow, and solid friction), meshless methods have yet to gain notoriety as a viable alternative to more traditional solution approaches in general solution domains. Although this may be due (at least in part) to the relative youth of the techniques, another potential cause is the lack of focus on developing robust methodologies. The failure to approach development from a practical perspective has prevented researchers from obtaining commercially relevant meshless methodologies which reach the full potential of the approach. The primary goal of this research is to present a novel meshless approach called MIMS (Model Integrated Meshless Solver) which establishes the method as a generalized solution technique capable of competing with more traditional PDE methodologies (such as the finite element and finite volume methods). This was accomplished by developing a robust meshless technique as well as a comprehensive model generation procedure. By closely integrating the model generation process into the overall solution methodology, the presented techniques are able to fully exploit the strengths of the meshless approach to achieve levels of automation, stability, and accuracy currently unseen in the area of engineering analysis. Specifically, MIMS implements a blended meshless solution approach which utilizes a variety of shape functions to obtain a stable and accurate iteration process. This solution approach is then integrated with a newly developed, highly adaptive model generation process which employs a quaternary triangular surface discretization for the boundary, a binary-subdivision discretization for the interior, and a unique shadow layer discretization for near-boundary regions. Together, these discretization techniques are able to achieve directionally independent, automatic refinement of the underlying model, allowing the method to generate accurate solutions without need for intermediate human involvement. In addition, by coupling the model generation with the solution process, the presented method is able to address the issue of ill-constructed geometric input (small features, poorly formed faces, etc.) to provide an intuitive, yet powerful approach to solving modern engineering analysis problems.

In computational fluid dynamics, we often need to solve complex problems with high precision and efficiency. We propose a three-pronged approach to attain this goal. First, we use the discontinuous Galerkin spectral element method (DG-SEM) for its high accuracy. Second, we use graphics processing units (GPUs) to perform our computations to exploit available parallel computing power. Third, we implement a parallel adaptive mesh refinement (AMR) algorithm to efficiently use our computing power where it is most needed. We present a GPU DG-SEM solver with AMR and dynamic load balancing for the 2D wave equation. The DG-SEM is a higher-order method that splits a domain into elements and represents the solution within these elements as a truncated series of orthogonal polynomials. This approach combines the geometric flexibility of finite-element methods with the exponential convergence of spectral methods. GPUs provide a massively parallel architecture, achieving a higher throughput than traditional CPUs. They are relatively new as a platform in the scientific community, therefore most algorithms need to be adapted to that new architecture. We perform most of our computations in parallel on multiple GPUs. AMR selectively refines elements in the domain where the error is estimated to be higher than a prescribed tolerance, via two mechanisms: p-refinement increases the polynomial order within elements, and h-refinement splits elements into several smaller ones. This provides a higher accuracy in important flow regions and increases capabilities of modeling complex flows, while saving computing power in other parts of the domain. We use the mortar element method to retain the exponential convergence of high-order methods at the non-conforming interfaces created by AMR. We implement a parallel dynamic load balancing algorithm to even out the load imbalance caused by solving problems in parallel over multiple GPUs with AMR. We implement a space-filling curve-based repartitioning algorithm which ensures good locality and small interfaces. While the intense calculations of the high order approach suit the GPU architecture, programming of the highly dynamic adaptive algorithm on GPUs is the most challenging aspect of this work. The resulting solver is tested on up to 64 GPUs on HPC platforms, where it shows good strong and weak scaling characteristics. Several example problems of increasing complexity are performed, showing a reduction in computation time of up to 3× on GPUs vs CPUs, depending on the loading of the GPUs and other user-defined choices of parameters. AMR is shown to improve computation times by an order of magnitude or more.

No description available.

[pt] A otimização topológica tem como objetivo encontrar a distribuição mais eficiente de material (ótima topologia) em uma determinada região, satisfazendo as restrições de projeto estabelecidas pelo usuário. Na abordagem tradicional atribui-se uma variável de projeto, constante, denominada densidade, para cada elemento finito da malha. Dessa forma, a qualidade da representação dos novos contornos da estrutura depende do nível de discretização da malha: quanto maior a quantidade de elementos, mais bem definida será a topologia da estrutura otimizada. No entanto, a utilização de malhas super-refinadas implica em um elevado custo computacional, principalmente na etapa de solução numérica das equações de equilíbrio pelo método dos elementos finitos. Este trabalho propõe uma nova estratégia computacional para o refinamento adaptativo local de malhas utilizando elementos finitos poligonais em domínios bidimensionais arbitrários. A ideia consiste em realizar um refinamento da malha nas regiões de concentração de material, sobretudo nos contornos internos e externos, e um desrefinamento nas regiões de baixa concentração de material, como por exemplo, nos furos internos. Desta forma, é possível obter topologias ótimas, com alta resolução e relativamente baixo custo computacional. Exemplos representativos são apresentados para demonstrar a robustez e a eficiência da metodologia proposta por meio de comparações com resultados obtidos com malhas super-refinadas e mantidas constantes durante todo o processo de otimização topológica. / [en] Topology optimization aims to find the most efficient distribution of material (optimal topology) in a given domain, subjected to design constraints defined by the user. The quality of the new boundary representation depends on the level of mesh refinement: the greater the number of elements in the mesh, the better will be the representation of the optimized structure. However, the use of super refined meshes implies in a high computational cost, especially regarding the numerical solution of the linear systems of equations that arise from the finite element method. This work proposes a new computational strategy for adaptive local mesh refinement using polygonal finite elements in arbitrary two-dimensional domains. The idea is to perform a mesh refinement in regions of material concentration, mostly in inner and outer boundaries, and a mesh derefinement in regions of low material concentration such as the internal holes. Thus, it is possible to obtain optimal topologies with high resolution and relatively low computational cost. Representative examples are presented to demonstrate the robustness and efficiency of the proposed methodology by comparing the results obtained herein with the ones from the literature where super refined meshes are held constant throughout all topology optimization process.

This thesis presents a study on two types of X-ray diffraction methodologies applied to the characterization of amorphous materials. The purpose of this study was to assess the feasibility of measuring the diffractive spectrum of amorphous materials by Energy-Dispersive X-ray Diffraction (EDXRD) utilizing Cadmium Zinc Telluride detectors. The total scattering intensity (coherent plus incoherent scatter) spectra precisely measured by high-energy Wide-Angle X-ray Scattering (WAXS) were compared to the EDXRD spectra to determine the level of agreement between the two techniques. The EDXRD spectra were constructed by applying a spectra fusing technique which combined the EDXRD spectra collected at different scattering angles rendering a continuous total scattering spectrum. The spectra fusing technique extended the momentum transfer range of the observed scattered spectrum beyond the limitations of the X-ray source and CZT detection efficiencies. Agreement between the WAXS and fused EDXRD spectra was achieved. In addition, this thesis presents the atomic pair correlation functions and coordination numbers of the first coordination shell for four hydrogen peroxide solutions of varying mass concentrations using Empirical Potential Structural Refinement (EPSR). The results are compared to the state-of-the art ad initio quantum mechanical charge field molecular dynamics (QMCF MD) model of the hydrogen peroxide in solution to support the model's predictions on why hydrogen peroxide is stable in water. The EPSR results using the coherent scattering intensity calculated from the WAXS data set predicts a hydration shell of 6.4 molecules of water surrounding hydrogen peroxide. The results also indicate that hydrogen peroxide is more likely to behave as a proton donor than acceptor. These findings are in agreement with QMCF MD model of aqueous hydrogen peroxide. / Physics

In recent years, deep learning based sequence modeling (neural sequence modeling) techniques have made substantial progress in many tasks, including information retrieval, question answering, information extraction, machine translation, etc. Benefiting from the highly scalable attention-based Transformer architecture and enormous open access online data, large-scale pre-trained language models have shown great modeling and generalization capacity for sequential data. However, not all domains benefit equally from the rapid development of neural sequence modeling. Domains like healthcare and software engineering have vast amounts of sequential data containing rich knowledge, yet remain under-explored due to a number of challenges: 1) the distribution of the sequences in specific domains is different from the general domain; 2) the effective comprehension of domain-specific data usually relies on domain knowledge; and 3) the labelled data is usually scarce and expensive to get in domain-specific settings. In this thesis, we focus on the research problem of applying neural sequence modeling methods to address both common and domain-specific challenges from the healthcare and software engineering domains. We systematically investigate neural-based machine learning approaches to address the above challenges in three research directions: 1) learning with long sequences, 2) learning from domain knowledge and 3) learning under limited supervision. Our work can also potentially benefit more domains with large amounts of sequential data. / Doctor of Philosophy / In the last few years, computer programs that learn and understand human languages (an area called machine learning for natural language processing) have significantly improved. These advances are visible in various areas such as retrieving information, answering questions, extracting key details from texts, and translating between languages. A key to these successes has been the use of a type of neural network structure known as a "Transformer", which can process and learn from lots of information found online. However, these successes are not uniform across all areas. Two fields, healthcare and software engineering, still present unique challenges despite having a wealth of information. Some of these challenges include the different types of information in these fields, the need for specific expertise to understand this information, and the shortage of labeled data, which is crucial for training machine learning models. In this thesis, we focus on the use of machine learning for natural language processing methods to solve these challenges in the healthcare and software engineering fields. Our research investigates learning with long documents, learning from domain-specific expertise, and learning when there's a shortage of labeled data. The insights and techniques from our work could potentially be applied to other fields that also have a lot of sequential data.

Serviceevolution erlaubt es, einen Service durch einen anderen Service zu verfeinern oder zu ersetzen. Der Austausch durch einen anderen Service sollte garantieren, dass alle oder ausgewählte Partner des Originalservices erhalten bleiben. In dieser Arbeit entwickeln wir einen Ansatz welcher einem Serviceentwickler helfen soll, Analyse- und Syntheseaufgaben für den Serviceaustausch so durchzuführen, dass jeder Partner eines gegebenen Services beim Austausch erhalten bleibt. Wir modellieren einen Kontrollfluss eines Services als Beschreibung der Reihenfolge von asynchron kommunizierenden Ereignissen mittels eines impliziten ungeordneten Nachrichtenspeichers. Weiterhin studieren wir den Verhaltensaspekt von korrekter Interaktion zwischen Services und konzentrieren uns auf zwei Varianten von Verklemmungsfreiheit als Korrektheitskriterien von Serviceersetzung. Der wichtigste Beitrag ist ein Ansatz zur Charakterisierung jedes möglichen Austausches eines gegebenen Services. Die zentrale Idee dieses Ansatzes ist eine systematische Untersuchung der Verbindung zwischen einem Service und all seiner Partner bzgl. eines gegebenen Korrektheitskriteriums. Wir nutzen diese Verbindung um von einem gegebenen Service einen kanonischen Partner und einen kanonischen Austausch bzgl. aller Partner zu synthetisieren. Ein Service welcher den kanonischen Austausch eines gegebenen Services verfeinert wird als Austausch des gegebenen Services angesehen, wenn die Menge all seiner Partner jeden Partner des gegebenen Services enthält. Mit dem kanonischen Austausch eines gegebenen Services identifizieren wir die Menge der möglichen austauschenden Services eines gegebenen Services bei der jeder exakt die gleichen Partner wie der gegebene Service hat. Einige Ergebnisse dieser Arbeit fundieren auf früheren Arbeiten zu Austausch und Korrektheit von Services und können daher mit diesen verbunden werden um schwierigere Analyse- und Syntheseaufgaben für den Serviceaustausch durchzuführen. / Service evolution allows one service to be refined into or substituted by another service. Substituting one service by another service should guarantee to preserve all or selected partners of the original service. In this thesis, we develop an approach that shall assist a service designer, such as a domain expert, to perform analysis and synthesis tasks on service substitution. We model a control flow of services that describes the ordering of asynchronously communicating events over an implicit unordered message buffer. We study the behavioral aspect of correct interaction between services and concentrate on two variants of deadlock freedom as correctness criteria of service substitution. The major contribution of this thesis is an approach for characterizing the set of all substitutes for a given service. We systematically investigate the relationship between a service and all its partners under a given correctness criterion and employ this relationship to synthesize from a given service its canonical partner and its canonical substitute with respect to all partners. A service that refines the canonical substitute for a given service is regarded as a substitute for the given service if the set of all its partners includes every partner of the given service. With the canonical substitute of a given service, we identify a specific subset of the set of all substitutes for the given service, each of which has exactly the same set of partners as that of the given service. Parts of the results in this thesis have been established upon previous works on service substitution and correctness of services. Consequently, we can also combine our results with the related existing techniques to perform more sophisticated analysis and synthesis tasks on service substitution.