DNA-encoded chemical libraries (DEL) allows an exhaustive chemical space sampling with a large-scale data consisting of compounds produced through combinatorial synthesis. This novel technology was utilized in the early drug discovery stages for robust hit identification and lead optimization. In this project, the aim was to build a Machine Learning- based QSAR model with conformal prediction for hit identification on two different target proteins, the DEL was assayed on. An initial investigation was conducted on a pilot project with 1000 compounds and the analyses and the conclusions drawn from this part were later applied to a larger dataset with 1.2 million compounds. With this classification model, the prediction of the compound activity in the DEL as well as in an external dataset was aimed to be analyzed with identification of the top hits to evaluate model’s performance and applicability. Support Vector Machine (SVM) and Random Forest (RF) models were built on both the pilot and the main datasets with different descriptor sets of Signature Fingerprints, RDKIT and CDK. In addition, an Autoencoder was used to supply data-driven descriptors on the pilot data as well. The Libsvm and the Liblinear implementations were explored and compared based on the models’ performances. The comparisons were made by considering the key concepts of conformal prediction such as the trade-off between validity and efficiency, observed fuzziness and the calibration against a range of significance levels. The top hits were determined by two sorting methods, credibility and p-value differences between the binary classes. The assignment of correct single-labels to the true actives over a wide range of significance levels regardless of the similarity of the test compounds to the training set was confirmed for the models. Furthermore, an accumulation of these true actives in the models’ top hit selections was observed according to the latter sorting method and additional investigations on the similarity and the building block enrichments in the top 50 and 100 compounds were conducted. The Tanimoto similarity demonstrated the model’s predictive power in selecting structurally dissimilar compounds while the building block enrichment analysis showed the selectivity of the binding pocket where the target protein B was determined to be more selective. All of these comparison methods enabled an extensive study on the model evaluation and performance. In conclusion, the Liblinear model with the Signature Fingerprints was concluded to give the best model performance for both the pilot and the main datasets with the considerations of the model performances and the computational power requirements. However, an external set prediction was not successful due to the low structural diversity in the DEL which the model was trained on.

No description available.

Dans cette thèse sont principalement étudiés des systèmes quantiques intégrables critiques avec l’ansatz de Bethe qui ont la propriété particulière d’être non-unitaires ou non-compacts. Ceci concerne des modèles de physique statistique non-locaux tels que la percolation, mais aussi par exemple les systèmes désordonnés. Ce manuscrit présente à la fois des études détaillées de la limite continue de modèles intégrables sur réseau, et développe de nouvelles techniques pour étudier cette correspondance. Dans une première partie nous étudions en détail la limite continue de chaînes de superspins non-unitaires (et parfois non-compactes) qui ont une symétrie orthosymplectique. Nous montrons qu’il s’agit de modèles sigma sur supersphère en calculant leur spectre avec la théorie des champs, avec l’ansatz de Bethe, et numériquement. Leur non-unitarité autorise une brisure spontanée de symétrie habituellement interdite par le théorème de Mermin-Wagner. Leur caractère de perturbation marginale d’une théorie conforme des champs logarithmique est particulièrement étudié. Nous établissons également une correspondance précise entre le spectre et des configurations de boucles avec intersections, et obtenons de nouveaux exposants critiques pour les chemins non-recouvrants compacts ainsi que leurs corrections logarithmiques multiplicatives. Cette étude fut par ailleurs l’occasion de développer une nouvelle méthode pour calculer le spectre d’excitation d’une chaîne de spin quantique critique à partir de l’ansatz de Bethe, incluant les corrections logarithmiques, également en présence de racines de Bethe dites ’en chaînes’, et qui évite les méthodes de Wiener-Hopf et les équations intégrales non-linéaires. Dans une deuxième partie nous abordons l’influence d’un champ magnétique sur une chaîne de spin quantique et montrons que des séries convergentes peuvent être obtenues pour plusieurs quantités physiques telles que l’aimantation acquise ou les exposants critiques, dont les coefficients peuvent être calculés efficacement par récurrence. La structure de ces relations de récurrence permet d’étudier génériquement le spectre d’excitation, et elles sont applicables y compris dans certains cas où les racines de Bethe sont sur une courbe dans le plan complexe. Nous espérons que l’étude de la continuation analytique de ces séries puisse être utile pour les chaînes non-compactes. Par ailleurs, nous montrons que les fluctuations à l’intérieur de la courbe arctique du modèle à six vertex avec conditions aux bords de type mur sont décrites par un champ Gaussien libre avec une constante de couplage dépendant de la position, qui peut être calculée à partir de l’énergie libre de la chaîne XXZ avec une torsion imaginaire dans un champ magnétique. / This thesis mainly deals with integrable quantum critical systems that exhibit peculiar features such as non-unitarity or non-compactness, through the technology of Bethe ansatz. These features arise in non-local statistical physics models such as percolation, but also in disordered systems for example. The manuscript both presents detailed studies of the continuum limit of finite-size lattice integrable models, and develops new techniques to study this correspondence. In a first part we study in great detail the continuum limit of non-unitary (and sometimes non-compact) super spin chains with orthosymplectic symmetry which is shown to be supersphere sigma models, by computing their spectrum from field theory, from the Bethe ansatz, and numerically. The non-unitarity allows for a spontaneous symmetry breaking usually forbidden by the Mermin-Wagner theorem. The fact that they are marginal perturbations of a Logarithmic Conformal Field Theory is particularly investigated. We also establish a precise correspondence between the spectrum and intersecting loops configurations, and derive new critical exponents for fully-packed trails, as well as their multiplicative logarithmic corrections. During this study we developed a new method to compute the excitation spectrum of a critical quantum spin chain from the Bethe ansatz, together with their logarithmic corrections, that is also applicable in presence of so-called ’strings’, and that avoids Wiener-Hopf and Non-Linear Integral Equations. In a second part we address the problem of the behavior of a spin chain in a magnetic field, and show that one can derive convergent series for several physical quantities such as the acquired magnetization or the critical exponents, whose coefficients can be efficiently and explicitely computed recursively using only algebraic manipulations. The structure of the recurrence relations permits to study generically the excitation spectrum content – moreover they are applicable even to some cases where the Bethe roots lie on a curve in the complex plane. It is our hope that the analytic continuation of such series might be helpful the study non-compact spin chains, for which we give some flavour. Besides, we show that the fluctuations within the arctic curve of the six-vertex model with domain-wall boundary conditions are captured by a Gaussian free field with space-dependent coupling constant that can be computed from the free energy of the periodic XXZ spin chain with an imaginary twist and in a magnetic field.

In this thesis an analysis is performed on operation times for different sized products in a manufacturing company. The thesis will introduce and summarise most of the theory needed to perform regression and also cover a worked example where three different regression models are learned, evaluated and analysed. Conformal prediction, which at the moment is a hot topic in machine learning, will also be introduced and will be used in the worked example.

Die perturbative Quantenfeldtheorie ist das am weitesten entwickelte Modell, zur präzisen Analyse vieler verschiedener physikalischer Phänomene. Das Standardwerkzeug der perturbativen QFT sind Feynman-Diagramme. Bei Rechnungen bis zu einer Schleife ist aber auch der Wärmekern eine mächtige Technik. Um in der QFT, über die Störungstheorie hinauszugehen ist Symmetrie von großer Bedeutung. Insbesondere die konforme Symmetrie schränkt die Korrelatoren stark ein. In dieser Arbeit wird sie mit der Average Null Energy Condition (ANEC) kombiniert, um die Hofman-Maldacena-Schranken für die Anomaliekoeffizienten in vier Dimensionen abzuleiten. Im Folgenden untersuchen wir verschiedene Probleme der perturbativen Quantenfeldtheorie. Zunächst studieren wir die Weyl-Anomalie für einen nicht-konformen freien Skalar in einer vierdimensionalen gekrümmten Raumzeit. Wir verstehen die Definition der Anomalie diagrammatisch ohne klassische Symmetrie und wir interpretieren die bekannte Wärmekernberechnung präzise. Dann untersuchen wir Eichtheorien mit höheren Ableitungen in sechs Dimensionen. Diese sind natürliche Kandidaten, um perturbativ nicht-unitäre konforme Theorien zu konstruieren. Die Berechnung erfolgt mit der Wärmekernmethode und wir leiten den allgemeinen Ausdruck des, zuvor nicht Bekannten, relevanten Koeffizienten her. Supersymmetrie sowie das Hinzufügen des Yang-Mills-Terms werden ebenfalls berücksichtigt. Schließlich beginnen wir die Untersuchung der Implikationen der ANEC auf nicht-konforme Feldtheorien, angefangen mit dem selbst-wechselwirkenden Skalar in vier Dimensionen. Der Energiefluss eines Zustands mit einer einzelnen Feldeinfügung wird berechnet. Ausgehend von den perturbativen Euklidischen Korrelatoren im Impulsraum konstruieren wir die relevante Wightman-Funktion, um den Energiefluss auszuwerten. Die Berechnung ist kompliziert, aber wir erhalten das erwartete Ergebnis und eröffnen so die Möglichkeit, interessantere Zustände zu untersuchen. / Perturbative quantum field theory is our most developed framework to accurately analyse many physical phenomena. The standard tool is Feynman diagrams, but at one-loop the heat kernel is also a powerful technique. It is however difficult to go beyond perturbation theory, and symmetry is a key factor. In particular, conformal symmetry strongly restricts the correlators, and have been combined with the average null energy condition (ANEC) to derive the Hofman-Maldacena bounds on the anomaly coefficients in four dimensions. In this thesis we study different problems in perturbative quantum field theory. First, we study the Weyl anomaly for a non-conformal free scalar in a four-dimensional curved spacetime. We diagrammatically understand the definition of the anomaly without classical symmetry, and we precisely interpret the well-known heat kernel calculation. Then, we study higher-derivative gauge theories in six dimensions. These theories are the natural candidate to perturbatively construct non-unitary conformal theories. The calculation is done with the heat kernel method and we derive the general expression of the relevant coefficient, which was previously unknown. Supersymmetry or the addition of a Yang-Mills term are also considered. Finally, we initiate the study of the consequence of the ANEC on non-conformal field theories with the example of the self-interacting scalar in four dimensions. The energy flux of a state with a single field insertion is computed. Starting from the perturbative momentum-space Euclidean correlators, we construct the relevant Wightman function to evaluate the energy flux. The calculation is considerably complicated, but we recover the expected result, opening the possibility of studying more interesting states.

No description available.

In this thesis we employ string perturbation theory in toroidal orbifold models to study aspects of supersymmetry breaking in type IIB string theory. First, we determine the dependence of physical Yukawa couplings on blow-up moduli in models with D3-branes at orbifold singularities. Blow-up moduli are scalar fields describing the size of small blow-up cycles in the compactification geometry. In models implementing moduli stabilisation these fields can acquire F-terms and break supersymmetry. We examine the moduli-dependence of physical Yukawa couplings at string tree-level by computing disk correlation functions involving a Yukawa interaction of visible sector fields and an arbitrary number of blow-up moduli. We perform the calculation for one blow-up insertion explicitly and find that the correlation function vanishes if the blow-up modulus is associated with a small cycle distant to the visible sector. For more than one blow-up insertion we show that all such correlation functions are exponentially suppressed by the compactification volume. We explain how these results are relevant to suppressing soft terms to scales parametrically below the gravitino mass. Further, we determine corrections to holomorphic Yukawa couplings on D3-branes at an orbifold singularity due to non-perturbative effects such as gaugino condensation on a stack of D7-branes. This can be done by calculating a one-loop threshold correction to the gauge coupling on the D7-branes. We show that, if present, the new contributions to Yukawa couplings are not aligned with the tree-level couplings. As the new Yukawa couplings contribute to soft A-terms they are sources of flavour-changing neutral currents. Last we discuss an effect unrelated to supersymmetry breaking. We show that orbifold models with D3-branes at orbifold singularities can exhibit kinetic mixing of different massless Abelian factors. For this to be possible, the relevant U(1) factors have to be associated with more than one orbifold singularity.

Thesis (PhD (Industrial Engineering))--University of Stellenbosch, 2011. / ENGLISH ABSTRACT: Injection moulding of polymer components is subject to ever increasing demands for improved part quality and production rate. It is widely recognised that the mould cooling strategy employed is crucial to achieving these goals. A brief overview of injection moulding units and different types of injection moulds is given. The modern Additive Manufacturing (AM) technology for processing metal powders such as Direct Metal Laser Sintering (DMLS) and Selective Laser Melting (SLM) offers almost full freedom to the mould designer. Some of these modern manufacturing methods based on metal powders, which are able to produce complex cooling channels are analysed. A drastic change has entered the mould design domain - shifting the paradigm from design for manufacture to manufacture for design. In combination with suitable AM methods the concept of surface cooling moulds can now be efficiently implemented. This study presents a new approach of predicting the minimum cooling time required for the produced part. Different cooling layouts are analysed taking the heat transfer into consideration. The lumped heat capacity method is implemented in this research in order to determine the minimum cooling cycle time required. A new approach was developed to determine the most suitable cooling layout configuration, such as conventional cooling, conformal cooling or surface cooling, required for a moulded part based on its characteristics such as shape complexity, space available for the cooling layout, part quality requirements, production volume, and product life cycle. A mould cooling design process including simulation, reverse engineering and manufacturing of the mould insert was implemented in this study. In order to validate the generic model developed during the course of this research comparative experiments were carried out to determine the difference in performance of injection moulding using conventional or surface cooling methods. The experimental results showed a significant improvement in part quality produced with reduced cycle times using the surface cooling method. / AFRIKAANSE OPSOMMING: ‘Injection Moulding’ van polimeer komponente word al meer gedruk vir verbeterde kwaliteit en vinniger produksie tyd. Dit is orals bekend dat die gietvorm afkoeling strategie ‘n groot rol speel om hierdie twee doelwitte te bereik. Eers word ‘n kort oorslag gegee van ‘Injection Moulding’ eenhede en van verskillende ‘Injection Moulding’ vorms. Die moderne Aditatiewe Vervaardigingstegnologie vir die prosessering van metaal poeiers soos bv. Direkte Metaal Laser Sintering (DMLS) en Selektiewe Laser Smelting (SLM) bied basies volle vryheid ten opsigte van gietvorm ontwerp. Party van die moderne vervaadigings metodes, wat op metaal poeiers gebaseer is, wat komplekse koelings kanale kan produseer word geanaliseer. Die ontwerpers arena het ‘n groot verandering ondergaan deurdat die fokus van ontwerp vir vervaardiging verskuif het na vervaardiging vir ontwerp. In kombinasie met toepaslike aditatiewe vervaardigings metodes kan oppervlak verkoeling nou effektief geïmplementeer word. Hierdie studie bied a nuwe manier om die minimum verkoelings tyd benodig vir ‘n part te voorspel. Verskeie verkoelings uitlegte word geanaliseer waar hitte oordrag in ag geneem word. Die “lumped heat capacity” metode word gebruik om die minimum siklus tyd te bepaal. ‘n Nuwe benadering is ontwikkel om die mees geskikste verkoelings uitleg soos bv. konvensionele verkoeling, konvorme verkoeling of oppervlak verkoeling te bepaal vir ‘n spesifieke part gebaseer op die part se vorm kompleksiteit, spasie beskikbaar vir verkoelings kanale, kwaliteit vereistes en produk lewensiklus. Die volgende is in die studie geïmplementeer: ‘n vorm verkoelings ontwerp proses met simulasie, ‘reverse engineering’ en vervaardiging van die vorm insetsel. Om die generiese model te verifieer gedurende die studie is vergelykende eksperimente uitgevoer om die verskil in prestasie te bepaal tussen die gebruik van konvensionele en oppervlak verkoelings metodes. Die eksperimentele resultate het ‘n beduidende verbetering in part kwaliteit getoon met ‘n verkorte siklus tyd tydens die gebruik van die oppervlag verkoelings metode.

This thesis is a review of research done over the course of the past 4 years, divided into two unrelated parts. The rst is set in the context of Bagger-Lambert-Gustavsson models, based on 3-Lie algebras. In particular I will describe theories with metric 3-algebras of inde nite signature: these present elds with negative kinetic terms. The problem can be solved by gaugeing away the non-physical degrees of freedom, to obtain other well understood theories. I will show how this procedure can be easily applied for 3-algebra metrics of any inde nite signature. Part II of this thesis focuses on solutions of topologically massive gravity (TMG): particular attention is devoted to warped AdS3 black holes, which are discussed in great detail. I will present a novel analysis of the near horizon geometries of these solutions. I further propose an approach for searching for new solutions to 3-dimensional gravity based on conformal symmetry. This approach is able to yield most of the known axisymmetric stationary TMG backgrounds.

Cette thèse étudie l'utilisation de certains invariants asymptotiques en géométrie conforme et géométrie CR.La première partie est consacrée à la géométrie conforme. Nous calculons les premiers termes du développement asymptotique de la fonction de Green des opérateurs GJMS au voisinage de la diagonale, pour un facteur conforme normal au sens de Lee et Parker. Nous montrons que le terme constant de ce développement est covariant sous un changement de facteur conforme normal. Nous le rattachons à un invariant à l'infini de type masse ADM d'une métrique non compacte obtenue par projection stéréographique.La deuxième partie est consacrée à la géométrie CR. Nous calculons les premiers termes du développement asymptotique de la fonction de Green de l'opérateur de Yamabe CR au voisinage de sa singularité,dans le cas CR sphérique, et en dimension 3 dans une carte CR-normale au sens de Jerison et Lee, lorsque la constante de Yamabe-CR est strictement positive. Nous montrons la covariance pseudo-conforme du terme constant sous les changements de cartes respectivement CR-sphériques et CR-normales.La troisième partie donne une explication formelle à une annulation algébrique sur laquelle repose la définition de plusieurs invariants à l'infini de type masse ADM, qui n'avait pu jusqu'à présent qu'être constatée par un calcul direct. / In this thesis we study the use of some asymptotic invariants in conformal and CR geometry.The first chapter is devoted to conformal geometry. We compute an asymptotic expansion ofthe Green function of GJMS operators near the diagonal, for a normal conformal factorin the sense of Lee and Parker. We show that the constant term in this expansion is covariant through achange of normal conformal factor. We relate it to an invariant at infinity of the type of the ADM massof a non-compact metric obtained by some kind of stereographic projection.In the second chapter we study CR geometry. We compute the first terms of the asymptotic expansion of the Greenfunction of the Yamabe-CR operator near its singularity, when the Yamabe-CR constant is positive, in the CR-sphericalcase, and in dimension 3 in a CR-normal chart in the sense of Jerison and Lee.We show the pseudo-conformal covariance of the constant term in this asymptotic expansion through a change of spherical chart andof CR-normal chart respectively.In the third chapter we give a formal explanation to an algebraic cancellationon which the defintion of some invariants at infinity such as the ADM mass relies.