Data collected at the LHC are analyzed by the ATLAS collaboration for evidence of dark matter. In this thesis, a fast simulation of the ATLAS detector response using the Delphes software is assessed for dark matter models with a leptonically decaying Z boson and a pair of dark matter particles in the final state. Limits for the Two Higgs Doublet plus pseudoscalar dark matter model are obtained using simplified systematics, and found to be nearly indistinguishable to limits obtained using the more complex standard ATLAS analysis. / Graduate

Nuclear logging-while-drilling (LWD) measurements acquired in high-angle and horizontal (HA/HZ) wells are influenced by tool, geometrical, and petrophysical effects. Reliable interpretation of petrophysical and geometrical properties from LWD measurements acquired in thinly-bedded formations requires that gamma ray, density, photoelectric (PEF), and neutron measurements be quantitatively integrated with explicit consideration of their effective volume of investigation (EVOI). One of the effects of different tool EVOIs is false gas density-neutron crossovers across thinly-bedded formations. Also, in the presence of tool eccentricity, azimuthally-varying standoff gives rise to an azimuthally-varying effective depth of investigation (EDOI), which introduces errors in the inference of formation dip. Conventional Monte Carlo simulations of nuclear measurements are computationally expensive in reproducing multi-sector LWD responses in HA/HZ wells. Using linear iterative refinement of pre-calculated flux sensitivity functions (FSFs), we introduce a fast method for numerical simulation of LWD nuclear images in the presence of tool eccentricity along any well trajectory. Our investigation of measurement responses from FSFs motivates techniques to explicitly consider the EVOI of LWD nuclear measurements. Simple radial DOI and standoff corrections suffice for interpretation of gamma-gamma images but are inadequate for neutron responses due to larger EVOI and azimuthal aperture. We introduce a new azimuthal deconvolution method of neutron images to improve bed-boundary detection. Neutron DOI varies significantly with porosity, whereby we correct neutron images for penetration length due to changes of porosity along the well trajectory. In addition, we implement a new method of separate linear iterative refinement on neutron thermal group responses to improve the resolution of neutron images across heterogeneous and thinly-bedded formations. The method reduces shoulder-bed effects and false neutron-density gas crossovers. We corroborate these techniques with rigorous Monte Carlo simulations in vertical and deviated wells. A field example of application conclusively indicates that numerical simulation of LWD nuclear measurements is necessary for reliable estimation of petrophysical properties. / text

Microwave lenses support low-phase error, wideband, wide-angle scanning, and true-time delay (TTD) beam forming. They provide ideal performance for applications such as satellites, remote-piloted vehicles, collision-avoidance radars and ultra-wideband communications systems. The emerging printed lenses in recent years have facilitated the advancement of designing high performance but low-profile, light-weight, and small-size beam-forming networks (BFNs). The microwave lens adopts a few beam ports to illuminate the prescribed receiving ports that feed energy into radiating antennas. Multi-beam patterns can be achieved by exciting multiple beam ports at a time. The design process starts with path-length equations from a limited number of beam-port foci assumptions. This constraint does not take into account the amplitude information; however, it allows an initial lens geometry to be solved. The resulted scanning angle of microwave lens is limited by the beam port contour, as such ± 90 degrees. In this dissertation, three contributions are made from the aspects of minimized phase errors, accurate and efficient simulation algorithms, and 360-degree scanning range extension. First, a minimum-phase-error, non-focal lens design method is proposed. It does not require a specific number of foci along the beam contour; however, minimum phase errors for all beam ports are able to be achieved. The proposed method takes into account flexible prescribed geometrical design parameters, and adopts numerical optimization algorithms to perform phase error minimization. Numerical results compared with the published tri-focal and quadru-focal lenses demonstrate the merits of the proposed method. Second, an accurate and fast simulation method for the microwave lens has been developed to predict the phase, amplitude, array factor, and power efficiency performance. The proposed method is compared to both full-wave simulation and measurement. Comparable results have been achieved. Third, a novel method for a 360-degree scanning microwave lens is proposed. This concept uses the beam ports and the receive ports in an interleaving sequence such that adjacent ports alternate beam and receive functions. The result is a lens that produces scanned beams on opposite sides of the structure resulting in a 360-degree scanning range. The structure can use multiple opposing facets or continuous circular-port and radiating-element contours. To prove the concept, a four-facet microstrip lens has been designed, simulated, fabricated, and tested. The comparison between full-wave simulation and measurement has demonstrated good agreement. / Ph. D.

The resolution of CMOS image sensor is becoming higher and higher, while for identifying its performance, designers need to do a series of simulations, and this work consumes large CPU time in classical design environment. This thesis titled "Fast Scalable and Variability Aware CMOS Image Sensor Simulation Methodology" is dedicated to explore a new simulation methodology for improving the simulation capability. This simulation methodology is used to study the image sensor performance versus low level design parameter, such as transistor size and process variability. The simulation methodology achieves error less than 0.4% on 3T-APS architecture. The methodology is tested in various pixel architectures, and it is used in simulating image sensor with 15 million pixels, the simulation capability is improved 64 times and time consumption is reduced from days to minutes. The potential application includes simulating array-based circuit, such as memory circuit matrix simulation. / L’amélioration de la résolution de ces capteurs implique la nécessité pour les concepteurs de réaliser des séries de simulation de plus en plus longue dans le but de caractériser leurs performances, et ces simulations qui génèrent des résultats difficiles à analyser requièrent de très grandes ressources de calcul ainsi qu’une grande quantité de mémoire. Cette thèse intitulée "Méthode de simulation rapide de capteur d'image CMOS prenant en compte les paramètres d'extensibilité et de variabilité" explore une nouvelle méthodologie de simulation pour améliorer les capacités de traitement actuelles. La méthode qui a été développée est utilisée pour étudier et comparer les performances d’un capteur d’images avec les paramètres de bas niveau de conception de tels circuits ; par exemple la taille des transistors ainsi que la variabilité. La méthodologie obtient l'erreur de sortie moins de 0,4% sur le capteur d’image de style APS-3T. La méthode a été testée avec diverses architectures de pixel, et elle a permis de simuler un capteur d'image de 15 millions de pixels. La vitesse de simulation est améliorée 64 fois, passant de plusieurs jours à plusieurs minutes. La simulation des circuits présentant une structure en matrice comme les mémoires est une autre application potentielle de ce type de méthodologie.