Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal.

In this thesis, we first present a brief overview of various theoretical approaches used to examine the electronic structure of defects in GaN. Using the recently developed hybrid density functional theory (HSE06) along with the experimental measurements, we propose a new explanation of the nature of the yellow luminescence band in carbon-doped GaN. We conduct a systematic study of electronic and optical properties of defects (Carbon, Oxygen, Silicon related) that are candidates for the origin of yellow luminescence. We show that the CN-ON complex is significantly more likely to form compared to isolated carbon configurations. In contrast to the properties of the isolated carbon acceptor, calculated defect levels and optical transitions involving deep level of the CN-ON complex agree quite well with our thermal luminescence quenching data as well as with the experimentally measured C-doped GaN luminescence spectra. Hence, the CN-ON complex, rather than isolated C impurity, is more likely to resolve a long-standing problem of the yellow luminescence in GaN.

Real-time computing is not just fast computing but time-predictable computing. Many tasks in safety-critical embedded real-time systems have hard real-time characteristics. Failure to meet deadlines may result in the loss of life or in large damages. Known of Worst Case Execution Time (WCET) is important for reliability or correct functional behavior of the system. As multi-core processors are increasingly adopted in industry, it has become a great challenge to accurately bound the worst-case execution time (WCET) for real-time systems running on multi-core chips. This is particularly true because of the inter-thread interferences in accessing shared resources on multi-cores, such as shared L2 caches, which can significantly affect the performance but are very difficult to be estimate statically. We propose an approach to analyzing Worst Case Execution Time (WCET) for multi-core processors with shared L2 instruction caches by using a model checking based method. Our experiments indicate that compared to the static analysis technique based on extended ILP (Integer Linear Programming), our approach improves the tightness of WCET estimation more than 31.1% for the benchmarks we studied. However, due to the inherent complexity of multi-core timing analysis and the state explosion problem, the model checking based approach currently can only work with small real-time kernels for dual-core processors. At the same time, improving the average-case performance and energy efficiency has also been important for real-time systems. Recently, Hybrid SPM-Cache (HSC) architectures by combining caches and Scratch-Pad Memories (SPMs) have been increasingly used in commercial processors and research prototypes. Our research explores HSC architectures for real-time systems to reconcile time predictability, performance, and energy consumption. We study the energy dissipation of a number of hybrid on-chip memory architectures by combining both caches and Scratch-Pad Memories (SPM) without increasing the total on-chip memory size. Our experimental results indicate that with the equivalent total on-chip memory size, several hybrid SPM-Cache architectures are more energy-efficient than either pure software controlled SPMs or pure hardware-controlled caches. In particular, using the hybrid SPM-cache to store both instructions and data can achieve the best energy efficiency. However, the SPM allocation for the HSC architecture must be aware of the cache to harness the full potential of the HSC architecture. First, we propose and evaluate four SPM allocation strategies to reduce WCET for hybrid SPM-Caches with different complexities. These algorithms differ by whether or not they can cooperate with the cache or be aware of the WCET. Our evaluation shows that the cache aware and WCET-oriented SPM allocation can maximally reduce the WCET with minimum or even positive impact on the average-case execution time (ACET). Moreover, we explore four SPM allocation algorithms to maximize performance on the HSC architecture, including three heuristic-based algorithms, and an optimal algorithm based on model checking. Our experiments indicate that the Greedy Stack Distance based Allocation (GSDA) can run efficiently while achieving performance either the same as or close to the optimal results got by the Optimal Stack Distance based Allocation (OSDA). Last but not the least, we extend the two stack distance based allocation algorithms to GSDA-E and OSDA-E to minimize the energy consumption of the HSC architecture. Our experimental results show that the GSDA-E can also reduce the energy either the same as or close to the optimal results attained by the OSDA-E, while achieving performance close to the OSDA and GSDA.

The goal of this research is to cover a broad set of scientific investigations of elastomeric materials based on polydimethylsiloxane (PDMS) and poly((3,3,3-trifluoroethoxymethyl)methyloxetane) diol. The scope of study covers five areas, well correlated with each other. The first study investigates the near surface morphology of condensation cured PDMS as a function of increasing the amount of siliceous phase. The appearance, disappearance and reappearance of untreated fumed silica nanoparticles at the PDMS near surface and their correlation with the volume fraction of siliceous phase have been studied. This research with PDMS nanocomposites has led to the development of an alternative route for improving mechanical strength of PDMS elastomers, conventionally known to have weak mechanical properties. The second study involves synthesis of a triblock copolymer comprising of four mutually immiscible phases, namely, soft segments comprising of fluorous and silicone domains, a diisocyanate hard segment and a glassy siliceous phase. Structure-property relationship has been established with an investigation of the interesting surface and bulk morphology. The highly improved mechanical strength of these soft materials is noteworthy. The dominance of silicone soft block at the triblock near surface has led to the third study which investigates their potential non-adhesive or abhesive characteristic in both a laboratory scale and in a marine environment. The peak removal stress and the removal energy associated with the detachment of a rigid object from the surface of these triblock copolymers have been measured. Results obtained from laboratory scale experiments have been verified by static immersion tests performed in the marine environment, involving the removal of adhered soft and hard fouling organisms. Gaining insights on the characteristics of an easy release surface, namely low surface energy and a low near surface modulus, a new way for controlling the near surface composition for elastomeric coatings have been developed. This technique involves an elastomer end-capped with a siliceous crosslinking agent and a tough, linear polyurethane. The basic concept behind the hybrid compositions is to develop a coating suitable for foul release applications, having a low energy surface, low surface modulus but good bulk mechanical strength. Henceforth, the fourth study deals with synthesis and characterization of the hybrid polymers over a wide range of composition and investigates their foul release characteristic in laborartory scale experiments. In our final study, attempts have been made in generating a silicone coating with antimicrobial property. A quaternary alkylammonium in different weight percents have been incorporated into a conventional, condensation cured polydimethylsiloxane (PDMS) elastomer. Antimicrobial assay has been performed on these modified silicone coatings to assess their biocidal activity against strains of Escherichia coli, Staphylococcus aureus and Pseudomonas aeruginosa. Surface accessibility of quaternary charges has been quantified by measuring the streaming potential of a modified coating. An effort has been made in improving the mechanical strength of the weak PDMS elastomers by adding treated fumed silica nanoparticles as reinforcements. The effect of adding fillers on the mechanical property (tensile), surface concentration of quaternary charge and on the biocidal activity of a representative sample has been investigated.

In recent years, corpus based machine translation systems produce significant results for a number of language pairs. However, for low-resource languages like Urdu the purely statistical or purely example based methods are not performing well. On the other hand, the rule-based approaches require a huge amount of time and resources for the development of rules, which makes it difficult in most scenarios. Hybrid machine translation systems might be one of the solutions to overcome these problems, where we can combine the best of different approaches to achieve quality translation. The goal of the thesis is to explore different combinations of approaches and to evaluate their performance over the standard corpus based methods currently in use. This includes: 1. Use of syntax-based and dependency-based reordering rules with Statistical Machine Translation. 2. Automatic extraction of lexical and syntactic rules using statistical methods to facilitate the Transfer-Based Machine Translation. The novel element in the proposed work is to develop an algorithm to learn automatic reordering rules for English-to-Urdu statistical machine translation. Moreover, this approach can be extended to learn lexical and syntactic rules to build a rule-based machine translation system.

Modelling of biochemical systems has received considerable attention over the last decade from bioengineering, biochemistry, computer science, and mathematics. This thesis investigates the applications of computational techniques to computational systems biology, for the construction of biochemical models in terms of topology and kinetic rates. Due to the complexity of biochemical systems, it is natural to construct models representing the biochemical systems incrementally in a piecewise manner. Syntax and semantics of two patterns are defined for the instantiation of components which are extendable, reusable and fundamental building blocks for models composition. We propose and implement a set of genetic operators and composition rules to tackle issues of piecewise composing models from scratch. Quantitative Petri nets are evolved by the genetic operators, and evolutionary process of modelling are guided by the composition rules. Metaheuristic algorithms are widely applied in BioModel Engineering to support intelligent and heuristic analysis of biochemical systems in terms of structure and kinetic rates. We illustrate parameters of biochemical models based on Biochemical Systems Theory, and then the topology and kinetic rates of the models are manipulated by employing evolution strategy and simulated annealing respectively. A new hybrid modelling framework is proposed and implemented for the models construction. Two heuristic algorithms are performed on two embedded layers in the hybrid framework: an outer layer for topology mutation and an inner layer for rates optimization. Moreover, variants of the hybrid piecewise modelling framework are investigated. Regarding flexibility of these variants, various combinations of evolutionary operators, evaluation criteria and design principles can be taken into account. We examine performance of five sets of the variants on specific aspects of modelling. The comparison of variants is not to explicitly show that one variant clearly outperforms the others, but it provides an indication of considering important features for various aspects of the modelling. Because of the very heavy computational demands, the process of modelling is paralleled by employing a grid environment, GridGain. Application of the GridGain and heuristic algorithms to analyze biological processes can support modelling of biochemical systems in a computational manner, which can also benefit mathematical modelling in computer science and bioengineering. We apply our proposed modelling framework to model biochemical systems in a hybrid piecewise manner. Modelling variants of the framework are comparatively studied on specific aims of modelling. Simulation results show that our modelling framework can compose synthetic models exhibiting similar species behaviour, generate models with alternative topologies and obtain general knowledge about key modelling features.

(no abstract available)

Master of Science / Department of Electrical and Computer Engineering / Balasubramaniam Natarajan / Noel Schulz / With increasing costs for fuel there is a growing interest in improving fuel efficiency and performance of military vehicles by employing (1) hybrid drive train architecture; (2) reliable vehicle power system structure, and (3) effective power management strategies of multiple power sources (engine, battery and ultracapacitor) and vehicle electrical loads. However, current ruled-based power management strategies that focus primarily on traction fail to meet the rapidly increasing requirements of military vehicles, including: (1) better fuel economy; (2) the ability to support pulsed power weapon loads; (3) maintaining battery SOC for power offloading applications, and (4) the ability to perform load scheduling of vehicle non-traction electrical loads to save energy. In this thesis, we propose an optimal control based algorithm in conjunction with a rule-based control strategy to optimally manage three power sources (engine, battery and pulsed power supply module) and an effective power management solution for vehicle non-traction electrical loads such that: (1) all traction, non-traction and pulsed power needs are met; (2) power drawn from the engine for specific mission is minimized; (3) a certain desired battery SOC is guaranteed for offloading power, and (4) the ability to perform load scheduling based on different mission requirements. The proposed approach is validated using simulation of a mission specific profile and is compared with two other popular control strategies. The improvements in power efficiency, desired SOC level and ability to perform optimal load scheduling are demonstrated.

Master of Agribusiness / Department of Agricultural Economics / John Crespi / Hybrid seed corn pricing has increased significantly over the past six or seven years and continues to be a topic of conversation amongst farmers. This issue is also an area of concern for Monsanto. The hybrid corn pricing team at Monsanto is concerned that they price current products at a point to maximize profits while continuing to grow market share. The key is to price at a point that captures all the value of the differentiated products Monsanto offers. The objective for this study is to estimate a demand model for the hybrid seed corn industry. The demand model will allow us to look at many different aspects of the hybrid seed corn industry and also evaluate the own-price and cross-price elasticities. The own-price elasticity is especially important because it will be used to determine if current pricing is revenue or profit-maximizing. A hedonic pricing model was also estimated in this study to complement the demand model. It is important for Monsanto to understand what attributes or traits are significant in pricing and demand.

Doctor of Philosophy / Department of Agronomy / Tesfaye Tesso / The approach used to identify inbred lines that can produce superior hybrids is costly and time-consuming. It requires creation of all possible crosses and evaluation of the crosses to estimate combining abilities for the desired traits. Predicting heterosis or hybrid performance in any way possible may help to reduce the number of crosses to be made and evaluated. In this study, four sets of experiments were conducted to determine whether heterosis can be predicted based on inbred line performance, genetic distance between parents and genomic prediction model. The first experiment was aimed at assessing the levels of genetic diversity, population structure and linkage disequilibrium (LD) in 279 public sorghum inbred lines, based on 66,265 SNPs generated using the genotyping-by-sequencing (GBS) platform. The inbred lines were developed at different times over the last two decades and harbor robust diversity in pedigree and agronomic characteristics. Some of the inbreds are resistant to Acetolactate synthase (ALS) and Acetyl co-enzyme-A carboxylase (ACC) inhibitor herbicides. The mean polymorphic information content (PIC) and gene diversity across the entire inbreds were 0.35 and 0.46, respectively with non-herbicide resistant inbreds harboring more diversity than the herbicide resistant ones. The population structure analysis clustered the inbred lines into three major subgroups according to pedigree and fertility-reaction with the maintainer lines (B-lines) distinctly forming a separate cluster. Analysis of molecular variance (AMOVA) revealed more variation within subgroups than among subgroups. Substantial linkage disequilibrium (LD) was detected between the markers in the population with marked variation between chromosomes. This information may facilitate the use of the inbreds in sorghum breeding programs and provide perspectives for optimizing marker density for gene mapping and marker-assisted breeding. The second experiment, based on 102 F1 hybrids developed by intercrossing closely and distantly related inbreds, was conducted to investigate the relationship of genetic distance between parents with hybrid vigor or heterosis. The F1 hybrids alongside their parents were evaluated at two environments in a randomized complete block design with three replications. The results show that correlations of genetic distance between parents with hybrid performance and heterosis were variable and dependent on the trait. Though most were statistically non-significant and not strong to be used as predictor for heterosis, the results tend to show that certain level of genetic distance between parents is needed to capture maximum heterosis and hybrid performance. The objective of the third research study was to determine whether traits measured on parents can be used to predict hybrid performance in sorghum and to assess the combining ability of selected inbreds. Forty-six parental inbred lines and 75 F1 hybrids generated from intercrossing the inbreds were evaluated in four environments in a randomized complete block design with three replications. The average performance of the parents (mid-parent) was significantly correlated with hybrid performance for thousand kernel weight, days to flowering and plant height. Significant general (GCA) and specific (SCA) combining abilities were observed for most traits, with highly significant GCA effects observed for most traits as compared to SCA indicating that additive genetic effects are more important in affecting the inheritance of the traits measured. Results show that studying parental inbred line performance could generate important information for predicting hybrid performance in sorghum. The fourth experiment was aimed at assessing the efficacy of genomic prediction of hybrid performance in sorghum. Genomic prediction was performed with five-fold cross-validation procedure on 204 F1 hybrids developed using 102 inbred lines. A total of 66,265 SNP markers generated using genotyping-by-sequencing were used in this study. Results showed that increasing training population size increased prediction accuracies for all traits with the effect being different for different traits. Also, considering additive effects alone versus additive and dominance effects in the model showed similar trend of prediction accuracy but the full model (considering both additive and dominance effects of the markers) provided better prediction at least for some of the traits. The results suggest that genomic prediction could become an effective tool for predicting the performance of untested sorghum hybrids thus adding efficiency to hybrid selection.