-
Prospective Students
-
Faculty and Staff
-
Current Students
-
Alumni and Friends
We propose methods for functional predictor selection and the estimation of smooth functional coefficients simultaneously in a scalar-on-function regression problem under a high-dimensional multivariate functional data setting. In particular, we develop two methods for functional group-sparse regression under a generic Hilbert space of infinite dimension. We show the convergence of algorithms and the consistency of the estimation and the selection (oracle property) under infinite-dimensional Hilbert spaces. Simulation studies show the effectiveness of the methods in both the selection and the estimation of functional coefficients. The applications to functional magnetic resonance imaging (fMRI) reveal the human brain regions related to ADHD and IQ. In addition, we apply the proposed methods to an econometric data set to find the related functional covariates to GDP of a country. To extend the results, we propose numerical algorithms for more complex models, such as nonlinear (via RKHS), logistic, sparse function--on--function, and standardization of the results of the sparse scalar--on--function models before we list the applications of these extensions to the brain image data analysis.
The modern economy runs with heavy reliance on the free flow of goods across the international logistic and supply chain. Advances in international freight transportation systems supported by intermodal integration, freight containerization, hub-and-spoke shipping system, and supply chain security, has reduced the distance friction of flow of goods and drastically lowered physical barriers of commercial activities. However, it is little known yet how spatial interactions of trade and shipping take place under the complex logistic chain process and what spatial phenomena ensue from such processes. In this dissertation, I study the nature of spatial interaction phenomena in the context of the contemporary state of the international transportation system. First, I study how the spatial structure of the port system is formed with intermodal integration of the modern international logistic system across land and water. Second, I explore how the hub-and-spoke system in the international transportation network contributes to the global shrinkage of space. Third, I investigate the effect of domestic armed conflicts developed by political instability on freight mobility and ensuing differential openness of regions to the global market. Results of the three pieces of research are as follows. First, the spatial structure of the port system is found to comprise interdependent collections of hinterlands, feeder and hub ports, and forelands along a logistical continuum, which mirror the functional division of logistic processes across space. Second, the hub-and-spoke shipping system reduces the distance friction of shipping flows and is the main driver of global shrinkage of space in terms of long-distance trade. Third, freight mobility is found to be greatly compromised by the lack of logistic chain security stemming from prevailing armed violence along inland transportation corridors. The findings confirm that intermodal logistic integration, hub-and-spoke distribution system and supply chain security are important key components of the modern international transportation system that determine global-scale spatial organization, shipping flow and freight mobility.
To stay competitive within the gas turbine community, turbine aero designers strive to maximize the total work output of each turbine stage through a combination of airfoil design improvements and increased total pressure ratio. Although increasing the mass flow rate could achieve a higher power target, the resultant increase in turbine annulus would result in structural limitations due to longer blades which cause increased strain on the blade root as well as amplified flutter and rotor dynamic excitation. An alternative path to achieving higher power output is to maximize the loading of each turbine stage through increased pressure ratio, but this may lead to airfoil limit loading and high aerodynamic losses.
This research systematically develops a detailed methodology to simulate the prediction of airfoil limit loading as well as provides a thorough investigation into the factors that influence the limit loading condition. A computational baseline was established using data previously collected at the Pratt & Whitney Canada High-Speed Wind Tunnel at Carleton University near design conditions using the Reynolds-Averaged Naiver-Stokes shear stress transport k-ω turbulence model (SST) with γ transition. An adaptive mesh refinement algorithm was developed based on the normalized local cell gradients of total pressure, total temperature, density, turbulent kinetic energy, turbulent eddy viscosity and the specific dissipation rate of turbulence. An overall reduction in computational cost was determined as 50% per simulation. The SST turbulence model with Gamma transition was found to have superior predictive veracity compared to other eddy viscosity turbulence models for the limit loading condition.
Variation of turbine inflow conditions were analyzed for four different transonic turbine airfoils based on the potential flow conditions exhausted by an upstream combustor. Influence of inflow conditions was found to be minimal on the exit flow profile with the exception of the mass-flow averaged total pressure loss coefficients. Results show incidence variation to change the total pressure loss coefficient differently for each airfoil, whereas turbulence intensity and turbulent length scale predicted a drastic rise in loss with increased turbulence level for all airfoils considered. The geometric characteristics of each airfoil were also investigated for influence on the stages to limit loading. Similar to previous experimental work, the limit loading pressure ratio and the mass-flow averaged outlet flow angle were strongly correlated with the airfoil outlet metal angle. It was also determined that the airfoil stagger and trailing edge blockage ratio play a role in the determination of the sublimit loading range, although no definitive parameter could be isolated due to lack of specific geometric constraints.
Lastly, the effect of transient vortex shedding on the nature of the trailing edge shock system and subsequent influence on the stages towards limit loading were investigated. A detailed review of the boundary layer states at the trailing edge were performed showing that all of the modeling approaches predicted laminar boundary layer profiles along the pressure surface trailing edge and turbulent profiles along the suction surface. Each modeling strategy (unsteady Reynolds-Averaged Navier Stokes, Delayed Detached Eddy Simulation and turbulence model free) predicted separation along the suction surface during limit loading due to acoustic wave propagation caused by the shock-base pressure interaction, although with varying degrees of size and magnitude. Temporal evolution of the mass flow averaged total pressure loss coefficient downstream of the airfoil allowed for the dominant vortex shedding frequency to be determined and subsequent Strouhal number to be calculated. It was found that each transient modeling strategy predicted the vortex frequency differently. A formal documentation and review were made outlining the required simulation time step to achieve accurate temporal resolution as well as approximate vortex shedding period. Qualitative images of numerical Schlieren (normalized density gradient) contours were presented and reviewed showing large differences in the prediction of vortex shape, size, and subsequent shock influence. Although conclusions were made on modeling ability, without extensive experimental documentation no concrete justification can be made at this time, outlining the importance of an experimental investigation.
The development of in-memory computing has fueled the emergence of in-memory computing systems. Data explosion is also posing an unprecedented demand for memory capacity to handle the ever-growing data size. Thus, in-memory computing systems are increasingly looking inward at hybrid memory caches of under-processed data as resources to be mined. Our preliminary study finds that some existing data management strategies often trade application performance for low memory utilization, and hence can induce frequent I/O operations between memory system and storage system.
To achieve this goal, we propose to design a hybrid memory system that includes fast and relatively slow memory hardware. In order to realize a runtime system that automatically optimizes data management on hybrid memory, we will (1) propose a new shared in-memory cache layer among parallel executors that are co-hosted on the same computing node, which aims to improve the overall hit rate of data blocks; (2) develop a middleware layer built on top of existing deep learning frameworks that streamlines the support and implementation of online learning applications; (3) design a unified in-memory computing architecture with efficient data management strategy to optimize memory allocation and recycle for ML applications.
Advancement in terahertz technologies have drawn interests in optical components suitable for the terahertz spectral range. Stereolithography, with its superior resolution, could be an efficient way of fabricating such terahertz elements with sub-wavelength scale architectures. However, stereolithographically fabricated terahertz optical elements or metamaterials have not yet been studied extensively. In this thesis, we sought to explore the terahertz optical properties of stereolithographically fabricated optics and novel metamaterials. Terahertz optical properties of materials commonly available for stereolithography have been accurately determined. Utilizing the determined properties, one-dimensional terahertz photonic crystals and defect modes within such crystals have been demonstrated for the first time through a single-step stereolithography from a single dielectric material. Mechanical tunability of the photonic bandgap and defect modes of the photonic crystals was experimentally realized. In addition, stereolithographically fabricated anisotropic metamaterial composed of slanted columnar structures have been investigated as a single layer, as well as constituent layers of one-dimensional photonic crystal structures for the first time. Off-axis parabolic reflectors have been demonstrated by metalizing a stereolithographically fabricated polymer base and by employing one-dimensional photonic crystal structure into design. In conclusion, stereolithography has been introduced as a new paradigm for fabrication of custom terahertz elements and novel metamaterials with tailored optical properties.
Climate change has resulted in warming of coastal aquatic habitats around the world at almost every latitude, threatening ecosystems with a significant loss in biodiversity and occurring at a rate that may exceed species’ ability to adapt. Understanding how reef species survive in habitats that currently experience extreme temperatures will help identify the biological processes that will govern future responses to climate change. The Persian/Arabian Gulf experiences the warmest coral reef temperatures on the planet (summer maxima ~35-36°C but can exceed 37°C) and connects to the neighboring Gulf of Oman, which experiences more benign environmental conditions (summer maxima of ~30-32°C). Here, we leverage this unique environmental gradient as a natural laboratory to better understand how the keystone sea urchin Echinometra sp. EZ copes with thermal extremes. Species survival in extreme habitats is dependent on their ability to acclimate over the course of an organisms’ lifetime and adapt over the course of many generations. Two complementary mechanisms for coping with environmental change are shifts in the host-associated microbial community, which can happen on a timescale of hours to days, and classic Darwinian evolution in which selection results in different patterns of alleles between populations over many generations. Here, we identify temperature as a robust predictor of community-level microbial variation and highlight specific bacterial taxa that may be crucial for maintenance of host homeostasis during thermal extremes. Next, we show that while there is a high degree of genetic admixture between all sites and bidirectional gene flow between the two seas, there is also significant population differentiation. We describe nine candidate loci that are under putative selection, including one collagen gene. Finally, we sequence, assemble, and annotate a chromosome-level genome that will add substantial value to future functional genomic datasets. Together, the research composing my dissertation has identified the importance of novel microbiome and genomic variation in the adaptation of a dominant ecosystem engineer to the warmest marine environment on Earth. These integrative results provide a foothold for understanding shared and unique mechanisms for the adaptation of marine species to historic and ongoing climate change.
Background: Current popular conceptualizations of the psychological process Repetitive Thought (RT) appear of limited accuracy due to ample construct proliferation (e.g. equating RT with rumination or worry), tautological definitions, and the construct being studied primarily in mentally disordered populations. This paper sought to unite current disparate lines of research surrounding RT, in order to illuminate and clarify the nature of RT.
Methods: Two studies were completed: First, a systematic literature review was conducted in order to develop a more comprehensive and conceptually coherent model of RT. Second, the structural validity of the model produced by the first study was empirically tested using factor analytic and multiple regression techniques.
Results: Partially Exploratory Factor Analyses revealed a strong general Repetitive Thinking factor, as well as a three-factor model that was empirically most appropriate (Intrusive Repetitive Thought, Deliberate Processing, and Self-Conscious Repetitive Thought). Additional validation analyses confirmed these findings.
Conclusions: This study contributes to our understanding of the nature of Repetitive Thought. Importantly, the three RT factors can be conceptualized as independent dimensions that are all part of a larger RT trait. The empirical and applied implications of the conceptualization of RT, as well as development of a preliminary measure of RT, are discussed.
Ocean Current Turbines (OCTs), which function similarly to wind and tidal turbines, represent a promising technology for harnessing the energy from oceanic currents such as the Gulf Stream. In planning the deployment of arrays of OCT devices, it is critical to consider the two-way interactions between the turbines and the ocean environment: temporally and spatially nonuniform flow fields are expected in the dynamic flow environments of western boundary currents, and include the presence of upstream shear and turbulence. These nonuniform flow conditions will affect power extraction, and the efficiency of the turbines when operating in isolation or as part of an array. Furthermore, models that are used in a predictive capability to compute the levelized cost of energy obtainable from such devices, or to optimize the layout of an array of turbines must be modified to account for the effects of such spatially and temporally inhomogeneous conditions. Similarly, the operation of OCT arrays can in turn influence the background flow in two significant ways, namely by contributing to the production of turbulence and through the generation of internal gravity waves that are radiated away from the point of origin. In this thesis, we have studied using detailed numerical simulations, the above two-way interaction between arrays of OCTs and the ocean environment. Insights developed from the simulations have guided the development of low-order wake interaction models capable of describing the effects of inhomogeneous flow conditions on array performance.
A new, wake interaction modeling framework capable of capturing the detailed effects of turbulence and upstream shear on various performance parameters associated with OCTs arranged in any arbitrary configuration has been developed. The model accounts for the effects of turbulence and shear on the structure of the turbine wakes, specifically the extents of near- and far-wake regions. The analytical description for turbine wake is combined with an existing wake interaction model, the Unrestricted Wind Farm Layout Optimization model to predict the global power output from an array of OCTs. The resulting modelling framework accurately captures the effect of inlet turbulence and shear on the OCT farm power and efficiency, and can be applied to any array configuration. Results from the model were validated against both Large Eddy Simulations and Reynolds Averaged Navier-Stokes simulations, in which the OCTs were modeled using a Blade Element Momentum model. The dispersion of OCT wake turbulence through the background stratification of the ocean was investigated using Large Eddy Simulations for different levels of the density stratification. The effects of varying the strength of the stratification as well as the turbulent forcing were studied. Finally, the wake turbulence associated with OCT operation can drive the formation and radiation of internal gravity waves in the density-stratified background flow of ocean currents. Through detailed numerical simulations, the effect of the propagation of the internal waves on the background turbulent diffusivity was studied, and found to alter the transport properties of the ambient flow. The properties of the internal wave field, and its impact on background turbulent mixing was found to depend both on the Richardson number and the ambient, upstream turbulence.
Careful control of intracellular signaling pathways plays an important role in a cell’s ability to maintain stable internal conditions in the face of an ever-changing extracellular environment. This is particularly true as it relates to the process of cellular self-eating or autophagy. Macroautophagy (herein referred to as autophagy) is a catabolic process by which unneeded or damaged cellular components are sequestered as cargo into unique double-membrane vesicles called autophagosomes which fuse to the vacuole (yeast lysosome) to be metabolized. The autophagy-related (Atg) proteins that mediate and regulate the process are evolutionarily conserved across all autophagy pathways, including starvation-induced bulk autophagy and cargo-selective autophagy pathways. The central theme of this thesis is to understand how autophagy is affected by lipids and regulatory proteins in yeast. In this thesis, we have summarized the field’s understanding of lipid homeostasis and trafficking during autophagy and autophagosome formation. Furthermore, we have extended this knowledge by discovering a clear interplay between autophagy and the SNX-BAR protein subfamily. In recent years, the SNX-BARs have been reported to have emerging roles in autophagy, however, such mechanisms of action have been primarily indirect. In this thesis, we have characterized a novel SNX-BAR protein, we have termed Vps501 and have found it directly affects autophagy which brings to light a new role of SNX-BAR proteins in autophagy regulation.
