Thesis (Ph.D.)--University of Delaware, 2009. / Principal faculty advisor: Patrick M. Gaffney, College of Earth, Ocean, & Environment. Includes bibliographical references.

While transistor size continues to shrink every technology generation increasing the amount of transistors on a die, the reduction in energy consumption is less significant. Furthermore, newer technologies induce fabrication challenges resulting in uncertainties in transistor and wire properties. Therefore to ensure correctness, design margins are introduced resulting in significantly sub-optimal energy efficiency. While increasing parallelism and the use of gating methods contribute to energy consumption reduction, ultimately, more radical changes to the architecture and better integration of architectural and circuit techniques will be necessary. This dissertation explores one such approach, combining a highly-efficient massively-parallel processor architecture with a design methodology that reduces energy by trimming design margins. Using a massively-parallel GPU-like (graphics processing unit) base- line architecture, we discuss the different components of process variation and design microarchitectural approaches supporting efficient margins reduction. We evaluate our design using a cycle-based GPU simulator, describe the conditions where efficiency improvements can be obtained, and explore the benefits of decoupling across a wide range of parameters. We architect a test-chip that was fabricated and show these mechanisms to work. We also discuss why previously developed related approaches fall short when process variation is very large, such as in low-voltage operation or as expected for future VLSI technology. We therefore develop and evaluate a new approach specifically for high-variation scenarios. To summarize, in this work, we address the emerging challenges of modern massively parallel architectures including energy efficient, reliable operation and high process variation. We believe that the results of this work are essential for breaking through the energy wall, continuing to improve the efficiency of future generations of the massively parallel architectures. / text

The field of evolutionary developmental biology (evo-devo) focuses on understanding the evolution of ontogeny and mechanisms of evolutionary change. Recently, taxonomic comparisons of the sequence of skeletal ossification have become prominent in evo-devo. However, most of these studies fail to consider two major issues: how the technique used to assay ossification and ontogenetic variation and variability may affect comparisons among taxa. This study focuses on the onset of ossification in the skeleton of Monodelphis domestica and quantifies the affects of variation, variability, and technique bias on reconstructions of ontogeny. Previous comparisons among mammalian taxa have used both computed tomography (CT) and clearing-and-staining (CS) to assess the presence or absence of skeletal elements (i.e., skeletal maturity). In this study, CT and CS were used on the same specimen to compare how these methods assess skeletal maturity. The comparisons of the same individual under reveal significant differences in how skeletal maturity is assessed by CT and CS techniques. Further, significant biases were recovered between techniques. CT is more likely to reveal cranial elements that CS does not, whereas CS is more likely to reveal appendicular elements that CT does not. To assess levels of variation and variability, Ontogenetic Sequence Analysis (OSA) was used to characterize the ontogeny of Monodelphis domestica. This revealed significant levels of variation with over 800 different ontogenetic pathways recovered for the onset of ossification of all skeletal elements studied. Additionally, high levels of variability were also reconstructed because the majority of specimens were found to exhibit non-modal ontogenetic sequences. This variability is more highly concentrated in the sequence of cranial ossification, suggesting potential modularity in ontogenetic variation and variability. Finally, OSA revealed that technique bias could importantly affect reconstructions of skeletal ossification sequences because no identical sequences were recovered by the CT and CS datasets. The results of this study demonstrate the importance of considering the primary nature of developmental studies, the specimen. Only by recognizing and quantifying the complexities of evo-devo research, especially natural variation and methodological biases, can more complete understandings of the evolution of ontogeny be had. / text

Individual specialization occurs when individuals selectively consume a subset of their population's diet. Intraspecific diet variation can stabilize population and community dynamics, promote species coexistence, and increase ecosystem productivity. Ecological variation also provides the variability necessary for natural or sexual selection to act. Individual threespine stickleback select different prey from a shared environment, and this variation is not simply a result of sex, size, or spatial heterogeneity. I use longitudinal observation of stickleback foraging microhabitat to support more commonly used cross-sectional metrics. Among recaptured individuals there were correlations between microhabitat use and functional morphology, and microhabitat use and long term dietary differences between individuals. I quantify individual specialization across populations using cross-sectional sampling to understand how and why ecological variation may itself be variable. All populations showed significant individual specialization. Specialization varied between populations and this variation seems to be a long-term property of populations. Overall morphological variance was positively correlated with ecological variation. Ecological variation, like all types of heritable variation, provides raw material for evolutionary change. For example, lacustrine populations of stickleback are commonly under disruptive selection due to intraspecific competition for prey resources. Speciation with gene flow may be driven by a combination of positive assortative mating and disruptive selection, particularly if selection and assortative mating act on the same trait. We present evidence that stickleback exhibit assortative mating by diet, using the isotopes of males and eggs within their nests. In concert with disruptive selection, this assortative mating should facilitate divergence. However, the population remains phenotypically unimodal, highlighting the fact that assortative mating and disruptive selection do not guarantee evolutionary divergence and speciation. There are several not-mutually-exclusive mechanisms by which assortative mating by diet may occur in these populations, such as shared microhabitat preference among individuals of similar diet. Stable isotopes reveal diet differences between different nesting areas and among individuals using different nest habitat within a nesting area. Spatial segregation of diet types may generate some assortative mating, but is insufficient to explain the observed assortment strength. We therefore conclude that sticklebacks' diet-assortative mating arises primarily from behavioral preference rather than from spatial isolation. / text

Undersökningar visar att elever har svårt att ta till sig evolutionsteorin och i stället väljer att förklara evolutionen med icke vetenskapliga argument. Till exempel väljer de att förklara fjällharens vita vinterpäls med att en vit päls gör att den syns sämre. Det här är ett exempel på att eleverna gärna förklarar evolutionen som behovsstyrd. En laboration utformades för att illustrera hur variation och urval ger evolution i en plantpopulation. Laborationen går ut på att eleverna under en tid studerar en population med plantor som utsätts för torka. Plantorna har olika simulerade genotyper och kommer därför att reagera olika på torkstressen. Tanken med laborationen var att få elever att använda sig av mer vetenskapliga förklaringar till evolutionen istället för icke veteskapliga. Resultaten visar att eleverna ändrat uppfattning, men inte riktigt till en mer vetenskaplig förklaring utan de har gått från sin tidigare förklaringsmodell till att inte riktigt kunna förklara vad som händer vid evolution.

The works contained in this dissertation were motivated by a desire to better understand the variability of patient performance on videofluoroscopic assessments of swallowing. Specifically, the variation present in kinematic and temporal measures of swallowing was investigated in three main phases: narrative literature review, healthy swallowing, and disordered swallowing. The primary goals were to identify which factors explain (or do not explain) variation, to develop methods to control for variation and to investigate the association between swallowing physiology and swallowing impairment. The literature reviews revealed wide ranges of variation for kinematic (Chapter 2) and temporal (Chapter 3) measures of swallowing in the existing literature on healthy deglutition. The kinematics (Chapter 4) and timing (Chapter 5) of swallowing were investigated in a prospectively collected sample of young healthy participants stratified by height. One main objective was to investigate the impact of participant size on physiological parameters of swallowing. Finally, kinematic and temporal measures of swallowing were investigated in a sample of patients referred for swallowing assessment (Chapter 6) to explore associations between swallowing physiology and impairment. The findings of this dissertation make several unique contributions to the dysphagia literature. It has demonstrated that inherent variation appears to exist in physiological measures of both healthy and disordered swallowing. Further, when certain sources of variation are controlled (such as participant size), men and women do not demonstrate significant differences for any of the parameters tested. Importantly, this work has demonstrated that participant size impacts the expected extent of hyoid excursion and that this variation can be controlled through normalization of hyoid movement to internal anatomical scalars. Two temporal variables trended toward detecting functional swallowing impairment. Finally, this dissertation provides the first set of normative reference values for parameters of swallowing with an ultra-thin liquid barium. Limitations are acknowledged and future work is suggested.

Understanding the factors and processes that influence intraspecific genetic variation are essential to better understand evolutionary processes. In this research, I examined patterns of gene flow and their effects on the distribution of genetic variation and spatial genetic structuring at different spatial scales. I used a combination of population genetics, spatial analysis, morphometrics and phylogeography in order to understand the patterns of genetic variation and their resulting phenotypic variations in a semi-aquatic species, the muskrat (Ondatra zibethicus).

The ATP-binding cassette transporter, sub-family A, member 1 (ABCA1) mediates the major pathway for cholesterol exit from non-hepatic cells and thereby controls the rate-limiting step in the biogenesis of high density lipoprotein (HDL) particles. In humans,ABCA1 deficiency results in Tangier disease, characterized by low levels of HDL cholesterol, cellular cholesterol accumulation and increased risk for atherosclerosis. More than 100 coding variants have been described in the ABCA1 gene. We attempted to understand how both naturally occurring and engineered mutations in ABCA1 impact its role in cholesterol transport in a variety of in vitro and in vivo systems. We attempted to correlate specific genetic variants in ABCA 1 with phenotypes in patients carrying the sevariants, and used an evolutionary approach to predict which specific variants in ABCA1would impact its function. We then turned to the study of tissue-specific genetic deletion of ABCA1 in mice to study its role in HDL biogenesis, atherosclerosis and glucose metabolism. We found that intestinal ABCA1 is an important site of HDL biogenesis and that activation of intestinal ABCA1 raises HDL levels in vivo. Hepatic ABCA1, which is a major site of HDL biogenesis, was shown to significantly contribute to susceptibility to atherosclerosis. Finally, we show that ABCA1 plays an unsuspected role in B-cell function and insulin secretion. These studies have contributed to our understanding of the impact of genetic variation in ABCA1 on diverse biological and pathological processes, and have identified novel aspects of ABCA 1 function in specific cell types.

A long term study of fish populations on the Bonaparte Plateau, B.C., has revealed the possibility of 2-year cyclic recruitment variation in the pygmy pikeminnow (Ptychocheilus oregonensis). This thesis reviews possible causes of the cycle, and evaluates an inexpensive and non-destructive visual survey method to estimate juvenile population numbers over short summer periods while pikeminnow juveniles are recruiting to the lakes and dying rapidly, for use in future studies of the cyclic pattern. To provide an index of the abundance of juvenile pikeminnow, strip-transect surveys were conducted from shore. Each strip transect area was a full circumference of each of three study lakes, and extended from shore to a distance of up to 3 meters offshore. The visual surveys were shown to be accurate and repeatable when estimating the number of fish and the lengths of the fish in a shoal. The strip transects provide an index of abundance for juvenile northern pikeminnow and this index of abundance can be used to calculate daily summer and interannual mortality rates, providing estimates close to those predicted from the Lorenzen model for size-dependent mortality rates of fish in general. Stationary point surveys were also conducted to investigate juvenile pikeminnow behaviour, and these point surveys reveal a link between sunlight intensity and pikeminnow activity, and provide evidence of direct agonistic interactions between age 0 and age 1 juveniles, which could result in higher mortality rates of age 0 fish when age 1 fish are abundant. Analysis of survey data collected to date over five years from three lakes supports the possibility of a recruitment cycle with lakes out of phase with each other, suggesting that ecological interactions drive the cycles. These interactions are worthy of further investigation.

This thesis reports on a design and modeling of a micro-machined gyroscope. The proposed sensor is a dual mass type, electro-statically driven to primary mode oscillation and senses, capacitively, the output signal. Full decoupling between drive and sense modes minimizes the mechanical crosstalk and based on this a novel gyroscope is designed and modeled which has separate sensing and driving masses. The dual mass gyroscope is designed such that driving and sensing resonant frequency is 23101 Hz with 0% mismatch (in simulation)with quality factor of 31.6227 and bandwidth of 730.51Hz. The gyroscope when actuated in simulation with 25V ac and 10V dc showed sensing capacitance variation of 126aF for 1 rad/s with base capacitance of 244.16fF. To the design of the gyroscope a new semi automatic tool was formulated for the noise analysis and noise based optimization of the resonant MEMS structures. Design of a sensitive gyroscope needs to take into account the noise shaping induced by damping phenomena at micro scale and is critical for optimization. The analysis was further extended to the design of the gyroscope and estimation shows that there is a trade of between the S/N ratio and the sensitivity and the design could be made much better in-terms of S/N by tuning its resonant frequency to 10⁶Hz.