No description available.

No description available.

Iterative CT reconstruction algorithms are superior to the standard convolution backpropagation (CBP) methods when reconstructing from a small number of views (hence less radiation), but are computationally costly. To reduce the execution time, this work implements and tests a parallel approach to iterative algorithms using a cluster of workstations, which is a low cost system found in many offices and non-academic sites. A previous implementation showed little speedup because of the significant cost of inter-processor communication. In this thesis, several data partitioning methods are examined, including some image tiling methods that exploit the spatial locality demonstrated by local CT. Using these methods, computation can proceed locally, without the need for inter-processor communication during every iteration. A relative speedup of up to 17 times is obtained using 25 processors, demonstrating that good performance can be obtained running computationally intensive CT reconstruction algorithms on distributed memory hardware.

Iterative CT reconstruction algorithms are superior to the standard convolution backpropagation (CBP) methods when reconstructing from a small number of views (hence less radiation), but are computationally costly. To reduce the execution time, this work implements and tests a parallel approach to iterative algorithms using a cluster of workstations, which is a low cost system found in many offices and non-academic sites. A previous implementation showed little speedup because of the significant cost of inter-processor communication. In this thesis, several data partitioning methods are examined, including some image tiling methods that exploit the spatial locality demonstrated by local CT. Using these methods, computation can proceed locally, without the need for inter-processor communication during every iteration. A relative speedup of up to 17 times is obtained using 25 processors, demonstrating that good performance can be obtained running computationally intensive CT reconstruction algorithms on distributed memory hardware.

The objective of this thesis is to calibrate two oxygen barometers for kimberlite magmas in the system CaO-MgO-Al2O3-SiO2-TiO2-FeO based on the Fe and V content of monticellite, CaMgSiO4, that may be utilized in cases where oxides in olivine phenocrysts and perovskite are absent from a kimberlite pipe. I first calibrate a new oxygen barometer for kimberlite magmas based on the Fe content of monticellite in equilibrium with kimberlite liquids in experiments at 100kPa from 1230 to 1350C and at fO2 from NNO-4.1 to NNO+5.3 (where NNO is the nickel-nickel oxide buffer). The XFeMtc/XFeliq (where XFeMtc/XFeliq is the ratio of mole fraction of total Fe in monticellite and Fe in liquid) decreases with increasing fO2, consistent with only Fe2+ entering the monticellite structure. Although the XFe in monticellite varies with temperature and bulk composition, these dependencies are small (0.03) compared to that with fO2. The experimental data were fitted by weigted least square regression to the following relationship: DNNO= (log (0.858(0.021)*XFeliq/XFeMtc-1)-0.139(0.022))/0.193(0.004) (uncertainties at 2 sigma). I apply this oxygen barometer to natural kimberlite assuming the bulk rock FeO is that of their liquid FeO. Monticellite compositions of five kimberlites from both literature and my own investigations revealed a range in fO2 from NNO-3.5 to NNO+1.7. I finally use my well-defined monticellite-liquid Kd Fe2+-Mg to derive a range of Mg/(Mg+Fe2+) (Mg number) for kimberlite melts of 0.40-0.90. This range in composition is broader than previous estimates of 'primary' kimberlites, reflecting the diverse mantle sources and processes that occur during generation and ascent of kimberlites. Second, I calibrate a new oxygen barometer for kimberlite magmas based on the V content of monticellite in equilibrium with kimberlite liquids doped with 0.5 wt% V2O5 at 100kPa at 1280 and 1350C and at fO2 from NNO-4.1 to NNO_0.5. The DV Mtc/liq (DV Mtc/liq = V (ppm) in monticellite/V (ppm) in liquid) decreases with increasing fO2. The partitioning data can be fitted to a model consistent with V5+ as the dominant species in the melt phase above NNO whereas V4+ dominates below those conditions in kimberlitic magmas. The total DV Mtc/liq, which embodies both DV3+ Mtc/liq and DV4+ Mtc/liq, shows a very slight temperature and bulk composition dependence. The experimental data can be fitted by weighted least square regression to the following relationship: DNNO= (log(0.354(1.785)*Vliq/VMtc-1)-1.172(2.302))/0.111(0.071) (uncertainties at 2 sigma and V in ppm). In order to apply this oxygen barometer rigorously, the V concentrations of the kimberlite melt coexisting with monticellite need to be constrained. In contrast to the Fe-in-monticellite oxygen barometer for which the concentration of Fe in monticellite was close to that of the whole rock composition, the concentration of V in the bulk rock composition reflects mostly the large accumulation of olivine xenocrysts which contain low V concentrations. For that reason, the V-in-monticellite oxygen barometer cannot be applied to natural kimberlites until we find a way to overcome this problem. The vanadium concentrations of kimberlite melts are likely higher than the V concentrations of the whole rock compositions leading to underestimated fO2 values. / Graduate

Understanding the origins of biological diversity is a fundamental goal of evolutionary biology. A large body of theory attributes ecological and genetic diversification to divergent natural selection for resource specialization. This thesis examines adaptive radiation in response to selection for resource specialization in microcosm populations of the asexual bacterium Pseudomonas fluorescens. The general protocol for these experiments is to introduce a clonal population of Pseudomonas into a novel environment and to allow evolution to occur through the spontaneous appearance of novel genotypes carrying beneficial mutations. Adaptation can then be quantified through direct comparisons between evolved populations and their clonal ancestors. These experiments show that resource heterogeneity generates divergent natural selection for specialization on alternative resources, irrespective of the spatial structure of the environment. Adaptive radiation is possible in sympatry because of genetic trade-offs in the ability to exploit different resources, but these trade-offs are often not the result of antagonistic pleiotropy among loci that determine fitness on alternative resources. The rate of phenotypic diversification declines during adaptive radiation, apparently because the ecological opportunities required to support specialist lineages disappear as a consequence of initial diversification. The ultimate outcome of repeated instances of adaptive radiation is the evolution of a community of ecologically equivalent specialists that share similar adaptive traits, despite differences in the underlying genetic basis of specialization in replicate radiations. Comparisons with the literature on experimental evolution in microbial populations illustrate the results of this thesis are well-supported by experiments in a wide range of microbial microcosms.

The statistical properties of training, validation and test data play an important role in assuring optimal performance in artificial neural networks (ANN). Re-searchers have proposed randomized data partitioning (RDP) and stratified data partitioning (SDP) methods for partition of input data into training, vali-dation and test datasets. RDP methods based on genetic algorithm (GA) are computationally expensive as the random search space can be in the power of twenty or more for an average sized dataset. For SDP methods, clustering al-gorithms such as self organizing map (SOM) and fuzzy clustering (FC) are used to form strata. It is assumed that data points in any individual stratum are in close statistical agreement. Reported clustering algorithms are designed to form natural clusters. In the case of large multivariate datasets, some of these natural clusters can be big enough such that the furthest data vectors are statis-tically far away from the mean. Further, these algorithms are computationally expensive as well. Here a custom design clustering algorithm (CDCA) has been proposed to overcome these shortcomings. Comparisons have been made using three benchmark case studies, one each from classification, function ap-proximation and prediction domain respectively. The proposed CDCA data partitioning method was evaluated in comparison with SOM, FC and GA based data partitioning methods. It was found that the CDCA data partitioning method not only performed well but also reduced the average CPU time. / Engineering Management

Climate change can affect habitat availability and species interactions at several spatial and temporal scales. I explored niche partitioning and spatial variation of Rock (Lagopus muta) and White-tailed Ptarmigan (L. leucura) in southwest Yukon. I examined habitat selection of foraging areas within a population and patches within foraging areas in a sympatric population of Rock and White-tailed Ptarmigan. At the larger foraging area scale, Rock Ptarmigan used areas with greater shrub cover compared to White-tailed Ptarmigan. At the smaller patch scale, both species selected patches with greater rock cover, but differed in other patch features. Second, I examined spatial variation in abundance of both ptarmigan species between the Ruby and Kluane Ranges using pellet count and transect surveys. Relative abundance was lower in the Kluane Range based on pellet counts, but transect surveys proved inadequate as a measure of population density. The Kluane Range also had fewer positive degree days above 0 C and a greater mean standard deviation of NDVI, and was composed of finer textured colluvium compared to the Ruby Range, which could influence relative abundance of ptarmigan. / Ecology

In order to accelerate logic simulation, it is highly beneficial to simulate the circuit design on FPGA hardware. However, limited hardware resources on FPGAs prevent large designs from being implemented on a single FPGA. Hence there is a need to partition the design and simulate it on a multi-FPGA platform. In contrast to existing FPGA-based post-synthesis partitioning approaches which first completely flatten the circuit and then possibly perform bottom-up clustering, we perform a selective top-down flattening and thereby avoid the potential netlist blowup. This also allows us to preserve the design hierarchy to guide the partitioning and to make subsequent debugging easier. Our approach analyzes the hierarchical design and selectively flattens instances using two metrics based on slack. The resulting partially flattened netlist is converted to a hypergraph, partitioned using a public domain partitioner (hMetis), and reconverted back to a plurality of FPGA netlists, one for each FPGA of the FPGA-based accelerated logic simulation platform. We compare our approach with a partitioning approach that operates on a completely flattened netlist. Static timing analysis was performed for both approaches, and over 15 examples from the OpenCores project, our approach yields a 52% logic simulation speedup and about 0.74x runtime for the entire flow, compared to the completely flat approach. The entire tool chain of our approach is automated in an end-to-end flow from hierarchy extraction, selective flattening, partitioning, and netlist reconstruction. Compared to an existing method which also performs slack-based partitioning of a hierarchical netlist, we obtain a 35% simulation speedup.

This dissertation proposes a novel theoretical framework for the data partitioning problem in computer vision and machine learning. The framework is based on level set methods that are derived from variational calculus and involve a curve-based objective function which integrates both boundary and region based information in a generic form. The proposed approaches within the framework provide original solutions to two important problems in variational methods, namely parameter tuning and information fusion, collectively termed Learning Level Sets in this thesis. Moreover, a novel pattern classification algorithm, namely Level Learning Sets, is proposed to classify any general dataset, including sparse and non sparse data. It is based on the same optimisation process of the objective function directly related to the curve propagation theory used in level set theory. The proposed approach learns the knowledge required for parameter tuning and information fusion in level set methods using machine learning techniques. It uses acquired knowledge to automatically perform parameter tuning and information fusion in level set methods. In the case of pattern classification, variational methods using level set theory optimise decision boundary construction in feature space. Consequently, the optimised values of the objective level set function over the feature space represent the model for pattern classification. The proposed automatic parameter tuning and information fusion method embedded in the level set method framework has been employed to provide original solutions to image segmentation and object extraction in computer vision. On the other hand, the Level Learning Set has been extended and applied to a variety of pattern classification problems". Several experimental results for each of the above methods are provided, demonstrating the effectiveness of the proposed solutions and indicating the potential of the automatic and dynamic tuning and fusion approaches as well as the Level Learning Set model.