We use cosmogenic 10Be analysis of fluvial sediments and bedrock to estimate erosion rates (103 – 106 year timescale) and to infer the distribution of post-orogenic geomorphic processes in the Blue Ridge Province in and around Shenandoah National Park, VA. Our sampling plan was designed to investigate relationships between erosion rate, lithology, slope, and basin area. Fifty-nine samples were collected from a variety of basin sizes (<1 – 3351 km2) and average basin slopes (7 - 26°) in each of four different lithologies that crop out in the Park: granite, metabasalt, quartzite, and siliciclastic rocks. The samples include bedrock (n = 5), fluvial sediment from single-lithology basins (n = 43), and fluvial sediment from multilithology basins (n = 11): two of these samples are from rivers draining streams exiting the eastern and western slopes of the Park (Rappahannock and Shenandoah Rivers). Inferred erosion rates for all lithologies for fluvial samples range from 3.8 to 24 m/My. The mean erosion rate for single-lithology basins in the Park is 11.6 ± 4.8 m/My. Singlelithology erosion rates ranges for fluvial samples are: granite (basin size = ~0.4-40 km2 and slope = 11-23°), 7.9–22 m/My; metabasalt (basin size = ~1-25 km2 and slope = 11-19°), 4.8–24 m/My; quartzite (basin size = ~0.1-9 km2 and slope = 12-23°), 4.7–17 m/My; and siliciclastic rocks (basin size = ~0.3-13 km2 and slope = 18-26°), 6.2–17 m/My. The mean erosion rate for multilithology basins (basin size = ~1-3351 km2 and slope = 7-22°) is 10.2 m/My, and individually for the Shenandoah River 7.3 m/My and the Rappahannock River 13.8 m/My. Bedrock erosion rates range from 2.4-13 m/My across all lithologies, with a mean erosion rate of 7.9 ± 5.0 m/My. Grain-size specific 10Be analysis of four samples showed no consistent trend of concentration with grain size. These data support Hack’s dynamic equilibrium model. Slope and erosion rate are not well correlated, and mean erosion rates are similar for different lithologies. Cosmogenicallydetermined erosion rates in Shenandoah Park are similar to or lower than those reported elsewhere in the Appalachians including those of Matmon and others (2003), 25 to 30 m/My for metaclastic rocks in the steep Great Smoky Mountains, Reuter and others (2004), 4 – 54 m/My in Susquehanna River basin for shale, sandstone, and schist, and Sullivan and others (2006), 6-38 m/My in the micaceous schist and gneiss of the Blue Ridge Escarpment. Cosmogenic erosion rates (integration over 104 yrs) in the Blue Ridge province of Shenandoah National Park are consistent with long-term unroofing rates (integration over 107 yrs) estimated from U-Th/He measurements (11-18 m/My) in samples collected near the Blue Ridge Escarpment by Spotila and others (2004), and fission tracks (20 m/My) in the Appalachians by Naeser and others (2005). The consistency of denudation rates integrated over very different periods of time suggests steady erosion most likely in balance with, and driving isostatic uplift of rock.

Variations in species occurrence and distribution across the landscape over time provide fundamental information concerning population dynamics. How this relates to habitat characteristics at multiple scales can elucidate the process of habitat selection. I evaluated these processes for a montane fir (Abies) forest specialist, Bicknell’s Thrush (Catharus bicknelli) in Vermont. This species is threatened by a suite of anthropogenic disturbances on its breeding grounds and quantifying the effects of environmental change at the population level for this songbird has not been addressed. The naturally fragmented breeding habitat of varying size, quality, and connectivity warranted a metapopulation approach and a robust occupancy analysis. Detection/non-detection data was collected for Bicknell’s Thrush across 88 sites during the breeding seasons in 2006 and 2007. Local habitat characteristics were measured for each site and landscape-level features were calculated using a predictive habitat model. The six local habitat variables were combined using a principal component analysis. Principal component 1 (PC1) described a gradient of increasing coniferous shrub density and proportion of coniferous dominated forest with decreasing canopy height. The landscape covariate was calculated by dividing patch size by patch isolation creating a continuum of small, isolated patches to large, less isolated patches. Thus each site was characterized by a single local habitat (PC1) and landscape metric. From these data, 67 models considering all combinations of landscape and local habitat scores (univariate, additive and interaction) were evaluated for individual estimates of the following parameters: (1) probability of detection, (2) probability of initial site occupancy, (3) probability of site colonization, and (4) probability of local site extinction. AIC model selection techniques were used to rank the models, which represented ecologically plausible hypotheses that compared the strength of local habitat characteristics to large-scale landscape features. Models within 4 AICc points of the top model were considered plausible. The top eight models were all plausible. Landscape characteristics alone were not significant in driving population dynamics. The relative importance of landscape + local habitat was highest for both probability of initial occupancy and local site extinction. Probability of occupancy increased and extinction decreased with the combination of increased patch size and decreased patch isolation (landscape) and increased coniferous shrub density, proportion of coniferous dominated forest and decreased canopy height (local habitat). Probability of site colonization was driven mainly by local habitat features and increased with increasing habitat quality. These results indicate a complex system with intricate links between landscape and local scales. Preserving large tracts of habitat may not be sufficient in assuring future species persistence, but could minimize local extinction risk. Careful consideration should be given to local habitat features within habitat fragments, particularly to maintain adequate colonization rates. Because important features from both scales are correlated, in intact montane forest patches, landscape-scale attributes alone may serve as a surrogate for identifying quality breeding habitat, assuming processes of natural disturbance can be maintained.

Large scale machine learning has many characteristics that can be exploited in the system designs to improve its efficiency. This dissertation demonstrates that the characteristics of the ML computations can be exploited in the design and implementation of parameter server systems, to greatly improve the efficiency by an order of magnitude or more. We support this thesis statement with three case study systems, IterStore, GeePS, and MLtuner. IterStore is an optimized parameter server system design that exploits the repeated data access pattern characteristic of ML computations. The designed optimizations allow IterStore to reduce the total run time of our ML benchmarks by up to 50×. GeePS is a parameter server that is specialized for deep learning on distributed GPUs. By exploiting the layer-by-layer data access and computation pattern of deep learning, GeePS provides almost linear scalability from single-machine baselines (13× more training throughput with 16 machines), and also supports neural networks that do not fit in GPU memory. MLtuner is a system for automatically tuning the training tunables of ML tasks. It exploits the characteristic that the best tunable settings can often be decided quickly with just a short trial time. By making use of optimization-guided online trial-and-error, MLtuner can robustly find and re-tune tunable settings for a variety of machine learning applications, including image classification, video classification, and matrix factorization, and is over an order of magnitude faster than traditional hyperparameter tuning approaches.

Research in the field of divorce has primarily focused on women and children, and there is a definite lack of literature that addresses divorce from a male perspective. This lack of knowledge has influenced the clinical treatment and general social support available to men who get divorced. This study aims to explore which, if any, of the personality factors in the 16PF are correlated with adjustment to divorce using the Fisher Divorce Adjustment scale in South African men. The divorce-stressadjustment perspective was used to conceptualise adjustment to divorce and the factors that influence adjustment to divorce (Amato, 2000). The sample consisted of South African men who were divorced. A sample size of 40 participants was obtained using convenient sampling. The results of the study indicated that the personality factors ‘Anxiety’ and ‘Emotional Sensitivity’ are significantly and negatively correlated to adjustment to divorce. This indicates that aspects of personality that relate to a robustness of ego, low anxiety and a self-confident individual who is emotionally mature, is more likely to adjust significantly better to divorce and thus contribute significantly to the well-being of an individual post divorce.

This dissertation concerns the the problem that natural science teachers with limited astronomy backgrounds have to teach new curriculum content about lunar motion, phases and eclipses. My study aims to establish: 60 teachers’ knowledge of lunar phenomena through surveys; whether an intervention incorporating models and activities is effective at improving a case study group’s understandings; how the case study teachers use these activities and models in the classroom. My results indicate that the majority of natural science teachers have little formal astronomy education. Only two teachers held a scientific understanding of lunar phenomena. The intervention led towards a more scientific understanding amongst the case study group. Scale is essential for developing an understanding of lunar phenomena and models are extremely beneficial, but participants experience spatial problems when viewing models from an external perspective. I propose in-service training in small groups for building knowledge and increasing confidence for teaching this content.

Student Number : 0100281N - MSc(Eng) Research Report - School of Civil and Environmental Engineering - Faculty of Engineering and the Built Environment / Many environmental and engineering projects require prediction of the velocity of flow in river channels, in terms of those channel properties and flow characteristics which induce resisting forces or an energy loss to the flow. Relationships such as the Manning, Chézy and Darcy-Weisbach equations have been in use for a century or more. All of them account for resistance with a single coefficient of resistance, and the central problem is evaluation of this coefficient. Experimental results by different researchers have shown that Manning’s n varies strongly with the ratio of flow depth to roughness height. It is constant for values of this ratio above about 4, but increases significantly for lower values. This suggests that the equation is not suitable in its original form for the case of intermediate-scale roughness. The roughness is intermediate-scale if the relative submergence ratio of flow depth to roughness elements height lies between 1 and 4. The influence of the roughness elements on flow resistance in this regime is caused by a combination of both element drag and boundary shear, or friction. The results of an experimental study with hemispherical roughness elements are presented, showing how the roughness element size, spacing and pattern influence flow resistance. For the range of conditions tested, Manning’s n appears to depend on roughness element size, spacing and pattern.

A presente dissertação de mestrado traz como foco a questão da escala em obras de arte, arquitetura e design. Está estruturada em três partes principais. Na primeira parte são apresentadas abordagens teóricas sobre o conceito de escala, percorrendo estudos sobre seus principais significados e aplicações nestes campos do conhecimento, a relação do corpo humano como parâmetro dimensional e os conceitos de massa, volume e espaço, além de abordar a teoria do estranhamento, idealizada por Chklóvski e o conflito entre conhecimento e experiência, proposto por Maderuelo. Na segunda parte, são expostas obras de arte, arquitetura e design que evidenciam algumas das possibilidades fornecidas pela manipulação da escala. Na terceira e última parte foram destacadas três obras de arte cuja forma é um cubo, sendo elas; Die (1962), de Tony Smith; a obra sem título (quatro cubos espelhados, 1965), de Robert Morris e Cruzeiro do Sul (1969-1970), de Cildo Meireles. A partir desta seleção sugeriu-se a sistematização das ideias apresentadas anteriormente como ferramentas de análise e reflexão aplicáveis a processos criativos. / This dissertation brings focus on the issue of scale in works of art, architecture and design. Is structured into three main parts. In the first part theoretical approaches to the concept of scale are presented, covering studies on its main meanings and applications in these fields of knowledge, the relation of the human body as a dimensional parameter and the concepts of mass, volume and space, in addition, presents the theory of estrangement, idealized by Chklóvski and the conflict between knowledge and experience, proposed by Maderuelo. In the second part are exhibited works of art, architecture and design which highlight some of the possibilities provided by the manipulation of scale. In the third and final part, three works of art whose form is a cube are shown, they are Die (1962), of Tony Smith, the Untitled work (Four mirrored cubes, 1965) of Robert Morris and Cruzeiro do Sul (1969- 1970), of Cildo Meireles, and from this selection suggest the systematization of ideas previously presented as tools for analysis and reflection applicable in creative processes.

by Ho Kwong-shing Lawrence. / Thesis (M.B.A.)--Chinese University of Hong Kong, 1994. / Includes bibliographical references (leaves 43-44). / ABSTRACT --- p.ii / TABLE OF CONTENTS --- p.iii / ACKNOWLEDGEMENT --- p.iv / CHAPTER / Chapter I. --- INTRODUCTION AND LITERATURE REVIEW --- p.1 / Chapter II. --- COMBINING SALES FORECASTS --- p.7 / Chapter III. --- EXPERIMENTAL DESIGN --- p.14 / Chapter IV. --- SIMULATION RESULTS --- p.19 / Chapter V. --- SUMMARY AND CONCLUSION --- p.27 / APPENDIX --- p.31 / BIBLIOGRAPHY --- p.43

by Li Wai Ting. / Thesis (M.Ph.)--Chinese University of Hong Kong, 1987. / Includes bibliographical references.

Large-scale distributed systems and distributed computing are the pillars of IT infrastructure and society nowadays. Robust theoretical principles for designing, building, managing and understanding the interactive behaviours of such systems need to be explored. A promising approach for establishing such principles is to view the session as the key unit for design, execution and verification. Governance is a general term for verifying whether activities meet the specified requirements and for enforcing safe behaviours among processes. This thesis, based on the asynchronous -calculus and the theory of session types, provides a monitoring framework and a theory for validating specifications, verifying mutual behaviours during runtime, and taking actions when noncompliant behaviours are detected. We explore properties and principles for governing large-scale distributed systems, in which autonomous and heterogeneous system components interact with each other in the network to accomplish application goals. This thesis, incorporating lessons from my participation in a substantial practical project, the Ocean Observatories Initiative (OOI), proposes an asynchronous monitoring framework and the process calculus for dynamically governing the asynchronous interactions among distributed multiple applications. We prove that this monitoring model guarantees the satisfaction of global assertions, and state and prove theorems of local and global safety, transparency, and session fidelity. We also study and introduce the semantic mechanisms for runtime session-based governance and the principles of validation of stateful specifications through capturing the runtime asynchronous interactions.