There are multiple ways of creating a modern mobile application and differentcombinations of programming languages and tools can be used to suite specificneeds.Typically, open market products need to support various platforms which, due tomultiple code bases, might lead to difficulties with maintenance and new featuredeployment.React Native is a JavaScript library, announced by Facebook in 2015, that provides auniversal pattern for creating mobile applications for Android and iOS. Theframework handles all data models and business logic using JavaScript whichenables large amounts of code to be shared between different operating systems.The cost of these advantages, however, is not well documented and it is possiblethat React Native may not be suitable in certain application scenarios.In this thesis, React Native is benchmarked in terms of performance and userexperience to evaluate which application scenarios that are suitable for theframework and which are not. Benchmarks show that React Native deliversanimation performance close to its native counterpart while computationallyintensive tasks took at least 27x more time to execute. The measurements alsoshows that React Native applications consumed 124% more battery and required25% more time to launch whilst also requiring significantly more memory.Based on the results, and evaluated differences in the development process, thisthesis has concluded that React Native is a viable option to native development inscenarios which does not rely on heavy background calculations, have strict limitson resource usage or need very specific Android features or components. Ideally,React Native is suitable for developers with web experience or scenarios whererapid deployment changes are desirable.

Energy saving techniques on the hardware level are not always capable of adapting to the applications. Thus energy saving techniques at software level are needed to overcome the limitations of the hardware. Dynamic Voltage Frequency Scaling (DVFS)is one of the most widely used techniques to scale down the voltage and frequency atruntime as needed. However, in modern processors the current leakage becomessignificant in low voltage (breakdown of Dennard’s scaling), leaving frequency scalingas the only solution. But frequency changes can take many CPU cycles, making DVFSnon applicable at instruction level granularity. Decoupled Access/Execute (DAE) is asolution to the above problem. DAE is an energy efficiency model with minimalperformance loss and has been evaluated on x86 platforms, showing significantimprovements in energy consumption, on task-based parallel programs and complexsequential applications. In this project we evaluate the DAE model on ARMv8 platforms using the SPEC CPU2006 benchmark suit. The results show energy benefitsand performance improvements up to 40%. Furthermore we discuss the behaviour ofDAE on x86 and ARMv8 architectures.

This study explored the attitudes and perceptions of faculty members in Rajamangala Institute of Technology (RIT), Thailand. The purpose of this study was to (a) determine which demographic variables are associated with faculty members attitudes toward the implementation of computers in higher education in Thailand, and (b) determine perceptions among faculty members toward barriers to the widespread use of computers.

Total collapses of buildings are a rare occurrence in modern building construction, and while they are scarcely seen, their consequences are often extensive enough to encourage building projectors to invest in a structural framework capable of sustaining its stability through unforeseen events. This individually completed degree project aims to provide insight into the principles and techniques used to prevent these kinds of collapses and how they can be applied to future structural design. This is achieved by studying the expected reactions of precast concrete structures and by analyzing the building codes currently applied to concrete buildings in Sweden. The study analyses a fictional four-story concrete building and the resulting deflection of its precast walls as a loadbearing column is removed from its bottom floor. Results are generated using the software Strusoft FEM-design 3D structure, and the deflection is calculated for walls spanning 6, 7, 8 or 8.5 m. Different concrete qualities ranging from 20/25 to 45/55 are applied to each span with the objective of examining the properties of a precast concrete buildings ability to remain stable without the use of extensive attachment methods. The biggest impact is presented by increasing the concrete quality from 25/30 to 30/37, which further justifies the usage of 30/37 in most modern construction. Generally the impact of increasing the structures concrete quality increases as its original deflection increases, which suggests an increase of quality is more profitable for high-risk structures. Furthermore walls constructed with high quality concrete will allow for the usage of further spanning components, as the study shows a lesser increase of deflection under these circumstances. While some results presented in the study may appear minor, it is proven crucial to provide proper attachment of precast concrete components in order to reduce the potential of a disastrous collapse. Keywords: Collapse, deflection, attachment, deformation, concrete quality, diaphragm action, membrane action, span, FEM-design.

As robots become more prolific in modern society, advances must be made in order to ensure understanding during human robot interactions. In this thesis we implement an aspect of a candidate system, Millstream systems, that could represent natural language commands as well as generate new representations for commands based on previously generated data. This thesis presents the results of using two well known syntactic and semantic parsers to generate data and implements a method of mining the data for “production rules” that dictate how to represent an uttered sentence based on the words used. These rules are then generalized using a naive method, allowing them to be applied to a larger set of inputs. Results indicate that from a corpus of 50 imperative sentences 37could be used to generate productions rules which resulted in 187 rules. These rules could then be generalized, resulting in 147 generalized rules, a compression rate of 21.3%. Finally the entire generation process was evaluated and suggestions for extensions to the system, such as gesture recognition, are presented.

<p> This paper will examine how the organizational structures of the U.S. military and the North Atlantic Treaty Organization (NATO) evolved to address the proliferating array of cyber threats. One of the gaps in military cybersecurity is the growing demand for qualified cyber security personnel. Estonia and other countries have employed all-volunteer cyber militias to help protect national security during emergencies. However, the U.S. and NATO are not in the political or legal position to rely upon an all-civilian group. Utilizing Thomas Malone's framework for business, the U.S. military and NATO could abandon their traditional hierarchies and embrace a centralized market model in order to delegate some of their cybersecurity tasks. One way to decentralize their structures is to modify the Estonian Cyber Defense League model by further implementing their reserves systems. They could hire skilled civilian technology professionals through their reserves, eventually creating Cyber Defense Leagues that are primarily comprised of non-commissioned officers.</p>

<p> Lack of information is a serious concern for clinicians. Information resources can address this problem, leading to improvements in decision making and patient outcomes. Genomics is an information-rich domain where searching for information can be complex. For example, most physicians agree that pharmacogenomics can be used to improve the quality of care, and there is evidence that many patients harbor actionable pharmacogenomic variation. However, surveys have shown that physicians feel their knowledge of pharmacogenomics to be inadequate. This represents an information need. A natural approach to meet this need is to provide context-aware access to the precise information needed. The Health Level 7 Context-Aware Knowledge Retrieval Standard, a.k.a the Infobutton, offers a modality to deliver context-aware knowledge into electronic health record (EHR) systems. OpenInfobutton is a reference implementation of this standard that offers an open-source instantiation. In this thesis, we aimed to provide insight into pharmacogenomics information needs and an automated mechanism for addressing these needs. Such work can aid the design of tools that support clinical decisions in genomics.</p>

Thèse numérisée par la Division de la gestion de documents et des archives de l'Université de Montréal.

No description available.

An apparatus was developed to project spinning golf balls directly onto golf greens. This employed a modified baseball/practice machine with two counter-rotating pneumatic wheels. The speed of the wheels could be varied independently allowing backspin to be given to the ball. The ball was projected into a darkened enclosure where the motion of the ball before and after impacting with the turf was recorded using a still camera and a stroboscope. The resulting photographs contained successive images of the ball on a single frame of film. The apparatus was tested on eighteen golf courses resulting in 721 photographs of impacts. Statistical analysis was carried out on the results of the photographs and from this, two types of green emerged. On the first, the ball tended to rebound with topspin, while on the second, the ball retained backspin after impact if the initial backspin was greater than about 350 rads-1. Eleven tests were devised to determine the characteristics of greens and statistical techniques were used to analyse the relationships between these tests. These showed the effects of the green characteristics on ball/turf impacts. It was found that the ball retained backspin on greens that were freely drained and had less than 60% of Poa annua (annual meadow grass) in their swards. Visco-elastic models were used to simulate the impact of the ball with the turf. Impacts were simulated by considering the ball to be rigid and the turf to be a two layered system consisting of springs and dampers. The model showed good agreement with experiment and was used to simulate impacts from two different shots onto two contrasting types of green.