No description available.

No description available.

No description available.

No description available.

With the emergence of various big data platforms in recent years, Apache Spark - a distributed large-scale computing platform, is perceived as a potential substitute for Message Passing Interface (MPI) in High Performance Computing (HPC). Due to the limitations in fault-tolerance, dynamic resource handling and ease of use, MPI, as a dominant method to achieve parallel computing in HPC, is often associated with higher development time and costs in enterprises such as Scania IT. This thesis project aims to examine Apache Spark as an alternative to MPI on HPC clusters and compare their performance in various aspects. The test results are obtained by running a compute- intensive application on both platforms to solve a Bayesian inference problem of a extended Lotka-Volterra model using particle Markov chain Monte Carlo methods. As is confirmed by the tests, Spark is demonstrated to be superior in fault tolerance, dynamic resource handling and ease of use, whilst having its shortcomings in performance and resource consumption compared with MPI. Overall, Spark proves to be a promising alternative of MPI on HPC clusters. As a result, Scania IT continues to explore Spark on HPC clusters for use in different departments.

Responsive graphical user interfaces, which can adapt to changes in display size, improve the user experience and are easier to developand maintain compared to ones relying on absolute (pixel) placement. Based on a survey of several responsive frameworks across desktop, mobile and web platforms, we design and develop a backwards- compatible responsive extension of an existing non-responsive Lua GUI framework. We also provide the specification that describes the features selected to be part of the new responsive framework, as well as some implementation details pertaining to those features.

Electricity can be considered as the backbone of the modern society. It has become anessential component of the day to day life of all human beings in the world. As a result, thedemand of electricity is growing at a higher rate each year. To fulfill this demand a largenumber of fossil fuels are being consumed all around the world.Sri Lanka, as a developing country in the Asian region, is currently facing this issueof having difficulties in meeting the electricity demand of the country’s population. As acountry having a lower economic growth rate, it is an essential choice to reduce the usage offossil fuel, since it is not only reducing the wastage of money for exportations of fossil fuelfrom other countries but also helps to reduce the harmful gas emissions. Compared to theEuropean countries, the involvement of Sri Lanka in utilizing renewable energy resources forpower generation is observed to be very less. Hence this project aims on analyzing andproviding an overview of the possibility in fulfilling the country’s energy demand bycompletely renewable resources.Three renewable energy components have been taken in to account for this study. Thepower output from solar, wind and hydropower resources have been considered and theanalysis has been carried out with MATLAB software. It is an obvious fact that the electricitydemand varies within the day due to the different lifestyles of people. Hence hourly demandis the most important fact to be considered during a demand-supply analysis. With the dataobtained from Ceylon electricity board, Ministry of power and Renewable Energy, theanalysis has been done for all three resources assuming the wind and solar power plants arebeing constructed at the perfect locations after a careful study of the wind maps and solarirradiation data throughout a year.The Analysis has been carried out with solar power alone and then with wind powerand then with the combined solar and wind power. The results have shown that it is notpossible to fulfill the hourly demand solely from solar and wind power since the solarradiation and the wind flow is not constant throughout the day in each location. Hence forcertain hours the power production is excess and for certain hours power production is lessthan the load requirement. Then Hydro power has been fitted in to place to compensate theshortage in the load. Calculations have been carried out considering only the current hydropower plants which are on operation.Regarding Hydro power plants, the analysis was mainly based on the assumption thatpumps are being used to handle the excess electricity produced by solar and wind plants.Hence the final outcome has been considered based on two main facts which are the currentavailable storage span of all hydro plants as a whole and the required pump power and thehydro turbine power to handle the excess power production from solar and wind.The results have shown that it is possible to meet the total hourly energy demand ofthe whole population of the country by completely renewable resources under certainconditions. The combination and sometimes correlation of solar and wind during the day hasthe capacity to meet the different hourly demands within the day for certain days. However tomeet the hourly demand, hydro power has to be included to balance the shortage. Consideringthe two facts mentioned earlier it has been proven that the current storage span and pumppower is sufficient to balance the excess electricity produced from certain value of installedPV power and certain number of turbines in a wind plant, however for certain value ofinstalled PV power and certain number of turbines ,the pump power requirement and storagelevel to handle the excess power is slightly more than the current power production fromhydro. Hence it is possible to meet the hourly load if more hydro power plants could beconstructed for those cases.Since this is a positive and bright outcome, the next step required is the attention ofthe authorities to execute the construction and implementation of projects related to solar,wind and Hydro plants and thereby opening a new era in the technology and the economy ofSri Lanka.

No description available.

No description available.

Impressive results can be achieved when stacking deep neural networks hierarchies together. Several machine learning papers claim state-of-the-art results when evaluating their models with different accuracy metrics. However, these models come at a cost, which is rarely taken into consideration. This thesis aims to shed light on the resource consumption of machine learning algorithms, and therefore, five efficiency metrics are proposed. These should be used for evaluating machine learning models, taking accuracy, model size, and time and energy consumption for both training and inference into account. These metrics are intended to allow for a fairer evaluation of machine learning models, not only looking at accuracy. This thesis presents an example of how these metrics can be used by applying them to both text and image classification tasks using the algorithms SVM, MLP, and CNN.