Dissertations / Theses on the topic 'Benchmark functions'

O desenvolvimento de aplicações móveis é guiado por uma especial atenção aos requisitos não funcionais (do inglês, NFR - Non Functional Requirements), sendo o principal objetivo proporcionar uma boa experiência ao usuário final. Entretanto, a avaliação de NFRs é ainda uma tarefa manual, não estruturada e que consome muito tempo. Esta dissertação apresenta um estudo de várias abordagens relacionadas à avaliação de desempenho (por exemplo, o uso de aplicações de benchmark) e de NFRs no âmbito de sistemas móveis. No entanto, os benchmarks atuais são genéricos, geralmente, voltados para a plataforma de execução e nem sempre instituem um consenso na classificação de dispositivos. Visando uma melhor avaliação de NFRs e uma classificação de dispositivos com base nas necessidades de aplicações reais, este trabalho propõe um framework para gerar benchmarks orientados às necessidades de cada aplicação e, assim, fornecer uma forma eficiente e eficaz para estimar requisitos não funcionais em sistemas móveis. Essa ferramenta é composta por uma biblioteca de testes parametrizáveis, métricas e uma estrutura para geração rápida de benchmarks orientados à aplicação. O framework foi construído utilizando o paradigma de programação orientada a aspectos para coleta das métricas por fornecer uma maior modularidade e separação de interesses, de modo que a sua evolução, através da adição de outras métricas ou testes, seja facilitada. Para validação da proposta, foram realizados experimentos com cinco aplicações Android reais disponíveis na Play Store, sendo que para cada aplicação foi gerado um benchmark específico cujos resultados foram comparados com os obtidos para as aplicações móveis reais. Os resultados são promissores, mostrando que é possível criar aplicações de teste com comportamento semelhante ao de aplicações reais e, assim, classificar dispositivos com base nas necessidades das aplicações, através da análise das métricas presentes no framework. Essas métricas podem, ainda, orientar o desenvolvedor na otimização de suas aplicações ou ainda na escolha de dispositivos com melhor custo benefício para executar seus aplicativos.<br>The mobile application development is guided by a special attention to non-functional requirements (NFRs), where a good experience for the end user is the primary goal. However, NFRs evaluation is still a manual, unstructured and time-consuming task. This thesis presents a study of several approaches related to performance and NFR evaluation within mobile systems. Among these approaches is the use of benchmark applications. Currently available benchmarks are generic, usually focused on the execution platform and do not always establish a consensus on the classification of devices. For a better NFRs assessment and classification of devices based on real application needs, this work proposes a framework for generating application-oriented benchmarks for the early estimation of non-functional requirements in mobile systems. This framework is composed of a configurable test library, a set of metrics and an engine the assembling of the test program. The framework uses aspect-oriented programming to collect the metrics of interest. This approach provides increased modularity and separation of concerns, thus facilitating the improvement of the framework itself, by adding other metrics or testing operations. In order to validate the proposed framework we used five application from the Android Play store. For each application, a specific benchmark is generated and executed in different devices. The results are compared to those of the execution of the actual applications in the same devices. Experimental results are promising, showing that it is possible to create test applications with similar behavior to that of real applications and thus classify devices based on the actual application needs, by analyzing the metrics present in the framework. These metrics can also guide the developer in optimizing her applications or in choosing devices with the best trade-off between cost and performance to run a given application.

Conselho Nacional de Desenvolvimento CientÃfico e TecnolÃgico<br>Quando um conversor opera com uma larga diferenÃa entre os nÃveis de tensÃo de entrada e de saÃda, normalmente à possÃvel identificar uma reduÃÃo significante no seu desempenho. A razÃo disto à o aumento da quantidade de energia processada, que primeiramente precisa ser armazenada em um elemento passivo (indutor), antes de ser entregue à carga. à possÃvel dizer que quÃo maior for a quantidade dessa energia âindiretaâ, menor serà a eficiÃncia do sistema. Tal situaÃÃo à especialmente crÃtica para inversores e retificadores com correÃÃo de fator de potÃncia (PFC), visto que a razÃo cÃclica dos interruptores abrange praticamente todos os possÃveis valores. Em casos em que a diferenÃa entre o valor de pico da tensÃo CA e o valor mÃdio da tensÃo CC à grande, o Ãndice de modulaÃÃo à desviado consideravelmente de 1 e, consequentemente, maiores perdas sÃo esperadas. Para lidar com tal situaÃÃo, à proposto um retificador com caracterÃstica abaixadora e elevadora utilizando uma funÃÃo chamada de bypass. Esta funÃÃo permite que o retificador escolha entre os dois barramentos o que melhor se adapta ao nÃvel de saÃda desejado em funÃÃo do ponto de operaÃÃo da tensÃo senoidal de entrada, maximizando a eficiÃncia. Algumas das vantagens esperadas com a utilizaÃÃo da funÃÃo bypass sÃo: reduÃÃo das perdas totais por meio da operaÃÃo parcial com estÃgio Ãnico; maior nÃmero de nÃveis de tensÃo, reduzindo a corrente de modo comum e o volume do filtro de entrada, alÃm de resultar em menores esforÃos de tensÃo sobre os semicondutores; e, finalmente, reduÃÃo de perdas adicionais e no tamanho do sistema, visto que o conversor Buck de saÃda à projetado para apenas uma fraÃÃo da potÃncia total. Em adiÃÃo, sÃo apresentadas possÃveis configuraÃÃes de conversores de potÃncia para o carregamento de baterias de veÃculos elÃtricos. Topologias de retificadores trifÃsicos com PFC com corrente de entrada senoidal e tensÃo de saÃda controlada sÃo analisadas e propostas, e suas funcionalidades e caracterÃsticas bÃsicas descritas brevemente. FÃrmulas analÃticas para o cÃlculo dos esforÃos de tensÃo e corrente sobre os semicondutores de potÃncia sÃo fornecidas. A fim de avaliar comparativamente o desempenho das topologias selecionadas, fatores adimensionais de referÃncia sÃo definidos com base nos esforÃos elÃtricos sobre os semicondutores e no volume dos indutores. As caracterÃsticas do sistema proposto, incluindo princÃpio de operaÃÃo, estratÃgia de modulaÃÃo, equaÃÃes de dimensionamento e cÃlculos de perdas e eficiÃncia, sÃo descritas em detalhes. Finalmente, a viabilidade do conceito de bypass à demonstrada por meio de resultados experimentais obtidos a partir de um protÃtipo de 22 kW.<br>When operating with large differences between input and output voltage levels, it is normally possible to identify a significant reduction on the performance of practically all topologies in regard of conversion efficiency. Reason for this is the increasing amount of processed energy that needs to be firstly stored in a passive element (inductor) before reaching the load. It is therefore possible to say that the higher the amount of such âindirectâ energy is; the lower will be the converter efficiency. Such situation is especially critical when considering the operation of inverters or controlled rectifiers with power factor correction (PFC) because the converter sweeps practically all the possible values of duty cycle. In the case the difference between peak AC value and DC value is large; the modulation index will strongly deviate from 1. Consequently, even higher amount of losses are expected. In order to deal with the above referred drawbacks, it is proposed a 5-level three-phase PFC rectifier with an innovative approach, named bypass concept. This function allows the converter to switch between one of the available DC-links that best matches the required output levels depending on the operating point of the sinus wave, thus maximizing the efficiency. Furthermore, the referred bypass function enables direct access to the required lower voltage level by the load, reducing significantly the amount of losses. The 5-level operation allows the voltage steps to be lower than those found in three and two-level topologies, from where lower harmonic contents, reduction of common mode current and EMI are observed. Finally, the output DC-DC converters are designed for only a fraction of the nominal power, having a rather reduced impact on additional losses and also on the converter size. In addition, possible power electronics configurations for charging of EVs are presented. Suitable three-phase PFC rectifier topologies with sinusoidal input currents and controlled output voltage are analyzed and proposed, and their functionality and basic characteristics briefly described. Analytical formulas for calculating the current stresses on the power semiconductors are provided, and in order to evaluate comparatively the performance of selected topologies, dimensionless benchmark factors are derived concerning the semiconductor stresses and the volume of the main inductive components. The characteristics of the proposed system, including the principle of operation, modulation strategy, dimensioning equations and calculated losses and efficiency, are described in detail. Finally, the feasibility of the bypass concept is demonstrated by means of experimental results obtained from a 22 kW hardware prototype.

Ce travail de thèse démontre l’importance d’intégrer des outils théoriques à des observations expérimentales dans le but d’étudier le rôle des interactions non-covalentes et plus précisément de la dispersion dans la chimie des métaux de transition. Plusieurs thèmes ont ainsi été abordés comme les interactions d’empilement entre chélates de métaux de transition à l’état solide; l’influence de la chiralité sur l’oligomérisation en solution de complexes plans carrés de Rh(I) isonitrile; la stabilité et inactivité inhabituelles de complexes de type cis-platine en solution concentrée. Les résultats obtenus par titration calorimétrique isotherme ont permis d’évaluer la capacité de méthodes théoriques à reproduire avec précision les résultats expérimentaux. Les calculs ont démontré qu’un traitement théorique approprié des effets de la dispersion et de la solvatation, donne des valeurs cohérentes avec les résultats expérimentaux. Cependant, des améliorations supplémentaires sont nécessaires<br>This thesis has shown the importance of integration of theoretical calculations and experimental investigations in studying the role of non-covalent interactions and particularly dispersion interactions in transition metal chemistry. Several subjects were addressed, such as stacking interactions of chelates in transition metal complexes in solid state, influence of chirality on the oligomerization of Rh(I) isonitrile complexes in solution and the stability of the cis-platin type complexes in concentrated solutions. Isothermal titration calorimetry proved to be very useful in the studies by providing accurate experimental data on the thermochemistry of addressed processes. This data was used to gauge the ability of the theoretical methods to accurately reproduce the experimental results. Calculations have shown that the proper treatment of dispersion effects and solvation by theoretical models gives values in relatively good agreement with experiments, but further improvements are needed

Multi functional health monitoring wearable devices are quite prominent these days. Usually these devices are battery-operated and consequently are limited by their battery life (from few hours to a few weeks depending on the application). Of late, it was realized that these devices, which are currently being operated at fixed voltage and frequency, are capable of operating at multiple voltages and frequencies. By switching these voltages and frequencies to lower values based upon power requirements, these devices can achieve tremendous benefits in the form of energy savings. Dynamic Voltage and Frequency Scaling (DVFS) techniques have proven to be handy in this situation for an efficient trade-off between energy and timely behavior. Within imec, wearable devices make use of the indigenously developed MUSEIC v2 (Multi Sensor Integrated circuit version 2.0). This system is optimized for efficient and accurate collection, processing, and transfer of data from multiple (health) sensors. MUSEIC v2 has limited means in controlling the voltage and frequency dynamically. In this thesis we explore how traditional DVFS techniques can be applied to the MUSEIC v2. Experiments were conducted to find out the optimum power modes to efficiently operate and also to scale up-down the supply voltage and frequency. Considering the overhead caused when switching voltage and frequency, transition analysis was also done. Real-time and non real-time benchmarks were implemented based on these techniques and their performance results were obtained and analyzed. In this process, several state of the art scheduling algorithms and scaling techniques were reviewed in identifying a suitable technique. Using our proposed scaling technique implementation, we have achieved 86.95% power reduction in average, in contrast to the conventional way of the MUSEIC v2 chip’s processor operating at a fixed voltage and frequency. Techniques that include light sleep and deep sleep mode were also studied and implemented, which tested the system’s capability in accommodating Dynamic Power Management (DPM) techniques that can achieve greater benefits. A novel approach for implementing the deep sleep mechanism was also proposed and found that it can obtain up to 71.54% power savings, when compared to a traditional way of executing deep sleep mode.<br>Nuförtiden så har multifunktionella bärbara hälsoenheter fått en betydande roll. Dessa enheter drivs vanligtvis av batterier och är därför begränsade av batteritiden (från ett par timmar till ett par veckor beroende på tillämpningen). På senaste tiden har det framkommit att dessa enheter som används vid en fast spänning och frekvens kan användas vid flera spänningar och frekvenser. Genom att byta till lägre spänning och frekvens på grund av effektbehov så kan enheterna få enorma fördelar när det kommer till energibesparing. Dynamisk skalning av spänning och frekvens-tekniker (såkallad Dynamic Voltage and Frequency Scaling, DVFS) har visat sig vara användbara i detta sammanhang för en effektiv avvägning mellan energi och beteende. Hos Imec så använder sig bärbara enheter av den internt utvecklade MUSEIC v2 (Multi Sensor Integrated circuit version 2.0). Systemet är optimerat för effektiv och korrekt insamling, bearbetning och överföring av data från flera (hälso) sensorer. MUSEIC v2 har begränsad möjlighet att styra spänningen och frekvensen dynamiskt. I detta examensarbete undersöker vi hur traditionella DVFS-tekniker kan appliceras på MUSEIC v2. Experiment utfördes för att ta reda på de optimala effektlägena och för att effektivt kunna styra och även skala upp matningsspänningen och frekvensen. Eftersom att ”overhead” skapades vid växling av spänning och frekvens gjordes också en övergångsanalys. Realtidsoch icke-realtidskalkyler genomfördes baserat på dessa tekniker och resultaten sammanställdes och analyserades. I denna process granskades flera toppmoderna schemaläggningsalgoritmer och skalningstekniker för att hitta en lämplig teknik. Genom att använda vår föreslagna skalningsteknikimplementering har vi uppnått 86,95% effektreduktion i jämförelse med det konventionella sättet att MUSEIC v2-chipets processor arbetar med en fast spänning och frekvens. Tekniker som inkluderar lätt sömn och djupt sömnläge studerades och implementerades, vilket testade systemets förmåga att tillgodose DPM-tekniker (Dynamic Power Management) som kan uppnå ännu större fördelar. En ny metod för att genomföra den djupa sömnmekanismen föreslogs också och enligt erhållna resultat så kan den ge upp till 71,54% lägre energiförbrukning jämfört med det traditionella sättet att implementera djupt sömnläge.

碩士<br>國立臺灣科技大學<br>營建工程系<br>101<br>Construction site layout (CSL) represents multi-criteria approach to solving problems which related to site planning and design. Arrange a set of predetermined facilities into appropriate locations is a difficult problem as there are many possible alternatives. Due to the high complexity of site layout problems, many algorithm based on metaheuristic methods have been developed to generate solutions for the problems. Previous metaheuristic methods such as particle swarm optimization (PSO), genetic algorithm (GA), differential evolution (DE), and firefly algorithm (FA), designate a computational method to optimize a problem, but these methods have their own drawbacks. To lessen those drawbacks, this study propose a new hybrid meta-heuristic model namely particle firefly differential algorithm (PFDA). This algorithm combines the advantages PSO, FA, and DE. This hybrid integrates the local search ability of PSO and global search ability of FA and DE. There are three phases in PFDA, first is PSO phase, which stores the best value and focus on exploitation. Second and third phase are proceed as parallel way, FA and DE. Both of them focus on exploration. This study compares the performance of PFDA with GA, PSO, FA, DE, bee algorithm (BA), and particle bee algorithm (PBA) for multidimensional benchmark function problems. Moreover, this study compares PFDA performance against original PSO, DE, FA, and the previous research works in site facility layout problems. The results show that PFDA&apos;s performance is better than those mentioned algorithms in the benchmark functions and outperforms the existing optimization algorithms in solving constructions site layout problem.

Dissertação de Mestrado em Engenharia Informática apresentada à Faculdade de Ciências e Tecnologia<br>Existem várias plataformas que suportam o paradigma de computação serverless na cloud. Perceber se estas plataformas são suficientemente maduras e oferecem uma performance que as permita aplicar a contextos de utilização real, é essencial para que se possa optar por uma delas. Estudos de performance e benchmarks existentes na área revelam-se lacunares e não suficientemente abrangentes, pelo que surgiu a necessidade de abordar esta temática. No decorrer deste trabalho, é realizada a identificação e estudo das plataformas serverless existentes, sendo estas comparadas em termos de funcionalidades e características, através da utilização de um conjunto de parâmetros apresentado. É desenhado e proposto um grupo de testes para um benchmark de utilização genérica e com vista à automatização da sua execução, é apresentada uma arquitetura de um sistema. Procedeu-se à implementação do sistema, que foi utilizado para a execução do conjunto de testes que constitui o benchmark. Foram testadas os serviços de computação serverless AWS Lambda, Azure Functions, Google Cloud Functions e OpenWhisk. Com os resultados obtidos foi efetuada uma observação de como estas se comportam em diferentes cenários, tendo também sido efetuada uma comparação entre elas. Por último, procedeu-se à criação de uma demonstração da utilização da OpenWhisk para a formação de um serviço de mobilidade que permite a otimização de rotas de um fluxo de veículos. Pensa-se que o conjunto de testes proposto e o sistema de execução do benchmark consistirá num bom auxiliar para pessoas que pretendam realizar testes de performance às plataformas de computação serverless, de forma a tomarem as melhores decisões, conforme as necessidades, no momento de utilizarem uma delas.<br>There are several platforms that supports the paradigm of serverless cloud computing.Realizing if these platforms are sufficiently mature and deliver performance that allows them to be applied to real-life contexts is essential so that we can chose one.Existing performance studies and benchmarks in the area presented a gap and are not very comprehensive, and therefore the need to address this issue has arisen.In the course of this internship, an identification and a study of the existing serverless platforms is carried out, being these compared in terms of functionalities and characteristics, through the use of a presented set of parameters. It is also designed and proposed a test suite for a benchmark of generic us and in order to automate its execution, an architecture of a system is presented.The implementation of the system was performed, which was used to run the test suite that constitutes the benchmark. Serverless computing platforms AWS Lambda, Azure Functions, Google Cloud Functions and OpenWhisk were tested. With the obtained results, an observation was made of how the platforms behave in different scenarios and a comparison was also made between them.Finally, a demonstration of the use of OpenWhisk for the creation of a mobility service that allows the optimization of routes of a flow of vehicles has been created.It is thought that the proposed test suite and the benchmark execution system will be a good helper for people who want to perform performance tests on serverless computing platforms in order to make the best decisions, according to the needs, when is the moment using one of them.

碩士<br>國立成功大學<br>資訊工程學系碩博士班<br>100<br>An objective novel evaluation approach, implemented by the benchmark function-based peak signal-to-noise ratio (PSNR), for evaluating the performance of image enlargement is proposed in this thesis. Also, a fast large-scale image enlargement method via the improved discrete cosine transform (DCT) is proposed to improve the quality and speed of image zooming. The proposed image enlargement algorithm based on DCT saves computation time by multiplication of the DCT matrix. Comparing to the traditional DCT approach, the improved approach overcomes the image shifting and blocky effects. In comparisons to other interpolation methods, DCT enlargement outperforms them in edge details because it considers the global frequency information of the whole image. The DCT enlargement is easy to implement the arbitrary pixel-size-based zooming of an image by employing the different size of transform matrix. Illustrative examples show the effectiveness of the proposed approach.

The non-adiabatic quantum molecular dynamics (NA-QMD) method couples self-consistently classical nuclear motion with time-dependent density functional theory (TDDFT) in basis expansion for the electron dynamics. It has become a versatile approach to study the dynamics of atoms, molecules and clusters in a wide range of scenarios. This work presents applications of the NA-QMD method to important benchmark systems and its systematic extension to include quantum effects in the nuclear motion. Regarding the first objective, a complete study of the strong-field ionization and dissociation dynamics of nature’s simplest molecule H2+ is performed. By including all electronic and nuclear degrees of freedom and all reaction channels, molecular rotation is shown to play an important role in the ionization process. In addition, strong orientation effects in the energy deposition process of the Buckminster fullerene C60 in short intense laser pulses are surprisingly found in full dimensional calculations. Their consequences on the subsequent nuclear relaxation dynamics shed new light on available experimental data and future experiments are proposed to confirm the detailed predictions. Regarding the second objective, the NA-QMD formalism is basically extended to take electron-nuclear correlations into account. This extension is achieved by means of a trajectory surface hopping scheme in the adiabatic Kohn-Sham framework. First studied examples from collision physics and photochemistry illustrate the relevance and importance of quantum effects in the nuclear dynamics.

This research presents a Hybrid Particle Swarm Optimization with Bat Algorithm (HPSOBA) based approach to solve Optimal Reactive Power Dispatch (ORPD) problem. The primary objective of this project is minimization of the active power transmission losses by optimally setting the control variables within their limits and at the same time making sure that the equality and inequality constraints are not violated. Particle Swarm Optimization (PSO) and Bat Algorithm (BA) algorithms which are nature-inspired algorithms have become potential options to solving very difficult optimization problems like ORPD. Although PSO requires high computational time, it converges quickly; while BA requires less computational time and has the ability of switching automatically from exploration to exploitation when the optimality is imminent. This research integrated the respective advantages of PSO and BA algorithms to form a hybrid tool denoted as HPSOBA algorithm. HPSOBA combines the fast convergence ability of PSO with the less computation time ability of BA algorithm to get a better optimal solution by incorporating the BA’s frequency into the PSO velocity equation in order to control the pace. The HPSOBA, PSO and BA algorithms were implemented using MATLAB programming language and tested on three (3) benchmark test functions (Griewank, Rastrigin and Schwefel) and on IEEE 30- and 118-bus test systems to solve for ORPD without DG unit. A modified IEEE 30-bus test system was further used to validate the proposed hybrid algorithm to solve for optimal placement of DG unit for active power transmission line loss minimization. By comparison, HPSOBA algorithm results proved to be superior to those of the PSO and BA methods. In order to check if there will be a further improvement on the performance of the HPSOBA, the HPSOBA was further modified by embedding three new modifications to form a modified Hybrid approach denoted as MHPSOBA. This MHPSOBA was validated using IEEE 30-bus test system to solve ORPD problem and the results show that the HPSOBA algorithm outperforms the modified version (MHPSOBA).<br>Electrical and Mining Engineering<br>M. Tech. (Electrical Engineering)

碩士<br>國立臺灣大學<br>物理研究所<br>101<br>This thesis primarily consists of two parts. In the first one, we benchmarked the performance of the functionals in density functional theory which have the asymptotically correct behavior. We compare the performance of the two different asymptotically corrected schemes, the long-range corrected (LC) hybrid scheme and the asymptotically corrected (AC) model potential scheme, in a wide range of applications. These schemes aim at producing the correct asymptotic behavior, which is known to be important to the Rydberg excitations and the frontier orbital energies of molecules. Based on our investigation, we found that the LC hybrid scheme outperforms the AC model potential scheme almost on all the properties studied, and we conclude that the LC hybrid scheme provides a promising direction for the future development of new exchange-correlation functionals. Besides, in the process of the benchmark, I established a molecular database ``IP131 full set&apos;&apos;, which consists of 131 experimental geometries of atoms and molecules, and their reference values of 131 experimental vertical ionization energies, 131 CCSD(T) calculated vertical electron affinities, 131 CCSD(T) calculated fundamental gaps and 113 CCSD(T) calculated atomization energies. The IP131 full set can be used in the benchmarks for other Ab Initio calculation methods, or in the parameter fitting in the development of density functionals in the future. In the second part, we adopted the Schrodinger equation numerical solutions to improve the applicability of the normal mode analysis in the case of TMA-H+-H2O, for the motivation that the theoretical simulated infrared spectrum of TMA-H+-H2O given by the normal mode analysis is far from the experiment results. Normal mode analysis is a convenient and widely adopted method that people usually use it to analyse the vibrational motions of multi-body mechanical systems, both classical and quantum. However, normal mode analysis is a approximate method and its assumptions, which make the analysis procedures simple and systematically solvable, also make the normal mode analysis not applicable to some vibrational motions in some mechanical systems. Our main modification method in this work is to regard the vibrational normal modes as some low dimensional quantum systems and adopt the Schrodinger equation numerical solutions in these quantum systems to replace the results from the normal mode analysis. By doing so, we can not only get a closer result to the experimental infrared spectrum, but also have a deeper understanding of the capability and applicability of normal mode analysis. Once in the future we encounter same problems when using the normal mode analysis in different systems, we may again adopt the procedures developed here to make it better.

The thesis compares several non-linear regression methods on synthetic data sets gen- erated using standard benchmarks for a continuous black-box optimization. For that com- parison, we have chosen the following regression methods: radial basis function networks, Gaussian processes, support vector regression and random forests. We have also included polynomial regression which we use to explain the basic principles of regression. The com- parison of these methods is discussed in the context of black-box optimization problems where the selected methods can be applied as surrogate models. The methods are evalu- ated based on their mean-squared error and on the Kendall's rank correlation coefficient between the ordering of function values according to the model and according to the function used to generate the data. 1