Verifying the power grid requires checking if the voltage drops on all the nodes do not exceed the threshold. We aim to correct an RC model of the grid when some voltage drops violate the threshold condition, by making minor changes to the original design. We first propose an accurate approach to correct the grid which turns out to be too slow. We therefore propose another approach, more suitable for large grids and which can be summarized as follows. The voltage drop is estimated as a function of the metal widths on the grid. A non-linear optimization problem is then formulated and the required metal line width changes that reduce the voltage drops by a sufficient amount are determined. There is a tradeoff between accuracy and speed of the algorithm. However the results show that the gain in speed achieved by the second method, outweighs greatly the loss in accuracy.

Verifying the power grid requires checking if the voltage drops on all the nodes do not exceed the threshold. We aim to correct an RC model of the grid when some voltage drops violate the threshold condition, by making minor changes to the original design. We first propose an accurate approach to correct the grid which turns out to be too slow. We therefore propose another approach, more suitable for large grids and which can be summarized as follows. The voltage drop is estimated as a function of the metal widths on the grid. A non-linear optimization problem is then formulated and the required metal line width changes that reduce the voltage drops by a sufficient amount are determined. There is a tradeoff between accuracy and speed of the algorithm. However the results show that the gain in speed achieved by the second method, outweighs greatly the loss in accuracy.

Grid computing has emerged as a global cyber-infrastructure for the next-generation of e-Science applications by integrating large-scale, distributed and heterogeneous resources. Scientific communities are utilizing Grids to share, manage and process large data sets. In order to support complex scientific experiments, distributed resources such as computational devices, data, applications, and scientific instruments need to be orchestrated while managing the application workflow operations within Grid environments. This thesis investigates properties of Grid workflow management systems, presents a workflow engine and algorithms for mapping scientific workflow applications to Grid resources based on specified QoS (Quality of Service) constraints. (For complete abstract open document)

Thesis (M.S. in Information Technology Management)--Naval Postgraduate School, June 2008. / Thesis Advisor(s): Bordetsky, Alex, "June 2008." Description based on title screen as viewed on August 26, 2008. Includes bibliographical references (p. 69-74). Also available in print.

Zugl.: Paderborn, Univ., Diss., 2006 / Hergestellt on demand

Berlin, Techn. Univ., Diss., 2009.

Zugl.: Hannover, Univ., Diss., 2009

Submitted by Pedro Barros (pedro.silvabarros@ufpe.br) on 2018-06-21T19:25:56Z No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) TESE Luiz Felipe Gomes Rebello Ferreira.pdf: 10574681 bytes, checksum: 88fb46b2af0638e486a62e4ca28fdd89 (MD5) / Made available in DSpace on 2018-06-21T19:25:56Z (GMT). No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) TESE Luiz Felipe Gomes Rebello Ferreira.pdf: 10574681 bytes, checksum: 88fb46b2af0638e486a62e4ca28fdd89 (MD5) Previous issue date: 2017-02-17 / FACEPE / Estima-se que o uso de métodos in silico pode reduzir os custos associados ao desenvolvimento de um novo fármaco em até 50%. Esta redução ocorre porque o número de moléculas que precisam ser testadas e sintetizadas experimentalmente passa a ser drasticamente reduzido devido a alta confiabilidade dos métodos computacionais. Porém, estes métodos podem apresentar uma alta demanda computacional quando o número de moléculas a ser testados é alto e quando se busca maior precisão nos resultados numéricos. Sendo assim, este trabalho apresenta o desenvolvimento do programa GriDoMol, uma plataforma unificada para realizar cálculos de docking molecular em um sistema distribuído, através de um grid computacional, com foco em alto desempenho e precisão. Utilizando o GriDoMol e configurações avançadas de docking, foi possível realizar o docking molecular de um conjunto de 213 complexos em um tempo até 9,91 vezes mais rápido e encontrando soluções de docking mais estáveis, com reduções de até 1,3 Kcal/mol, quando comparado com a execução sequencial deste mesmo conjunto em um único computador, utilizando apenas um núcleo de processamento e a configuração padrão de docking. O GriDoMol também oferece a opção de realizar estudos de vacinologia reversa permitindo cálculos de docking molecular entre candidatos a epítopos e alelos de MHCs de Classe I e II humanos com o intuito de encontrar epítopos que possuam uma boa afinidade por estes alelos, aumentando as chances de apresentar uma resposta imunológica significativa. / It is estimated that the use of in silico methods can reduce the costs spent at the development stage of a new drug by up to 50%. This happens because the number of molecules that need to be experimentally synthesized and tested becomes drastically reduced due to the high predictability and reliability of the computational methods. Nevertheless, these methods may present a high computational demand when the number of molecules to be tested is high and when it’s seeking for a higher precision in the numerical results. Therefore, this work presents the development of the program GriDoMol, a unified platform for performing molecular docking calculations in a distributed system, through a computational grid environment, with focus in high performance and precision. By using GriDoMol and higher docking settings values, it was possible to execute the molecular docking of a set containing 213 complexes in a time up to 9.91 times faster and finding more stable complexes, with energy reduction by up to 1.3 Kcal/mol, when compared with the sequential execution of this same set on a single computer, using a single processor core and the default docking settings values. The program GriDoMol also offers a reverse vaccinology option, allowing the molecular docking of candidate epitopes on selected human MHC’s Class I and II alleles, in order to find the most promising epitopes which have a good binding affinity on a large set of alleles and, thus, better chances of having a significant immunogenic response.

Aiming at aggregating numerous distributed resources to provide immense computing power, Grid computing has emerged as a promising paradigm to run complex composite applications such as workflows. However, the inherent uncertainties of grid systems as well as the structural complexity of workflow applications make it extremely challenging to schedule workflows in an efficient way, regardless of whether the objective is to minimize execution time or meet specific user and/or system Quality of Service (QoS) requirements. For both these cases, this thesis considers scheduling problems motivated by grid uncertainties and advances the state-of-the-art by developing new techniques to address these problems.First, based on existing scheduling heuristics, a Monte-Carlo approach is developed to minimize the average makespan (i.e., the overall execution time) in the presence of task estimates exhibiting limited uncertainty in the form of (controlled) random behaviour. Next, a scenario where performance prediction is difficult to obtain and resource availability may vary over time, is considered. A low-cost efficient just-in-time heuristic is proposed to cope with grid uncertainties.After addressing these performance-driven scheduling problems, a QoS-driven problem, which considers not only the aforementioned uncertainties but also the uncertainty caused by queue-based scheduling, is examined. In order to tackle all these uncertainties, an integrated scheduling model consisting of three supportive techniques is developed. Extensive evaluation using simulation shows that the proposed techniques can achieve substantial improvements towards the ultimate goal of providing a good solution for QoS-driven workflow scheduling on the Grid.

Nous traitons dans cette thèse des différents aspects de la manipulation de maillages, et de la façon dont ces opérations peuvent être effectuées en parallèle, ou en mémoire distribuée, à l’heure où les GPUs comme les super-calculateurs deviennent de plus en plus utilisés. Ainsi, nous présentons des algorithmes de lissage surfacique et volumique, inspirés d’algorithmes de traitement d’images (filtre bilatéral, histogrammes locaux). Après ces manipulations de géométrie, nous abordons des problématiques topologiques comme le remaillage local, dans le but de générer, à partir d’un maillage tétraédrique, une couche limite de cellules prismatiques et hexaédriques de bonne qualité, permettant des simulations de mécanique des fluides dans ces ones proches de la surface. Enfin, nous décrivons une technique de maillage basée sur des interactions particulaires, permettant de générer des maillages à dominante quadrangulaire. / This thesis deals with different aspects of mesh processing, and the way those operations can be done in parallel, or using distributed memory, when GPUs and supercomputers are more and more commonly used. We present surfacic and volumetric mesh smoothing algorithms, based upon image processing techniques (bilateral filter, local histograms). After those geometric considerations, we talk about topologic methods, as local remeshing, enabling one to generate, from a tetrahedral mesh, one layer of good quality prisms and hexahedron, allowing fluid mechanics simulations in those near-surface areas. Finally, we present a meshing technique based upon particular interactions, in order to construct quad-dominant meshes.