Distributed processing of large remote sensing images using MapReduce - A case of Edge Detection

Dissertation submitted in partial fulfillment of the requirements for the Degree of Master of Science in Geospatial Technologies.

Multi-GPU support on the marrow algorithmic skeleton framework

Dissertação para obtenção do Grau de Mestre em Engenharia Informática

Using the GPU to design complex profile extrusion dies

In the present work the benefits of using graphics processing units (GPU) to aid the design of complex geometry profile extrusion dies, are studied. For that purpose, a3Dfinite volume based code that employs unstructured meshes to solve and couple the continuity, momentum and energy conservation equations governing the fluid flow, together with aconstitutive equation, was used. To evaluate the possibility of reducing the calculation time spent on the numerical calculations, the numerical code was parallelized in the GPU, using asimple programing approach without complex memory manipulations. For verificationpurposes, simulations were performed for three benchmark problems: Poiseuille flow, lid-driven cavity flow and flow around acylinder. Subsequently, the code was used on the design of two real life extrusion dies for the production of a medical catheter and a wood plastic composite decking profile. To evaluate the benefits, the results obtained with the GPU parallelized code were compared, in terms of speedup, with a serial implementation of the same code, that traditionally runs on the central processing unit (CPU). The results obtained show that, even with the simple parallelization approach employed, it was possible to obtain a significant reduction of the computation times.

Development of an integral viscoelastic flow solver in OpenFOAM®

The usual high cost of commercial codes, and some technical limitations, clearly limits the employment of numerical modelling tools in both industry and academia. Consequently, the number of companies that use numerical code is limited and there a lot of effort put on the development and maintenance of in-house academic based codes. Having in mind the potential of using numerical modelling tools as a design aid, of both products and processes, different research teams have been contributing to the development of open source codes/libraries. In this framework, any individual can take advantage of the available code capabilities and/or implement additional features based on his specific needs. These type of codes are usually developed by large communities, which provide improvements and new features in their specific fields of research, thus increasing significantly the code development process. Among others, OpenFOAM® multi-physics computational library, developed by a very large and dynamic community, nowadays comprises several features usually only available in their commercial counterparts; e.g. dynamic meshes, large diversity of complex physical models, parallelization, multiphase models, to name just a few. This computational library is developed in C++ and makes use of most of all language capabilities to facilitate the implementation of new functionalities. Concerning the field of computational rheology, OpenFOAM® solvers were recently developed to deal with the most relevant differential viscoelastic rheological models, and stabilization techniques are currently being verified. This work describes the implementation of a new solver in OpenFOAM® library, able to cope with integral viscoelastic models based on the deformation field method. The implemented solver is verified through the comparison of the predicted results with analytical solutions, results published in the literature and by using the Method of Manufactured Solutions.

A new integral viscoelastic flow solver in OpenFOAM®

The usual high cost of commercial codes, and some technical limitations, clearly limits the employment of numerical modelling tools in both industry and academia. Consequently, the number of companies that use numerical code is limited and there a lot of effort put on the development and maintenance of in-house academic based codes . Having in mind the potential of using numerical modelling tools as a design aid, of both products and processes, different research teams have been contributing to the development of open source codes/libraries. In this framework, any individual can take advantage of the available code capabilities and/or implement additional features based on his specific needs. These type of codes are usually developed by large communities, which provide improvements and new features in their specific fields of research, thus increasing significantly the code development process. Among others, OpenFOAM® multi-physics computational library, developed by a very large and dynamic community, nowadays comprises several features usually only available in their commercial counterparts; e.g. dynamic meshes, large diversity of complex physical models, parallelization, multiphase models, to name just a few. This computational library is developed in C++ and makes use of most of all language capabilities to facilitate the implementation of new functionalities. Concerning the field of computational rheology, OpenFOAM® solvers were recently developed to deal with the most relevant differential viscoelastic rheological models, and stabilization techniques are currently being verified. This work describes the implementation of a new solver in OpenFOAM® library, able to cope with integral viscoelastic models based on the deformation field method. The implemented solver is verified through the comparison of the predicted results with analytical solutions, results published in the literature and by using the Method of Manufactured Solutions

A linear algebra approach to OLAP

Inspired by the relational algebra of data processing, this paper addresses the foundations of data analytical processing from a linear algebra perspective. The paper investigates, in particular, how aggregation operations such as cross tabulations and data cubes essential to quantitative analysis of data can be expressed solely in terms of matrix multiplication, transposition and the Khatri–Rao variant of the Kronecker product. The approach offers a basis for deriving an algebraic theory of data consolidation, handling the quantitative as well as qualitative sides of data science in a natural, elegant and typed way. It also shows potential for parallel analytical processing, as the parallelization theory of such matrix operations is well acknowledged.

Adaptive numerical simulation of reaction - diffusion systems

Fluidized beds, granulation, heat and mass transfer, calcium dynamics, stochastic process, finite element methods, Rosenbrock methods, multigrid methods, parallelization

Software for Explicitly Parallel Memory-Centric Processor Architecture

Advances in computer memory technology justify research towards new and different views on computer organization. This paper proposes a novel memory-centric computing architecture with the goal to merge memory and processing elements in order to provide better conditions for parallelization and performance. The paper introduces the architectural concepts and afterwards shows the design and implementation of a corresponding assembler and simulator.

A note on parallelizing the parameterized expectations algorithm

The parameterized expectations algorithm (PEA) involves a long simulation and a nonlinear least squares (NLS) fit, both embedded in a loop. Both steps are natural candidates for parallelization. This note shows that parallelization can lead to important speedups for the PEA. I provide example code for a simple model that can serve as a template for parallelization of more interesting models, as well as a download link for an image of a bootable CD that allows creation of a cluster and execution of the example code in minutes, with no need to install any software.

User-friendly parallel computations with econometric examples

This paper shows how a high level matrix programming language may be used to perform Monte Carlo simulation, bootstrapping, estimation by maximum likelihood and GMM, and kernel regression in parallel on symmetric multiprocessor computers or clusters of workstations. The implementation of parallelization is done in a way such that an investigator may use the programs without any knowledge of parallel programming. A bootable CD that allows rapid creation of a cluster for parallel computing is introduced. Examples show that parallelization can lead to important reductions in computational time. Detailed discussion of how the Monte Carlo problem was parallelized is included as an example for learning to write parallel programs for Octave.

Solució paral·lelitzada d'interpolació kriging amb ajust automatitzat del variograma

El principal objectiu d'aquest treball és proporcionar una metodologia per a reduir el temps de càlcul del mètode d'interpolació kriging sense pèrdua de la qualitat del model resultat. La solució adoptada ha estat la paral·lelització de l'algorisme mitjançant MPI sobre llenguatge C. Prèviament ha estat necessari automatitzar l'ajust del variograma que millor s'adapta a la distribució espacial de la variable d'estudi. Els resultats experimentals demostren la validesa de la solució implementada, en reduir de forma significativa els temps d'execució final de tot el procés.

Estudi GPGPU dedicat al processament de neuroimatges

La tecnologia GPGPU permet paral∙lelitzar càlculs executant operacions aritmètiques en els múltiples processadors de que disposen els xips gràfics. S'ha fet servir l'entorn de desenvolupament CUDA de la companyia NVIDIA, que actualment és la solució GPGPU més avançada del mercat. L'algorisme de neuroimatge implementat pertany a un estudi VBM desenvolupat amb l'eina SPM. Es tracta concretament del procés de segmentació d'imatges de ressonància magnètica cerebrals, en els diferents teixits dels quals es composa el cervell: matèria blanca, matèria grisa i líquid cefaloraquidi. S'han implementat models en els llenguatges Matlab, C i CUDA, i s'ha fet un estudi comparatiu per plataformes hardware diferents.

Estudi de comparació de rendiment de Benchmark multicore i multithreading amb OpenMP 2.5 i OpenMP 3.0

Estudi comparatiu amb benchmark del rendiment en dues plataformes multicore multithreading de diferents modalitats de paral·lelització de multiplicacions de matrius de nombres enters i de nombres en coma flotant mitjançant el model de memòria compartida OpenMP versió 2.5 i OpenMP versió 3.0.

Implementation of zonal control algorithms in adaptive optical systems with sparse control matrices

Miralls deformables més i més grans, amb cada cop més actuadors estan sent utilitzats actualment en aplicacions d'òptica adaptativa. El control dels miralls amb centenars d'actuadors és un tema de gran interès, ja que les tècniques de control clàssiques basades en la seudoinversa de la matriu de control del sistema es tornen massa lentes quan es tracta de matrius de dimensions tan grans. En aquesta tesi doctoral es proposa un mètode per l'acceleració i la paral.lelitzacó dels algoritmes de control d'aquests miralls, a través de l'aplicació d'una tècnica de control basada en la reducció a zero del components més petits de la matriu de control (sparsification), seguida de l'optimització de l'ordenació dels accionadors de comandament atenent d'acord a la forma de la matriu, i finalment de la seva posterior divisió en petits blocs tridiagonals. Aquests blocs són molt més petits i més fàcils de fer servir en els càlculs, el que permet velocitats de càlcul molt superiors per l'eliminació dels components nuls en la matriu de control. A més, aquest enfocament permet la paral.lelització del càlcul, donant una com0onent de velocitat addicional al sistema. Fins i tot sense paral. lelització, s'ha obtingut un augment de gairebé un 40% de la velocitat de convergència dels miralls amb només 37 actuadors, mitjançant la tècnica proposada. Per validar això, s'ha implementat un muntatge experimental nou complet , que inclou un modulador de fase programable per a la generació de turbulència mitjançant pantalles de fase, i s'ha desenvolupat un model complert del bucle de control per investigar el rendiment de l'algorisme proposat. Els resultats, tant en la simulació com experimentalment, mostren l'equivalència total en els valors de desviació després de la compensació dels diferents tipus d'aberracions per als diferents algoritmes utilitzats, encara que el mètode proposat aquí permet una càrrega computacional molt menor. El procediment s'espera que sigui molt exitós quan s'aplica a miralls molt grans.

Introducing massively parallel computers into operational weather forecasting

Numerical weather prediction and climate simulation have been among the computationally most demanding applications of high performance computing eversince they were started in the 1950's. Since the 1980's, the most powerful computers have featured an ever larger number of processors. By the early 2000's, this number is often several thousand. An operational weather model must use all these processors in a highly coordinated fashion. The critical resource in running such models is not computation, but the amount of necessary communication between the processors. The communication capacity of parallel computers often fallsfar short of their computational power. The articles in this thesis cover fourteen years of research into how to harness thousands of processors on a single weather forecast or climate simulation, so that the application can benefit as much as possible from the power of parallel high performance computers. The resultsattained in these articles have already been widely applied, so that currently most of the organizations that carry out global weather forecasting or climate simulation anywhere in the world use methods introduced in them. Some further studies extend parallelization opportunities into other parts of the weather forecasting environment, in particular to data assimilation of satellite observations.