Sparse matrix multiplication on a reconfigurable many-core architecture

Sparse matrix-vector multiplication (SMVM) is a fundamental operation in many scientific and engineering applications. In many cases sparse matrices have thousands of rows and columns where most of the entries are zero, while non-zero data is spread over the matrix. This sparsity of data locality reduces the effectiveness of data cache in general-purpose processors quite reducing their performance efficiency when compared to what is achieved with dense matrix multiplication. In this paper, we propose a parallel processing solution for SMVM in a many-core architecture. The architecture is tested with known benchmarks using a ZYNQ-7020 FPGA. The architecture is scalable in the number of core elements and limited only by the available memory bandwidth. It achieves performance efficiencies up to almost 70% and better performances than previous FPGA designs.

Uma aplicação Web para apoio a laboratórios remotos reconfiguráveis baseados na norma IEEE 1451.0

Nos últimos anos, o processo de ensino e aprendizagem tem sofrido significativas alterações graças ao aparecimento da Internet. Novas ferramentas para apoio ao ensino têm surgido, nas quais se destacam os laboratórios remotos. Atualmente, muitas instituições de ensino disponibilizam laboratórios remotos nos seus cursos, que permitem, a professores e alunos, a realização de experiências reais através da Internet. Estes são implementados por diferentes arquiteturas e infraestruturas, suportados por vários módulos de laboratório acessíveis remotamente (e.g. instrumentos de medição). No entanto, a sua inclusão no ensino é ainda deficitária, devido: i) à falta de meios e competências técnicas das instituições de ensino para os desenvolverem, ii) à dificuldade na partilha dos módulos de laboratório por diferentes infraestruturas e, iii) à reduzida capacidade de os reconfigurar com esses módulos. Para ultrapassar estas limitações, foi idealizado e desenvolvido no âmbito de um trabalho de doutoramento [1] um protótipo, cuja arquitetura é baseada na norma IEEE 1451.0 e na tecnologia de FPGAs. Para além de garantir o desenvolvimento e o acesso de forma normalizada a um laboratório remoto, este protótipo promove ainda a partilha de módulos de laboratório por diferentes infraestruturas. Nesse trabalho explorou-se a capacidade de reconfiguração de FPGAs para embutir na infraestrutura do laboratório vários módulos, todos descritos em ficheiros, utilizando linguagens de descrição de hardware estruturados de acordo com a norma IEEE 1451.0. A definição desses módulos obriga à criação de estruturas de dados binárias (Transducer Electronic Data Sheets, TEDSs), bem como de outros ficheiros que possibilitam a sua interligação com a infraestrutura do laboratório. No entanto, a criação destes ficheiros é bastante complexa, uma vez que exige a realização de vários cálculos e conversões. Tendo em consideração essa mesma complexidade, esta dissertação descreve o desenvolvimento de uma aplicação Web para leitura e escrita dos TEDSs. Para além de um estudo sobre os laboratórios remotos, é efetuada uma descrição da norma IEEE 1451.0, com particular atenção para a sua arquitetura e para a estrutura dos diferentes TEDSs. Com o objetivo de enquadrar a aplicação desenvolvida, efetua-se ainda uma breve apresentação de um protótipo de um laboratório remoto reconfigurável, cuja reconfiguração é apoiada por esta aplicação. Por fim, é descrita a verificação da aplicação Web, de forma a tirar conclusões sobre o seu contributo para a simplificação dessa reconfiguração.

Sistema de apoio à racionalização da utilização de energia eléctrica

Esta dissertação descreve o sistema de apoio à racionalização da utilização de energia eléctrica desenvolvido no âmbito da unidade curricular de Tese/Dissertação. O domínio de aplicação enquadra-se no contexto da Directiva da União Europeia 2006/32/EC que declara ser necessário colocar à disposição dos consumidores a informação e os meios que promovam a redução do consumo e o aumento da eficiência energética individual. O objectivo é o desenvolvimento de uma solução que permita a representação gráfica do consumo/produção, a definição de tectos de consumo, a geração automática de alertas e alarmes, a comparação anónima com clientes com perfil idêntico por região e a previsão de consumo/produção no caso de clientes industriais. Trata-se de um sistema distribuído composto por front-end e back-end. O front-end é composto pelas aplicações de interface com o utilizador desenvolvidas para dispositivos móveis Android e navegadores Web. O back-end efectua o armazenamento e processamento de informação e encontra-se alojado numa plataforma de cloud computing – o Google App Engine – que disponibiliza uma interface padrão do tipo serviço Web. Esta opção assegura interoperabilidade, escalabilidade e robustez ao sistema. Descreve-se em detalhe a concepção, desenvolvimento e teste do protótipo realizado, incluindo: (i) as funcionalidades de gestão e análise de consumo e produção de energia implementadas; (ii) as estruturas de dados; (iii) a base de dados e o serviço Web; e (iv) os testes e a depuração efectuados. (iv) Por fim, apresenta-se o balanço deste projecto e efectuam-se sugestões de melhoria.

Environmental analyses of the parasitic profile found in the sandy soil from the Santos municipality beaches, SP, Brazil

The environmental contamination by geohelminths represents a world public health problem and has been well documented by several authors. However, few papers describe the presence of such contamination in saline soils of coastal beaches. A study was performed on the beaches of the municipality of Santos in the period between May 2004 to April 2005 with the aim of determining the degree of contamination, and the correlation between contamination level and seasonal conditions and characteristics of the environment. Of the 2,520 samples analyzed, 18.2% (458) were contaminated, 32.3% (148) of which were localized in children's recreational areas (playgrounds). The parasite profile found in the analyzed samples indicated the presence of several zoonotic parasites: Ancylostoma larvae (82.5%), Toxocara sp. eggs (59.4%), Ancylostomidae-like eggs (37.1%), coccid oocysts (13.5%), Trichostrongylus sp. eggs and larvae, Ascaris lumbricoides eggs, (11.6%), Entamoeba sp. cysts (10.0%), Strongyloides sp. (4.8%), several free nematoids and some non-identified parasitic structures (3.3%). It was established that the highest frequency of parasitic structures occurred in the months between May and October 2004, and from February to March 2005. An increase in the diversity of parasitic forms was documented in the months between February to December 2004 and from January to April 2005, these periods having the highest rainfall.

Algorithmic skeleton framework for the orchestration of GPU computations

Dissertação para obtenção do Grau de Mestre em Engenharia Informática

A type system for value-dependent information flow analysis

Information systems are widespread and used by anyone with computing devices as well as corporations and governments. It is often the case that security leaks are introduced during the development of an application. Reasons for these security bugs are multiple but among them one can easily identify that it is very hard to define and enforce relevant security policies in modern software. This is because modern applications often rely on container sharing and multi-tenancy where, for instance, data can be stored in the same physical space but is logically mapped into different security compartments or data structures. In turn, these security compartments, to which data is classified into in security policies, can also be dynamic and depend on runtime data. In this thesis we introduce and develop the novel notion of dependent information flow types, and focus on the problem of ensuring data confidentiality in data-centric software. Dependent information flow types fit within the standard framework of dependent type theory, but, unlike usual dependent types, crucially allow the security level of a type, rather than just the structural data type itself, to depend on runtime values. Our dependent function and dependent sum information flow types provide a direct, natural and elegant way to express and enforce fine grained security policies on programs. Namely programs that manipulate structured data types in which the security level of a structure field may depend on values dynamically stored in other fields The main contribution of this work is an efficient analysis that allows programmers to verify, during the development phase, whether programs have information leaks, that is, it verifies whether programs protect the confidentiality of the information they manipulate. As such, we also implemented a prototype typechecker that can be found at http://ctp.di.fct.unl.pt/DIFTprototype/.

The effects of cultural heritage on residential property values: Evidence from Lisbon, Portugal

This paper examines the impact of historic amenities on residential housing prices in the city of Lisbon, Portugal. Our study is directed towards identifying the spatial variation of amenity values for churches, palaces, lithic (stone) architecture and other historic amenities via the housing market, making use of both global and local spatial hedonic models. Our empirical evidence reveals that different types of historic and landmark amenities provide different housing premiums. While having a local non-landmark church within 100 meters increases housing prices by approximately 4.2%, higher concentrations of non-landmark churches within 1000 meters yield negative effects in the order of 0.1% of prices with landmark churches having a greater negative impact around 3.4%. In contrast, higher concentration of both landmark and non-landmark lithic structures positively influence housing prices in the order of 2.9% and 0.7% respectively. Global estimates indicate a negative effect of protected zones, however this significance is lost when accounting for heterogeneity within these areas. We see that the designation of historic zones may counteract negative effects on property values of nearby neglected buildings in historic neighborhoods by setting additional regulations ensuring that dilapidated buildings do not damage the city’s beauty or erode its historic heritage. Further, our results from a geographically weighted regression specification indicate the presence of spatial non-stationarity in the effects of different historic amenities across the city of Lisbon with variation between historic and more modern areas.

Epidemic store for massive scale systems

The MAP-i doctoral program of the Universities of Minho, Aveiro and Porto

BIOMedical search engine framework: lightweight and customized implementation of domain-specific biomedical search engines

The Smart Drug Search is publicly accessible at http://sing.ei.uvigo.es/sds/. The BIOMedical Search Engine Framework is freely available for non-commercial use at https://github.com/agjacome/biomsef

Constrained school choice

Recently, several school districts in the US have adopted or consider adopting the Student-Optimal Stable Mechanism or the Top Trading Cycles Mechanism to assign children to public schools. There is clear evidence that for school districts that employ (variants of) the so-called Boston Mechanism the transition would lead to efficiency gains. The first two mechanisms are strategy-proof, but in practice student assignment procedures impede students to submit a preference list that contains all their acceptable schools. Therefore, any desirable property of the mechanisms is likely toget distorted. We study the non trivial preference revelation game where students can only declare up to a fixed number (quota) of schools to be acceptable. We focus on the stability of the Nash equilibrium outcomes. Our main results identify rather stringent necessary and sufficient conditions on the priorities to guaranteestability. This stands in sharp contrast with the Boston Mechanism which yields stable Nash equilibrium outcomes, independently of the quota. Hence, the transition to any of the two mechanisms is likely to come with a higher risk that students seek legal actionas lower priority students may occupy more preferred schools.

Equilibrium play and best response to (Stated) beliefs in constant sum games

We report experimental results on one-shot two person 3x3 constant sum games played by non-economists without previous experience in the laboratory. Although strategically our games are very similar to previous experiments in which game theory predictions fail dramatically, 80% of actions taken in our experiment coincided with the prediction of the unique Nash equilibrium in pure strategies and 73% of actions were best responses to elicited beliefs. We argue how social preferences, presentation effects and belief elicitation procedures may influence how subjects play in simple but non trivial games and explain the diferences we observe with respect to previous work.

Integració de tècniques per a la comprovació eficient de restriccions d'integritat en la generació de programari per entorns web

Estudi elaborat a partir d’una estada al Politecnico de Milano, Itàlia, entre gener i juny del 2006. Un dels principals objectius de l’Enginyeria del Programari és automatitzar el màxim possible el procés de desenvolupament del programari, reduint costos mitjançant la generació automàtica del programari a partir de la seva especificació. Per assolir-ho, entre altres, cal resoldre el problema de la comprovació eficient de restriccions, que són una part fonamental de l’especificació del programari. Aquest és precisament l’àmbit en què s’està desenvolupant una tesi que presentarà un mètode que poden integrar totes les eines generadores de codi per tal d’assolir una implementació eficient de les restriccions d’integritat. En l’actual fase del projecte s’ha treballat per validar el mètode de la tesi, optimitzant-lo pel cas específic de les aplicacions web i estendre’l per poder tractar també aplicacions basades en workflows. Pel que fa a l’optimització del mètode per aplicacions web, s’han definit una sèrie de paràmetres que permeten configurar la implementació del mètode tenint en compte les necessitats específiques de rendiment de cada aplicació web en particular. Respecte als workflows (cada cop més populars i que s’usen com a definició d’alt nivell per a les aplicacions a desenvolupar) s’ha estudiat quins són els tipus de restriccions que impliquen i com després es pot aplicar el mètode de la tesi sobre aquestes restriccions per tal de generar de forma eficient també les aplicacions basades en workflows.

The Inflation Risk Premium on Government Debt in an Overlapping Generations Model

This paper presents a general equilibrium model in which nominal government debt pays an inflation risk premium. The model predicts that the inflation risk premium will be higher in economies which are exposed to unanticipated inflation through nominal asset holdings. In particular, the inflation risk premium is higher when government debt is primarily nominal, steady-state inflation is low, and when cash and nominal debt account for a large fraction of consumers' retirement portfolios. These channels do not appear to have been highlighted in previous models or tested empirically. Numerical results suggest that the inflation risk premium is comparable in magnitude to standard representative agent models. These findings have implications for management of government debt, since the inflation risk premium makes it more costly for governments to borrow using nominal rather than indexed debt. Simulations of an extended model with Epstein-Zin preferences suggest that increasing the share of indexed debt would enable governments to permanently lower taxes by an amount that is quantitatively non-trivial.

The Inflation Bias under Calvo and Rotemberg Pricing.

New Keynesian models rely heavily on two workhorse models of nominal inertia - price contracts of random duration (Calvo, 1983) and price adjustment costs (Rotemberg, 1982) - to generate a meaningful role for monetary policy. These alternative descriptions of price stickiness are often used interchangeably since, to a first order of approximation they imply an isomorphic Phillips curve and, if the steady-state is efficient, identical objectives for the policy maker and as a result in an LQ framework, the same policy conclusions. In this paper we compute time-consistent optimal monetary policy in bench-mark New Keynesian models containing each form of price stickiness. Using global solution techniques we find that the inflation bias problem under Calvo contracts is significantly greater than under Rotemberg pricing, despite the fact that the former typically significant exhibits far greater welfare costs of inflation. The rates of inflation observed under this policy are non-trivial and suggest that the model can comfortably generate the rates of inflation at which the problematic issues highlighted in the trend inflation literature emerge, as well as the movements in trend inflation emphasized in empirical studies of the evolution of inflation. Finally, we consider the response to cost push shocks across both models and find these can also be significantly different. The choice of which form of nominal inertia to adopt is not innocuous.

A Multiple Break Panel Approach to Estimating United States Phillips Curves

Phillips curves are often estimated without due attention being paid to the underlying time series properties of the data. In particular, the consequences of inflation having discrete breaks in mean have not been studied adequately. We show by means of simulations and a detailed empirical example based on United States data that not taking account of breaks may lead to biased, and therefore spurious, estimates of Phillips curves. We suggest a method to account for the breaks in mean inflation and obtain meaningful and unbiased estimates of the short- and long-run Phillips curves in the United States.