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.

Os efeitos das variações das taxas de juro na volatilidade dos mercados bolsistas

Mestrado em Controlo de Gestão e dos Negócios

Water quality issues in rivers, lakes and reservoirs in Asian countries: an approach to the problems and the methodologies

Water covers over 70% of the Earth's surface, and is vital for all known forms of life. But only 3% of the Earth's water is fresh water, and less than 0.3% of all freshwater is in rivers, lakes, reservoirs and the atmosphere. However, rivers and lakes are an important part of fresh surface water, amounting to about 89%. In this Master Thesis dissertation, the focus is on three types of water bodies – rivers, lakes and reservoirs, and their water quality issues in Asian countries. The surface water quality in a region is largely determined both by the natural processes such as climate or geographic conditions, and the anthropogenic influences such as industrial and agricultural activities or land use conversion. The quality of the water can be affected by pollutants discharge from a specific point through a sewer pipe and also by extensive drainage from agriculture/urban areas and within basin. Hence, water pollutant sources can be divided into two categories: Point source pollution and Non-point source (NPS) pollution. Seasonal variations in precipitation and surface run-off have a strong effect on river discharge and the concentration of pollutants in water bodies. For example, in the rainy season, heavy and persistent rain wash off the ground, the runoff flow increases and may contain various kinds of pollutants and, eventually, enters the water bodies. In some cases, especially in confined water bodies, the quality may be positive related with rainfall in the wet season, because this confined type of fresh water systems allows high dilution of pollutants, decreasing their possible impacts. During the dry season, the quality of water is largely related to industrialization and urbanization pollution. The aim of this study is to identify the most common water quality problems in Asian countries and to enumerate and analyze the methodologies used for assessment of water quality conditions of both rivers and confined water bodies (lakes and reservoirs). Based on the evaluation of a sample of 57 papers, dated between 2000 and 2012, it was found that over the past decade, the water quality of rivers, lakes, and reservoirs in developing countries is being degraded. Water pollution and destruction of aquatic ecosystems have caused massive damage to the functions and integrity of water resources. The most widespread NPS in Asian countries and those which have the greatest spatial impacts are urban runoff and agriculture. Locally, mine waste runoff and rice paddy are serious NPS problems. The most relevant point pollution sources are the effluents from factories, sewage treatment plant, and public or household facilities. It was found that the most used methodology was unquestionably the monitoring activity, used in 49 of analyzed studies, accounting for 86%. Sometimes, data from historical databases were used as well. It can be seen that taking samples from the water body and then carry on laboratory work (chemical analyses) is important because it can give an understanding of the water quality. 6 papers (11%) used a method that combined monitoring data and modeling. 6 papers (11%) just applied a model to estimate the quality of water. Modeling is a useful resource when there is limited budget since some models are of free download and use. In particular, several of used models come from the U.S.A, but they have their own purposes and features, meaning that a careful application of the models to other countries and a critical discussion of the results are crucial. 5 papers (9%) focus on a method combining monitoring data and statistical analysis. When there is a huge data matrix, the researchers need an efficient way of interpretation of the information which is provided by statistics. 3 papers (5%) used a method combining monitoring data, statistical analysis and modeling. These different methods are all valuable to evaluate the water quality. It was also found that the evaluation of water quality was made as well by using other types of sampling different than water itself, and they also provide useful information to understand the condition of the water body. These additional monitoring activities are: Air sampling, sediment sampling, phytoplankton sampling and aquatic animal tissues sampling. Despite considerable progress in developing and applying control regulations to point and NPS pollution, the pollution status of rivers, lakes, and reservoirs in Asian countries is not improving. In fact, this reflects the slow pace of investment in new infrastructure for pollution control and growing population pressures. Water laws or regulations and public involvement in enforcement can play a constructive and indispensable role in environmental protection. In the near future, in order to protect water from further contamination, rapid action is highly needed to control the various kinds of effluents in one region. Environmental remediation and treatment of industrial effluent and municipal wastewaters is essential. It is also important to prevent the direct input of agricultural and mine site runoff. Finally, stricter environmental regulation for water quality is required to support protection and management strategies. It would have been possible to get further information based in the 57 sample of papers. For instance, it would have been interesting to compare the level of concentrations of some pollutants in the diferente Asian countries. However the limit of three months duration for this study prevented further work to take place. In spite of this, the study objectives were achieved: the work provided an overview of the most relevant water quality problems in rivers, lakes and reservoirs in Asian countries, and also listed and analyzed the most common methodologies.

Anomaly-free U(1) gauge symmetries in neutrino seesaw flavor models

Agência Financiadora: Fundação para a Ciência e a Tecnologia (FCT) - PEst-OE/FIS/UI0777/2013; CERN/FP/123580/2011; PTDC/FIS-NUC/0548/2012

Energy-Aware Routing And Grooming For IP Transport Over WDM MLR Networks

In this work a mixed integer optimization linear programming (MILP) model was applied to mixed line rate (MLR) IP over WDM and IP over OTN over WDM (with and without OTN grooming) networks, with aim to reduce network energy consumption. Energy-aware and energy-aware & short-path routing techniques were used. Simulations were made based on a real network topology as well as on forecasts of traffic matrix based on statistical data from 2005 up to 2017. Energy aware routing optimization model on IPoWDM network, showed the lowest energy consumption along all years, and once compared with energy-aware & short-path routing, has led to an overall reduction in energy consumption up to 29%, expecting to save even more than shortest-path routing. © 2014 IEEE.

Probabilistic consensus clustering using evidence accumulation

Clustering ensemble methods produce a consensus partition of a set of data points by combining the results of a collection of base clustering algorithms. In the evidence accumulation clustering (EAC) paradigm, the clustering ensemble is transformed into a pairwise co-association matrix, thus avoiding the label correspondence problem, which is intrinsic to other clustering ensemble schemes. In this paper, we propose a consensus clustering approach based on the EAC paradigm, which is not limited to crisp partitions and fully exploits the nature of the co-association matrix. Our solution determines probabilistic assignments of data points to clusters by minimizing a Bregman divergence between the observed co-association frequencies and the corresponding co-occurrence probabilities expressed as functions of the unknown assignments. We additionally propose an optimization algorithm to find a solution under any double-convex Bregman divergence. Experiments on both synthetic and real benchmark data show the effectiveness of the proposed approach.

A MAP approach to evidence accumulation clustering

The Evidence Accumulation Clustering (EAC) paradigm is a clustering ensemble method which derives a consensus partition from a collection of base clusterings obtained using different algorithms. It collects from the partitions in the ensemble a set of pairwise observations about the co-occurrence of objects in a same cluster and it uses these co-occurrence statistics to derive a similarity matrix, referred to as co-association matrix. The Probabilistic Evidence Accumulation for Clustering Ensembles (PEACE) algorithm is a principled approach for the extraction of a consensus clustering from the observations encoded in the co-association matrix based on a probabilistic model for the co-association matrix parameterized by the unknown assignments of objects to clusters. In this paper we extend the PEACE algorithm by deriving a consensus solution according to a MAP approach with Dirichlet priors defined for the unknown probabilistic cluster assignments. In particular, we study the positive regularization effect of Dirichlet priors on the final consensus solution with both synthetic and real benchmark data.

Hyperspectral unmixing algorithm via dependent component analysis

This paper introduces a new method to blindly unmix hyperspectral data, termed dependent component analysis (DECA). This method decomposes a hyperspectral images into a collection of reflectance (or radiance) spectra of the materials present in the scene (endmember signatures) and the corresponding abundance fractions at each pixel. DECA assumes that each pixel is a linear mixture of the endmembers signatures weighted by the correspondent abundance fractions. These abudances are modeled as mixtures of Dirichlet densities, thus enforcing the constraints on abundance fractions imposed by the acquisition process, namely non-negativity and constant sum. The mixing matrix is inferred by a generalized expectation-maximization (GEM) type algorithm. This method overcomes the limitations of unmixing methods based on Independent Component Analysis (ICA) and on geometrical based approaches. The effectiveness of the proposed method is illustrated using simulated data based on U.S.G.S. laboratory spectra and real hyperspectral data collected by the AVIRIS sensor over Cuprite, Nevada.