Optimal Piecewise Linear Function Approximation for GPU-based Applications

Many computer vision and human-computer interaction applications developed in recent years need evaluating complex and continuous mathematical functions as an essential step toward proper operation. However, rigorous evaluation of this kind of functions often implies a very high computational cost, unacceptable in real-time applications. To alleviate this problem, functions are commonly approximated by simpler piecewise-polynomial representations. Following this idea, we propose a novel, efficient, and practical technique to evaluate complex and continuous functions using a nearly optimal design of two types of piecewise linear approximations in the case of a large budget of evaluation subintervals. To this end, we develop a thorough error analysis that yields asymptotically tight bounds to accurately quantify the approximation performance of both representations. It provides an improvement upon previous error estimates and allows the user to control the trade-off between the approximation error and the number of evaluation subintervals. To guarantee real-time operation, the method is suitable for, but not limited to, an efficient implementation in modern Graphics Processing Units (GPUs), where it outperforms previous alternative approaches by exploiting the fixed-function interpolation routines present in their texture units. The proposed technique is a perfect match for any application requiring the evaluation of continuous functions, we have measured in detail its quality and efficiency on several functions, and, in particular, the Gaussian function because it is extensively used in many areas of computer vision and cybernetics, and it is expensive to evaluate.

Modelo de escoamento multifásico para estudo da interação onda/sedimento

Modelos de escoamento multifásico são amplamente usados em diversas áreas de pesquisa ambiental, como leitos fluidizados, dispersão de gás em líquidos e vários outros processos que englobam mais de uma propriedade físico-química do meio. Dessa forma, um modelo multifásico foi desenvolvido e adaptado para o estudo do transporte de sedimentos de fundo devido à ação de ondas de gravidade. Neste trabalho, foi elaborado o acoplamento multifásico de um modelo euleriano não-linear de ondas do tipo Boussinesq, baseado na formulação numérica encontrada em Wei et al. (1995), com um modelo lagrangiano de partículas, fundamentado pelo princípio Newtoniano do movimento com o esquema de colisões do tipo esferas rígidas. O modelo de ondas foi testado quanto à sua fonte geradora, representada por uma função gaussiana, pá-pistão e pá-batedor, e quanto à sua interação com a profundidade, através da não-linearidade e de propriedades dispersivas. Nos testes realizados da fonte geradora, foi observado que a fonte gaussiana, conforme Wei et al. (1999), apresentou melhor consistência e estabilidade na geração das ondas, quando comparada à teoria linear para um kh   . A não-linearidade do modelo de ondas de 2ª ordem para a dispersão apresentou resultados satisfatórios quando confrontados com o experimento de ondas sobre um obstáculo trapezoidal, onde a deformação da onda sobre a estrutura submersa está em concordância com os dados experimentais encontrados na literatura. A partir daí, o modelo granular também foi testado em dois experimentos. O primeiro simula uma quebra de barragem em um tanque contendo água e o segundo, a quebra de barragem é simulada com um obstáculo rígido adicionado ao centro do tanque. Nesses experimentos, o algoritmo de colisão foi eficaz no tratamento da interação entre partícula-partícula e partícula-parede, permitindo a evidência de processos físicos que são complicados de serem simulados por modelos de malhas regulares. Para o acoplamento do modelo de ondas e de sedimentos, o algoritmo foi testado com base de dados da literatura quanto à morfologia do leito. Os resultados foram confrontados com dados analíticos e de modelos numéricos, e se mostraram satisfatórios com relação aos pontos de erosão, de sedimentação e na alteração da forma da barra arenosa

Heat recovery from wastewater: numerical modelling of sewer systems

Dissertação para obtenção do Grau de Mestre em Engenharia do Ambiente, perfil Engenharia Sanitária

Estimating the colors of paintings

Observers can adjust the spectrum of illumination on paintings for optimal viewing experience. But can they adjust the colors of paintings for the best visual impression? In an experiment carried out on a calibrated color moni- tor images of four abstract paintings obtained from hyperspectral data were shown to observers that were unfamiliar with the paintings. The color volume of the images could be manipulated by rotating the volume around the axis through the average (a*, b*) point for each painting in CIELAB color space. The task of the observers was to adjust the angle of rotation to produce the best subjective impression from the paintings. It was found that the distribution of angles selected for data pooled across paintings and observers could be de- scribed by a Gaussian function centered at 10o, i.e. very close to the original colors of the paintings. This result suggest that painters are able to predict well what compositions of colors observers prefer.

Sintonização de estados quânticos: um estudo numérico do oscilador harmônico quântico

The quantum harmonic oscillator is described by the Hermite equation.¹ The asymptotic solution is predominantly used to obtain its analytical solutions. Wave functions (solutions) are quadratically integrable if taken as the product of the convergent asymptotic solution (Gaussian function) and Hermite polynomial,¹ whose degree provides the associated quantum number. Solving it numerically, quantization is observed when a control real variable is "tuned" to integer values. This can be interpreted by graphical reading of Y(x) and |Y(x)|², without other mathematical analysis, and prove useful for teaching fundamentals of quantum chemistry to undergraduates.

Constructing the frequency and wave normal distribution of whistler-mode wave power

We introduce a new methodology that allows the construction of wave frequency distributions due to growing incoherent whistler-mode waves in the magnetosphere. The technique combines the equations of geometric optics (i.e. raytracing) with the equation of transfer of radiation in an anisotropic lossy medium to obtain spectral energy density as a function of frequency and wavenormal angle. We describe the method in detail, and then demonstrate how it could be used in an idealised magnetosphere during quiet geomagnetic conditions. For a specific set of plasma conditions, we predict that the wave power peaks off the equator at ~15 degrees magnetic latitude. The new calculations predict that wave power as a function of frequency can be adequately described using a Gaussian function, but as a function of wavenormal angle, it more closely resembles a skew normal distribution. The technique described in this paper is the first known estimate of the parallel and oblique incoherent wave spectrum as a result of growing whistler-mode waves, and provides a means to incorporate self-consistent wave-particle interactions in a kinetic model of the magnetosphere over a large volume.

Thickness and Annealing Temperature Effects on the Optical Properties and Surface Morphology of Layer-by-Layer Poly(p-phenyline vinylene) plus Dodecylbenzenesulfonate Films

The fabrication of controlled molecular architectures is essential for organic devices, as is the case of emission of polarized light for the information industry. In this study, we show that optimized conditions can be established to allow layer-by-layer (LbL) films of poly(p-phenylene vinylene) (PPV)+dodecylbenzenesulfonate (DBS) to be obtained with anisotropic properties. Films with five layers and converted at 110 degrees C had a dichroic ratio delta = 2.3 and order parameter r = 34%, as indicated in optical spectroscopy and emission ellipsometry data. This anisotropy was decreased with the number of layers deposited, with delta = 1.0 for a 75-layer LbL PPV + DBS film. The analysis with atomic force microscopy showed the formation of polymer clusters in a random growth process with the normalized height distribution being represented by a Gaussian function. In spite of this randomness in film growth, the self-covariance function pointed to a correlation between clusters, especially for thick films. In summary, the LbL method may be exploited to obtain both anisotropic films with polarized emission and regular, nanostructured surfaces. (c) 2010 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys 49: 206-213, 2011

An analog implementation of radial basis neural networks (RBNN) using BiCMOS technology

This paper describes a analog implementation of radial basis neural networks (RBNN) in BiCMOS technology. The RBNN uses a gaussian function obtained through the characteristic of the bipolar differential pair. The gaussian parameters (gain, center and width) is changed with programmable current source. Results obtained with PSPICE software is showed.

Investigation of the dipole response of nickel isotopes in the presence of a high-frequency electromagnetic field

The electric dipole response of neutron-rich nickel isotopes has been investigated using the LAND setup at GSI in Darmstadt (Germany). Relativistic secondary beams of 56−57Ni and 67−72Ni at approximately 500 AMeV have been generated using projectile fragmentation of stable ions on a 4 g/cm2 Be target and subsequent separation in the magnetic dipole fields of the FRagment Separator (FRS). After reaching the LAND setup in Cave C, the radioactive ions were excited electromagnetically in the electric field of a Pb target. The decay products have been measured in inverse kinematics using various detectors. Neutron-rich 67−69Ni isotopes decay by the emission of neutrons, which are detected in the LAND detector. The present analysis concentrates on the (gamma,n) and (gamma,2n) channels in these nuclei, since the proton and three-neutron thresholds are unlikely to be reached considering the virtual photon spectrum for nickel ions at 500 AMeV. A measurement of the stable 58Ni isotope is used as a benchmark to check the accuracy of the present results with previously published data. The measured (gamma,n) and (gamma,np) channels are compared with an inclusive photoneutron measurement by Fultz and coworkers, which are consistent within the respective errors. The measured excitation energy distributions of 67−69Ni contain a large portion of the Giant Dipole Resonance (GDR) strength predicted by the Thomas-Reiche-Kuhn energy-weighted sum rule, as well as a significant amount of low-lying E1 strength, that cannot be attributed to the GDR alone. The GDR distribution parameters are calculated using well-established semi-empirical systematic models, providing the peak energies and widths. The GDR strength is extracted from the chi-square minimization of the model GDR to the measured data of the (gamma,2n) channel, thereby excluding any influence of eventual low-lying strength. The subtraction of the obtained GDR distribution from the total measured E1 strength provides the low-lying E1 strength distribution, which is attributed to the Pygmy Dipole Resonance (PDR). The extraction of the peak energy, width and strength is performed using a Gaussian function. The minimization of trial Gaussian distributions to the data does not converge towards a sharp minimum. Therefore, the results are presented by a chi-square distribution as a function of all three Gaussian parameters. Various predictions of PDR distributions exist, as well as a recent measurement of the 68Ni pygmy dipole-resonance obtained by virtual photon scattering, to which the present pygmy dipole-resonance distribution is also compared.

Measurement of the distributions of event-by-event flow harmonics in lead-lead collisions at = 2.76 TeV with the ATLAS detector at the LHC

The distributions of event-by-event harmonic flow coefficients v_n for n=2-4 are measured in sqrt(s_NN)=2.76 TeV Pb+Pb collisions using the ATLAS detector at the LHC. The measurements are performed using charged particles with transverse momentum pT> 0.5 GeV and in the pseudorapidity range |eta|<2.5 in a dataset of approximately 7 ub^-1 recorded in 2010. The shapes of the v_n distributions are described by a two-dimensional Gaussian function for the underlying flow vector in central collisions for v_2 and over most of the measured centrality range for v_3 and v_4. Significant deviations from this function are observed for v_2 in mid-central and peripheral collisions, and a small deviation is observed for v_3 in mid-central collisions. It is shown that the commonly used multi-particle cumulants are insensitive to the deviations for v_2. The v_n distributions are also measured independently for charged particles with 0.51 GeV. When these distributions are rescaled to the same mean values, the adjusted shapes are found to be nearly the same for these two pT ranges. The v_n distributions are compared with the eccentricity distributions from two models for the initial collision geometry: a Glauber model and a model that includes corrections to the initial geometry due to gluon saturation effects. Both models fail to describe the experimental data consistently over most of the measured centrality range.

Identification of visual field progression

The study developed statistical techniques to evaluate visual field progression for use with the Humphrey Field Analyzer (HFA). The long-term fluctuation (LF) was evaluated in stable glaucoma. The magnitude of both LF components showed little relationship with MD, CPSD and SF. An algorithm was proposed for determining the clinical necessity for a confirmatory follow-up examination. The between-examination variability was determined for the HFA Standard and FASTPAC algorithms in glaucoma. FASTPAC exhibited greater between-examination variability than the Standard algorithm across the range of sensitivities and with increasing eccentricity. The difference in variability between the algorithms had minimal clinical significance. The effect of repositioning the baseline in the Glaucoma Change Probability Analysis (GCPA) was evaluated. The global baseline of the GCPA limited the detection of progressive change at a single stimulus location. A new technique, pointwise univariate linear regressions (ULR), of absolute sensitivity and, of pattern deviation, against time to follow-up was developed. In each case, pointwise ULR was more sensitive to localised progressive changes in sensitivity than ULR of MD, alone. Small changes in sensitivity were more readily determined by the pointwise ULR than by the GCPA. A comparison between the outcome of pointwise ULR for all fields and for the last six fields manifested linear and curvilinear declines in the absolute sensitivity and the pattern deviation. A method for delineating progressive loss in glaucoma, based upon the error in the forecasted sensitivity of a multivariate model, was developed. Multivariate forecasting exhibited little agreement with GCPA in glaucoma but showed promise for monitoring visual field progression in OHT patients. The recovery of sensitivity in optic neuritis over time was modelled with a Cumulative Gaussian function. The rate and level of recovery was greater in the peripheral than the central field. Probability models to forecast the field of recovery were proposed.

The development and stability analysis of a nonlinear growth model for microorganisms

A nonlinear dynamic model of microbial growth is established based on the theories of the diffusion response of thermodynamics and the chemotactic response of biology. Except for the two traditional variables, i.e. the density of bacteria and the concentration of attractant, the pH value, a crucial influencing factor to the microbial growth, is also considered in this model. The pH effect on the microbial growth is taken as a Gaussian function G0e-(f- fc)2/G1, where G0, G1 and fc are constants, f represents the pH value and fc represents the critical pH value that best fits for microbial growth. To study the effects of the reproduction rate of the bacteria and the pH value on the stability of the system, three parameters a, G0 and G1 are studied in detail, where a denotes the reproduction rate of the bacteria, G0 denotes the impacting intensity of the pH value to microbial growth and G1 denotes the bacterial adaptability to the pH value. When the effect of the pH value of the solution which microorganisms live in is ignored in the governing equations of the model, the microbial system is more stable with larger a. When the effect of the bacterial chemotaxis is ignored, the microbial system is more stable with the larger G1 and more unstable with the larger G0 for f0 > fc. However, the stability of the microbial system is almost unaffected by the variation G0 and G1 and it is always stable for f0 < fc under the assumed conditions in this paper. In the whole system model, it is more unstable with larger G1 and more stable with larger G0 for f0 < fc. The system is more stable with larger G1 and more unstable with larger G0 for f0 > fc. However, the system is more unstable with larger a for f0 < fc and the stability of the system is almost unaffected by a for f0 > fc. The results obtained in this study provide a biophysical insight into the understanding of the growth and stability behavior of microorganisms.

Comparing Support Vector Machines with Gaussian Kernels to Radial Basis Function Classifiers

The Support Vector (SV) machine is a novel type of learning machine, based on statistical learning theory, which contains polynomial classifiers, neural networks, and radial basis function (RBF) networks as special cases. In the RBF case, the SV algorithm automatically determines centers, weights and threshold such as to minimize an upper bound on the expected test error. The present study is devoted to an experimental comparison of these machines with a classical approach, where the centers are determined by $k$--means clustering and the weights are found using error backpropagation. We consider three machines, namely a classical RBF machine, an SV machine with Gaussian kernel, and a hybrid system with the centers determined by the SV method and the weights trained by error backpropagation. Our results show that on the US postal service database of handwritten digits, the SV machine achieves the highest test accuracy, followed by the hybrid approach. The SV approach is thus not only theoretically well--founded, but also superior in a practical application.

Robust initialisation of Gaussian radial basis function networks using partitioned k-means clustering

Radial basis function networks can be trained quickly using linear optimisation once centres and other associated parameters have been initialised. The authors propose a small adjustment to a well accepted initialisation algorithm which improves the network accuracy over a range of problems. The algorithm is described and results are presented.