A radial basis function network (RBFN) for function approximation

A radial basis function network (RBFN) circuit for function approximation is presented. Simulation and experimental results show that the network has good approximation capabilities. The RBFN was a squared hyperbolic secant with three adjustable parameters amplitude, width and center. To test the network a sinusoidal and sine function,vas approximated.

Comparative study between powers of sigmoid functions, MLP-backpropagation and polynomials in function approximation problems

Function approximation is a very important task in environments where the computation has to be based on extracting information from data samples in real world processes. So, the development of new mathematical model is a very important activity to guarantee the evolution of the function approximation area. In this sense, we will present the Polynomials Powers of Sigmoid (PPS) as a linear neural network. In this paper, we will introduce one series of practical results for the Polynomials Powers of Sigmoid, where we will show some advantages of the use of the powers of sigmiod functions in relationship the traditional MLP-Backpropagation and Polynomials in functions approximation problems.

Monotonicity of zeros of Jacobi polynomials

Denote by x(nk)(alpha, beta), k = 1...., n, the zeros of the Jacobi polynornial P-n((alpha,beta)) (x). It is well known that x(nk)(alpha, beta) are increasing functions of beta and decreasing functions of alpha. In this paper we investigate the question of how fast the functions 1 - x(nk)(alpha, beta) decrease as beta increases. We prove that the products t(nk)(alpha, beta) := f(n)(alpha, beta) (1 - x(nk)(alpha, beta), where f(n)(alpha, beta) = 2n(2) + 2n(alpha + beta + 1) + (alpha + 1)(beta + 1) are already increasing functions of beta and that, for any fixed alpha > - 1, f(n)(alpha, beta) is the asymptotically extremal, with respect to n, function of beta that forces the products t(nk)(alpha, beta) to increase. (c) 2007 Elsevier B.V. All rights reserved.

Almost strict total positivity and a class of Hurwitz polynomials

We establish sufficient conditions for a matrix to be almost totally positive, thus extending a result of Craven and Csordas who proved that the corresponding conditions guarantee that a matrix is strictly totally positive. Then we apply our main result in order to obtain a new criteria for a real algebraic polynomial to be a Hurwitz one. The properties of the corresponding extremal Hurwitz polynomials are discussed. (C) 2004 Elsevier B.V. All rights reserved.

On the behaviour of zeros of Jacobi polynomials

Denote by x(n,k)(alpha, beta) and x(n,k) (lambda) = x(n,k) (lambda - 1/2, lambda - 1/2) the zeros, in decreasing order, of the Jacobi polynomial P-n((alpha, beta))(x) and of the ultraspherical (Gegenbauer) polynomial C-n(lambda)(x), respectively. The monotonicity of x(n,k)(alpha, beta) as functions of a and beta, alpha, beta > - 1, is investigated. Necessary conditions such that the zeros of P-n((a, b)) (x) are smaller (greater) than the zeros of P-n((alpha, beta))(x) are provided. A. Markov proved that x(n,k) (a, b) < x(n,k)(α, β) (x(n,k)(a, b) > x(n,k)(alpha, beta)) for every n is an element of N and each k, 1 less than or equal to k less than or equal to n if a > alpha and b < β (a < alpha and b > beta). We prove the converse statement of Markov's theorem. The question of how large the function could be such that the products f(n)(lambda) x(n,k)(lambda), k = 1,..., [n/2] are increasing functions of lambda, for lambda > - 1/2, is also discussed. Elbert and Siafarikas proved that f(n)(lambda) = (lambda + (2n(2) + 1)/ (4n + 2))(1/2) obeys this property. We establish the sharpness of their result. (C) 2002 Elsevier B.V. (USA).

Sharp bounds for the extreme zeros of classical orthogonal polynomials

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Asymptotics for Jacobi-Sobolev orthogonal polynomials associated with non-coherent pairs of measures

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Blumenthal's theorem for Laurent orthogonal polynomials

We investigate polynomials satisfying a three-term recurrence relation of the form B-n(x) = (x - beta(n))beta(n-1)(x) - alpha(n)xB(n-2)(x), with positive recurrence coefficients alpha(n+1),beta(n) (n = 1, 2,...). We show that the zeros are eigenvalues of a structured Hessenberg matrix and give the left and right eigenvectors of this matrix, from which we deduce Laurent orthogonality and the Gaussian quadrature formula. We analyse in more detail the case where alpha(n) --> alpha and beta(n) --> beta and show that the zeros of beta(n) are dense on an interval and that the support of the Laurent orthogonality measure is equal to this interval and a set which is at most denumerable with accumulation points (if any) at the endpoints of the interval. This result is the Laurent version of Blumenthal's theorem for orthogonal polynomials. (C) 2002 Elsevier B.V. (USA).

The stability of the Johnson-Mehl-Avrami equation parameters

The stability of the parameters of the Johnson-Mehl-Avrami equation was studied using two parametrizations of the sigmoidal function and its fit to some kinetic data. The results indicate that one of the forms of the function has more stable parameters and only for this form it is reasonable to use, as an approximation, the linear regression theory to analyse the parameters. © 1995 Chapman & Hall.

Modelling of the influence of the layer pollution thickness in high voltage polluted insulators

An experimental model and a mathematical model with the introduction of a ramp in the channel of Obenaus model are presented. The aim is to present a better reproduction of the real layer pollution deposited on the HV insulators. This better reproduction is obtained from two types of thickness variation: the introduction of a ramp (soft variation) and the introduction of a step (sudden variation). The computational simulations and the experimental data suggest that the introduction of the ramp is the better reproduction of the layer pollution. The ramp approximates to the real layer pollution more than the step.

Numerical simulation of viscous three-dimensional flow over aerospace vehicles using Euler/boundary-layer zonal approach

This paper presents a viscous three-dimensional simulations coupling Euler and boundary layer codes for calculating flows over arbitrary surfaces. The governing equations are written in a general non orthogonal coordinate system. The Levy-Lees transformation generalized to three-dimensional flows is utilized. The inviscid properties are obtained from the Euler equations using the Beam and Warming implicit approximate factorization scheme. The resulting equations are discretized and approximated by a two-point fmitedifference numerical scheme. The code developed is validated and applied to the simulation of the flowfield over aerospace vehicle configurations. The results present good correlation with the available data.

Judd-Ofelt analysis of Pr3+ ions in fluoroindate glasses: Influence of odd third order intensity parameters

In this work an analysis of the Judd-Ofelt phenomenological Ωλ intensity parameters for the Pr3+ ion in fluoroindate glass is made. Different Pr3+ concentrations, namely 1, 2, 3 and 4 mol% are used. The experimental oscillator strengths have been determined from the absorption spectra. A consistent set of parameters is obtained only with the inclusion of odd rank third order intensity parameters and if the band at 21 470 cm-1 is assigned to the 3H4 → 3P1 transition and the 1I6 component is incorporated in the 3H4 → 3P2 transition at 22 700 cm-1.

Numerical and experimental analysis of natural convection in a cavity with flush mounted heat sources on a side wall

This work considers a problem of interest in several technological applications such as the thermal control of electronic equipment. It is also important to study the heat transfer performance of these components under off-normal conditions, such as during failure of cooling fans. The effect of natural convection on the flow and heat transfer in a cavity with two flush mounted heat sources on the left vertical wall, simulating electronic components, is studied numerically and experimentally. The influence of the power distribution, spacing between the heat sources and cavity aspect ratio have been investigated. An analysis of the average Nusselt number of the two heat sources was performed to investigate the behavior of the heat transfer coefficients. The results obtained numerically and experimentally, after an error analysis, showed a good agreement.

Improved Conductivity Induced by Photodesorption in SnO2 Thin Films Grown by a Sol-Gel Dip Coating Technique

Thin films of undoped and Sb-doped SnO2 have been prepared by a sol-gel dip-coating technique. For the high doping level (2-3 mol% Sb) n-type degenerate conduction is expected, however, measurements of resistance as a function of temperature show that doped samples exhibit strong electron trapping, with capture levels at 39 and 81 meV. Heating in a vacuum and irradiation with UV monochromatic light (305 nm) improve the electrical characteristics, decreasing the carrier capture at low temperature. This suggests an oxygen related level, which can be eliminated by a photodesorption process. Absorption spectral dependence indicates an indirect bandgap transition with Eg ≅ 3.5 eV. Current-voltage characteristics indicate a thermionic emission mechanism through interfacial states.

Dynamical calculation of direct muon-transfer rates from thermalized muonic hydrogen to C6+ and O8+

Moun-transfer reactions from muonic hydrogen to carbon and oxygen nuclei employing a full quantum-mechanical few-body description of rearrangement scattering were studied by solving the Faddeev-Hahn-type equations using close-coupling approximation. The application of a close-coupling-type ansatz led to satisfactory results for direct muon-transfer reactions from muonic hydrogen to C6+ and O8+.