New maximum entropy-based algorithm for structural design optimization

A new algorithm based on nonlinear transformation is proposed to improve the classical maximum entropy method and solve practical problems of reliability analysis. There are three steps in the new algorithm. Firstly, the performance function of reliability analysis is normalized, dividing by its value when each input is the mean value of the corresponding random variable. Then the nonlinear transformation of such normalized performance function is completed by using a monotonic nonlinear function with an adjustable parameter. Finally, the predictions of probability density function and/or the failure probability in reliability analysis are achieved by looking the result of transformation as a new form of performance function in the classical procedure of maximum entropy method in which the statistic moments are given through the univariate dimension reduction method. In the proposed method, the uncontrollable error of integration on the infinite interval is removed by transforming it into a bounded one. Three typical nonlinear transformation functions are studied and compared in the numerical examples. Comparing with results from Monte Carlo simulation, it is found that a proper choice of the adjustable parameter can lead to a better result of the prediction of failure probability. It is confirmed in the examples that result from the proposed method with the arctangent transformation function is better than the other transformation functions. The error of prediction of failure probability is controllable if the adjustable parameter is chosen in a given interval, but the suggested value of the adjustable parameter can only be given empirically.     

Inducing normality from non‐Gaussian long memory time series and its application to stock return data

Motivated by Lee and Ko (Appl. Stochastic Models. Bus. Ind. 2007; 23 :493–502) but not limited to the study, this paper proposes a wavelet‐based Bayesian power transformation procedure through the well‐known Box–Cox transformation to induce normality from non‐Gaussian long memory processes. We consider power transformations of non‐Gaussian long memory time series under the assumption of an unknown transformation parameter, a situation that arises commonly in practice, while most research has been devoted to non‐linear transformations of Gaussian long memory time series with known transformation parameter. Specially, this study is mainly focused on the simultaneous estimation of the transformation parameter and long memory parameter. To this end, posterior estimations via Markov chain Monte Carlo methods are performed in the wavelet domain. Performances are assessed on a simulation study and a German stock return data set. Copyright © 2009 John Wiley & Sons, Ltd.     

The practical use of the Euler transformation

The generalised Euler transformation is a powerful transformation of infinite series which can be used, in theory, for the acceleration of convergence and for analytic continuation. When the transformation is applied to a series with rounded coefficients, its behaviour can differ substantially from that predicted theoretically. In general, analytic continuation is impossible in this case. It is still possible, however, to use the transformation for acceleration of convergence, but some changes are necessary in the method of choosing the optimum parameter value.     

Percentile estimation based on the power-normal distribution

The authors propose several estimation methods for the confidence interval of the percentile of the power-normal distribution, which specifies the distribution of the observed values before Box–Cox transformation and assumes that transformed distribution is the truncated normal distribution. Their approaches are based on the delta method and the inverse transformation method. Then the finite sample performance of each estimation method is compared through a simulation, and it is shown that the performance of the inverse transformation method is superior to the delta method. A reparametrization method of the power-normal distribution, which is useful parameter setting tool in simulations, is suggested. The authors also investigate the asymptotic behavior of the coverage probability of the confidence interval of the percentile of the power-normal distribution in the case wherein the variance inflation of the MLE associated with the estimation of the transformation parameter of the power-normal distribution is ignored.     

生长曲线模型的变量选择

生长曲线模型是一个典型的多元线性模型, 在现代统计学上占有重要地位. 文章首先基于Potthoff-Roy变换后的生长曲线模型, 采用自适应LASSO为惩罚函数给出了参数矩阵的惩罚最小二乘估计, 实现了变量的选择. 其次, 基于局部渐近二次估计, 对生长曲线模型的惩罚最小二乘估计给出了统一的近似估计表达式. 接着, 讨论了经过Potthoff-Roy变换后模型的惩罚最小二乘估计, 证明了自适应LASSO具有Oracle性质. 最后对几种变量选择方法进行了数据模拟. 结果表明自适应LASSO效果比较好. 另外, 综合考虑, Potthoff-Roy变换优于拉直变换.     

Influence Diagnostics in Common Principal Components Analysis

In principal components analysis, the influence function and local influence approaches have been well established as important diagnostic tools. In this article, we first review the generalized local influence approach in the restricted likelihood framework. We then apply the restricted likelihood local influence diagnostic in the common principal components analysis. One special part of this local influence result is an elliptical norm of the empirical influence function, which is comparable to the deletion diagnostic scaled by the same matrix which requires iterative solutions for parameter estimates with every case deleted. Local influence diagnostics are constructed by some basic building blocks that are obtained directly from the maximum likelihood estimates of the parameters, and which are based on the original data and thus require less computation. A numerical example illustrates the technique and some joint influence effects are identified by the proposed method.     

Interpolation between sequence transformations

Levin's sequence transformation [1] and a structurally very similar sequence transformation [4] behave quite differently in convergence acceleration and summation processes. In particular, it was found recently that Levin's transformation fails completely in the case of the strongly divergent Rayleigh-Schrödinger and renormalized perturbation expansions for the ground state energies of anharmonic oscillators, whereas the structurally very similar sequence transformation gives very good results [14,17]. For a more detailed investigation of these phenomena, a sequence transformation is constructed which — depending on a continuous parameter — is able to interpolate between Levin's transformation and the other sequence transformation. Some numerical examples, which illustrate the properties of the interpolating sequence transformation, are presented.     

Application of the differential transformation method to the free vibrations of strongly non-linear oscillators

This paper adopts the differential transformation method to obtain the free vibration behavior of an oscillator with fifth-order non-linearities. The principle of differential transformation is briefly introduced, and is then applied in the derivation of a set of difference equations for the free vibration oscillator problem. The solutions are subsequently solved by a process of inverse transformation. The time responses of the oscillator are presented under different parameter conditions, and the current results are then compared with those derived from the established Runge–Kutta method in order to verify the accuracy of the proposed method. It is shown that there is excellent agreement between the two sets of results. This finding confirms that the proposed differential transformation method is a powerful and efficient tool for solving non-linear problems.     

Réduction de la conjecture de Hodge à une continuité

The obstructions, for a closed differential form on a projective manifold, to be cohomologous to an algebraic cycle with complex coefficients, are computed in terms of the Chow transformation. They can be expressed as an orthogonality condition, on the manifold itself, with families parametrized by the Grassmannian of currents which are completely determined. A parameter does not yield any obstruction if the associated projective subspace meets properly the manifold. The embedding of the manifold is degenerated, in view of applying the characterization of currents associated to algebraic cycles by the Chow transformation. We study the set of periods obtained when the parameter varies, in particular, we prove a continuity result, thanks to the constructibility of the Bernstein polynomial. When the cohomology class is rational, we conjecture that this set is connected.     

Holomorphic mappings of a strip into itself with bounded distortion at infinity

A class of holomorphic self-mappings of a strip which is symmetric with respect to the real axis is studied. It is required that the mappings should boundedly deviate from the identity transformation on the real axis. Distortion theorems for this class of functions are obtained, and domains of univalence are found that arise for certain values of the parameter characterizing the deviation of the mappings from the identity transformation on the real axis.     

生长曲线模型的协变量选择与参数估计

本文研究了生长曲线模型的Ｐｏｔｔｈｏｆｆ－Ｒｏｙ变换，给出了这种变换在矩阵范数最小规则下的最佳选择．进一步本文给出了利用协变量来改进Ｐｏｔｔｈｏｆｆ－Ｒｏｙ变换的途径并研究了协变量的选择及对参数估计的影响．     

Optimal solution for novel grey polynomial prediction model

The grey prediction model, as a time-series analysis tool, has been used in various fields only with partly known distribution information. The grey polynomial model is a novel method to solve the problem that the original sequence is in accord with a more general trend rather than the special homogeneous or non-homogeneous trend, but how to select the polynomial order still needs further study. In this paper the tuned background coefficient is introduced into the grey polynomial model and then the algorithmic framework for polynomial order selection, background coefficient search and parameter estimation is proposed. The quantitative relations between the affine transformation of accumulating sequence and the parameter estimates are deduced. The modeling performance proves to be independent of the affine transformation. The numerical example and application are carried out to assess the modeling efficiency in comparison with other conventional models.     

The cauchy problem for mechanical systems with a finite number of degrees of freedom as a problem of continuation on the best parameter

The Cauchy problem for a system of ordinary differential equations is formulated as a problem of continuation on the best parameter. It is proved that the length of an integral curve of the problem is such a parameter. The merits of the proposed transformation are demonstrated by a test example in which a stiff system of equations describing the perturbed motion of an aircraft is solved numerically.     

Unsteady convective boundary layer flow of a viscous fluid at a vertical surface with variable fluid properties

In this paper we present numerical solutions to the unsteady convective boundary layer flow of a viscous fluid at a vertical stretching surface with variable transport properties and thermal radiation. Both assisting and opposing buoyant flow situations are considered. Using a similarity transformation, the governing time-dependent partial differential equations are first transformed into coupled, non-linear ordinary differential equations with variable coefficients. Numerical solutions to these equations subject to appropriate boundary conditions are obtained by a second order finite difference scheme known as the Keller-Box method. The numerical results thus obtained are analyzed for the effects of the pertinent parameters namely, the unsteady parameter, the free convection parameter, the suction/injection parameter, the Prandtl number, the thermal conductivity parameter and the thermal radiation parameter on the flow and heat transfer characteristics. It is worth mentioning that the momentum and thermal boundary layer thicknesses decrease with an increase in the unsteady parameter.     

A perturbation treatment for optimal slightly nonlinear systems with linear control and quadratic criteria

Recently, it has been discovered that symbolic algebraic manipulations can be performed on the computer with input/output data in symbolic form. Accordingly, and for slightly nonlinear two-point boundary-value problems, it is feasible to obtain approximate analytical solutions in the form of power series in a small parameter. In these solutions, the boundary values are presented in a literal form. In this paper, the Lie canonical transformation is applied to derive approximate optimal solutions for slightly nonlinear systems with quadratic criteria. The transformation generator is determined by simple partial differential equations of the first order. To determine the arbitrary constants of the transformation in terms of the two-point boundary values, inversion of a vectorial power series in a small parameter is required, and a recursive algorithm for this inversion is given. To express the final solution in terms of these boundary values, a substitution of the inverted vectorial power series into another vectorial power series is also necessary, and a recursive algorithm for this substitution is presented.The authors are indebted to Professor J. V. Breakwell for his suggestions.     

Linear system of differential equations with turning point

A system of linear differential equations with small parameter as a coefficient of a part of derivatives is reduced to the canonical form and the properties of the transformation matrix are investigated.     

Instance-specific multi-objective parameter tuning based on fuzzy logic

Finding good parameter values for meta-heuristics is known as the parameter setting problem. A new parameter tuning strategy, called IPTS, is proposed that is a novel instance-specific method to take the trade-off between solution quality and computational time into consideration. Two important steps in the method are an a priori statistical analysis to identify the factors that determine heuristic performance in both quality and time for a specific type of problem, and the transformation of these insights into a fuzzy inference system rule base which aims to return parameter values on the Pareto-front with respect to a decision maker’s preference.Applied to the symmetric Travelling Salesman Problem and the meta-heuristic Guided Local Search, the approach is consistently faster than a traditional non-instance-specific parameter tuning strategy without significantly affecting solution quality; optimised for speed, computational times are shown to be on average 20 times faster while producing solutions of similar quality. A number of interesting areas for further research are discussed.     

Sturm-Liouville problems with boundary conditions depending quadratically on the eigenparameter

We study Sturm-Liouville problems with right-hand boundary conditions depending on the spectral parameter in a quadratic manner. A modified Crum-Darboux transformation is used to produce chains of problems almost isospectral with the given one. The problems in the chain have boundary conditions which in various cases are affine or bilinear in the spectral parameter, and in all cases culminate in a problem with constant boundary conditions. This extends recent work of Binding, Browne, Code and Watson when the right-hand condition is either an affine function of the spectral parameter with negative leading coefficient or a Herglotz function.     

数据变换参数的渐近置信域的曲率表示

该文首次用几何方法研究回归模型中数据变换参数及其子集参数的渐近置信域问题．由于文中讨论的是一般的数据变换多参数的渐近置信域的曲率表示，从而有关结论适用于各种数据变换如著名的Box-Cox变换、带有漂移参数的幂变换等变换中变换参数的渐近置信域的曲率表示．     

On the choice of parameters for power-series interior point algorithms in linear programming

In this paper we study higher-order interior point algorithms, especially power-series algorithms, for solving linear programming problems. Since higher-order differentials are not parameter-invariant, it is important to choose a suitable parameter for a power-series algorithm. We propose a parameter transformation to obtain a good choice of parameter, called ak-parameter, for general truncated powerseries approximations. We give a method to find ak-parameter. This method is applied to two powerseries interior point algorithms, which are built on a primal—dual algorithm and a dual algorithm, respectively. Computational results indicate that these higher-order power-series algorithms accelerate convergence compared to first-order algorithms by reducing the number of iterations. Also they demonstrate the efficiency of thek-parameter transformation to amend an unsuitable parameter in power-series algorithms.Work supported in part by the DFG Schwerpunktprogramm Anwendungsbezogene Optimierung und Steuerung.