薄体结构温度场的高阶边界元分析

分析了二维问题边界元法3节点二次单元的几何特征,区分和定义了源点相对高阶单元的Ⅰ型和Ⅱ型接近度.针对二维位势问题高阶边界元中奇异积分核,构造出具有相同Ⅱ型几乎奇异性的近似核函数,在几乎奇异积分单元上分离出积分核中主导的奇异函数部分.原积分核扣除其近似核函数后消除几乎奇异性,成为正则积分核函数,并采用常规Gauss数值方法计算该正则积分;对奇异核函数的积分推导出解析公式,从而建立了一种新的边界元法高阶单元几乎奇异积分半解析算法.应用该算法计算了二维薄体结构温度场算例,计算结果表明高阶单元半解析算法能充分发挥边界元法优势,显著提高计算精度.     

贝塞尔函数在计算几种特殊三角函数积分中的应用

<正>在工程上特别是科技仪器理论中有时会遇到一些被积函数中三角函数又含有三角函数的在工程上特别是科技仪器理论中有时会遇到一些被积函数中三角函数又含有三角函数的份殊二用幽数附积分,.而这些积分一般是不能直接用求何应用贝塞尔函数来计算下面两种特殊的三角函数积分特殊三角函数的积分,而这些积分;能     

轴对称弹性体的有限元分析

轴对称弹性力学问题的有限元分析长期以来都是采用三角圆环有限元和线性形状函数.由于积分困难,常用近似积分求得刚度矩阵,这种近似积分对于靠近旋转对称轴的元素,误差很大,所以,长期以来,被认为不满意的办法.也有用精确积分计算刚度矩阵的.但本文指出,这种积分只适用于有中孔的轴对称体.对于实心的轴对称体而言,这种刚度矩阵都不收敛,计算是无效的.本文提出了一种新的形状函数,当径向座标r接近于零时,这种形状函数的径向位移u自然地接近于零.如果用这种新的形状函数,则由此计算求得的刚度矩阵,不论三角圆环有限元的位置是否靠近轴线.都是存在的.这种有限元,就能用于计算实心的轴对称体的问题.     

一种有效的分析弹塑性问题的边界元法

本文提出了一组有效的边界元公式.该公式通过利用一个新的变量,使核函数仅具有lnr(r为源点和场点的距离)的较低阶奇异性,从而,在积分点的传统位移和应力公式的奇异性得到降低,且原公式中影响应力计算精度的边界层效应得到消除.同时,也避免了难于计算的参数C.将该方法应用到弹塑性分析中,数值分析结果表明该公式具有明显的优势.     

第二类端点奇异Fredholm积分方程的分数阶退化核方法

本文针对第二类端点奇异Fredholm积分方程构造基于分数阶Taylor展开的退化核方法，设计了两种计算格式，一是在全区间上使用分数阶Taylor展开式近似核函数，二是在包含奇点的小区间上采用分数阶插值，在剩余区间上采用分段二次多项式插值逼近核函数.讨论了两种退化核方法收敛的条件，并给出了混合插值法的收敛阶估计.数值算例表明对于非光滑核函数分数阶退化核方法有着良好的计算效果，且混合二次插值法比全区间上的分数阶退化核方法有着更广泛的适用范围.     

面向大数据流的多任务在线学习算法研究

大数据时代背景下,越来越多领域对大数据计算提出了高要求,尤其各行各业产生的大数据更多地是一种动态的流式数据形态,因此,实现实时、快速、高效的大数据流计算与分析日益紧要.在线机器学习算法是解决实时大数据流分析的有效方案.在机器学习算法中,通过核学习能够获得有效的核函数,而所选核函数又对核学习器的性能有很大影响.结合在线机器学习与核函数研究一种适用于大数据流环境下的多任务在线学习算法,探讨了算法过程中可能出现的扰动项,应用数据依赖核的构建方法提高了算法的广泛性.算法不需要对历史数据流进行存储和重新扫描,只需选择一个数据集样本,在分析新的流式大数据时能够在可接受时间内直接将当前核函数更新为最合适的核函数,非常适合应用于流式大数据环境下的核学习问题.     

用凑微分法解三角函数有理式的积分

在不定积分的计算中,凑微分法是一种极为重要的方法.它的运用范围广泛,而且计算量较小,许多类型函数的积分都可以优先考虑应用这种方法.三角函数有理式的积分,用凑微分法通常是有效而较为简便的.     

第二类两端奇异Fredholm积分方程的分数阶线性插值方法

考虑第二类两端奇异的Fredholm积分方程,假设核函数在区间的两个端点非光滑,存在分数阶的Taylor展开式.对于这种类型的核函数,在包含端点的小区间上采用分数阶插值,在剩余区间上采用分段线性插值逼近,由此得到一种分数阶线性插值退化核方法.本文讨论该方法收敛的条件,给出收敛阶估计.数值算例表明这种分数阶混合线性插值方法对于两端奇异核函数有着较好的计算效果.     

一类积分方程组正解的对称性和单调性

本文讨论积分方程组（？）解的性质,其中G_α是α阶贝塞尔位势核,0≤β〈α（n-α＋β）/n,1/（q＋1）＋1/（r＋1）〉（n-α＋β）/n,1/（r＋1）＋1/（p＋1）〉（n-α＋β）/n.我们用积分形式的移动平面法证明上述积分方程组的正解是径向对称且单调的.     

Cn空间中有界域上一种积分表示

本文应用单位分解的观点及积分表示中核函数的构造理论,得到ln空间中有界域上积分表示的一种抽象的一般形式,根据这种一般形式,可以得到至今许多区域上光滑函数和全纯函数种种已有的抽象公式和具体的积分公式.     

Evaluation of Fresnel integrals based on the continued fractions method

We present a simple algorithm for evaluating Fresnel integrals based on the continued fractions method: we use the relation between these integrals and first-kind Bessel functions of fractional order, and we apply a fast code to calculate them based on the continued fractions method. This latter code is especially useful for evaluating high order Bessel functions because it does not require recalculations using normalization relations. Comments on the same procedure but using Miller’s algorithm to evaluate the required Bessel functions are presented and a comparison with a standard code for evaluating Fresnel integrals (Numerical Recipes program FRENEL) is provided.     

Modified clenshaw-curtis method for the computation of Bessel function integrals

The numerical evaluation of Bessel function integrals may be difficult when the Bessel function is rapidly oscillating in the interval of integration. In the method presented here, the smooth factor of the integrand is replaced by a truncated Chebyshev series approximation and the resulting integral is computed exactly. The numerical aspects of this exact integration are discussed.     

Bessel,sine and cosine functions and extrapolation methods for computing molecular multi-center integrals

In this work, we present an extremely efficient approach for a fast numerical evaluation of highly oscillatory spherical Bessel integrals occurring in the analytic expressions of the so-called molecular multi-center integrals over exponential type functions. The approach is based on the Slevinsky-Safouhi formulae for higher derivatives applied to spherical Bessel functions and on extrapolation methods combined with practical properties of sine and cosine functions. Recurrence relations are used for computing the approximations of the spherical Bessel integrals, allowing for a control of accuracy and the stability of the algorithm. The computer algebra system Maple was used in our development, mainly to prove the applicability of the extrapolation methods. Among molecular multi-center integrals, the three-center nuclear attraction and four-center two-electron Coulomb and exchange integrals are undoubtedly the most difficult ones to evaluate rapidly to a high pre-determined accuracy. These integrals are required for both density functional and ab initio calculations. Already for small molecules, many millions of them have to be computed. As the molecular system gets larger, the computation of these integrals becomes one of the most laborious and time consuming steps in molecular electronic structure calculation. Improvement of the computational methods of molecular integrals would be indispensable to a further development in computational studies of large molecular systems. Convergence properties are analyzed to show that the approach presented in this work is a valuable contribution to the existing literature on molecular integral calculations as well as on spherical Bessel integral calculations.     

Indefinite integrals of quotients of special functions

A new method is presented for deriving indefinite integrals involving quotients of special functions. The method combines an integration formula given previously with the recursion relations obeyed by the function. Some additional results are presented using an elementary method, here called reciprocation, which can also be used in combination with the new method to obtain additional quotient integrals. Sample results are given here for Bessel functions, Airy functions, associated Legendre functions and the three complete elliptic integrals. All results given have been numerically checked with Mathematica.     

Indefinite integrals of products of special functions

A method is given for deriving indefinite integrals involving squares and other products of functions which are solutions of second-order linear differential equations. Several variations of the method are presented, which applies directly to functions which obey homogeneous differential equations. However, functions which obey inhomogeneous equations can be incorporated into the products and examples are given of integrals involving products of Bessel functions combined with Lommel, Anger and Weber functions. Many new integrals are derived for a selection of special functions, including Bessel functions, associated Legendre functions, and elliptic integrals. A number of integrals of products of Gauss hypergeometric functions are also presented, which seem to be the first integrals of this type. All results presented have been numerically checked with Mathematica.     

Indefinite integrals of Lommel functions from an inhomogeneous Euler–Lagrange method

A method given recently for deriving indefinite integrals of special functions which satisfy homogeneous second-order linear differential equations has been extended to include functions which obey inhomogeneous equations. The extended method has been applied to derive indefinite integrals for the Lommel functions, which obey an inhomogeneous Bessel equation. The method allows integrals to be derived for the inhomogeneous equation in a manner which closely parallels the homogeneous case, and a number of new Lommel integrals are derived which have well-known Bessel analogues. Results will be presented separately for other special functions which obey inhomogeneous second-order linear equations.     

On the summation of Schlömilch's series

ABSTRACT

Schlömilch's series is named after the German mathematician Oscar Xavier Schlömilch, who derived it in 1857 as a Fourier series type expansion in terms of the Bessel function of the first kind. However, except for Bessel functions, here we consider an expansion in terms of Struve functions or Bessel and Struve integrals as well. The method for obtaining a sum of Schlömilch's series in terms of the Bessel or Struve functions is based on the summation of trigonometric series, which can be represented in terms of the Riemann zeta and related functions of reciprocal powers and in certain cases can be brought in the closed form, meaning that the infinite series are represented by finite sums. By using Krylov's method we obtain the convergence acceleration of the trigonometric series.     

On the coefficients of integrated expansions of Bessel polynomials

A new formula expressing explicitly the integrals of Bessel polynomials of any degree and for any order in terms of the Bessel polynomials themselves is proved. Another new explicit formula relating the Bessel coefficients of an expansion for infinitely differentiable function that has been integrated an arbitrary number of times in terms of the coefficients of the original expansion of the function is also established. An application of these formulae for solving ordinary differential equations with varying coefficients is discussed.     

Analysis of a collocation method for integrating rapidly oscillatory functions

A collocation method for approximating integrals of rapidly oscillatory functions is analyzed. The method is efficient for integrals involving Bessel functions J_v(rx) with a large oscillation frequency parameter r, as well as for many other one- and multi-dimensional integrals of functions with rapid irregular oscillations. The analysis provides a convergence rate and it shows that the relative error of the method is even decreasing as the frequency of the oscillations increases.     

Error Bounds for Uniform Asymptotic Expansions—Modified Bessel Function of Purely Imaginary Order

The authors modify a method of Olde Daalhuis and Temme for representing the remainder and coefficients in Airy-type expansions of integrals. By using a class of rational functions, they express these quantities in terms of Cauchy-type integrals; these expressions are natural generalizations of integral representations of the coefficients and the remainders in the Taylor expansions of analytic functions. By using the new representation, a computable error bound for the remainder in the uniform asymptotic expansion of the modified Bessel function of purely imaginary order is derived.