预给极点的向量连分式插值
2016-12-19赵前进王本强
赵前进,王本强
(安徽理工大学理学院,安徽 淮南 232001)
预给极点的向量连分式插值
赵前进,王本强
(安徽理工大学理学院,安徽 淮南 232001)
为了保证函数在预给极点处的重数,给出了一种新算法计算预给极点的向量连分式插值。由预给的极点信息构造插值函数分母多项式的一个因式,通过每个向量值乘以一个确定的数,将预给极点的向量插值转化为无预给极点的向量插值,基于向量的Samelson逆构造Thiele型向量连分式插值,最终通过除以一个确定的函数获得预给极点的向量连分式插值。具有预给的极点且保持原有的重数。通过数值实例对比不同方法在极点附近的插值误差,说明了新方法的有效性。
预给极点;重数;向量有理插值;算法
在工程实践和科学研究领域经常遇到有极点的奇异函数的计算问题,连分式插值与逼近是解决此类问题的有效途径之一[1-21]。文献[2]中提出了一个计算预给极点的二元向量有理插值方法,通过设定极点处的向量函数值为无穷大向量,将预给极点和原有的插值节点都作为新的插值节点,基于向量的Samelson逆来计算预给极点的二元向量有理插值,但是无法区分和保持极点的重数。本文研究预给极点的向量连分式插值。由预给的极点信息构造插值函数分母多项式的一个因式,通过每个向量值乘以一个确定的数,将预给极点的向量插值转化为常规的无预给极点的向量插值,进一步基于向量的Samelson逆计算Thiele型向量连分式插值,最终通过除以一个确定的函数获得预给极点的向量有理插值函数,它具有预给的极点且每个预给极点保持原有的重数。给出的数值实例说明了新方法的的有效性。
1 向量值连分式插值
文献[21]给出了向量值连分式插值算法。给定数据(xi,v(i)), i=0,1,…,n,则Thiele型向量值连分式插值函数为
(1)
其中
φk(x0,x1,…,xk)=
k=1,2,…,n。
(2)
由文献[21]知,如果n为偶数,则式(1)中向量值插值有理函数为[n/n]型;如果n为奇数,则式(1)中向量值插值有理函数为[n/(n-1)]型。
2 预给极点的向量连分式插值
T(x)=R(x)d(x)
(3)
满足
T(xi)=R(xi)d(xi)=v(i)d(xi)
(i=0,1,…,n)
(4)
基于向量的Samelson逆, 计算得满足插值条件 (4)的Thiele型向量值连分式插值函数T(x), 从而得到预给极点的向量值连分式插值
(5)
显然,当T(x)的分子、分母中每个多项式在各极点处的值均不等于零时,有理插值函数R(x)具有预给的极点且每个预给极点保持原有的重数。算法流程图如下
图1 算法流程图
3 数值例子
取x0=0,x1=1,x2=2三个插值节点,它们对应的函数值分别是f(x0)=(0.685,0),f(x1)=(1.879,0.582),f(x2)=(5.916,0.728)。
显然d(x)=(x+1)(x-3)2。
根据上文中的算法,求得
T(x0)=(6.165,0),T(x1)=(15.032,4.656),T(x2)=(17.748,2.184),基于Samelson逆,计算得满足插值条件的Thiele型向量连分式插值函数
利用文献[21]中朱功勤提供的方法,取x0=0,x1=1, x2=2, x3=-1, x4=3, 设f(x) 在
x3= -1和x4=3处的向量值为无穷大向量,基于Samelson逆,计算得满足插值条件的Thiele型向量连分式插值函数
两种插值方法在点x=2.6和x=3.6处的相对误差对比如表1所示。
表1 两种插值方法的相对误差对比
从两个插值方法相对差误差对比可以看出本文中给出的方法在二重极点x=3附近相对误差较小,且保持其极点的重数,说明上文中的方法的可行性和有效性。
4 结论
由预给的极点位置和重数信息构造一个新的向量值插值函数,将预给极点的向量值有理插值转化为无预给极点的向量值有理插值,基于向量的Samelson逆构造Thiele型向量值连分式插值,最终通过极点的信息,将其转换为预给极点的向量值连分式插值,具有预给的极点且极点保持原有的重数。在实际问题中,靠近极点处的函数精度将会大大提高,给出的数值实例说明了新方法的优点。
[1] 朱功勤,檀结庆,王洪燕. 预给极点的向量有理插值及性质[J].高等学校计算数学学报,2000,22(2):97-104.
[2] 檀结庆.连分式理论及其应用[M]. 北京:科学出版社,2007:32-100.
[3] JIEQING TAN,BAORUI SONG,GONGQIN ZHU.Vector Valued Rational Interpolants Over Triangular Grids[J].Computers and Mathematics with Applicatigns,2002,44(44):1 357-1 367.
[4] QIANJIN ZHAO, JIEQING TAN.Block-based Thiele-like blending rational interpolation [J].Journal of Computational and Applied Mathematics,2006,195 (1):312-325.
[5] TAN J Q. Bivariate blending rational interpolants [J]. Approx Theory & Its Appl, 1999, 15(2): 74-83.
[6] TAN J Q, FANG Y.Newton-Thiele’s rational interpolants [J]. Numerical Algorithms, 2000, 24(1-2):141-157.
[7] P.R.GRAVES-MORRIS.Vector valued rational interpolants I [J]. Numerische Mathemalik,1983,42(3):331-348.
[8] P. R. GRAVES-MORRIS , C. D. JENKINS.Vector-Valued, Rational Interpolants III [J]. Constructive Approxiration,1986,2(1):263-289.
[9] M.GASCA,J.J.MARTINEZ.Computation of rational interpolants with prescribed poles [J]. Journal of Computational and Applied Mathematics,1989,26(3):297-309.
[10] ERIK MEIJERING.A Chronology of Interpolation [J].Proc IEEE-2002,90(3):319-342.
[11] AVRAM SIDI.Algebraic properties of some new vector-valued rational interpolants [J]. Journal of Approximation Theory,2006,141(2):142-161.
[12] AVRAM SIDI. A new approach to vector-valued rational interpolation [J].Journal of Approximation Theory,2004,130(2):179-189.
[13] LISA LORENTZEN.A convergence theorem for random continued fractions [J].Journal of Approximation Theory, 2015,197:1-8.
[14] WOJCIECH SIEMASZKO.Thiele-type branched continued fractions for two-variable functions [J]. Journal of Computational and Applied Mathematics,1983,9(2): 137-153.
[15] DAWEI LU, ZEXI SONG.Some new continued fraction estimates of the Somos' quadratic recurrence constant [J]. Journal of Number Theory,2015,155:36-45.
[16] GU CHUANQING.Multivariate generalized inverse vector-valued rational interpolants [J].Journal of Computational and Applied Mathematics,1997,84(2):137-146.
[17] ANGEL RIBALTA.State space realizations of rational interpolants with prescribed poles [J]. Systems & Control Letters,2001,43(43):379-386.
[18] BERRUT J.-P.. The barycentric weights of rational interpolation with prescribed poles [J]. Journal of Computational and Applied Mathematics,1997,86(1):45-52.
[19] BERRUT J P, MITTELMANN H D.Lebesgue constant minimizing linear rational interpolation of continuous functions over the interval [J]. Computers and Mathematics with Applications,1997,33(6):77-86.
[20] BERRUT J.-P.,MITTELMANN H.Rational interpolation through the optimal attachment of poles to the interpolation polynomial [J]. Numerical Algorithms,2000, 23(4):315-328.
[21] 王仁宏,朱功勤.有理函数逼近及其应用[M]. 北京:科学出版社,2004:139-142.
(责任编辑:李 丽,吴晓红,编辑:丁 寒)
Vector valued continued fraction interpolation with prescribed poles
ZHAO Qian-jin, WANG Ben-qiang
(School of Science, Anhui University of Science and Technology, Huainan Anhui 232001, China)
In order to guarantee the number of functions in the prescribed poles, this paper presents an algorithm developed to calculate the vector valued continued fraction interpolant with prescribed poles. In the vector valued interpolant, a factorization of the denominator polynomial is constructed based on the information about the prescribed poles. By means of multiplying each interpolated vector value by a certain number, vector valued interpolation with prescribed poles is transformed into the one without prescribed poles. The vector valued continued fraction interpolant is constructed based on the Samelson inverse. Finally, by dividing a defined function, the vector valued continued fraction interpolant with prescribed poles is obtained and has prescribed poles with intrinsic multiplicity. Finally, an example is given in the text, by comparing different methods in interpolation error pole nearby, and shows the effectiveness of the new method.
prescribed poles; multiplicity; vector valued rational interpolation; algorithm
2016-01-13
国家自然科学基金(60973050); 安徽省教育厅自然科学基金项目(KJ2009A50)
赵前进(1967-),男,安徽凤阳人,教授,博士,硕士生导师,研究方向:有理插值与逼近, 数字图像处理。
O241
A
1672-1098(2016)05-0001-04