和文強(qiáng), 秦國(guó)良, 包振忠
(西安交通大學(xué) 流體機(jī)械研究所 陜西 西安 710049)
?
Chebyshev-Gauss-Lobatto節(jié)點(diǎn)的一個(gè)注記
和文強(qiáng), 秦國(guó)良, 包振忠
(西安交通大學(xué) 流體機(jī)械研究所 陜西 西安 710049)
利用Lagrange插值基函數(shù)和Chebyshev多項(xiàng)式的性質(zhì),推導(dǎo)以Chebyshev-Gauss-Lobatto點(diǎn)為插值點(diǎn)構(gòu)造的插值基函數(shù)的一階、二階微分矩陣的顯示格式,并由插值點(diǎn)的性質(zhì)得出兩者之間的關(guān)系.通過(guò)對(duì)具有解析解的一維對(duì)流擴(kuò)散方程進(jìn)行數(shù)值求解,驗(yàn)證了一階、二階微分矩陣顯式格式的正確性.數(shù)值結(jié)果表明:由微分矩陣顯式格式可以方便地構(gòu)造配置點(diǎn)譜方法中的擬譜算子,利用其求解微分方程,在較少的網(wǎng)格點(diǎn)時(shí),即可得到快速收斂的高精度的數(shù)值結(jié)果.研究工作對(duì)配置點(diǎn)譜方法的應(yīng)用具有一定的理論指導(dǎo)意義.
譜方法; Chebyshev-Gauss-Lobatto節(jié)點(diǎn); Lagrange插值基函數(shù); 微分矩陣
1977年,Gottlieb和Orszag[1]系統(tǒng)地從數(shù)學(xué)上對(duì)譜方法進(jìn)行了理論闡述.從此,譜方法以高精度和快速收斂的特點(diǎn)被廣泛地應(yīng)用到了分子動(dòng)力學(xué)[2]、流體流動(dòng)數(shù)值模擬[3-4]、非線性波及孤立子計(jì)算[5]、數(shù)值天氣預(yù)報(bào)[6]等不同領(lǐng)域中,并得到了快速的發(fā)展.利用譜方法進(jìn)行計(jì)算時(shí),首先定義計(jì)算區(qū)域上的一組正交函數(shù)為基函數(shù),然后利用基函數(shù)將求解變量表示為其譜逼近形式PN(x).按照逼近方式的不同,譜方法可分為Galerkin法和配置點(diǎn)法(擬譜方法).
利用配置點(diǎn)法進(jìn)行求解時(shí),變量的n階導(dǎo)數(shù)是以PN(x)的n階微分矩陣形式給出,所以微分矩陣在配置點(diǎn)法中有著十分重要的作用.文獻(xiàn)[7]分析了利用Legendre-Gauss-Lobatto插值點(diǎn)構(gòu)造的Lagrange插值基函數(shù)[8]微分矩陣及其相互關(guān)系,本文將推導(dǎo)基于Chebyshev-Gauss-Lobatto插值點(diǎn)的插值基函數(shù)一階、二階微分矩陣的顯式格式,并說(shuō)明兩者間相互關(guān)系.結(jié)合導(dǎo)出的微分矩陣顯式格式,利用Chebyshev配置點(diǎn)法求解一維對(duì)流擴(kuò)散方程,通過(guò)數(shù)值算例驗(yàn)證推導(dǎo)的正確性,為Chebyshev配置點(diǎn)譜方法的應(yīng)用奠定一定理論基礎(chǔ).
令Tn(x)表示N階Chebyshev多項(xiàng)式,取其極值點(diǎn)xi=-cos(iπ/N)(0≤i≤N),為Chebyshev-Gauss-Lobatto插值點(diǎn),則以xi為插值點(diǎn)的Lagrange插值基函數(shù)為
(1)
根據(jù)文獻(xiàn)[9],lj(x)可以表示為
(2)
由Lagrange插值基函數(shù)的性質(zhì),lj(xi)=δji.
Chebyshev多項(xiàng)式TN(x)=cos(N·arccosx),對(duì)TN(x)關(guān)于x求導(dǎo),即可得其在極值點(diǎn)的一階導(dǎo)數(shù)值.又TN(x)是Chebyshev微分方程的解,滿足
(3)
由式(3)可得
(4)
將xi代入式(4),可得TN(x)在其極值點(diǎn)的二階導(dǎo)數(shù)值,對(duì)式(4)兩側(cè)關(guān)于x求導(dǎo),即可得TN(x)在xi點(diǎn)的三階導(dǎo)數(shù)值.
定理1設(shè)以Chebyshev-Gauss-Lobatto點(diǎn)為節(jié)點(diǎn)的插值基函數(shù)的一階微分矩陣為D=(dij)(N+1)×(N+1),則有
證明(Ⅰ) 由式(2)得
(5)
對(duì)式(5)兩側(cè)關(guān)于x求導(dǎo),并化簡(jiǎn)得
(6)
將x=xi代入式(6)得
(7)
(Ⅱ) 對(duì)式(6)兩側(cè)再一次求導(dǎo)得
(8)
當(dāng)i=j0,N時(shí),將x=xi代入式(8),可得.
綜合上述(Ⅰ)、(Ⅱ)、(Ⅲ),可得定理1.
證明(Ⅰ) 當(dāng)ij0,N時(shí),將x=xi代入式(8),可得
(Ⅱ) 對(duì)式(8)兩側(cè)再一次求導(dǎo)得
(9)
當(dāng)i=j≠0,N時(shí),將x=xi代入式(9),可得
綜合上述(Ⅰ)、(Ⅱ),知定理2得證.
為了推導(dǎo)一階微分矩陣和二階微分矩陣的關(guān)系,先給出Chebyshev-Gauss-Lobatto節(jié)點(diǎn)的如下性質(zhì):定理3,4.
定理3設(shè)x0,x1,…,xN為Chebyshev-Gauss-Lobatto節(jié)點(diǎn),則有
(10)
(11)
綜合(Ⅰ)、(Ⅱ),知定理3得證.
(Ⅱ) 當(dāng)i=j時(shí),由定理1得
(12)
對(duì)式(1)兩側(cè)取對(duì)數(shù),并求二階導(dǎo)數(shù)得
(13)
在式(13)中,令x=xj,j=0,1,…,N-1,N,由Lagrange插值基函數(shù)性質(zhì)得
(14)
綜合(Ⅰ)、(Ⅱ)知,定理4得證.
為了驗(yàn)證上述關(guān)于Chebyshev-Gauss-Lobatto節(jié)點(diǎn)生成的Lagrange插值基函數(shù)一階、二階微分矩陣顯式格式的正確性,考慮一維對(duì)流擴(kuò)散方程
(15)
相應(yīng)的初始條件和邊界條件:u(x,0)=u0(x),u(0,t)=g1(t),u(1,t)=g2(t).其中,Re=aL/ν為無(wú)量綱數(shù),L為特征長(zhǎng)度,a為特征速度,ν為擴(kuò)散系數(shù),S為無(wú)量綱源項(xiàng).
利用插值基函數(shù)一階、二階微分矩陣,式(15)可以表示為
?tUN=LNUN+SN,t>0,
(16)
其中,uN(x,t)為u的配置點(diǎn)譜逼近形式,UN(t)為關(guān)于uN的列向量,它由N-1個(gè)分量(內(nèi)配置點(diǎn)上的變量值)組成,在兩邊界點(diǎn)上,uN(x0,t)、uN(xN,t)由邊界條件計(jì)算出;SN為離散源項(xiàng),LN為擬譜算子.
采用四階Runge-Kutta方法(RK4)對(duì)式(16)中的時(shí)間項(xiàng)進(jìn)行離散,可得式(16)的離散形式為
(17)
式(17)中,f為包含源項(xiàng)在內(nèi)的空間導(dǎo)數(shù)項(xiàng),f(UN)=LNUN+SN.
圖1 u的數(shù)值解和解析解的比較
圖2 計(jì)算誤差和插值階數(shù)N的關(guān)系
推導(dǎo)了配置點(diǎn)譜方法中由Chebyshev-Gauss-Lobatto點(diǎn)為插值節(jié)點(diǎn)構(gòu)造的Lagrange插值基函數(shù)的一階、二階微分矩陣的顯式格式,利用插值點(diǎn)的性質(zhì),導(dǎo)出了兩微分矩陣之間的關(guān)系.通過(guò)對(duì)一維對(duì)流擴(kuò)散方程數(shù)值求解,驗(yàn)證了插值基函數(shù)微分矩陣顯式格式正確性,為Chebyshev配置點(diǎn)譜方法的數(shù)值應(yīng)用提供了方便.
[1] Gottlieb D,Orszag S A.Numerical Analysis of Spectral Method:Theory and Application[M].Philadelphia:SIAM,1977:7-19.
[2] Sgorrakis N G,Merced-Serrano M,Boutsidis C,et al.Atomic-level characterization of the ensemble of the Aβ(1-42) monomer in water using unbiased molecular dynamics simulations and spectral algorithms[J].Journal of Molecular Biology,2011,405(2):570-583.
[3] Hessari P,Shin B C.The least-squares pseudo-spectral method for Navier-Stokes equations[J]. Computers and Mathematics with Applications,2013,66(3):319-329.
[4] Schneider K,Neffaa S,Bos W J T.A pseudo-spectral method with volume penalisation for magnetohydrodynamic terbulence in confined domains[J].Computer Physics Communications,2011, 182(1):2-7.
[5] Mokhtari R,Mohammadi M.Numerical solution of GRLW equation using Sinc-collocation method[J]. Computer Physics Communications,2010,181(7):1266-1274.
[6] Shizgal B D.Spectral methods based on nonclassical basis functions: the advection-diffusion equation[J].Computers & Fluids,2002,31(4/5/6/7):825-843.
[7] 王天軍,殷政偉.Legendre-Gauss-Lobatto節(jié)點(diǎn)的一個(gè)注記[J].河南科技大學(xué)學(xué)報(bào):自然科學(xué)版,2012,33(1):71-74.
[8] 王全來(lái),張薇.對(duì)波萊爾改進(jìn)拉格朗日插值公式思想方法的研究[J].鄭州大學(xué)學(xué)報(bào):理學(xué)版,2007,39(2):7-11.
[9] Gottlieb D,Hesthaven J S.Spectral methods for hyperbolic problems[J].Journal of Computational and Applied Mathematics,2001,128(1/2):83-131.
An Annotation of Chebyshev-Gauss-Lobatto Node
HE Wen-qiang, QIN Guo-liang, BAO Zhen-zhong
(InstituteofFluidMachinery,Xi’anJiaotongUniversity,Xi’an710049,China)
The explicit formula of first-order and second-order differential matrix of interpolation function based on Chebyshev-Gauss-Lobatto node was deduced by taking advantage of the properties of Lagrange interpolation function and Chebyshev polynomial. The relationship between the two differential matrixes was obtained by using the characters of the interpolation node. The correctness of the explicit formula of the two differential matrixes was verified by the numerical solution of one-dimensional convection-diffusion equation. It was indicated that numerical results with high accuracy and fast convergence could be obtained by the use of fewer nodes when the spectral operator constructed by the explicit formula of the two differential matrixes was used to solve differential equation. The present study could be of theoretical importance to the application of collocation spectral method.
spectral method; Chebyshev-Gauss-Lobattonode; Lagrange interpolation function; differential matrix
2014-05-29
國(guó)家重點(diǎn)基礎(chǔ)研究發(fā)展計(jì)劃(973計(jì)劃)項(xiàng)目,編號(hào)2012CB026004.
和文強(qiáng)(1982-),男,河南焦作人,博士研究生,主要從事計(jì)算流體力學(xué)研究,E-mail:wenqianghe2011@stu.xjtu.edu.cn;通訊作者:秦國(guó)良(1964-),男,安徽阜陽(yáng)人,教授,主要從事計(jì)算流體力學(xué)研究,E-mail:glqin@mail.xjtu.edu.cn.
O368
A
1671-6841(2015)01-0028-05
10.3969/j.issn.1671-6841.2015.01.006