数值分析第五_第1页
数值分析第五_第2页
数值分析第五_第3页
数值分析第五_第4页
数值分析第五_第5页
已阅读5页,还剩45页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

1、(完好版)数值分析第五版答案(全)(完好版)数值分析第五版答案(全)50/50(完好版)数值分析第五版答案(全)第一章绪论1设x0,x的相对偏差为,求lnx的偏差。解:近似值x*的相对偏差为=er*e*x*xx*x*而lnx的偏差为elnx*lnx*lnx1e*x*从而有(lnx*)2设x的相对偏差为2%,求xn的相对偏差。解:设f(x)xn,则函数的条件数为Cp|xf(x)|f(x)又Qf(x)nxn1,Cp|xnxn1|nn又Qr(*)n)Cpr(*)xx且er(x*)为2r(x*)n)0.02n3以下各数都是经过四舍五入获取的近似数,即偏差限不超出最后一位的半个单位,试指出它们是几位有效

2、数字:x1*1.1021,x2*0.031,x3*385.6,x4*56.430,x5*71.0.解:x1*1.1021是五位有效数字;x2*0.031是二位有效数字;x3*385.6是四位有效数字;x4*56.430是五位有效数字;x5*71.0.是二位有效数字。4利用公式(2.3)求以下各近似值的偏差限:(1)x1*x2*x4*,(2)x1*x2*x3*,(3)x2*/x4*.此中x1*,x*2,x3*,x4*均为第3题所给的数。解:(x1*)11042(x2*)1102(x3*)1102(x4*)1102(x5*)11023131(1)(x1*x2*x4*)(x1*)(x2*)(x4*)

3、1104110311032103221.05(x1*x*2x3*)x1*x2*(x3*)x2*x3*(x1*)x1*x3*(x2*)1.10210.03111010.031385.611041.1021385.611032220.215(3)(x*2/x*4)x2*(x*4)x*4(x*2)x*420.031110356.4301103256.43056.43021055计算球体积要使相对偏差限为1,问胸襟半径R时同意的相对偏差限是多少?解:球体体积为V4R33则何种函数的条件数为RVR4R2Cpgg3V4R33r(V*)Cpgr(R*)3r(R*)又Qr(V*)1%1故胸襟半径R时同意的相对

4、偏差限为(?)=?1%=1?133006设Y028,按递推公式YnYn1783(n=1,2,)1100计算到Y100。若取78327.982(5位有效数字),试问计算Y100将有多大偏差?解:QYnYn117831100Y100Y99783100Y99Y981783100Y98Y971783100Y1Y017831001挨次代入后,有Y100Y0100783100即Y100Y0783,若取78327.982,Y100Y027.982(Y*)(Y)(27.982)110310002Y100的偏差限为1103。27求方程x256x10的两个根,使它最少拥有4位有效数字(78327.982)。解:x

5、256x10,故方程的根应为x1,228783故x1287832827.98255.982x1拥有5位有效数字x2287831117832827.9820.0178632855.982x2拥有5位有效数字N118当N充分大时,如何求2dx?11NxN12dxarctan(N1)arctanN解1xN设arctan(N1),arctanN。则tanN1,tanN.N11dxN1x2arctan(tan()tantanarctantang1tanarctanN1N(N1)N1arctanN21N19正方形的边长大体为了100cm,应如何丈量才能使其面积偏差不超出1cm2?解:正方形的面积函数为A(

6、x)x2(A*)2A*g(x*).当x*100时,若(A*)1,则(x*)11022故丈量中边长偏差限不超出0.005cm时,才能使其面积偏差不超出1cm210设S1gt2,假设g是正确的,而对t的丈量有0.1秒的偏差,证明当t增添时S的2绝对偏差增添,而相对偏差却减少。解:QS1gt2,t02(S*)gt2g(t*)当t*增添时,S*的绝对偏差增添(S*)r(S*)S*gt2g(t*)g(t*)2(t*)2t*当t*增添时,(t*)保持不变,则S*的相对偏差减少。11序列yn满足递推关系yn10yn11(n=1,2,),若y021.41(三位有效数字),计算到y10时偏差有多大?这个计算过程

7、稳固吗?解:Qy021.41(y0*)11022又Qyn10yn11y110y01(y1*)10(y0*)又Qy210y11(y2*)10(y1*)(y2*)102(y0*)(y10*)1010(y0*)101011022108计算到y10时偏差为1108,这个计算过程不稳固。212(21)6,取2,利用以下等式计算,哪一个获取的结果最好?计算f16,(322)3,12)3,99702。(21)(32解:设y(x1)6,若x2,x*1.4,则x*1101。2若经过16计算y值,则2(1)y*(x*1gx*1)7*67y*x*(x1)y*x*若经过(322)3计算y值,则y*(32x*)2gx*

8、36*y*gx*2xy*x*若经过1计算y值,则22)3(3y*(31*)4gx*2x1*(32x*)7yxy*x*经过(312)3计算后获取的结果最好。213f(x)ln(xx21),求f(30)的值。若开平方用6位函数表,问求对数时偏差有多大?若改用另一等价公式。ln(xx21)ln(xx21)计算,求对数时偏差有多大?解Qf(x)ln(xx21),f(30)ln(30899)设u899,yf(30)则u*u*142故y*u*u*gu*0.01673若改用等价公式ln(xx21)ln(xx21)则f(30)ln(30899)此时,y*u*u*u*59.98337第二章插值法1当x1,1,2

9、时,f(x)0,3,4,求f(x)的二次插值多项式。解:x01,x11,x22,f(x0)0,f(x1)3,f(x2)4;l0(x)(xx1)(xx2)1(x1)(x2)(x0 x1)(x0 x2)2l1(x)(xx0)(xx2)1(x1)(x2)(x1x0)(x1x2)6l2(x)(xx0)(xx1)1(x1)(x1)(x2x0)(x2x1)3则二次拉格朗日插值多项式为2L2(x)k0yklk(x)3l0(x)4l2(x)1(x1)(x2)42(x1)(x1)35x23x76232给出f(x)lnx的数值表X0.40.50.60.70.8lnx用线性插值及二次插值计算ln0.54的近似值。解

10、:由表格知,x00.4,x10.5,x20.6,x30.7,x40.8;f(x0)0.916291,f(x1)0.693147f(x2)0.510826,f(x3)0.356675f(x4)0.223144若采纳线性插值法计算ln0.54即f(0.54),则0.50.540.6xx210(x0.6)l1(x)x2x1l2xx110(x0.5)(x)x1x2L1(x)f(x1)l1(x)f(x2)l2(x)6.93147(x0.6)5.10826(x0.5)L1(0.54)0.62021860.620219若采纳二次插值法计算ln0.54时,l0(xx1)(xx2)50(x0.5)(x0.6)(

11、x)x1)(x0 x2)(x0(xx0)(xx2)100(x0.4)(x0.6)l1(x)x0)(x1x2)(x1l2(xx0)(xx1)50(x0.4)(x0.5)(x)x0)(x2x1)(x2L2(x)f(x0)l0(x)f(x1)l1(x)f(x2)l2(x)500.916291(x0.5)(x0.6)69.3147(x0.4)(x0.6)0.51082650(x0.4)(x0.5)L2(0.54)0.615319840.6153203给全cosx,0ox90o的函数表,步长h1(1/60)o,若函数表拥有5位有效数字,研究用线性插值求cosx近似值时的总偏差界。解:求解cosx近似值时

12、,偏差可以分为两个部分,一方面,x是近似值,拥有5位有效数字,在此后的计算过程中产生必定的偏差流传;另一方面,利用插值法求函数cosx的近似值时,采纳的线性插值法插值余项不为0,也会有必定的偏差。所以,总偏差界的计算应综合以上双方面的要素。当0ox90o时,令f(x)cosx取x00,h(1)o1180108006060令xix0ih,i0,1,.,5400则x5400290o当xxk,xk1时,线性插值多项式为L1(x)f(xk)xxk1f(xk1)xxkxkxk1xk1xk插值余项为R(x)cosxL1(x)1f()(xxk)(xxk1)2又Q在建立函数表时,表中数据拥有5位有效数字,且c

13、osx0,1,故计算中有偏差流传过程。(f*(xk)11052R2(x)(f*(xk)xxk1(f*(xk1)xxk1xkxk1xk1xk(f*(xk)(xxk1xxk1)xkxk1xk1xk(f*(x)1(xk1xxx)khk(f*(xk)总偏差界为RR1(x)R2(x)1(cos)(xxk)(xxk1)(f*(xk)21(xxk)(xk1x)(f*(xk)21(1h)2(f*(xk)2211.0610810520.501061054设为互异节点,求证:n(1)xkjlj(x)xk(k0,1,L,n);j0nx)klj(x)(2)(xj0(k0,1,L,n);j0证明(1)令f(x)xknx

14、kjlj(x)。若插值节点为xj,j0,1,L,n,则函数f(x)的n次插值多项式为Ln(x)j0Rn(x)f(x)f(n1)()n1(x)插值余项为Ln(x)1)!(n又Qkn,f(n1)()0Rn(x)0nxkjlj(x)xkj0(k0,1,L,n);nx)klj(x)(2)(xjj0nn(Ckjxij(x)ki)lj(x)j0i0nnCki(x)ki(xijlj(x)i0j0又Q0in由上题结论可知nxkjlj(x)xij0n原式Cki(x)kixii0(xx)k0得证。5设f(x)C2a,b且f(a)f(b)0,求证:maxf(x)1(ba)2maxf(x).axb8axb解:令x0a

15、,x1b,以此为插值节点,则线性插值多项式为L1(x)f(x0)xx1f(x1)xx0 x0 x1xx0=f(a)xbf(b)xaabxa又Qf(a)f(b)0L1(x)0插值余项为R(x)f(x)L1(x)1f(x)(xx0)(xx1)2f(x)1f(x)(xx0)(xx1)2又Q(xx0)(xx1)12x)(xx0)(x121(x1x0)241(ba)24maxf(x)1(ba)2maxf(x).axb8axb6在4x4上给出f(x)ex的等距节点函数表,若用二次插值求ex的近似值,要使截断偏差不超出106,问使用函数表的步长h应取多少?解:若插值节点为xi1,xi和xi1,则分段二次插值

16、多项式的插值余项为R2(x)1f()(xxi1)(xxi)(xxi1)3!R2(x)1(xxi1)(xxi)(xxi1)maxf(x)64x4设步长为h,即xi1xih,xi1xihR2(x)1e42h33e4h3.63327若截断偏差不超出106,则R2(x)106e4h310627h0.0065.7若yn2n,求4yn及4yn.,解:依据向前差分算子和中心差分算子的定义进行求解。yn2n4yn(E1)4yn4j44j(1)Eynjj04j4(1)y4njjj04(1)j424jynj0j(21)4ynyn2n114yn(E2E2)4yn1(E2)4(E1)4ynE24ynyn22n28假如

17、f(x)是m次多项式,记f(x)f(xh)f(x),证明f(x)的k阶差分kf(x)(0km)是mk次多项式,而且m1f(x)0(l为正整数)。解:函数f(x)的Taylor展式为f(xh)f(x)f(x)h1f(x)h2L1f(m)(x)hm1f(m1)()hm12m!(m1)!此中(x,xh)又Qf(x)是次数为m的多项式f(m1)()0f(x)f(xh)f(x)f(x)h1f(x)h2L1f(m)(x)hm2m!f(x)为m1阶多项式2f(x)(f(x)2f(x)为m2阶多项式依此过程递推,得kf(x)是mk次多项式f(x)是常数当l为正整数时,m1f(x)09证明(fkgk)fkgkg

18、k1fk证明(fkgk)得证fk1gk1fk1gk1gk1(fk1gk1fkfkgkfkgkfkgk1fkgk1fkgkfk)fk(gk1gk)fkgkgk1fkn1n110证明fkgkfngnf0g0gk1fkk0k0证明:由上题结论可知fkgk(fkgk)gk1fkn1fkgk0n1(fkgk)gk1fk)k0n1n1(fkgk)gk1fkk0k0Q(fkgk)fk1gk1fkgkn1(fkgk)k0(f1g1f0g0)(f2g2f1g1)L(fngnfn1gn1)fngnf0g0n1n1fkgkfngnf0g0gk1fkk0k0得证。n12yj11证明yny0j0n1n1证明2yj(yj

19、1yj)j0j0(y1y0)(y2y1)L(ynyn1)yny0得证。12若f(x)a0a1xLan1xn1anxn有n个不一样实根x1,x2,L,xn,nk0,0kn2;xj证明:j1f(xj)n01,kn1证明:Qf(x)有个不一样实根x1,x2,L,xn且f(x)a0a1xLan1xn1anxnf(x)an(xx1)(xx2)L(xxn)令n(x)(xx1)(xx2)L(xxn)nknk则xjxjj1f(xj)j1ann(xj)而n(x)(xx2)(xx3)L(xxn)(xx1)(xx3)L(xxn)L(xx1)(xx2)L(xxn1)n(xj)(xjx1)(xjx2)L(xjxj1)(

20、xjxj1)L(xjxn)令g(x)xk,nkgx1,x2,L,xnxjj1n(xj)nk则gx1,x2,L,xnxjj1n(xj)nk1又xjgx1,x2,L,xnf(xj)j1annk0,0kn2;xjj1f(xj)n01,kn1得证。13证明n阶均差有以下性质:(1)若F(x)cf(x),则Fx0,x1,L,xncfx0,x1,L,xn;(2)若F(x)f(x)g(x),则Fx0,x1,L,xnfx0,x1,L,xngx0,x1,L,xn.证明:nj)(1)Qfx1,x2,L,xnf(x(xjx0)L(xjxj1)(xjxj1)L(xjxn)j0nFx1,x2,L,xnj0nj0(x(x

21、F(xj)jx0)L(xjxj1)(xjxj1)L(xjxn)cf(xj)jx0)L(xjxj1)(xjxj1)L(xjxn)nf(xj)c()j0(xjx0)L(xjxj1)(xjxj1)L(xjxn)cfx0,x1,L,xn得证。(2)QF(x)f(x)nFx0,L,xnj0nj0g(x)(xjx0)L(xj(xjx0)L(xjF(xj)xj1)(xjxj1)L(xjxn)f(xj)g(xj)xj1)(xjxj1)L(xjxn)nf(xj)j0(xjx0)L(xjxj1)(xjxj1)L(xjxn)ng(xj)+)(xjx0)L(xjxj1)(xjxj1)L(xjj0 xn)fx0,L,x

22、ngx0,L,xn得证。14f(x)x7x43x1,求F20,21,L,27及F20,21,L,28。解:Qf(x)x7x43x1若ixi2,i0,1,8L则fx0,x1,L,xnf(n)()n!fx0,x1,L,x7f(7)()7!17!7!f(8)()0fx0,x1,L,x88!15证明两点三次埃尔米特插值余项是R3(x)f(4)()(xxk)2(xxk1)2/4!,(xk,xk1)解:若xxk,xk1,且插值多项式满足条件H3(xk)f(xk),H3(xk)f(xk)H3(xk1)f(xk1),H3(xk1)f(xk1)插值余项为R(x)f(x)H3(x)由插值条件可知R(xk)R(xk

23、1)0且R(xk)R(xk1)0R(x)可写成R(x)g(x)(xxk)2(xxk1)2此中g(x)是关于x的待定函数,现把x看作xk,xk1上的一个固定点,作函数(t)f(t)H3(t)g(x)(txk)2(txk1)2依据余项性质,有(xk)0,(xk1)0(x)f(x)H3(x)g(x)(xxk)2(xxk1)2f(x)H3(x)R(x)0(t)f(t)H3(t)g(x)2(txk)(txk1)22(txk1)(txk)2(xk)0(xk1)0由罗尔定理可知,存在(xk,x)和(x,xk1),使(1)0,(2)0即(x)在xk,xk1上有四个互异零点。依据罗尔定理,(t)在(t)的两个零

24、点间最罕有一个零点,故(t)在(xk,xk1)内最罕有三个互异零点,依此类推,(4)(t)在(xk,xk1)内最罕有一个零点。记为(xk,xk1)使(4)()f(4)()H3(4)()4!g(x)0又QH3(4)(t)0g(x)f(4)(),(xk,xk1)4!此中依赖于xR(x)f(4)()(xxk)2(xxk1)24!分段三次埃尔米特插值时,若节点为xk(k0,1,L,n),设步长为h,即xkx0kh,k0,1,L,n在小区间xk,xk1上R(x)f(4)()2(xxk1)24!(xxk)R(x)1f(4)()(xxk)2(xxk1)24!(x4!(x4!14!24xk)2(xk1x)2m

25、axf(4)(x)axbxkxk1x)22maxf(4)(x)2axbh4maxf(4)(x)axbh4maxf(4)(x)384axb16求一个次数不高于4次的多项式P(x),使它满足P(0)P(0)0,P(1)P(1)0,P(2)0解:利用埃米尔特插值可获取次数不高于4的多项式x00,x11y00,y11m00,m1111H3(x)yjj(x)mjj(x)j0j00(x)(12xx0)(xx1)2x0 x1x0 x1(12x)(x1)21(x)(12xx1)(xx0)2x1x0 x1x0(32x)x20(x)x(x1)21(x)(x1)x2H3(x)(32x)x2(x1)x2x32x2设P

26、(x)H3(x)A(xx0)2(xx1)2此中,A为待定常数QP(2)1P(x)x32x2Ax2(x1)2A14从而P(x)1x2(x3)2417设f(x)1/(1x2),在5x5上取n10,按等距节点求分段线性插值函数Ih(x),计算各节点间中点处的Ih(x)与f(x)值,并预计偏差。解:若x05,x105则步长h1,xix0ih,i0,1,L,101f(x)1x2在小区间xi,xi1上,分段线性插值函数为Ih(x)xxi1f(xi)xxif(xi1)xixi1xi1xi(xi1x)1(xxi)11xi21xi12各节点间中点处的Ih(x)与f(x)的值为当x4.5时,f(x)0.0471,

27、Ih(x)0.0486当x3.5时,f(x)0.0755,Ih(x)0.0794当x2.5时,f(x)0.1379,Ih(x)0.1500当x1.5时,f(x)0.3077,Ih(x)0.3500当x0.5时,f(x)0.8000,Ih(x)0.7500偏差maxf(x)Ih(x)h2maxf()8xixxi15x51又Qf(x)1x22xf(x)(1x2)2,6x22f(x)(1x2)324x24x3f(x)(1x2)4令f(x)0得f(x)的驻点为x1,21和x30f(x1,2)1,f(x3)22maxf(x)Ih(x)145x518求f(x)x2在a,b上分段线性插值函数Ih(x),并预

28、计偏差。解:在区间a,b上,x0a,xnb,hixi1xi,i0,1,L,n1,hmaxhi0in1Qf(x)x2函数f(x)在小区间xi,xi1上分段线性插值函数为Ih(x)xxi1f(xi)xxif(xi1)xixi1xi1xi1xi2(xi1x)xi12(xxi)hi偏差为maxf(x)Ih(x)1maxf()ghi2xixxi18abQf(x)x2f(x)2x,f(x)2maxf(x)Ih(x)h24axb19求f(x)x4在a,b上分段埃尔米特插值,并预计偏差。解:在a,b区间上,x0a,xnb,hixi1xi,i0,1,L,n1,令hmaxhi0in1Qf(x)x4,f(x)4x3

29、函数f(x)在区间xi,xi1上的分段埃尔米特插值函数为Ih(x)(xxi1)2(12xxi)f(xi)xixi1xi1xi(xxi)2(12xxi1)f(xi1)xi1xixixi1(xxi1)2(xxi)f(xi)xixi1(xxixi1xixi4(xhi3xi431(xhi)2(xxi1)f(xi1)xi1)2(hi2x2xi)x)2(h2x2x1)iii34x2i(xxi1)2(xxi)hi4xi13(xxi)2(xxi1)h2i偏差为f(x)Ih(x)1f(4)()(xxi)2(xxi1)24!1maxf(4)()(hi)424axb2又Qf(x)x4f(4)(x)4!24maxf(

30、x)Ih(x)maxhi4h4axb0in1161620给定数据表以下:Xj0.250.300.390.450.53Yj0.50000.54770.62450.67080.7280试求三次样条插值,并满足条件:(1)S(0.25)1.0000,S(0.53)0.6868;(2)S(0.25)S(0.53)0.解:h0 x1x00.05h1x2x10.09h2x3x20.06h3x4x30.08Qhj1,hjjhj1hjjhjhj115,23,33,41145719,22,34,011457fx0,x1f(x1)f(x0)0.9540 x1x0fx1,x20.8533fx2,x30.7717fx

31、3,x40.7150(1)S(x0)1.0000,S(x4)0.6868d06f0)5.5200(fx1,x2h0d1fx1,x2fx0,x14.31576h0h1d2fx2,x3fx1,x23.26406h1h2d3fx3,x4fx2,x32.43006h2h3d46(f4fx3,x4)2.1150h3由此得矩阵形式的方程组为21M0529M11414322M25534M327712M45.52004.31573.26402.43002.1150求解此方程组得M02.0278,M11.4643M21.0313,M30.8070,M40.6539三次样条表达式为(xj1x)3Mj1(xxj)3

32、S(x)Mj6hj6hjMjhj2xj1xMj1hj2)xxj(j0,1,L,n1)(yj)(yj1hj6hj6将M0,M1,M2,M3,M4代入得6.7593(0.30 x)34.8810(x0.25)310.0169(0.30 x)10.9662(x0.25)x0.25,0.302.7117(0.39x)31.9098(x0.30)36.1075(0.39x)6.9544(x0.30)S(x)x0.30,0.392.8647(0.45x)32.2422(x0.39)310.4186(0.45x)10.9662(x0.39)x0.39,0.451.6817(0.53x)31.3623(x0.

33、45)38.3958(0.53x)9.1087(x0.45)x0.45,0.53(2)S(x0)0,S(x4)0d02f00,d14.3157,d23.2640d32.4300,d42f40040由此得矩阵动工的方程组为M0M4029014M14.3157322M23.2640553M32.4300027求解此方程组,得M00,M11.8809M20.8616,M31.0304,M40又Q三次样条表达式为S(x)Mj(xj1x)3(xxj)36hjMj16hj(yjMjhj2)xj1x(yj1Mj1hj2)xxj6hj6hj将M0,M1,M2,M3,M4代入得6.2697(x0.25)310(

34、0.3x)10.9697(x0.25)x0.25,0.303.4831(0.39x)31.5956(x0.3)36.1138(0.39x)6.9518(x0.30)x0.30,0.39S(x)2.3933(0.45x)32.8622(x0.39)310.4186(0.45x)11.1903(x0.39)x0.39,0.452.1467(0.53x)38.3987(0.53x)9.1(x0.45)x0.45,0.53221若f(x)Ca,b,S(x)是三次样条函数,证明:bf(x)abf(x)a2bS(x)2dxdxa2b2S(x)dx2aS(x)f(x)S(x)dx(2)若f(xi)S(xi)

35、(i0,1,L,n),式中xi为插值节点,且ax0 x1Lxnb,则bS(x)f(x)S(x)dxaS(b)f(b)S(b)S(a)f(a)S(a)证明:bf(x)2dx(1)S(x)abf(x)2bS(x)2bf(x)S(x)dxadxadx2abf(x)2b2bS(x)f(x)S(x)dxadxS(x)dx2aa从而有b2b2dxf(x)dxS(x)aab2bf(x)S(x)dx2S(x)f(x)S(x)dxaab(2)S(x)f(x)S(x)dxabS(x)df(x)S(x)abbS(x)f(x)S(x)f(x)aaS(x)dS(x)S(b)f(b)S(b)S(a)f(a)S(b)f(b

36、)S(b)S(a)f(a)S(b)f(b)S(b)S(a)f(a)S(b)f(b)S(b)S(a)f(a)bS(a)S(x)an1S(a)S(xkk0n1xkS(a)S(k0S(a)f(x)S(x)dxxk1)gxk1f(x)S(x)dx2xkxk1)gf(x)xk12S(x)xk第三章函数迫近与曲线拟合1f(x)sin2x,给出0,1上的伯恩斯坦多项式B1(f,x)及B3(f,x)。解:Qf(x)sin,x0,12伯恩斯坦多项式为nkBn(f,x)f()Pk(x)k0n此中Pk(x)nxk(1x)nkk当n1时,P0(x)1(1x)0P1(x)xB1(f,x)f(0)P0(x)f(1)P1(

37、x)1(1x)sin(0)xsin022x当n3时,P0(x)1(1x)30P1(x)1x(1x)23x(1x)20P(x)3x2(1x)3x2(1x)21P3(x)3x3x33B(f,x)3f(k)P(x)3k0nk03x(1x)2gsin63x2(1x)gsinx3sin323x(1x)233x2(1x)x322533x3336x23x2221.5x0.402x20.098x32当f(x)x时,求证Bn(f,x)x证明:若f(x)x,则nf(k)Pk(x)Bn(f,x)k0nnknxk(1x)nk0nkknkn(n1)L(nk1)k(1x)nk0nk!xkn(nL(n1)(k1)1xk(1

38、x)nk1)k1(k1)!nn1x)nkkxk(1k11nn1xk1(1x)(n1)(k1)xk1k1xx(1x)n1x3证明函数1,x,L,xn线性没关证明:若a0a1xa2x2Lanxn0,xR分别取xk(k0,1,2,L,n),对上式两端在0,1上作带权(x)1的内积,得1Ln11a00MOa10MMM11Lan0n12n1此方程组的系数矩阵为希尔伯特矩阵,对称正定非奇异,只有零解a=0。函数1,x,L,xn线性没关。4。计算以下函数f(x)关于C0,1的f,f1与f2:f(x)(x1)3,x0,1f(x)x1,2(3)f(x)xm(1x)n,m与n为正整数,(4)f(x)(x1)10e

39、x解:(1)若f(x)(x1)3,x0,1,则f(x)3(x1)20f(x)(x1)3在(0,1)内单调递加fmaxf(x)0 x1maxf(0),f(1)max0,11fmaxf(x)0 x1maxf(0),f(1)max0,1111f(x)6dx)22(101x)711(127077(2)若f(x)x1,x0,1,则2fmax1f(x)0 x121f10f(x)dx11)dx21(x2214f2(1(x03611f2(x)dx)201)2dx2(3)若f(x)xm(1x)n,m与n为正整数当x0,1时,f(x)0f(x)mxm1(1x)nxmn(1x)n1(1)xm1(1x)n1m(1nm

40、x)m当x(0,m)时,f(x)0nmmf(x)在(0,)内单调递减mnm当x(,1)时,f(x)0nmmf(x)在(,1)内单调递减。nmx(m,1)f(x)0nmfmaxf(x)0 x1maxf(0),f(m)nmmmgnn(mn)mnf1f(x)dx101xm(1x)ndx0(sin2t)m(1sin2t)ndsin2t02sin2mtcos2ngg0tcost2sintdtn!m!(nm1)!11fx)2ndx22x2m(10102sin4mtcos4ntd(sin2t)21022sin4m1tcos4n1tdt2(2n)!(2m)!2(nm)1!(4)若f(x)(x1)10ex当x0

41、,1时,f(x)0f(x)10(x1)9ex(x1)10(ex)(x1)9ex(9x)0f(x)在0,1内单调递减。fmaxf(x)0 x1maxf(0),f(1)210e1f10f(x)dx11)10exdx(x0(x10 x119xdx1)e010(x1)e010e11f2xdx22(x1)20e07(342)4e5。证明fgfg证明:f(fg)gfggfgfg6。对f(x),g(x)C1a,b,定义(1)(f,g)b(x)g(x)dxfa(2)(f,g)b(x)g(x)dxf(a)g(a)fa问它们能否构成内积。解:(1)令f(x)C(C为常数,且C0)则f(x)0而(,)b()()ff

42、fxfxdxa这与当且仅当f0时,(f,f)0矛盾不可以构成C1a,b上的内积。(2)若(f,g)bf(a)g(a),则f(x)g(x)dxa(g,f)b(x)dxg(a)f(a)(f,g),Kg(x)fa(f,g)baf(a)g(a)af(x)g(x)dxbf(x)g(x)dxf(a)g(a)a(f,g)hC1a,b,则(fg,h)bg(x)h(x)dxf(a)g(a)h(a)f(x)abf(x)h(x)dxf(a)h(a)bg(a)h(a)af(x)h(x)dxa(f,h)(h,g)(f,f)b2dxf2(a)0f(x)a若(f,f)0,则b(x)2dx0,且f2(a)0faf(x)0,f

43、(a)0f(x)0即当且仅当f0时,(f,f)0.故可以构成C1a,b上的内积。7。令Tn*(x)Tn(2x1),x0,1,试证Tn*(x)是在0,1上带权(x)1的正交多xx2项式,并求T0*(x),T1*(x),T2*(x),T3*(x)。解:若Tn*(x)Tn(2x1),x0,1,则1Tn*(x)Tm*(x)P(x)dx011Tn(2x1)Tm(2x1)2dx0 xx令t(2x1),则t1,1,且xt12,故1Tn*(x)Tm*(x)(x)dx01d(t1)Tn(t)Tm(t)11t1(t1)222211dtTn(t)Tm(t)112t又Q切比雪夫多项式Tk*(x)在区间0,1上带权(x

44、)1正交,且1x2x0,nm11Tn(x)Tm(x)dt2,nm012,nm0Tn*(x)是在0,1上带权(x)1的正交多项式。xx2又QT0(x)1,x1,1T0*(x)T0(2x1)1,x0,1QT1(x)x,x1,1T1*(x)T1(2x1)2x1,x0,1QT2(x)2x21,x1,1T*(x)T(2x1)222(2x1)218x28x1,x0,1QT3(x)4x33x,x1,1T3*(x)T3(2x1)4(2x1)33(2x1)32x348x218x1,x0,18。对权函数(x)1x2,区间1,1,试求首项系数为1的正交多项式n(x),n0,1,2,3.解:若(x)1x2,则区间1,

45、1上内积为(f,g)1f(x)g(x)(x)dx1定义0(x)1,则n1(x)(xn)n(x)nn1(x)此中n(xn(x),n(x)/(n(x),n(n(x),n(x)/(n1(x),(x,1)/(1,1)1x(1x2)dx11(1x2)dx101(x)x(x2,x)/(x,x)x3(1x2)dx1x2(1x2)dx10(x,x)/(1,1)1x2(1x2)dxn(x)1(x)11(1x2)dx11628532(x)x2252(x32x,x22)/(x22,x22)555512x)(x22)(1x2)dx(x3155122)(x22)(1x2)dx(x15502(x22,x22)/(x,x)

46、5512)(x22)(1x2)dx(x21551x2)dxx2(11136171670153(x)x32x217xx39x570149。试证明由教材式(2.14)给出的第二类切比雪夫多项式族n(x)是0,1上带权u(x)1x2的正交多项式。证明:sin(n1)arccosx若Un(x)1x2令xcos,可得1Um(x)Un(x)1x2dx11sin(m1)arccosxsin(n1)arccosx11x2dx0sin(m1)sin(n1)1cos2dsin(m1)sin(n1)d0当mn时,sin2(m1)d01cos2(m1)d022当mn时,sin(m1)sin(n1)d01cos(n1)

47、0sin(m1)dn11cos(n1)dsin(m1)n1m10n1cos(n1)cos(m1)dm1cos(m1)d1sin(n1)0n1n1m11)(n2sin(n1)dcos(m01)(m12sin(n1)sin(m1)dn)0101(m1)2sin(n1)sin(m1)d0n10又Qmn,故(m1)21n1sin(n1)sin(m1)d00得证。10。证明切比雪夫多项式Tn(x)满足微分方程(1x2)Tn(x)xTn(x)n2Tn(x)0证明:切比雪夫多项式为Tn(x)cos(narccosx),x1从而有TnTn(x)gg1sin(narccosx)n()1x2nsin(narcco

48、sx)1x2nsin(narccosx)n2cos(narccosx)(x)3x2(1x2)21(1x2)T(x)xT(x)n2T(x)nnnnxsin(narccosx)n2cos(narccosx)1x2nxsin(narccosx)n2cos(narccosx)1x20得证。11。假设f(x)在a,b上连续,求f(x)的零次最正确一致迫近多项式?解:Qf(x)在闭区间a,b上连续存在x1,x2a,b,使f(x1)minf(x),axbf(x2)maxf(x),axb取P1f(x1)f(x2)2则x1和x2是a,b上的2个轮流为“正”、“负”的偏差点。由切比雪夫定理知P为f(x)的零次最正

49、确一致迫近多项式。12。采纳常数a,使maxx3ax达到极小,又问这个解能否独一?0 x1解:令f(x)x3ax则f(x)在1,1上为奇函数maxx3ax0 x1maxx3ax1x1f又Qf(x)的最高次项系数为1,且为3次多项式。3(x)123T3(x)与0的偏差最小。(x)1T(x)x33x3434从而有a3413。求f(x)sinx在0,上的最正确一次迫近多项式,并预计偏差。2解:Qf(x)sinx,x0,2f(x)cosx,f(x)sinx0a1f(b)f(a)2,bacosx22,x2arccos20.88069f(x2)0.77118a0f(a)f(x2)f(b)f(a)gax22

50、ba20.10526于是得f(x)的最正确一次迫近多项式为P1(x)0.105262x即sinx0.105262x,0 x2偏差限为sinxP1(x)sin0P1(0)0.1052614。求f(x)ex0,1在0,1上的最正确一次迫近多项式。解:Qf(x)ex,x0,1f(x)ex,f(x)ex0a1f(b)f(a)e1baex2e1x2ln(e1)f(x2)ex2e1a0f(a)2f(x2)f(b)f(a)gax2ba21(e1)(eln(e1)21)2ln(e1)于是得f(x)的最正确一次迫近多项式为P1(x)e(e1)x1ln(e1)22(e1)x1e(e1)ln(e1)215。求f(x

51、)x43x31在区间0,1上的三次最正确一致迫近多项式。解:Qf(x)x43x31,x0,1令t2(x1),则t1,12且x1t122f(t)(1t1)43(1t1)3122221(t410t324t222t9)16令g(t)16f(t),则g(t)t410t324t222t9若g(t)为区间1,1上的最正确三次迫近多项式P3*(t)应满足maxg(t)P3*(t)min1t1当g(t)P*(t)1T(t)1(8t48t21)32348时,多项式g(t)P3*(t)与零偏差最小,故3*(t)g(t)13T4(t)210t325t222t738从而,f(x)的三次最正确一致迫近多项式为1*,则f

52、(x)的三次最正确一致迫近多项式为16P3(t)P3*(t)110(2x1)325(2x1)222(2x1)731685x35x21x1294412816。f(x)x,在1,1上求关于span1,x2,x4的最正确平方迫近多项式。解:Qf(x)x,x1,1若(f,g)1f(x)g(x)dx1且01,1x2,2x4,则22022,12,222,9(f,0)1,(f,1)1,(f,2)1,23(0,1)1,(0,2)21,2)2,(,57则法方程组为222135a02221a13572a222215793解得a00.1171875a11.640625a20.8203125故f(x)关于span1,

53、x2,x4的最正确平方迫近多项式为S*(x)a0a1x2a2x40.11718751.640625x20.8203125x417。求函数f(x)在指定区间上关于span1,x的最正确迫近多项式:(1)f(x)1,1,3;(2)f(x)ex,0,1;x(3)f(x)cosx,0,1;(4)f(x)lnx,1,2;解:(1)Qf(x)1,1,3;x若(f,g)3f(x)g(x)dx1且01,1x,,则有22,2260212,3(0,1)4,(f,0)ln3,(f,1)2,则法方程组为24a0ln3264a123从而解得a01.1410a10.2958故f(x)关于span1,x的最正确平方迫近多项

54、式为*S(x)a0a1x(2)Qf(x)ex,0,11若(f,g)f(x)g(x)dx0且01,1x,,则有21,121022,3(0,1)1,2(f,0)e1,(f,1)1,则法方程组为11a02e111a1123从而解得a00.1878a11.6244故f(x)关于span1,x的最正确平方迫近多项式为*S(x)a0a1x(3)Qf(x)cosx,x0,11若(f,g)f(x)g(x)dx0且01,1x,,则有21,21,02123(0,1)1,22(f,0)0,(f,1)2,则法方程组为11a002211a1223从而解得a01.2159a10.24317故f(x)关于span1,x的最

55、正确平方迫近多项式为S*(x)a0a1x1.21590.24317x(4)Qf(x)lnx,x1,2若(f,g)2f(x)g(x)dx1且01,1x,则有21,27,02123(0,1)3,2(f,0)2ln21,(f,1)2ln23,4则法方程组为13a02ln212337a12ln2423从而解得a00.6371a10.6822故f(x)关于span1,x最正确平方迫近多项式为S*(x)a0a1x0.63710.6822x18。f(x)sinx,在1,1上按勒让德多项式睁开求三次最正确平方迫近多项式。2解:Qf(x)sinx,x1,12按勒让德多项式P0(x),P1(x),P2(x),P3(x)睁开121(f(x),P0(x)sinxdxcosx012211xsinxdx8(f(x),P1

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论