mo © & one o e e xan Ne CHAPTER | ANSWERS TO SELECTED PROBLEMS Section 1.1 2a). 346). Xa). pid = 144-0 p(x) = 1+ Hx 1) = fF paix) = 1+ ('P)x + (PF pix) lat bt at glx) © 1+ bx + Le, for small values of x 4G) = Fla) + [x + a)/(0 ~ aILs) ~ f(@)) (0); solve for ba, bs from the linear system by~ by "(y= by +22534 ANSWERS TO SELECTED PROBLEMS Section 1.2 1 = psx) = Set cbetween Oand r - 1 max le ~ yO] = wy 20.133 FromTable 1.1, e! — py(1) = 0.05161. co 4. MIA) coor for xi < 0.245 5. Usedegree = 7. 9a). 3 1b) na) =x Fe SO yayint! + Roneitx) a2 gy rye aos wood a] 19. Withnodd, fx) =2)e pS eee gS 305 n Repsile)= (-1! [ lo Section 13 4. Use degree = 8. 6. Using (1.35)1.38), ae ote trth by (x = 1a) Form p'(x) by direct differentiation in (1.38), and then let x = z, 9. Seven multiplications are needed, if we assume that all coefficients have been calculated already. WW. fX)=l+e(St2(-1422)), < on 2) 1(b), (12)yohasthe IEEE double precisionhexadecimal representation 4028000000000000. The significand (in binary) is 1.100... The biased exponent (in binary) is 10000000010. 2. m= 31 bitsin signticand, M = 2" = 2.15 x 10, -2" <¢ <2 Seetion 1.2 Me. x x 7182818... (not repeating) 271428571... (Gepeating)Oo og © gare @ ora ANSWERS TO SELECTED PROBLEMS 535 ar — 44 = 1000996... Re(x,) = 0.00147 214 has three significant digits. 3 im N(@) = Ne=2N(O) S(a). Sid). nae Ge). altats = cos(4" 7. Using a "Taylor polynomial of degree 4 for cosu about u = 0, obtain ——~ cole ee es oct ee pe agit gto Osa 10. The roots are x; = 20+ 399, xy = 20 ~ 399. To calculate x2, use 399 = 19.975 and then form x3 = =~ a 0.025016 20+ Viv 39875 19. ForlEEE normalized double precision floating-point representation, itunderfiows if Lx] < M2 = 1.72 107 Section 2.3 a), 2.05265 < x7 + yr ¢ 2.05375 8.4725 xr _ 8.4735 MO D685 * yy * 0.0685 ta1as6s F< ian S{a). Use (242), with x4 5 1.473, lar — x 0.0005, ‘Then costs) ~ e0s(t4) = —sin(e) (er ~ x1) looser) ~¢0s¢x4)|< [sin( 1 4735)] (0.0005) < 0.0005, 50. x4 = LATI2, lar —x4l $ 0.00005, 20005 r a Hog(s7) ~ logtsa)l = 1 15(e), Hint: Use the floating-point representation of z 2a), 1.839286755 2c}. 0424031039 1 « : : 4 wet fin= +25] on: 8 Root 1 ~0.5884017765, 0 03264020101 | 30 -0.1832913333 BUD dost =a Fae 2 Use [a,b] = [0,05]. Then |g'(x)] < 032 fora 0 for all x, the solution a must be positive. Thus considering x > 0, we ~ haveO < g(x) < I. Then{a, 6] = (0, 1] satisfies (3.37) in Lemma 3.4.1, but not (3.38) in ‘Theorem 3.4.2. For the later, consider a= min g(x) b= mangle) 5. Use(a,b]=fe-Idl,c+ ill 8(b). Converges; order of convergence u 4 for quadratic convergence,& a eeeraer 2 CHAPTER 4 ANSWERS TO SELECTED PROBLEMS 337 13(a). The Aitken error estimate is -9.175 x 10%, 83 4 Section 3, 4, Fora(0) =3,a(¢) ¥3-15.2€ = 303. 7. ale) — 5% 312Se, anill-conditioned problem 8, The root has multiplicity 4 1@). Ai) =2-x 1b). f(x) 2a). p(x) = 3 ~ costrex) + 2sin(x2) 2 xt tax -3 Bla). x(x) = Px) art! 10. Lox) = Go ANG9 = m2)(%0— 5) 16 gx) =4v-ar-1 19. flav, 4) = 2.363200, r(: £2) =236u61 Flt0, 51, #2) = 119349521 2(a). px) = f(z) — f lao) is a polynomial of degree m, and plxo) = 0. Thus, (x ~ x0) isa factor of p(x), and p(x) (x — sa) isa polynomial of degree m ~ I £0) = 2feo, x1 + f'G0} 2B). (= 20)? ue ee oe 102028 1 06 2 03 030 with Dy, and D®y, denoting frst-andsecond-order divideddifferences. Then P, (0.15) = 0.22, Py(0.15) = 0.2175. at 3A. Using (4.24), with m =n, we get fl¥0, 16.6.4) = =538 ANSWERS TO SELECTED PROBLEMS Section 4.2 Ifa), 1.25 x 10-5 1b). 6.42 x 10% Ha). 1.07 x 107 3b). 1.28 x 107? S(a). fh < 0.00632, say, A = 0.006 or 0.005. 10, Ate/24 Wa le - Pela) s Osxsl Section 4.3) Pelion Me). 50) = (Q-x)) + (4x-9-Q-2), 3(e). The equations for My = s(x) ate IMs + SMa =1 biti} Ms + Me = The solution is Mh = 8 =—8.Ma= 7. The equations for M, 5M + Lh = EM, + 3M + 1M = 4 Lats + EM + Me fis + Ma ‘The solution is, 2 (a) are My = 173/120, Ma=7/60, My = 11/120, Using the resulting value of s(x), we obtain 5 (3) = 0.652604 True f (3) = s (3) £0.40364 Tre f(}) =04 5 (3) $0.284896 True f (3) = 10. Yes 15. 8 = 20, Nota natural cubie spline. Section 4.4 24a). max, jtan“!(x) ~ 1)(2)| + 0.215 (x) aly xode Mz = 1/60oon ANSWERS TO SELECTED PROBLEMS 539) 210). max, jan-"(x) ~ ma(x)] = 0.00485 4, Use trigonomettic identities. For example, cos(~x) = cos(x), cos(x + 2nm) = cos(x), any integer n Sfa), From (4.82), px(f) < 0.00198. 0.00137 S(b). max _Jeos(x) ~ my(x 8(a). Leto = tan“!(x), x > 0. Then <8 < 1/2, and tan(@) = x, Fromusing aright triangle with angle 4 itis clear that tan|(r/2) ~ 8] = 1/x, or (x /2) ~ @ = tan”"(1/x), as desired Section 4.5 L Ts(x) = 16x5 = 20x) + Sx 3 my cos (mm jn), Serer SeeeaTE aoe BC Sag (x) = 2Sy(x) = Sat), 2B Section 4.6 (V3, xq = cost3n/4) = 1/2 2 x = cos(n/4) ene © 1.2606 + 1.08544 Hom 3(b). Using the zeros of Ts(x), Py(x) * 0,9670576x — 0.1871296x° BB, |tan~' (x) — P3(x)| = 0.00590 Pix) sel + (x a1) 4 User = 3 +.4),-1 x 5 1,and approximate f(x) = 7 on [1,1] h pe ji 2 3 4 s 6 | | | 14E-S | 10. Evaluate p(x) and equate itto px). Section 4.7 1. (x)= (de - 10) +4 (~6e +18) = 0.873127 + 1.690309"540 CHAPTER 5 ANSWERS TO SELECTED PROBLEMS 1535). 3 (35 3a). (x)= Aa) =(-2 + =) +8 e - = 2,6922925x - 2.8956048x? 6& 1= PQ) 1 =P) x= 5 (2PA(3) + Pale) a x $(2Py(x) + 3P,(x)) 1. pla) = (ao + fan) Pol) + a) Pi(x) + JanPr(x) @ 525 ‘ a Section 5.1 : LT) S0697004 1-2) Sf) = 0683254 1 -S\(4) 2a), : * th Error Ratio 2 16sIe6 282 4 3250505 195 80 8 lees -3E-1 608 Wists 7338-2439 ! 2 ists 19-2408 61306879 AMSE-3 402 3038057 LE-3- 26 Laer 278-4400 si 1065-3 400 3a). 5. Error Ratio masa 214 a ee ee -369 8 LossoN6ss1s 29-126 1 asemmeanas 1-3 . 2 1BOTBAN66S3—SSNE~8 % —LsorsseneTy BRE S163 1 130R 1S E—6 BL 256 1.3028035531 131-7160 sig 13025936761 8206-9 160 10(b). 4(f) = 0.69817460, 1 — B4C/) = -0.0000274 tee Bo€ Cy © e 6 € ANSWERS TO SELECTED PROBLEMS 54 Section 5.2 1a). 100), 3a). 6a). 6(d). ‘13(b). 14(a). (a). ah?/24 H/6 EN = Fyle ou Ef) = -0.01778, True Efy(f) + -0.0179 EX = Be han D> 75" Choose h < 0.031, orn > 65. The number of nodes = n-+ 1 > 66. 182 396 for Simpson's rule he Te ‘Apply (5.39) and simplify pals From Problem 2(a) in Section $.1, with m = 32, 1 ~ Typ © 4(Tsz ~ Tis) + -0.01847 True J - Ty; = —0.0179. Section 5.3 L Ya). 10(a). 1000). 10(c). fy = 2,3508369, 1 ~ ty = 0.000655, n 1, Error 2 =19,244871326 205 3 9.ogv728708¢ 7.79. 4 09220525215 3.806 = 1 5 11462068487 166-1 6 13298098510 ~2.74E -2 7 13003438200 205E-3 8 Laena7e10s1 844 ~5 9 1.3023918529 LBB 6 0 13023936884 4.108 ~ 9 [ roe gars eee wt) twat = f : wut + wad = ‘ wx} + wx} = 968912, error = 0.0228ANSWERS TO SELECTED PROBLEMS ion 5. Ha). Ha). 4 Error h D,fO) Estimate Exror Ratio 7 1051709 =S.008-2 5.17 -2 os 1.008422 -250B-2 2548-2203 0.9025 L.oiz6os-1.25E-2 1268-2 2.02 ois 1.006276 25E-3 -6.28E-3 2.01 0062s 1.003132 -32E-3—-3.3E-3 2.00 Error h DFO) Estimate Error Ratio ou Looie6r7S = L67E-3 -L.6TE=3 ons Looser? = 417-4 4.1784 4.00 0.025 10001047 L044 1088-4 4.00, 00125 1,00002604 260-5 =2.60E-5 4.00 0.00625 —_1.00000651 ~6SIE~6 -651E~6 5. f'G2) ® Dy flan) = (I/12AL Fao) ~ Bf) + Bf es) ~ fet £32) ~ Dy fn) = (8/30) f(O), 20S ¢ a 1 Denote By f(x) = 2D, f(x) = Day f(x), with Dy, f(x) defined by (5.76). A Ds 01 0.58243 0.05 052184 oms = 951077 ois §— -0.50s40 Dif) 0.50045 0.50010 0.50003 Far by f09 o.00045 0.00010 6.00003 10d). fe ® Hi-se+3m 44s +2m ~ Spee +2F001 angel # (x) - Bf s a | ‘Dj f(x) with rounded entries, and 6 = max{teo|, |¢1|I- M4. — With = 02, DP? f(0.5) + 00775. The part ofthe etor due to rounding in table entries Aa). Write p(x) = ap +.ayx + apx? +ayx9, Then ap, a), 2, a5 satisfy ay + ax) + ax? + asx} 48 4(0.00005) G=0,1,23 4.00 3 - "tol + Prk sess th, with D(x) the version of5. e ANSWERS TO SELECTED PROBLEMS 543 I te=4 xi +34 +a Sy) +3ty =4 sign = 1; b = zeros(n,1) A = zeros(n,n); tor i DG) = sign; sign = ~sign; AGED = G/DeLt AQH-1a) = AG 1D) end Section 6.2 1a, I 4] a b+2c 10). [5 “ | ea we. a=|-; 4-3], 8 ‘ 44 44 e rs Write A = [Asi, Asa. Aas], with A,, denoting the jth column of A, a column matrix. Then AD = [iyAyisdaAcq.oAus). Column j of Ais mukiplied by 4 € 8 A sun’, Bs A= (ww! ww") = ww" ww" = (ut = wal = A & mp 1. a=db e 28. (1, 1,-2.-2F € Section 6.3 & 21 -t ty 6 10). 40 -1 1] -6 82 2 -2| 16 ns-2n=2n Ifthe third equationisantultipleofthe second, then there are an infinite number of solution. Solve for xy and then substitute into the ist equation, to determine the relation between x, and xy. If the third equation isnot « multiple ofthe second, then the system i inconsistent. and there is no solution. To betterunderstand the possible behavior, try particular systems ‘ofthe given special form.S44 ANSWERS TO SELECTED PROBLEMS 0.1 -0.1 04 -01 . 01 4. eet eer -05 -05 -05 -1S -15 -15 -15 -25 -25 “15 7 MDA U) MD(b—> 8) MD(g > x) s I 1 oo | 2@). Ls to}, uso -2 1 . 4-3 1 oo 1 Lo oo. [ {3} ‘This is not the only possible value for L & fi-hpche 1a). MDA LU) = 2-2 1b), MD(x) = 3n-2 Section 8.5 Bla). cond) = 289 5. 6 1b). 2, For Jacobi iteration, the ratios approach 0.4069, and {| For Gauss-Seidel, the ratios donot converge; but | M\| = 1/3. Ma). |My = 1/2 4b). My = 1/2 x For Jacobi, the ratios alternate between 0 and 1/2. Me ele v2( mm @ @ Qe a a foo ae oo CHAPTER 7 CHAPTER 8& ANSWERS TO SELECTED PROBLEMS Section 7. 1. f*(x) = 2.0810909r — 1,0837273, 8 = 0.236 3. f(x) 7.3301079x? ~ 2,2158961x ~ 0.98132527 Vs Py » 4 Ed Yl [=] m syle » 5 wlya@yds, 651,23 10. na tb re = Dy i iat eens yete Lye = i= = ‘ Section 7.2 Ha). f0) =¥7- 52-6, =6, = of [25], 0 2 v@=[1,0,-17, 2 oa 11b), Ay = 12, =H EV RL LE IMA). Ay = 3 A= 2 AHH L4H IMD.) =15, = 5, = SMe = HT, VO aL Section 7.4 L Another root is a = (2.1372167369636, 1.05265196281 14). 2a). The Four exact roots are (V2.5, £VTS) and (£23, ¢VTS), Ab). One of the roots is (1.7701689217513, 0.4654304428338). Section 8.1 Ufa). ¥(x) = sin(x) +e" 1b). ¥(x) = x/(x-0.5) -546 ANSWERS TO SELECTED PROBLEMS. Ba). ¥(x)=ce¥ + fle — Ij, general solution. ¥(x) = U1 + (1 )]e — 1 4a). Yar ee S(a). af /8z = —1, equation well-conditioned. Section 8.2 lt ae e y: Sere enna Yo) YG) yo200) z 2 ons -3308-2 ae iver (are 2 & 6 140565 9.708 -3 8 1.44684 6.386 -3 . . 1 ams 44-3 2). ee x Yor) Y¥ea)—yost) YG) - yaoas() 2 assis 2BE-1 AME 3 4 1.41045, 1ME+0 8.67E-3 & 6 0.68075, 6.39E +0 ~LBE-3 € 8 0.84386 3.25E +1 "6366-3 1 1.38309 LSE 42 807-3 Section 8.3 = 3 sop (I/te +1) = 1, foranyb >0 paxlY"(@)] = max Qa/lx +1) = 20/06 +1) 2b YD) = wis Yb) — yp(b)i Sh Al For h = 0.05, b —_¥(6)~ (0) __ Error Bound i 1 01s 0.0480 2 ness 0213 i 3 as one ‘ 0.199 aie 5(b). Use K 2max YQ) =4, forb > 1. t 1) = may (roe 28 a5 001 = as a 1.46, forb > 0.41,HGe rene eae ANSWERS TO SELECTED PROBLEMS 547 7. Solve Di(x) =AD(x)~ fsintx), DO ‘The solution D(x) = {[2.sin(x) + cos(x)) ~ e*) /1202 + 1)] and ¥ (a) — yale) & Dah For = —1 and h = 0.1, the answers areas follows; 10(a). x Vix) ysl) val) = yas) x) Yea) = jaw 2 -8IE~3 -82E-3 1.1070296 LIVE — 4 4 -417E=3 4166-3 13258165, LE 6 6 2448-3 -2438-3 14056549 -7.28E-6 8 ~160E-3 1.608 -3 Lagsagr2 SBE -6 10 = 1146-3 1-3 Lani317—~406E-6 Section 8.4 7. Wegive the values obtained with the backward Euler method when, = ~S0. Compare these with the values for A = —1, ~10 for the same values of A. Note that the errors decrease by a factor of about 10 when h decreases from 0.1 to 0.01; this is consistent with the method being of order | in ts convergence rate. Section 8.5 2a), - Following are some results for h = 0.05 and 2h = 0.1. The Richardson error estimate (8.76) is also included.ANSWERS TO SELECTED PROBLEMS x Y(x)—yoalz) — ¥()— yale) — Ratio SlyaCx) — yale) 110SE-4 58-5 40 5.128 — LSE 4 4288-5 41 4408-5 6 1o1e—4 2488-5 41 2558-5 8 640B~S 1sTB-5 41 161g 5 0 asre-s. O78 -s 41 8 x F(x) yale) dys) — peal) jaa) ¥(x) = 2 1BIB-4 2318-4 04931513 7286-7 4 2-4 OE-4—=1.8108900—=1.61B-S 6 40RE 4 4196-4 6807832 1.16E~s 8 5.68E-5 5.08E - 5 0.843852 6.05E-6 WW 382E-4 378E-4 ~1.3830760 1.6TE 5 hh 6. With yy = 1, the Runge-Kutta method is yey = Ya + Af ( ty det Flt v0) 7. Using the formula for y: = | in Problem 6, we obtain the error results below for f= 0.05, 2h = 0.1. Richardson's error estimate is includes. x YOe)— yale) YO) yalx)— Ratio Fyn) — yaa] 2 1466-3 350E-4 4 3.098 -4 4 6.6864 -161E 4 40 1676-4. 6 -831E-4 -1g1B-4 42 “2-47 8 La7E-3 3B - 4 3466 - 4 10 ~30RE-4 -801E-5 38 = 1608-5 8a), For & = 0.05 and 2h = 0.1, x) — yaar) Px) = yal) Ratio fy) ~ yaad] 2 6468 - 4 1S6E-4 y 4 Lee —4 4 251E=4 6.085 4 6348-5 6 1268-4 307E-5 4 3.8E 5 8 7498-5 192-5 4 189-5 1208-5 4a 125E~5 9, The results for h = 0.125 and 2h = 0.25 are given below, along with Richardson's error ‘estimate for yy (x), taken ftom Problem 4ANSWERS TO SELECTED PROBLEMS 549 C 1 . x ¥(x} = yl) (x) ~ yal) Ratio Lyne) = yaad] eee nn _ ~ pesesabe terete a 300-5 1698-6 8 1908 ~ 6 € 4 3648-5 2.008 - 6 ” ~208E-6 e 6 =1MME-6 3AIE-8 — ~42" ~985E -8 . 8 34E-5 2.168 -6 " 2356 ~6 & 0 297-5 -183E-6 6 ~ 1868 ~6 € ‘The Richardson extrapolation isnot justified, since Ratio < 0. 13(a). The true solution is u(e) = =98{1 - e°*"], r= 0. é & Section 8.6 Mla), Leth = 0.05, 2h = O11 x Y(q)— yao) Y(x) — yw) Ratio Ghat = yard] : 2 S.10E~4 1286-4 40 124 4 4sB-4 LOTE~4 4a 110E-4 6 215-4 628-5 4 6.408 -~§ & 8 LIE =4 3938-5 4 4088-5 € 1.09 ~4 2.688 ~ 5 41 2768-5 6 Ya). Let h =005, 2h = 0.1. Richardson's error formula is from Problem 4 of Section 8.5. ‘The orderof convergence is p= 3. ¥, and ¥, obtained using RK method of order 2 with nel x Y(x) - yal) Ya) — yal) Ratio HL%0) — Yon] Lae 4 LES BS 1.605 ~$ 6 4 SITE ~6 658-7 19 645E-7 € 6 1556-6 1656-7 94 1986-7 8 =I E-6 1.9467 89 27-7 & 10 -146E -6 =156E-7 87 -1.E-7 a). Leth = 0.05, 2h = O11 c 6 . a x Yx)— yo) FO)— yale) — Ratio Lyle) — yaa) 2 -14TE 4 -34E-5 ar 3866-5 © 4 = 1,08E - 4 -2ME-5 45 ~2.80E - 5 i 6 ~6.1E-5 ~13TE-S 45 -1.58E-5 8 -38IE-5 -8.61E-6 44 9.846 ~ 6 & 2586-5 ~S85E-6 44 6.968 ~6ANSWERS TO SELECTED PROBLEMS. | | (a), The errors forthe three stepsizes are given below. Sharply differing values of the erors are due to differen stability regions ofthe methods, (x) = youl) ¥@)~ yomls) 381B +13 13TE~4 ie 206E +30 L33E-4 | “ VE +47 1B =4 1é S.98E +68 LTE 4 ¢ 0 3.03 + 80 LME=4 Section 8.7 Lo Wsh--2et42, HOT YysM-h-2et +l, Hs] Mb). Let) = ¥, Ya = ¥", Then ,f or 0 fa rls a} +fncow), 7=[3] ‘7(b). Use the notation of 3(0). The error estimate is YE) = yale) yale) ~ yas) Errors in ¥i(x) = y(x) Brorestimate e005 A=O0S toch = 0.005 2 “31E-3-16SE-3——347E-3. ‘ 200-2 4.068=2 108-2 s ‘ 258-2 1288-2 1288-2 8 G6 MEd aed was 5 1 SN0E~22SIE~2 1.20282 8(b). Use the notation of 3(b). The error estimate is YQ) = wile ® shoe) = yA] ee fANSWERS TO SELECTED PROBLEMS 551 e 6 - - v0) € Error estimate gr A= 0.025 fork = 0,025 s & 2 4,06E - 5 4aMME- 5 a es 4 -7.91E-3 ~202E~3 5.045 = 4 -S.06E = 4 ~ 6 4.778 -3 1326-3 3.44E ~ 4 3.278 = 4 ¢ 8 3986-3 90-4 AISE-4 2B 10 -80SE-3 2088-3 S224 5.194 = 9a). CG =3G a Section 8.8 3a) : 3b). 3. € 7. The true solution is ¥(x) = xe, Numerical errors ¥(x) — ya(x) at selected points are. listed in the following table. € x hall h=1/20 k= 1/40 € 12 (3408-3 «BASE—4 DE - 14 6306-3 157-3 3.914 e 16 TMIE=3 L923 484 18 GME-3—1SSE-3—387B-4 & © = 5, the numerical values are @ - € 1.1888 € 1.3294 13639 1.5294 1.5639 v 758.76 oo552 ANSWERS TO SELECTED PROBLEMS 7. Themaximum numerical solution errors are n MaxError Ratio 4 276IE-1 8 S8NE-2 373 16 L4E-2 3.93 2 37087-3400 8 For? , © S & oc. Approximate vertical displacement ec. aon | |55a ANSWERS TO SELECTED PROBLEMS Poa 6S é APPENDIX E 1b). (5.8125)i9 3b). (100.012 Xd). = (1.001001); 4, (08) = (0.110011001100....) ' 8 67€ e © ” eaeeego Qa mn > BIBLIOGRAPHY [1] M. Abramowitz and I, Stegun. Handbook of Mathematical Functions, Dover Press, 1964, New York, [2] M. Allen III and E. Isaacson. Numerical Analysis for Applied Science, John Wiley, 1998, New York. [3] E. Anderson, Z Bai, C. Bischof, J. Demmel, 1. Dongarra, J. DuCro2, A. Greenbaum, S Hammarling, A. McKenney, S. Ostrouchoy, and D, Sorenson. LAPACK Users’ Guide, SIAM, 1992, Philadelphia [4] U. Ascher, R. Matthelj, and R. Russell. Numerical Solution of Boundary Value Problems for Ordinary Differential Equations, Prentice-Hall, 1988, Englewood Cliffs, New Jersey. {5} K. Atkinson. An Induction to Numerical Analysis, 2nd ed., John Wiley, 1993, New York [6] 0. Axelsson. erative Solution Methods, Cambridge University Press, 1994, Cambridge, UK. [7] A. Bjorck. Numerical Methods for Least Squares Problems, SIAM, 1996, Philadelphia. {8} P. Bogacki and L. Shampine. A 3(2) Pair of Runge-Kutta Formulas, Appl. Math. Let, 2 (1989), 321-325, 555556 Bibliography [9] C. de Boor. A Practical Guide to Splines, Springer-Verlag, 1978, New York. {10} C. Chui, An Jniroduction 10 Wavelets, Academic Press, 1992, Burlington, Massachusets [11] P, Davis and P. Rabinowitz, Methods of Numerical Integration, 2nd ed., Academic Press, New York. {12} J. Demmel, Applied Numerical Linear Algebra, SIAM, 1997, Philadelphia. {13} J. Dongarra, I. Duff, D. Sorensen, and H. van der Vorst. Numerical Linear Algebra for High-Performance Computers, SIAM, 1998, Philadelphia, [14] J. Dormand and P. Prince. A Family of Embedded Runge-Kutta Formulae, J. Comp. Appl ‘Math 6 (1980), 19-26. {15] W. Gander and J. Hrebicek. Solving Problems in Scientific Computing Using Maple and Matlab, 3rd ed. Springer, 1997, New York. [16] W. Gautschi. Numerical Analysis: An Introduction, Birkhiuser, 1997, Boston. (171 G Golub and C. Van Loan. Matrix Computations, 3d ed, John Hopkins University Press, 1996, Baltimore. [18] D. Higham and N. Higham. Matlab Guide, SIAM, 2000, Philadelphia, [19] A. Householder, The Numerical Treatment of a Single Nonlinear Equation, McGraw-Hill, 1970, New York. ‘ (20) A. Iserles. A First Course in the Numerical Analysis of Differential Equations, Cambridge University Press, 1996, Cambridge, U.K. [21] C.Kelley erative Methods for Linear and Nonlinear Equations, SIAM, 1995, Philadelphia, [22] R. Kress. Numerical Analysis, Springer-Verlag, 1998, New York. (23] A. Krommer and C. Ueberhuter. Computational Integration, SLAM, 1998, Philadelphia (24) P, Marchand. Graphics and GUIs with MATLAB, 2nd ed., CRC Press, 1999, Boca Raton, Florida, (25] K. W. Morton and D. F. Mayers. Numerical Solution of Partial Differential Equations: An Inroduction, Cambridge University Press, 1994, Cambridge, U.K. {26] J. Nocedal and S. Wright. Numerical Optimization, Springer-Veriag, 1999, New York [27] M. Overton. Numerical Computing with IEEE Floating Point Arithmetic, SLAM, 2001, Philadelphia. {28] B, Parlet, The Symmetric Eigenvalue Problem, Classics in Applied Mathematics, SIAM, 1998, Philadelphia, [29] M. Powell. Approximation Theory and Methods, Cambridge University Press, 1981, Cam. bridge, U.K. [0] A. Quarteroni, R. Sacco, and F. Saleri, Numerical Mathematics, Springer-Verlag, 2000, New York (31) L. Shampine. Numerical Solution of Ordinary Differential Equations, Chapman & Hall, 1994, New York. (32) A. Stroud and D. Secrest. Gaussian Quadrature Formulas, Prentice-Hall, 1966, Englewood Cliffs, New Jersey. [53] L. Trefethen and D. Bau, Numerical Linear Algebra, SIAM, 1997, PhiladelphiaLU factorization, 283 =6 Absolute stability, 395 ‘Adams-Bashforth (AB), 423 ‘Adams-Moulton (AM) methods, 423, Aitken's etor estimation formula, 345 Aitken's extrapolation formula, 245 ‘Appeoprximation of functions best approximation, 159 Approximation of functions, 159 leastsquares approximation, 178 minimax approximation, 160 minimax error, 160, 163 near-minimax approximation, 171 Taylor polynomial, 18 Average value, 498 Back substitution, 265 Backward Euler method absolute stability, 397 convergence result, 397 error expansion, 397, euler back, 399 firstorderaccuracy, 397 predictor formula, 398 Binary number sytem, $25 arithmetic operations, 527 conversion from decimal, $28 ‘conversion from hexadecimal, $30 ‘conversion to decimal, $26, conversion tp hexadecimal, 530 Binomial coeffcient, 13 Bisection n bisect, 75 convergence, 4 errr bound, 74 bits, 525, Chebyshev polynomials, 165 definition, 166 least-squares date iting, 226 rminimum sie propety, 168 orthogonality relation, 169 second kind, !70 triple recursion relation 168 ers, 174 Chopping, 39 error, 39 CConttion number, 62 Diresin elds, 376 Divided ifferences, 124 aivaits 127 festorder, 124 igher order, 125 properties, 126 second-order, 125 thiedorder, 125 Dot product, 68 Eigenvalue, 239 ‘dominant, 337 INDEX Ejgewale problem, 327 characters polynomial, 235 01g, 45 ronsymmetic matrices, 339 power method, 340 symmeti mates, 37 eigenvector, 323 Elliptic equations, Se ao Poisson equation’52 Bro 8 absolute eos 42 blunder, 45 discretization, 47 Joop evar. 67 Jos. significance ero, 48 machine, 46 mathematical approximation, 47 rmodelingerrors, 45 snise, ST cverfow errors, $3 physical dat, 46 programming ero, 46 propagated ero, 57 relative err, 43 roat-mean square, 178 truncation, 47 underflow ero, 52 Euler's method, 379 ssymprotc error analysis, 90 teckward Euler metho, 394 convergence tee, 288 error bound, 388558 Index Euler's method (Cont) ‘euler.tor, 384 forward Euler method, 383, Richardson'serrorestimae, 391 Richardson's extrapolation formula, 391 truncation ero, 386 Euler-MacLaurin expansion, 212 Explicit method, 469 Fibonacei numbers, 97 Finite cifference method, 453 ‘heat equation, 466 Poisson equation, 453 wave equation, 487 Fixed point, 98 existence, 98 Fixed point iteration, 97 Aiken's eror estimate, 104 Aitken's extrapolation formula, 104 ‘contraction mapping theorem, 98 convergence, 9 ceabic convergence, 10) definition, 98 ror bound, 99 higher-order, 195 linear convergence, 101 order of convergence, [01 ‘quadratic convergence, 101 Floating-point numbers, 34 accuracy, 38 binary, 34 chopping, 39 decimal, 34 exponent, 34 hexadecimal, 36 IEEE double precision, 35 IEEE single precision, 35 IEEE standard, 35 machine epsilon, 38 mantissa, 34 normalized, 36 representation, 34 rounding, 39 scientific notation, 34 sien, 34 significant, 34 ‘unnormalized, 36 Forward difference, 137 Forward substitution, 284 Frechet derivative, 357 Fundamental theorem of ealelus, 139. 509 Gauss-Seidel method, 305 Gaussian elimination, 264 ‘LU factorization, 283 back substiucon, 265, Cholesky factorization, 293 ‘compact variants, 285 Dooltle’s method, 286 elimination, 265, error analysis, 294 error estimation, 24 forvard substitution, 284 Gepivor, 279 18,291 aipliers, 268 operations count, 276, patil pivoting, 277 pivoting, 277 residual correction method, 296 ‘ridiag, 289 tridiagonal systems, 287 Gaussian numerical integration, 219 change of integration interval, 224 defiton, 223, error bound, 225 Legendre polynomial, 223 odes, 23 periodic integrand, 225, ‘weight function, 226 weight function I, 227 ‘weight fonction J. 237 weighted, 226 weights, 223 Geometric series, 15 Global eran 415 Golden mean, 97 Heat equation, 452 ; backward Euler metod, 475 Catk-Nichoson scheme, 479 error bound, #70, 475 explicit fl discretization, 4658 explicit metho, 469 forward Euler formu, 468 Heat, 470 ‘plc ful iscrezation 473 implicit metho, 475 inal boundary ‘valve problem, 466 semiseretizaton 467 stability condition, 469 truncation eae, 468 unconditional stability, 475 Heimiolz equation, 463 Hean’s method, 402 Hilbert matrix, 300 Hyperbolic equation, Se also Wave equations 452 IBEE standard, 35 TEE double precision, 25, IEEE single pression, 35 UUkconditioned, 374 Implicit method, «75 Infinite series, 15 Inner product, 68 Integral equation, 318 Integral Mean Value Theorem, 495 Intermediate Valve Theorem, 491 Interpolation, 118 fncerp, 13! error, polynomial, 138 Hermite 134 higher depres, 123 Lagrange's formula, 121, 123 linea, 119 Newton's formula, 129 piecewise linear, 147 Piecewise quadratic, 48 squadati, 120 spline interpolation, 149 Interval arithmetic, 57 Inverse, 255, Jacobi method, 304 Laplce equation, 452 Least squares approximation formule, 184 weighted, 185 Leastsquared data fiting leassquares approximation, 32) Least squares approximation, 178 root mean-square-eroe, 78 Least-squares dta iting, 219 ilkconitoning, 325 linear polyno 322 poly, 329 polynomial fi, 224 root-mear-squar-errr, 321 Legendre polynomials, 182 definition, 187 least-squares data fiting, 326 onthogonaliy, 182 tiple recursion relation, 182 eros, 183 Linear systems, 244 cond, 300 condition amber, 299 homogeneous, 238 ileoritoned. 299 matsivector notation, 245 tonhomogeneous, 258 solvability theory. 258 sparse, 289 stability, 297 welleontitiond, 209 Linear systems diret methods Gaussian elimination, 254 Linear systems, diret methods, 243 LU faxorizaton, 283 (Ghoesky Factorization, 293 compact schemes, 285 Doolite’s method, 286 n,2) . residual corteton method, 296 tridiag, 289 tridiagonal systems, 287 Linear systems, tration, 303 convergence, 107 izgonlly domizat, 208ee eo Qoor ae mm Index Gauss Seidel method, 305 general sche, 306 63,313 “aco, 4/2 ‘eabi metho, 304 resid earrecion method, 296,310 Local error, 415. lop(x),2 “Lota Voter predtorgrey model 435 Machine epsiion, 38 Mathematical Software, 517 Mathematical software cominecil, $12 interactive numerical computation, 515 literature, 516 public domain, $12 symbolic computation, 515 Matlab, $17 Mate, 245, 248 LU factoriaation, 283 arithmetic operations, 249 CCholesy factorization, 293 column, 248 eterminant, 255 diagonal, 260 diagonally dominan, 308 elementary, 264 Hien, 300 idemity, 254 inverse, 255, 273 linear function, 252 lower triangular, 260 negative, 290 nonsingular, 255, norms, 298 ‘orthogonal, 338 rectangular, 248, rotation, 350, row, 248 singular, 255, spare, 248 symmetic, 260 transpose, 249 tridiagonal, 287, upper triangular, 260, zero, 250 Matrix arithmetic, 249 aditon, 250 algebra ules, 256 clementary row operations, 253 inversion, 255 rmtpicaton, 257 negative, 290, scalar maliplication, 249 sobration, 250 Mean Vale Theorem, 494 Imueprl, 495 Mulistep methods, 423 AB, 20 absolute stabi, 430 ‘Adams-Bashforth (AB) methods, 423 ‘Adaras-Mouiton (AM) methods, 423 asymptotic errr formula, 427 label eror, 427 | tes, 425 odet13, 420 odetis, 430 Richardson's extrapolation, 427 trapeation ear, 426 vs Runge-Kutta methods, 425 Neat-minimax approximation, !71 chebyshev.interp,!75 construction, 173 even functions, 176 od functions, 176 Newcon divide diferece, Se lso Divided ferences 125 Newon's method, 79 convergence, 84 cero estimation, 86 error formula, 8 iteration formula, 60 nevton, 86 sonlnea systems, 352 polynomial defition, 89 square 100, 88 Nonlinear systems, 352 convergence theory, 360 rnnifed Newton's method, 361 Newion's method, 352,357 newton.sys, 362 Norms, 298 ‘Numerical differentiation, 252 ‘backward difference formu, 234 cenealdifference formula, 235 -ertorin function values, 238 forward difference formula, 234 interpolation. 234 method of undetermined coefficients, 236 Richardson's extrapolation, 247 second derivative, 237 Namerial Integration, 189 Numerical integration asymptotic err estimate, 206, asymptotic eror formula, 208 automatic, 217 corrected Simpson's rule, 209 corrected trapencidal rue, 207 degree of precision, 202 ertor, 204,208,216 Gaussian, 219 midpoint rule, 202 node points 93 periodic Function, 20) Richardson’ error estimate, 2/0 Richardson's extraolation formula, 210 ‘Simpson's rule, 198 trapezoidal, 19 trapezoidal ru, 193 Operations count, 276 Ocdinay differential equations, 367 closed-form solutions 368 ‘rection field, 376 ‘existence and unique, 372 first-order, 367 general solution, 368 higher-order, 367,434 illconditioned probier, 374 inital value probe, 370 linear, 369 method of ineratng factors 369 particular solutions, 368 separable, 378 stability, 373 systems, 432 theory, 468 ‘worpoint boundary value problems, “2 ‘wel-conditoned problem, 376 Ordinary sifferential equation, numerical solution absolute stability, 395,396, backward Euler method, 394 convergence order, 388 Euler's method, 379 explicit methods, 397 generalized midpoint method, 406 eun's method, 402 implicit method, 297 midpoint method, 406 model equation, 394 rmittep methods, 423 region of abso silty, 395 Runge-Kutta methods, 417 single-step methods, 423 stability, 394 if equations, 405 Tayor method, 409 trapezoidal method 400 ‘Ordinary differential equations, systems, 42 asymptotic error formula, 437 ror bound, 437 ‘ror estimation, 437 Euler's method, 437 eulersys, 48 firstorder, 433 higher-order equations 434 Richardson's extrapolation 437 Ronge-Kutta method, 440 Parabolic equations, See also Heat equation 452 Partial differential equations, 451 boundary vale problem, 453 elliptic equations, 52 heat equation, 452560 Index Paria ferential equations (Cont) Hyperbole equations, 52 inital boundary valve problem, 453 inital vee problem, 453 Laplace equation, 452 parabolic equations, 452 Poisson equation, 452 wave equation, 452 Pendulum equation, 435 Poisson equation, 452 asymptotic error expansion, 460 ‘convergence order. 456 Dirichlet boundary condition, 453 five-point scheme, 455 Neumann boundary condition, 453 Poisson, 456 Poissont, 454 Richardson extrapolation, 460 Robin boundary condition, 453 truncation ero, 455 Polynomial polynomial evaluation, 23 ‘Taylor polynomial, 1 Polynomial evluation, 23 Horner's method, 25 synthetic division, 25° Polynomial interpolation, Se also Interpolation 118 Polynomials chebyshev, 165 Legendre, 181 monic, 169 Power method, 340 Aitken's error estimation formula, 345 ‘Aitken’ extrapolation formula, 345 convergence, 342 power eethod, 346 Power series, 7 Predictor formula, 398 Propogated err, 57 addition, 39 division, 2 function evaluation, 60, rmuliplication, 38 subtraction, $9 (Quadrature, See also Numerical integration, 189 Region of absolute sabilty, 395 Residual eoretion method, 310 Richardson's ror estimate, 391 Richardson's extapolation formula, 391 Rootinging, 77 bisection method, 72 cubic convergencs, 107 Fixed point iteration, 97 aero, 95 itbehavior, 109 ilbconditioned, 112 linea convergence, 107 rile root, 109 Newon’s method, 7 ‘Newion-Rapison method, 80 conde of convergence, 107 quatratic convergence, 10! Secant method, 90 simple rt, 109 stability, 12 Rounding, 39 eran, 39 Runge-Kutta methods, 47 asympiatic error formula, 414 convergence order, 4/4 ‘eror control, 415 ‘eror estimate, 415 or prediction, 415 Feiberg methods, 416 fourthonder methods, 414 e23, 420 046, 8 Richardson's extrapolation, 4/5 second-order methods, 417 6. multistep method, 425 Scentfc notation, 34 Secant method, 90 ‘convergence, 92 ero estimate, 93 «eror formula, 92 iteration formula, 91 secant, 93 vs Newton's method, 94 Significant digits, 45 Simpson's rue, 198 asymptotic error formula, 208, corrected Simpson's rule, 209 ro, 208 simpson, 199 Sint 1,25 Software packages, $1! commercial, 5/2 interactive numerical compstation, sis literature, $16 public domsin, $12, + symbolic computation, $15 Spline, 149 Spine functions 147 spline, [54 ‘bic spline, 152 natural cubic spline, 149 ot interpolation, 153 terpolation Spline construction, 149 error bound, 153 Stability, 292, 373, absolute stabiy, 395 region of absolute stability, 395 Systems nonlinear, See also Nonlinear systems, 352 ‘Systems of linear equations, See also Linear systems, 244 “Tylor polynomial ‘common Functions, 13 eon J) remainder, 17 Sint x,25 Taylor polynomial of degree, 5 “Taylor polynomials, 1 Taylor series expansion, 16 ‘Trapezoidal method, 400 absolut stability, 40! second-order accuracy, 40! trapezoidal, 402 ‘Trapezoid rule, 193 asymptotic errr estimate, 206, corected trapezoidal rl, 207 error, 204 trapezoidal, 194 ‘Tridiagonal systems, 287, “Teuncaton error, 386, ‘Two point boundary value problem, 442 ‘Two-point boundary value problems asympttic error expansion, 444,447 boundary conditions, 42 error bound, 444 existence and uniqueness, 42 finite difference method, 442 rode points, 442 cDEBYP, 445 Richardson's extrapolation, #44 Vestor, 245, 248 column, 248 orm, 298 ro, 248 ‘Wave equation, 452 D'Alember' formula, 488 ‘error bound, 484 explicit central difference equation, 482 homogeneous, 487 intial boundary value problem, 481 stability condition, 483, truncation eo, 482 Wave, #84 ‘Weight function, 185, 226