18.085 Computational Science and Engineering Answers to Problem Set 10 April 26, 2010 TAs: Dorian Croitoru, Nan Li 4.2.1.a) We have 2 ˆ π ˆ π 1 F (x, y)e−imx e−iny dxdy 2π −π −π ˆ πˆ π ˆ π ˆ π 1 1 −imx −iny −imx e e dxdy = e dx e−iny dy. 4π 2 0 0 4π 2 0 0 cmn = = ´π −imπ otherwise, we c00 = e−imx dxequals π if m = 0, and 1−eim get −1 1/(2πim) if m odd if m, n odd 1/4, cm0 = c0m = for m 6= 0, and cmn = π2 mn 0 otherwise 0 otherwise for m, n 6= 0. ´ ´ ´0 ´0 π π b) cmn = 4π1 2 0 0 e−imx e−iny dxdy + −π −π e−imx e−iny dxdy . Simi- Since 0 larly to a), we obtain c00 = 1/2, cm0 = c0m = 0 for m 6= 0 and cmn = −2 π 2 mn if m, n odd for m, n 6= 0. otherwise 4.2.2. Since sin mx andPsin Pny are odd in x and y respectively, S(x, y) will have a double sine series bmn sin mx sin ny if S(x, y) is odd in x and y : −S(x, y) = S(−x, y) = S(x, −y). The double sine functions are orthogonal: ˆ π ˆ π 1 1 if k = m and l = n (sin kx sin ly)(sin mx sin ny)dxdy = . 0 otherwise π 2 −π −π 0 4.2.3. Using the orthogonality from the previous problem, bmn 1 = 2 π ˆ π ˆ π S(x, y) sin mx sin nydxdy. −π −π 4.2.7. FromPthe (1-dimensional) Fourier series of the delta function we obtain P 1 1 −im(x+y) −imx −imy δ(x + y) = 2π e = e . m∈Z m∈Z e 2π 1 1 4 4 2 2 4.2.8. x = 8 (8x − 8x + 1) + 2 (2x − 1) + 38 = 18 T4 + 21 T2 + 38 T0 . x −1 ... −1 2x −1 4.2.10. Let Bn (x) = (n by n matrix). We compute . . . −1 −1 −1 2x its determinant by expanding along the last row: det Bn = 2x det Bn−1 − (−1) det Bn−2 0 ... − 1 0 −1 = 2x det Bn−1 − det Bn−2 . Thus det Bn is a polynomial in x satisfying the same recurrence as Tn . Since det B1 = x = T1 , det B2 = 2x2 − 1 = T2 , it follows that det Bn = Tn for all n. 1 4.2.20. We have (pu0m )0 + (q + λm w)um = 0 (pu0n )0 + (q + λn w)un = 0 Multiplying the rst equation by un ,the second by um and then subtracting, ´b we obtain un (pu0m )0 − um (pu0n )0 + (λm − λn )wum un = 0. Integrating, a (λm − ´b b λn )wum un dx = a un (pu0m )0 − um (pu0n )0 dx = [un (pu0m ) − um (pu0n )]a = 0, by ´b the boundary conditions. Since λm 6= λn , we get a wum un dx = 0. 4.2.21. Bessel's equation: r2 B 00 + rB 0 + λr2 B = n2 B ⇒ rB” + B 0 + (λr − 2 n /r)B = 0 ⇔ (rB 0 )0 + (λr − n2 /r)B = 0 which is a Sturm-Liouville equation with u = B, p = r, q = −n2 /r, w = r. Alternatively, one can try u = B(r/2) and p = r2 . Legendre's equation: (1−x2 )P ”−2xP 0 +λP = 0 ⇔ ((1−x2 )P 0 )0 +λP = 0 is a Sturm-Liouville equation with u = P, p = 1 − x2 , w = 1, q = 0. 2