PHZ 3113 Fall 2010 Homework #9, Due Monday, December 6 1. Let the differential operator d d (1 − x2 ) dx dx act on functions defined on the interval (−1, 1). D= (a) Suppose DPm (x) = −m(m + 1) Pm (x) and DPn (x) = −n(n + 1) Pn (x) Show that for n 6= m Pm (x) and Pn (x) are orthogonal on (−1, 1). Take the differential equation DPm = −m(m + 1)Pm , multiply by Pn , and integrate from −1 to 1: Z 1 Z 1 d dPm 2 (1 − x ) Pm Pn dx. Pn dx = −m(m + 1) dx −1 dx −1 Integrate the left-hand side by parts, casting the derivative onto Pn , and obtain Z 1 Z 1 2 dPm dPn (1 − x ) Pm Pn dx dx = m(m + 1) dx dx −1 −1 (because of the factor 1−x2 the surface term vanishes). Then, do the same thing, exchanging m and n, Z 1 Z 1 dPn dPm 2 dx = n(n + 1) (1 − x ) Pn Pm dx, dx dx −1 −1 and subtract: [m(m + 1) − n(n + 1)] Z 1 −1 Pm Pn dx = 0. If m 6= n, the factor in brackets is nonzero, and so Z 1 −1 Pm Pn dx = 0. (b) Verify that P0 = 1 and P1 = x are solutions to DPm = −m(m + 1) Pm . What are the corresponding values of m? d d 2 DP0 = (1 − x ) (1) = 0 = −0(0 + 1)P0 . dx dx i d d h 2 d 2 (1 − x ) DP1 = (x) = (1 − x )(1) = −2x = −1(1 + 1)P1 . dx dx dx So, P0 and P1 are solutions to DPm = −m(m + 1)Pm , with m = 0 and m = 1, respectively. (c) Let P2 = c0 + c1 x + c2 x2 be a polynomial of degree 2. Choose c0 and c1 so that P2 is orthogonal to P0 and P1 . Choose c2 so that P2 (1) = 1. Write your resulting P2 (x). Is your P2 a solution of DP2 = −m(m + 1) P2 for some m? If so, what is the your value for m? The inner products are Z 1 Z 1 −1 P0 P2 dx = −1 Z 1 −1 (c0 + c1 x + c2 x2 ) dx = 2c0 + Z 1 P1 P2 dx = −1 x(c0 + c1 x + c2 x2 ) dx = 2 c . 3 2 2 c . 3 1 For P2 to be orthogonal to P0 and P1 , both of these must vanish: c1 = 0, c0 = − 13 c2 , leaving P2 = c2 (− 13 + x2 ). Evaluated at x = 1, P2 = 23 c2 = 1 gives c2 = 32 . Thus, P2 (x) = 3 2 1 x − . 2 2 Acting with D gives i d 3 2 1 d d h 2 DP2 = (1 − x ) x − = (1 − x2 )(3x) = 3 − 9x2 = −2(2 + 1)P2 . dx dx 2 2 dx So, m = 2 (perhaps you anticipated this). This construction will produce any Legendre polynomial Pm you need, but it is not the most efficient way to accomplish this. 2. Let f (x) = sin x 0 < x < π , 0 π < x < 2π . (a) Compute the Fourier series coefficients an , bn . For n 6= 1, the cosine coefficients are 2/π Z π 1 1 + cos nπ − 2 n −1 an = = sin x cos nx dx = − π 0 π(n2 − 1) 0 (n even) (n odd) (this also holds for a0 ), and the sine coefficients are Z 1 π sin nπ bn = sin x sin nx dx = 2 = 0. π 0 n −1 For n = 1, Thus, Z 1 π a1 = sin x cos x dx = 0, π 0 Z 1 π 2 1 b1 = sin x dx = . π 0 2 ∞ 1 1 2 X cos 2kx f (x) = + sin x − . π 2 π (2k)2 − 1 k=1 ∞ X (−1)k+1 (b) Compute the value of the sum . (2k)2 − 1 k=1 [Hint: what is f ( π2 )?] As it says, evaluate f at π2 : ∞ π 1 1 2 X cos kπ f( ) = 1 = + − . 2 π 2 π (2k)2 − 1 k=1 So, ∞ X π (−1)k+1 1 1 π 1 = 1− − = − . 2 2 2 π 4 2 (2k) − 1 k=1 (c) Compute the value of the sum ∞ X k=1 1 [(2k)2 − 1]2 . The square of coefficients appears in Parseval’s theorem, Z 2π 0 and so f 2 (x) = Z π 0 sin2 x dx = ∞ X π π = a20 + π (a2n + b2n ), 2 2 n=1 ∞ 1 2 2 1 2 1 4 X 1 + + 2 = , 2 π 2 2 π [(2k)2 − 1]2 k=1 ∞ X k=1 2 π2 1 π2 1 1 = − − − . = 4 2 4 π2 16 2 [(2k)2 − 1]2 1 3. Let f (x) = cos(k0 x) −a < x < a, 0 otherwise. with k0 a = π2 . (a) Compute the Fourier transform f̃ (k). What happens at k → k0 ? Z a Z a 1 ik0 x f̃(k) = (e cos k0 x e dx = + e−ik0 x ) e−ikx dx −a −a 2 " #a i(k −k)x −i(k +k)x 0 0 1 e e sin[(k0 − k)a] sin[(k0 + k)a] = + + = 2 i(k0 − k) −i(k0 + k) k0 − k k0 + k −ikx −a = sin( π2 − ka) sin( π2 + ka) 2k cos ka + = 02 . k0 − k k0 + k k0 − k 2 Although the denominator vanishes at k = k0 , the numerator does too, and the function goes as 2k0 cos ka π π ≈ − (k − k0 ) + · · · . 2k0 4k02 k02 − k 2 (b) Compute the integral (there is an easy way and an easier way to do this) Z ∞ cos y dy . π 2 2 −∞ [( 2 ) − y ] The “easy” way is as a contour integral, with cos y = Re[eiy ] and with two poles (on the axis) at y = ± π2 ; but the “easier” way is to evaluate Fourier transform back to f (x) at x = 0: Z ∞ Z d(ka) 2k0 a cos ka dk 2k0 cos ka ikx f (0) = e = = 1. 2 2 2π (k0 a)2 − (ka)2 x=0 −∞ 2π k0 − k Now, let ka = y, with k0 a = π2 , and Z 1= So the value of the desired integral is 2. dy π cos y . 2π ( π2 )2 − y 2 (c) Compute the integral Z ∞ cos2 y dy . π 2 22 −∞ [( 2 ) − y ] This time, use Parseval’s theorem for the Fourier integrals, Z a Z 2 cos2 k0 x dx = a, |f (x)| dx = −a Z 2k0 a cos ka 2 πa ∞ cos2 y dy d(ka) = . 2π 2 −∞ [( π2 )2 − y 2 ]2 (k0 a)2 − (ka)2 By theorem these are the same, and so Z dk 2k0 cos ka 2 =a 2π k02 − k 2 Z Z ∞ cos2 y dy 2 = . π 2 2 2 π −∞ [( 2 ) − y ]