Math 6350 Homework #6 Solutions 1. Suppose the branch of the logarithm is chosen so that 0 < arg z < 2π. Under what circumstances is it true that log z1 z2 = log z1 + log z2 ? Under what circumstances is it true that log z p = p log z for a real number p? Solution: The real parts certainly work out fine since they use only the properties of the real logarithm: we have ln |z1 z2 | = ln |z1 ||z2 | = ln |z1 | + ln |z2 | and ln |z p | = ln |z|p = p ln |z|. Thus we only need to worry about the imaginary part. First we ask when arg z1 z2 = arg z1 + arg z2 . The argument is assumed to be in (0, 2π), which means the left side is in (0, 2π) while the right side can be anything between 0 and 4π. We will therefore have ( arg z1 + arg z2 arg z1 + arg z2 < 2π arg z1 z2 = arg z1 + arg z2 − 2π 2π < arg z1 + arg z2 < 4π. The condition is therefore that arg z1 + arg z2 < 2π. I’m not sure of a simpler way to express this. As for arg z p = p arg z, suppose z = reiθ for θ ∈ (0, 2π). Then obviously arg z = θ so that p arg z = pθ. On the other hand z p = rp eipθ and arg z p = pθ + 2nπ for some integer n which makes pθ + 2nπ ∈ (0, 2π). So as long as 0 < p arg z < 2π we will have arg z p = p arg z. 2. Use the method from class to evaluate the integral Z 2π dt . 5 − 3 sin t 0 Solution: We express this as the integral around the closed curve γ(t) = eiπt , using the fact that on this curve sin t = z − 1/z eit − e−it = . 2i 2i We have dz = ieit dt = iz dt so that dt = dz/(iz). Therefore we can turn the integral into Z 2π Z Z dt dz 2i dz = − 2 = . 3 5 − 3 sin t 3iz + 10z − 3i 0 γ iz 5 − 2i (z − 1/z) γ Now factor this quadratic to get z= −10 ± √ 100 − 36 −10 ± 8 = 6i 6i 1 which yields z = 3i and z = i/3. Only i/3 is inside the unit disc, so we may express this integral as Z Z 2π f (z) dt = 5 − 3 sin t γ z − i/3 0 where f (z) = − 2i . 3i(z − 3i) The integral is therefore equal to Z 2π dt 4π 4π π = 2πif (i/3) = = = . 5 − 3 sin t 3i(i/3 − 3i) 8 2 0 3. Ahlfors pg. 120 #2: Compute Z |z|=2 dz +1 z2 by decomposing it into partial fractions. Solution: Assuming a decomposition of the form z2 1 A B = + , +1 z−i z+i we must have A(z − i) + B(z + i) = 1 for all z, and thus B = −A and −2iA = 1 so that A = i/2. We obtain i 1 i 1 1 = − . z2 + 1 2 z−i 2 z+i Hence the given integral is Z γ 4. Compute R |z|=1 Z Z dz i dz i dz = − 2 z +1 2 γ z−i 2 γ z+i i i = (2πi) − (2πi) = 0. 2 2 cos z/z n dz where n is any integer. Solution: Using the general formula Z f (z) 2πif (n−1) (0) = n (n − 1)! γ z if n ≥ 1, we obtain Z |z|=1 0 n even and positive, 2πi cos z dz = 1 n = 4k + 1 for k ≥ 0, zn (n − 1)! −1 n = 4k + 3 for k ≥ 0. On the other hand if n ≤ 0 then z −n cos z is analytic at 0 and hence everywhere, so that the integral is zero. 2 5. Ahlfors pg. 123 #2: Prove that a function f which is analytic on the whole plane and satisfies an inequality |f (z)| < |z|n for some n and all sufficiently large z reduces to a polynomial. Solution: We will prove that f (n+1) (z) = 0 for all z ∈ C, which implies that f is a polynomial of degree n. Using formula (24) we have Z f (ζ) dζ (n + 1)! (n+1) f (z) = n+2 2πi C (ζ − z) whenever f is analytic inside C. Choose C to be a circle of large radius R; then we have Z (n + 1)! 2π Rn+1 dt 1 (n + 1)! (n+1) |f (z)| ≤ = . n+2 2π (|R| − |z|) R (1 − |z/R|)n+2 0 For fixed z, let R go to infinity, and we get |f (n+1) (z)| ≤ 0, which implies f (n+1) (z) = 0 as desired. 6. From “Counterexamples in Analysis” by Gelbaum and Olmsted. Define ( 2 x3 /y 2 e−x /y y > 0 f (x, y) = 0 y ≤ 0. (a) Show that f is smooth (infinitely differentiable) in x for fixed y. Solution: If y > 0 then obviously the function is smooth since it’s just a rescaling 2 of x3 e−x which is smooth. If y ≤ 0 then it’s the zero function which is smooth in x. (b) Show that f is continuous in y for any fixed x. Solution: It’s clearly continuous at points y > 0 and at points y < 0, so we just have to check that it’s continuous at y = 0. We have lim+ f (x, y) = lim+ y→0 y→0 x3 z 2 1 u2 x3 = lim = lim =0 x u→+∞ eu y 2 ex2 /y z→+∞ ex2 z by L’Hopital’s rule. We also have limy→0− f (x, y) = f (x, 0) = 0, and since these all match we have continuity in y. (c) Show that Z 1 Z 1 d ∂ f (x, y) dy 6= f (x, y) dy. dx x=0 0 0 ∂x x=0 Solution: First evaluate the left side. We have Z 1 Z 1 2 f (x, y) dy = x3 /y 2 e−x /y dy 0 Z0 1 ∂ 2 = xe−x /y dy 0 ∂y y=1 2 = xe−x /y y=0 2 = xe−x − lim+ xe−x y→0 = xe 3 −x2 2 /y whenever x 6= 0. Of course if x = 0 then the integral is zero, so this formula always works. Differentiating we get Z 1 d 2 2 f (x, y) dy = (e−x − 2x2 e−x )x=0 = 1. dx x=0 0 Now we work on the right side. We compute (if y > 0) that ∂ ∂ 3 2 −x2 /y 2 2 f (x, y) = x /y e = 3x2 /y 2 e−x /y − 2x4 /y 3 e−x /y , ∂x ∂x ∂ and evaluating at x = 0 we obtain ∂x f (x, y) = 0. On the other hand if y ≤ 0 x=0 then the derivative is clearly zero for any x (and thus for x = 0). Thus the integral of the zero function is zero, and the right side is always zero. (d) Show that f is not continuous at (0, 0). Solution: We have f (0, 0) = 0 and we want to find a sequence (xn , yn ) such that (xn , yn ) → 0 and f (xn , yn ) does not converge to zero. Suppose yn = x2n /k for some k; then we have k 2 e−k . f (xn , yn ) = xn Since (xn , yn ) → (0, 0), we must have xn → 0, so f (xn , yn ) → ∞. 4
© Copyright 2024