MAT2700 - EXAM, FALL 2009 (DEC 16, 4 HOURS) - FASIT Problem 1. a. Solve the equation z 2 + iz + 2 = 0. Express its two solutions z1 , z2 in polar form: zj = rj eiθj with θj ∈ (−π, π], j = 1, 2. π π Answer. The two solutions are z1 = i = ei 2 , z2 = −2i = 2e−i 2 . b. Compute the cube root of −2i, i.e., (−2i)1/3 . π π 2 Answer. (−2i)1/3 = |(−2i)|1/3 ei(− 2 +2πk)/3 = 21/3 ei(− 6 + 3 πk) , k = 0, 1, 2. Hence 1/3 −i π 2 e 6 , k = 0, π (−2i)1/3 = 21/3 ei 2 , k = 1, 21/3 ei 7π6 , k = 2. c. Determine the Laurent series for the function 1 2 z + iz + 2 in the annulus 1 < |z| < 2. Answer. By a) and partial fractions, z2 1 i i 1 = = − . + iz + 2 (z + i)(z + 2i) 3(z + 2i) 3(z − i) First, for |z| < 2, i i 1 1 1 1 = , = z 3(z + 2i) 3 2i 1 − (− 2i ) 61−c c=− z iz = . 2i 2 Since |c| = | z2 | < 1, we can apply the results on geometric series: j ∞ X i 1 i = zj . 3(z + 2i) 6 2 j=0 Next, for |z| > 1, i i 1 = , 3(z − i) 3z 1 − zi 1 2 MAT2700 - EXAM, FALL 2009 (DEC 16, 4 HOURS) - FASIT and as before, since |i/z| < 1, the results on geometric series apply: ∞ ∞ ∞ i X j −j X 1 j+1 −(j+1) X 1 k −k i = iz = i z = i z . 3(z − i) 3z j=0 3 3 j=0 k=1 Hence, the Laurent series is ∞ X1 1 = z 2 + iz + 2 6 j=0 j ∞ X i 1 j −j j iz . z − 2 3 j=1 Problem 2. Consider the contour integral Z 1 I := dz, 3 2 C2 z − z where C2 is the set {z ∈ C : |z| = 2} traversed in the counter-clockwise direction. a. Draw a figure of the contour C2 and signify the poles of the integrand. Compute the integral I without using the residue theorem (justify all steps R in your calculations). [Hint: As a part of your computations, explain why C2 1/(z − 1) dz = R 1/(z − 1) dz, with the circle |z − 1| = 1 being oriented in the counter|z−1|=1 clockwise direction.] 2 1 -2 1 -1 2 -1 -2 Figure 1. Contour and poles for Problem 2 a). Answer. By partial fractions, we can write the integrand as 1 1 1 1 1 = 2 =− 2 − + . 3 2 z −z z (z − 1) z z z−1 We then compute three contour integrals. First, Z 1 dz = 0, 2 C2 z since since 1/z 2 has an antiderivative (−1/z) in a domain containing C2 . MAT2700 - EXAM, FALL 2009 (DEC 16, 4 HOURS) - FASIT 3 Next, parameterizing C2 by z(t) = 2eit for t ∈ [0, 2π], we compute Z 2π Z 2π Z 1 0 1 i dt = 2πi. dz = z (t) dt = z(t) 0 0 C2 z Similarly, Z 1 dz = z−1 C2 Z |z−1|=1 1 dz = 2πi. z−1 where the first equality comes as a consequence of the Deformation Invariance Theorem. Summarizing, I = −0 − 2πi + 2πi = 0. b. For each R > 2, set Z I(R) := CR 1 dz, z3 − z2 where CR denotes the set {z ∈ C : |z| = R} traversed in the counter-clockwise direction. Explain why we have I = I(R), for any R > 2. Answer. The singularities of the integrand are inside C2 , and thus the integrand is analytic in a suitable domain D containing the contours C2 and CR . The contour C2 can be continuously deformed in D into CR . The result then follows from the Deformation Invariance Theorem. c. Prove that |I(R)| ≤ 2π , −R R2 for R > 2. Answer. For z ∈ CR , by triangle inequality |z 3 − z 2 | ≥ |z|3 − |z|2 = R3 − R2 , and hence |z 3 The result now follows: Z Z 1 ≤ dz z3 − z2 CR CR |z 3 1 1 ≤ 3 . 2 −z | R − R2 1 length(CR ) 2πR 2π dz ≤ ≤ 3 = 2 2 3 2 2 −z | R −R R −R R −R d. Compute I by utilizing b) and c) [hint: let R → ∞]. 4 MAT2700 - EXAM, FALL 2009 (DEC 16, 4 HOURS) - FASIT Answer. According to c), |I(R)| ≤ 2π → 0 as R → ∞, R2 − R so, by b), I = lim I(R) = 0. R→∞ e. Use the residue theorem to compute I. 1 1 Answer. The integrand f (z) = z3 −z 2 = z 2 (z−1) has singularities at z = 0 (pole of order 2) and z = 1 (simple pole), elsewhere it is analytic. We compute d 2 d 1 1 Res(0) = lim (z f (z) = lim = − lim = −1 z→0 dz z→0 dz z→0 (z − 1)2 z−1 and 1 Res(1) = lim(z − 1)f (z) = lim 2 = 1. z→1 z→1 z Consequently, I = 2πi (Res(0) + Res(1)) = 2πi(−1 + 1) = 0. Problem 3. The Laplace equation is a second-order partial differential equation of the form ∂ 2φ ∂ 2φ + = 0. (1) ∂x2 ∂y 2 A real-valued function φ = φ(x, y) is said to be harmonic in a domain D ⊂ R2 if φ 2 2 2 2 is twice continuously differentiable (i.e., ∂φ , ∂φ , ∂ φ , ∂ φ , ∂ φ , ∂ φ exist and are ∂x ∂y ∂x∂y ∂y∂x ∂x2 ∂y 2 continuous in D) and satisfies the Laplace equation (1) at each point (x, y) in D. a. Suppose f (z) = u(x, y) + i v(x, y) is analytic in a domain D. Show that u(x, y) and v(x, y) are harmonic in D. [Hint: Differentiate the Cauchy-Riemann equations.] Answer. Since f is analytic, all partial derivatives of u and v up to order 2 exist and are continuous in D, so it remains to show that the Laplace equation holds. Since f is analytic, the Cauchy-Riemann equations demand that ∂v ∂u ∂v ∂u = , =− . ∂x ∂y ∂y ∂x Taking partial derivatives with respect to x yields ∂ 2u ∂ 2v ∂ 2u ∂ 2v = , = − . ∂x2 ∂y∂x ∂y∂x ∂x2 Similarly, taking partial derivatives with respect to y yields ∂ 2u ∂ 2v ∂ 2u ∂ 2v = 2, = − . ∂x∂y ∂y ∂y 2 ∂x∂y MAT2700 - EXAM, FALL 2009 (DEC 16, 4 HOURS) - FASIT 5 Since all the partial derivatives are continuous, it is a calculus fact that ∂ 2v ∂ 2v = , ∂y∂x ∂x∂y ∂ 2u ∂ 2u = . ∂x∂y ∂y∂x Hence, it follows that ∂ 2v ∂ 2u ∂ 2v ∂ 2u ∂ 2u ∂ 2u = = − = =⇒ + =0 ∂x2 ∂y∂x ∂x∂y ∂y 2 ∂x2 ∂y 2 and − ∂ 2u ∂ 2v ∂ 2v ∂ 2u ∂ 2v ∂ 2v = = = =⇒ + = 0; ∂x2 ∂y∂x ∂x∂y ∂y 2 ∂x2 ∂y 2 Therefore, u and v are both harmonic functions in D. b. Suppose φ(x, y) is a harmonic function in a simply connected domain D. Set g(z) := ∂φ ∂φ −i . ∂x ∂y Show that g is analytic. Answer. Let us show that the Cauchy-Riemann equations are satisfied for g. Indeed, since φ is harmonic, ∂ ∂φ ∂ ∂φ = − . ∂x ∂x ∂y ∂y Moreover, since ∂2φ ∂y∂x = ∂2φ , ∂x∂y ∂ ∂y ∂φ ∂x ∂ =− ∂x ∂φ − ∂y . (all the displayed second order partial derivatives exist and are continuous since φ is harmonic). c. Let g(z) be the function defined in b). Since g is analytic in the simply connected domain D we know that there exists an antiderivative G(z) = u(x, y) + i v(x, y) for some functions u and v, G0 = g in D. Define the function f (z) := G(z) + c, where c is a constant. Show that c can be chosen so that the real part of f is φ: Re(f ) = φ. 6 MAT2700 - EXAM, FALL 2009 (DEC 16, 4 HOURS) - FASIT Answer. We have ∂v ∂u ∂u ∂u +i = −i , ∂x ∂x ∂x ∂y where the last equality comes from the Cauchy-Riemann equations. At the same time we also have ∂φ ∂φ G0 (z) = g(z) = −i , ∂x ∂y so necessarily ∂u ∂φ ∂u ∂φ = , = ∂x ∂x ∂y ∂y Hence, u = φ + constant, and choosing c = −constant it follows that the real part of f 0 (z) = G0 (z) = f = G + c = u + i v + c = φ + constant + i v + c = φ + i v is indeed φ. d. The Maximum Modulus Principle (MMP) says that a complex-valued function f (z) that is analytic in a bounded domain D and continuous up to and including its boundary attains its maximum modulus |f | on the boundary. Use the MMP to show the maximum principle for solutions of the Laplace equation, namely that a function φ(x, y) that is harmonic in a bounded simply connected domain D and continuous up to and including the boundary attains its maximum on the boundary [Hint: Apply MMP to the function |ef | with f from c).] Answer. With f = φ + i v, it follows that f e = eφ ei v = eφ . According to the MMP, ef attains its maximum at the boundary and consequently the same must be true of eφ , and thus also of φ (since the real-valued exponential function is monotone).