Applied Mathematics Comprehensive Exam August 2011 Instructions: Answer 3 of the problems from Part A, and answer 3 of the problems from Part B. Indicate clearly which questions you wish to be graded. Part A A.1 (a) Determine conditions on f, α and β for which there are solutions of u00 = f (x), 0<x<π u0 (0) = α, u0 (π) = β (b) Find the best possible (least squares) solution of u00 = sin2 (x), u0 (0) = 1, u0 (π) = 1 A.2 Let A ∈ lRN ×N be a Symmetric Positive Definite matrix. Recall the quadratic form q(x) = hAx, xi. Show that the level surface q(x) = 1 is an ellipsoid. A.3 Use the Green’s function to solve u00 = f (x), 0 < x < 1 Z 1 u(x)dx = 4. u(0) = 3, 0 A.4 Find solvability conditions for u00 + u = f (x), 0 < x < 2π u(0) − u(2π) = α, u0 (0) − u0 (2π) = −β A.5 (a) Find the resolvent (or pseudo-resolvent) kernel for the problem Z 1 u(x) = f (x) + λ u(y)dy 0 and for any choice of λ. (b) Solve the integral equation for f (x) = x. 1 Part B B.1 Define the bounded linear operator T on L2 (0, 1) by Z 1 e−|x−y| u(y) dy T u(x) = 0 (a) Show that if T u = v then v(x) satisfies v 00 − v = −2u, 0 < x < 1 and v(0) − v 0 (0) = v(1) + v 0 (1) = 0 (b) Show that if λ is a nonzero eigenvalue of T , with eigenfunction u(x) then 2 u00 + ( − 1)u = 0, λ 0 < x < 1 and u(0) − u0 (0) = u(1) + u0 (1) = 0 (c) Show that the eigenvalues of T are real and lie in the interval (0, 2). B.2 Let p(x) > 0 be smooth and define the Sturm-Liouville operator L with domain D(L) as follows: du d (p(x) ), Lu ≡ dx dx 2 2 D(L) ≡ {u ∈ L (0, π): Lu ∈ L (0, π), u(0) = u(π) = 0} (a) Assume that L has a complete orthonormal set of eigenfunctions φn (x), n = 1, 2, . . . with associated eigenvalues λn , for n = 1, 2, . . .. Show that the Green’s function g(x, y) solving Lg = δ(x − y) has the representation g(x, y) = ∞ X φn (x)φn (y) λn n=1 (b) For the case p(x) = 1, compute g(x, y) and then use the result in part a) to find a formula for the following sum: ∞ X sin nx sin ny n2 n=1 B.3 Find the modified Green’s function for u00 = f (x), u(0) = u(1), u0 (0) = u0 (1) 0 < x < 1, B.4 Let f (x) be a continuously differentiable function except at a discrete set of points x1 , x2 . . . , xn , where f has finite jump discontinuities ∆f1 , ∆f2 , . . . , ∆fn . Show that the derivative of f (x) in the sense of distributions is given by n df X + f = ∆fj δ(x − xj ), dx j=1 0 where df /dx is the classical derivative of f wherever it exists. 2 B.5 Use Neumann iterates to solve the following integral equation using u0 (x) = 1. Z x (y − x)u(y)dy u(x) = 1 + 0 B.6 (a) Consider the variational problem of minimizing J(y1 (t), y2 (t)), where Z t1 f (s, y1 (s), y2 (s), ẏ1 (s), ẏ2 (s)) ds J(y1 , y2 ) = t0 subject to y1 (t0 ) = y1,0 y1 (t1 ) = y1,1 y2 (t0 ) = y2,0 y2 (t1 ) = y2,1 Assume that f (s, y1 , y2 , v1 , v2 ) is a C 2 function. Derive the set of Euler-Lagrange equations for this problem. (b) Find extremals of the functional π/2 Z [y 0 ]2 + [z 0 ]2 + 2yz ds J(y, z) = 0 subject to y(0) = 0, y(π/2) = 1, z(0) = 0, z(π/2) = 1. 3