MIT OpenCourseWarehttp://ocw.mit.edu 18.034 Honors Differential Equations Spring 2009 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms. � � � � � � � � � � � � � LECTURE 20. TRANSFORM AND DIFFERENTIAL EQUATIONSProperties of Laplace transform. We derive several important properties of Laplace transforms. Theorem 20.1. Let L[f(t)](s)= F (s). For f ∈ E, (i) (s-shift) L[e−ctf(t)] = F (s + c). (ii) (t-shift) L[f(t − c)] = e−scF (s) if c � 0 and f(t)=0 for t< 0. (iii) (s-derivative) L[tf(t)] = −F �(s). (iv) (t-derivative) L[f�(t)] = sF (s) − f(0), if f is continuous. �� ��� (v) (scaling) L[f(ct)] = 1 c F sc , F (sc)= 1 c L f ct if c> 0. Proof. 1. It follows from ∞ e−st e−ctf(t) dt = ∞ e−(s+c)tf(t) dt. 00 2. Let u = t − c. Then, ∞ e−stf(t − c) dt = ∞∞ e−sc e−s(u+c)f(u) du = e−suf(u) du. 0 −c 0 The limits (−c, ∞) can be changed to (0, ∞) by hypothesis. 3. It follows from d ∞ e−stf(t) dt = ∞ ∂ � e−stf(t) � dt = ∞ −te−stf(t) dt. ds 00 ∂s 0 4. It is proved in Lecture 19. 5. By change of variables. � Exercise. Show that 1. L f(t)= ∞ F (s) ds if f(t)/t ∈ E. s�� t � 2. L t f(t) dt = F (s) if f ∈ E. 0 s Example 20.2. Compute the Laplace transform of tet . 1SOLUTION. Applying the s-derivative property to L[et]= s − 1 gives 1 � 1 L[tet]= − s − 1 =(s − 1)2 . n!Exercise. Show that L[tneat]= (s − a)n+1 , n =0, 1, 2,... and a ∈ R. Example 20.3. Compute the Laplace transform of e3t sin t. 1SOLUTION. Applying the s-shift property to L[sin t]= s2 +1 gives 1 L[e 3t sin t]= (s − 3)2 +1. 1 � � � � � � � � s + cbExercise. Show that L[e−ct cos bt]= (s + c)2 + b2 and L[e−ct sin bt]= (s + c)2 + b2 . 4sExample 20.4. What function has Laplace transform ?(s2 + 4)2 SOLUTION. Observe that 2 � 2 = −4s and L[sin2t]= . s2 +4 (s2 + 4)2 s2 +4Hence, by the s-derivative property, 4s L−1 (s2 + 4)2 = t sin2t. 1Example 20.5. What function has Laplace transform ? s2 +4s +9SOLUTION. It involves completing the square. We write 11 = . s2 +4s +9 (s + 2)2 +5Since L � sin √5t � = √5 , by the s-shift property s2 +5L−1 s2 + 41 s +9 = √15 e 2t sin √5t. Exercise. Show that L−1 s2 +4ss +9 = √15 e 2t cos √5t + √25 e 2t sin √5t. Generalized solutions. We study (20.1) P (D)y := y(n) + a1y(n−1) ++ any = f(t),··· where aj are constants and f ∈ E. That is, f is allowed to have discontinuities. a very effective method of dealing with such problems. First, we must come to grips with the following. Theorem 20.6. Let n � 1 and t ∈ I be an open interval. If f has a simple discontinuity at some point in I, then (20.1) has no classical solution on I. At a simple discontinuity the left limit and the right limit exist but they do not agree. A classical solution on I is a function y = φ(t) which satisfies the differential equation at every point of I. The condition n � 1 ensures that (20.1) actually involves differentiation. The proof uses a theorem of Darboux and it is omitted here. Here we extend the notion of “solution” to allow discontinuous inputs and we develop the theory of the Laplace transform within the context of the extension. Definition 20.7. A function y = φ(t) is called a generalized solution of (20.1) on the interval I if (i) φ,φ�,...,φ(n−1) are continuous on I, and (ii) P (D)φ(t)= f(t) on I wherever f is continuous. Lecture 20 2 18.034 Spring 2009 � The condition (i) means all bad behaviors of f are absolved by φ(n). At discontinuities of f, the equation may fail. y(n) doesn’t need to exist. Exercise. Establish the existence and uniqueness theorem of the initial value problem of (20.1) and y(0) = y0,y�(0) = y1,...,y(n−1)(0) = yn−1, where f ∈ E in the class of generalized solutions. Theorem 20.8. If y is a generalized solution of (20.1) on t ∈ [0, ∞) and f ∈ E, then y,y�,...,y(n) ∈ E. Sketch of proof. First, show that the class � E �is closed under addition, multiplication, and integratiion That is, if f,g ∈ E, then f + g,fg, f, g ∈ E. Next, if P (D)= D − a, then the solution of P (D)= f is given by t y(t)= e at e−asf(s) ds + ce at . 0 Therefore, if f ∈ E then y ∈ E. To show that the derivative belongs to E, we write the equation as y� = f − p0y. Since f,y ∈ E, so is y�. Finally, we use induction on n. � The transformed equation. Let us consider (20.2) P (D)y := y(n) + a1y(n−1) ++ any = f ∈ E, ··· where aj are constants. The characteristic polynomial of P (D) is P (s) := s(n) + a1s(n−1) ++ an.··· When f has a discontinuity, then the word “solution” will be used to mean “generalized solution”. Using the formula nL[f(n)(t)](s)= s L[f(t)](s) − s n−1f(0) −···− f(n−1)(0) recursively, the Laplace transform of (20.2) gives, s nLy − (s n−1 y0 ++ yn−1)+ a1s n−1Ly − a1(s n−2 y0 ++ yn−2)+ + anLy = Lf, ··· ··· ··· which can be written as P (s)Y (s)= F (s)+ P0(s), where Y (s)= Ly, F (s)= Lf, and P0 is a polynomial of degree � n − 1 whose coefficients depend on the initial conditions. A formula of P0 is readily obtained by (19.4). Therefore, we arrive that F (s) P0(s)Y (s)= + . P (s) P (s) Then, by finding the inverse transform of Y (s) we obtain the solution of (20.2). Suppose that the inhomogeneous term f(t) is a finite sum of functions of the form tm e at(A cos bt + B sin bt), where m � 0 and a,b,A,B are constants. Inputs of this type are encountered frequently and in diverse contexts. The Laplace transform of f(t) is a sum of rational functions, that is, quotients of two polynomials. For example, L[tmeat]= m!/(s − a)m+1. The discussion above then tells us that the Laplace transform of the output y(t) is again a sum of rational functions. When F (s) is a rational function, the basic method of recovering f(t), where Lf(t)= F (s), is by expanding F (s) into partial fractions. Lecture 20 3 18.034 Spring 2009 Example 20.9. Solve the initial value problem y�� − 2y� +2y =2e t ,y(0) = 0,y�(0) = 1 By means of the Laplace transform. SOLUTION. Let Y (s)= Ly and F (s)= Lf. Taking the transform, we obtain 1 s +1 s 2Y (s) − 1 − 2sY (s)+2Y (s)=2 s − 1,Y (s)= (s − 1)(s2 − 2s + 2). We expand the rational function in partial fractions and then complete the square to obtain s +1 =2+ −2s +3 (s − 1)(s2 − 2s + 2) s − 1 s2 − 2s +2 2 s − 11 = s − 1 − 2(s − 1)2 +1 + (s − 1)2 +1 By finding the inverse transform of each term, y(t)=2e t − 2e t cos t + e t sin t. Exercise. (The initial and the final value theorems) 1. If f ∈ E, show that lim F (s)=0. 2. If f� ∈ E and f is continuous, show that lim sF (s)= f(0). s→∞ s→∞3. If f ∈ E and lim f(t)= k, show that lim sF (s)= k. t→∞ s→0+ Lecture 20 4 18.034 Spring 2009