Published on: **Mar 3, 2016**

Published in:
Technology Education

- 1. Contents CHAPTER 1: Introduction 1 EXERCISES 1.1: Background, page 5 . . . . . . . . . . . . . . . . . . . . . . . . . . 1 EXERCISES 1.2: Solutions and Initial Value Problems, page 14 . . . . . . . . . . . 3 EXERCISES 1.3: Direction Fields, page 22 . . . . . . . . . . . . . . . . . . . . . . . 10 EXERCISES 1.4: The Approximation Method of Euler, page 28 . . . . . . . . . . . 17 CHAPTER 2: First Order Diﬀerential Equations 27 EXERCISES 2.2: Separable Equations, page 46 . . . . . . . . . . . . . . . . . . . . 27 EXERCISES 2.3: Linear Equations, page 54 . . . . . . . . . . . . . . . . . . . . . . 41 EXERCISES 2.4: Exact Equations, page 65 . . . . . . . . . . . . . . . . . . . . . . 59 EXERCISES 2.5: Special Integrating Factors, page 71 . . . . . . . . . . . . . . . . . 72 EXERCISES 2.6: Substitutions and Transformations, page 78 . . . . . . . . . . . . 79 REVIEW PROBLEMS: page 81 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 CHAPTER 3: Mathematical Models and Numerical Methods Involving First Order Equations 103 EXERCISES 3.2: Compartmental Analysis, page 98 . . . . . . . . . . . . . . . . . . 103 EXERCISES 3.3: Heating and Cooling of Buildings, page 107 . . . . . . . . . . . . 116 EXERCISES 3.4: Newtonian Mechanics, page 115 . . . . . . . . . . . . . . . . . . . 123 EXERCISES 3.5: Electrical Circuits, page 122 . . . . . . . . . . . . . . . . . . . . . 137 EXERCISES 3.6: Improved Euler’s Method, page 132 . . . . . . . . . . . . . . . . . 139 EXERCISES 3.7: Higher Order Numerical Methods: Taylor and Runge-Kutta, page 142 153 CHAPTER 4: Linear Second Order Equations 167 EXERCISES 4.1: Introduction: The Mass-Spring Oscillator, page 159 . . . . . . . . 167 EXERCISES 4.2: Homogeneous Linear Equations; The General Solution, page 167 . 169 EXERCISES 4.3: Auxiliary Equations with Complex Roots, page 177 . . . . . . . . 177 iii
- 2. EXERCISES 4.4: Nonhomogeneous Equations: The Method of Undetermined Coeﬃcients, page 186 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 EXERCISES 4.5: The Superposition Principle and Undetermined Coeﬃcients Revisited, page 192 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 EXERCISES 4.6: Variation of Parameters, page 197 . . . . . . . . . . . . . . . . . . 211 EXERCISES 4.7: Qualitative Considerations for Variable-Coeﬃcient and Nonlinear Equations, page 208 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226 EXERCISES 4.8: A Closer Look at Free Mechanical Vibrations, page 219 . . . . . . 232 EXERCISES 4.9: A Closer Look at Forced Mechanical Vibrations, page 227 . . . . 241 REVIEW PROBLEMS: page 228 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246 CHAPTER 5: Introduction to Systems and Phase Plane Analysis 259 EXERCISES 5.2: Elimination Method for Systems, page 250 . . . . . . . . . . . . . 259 EXERCISES 5.3: Solving Systems and Higher–Order Equations Numerically, page 261 282 EXERCISES 5.4: Introduction to the Phase Plane, page 274 . . . . . . . . . . . . . 293 EXERCISES 5.5: Coupled Mass-Spring Systems, page 284 . . . . . . . . . . . . . . 307 EXERCISES 5.6: Electrical Circuits, page 291 . . . . . . . . . . . . . . . . . . . . . 317 EXERCISES 5.7: Dynamical Systems, Poincar`e Maps, and Chaos, page 301 . . . . . 325 REVIEW PROBLEMS: page 304 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331 CHAPTER 6: Theory of Higher Order Linear Diﬀerential Equations 341 EXERCISES 6.1: Basic Theory of Linear Diﬀerential Equations, page 324 . . . . . . 341 EXERCISES 6.2: Homogeneous Linear Equations with Constant Coeﬃcients, page 331 351 EXERCISES 6.3: Undetermined Coeﬃcients and the Annihilator Method, page 337 361 EXERCISES 6.4: Method of Variation of Parameters, page 341 . . . . . . . . . . . . 375 REVIEW PROBLEMS: page 344 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384 CHAPTER 7: Laplace Transforms 389 EXERCISES 7.2: Deﬁnition of the Laplace Transform, page 359 . . . . . . . . . . . 389 EXERCISES 7.3: Properties of the Laplace Transform, page 365 . . . . . . . . . . . 396 EXERCISES 7.4: Inverse Laplace Transform, page 374 . . . . . . . . . . . . . . . . 402 EXERCISES 7.5: Solving Initial Value Problems, page 383 . . . . . . . . . . . . . . 413 EXERCISES 7.6: Transforms of Discontinuous and Periodic Functions, page 395 . . 428 EXERCISES 7.7: Convolution, page 405 . . . . . . . . . . . . . . . . . . . . . . . . 450 EXERCISES 7.8: Impulses and the Dirac Delta Function, page 412 . . . . . . . . . 459 EXERCISES 7.9: Solving Linear Systems with Laplace Transforms, page 416 . . . . 466 REVIEW PROBLEMS: page 418 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481 iv
- 3. CHAPTER 8: Series Solutions of Diﬀerential Equations 491 EXERCISES 8.1: Introduction: The Taylor Polynomial Approximation, page 430 . . 491 EXERCISES 8.2: Power Series and Analytic Functions, page 438 . . . . . . . . . . . 496 EXERCISES 8.3: Power Series Solutions to Linear Diﬀerential Equations, page 449 505 EXERCISES 8.4: Equations with Analytic Coeﬃcients, page 456 . . . . . . . . . . . 520 EXERCISES 8.5: Cauchy-Euler (Equidimensional) Equations Revisited, page 460 . 529 EXERCISES 8.6: Method of Frobenius, page 472 . . . . . . . . . . . . . . . . . . . 534 EXERCISES 8.7: Finding a Second Linearly Independent Solution, page 482 . . . . 547 EXERCISES 8.8: Special Functions, page 493 . . . . . . . . . . . . . . . . . . . . . 559 REVIEW PROBLEMS: page 497 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563 CHAPTER 9: Matrix Methods for Linear Systems 569 EXERCISES 9.1: Introduction, page 507 . . . . . . . . . . . . . . . . . . . . . . . . 569 EXERCISES 9.2: Review 1: Linear Algebraic Equations, page 512 . . . . . . . . . . 570 EXERCISES 9.3: Review 2: Matrices and Vectors, page 521 . . . . . . . . . . . . . 573 EXERCISES 9.4: Linear Systems in Normal Form, page 530 . . . . . . . . . . . . . 577 EXERCISES 9.5: Homogeneous Linear Systems with Constant Coeﬃcients, page 541 584 EXERCISES 9.6: Complex Eigenvalues, page 549 . . . . . . . . . . . . . . . . . . . 596 EXERCISES 9.7: Nonhomogeneous Linear Systems, page 555 . . . . . . . . . . . . 602 EXERCISES 9.8: The Matrix Exponential Function, page 566 . . . . . . . . . . . . 617 CHAPTER 10: Partial Diﬀerential Equations 629 EXERCISES 10.2: Method of Separation of Variables, page 587 . . . . . . . . . . . 629 EXERCISES 10.3: Fourier Series, page 603 . . . . . . . . . . . . . . . . . . . . . . . 635 EXERCISES 10.4: Fourier Cosine and Sine Series, page 611 . . . . . . . . . . . . . 639 EXERCISES 10.5: The Heat Equation, page 624 . . . . . . . . . . . . . . . . . . . . 644 EXERCISES 10.6: The Wave Equation, page 636 . . . . . . . . . . . . . . . . . . . 653 EXERCISES 10.7: Laplace’s Equation, page 649 . . . . . . . . . . . . . . . . . . . . 660 CHAPTER 11: Eigenvalue Problems and Sturm-Liouville Equations 675 EXERCISES 11.2: Eigenvalues and Eigenfunctions, page 671 . . . . . . . . . . . . . 675 EXERCISES 11.3: Regular Sturm-Liouville Boundary Value Problems, page 682 . . 683 EXERCISES 11.4: Nonhomogeneous Boundary Value Problems and the Fredholm Al- ternative, page 692 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687 EXERCISES 11.5: Solution by Eigenfunction Expansion, page 698 . . . . . . . . . . 690 EXERCISES 11.6: Green’s Functions, page 706 . . . . . . . . . . . . . . . . . . . . 694 EXERCISES 11.7: Singular Sturm-Liouville Boundary Value Problems, page 715 . . 701 v
- 4. EXERCISES 11.8: Oscillation and Comparison Theory, page 725 . . . . . . . . . . . 705 CHAPTER 12: Stability of Autonomous Systems 707 EXERCISES 12.2: Linear Systems in the Plane, page 753 . . . . . . . . . . . . . . . 707 EXERCISES 12.3: Almost Linear Systems, page 764 . . . . . . . . . . . . . . . . . 709 EXERCISES 12.4: Energy Methods, page 774 . . . . . . . . . . . . . . . . . . . . . 716 EXERCISES 12.5: Lyapunov’s Direct Method, page 782 . . . . . . . . . . . . . . . 718 EXERCISES 12.6: Limit Cycles and Periodic Solutions, page 791 . . . . . . . . . . 719 EXERCISES 12.7: Stability of Higher-Dimensional Systems, page 798 . . . . . . . . 722 CHAPTER 13: Existence and Uniqueness Theory 725 EXERCISES 13.1: Introduction: Successive Approximations, page 812 . . . . . . . . 725 EXERCISES 13.2: Picard’s Existence and Uniqueness Theorem, page 820 . . . . . . 733 EXERCISES 13.3: Existence of Solutions of Linear Equations, page 826 . . . . . . 741 EXERCISES 13.4: Continuous Dependence of Solutions, page 832 . . . . . . . . . . 743 vi
- 5. CHAPTER 1: Introduction EXERCISES 1.1: Background, page 5 1. This equation involves only ordinary derivatives of x with respect to t, and the highest deriva- tive has the second order. Thus it is an ordinary diﬀerential equation of the second order with independent variable t and dependent variable x. It is linear because x, dx/dt, and d2 x/dt2 appear in additive combination (even with constant coeﬃcients) of their ﬁrst powers. 3. This equation is an ODE because it contains no partial derivatives. Since the highest order derivative is dy/dx, the equation is a ﬁrst order equation. This same term also shows us that the independent variable is x and the dependent variable is y. This equation is nonlinear because of the y in the denominator of the term [y(2 − 3x)]/[x(1 − 3y)] . 5. This equation is an ODE because it contains only ordinary derivatives. The term dp/dt is the highest order derivative and thus shows us that this is a ﬁrst order equation. This term also shows us that the independent variable is t and the dependent variable is p. This equation is nonlinear since in the term kp(P − p) = kPp − kp2 the dependent variable p is squared (compare with equation (7) on page 5 of the text). 7. This equation is an ordinary ﬁrst order diﬀerential equation with independent variable x and dependent variable y. It is nonlinear because it contains the square of dy/dx. 9. This equation contains only ordinary derivative of y with respect to x. Hence, it is an ordi- nary diﬀerential equation of the second order (the highest order derivative is d2 y/dx2 ) with independent variable x and dependent variable y. This equation is of the form (7) on page 5 of the text and, therefore, is linear. 1
- 6. Chapter 1 11. This equation contains partial derivatives, thus it is a PDE. Because the highest order deriva- tive is a second order partial derivative, the equation is a second order equation. The terms ∂N/∂t and ∂N/∂r show that the independent variables are t and r and the dependent variable is N. 13. Since the rate of change of a quantity means its derivative, denoting the coeﬃcient propor- tionality between dp/dt and p(t) by k (k > 0), we get dp dt = kp. 15. In this problem, T ≥ M (coﬀee is hotter than the air), and T is a decreasing function of t, that is dT/dt ≤ 0. Thus dT dt = k(M − T), where k > 0 is the proportionality constant. 17. In classical physics, the instantaneous acceleration, a, of an object moving in a straight line is given by the second derivative of distance, x, with respect to time, t; that is d2 x dt2 = a. Integrating both sides with respect to t and using the given fact that a is constant we obtain dx dt = at + C. (1.1) The instantaneous velocity, v, of an object is given by the ﬁrst derivative of distance, x, with respect to time, t. At the beginning of the race, t = 0, both racers have zero velocity. Therefore we have C = 0. Integrating equation (1.1) with respect to t we obtain x = 1 2 at2 + C1 . For this problem we will use the starting position for both competitors to be x = 0 at t = 0. Therefore, we have C1 = 0. This gives us a general equation used for both racers as x = 1 2 at2 or t = 2x a , 2
- 7. Exercises 1.2 where the acceleration constant a has diﬀerent values for Kevin and for Alison. Kevin covers the last 1 4 of the full distance, L, in 3 seconds. This means Kevin’s acceleration, aK, is determined by: tK − t3/4 = 3 = 2L aK − 2(3L/4) aK , where tK is the time it takes for Kevin to ﬁnish the race. Solving this equation for aK gives, aK = √ 2 − 3/2 2 9 L. Therefore the time required for Kevin to ﬁnish the race is given by: tK = 2L √ 2 − 3/2 2 L/9 = 3 √ 2 − 3/2 √ 2 = 12 + 6 √ 3 ≈ 22.39 sec. Alison covers the last 1/3 of the distance, L, in 4 seconds. This means Alison’s acceleration, aA, is found by: tA − t2/3 = 4 = 2L aA − 2(2L/3) aA , where tA is the time required for Alison to ﬁnish the race. Solving this equation for aA gives aA = √ 2 − 4/3 2 16 L. Therefore the time required for Alison to ﬁnish the race is given by: tA = 2L √ 2 − 4/3 2 (L/16) = 4 √ 2 − 4/3 √ 2 = 12 + 4 √ 6 ≈ 21.80 sec. The time required for Alison to ﬁnish the race is less than Kevin; therefore Alison wins the race by 6 √ 3 − 4 √ 6 ≈ 0.594 seconds. EXERCISES 1.2: Solutions and Initial Value Problems, page 14 1. (a) Diﬀerentiating φ(x) yields φ (x) = 6x2 . Substitution φ and φ for y and y into the given equation, xy = 3y, gives x 6x2 = 3 2x3 , 3
- 8. Chapter 1 which is an identity on (−∞, ∞). Thus φ(x) is an explicit solution on (−∞, ∞). (b) We compute dφ dx = d dx (ex − x) = ex − 1. Functions φ(x) and φ (x) are deﬁned for all real numbers and dφ dx +φ(x)2 = (ex − 1)+(ex − x)2 = (ex − 1)+ e2x − 2xex + x2 = e2x +(1−2x)ex +x2 −1, which is identically equal to the right-hand side of the given equation. Thus φ(x) is an explicit solution on (−∞, ∞). (c) Note that the function φ(x) = x2 − x−1 is not deﬁned at x = 0. Diﬀerentiating φ(x) twice yields dφ dx = d dx x2 − x−1 = 2x − (−1)x−2 = 2x + x−2 ; d2 φ dx2 = d dx dφ dx = d dx 2x + x−2 = 2 + (−2)x−3 = 2 1 − x−3 . Therefore x2 d2 φ dx2 = x2 · 2 1 − x−3 = 2 x2 − x−1 = 2φ(x), and φ(x) is an explicit solution to the diﬀerential equation x2 y = 2y on any interval not containing the point x = 0, in particular, on (0, ∞). 3. Since y = sin x+x2 , we have y = cos x+2x and y = − sin x+2. These functions are deﬁned on (−∞, ∞). Substituting these expressions into the diﬀerential equation y + y = x2 + 2 gives y + y = − sin x + 2 + sin x + x2 = 2 + x2 = x2 + 2 for all x in (−∞, ∞). Therefore, y = sin x + x2 is a solution to the diﬀerential equation on the interval (−∞, ∞). 5. Diﬀerentiating x(t) = cos 2t, we get dx dt = d dt (cos 2t) = (− sin 2t)(2) = −2 sin 2t. 4
- 9. Exercises 1.2 So, dx dt + tx = −2 sin 2t + t cos 2t ≡ sin 2t on any interval. Therefore, x(t) is not a solution to the given diﬀerential equation. 7. We diﬀerentiate y = e2x − 3e−x twice: dy dx = d dx e2x − 3e−x = e2x (2) − 3e−x (−1) = 2e2x + 3e−x ; d2 y dx2 = d dx dy dx = d dx 2e2x + 3e−x = 2e2x (2) + 3e−x (−1) = 4e2x − 3e−x . Substituting y, y , and y into the diﬀerential equation and collecting similar terms, we get d2 y dx2 − dy dx − 2y = 4e2x − 3e−x − 2e2x + 3e−x − 2 e2x − 3e−x = (4 − 2 − 2)e2x + (−3 − 3 − 2(−3))e−x = 0. Hence y = e2x − 3e−x is an explicit solution to the given diﬀerential equation. 9. Diﬀerentiating the equation x2 + y2 = 6 implicitly, we obtain 2x + 2yy = 0 ⇒ y = − x y . Since there can be no function y = f(x) that satisﬁes the diﬀerential equation y = x/y and the diﬀerential equation y = −x/y on the same interval, we see that x2 + y2 = 6 does not deﬁne an implicit solution to the diﬀerential equation. 11. Diﬀerentiating the equation exy + y = x − 1 implicitly with respect to x yields d dx (exy + y) = d dx (x − 1) ⇒ exy d dx (xy) + dy dx = 1 ⇒ exy y + x dy dx + dy dx = 1 ⇒ yexy + dy dx (xexy + 1) = 1 ⇒ dy dx = 1 − yexy 1 + xexy = exy (e−xy − y) exy (e−xy + x) = e−xy − y e−xy + x . 5
- 10. Chapter 1 Therefore, the function y(x) deﬁned by exy + y = x − 1 is an implicit solution to the given diﬀerential equation. 13. Diﬀerentiating the equation sin y + xy − x3 = 2 implicitly with respect to x, we obtain y cos y + xy + y − 3x2 = 0 ⇒ (cos y + x)y = 3x2 − y ⇒ y = 3x2 − y cos y + x . Diﬀerentiating the second equation above again, we obtain (−y sin y + 1)y + (cos y + x)y = 6x − y ⇒ (cos y + x)y = 6x − y + (y )2 sin y − y = 6x − 2y + (y )2 sin y ⇒ y = 6x − 2y + (y )2 sin y cos y + x . Multiplying the right-hand side of this last equation by y /y = 1 and using the fact that y = 3x2 − y cos y + x , we get y = 6x − 2y + (y )2 sin y cos y + x · y (3x2 − y)/(cos y + x) = 6xy − 2(y )2 + (y )3 sin y 3x2 − y . Thus y is an implicit solution to the diﬀerential equation. 15. We diﬀerentiate φ(x) and substitute φ and φ into the diﬀerential equation for y and y . This yields φ(x) = Ce3x + 1 ⇒ dφ(x) dx = Ce3x + 1 = 3Ce3x ; dφ dx − 3φ = 3Ce3x − 3 Ce3x + 1 = (3C − 3C)e3x − 3 = −3, which holds for any constant C and any x on (−∞, ∞). Therefore, φ(x) = Ce3x + 1 is a one-parameter family of solutions to y −3y = −3 on (−∞, ∞). Graphs of these functions for C = 0, ±0.5, ±1, and ±2 are sketched in Figure 1-A. 6
- 11. Exercises 1.2 –10 10 –0.5 0.5 C=2 C=1 C=0.5 C=0 C=−0.5 C=−1 C=−2 Figure 1–A: Graphs of the functions y = Ce3x + 1 for C = 0, ±0.5, ±1, and ±2. 17. Diﬀerentiating φ(x), we ﬁnd that φ (x) = 2 1 − cex = 2 (1 − cex )−1 = 2(−1) (1 − cex )−2 (1 − cex ) = 2cex (1 − cex )−2 . (1.2) On the other hand, substitution of φ(x) for y into the right-hand side of the given equation yields φ(x)(φ(x) − 2) 2 = 1 2 2 1 − cex 2 1 − cex − 2 = 2 1 − cex 1 1 − cex − 1 = 2 1 − cex 1 − (1 − cex ) 1 − cex = 2cex (1 − cex)2 , which is identical to φ (x) found in (1.2). 19. Squaring and adding the terms dy/dx and y in the equation (dy/dx)2 + y2 + 3 = 0 gives a nonnegative number. Therefore when these two terms are added to 3, the left-hand side will always be greater than or equal to three and hence can never equal the right-hand side which is zero. 7
- 12. Chapter 1 21. For φ(x) = xm , we have φ (x) = mxm−1 and φ (x) = m(m − 1)xm−2 . (a) Substituting these expressions into the diﬀerential equation, 3x2 y + 11xy − 3y = 0, gives 3x2 m(m − 1)xm−2 + 11x mxm−1 − 3xm = 0 ⇒ 3m(m − 1)xm + 11mxm − 3xm = 0 ⇒ [3m(m − 1) + 11m − 3] xm = 0 ⇒ 3m2 + 8m − 3 xm = 0. For the last equation to hold on an interval for x, we must have 3m2 + 8m − 3 = (3m − 1)(m + 3) = 0. Thus either (3m − 1) = 0 or (m + 3) = 0, which gives m = 1 3 , −3. (b) Substituting the above expressions for φ(x), φ (x), and φ (x) into the diﬀerential equa- tion, x2 y − xy − 5y = 0, gives x2 m(m − 1)xm−2 − x mxm−1 − 5xm = 0 ⇒ m2 − 2m − 5 xm = 0. For the last equation to hold on an interval for x, we must have m2 − 2m − 5 = 0. To solve for m we use the quadratic formula: m = 2 ± √ 4 + 20 2 = 1 ± √ 6 . 23. In this problem, f(x, y) = x3 − y3 and so ∂f ∂y = ∂ (x3 − y3 ) ∂y = −3y2 . Clearly, f and ∂f/∂y (being polynomials) are continuous on the whole xy-plane. Thus the hypotheses of Theorem 1 are satisﬁed, and the initial value problem has a unique solution for any initial data, in particular, for y(0) = 6. 8
- 13. Exercises 1.2 25. Writing dx dt = − 4t x = −4tx−1 , we see that f(t, x) = −4tx−1 and ∂f(t, x)/∂x = ∂(−4tx−1 )/∂x = 4tx−2 . The functions f(t, x) and ∂f(t, x)/∂x are not continuous only when x = 0. Therefore, they are continuous in any rectangle R that contains the point (2, −π), but does not intersect the t-axis; for instance, R = {(t, x) : 1 < t < 3, −2π < x < 0}. Thus, Theorem 1 applies, and the given initial problem has a unique solution. 26. Here f(x, y) = 3x − 3 √ y − 1 and ∂f(x, y)/∂y = −1 3 (y − 1)−2/3 . Unfortunately, ∂f/∂y is not continuous or deﬁned when y = 1. So there is no rectangle containing (2, 1) in which both f and ∂f/∂y are continuous. Therefore, we are not guaranteed a unique solution to this initial value problem. 27. Rewriting the diﬀerential equation in the form dy/dx = x/y, we conclude that f(x, y) = x/y. Since f is not continuous when y = 0, there is no rectangle containing the point (1, 0) in which f is continuous. Therefore, Theorem 1 cannot be applied. 29. (a) Clearly, both functions φ1(x) ≡ 0 and φ2(x) = (x − 2)3 satisfy the initial condition, y(2) = 0. Next, we check that they also satisfy the diﬀerential equation dy/dx = 3y2/3 . dφ1 dx = d dx (0) = 0 = 3φ1(x)2/3 ; dφ2 dx = d dx (x − 2)3 = 3(x − 2)2 = 3 (x − 2)3 2/3 = 3φ2(x)2/3 . Hence both functions, φ1(x) and φ2(x), are solutions to the initial value problem of Exapmle 9. (b) In this initial value problem, f(x, y) = 3y2/3 ⇒ ∂f(x, y) ∂y = 3 2 3 y2/3−1 = 2 y1/3 , x0 = 0 and y0 = 10−7 . The function f(x, y) is continuous everywhere; ∂f(x, y)/∂y is continuous in any region which does not intersect the x-axis (where y = 0). In particular, 9
- 14. Chapter 1 both functions, f(x, y) and ∂f(x, y)/∂y, are continuous in the rectangle R = (x, y) : −1 < x < 1, (1/2)10−7 < y < (2)10−7 containing the initial point (0, 10−7 ). Thus, it follows from Theorem 1 that the given initial value problem has a unique solution in an interval about x0. 31. (a) To try to apply Theorem 1 we must ﬁrst write the equation in the form y = f(x, y). Here f(x, y) = 4xy−1 and ∂f(x, y)/∂y = −4xy−2 . Neither f nor ∂f/∂y are continuous or deﬁned when y = 0. Therefore there is no rectangle containing (x0, 0) in which both f and ∂f/∂y are continuous, so Theorem 1 cannot be applied. (b) Suppose for the moment that there is such a solution y(x) with y(x0) = 0 and x0 = 0. Substituting into the diﬀerential equation we get y(x0)y (x0) − 4x0 = 0 (1.3) or 0 · y (x0) − 4x0 = 0 ⇒ 4x0 = 0. Thus x0 = 0, which is a contradiction. (c) Taking C = 0 in the implicit solution 4x2 − y2 = C given in Example 5 on page 9 gives 4x2 − y2 = 0 or y = ±2x. Both solutions y = 2x and y = −2x satisfy y(0) = 0. EXERCISES 1.3: Direction Fields, page 22 1. (a) For y = ±2x, dy dx = d dx (±2x) = ±2 and 4x y = 4x ±2x = ±2, x = 0. Thus y = 2x and y = −2x are solutions to the diﬀerential equation dy/dx = 4x/y on any interval not containing the point x = 0. (b) , (c) See Figures B.1 and B.2 in the answers of the text. 10
- 15. Exercises 1.3 (d) As x → ∞ or x → −∞, the solution in part (b) increases unboundedly and has the lines y = 2x and y = −2x, respectively, as slant asymptotes. The solution in part (c) also increases without bound as x → ∞ and approaches the line y = 2x, while it is not even deﬁned for x < 0. 3. From Figure B.3 in the answers section of the text, we conclude that, regardless of the initial velocity, v(0), the corresponding solution curve v = v(t) has the line v = 8 as a horizontal asymptote, that is, limt→∞ v(t) = 8. This explains the name “terminal velocity” for the value v = 8. 5. (a) The graph of the directional ﬁeld is shown in Figure B.4 in the answers section of the text. (b), (c) The direction ﬁeld indicates that all solution curves (other than p(t) ≡ 0) will approach the horizontal line (asymptote) p = 1.5 as t → +∞. Thus limt→+∞ p(t) = 1.5 . (d) No. The direction ﬁeld shows that populations greater than 1500 will steadily decrease, but can never reach 1500 or any smaller value, i.e., the solution curves cannot cross the line p = 1.5 . Indeed, the constant function p(t) ≡ 1.5 is a solution to the given logistic equation, and the uniqueness part of Theorem 1, page 12, prevents intersections of solution curves. 6. (a) The slope of a solution to the diﬀerential equation dy/dx = x + sin y is given by dy/dx . Therefore the slope at (1, π/2) is equal to dy dx = 1 + sin π 2 = 2. (b) The solution curve is increasing if the slope of the curve is greater than zero. From part (a) we know the slope to be x + sin y. The function sin y has values ranging from −1 to 1; therefore if x is greater than 1 then the slope will always have a value greater than zero. This tells us that the solution curve is increasing. (c) The second derivative of every solution can be determined by ﬁnding the derivative of 11
- 16. Chapter 1 the diﬀerential equation dy/dx = x + sin y. Thus d dx dy dx = d dx (x + sin y); ⇒ d2 y dx2 = 1 + (cos y) dy dx (chain rule) = 1 + (cos y)(x + sin y) = 1 + x cos y + sin y cos y; ⇒ d2 y dx2 = 1 + x cos y + 1 2 sin 2y. (d) Relative minima occur when the ﬁrst derivative, dy/dx, is equal to zero and the second derivative, d2 y/dx2 , is greater than zero. The value of the ﬁrst derivative at the point (0, 0) is given by dy dx = 0 + sin 0 = 0. This tells us that the solution has a critical point at the point (0, 0). Using the second derivative found in part (c) we have d2 y dx2 = 1 + 0 · cos 0 + 1 2 sin 0 = 1. This tells us the point (0, 0) is a relative minimum. 7. (a) The graph of the directional ﬁeld is shown in Figure B.5 in the answers section of the text. (b) The direction ﬁeld indicates that all solution curves with p(0) > 1 will approach the horizontal line (asymptote) p = 2 as t → +∞. Thus limt→+∞ p(t) = 2 when p(0) = 3. (c) The direction ﬁeld shows that a population between 1000 and 2000 (that is 1 < p(0) < 2) will approach the horizontal line p = 2 as t → +∞. (d) The direction ﬁeld shows that an initial population less than 1000 (that is 0 ≤ p(0) < 1) will approach zero as t → +∞. (e) As noted in part (d), the line p = 1 is an asymptote. The direction ﬁeld indicates that a population of 900 (p(0) = 0.9) steadily decreases with time and therefore cannot increase to 1100. 12
- 17. Exercises 1.3 9. (a) The function φ(x), being a solution to the given initial value problem, satisﬁes dφ dx = x − φ(x), φ(0) = 1. (1.4) Thus d2 φ dx2 = d dx dφ dx = d dx (x − φ(x)) = 1 − dφ dx = 1 − x + φ(x), where we have used (1.4) substituting (twice) x − φ(x) for dφ/dx. (b) First we note that any solution to the given diﬀerential equation on an interval I is continuously diferentiable on I. Indeed, if y(x) is a solution on I, then y (x) does exist on I, and so y(x) is continuous on I because it is diﬀerentiable. This immediately implies that y (x) is continuous as the diﬀerence of two continuous functions, x and y(x). From (1.4) we conclude that dφ dx x=0 = [x − φ(x)] x=0 = 0 − φ(0) = −1 < 0 and so the continuity of φ (x) implies that, for |x| small enough, φ (x) < 0. By the Monotonicity Test, negative derivative of a function results that the function itself is decreasing. When x increases from zero, as far as φ(x) > x, one has φ (x) < 0 and so φ(x) decreases. On the other hand, the function y = x increases unboundedly, as x → ∞. Thus, by intermediate value theorem, there is a point, say, x∗ > 0, where the curve y = φ(x) crosses the line y = x. At this point, φ(x∗ ) = x∗ and hence φ (x∗ ) = x∗ − φ(x∗ ) = 0. (c) From (b) we conclude that x∗ is a critical point for φ(x) (its derivative vanishes at this point). Also, from part (a), we see that φ (x∗ ) = 1 − φ (x∗ ) = 1 > 0. Hence, by Second Derivative Test, φ(x) has a relative minimum at x∗ . (d) Remark that the arguments, used in part (c), can be applied to any point x, where φ (x) = 0, to conclude that φ(x) has a relative minimum at x. Since a continuously 13
- 18. Chapter 1 diﬀerentiable function on an interval cannot have two relative minima on an interval without having a point of relative maximum, we conclude that x∗ is the only point where φ (x) = 0. Continuity of φ (x) implies that it has the same sign for all x > x∗ and, therefore, it is positive there since it is positive for x > x∗ and close to x∗ (φ (x∗ ) = 0 and φ (x∗ ) > 0). By Monotonicity Test, φ(x) increases for x > x∗ . (e) For y = x − 1, dy/dx = 1 and x − y = x − (x − 1) = 1. Thus the given diﬀerential equation is satisﬁed, and y = x − 1 is indeed a solution. To show that the curve y = φ(x) always stays above the line y = x − 1, we note that the initial value problem dy dx = x − y, y(x0) = y0 (1.5) has a unique solution for any x0 and y0. Indeed, functions f(x, y) = x−y and ∂f/∂y ≡ −1 are continuous on the whole xy-plane, and Theorem 1, Section 1.2, applies. This implies that the curve y = φ(x) always stays above the line y = x − 1: φ(0) = 1 > −1 = (x − 1) x=0 , and the existence of a point x with φ (x) ≤ (x − 1) would imply, by intermediate value theorem, the existence of a point x0, 0 < x0 ≤ x, satisfying y0 := φ(x0) = x0 − 1 and, therefore, there would be two solutions to the initial value problem (1.5). Since, from part (a), φ (x) = 1−φ (x) = 1−x+φ(x) = φ(x)−(x−1) > 0, we also conclude that φ (x) is an increasing function and φ (x) < 1. Thus there exists limx→∞ φ (x) ≤ 1. The strict inequality would imply that the values of the function y = φ(x), for x large enough, become smaller than those of y = x − 1. Therefore, lim x→∞ φ (x) = 1 ⇔ lim x→∞ [x − φ(x)] = 1, and so the line y = x − 1 is a slant asymptote for φ(x). (f), (g) The direction ﬁeld for given diﬀerential equation and the curve y = φ(x) are shown in Figure B.6 in the answers of the text. 14
- 19. Exercises 1.3 11. For this equation, the isoclines are given by 2x = c. These are vertical lines x = c/2. Each element of the direction ﬁeld associated with a point on x = c/2 has slope c. (See Figure B.7 in the answers of the text.) 13. For the equation ∂y/∂x = −x/y, the isoclines are the curves −x/y = c. These are lines that pass through the origin and have equations of the form y = mx, where m = −1/c , c = 0. If we let c = 0 in −x/y = c, we see that the y-axis (x = 0) is also an isocline. Each element of the direction ﬁeld associated with a point on an isocline has slope c and is, therefore, perpendicular to that isocline. Since circles have the property that at any point on the circle the tangent at that point is perpendicular to a line from that point to the center of the circle, we see that the solution curves will be circles with their centers at the origin. But since we cannot have y = 0 (since −x/y would then have a zero in the denominator) the solutions will not be deﬁned on the x-axis. (Note however that a related form of this diﬀerential equation is yy + x = 0. This equation has implicit solutions given by the equations y2 + x2 = C. These solutions will be circles.) The graph of φ(x), the solution to the equation satisfying the initial condition y(0) = 4, is the upper semicircle with center at the origin and passing through the point (0, 4) (see Figure B.8 in the answers of the text). 15. For the equation dy/dx = 2x2 −y, the isoclines are the curves 2x2 −y = c, or y = 2x2 −c. The curve y = 2x2 − c is a parabola which is open upward and has the vertex at (0, −c). Three of them, for c = −1, 0, and 2 (dotted curves), as well as the solution curve satisfying the initial condition y(0) = 0, are depicted in Figure B.9. 17. The isoclines for the equation dy dx = 3 − y + 1 x are given by 3 − y + 1 x = c ⇔ y = 1 x + 3 − c, which are hyperbolas having x = 0 as a vertical asymptote and y = 3 − c as a horizontal asymptote. Each element of the direction ﬁeld associated with a point on such a hyperbola has slope c. For x > 0 large enough: if an isocline is located above the line y = 3, then c ≤ 0, 15
- 20. Chapter 1 0 5 5 10 c=−5 c=−4 c=−3 c=−2 c=−1 c=1 c=2 c=3 c=4 3 Figure 1–B: Isoclines and the direction ﬁeld for Problem 17. and so the elements of the direction ﬁeld have negative or zero slope; if an isocline is located below the line y = 3, then c > 0, and so the elements of the direction ﬁeld have positive slope. In other words, for x > 0 large enough, at any point above the line y = 3 a solution curve decreases passing through this point, and any solution curve increases passing through a point below y = 3. The direction ﬁeld for this diﬀerential equation is depicted in Figure 1-B. From this picture we conclude that any solution to the diﬀerential equation dy/dx = 3 − y + 1/x has the line y = 3 as a horizontal asymptote. 19. Integrating both sides of the equation dy/y = −dx/x yields 1 y dy = − 1 x dx ⇒ ln |y| = − ln |x| + C1 ⇒ ln |y| = ln eC1 |x| ⇒ |y| = eC1 |x| ⇒ |y| = C2 |x| , where C1 is an arbitrary constant and so C2 := eC1 is an arbitrary positive constant. The last equality can be written as y = ± C2 x = C x , 16
- 21. Exercises 1.4 where C = ±C2 is any nonzero constant. The value C = 0 gives y ≡ 0 (for x = 0), which is, clearly, also a solution to the given equation. EXERCISES 1.4: The Approximation Method of Euler, page 28 1. In this initial value problem, f(x, y) = x/y, x0 = 0, and y0 = −1. Thus, with h = 0.1, the recursive formulas (2) and (3) on page 25 of the text become xn+1 = xn + h = xn + 0.1 , yn+1 = yn + hf(xn, yn) = yn + 0.1 · xn yn , n = 0, 1, . . .. We set n = 0 in these formulas and obtain x1 = x0 + 0.1 = 0 + 0.1 = 0.1 , y1 = y0 + 0.1 · x0 y0 = −1 + 0.1 · 0 −1 = −1. Putting n = 1 in the recursive formulas yields x2 = x1 + 0.1 = 0.1 + 0.1 = 0.2 , y2 = y1 + 0.1 · x1 y1 = −1 + 0.1 · 0.1 −1 = −1.01 . Continuing in the same manner, we ﬁnd for n = 2, 3, and 4: x3 = 0.2 + 0.1 = 0.3 , y3 = −1.01 + 0.1 · 0.2 −1.01 = −1.02980 ; x4 = 0.3 + 0.1 = 0.4 , y4 = −1.02980 + 0.1 · 0.3 −1.02980 = −1.05893 ; x5 = 0.4 + 0.1 = 0.5 , y5 = −1.05893 + 0.1 · 0.4 −1.05893 = −1.09671 , where we have rounded oﬀ all answers to ﬁve decimal places. 2. In this problem, x0 = 0, y0 = 4, h = 0.1, and f(x, y) = −x/y. Thus, the recursive formulas given in equations (2) and (3) on page 25 of the text become xn+1 = xn + h = xn + 0.1 , 17
- 22. Chapter 1 yn+1 = yn + hf(xn, yn) = yn + 0.1 · − xn yn , n = 0, 1, 2, . . . . To ﬁnd an approximation for the solution at the point x1 = x0 + 0.1 = 0.1, we let n = 0 in the last recursive formula to ﬁnd y1 = y0 + 0.1 · − x0 y0 = 4 + 0.1 · (0) = 4. To approximate the value of the solution at the point x2 = x1 + 0.1 = 0.2, we let n = 1 in the last recursive formula to obtain y2 = y1 + 0.1 · − x1 y1 = 4 + 0.1 · − 0.1 4 = 4 − 1 400 = 3.9975 ≈ 3.998 . Continuing in this way we ﬁnd x3 = x2 + 0.1 = 0.3 , y3 = y2 + 0.1 · − x2 y2 = 3.9975 + 0.1 · − 0.2 3.9975 ≈ 3.992 , x4 = 0.4 , y4 ≈ 3.985 , x5 = 0.5 , y5 ≈ 3.975 , where all of the answers have been rounded oﬀ to three decimal places. 3. Here f(x, y) = y(2 − y), x0 = 0, and y0 = 3. We again use recursive formulas from Euler’s method with h = 0.1. Setting n = 0, 1, 2, 3, and 4 and rounding oﬀ results to three decimal places, we get x1 = x0 + 0.1 = 0.1 , y1 = y0 + 0.1 · [y0(2 − y0)] = 3 + 0.1 · [3(2 − 3)] = 2.700; x2 = 0.1 + 0.1 = 0.2 , y2 = 2.700 + 0.1 · [2.700(2 − 2.700)] = 2.511; x3 = 0.2 + 0.1 = 0.3 , y3 = 2.511 + 0.1 · [2.511(2 − 2.511)] ≈ 2.383; x4 = 0.3 + 0.1 = 0.4 , y4 = 2.383 + 0.1 · [2.383(2 − 2.383)] ≈ 2.292; x5 = 0.4 + 0.1 = 0.5 , y5 = 2.292 + 0.1 · [2.292(2 − 2.292)] ≈ 2.225 . 5. In this problem, f(x, y) = (y2 + y)/x, x0 = y0 = 1, and h = 0.2. The recursive formulas (2) and (3) on page 25 of the text, applied succesively with n = 1, 2, 3, and 4, yield x1 = x0 + 0.2 = 1.2 , y1 = y0 + 0.2 y2 0 + y0 x0 = 1 + 0.2 12 + 1 1 = 1.400; 18
- 23. Exercises 1.4 x2 = 1.2 + 0.2 = 1.4 , y2 = 1.400 + 0.2 1.4002 + 1.400 1.2 ≈ 1.960; x3 = 1.4 + 0.2 = 1.6 , y3 = 1.960 + 0.2 1.9602 + 1.960 1.4 ≈ 2.789; x4 = 1.6 + 0.2 = 1.8 , y4 = 2.789 + 0.2 2.7892 + 2.789 1.6 ≈ 4.110 . 7. For this problem notice that the independent variable is t and the dependent variable is x. Hence, the recursive formulas given in equations (2) and (3) on page 25 of the text become tn+1 = tn + h and φ(tn+1) ≈ xn+1 = xn + hf(tn, xn), n = 0, 1, 2, . . . . For this problem, f(t, x) = 1+t sin(tx), t0 = 0, and x0 = 0. Thus the second recursive formula above becomes xn+1 = xn + h [1 + tn sin(tnxn)] , n = 0, 1, 2, . . . . For the case N = 1, we have h = (1 − 0)/1 = 1 which gives us t1 = 0 + 1 = 1 and φ(1) ≈ x1 = 0 + 1 · (1 + 0 · sin 0) = 1. For the case N = 2, we have h = 1/2 = 0.5 . Thus we have t1 = 0 + 0.5 = 0.5 , x1 = 0 + 0.5 · (1 + 0 · sin 0) = 0.5 , and t2 = 0.5 + 0.5 = 1, φ(1) ≈ x2 = 0.5 + 0.5 · [1 + 0.5 · sin(0.25)] ≈ 1.06185 . For the case N = 4, we have h = 1/4 = 0.25 , and so the recursive formulas become tn+1 = tn + 0.25 and xn+1 = xn + 0.25 · [1 + tn sin(tnxn)] . Therefore, we have t1 = 0 + 0.25 = 0.25 , x1 = 0 + 0.25 · [1 + 0 · sin(0)] = 0.25 . 19
- 24. Chapter 1 Plugging these values into the recursive equations above yields t2 = 0.25 + 0.25 = 0.5 and x2 = 0.25 + 0.25 · [1 + 0.25 · sin(0.0625)] = 0.503904 . Continuing in this way gives t3 = 0.75 and x3 = 0.503904 + 0.25 · [1 + 0.5 · sin(0.251952)] = 0.785066 , t4 = 1.00 and φ(1) ≈ x4 = 1.13920 . For N = 8, we have h = 1/8 = 0.125 . Thus, the recursive formulas become tn+1 = tn + 0.125 and xn+1 = xn + 0.125 · [1 + tn sin(tnxn)] . Using these formulas and starting with t0 = 0 and x0 = 0, we can ﬁll in Table 1-A. From this we see that φ(1) ≈ x8 = 1.19157, which is rounded to ﬁve decimal places. Table 1–A: Euler’s method approximations for the solution of x = 1+t sin(tx), x(0) = 0, at t = 1 with 8 steps (h = 1/8). nnn tttnnn xxxnnn 1 0.125 0.125 2 0.250 0.250244 3 0.375 0.377198 4 0.500 0.508806 5 0.625 0.649535 6 0.750 0.805387 7 0.875 0.983634 8 1.000 1.191572 9. To approximate the solution on the whole interval [1, 2] by Euler’s method with the step h = 0.1, we ﬁrst approximate the solution at the points xn = 1 + 0.1n, n = 1, . . ., 10. Then, on each subinterval [xn, xn+1], we approximate the solution by the linear interval, connecting 20
- 25. Exercises 1.4 (xn, yn) with (xn+1, yn+1), n = 0, 1, . . ., 9. Since f(x, y) = x−2 − yx−1 − y2 , the recursive formulas have the form xn+1 = xn + 0.1 , yn+1 = yn + 0.1 1 x2 n − yn xn − y2 n , n = 0, 1, . . ., 9 , x0 = 1, y0 = −1. Therefore, x1 = 1 + 0.1 = 1.1 , y1 = −1 + 0.1 1 12 − −1 1 − (−1)2 = −0.9 ; x2 = 1.1 + 0.1 = 1.2 , y2 = −0.9 + 0.1 1 1.12 − −0.9 1.1 − (−0.9)2 ≈ −0.81653719 ; x3 = 1.2 + 0.1 = 1.3 , y3 = −0.81653719 + 0.1 1 1.22 − −0.81653719 1.2 − (−0.81653719)2 ≈ −0.74572128 ; x4 = 1.3 + 0.1 = 1.4 , y4 = −0.74572128 + 0.1 1 1.32 − −0.74572128 1.3 − (−0.74572128)2 ≈ −0.68479653 ; etc. The results of these computations (rounded to ﬁve decimal places) are shown in Table 1-B. Table 1–B: Euler’s method approximations for the solutions of y = x−2 − yx−1 − y2 , y(1) = −1, on [1, 2] with h = 0.1. nnn xxxnnn yyynnn nnn xxxnnn yyynnn 0 1.0 −1.00000 6 1.6 −0.58511 1 1.1 −0.90000 7 1.7 −0.54371 2 1.2 −0.81654 8 1.8 −0.50669 3 1.3 −0.74572 9 1.9 −0.47335 4 1.4 −0.68480 10 2.0 −0.44314 5 1.5 −0.63176 The function y(x) = −1/x = x−1 , obviously, satisﬁes the initial condition, y(1) = −1. Further 21
- 26. Chapter 1 –1 0 1.2 1.4 1.6 1.8 2 Polygonal approximation y=−1/x Figure 1–C: Polygonal line approximation and the actual solution for Problem 9. we compute both sides of the given diﬀerential equation: y (x) = −x−1 = x−2 , f(x, y(x)) = x−2 − −x−1 x−1 − −x−1 2 = x−2 + x−2 − x−2 = x−2 . Thus, the function y(x) = −1/x is, indeed, the solution to the given initial value problem. The graphs of the obtained polygonal line approximation and the actual solution are sketched in Figure 1-C. 11. In this problem, the independent variable is t and the dependent variable is x; f(t, x) = 1+x2 , t0 = 0, and x0 = 0. The function φ(t) = tan t satisﬁes the initial condition: φ(0) = tan 0 = 0. The diﬀerential equation is also satisﬁed: dφ dt = sec2 t = 1 + tan2 t = 1 + φ(t)2 . Therefore, φ(t) is the solution to the given initial value problem. 22
- 27. Exercises 1.4 For approximation of φ(t) at the point t = 1 with N = 20 steps, we take the step size h = (1 − t0)/20 = 0.05. Thus, the recursive formulas for Euler’s method are tn+1 = tn + 0.05 , xn+1 = xn + 0.05 1 + x2 n . Applying these formulas with n = 0, 1, . . ., 19, we obtain x1 = x0 + 0.05 1 + x2 0 = 0.05 , x2 = x1 + 0.05 1 + x2 1 = 0.05 + 0.05 1 + 0.052 = 0.100125 , x3 = x2 + 0.05 1 + x2 2 = 0.100125 + 0.05 1 + 0.1001252 ≈ 0.150626 , ... x19 = x18 + 0.05 1 + x2 18 ≈ 1.328148 , φ(1) ≈ x20 = x19 + 0.05 1 + x2 19 = 1.328148 + 0.05 1 + 1.3281482 ≈ 1.466347 , which is a good enough approximation to φ(1) = tan 1 ≈ 1.557408. 13. From Problem 12, yn = (1 + 1/n)n and so limn→∞ [(e − yn)/(1/n)] is a 0/0 indeterminant. If we let h = 1/n in yn and use L’Hospital’s rule, we get lim n→∞ e − yn 1/n = lim h→0 e − (1 + h)1/h h = lim h→0 g(h) h = lim h→0 g (h) 1 , where g(h) = e − (1 + h)1/h . Writing (1 + h)1/h as eln(1+h)/h the function g(h) becomes g(h) = e − eln(1+h)/h . The ﬁrst derivative is given by g (h) = 0 − d dh eln(1+h)/h = −eln(1+h)/h d dh 1 h ln(1 + h) . Substituting Maclaurin’s series for ln(1 + h) we obtain g (h) = −(1 + h)1/h d dh 1 h h − 1 2 h2 + 1 3 h3 − 1 4 h4 + · · · 23
- 28. Chapter 1 = −(1 + h)1/h d dh 1 − 1 2 h + 1 3 h2 − 1 4 h3 + · · · = −(1 + h)1/h − 1 2 + 2 3 h − 3 4 h2 + · · · . Hence lim h→0 g (h) = lim h→0 −(1 + h)1/h − 1 2 + 2 3 h − 3 4 h2 + · · · = − lim h→0 (1 + h)1/h · lim h→0 − 1 2 + 2 3 h − 3 4 h2 + · · · . From calculus we know that e = lim h→0 (1 + h)1/h , which gives lim h→0 g (h) = −e − 1 2 = e 2 . So we have lim n→∞ e − yn 1/n = e 2 . 15. The independent variable in this problem is the time t and the dependent variable is the temperature T(t) of a body. Thus, we will use the recursive formulas (2) and (3) on page 25 with x replaced by t and y replaced by T. In the diﬀerential equation describing the Newton’s Law of Cooling, f(t, T) = K(M(t) − T). With the suggested values of K = 1 (min)−1 , M(t) ≡ 70◦ , h = 0.1, and the initial condition T(0) = 100◦ , the initial value problem becomes dT dt = 70 − T, T(0) = 100, and so the recursive formulas are tn+1 = tn + 0.1 , Tn+1 = Tn + 0.1(70 − Tn). For n = 0, t1 = t0 + 0.1 = 0.1 , T1 = T0 + 0.1(70 − T0) = 100 + 0.1(70 − 100) = 97 ; 24
- 29. Exercises 1.4 for n = 1, t2 = t1 + 0.1 = 0.2 , T2 = T1 + 0.1(70 − T1) = 97 + 0.1(70 − 97) = 94.3 ; for n = 2, t3 = t2 + 0.1 = 0.3 , T3 = T2 + 0.1(70 − T2) = 94.3 + 0.1(70 − 94.3) = 91.87 . Table 1–C: Euler’s method approximations for the solutions of T = K(M − T), T(0) = 100, with K = 1, M = 70, and h = 0.1. nnn tttnnn TTTnnn nnn tttnnn TTTnnn 0 0.0 100.00 11 1.1 79.414 1 0.1 97.000 12 1.2 78.473 2 0.2 94.300 13 1.3 77.626 3 0.3 91.870 14 1.4 76.863 4 0.4 89.683 15 1.5 76.177 5 0.5 87.715 16 1.6 75.559 6 0.6 85.943 17 1.7 75.003 7 0.7 84.349 18 1.8 74.503 8 0.8 82.914 19 1.9 74.053 9 0.9 81.623 20 2.0 73.647 10 1.0 80.460 By continuing this way and rounding results to three decimal places, we ﬁll in Table 1-C. From this table we conclude that (a) the temperature of a body after 1 minute T(1) ≈ 80.460◦ and (b) its temperature after 2 minutes T(2) ≈ 73.647◦ . 16. For this problem notice that the independent variable is t and the dependent variable is T. Hence, in the recursive formulas for Euler’s method, the t will take the place of the x and the 25
- 30. Chapter 1 T will take the place of the y. Also we see that h = 0.1 and f(t, T) = K (M4 − T4 ), where K = 40−4 and M = 70. The recursive formulas (2) and (3) on page 25 of the text become tn+1 = tn + 0.1 , Tn+1 = Tn + hf (tn, Tn) = Tn + 0.1 40−4 704 − T4 n , n = 0, 1, 2, . . . . From the initial condition, T(0) = 100, we see that t0 = 0 and T0 = 100. Therefore, for n = 0, t1 = t0 + 0.1 = 0 + 0.1 = 0.1 , T1 = T0 + 0.1 40−4 704 − T4 0 = 100 + 0.1 40−4 704 − 1004 ≈ 97.0316, where we have rounded oﬀ to four decimal places. For n = 1, we have t2 = t1 + 0.1 = 0.1 + 0.1 = 0.2 , T2 = T1 + 0.1 40−4 704 − T4 1 = 97.0316 + 0.1 40−4 704 − 97.03164 ≈ 94.5068 . By continuing this way, we ﬁll in Table 1-D. Table 1–D: Euler’s method approximations for the solution of T = K (M4 − T4 ), T(0) = 100, with K = 40−4 , M = 70, and h = 0.1. nnn tttnnn TTTnnn nnn tttnnn TTTnnn nnn tttnnn TTTnnn 0 0 100 7 0.7 85.9402 14 1.4 79.5681 1 0.1 97.0316 8 0.8 84.7472 15 1.5 78.9403 2 0.2 94.5068 9 0.9 83.6702 16 1.6 78.3613 3 0.3 92.3286 10 1.0 82.6936 17 1.7 77.8263 4 0.4 90.4279 11 1.1 81.8049 18 1.8 77.3311 5 0.5 88.7538 12 1.2 80.9934 19 1.9 76.8721 6 0.6 87.2678 13 1.3 80.2504 20 2.0 76.4459 From this table we see that T(1) = T(t10) ≈ T10 = 82.694 and T(2) = T(t20) ≈ T20 = 76.446 . 26
- 31. CHAPTER 2: First Order Diﬀerential Equations EXERCISES 2.2: Separable Equations, page 46 1. This equation is separable because we can separate variables by multiplying both sides by dx and dividing by 2y3 + y + 4. 3. This equation is separable because dy dx = yex+y x2 + 2 = ex x2 + 2 yey = g(x)p(y). 5. Writing the equation in the form ds dt = s + 1 st − s2 , we see that the right-hand side cannot be represented in the form g(t)p(s). Therefore, the equation is not separable. 7. Multiplying both sides of the equation by y2 dx and integrating yields y2 dy = (1 − x2 )dx ⇒ y2 dy = (1 − x2 )dx ⇒ 1 3 y3 = x − 1 3 x3 + C1 ⇒ y3 = 3x − x3 + C ⇒ y = 3 √ 3x − x3 + C , where C := 3C1 is an arbitrary constant. 9. To separate variables, we divide the equation by y and multiply by dx. This results dy dx = y(2 + sin x) ⇒ dy y = (2 + sin x)dx ⇒ dy y = (2 + sin x)dx ⇒ ln |y| = 2x − cos x + C1 ⇒ |y| = e2x−cos x+C1 = eC1 e2x−cos x = C2e2x−cos x , 27
- 32. Chapter 2 where C1 is an arbitrary constant and, therefore, C2 := eC1 is an arbitrary positive constant. We can rewrite the above solution in the form y = ±C2e2x−cos x = Ce2x−cos x , (2.1) with C := C2 or C = −C2. Thus C is an arbitrary nonzero constant. The value C = 0 in (2.1) gives y(x) ≡ 0, which is, clearly, is also a solution to the diﬀerential equation. Therefore, the answer to the problem is given by (2.1) with an arbitrary constant C. 11. Separating variables, we obtain dy sec2 y = dx 1 + x2 . Using the trigonometric identities sec y = 1/ cos y and cos2 y = (1 + cos 2y)/2 and integrating, we get dy sec2 y = dx 1 + x2 ⇒ (1 + cos 2y)dy 2 = dx 1 + x2 ⇒ (1 + cos 2y)dy 2 = dx 1 + x2 ⇒ 1 2 y + 1 2 sin 2y = arctan x + C1 ⇒ 2y + sin 2y = 4 arctan x + 4C1 ⇒ 2y + sin 2y = 4 arctan x + C. The last equation deﬁnes implicit solutions to the given diﬀerential equation. 13. Writing the given equation in the form dx/dt = x − x2 , we separate the variables to get dx x − x2 = dt . Integrate (the left side is integrated by partial fractions, with 1/(x − x2 ) = 1/x + 1/(1 − x)) to obtain: ln |x| − ln |1 − x| = t + c ⇒ ln x 1 − x = t + c ⇒ x 1 − x = ±et+c = Cet , where C = ec ⇒ x = Cet − xCet ⇒ x + xCet = Cet 28
- 33. Exercises 2.2 ⇒ x 1 + Cet = Cet ⇒ x = Cet 1 + Cet . Note: When C is replaced by −K, this answer can also be written as x = Ket /(Ket − 1). Further we observe that since we divide by x − x2 = x(1 − x), then x ≡ 0 and x ≡ 1 are also solutions. Allowing K to be zero gives x ≡ 0, but no choice for K will give x ≡ 1, so we list this as a separate solution. 15. To separate variables, we move the term containin dx to the right-hand side of the equation and divide both sides of the result by y. This yields y−1 dy = −yecos x sin x dx ⇒ y−2 dy = −ecos x sin x dx. Integrating the last equation, we obtain y−2 dy = (−ecos x sin x) dx ⇒ −y−1 + C = eu du (u = cos x) ⇒ − 1 y + C = eu = ecos x ⇒ y = 1 C − ecos x , where C is an arbitrary constant. 17. First we ﬁnd a general solution to the equation. Separating variables and integrating, we get dy dx = x3 (1 − y) ⇒ dy 1 − y = x3 dx ⇒ dy 1 − y = x3 dx ⇒ − ln |1 − y| + C1 = x4 4 ⇒ |1 − y| = exp C1 − x4 4 = Ce−x4/4 . To ﬁnd C, we use the initial condition, y(0) = 3. Thus, substitution 3 for y and 0 for x into the last equation yields |1 − 3| = Ce−04/4 ⇒ 2 = C. Therefore, |1 − y| = 2e−x4/4 . Finally, since 1 − y(0) = 1 − 3 < 0, on an interval containing x = 0 one has 1 − y(x) < 0 and so |1 − y(x)| = y(x) − 1. The solution to the problem is then y − 1 = 2e−x4/4 or y = 2e−x4/4 + 1. 29
- 34. Chapter 2 19. For a general solution, separate variables and integrate: dy dθ = y sin θ ⇒ dy y = sin θ dθ ⇒ dy y = sin θ dθ ⇒ ln |y| = − cos θ + C1 ⇒ |y| = e− cos θ+C1 = Ce− cos θ ⇒ y = −Ce− cos θ (because at the initial point, θ = π, y(π) < 0). We substitute now the initial condition, y(π) = −3, and obtain −3 = y(π) = −Ce− cos π = −Ce ⇒ C = 3e−1 . Hence, the answer is given by y = −3e−1 e− cos θ = −3e−1−cos θ . 21. Separate variables to obtain 1 2 (y + 1)−1/2 dy = cos x dx. Integrating, we have (y + 1)1/2 = sin x + C. Using the fact that y(π) = 0, we ﬁnd 1 = sin π + C ⇒ C = 1. Thus (y + 1)1/2 = sin x + 1 ⇒ y = (sin x + 1)2 − 1 = sin2 x + 2 sin x . 23. We have dy dx = 2x cos2 y ⇒ dy cos2 y = 2x dx ⇒ sec2 y dy = 2x dx ⇒ sec2 y dy = 2x dx ⇒ tan y = x2 + C. Since y = π/4 when x = 0, we get tan(π/4) = 02 + C and so C = 1. The solution, therefore, is tan y = x2 + 1 ⇔ y = arctan x2 + 1 . 30
- 35. Exercises 2.2 25. By separating variables we obtain (1 + y)−1 dy = x2 dx. Integrating yields ln |1 + y| = x3 3 + C . (2.2) Substituting y = 3 and x = 0 from the initial condition, we get ln 4 = 0 + C, which implies that C = ln 4. By substituting this value for C into equation (2.2) above, we have ln |1 + y| = x3 3 + ln 4 . Hence, eln |1+y| = e(x3/3)+ln 4 = ex3/3 eln 4 = 4ex3/3 ⇒ 1 + y = 4ex3/3 ⇒ y = 4ex3/3 − 1 . We can drop the absolute signs above because we are assuming from the initial condition that y is close to 3 and therefore 1 + y is positive. 27. (a) The diﬀerential equation dy/dx = ex2 separates if we multiply by dx. We integrate the separated equation from x = 0 to x = x1 to obtain x1 0 ex2 dx = x=x1 x=0 dy = y x=x1 x=0 = y(x1) − y(0). If we let t be the variable of integration and replace x1 by x and y(0) by 0, then we can express the solution to the initial value problem as y(x) = x 0 et2 dt. (b) The diﬀerential equation dy/dx = ex2 y−2 separates if we multiply by y2 and dx. We integrate the separated equation from x = 0 to x = x1 to obtain x1 0 ex2 dx = x1 0 y2 dy = 1 3 y3 x=x1 x=0 = 1 3 y(x1)3 − y(0)3 . 31
- 36. Chapter 2 If we let t be the variable of integration and replace x1 by x and y(0) by 1 in the above equation, then we can express the initial value problem as x 0 et2 dt = 1 3 y(x)3 − 1 . Solving for y(x) we arrive at y(x) = 1 + 3 x 0 et2 dt 1/3 . (2.3) (c) The diﬀerential equation dy/dx = √ 1 + sin x(1 + y2 ) separates if we divide by (1 + y2 ) and multiply by dx. We integrate the separated equation from x = 0 to x = x1 and ﬁnd x1 0 √ 1 + sin x dx = x=x1 x=0 (1 + y2 )−1 dy = tan−1 y(x1) − tan−1 y(0). If we let t be the variable of integration and replace x1 by x and y(0) by 1 then we can express the solution to the initial value problem by y(x) = tan x 0 √ 1 + sin t dt + π 4 . (d) We will use Simpson’s rule (Appendix B) to approximate the deﬁnite integral found in part (b). (Simpson’s rule is implemented on the website for the text.) Simpson’s rule requires an even number of intervals, but we don’t know how many are required to obtain the desired three-place accuracy. Rather than make an error analysis, we will compute the approximate value of y(0.5) using 2, 4, 6, . . . intervals for Simpson’s rule until the approximate values for y(0.5) change by less than ﬁve in the fourth place. For n = 2, we divide [0, 0.5] into 4 equal subintervals. Thus each interval will be of length (0.5 − 0)/4 = 1/8 = 0.125. Therefore, the integral is approximated by 0.5 0 ex2 dx = 1 24 e0 + 4e(0.125)2 + 2e(0.25)2 + 4e(0.325)2 + e(0.5)2 ≈ 0.544999003 . 32
- 37. Exercises 2.2 Substituting this value into equation (2.3) from part (b) yields y(0.5) ≈ [1 + 3(0.544999003)]1/3 ≈ 1.38121 . Repeating these calculations for n = 3, 4, and 5 yields Table 2-A. Table 2–A: Successive approximations for y(0.5) using Simpson’s rule. Number of Intervals yyy(0.5) 6 1.38120606 8 1.38120520 10 1.38120497 Since these values do not change by more than 5 in the fourth place, we can conclude that the ﬁrst three places are accurate and that we have obtained an approximate solution y(0.5) ≈ 1.381 . 29. (a) Separating variables and integrating yields dy y1/3 = dx ⇒ dy y1/3 = dx ⇒ 1 2/3 y2/3 = x + C1 ⇒ y = 2 3 x + 2 3 C1 3/2 = 2x 3 + C 3/2 . (b) Using the initial condition, y(0) = 0, we ﬁnd that 0 = y(0) = 2(0) 3 + C 3/2 = C3/2 ⇒ C = 0 , and so y = (2x/3 + 0)3/2 = (2x/3)3/2 , x ≥ 0, is a solution to the initial value problem. (c) The function y(x) ≡ 0, clearly, satisﬁes both, the diﬀerential equation dy/dx = y1/3 and the initial condition y(0) = 0. (d) In notation of Theorem 1 on page 12, f(x, y) = y1/3 and so ∂f ∂y = d dy y1/3 = 1 3 y−2/3 = 1 3y2/3 . 33
- 38. Chapter 2 Since ∂f/∂y is not continuous when y = 0, there is no rectangle containing the point (0, 0) in which both, f and ∂f/∂y, are continuous. Therefore, Theorem 1 does not apply to this initial value problem. 30. (a) Dividing the equation by (y + 1)2/3 and multiplying by dx separate variables. Thus we get dy dx = (x − 3)(y + 1)2/3 ⇒ dy (y + 1)2/3 = (x − 3)dx ⇒ dy (y + 1)2/3 = (x − 3)dx ⇒ 3(y + 1)1/3 = x2 2 − 3x + C1 ⇒ y + 1 = x2 6 − x + C1 3 3 ⇒ y = −1 + x2 6 − x + C 3 . (2.4) (b) Substitution y(x) ≡ −1 into the diﬀerential equation gives d(−1) dx = (x − 3)[(−1) + 1]2/3 ⇒ 0 = (x − 3) · 0, which is an identity. Therefore, y(x) ≡ −1 is, indeed, a solution. (c) With any choice of constant C, x2 /6 − x + C is a quadratic polynomial which is not identically zero. So, in (2.4), y = −1 + (x2 /6 − x + C) 3 ≡ − 1 for all C, and the solution y(x) ≡ −1 was lost in separation of variables. 31. (a) Separating variables and integrating yields dy y3 = x dx ⇒ dy y3 = x dx ⇒ 1 −2 y−2 = 1 2 x2 + C1 ⇒ y−2 = −x2 − 2C1 ⇒ x2 + y−2 = C, (2.5) where C := −2C1 is an arbitrary constant. (b) To ﬁnd the solution satisfying the initial condition y(0) = 1, we substitute in (2.5) 0 for x and 1 for y and obtain 02 + 1−2 = C ⇒ C = 1 ⇒ x2 + y−2 = 1. 34
- 39. Exercises 2.2 Solving for y yields y = ± 1 √ 1 − x2 . (2.6) Since, at the initial point, x = 0, y(0) = 1 > 1, we choose the positive sign in the above expression for y. Thus, the solution is y = 1 √ 1 − x2 . Similarly we ﬁnd solutions for the other two initial conditions: y(0) = 1 2 ⇒ C = 4 ⇒ y = 1 √ 4 − x2 ; y(0) = 2 ⇒ C = 1 4 ⇒ y = 1 (1/4) − x2 . (c) For the solution to the ﬁrst initial problem in (b), y(0) = 1, the domain is the set of all values of x satisfying two conditions 1 − x2 ≥ 0 (for existence of the square root) 1 − x2 = 0 (for existence of the quotient) ⇒ 1 − x2 > 0. Solving for x, we get x2 < 1 ⇒ |x| < 1 or − 1 < x < 1. In the same manner, we ﬁnd domains for solutions to the other two initial value problems: y(0) = 1 2 ⇒ −2 < x < 2 ; y(0) = 2 ⇒ − 1 2 < x < 1 2 . (d) First, we ﬁnd the solution to the initial value problem y(0) = a, a > 0, and its domain. Following the lines used in (b) and (c) for particular values of a, we conclude that y(0) = a ⇒ 02 + a−2 = C ⇒ y = 1 √ a−2 − x2 and so its domain is a−2 − x2 > 0 ⇒ x2 < a−2 ⇒ − 1 a < x < 1 a . As a → +0, 1/a → +∞, and the domain expands to the whole real line; as a → +∞, 1/a → 0, and the domain shrinks to x = 0. 35
- 40. Chapter 2 –4 –2 0 2 4 –2 2 a= 1 2 a=1 a=2 a=− 1 2 a=−1 a=−2 Figure 2–A: Solutions to the initial value problem y = xy3 , y(0) = a, a ± 0.5, ±1, and ±2. (e) For the values a = 1/2, 1, and 2 the solutions are found in (b); for a = −1, we just have to choose the negative sign in (2.6); similarly, we reverse signs in the other two solutions in (b) to obtain the answers for a = −1/2 and −2. The graphs of these functions are shown in Figure 2-A. 33. Let A(t) be the number of kilograms of salt in the tank at t minutes after the process begins. Then we have dA(t) dt = rate of salt in − rate of salt out. rate of salt in = 10 L/min × 0.3 kg/L = 3 kg/min. Since the tank is kept uniformly mixed, A(t)/400 is the mass of salt per liter that is ﬂowing out of the tank at time t. Thus we have rate of salt out = 10 L/min × A(t) 400 kg/L = A(t) 40 kg/min. 36
- 41. Exercises 2.2 Therefore, dA dt = 3 − A 40 = 120 − A 40 . Separating this diﬀerential equation and integrating yield 40 120 − A dA = dt ⇒ −40 ln |120 − A| = t + C ⇒ ln |120 − A| = − t 40 + C, where − C 40 is replaced by C ⇒ 120 − A = Ce−t/40 , where C can now be positive or negative ⇒ A = 120 − Ce−t/40 . There are 2 kg of salt in the tank initially, thus A(0) = 2. Using this initial condition, we ﬁnd 2 = 120 − C ⇒ C = 118 . Substituting this value of C into the solution, we have A = 120 − 118e−t/40 . Thus A(10) = 120 − 118e−10/40 ≈ 28.1 kg. Note: There is a detailed discussion of mixture problems in Section 3.2. 35. In Problem 34 we saw that the diﬀerential equation dT/dt = k(M − T) can be solved by separation of variables to yield T = Cekt + M. When the oven temperature is 120◦ we have M = 120. Also T(0) = 40. Thus 40 = C + 120 ⇒ C = −80. Because T(45) = 90, we have 90 = −80e45k + 120 ⇒ 3 8 = e45k ⇒ 45k = ln 3 8 . Thus k = ln(3/8)/45 ≈ −0.02180. This k is independent of M. Therefore, we have the general equation T(t) = Ce−0.02180t + M. 37
- 42. Chapter 2 (a) We are given that M = 100. To ﬁnd C we must solve the equation T(0) = 40 = C +100. This gives C = −60. Thus the equation becomes T(t) = −60e−0.02180t + 100. We want to solve for t when T(t) = 90. This gives us 90 = −60e−0.02180t + 100 ⇒ 1 6 = e−0.02180t ⇒ −0.0218t = ln 1 6 ⇒ 0.0218t = ln 6 . Therefore t = ln 6/0.0218 ≈ 82.2 min. (b) Here M = 140, so we solve T(0) = 40 = C + 140 ⇒ C = −100. As above, solving for t in the equation T(t) = −100e−0.02180t + 140 = 90 ⇒ t ≈ 31.8 . (c) With M = 80, we solve 40 = C + 80, yielding C = −40. Setting T(t) = −40e−0.02180t + 80 = 90 ⇒ − 1 4 = e−0.02180t . This last equation is impossible because an exponential function is never negative. Hence it never attains desired temperature. The physical nature of this problem would lead us to expect this result. A further discussion of Newton’s law of cooling is given in Section 3.3. 37. The diﬀerential equation dP dt = r 100 P 38
- 43. Exercises 2.2 separates if we divide by P and multiply by dt. 1 P dP = r 100 dt ⇒ ln P = r 100 t + C ⇒ P(t) = Kert/100 , where K is the initial amount of money in the savings account, K = $1000, and r% is the interest rate, r = 5. This results in P(t) = 1000e5t/100 . (2.7) (a) To determine the amount of money in the account after 2 years we substitute t = 2 into equation (2.7), which gives P(2) = 1000e10/100 = $1105.17 . (b) To determine when the account will reach $4000 we solve equation (2.7) for t with P = $4000: 4000 = 1000e5t/100 ⇒ e5t/100 = 4 ⇒ t = 20 ln 4 ≈ 27.73 years. (c) To determine the amount of money in the account after 31 2 years we need to determine the value of each $1000 deposit after 31 2 years has passed. This means that the initial $1000 is in the account for the entire 31 2 years and grows to the amount which is given by P0 = 1000e5(3.5)/100 . For the growth of the $1000 deposited after 12 months, we take t = 2.5 in equation (2.7) because that is how long this $1000 will be in the account. This gives P1 = 1000e5(2.5)/100 . Using the above reasoning for the remaining deposits we arrive at P2 = 1000e5(1.5)/100 and P3 = 1000e5(0.5)/100 . The total amount is determined by the sum of the Pi’s. P = 1000 e5(3.5)/100 + e5(2.5)/100 + e5(1.5)/100 + e5(0.5)/100 ≈ $4, 427.59 . 39. Let s(t), t > 0, denote the distance traveled by driver A from the time t = 0 when he ran out of gas to time t. Then driver A’s velocity vA(t) = ds/dt is a solution to the initial value problem dvA dt = −kv2 A , vA(0) = vB , 39
- 44. Chapter 2 where vB is driver B’s constant velocity, and k > 0 is a positive constant. Separating variables we get dvA v2 A = −k dt ⇒ dvA v2 A = − k dt ⇒ 1 vA(t) = kt + C . From the initial condition we ﬁnd 1 vB = 1 vA(0) = k · 0 + C = C ⇒ C = 1 vB . Thus vA(t) = 1 kt + 1/vB = vB vBkt + 1 . The function s(t) therefore satisﬁes ds dt = vB vBkt + 1 , s(0) = 0. Integrating we obtain s(t) = vB vBkt + 1 dt = 1 k ln (vBkt + 1) + C1 . To ﬁnd C1 we use the initial condition: 0 = s(0) = 1 k ln (vBk · 0 + 1) + C1 = C1 ⇒ C1 = 0. So, s(t) = 1 k ln (vBkt + 1) . At the moment t = t1 when driver A’s speed was halved, i.e., vA(t1) = vA(0)/2 = vB/2, we have 1 2 vB = vA(t1) = vB vBkt1 + 1 and 1 = s(t1) = 1 k ln (vBkt1 + 1) ⇒ vBkt1 + 1 = 2 and so k = ln (vBkt1 + 1) = ln 2 ⇒ s(t) = 1 ln 2 ln (vBt ln 2 + 1) . 40
- 45. Exercises 2.3 Since driver B was 3 miles behind driver A at time t = 0, and his speed remained constant, he ﬁnished the race at time tB = (3 + 2)/vB = 5/vB. At this moment, driver A had already gone s(tB) = 1 ln 2 ln (vBtB ln 2 + 1) = 1 ln 2 ln 5 vB vB ln 2 + 1 = 1 ln 2 ln (5 ln 2 + 1) ≈ 2.1589 > 2 miles, i.e., A won the race. EXERCISES 2.3: Linear Equations, page 54 1. Writing dy dx − x−2 y = −x−2 cos x , we see that this equation has the form (4) on page 50 of the text with P(x) = −x−2 and Q(x) = −x−2 cos x. Therefore, it is linear. Isolating dy/dx yields dy dx = y − cos x x2 . Since the right-hand side cannot be represented as a product g(x)p(y), the equation is not separable. 3. In this equation, the independent variable is t and the dependent variable is x. Dividing by x, we obtain dx dt = sin t x − t2 . Therefore, it is neither linear, because of the sin t/x term, nor separable, because the right- hand side is not a product of functions of single variables x and t. 5. This is a linear equation with independent variable t and dependent variable y. This is also a separable equation because dy dt = y(t − 1) t2 + 1 = t − 1 t2 + 1 y = g(t)p(y). 41
- 46. Chapter 2 7. In this equation, P(x) ≡ −1 and Q(x) = e3x . Hence the integrating factor µ(x) = exp P(x)dx = exp (−1)dx = e−x . Multiplying both sides of the equation by µ(x) and integrating, we obtain e−x dy dx − e−x y = e−x e3x = e2x ⇒ d (e−x y) dx = e2x ⇒ e−x y = e2x dx = 1 2 e2x + C ⇒ y = 1 2 e2x + C ex = e3x 2 + Cex . 9. This is a linear equation with dependent variable r and independent variable θ. The method we will use to solve this equation is exactly the same as the method we use to solve an equation in the variables x and y since these variables are just dummy variables. Thus we have P(θ) = tan θ and Q(θ) = sec θ which are continuous on any interval not containing odd multiples of π/2. We proceed as usual to ﬁnd the integrating factor µ(θ). We have µ(θ) = exp tan θ dθ = e− ln | cos θ|+C = K · 1 | cos θ| = K| sec θ|, where K = eC . Thus we have µ(θ) = sec θ, where we can drop the absolute value sign by making K = 1 if θ is in an interval where sec θ is positive or by making K = −1 if sec θ is negative. Multiplying the equation by the integrating factor yields sec θ dr dθ + (sec θ tan θ)r = sec2 θ ⇒ Dθ(r sec θ) = sec2 θ . Integrating with respect to θ yields r sec θ = sec2 θ dθ = tan θ + C ⇒ r = cos θ tan θ + C cos θ ⇒ r = sin θ + C cos θ . Because of the continuity of P(θ) and Q(θ) this solution is valid on any open interval that has end points that are consecutive odd multiples of π/2. 42
- 47. Exercises 2.3 11. Choosing t as the independent variable and y as the dependent variable, we put the equation put into standard form: t + y + 1 − dy dt = 0 ⇒ dy dt − y = t + 1. (2.8) Thus P(t) ≡ −1 and so µ(t) = exp (−1)dt = e−t . We multiply both sides of the second equation in (2.8) by µ(t) and integrate. This yields e−t dy dt − e−t y = (t + 1)e−t ⇒ d dt e−t y = (t + 1)e−t ⇒ e−t y = (t + 1)e−t dt = −(t + 1)e−t + e−t dt = −(t + 1)e−t − e−t + C = −(t + 2)e−t + C ⇒ y = et −(t + 2)e−t + C = −t − 2 + Cet , where we have used integration by parts to ﬁnd (t + 1)e−t dt. 13. In this problem, the independent variable is y and the dependent variable is x. So, we divide the equation by y to rewrite it in standard form. y dx dy + 2x = 5y2 ⇒ dx dy + 2 y x = 5y2 . Therefore, P(y) = 2/y and the integrating factor, µ(y), is µ(y) = exp 2 y dy = exp (2 ln |y|) = |y|2 = y2 . Multiplying the equation (in standard form) by y2 and integrating yield y2 dx dy + 2y x = 5y4 ⇒ d dy y2 x = 5y4 ⇒ y2 x = 5y4 dy = y5 + C ⇒ x = y−2 y5 + C = y3 + Cy−2 . 15. To put this linear equation in standard form, we divide by (x2 + 1) to obtain dy dx + x x2 + 1 y = x x2 + 1 . (2.9) 43
- 48. Chapter 2 Here P(x) = x/(x2 + 1), so P(x) dx = x x2 + 1 dx = 1 2 ln(x2 + 1). Thus the integrating factor is µ(x) = e(1/2) ln(x2+1) = eln[(x2+1)1/2 ] = (x2 + 1)1/2 . Multiplying equation (2.9) by µ(x) yields (x2 + 1)1/2 dy dx + x (x2 + 1)1/2 y = x (x2 + 1)1/2 , which becomes d dx (x2 + 1)1/2 y = x (x2 + 1)1/2 . Now we integrate both sides and solve for y to ﬁnd (x2 + 1)1/2 y = (x2 + 1)1/2 + C ⇒ y = 1 + C(x2 + 1)−1/2 . This solution is valid for all x since P(x) and Q(x) are continuous for all x. 17. This is a linear equation with P(x) = −1/x and Q(x) = xex which is continuous on any interval not containing 0. Therefore, the integrating factor is given by µ(x) = exp − 1 x dx = e− ln x = 1 x , for x > 0. Multiplying the equation by this integrating factor yields 1 x dy dx − y x2 = ex ⇒ Dx y x = ex . Integrating gives y x = ex + C ⇒ y = xex + Cx. Now applying the initial condition, y(1) = e − 1, we have e − 1 = e + C ⇒ C = −1. 44
- 49. Exercises 2.3 Thus, the solution is y = xex − x, on the interval (0, ∞). Note: This interval is the largest interval containing the initial value x = 1 in which P(x) and Q(x) are continuous. 19. In this problem, t is the independent variable and x is the dependent variable. One can notice that the left-hand side is the derivative of xt3 with respect to t. Indeed, using product rule for diﬀerentiation, we get d dt xt3 = dx dt t3 + x d (t3 ) dt = t3 dx dt + 3t2 x. Thus the equation becomes d dt xt3 = t ⇒ xt3 = t dt = t2 2 + C ⇒ x = t−3 t2 2 + C = 1 2t + C t3 . (Of course, one could divide the given equation by t3 to get standard form, conclude that P(t) = 3/t, ﬁnd that µ(t) = t3 , multiply by t3 back, and come up with the original equation.) We now use the initial condition, x(2) = 0, to ﬁnd C. 0 = x(2) = 1 2(2) + C 23 ⇒ 1 4 + C 8 = 0 ⇒ C = −2. Hence, the solution is x = 1/(2t) − 2/(t3 ). 21. Putting the equation in standard form yields dy dx + sin x cos x y = 2x cos x ⇒ dy dx + (tan x)y = 2x cos x. Therefore, P(x) = tan x and so µ(x) = exp tan x dx = exp (− ln | cos x|) = | cos x|−1 . 45
- 50. Chapter 2 At the initial point, x = π/4, cos(π/4) > 0 and, therefore, we can take µ(x) = (cos x)−1 . Multiplying the standard form of the given equation by µ(x) gives 1 cos x dy dx + sin x cos2 x y = 2x ⇒ d dx 1 cos x y = 2x ⇒ 1 cos x y = 2x dx = x2 + C ⇒ y = cos x x2 + C . From the initial condition, we ﬁnd C: −15 √ 2π2 32 = y π 4 = cos π 4 π 4 2 + C ⇒ C = −π2 . Hence, the solution is given by y = cos x (x2 − π2 ). 23. We proceed similarly to Example 2 on page 52 and obtain an analog of the initial value problem (13), that is, dy dt + 5y = 40e−20t , y(0) = 10. (2.10) Thus P(t) ≡ 5 and µ(t) = exp 5dt = e5t . Multiplying the diﬀerential equation in (2.10) by µ(t) and integrating, we obtain e5t dy dt + 5e5t y = 40e−20t e5t = 40e−15t ⇒ d (e5t y) dt = 40e−15t ⇒ e5t y = 40e−15t dt = 40 −15 e−15t + C. Therefore, a general solution to the diﬀerential equation in (2.10) is y = e−5t 40 −15 e−15t + C = Ce−5t − 8 3 e−20t . Finally, we ﬁnd C using the initial condition. 10 = y(0) = Ce−5·0 − 8 3 e−20·0 = C − 8 3 ⇒ C = 10 + 8 3 = 38 3 . Hence, the mass of RA2 for t ≥ 0 is given by y(t) = 38 3 e−5t − 8 3 e−20t . 46
- 51. Exercises 2.3 25. (a) This is a linear problem and so an integrating factor is µ(x) = exp 2x dx = exp x2 . Multiplying the equation by this integrating factor yields ex2 dy dx + 2xex2 y = ex2 ⇒ Dx yex2 = ex2 ⇒ x 2 Dt yet2 dt = x 2 et2 dt, where we have changed the dummy variable x to t and integrated with respect to t from 2 (since the initial value for x in the initial condition is 2) to x. Thus, since y(2) = 1, yex2 − e4 = x 2 et2 dt ⇒ y = e−x2 e4 + x 2 et2 dt = e4−x2 + e−x2 x 2 et2 dt . (b) We will use Simpson’s rule (page A.3 of the Appendix B) to approximate the deﬁnite integral found in part (a) with upper limit x = 3. Simpson’s rule requires an even number of intervals, but we don’t know how many are required to obtain the desired 3 place accuracy. Rather than make an error analysis, we will compute the approximate value of y(3) using 4, 6, 8, 10, 12, . . . intervals for Simpson’s rule until the approximate values for y(3) change by less than 5 in the fourth place. For n = 2 we divide [2, 3] into 4 equal subintervals. Thus, each subinterval will be of length (3 − 2)/4 = 1/4. Therefore, the integral is approximated by 3 2 et2 dt ≈ 1 12 e(2)2 + e(2.25)2 + e(2.5)2 + e(2.75)2 + e(3)2 ≈ 1460.354350 . Dividing this by e(3)2 and adding e4−32 = e−5 , gives y(3) ≈ 0.186960 . Doing calculations for 6, 8, 10, and 12 intervals yields Table 2-B. 47
- 52. Chapter 2 Table 2–B: Successive approximations for y(3) using Simpson’s rule. Number of Intervals yyy(3) 6 0.183905 8 0.183291 10 0.183110 12 0.183043 Since the last 3 approximate values do not change by more than 5 in the fourth place, it appears that their ﬁrst three places are accurate and the approximate solution is y(3) ≈ 0.183 . 27. (a) The given diﬀerential equation is in standard form. Thus P(x) = √ 1 + sin2 x. Since we cannot express P(x) dx as an elementary function, we use fundamental theorem of calculus to conclude that, with any ﬁxed constant a, x a P(t)dt = P(x), that is, the above deﬁnite integral with variable upper bound is an antiderivative of P(x). Since, in the formula for µ(x), one can choose any antiderivative of P(x), we take the above deﬁnite integral with a = 0. (Such a choice of a comes from the initial point x = 0 and makes it easy to satisfy the initial condition.) Therefore, the integrating factor µ(x) can be chosen as µ(x) = exp x 0 1 + sin2 t dt . Multiplying the diﬀerential equaion by µ(x) and integrating from x = 0 to x = s, we obtain d[µ(x)y] dx = µ(x)x ⇒ d[µ(x)y] = µ(x)x dx 48
- 53. Exercises 2.3 ⇒ s 0 d[µ(x)y] = s 0 µ(x)x dx ⇒ µ(x)y(x) x=s x=0 = s 0 µ(x)x dx ⇒ µ(s)y(s) − µ(0)y(0) = s 0 µ(x)x dx . From the initial condition, y(0) = 2. Also, note that µ(0) = exp 0 0 1 + sin2 t dt = e0 = 1. This yields µ(0)y(0) = 2 and so µ(s)y(s) = s 0 µ(x)x dx + 2 . Dividing by µ(s) and interchanging x and s give the required. (b) The values of µ(x), x = 0.1, 0.2, . . ., 1.0, approximated by using Simpson’s rule, are given in Table 2-C. Table 2–C: Approximations of ν(x) = x 0 √ 1 + sin2 t dt and µ(x) = eν(x) using Simpson’s rule. xxx ν(x)ν(x)ν(x) µ(x)µ(x)µ(x) xxx ν(x)ν(x)ν(x) µ(x)µ(x)µ(x) 0.0 0.0 1.0000 0.6 0.632016 1.881401 0.1 0.100166 1.105354 0.7 0.748903 2.114679 0.2 0.201315 1.223010 0.8 0.869917 2.386713 0.3 0.304363 1.355761 0.9 0.994980 2.704670 0.4 0.410104 1.506975 1.0 1.123865 3.076723 0.5 0.519172 1.680635 We now use these values of µ(x) to approximate 1 0 µ(s)s ds by applying Simpson’s rule again. With n = 5 and h = 1 − 0 2n = 0.1 49
- 54. Chapter 2 the Simpson’s rule becomes 1 0 µ(s)s ds ≈ 0.1 3 [µ(0)(0) + 4µ(0.1)(0.1) + 2µ(0.2)(0.2) + 4µ(0.3)(0.3) +2µ(0.4)(0.4) + 4µ(0.5)(0.5) + 2µ(0.6)(0.6) + 4µ(0.7)(0.7) +2µ(0.8)(0.8) + 4µ(0.9)(0.9) + µ(1.0)(1.0)] ≈ 1.064539 . Therefore, y(1) ≈ 1 µ(1) 1 0 µ(s)s ds + 2 µ(1) = 1 3.076723 · 1.064539 + 2 3.076723 = 0.9960 . (c) We rewrite the diﬀerential equation in the form used in Euler’s method, dy dx = x − 1 + sin2 x y , y(0) = 2, and conclude that f(x, y) = x − √ 1 + sin2 xy. Thus the recursive formulas (2) and (3) on page 25 of the text become xn+1 = xn + h, yn+1 = yn + h xn − 1 + sin2 xn yn , n = 0, 1, . . . , x0 = 0, y0 = 2. With h = 0.1 we need (1 − 0)/0.1 steps to get an approximation at x = 1. n = 0 : x1 = 0.1 , y1 = (2) + 0.1[(0) − 1 + sin2 (0) (2)] = 1.8000; n = 1 : x2 = 0.2 , y2 = (1.8) + 0.1[(0.1) − 1 + sin2 (0.1) (1.8)] ≈ 1.6291; n = 2 : x3 = 0.3 , y3 = (1.6291) + 0.1[(0.2) − 1 + sin2 (0.2) (1.6291)] ≈ 1.4830; ... Results of these computations, rounded oﬀ to four decimal places, are given in Table 2-D. Thus Euler’s method with step h = 0.1 gives y(1) ≈ 0.9486 . Next we take h = 0.05 and ﬁll in the Table 2-E. So, with step h = 0.05, we have y(1) ≈ 0.9729 . 50
- 55. Exercises 2.3 Table 2–D: Euler’s method approximations for the solution of y + y √ 1 + sin2 x = x, y(0) = 2, at x = 1 with h = 0.1. kkk xxxkkk yyykkk kkk xxxkkk yyykkk kkk xxxkkk yyykkk 0 0.0 2.0000 4 0.4 1.3584 8 0.8 1.0304 1 0.1 1.8000 5 0.5 1.2526 9 0.9 0.9836 2 0.2 1.6291 6 0.6 1.1637 10 1.0 0.9486 3 0.3 1.4830 7 0.7 1.0900 Table 2–E: Euler’s method approximations for the solution of y + y √ 1 + sin2 x = x, y(0) = 2, at x = 1 with h = 0.05. nnn xxxnnn yyynnn nnn xxxnnn yyynnn nnn xxxnnn yyynnn 0 0.00 2.0000 7 0.35 1.4368 14 0.70 1.1144 1 0.05 1.9000 8 0.40 1.3784 15 0.75 1.0831 2 0.10 1.8074 9 0.45 1.3244 16 0.80 1.0551 3 0.15 1.7216 10 0.50 1.2747 17 0.85 1.0301 4 0.20 1.6420 11 0.55 1.2290 18 0.90 1.0082 5 0.25 1.5683 12 0.60 1.1872 19 0.95 0.9892 6 0.30 1.5000 13 0.65 1.1490 20 1.00 0.9729 29. In the presented form, the equation dy dx = 1 e4y + 2x is, clearly, not linear. But, if we switch the roles of variables and consider y as the independent variable and x as the dependent variable (using the connection between derivatives of inverse functions, that is, the formula y (x) = 1/x (y)), then the equation transforms to dx dy = e4y + 2x ⇒ dx dy − 2x = e4y . This is a linear equation with P(y) = −2. Thus the integrating factor is µ(y) = exp (−2)dy = e−2y 51
- 56. Chapter 2 and so d dy e−2y x = e−2y e4y = e2y ⇒ e−2y x = e2y dy = e2y 2 + C. Solving for x yields x = e2y e2y 2 + C = e4y 2 + Ce2y . 31. (a) On the interval 0 ≤ x ≤ 2, we have P(x) = 1. Thus we are solving the equation dy dx + y = x, y(0) = 1. The integrating factor is given by µ(x) = exp dx = ex . Multiplying the equation by the integrating factor, we obtain ex dy dx + ex y = xex ⇒ Dx [ex y] = xex ⇒ ex y = xex dx . Calculating this integral by parts and dividing by ex yields y = e−x (xex − ex + C) = x − 1 + Ce−x . (b) Using the initial condition, y(0) = 1, we see that 1 = y(0) = 0 − 1 + C = −1 + C ⇒ C = 2. Thus the solution becomes y = x − 1 + 2e−x . (c) In the interval x > 2, we have P(x) = 3. Therefore, the integrating factor is given by µ(x) = exp 3 dx = e3x . Multiplying the equation by this factor and solving yields e3x dy dx + 3e3x y = xe3x ⇒ Dx e3x y = xe3x ⇒ e3x y = xe3x dx . Integrating by parts and dividing by e3x gives y = e−3x 1 3 xe3x − 1 9 e3x + C = x 3 − 1 9 + Ce−3x . 52
- 57. Exercises 2.3 (d) We want the value of the initial point for the solution in part (c) to be the value of the solution found in part (b) at the point x = 2. This value is given by y(2) = 2 − 1 + 2e−2 = 1 + 2e−2 . Thus the initial point we seek is y(2) = 1 + 2e−2 . Using this initial point to ﬁnd the constant C given in part (c) yields 1 + 2e−2 = y(2) = 2 3 − 1 9 + Ce−6 ⇒ C = 4 9 e6 + 2e4 . Thus, the solution of the equation on the interval x > 2 is given by y = x 3 − 1 9 + 4 9 e6 + 2e4 e−3x . Patching these two solutions together gives us a continuous solution to the original equa- tion on the interval x ≥ 0: y = x − 1 + 2e−x , 0 ≤ x ≤ 2; x 3 − 1 9 + 4 9 e6 + 2e4 e−3x , 2 < x. (e) The graph of the solution is given in Figure B.18 of the answers in the text. 33. (a) Writing the equation in standard form yields dy dx + 2 x y = 3. Therefore, P(x) = 2/x and µ(x) = exp 2 x dx = exp (2 ln |x|) = |x|2 = x2 . Hence d dx x2 y = 3x2 ⇒ x2 y = 3x2 dx = x3 + C ⇒ y = x + C x2 53
- 58. Chapter 2 is a general solution to the given diﬀerential equation. Unless C = 0 and so y = x, the function y = x + C/x2 is not deﬁned when x = 0. Therefore, among all solutions, the only function deﬁned at x = 0 is φ(x) = x, and the initial value problem with y(0) = y0 has a solution (and unique) if and only if y0 = φ(x) x=0 = 0. (b) Standard form of the equation xy − 2y = 3x is dy dx − 2 x y = 3. This gives P(x) = −2/x, µ(x) = exp (−2/x)dx = x−2 , and d dx x−2 y = 3x−2 ⇒ x−2 y = 3x−2 dx = −3x−1 + C ⇒ y = −3x + Cx2 . Therefore, any solution is a polynomial and so is deﬁned for all real numbers. Moreover, any solution satisﬁes the initial condition y(0) = 0 because −3x + Cx2 x=0 = −3(0) + C(0)2 = 0 and, therefore, is a solution to the initial value problem. (This also implies that the initial value problem with y(0) = y0 = 0 has no solution.) 35. (a) This part of the problem is similar to Problem 33 in Section 2.2. So, we proceed in the same way. Let A(t) denote the mass of salt in the tank at t minutes after the process begins. Then we have rate of input = 5 L/min × 0.2 kg/L = 1 kg/min , rate of exit = 5 L/min × A(t) 500 kg/L = A(t) 100 kg/min , dA dt = 1 − A 100 = 100 − A 100 . Separating this diﬀerential equation yields dA/(100 − A) = dt/100. Integrating, we obtain − ln |100 − A| = t 100 + C1 ⇒ |100 − A| = e−t/100−C1 = e−C1 e−t/100 54
- 59. Exercises 2.3 ⇒ 100 − A = Ce−t/100 C = ±e−C1 ⇒ A = 100 − Ce−t/100 . The initial condition, A(0) = 5 (initially, there were 5 kg of salt in the tank) implies that 5 = A(0) = 100 − C ⇒ C = 95. Substituting this value of C into the solution, we have A(t) = 100 − 95e−t/100 . Thus the mass of salt in the tank after 10 min is A(10) = 100 − 95e−10/100 ≈ 14.04 kg , which gives the concentration 14.04 kg/500 L ≈ 0.0281 kg/L. (b) After the leak develops, the system satisﬁes a new diﬀerential equation. While the rate of input remains the same, 1 kg/min, the rate of exit is now diﬀerent. Since, every minute, 5 liters of the solution is coming in and 5 + 1 = 6 liters are going out, the volume of the solution in the tank decreases by 6 − 5 = 1 liter per minute. Thus, for t ≥ 10, the volume of the solution in the tank is 500 − 1 · (t − 10) = 510 − t liters. This gives the concentration of salt in the tank A(t) 510 − t kg/L (2.11) and rate of exit = 6 L/min × A(t) 510 − t kg/L = 6A(t) 510 − t kg/min . Hence, the diﬀerential equation, for t > 10, becomes dA dt = 1 − 6A 510 − t ⇒ dA dt + 6A 510 − t = 1 with the initial condition A(10) = 14.04 (the value found in (a) ). This equation is a linear equation. We have µ(t) = exp 6 510 − t dt = exp (−6 ln |510 − t|) = (510 − t)−6 55
- 60. Chapter 2 ⇒ d dt (510 − t)−6 A = 1 · (510 − t)−6 = (510 − t)−6 ⇒ (510 − t)−6 A = (510 − t)−6 dt = 1 5 (510 − t)−5 + C ⇒ A = 1 5 (510 − t) + C(510 − t)6 . Using the initial condition, A(10) = 14.04, we compute C. 14.04 = A(10) = 1 5 (510 − 10) + C(510 − 10)6 ⇒ C = − 85.96 (500)6 . Therefore, A(t) = 1 5 (510 − t) − 85.96 (500)6 (510 − t)6 = 1 5 (510 − t) − 85.96 510 − t 500 6 and, according to (2.11), the concentration of salt is given by A(t) 510 − t = 1 5 − 85.96 510 − t · 510 − t 500 6 . 20 minutes after the leak develops, that is, when t = 30, the concentration will be 1 5 − 85.96 510 − 30 · 510 − 30 500 6 ≈ 0.0598 kg/L . 37. We are solving the equation dx dt + 2x = 1 − cos πt 12 , x(0) = 10. This is a linear problem with dependent variable x and independent variable t so that P(t) = 2. Therefore, to solve this equation we ﬁrst must ﬁnd the integrating factor µ(t). µ(t) = exp 2 dt = e2t . Multiplying the equation by this factor yields e2t dx dt + 2xe2t = e2t 1 − cos πt 12 = e2t − e2t cos πt 12 ⇒ xe2t = e2t dt − e2t cos πt 12 dt = 1 2 e2t − e2t cos πt 12 dt. 56
- 61. Exercises 2.3 The last integral can be found by integrating by parts twice which leads back to an integral similar to the original. Combining these two similar integrals and simplifying, we obtain e2t cos πt 12 dt = e2t 2 cos πt 12 + π 12 sin πt 12 4 + ( π 12 )2 + C. Thus we see that x(t) = 1 2 − 2 cos πt 12 + π 12 sin πt 12 4 + ( π 12 )2 + Ce−2t . Using the initial condition, t = 0 and x = 10, to solve for C, we obtain C = 19 2 + 2 4 + ( π 12 )2 . Therefore, the desired solution is x(t) = 1 2 − 2 cos πt 12 + π 12 sin πt 12 4 + ( π 12 )2 + 19 2 + 2 4 + ( π 12 )2 e−2t . 39. Let Tj(t), j = 0, 1, 2, . . ., denote the temperature in the classroom for 9 + j ≤ t < 10 + j, where t = 13 denotes 1 : 00 p.m., t = 14 denotes 2 : 00 p.m., etc. Then T(9) = 0, (2.12) and the continuity of the temperature implies that lim t→10+j = Tj+1(10 + j), j = 0, 1, 2, . . . . (2.13) According to the work of the heating unit, the temperature satisﬁes the equation dTj dt = 1 − Tj , if j = 2k −Tj , if j = 2k + 1 , 9 + j < t < 10 + j k = 0, 1, . . . . The general solutions of these equations are: for j even dTj dt = 1 − Tj ⇒ dTj 1 − Tj = dt ⇒ ln |1 − Tj| = −t + cj ⇒ Tj(t) = 1 − Cje−t ; 57
- 62. Chapter 2 for j odd dTj dt = −Tj ⇒ dTj −Tj = dt ⇒ ln |Tj| = −t + cj ⇒ Tj(t) = Cje−t ; where Cj = 0 are constants. From (2.12) we have: 0 = T0(9) = 1 − C0e−t t=9 = 1 − C0e−9 ⇒ C0 = e9 . Also from (2.13), for even values of j (say, j = 2k) we get 1 − C2ke−t t=9+(2k+1) = C2k+1e−t t=9+(2k+1) ⇒ 1 − C2ke−(10+2k) = C2k+1e−(10+2k) ⇒ C2k+1 = e10+2k − C2k . Similarly from (2.13) for odd values of j (say, j = 2k + 1) we get C2k+1e−t t=9+(2k+2) = 1 − C2k+2e−t t=9+(2k+2) ⇒ C2k+1e−(11+2k) = 1 − C2k+2e−(11+2k) ⇒ C2k+2 = e11+2k − C2k+1 . In general we see that for any integer j (even or odd) the following formula holds: Cj = e9+j − Cj−1. Using this recurrence formula we successively compute C1 = e10 − C0 = e10 − e9 = e9 (e − 1) C2 = e11 − C1 = e11 − e10 + e9 = e9 (e2 − e + 1) ... Cj = e9 j k=0 (−1)j−k ek . 58
- 63. Exercises 2.4 Therefore, the temperature at noon (when t = 12 and j = 3) is T3(12) = C3e−12 = e−12 e9 3 k=0 (−1)3−k ek = 1 − e−1 + e−2 − e−3 ≈ 0.718 = 71.8◦ F. At 5 p.m.(when t = 17 and j = 8), we ﬁnd T8(17) = 1 − C8e−17 = 1 − e−17 e9 8 k=0 (−1)8−k ek = 8 k=1 (−1)k+1 e−k = e−1 · 1 − (−e−1 )8 1 + e−1 ≈ 0.269 = 26.9◦ F. EXERCISES 2.4: Exact Equations, page 65 1. In this equation, M(x, y) = x2 y + x4 cos x and N(x, y) = −x3 . Taking partial derivatives, we obtain ∂M ∂y = ∂ ∂y x2 y + x4 = x2 = −3x2 = ∂N ∂x . Therefore, according to Theorem 2 on page 61 of the text, the equation is not exact. Rewriting the equation in the form dy dx = x2 y + x4 cos x x3 = 1 x y + x cos x, (2.14) we conclude that it is not separable because the right-hand side in (2.14) cannot be factored as p(x)q(y). We also see that the equation is linear with y as the dependent variable. 3. Here M(x, y) = yexy + 2x, N(x, y) = xexy − 2y. Thus ∂M ∂y = ∂ ∂y (yexy + 2x) = exy + y ∂ ∂y (exy ) = exy + yexy x = exy (1 + yx), ∂N ∂x = ∂ ∂x (xexy − 2y) = exy + x ∂ ∂x (exy ) = exy + xexy y = exy (1 + xy), ∂M/∂y = ∂N/∂x, and the equation is exact. We write the equation in the form dy dx = − yexy + 2x xexy − 2y 59
- 64. Chapter 2 and conclude that it is not separable because the right-hand side cannot be represented as a product of two functions of single variables x and y. Also, the right-hand side is not linear with respect to y which implies that the equation is not linear with y as the dependent variable. Similarly, choosing x as the dependent variable (taking the reciprocals of both sides) we conclude that the equation is not linear either. 5. The diﬀerential equation is not separable because (2xy + cos y) cannot be factored. This equation can be put in standard form by deﬁning x as the dependent variable and y as the independent variable. This gives dx dy + 2 y x = − cos y y2 , so we see that the diﬀerential equation is linear. If we set M(x, y) = y2 and N(x, y) = 2xy + cos y we are able to see that the diﬀerential equation is also exact because My(x, y) = 2y = Nx(x, y). 7. In this problem, the variables are r and θ, M(r, θ) = θ, and N(r, θ) = 3r − θ − 1. Because ∂M ∂θ = 1 = 3 = ∂N ∂r , the equation is not exact. With r as the dependent variable, the equation takes the form dr dθ = − 3r − θ − 1 θ = − 3 θ r + θ + 1 θ , and it is linear. Since the right-hand side in the above equation cannot be factored as p(θ)q(r), the equation is not separable. 9. We have that M(x, y) = 2xy + 3 and N(x, y) = x2 − 1. Therefore, My(x, y) = 2x = Nx(x, y) and so the equation is exact. We will solve this equation by ﬁrst integrating M(x, y) with respect to x, although integration of N(x, y) with respect to y is equally easy. Thus F(x, y) = (2xy + 3) dx = x2 y + 3x + g(y). 60
- 65. Exercises 2.4 Diﬀerentiating F(x, y) with respect to y gives Fy(x, y) = x2 + g (y) = N(x, y) = x2 − 1. From this we see that g = −1. (As a partial check we note that g (y) does not involve x.) Integrating gives g(y) = (−1) dy = −y. Since the constant of integration will be incorporated into the parameter of the solution, it is not written here. Substituting this expression for g(y) into the expression that we found for F(x, y) yields F(x, y) = x2 y + 3x − y. Therefore, the solution of the diﬀerential equation is x2 y + 3x − y = C ⇒ y = C − 3x x2 − 1 . The given equation could be solved by the method of grouping. To see this, express the diﬀer- ential equation in the form (2xy dx + x2 dy) + (3 dx − dy) = 0. The ﬁrst term of the left-hand side we recognize as the total diﬀerential of x2 y. The second term is the total diﬀerential of (3x − y). Thus we again ﬁnd that F(x, y) = x2 y + 3x − y and, again, the solution is x2 y + 3x − y = C. 11. Computing partial derivatives of M(x, y) = cos x cos y + 2x and N(x, y) = −(sin x sin y + 2y), we obtain ∂M ∂y = ∂ ∂y (cos x cos y + 2x) = − cos x sin y , ∂N ∂x = ∂ ∂x [− (sin x sin y + 2y)] = − cos x sin y , ⇒ ∂M ∂y = ∂N ∂x , and the equation is exact. 61
- 66. Chapter 2 Integrating M(x, y) with respect to x yields F(x, y) = M(x, y)dx = (cos x cos y + 2x) dx = cos y cos x dx + 2x dx = sin x cos y + x2 + g(y). To ﬁnd g(y), we compute the partial derivative of F(x, y) with respect to y and compare the result with N(x, y). ∂F ∂y = ∂ ∂y sin x cos y + x2 + g(y) = − sin x sin y + g (y) = − (sin x sin y + 2y) ⇒ g (y) = −2y ⇒ g(y) = (−2y)dy = −y2 . (We take the integration constant C = 0.) Therefore, F(x, y) = sin x cos y + x2 − y2 = c is a general solution to the given equation. 13. In this equation, the variables are y and t, M(y, t) = t/y, N(y, t) = 1 + ln y. Since ∂M ∂t = ∂ ∂t t y = 1 y and ∂N ∂y = ∂ ∂y (1 + ln y) = 1 y , the equation is exact. Integrating M(y, t) with respect to y, we get F(y, t) = t y dy = t ln |y| + g(t) = t ln y + g(t). (From N(y, t) = 1 + ln y we conclude that y > 0.) Therefore, ∂F ∂t = ∂ ∂t [t ln y + g(t)] = ln y + g (t) = 1 + ln y ⇒ g (t) = 1 ⇒ g(t) = t ⇒ F(y, t) = t ln y + t, and a general solution is given by t ln y + t = c (or, explicitly, t = c/(ln y + 1)). 62
- 67. Exercises 2.4 15. This diﬀerential equation is expressed in the variables r and θ. Since the variables x and y are dummy variables, this equation is solved in exactly the same way as an equation in x and y. We will look for a solution with independent variable θ and dependent variable r. We see that the diﬀerential equation is expressed in the diﬀerential form M(r, θ) dr + N(r, θ) dθ = 0, where M(r, θ) = cos θ and N(r, θ) = −r sin θ + eθ . This implies that Mθ(r, θ) = − sin θ = Nr(r, θ), and so the equation is exact. Therefore, to solve the equation we need to ﬁnd a function F(r, θ) that has cos θ dr + (−r sin θ + eθ ) dθ as its total diﬀerential. Integrating M(r, θ) with respect to r we see that F(r, θ) = cos θ dr = r cos θ + g(θ) ⇒ Fθ(r, θ) = −r sin θ + g (θ) = N(r, θ) = −r sin θ + eθ . Thus we have that g (θ) = eθ ⇒ g(θ) = eθ , where the constant of integration will be incorporated into the parameter of the solution. Substituting this expression for g(θ) into the expression we found for F(r, θ) yields F(r, θ) = r cos θ + eθ . From this we see that the solution is given by the one parameter family r cos θ + eθ = C, or, solving for r, r = C − eθ cos θ = (C − eθ ) sec θ. 17. Partial derivatives of M(x, y) = 1/y and N(x, y) = − (3y − x/y2 ) are ∂M ∂y = ∂ ∂y 1 y = − 1 y2 and ∂N ∂x = ∂ ∂x −3y + x y2 = 1 y2 . Since ∂M/∂y = ∂N/∂x, the equation is not exact. 63
- 68. Chapter 2 19. Taking partial derivatives of M(x, y) = 2x + y/(1 + x2 y2 ) and N(x, y) = −2y + x/(1 + x2 y2 ) with respect to y and x, respectively, we get ∂M ∂y = ∂ ∂y 2x + y 1 + x2y2 = (1)(1 + x2 y2 ) − yx2 (2y) (1 + x2y2)2 = 1 − x2 y2 (1 + x2y2)2 , ∂N ∂x = ∂ ∂x −2y + x 1 + x2y2 = (1)(1 + x2 y2 ) − xy2 (2x) (1 + x2y2)2 = 1 − x2 y2 (1 + x2y2)2 . Therefore, the equation is exact. F(x, y) = 2x + y 1 + x2y2 dx = x2 + d(xy) 1 + (xy)2 = x2 + arctan(xy) + g(y) ∂F ∂y = ∂ ∂y x2 + arctan(xy) + g(y) = x 1 + (xy)2 + g (y) = −2y + x 1 + x2y2 ⇒ g (y) = −2y ⇒ g(y) = −y2 ⇒ F(x, y) = x2 − y2 + arctan(xy) and a general solution then is given implicitly by x2 − y2 + arctan(xy) = c. 21. We check the equation for exactness. We have M(x, y) = 1/x + 2y2 x, N(x, y) = 2yx2 − cos y, ∂M ∂y = ∂ ∂y 1 x + 2y2 x = 4yx, ∂N ∂x = ∂ ∂x 2yx2 − cos y = 4yx. Thus ∂M/∂y = ∂N/∂x. Integrating M(x, y) with respect to x yields F(x, y) = 1 x + 2y2 x dx = ln |x| + x2 y2 + g(y). Therefore, ∂F ∂y = ∂ ∂y ln |x| + x2 y2 + g(y) = 2x2 y + g (y) = N(x, y) = 2yx2 − cos y ⇒ g (y) = − cos y ⇒ g(y) = (− cos y)dy = − sin y ⇒ F(x, y) = ln |x| + x2 y2 − sin y, and a general solution to the given diﬀerential equation is ln |x| + x2 y2 − sin y = c. 64
- 69. Exercises 2.4 Substituting the initial condition, y = π when x = 1, we ﬁnd c. ln |1| + 12 π2 − sin π = c ⇒ c = π2 . Therefore, the answer is given implicitly by ln |x| + x2 y2 − sin y = π2 . (We also used the fact that at the initial point, (1, π), x > 0 to skip the absolute value sign in the logarithmic term.) 23. Here M(t, y) = et y + tet y and N(t, y) = tet + 2. Thus My(t, y) = et + tet = Nt(t, y) and so the equation is exact. To ﬁnd F(t, y) we ﬁrst integrate N(t, y) with respect to y to obtain F(t, y) = (tet + 2) dy = (tet + 2)y + h(t), where we have chosen to integrate N(t, y) because this integration is more easily accomplished. Thus Ft(t, y) = et y + tet y + h (t) = M(t, y) = et y + tet y ⇒ h (t) = 0 ⇒ h(t) = C. We will incorporate this constant into the parameter of the solution. Combining these results gives F(t, y) = tet y + 2y. Therefore, the solution is given by tet y + 2y = C. Solving for y yields y = C/(tet + 2). Now we use the initial condition y(0) = −1 to ﬁnd the solution that passes through the point (0, −1). Thus y(0) = C 0 + 2 = −1 ⇒ C 2 = −1 ⇒ C = −2. This gives us the solution y = − 2 tet + 2 . 25. One can check that the equation is not exact (∂M/∂y = ∂N/∂x), but it is separable because it can be written in the form y2 sin x dx + 1 − y x dy = 0 ⇒ y2 sin x dx = y − 1 x dy ⇒ x sin x dx = y − 1 y2 dy. 65
- 70. Chapter 2 Integrating both sides yields x sin x dx = y − 1 y2 dy ⇒ x(− cos x) − (− cos x)dx = 1 y − 1 y2 dy ⇒ −x cos x + sin x = ln |y| + 1 y + C, where we applied integration by parts to ﬁnd x sin x dx. Substitution of the initial condition, y(π) = 1, results −π cos π + sin π = ln |1| + 1 1 + C ⇒ C = π − 1. So, the solution to the initial value problem is −x cos x + sin x = ln y + 1/y + π − 1 . (Since y(π) = 1 > 0, we have removed the absolute value sign in the logarithmic term.) 27. (a) We want to ﬁnd M(x, y) so that for N(x, y) = sec2 y − x/y we have My(x, y) = Nx(x, y) = − 1 y . Therefore, we must integrate this last expression with respect to y. That is, M(x, y) = − 1 y dy = − ln |y| + f(x), where f(x), the “constant” of integration, is a function only of x. (b) We want to ﬁnd M(x, y) so that for N(x, y) = sin x cos y − xy − e−y we have My(x, y) = Nx(x, y) = cos x cos y − y. Therefore, we must integrate this last expression with respect to y. That is M(x, y) = (cos x cos y − y) dy = cos x cos y dy − y dy = cos x sin y − y2 2 + f(x), where f(x), a function only of x, is the “constant” of integration. 66
- 71. Exercises 2.4 29. (a) We have M(x, y) = y2 + 2xy and N(x, y) = −x2 . Therefore My(x, y) = 2y + 2x and Nx(x, y) = −2x. Thus My(x, y) = Nx(x, y), so the diﬀerential equation is not exact. (b) If we multiply (y2 + 2xy)dx − x2 dy = 0 by y−2 , we obtain 1 + 2x y dx − x2 y2 dy = 0. In this equation we have M(x, y) = 1 + 2xy−1 and N(x, y) = −x2 y−2 . Therefore, ∂M(x, y) ∂y = − 2x y2 = ∂N(x, y) ∂x . So the new diﬀerential equation is exact. (c) Following the method for solving exact equations we integrate M(x, y) in part (b) with respect to x to obtain F(x, y) = 1 + 2 x y dx = x + x2 y + g(y) . To determine g(y), take the partial derivative of both sides of the above equation with respect to y to obtain ∂F ∂y = − x2 y2 + g (y) . Substituting N(x, y) (given in part (b)) for ∂F/∂y, we can now solve for g (y) to obtain N(x, y) = − x2 y2 = − x2 y2 + g (y) ⇒ g (y) = 0 . The integral of g (y) will yield a constant and the choice of the constant of integration is not important so we can take g(y) = 0. Hence we have F(x, y) = x + x2 /y and the solution to the equation is given implicitly by x + x2 y = C . Solving the above equation for y, we obtain y = x2 C − x . 67
- 72. Chapter 2 (d) By dividing both sides by y2 we lost the solution y ≡ 0. 31. Following the proof of Theorem 2, we come to the expression (10) on page 63 of the text for g (y), that is g (y) = N(x, y) − ∂ ∂y x x0 M(s, y) ds (2.15) (where we have replaced the integration variable t by s). In other words, g(y) is an antideriva- tive of the right-hand side in (2.15). Since an antiderivative is deﬁned up to an additive constant and, in Theorem 2, such a constant can be chosen arbitrarily (that is, g(y) can be any antiderivative), we choose g(y) that vanishes at y0. According to fundamental theorem of calculus, this function can be written in the form g(y) = y y0 g (t) dt = y y0 N(x, t) − ∂ ∂t x x0 M(s, t) ds dt = y y0 N(x, t) dt − y y0 ∂ ∂t x x0 M(s, t) ds dt = y y0 N(x, t) dt − x x0 M(s, t) ds t=y t=y0 = y y0 N(x, t) dt − x x0 M(s, y) ds + x x0 M(s, y0) ds . Substituting this function into the formula (9) on page 63 of the text, we conclude that F(x, y) = x x0 M(t, y) dt + y y0 N(x, t) dt − x x0 M(s, y) ds + x x0 M(s, y0) ds = y y0 N(x, t) dt + x x0 M(s, y0) ds . (a) In the diﬀerential form used in Example 1, M(x, y) = 2xy2 + 1 and N(x, y) = 2x2 y. 68
- 73. Exercises 2.4 Thus, N(x, t) = 2x2 t and M(s, y0) = 2s · 02 + 1 = 1, and (18) yields F(x, y) = y 0 2x2 t dt + x 0 1 · ds = x2 y 0 2t dt + x 0 ds = x2 t2 t=y t=0 +s s=x s=0 = x2 y2 + x. (b) Since M(x, y) = 2xy − sec2 x and N(x, y) = x2 + 2y, we have N(x, t) = x2 + 2t and M(s, y0) = 2s · 0 − sec2 s = − sec2 s, F(x, y) = y 0 x2 + 2t dt + x 0 − sec2 s ds = x2 t + t2 t=y t=0 − tan s s=x s=0 = x2 y + y2 − tan x. (c) Here, M(x, y) = 1 + ex y + xex y and N(x, y) = xex + 2. Therefore, N(x, t) = xex + 2 and M(s, y0) = 1 + es · 0 + ses · 0 = 1, F(x, y) = y 0 (xex + 2) dt + x 0 1 · ds = (xex + 2) t t=y t=0 +s s=x s=0 = (xex + 2) y + x, which is identical to F(x, y) obtained in Example 3. 32. (a) The slope of the orthogonal curves, say m⊥, must be −1/m, where m is the slope of the original curves. Therefore, we have m⊥ = Fy(x, y) Fx(x, y) ⇒ dy dx = Fy(x, y) Fx(x, y) ⇒ Fy(x, y) dx − Fx(x, y) dy = 0. (b) Let F(x, y) = x2 + y2 . Then we have Fx(x, y) = 2x and Fy(x, y) = 2y. Plugging these expressions into the ﬁnal result of part (a) gives 2y dx − 2x dy = 0 ⇒ y dx − x dy = 0. 69
- 74. Chapter 2 To ﬁnd the orthogonal trajectories, we must solve this diﬀerential equation. To this end, note that this equation is separable and thus 1 x dx = 1 y dy ⇒ ln |x| = ln |y| + C ⇒ eln |x|−C = eln |y| ⇒ y = kx, where k = ±e−C . Therefore, the orthogonal trajectories are lines through the origin. (c) Let F(x, y) = xy. Then we have Fx(x, y) = y and Fy(x, y) = x. Plugging these expres- sions into the ﬁnal result of part (a) gives x dx − y dy = 0. To ﬁnd the orthogonal trajectories, we must solve this diﬀerential equation. To this end, note that this equation is separable and thus x dx = y dy ⇒ x2 2 = y2 2 + C ⇒ x2 − y2 = k , where k := 2C. Therefore, the orthogonal trajectories are hyperbolas. 33. We use notations and results of Problem 32, that is, for a family of curves given by F(x, y) = k, the orthogonal trajectories satisfy the diﬀerential equation ∂F(x, y) ∂y dx − ∂F(x, y) ∂x dy = 0. (2.16) (a) In this problem, F(x, y) = 2x2 + y2 and the equation (2.16) becomes ∂(2x2 + y2 ) ∂y dx − ∂(2x2 + y2 ) ∂x dy = 0 ⇒ 2y dx − 4x dy = 0. (2.17) Separating variables and integrating yield 2y dx = 4x dy ⇒ dx x = 2dy y ⇒ dx x = 2dy y ⇒ ln |x| = 2 ln |y| + c1 ⇒ eln |x| = e2 ln |y|+c1 ⇒ |x| = ec1 |y|2 = c2y2 ⇒ x = ±c2y2 = cy2 , 70