14 CHAPTER I First-Order Diferential Equations EE Posies In each of Problems 1 through 6, determine whether ‘9(2) is a solution of the differential equation. C is ‘constant wherever it appears. 2ytet 2 ox) for x>0; p(x) Ps In each of Problems 7 through 16, determine if the df= ferential equation is separable. If itis, find the general solution (perhaps implicitly defined) and also any singu- Jar solutions the equation might have. IF tis not separable, do not attempt a solution. 7. xsin(y)y’=e0s) 8. [cost +y) +sin(x —y)]y’ =cés(2) IL. 12 y 13. 14 15, 16. problem. 1. 18. 19, 20. 2. xy'y'eythiyGe)=2 22. An object having a temperature of 90° Fahrenheit is placed in an environment kept at 60°. Ten minutes later the object has cooled to 88", What will be the temperature ofthe object aftr it has been inthis envi- ronment for 20 minutes? How long wil it take forthe Object to ool to 65°? Evaluate egy Hint: Lot 1) [ ee Calculate 1"(x) and find a differential equation for T(x). Use the standard integral /—"e-" dt = ft /2 to determine 1(0), and use this initial condition to solve for I(x). Finally, evaluate 1(3) (Draining a Hot Tub) Consider a cylindsical hot tub witha 5-foot radius and a height of 4 feet placed on ‘one ofits circular ends, Water is draining from the tub through a circular hole 5/8 inches in diameter in the base ofthe tub, (@) With k=06, determine the rate at which the pth of the water is changing. Here itis useful towrite ahah dv_avjde a av de avai (©) Cateuate the time required to drain the hot tub AF iis nally fu. Hin: One way to do ths is to wtite at , ai (©) Determine how much longer takes to drain the lower half than the upper half of the tub. Hin: ‘Use the integral of pat (2) with diferent limits foreach hal. Calculate the time required to empty the hemispheti- cal tank of Example 1.7 if the tak is inverted to Tie ‘on a fa cap across the open part of the Hemisphere “The drain hole is in this cap. Take k= 0.8 as in the example. Determine the time i takes to drain aspherical tank with a rads of 18 fee ft is inially Fall of water, hich dans through a circular hole with radius of 3 inches inthe bottom ofthe tank, Use k= 08 The halFlife of Uranium-238 is approximately 45(10") years. How much ofa 10 kilogram block of UV ~238 willbe present one billion year from now’? Given that 12 grams ofa radioactive element decays {0 91 grams in minutes, what isthe halite ofthis element? dh . A thermometers carried outside a house whose ambi- ent temperature is 70” Fahrenheit. After five minutes, the thermometer reads 60", and fifteen minutes after30, 31. 32, 33 this, it reads 50.4°. What is the outside temperature (hich is assumed to be constant)? ‘A radioactive element has a half-life of In(2) weeks TE2 toas ate present at 2 given time, how much will be left dhree weeks later? ‘A tank shaped like a right circular cone, vertex down, is 9 feet high and has a diameter of 8 feet. Is initially full of water (@) Determine the time required to drain the tank through a cicalar hole witha diameter of 2 inches atthe vertex. Take k= 0.6, () Determine the time it takes to drain the tank itis inverted and the dain hole is ofthe same size and shape asin (a, but now located in the new (fat) base, Determine the rate of change of the depth of water in the tank of Problem 31 (vertex at the bottom) if the drain hole is located in the side of the cone 2 feet above the bottom of the tank. What is the rate of change in the depth ofthe water when the drain hole is located in the bottom of the tank? Is it possible to determine the location of the drain hoe if we are told the rate of change of the depth and the depth of the ‘water in the tank? Can this be done without knowing the size of te drain opening? (Logistic Model of Population Growth) In 1837, the Dutch biologist Verhulst developed a differential ‘equation to model changes in a population (he was studying fish populations in the Adriatic Sea). Ver~ hulst reasoned thatthe rate of ehange of a population LL Terminology and Separable Equations 15, 34 ‘P() with respect to time should be influenced by frowth factors (for example, current population) and also factors tending to retard the population (such as limitations on food and space). He formed a model by assuming that growth factors can be incorporated into a term a P(t) and retarding factors into a term bP (0) witha and bas postive constants whose val- ues depend onthe particular population. This led to his logistic equation PQ =aP(o) —bP 0 [Note that, when b = 0, this is the exponential model Solve the logistic model, subject to the initial condition P(0) = po to obtain apo Bea + base “This is the logistic model of population growth. Show that, unlike exponential growth, the logistic mode] ‘produces a population function P(t) that is bounded above and increases asymptotically toward a/b as 1+ 00. Thus a logistic model produces a population function that never grows beyond a certain valu. Continuing Problem 33, a 1920 study by Pearl and Reed (appearing in the Proceedings of the National ‘Academy of Sciences) suggested the values a =0,03134, b= (1.5887)10-” for the population of the United States. Table 1.1 gives the census data for the United States in ten year MRE est for Poon 33d 34, Seton 1 ‘Year Population Po 1790 3929;214 1800 Y 1810 1820 1830 1840) 1850 -23191,876 186031483321 17038588371 1880 30,189,209 190 @2879.766 1900 76212,168, 1910 92228,496 1920 106021,537 1930 123,202,624 1M0 132,164,568 1950151325798 1960 199323,175 197 203,302,031, 1980 226567,042 Percent error Percent error ao ee eeeeeeeeeSFSFSFF16 CHAPTER 1 First-Order Differential Equations increments from 1790 through 1980. Taking 1790 as ‘year zero to determine po, show thatthe logistic model forthe United States population is 123, 141.5668 aos 103072 -oo006z<0=m%° Calculate P(r) in ten year increments from 1790 to fil in the P(#) column in the table. Remember that (with 1790 asthe base year) 1800 is year = 10 inthe model, 1810 is ¢=20, and so on. Also, calculate the percentage error in the model and ill inthis column, Plot the ceasus figures and the numbers predicted by the logistic model onthe same set of axes. You should jobserve that the model is fairly accurate for a long. period of time, then diverges from the actual census ‘numbers. Show that the limit of the population in this PC 12 Linear Equations ‘A first-order differential equatio ‘model is about 197, 300,000, which the United States actually exceeded in 1970, ‘Sometimes an exponential model 1) =kQ(0) is used for population growth. Use the census dat (again with 1790 as year zero) to solve for Q(0), ‘Compute Q(t) for the years of the census data and the percentage error in this exponential prediction of population, Plot the census data and the exponential ‘model predicted data on the same set of axes. It should be clear that Q(0) diverges rapidly from the actual census figures. Exponential models are useful for very simple populations (such as bacteria in a dish) but are not sophisticated enough for humen or (in gen- ral) animal populations, despite occasional claims by experts that the population of the world is increasing exponentially, is linear if it has the form ¥ + P@y=ae) for some functions p and g. ‘There is a general approach to solving linear equation. Let, and notice that g@)= Je #1) = pixel = plxyg). a3) Now multiply y+ p(x)y=9(x) by g(x) to obtain 800)" + PO)E@y =a). In view of equation (1.3), this is ay’ +2'@y =4@)g@0. Now we see the point to multiplying the differential equation by g(x). The left side of the new ‘equation is the derivative of g(x)y. The differential equation has become a qe SOn=awa. which we can integrate to obtain sQ)y= f g@)g@dxte. If g(x) £0, we can solve this equation for y: ye) 1 ah, [acouer+ cae 8G)”20 CHAPTER 1 First-Order Differential Equations settings. For example, the current in a circuit is often written as a sum of a steady-state term and a transient term, ‘The initial ratio of salt to brine in the tank is 100 pounds per 200 gallons or 1/2 pound per gallon, Since the mixture pumped in has a constant ratio of 1/8 pound per gallon, we expect the brine mixture to dilute toward the incoming ratio with a terminal amount of salt in the tank of 1/8 pound per gallon times 200 gallons. This leads to the expectation (in the long term) that the amount of salt inthe tank should approach 25, as the model verifies. EERONEN Peoszeus Ineach of Problems I through 5, find the general solution, In each of Problems 6 through 10, solve the initial value problem. 10, y'-+3y=Se" —6; y()=2 11. Two tanks are connected a in Figure 1.6, Tank 1 initially contains 20 pounds of salt dissolved in 100, gallons of brine. Tank 2 initially contains 150 gallons ‘of brine in which 90 pounds of salt are dissolved, At 5 gain; 121 time zero, a brine solution containing 1/2 pound of salt per gallon is added to tank atthe rate ofS gallons per minute. Tank 1 has an output that discharges brine info tank 2 a the ate ofS gallons per minute, and tank 2 also has an output of 5 gallons per minute. Deter- ‘mine the amount of salt in each tank at any time. Also, determine when the concentration of salt in tank 2 is @ ‘minimum and how much salts inthe tank at that ime. Hint: Solve for the amount of salt in tank 1 at time + and use this solution to help determine the amount in tank 2 12, A 500 gallon tank initially contains 50 gallons of brine solution in which 28 pounds of salt have been dissolved. Beginning at time zero, brine containing 2 pounds of salt per gallon is added at the rate of 3 gal- Tons per minute, and the mixture is poured out ofthe tank tthe rate of 2 gallons per minute. How much salt is in the tank when it contains 100 gallons of brine? Hint: The amount of brine in the tank at time ¢ is 50-0 13, Findall functions with the property thatthe y intercept ofthe tangent tothe graph at (x,y) is 2 FIGURE 1.6 Storage rank in Problem 11, Section 1.2.15 Bxact Equations 25 ‘The “constant of integration” is h(x), because x is held fixed in a partial derivative with respect, to y. Now we know g(r, y) to within h(x). Next we need $8 cos(x) -20y try +h). ‘This requires that HQ) =cos(x), so h(x) in(x). A potential function is (x,y) ‘The general solution is implicitly defined by e—aty +sin(x), ef xy-+sin@a) =e. For the initial condition, choose c so that y(1)=4. We need 4+sin(1) ‘The solution of the initial value problem is implicitly defined by +sin(x)=e!—4+sin(1). PROBLEMS In each of Problems 1 through 5, test the differential equa- tion for exactness. If itis exact (on some region of the plane), find a potential function and the general solution (periapsimplicidy defined), If is not exact anywhere, do not attempt a solution, Irty+Gy t0y=' 2eos(x +y) ~2x singx + y) 2x sinks +»); DvP ye + Gay x07 + 2y)y/ dry 42e $22 43))y'=0 4uy + 2ey 4 Qe 43y?py'=0 Jn each of Problems 6 and 7, determine aso thatthe equation is exact Obtain the general solution ofthe exact equation 6. 38 bay ety = 1. day’ =3y— Grtar'y?—2ay)y'=0 In each of Problems 8 through 12, determine ifthe differ. ential equation is exact in some rectangle containing the point where the initial condition is given. If it is exact, solve the initial value problem. If not, do not attempt a solution 8 LHe Ler ery! 0; x) =—5 9, xc0s(2y —x) ~ sin(2y —x) ~ 2x cos(2y —s)y! yor/12)=x/8 1B. 4 1s. Show that ey tysy isnot exact. Solve this equation by finding an integrat- ing factor of the form (x, y) =x*y?. Hint: Consider the differential equation werty basy ay? and solve for and b so that dhe resulting equation is Try the strategy of Problem 13 on the differential equation 2y' xy + Gay 6°) Let be potential function for M-+ Ny’ =0. Show that +e is also a potential function for any con- stant c. How does the general solution obtained using 6 differ from that obtained using g +c? If Af-+ Ny’ =0 is not exact, it might be possible to find a nonzero function j(x,y) such that uM + uNy’ = 0 is exact. The benefit to this is that M+ Ny’ =0 and p(M + Ny’) =0 have the same solutions if (x.y) #0 for any x and6 CHAPTER 1 First-Order Differential Equations 4 and the Tater equation is exact (hence is solv able if we can find a potential function). Such function (x,y) is ealled an integrating factor for MeNy’ (@) Show that 6(x,y) =xy7 is also an integrating factor on any rectangle where x #0 and y #0. Use this integrating factor to find the general solution, 16. (2) Show that y—xy’=0iisnotexacton any rectangle (e) Write the ditferential equation as inthe plane, () Show that (x, is an integrating fac- i tor on any rectangle over which x #0. Use this yozy=0 {o find the general solution of the differential : ‘equation and solve it asa linear differential equation. (©, How do the solutions found in parts (b) through (@) diffe from each other? (©) Show that v(x, y) ="? is also an integrating fac toron any rectangle where y #0, and use this to solve the differential equation Integrating Factors “Most” differential equations are not exact on any rectangle. But sometimes we can multiply the differential equation by 2 nonzero function (x, y) to obtain an exact equation. Here is an example that suggests why this might be useful. EXAMPLE 1.15 The equation yP Gry + Gry -6r°)y'=0 an {snot exact on any rectangle. Multiply it by (x. y) = to get, yxy? + Bry? — 6x7y)y" =O. (1.8) ‘Wherever y 0, equations (1.7) and (1.8) have the same solution. The reason for this is that ‘equation (1.8) is just y[y? -6xy + Gry — 6x*)y']=0, and ify £0, then necessarily y* — 6xy + Gry —6x")y’ =0. [Now notice that equation (1.8) is exact (over the entire plane), having potential function (sy =ay? —3x7y* ‘Thus the general solution of equation (1.8) is defined implicitly by ay 38 =C, and, wherever y #0, this defines the general solution of equation (1.7) as well. # ‘To review what has just occurred, we began with a nonexact differential equation. We multi- plied it by a function jz chosen so that the new equation was exact. We solved this exact equation, then found that this solution also worked for the original, nonexact equation. The function 4. therefore enabled us to solve a nonexact equation by solving an exact one. This idea is worth pursuing, and we begin by giving a name to 1CHAPTER 1 First-Order Diflerential Equations 1.4.1 Separable Equations and Integrating Factors ‘We will point out a connection between separable equations and separable equations. ‘The separable equation y’ = A(x) B(y) isin general not exact. To see this, write it as, A@B)—y'=0. so in the present context we have M(x, y) = A(x)B(y) and N(x, y) =—1. Now a a 7 Fe CVHO and 5 LAG)BON=AG)B'O), and in general A(x)B'(y) #0. However, 1(y) = 1/B(y) is an integrating factor for the separable equation. If we multiply the differential equation by 1/B(y), we get 1 AW) ~ fan exact equation because a: [se ax [BOD (0) [A@)1=0. ay ‘The act of separating the variables is the same as multiplying by the integrating factor 1/(y).. 1.4.2. Linear Equations and Integrating Factors Consider the linear equation ‘+ p(x)y = q(x). We can write this as [p(x)y —q(x)] + in the present context, M(x, y) a art af so the linear equation is not exact unless p(x) is identically zero. However, 1x, and. 0, s0 (x) g(a) and N(x, y) = 1. Now a iy (PCy = 968] = PUD, elvinee ig ‘an integrating factor. Upon multiplying the linear equation by j., we get [pay —qlxylel 4 ef mae is is exact because a 2 esoiner ox PROBLEMS 1. Determine a test involving M and N to tell when M+ Ny’ =O has an integrating factor that is a funtion of y only. 2. Determine a test to determine when M +N} integrating factor of the form s(x, y) constants @ and b Ohas an for some 3. Consider y— ay’ (@) Show that this equation is not exact on any rectan- ale. () Find an integrating factor (x) that isa function of alone. pore! =0, [tee@y -gcne!"] (©) Find an integrating factor v(y) that is @ function of yy alone (@) Show that there is also an integrating factor ‘Gs, y)=s"y? for some constants @ and b. Find all such integrating factors15 Homogeneous, Bernoulli, and Riccati Equations 31. Ineach of Problems 4 through 12, (a) show that the differ- ential equation isnot exact, (6) find an integrating factor, (€) find the general solution (perhaps implicitly defined), and (A) determine any singular solutions the differential ‘equation might have. 4 S.14Gx-e™y 6, 6xty + 12ay +9? + (6x +2y)9/=0 7, Axy+6)+ QP + 6xy)y"=0 8. yty-sy 9, day? bey HOP yay’ 0. 2y?~9xy + Gay = 6 Y xy) 1 yyy! Hint ty nla.) 12, By bay sy? Hine ty 4OR 9) 0 (efits try Hox, y= In each of Problems 13 through 20, find an integrating factor, use it to find the general solution of the differen~ tial equation, and then obtain the solution of the initial value problem. B. M4 1s. 16. 17. 18. 19. 20. 3x°y by +287" 2. 20° —2) +3xy°y" y(L-+2)42ay =05 9) =6 day +3y'=0:90) 2yhbx) tay sin(x—y) +eas(x —y) —cos(x —y)y"=0; 0) = 78/6 ©) ‘Show that any nonzero constant multiple of an inte- trating factor for M + Ny =0 is also an integrating factor. Let u(x, ») be an integrating factor for M 4+ Ny’ and suppose that the general solution is defined by ‘oC, 9) =C. Show that (x, y)G¢@Cx, 99) is also an integrating factor, for any differentiable function G of one variable. Homogeneous, Bernoulli, and Riccati Equations “We will discuss three other types of first-order differential equations for which techniques of solution are available. 1.5.1 The Homogeneous Differential Equation A homogeneous differential equation is one of the form y= FO) with y’ isolated on one side and on the other an expression in which x and y always occur in the combination y/x. Examples are y” in(y/x) —x/y and y'=x?/y". In some instances, a differential equation can be manipulated into homogenous form. For example, with, » aty we can divide numerator and denominator on the right by x to obtain the homogeneous equation yi T+y/x ‘This manipulation requires the assumption that x #0.16 Additional Applications 35 or eQ)= ‘The general solution of the Riccati equation is. 1+ 1/3 +e)" riscnvewitns y@)= Tos + BEGUM reosieus In each of Problems 1 though 14, fn he pros ssh a. yp Zym yt (esis in which a,b, ¢,d, p, and r are constants. Show that this equation is homogeneous if and only ife=r =0. Thus, suppose at least one of c and r is not zero. ‘Then this differential equation is called nearly homo. ‘geneous. Show that if ap — bd #0 it is possible to choose constants f and k such that the transforma- tion x= X +H, y=Y +4 results in a homogeneous equation In each of Problems 16 through 19, use the idea from Problem 15 to solve the differential equation ‘Additional Applications ‘This section is devoted to some additional applications of first-order differential equations. We will need Newton's second law of motion, which states that the sum of the external forces acting44° CHAPTER 1 First-Order Diferential Equations 64, 000/3. Therefore, + 64 1000 vor Sf 22] ‘The chain leaves the support when x =40, so at this time, 1_ 1000" 49 — 1000 3 [: 0 600. (210). ‘The velocity t this ime is = 2270, which is about 29 fet per second. PROBLEMS 1. Ina conta elecuomatve fore RL cic, we find she opes the parachute, producing a rg equ thatthe cents given by tod tines the square of te vlc, Deter le : velocity and how fate sydver a fallen a ine. i Zire) +e a ‘What is the terminal velocity? 6. A 48 pound box is given an inal push of 16 feet Leti(0) per second down an inclined plane that hes © gradi- (@) Show that the cument increases with time. ent of 7/24. I there is coefficient of friction of 1/3 ‘between the box andthe plane and a force of air esis- tance equal in magnitude © 3/2 ofthe velocity ofthe box, determine how far the box wil travel down dhe plane before coming to rest. (©) Does the inductive time constant depend on i(0)? ‘Suppose the box in Problem 4 cracks open upon hit- so in what way? ting the botiom ofthe lke, and 32 pound of ts con- tents spill out. Approximate the velocity with which the box surfaces. () Find 2 time fp at which the current is 63 percent ‘of E/R. This time is called the inductive time ‘constant of the citcuit. ‘The acceleration due to gravity inside the earth is pro- portional to the distance from the center ofthe earth ‘An object is dropped from the surface ofthe earth into Determine the curents inthe circuit of Figure 1.13. 8 hole extending straight through the planets center. In the circuit of Figure 1.14, the capacitors initially Calculate the speed the object achieves by the time it discharged. How long after the switch is closed will reaches the center 3. A particle stats from rest at the highest point of a Vertical circle and slides under only the influence of woo ‘gravity along a chord to another point on the circle. ‘Show thatthe time taken is independent of the choice of the terminal point. What is this common time? 4. Archimedes’ principle of buoyancy states that an object submerged in a fluid is buoyed up by a force equal to the weight of the fluid that is displaced by the object. A rectangular box of 1 x 2.x 3 feet and ‘weighing 384 pounds is dropped into a 100-foot deep freshwater lake, The box begins to sink with a drag {due to the water having a magnitude equal to 1/2 the velocity. Calculate the terminal velocity of the box. Will the box have achieved a velocity of 10 feet per second by the time it reaches the bottom? Assume that FIGURE 1.13 Circuit of Problem 8, Section 1.6 the density of water is 62.5 pounds per cubic Foo. 5. A skydiver and her equipment together weigh 192 the capacitor voltage be 76 volts? Determine the eur- ‘pounds. Before the parachute is opened, there is an air rent in the resistor at that time. The resistances are drag force equal in magnitude to six times her veloc- in thousands of ohms, and the capacitor is in micro- ity. Four seconds after stepping from the plane, the farads (10°* ferads) 150 soa)FIGURE 1.18. Circuit of Problem 9, Section 1.6. 10, For the circuit in Figure 1.15, find all currents imme- iately after the switch is closed, assuming that all of these currents and the charges on the capaci tors are zero just prior to closing the switch. Resis- tances are in ohms, the capacitor in farads, and the {inductor in henrys. wa 30a sa — mor 410% FIGURE 1.15 Circuit of Problem 10, Section 16. 1. A 10-pound ballast beg is dropped from a hot air bal- Joon which is at an altitude of 342 feet and ascending at 4 feet per second. Assuming that air resistance is not a factor, determine the maximum height reached by the bag, how long it remains aloft and the speed with which i eventually trikes the ground. 12. A man stands at the junction of two perpendicular roads, and his dog is watching him from one of the roads at a distance A feet away. At some time, the rman stars to walk with constant speed v along the other road, and at the same time, the dog begins 10 run toward the man with a speed of 2v. Determine the path the dog will tke, assuming that it always moves so tha tis facing the man. Also determine when the dog will eventually catch the man. In each of Problems 13 tough 17, find the family of orthogonal trajectories of the given family of curves. If software is available, graph some curves of both families. 13. yoke +1 14. x 42y 15. 16. 1. 18. 19. 20, aL 16 Additional Applications 45 yee" 2 42y 2x? —3y=k Recall thet the charge q(t in an RC circuit satisfies the linear ciffereatal equation 11 a+ Bets REO: (@) Solve for the charge in the case that £0) Which is constant. Evaluate the constant of integration in this solution process by using the condition (0) (©) Determine lim... (0), and show that this limit is. independent of gy. (©) Determine at what time (1) is within I percent of its steady-state value the limiting value requested in par (6) A 24 foot chain weighing p pounds per foot is stretched out on a very tall ftictionless table with 6 feet hanging off the edge. Ifthe chain is released from rest, determine the time it takes for the end of the chain to fal off the table and also the velocity of the chain at this instant ‘Suppose the chain in Problem 19 is placed on a table that is only 4 feet high, so that the chain accumulates ‘nthe floor as it slides ff the table. Two feet of chain ae already piled up on the floor atthe time thatthe ‘est of the chain is released, Determine the velocity of the moving end of the chain a the instant it leaves the table top. Hint: Newton's law applies to the center of ‘mass ofthe moving system, ‘A bug is located at each comer of a square table of side length a. Ata given time, the bugs begin moving at constant speed v with each pursuing the neighbor ‘0 the right (2) Determine the curve of pursuit of each bug. Hint: Use polar coordinates with the origin atthe center of the table and the polar axis containing one of the corners. When a bugis at (f() 0), its target is at (F(0,0-+ /2)). Use the chain rule to waite dy _dy/do ax” dx/d6 where (6) = F(O)sin(O) (©) Determine the distance traveled by each bug. (©) Does any bug actually reach its quarry? ‘A bug steps onto the edge of a disk of radius a that is spinning at a constant angular speed of w. The ‘bug moves toward the center of the disk st constant speed v,46 CHAPTER 1 First-Order Differential Equations (2) Derive a differential equation forthe path of the in terms of the angular speed and radius of the ‘bug using polar coordinates dist. (©) How many revolutions will the disk make before (©) Referring to pat (b), what isthe total distance the the bug reaches the comer? (The solution will be ‘bag will travel, taking into account the motion of the disk? 17 Existence and Uniqueness Questions There are initial value problems having no solution. One example is yf =2YF: (0) =—1. The differential equation has general solution y = (x +c), but there is no real number ¢ such that y(O)==1. ‘An initial value problem also may have more than one solution. In particular, the initial value problem VF YY =O has the zero solution y = (x) =0 forall x. But it also has the solution forx <1 1 forx> 1 Because existence and/or uniqueness can fail for even apparently simple initial value prob- Jems, we look for conditions that are sufficient to guarantee both existence and uniqueness of a solution, Here is one such result ° yaV@=]_ — THEOREM 12 Existence and Uniqueness Let f(x, y) and af/2y be continuous for all (x,y) in a rectangle R centered at (19, y9). Then there is a positive number t such that the intial value problem Y= foi ye) = 30 has a unique solution defined at least for xy—h 0 find two solutions (&) Explain why the conclusion of part (8) does not ‘violate Theorem 1.2 “Theorem 1.2 can be proved using Picard iterates. Here is the idea. Consider the inital value problem y For each positive integer n, define (eye WE paedare [seaside ‘This is recursive defizition, giving yy(x) in terms of yo, then y(t) in terms of y;(x), and so on. The functions yl) are called Picard iterates forthe initial value prob- Jem. Under the assurmptions of the theorem, the sequence fof functions yq(x) converges for all x in some interval bout xp, and the limit ofthis sequence is the solution of the intial value problem on this interval. Inach of Problems 6 trough 9: (@) Use Theorem 1.2 t0 show that the problem has a solution in some interval about x. () Find this solution () Compote Picard iterates ys (x) through ys(x), and from these, guess y,(x) in general (@) Find the Taylor series of the solution from part (b) about x. ‘You should find that the iterates computed in part (¢) are exactly the partial sums of the series solution of part (@). Conclude that in these examples the Picard iter fates converge 0 an infinite series representation of the solution56 MEE Peoetens CHAPTER 2. Linear Second-Order Equations In each of Problems I through 5, verify that y) and y, are solutions of the homogeneous differential equation, caleu- late the Wronskian of these solutions, write the general solution, and solve the initial value problem. i 1. y'—2y'+29=0; 90) =6,¥0) (3) = cos(s), Cx) =e" sins) 2 y'-6y + By =0; 0) =—1, YO) yyCa) =e e0s(23), ya 4, y*=16y=0; (0) = 12, Q)=3 la) =e 2G) se 5. y+ 36y=0; y(0)=~5,¥O)=2 s(x) =sin(Gx), y() = cos(6x) In Problems 6 through 10, use the results of Problems 1 through 5, respectively, and the given particular solution, Y, to write the general solution of the nonhomogeneous equation 6. 9" 2y' $2y= Sx Yo) Yo) 8. 439 $2y= 15: Yy(0) = 15/2 9. y= Mey mss Yon) = 3/44 1/2 10. "436 =2~1LY,@)=¢ ~ D/36 11, Let g be a solution of y"-+ py'-+gy=0 on an open imterval /. Suppose p(x) =0 for some x inthis inter- val, Suppose (x) is not identically zero. Prove that (40) #0. 12, Suppose y\ and ys are solutions of equation (2.2) on (a,b) and that p and q are continuous. Suppose yi, and 3 both have a relative extremum at some point between @ and b. Show that yy and y> are linearly dependent. 13, Show that yy(x)=x and yo(x) =x" are linearly inde- pendent solutions of x*y* —2xy‘ +2y=0 on (I, 1) but that W(0) = 0. Why does this not contradict ‘Theorem 2.3 conclusion (1)? Sx/2—Sx-4 1. y-6y +13 Bet 16. 22 Reduction of Order Given y" + p@y’ +q@y Let y1 and yz be distinct solutions of equation (2.2) fon an open interval 7. Let x be in I, and suppose ‘yC%) = 39%) =, Prove that yy and ys are linearly “dependent on 7. Thus, linearly independent solutions cannot share a common 2er0, Here isa sketch ofa proof of Theorem 2.3. Fill in the etails. Denote W(y,,y:) = W for convenience. For conclusion (1), use the fact that y, and y; are solutions of equation (22) to write vet Pv Fa) weet: ‘Multiply the first equation by y» and the second by —y and add. Use the resulting equation to show that w+ pw. Solve this linear equation to verify the conclusion of part (1). To prove conclusion (2), show first that, if yeCe) = byy(x) for allx in 7, then W(x) =0. A sim- ilar conclusion holds if ys(+) = ky3(x). Ths, linear dependence implies vanishing ofthe Wronskian. Conversely, suppose W(x) = 0 on J. Suppose first that y9(2) does not vanish on I. Differentiate vile fo show that i (2)=-we fon I, This means that ys/ys has a zero derivative on Thence y,/2=c, 50 y= cys and these solutions are linearly dependent. technical argument, which we omit, covers the case that s(x) can vanish at points of Let y\(x) =x? and ya(x)=2°. Show that W(x) Now W(0) =0, but W(x) #0 if x #0. Why does this not violate Theorem 2.3 conclusion (1)? ), we want two independent solutions. Reduction of order is a technique for finding a second solution, if we can somehow produce a frst solution.22 Reduction of Order 59 so u=In@) +4, and we choose d=0 because we need only one suitable u. Then ys(x) = x?ln(x) is a second solution, Further, for x > 0, wa=(* P) ogo Tae deiner) +x) ‘Then x? and x?In(x) form a fundamental set of solutions for x > 0. The general solution is for x>0is In each of Problems 1 through 10, verify that the given function is a solution of the differential equation, find a second solution by reduction of order, and finally write the general solution, 0: 2) 3, y"— 10y' +259 ay! Tay + 16y= ay! ~3ry'+4y=0, 90) Ox by" Ary +4y=0% 10) forx>0 y= Bayt dry =O) = vty + (ak) y=) (x4 3x4 I)y" + 2x7 = 2y =0; 1G) interval not containing —1 or —! 4 5. 6 1. yi—ty'=Ay=0: 1) 8 9. -eostx) for x> 0 0. on any 11, Verify that, for any nonzero constant a, (x) =e" isa solution of y" + 2ay’ + ay =0. Write the general solution, 12. A second-order equation F(x, y. ',7")=0 in which y is not explicitly present can sometimes be solved by puiting w= y’. This results in a firs-order equation G(x,u,w’) = 0. If this can be solved for u(x), then yiCa) = fu(x)dx is a solution of the given second order equation. Use this method to find one solution, then find a second solution, and finally the general solution ofthe following (@) xy"'=24y' (b) ay" #29 8. 18. © 1-y=4y" © ¥ +07 © y=1407 {A second-order equation in which x does not explic~ itly appear can sometimes be solved by putting w= and thinking of y asthe independent variable and w as afunetion of y. Write ¢ fas) a y-al3ht tecanen £7) = a fe va oon oinnans Sets en Ho) Bo gt ry eno ire teed sie Gets ney Sac tears ay du dy _ du ay ax ay @ 4307 ©) yy" +O+D0F © yay + OF @ y'=14+0/F © y+07 Consider y’ + Ay'-+ By =0, in which A and B are con- stants and A? — 4B =0, Show that y(x)=e"*? is one solution, and use reduction of order to find the second solution yx) =e? Consider y" + (A/x)y' + (B/2°)y =0 for x >0, with ‘A and B constants such that (A — 1)? 4B =0. Verify that y\(x) = x4 is one solution, and use reduc- tion of order to derive the second solution y2(x nie23 The Constant Coefficient Case 63 Let A; and A, be the roots of the characteristic equation Then: 1, If; and Ay are real and distinct, 2. Wada, Veah+b ie teed yO) Scie tenet, 3, Ifthe roots are complex «+ if, 3) PROBLEMS In each of Problems 1 through 10, write the general solution. 1. y"43y 418) 2. y"46y'—40y = 3. y"4 10y' +26y=0 4. y'+16y +64y=0 5. y'= My +49y=0 6. y= 6y-+7y=0 1. 8 o Y#6y'+9y=0 9. y’-y'-6 10, y*~2y'+10y In each of Problems 11 through 20, solve the initial value problem, IL, y'by'=12y=0;y2)=2, 2 12. y'—4y+4y=05 910) =3, 70) 1B. y-2y+ 4 15, 16 1", 18. y"-2y ~Sy=05 0) =0, (0) = 19. y"43y'=0; ¥(0)=3, (0) =6 20. y"+2y—3y=0; 900) =6, ¥'0) 21, Suppose gis a solution of ¥ tay’ +by=0; ya) = AY) =B = ce" cos(Bx) +ex¢" sin( Bx). with «,b,A, and B as given numbers and a and b postive, Show that lim gta 22. Use power series expansions to derive Euler's for- mula, Hints Wete e-Die. sin(x) and 0s (8 with B real, and use the fact that 7 Dv 1)" and "4 for every positive integer n 23, This problem illustrates how small changes in the coefficients of a differential equation may cause dra- ‘matic changes inthe solution, (2) Find the general solution (2) of y'~2ay' +a%y=0 with a #0. (b) Find the general solution , (x) of y'=2ay'+@"- ey with ¢ a positive constant. (©) Show that, as € > O, the solution in part (b) does ‘not approach the solution in part (a) even though the differential equation in part (b) would appear to more closely resemble that of part (a) s« is chosen smaller.“4 CHAPTER 2. Linear Second-Order Equations 24, (a) Find the solution y ofthe initial value problem (©) Find the solution y, of YOxd y'-2ay +@ ey =0; 10) =e, ¥O=d. (©) Isit true that YeCe) > Ya) ase 07 The Nonhomogeneous Equation From Theorem 2.4, the keys to solving the nonhomogeneous linear equation (2.1) are to find two linearly independent solutions of the associated homogeneous equation and a particular solution Y, for the nonhomogeneous equation. We can perform the first task atleast when the coefficients are constant, We will now focus on finding ¥,, considering two methods for doing this. 2.4.1 Variation of Parameters Suppose we know two linearly independent solutions y, and ys of the associated homogeneous ‘equation. One strategy for finding Y, is called the method of variation of parameters. L.0ok for functions u; and w; $0 that ¥, (x) Sus (x) (x) + ua(2)y2@). ‘To see how to choose u; and #2, substitute ¥, ino the differential equation, We must compute two derivatives. First, Ypsuny) bays bay tat. Simplify this derivative by imposing the condition that (2.8) iy ty MY, aYs so Yo muy, tulys + uyl tut Substitute ¥, into the differential equation to get wi bay bay tay + PONay, Hays) +4 Ce) uays Haye Fx) Rearrange terms to write wibl + Pedy, +a@y] tualys + Pedy, +9@ys Hayy tuys = FO). ‘The two terms in square brackets are zero, because y, and y: are solutions of y’ + p(x)y' + q(2)y =0. The last equation therefore reduces to ayy bugys= FO). 29) Equations (2.8) and (2.9) can be solved for wand w; to get WOOF) og yg = MOLE) way Mt BO Wey u\@) (2.10)n CHAPTER 2 Linear Second-Order Equations Using this, the general solution is ye cost) Hersinea+ Flee"). # BE rropiems Ineach of Problems | through 6 find the general solution, AL. using the method of variation of parameters fora particular 49 r—2y 4 10y=200 4 2x8 Ay +5y=21e sohtion 13, yy 48 —30" 1. y49y= 1296632) Pi pac 2. y= 2y'—3y—2sin's) eee ee eee 3, y'—3y'+2y=c0s"e-?) eee: 4. yf -5y'46y= Bsn) In cach of Problems 17 txough 24, solve the ntl value 5. problem. ° pages 11 y48y + 1y=e*+7. yO) =LYO=O In cach of Problems 7 through 16, nd the general solu. 18: y" Fy =tanGx): yO) =4, ¥() tio, using the metiod of undetemined coctcens for © 19, °—2y/ By 0e"™+ 80 y(0)—=1ey O=e particular solution. 7. y"=3y-42y=10sin(x) 21. y= y=Ssin?hs 1 =2.7O= 8, y’-2y-ty=3x +25sinGe) 22. 2 sin(x); yO) 9. yay + 13y = 30" — Se 2 10. y’—4y'= Br? 420 24, 2.5 2. y= y'4y=hy(D=4,y(=-2 Spring Motion A spring suspended vertically and allowed to come to rest has a natural lengih-L. An object, (bob) of mass m is attached at the lower end, pulling the spring d units past its natural length. ‘The bob comes to rest in its equilibrium position and is then displaced vertically a distance yo units (Figure 2.3) and released from rest or with some initial velocity. We want to construct 2 mode! allowing us to analyze the motion of the bob. Let y(t) be the displacement of the bob from the equilibrium position at time r, and take this equilibrium position to be y= 0. Down is chosen as the positive direction. Now consider the forces acting on the bob. Gravity pulls it downward with a force of magnitude mg. By Hooke’s law, the spring exerts a force ky on the object. k is the spring constant, which is a number ‘quantifying the “stiffness” of the spring. At the equilibrium position, the force of the spring is kd, which is negative because it acts upward. If the object is pulled downward a distance y from this position, an additional force —ky is exerted on it. The total force due to the spring is therefore —kd — ky. The total force due to gravity and the spring is mg — kd — ky. Since at the equilibrium point this force is zero, then mg = kd. The net force acting on the object due to ‘gravity and the spring is therefore just —ky. There are forces tending to retard or damp out the motion. These include air resistance or perhaps viscosity of a medium in which the object is suspended. A standard assumption (verified by observation) is that the retarding forces have magnitude proportional to the velocity y’. Then80 CHAPTER? Linear Second-Order Equations Since Rag ayy tet 4’, this is a second-order differential equation for the charge: 1 pho. Assuming that the resistance, inductance, and capacitance are constant, this equation is exactly ‘analogous to the spring equation with a driving force, which has the form ce, kd Sey tye TsO. ‘This means that solutions of the spring equation immediately translate into solutions ofthe circuit ‘equation with the following identifications: Displacement function y(t) <=> charge q(¢) Velocity y'(t) <=> current (2) Driving force f(0) <=> electromotive force E(t) ‘Mass m <=> inductance L Damping constant c <=> resistance R Spring modulus k <= reciprocal 1/C of the capacitance. PROBLEMS 1, This problem gauges the relative effects of initial position and velocity on the motion in the unforced, ‘overdamped case. Solve the initial value problems yay +2y=0; yO) =5,9°0) and yf $4y 42y=0; yO)=0,9'0) Graph the solutions on the same set of axes. 2, Repeat the experiment of Problem 1, except now use te eitcally damped, unforced equation an 3. Repeat the experiment of Problem 1 forthe under damped, unforced equation yit2y'+5y=0 Problems 4 Uuough 9 explore the effects of changing the inal position oF inital velocity on the mation of the abject. In each, use the same set of axes to graph the slu- tion of the inital vale problem forthe given values of ‘A and observe the effect that these changes cause in the solution, 4 y"+Ay +4y= 0:90) 1,3,6, 10, —4 and ~7. 5. y*+4y'+4y=05 0) 1,3,6,10,-4 and ~7. 6. y"#2)'+5y=0, 10) = 1,3,6,10,~4 and ~7. 1. y"2)'+5y=0, 10) = 1,3,6,10, ~4 and ~7. 170 Aas values OSA; A has values .¥(0) =0; A has values Y(O)=A; A has values 9. 2. 13. Yi Ay’ +2y = 0; 0) =A, (0) =0; A has values 1,3,6,10, ~4 and ~7. yf $49" +27 =0; 9(0) =0, ¥(O)=A; A bas values 1,3,6,10, 4 and =7, How many times can the mass pass through equi- librium in critical damping? What condition can be placed on (0) to ensure that the mass never passes through the equilibrium point? How does the initial velocity influence whether the mass passes through the equilibrium point? |. Consider overdamped forced motion governed by V+ 6y'42 4c05(3). (@) Find the solution satisfying (0) = (0) Find the solution satistying (0) Graph these solutions on the same set of axes to com- aro the effects of inital displacement and velocity on the motion. Suppose y(0) = y/(0) #0. Determine the maxi- ‘mum displacement of the mass in eritially damped, ‘unforced motion. Show that the time at which this ‘maximum occurs is independent of the initial dis- placement. How many times can the mass pass through the equi- Librium point in overdamped motion? What condition can be placed onthe inital displacement to ensure that it never passes through equilibrium?14, An object having s mass of 1 gram is attached tothe lower end of a spring having a modulus of 29 dynes per centimeter. The mass in tur is adhered toa dash ‘pot that imposes a damping force of 10v dynes, where ‘(0s the velocity ofthe mass at time tin cenimet per second, Determine the motion of the massif itis pulled down 3 centimeters from equilibrium and then struck upward with a blow sufficient o impart veloc- ity of I centimeter per second. Graph the solution Solve the problem when the initial velocity i (in turn) 2,4,7, and 12 centimeters per second. Graph these 26 Euler's Differential Equation 81 15. In underdamped, unforced motion, what effect does the damping constant have on the frequency of the oscillations? 16. Carry out the program of Problem 11 forthe critically damped, forced system governed by yi bay $4y = 409030). 17. Carry out the program of Problem 11 for the under- damped, forced system governed by solutions on the same axes to ‘of the initial velocity on the motion. ualize the influence yi by! +3y = 400830). Euler’s Differential Equation If A and B are constants, the second-order differential equation xy" + Any +By=0 (2s) is called Euler's equation. Euler’s equation is defined on the half-lines x > 0 and x <0. We ‘will find solutions on x > 0, and a simple adjustment will yield solutions on x <0. ‘A change of variables will convert Euler’s equation to a constant coefficient linear second order homogeneous equation, which we can always solve. Let xed or, equivalently, ¢ =In(x). If we substitute x =e" into y(x), we obtain a function of ¢ as Yo=ye). ‘To convert Euler's equation to an equation in r, we need to convert derivatives of y(x) to derivatives of ¥ (2). First, by the chain rule, we have no td y¥@) GOO = VO) ava _ iy, as 37 Next, rato Y@)=F0'@) =i (tro) ratty + xa” o) = yoy toe ~ Brot, dt dx ratty wr" -Y'O).26 Euler's Differential Equation 83 ¥() =ce7 c0s(31) +2 ‘The transformed equation is ¥" + 2Y' + 10Y =0 with the general solution 1). ‘Then 12 cos(3n(x)) + oe sin(3in(x)) 1 y= EXAMPLE 2.25 | Solve x*y" +3xy + 10y= (c\cos(in(x)) +¢sin3In@))). As usual, we solve an initial value problem by finding the general solution of the differential ‘equation and then using the initial conditions to determine the constants. EXAMPLE 2.26 Solve xy’ — Sry! + 10y =0; y(1) =4, y’(I) = 6. ‘The Euler equation transforms to Y" — 6Y’ + 10Y =0 with the general solution Y() =cye* cos(t) +cxe" sin(t) for x > 0. Then (a) = 2° (€,cos(in(x)) + er sin(in(x)) Then y= ‘Thus far, lx) =4x7cos(tnGe)) +624? sin(ln(s)) Compute (a) = 12? cos(In(x)) — 4x" sin(in(s)) + 3ex*sin(In(x)) + ¢2x* e0s(ta(x). Then yQM=2+e=-6, $0 cy = 18, The solution of the initial value problem is y() =4x" cos(In(x)) — 18x? sin(n@a)). # EE PeonLeus Ineach of Problems I through 10, find the general solution. 3. x"y"-425xy/-+ 144y=0. 1. sy"+6xy'+6y=0 4. xy" Bay" + 10y 2. atyr—ay +3520 5. xty"-+xy'—l6y=084 CHAPTER2 Linear Second-Order Equations 6, ay" —Suy+58y=0 1. aty tay +4y=0 r4xy—4y=0 2x) 43x’ +y=0 In each of Problems 11 through 16, solve the initial value problem. AL 12, xy" +25ry’ + 14dy =0; y(1) 13, x¥y"—9ny/-+24y=04 y(D) 14 15. 16. V. 18, 0; 92) =5,¥2) Here is another approach to solving an Euler eque tion. For x > 0, substitute y=" into the differential ‘equation to obtain a quadratic equation for Roots of this quadratic equation yield solutions y=x" Use this approach to solve the Euler equations of Examples 2.23,2.24, and 2.25 ‘Outline « procedure for solving the Euler equation for <0. Hint: Let t= In |x| inthis cas.41 Definition and Notation 87 to open the integral transforms package of subroutines. For the Laplace transform of (enter laplace(£(t),t, s): to obtain F(s). “The Laplace transform is linear, which means that the transform of a sum is the sum of the transforms and that constants factor through the transform: LU +g)=F +G and Clef =cF for all s such that F(3) and G(s) are both defined and for any number ¢ Given F(3), we sometimes need to find f(t) such that £[}= F. This is the reverse process of computing the transform of f, and we refer to itas taking an inverse Laplace transform. ‘This is denoted £-', and LOLF] = f exactly when £if]=F. For example, the inverse Laplace transform of 1/(s—a) is e* If we use Table 3.1 to find an inverse transform, read from the right column to the left column. For example, using the table and the linearity of the Laplace transform, we can read that [3 £ [; +16~ Br 1 3 opi lesen re ‘The inverse Laplace transform C~ is linear because £ is. This means that COUR +GJ=LFI+ LG =f +8, and for any number c, L(eF]=cL"Fl=cf. ‘To use MAPLE to compute the inverse Laplace transform of Fs), enter invlaplace (F(s),5,t): to obtain f(t). This assumes that the integral transforms package has been opened. EE reosLeus Ineach of Problems 1 through 5, use Table 3.1 to determine the Laplace transform of the function. 1. kQ)= sre + sine) 2. w(t) = e05(31) — 0s(71) pl meree 4. g() =e sint8) 5. (= 144 — sing) In each of Problems 6 through 10, use Table 3.1 t0 determine the inverse Laplace transform of the function. FFor Problems 11 through 14, suppose that f(¢) is defined forall > Oand has a period T. This means that f(¢++7) = £(0 forall 0. 11. Show that cine Ss [" ereoae88 © CHAPTERS The Laplace Transform 12, Show that [Co roa 13. From Problems 11 and 12, show that enno=[Se] fl ersioa ~ [leroa 14, Recall he geometric series Pie for |r| < 1. With this and the result of Problem 13, show that LIfe) = oF [fe "Fond In each of Problems 15 through 22, a periodic function is, given (sometimes by a graph). Use the result of Problem 14 0 compute its Laplace transform, 15, f has period of 6 and js ford form to solve the intial value problem. yf Ay =c0s(); O)=O e's yQ=1 11. Prove Theorem 3.1. Hint: Write cyo=[ ersoa YF 1Gy=145 yO =-2.¥O) Sy'+6y =e (0) =0.y'0) and integrate by pars 12, Derive equation (3.3). Hint: Integrate by parts twice 33 ‘Shifting and the Heaviside Function ‘The shifting theorems of this section will enable us to solve problems involving pulses and other discontinuous forcing functions. 3.3.1 The First Shifting Theorem ‘We will show that the Laplace transform of e f(t) is the transform of f(t), shifted a units tothe right. This shift is achieved by replacing s by s —a in F(s) to obtain F(s—a).35 Shifting and the Heaviside Function 103 Here is a rationale for Heaviside’s formula. The partial fractions expansion of p(s)/q(s) has the form| All we need are the numbers Ay, Ay to write LTE) = Aye ot Ae We will ind A. The other 4,’ ae found sinilrly. Notice that 26s) 5 50) 2 = A+ Ay a S90) soa TFA, Because thea,’ are assumed tobe distinc, then PO) Lime -ay with all the other terms on the right having zero limit as s —> ay. But in this limit, (s — a;)p(6)/q(s) is exactly the quotient of p(s) with the polynomial obtained by deleting s — ay from q(s). This yields Heaviside’s formula. For those familiar with complex analysis, in Section 22.4, we will present a general formula for C"'(F] as a sum of residues of e F(z) at singularities of F(z). In that context, Heaviside’s {ormula is the special case that F(z) is a quotient of polynomials with simple poles ata, ++ dy. PROBLEMS Ineach of Problems 1 through 15, nd the Laplace trans yy p.,)_ [1—cos(2t) for st <3 form ofthe function 2 f0=19 fort > 3x stg 2e _ feos) fords <2 1 eG 42249 13. r= { re oe 2 e*—cost)) 14, e'(1 —coshie) 3, tos " oo 15. ( —31+2)e* tb pon [2t 800 fort fO=19 fort>x In each of Problems 16 through 25, find the inverse Laplace transform. 5. ee +sin(o) 6 e¢-2) 16 1 ferder<3 1 f= s ro hag tee " 4 fordt3 i (-2 fordsr<16 pe E ford<1<2 0. f= s 21, roel se tds 2. for0st<7 1 GOs [cos(t) fort = 7 2,10§ CHAPTERS The Laplace Transform SyaseR 1 F=a35 26, Determine Lle-* ' e* cos(3w)dw}. Hint: Use the fst shifting thor 25, In each of Problems 27 through 32, solve the inital value problem. 21. yr —8y=g0: YO =¥O=y"O=0, with ( for0<1<6 = Vy toria6 28. y"—4y +4y= FO: VO (0) = 1 with refs SS” 29. eae YO) =0, y"(0) = 1, onl tee 30. y"—2y' —3y= FN; yO) =1, yO) =0, with snl, wes 31. y" +4y= FO; xO) ¥(0): with 10) { ford 0 such that this integral converges.54 Convolution 109 FIGURE 3.21 Replacement function. Figure 3.21 is a graph of this replacement function for A= B=0.001, and k expected, r(¢) is a strictly increasing function, because it measures total replacements up to a given time. The graph gives an indication of how the drug needs to be replenished to maintain F(0) doses at time Me Prosiems In each of Problems 1 through 8, use the convolution the- ‘orem to help compute the inverse Laplace transform of the function. Wherever they occur, a and b are postive constants. 1 seed) way 2 + 3D yer ae Frae=H 1 +16" In each of Problems 9 through 16, use the convolution theorem to write a formula forthe solution in terms of f. 9. 10. nL y= yt — ay dy yO) 12, y® — Lly" + 18) 7m By” By 4 2y= SO; ¥O) -32YO=2 In each of Problems 17 through 22, solve the integral ‘equation. 1, fe 4 ff fe—de 18 f(D=—141=2f fev sinteyde0 CHAPTER 3 The Laplace Transform 1. f=I+ ff f(epeas(at— ode 20. 14 ff f—esineonde 2 1+ ff fe—ne*ae D. fiymcosty +e ff fear 23, Solve for the replacement function r(t)if f(t) = A+ Br-+CP and m(0) =e". Graph r(0). 24, Prove the convolution theorem, Hint: Fist write [Proetsae FOG) reac ["aue—nse—syontee Use the definitions of the Heaviside function and of the transform t obtain FO)GE)= * [erases Reverse the order of integration to obtain roof [ = [eusnoe. (Of —dede From this, show that CLF #g](8) = F()G() 25. Solve forthe replacement function rt) if F() =A, constant, and mie) =e-" with k positive constant oh (0 26, Solve forthe replacement function (0) if f(Q= A+ Br and m(o) =e-*. Graph r( " Impulses and the Delta Function Informally, an impulse is a force of extremely large magnitude applied over an extremely short period of time (imagine hitting your thumb with a hammer). We can model this idea as follows. First, for any positive number € consider the pulse 6, defined by 8.() = 21H — H¢-))- ‘This pulse, which is graphed in Figure 3.22, has magnitude (height) of 1/e and dura- tion of e. The Dirac delta function is thought of as a pulse of infinite magnitude over an infinitely short duration and is defined to be 3() = lim 8.0. 80 ve FIGURE 3.22 5,(:)444 CHAPTERS The Laplace Transform FIGURE 3.25 Graph of the output voltage in Example 3.18. BEE Peosieus In each of Problems 1 through 5, solve the initial value problem and graph the solution. 1. y"F4y"45y42y= 6500; Y= O=y'O=0 2. YP 16y! = 128(¢— 52/8); 0 3. y+ 59" +6y= BECO yO) =: ¥O 4, y"=4y" 4 1By = 450-3, YO =Y'O 5. yHSy +6y: Bt -2) — 4500-5}: YO y¥=0 66. Am object of mass m is attached tothe lower end of a spring of modulus k. Assume that there is no damping. Derive and solve an equation of motion forthe object, assuming that at time zero itis pushed down from the equilibrium position with an initial velocity wp. With ‘what momentum does the object leave the equilibrium position? 3.6 Solution of Systems Prove the filtering property ofthe delta function (The- orem 3.6). Hint: Replace 6(¢ a) with uot —e-0- 0-0) in dhe integral and interchange the limit and the integral A.2 pound weight is attached to the lower end of a spring, stretching it 8/3 inches. The weight is allowed to come to rest in the equilibrium position. At some later time, which we call time 0, the weight is struck a ‘downward blow of magnitude 1/4 pound (an impulse. ‘Assume no damping in the system. Determine the velocity with which the weight leaves the equilibrium position as well asthe frequency and magnitude ofthe oscillations. ‘Suppose, inthe seting of Problem 6, the objects struck downward blow of magnitude mvp at time 0. How does the postion of this object compare with that ofthe ‘object in Problem 6 at any postive time 1? Physical systems, such as circuits with multiple loops, may be modeled by systems of linear differential equations. These are often solved using the Laplace transform (and later by matrix methods), We will illustrate the idea with a system having no particular significance, then look at a problem in mechanics and one involving a circuit.418 CHAPTERS The Laplace Transform Put these into the circuit equations to get i+ 120 fs) ~Acer de +120(q(0) 4x0) 0, 6ois+ 120 fsa + 10,0) = 120 f(s) tarde + 1200410) ~ 2), Setting q:(0) =q2(0) =0 in this system, we obtain, 40i, +120 fics) —itepar=10 6oie+ 100 fieyde=120 fe) —Hfe) de Apply the transform to obtain 120, 120, _ 10 40h +—Sh- ha ‘These can be written 1 (+9h-3h=3 2 -(s+4)h=0. Solve these to obtain ses = Tee NETO Wst1 and 1 1414 b= 3 Then aod MEE Pontes In each of Problems 1 through LI, use the Laplace trans- orm to solve the initial value problem, 1 Hn D_+3y) 0) = 320) = p20) = 2. 242)" 4x $3y +y=-6 x =(=0 Bx -y=2,x ty vty 8(t),2" +29 =0; (0) =y()=012, Solve for the currents in the circuit of Figure 3.28 assuming that the currents and charges are initially zero and that E(®) =2H(¢—4)—H(—5). 20 19 WW BO GN GN WW 3a FIGURE 3.28 Circuit in Problems 12 and 13, Section 3.6. 13, Solve for the currents in the circuit of Figure 3.28 if the currents and charges are initially zero and E(e 1-H¢—4)sin@a~4)) 14, Solve for the displacement functions of the masses in the system of Figure 3.29. Neglect damping and assume zero initial displacements and velocities and external forces f(t) = fa(0)=0. FIGURE 3.29 Masv/spring system in Problems 14 and 15, Section 3.6 56 Solution of Systems 119 15, Solve for the displacement functions in the system of Figure 329 if Al) =1-HO=2), Ald and the initial displacements and vel 16. Consider the system of Figure 3.30. Let M be sub- Jected to a periodic driving force f(4) = Asin(ot) ‘The masses are initially at rest in the equilibrium position, o tes are zero, [ « ” » FIGURE 330. Massspring system in Problem 16, Section 36. (2) Derive and solve the inital value problem for the displacement functions forthe masses. (b) Show that, if'm and fy are chosen so that fz], ten the mass m cancels the forced vibra tions of Af. In this ease, we eall m a vibration absorber. 17, Two objects of masses m, and my are attached to ‘opposite ends of a spring having spring constant k (Figure 3.31). The entire apparatus is placed on a highly varnished table. Show that, if the spring is stretched and released from rest, the masses oscillate with respect to each other with period 2 [a Kom +m) NN | k FIGURE 881 Mass/spring system in Problem 17, Section 3.6. 18, Solve for the currents in the circuit of Figure 3.32 if E(t) =5H(t 2) and the initial cutents are 220 20H 30# COO WO a 3 wo FIGURE 82 Circuit of Problem 18, Section 3.6a0. CHAPTERS The Laplace Transform 2 galimin 3 galiin FIGURE 8.33 System of tanks in Problem 20, Section 3.6. 3 gabinin 3 alma FIGURE 8:34 System oftanks in Problem 21, Section 3.6. 19, Solve for the currents in the E()=55(0— 1). 20, ‘Two tanks are connocted by a series of pipes as shown in Figure 3.33. Tank 1 initially contains 60 gallons of brine in which 11 pounds of salt are dissolved. ‘Tank 2 initially contains 7 pounds of salt dissolved in 18 gallons of brine. Beginning at time zero, 2 mix- ture containing 1/6 pound of salt for each gallon of ‘water is pamped into tank I at the rate of 2 gallons ‘per minute, while salt water solutions are interchanged Detween the two tanks and also flow out of tank 2 at the rates shown in the diagram. Four minutes after time zero, salt is poured into tank 2 at the rate of 11 pounds per minute for a period of 2 minutes. uit of Figure 3.32 if a 3.7 Polynomial Coefficients 3 21 Determine the amount of salt in each tank for any time 1>0. ‘Two tanks are connected by a series of pipes as shown in Figure 334. Tank 1 initially contains 200 gallons of brine in which 10 pounds of salt are dissolved. Tank 2 initially contains 5 pounds of salt dissolved in 100, gallons of water. Beginning at time zero, pure water is. pumped into tank I atte rate of3 gallons per minute, While brine solutions are interchanged between the tanks atthe rates shown in the diagram. Taree min- tutes after time zero, 5 pounds of salt are dumped into tank 2. Determine the amount of sat in each tank for any time # > 0. 1 Differential Equations with Polynomial Coefficients If a differential equation has polynomial coefficients, we can use the Laplace transform if we ‘know how to take the transform of a function of the form r* f(t). Begin with the case n326 CHAPTERS The Laplace Transform EE Posies Solve each of the following problems using the Laplace transform. ty" +0 - Dy + y=0;:yO=0 yi +2 —4y=6: yO) =YO=0 (oy + ¥=0; 1. y= By! +16 =3: (0) =9'(0)=0 9. y= Nbry +32y=0, 90) =y'O) 3.¥O 10. y"-+4ty —4y=0; 7) =0, 0) CO a eeere 11. Review the derivation of the solution of Besse's equa 0;(0)=0, (@=10 tion of order m for m positive integer. Are any steps taken that would prevent m being an arbitrary post- tive number, not necessarily an integer? Could n be negative? 0 -1 Appendix on Partial Fractions Decompositions Partial fractions decomposition is an algebraic manipulation designed to write a quotient P(x)/Q(x) of polynomials as a sum of simpler quotients, where simpler will be defined by the process Let P have degree m and let Q have degree m and assume that n > m. If this is not the ease, divide Q into P. Assume that P and Q have no common roots, and that Q has been completely factored into linear and/or irreducible quadratic factors. A factor is irreducible quadratic if it is second degree with complex roots, hence it cannot be factored into linear factors with real coefficients. An example ofan irreducible quadratic factor is x7+4. ‘The partial fractions decomposition consisting of writing P(s)/Q(z) as a sum S(x) of simpler quotients is given in the following rules. 1. If x —a is a factor of Q(x) but (x —a)? i not, then include in $(x) a term of the form A 2.1 @ —a)* is a factor of Q(x) with k> 1 but (x — a)** is not a facto, then include in (x) a sum of terms of the form Boy BB @=aF Gray 3, Ifax?-+bx-+c is an ieducible quadratic factor of Q(x) but no higher power is a factor of Qs), then include in S(x) a term of the form Cx+D_ ax? thx Fe 4. f (ax? +-bx +0} isa product of ireducible factors of Q(x) but (ax? +x +0)! is not a factor of 0(4), then include in S(x) a sum of terms of the forma Crt Dy, CxtDs |) Gxt De aetbrte! @atbe+ oF G@erixto When each factor of Q(x) has contributed one or more terms to S(x) according to these rules, we have an expression ofthe form57 Polynomial Coefficients 127 with the coefficients to be determined. One way to do this is to add the terms in S(x), set the ‘numerator ofthe resulting quotient equal to P(x), which is known, and solve forthe coefficients of the terms in S(x) by equating coefficients of like powers of x EXAMPLE 3.23, ‘We will decompose 2-1 Py6e 45x12 {nto a sum of simpler fractions. First factor the denominator, then use the rules 1 through 4 to ‘write the form of a partial fractions decomposition: 2-1 2e-1 PF6845x- 12” -DEFDEFD Menara, tras tree (Once we have this template, the rest is routine algebra. Ifthe fractions on the right are added, the numerator of the resulting quotient must equal 2x — 1, which is the numerator of the original quotient. Therefore, AC +3044) + Bex +4) + Cle G43) =2e— ‘There are at least two ways we can find A, B, and C. ‘Method 1 Multiply the factors on the left and collect the coefficients of each power of x to write AG? 47x +12) + BO? +32 —4) + C(? +2x-3) = (A+ B+ C)x74+ (7A+3B + 2C)x + (12A 4B - 30) =2r = quate the coefficient of each power of x on the left to the coefficient of that power of x on the right, obtaining a system of three linear equations in three unknowns: A+B+C=0 from the coefficients of x2, 7A +3B +2C =? from the coefficients of x, and 124 —4B —3C =—1 from the constant term, Solve these three equations obtaining A= 1/20, B=7/4, and C = -9/5. Then 2x-1 Laij7i gt P+G 45x12 2x-1 4x43 5x4a Method 2 Begin with AC -+3)(¢-44) + BG = 1) +4) + CG = E43) =2x 1, and assign values of x that make it easy to determine A, B, and C. Put x =1 to get 20A 50 A= 1/20. Put x =—3 to get —4B =—7, s0 B=7/4, And put x = —4 to get SC =—9, 50 9/5. This yields the same result as method 1, but in this example, method 2 is probably easier and quicker. #128 CHAPTERS The Laplace Transform EXAMPLE 3.26 Decompose i P4243 . wee into partial fractions. First observe that x? +x +5 has complex roots and so is irreducible. Thus, ee use the form +2043 A B Cxt+D +r Ne + Gap aad ‘If we add the fractions on the right, the numerator must equal x* + 2x +3. Therefore, AG =2)(4" 44-+5)-4 BU? +4 +5) + (Cx+ DY 2 a3? $2 $3, Expand the left side, and collect terms to write this equation as (A+OP + (A+ B-4C + D)x? + GA4B+AC—4D)x— 104 +5B 44D SP +2 +3, Equate coefficients of like powers of x to get A+C=0, ~A+B-4C+D=1, 3A+B+4C—4D —10A+5B4+4D=3. Solve these to obtain A = 1/11, B=1,C =—1/11, and D =—3/11. The partial fractions decomposition is P2043 @4r+G—F x43 B4at5134 CHAPTER 4. Series Solutions ) This is the general solution, since aj and a, are arbitrary constants, Because ap = y(0) and (0), a unique solution is determined by specifying these two constants. EE PkoeLems In each of Problems 1 through 10, find the recurrence rela tion and use it to generate the frst ive terms of a power series solution about 0 9. ye yyer 10. y"42y +33 y bay costs) ‘We will focus on the differential equation, PO)y"+ Oly’ + R@Y= FC). an If P(x) £0 on some interval, then we can divide by P(x) to obtain the standard form y+ Py +al)y = fx), 68) If P(4) =0, we call xg a singular point of equation (4.7) This singular point regular if 2@) + R(x) and (&— x9) 2O) Poa) Gy are analytic at xy. A singular point that is not regular is an irregular singular point. @—x) EXAMPLE 43 PUE=D?y" $5 +DOE—Dy' +3x7y=0 has singular points at 0 and 2. Now (0 20) _ Sx +20 ~2) 5 (2) PG) e@-2 LE? is not analytic (or even defined) at 0, so O is an irregular singular point. But Q(x) _ 5x +2) «05542 Frobenius Solutions 143 Furthermore, the recurrence relation is forn=1,2, particular solution, we may also choose c} va HEE Posies In each of Problems 1 through 10, find the frst five terms, ‘of each of two linearly independent solutions. aly! —2xy'~ (72) ay" $x(e + Dy |. Any" +2) +2y: ay"-y+2y=0 ances) yiln(x) +1 ant nero] on F DI Since ¢ can be any nonzero number, we wil for convenience let = 1. For a (0. These give us 3 5 P36" et xQ— ay" —2¢e— Ny +2y Any" + dry’ + (4x? 9)y xr Dy" +3y'-2) ay" —2ay'+2y=0 ay (=a)! by. 4x'y" +4ry'-yME reoeiems 1 Let pla) bea polynomial. Prove tha there is no finite sum 5-* , by sin(nx) that is equal to p(x) for Zea sin(nx)460 CHAPTER 13 Fourier Series RENE "1001005 Let f(3)=2Ssin(x) for a , with a similar inequality for B,. Add these two inequalities to obtain 1 eet ital + IBD 5043 + BD += 135 Phase Angle Form ‘A function f has period p if f+ p)= Hence show that L la als UAE + BD +E. “Thus show by comparison that Lian + iad converges. Finally, show that lay costs) +b, singexx/ 23]

} dees 5 x (fev +4 ye) Lec ee => Zen sinter, rss) ‘This is the Fourier series for f(x) on{[-1,1]. @ ‘The amplitude spectrum of a complex Fourier series of a periodic function is a graph of the points (na, |dg|)- Sometimes this graph is also referred to as a frequency spectrum. MEE Pxonze0s Ineach of Problems I through 7, wre thecomplex Fourier, [1 for0sx<2 series off, determine the sum ofthe seis, and plot some FO=)) tecrex1. ‘There is another expression for the Fourier integral of a function that is sometimes convenient to use. Insert the coefficients into the Fourier integra: f [Ax cos(eox) + By sin(ox)] de f [GS Feeyeoxetyat coon + (=f 1st) a) sao] do aie =i f f _, FE Dleos(ast costo) + sin(of) sin(o))d5 der $f [re cosoe—srae aos This gives us the equivalent Fourier integral representation EL [sercosioe -yat dw asa) of f(x) on the real line. Incach of Problems | through 10, write the Fourier integral this integral converges to. 1 so-{ bel foc sx<2n See ace ee oe ee x for—cxem for -100-m 0 for|x1> 100Bane... CHAPTER 14 The Fourier Integral and Transforms eee ors Caer Se Fine) rte see 5. fee)mer! “|0 torx-<—3x and forx > x ess es eee eee es ee ete aes users tnt [rote We can define Fourier cosine and sine integral expansions for functions defined on the half-Line in a manner completely analogous to Fourier cosine and sine expansions of functions defined on ahalf interval, ‘Suppose (x) is defined for x > 0. Extend f to an even function fon the real line, where _[7@) forxeo, son [fe, tree This reflects the graph of f(x) for x > 0 back across the vertical axis to a function J, defined on the entire line. Because fis an even function, its Fourier coefficients are f LAG) cos(wt) dé 2 f ” Fe eostos) ds and f FelG)sin(ws) dg =0. ‘The Fourier integral of f.(x) contains only cosine terms. Since f(x) = f(x) for x > 0, this expansion may be thought of as a cosine expansion of f(x), on the half-line x > 0. This leads us to define the Fourier cosine integral of f(x) on x =0 0 be f * Ancoe(on) do 45) in which : ak [” peyeosios at 46) is the Fourier integral cosine coefficient.478 CHAPTER 14 The Fourier Integral and Transforms ‘Next compute the sine coefficients aon? [er sintotte ‘Then, for x > 0, we also have 2 £ egitinor)do. However, tis integral is zero for x = 0 and so does not represent f(x) there. ‘These integral representations are called Laplace's integrals because A, is 2/zr times the Laplace transform of sin(x), while B, is 2/1 times the Laplace transform of cos(kx). # BEE reonsens In each of Problems I through 10, find the Fourier cosine and sine integral representations of the function, Deter- ‘mine what each integral representation converges to. 2r+1 fordsxen 1 fO)=]2 forn 3x, x forOsx2 3. FG) =e" cos(x) for x >0 4, fx) sxe" for x20 1 ford4 14.3 The Fourier Transform 6 f@ 10. n lcosh(x) for05 Let k be a nonzero number and c a positive number, and £6) Jk fordexe, fla) =e cox) forx>0. ford1 He-de* ental 3. f0) 4 £0) 5. £0 6 1. FS FQ=HO be _ fine). for—kersk » r0-( Sa 9% fO=/A+F) 10, F0)=3H-2e* Incach of Problems 1 through 15, find the inverse Fourier transform ofthe function. 11, (14a) (6—e? Si) int Factor the denominator and use pal factions 12, 10sin@0)/(o+z) 13. geen 14, ®-9/3—(5—a)i) 15, 9/(5— (3 —w)i) Ineach of Problems 16, 17, and 18, use convolution to find the inverse Fourier transform of the function. 16, sin(30)/o(2-4i0) 17. 1/0+i0y 18. 1/(1+ia)2+i0)) 19, Prove the following version of Parseval's theorem: [Luora= [fora 20. Use the Fourier transform to solve ¥'+6y/-45y = 3). 21. Compute the total energy of the signal (1) = (1/1) sin@). Hint: Use Parseval's theorem, Problem 19. 22. Compute the total energy of the signal f(0) = Hoe, Ineach of Problems 23 through 28, compute the windowed Fourier transform of f for the given window function ww. Also compute the center and RMS bandwidth of the ‘window function. 2B. f=C427, we) 1 for—2<1=2, 0 for el>2.wm. fO=e'sinae, we 26. f=HE-m, wl 498 CHAPTER 14 The Fourier Integral and Transforms 0 for |e] > 5. 28. f(t) =cos(at), of, for | > 4r. 1 for3n Sa. Fourier Cosine and Sine Transforms If f is piecewise smooth on each interval (0, L] and (-° |f(¢)| dt converges, then at each where 4 is continuous, the Fourier cosine integral for fi 10= [ aucontony den, where a f ” paycos(an) dt. We define the Fourier cosine transform of f by Fel fo)= Often we denote Fel fw) = fel). Notice that few) and that s0=2 [” jeroosionde. F(C)cos(or) dt. 14.16) a4.a7 Equations (14.16) and (14.17) form a transform pair for tbe Fourier cosine transform. Equa- tion (14.17) is the inverse Fourier cosine transform, reproducing fom . This inverse is denoted fe". EXAMPLE 14.12 Let K be a positive number and let 1 ford K. felwy= f° foocostonyat [(econar sin(Ke) 4144 Fourier Cosine and Sine Transforms 499 Using the Fourier sine integral instead of the cosine integral, this discussion leads us to define the Fourier sine transform of f by Fifiey= [ fosinwwn at We also denote this as f(a). If f is continuous at r > 0 then the Fourier sine integral representation is 10= [” basinvon de, where 2 fF feo sinonar= ‘This means that [ florsinen do ‘which provides a way of retrieving f from, This integral isthe inverse Fourie sine transform As. EXAMPLE 14.13 ‘With f as in Example 14.12, fioy= f° Fesigonya sinot)dt = ‘The following operational formulas are needed when these transforms are used to solve differential equations. Operational Formulas Let f and ff’ be continuous on every interval (0, L] and let f° |f(t)|dr converge. Suppose f(1)—> 0 and "() > 0 as t+ 00. Suppose f" is piecewise continuous on every (0, L]. Then 1. Fel fo) =o" fotw) — FO). 2. FalLf" OMe) =-a" fslo) taf). + ae Posies In each of Problems 1 through 6, determine the Fourier sinh(®) for K <0 <2K cosine transform and the Fourier sine transform of the “Vo forOsr 1.04832. Then es mL = 0.47694, This gives the 128 point DFT approximation of 0.47694 to the sampled partial sum Syo(1/2), which was computed to be 0.45847. The difference is 0.0185. ‘The actual sum of the Fourier series at r= 1/2 is f (1/2) =0.50000. In practice we would achieve greater accuracy by choosing I larger, allowing larger M. PROBLEMS function is given 1sin(1) forO=1 <4, p=4, # for0sr0 teos(t) for 0d,use the generating function to shove that 1 om Yer reosoyr Use Rodrigues’s formula to derive Py(x) through Px). Use the recurrence relation to derive P(x), Py(x), and PG), [Expand each of the following polynomials in a series of Legendre polynomials, (@) 142e— () 2rtx? 5x" © 2-0 44x5. Iean be show that Ry On? de ee ‘where, for any number f, [F] denotes the largest into- ‘ger not exceeding 1. Use this formula to generate P(x) through Ps(x) (2) Pe). Solve the resulting equation for us) to derive the second solution 1 on = Re) f tar of Legend's equation. These functions Q(x) are called Legendre functions ofthe second Kind, and they are defined bt unbounded on (~1. 1). Show that 1 (ute (7). ss) lee —pm(t#)_3 2.09 = Zoe —pin( 2) — Bessel Functions 155 Bessel Functions 541 7. Letn be a nonnegative integer. Prove that Poy s(O) =O and PO) = 8. Show that X (sr) 209-35 In each of Problems 9 through 14, find the fist ve coef ficients in the Fourier Legendre expansion of f(x) on [1,1], Graph the function and the partial sum of these first five terms on the same set of axes, for ~3 0. We will use the following property of P(x).568 CHAPTER 15 Special Functions and igenfunction Expansions We obtain by 2 [w -* cos(ny) ¥ is cs) ay r e230) yy cos(ny)dg 2 2 f cos(ng —nesin()) dp 2 qian). ‘The last line comes from Bessel’s integrals, The solution to Kepler’s problem is therefore OHV) 2 Ulne)sininy), BEE Proviems y-(Et)ye(vems In each of Problems 2 through 9, write the general solu- tion of the differential equation in terms of 2*J,(bx*) and. xJ(bx*) for appropriate a, band c 3 Saree pi 3y. pak a vedy+(ie- 2 1 4 yore 10, Use the change of variables Law uae to transform the differential equation 2 year” Db aes! into the differential equation eu we bextu= Find the general solution of this differential equation in terms of Bessel functions and use this to solve the original differential equation. Assume that b is a positive constant. In each of Problems 11 through 16, use the given change of variables to transform the differential equation into one ‘whose general solution can be written in terms of Bessel functions. Use this to write the general solution of the coviginal differential equation 11. 9x2y"49xy' + (be ~ 16) 12, dxty" 4 Bry’ + (de? —35)y =05, 2x" ye13. 4 15. 16. 1. 20. 2 36x*y"— IDey! + Bax? + Dy =O; Arty" +41 + OP ~36)y=0; 4xty"+4ny +0 9y=0. Sey" —27xy' + Ox +35)y=0,u= yx? Show that, for any postive integer n, [Ptnscrdr=s"40 and Ins) yp te) (Here we have omitted the constants of integration) Hint: Use Theorem 15.10. Alternatively, one can inte- arate the series for Jyay(x)/x term by term, Show that for any positive integer and any nonzero number a, and Hint: The result of Problem 17 can be used, or the series for J(x) can be used, Leta be a positive zero of Je. Show that il Jaxyde= + Let u(x) = Jo(ax) and v(x) = Jo(Bx), with @ and postive constants. (@ Show that xu" +0’ + aeeu = 0, with a similar equation for v (©) Multiply the diferential equation for u by v and the equation for v by w and subsract to show that fw'v- vw a)xuv. (©) Use the conclusion of part (6) show that =a) J syexyspnas [a J(ex) Jo(Bx) - BIg(B2)J0(a)], ‘This is one of a class of integrals known as Lonimel’s integrals. For @ any nonzero number, and » and & nonnegative integers, define tas [ dys glade. (2 Show that 153 Bessel Functions 368 Hint: Use the first integral in Problem 18. (©) Show that tax fe £ (aut) es ‘Hing: Use the first integral from Problem 18 in the definition of fn. (© Show that This provides a recurrence relation for the quantities Ina. Hint: Integrate by part in (). @ Show that ks “Hint: Apply part (c) in repetition. (©) Show that 2re+1), Hint: Use the result of pat (@), (® Show that ares), (g) Show that, ifn is a nonnegative integer and m is a positive integee with n 0, then for any positive x, twee fe Hint: Let try in the definition of P (2), ‘Show that 1(1/2)=/m, using the fact from statistics x= Je/2. at.162 Wave Motion on an Interval 575 Ifthe string is released from rest (zero initial velocity), then g(x) is identically ero. Further, if the sting has its ends fixed at the same level, then the intial position function f must satisfy the compatibility condition (0) = f(L) =0. Variations on this problem can allow for moving ends with conditions such as ¥(0, 1) =a(t) and y(L, 1) = BC), ‘We can also have a forcing term in which an external force drives the motion of the string. In this cease the wave equation is ey ety ae +FO,1) (16.2) in which (x, y, 1) is the displacement funetion, Itis a routine exercise in chain rule differentiation to convert this two-dimensional wave equation to polar coordinates. Ifthe displacement function is u(r, 0, , this wave equation is ): (163) Su _ a (Pu Ld, 1B ar "rar BEESON Proscems 1. Let f bea twice-ifferetiable function of one variable. 4. Show that z(x,y. 1) =sin(nx) cos(my) eos (V@™-E mF) Show that satisfies the two-dimensional wave equation for any 1 positive integers n and m. WED Uebel + fle— ot] 5. Let y(x.1)= sin(nx/L)cos(nmet/L) for Ox 0 ae ¥10,1)=y3,0) =O for =0 4 FF ford0 y(2,0)=0, 2G,0)=16-x fordsxs3 00,0) y(4,1)=0 for 0 vat) mteen, eso) st forO0 -xlorOsx0 W0.1)=y2x.1) =O for =0 ay yE2,0)= fla), (x, 0)=0 for Osx <2" 3x forO0 0.1) y(5.9=0for1=0 3 6.00.2 ,0)=e6) fe0n 5 fordex<4 were (8) = - BO N55 tordexss, Solve ay vy $2 = FY cose) for 0 310.1) = yx, 1)=0 for =0 a ybe,0)=0, (2,0) =0 for 0 0, Oford0, 13, 14 (©) Solve for the separation constant and for X and T in the case of supported ends in which au HO.) =H0) = F500) = Solve the zlegraphh equation eu Er for00. A and B are positive constants ‘The boundary conditions are atte ar 14(0,1)=u(L, 1) =0 for > 0, and the initial conditions axe u(x,0) 40,86, =0fe 0x ‘Assume that A*L? <4(BL? +n?) (@) Write a series solution for 5 92 452 order 4,10 ae (0,1) = y4,1) =0 fort >0 sO) mete, 0) =O 025 <4 (©) Write a series solution when the forcing tem Sx? is removed. (©) In order to gauge the effect of the forcing term onthe motion, graph the fortieth partial sum of the solutions in pars (a) and (b) on the same set of axes when 1 = 04 seconds, Repeat this for 08,14,2,25,3, and (a) Write a series solution for 0, ay, 94,0) =sin(ox), P(x, 0)=0 for Ox <4. (©) Write a series solution when the foreing term e-* is removed, (©) In order to gauge the effect of the forcing term on the motion, graph the fortieth partial sum of the solutions in parts (a) and (b) on the same set cof axes when 1 = 0.4 seconds. Repeat this for 0.8, 14, 2,2.5,3, and 4 (@) Write series solution for ay oos(erx) for0-0 for t>0 _ Oe ye 9,0) =24—2), 26,0) =0 for00 O for >0 ¥(0,0)=0, 2,0)=0 for 02x 4 2x ford-0 an : Graph the fortieth partial sum of the solution for y(O,1)=y(2,)=0 fore =0 selected values oft, 16.3 Wave Motion in an Infinite Medium If great distances are involved (as with sound waves through the ocean or cosmic background radiation across the universe), wave motion is often modeled by the wave equation on —co < x <0, In this case, there is no boundary, hence no boundary condition. However, we sek bounded solutions ‘The analysis is similar to that for solutions on a closed interval, except that J, --da replaces J". As we did on a bounded interval, we will consider separately the cases of zero initial velocity and no initial displacement. Zero Initial Velocity The initial-boundary value problem is, = F(x), 2%, 0)=0 for ~ 00 0 ‘Then X(x) = ce“ +cxe-**, and this function is unbounded on the entire real line if either constant is nonzero, This problem has no negative eigenvalue. Case 3: If > 0, write 4 = 07 with o> 0 Now X(x) =c:cos(wx) +eysin(wx), @ bounded function for any choices of positive w. Every positive number 4 =? is an eigenvalue, with eigenfunctions of the form X,, = cos(wx) + esin(ox).592 CHAPTER 6 The Wave Equation EXAMPLE 16.10 ‘We will use the Fourier transform to solve the wave equation on the real-line subject to the intial conditions g(x) =0 and costs) for—n/2/2. With zero initial velocity, we can use the solution (16.11). We need the Fourier transform of the initial position function: fe= f Fee de =f cosceyeni ag = {2008/2)/1 a") forwA L =f comes dea a F is continuous because sm 2008(0/2) _ mI Tat 72 ‘The solution is ee a re(t 2cos(zre/2) ): ‘The integral is complex because the Fourier transform is a complex quantity. The real parts the solution ofthe problem. In this case, we can extrac the real part by writing = cos(ax) +i sin(wx), obtaining [22 2P costox)cosiwer) do. & In each of Problems 1 Unrough 6, solve the wave equa- foes tion on the real line for the given initial position f and initial velocity g, fist by separation of variables and the Fourier integral, and then by using the Fourier transform. The same solution should result from both methods. 3. c=3, f(x) =0, and g(x) = het 4. o=2, f(x) =0, and for02 and forx <2 sins) form cx 8, i] for22 2. c= 1st) ome so0~ {>