404. Chapter? Techniques of Integration and has the property that the string is always tangent to the curve. Set up a differential equation for the curve and solve it, 75 Integration of Rational Functions Using Partial Fractions 2-3 45x SE ON eT a+ Se = 38 - x 31S TarsT Figure 1 Answers to Concepts Review: 1 V3 2 2sine 3. lant 4. 2sece A rational funetion is by definition the quotient of two polynomial functions. Examples are 2 r+? xt Ha (x +1) a) = vn 4e +8 ma) e+ Sx Of these, f and g are proper rational functions, meaning that the degree of the ‘numerator is less than that of the denominator, An improper (not proper) rational function can always be written as a sum of a polynomial function and a proper ra- tional funetion. Thus, for example, ee eee Mx +1 tox , Yt Se Aix) = 1 result obtained by long division (Figure 1). Since polynomials are easy to inte- grate, the problem of integrating rational functions is really that of integrating, Proper rational functions But can we always integrate proper rational functions? In theory, the answer is yes, though the practical details may be messy. Consider first the integrals of fand g above. MERASILED es f 2? SOLUTION Think of the substitution u = x +L ary 3 2x +1) @p (eet 1 (x41 . ERATED] ra f 22 a SOLUTION Think first of the substitution «=x? 4x +8 for which du = (2x ~ 4) dx, Then write the given integral as a sum of two integrals. ete: [ea [ec =4e+8 va4x+8 wade +8 . 1 =tnle? = 4x +81 +6 fae widths | Saas In the second integral, complete the square. fren [a Laces wre +) +6 3 “Se We conclude thatSolve This DE “Olen, there is hile resemblance between a differential equation and is solution, Who would suppose that an expression as simple as dy ae could be transformed into This resembles the transformation of a chrysalis into a butterfly" Silvanus P. Thompson ‘The method of partial fractions sakes this an easy transformation Do you see how iis done? Section 7.8 Integration of Rational Functions Using Partial Fractions 405. I is a remarkable fact that any proper rational function can be written as a sum of simple proper rational functions like those illustrated in Examples 1 and 2. Partial Fraction Decomposition (Linear Factors) To add fractions is a standard algebraic exercise: find a common denominator and add. For example, 23 meee) set set y-1x+1 G@-D@+) @-NeF) ¥-1 Itis the reverse process of decomposing a fraction into a sum of simpler fractions that interests us now. We focus on the denominator and consider cases. Distinct Linear Factors Decompose (3x ~ V)/(x* ~ x and then find its indefinite integral 6) SOLUTION Since the denominator factors as (x+2)(x—3). it seems reasonable to hope fora decomposition ofthe following form: —3x-1____a B Ge “ ) eta eH Our job is, of course, to determine A and B so that (1) is an identity, a task that we find easier after we have multiplied both sides by (x + 2)(x ~ 3). We obtain @ 3x1 = A(x = 3) + B(x +2) or.equivalenty, 0 3-15 (A+ B)x + (-34 +28) However, (3) is an identity if and only if coefficients of like powers of x on both sides are equal: that is, A+B= “3A +2B=-1 By solving this palr of equations for A and B, we obtain A=1,B=§ Consequently, and \ 8 [ref Zine +21 + $inbe = 3) +e . there was anything difficult about this process, it was the determination of A and B. We found their values by “brute force.” but there is an easier way. In (2), which we wish to be an identity (that is, true for every value of x), substitute the convenient values x = 3 and x = ~2, obtaining 8=AO+ BS -1= As(-5) + 8-0 This immediately gives B= and A= 2 ‘You have just witnessed an odd, but correct, mathematical maneuver. Equa tion (1) turns out to he an identity (true forall x except ~2 and 3) if and only if the406 Chapter? Techniques of Integration this isso. Ultimately it depends om the fact that the two sides of equation (2), both linear polynomials are identical if they have the same Values at any two points, EEXAMPIES) Distinct Linear Factors Find {738 SOLUTION Since the denominator factors as x(x + 1)(x ~ 3), we write Sx+3_ ALB X@te-3) © etd and seek to determine A, B, and C.Clearing the fractions gives Sx +3 = Ax + 1)(x 3) + Bx(x ~ 3) + x(x + 1) Substitution of the values = 0, x = —1,and x = 3 results in 35 Al-3) -2= Dis) 18 = c(12) Thus, 1,-1f4 3 freafeedfe 1 3 : = -tnls| — inks #1 +3ins—3] +e NEXAGIPTES) nepnt Lear taco Ff = dx, (x3 SOLUTION Now the decomposition takes the form _x 4A, 8B Gy 3 Ga with A and B to be determined. After clearing the fractions, we get x= AWe-3) +B If we now substitute the convenient value x ~3 and any other value, such as x= 0, we obtain B = 3and A= 1, Thus, fee] [EEXAMPLE 6) Some Distinct, Some Repeated Linear Factors. Find ae - aren ia Fae SOLUTION We decompose the integrand in the following way: ae-sr+ A c (e+ aap +3 Gy Clearing the fractions changes this to Bx? — Bx +13 = A(x ~ 1)? + Bx + 3)(n - 1) + CU + 3)Section 7.5 Integration of Rational Functions Using Partial Fractions 407 Substitution of x = 1, x = -3, and x = O yields C = 2, A= 4, and B = =1. Thus, [eect ss- [Flap +c oo8 = 4 in|x + 3] = nlx = 1) = Be sure to note the inclusion of the two fractions B/(x ~ 1) and C/(x ~ 1)? in the decomposition above. The general rule for decomposing fractions with re- peated linear factors in the denominator is this: for euch factor (ax + b)* of the Jenominator, there are k terms inthe partial fraction decomposition: AL A: As Ae a+b” (ars bp” (ax toy (ax + 5 Partial Fraction Decomposition (Quadratic Factors) In factoring the denominator of a fraction, we may well get some quadratic factors, such as x + 1, that cannot be factored into linear factors without introducing complex mambers x +1 5 é 6 GETXAWPLE7) A Single Quadra Factor Decompose sy and then find its indefinite integral. SOLUTION The best we can hope for is a decomposition of the form 6x -3r+1 A, Be+e Grete) et et To determine the constants A, B, and C, we multiply both sides by (4x +1) + 1) and obtain 6x2 — 3x41 = AGE +1) + (Be + Ox +1) Substitution of x =~. = 0,and.x = 1 yields &titt=aQ) = A=2 leat = ce- 4544 (8-1) > BH ” fatstges fade [3 +1 afesital ser Lei 1 4, a Pinte + 1h + dance +i)-urtx+¢ nieem EXAMPLES) {& EXAMPLES) A Repeated Quadratic Factor Find {735 st SOLUTION Here the appropriate decomposition is -1x+2 A Br+C, Det (e+ 38 +2)? va2 '(i+ay After considerable work, we discover that A = 1, B = ~1,C = 3,D = ~S, and E = 0.Thus,408 Chapter 7 Techniques of Integration TEI A Bound for the Answer “The inital population size is 800, and the rate of change in population size is pasitive,so the population _gfows. Asit mears 2000, the rate ff change pets close 107270, $038 1400, we have y+ 2000. The Population atime ¢ = 2 should he somewhere between 800 and 2000. 6x2 = 152 +22 (ear +a [Bs aaa ts dx 1 f_2e 5 f_2xdx + ne “_S x+3 {ea 3 +2 2S (x? +2) 51'S) + aermte Summary To decompose a rational function f(x) = p(x)/q(x) into partial fractions proceed as follows: = Inlx + 31 lig ghd Bot zinta? +2) v2 Step 1: If F(x) is improper, that is, if p(x) is of degree at least that of q(x). divide (2) by q(x), obtaining N(x) Da) f(x) = apolynomial + ‘Step 2: Factor D(x) into a product of linear and irreducible quadratic factors with real coefficients, By a theorem of algebra, this is always (theoretically) possible. Step 3: For each factor of the form (ax + 6), expect the decomposition to have the terms a er (ax +b)” (ax + BF (ax + 5 ‘Step 4: For each factor of the form (ax? + bx +)". expect the decomposition to have the terms Bx tc, ae + bet ae + (ae + be Bak + Co (ae + bet Step 8: Set N(x)/D(x) equal tothe sum of all the terms found in Steps 3 and 4 ‘The number of constants to be determined should equal the degree ofthe denom- inator, DQ). ‘Step 6: Multiply both sides of the equation found in Step S by D(x) and solve for the unknown constants This ean be done by either of two methods: (1) Equate co efficients of like-degree terms or (2) assign convenient values to the variable x. The Logistic Differential Equation In the last chapter, we saw that the assumption that the rate of growth of a population is proportional to its size, that iy! = ky, leads to exponential growth. This assumption may be realistic until the available resources in the system are unable to sustain the population. In such @ ‘case, more reasonable assumptions are that there is a maximum capacity Z. that the system can sustain, and that the rate of growth is proportional to the product ‘of the population size y and the “available room” £. ~ y. These assumptions lead. to the differential equation kL») ‘This is called the logistic differential equation. It is separable and now that we have covered the method of partial fractions, we can perform the necessary inte- ‘aration to solve it, HEEXAMPLES) A population grows according to the logistic differential ‘equation y’ = 0.0003y(2000 ~ y). The initial population size is 800, Solve this di ferential equation and use the solution to predict the population size at time ¢ ~ 2.Section 7.8 Integration of Rational Functions Using Partial Fractions 409 SOLUTION Writing y’ as dy/dr, we see that the differential equation can be written as. * ~ go1m3y(2000- 9) out y) oom Isai ~ [oma ‘The integral on the left can he evaluated using the method of partial fractions. We write 1 A,B s2000— yy ~ 2000 —y which leads to 1 = A(2000 = y) + By Substituting y = 0 and y = 2000 yields = 20004 20008 1 1 Thus, A = and 7 2000" 2000 1 1 J es * ssrwo=55) = 00+ € 1 : Fann!" ¥ ~ Fqp!(2000 ~ y) = 0003 + leading to 460 + 2000€ 2000 y we 2000 y Here, C, = &, At this point we can use the initial condition y(0) = 800 to determine C, 300 oso 000 — 800 ~ ~ _ 80 2 = T00 = 3 Thus: — ad 2000-3 a2 ose F (2000 - ye 4000 se 3 (4000/32 4000/3 2 1+ (ase ~ 2a + eH410 Chapter? Techniques of Integration Figure 2 A sketch of the population size as a function of ris given in Figure 2 “The population at time ¢ = 2s thus 4000/3 B+ Concepts Review 1. If the degree of the polynomial p(x) is es than the de- poe of) then (2) = p(x) i ced 8 ronal 2 Tojntegrate the improper rational function Fa) = (2 + A)/C4 + He fs eerie tas f(8) = B I(x ort) tarts and + be + e.then 4 (ax + 1)/MCa — 1)? + 1)] Gam be decomposed in the form Problem Set 7.5 In Problems 1-10, use the method of partial fraction decomposi= ‘ion to perform the required integration. Lf 2 [ane “fz Ms (Qe — 1)(3x + 2)(x - 3) “= 8 [atin o* PG + 114-6 ae Fe — ae + Ser — 30 * vf am [tha PHF Fraere we fsa om PE Faat a a [hw Feared 3e¢2 oe Jaana » [aa a PEBEBe m feats an [tote = fare aay * [zt : » [aaaerRe yoo Gine + fein — asin e + 5) cory oe * [oaraaern" west sty, ag f 2 Be + Ie 7 “Ls wen rr a Taare 0.In Problems 41-4, solve the logistic differential equation repre. senting population growth with the given intial condition. Then use the solution to predict the population size at tine t= 3. 4b y= y(L — 9), 0) = 05 a gal v2 v0) 6 48, y* = 010003» (8000 ~ y),»(0) = 1000 ‘.001y (4000 = »), »(0) 45, Solve the logistic diferemtial equation for an arbitrary constant of proportionality k, capacity L, and initial condition (0) = Bo 46, Explain what happens to the solution of the logistic dit. ferential equation if the initial population size is larger than the v 47, Without solving the logistic equation or reterring to as Solution, explain how you know that if < £, then the popula tion size is increasing. 48, Assuming yy < L, for what values of (isthe graph of the population size y(@) concave up? ay 00 49. Suppose thatthe earth will not support a population of ‘more than 16 billion and that there were ?illion people in 1925 and 4 billion people in 1975, Then. i ys the population ¢ years after 1925, an appropriate model is the logistic differential equation dy a 7 RYO ~ 9) (a) Solve this differential equation, (b) Predict the population in 2015, (©), When will the population be 9 billion? 50, Do Proiblem 49 assuming thatthe upper limit for the pop lation is 10 billion, 7.6 Strategies for Integration Section 7.6 Strategies for Integration 411 1. The Law of Mass Action in chemistry result inthe differ- cetial equation 4 aka = 3) 9), k>0, a>, b>O where xs the amount of a substance at time f resulting from the reaction of two others Assume that x = Owhen 1 = 0, (a) Solve this differential equation in the case b > a (b) Show that x +a as 1 00 (ib > a). (©) Suppose that « = 2 and b = 4, and that 1 gram of the sub- stance is formed in 20/mimutes. How much will be present in hour? (€) Solve the differential equation if 52, The differential equation dy Seo ky mm ~ ») with & > Dand =m < yy < AP is used to model some growth problems. Solve the equation and find lim: 553. Asa model forthe produetion of trypsin from trypsino- em in digestion, biochemists have proposed the model as GMAW MBH) where k > 0, Ais the initial amount of trypsinogen, and B isthe original amount of trypsin Solve this differential equation, 54, Evaluate Answers to Concepts Review: Rae-i+ RIK ‘Throughout this chapter we have presented a number of techniques for finding an antiderivative (or indefinite integral) of a given function. By now it should be clear that while differentiation is a straightforward process, antidifferentiation is, not. The Sum Rule, Product Rule, Quotient Rule, and Chain Rule can be used to find the derivative of almost any function, but there is no sure-fire method for finding antiderivatives. There is only a set of techniques that one might apply. ‘Thus to a large extent, an ifferentiation is a trial and error process; when one ‘method fails, look for another. This being said, however, we can give the following strategies for finding antiderivatives, 1. Look for a substitution that makes the integral look like one of the basic integration formulas from the first section of this chapter. For example, in2xdx, f xe dx, [VF Tae ‘can be evaluated using simple substitutions, 2. Look for situations where you have the product of two fun ive of one of them times the antiderivative of the other is one of the deriv i, where the basic integration formulas from Section 7.1. Integration by parts can be used412. Chapter 7 Techniques of Integration forthese inteprals For example xe*dvand [xsinh.x dv can both be eval uated using integration by parts 3. Trigonometrc Substitutions Ifthe integrand contains Vi? — the integrand contains Vi + a, consider the substitution x = a tans. If the imegrand contains Vi — a, consider the substitution x 4. If the integrand is a proper rational function, that is, the degree of the numer- ators ess than that of the denominator, then decompose the integrand using the method of parca! fractions. Often the terms inthe resulting sum can be in tegrated one ata time. Ifthe integrand isan improper rational function, apply long division to write it as the sum of a polynomial and a proper rational func- tion. Then apply the method of partial fractions to the proper rational function. consider the substitution x = a sin sect, ‘These suggestions, along with a bit of ingenuity, will go a long way in evaluat- Ing antiderivatives, Tables of Integrals The inside back cover of the book contains 110 (indeti- nite) integration formulas. There are larger tables such as those found in the CRC Standard Mathematical Tables and Formulae (published by CRC Press) and Handbook of Mathematical Functions (edited by Abramowitz and Stegun, pub- lished by Dover), but the lst of 110 will sufice for our purposes. The important thing to keep in mind is that you must often use these formulas along with the method of substitution to evaluate an indefinite integral. This is why many tables of integrals including those at the end of this book, use u forthe variable of in- tegration. rather than x. You should think of w as being some function of x (maybe just x itself), The next example shows how one formula can be used to evaluate several integrals using the method of substitution ‘Use Formula (54) (54), [ve = du = ve to evaluate the following integrals: @ [vreau ) [ Vie=H ay © fyvizway (a fevio= ear SOLUTION (a) Inthisintegral we have a = 3 and w = x, 50 [Vreu-ere For part (b), we have to recognize 4y* as (2y)*, so the appropriate substitution is uw 2yand du = 2dy. Thus [Ve-Rar =} [VET BF ean 73 apt aes. VE= Gay + Ssin-!22) +c = 3Vis= ay + 4sintt +cTables and Substitution Tn Example 1 we were able to evalu ate four seemingly unrelated inte- gral using the same formula from the table of integrals, Each required different substitution, When you tse a table of integrals to help you evaluate an integral, Keep in the back of your mind the method of substitution Section 7.6 Strategies for Integration 413 Part (c) requires a litle foresight. We might be tempted to make the substitution u =~ 4y‘, but then du = ~16y" dy. The presence of y* in the expression du is troubling because we have just inthe remainder ofthe integrand. For this part we ‘must see the radical as Vi — (2y")* making Formula (54) applicable with the sub- stitution u = 2y? and du = 4y dy. Thus [vir Way} [Vir @F wan CE : oP yrmo + lta? =v 4y" zg 2y? +c For part (d) we must recognize that the radical can be written as W100 ~ (@) and that we should make the substitution u = e and du = ef dt. Thus [eviarFa~ [Vie ean svr=re Vi00 = + 50 sin Computer Algebra Systems and Calculators Today a computer algebra system such as Maple, Mathemauica. or Derive, can be used to evaluate indefinite or definite integra, Many calculators are also capable of evaluating integrals I such systems are used to evaluate definite integrals, it is important to distinguish ‘whether the system is giving you an exacr answer, usually obtained by applying the Second Fundamental Theorem of Calculus, or whether it is giving an approxi- nation (using something similar to, but probably a little more sophisticated than, the Parabolie Method of Section 4.6) It might seem lke the two are equally good in practical situations, and if we were presented with just one integral {0 evaluate, this may well be correct. However, in many cases the result of the definite integral Will be used in subsequent calculations. Ina case like this itis more accurate, and often easier, to find the exact answer and then use the exact answer in further cal- ‘culuations. For example, if dx is needed in subsequent calculations, it ‘would be better to find an antiderivative and use the Second Fundamental Theo- em of Calculus to obt dx + Using /4 in subsequent calculations would be preferable to 0.785398, which is what Mathematica gives for a numerical approximation to the integral. In some cases, however, itis not possible to evaluate a definite integral by ap- plying the Second Fundamental Theorem of Calculus, because some functions do not have antiderivatives that can be expressed in terms of elementary functions. ‘Our inability to find a simple formula for an antiderivative does not absolve us of the task of finding the value of the definite integral. It just means that we must use numerical method to approximate the definite integral. Many practical problems, lead to just this situation, where the necessary integral is intractable and we must resort to a numerical method. We discussed numerical integration in Section 4.6. ‘Usually a CAS will use a method similar to the Parabolic Rule, but a little more sophisticated.414 Chapter? Techniques of Integration E An Approximate Answer “The mass isthe integral of the dows ty.s0 the mass ean be thought of as the area under the density curve. At position x = 0, the density is I, and itdecreases slowly asx increases. In ‘order to make the area under the density curve equal to I, we would ‘expect to have 10 choose the cut-off point to be slightly larger than 1. yationes Find the center of mass of the homogeneous lamina shown in Figure I. SOLUTION Using the formulas from Section 5.6, we have Ae maa [oneae a an=8 [rsa ve 8M a 2 My af sin? a? de Among these integrals, only the second can be evaluated using the Second Funda- mental Theorem of Calculus. For the first and the third, there is no antiderivative that can be expressed in terms of elementary functions, We must therefore resort ‘oan approximation for the integrals. A CAS gives the following values for these integrals Nottce that the CAS was able to give an exact value for the second integral and approximations for the first and third, From these results we can calculate Ms 3 ote tag ts M, _ 0.334948 Fn ry = 03748 . ‘There are also situations where the upper limit of an integral is an unknown. If this is the ease, then use of the Second Fundamental Theorem of Calculus is pre- ferred aver the use of a numerical approximation. The next two examples illustrate thie The twa problems are in principla thecame, hut the methods of a necessity, different ‘A rod has density equal to 8(x) = exp(—a/4) for x > 0. Where should the rod be cut off so that the mass from 0 to the cutis equal to one? SOLUTION Let «denote the cut-off point. We then require [oonde= Feriansyac =a — ser" Solving for a gives La4—4eu4 dew =3 3 a= -4Ing = 1.1307 Here we obtained the exact answer, a = ~4 In(3/4). which we could approximate as 1.1507 if we needed an approximation. .TE] Another Approximation Section 7.6 Strategies for Integration 415, Using the fact that mass is area tunder the density curve, we see from the figure below that the cut-off ‘ust he somewhere between 0.8 and This give us a starting point to approximate the answer. LEEIGIPTET 4 soning ts) ~ pS”) o> 0 Where should the rod be cut off so that the mass from 0 to the cut is equal to one? Use the Bisection Method to approximate the cut-off point accurate to two signif- ‘cant places SOLUTION Again, let a denote the position of the cut. We then require that - [ora [eo($ee) ax ‘The antiderivative of exp($x22) cannot be expressed in terms of elementary functions so we cannot use the Second Fundamental Theorem of Calculus to eval uate the definite integral, We ae forced to approximate the integral using numer Cal methods The problem is that we must have fixed upper and lower Hints forthe integral in order to approximate i, but in this ase the upper limits the variable a. A litte tral and error, along with a program to approximate definite integrals leads tothe following [o0(e2)aesosne aos tom At this point we know that the desired value of ais between 0.5 and 1.0. "The mid point of [05, 1.0] is 0.75, 0 we try 0.75: ars, [aol] ae 0098 «= o84S 00s ans, [col ie asia [e(Le) arm Las «= oassasiin fo large osnases [aol Le") ac . = mh @ mm ng 0 ass 16 tre 20 a 219 157 Figure? ETA Rough Approximation “The figure suggests thatthe average fol consumption is about 20 miles per gallon and that the average velocity is about 50 miles per how. After? hours (120 minutes) the ear Would have traveled about 100, miles and, at roughly 20 miles per gallon the fuel consumed would be 100) miles per gallon expect our answer to be near S gallons 5 gallons. We SOLUTION We will use the Tiapezoidal Rule to approximate the integrals, The distance traveled is the definite integral of instantaneous speed, so 20 b= [Sar 2p te rae 6 +40) +41] = 109.4 miles af ‘The total arount of fuel consumed isthe integral of where /() is the amount of fuel in the car's tank at time ¢, Notice that fuel consumption is given in miles per gallon, which is ds/df. The last column in the above table is the speed ds/de divided by ds/df. The fuel consumed is therefore taf * ds/dt bie ~ fy asjap 2 8 i 90 + 2(1.66 + 152 + + + 161) +157] 2130 = 530 gallons . Special Functions _ Many definite integrals that cannot be evaluated using the Second Fundamental Theorem of Calculus arise so often in applied mathematics that they are given special names, Here are some of these accumulation functions, along with their common names and abbreviations: the error funetion the sine integral the Fresnel sine integral 2 ‘There are numerous others; see Handbook of Mathematical Functions for ‘many more. Algorithms, often involving infinite series (a topic we will take up in ‘Chapter 9), have been developed to approximate these functions. These algorithms are usually accurate and efficient. In fact, itis no mote difficult (for a computer. anyway) to approximate the Fresnel integral §(1) than itis to approximate the sine ‘of I, Since many practical problems work out to involve such funitionsit isimpor- tant to know that they exist and how to find approximations for them, (EXAMPLE 7) Express the mass of the la the Fresnel sine integral. the Fresnel cosine integral cw = f “eos(2) a 1a from Example 2 in terms of418. Chapter? Techniques of Integration SOLUTION The mass was found to be If we make the substitution x = 1-V/2, then x? = 'n/2 and du = Vap2dt. “The limits on the definite integral must also be transformed ‘Thus, x=050=0 xevesrevi maa [so(®) (Ea FE [so 22) Noh 2 Fs v2) = 0.895 5 . Concepts Review 1 Tithe of tegris areson ei when wel in cone juiction wit hn mattod of 2 Both {(x? + 9)2dx and f(sin? x + 1)" cos xd can be evaluated ising Formula Number 3. When using a CAS to evaluate a definite integea i is Important to know whether the system is giving us an exact answer or a(n) Ae The ogra evaluated at ¢ — Oe $(0) ~ Problem Set 7.6 1m Problems 1-12, evaluate the given integral L f wea “ine WE ay . ~ 5 » [aren [wre x xsinedy 1 In Problems 13-30, use the table of integrals om the inside back ‘cover perhaps combined with a substitution, to evaluat the given integrals 18 [artis © f Za 0) foo V Vie et dx eos sin dt 11) [evo Bts (0) so? xox I= FT ats wof Fao [Me offs © lea m9. @ fevarse ride (by de BVaaS dt aofraa © ha 2 [aw [YES sofa o ji aofTie % lant sech VE OE as 2. Jot sea» [etn View oT ae fooersine Vicon w [sine Use a CAS 0 evaluate the definite invegrats in Problems 31-40. 10, the CAS does mo give an exact answer in terms. of elementary functions, give a numerical upproximation. a [am [savias a [Cater ow [otter « [» uVvau—1 {In Problems 41-48, the density of a rod is given. Find cso thatthe ‘mass from Oo € & equal to J. Whenever possible find an exact solution. If this ix not possible, find an approximation fore. (See Examples 4and 5). 1 41. (0) = 5 0) 43. (x) = In(x +1) 480)" S45 48, (x) = 207 46. 5{x) = InGet + 1) 2) wy et 1.54) = bn: 8. Findesoat [ r= ede = 0.90, h sve Lege = 095. Hoe Ue 58 Find © so that “Vie symmetry. Section 7.7 Chapter Review 419 {In Problems 51-54, the graph of y = f(x) is given along with the graph of line. Find c so that the x component of the conter of ‘mass ofthe shaded homogencous lamina is equal to 2 st. 455, Find the Following derivatives, q 4 (a) ett) 0) Ss 86, Find the d 4 (a) Bia) tives ofthe Fresnel functions “ @ Ze S57. Over what intervals (the nonnegative side ofthe mum ‘ber line) is the error function increasing? Concave up? S58. Over what subimervals of [0,2] isthe Fresnel function S42) inereasing? Concave up? 489, Over what subintervals of [0,2] is the Fresnel function ‘C(x increasing? Concave up? 0, Find the coordinates of the fst inilection point of the Fresnel function S(3) that st the right ofthe origin ‘wubstitation 2, 53 3. approximation 4. 0 7.7 Chapter Review Concepts ‘Test Respond with re or false wo each of the following asertions Be Prepared to jusify your answer 4. To evaluate [since ds make the substitution oe mos square ofthe denominator. 3 s Toortate [2 ten coming te Vix,CHAPTER 8 8.1 Indeterminate Forms of Type 0/0 8.2 Other Indeterminate Forms 8.4 Improper Integrals: Infinite Limits of Integration 8.4 Improper Integrals: Infinite Integrands Geometric Interpretation ‘of Hépital’s Rule Study the diagrans below They should make THopita’s Rule scom quite reasonable, (See Prab- Jems 38-42.) ti fi 13-0 iy £2 «lg FD nga" Sb yar Indeterminate Forms and Improper Integrals 8.1 Indeterminate Forms of Type 0/0 Here are three famil + Limit problems: ti SE, ‘The frst was treated at length in Section 14, andthe third actually defines the de- rivatve f'(a), The three limits have @ common feature. In each ease, a quotient is involved and, in each ease, both numerator and denominator have O as their limits An attempt fo apply part 7 of the Main Limit Theorem (Theorem 1.3), which Sys that the limit of quotient is equal tothe quotient of the mits. leads to the nonsensical result 0/0. Infact, the theorem does not apply. since it requires that the limit ofthe denominator be diferent from 0,WWe are not saying tha these limits do not exist only that the Main Limit Theorem will not determine them, You may recall that an intricate geometric argument led us tothe conclusion limy(sinx)/x = 1 (Theorem 1.4B). On the other hand, the algebraic technique of factoring yields (=3r+3)_ 2436 Bet2) ext? 5 ‘Would i not be nice to have a standard procedure for handling al problems for which the limits ofthe numerator and denominator are bath (? Thats too much 19 hope for. However. there is a simple rule that works beautifully on a wide variety of such problems LHopital’s Rule In 1696, Guillaume Frangois Antoine de I'Hépital published the first textbook on differential calculus; it included the following rule, which he had learned from his teacher Johann Bernoulli UHopital’s Rule for forms of type 0/0 Suppose that lim f(x) = tim g(x) = 0. If im["(x)/a"(x)] exists in either the if this limit tim 22 — jg 2D Sh gtx) eG) finite or infinite sense (j a finite number or ~00 or +00), then tals Rule allows us to replace one limit by another, which may be simpler and, in par- ticular, may not have the 0/0 form, (EXAMPLE 1) Use rHopital's Rule to show that tig = 1 and tim = 8 9 Before attempting to prove this theorem, we illustrate it. Note that Hd 423424. Chapter8 Indeterminate Forms and Improper Integrals SOLUTION We worked pretty hard to demonstrate these two faets in Section 14. After noting that trying to evaluate both limits by substitution leads to the form 0/0, we can now establish the desired results in two lines (but see Problem 25). By 'Hopital’s Rule, Das De EXAMPLE 2) Find tim rn2x = 2x 2x +3 acd haya “The first of these limits was handled at the beginning of this section by factoring and simplifying, Of course, we get the same answer either way. ESAT) in 2, SOLUTION Both numerator and denominator have limit 0. Hence, fmm . Ifonly ¢ and es were equal, we could divide the first equality by the second and be done: but there is mo reason to hope for such ‘coincidence. However, this attempt is not a complete failure since (2) yields the valuable information that ¢() ~ g(a) # 0, a fact we will need later (this follows from the hypothesis that "(x) # 0 for all xin (a,)).. Recall that the proof of the Mean Value Theorem for Derivatives (Theorem 36A) rested on the introduetion of an auxiliary funetion s. IF we try to mimie that proof, we are led to the following choice for s(x). Let £0) = fs ) a) = fl) ~ fla) — 7 ay lax) ~ sad] No division by zero is involved since we earlier established that g(b) ~ g(a) # 0. Note further that s(@ (5). Also, s is continuous on [2, b] and differen- Liable on (a,b), this following from the corresponding facts for fand g. Thus, by the Mean Value Theorem for Derivatives there is a number c in (a, ) such that s(6) - sa) boa boa But Ac) «Fj LALO oy. HO =F ~ ay tay? O £0) _ f(b) ~ fla) gle) at) — sta) Which is what we wished t0 prove, . Proof of L'Hépital’s Rule Proof, Refer back to Theorem A, which actually states several theorems at once. We will prove only the case where Lis finite and the limit is the one-sided limit tim, The hypotheses for Theorem A imply more than they say explicitly. In particu- lar. the existence of lim {f"(x)/a"(x)] implies that both f’(x) and g'(x) exist in at least a small interval (a, b] and that g(x) # 0 there, Ata, we do not even know that f and g are defined, but we do know that lim, f(x) = O-and lim, g(x) = 0 “Thus, we may define (or redefine if necessary) hoth f(a) and g(a) tobe zero, there by making both fand g (right) continuous at a. All this is to say that fand g satisfy: the hypotheses of Cauchy’s Mean Value Theorem on [a,b]. Consequently, there is a number c in (a, b) such that fle) - fa) _ f°) 8) —s@) se) or,sinee f(a) sta), f(b) ath)Section 8.1 Indeterminate Forms of Type 0/0) 427 which is equivalent to what we wanted to prove. A very similar proof works for the case of left-hand limits and thus for two- sided limits. The proofs for limits at infinity and infinite limits are harder, and we omit them. 7 Concepts Review 1. Leptas Rule is useful in nding liml/(x)/e(s)). 3 From THpita's. Rule, we can conclude that where both ___and_are zero, . inna )/ = Fim, ; but FHOpitars Rule gives 2. LHopitas Rue says that under appropriate conditions Wn information about Fim(Gaya/ x been lim fxV/atx) = im —— 4. The proof of HOpita’s Rule depends on __ Theorem. Problem Set 8.1 In Problems 1-24, find the indicated lonit, Make sure that you /have an indeterminate form before you apply ! Hopital’ Rule sin ~ tn 15, lim, im cosh x [voma [Vicor 23, Jim, 24 im 25, In Section 14, we worked very hard to prove that lim (sin.x)/x = 1; Hopita's Rue allows us to show this in one line. However, even if we had I Hopita’s Rule, say at the end of Section 1.3, it would not have helped us Explain why. (We really [the way we didin Section 1.4.) isa toe in Hint: Begin by deciding, why 'HOpital’s Rule is not applicable. “Thon find the limit by other means. 27. For Figure 2, compute the following limits. area of triangle ABC (©) IBS area of curved region ABC area of curved region BCD (©) area of euved region ABC Figure 2428 Chapter 8 Indeterminate Forms and Improper Integrals 28, In Figure3,CD = DE = DI (0) him» () im AarO) Figure 3 29, Let sex) { ¢ its What value of e makes f(x) continuows atx 3 Let ms sla) = { a itx What value of e makes f(x) continuous at [BL Using the concepts of Section 5.4, you cam show that the [SU 42, Use the concept of the linear a surface area of the prolate spheroid goiten £. Find each limit x axt+ b+ 33. Lopital’s Rule in its 1696 form suid this If fim f(x) = Him a(x) Fay/ats) = f"(a)/e'(a). provided that /°(a) and g’(a) both exist and g(a) # 0. Prove this result without recourse t9 Cauchy's Mean Value Theorem. (Use a CAS to evalua the limits in Problems 34-37. cose 14/2 52 : me! arso (Gal For Problems 38-41, plot the numerator f(x) and the denom- <0 {nator g(x) in the same graph window for each of these domains: Hl sx 1,0. < x= O41, and -001 = x = O01. From the plot, estimate the values of {'(x) and g'(x) and use these 0 ap- proximate the given limit. “1 8 in ® mt Dyrotating the ellipse function (Sect 1m 2.9) to explain the geomet ja + y/b2 = 1 (a >) about the x-axis is [’Hopita’s Rule in the marginal box next toTheoter A. What should A approach as ab"? Use VHopital’: Rule to show that ths does happen. 8.2 Other Indeterminate Forms 2 spenyer(sy & Value In the solution to Example 6 of the pres problem. “Ths is typical of a class of problems of the form lim. (x)/a(x), where both nu- merator and denominator are growing indefinitely large; we call it an indetermi- hate form of type 20/20. Teturis out that Hépital’s Rule also applies inthis situ ation; that i, tim £9. jim LO BB g(a) Eg) A rigorous proof is quite difficult, but there is an intuitive way of seeing that the result has to be true. Imagine that f(¢) and g(¢) represent the positions of two ‘cars on the axis at time ¢ Figure 1). These two cars, the f-car and the g-ear,are on ‘endless journeys with respective velocities f”(r) and g'(). Now,Section 8.2 Other Indeterminate Forms 429 then ultimately the fcar travels about L. times as fast as the g-car. It is therefore reasonable to say that, in the long run, it wil travel about Z. times as far; that is, im £09 AB) We do not call this a proof, but it does lend plausi formally. ty toa result that we now state UHOpital’s Rule for Forms of Type 2/0 Suppose that fin |f(x)| = tim |g(2)] = 98.4 fim /"(x)/6°(x)] exists either the finite or infinite sense, then ‘The Indeterminate Form 9°/c2 We use Theorem A to finish Example 6 of the previous section. Exar) Here is a general result of the same type. [EEXAMPLE2) show that it ais any positive real number, im, 2 re SOLUTION Suppose as a special ease that a = 2.5. Then three applications of VHpital's Rule give © gg 8b ys fim 25115105) _ 9 wee ane et ae et A similar argument works for any a > 0. Let m denote the greatest integer less than a. Then m + | applications of HOpital’s Rule give430. Chapter8 Indeterminate Forms and Improper Integrals ‘See How They Grow Tn computer science, one pays eare- ful attention tothe amount of time needed to periorm a task, For exam= pile. to sort items using the "bubble Sort” algorithm takes time propor- tional toa”, whereas the “quick ‘sn algorithm dacs the same task in ime proportional to. In. major improvement, Here sa chart illustrating how some common func- tions prow asx inereases from 10 t0 100 10 1000. ws 23 a6 vi to a6 x 1001000 B one 1 tom iF 22st 27x10 10h SOLUTION Both In x and x" tend to 00 as x + ov, Hence, by one apy PHopital’s Rule, ©) Examples 2 and 3 say something that is worth remembering: for sufficiently large m ¢ sro fsteras x incraces thar ety Contr! power of: Wherods ds ares mate sey thas any constant power of For example; when a sutisear- Iylargee* grows faster than 1" and nx grows more lowly than 0’. The chart inthe agin apd Figure 2 offer additional iatration. Ins EXAMPLE 4) Find lim. 2 ote” SOLUTION As xO*Inx—> co and cotx—oo, so MHopital’s Rule applies. Us aatoe ore "yor [ese] ‘This is stil indeterminate as it stands, but rather than apply P'HOpital's Rule again (which only makes things worse), we rewrite the expression in brackets as Vx 0 . The Indeterminate Forms ()+ ee and 2 — 2 Suppose that A(x) +0, but &(.x) -* 2°, What is going to happen to the product A(x)B(x)? Two competing forces are at work, tending to pull the product in opposite directions, Which wil win this battle, A or B or neither? It depends on whether one is stronger (ve. doing its job at a faster rate) or whether they are evenly matched. L’HOpital’s Rule will help us to decide, but only after we have transformed the problem to a 0/0 or 0/26 form. WEEXAMPEE 5] Find im, (tan x- tnsin x). SOLUTION Since fim, insin.x = 0 and fim, ltan.x| = 9°, this is a 0-00 indeterminate form. We can rewrite it as a 0/0 form by the ‘changing tan x to 1/cot x. Thus, Tim (can.x+In sin x)= tim Insine rome iors fim, (608 sin) =0Section 8.2 Other Indeterminate Forms 431 EXAMPLE) Find im SOLUTION The first term is growing without bound so is the second. We say ‘that the limit is an 0 ~ oo indeterminate form. L’Hopital’s Rule will determine the result, but only after we rewrite the problem in a form for which the rule applies In this case, the two fractions must be combined, a procedure that changes the problem to a 0/0 form. Two applications of 'Hopital’s Rule yield © in fim xnxx xeliv+inx=1 ote athe (= Ding ahs (= De) + Ine © xing dene 1 m —*IX_ tim co deelaw lth Deine ‘The Indeterminate Forms 0, 00", 1” We wn now to three indeterminate forms of exponential type. Here the trick is to consider not the original expression, ‘but rather its logarithm. Usually, 'Hopital’s Rule will apply to the logarithm. (BEEXAMPLE7) Find Jim (x + 1)%*, SOLUTION This takes the indeterminate form 1*. Let y = (x + 1)*,s0. Iny = cotxin(x + 1) = MED Using Hépital’s Rule for 0/0 forms, we obtain = Fim tny = tim IM * DS jim F* AN ana nb ache and since the exponential function f(r) = e* is eontinuous, exp! im, espn y) = exp tm tn 9 EXAMPLES) Find tim, (tan x) SOLUTION This has the indeterminate rm oo". Let y = (tan x), 50 Intan x Iny = cos x Intanx = SER Then finn Iny= tim AREAMA Sigg Fa waite Me secx Ate sec xian = tim lim S28 me432. Chapter8 Indeterminate Forms and Improper Integrals Concepts Review 1. IF lim f(+) = tim g(x) = 20, then PHOpitas Rae says that fim F}/a(s) = fy “Therefore, At, . Summary We have clasifed certain limit problems as indeterminate forms using the seven symbols 0/0, 0/2¢, 0-20, 20 — 20, 0 09, and I™, Each involves a competition of opposing forces, which means that th results not obvious How ever, with the help of fHapital's Rule, which applies directly only tothe 0/0 and 56/20 form, we ean usual determine the lit There are many other possibilities symbolized by, for example, 0/2, 0/0 0 + 26, 20+ 00, 0, and 20, Why don't we call these indeterminate forms? Be- cause, in each of these cases. the forees work together, not in competition. EXAMPLES] Find fim (sin x) SOLUTION We might cal this a 0° form, but itis not indeterminate, Note that sin.xis approaching zero, and raising i to the exponent cot x, an increasingly large number, serves only (o make it approach zero faster. Thus, lrg (sin x)%* = 0 . fe discussed in this hook. ‘They are symbolized by 0/0, °/90, 0-6, and grows faster than any power of x, but grows an indeterminate Form. To apply FHépitals Rue, we may sewrite this later limit as Problem Set 8.2 Final cach limit in Problems 1-40, Be sure you have am indetermi- 28, fgg x" 24, ting (eos x)" ‘nate form before opplving /Hopita’s Rue. = = tail 28. fim (tan x)" 26. lim (ex) e 27. ig (sin) 28. fig(con x — sin it 1 1y 4 lim aide +e) 2 lim (sex 4) 30. Jin (2 + 4) In sin? x x 6 lim Sram SH fig (1 + 268)!" 2 in(- 4) i Ind ~ 88) 38 lim (cos 9%" Min (2! tn) 35 ime" 10. 36. Hin {Inc +1) ~ In(x ~ 1} 12. tm a x am ig 38 in nxt) 1% fim (ose x = cot?x) 4A, tim (tam x ~ sees) [vewe [smc 18. fim (34) 36 im(cos ay 2 tin $4 im ” (Seos.y""* 4, Find cach limit, Hine Transform to problems involving & saith ‘continuous variable x. Assume that a > 0 19. tim(x +e)" @) jim va ©) sim va (© jim (2-1) (fim MH~ 1)42, Find each limit (2) im, (©) img x0" 48, Graph» . (©) sim (24) (© ig. x (8) im (099 Section 8.3 Improper Integra: Infinite Limits of Integration 433 Here TT means product that i, [T, means a,» paniculata & x.andy ae postive and a +b lig (ax! + by!" = atv 47, Verify the last statement in Problem 46 by calculating tor x > 0, Show what happens for very each of the following, small x and very large x. Indicate the maximum value fa) im (52" + $5)" + ts) 44, Find cach mit. © inl ) ) (a) fi (1 29)" (6) im (1+ 25" (©) im (52! > 751)! (© tmcreane (a) sme (oe 48, Consider f(x) = wxe™. —_ ~ {a) Graph f(x) for n 4,5,6.0n [0, 1] in the same graph 45, Fork = 0,fnd Sie ig BB et (0) For x > Osfind fim, /(3), Hint: Though this as the 20/50 form, 1 helpful. Think of a Riemann sum. i at fdas bps a Jet 2, %..¢5 be positive numbers. Take natural logarithms and then use FHopital’s Rule to show that tn ( Son)!" 8 8.3 Improper Integrals: Infinite Limits of Integration (9 trae [foarien= 13.8488 ‘Hopital’s Rule Ws not (2) Guessat is [| 102) a-Trenjusity your answer rigorously. and (55149, Find the absolute maximum and! minimum points (i they exist) for f(x) = (x8 + 4° + eon (0,2). Answers to Concepts Review: _L Ptayva'(=) 2 im F(x)/L1/a(x)] oF tim, (29/10/03) ae The 4c Rol” dns Tin the definition of [ *f(x) da, it was assumed thatthe interval [a,b] was finite, However, in many applications in physics, economics, and probability we wish to allow a or b (or both) to be 2° or ~20, We must therefore find a way to give mean- ing to symbols like [ 1 ax, h ism ‘These integrals are called improper integrals with infinite limits, ‘One Infinite Limit Consider the function f(x) = xe, It makes perfectly good sense to avk for Jjxe~* dx or J xe-* dx, or indeed for fixe" dx, where bis any positive number. As the table on the next page indicates. as we increase the "upper limit in the definite integral the value of the integral (the area under the ‘curve) increases, but apparently not without bound (in this example, at least). To give meaning to J,~xe* dx we begin by integrating from 0 to an arbitrary upper limit, say b, which Using integration by parts gives [ ets = [set = [ (ed = Now, imagine that the value of 6 marches off to infinity, (See the accompanying table.) As the above calculation shows, if we let b—> 00, the value of the definite integral converges to 1. Thus, it seems natural 10 define be xe de = im (1 — €*