Corrected exercises: strategy for integration (MATH 141)
This is the corrected exercise set for section 7.5 of Stewart, strategy for integration, in MATH 141, Calculus 2, the second calculus course at McGill University.Up to this point each technique came with its own chapter: substitution, integration by parts, trigonometric integrals, trigonometric substitution, partial fractions.
What the set covers, the traps it targets
Up to this point each technique came with its own chapter: substitution, integration by parts, trigonometric integrals, trigonometric substitution, partial fractions.
On the midterm and the final the integrals come without a label, and choosing the technique is now the skill being tested. Every answer below is exact and computed by hand.
The thread running through the set: READ the integrand before choosing anything. Simplify first, with algebra, an identity or a conjugate; then look for a piece whose derivative is sitting next to it; only then name the technique that the new form calls for. Two integrals that look alike can part ways completely, one taking a line, the other a page, and a third having no elementary antiderivative at all.
The traps named in the solutions: substituting without the derivative being present, forgetting to rewrite dx or the bounds, splitting into partial fractions before the long division, clearing one root out of two, writing sin2x=sinx, dividing by the derivative of the inner function, declaring that a non-elementary integral does not exist, and rejecting a correct answer because it does not look like the one in the book.
•Step 1, simplify: expand, split a numerator along the denominator, factor (e2x−1=(ex−1)(ex+1)), use an identity (1+cos2x=2cos2x) or a conjugate.
•Step 2, look for g(x) AND g′(x) as a factor: then u=g(x). Otherwise note the form: product, power of trigonometric functions, a2±x2 or x2−a2, rational function, root of x or of ex.
•Step 3, the technique the form calls for: parts for a product (ln, arctan or a polynomial times ex, sin, cos), trigonometric substitution for a2−x2, long division then partial fractions for a rational function, t=x1/n with n the LCM of the root indices.
•Step 4, if the first try stalls, try again: substitution then parts (∫exdx), rationalizing then partial fractions, or a different first move.
•No elementary antiderivative: ex2, e−x2, sin(x2), xex, lnx1. The definite integral still exists and can be bounded.
•Every substitution replaces three things: the inner function, dx and the bounds. Every antiderivative is checked by differentiating it.
Part A: the basics (/50)
Exercise 1: Simplify first: five integrands that stop being hard
5 answers to check
On a MATH 141 exam the integral arrives naked: no section title tells you which technique to use. The first move is therefore not a technique at all, it is a look. Can the integrand be rewritten, by algebra or by one identity, into something that integrates at once?
Each integrand below looks as if it needs a substitution, integration by parts or partial fractions. None of them does. For each one, NAME the simplification in words, then integrate.
a) ∫x(1+x)2dx
b) ∫tanxsinx+secxdx
c) ∫x2+1(x+1)2dx
d) ∫ex+1e2x−1dx
e) ∫1+cos2xdx, then explain why the substitution u=1+cos2x fails here.
Show the solution
Answers
a)lnx+4x+x+C
b)sinx+ln∣cscx−cotx∣+C
c)x+ln(x2+1)+C
d)ex−x+C
e)21tanx+C; du=−2sin2xdx is not in the integrand.
a)
Simplification: EXPAND the square and divide term by term. (1+x)2=1+2x+x, so the integrand is x−1+2x−1/2+1. Then ∫(x−1+2x−1/2+1)dx=lnx+4x+x+C. The absolute value in ln∣x∣ can be dropped because x already forces x>0. Check by differentiating: x1+x2+1, which is the expanded integrand. The substitution u=1+x also works, but it gives x=(u−1)2, dx=2(u−1)du and the new integrand u−12u2, which then needs a long division: three extra lines for the same answer, and three extra places to drop a sign.
b)
Simplification: rewrite EVERYTHING in sines and cosines. tanxsinx=sinx⋅sinxcosx=cosx and tanxsecx=cosx1⋅sinxcosx=sinx1=cscx. So the integral is ∫(cosx+cscx)dx=sinx+ln∣cscx−cotx∣+C, where ∫cscxdx=ln∣cscx−cotx∣+C is the companion of ∫secxdx. Check: dxdln∣cscx−cotx∣=cscx−cotx−cscxcotx+csc2x=cscx. The trap is to see tanx and reach for u=tanx: its derivative sec2x is nowhere in the integrand, and the attempt goes in circles.
c)
Simplification: split the numerator along the denominator. (x+1)2=x2+2x+1=(x2+1)+2x, so x2+1(x+1)2=1+x2+12x. The second piece is of the form g(x)g′(x) with g(x)=x2+1. Hence ∫x2+1(x+1)2dx=x+ln(x2+1)+C, with no absolute value since x2+1>0. The tempting substitution u=x+1 turns the denominator into u2−2u+2, which is worse. Partial fractions would start with a long division, since numerator and denominator have the same degree, and that division is exactly the one line written here.
d)
Simplification: a DIFFERENCE OF SQUARES. e2x−1=(ex)2−1=(ex−1)(ex+1), so the fraction reduces to ex−1 and ∫(ex−1)dx=ex−x+C. The route u=ex, dx=udu gives ∫u(u+1)u2−1du=∫uu−1du=u−lnu+C, the same answer after the same cancellation, reached later. Whenever e2x sits next to ex, ask first whether it factors.
e)
Simplification: the double-angle identity cos2x=2cos2x−1 gives 1+cos2x=2cos2x. So ∫1+cos2xdx=21∫sec2xdx=21tanx+C, on any interval where cosx=0. The substitution u=1+cos2x fails because it needs du=−2sin2xdx in the numerator, and the numerator is 1: a substitution is not a change of name, it demands that the derivative of the new variable be present, up to a constant factor. That test, is the derivative there, is the one to run before every substitution in this chapter.
Points are earned before the solution; after it, you can keep practising.
For you:
Tick the exercises you have done or want to review: a free account, no password, keeps your ticks and your right answers from one visit to the next, tells you which chapter to tackle next and lets you ask for the exercise you are missing. Create your space, an email is enough.
Exercise 2: Find the derivative hiding in the integrand
5 answers to check
A substitution u=g(x) works when the integrand contains g(x) AND, up to a constant factor, g′(x). On a naked integral that pair is disguised: x1 is the derivative of lnx, x1 is twice the derivative of x, sinxcosx is half the derivative of sin2x.
In each integral below, find the pair, substitute, and then recognise the form that is left: an arctangent, a square to complete, a pair of partial fractions.
a) ∫1ex(1+(lnx)2)dx
b) ∫tan2x+4tanx+5sec2xdx
c) ∫13x(1+x)dx
d) ∫sin2x−3sinx+2cosxdx
e) ∫0π/21+sin4xsinxcosxdx. Say why u=sin2x is a better choice than u=sinx.
Show the solution
Answers
a)4π
b)arctan(tanx+2)+C
c)6π
d)ln(1−sinx2−sinx)+C
e)8π; u=sin2x reaches 1+u2du in one step.
a)
The pair: u=lnx, du=xdx, and the factor x1 is sitting in the denominator. Bounds: x=1 gives u=0, x=e gives u=1. Then ∫011+u2du=arctan1−arctan0=4π. Changing the bounds at the moment of the substitution means never returning to x. The trap is to read 1+(lnx)2 as the thing to substitute: u=1+(lnx)2 needs x2lnx in the numerator, and there is no lnx upstairs.
b)
The pair: u=tanx, du=sec2xdx, which is exactly the numerator. Seeing powers of tanx and a sec2x in the same integrand is the signal, and it has nothing to do with the trigonometric-integral recipes. The integral becomes ∫u2+4u+5du. The quadratic has discriminant 16−20=−4<0, so it does not factor: COMPLETE THE SQUARE, u2+4u+5=(u+2)2+1. Then ∫(u+2)2+1du=arctan(u+2)+C=arctan(tanx+2)+C. The classic wrong answer is ln(tan2x+4tanx+5): its derivative is tan2x+4tanx+5(2tanx+4)sec2x, and the numerator here is only sec2x. A logarithm needs the derivative of the WHOLE denominator on top.
c)
The pair: u=x, du=2xdx, so xdx=2du, and x=u2. Bounds: 1↦1 and 3↦3. Then ∫131+u22du=2(arctan3−arctan1)=2(3π−4π)=6π. Partial fractions cannot even start on x(1+x)1, which is not a rational function; the substitution is what makes it one, and here the rational function it produces is already an arctangent.
d)
The pair: u=sinx, du=cosxdx. The integral becomes ∫u2−3u+2du=∫(u−1)(u−2)du, a rational function with two distinct linear factors: partial fractions. Write (u−1)(u−2)1=u−1A+u−2B. Covering up: u=1 gives A=1−21=−1, u=2 gives B=2−11=1. So ∫=−ln∣u−1∣+ln∣u−2∣+C=lnsinx−1sinx−2+C. Since sinx−2<0 and sinx−1<0 wherever the integrand is defined, the quotient is positive and the answer reads ln(1−sinx2−sinx)+C. Two techniques, chained: the substitution produced the rational function, partial fractions finished it.
e)
With u=sin2x: du=2sinxcosxdx and sin4x=u2. Bounds: 0↦0, 2π↦1. So the integral is 21∫011+u2du=21⋅4π=8π. With u=sinx instead, one gets ∫011+u4udu, which still needs a second substitution w=u2 to reach the same arctangent. Both are correct; the better choice reads sin4x as (sin2x)2 and spots that sinxcosx is half the derivative of sin2x. Plausibility: the integrand lies between 0 and 21 on an interval of length 2π, and 8π≈0.39 is well inside.
Points are earned before the solution; after it, you can keep practising.
For you:
Exercise 3: Substitute first, then integrate by parts
4 answers to check
When a function is COMPOSED with an inner function (x, x2, sinx), integration by parts attempted directly makes things worse: differentiating ex drags the inner derivative along. The order that works is the reverse. Substitute the inner function first and rewrite dx completely; the new integral is a product that parts can finish.
The letter t is used for the substitution, so that u and dv stay free for the parts.
a) ∫01exdx. Check that the value lies between the smallest and largest value of the integrand times the length of the interval.
b) ∫0π/6sin2xesinxdx. Say which identity makes the substitution visible.
c) ∫0πx3cos(x2)dx. Explain the sign of the result.
d) A student writes ∫exdx=∫etdt=ex+C. Prove the answer wrong by differentiating it, and name the error.
e) Show what happens in a) if one integrates by parts FIRST, with u=ex and dv=dx.
Show the solution
Answers
a)2, antiderivative 2(x−1)ex; 1≤2≤e
b)2−e, via sin2x=2sinxcosx
c)−1; negative because cos(x2)<0 where x3 is largest
d)dxdex=2xex: the dx=2tdt was forgotten.
e)xex−21∫xexdx: harder than the start.
a)
Substitution: t=x, so x=t2 and dx=2tdt; bounds 0↦0, 1↦1. The integral becomes ∫012tetdt. Parts on this product: u=t (it simplifies when differentiated), dv=etdt, so du=dt, v=et, and ∫tetdt=tet−et=(t−1)et. Hence ∫01exdx=2[(t−1)et]01=2(0−(−1))=2. In terms of x, the antiderivative is 2(x−1)ex+C. Check: on [0,1] the integrand runs from e0=1 to e1=e and the interval has length 1, so the value must lie between 1 and e≈2.72. It does.
b)
As written, nothing pairs up: the exponent is sinx, whose derivative cosx does not appear. The double-angle identity sin2x=2sinxcosx makes it appear: the integrand is 2sinx⋅esinxcosx. Now t=sinx, dt=cosxdx, bounds 0↦0 and 6π↦21, and the integral becomes ∫01/22tetdt, the same product as in a). So the value is 2[(t−1)et]01/2=2(−21e1/2+1)=2−e. Three moves in a row, an identity, a substitution and integration by parts, and none of them is visible on the original integrand. Plausibility: e≈1.65 gives about 0.35, and the integrand stays below sin3πe1/2≈1.43 on an interval of length 6π≈0.52, so the value is below 0.75.
c)
Write x3dx=x2⋅xdx: the factor x is half the derivative of x2. So t=x2, dt=2xdx, bounds 0↦0 and π↦π, and the integral becomes 21∫0πtcostdt. Parts with u=t, dv=costdt, v=sint: ∫tcostdt=tsint+cost. So the value is 21[tsint+cost]0π=21((0−1)−(0+1))=−1. Sign: cos(x2)>0 only while x2<2π, and on the rest of the interval, where x3 is largest, it is negative; a negative total is expected. Integrating by parts first with u=x3 would need an antiderivative of cos(x2), which does not exist in elementary form: the substitution is not a convenience here, it is the only way in.
d)
dxdex=ex⋅2x1 by the chain rule, which is NOT ex. The error: the student renamed x as t but left dx as if it were dt. A substitution replaces three things, the inner function, the differential and (for a definite integral) the bounds. Here dx=2tdt, and that factor 2t is precisely what makes part a) an integration by parts. Differentiating the final answer takes ten seconds and catches this every time.
e)
With u=ex and dv=dx: du=2xexdx and v=x, so ∫exdx=xex−∫x⋅2xexdx=xex−21∫xexdx. The new integral is harder than the one we started with, and a second round of parts makes it worse again. The rule that comes out: when the difficulty is a composition, attack the inner function first; parts is for PRODUCTS, and the substitution is what turns the composition into one.
Points are earned before the solution; after it, you can keep practising.
For you:
Exercise 4: Rationalize, divide, then split into partial fractions
6 answers to check
Partial fractions only apply to a RATIONAL function, a quotient of two polynomials. A root or an exponential in the integrand blocks them, until a substitution removes it. The substitution is only the first of up to three moves: after it, compare the degrees, divide if the numerator is not of lower degree, and only then split.
No calculator: logarithms stay as logarithms, and plausibility checks use ln2≈0.69, ln3≈1.10, ln5≈1.61.
a) ∫916x−x−2xdx
b) ∫164x+3xdx. Choose the substitution that removes BOTH roots at once.
c) ∫0ln2e2x+5ex+6e3xdx
d) A student starts b) with t=x, and another starts c) by writing ∫(t+2)(t+3)t3dt after t=ex. What goes wrong in each case?
e) Without a calculator, check that the value found in a) is plausible, using the fact that the integrand decreases on [9,16].
Show the solution
Answers
a)9+4ln2+32ln5
b)11+6ln32
c)1+4ln34−9ln45
d)t=x leaves t2/3; after t=ex the dx=tdt was forgotten.
e)11.2≤9+4ln2+32ln5≈12.8≤15.75
a)
Rationalizing substitution: t=x, x=t2, dx=2tdt; bounds 9↦3, 16↦4. The integral becomes ∫34t2−t−2t2⋅2tdt=∫34(t−2)(t+1)2t3dt, and the denominator does not vanish on [3,4]. Degrees: 3 over 2, so DIVIDE first: 2t3=(t2−t−2)(2t+2)+(6t+4), that is t2−t−22t3=2t+2+(t−2)(t+1)6t+4. Then split: (t−2)(t+1)6t+4=t−2A+t+1B with A=316 (cover up at t=2) and B=−3−2=32 (at t=−1). So the value is [t2+2t+316ln(t−2)+32ln(t+1)]34=(24−15)+316ln2+32ln45=9+4ln2+32ln5, using 32ln45=32ln5−34ln2. Three techniques in a row, and skipping the division is the classic slip: the right side of a decomposition never produces the polynomial part 2t+2.
b)
The roots have indices 2 and 3, and their least common multiple is 6: set t=x1/6, so x=t6, x=t3, 3x=t2 and dx=6t5dt. Bounds: 1↦1 and 64↦2, since 26=64. The integral becomes ∫12t3+t26t5dt=6∫12t+1t3dt after cancelling t2. Degree 3 over degree 1: long division gives t+1t3=t2−t+1−t+11. So the value is [2t3−3t2+6t−6ln(t+1)]12=(16−12+12−6ln3)−(2−3+6−6ln2)=11−6ln3+6ln2=11+6ln32. Plausibility: 6ln32≈6(0.69−1.10)≈−2.4, so about 8.6; the integrand falls from 21 to 121 over an interval of length 63, so the value lies between 5.25 and 31.5. It does.
c)
Substitution t=ex, dx=tdt, bounds 0↦1 and ln2↦2. The integral becomes ∫12t2+5t+6t3⋅tdt=∫12(t+2)(t+3)t2dt. Degrees 2 and 2: divide, t2+5t+6t2=1−(t+2)(t+3)5t+6. Split: (t+2)(t+3)5t+6=t+2A+t+3B with A=1−4=−4 and B=−1−9=9. So the integrand is 1+t+24−t+39 and the value is [t+4ln(t+2)−9ln(t+3)]12=1+4ln34−9ln45. Plausibility: the integrand grows from 121 at x=0 to 208=52 at x=ln2, so the value lies between 12ln2≈0.06 and 52ln2≈0.28. The exact value is a small difference of larger terms, about 0.14, and two-decimal logarithms are too coarse to confirm it; the check that protects the answer here is on the decomposition itself, for instance at t=0: 1+24−39=0, which is indeed 2⋅302.
d)
With t=x in b), 3x=t2/3: the integrand is still not rational, and the substitution has bought nothing. The LCM of ALL the indices is the one that clears every root. In c), the factor t1 coming from dx=tdt was dropped: the student is integrating a different function, of degree 3 over 2, which needs a longer division and gives a wrong number. Writing dx in terms of dt on its own line before substituting is the gesture that prevents both.
e)
On [9,16] the integrand falls from 9−3−29=49 at x=9 to 16−4−216=58 at x=16. The interval has length 7, so the integral lies between 7⋅58=11.2 and 7⋅49=15.75. And 9+4ln2+32ln5≈9+2.77+1.07=12.84, inside the bracket. Dropping the polynomial part 2t+2 of the division, a very common slip, would remove the 9 and give about 3.8, far outside.
Points are earned before the solution; after it, you can keep practising.
For you:
Exercise 5: Three families of lookalikes: same shape, different tools
10 answers to check
The integrals below come in three families of three. Inside a family they look almost identical, and they call for different tools, or for none at all. This is the gesture the chapter trains: before writing anything, say in one line what the first move is and WHY.
For each integral: name the first move in one line, then evaluate it exactly. You may use the following fact, proved in more advanced courses: ex2 has no elementary antiderivative. The figure shows the region measured by the last integral of family b).
a) ∫01xex2dx, ∫01x3ex2dx and ∫01ex2dx (for the last one, give the best bounds you can from the monotonicity of the integrand).
b) ∫014−x2xdx, ∫014−x2dx and ∫014−x2dx.
c) ∫1exlnxdx, ∫1ex2lnxdx and ∫ee2xlnxdx.
d) State the single question that separates the three members of family a), and apply it to family c).
Show the solution
Answers
a)2e−1; 21; no elementary antiderivative, 1≤∫01ex2dx≤e
b)2−3; 6π; 3π+23
c)21; 1−e2; ln2
d)Is the derivative of the inner function a factor of the integrand, and what is left over?
a)
First: u=x2, because du=2xdx and the factor x is present. ∫01xex2dx=21∫01eudu=2e−1. Second: the same u=x2, writing x3dx=x2⋅xdx=u⋅2du; what is left, 21∫01ueudu, is a product, so integrate by parts, differentiating the polynomial factor u and integrating eu: ∫ueudu=ueu−∫eudu=(u−1)eu, and the value is 21[(u−1)eu]01=21(0+1)=21. Third: no factor x, so no substitution, and parts only trades it for something worse. The antiderivative is not elementary, yet the integral EXISTS, since ex2 is continuous on [0,1]. The integrand increases from 1 to e there, so 1≤∫01ex2dx≤e. Exercise 7 sharpens this bracket.
b)
First: u=4−x2, du=−2xdx, and the factor x is present: ∫014−x2xdx=[−4−x2]01=−3+2=2−3. Second: no factor x, but the form a2−x21 is in the table: [arcsin2x]01=arcsin21=6π. Third: the root is now in the NUMERATOR, neither trick applies, and the tool is the trigonometric substitution x=2sinθ. Faster still is the GEOMETRY of the figure: the region under the circle x2+y2=4 between x=0 and x=1 splits into a triangle T with vertices (0,0), (1,0), (1,3), of area 23, and a sector S of radius 2 between the directions of 60∘ and 90∘, of angle 6π and area 21⋅4⋅6π=3π. Total 3π+23, which the substitution confirms: [2x4−x2+2arcsin2x]01=23+3π.
c)
First: u=lnx, du=xdx: ∫01udu=21. Second: the x1 is there, but a second x1 is left over, and it is not a function of lnx unless one writes it e−u. The direct route is parts, with u=lnx (the factor that simplifies when differentiated) and dv=x2dx, v=−x1: [−xlnx]1e+∫1ex2dx=−e1+(1−e1)=1−e2. Third: u=lnx again, bounds e↦1 and e2↦2: ∫12udu=ln2. The first and third are both substitutions, but the first gives a power of u and the third a logarithm: once substituted, the form decides again.
d)
The question: what is the derivative of the piece that makes the integrand hard, and is it a FACTOR of the integrand, with what left over? In family a) the hard piece is x2 inside the exponential, with derivative 2x. Exactly present: substitution alone. Present with x2 left over: substitution, then parts. Absent: no elementary route. In family c) the hard piece is lnx, with derivative x1: present in the first and the third, so substitution; present in the second too, but with a second x1 left over that the substitution does not absorb, so parts. Asking this question before choosing is what the marker rewards: the first line of a correct solution names the choice.
Points are earned before the solution; after it, you can keep practising.
For you:
Part B: problems and reasoning (/50)
Exercise 6: One integral, three routes: find the short one
5 answers to check
The integral I=∫01/21+x1−xdx can be attacked in at least three ways, and they do not cost the same. The shaded region of the figure is the area that I measures; the two dashed boxes are there for part e).
All three routes must give the same number. The point of the exercise is to see why one of them takes four lines and another a full page.
a) Route 1, the conjugate: multiply the numerator and the denominator under the root by 1−x. Justify every step (in particular what happens to (1−x)2 on the interval), then evaluate I.
b) Route 2, a trigonometric substitution: set x=sinθ. Show that the integrand times dx becomes (1−sinθ)dθ, then evaluate I again.
c) Route 3, the rationalizing substitution t=1+x1−x. Express x and dx in terms of t, write the new integral with its bounds, and explain why you would not choose this route on an exam.
d) Check the antiderivative found in a) by differentiating it.
e) Use the two dashed boxes of the figure to bracket I, and check your exact value against the bracket without a calculator.
Show the solution
Answers
a)I=6π+23−1, antiderivative arcsinx+1−x2
b)[θ+cosθ]0π/6=6π+23−1
c)I=∫1/31(1+t2)24t2dt: a repeated irreducible quadratic, then a second substitution.
d)1−x21−1−x2x=1−x21−x=1+x1−x
e)231≈0.29≤I≈0.39≤0.5
a)
On [0,21] both 1−x and 1+x are positive. Multiply under the root by 1−x1−x: 1+x1−x=(1+x)(1−x)(1−x)2=1−x2(1−x)2=1−x2∣1−x∣=1−x21−x, the absolute value dropping because 1−x>0. Now SPLIT the numerator: 1−x21−1−x2x. The first piece is in the table, arcsinx; the second is the substitution u=1−x2 of exercise 5, with antiderivative −1−x2. So ∫1+x1−xdx=arcsinx+1−x2+C and I=(6π+23)−(0+1)=6π+23−1. Four lines, and every one of them is a simplification, not a technique.
b)
With x=sinθ, θ∈[0,6π], dx=cosθdθ. Then 1+sinθ1−sinθ=(1+sinθ)(1−sinθ)(1−sinθ)2=cos2θ(1−sinθ)2, whose square root is cosθ1−sinθ because 1−sinθ≥0 and cosθ>0 on this interval. Multiplying by dx=cosθdθ leaves (1−sinθ)dθ. Bounds: x=0↦θ=0, x=21↦θ=6π. So I=[θ+cosθ]0π/6=6π+23−1. Same value, and in fact the same computation: the conjugate of route 1 is hidden inside the identity 1−sin2θ=cos2θ. Route 2 is a fine choice when the conjugate does not come to mind.
c)
From t2=1+x1−x: t2+t2x=1−x, so x=1+t21−t2 and, by the quotient rule, dx=(1+t2)2−2t(1+t2)−(1−t2)2tdt=(1+t2)2−4tdt. Bounds: x=0↦t=1, x=21↦t=31. So I=∫11/3t⋅(1+t2)2−4tdt=∫1/31(1+t2)24t2dt. The integrand is rational, but its denominator is a REPEATED irreducible quadratic: partial fractions give 1+t24−(1+t2)24, and the second piece needs yet another substitution, t=tanφ. It works, and it is a page long. The lesson is not that rationalizing is wrong, it is that it should be the LAST resort once simplification and the standard forms have been tried.
d)
dxd(arcsinx+1−x2)=1−x21+21−x2−2x=1−x21−x. And for −1<x<1, 1−x21−x=1−x1+x1−x1−x=1+x1−x. The antiderivative is confirmed, on the whole interval (−1,1) and not only on [0,21].
e)
The integrand decreases on [0,21]: its value at 0 is 1 and at 21 it is 3/21/2=31. The region therefore sits above the lower box of height 31 and below the upper box of height 1, both of width 21: 231≤I≤21, that is about 0.29≤I≤0.5. With π≈3.14 and 3≈1.73: I≈0.523+0.865−1=0.388. Inside the bracket. A sign slip in route 1, giving 6π−23+1≈0.66, would land outside it.
Points are earned before the solution; after it, you can keep practising.
For you:
Exercise 7: No elementary antiderivative: recognise it, then work around it
11 answers to check
Some integrands have no antiderivative that can be written with the usual functions (powers, roots, exponentials, logarithms, trigonometric functions and their inverses, combined by the four operations and composition). You may use without proof that this is the case for ex2, e−x2, sin(x2), xex and lnx1.
Recognising such an integral saves the time you would lose trying every technique on it. And it does not end the question: the definite integral still exists, it can be bounded, and sometimes two non-elementary pieces cancel. The figure shows ex2 on [0,1] with two simpler curves.
a) Which of these have an elementary antiderivative? Give it when it exists.(i)∫xsin(x2)dx(ii)∫sin(x2)dx(iii)∫xexdx(iv)∫x2(x−1)exdx(v)∫lnxdx(vi)∫x2e1/xdx
b) Integrate by parts ONE of the two terms of ∫(1+2x2)ex2dx to show that the non-elementary pieces cancel. Deduce ∫01(1+2x2)ex2dx.
c) Same idea: evaluate ∫2e(lnx1−(lnx)21)dx and explain the sign of the result.
d) Using eu≥1+u for all u, and the fact that on [0,1] the convex function eu lies below its chord 1+(e−1)u, show that 34≤∫01ex2dx≤3e+2.
e) A student concludes from (iii) that ∫12xexdx does not exist. Correct him, and compute dxd∫1x2tetdt.
(i) The factor x is half the derivative of x2: u=x2 gives −21cos(x2)+C. (ii) The same composition without the factor x: no elementary antiderivative, and no technique will find one. (iii) None, it is on the list. (iv) It LOOKS like (iii) made worse, yet dxdxex=x2xex−ex=x2(x−1)ex: the antiderivative is xex+C. Splitting it into xex−x2ex would produce two non-elementary pieces that happen to combine, which is exactly the phenomenon of b). (v) None, on the list. (vi) u=x1, du=−x2dx: −e1/x+C. The test is again the derivative of the inner function: x2 has derivative 2x, present in (i) and absent in (ii); x1 has derivative −x21, present in (vi).
b)
Split: ∫(1+2x2)ex2dx=∫ex2dx+∫2x2ex2dx. In the second, write 2x2ex2=x⋅2xex2 and integrate by parts with u=x, dv=2xex2dx, so du=dx and v=ex2: ∫2x2ex2dx=xex2−∫ex2dx. Adding, the two copies of ∫ex2dx cancel: ∫(1+2x2)ex2dx=xex2+C. Check: dxdxex2=ex2+2x2ex2. So ∫01(1+2x2)ex2dx=[xex2]01=e. Never write ∫ex2dx and stop: carry it as an unevaluated symbol, it may cancel.
c)
Integrate ∫lnxdx by parts with u=lnx1, dv=dx: du=−(lnx)21⋅x1dx and v=x, so ∫lnxdx=lnxx+∫(lnx)2dx. Subtracting ∫(lnx)2dx from both sides: ∫(lnx1−(lnx)21)dx=lnxx+C. The interval [2,e] avoids x=1, where lnx=0. Value: lnee−ln22=e−ln22. Sign: on [2,e), 0<lnx<1, so (lnx)2<lnx and lnx1<(lnx)21; the integrand is negative and so is the value, e−ln22≈2.72−0.692≈−0.17.
d)
Lower bound: with u=x2, ex2≥1+x2, the dashed curve of the figure, so ∫01ex2dx≥∫01(1+x2)dx=1+31=34. Upper bound: for x∈[0,1], u=x2∈[0,1] and ex2≤1+(e−1)x2, the chord curve, which meets ex2 at x=0 and x=1. So ∫01ex2dx≤1+3e−1=3e+2≈1.57. The bracket [1.33,1.57] is much sharper than the crude [1,e] of exercise 5, and it uses nothing but comparison of integrals: this is what can still be said about a non-elementary integral in MATH 141 without series, which come later in the course.
e)
The integrand xex is continuous on [1,2], so the definite integral EXISTS: it is a perfectly good number, the area under the curve. What does not exist is a closed-form expression for it. The function F(x)=∫1xtetdt is well defined for x>0 and, by the Fundamental Theorem, F′(x)=xex. With the chain rule, dxd∫1x2tetdt=x2ex2⋅2x=x2ex2. Not elementary never means not integrable.
Points are earned before the solution; after it, you can keep practising.
For you:
Exercise 8: Five statements to correct
5 answers to check
Each statement below was written on a real exam paper and each one is false, or reaches a false conclusion. For each: say precisely what is wrong, give the correct result, and name the check that would have caught the error before the paper was handed in.
a) ∫x4+1xdx=41ln(x4+1)+C, by the substitution u=x4+1.
b) ∫ex2dx=2xex2+C, since the derivative of x2 is 2x.
c) ∫02π1−cos2xdx=∫02π2sin2xdx=2∫02πsinxdx=0.
d) With t=x: ∫04exdx=∫042tetdt=2[(t−1)et]04=6e4+2.
e) ∫x4+1x3dx needs partial fractions, since the denominator has degree 4: first factor x4+1=(x2+2x+1)(x2−2x+1).
d)The bounds become 0 and 2: the value is 2(e2+1).
e)The numerator is a quarter of the derivative of the denominator: 41ln(x4+1)+C.
a)
The substitution u=x4+1 requires du=4x3dx, and the integrand only carries xdx: the factor x2 is missing, so the logarithm form ∫gg′=ln∣g∣ does not apply. The piece whose derivative IS present is x2, since xdx is half of d(x2), and x4=(x2)2. With u=x2, du=2xdx, the integral becomes 21∫u2+1du=21arctanu+C=21arctan(x2)+C. The check: differentiate the proposed answer. dxd41ln(x4+1)=x4+1x3, not the integrand; that answer belongs to the integral of e), a lookalike that differs by the power in the numerator.
b)
Dividing by the derivative of the inner function is not a rule of integration: it only works when that derivative is a CONSTANT, as in ∫e3xdx=3e3x. Differentiating the proposed answer by the quotient rule gives 4x22x⋅2xex2−2ex2=ex2−2x2ex2, which is not ex2. In fact ex2 has no elementary antiderivative at all: the correct response on an exam is to say so, and, if a definite integral is asked, to bound it (exercise 7).
c)
1−cos2x=2sin2x is right, but sin2x=∣sinx∣, not sinx. On [π,2π], sinx≤0, and the error flips the sign of the whole second half. Correct computation: 2∫02π∣sinx∣dx=2(∫0πsinxdx−∫π2πsinxdx)=2(2+2)=42. The check that catches it is immediate: the integrand is a square root, so it is never negative, and it is not identically zero, so its integral must be POSITIVE. The figure of the solution shows the two humps of 2∣sinx∣ against the dashed arch of 2sinx that goes below the axis.
d)
The substitution changed the variable but not the bounds: t=x runs from 0=0 to 4=2, not to 4. Correct: ∫04exdx=∫022tetdt=2[(t−1)et]02=2(e2+1). Check: the integrand runs from 1 to e2≈7.4 on an interval of length 4, so the value is at most 4e2≈29.6; the proposed 6e4+2≈330 is about twenty times too big, whereas 2(e2+1)≈16.8 fits. Write the new bounds on the same line as the substitution, or go back to x before evaluating: never a mixture of the two.
e)
The integral does not NEED partial fractions, and reaching for them first is a costly mistake. Before any factorization, run the test of this chapter: the derivative of the denominator is 4x3, and the numerator is x3, a quarter of it. So u=x4+1, du=4x3dx and ∫x4+1x3dx=41ln(x4+1)+C, one line, no absolute value since x4+1>0. The factorization is correct, but it leads to a decomposition with four unknown constants, two logarithms and two arctangents that must all recombine into this single logarithm: half an hour of work and many chances to slip. Partial fractions are the LAST resort for a rational function, after simplification and after the test for g′ over g. Check: dxd41ln(x4+1)=4(x4+1)4x3.
Points are earned before the solution; after it, you can keep practising.
For you:
Exercise 9: A final exam question: four integrals, no technique announced
4 answers to check
This is the format of the integration question on a MATH 141 final: a list of integrals, the instruction Evaluate, and nothing else. Two marks each for a) to d), and the marks go to the METHOD: a first line that names the move (the substitution with its dx and its bounds, or the choice of u and dv), then the computation.
Work without a calculator. Each answer is an exact number.
a) ∫01ln(1+x2)dx
b) ∫0ln2e2x+1e3xdx
c) ∫eeexln(lnx)dx
d) ∫0ln2ex−1dx
e) Check b) by differentiating its antiderivative written in x, and check d) with the size of its integrand.
Show the solution
Answers
a)ln2−2+2π
b)1+4π−arctan2
c)1
d)2−2π
e)dxd(ex−arctanex)=e2x+1e3x; 0≤2−2π≈0.43≤ln2
a)
First line: integration by parts, u=ln(1+x2) (it simplifies when differentiated, to a rational function) and dv=dx, so du=1+x22xdx and v=x. Then ∫01ln(1+x2)dx=[xln(1+x2)]01−∫011+x22x2dx. The new integrand has degrees 2 over 2: divide, 1+x22x2=2−1+x22. So the value is ln2−[2x−2arctanx]01=ln2−2+2π. Plausibility: the integrand grows from 0 to ln2 on an interval of length 1, and 0.69−2+1.57=0.26 lies between 0 and 0.69. The substitution u=1+x2 is a trap here: its du=2xdx is not in the integrand.
b)
First line: t=ex, dx=tdt, bounds 0↦1 and ln2↦2. Then e2x+1e3xdx=t2+1t3⋅tdt=t2+1t2dt. Degrees are equal, so divide: t2+1t2=1−t2+11. Hence the value is [t−arctant]12=(2−arctan2)−(1−4π)=1+4π−arctan2. The answer contains arctan2, which is not a standard angle: on a no-calculator exam it stays as it is. Leaving e3x unsimplified and trying u=e2x+1 fails: du=2e2xdx leaves an ex that must still be expressed through u.
c)
First line: t=lnx, dt=xdx, bounds e↦1 and ee↦e. The integral becomes ∫1elntdt. Then parts with u=lnt, dv=dt, v=t: ∫lntdt=tlnt−∫t⋅t1dt=tlnt−t. So the value is [tlnt−t]1e=(e−e)−(0−1)=1. Two moves, substitution then parts, and the bounds ee↦e are where most marks are lost: ln(ee)=e, not ee−1 nor 1.
d)
First line: the rationalizing substitution t=ex−1, so ex=t2+1, x=ln(t2+1) and dx=t2+12tdt; bounds 0↦0 and ln2↦2−1=1. The integral becomes ∫01t⋅t2+12tdt=∫01t2+12t2dt. Divide: t2+12t2=2−t2+12. So the value is [2t−2arctant]01=2−2π. Setting t=ex instead leaves t−1: still a root, so the substitution must swallow the WHOLE root.
e)
For b): the antiderivative in x is ex−arctan(ex), and dxd(ex−arctanex)=ex−1+e2xex=1+e2xex(1+e2x)−ex=e2x+1e3x. For d): on [0,ln2] the integrand ex−1 runs from 0 to 2−1=1, so the value lies between 0 and ln2≈0.69; and 2−2π≈2−1.57=0.43 does. Had the division been forgotten, the answer 2ln2≈1.39 (from ∫t2+12t, a very common wrong reading) would exceed the bound.
Points are earned before the solution; after it, you can keep practising.
For you:
Exercise 10: Electric charge from a current: the formula is given, the technique is not
6 answers to check
The charge Q that passes through a cross-section of a wire between times a and b is the integral of the current: Q=∫abI(t)dt. With I in milliamperes and t in seconds, Q is in millicoulombs (mC). A negative current means charge flowing the other way, so the integral gives the NET charge.
Three devices are tested. The integrals are what an engineer meets in practice: nobody tells you which technique applies. The figure shows the current IA(t)=sint of the first device on [0,4π2], with the region between the curve and the axis shaded.
a) Device A: IA(t)=sint. Compute the charge delivered on [0,π2].
b) Still device A, now on [0,4π2]. Compute the net charge, then the total charge that crossed the section in either direction, and explain the difference using the figure.
c) Device B: IB(t)=te−t on [0,4]. Compute the charge exactly, then estimate it using e2≈7.4.
d) Device C: IC(t)=e−t2 on [0,1]. Explain why the charge has no closed form, then show that it lies between 32 mC and 1 mC.
e) Check the antiderivative used in a) by differentiating it with respect to t.
Show the solution
Answers
a)Q=2π≈6.3 mC
b)Net −4π mC; total 8π mC (2π forward, 6π back).
c)Q=12−76e−2≈1.7 mC
d)e−t2 is not elementary; 32≤Q≤1 from 1−t2≤e−t2≤1.
e)dtd2(sint−tcost)=sint
a)
The current is a composition with t, so substitute first: s=t, t=s2, dt=2sds, bounds 0↦0 and π2↦π. Then Q=∫0π2ssinsds, a product: parts with u=s, dv=sinsds, v=−coss, giving ∫ssinsds=−scoss+sins. So Q=2[−scoss+sins]0π=2(π+0)−0=2π mC, about 6.3 mC with π≈3.14. The value is positive, as it must be: on [0,π2], t∈[0,π] and sint≥0.
b)
Same antiderivative, bounds 0↦0 and 4π2↦2π: the net charge is 2[−scoss+sins]02π=2(−2π)=−4π mC. On [π2,4π2], t∈[π,2π] and the current is NEGATIVE: that stretch contributes 2[−scoss+sins]π2π=2(−2π−π)=−6π mC, charge flowing back. The net charge adds the two signed contributions, 2π−6π=−4π; the total charge that crossed the section adds their sizes, 2π+6π=8π mC. On the figure the second lobe, below the axis, is much wider than the first because the time needed for t to go from π to 2π is 3π2, three times the length of the first stretch.
c)
Again a composition with t: s=t, t=s2, dt=2sds, bounds 0↦0, 4↦2, so Q=∫02s2e−s⋅2sds=2∫02s3e−sds. Parts three times, conveniently in tabular form: derivatives of s3 are 3s2, 6s, 6, 0, and successive antiderivatives of e−s are −e−s, e−s, −e−s, e−s, with alternating signs. This gives ∫s3e−sds=−e−s(s3+3s2+6s+6). So Q=2[−e−s(s3+3s2+6s+6)]02=2(6−38e−2)=12−76e−2 mC. With e2≈7.4: 7.476≈10.3 and Q≈1.7 mC. Sanity: IB≥0 on [0,4], and IB(4)=4e−2≈0.54 is its largest value there, so Q≤4×0.54≈2.2.
d)
e−t2 has no elementary antiderivative: no substitution applies because the derivative −2t of the exponent is not a factor, and parts only trades it for something worse. The integral still exists, the integrand being continuous. For the bounds: eu≥1+u with u=−t2 gives e−t2≥1−t2, and e−t2≤e0=1 since −t2≤0. Integrating over [0,1]: ∫01(1−t2)dt=32≤Q≤1 mC. The revision sheet of this chapter draws this bracket.
e)
F(t)=2(sint−tcost) is the antiderivative of a) written back in t. By the chain rule, with dtdt=2t1: F′(t)=2(2tcost−2tcost+t⋅2tsint)=sint. The two cosine terms cancel exactly, which is the signature of an integration by parts done right.
Points are earned before the solution; after it, you can keep practising.
Something missing from this set?A type of exercise your teacher gives, a problem you are stuck on, a mistake you spotted: tell me, I read every request.
Sending goes through your free space: that is how I can answer you and tell you when the exercise is added. An email is enough, no password. Create your space.
Rather not create an account? Write to me directly.