This is the corrected exercise set for trigonometric substitution in MATH 141, Calculus 2, the integration course taken at McGill University in science, engineering and the life sciences. It follows section 7.3 of Stewart. Every answer is exact, as on the midterm and the final, where calculators are not allowed: the limits are chosen so that the angles are 6π, 4π or 3π, and the solutions name each substitution, each identity and each change of limits, because that is where the marks are.
The thread running through the whole set: a trigonometric substitution is a round trip. The FORM under the root chooses the identity, a2−x2 for the sine, a2+x2 for the tangent, x2−a2 for the secant. The INTERVAL of θ fixes the sign of the root once the square comes out. And the answer comes BACK to x through the right triangle, or never leaves θ because the limits were changed. Every mark lost in this chapter is a missing leg of that trip.
The traps named explicitly in the solutions: choosing the substitution from the letters instead of the sign pattern, forgetting dx, keeping the old limits after the variable changed, writing tanθ for the root on the negative branch of the secant, reading cosθ=9−x2 without the hypotenuse, writing sin2θ=32x, losing the minus sign when completing the square of 3+2x−x2, and reaching for a triangle when u=x2+a2 did the job in two lines.
•x2−a2: x=asecθ, dx=asecθtanθdθ, root =a∣tanθ∣: equal to atanθ only where tanθ≥0.
•Back to x: draw the triangle of the substitution, read every ratio as a ratio of two sides; sin2θ=2sinθcosθ first. Definite integral: move the limits instead.
•A quadratic with an x term: complete the square, 3+2x−x2=4−(x−1)2, and substitute for x−1.
•An odd power of x outside the root: try u= radicand before any triangle.
•Half-angle: cos2θ=21+cos2θ, sin2θ=21−cos2θ. Area of an ellipse: πab.
Part A: the basics (/50)
Exercise 1: Reading the form under the root: substitution, dx and the sign
A trigonometric substitution replaces x by a trigonometric function of a new angle θ, chosen so that a Pythagorean identity collapses the expression under the root. Which identity, and therefore which substitution, is dictated by the FORM of the radicand, not by the rest of the integrand.
The interval allowed for θ is not decoration: it is what decides the SIGN of the root once the square comes out, and a marker checks that you said so.
a) For each of 25−x2, x2+9 and x2−16, give the substitution, dx, an interval for θ, and the radical written without a root, justifying its sign.
b) Compute ∫x225−x2dx, and return to x with a right triangle.
c) Check your answer to b) by differentiating it.
d) A classmate does b) with x=5cosθ, 0≤θ≤π. Is that allowed? Carry it through and compare.
e) For which of the three radicands of a) does one interval of θ fail to cover the whole domain in x? What does that mean for a definite integral?
d)Allowed: sinθ≥0 on [0,π], and the same answer comes out.
e)x2−16: domain x≤−4 or x≥4, two pieces, two intervals of θ; the sign of tanθ must be checked on each.
a) Under 25−x2 a constant MINUS x2: the identity 1−sin2θ=cos2θ collapses it, so x=5sinθ with −2π≤θ≤2π, dx=5cosθdθ and 25−25sin2θ=5∣cosθ∣=5cosθ, because cosθ≥0 on that interval. Under x2+9 a SUM: 1+tan2θ=sec2θ, so x=3tanθ with −2π<θ<2π, dx=3sec2θdθ and 9tan2θ+9=3∣secθ∣=3secθ, since secθ>0 there. Under x2−16x2 MINUS a constant: sec2θ−1=tan2θ, so x=4secθ, dx=4secθtanθdθ and 16sec2θ−16=4∣tanθ∣, equal to 4tanθ on 0≤θ<2π (for x≥4) and on π≤θ<23π (for x≤−4), the two intervals of Stewart's convention, where tanθ≥0. The rule to memorise is the sign pattern, not the letters: which term is subtracted decides the identity.
b) The form is a2−x2 with a=5, and there is no odd power of x outside the root to hand to a plain substitution. So x=5sinθ, dx=5cosθdθ, 25−x2=5cosθ, and ∫25sin2θ⋅5cosθ5cosθdθ=251∫csc2θdθ=−251cotθ+C. To return to x, draw the triangle that sinθ=5x describes: opposite side x, hypotenuse 5, adjacent side 25−x2 by Pythagoras (figure of the solution, left). Then cotθ=oppositeadjacent=x25−x2, and ∫x225−x2dx=−25x25−x2+C. Leaving cotθ, or writing cot(arcsin5x), costs the return mark: the question is in x.
c) Write F(x)=−251⋅x25−x2. By the quotient rule, dxdx25−x2=x225−x2−x⋅x−25−x2=x225−x2−x2−(25−x2)=x225−x2−25. Multiplying by −251 gives F′(x)=x225−x21, the integrand. On an exam without a calculator this check costs one minute and is the only proof you have that the triangle was read correctly.
d) Yes. The identity 1−cos2θ=sin2θ works just as well, and on 0≤θ≤π the sine is non-negative, so 25−x2=5sinθ. Now dx=−5sinθdθ, note the MINUS, and the integral becomes ∫25cos2θ⋅5sinθ−5sinθdθ=−251∫sec2θdθ=−251tanθ+C. The triangle of cosθ=5x has x ADJACENT and 25−x2 opposite, so tanθ=x25−x2 and the answer is again −25x25−x2+C. The cosine choice is legal but carries two extra traps, the sign of dx and a triangle drawn the other way round; the sine is the default for a reason.
e) The domain of 25−x2 is [−5,5], swept once by 5sinθ on [−2π,2π]; that of x2+9 is every real, swept once by 3tanθ on (−2π,2π). But x2−16 lives on TWO pieces, x≥4 and x≤−4, and x=4secθ needs one interval of θ for each. For a definite integral on the negative piece, the new limits must be read in the interval chosen for that piece, and the sign of tanθ must be checked there: with the principal choice 2π<θ≤π the tangent is NEGATIVE and the root is −4tanθ. Exercise 4 shows what forgetting this costs.
Exercise 2: The circle behind the root of 9 minus x squared
The integral ∫9−x2dx is the model of the chapter: the substitution rule of section 5.5 has nothing to grip, since no factor x stands outside the root to absorb du=−2xdx. Its graph is the upper half of the circle x2+y2=9, and the figure shades the area ∫029−x2dx, up to the point P(2,5).
a) With x=3sinθ, write the integral in θ and compute it with a half-angle formula.
b) Return to x with a right triangle. Write sin2θ in terms of x.
c) Differentiate your antiderivative to check it.
d) Evaluate ∫029−x2dx exactly. Then cut the shaded region into a triangle and a circular sector, and show that each piece is one term of your answer. Bracket the value without a calculator.
e) Deduce ∫−339−x2dx and ∫039−x2dx from your antiderivative, and say which geometric fact each confirms.
Show the solution
Answers
a)∫9cos2θdθ=29θ+49sin2θ+C
b)sin2θ=92x9−x2, so ∫9−x2dx=29arcsin3x+2x9−x2+C
c)The derivative simplifies to 9−x29−x2=9−x2.
d)29arcsin32+5: sector 29arcsin32 plus triangle 5; between 25 and 6.
e)29π, half the disc; 49π, a quarter of it.
a) The form is a2−x2 with a=3: x=3sinθ, −2π≤θ≤2π, dx=3cosθdθ, and 9−9sin2θ=3∣cosθ∣=3cosθ because the cosine is non-negative on that interval. The integral becomes ∫3cosθ⋅3cosθdθ=9∫cos2θdθ. An even power of the cosine alone calls for the half-angle formula of section 7.2, cos2θ=21+cos2θ: 9∫21+cos2θdθ=29θ+49sin2θ+C. Forgetting dx here, and integrating 3cosθdθ, gives 3sinθ=x: an answer so simple it should raise an alarm.
b) sinθ=3x describes a right triangle with x opposite θ and hypotenuse 3, so the adjacent side is 9−x2 and cosθ=39−x2, with the 3 in the denominator. The term sin2θ cannot be read on the triangle as it stands: expand it first, sin2θ=2sinθcosθ=2⋅3x⋅39−x2=92x9−x2. Then 49sin2θ=2x9−x2, and θ=arcsin3x, which is legitimate because θ was taken in [−2π,2π], the range of arcsine. Final answer: ∫9−x2dx=29arcsin3x+2x9−x2+C. Writing sin2θ=32x is the classic slip: it doubles the angle's sine as if it were linear.
c) dxd[29arcsin3x]=29⋅1−x2/91/3=29⋅9−x21, and by the product rule dxd[2x9−x2]=219−x2−29−x2x2. Over the common denominator 9−x2 the numerator is 29+29−x2−2x2=9−x2, and 9−x29−x2=9−x2. The two terms of the antiderivative are not independent: neither one alone differentiates to anything useful, only their sum does.
d) ∫029−x2dx=[29arcsin3x+2x9−x2]02=29arcsin32+5. Now the geometry. Join the origin O to P(2,5), which is on the circle since 4+5=9. The segment OP cuts the shaded region into the right triangle O, (2,0), P, of area 21⋅2⋅5=5, and the sector between the positive y-axis and OP. That sector has radius 3 and an angle α measured from the vertical with sinα=32, so its area is 21⋅9⋅α=29arcsin32. The antiderivative IS triangle plus sector: θ is the angle at the centre. To bracket: 21<32<22, so 6π<arcsin32<4π, and the value lies between 43π+5≈4.6 and 89π+5≈5.8. That is consistent with the region itself: it contains the rectangle of height 5, area 25≈4.5, and fits in the rectangle of height 3, area 6.
e) ∫−339−x2dx=29(arcsin1−arcsin(−1))+0=29⋅π=29π: the area of the upper half disc, 21π⋅32. And ∫039−x2dx=29⋅2π=49π, the quarter disc. The product term vanishes at both ends because the root does. These two values are the standard sanity check of this antiderivative: if your formula does not give 2πa2 between −a and a, a coefficient is wrong, usually the 21 of the half-angle formula.
Exercise 3: Tangent substitution: a sum of squares and limits that move
When the radicand is a SUM x2+a2, the identity that collapses it is 1+tan2θ=sec2θ: the substitution is x=atanθ with −2π<θ<2π, where secθ>0. A power such as (x2+a2)3/2 is a root in disguise and is handled the same way.
For a definite integral there is a second decision: return to x, or move the limits with the variable and never come back. The second is shorter and is what the solutions below do.
a) Compute ∫x2x2+4dx and return to x with a triangle.
b) Compute ∫02(x2+4)3/2dx with x=2tanθ, changing the limits, without ever returning to x.
c) Show that 4x2+4x is an antiderivative of the integrand of b), and use it to confirm your value.
d) A student does b) with x=2tanθ but keeps the limits 0 and 2 on the θ integral. Why is her upper limit not even an admissible value of θ?
e) Explain why (x2+4)3/2=8sec3θ with no absolute value, and why the answer to b) must be less than 41 before any computation.
Show the solution
Answers
a)−4xx2+4+C
b)41∫0π/4cosθdθ=82
c)Its derivative is (x2+4)3/21, and 482=82.
d)θ=2>2π lies outside (−2π,2π); the limits are 0 and 4π.
e)secθ>0 on (−2π,2π); the integrand is at most 81 on an interval of length 2.
a) The form x2+a2 with a=2 gives x=2tanθ, dx=2sec2θdθ, x2+4=2secθ. So ∫4tan2θ⋅2secθ2sec2θdθ=41∫tan2θsecθdθ. Rewrite in sines and cosines: tan2θsecθ=cosθ1⋅sin2θcos2θ=sin2θcosθ, and with w=sinθ the integral is 41∫w−2dw=−4sinθ1+C. The triangle of tanθ=2x has x opposite, 2 adjacent and hypotenuse x2+4, so sinθ=x2+4x and the answer is −4xx2+4+C. Converting to sines and cosines is the reflex when a quotient of sec and tan does not match a formula of the table.
b) x=2tanθ, dx=2sec2θdθ and (x2+4)3/2=(4sec2θ)3/2=8sec3θ. The integrand becomes 8sec3θ2sec2θ=41cosθ. The limits move with the variable: x=0 gives tanθ=0, so θ=0; x=2 gives tanθ=1, so θ=4π. Then ∫02(x2+4)3/2dx=41∫0π/4cosθdθ=41sin4π=82. No triangle was needed: once the limits are in θ, the computation ends in θ. Writing the new limits next to the substitution, on the same line, is what earns the mark and prevents the error of part d).
c) With G(x)=4x(x2+4)−1/2, the product rule gives G′(x)=41(x2+4)−1/2−4x2(x2+4)−3/2=4(x2+4)3/2(x2+4)−x2=(x2+4)3/21. This is exactly what the triangle would have produced from 41sinθ, since sinθ=x2+4x. Then G(2)−G(0)=482=822=421=82. The two routes agree, and on an exam the route in θ is the one to take: fewer steps, fewer places to slip.
d) Keeping the limits gives 41[sinθ]02=41sin2, a number with no reason to be right. The deeper problem: the substitution was made with −2π<θ<2π, and 2>2π≈1.57. The value θ=2 is not in that interval, and it corresponds to x=2tan2, which is NEGATIVE since the tangent is negative in the second quadrant: the student is integrating over a stretch of x that has nothing to do with [0,2]. The limits of a definite integral always belong to the variable of integration; when the variable changes, they change with it.
e) (4sec2θ)3/2=8(sec2θ)3=8∣secθ∣3, and on −2π<θ<2π the cosine is positive, so secθ>0 and the absolute value drops. That is the reason for the interval, not a formality. For the size: on [0,2], x2+4≥4, so (x2+4)3/2≥8 and the integrand is at most 81; over an interval of length 2 the integral is at most 41. And 82≈81.41≈0.18 satisfies it. A bound like this takes twenty seconds and catches a lost factor of 2 or 4.
Exercise 4: Secant substitution: one integrand, two branches, two signs
For x2−a2 the identity is sec2θ−1=tan2θ and the substitution x=asecθ. The domain now has two pieces, x≥a and x≤−a, and the root a2tan2θ=a∣tanθ∣ equals atanθ only where tanθ≥0.
The figure shows y=xx2−1 on both pieces of its domain, with the region between the curve and the axis shaded over [−2,−1].
a) Compute ∫12xx2−1dx with x=secθ, 0≤θ<2π.
b) Read the sign of ∫−2−1xx2−1dx on the figure, then compute it with x=secθ, π≤θ<23π (the convention of Stewart).
c) Compute it again with the principal choice 2π<θ≤π, where tanθ≤0, and show that the answer is the same only if the absolute value ∣tanθ∣ is written.
d) Confirm b) by a symmetry argument, with no substitution at all.
e) A classmate finds 3−3π for b). Name the error and give a five-second check that catches it.
Show the solution
Answers
a)∫0π/3tan2θdθ=3−3π
b)Negative on the figure; ∫4π/3πtan2θdθ=3π−3
c)x2−1=−tanθ: ∫2π/3π(−tan2θ)dθ=3π−3
d)The integrand is odd, so the integral is −(3−3π).
e)He wrote tan2θ=tanθ where tanθ<0; a positive value for a negative integrand is impossible.
a) x=secθ, dx=secθtanθdθ, and on 0≤θ<2π the tangent is non-negative, so sec2θ−1=tanθ. The limits: x=1 gives cosθ=1, θ=0; x=2 gives cosθ=21, θ=3π. The integrand becomes secθtanθ⋅secθtanθdθ=tan2θdθ=(sec2θ−1)dθ. So the integral is [tanθ−θ]0π/3=3−3π. Positive, as it must be: 3≈1.73 and 3π≈1.05, and the integrand is positive on (1,2]. The identity tan2θ=sec2θ−1 is the only tool needed, and it comes straight from the table of section 7.2.
b) On [−2,−1] the numerator x2−1 is non-negative and the denominator x is negative: the curve is BELOW the axis and the integral is negative, as the shaded region shows. With Stewart's convention, x=−1 gives secθ=−1, so θ=π, and x=−2 gives cosθ=−21 with π≤θ<23π, so θ=34π. On that interval the tangent is non-negative (third quadrant), so again x2−1=tanθ and the integrand is tan2θdθ. The limits run from 34π (for x=−2) to π (for x=−1), in that order: ∫4π/3πtan2θdθ=[tanθ−θ]4π/3π=(0−π)−(3−34π)=3π−3. Negative. The upper limit is smaller than the lower one, and that is fine: the limits are wherever the substitution sends them, never reordered by hand.
c) With 2π<θ≤π: x=−2 gives θ=32π and x=−1 gives θ=π. In the second quadrant tanθ≤0, so tan2θ=∣tanθ∣=−tanθ. The integrand becomes secθ−tanθ⋅secθtanθdθ=−tan2θdθ, and ∫2π/3π(−tan2θ)dθ=−[tanθ−θ]2π/3π=−[(0−π)−(−3−32π)]=−(3−3π)=3π−3. The same number as in b). Both conventions are correct; what is never correct is writing tanθ for the root without looking at the quadrant.
d) Let f(x)=xx2−1. Then f(−x)=−xx2−1=−f(x): f is odd, and the figure shows it, the left branch is the right branch turned half a turn about the origin. The substitution x=−u turns ∫−2−1f(x)dx into −∫12f(u)du=−(3−3π)=3π−3. On an exam this is the fastest route of all, and the right one to use as a CHECK of b) when the secant has been used.
e) He took 2π<θ≤π, the principal range of arcsecant, and wrote x2−1=tanθ as on the positive branch. There tanθ<0, so the root he used is negative, and every value of the integrand has the wrong sign: he obtained exactly the opposite of the answer. The five-second check is the SIGN of the integrand: on [−2,−1] it is negative, so the integral must be negative, and 3−3π>0 is impossible. It costs the whole question, and it is the error the figure was drawn to prevent.
Exercise 5: The area of an ellipse, and a strip of it
The ellipse 16x2+9y2=1 has semi-axes 4 and 3. Solving for y, its upper half is the graph of y=4316−x2 and its lower half that of y=−4316−x2.
The figure shades the part of the ellipse between the vertical lines x=0 and x=2.
a) Explain why the area enclosed by the ellipse is 3∫0416−x2dx.
b) Compute this area with x=4sinθ, changing the limits.
c) Compute the area of the shaded strip exactly, and check its size against two rectangles.
d) Redo b) for the ellipse a2x2+b2y2=1, a,b>0, and check the formula on a circle.
e) A student writes the area as ∫−444316−x2dx. What region did she compute, and what is missing?
Show the solution
Answers
a)Four symmetric quarters, each ∫044316−x2dx.
b)48∫0π/2cos2θdθ=12π
c)24∫0π/6cos2θdθ=2π+33, between 63 and 12.
d)πab; for a=b=r, πr2.
e)The upper half only, 6π; the lower half doubles it to 12π.
a) The ellipse is symmetric about both axes: replacing x by −x or y by −y leaves its equation unchanged. So its area is four times the area of the part in the first quadrant, which is the region under y=4316−x2 for 0≤x≤4. The area is 4∫044316−x2dx=3∫0416−x2dx. Stating the symmetry is part of the answer: a formula like 4∫ without a reason is a formula the marker cannot check.
b) Form a2−x2 with a=4: x=4sinθ, dx=4cosθdθ, and 16−x2=4cosθ since cosθ≥0 on [0,2π]. The limits: x=0 gives θ=0, and x=4 gives sinθ=1, θ=2π. So 3∫0416−x2dx=3∫0π/216cos2θdθ=48∫0π/221+cos2θdθ=48[2θ+4sin2θ]0π/2=48⋅4π=12π. The term sin2θ vanishes at both limits, which is typical when the limits are 0 and 2π: no triangle, no return to x.
c) The strip is symmetric about the x-axis, so its area is 2∫024316−x2dx=23∫0216−x2dx. With x=4sinθ, x=2 gives sinθ=21, so θ=6π. The area is 23∫0π/616cos2θdθ=24[2θ+4sin2θ]0π/6=24(12π+83)=2π+33. Here sin2θ=sin3π=23 does NOT vanish, and dropping it is the usual loss. Size: the height of the strip decreases from 6 at x=0 to 2⋅4312=33 at x=2, so the area lies between 2⋅33=63≈10.4 and 2⋅6=12. And 2π+33≈6.28+5.20=11.48 fits.
d) By the same symmetry, the area is 4∫0aaba2−x2dx. With x=asinθ, dx=acosθdθ, a2−x2=acosθ and limits 0 and 2π: 4⋅ab∫0π/2a2cos2θdθ=4ab⋅4π=πab. For a=4, b=3 this is 12π, as in b). For a circle, a=b=r gives πr2. A general formula that does not reduce to the circle is wrong, and this check takes one line.
e) ∫−444316−x2dx is the area under the upper half only, between the curve and the x-axis: 43⋅2π⋅16=6π, half of the ellipse. The root always returns the non-negative square root, so it can never describe the lower half, which is y=−4316−x2. The enclosed area is ∫−442⋅4316−x2dx=12π, the height of the ellipse at x being the distance between its two halves. The missing factor 2 is the question to ask of every area built on a root.
Part B: problems and reasoning (/50)
Exercise 6: Complete the square first: the circle hidden in 3 + 2x minus x squared
A quadratic under the root that is not already of the form a2±x2 must be put in that form BEFORE any substitution. Completing the square turns it into ±(x−h)2±k2, and the trigonometric substitution is then made on u=x−h, not on x.
The figure shows the graph of y=3+2x−x2.
a) Complete the square in 3+2x−x2 and explain what the figure shows, with its centre and radius.
b) Compute ∫3+2x−x2xdx.
c) Evaluate ∫123+2x−x2xdx exactly, and bracket it without a calculator.
d) Compute ∫(x2+6x+13)3/2dx.
e) A student writes 3+2x−x2=(x−1)2−4 and sets x−1=2secθ. Show with one value of x that the square is wrong, and say what the wrong sign changes.
Show the solution
Answers
a)3+2x−x2=4−(x−1)2: upper half of the circle of centre (1,0) and radius 2.
b)arcsin2x−1−3+2x−x2+C
c)6π+2−3, between 21 and 32
d)4x2+6x+13x+3+C
e)At x=1: 4 against −4. The wrong sign swaps the sine form for the secant form, on a domain where the root does not exist.
a) Factor out the MINUS sign first, it is where the error hides: 3+2x−x2=−(x2−2x)+3=−[(x−1)2−1]+3=4−(x−1)2. Check at x=0: 4−1=3. So y=4−(x−1)2, that is (x−1)2+y2=4 with y≥0: the figure is the upper half of the circle of centre (1,0) and radius 2, defined on [−1,3]. The form is a2−u2 with a=2 and u=x−1, so the substitution is x−1=2sinθ, −2π≤θ≤2π: the same angle as in Exercise 2, on a circle moved one unit to the right.
b) x=1+2sinθ, dx=2cosθdθ, 3+2x−x2=4−4sin2θ=2cosθ (non-negative on the interval). The numerator is x=1+2sinθ, NOT 2sinθ: the shift must be carried everywhere x appears. The integral becomes ∫2cosθ(1+2sinθ)⋅2cosθdθ=∫(1+2sinθ)dθ=θ−2cosθ+C. Back to x: θ=arcsin2x−1 and 2cosθ=4−(x−1)2=3+2x−x2. Answer: arcsin2x−1−3+2x−x2+C. Check by differentiating: 4−(x−1)21−23+2x−x22−2x=3+2x−x21+(x−1)=3+2x−x2x.
c) With the limits moved: x=1 gives sinθ=0, θ=0; x=2 gives sinθ=21, θ=6π. So the integral is [θ−2cosθ]0π/6=(6π−3)−(0−2)=6π+2−3. Bracket: on [1,2] the numerator grows from 1 to 2 and the root falls from 2 to 3, so the integrand increases from 21 to 32≈1.15, and over an interval of length 1 the integral lies between 0.5 and 1.15. Since 6π≈0.52 and 2−3≈0.27, the value is about 0.79. The integrand stays bounded on [1,2], the root only vanishing at x=3, so nothing improper is involved here.
d) x2+6x+13=(x+3)2+4: a SUM of squares, so x+3=2tanθ, −2π<θ<2π, dx=2sec2θdθ and (x2+6x+13)3/2=(4sec2θ)3/2=8sec3θ. The integral is ∫8sec3θ2sec2θdθ=41∫cosθdθ=41sinθ+C. The triangle of tanθ=2x+3 has x+3 opposite, 2 adjacent and hypotenuse (x+3)2+4=x2+6x+13, so sinθ=x2+6x+13x+3 and the answer is 4x2+6x+13x+3+C. Label the sides of the triangle with x+3, never with x: the shift survives to the last line.
e) At x=1: 3+2−1=4, while (1−1)2−4=−4. The completed square is the negative of the right one, the minus sign in front of x2 having been lost. The consequence is not a small slip. The form (x−1)2−4 is u2−a2, which calls for a secant, and it is non-negative only for ∣x−1∣≥2, that is outside (−1,3): the student is working on exactly the set where the original root does not exist, apart from the two endpoints. One value of x substituted into both sides catches it in five seconds, and the figure shows it at a glance: the graph is a bounded arc, the signature of a2−u2.
Exercise 7: Trigonometric or plain substitution? The power outside the root decides
A trigonometric substitution is a heavy tool, and the substitution rule of section 5.5 is sometimes available at a fraction of the cost. What decides is the factor OUTSIDE the root: an odd power of x next to a2±x2 leaves one factor xdx to form du; an even power, or no power at all, leaves nothing to absorb.
a) Compute ∫x2+4x3dx with u=x2+4, then evaluate it from 0 to 2.
b) Compute the same indefinite integral with x=2tanθ and show that the two answers are the same function.
c) Evaluate ∫03x29−x2dx. Show why u=9−x2 gets nowhere.
d) Without computing, choose the method for each: ∫x16−x2dx, ∫16−x2x2dx, ∫x2+16xdx and ∫x2+16dx. Then compute the first one.
Show the solution
Answers
a)31(x2+4)3/2−4x2+4+C; from 0 to 2: 316−82
b)38sec3θ−8secθ+C, which is the same function once secθ=2x2+4.
c)1681π; u=9−x2 leaves ∫u9−u, no simpler.
d)u=16−x2; x=4sinθ; u=x2+16 (a logarithm); the arctangent formula. First one: −31(16−x2)3/2+C.
a) Write x3dx=x2⋅xdx. With u=x2+4, du=2xdx and x2=u−4, so ∫x2+4x3dx=21∫uu−4du=21∫(u1/2−4u−1/2)du=31u3/2−4u1/2+C=31(x2+4)3/2−4x2+4+C. From 0 to 2: at x=2, u=8 and 31⋅162−82=−382; at x=0, u=4 and 38−8=−316. The difference is 316−82≈316−11.3≈1.6, positive as the integrand is on (0,2]. The odd power made this a two-line computation.
b) x=2tanθ, dx=2sec2θdθ, x2+4=2secθ: the integral becomes ∫2secθ8tan3θ⋅2sec2θdθ=8∫tan3θsecθdθ. This is a trigonometric integral of section 7.2 with an odd power of the tangent: keep secθtanθdθ and write tan2θ=sec2θ−1, so with w=secθ it is 8∫(w2−1)dw=38sec3θ−8secθ+C. The triangle of tanθ=2x gives secθ=2x2+4, so 38⋅8(x2+4)3/2−8⋅2x2+4=31(x2+4)3/2−4x2+4: the same function, not merely the same up to a constant. The trigonometric route works, but it needed a second chapter's technique and a triangle, where a) needed neither.
c) Try u=9−x2: du=−2xdx, but the integrand has x2dx, and after using one x for du the other remains, x=9−u. The integral becomes −21∫9−uudu, a root of a product, no simpler than the start. With an even power, the trigonometric substitution is the tool: x=3sinθ, dx=3cosθdθ, 9−x2=3cosθ, limits 0 and 2π. Then ∫0π/29sin2θ⋅3cosθ⋅3cosθdθ=81∫0π/2sin2θcos2θdθ. Since sinθcosθ=21sin2θ, this is 481∫0π/2sin22θdθ=881∫0π/2(1−cos4θ)dθ=881⋅2π=1681π. Size check: the integrand is at most 63≈10.4 (at x2=6) on an interval of length 3, and 1681π≈15.9 is well below 31.
d) ∫x16−x2dx: odd power outside, so u=16−x2, du=−2xdx, and −21∫u1/2du=−31(16−x2)3/2+C. ∫16−x2x2dx: even power, form a2−x2, so x=4sinθ. ∫x2+16xdx: there is no root at all and the numerator is half the derivative of the denominator, so u=x2+16 gives 21ln(x2+16)+C; a tangent substitution would work and waste five minutes. ∫x2+16dx: this is in the table, 41arctan4x+C; the substitution x=4tanθ is precisely how that formula is proved, not something to redo on an exam. The rule: look for the plain substitution FIRST, and reach for the triangle only when it fails.
Exercise 8: Five statements to correct
Each statement below comes from a MATH 141 paper, and each is false. Say what is wrong, give the correct statement, and settle it with a computation or a counterexample.
a) With x=2sinθ, ∫014−x2dx=∫01dθ=1.
b) After x=3sinθ, the triangle gives cosθ=9−x2, so ∫(9−x2)3/2dx=91tanθ+C=279−x2x+C.
c) ∫x2−1xdx requires x=secθ, since the root has the form x2−a2.
d) A trigonometric substitution is only for integrands that contain a square root.
e) Since θ=arcsin3x, we have sin2θ=32x.
Show the solution
Answers
a)False: the limits become 0 and 6π, and the integral is 6π.
b)False: cosθ=39−x2, and the answer is 99−x2x+C.
c)False: u=x2−1 gives x2−1+C in one line.
d)False: x=2tanθ gives ∫02(x2+4)2dx=64π+321.
e)False: sin2θ=2sinθcosθ=92x9−x2; at x=3 it is 0, not 2.
a) FALSE. The integrand does become 2cosθ2cosθdθ=dθ, but the limits 0 and 1 are values of x. With x=2sinθ: x=0 gives θ=0 and x=1 gives sinθ=21, so θ=6π. Correct statement: ∫014−x2dx=∫0π/6dθ=6π, which the table confirms, [arcsin2x]01=6π. Size check: the integrand is at least 21 on [0,1], and 6π≈0.52; the value 1 would need an integrand averaging 1, which only happens near x=3.
b) FALSE. The triangle of sinθ=3x has hypotenuse 3, so cosθ=hypotenuseadjacent=39−x2: the division by the hypotenuse was dropped. The θ computation is right: 27cos3θ3cosθ=91sec2θ, so 91tanθ+C. But tanθ=adjacentopposite=9−x2x, and the correct answer is 99−x2x+C. The false one is off by a factor 3, and differentiating it at x=0 gives 811 instead of the integrand's 271. Correct statement: read every ratio on the triangle as a ratio of TWO sides; a single side is never a trigonometric value.
c) FALSE. The substitution x=secθ works, but it is not required, and on an exam it is a waste. The odd power outside the root hands over du: with u=x2−1, du=2xdx, so ∫x2−1xdx=21∫u−1/2du=x2−1+C, valid on both pieces of the domain at once, whereas the secant would force the two-branch discussion of Exercise 4. Correct statement: the form of the root chooses AMONG the three trigonometric substitutions; it does not decide whether one is needed. Check first for a plain substitution.
d) FALSE. The identity 1+tan2θ=sec2θ simplifies x2+a2 whatever power it carries. Counterexample: ∫02(x2+4)2dx with x=2tanθ, dx=2sec2θdθ, (x2+4)2=16sec4θ, limits 0 and 4π: 81∫0π/4cos2θdθ=81[2θ+4sin2θ]0π/4=81(8π+41)=64π+321. Correct statement: a trigonometric substitution applies whenever an expression a2−x2, a2+x2 or x2−a2 is raised to a power, root or not; the square root is only the most common case.
e) FALSE. The sine of a double angle is not twice the sine: sin2θ=2sinθcosθ. With sinθ=3x and cosθ=39−x2, sin2θ=92x9−x2. A single value refutes the statement: at x=3, θ=2π and sin2θ=sinπ=0, while 32x=2, a value no sine can take. Correct statement: expand sin2θ with the double-angle formula FIRST, then read both factors on the triangle. This slip turns the right antiderivative of 9−x2 into a wrong one, and a derivative check catches it.
Exercise 9: A fuel gauge for a tank lying on its side
A cylindrical fuel tank of radius 1 m and length 5 m lies on its side. A dipstick measures the depth h of fuel, 0≤h≤2 (in metres), and the operator needs the volume. Since every cross-section is the same, the volume is 5 times the area of fuel in one cross-section.
In the figure the cross-section is the disc x2+y2≤1, centre O, and the fuel fills the part below the level line y=h−1 (drawn for h=21).
a) Show that the area of fuel in a cross-section is A(h)=∫−1h−121−y2dy.
b) Compute A(h) in closed form with y=sinθ.
c) Give the volume of fuel for h=21 m and for h=23 m, and the fraction of the full tank each represents. Check that the two volumes add up to the full tank, and say why they must.
d) A careless operator marks the dipstick as if the volume were proportional to the depth. At h=21, by how much does he overestimate the fuel?
e) Find dhdV with the Fundamental Theorem of Calculus. When fuel is pumped in at a constant rate, at what depth does the level rise most slowly, and why does the figure make that obvious?
Show the solution
Answers
a)A horizontal strip at height y has width 21−y2 and thickness dy.
b)A(h)=arcsin(h−1)+(h−1)2h−h2+2π
c)V(21)=35π−453 m3, fraction 31−4π3≈0.2; V(23)=310π+453 m3; sum 5π.
d)By 125(33−π)≈0.86 m3 (he reads 45π).
e)dhdV=102h−h2, largest at h=1: the level rises slowest at half depth, where the surface is widest.
a) Slice the fuel horizontally. At height y the chord of the unit circle runs from x=−1−y2 to x=1−y2, so a strip of thickness dy has area about 21−y2dy. The fuel occupies the heights from the bottom of the tank, y=−1, up to the level y=h−1 (a depth h measured from the bottom). Adding the strips gives A(h)=∫−1h−121−y2dy. Slicing horizontally is the natural choice here: the level is horizontal, so it becomes a LIMIT of integration instead of a curve to intersect.
b) The form is 1−y2: y=sinθ, dy=cosθdθ, 1−y2=cosθ on [−2π,2π]. Then ∫2cos2θdθ=∫(1+cos2θ)dθ=θ+sinθcosθ+C, which returns to arcsiny+y1−y2+C. Here returning to y is the right choice, because the upper limit h−1 is a letter: moving it would give arcsin(h−1) anyway. So A(h)=[arcsiny+y1−y2]−1h−1=arcsin(h−1)+(h−1)1−(h−1)2−(−2π), and since 1−(h−1)2=2h−h2, A(h)=arcsin(h−1)+(h−1)2h−h2+2π. Checks: A(0)=−2π+0+2π=0, empty tank; A(2)=2π+0+2π=π, the whole disc; A(1)=2π, half.
c) For h=21: arcsin(−21)=−6π and (−21)1−41=−43, so A=−6π−43+2π=3π−43 and V=35π−453 m3. The full tank holds 5π m3, so the fraction is 31−4π3≈0.333−12.61.73≈0.20. For h=23: arcsin21=6π, so A=6π+43+2π=32π+43 and V=310π+453 m3, about 80 per cent. The two add up to 5π: the empty part at depth 23 is the mirror image, top to bottom, of the full part at depth 21.
d) Proportional marking would read 2h of the full tank, so at h=21 he reads a quarter, 45π m3. The overestimate is 45π−(35π−453)=453−125π=125(33−π) m3, about 125(5.20−3.14)≈0.86 m3, or roughly 860 litres. The reason is geometric: near the bottom the tank is narrow, so the first half metre of depth holds much less than a quarter of the volume. A gauge for a horizontal cylinder has to be graduated with A(h), never with a ruler.
e) V(h)=5∫−1h−121−y2dy, so by the Fundamental Theorem of Calculus (part 1) with the chain rule, dhdV=5⋅21−(h−1)2⋅1=102h−h2. Differentiating the closed form of b) gives the same, a useful check. This is the length 5 times the WIDTH of the fuel surface, 22h−h2. If fuel arrives at a constant rate r, then dtdh=102h−h2r, smallest when 2h−h2=1−(h−1)2 is largest, at h=1. The figure shows why: at half depth the surface is a full diameter wide, so each extra litre spreads over the largest area and raises the level the least.
Exercise 10: A final exam question: the hyperbola, the secant cubed and a hyperbolic sector
A long final exam question often chains a trigonometric substitution with a result of the previous chapter. The curve here is the right branch of the hyperbola x2−y2=1, and the figure shades the region between that branch and the line x=2; the dashed segments join the origin to the corners (2,±3).
You may quote from section 7.2: ∫sec3θdθ=21(secθtanθ+ln∣secθ+tanθ∣)+C and ∫secθdθ=ln∣secθ+tanθ∣+C.
a) With x=secθ, 0≤θ<2π, show that for x≥1: ∫x2−1dx=2xx2−1−21ln(x+x2−1)+C.
b) Check this formula by differentiating it.
c) Find the exact area of the shaded region.
d) Without a calculator, show that this area lies between 3 and 23, and explain from the figure which two shapes these numbers measure.
e) The two dashed segments and the arc of the hyperbola between them enclose a hyperbolic sector. Show that its area is ln(2+3), and compare with the circular sector of the unit circle cut out by the rays to (cosα,±sinα).
Show the solution
Answers
a)∫tan2θsecθdθ=21secθtanθ−21ln∣secθ+tanθ∣+C, then the triangle.
b)The derivative simplifies to x2−1x2−1=x2−1.
c)23−ln(2+3)
d)Triangle (1,0), (2,±3): 3; rectangle [1,2]×[−3,3]: 23; and 0<ln(2+3)<ln4<3.
e)Triangle from the origin, 23, minus the region of c): ln(2+3); for the circle the sector's area is the angle α.
a) x=secθ, dx=secθtanθdθ and sec2θ−1=tanθ, non-negative on 0≤θ<2π, which covers x≥1. The integral becomes ∫tanθ⋅secθtanθdθ=∫tan2θsecθdθ=∫(sec3θ−secθ)dθ, using tan2θ=sec2θ−1. With the two quoted results: 21secθtanθ+21ln∣secθ+tanθ∣−ln∣secθ+tanθ∣=21secθtanθ−21ln∣secθ+tanθ∣+C. The triangle of secθ=1x has hypotenuse x, adjacent 1 and opposite x2−1, so tanθ=x2−1, and x+x2−1≥1>0 lets the absolute value go: 2xx2−1−21ln(x+x2−1)+C. The two logarithms combining with coefficient 21−1=−21 is where signs are usually lost.
b) The first term: dxd[2xx2−1]=21x2−1+2x2−1x2. The second: dxdln(x+x2−1)=x+x2−11+x2−1x=x+x2−1x2−1x2−1+x=x2−11, a simplification worth remembering. So the derivative is 2x2−1(x2−1)+x2−1=2x2−12(x2−1)=x2−1, for x>1. The formula is right.
c) The region is symmetric about the x-axis: for 1≤x≤2 it runs from y=−x2−1 to y=x2−1. Its area is 2∫12x2−1dx=2[2xx2−1−21ln(x+x2−1)]12=2[(3−21ln(2+3))−(0−21ln1)]=23−ln(2+3). In θ the same computation runs from θ=0 to θ=3π, where secθ=2 and tanθ=3, and gives the same number. Here returning to x costs nothing, since the antiderivative in x is already known from a).
d) The triangle with vertices (1,0), (2,3) and (2,−3) has base 23 and height 1, area 3. The rectangle [1,2]×[−3,3] has area 23. The region contains the triangle because y=x2−1 is concave on (1,∞) (its second derivative is −(x2−1)3/21<0), so the arc lies above its chord from (1,0) to (2,3); and it sits inside the rectangle. Numerically: 3<23−ln(2+3)<23 means 0<ln(2+3)<3. The left inequality holds since 2+3>1; the right one since 2+3<4 and ln4=2ln2≈1.39<1.73≈3. No calculator, only ln2.
e) The triangle with vertices O, (2,3) and (2,−3) has base 23 and height 2, area 23. It contains the shaded region, since x2−1≤23x for 1≤x≤2 (square both sides: x2−1≤43x2 means x2≤4), and what remains is exactly the hyperbolic sector. Its area is 23−(23−ln(2+3))=ln(2+3). For the unit circle, the sector cut out by the rays to (cosα,±sinα) has angle 2α and area 21⋅2α=α. So ln(2+3) plays for the hyperbola the role the angle plays for the circle: with cosht=2et+e−t and t=ln(2+3), e−t=2−3, so cosht=2 and sinht=2et−e−t=3: the corner is (cosht,sinht), as the circle's is (cosα,sinα).