This is the corrected exercise set for the inverse trigonometric functions in MATH 203, Differential and Integral Calculus I, at Concordia University: section 1.6 of Thomas for sin−1 and cos−1, and section 3.9 for all six functions and their derivatives. The chapter comes late in the term, just before related rates, and it reuses everything before it: the chain rule, implicit differentiation, the derivative as a slope. A scientific calculator is allowed, and every exact answer is also given to four decimals.
The thread running through the set: an inverse trigonometric function returns ONE angle, the one in its restricted range, and that range decides every answer. It is why sin−1(sin65π) is not 65π, why the triangle of tan−1(−2) has a positive cosine, why the derivative of sec−1x carries an ∣x∣, and why an identity proved by a zero derivative has one constant PER INTERVAL. The algebraic gesture where the marks are really lost is named in every solution: bring 1+u2 or 1−u2 to one fraction before inverting or taking a root, and read the square root of a square as an absolute value, x2=∣x∣.
The traps named in the solutions: the degree mode, cos−1 answered with a negative angle, a domain written for x instead of the inner function, 9−x2 taken for 3−x, (3x)2 written 3x2, a chain factor dropped in tan−1x1, sin2x taken for sinx, a constant read at a point of the wrong interval, a candidate created by squaring and kept, and a complex fraction left unsimplified.
Self-checking setType your answers and check them question by question: the page tells you right or wrong before you open the solution. With an account, every answer right on the first try earns points, and your right answers stay filled in when you come back.
•cot−1x=2π−tan−1x and csc−1x=2π−sec−1x: their derivatives are the opposites.
•x2=∣x∣. A complex fraction is simplified by multiplying its top and bottom by the common denominator of the small fractions.
•If F′(x)=0 at every point of an interval, F is constant on that interval: read the constant at a point of THAT interval.
Part A: the basics (/50)
Exercise 1: Restricted ranges: the one angle an inverse function returns
Sine, cosine and tangent take each of their values infinitely often, so none of them has an inverse on its whole domain. Thomas (section 1.6) restricts each one to an interval where it is one-to-one and inverts the restriction. y=sin−1x=arcsinx is THE angle y of [−2π,2π] with siny=x, defined for −1≤x≤1. y=cos−1x=arccosx is the angle of [0,π] with cosy=x. y=tan−1x=arctanx is the angle of (−2π,2π) with tany=x, defined for every real x. The −1 of sin−1 names an inverse function, it is NOT an exponent.
The figure shows y=sinx for −2π≤x≤2π, the restricted piece on [−2π,2π] in blue, and the dashed line y=21. A scientific calculator is allowed, in RADIAN mode.
a) Find exactly, in radians: arcsin(−21), arccos(−23), arctan3, arccos0 and arctan(−1).
b) Using the figure, how many solutions does the equation sinx=21 have on [−2π,2π]? Give them all exactly, and say which one is arcsin21 and why.
c) Find the domain of f(x)=arcsin(2x−3), of g(x)=arccos3x−1 and of h(x)=arctan4−x.
d) In radian mode, a calculator gives cos−1(0.2)≈1.3694. Using only this value, the identity arccos(−x)=π−arccosx and the identity arcsinx+arccosx=2π of Thomas 1.6, find arccos(−0.2) and arcsin(0.2) to four decimals. Explain the first identity with the unit circle.
e) A student whose calculator is in degree mode types tan−1(1) and reads 45, then types sin−1(2) and reads an error message. Give arctan1 in radians, exactly and to four decimals, and explain the error message.
Show the solution
Answers
a)−6π, 65π, 3π, 2π, −4π
b)Four solutions: −611π, −67π, 6π, 65π; arcsin21=6π, the only one on the blue piece
c)f: [1,2]; g: [−2,4]; h: (−∞,4]
d)arccos(−0.2)≈1.7722, arcsin(0.2)≈0.2014
e)arctan1=4π≈0.7854; sin−1(2) does not exist, since 2 is not the sine of any angle
a)
Each answer is found by asking which angle OF THE RANGE has the given sine, cosine or tangent. sin(−6π)=−21 and −6π∈[−2π,2π], so arcsin(−21)=−6π; the angle 67π also has sine −21, but it is not in the range. cos65π=−23 with 65π∈[0,π], so arccos(−23)=65π: the answer −6π, which a student gets by copying the sine case, is outside [0,π] and its cosine is +23. tan3π=3, so arctan3=3π. cos2π=0, so arccos0=2π. tan(−4π)=−1, so arctan(−1)=−4π, and not 43π, which has the same tangent but lies outside (−2π,2π). The rule to memorize: a negative input sends arcsin and arctan below zero, and sends arccos into the second quadrant.
b)
The dashed line y=21 crosses the grey curve four times on [−2π,2π]. On [0,2π], the solutions are 6π and π−6π=65π, since sin(π−x)=sinx. Subtracting one period 2π gives the two others: 6π−2π=−611π and 65π−2π=−67π. Four solutions, and the calculator key sin−1 returns only ONE of them, 6π≈0.5236: the only one on the blue piece, because arcsin is by definition the inverse of the RESTRICTED sine. That is the whole chapter in one picture: an inverse trigonometric function answers the question 'which angle of my range', never 'which angles'.
c)
Each inverse function imposes a condition on what is inside it, and the domain is found by solving an inequality: this is the algebra that costs marks. arcsinu needs −1≤u≤1: −1≤2x−3≤1, add 3 to the three members, 2≤2x≤4, divide by 2: 1≤x≤2, so the domain of f is [1,2]. arccosu needs the same: −1≤3x−1≤1, multiply by 3, −3≤x−1≤3, so −2≤x≤4. arctan accepts every real number, so the only condition on h comes from the square root: 4−x≥0, that is x≤4, and the domain is (−∞,4]. The classic slip is to answer [−1,1] for f, the domain of arcsinx itself: the condition is on the INSIDE 2x−3, not on x.
d)
On the unit circle, the angle α=arccosx of [0,π] has the point (x,sinα); the angle π−α, also in [0,π], is its mirror image in the vertical axis, with first coordinate −x. So cos(π−α)=−x, and since π−α is in the range, arccos(−x)=π−arccosx. With arccos(0.2)≈1.369438: arccos(−0.2)=π−1.369438≈1.772154, that is 1.7722 to four decimals. Then arcsin(0.2)=2π−arccos(0.2)≈1.570796−1.369438=0.201358, that is 0.2014. Both are checked directly on the calculator. The false symmetry arccos(−x)=−arccosx gives a NEGATIVE answer, impossible for arccos, whose values are in [0,π]: that alone refutes it.
e)
In degree mode the calculator returns angles in degrees: 45 means 45°, and 45°=4π rad. In calculus every angle is in radians, because the derivative formulas of the chapter, such as dxdarctanx=1+x21, are only true in radians. So arctan1=4π≈0.7854; a 45 copied into a tangent line or a limit makes the whole answer wrong by a factor π180. The error message on sin−1(2) is correct: sin−1 is only defined on [−1,1], since no angle has a sine equal to 2. It is not the reciprocal sin21≈1.0998, which does exist: the reciprocal of the sine is csc. Check the mode before the first question of any test: sin−1(1) must display 1.5708.
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: Compositions and the reference triangle: the range fixes the sign
sin(arcsinx)=x for every x of [−1,1]: the inverse acts first and the sine undoes it. The other order is the trap: arcsin(sinx)=x ONLY when x is already in [−2π,2π], because arcsin can only return an angle of its range. To evaluate a trigonometric function of an inverse trigonometric function, draw the right triangle of the angle, find the missing side by Pythagoras, then fix the SIGN with the range.
The figure shows the triangle of θ=arcsin3x for 0<x<3: hypotenuse 3, side opposite θ equal to x, so that sinθ=3x.
a) Find exactly: arcsin(sin67π), arccos(cos35π) and arctan(tan(−43π)).
b) Find exactly, then to four decimals: cos(arctan(−2)), sin(arctan(−2)) and tan(arcsin(−135)).
c) Using the triangle of the figure, write cos(arcsin3x) and tan(arcsin3x) as algebraic expressions in x, and say for which x each one holds. With a triangle of your own, write sec(arctan2x) in the same way. Evaluate the second expression at x=1 and the third at x=1.
d) Thomas also defines sec−1x=cos−1x1 for ∣x∣≥1, with values in [0,π] except 2π. Find sec−12, sec−1(−2) and tan(sec−1(−2)). Show that tan(sec−1x)=x2−1 for x≥1 but −x2−1 for x≤−1.
e) Solve for x: arcsin(2x)=6π; arctan(x−1)=−3π; arccosx=34π.
Show the solution
Answers
a)−6π, 3π, 4π
b)51≈0.4472, −52≈−0.8944, −125≈−0.4167
c)39−x2 on [−3,3]; 9−x2x on (−3,3); 1+4x2 for every x; values 221≈0.3536 and 5≈2.2361
d)3π, 32π, −3; the sign of tan follows the quadrant of the angle
e)x=41; x=1−3; no solution, 34π is not in [0,π]
a)
Compute the inner value first, then ask for the angle OF THE RANGE. sin67π=−21, and the angle of [−2π,2π] with sine −21 is −6π. cos35π=21, and the angle of [0,π] with cosine 21 is 3π. tan(−43π)=1, since the tangent has period π and −43π+π=4π; the angle of (−2π,2π) with tangent 1 is 4π. None of the three answers is the angle we started from: each of 67π, 35π, −43π lies outside the range of its inverse, and writing it back as the answer is the most frequent zero of the chapter.
b)
Let θ=arctan(−2). Then tanθ=−2 and θ∈(−2π,0), a fourth-quadrant angle: cosine POSITIVE, sine NEGATIVE. The triangle with opposite side 2 and adjacent side 1 has hypotenuse 1+4=5, so cosθ=51=55≈0.4472 and sinθ=−52≈−0.8944. The triangle gives the SIZES, the range gives the SIGNS. Let φ=arcsin(−135), also in (−2π,0): the triangle 5, 13 has third side 169−25=12, cosφ=+1312 and tanφ=12/13−5/13=−125≈−0.4167. A student who takes cosθ=−51 because the input was negative has moved the angle to the second quadrant, where arctan never goes.
c)
In the figure, the adjacent side is 32−x2=9−x2 by Pythagoras. It is NOT 3−x: a square root does not distribute over a difference, and 9−4=5=1 settles it. So cos(arcsin3x)=39−x2. The triangle only shows 0<x<3, but the formula holds on all of [−3,3]: θ is in [−2π,2π], where the cosine is never negative, so the positive root is the right one. Then tan(arcsin3x)=9−x2x for −3<x<3 (the sign now comes from x, as it should). At x=1: 81=42≈0.3536. For β=arctan2x: opposite side 2x, adjacent 1, hypotenuse 1+(2x)2=1+4x2, and since cosβ>0 on the range, secβ=1+4x2 for every x. Watch the square: (2x)2=4x2, not 2x2. At x=1: 5≈2.2361.
d)
sec−12=cos−121=3π and sec−1(−2)=cos−1(−21)=32π. The angle 32π is in the second quadrant, so tan32π=−3≈−1.7321. In general, let y=sec−1x, so secy=x and tan2y=sec2y−1=x2−1, which gives tany=±x2−1, and the range decides. For x≥1, cosy=x1>0, so y∈[0,2π) and tany≥0: tan(sec−1x)=x2−1. For x≤−1, cosy<0, so y∈(2π,π] and tany≤0: tan(sec−1x)=−x2−1. Check at x=−2: −3, as found directly. This sign is exactly the one that will put an absolute value in the derivative of sec−1x.
e)
Apply the direct function to both sides, but only after checking that the right side is in the range. 6π is in [−2π,2π], so arcsin(2x)=6π gives 2x=sin6π=21 and x=41; check: arcsin21=6π. −3π is in (−2π,2π), so x−1=tan(−3π)=−3 and x=1−3≈−0.7321. For the third, 34π>π is NOT a value of arccos: the equation has no solution. Taking the cosine of both sides blindly gives x=cos34π=−21, and the check refutes it at once: arccos(−21)=32π=34π.
Exercise 3: Derivatives of arcsin, arccos and arctan: the formula is one line, the algebra is the rest
Thomas (section 3.9) gives, for a differentiable inner function u of x: dxdsin−1u=1−u2u′ and dxdcos−1u=−1−u2u′ for ∣u∣<1, and dxdtan−1u=1+u2u′ for every u.
Every one of these formulas contains 1−u2 or 1+u2. When u is a power, a fraction or a root, that expression is where the marks are lost: square u correctly, bring 1±u2 to ONE fraction, then take the root or invert. Exact answers, then the decimal asked.
a) Differentiate f(x)=arccos(x2), g(x)=arctan(x2−1) and h(x)=arcsinx, naming each inner function, simplify, and evaluate f′(21), g′(1) and h′(21).
b) Differentiate y=arcsin1+x2x and show that y′=1+x21. Which familiar function has the same derivative and the same value at x=0? What do you conclude?
c) Differentiate k(x)=(1+x2)arctanx and m(x)=xarcsinx, and evaluate k′(1) and m′(21).
d) Find the equation of the tangent line to y=arccos2x at x=1.
e) Show that y=arctanx satisfies (1+x2)y′′+2xy′=0, and give y′′(1).
f: inner function u=x2, u′=2x, and u2=(x2)2=x4, so f′(x)=−1−x42x for ∣x∣<1. At 21: 1−161=1615, 1615=415, and f′(21)=−1⋅154=−154≈−1.0328. The slip to avoid: 1−x2, the formula of arccosx copied without replacing x by u. g: inner function u=x2−1, u′=2x, 1+u2=1+x4−2x2+1=x4−2x2+2, so g′(x)=x4−2x2+22x and g′(1)=12=2. Expanding (x2−1)2 as x4−1 would give g′(1)=2 by luck here but a wrong function everywhere else, for example 164=41 instead of 104 at x=2. h: inner function u=x, u′=2x1, u2=x, so h′(x)=2x1−x1 for 0<x<1, and h′(21)=2⋅21⋅211=1.
b)
Write u=x(1+x2)−1/2 and use the product rule with the chain rule on the second factor: u′=(1+x2)−1/2−x2(1+x2)−3/2. Factor the SMALLEST power, (1+x2)−3/2: u′=(1+x2)−3/2[(1+x2)−x2]=(1+x2)−3/2. Then bring 1−u2 to one fraction: 1−1+x2x2=1+x21+x2−x2=1+x21, whose root is (1+x2)−1/2. Finally y′=(1+x2)−1/2(1+x2)−3/2=(1+x2)−1=1+x21. This is the derivative of arctanx, and both functions are 0 at x=0. Their difference has derivative 0 on the whole interval (−∞,∞), so it is constant (a fact used as a tool, the constant being read at a point), and that constant is 0: arcsin1+x2x=arctanx for every x. The triangle confirms it: the angle with opposite side x and adjacent side 1 has hypotenuse 1+x2.
c)
Product rule on k: k′(x)=2xarctanx+(1+x2)⋅1+x21=2xarctanx+1. The factor 1+x2 was chosen to cancel the denominator; simplify it rather than leaving 1+x21+x2. At x=1: k′(1)=2⋅4π+1=2π+1≈2.5708. Quotient rule on m, for x=0 in [−1,1]: m′(x)=x2x⋅1−x21−arcsinx⋅1. At x=21: 3/21/2=31, arcsin21=6π, so m′(21)=4(31−6π)=34−32π≈0.2150. The order of the numerator is 'derivative of the top times the bottom, minus the top times the derivative of the bottom': swapped, it gives −0.2150.
d)
Point: y(1)=arccos21=3π. Slope: inner function u=2x, u′=21, 1−u2=1−4x2=44−x2, so y′=−4−x2/21/2=−4−x21. At x=1: −31≈−0.5774. Tangent: y−3π=−31(x−1), that is y=−31x+3π+31, with y-intercept ≈1.6245. Two checks: the slope is negative, as it must be for an arccosine, and the point (1,3π) satisfies the tangent equation. Forgetting the inner derivative 21 gives the slope −32, twice too steep.
e)
y′=1+x21=(1+x2)−1, and by the chain rule y′′=−(1+x2)−2⋅2x=−(1+x2)22x. Then (1+x2)y′′=−1+x22x and 2xy′=1+x22x: the sum is 0 for every x. At x=1: y′′(1)=−42=−21. Rewriting 1+x21 as a power BEFORE differentiating spares the quotient rule, and it is the rewriting that the exam rewards: the student who differentiates 1+x21 as 2x1 has differentiated the denominator alone, and that is not a rule.
Exercise 4: The derivative of arcsec by implicit differentiation, and the absolute value it needs
Thomas defines y=sec−1x for ∣x∣≥1 as the angle y of [0,π], y=2π, with secy=x. Equivalently sec−1x=cos−1x1. The figure shows its graph: a right branch from (1,0) and a left branch ending at (−1,π), both approaching the dashed line y=2π, and BOTH rising from left to right.
Its derivative is the one formula of section 3.9 with an absolute value in it, and this exercise is about where that absolute value comes from.
a) Let y=sec−1x with ∣x∣>1. Differentiate secy=x implicitly, show that dxdy=secytany1, then express secytany in terms of x on each branch, and conclude that dxdsec−1x=∣x∣x2−11. Evaluate at x=2 and at x=−2.
b) Check the result by a second route: differentiate cos−1x1 with the chain rule, and simplify 1−x21 correctly.
c) Thomas obtains three derivatives from identities: cos−1x=2π−sin−1x, cot−1x=2π−tan−1x, csc−1x=2π−sec−1x. Deduce the derivatives of cot−1x and csc−1x, and evaluate the first at x=1, the second at x=−2.
d) Differentiate f(x)=sec−1(3x) for x<−31 and evaluate f′(−32). A student writes f′(x)=3x9x2−13: what value does he find, and how does the figure tell him he is wrong before any computation?
e) Find the equation of the tangent line to y=sec−1x at x=−2.
Show the solution
Answers
a)secytany⋅y′=1; secytany=∣x∣x2−1 on both branches; 231≈0.2887 at x=2 AND at x=−2
b)1−x21=∣x∣x2−1, and the chain rule gives ∣x∣x2−11 again
c)dxdcot−1x=−1+x21, −21 at 1; dxdcsc−1x=−∣x∣x2−11, −231 at −2
d)f′(x)=∣3x∣9x2−13, f′(−32)=23≈0.8660; the student finds −0.8660, a negative slope on a rising graph
e)y=32π+231(x+2)
a)
Differentiate secy=x with respect to x, y being a function of x (chain rule, inner function y): secytany⋅y′=1, so y′=secytany1 wherever tany=0, that is for ∣x∣>1. Now secy=x, and tan2y=sec2y−1=x2−1, so tany=±x2−1: the RANGE chooses. On the right branch, x>1, y∈(0,2π) and tany>0: secytany=xx2−1. On the left branch, x<−1, y∈(2π,π) and tany<0: secytany=x⋅(−x2−1)=−xx2−1. In both cases the product is ∣x∣x2−1, positive, so dxdsec−1x=∣x∣x2−11. At x=2: 231=63≈0.2887, and at x=−2 exactly the same value: the two branches rise, the derivative is positive on both.
b)
With u=x1, u′=−x21: dxdcos−1x1=−1−1/x2−1/x2=x21−x211. Bring the inside to one fraction: 1−x21=x2x2−1. Its square root is x2x2−1=∣x∣x2−1, because x2=∣x∣ and NOT x: for x=−2, 4=2. Then x2⋅∣x∣x2−11=x2x2−1∣x∣=∣x∣x2−11, since x2=∣x∣2. The same absolute value appears by a completely different road: it is not a convention of Thomas, it is forced by the algebra. Writing x2=x in this line is the most expensive slip of section 3.9, because it gives a formula that is wrong on a whole branch.
c)
Differentiate the identities; the constant 2π has derivative 0. dxdcot−1x=−1+x21, which is −21 at x=1. dxdcsc−1x=−∣x∣x2−11 for ∣x∣>1, which is −231≈−0.2887 at x=−2. The pattern of Thomas' table: each 'co' function has the derivative of its partner with a minus sign, because it is a constant minus the partner. Nothing new has to be memorized, and the signs are checked on the graphs: cot−1 decreases.
d)
Chain rule, inner function u=3x, u′=3, u2=9x2: f′(x)=∣3x∣9x2−13. At x=−32: ∣3x∣=2 and 9x2−1=4−1=3, so f′(−32)=233=23≈0.8660. The student's formula has 3x=−2 in the denominator and gives −23≈−0.8660. The figure refutes it without any computation: f(x)=sec−1(3x) is the left branch of the graph compressed horizontally, and a rising curve cannot have a negative slope. Keep the absolute value until the very last substitution, then replace ∣3x∣ by −3x when x<0.
e)
Point: sec−1(−2)=cos−1(−21)=32π. Slope from a): ∣−2∣31=231≈0.2887. Tangent: y−32π=231(x+2), that is y=231x+32π+31, with y-intercept 32π+31≈2.6717. The point is on the left branch of the figure, at height about 2.09, and the tangent rises gently, as it should.
Exercise 5: The chain rule through an inverse trigonometric function, and the square root of a square
In sin−1u, cos−1u or tan−1u, name the inner function u first, write u′ next to it, and only then apply the formula. When 1−u2 is itself a square, its root is an ABSOLUTE VALUE: 1−cos2x=sin2x=∣sinx∣.
Part d) uses a differentiable function g known only through a table: g(3)=21, g′(3)=6, g(4π)=5, g′(4π)=2.
a) Let y=arctan(e2x). Name the inner function, differentiate, and evaluate y′(0).
b) Differentiate y=(arcsinx)2 and z=arctan(lnx). Evaluate y′(21) and z′(e) to four decimals.
c) Let y=arcsin(cosx). Show that y′=−1 for 0<x<π, and find y′ for −π<x<0. Deduce a formula for y on [0,π], reading the constant at x=2π, and check it at x=3π.
d) With the table of the introduction, find h′(3) for h(x)=arcsin(g(x)) and k′(1) for k(x)=g(arctanx).
e) Differentiate y=arctan(sin3x), naming its three layers, and evaluate y′(9π).
c)y′=−∣sinx∣sinx: −1 on (0,π), +1 on (−π,0); y=2π−x on [0,π], and y(3π)=6π
d)h′(3)=3/26=43≈6.9282; k′(1)=g′(4π)⋅21=1
e)y′=1+sin23x3cos3x, y′(9π)=76≈0.8571
a)
Inner function u=e2x, whose derivative is u′=2e2x (chain rule once more, inner 2x). Then u2=(e2x)2=e4x by the law (ea)b=eab; it is not e4x2 and not e2x2. So y′=1+e4x2e2x, and y′(0)=1+12=1. The exponent law is the algebra of this line, and a wrong exponent gives the right value at 0 by coincidence and a wrong function everywhere else: at x=1 the correct slope is 1+e42e2≈0.266.
b)
For y=(arcsinx)2 the OUTER function is the square and the inner one is arcsinx: y′=2arcsinx⋅1−x21. Do not confuse it with arcsin(x2), whose outer function is arcsin. At x=21: 232⋅6π=332π≈1.2092. For z=arctan(lnx), x>0: inner function lnx, derivative x1, so z′=1+(lnx)21/x=x(1+ln2x)1. At x=e, lne=1: z′(e)=2e1≈0.1839. Writing ln2x means (lnx)2, not ln(x2)=2lnx; at x=e the confusion gives 3e1.
c)
Inner function u=cosx, u′=−sinx, 1−u2=1−cos2x=sin2x. So y′=sin2x−sinx=−∣sinx∣sinx. On (0,π), sinx>0, ∣sinx∣=sinx and y′=−1. On (−π,0), sinx<0, ∣sinx∣=−sinx and y′=+1. The student who writes sin2x=sinx finds −1 everywhere and misses half of the graph. On the interval (0,π) the function y+x has derivative 0, so it is constant there (the zero-derivative fact, used as a tool). Read it at x=2π: arcsin(cos2π)+2π=arcsin0+2π=2π. Hence y=2π−x on (0,π), and also at the endpoints, where arcsin1=2π and arcsin(−1)=−2π. Check at 3π: arcsin21=6π=2π−3π.
d)
Chain rule on h, inner function g: h′(x)=1−g(x)2g′(x). At x=3: 1−41=23, so h′(3)=3/26=312=43≈6.9282. Chain rule on k, inner function arctanx, outer function g: k′(x)=g′(arctanx)⋅1+x21. At x=1, arctan1=4π, so k′(1)=g′(4π)⋅21=2⋅21=1. The table is built with two traps: g(4π)=5 is there to be confused with g′(4π), which gives 2.5; and g′(1) is not in the table because the outer derivative is evaluated at the INNER value arctan1, not at 1.
e)
Three layers, from the outside: arctan, then sin, then 3x. Differentiate each at the layer below it: y′=1+sin23x1⋅cos3x⋅3=1+sin23x3cos3x. At x=9π, 3x=3π: cos3π=21 and sin23π=43, so y′=7/43/2=76≈0.8571. The factor 3 of the innermost layer is the one that disappears on exams; the check is to count the layers, three, and the factors in the answer, three.
Part B: problems and reasoning (/50)
Exercise 6: Two functions that agree on one side only: arcsec from the arctan key
Most scientific calculators, including the models approved for the course, have no sec−1 key. This exercise builds a formula that computes sec−1x with the tan−1 key, and proves it with the tool of the chapter: if F′(x)=0 at every point of an INTERVAL, then F is constant on that interval (admitted), and the constant is read at one convenient point OF THAT INTERVAL.
Let g(x)=tan−1x2−1 and h(x)=sec−1x, both defined for ∣x∣≥1. Recall from section 3.9 that h′(x)=∣x∣x2−11 for ∣x∣>1. The figure shows h in blue and g in dashed orange: they coincide on the right and separate on the left.
a) Show that g is even and that h is not. Compute g(2), g(−2), h(2) and h(−2) exactly.
b) Differentiate g for ∣x∣>1 and show that g′(x)=xx2−11. Compare g′ with h′ on each side, and evaluate both at x=−2.
c) Show that g−h is constant on (1,∞) and find the constant.
d) Show that g+h is constant on (−∞,−1) and find the constant. Why is it wrong to read this constant at x=2?
e) Conclude: write sec−1x in terms of tan−1 for x≥1 and for x≤−1 (check the endpoints ±1 directly), then compute sec−1(−3) to four decimals with the tan−1 key, and check it with cos−1(−31).
b)g′(x)=xx2−11: g′=h′ for x>1, g′=−h′ for x<−1; g′(−2)=−231, h′(−2)=231
c)(g−h)′=0 on (1,∞), constant g(2)−h(2)=0
d)(g+h)′=0 on (−∞,−1), constant g(−2)+h(−2)=π; x=2 is not in that interval
e)sec−1x=tan−1x2−1 for x≥1, π−tan−1x2−1 for x≤−1; sec−1(−3)≈1.9106
a)
g(−x)=tan−1(−x)2−1=tan−1x2−1=g(x): g only sees x2, so it is even and its graph is symmetric in the y-axis, as the orange curve of the figure is. h is not even: h(2)=cos−121=3π but h(−2)=cos−1(−21)=32π. And g(2)=g(−2)=tan−13=3π. The values already say it: g=h at x=2, but not at x=−2, where g(−2)+h(−2)=π.
b)
Inner function u=x2−1=(x2−1)1/2, u′=21(x2−1)−1/2⋅2x=x2−1x. Now the gesture of the chapter: 1+u2=1+(x2−1)=x2, since squaring a square root of a nonnegative number gives the number back. So g′(x)=x2x/x2−1=xx2−11, with x and not ∣x∣: here no root of a square was taken, and the sign of x stays. For x>1, x=∣x∣ and g′(x)=h′(x). For x<−1, x=−∣x∣ and g′(x)=−h′(x). At x=−2: g′(−2)=−231≈−0.2887 and h′(−2)=231≈0.2887. The figure agrees: on the left, the orange curve goes down while the blue one goes up.
c)
On (1,∞), (g−h)′(x)=g′(x)−h′(x)=0 at every point, and (1,∞) is an interval, so g−h is constant there. Read the constant at a point of THIS interval, x=2: g(2)−h(2)=3π−3π=0. So g(x)=h(x) for every x>1, which is the coincidence of the two curves on the right of the figure.
d)
On (−∞,−1), (g+h)′(x)=g′(x)+h′(x)=−h′(x)+h′(x)=0, so g+h is constant on this interval. Read it at x=−2: g(−2)+h(−2)=3π+32π=π. So h(x)=π−g(x) for x<−1: the left branches of the figure are mirror images in the dashed line y=2π. Reading the constant at x=2 would give g(2)+h(2)=32π, a wrong answer: the fact 'zero derivative gives a constant' holds on ONE interval, and x=2 belongs to the other one. The domain ∣x∣>1 is made of two separate intervals, and each gets its own constant, read at its own point.
e)
Collecting c) and d): sec−1x=tan−1x2−1 for x>1 and sec−1x=π−tan−1x2−1 for x<−1. The endpoints are checked directly: sec−11=0=tan−10 and sec−1(−1)=π=π−tan−10. With the calculator in radians: sec−1(−3)=π−tan−18≈3.141593−1.230959=1.910633, that is 1.9106. Check: cos−1(−31)≈1.9106, the same. The tempting shortcut tan−18≈1.2310 is the formula of the RIGHT branch used on the left; it gives an angle below 2π, impossible for sec−1 of a negative number.
Exercise 7: Horizontal tangents and prescribed slopes: one denominator, then factor
Let f(x)=sin−1x−21−x2 on [−1,1], whose graph is in the figure, and g(x)=4tan−1x−x2 for every real x.
A horizontal tangent is a point where the derivative is 0; a tangent parallel to a line y=mx+b is a point where the derivative is m. In both cases the equation is solved on ONE fraction: put the derivative over a common denominator, then factor the numerator. Exact answers first, then decimals.
a) Show that f′(x)=1−x21+2x for −1<x<1, and find the point of the graph of f with a horizontal tangent, exactly and to four decimals.
b) What does f′(x) do as x→1− and as x→−1+? Describe the tangent to the graph at each endpoint.
c) Find every point of the graph of f where the tangent has slope 2. Squaring is allowed, but every candidate must be checked.
d) Show that g′(x)=1+x24−2x−2x3 and that the graph of g has exactly one horizontal tangent. Give the point.
e) Find every point of the graph of g where the tangent is parallel to the line y=4x.
Show the solution
Answers
a)Horizontal tangent at (−21,−6π−3)≈(−0.5,−2.2556)
b)f′(x)→+∞ as x→1− and −∞ as x→−1+: vertical tangents at both endpoints
c)One point, x=4−1+7≈0.4114; x=4−1−7 rejected
d)x3+x−2=(x−1)(x2+x+2): only x=1, the point (1,π−1)
e)−2x(x+1)2=0: at (0,0) and (−1,−π−1)
a)
Differentiate term by term: dxdsin−1x=1−x21, and by the chain rule dxd(−21−x2)=−2⋅21−x2−2x=1−x22x. The two terms have the SAME denominator, so f′(x)=1−x21+2x. A fraction is 0 when its numerator is 0 and its denominator is not: 1+2x=0 gives x=−21, where 1−41=23=0. Then f(−21)=−6π−2⋅23=−6π−3≈−2.2556. The figure agrees: the lowest point of the curve is near (−0.5,−2.26). Leaving f′ as two fractions and setting each to 0 separately is the mistake that finds nothing.
b)
As x→1−, the numerator 1+2x→3 and the denominator 1−x2→0+, so f′(x)→+∞. As x→−1+, the numerator tends to −1 and the denominator to 0+, so f′(x)→−∞. The slopes grow without bound: the graph arrives VERTICALLY at both endpoints (1,2π) and (−1,−2π), which is what the figure shows. Numerically, f′(0.99)≈21.1 and f′(−0.99)≈−6.9.
c)
Solve 1−x21+2x=2, that is 1+2x=21−x2. The right side is positive, so any solution must have 1+2x>0: note it BEFORE squaring. Squaring: 1+4x+4x2=4−4x2, so 8x2+4x−3=0 and x=16−4±16+96=4−1±7. The candidate 4−1+7≈0.4114 has 1+2x≈1.82>0: kept, and f′(0.4114)≈2.000. The candidate 4−1−7≈−0.9114 has 1+2x≈−0.82<0: at that point f′=−2, not 2. Squaring merged the equations f′=2 and f′=−2; the check separates them again. One point only.
d)
g′(x)=1+x24−2x=1+x24−2x(1+x2)=1+x24−2x−2x3, over the common denominator 1+x2, never 0. Horizontal tangent: 4−2x−2x3=0, that is x3+x−2=0 after dividing by −2. x=1 is a root (1+1−2=0), so factor: x3+x−2=(x−1)(x2+x+2), and x2+x+2 has discriminant 1−8<0, no real root. Exactly one horizontal tangent, at x=1, where g(1)=4⋅4π−1=π−1≈2.1416.
e)
Parallel to y=4x means g′(x)=4: 1+x24−2x−2x3=4, so 4−2x−2x3=4+4x2, that is −2x3−4x2−2x=0. Factor −2x first, the common factor that a division by x would lose: −2x(x2+2x+1)=−2x(x+1)2=0. Two points: x=0, with g(0)=0, and x=−1, with g(−1)=4(−4π)−1=−π−1≈−4.1416. Check: g′(−1)=24+2=4. Dividing both sides by x at the start keeps only x=−1 and loses the tangent y=4x at the origin itself.
Exercise 8: Five statements to correct
Each statement below was written by a student on a MATH 203 assignment, and each is false. Say what is wrong, give a counterexample or the correct computation, and write the correct statement.
a) sin−1(sin65π)=65π, since sin−1 undoes sin.
b) dxdsec−1x=xx2−11 for every x with ∣x∣>1.
c) tan−1x=cos−1xsin−1x for −1<x<1, because tan=cossin.
d) dxdtan−1x1=1+x211=x2+1x2.
e) The domain of sin−1(3x) is [−1,1], and dxdsin−1(3x)=1−3x23.
Show the solution
Answers
a)False: sin65π=21 and sin−121=6π
b)False for x<−1: the derivative is ∣x∣x2−11, equal to +231 at x=−2
c)False: at x=21 the right side is 21, while tan−121≈0.4636
d)False: the chain factor −x21 is missing; dxdtan−1x1=−1+x21
sin−1 undoes sin only on [−2π,2π], and 65π is not in that interval. Compute from the inside: sin65π=21, and the angle of the range with sine 21 is 6π. So sin−1(sin65π)=6π. Correct statement: sin−1(sinx)=x for −2π≤x≤2π only, while sin(sin−1x)=x for every x of [−1,1]. The student's answer fails the one-second test: 65π>2π is not a possible value of sin−1.
b)
The formula is right for x>1 and wrong for x<−1. At x=−2 it gives −231<0, but sec−1 rises on its left branch, so its slope there is positive. The implicit derivation gives secytany1 with tany<0 on the left branch, and secytany=x⋅(−x2−1)=∣x∣x2−1. Correct statement: dxdsec−1x=∣x∣x2−11 for ∣x∣>1, which is 231≈0.2887 at x=−2.
c)
The identity tanθ=cosθsinθ is about ONE angle θ. The three inverse functions of the statement return three DIFFERENT angles for the same x, so nothing allows dividing them. Counterexample at x=21: cos−1(1/2)sin−1(1/2)=π/3π/6=21, while tan−121≈0.4636. At x=1 it is worse: cos−11=0, a division by 0. What IS true, by the triangle of the angle θ=sin−1x: tan(sin−1x)=1−x2x for −1<x<1, a trigonometric function of ONE angle.
d)
The formula 1+u21 was applied with u=x1, but the factor u′=−x21 of the chain rule was dropped. Correctly: dxdtan−1x1=1+1/x2−1/x2, and multiplying top and bottom by x2 (the complex fraction gesture) gives −x2+11 for x=0. At x=1 the student finds +21 and the true value is −21: a sign error that a single look at tan−1x1 reveals, since x1 decreases on (0,∞) and so does its arctangent.
e)
Two errors. The domain: sin−1u needs −1≤u≤1, here −1≤3x≤1, so −31≤x≤31. The derivative: u2=(3x)2=9x2, the square applies to the 3 as well, so dxdsin−1(3x)=1−9x23 for ∣x∣<31. At x=0.3 the student gets 0.733≈3.511 and the true value is 0.193≈6.8825. The student's formula is even defined at x=0.5, a point where the function itself does not exist: a derivative can never have a larger domain than its function.
Exercise 9: The angle under which a painting is seen, and its rate with distance
In a gallery, a painting 2 m tall hangs on a wall with its lower edge 1 m above the eye level of a visitor and its upper edge 3 m above. The visitor stands at a horizontal distance x metres from the wall, x>0. The angle θ under which the painting is seen is the angle at the eye E between the line of sight to the upper edge and the line of sight to the lower edge, as in the figure.
Scientific calculator allowed; angles in radians unless degrees are asked. For b), recall tan(A−B)=1+tanAtanBtanA−tanB. This exercise studies θ as a function of x and its rate dxdθ; where the angle is LARGEST is an optimization question for section 4.6, not asked here.
a) Explain why θ(x)=tan−1x3−tan−1x1. Compute θ(2) in radians, to four decimals, and in degrees, to two decimals.
b) Show that tanθ=x2+32x, and deduce that θ(x)=tan−1x2+32x. Why is it legitimate to apply tan−1 here? Give tanθ at x=2.
c) Show that dxdtan−1xa=−x2+a2a for a constant a>0, then write θ′(x) as a single fraction and evaluate θ′(1) and θ′(2).
d) Give the units of θ′(1) and its meaning in one sentence, then convert it to degrees per metre. At x=2, does the angle grow or shrink when the visitor steps back a little?
e) Differentiate the one-arctangent form of b) directly, and show that the two routes agree by factoring x4+10x2+9. Then evaluate θ′(3).
Show the solution
Answers
a)θ=β−α with tanβ=x3, tanα=x1; θ(2)≈0.5191 rad ≈29.74°
b)tanθ=1+3/x22/x=x2+32x, and 0<θ<2π; tanθ(2)=74
d)Radians per metre: at 1 m, the angle grows by about 0.2 rad per metre stepped back, ≈11.46° per metre; at 2 m it shrinks
e)x4+10x2+9=(x2+1)(x2+9); θ′(3)=−151≈−0.0667
a)
Let β be the angle between the horizontal line of sight and the line to the upper edge, and α the angle to the lower edge. In the right triangles of the figure, the side opposite β is 3 and the adjacent side is x, so tanβ=x3; likewise tanα=x1. Both angles are in (0,2π), the range of tan−1, so β=tan−1x3 and α=tan−1x1, and θ=β−α. At x=2: tan−11.5−tan−10.5≈0.982794−0.463648=0.519146, so θ(2)≈0.5191 rad. In degrees: 0.519146⋅π180≈29.74°. Keep the six-digit values until the end: subtracting two rounded angles loses a digit.
b)
With tanβ=x3 and tanα=x1: tanθ=1+x23x3−x1. This is a complex fraction, the gesture that costs marks in MATH 203: multiply its top and bottom by x2, the common denominator of every small fraction inside it. The top becomes 3x−x=2x and the bottom x2+3, so tanθ=x2+32x. Since 0<α<β<2π, the angle θ=β−α lies in (0,2π), inside the range of tan−1; so applying tan−1 gives back θ itself: θ=tan−1x2+32x. Without that check the step could be false, exactly as tan−1(tan43π)=43π. At x=2: tanθ=74≈0.5714, and tan−174≈0.5191, as in a).
c)
Inner function u=xa=ax−1, u′=−x2a. So dxdtan−1xa=1+a2/x2−a/x2; multiply top and bottom by x2: −x2+a2a. With a=3 and a=1: θ′(x)=−x2+93+x2+11. One fraction, over (x2+1)(x2+9): the numerator is −3(x2+1)+(x2+9)=6−2x2, so θ′(x)=(x2+1)(x2+9)2(3−x2). Watch the distribution of −3 over x2+1: writing −3x2+1 gives 10−2x2 and a wrong function. θ′(1)=2⋅104=0.2 and θ′(2)=5⋅13−2=−652≈−0.0308.
d)
θ is in radians and x in metres, so θ′(1) is in radians per metre. Meaning: when the visitor stands 1 m from the wall, the viewing angle increases at the rate of 0.2 rad for each metre of stepping back, that is, about 0.02 rad for a step of 10 cm. In degrees: 0.2⋅π180≈11.46° per metre. At x=2, θ′(2)≈−0.0308<0: stepping back a little makes the painting look slightly SMALLER. So the angle grows when one steps back from 1 m and shrinks when one steps back from 2 m; where exactly the change of behaviour happens, and why the angle is largest there, belongs to the optimization chapter.
e)
With u=x2+32x, the quotient rule gives u′=(x2+3)22(x2+3)−2x⋅2x=(x2+3)26−2x2, and 1+u2=(x2+3)2(x2+3)2+4x2=(x2+3)2x4+10x2+9. In the quotient 1+u2u′ the factors (x2+3)2 cancel: θ′(x)=x4+10x2+96−2x2. Factor the denominator as a quadratic in x2: two numbers with product 9 and sum 10 are 1 and 9, so x4+10x2+9=(x2+1)(x2+9), and the two routes agree. Then θ′(3)=10⋅186−18=−18012=−151≈−0.0667.
Exercise 10: A final exam question: one function, two corners, three formulas
Let f(x)=sin−11+x22x. The figure shows its graph: it climbs to 2π at x=1, where it turns sharply, and it is symmetric with respect to the origin. A final exam often takes one such function and asks every question of the chapter about it: domain, values, derivative with a square root to simplify, differentiability, and an identity proved by a zero derivative.
Exact answers, then decimals when asked. The fact 'a zero derivative on an interval gives a constant' is used as a tool.
a) Prove that −1≤1+x22x≤1 for every real x, so that f is defined on R. Compute f(1), f(−1) and f(3) exactly.
b) Show that 1−(1+x22x)2=(1+x2)2(1−x2)2, deduce that f′(x)=(1+x2)∣1−x2∣2(1−x2) for x=±1, and write f′ without absolute value on each interval. Evaluate f′(0) and f′(2).
c) Which slopes do the two pieces of the graph approach at x=1, from the left and from the right? Is f differentiable at x=1? Describe the graph there.
d) Compare f′ with the derivative of 2tan−1x. Deduce, interval by interval, that f(x)=2tan−1x on [−1,1] and f(x)=π−2tan−1x for x≥1, reading each constant at a point of its own interval. Find the formula for x≤−1.
e) Solve f(x)=3π and explain why there are two solutions although sin−1 returns only one angle. Then compute f(3) to four decimals in two ways.
b)(1+x2)2−4x2=(1−x2)2, root ∣1−x2∣; f′=1+x22 for ∣x∣<1, −1+x22 for ∣x∣>1; f′(0)=2, f′(2)=−52
c)Left slope 1, right slope −1: a corner, f is not differentiable at 1
d)f=2tan−1x on [−1,1]; f+2tan−1x=π on [1,∞); f=−π−2tan−1x on (−∞,−1]
e)x=31 and x=3; f(3)=sin−153=π−2tan−13≈0.6435
a)
Since 1+x2>0, the double inequality is equivalent to −(1+x2)≤2x≤1+x2. On the right: 1+x2−2x=(1−x)2≥0. On the left: 1+x2+2x=(1+x)2≥0. Both hold for every real x, so 1+x22x is always in [−1,1] and f is defined on R; the values ±1 are reached exactly at x=±1. Then f(1)=sin−11=2π, f(−1)=sin−1(−1)=−2π, and f(3)=sin−1423=sin−123=3π. Recognizing a perfect square is the whole proof: expanding and 'testing some values' proves nothing.
b)
1−(1+x2)24x2=(1+x2)2(1+x2)2−4x2, and the numerator is 1+2x2+x4−4x2=1−2x2+x4=(1−x2)2. Its square root is 1+x2∣1−x2∣: the root of a square is an ABSOLUTE VALUE, and here 1−x2 is negative as soon as ∣x∣>1. The inner derivative, by the quotient rule: u′=(1+x2)22(1+x2)−2x⋅2x=(1+x2)22(1−x2). So f′(x)=(1+x2)22(1−x2)⋅∣1−x2∣1+x2=(1+x2)∣1−x2∣2(1−x2) for x=±1. For ∣x∣<1, ∣1−x2∣=1−x2 and f′(x)=1+x22; for ∣x∣>1, ∣1−x2∣=x2−1 and f′(x)=−1+x22. f′(0)=2 and f′(2)=−52. The student who writes (1−x2)2=1−x2 finds 1+x22 everywhere, a positive slope at x=2 where the figure goes down.
c)
Just left of 1, f′(x)=1+x22, which approaches 22=1. Just right of 1, f′(x)=−1+x22, which approaches −1. The two pieces arrive at the point (1,2π) with slopes 1 and −1: the graph has a CORNER there, and f is continuous but not differentiable at x=1. In the formula of b), the division by ∣1−x2∣=0 at x=1 was the warning. By the symmetry of the graph, the same happens at x=−1.
d)
dxd(2tan−1x)=1+x22. On the interval (−1,1), f′−2(tan−1)′=0, so f(x)−2tan−1x is constant there; at x=0 it is 0−0=0. So f=2tan−1x on (−1,1), and also at ±1: f(1)=2π=2⋅4π. On (1,∞), f′+2(tan−1)′=0, so f(x)+2tan−1x is constant; read it at x=3, a point of THAT interval: 3π+2⋅3π=π. So f(x)=π−2tan−1x for x≥1. On (−∞,−1), same derivative relation, constant read at x=−3: f(−3)+2tan−1(−3)=−3π−32π=−π, so f(x)=−π−2tan−1x for x≤−1. Three intervals, three constants 0, π, −π: reading the first one at x=0 and extending it to the whole line would claim f(3)=32π, a value that sin−1 can never take, since it exceeds 2π.
e)
f(x)=3π means 1+x22x=sin3π=23, legitimate since 3π is in the range of sin−1. Then 4x=3(1+x2), that is 3x2−4x+3=0, with discriminant 16−12=4: x=234±2, so x=3≈1.7321 or x=31≈0.5774. The formulas of d) agree: 2tan−1x=3π gives 31 on [−1,1], and π−2tan−1x=3π gives 3 on [1,∞). sin−1 returns one angle, but the INNER function 1+x22x takes the value 23 twice, once on each side of x=1. Finally f(3)=sin−1106=sin−10.6≈0.6435, and π−2tan−13≈3.141593−2.498092=0.643501: the same, to four decimals 0.6435.
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.