Differential Calculus 201-SN2-RE • CEGEP, Montreal

Corrected exercises: derivatives of trigonometric, exponential and logarithmic functions (201-SN2-RE)

This is the corrected exercise set on the derivatives of the trigonometric functions and the other transcendental functions for Differential Calculus, 201-SN2-RE (formerly 201-NYA): sine, cosine, tangent, cotangent, secant and cosecant (written tg, cotg and cosec in several French-language Quebec textbooks), their inverses arcsine, arccosine and arctangent, then the exponential, the logarithm, logarithmic differentiation and the hyperbolic functions. This is the chapter that turns differentiation from a catalogue of polynomials into a tool you can use in physics, chemistry and biology.

Exercise 0.1 opens the set with twelve derivatives to compute directly from the formulas, as in textbook drill lists: composite cotangent, secant and cosecant, sec⁡2(3x)\sec^{2}(3x) and csc⁡(x2)\sqrt{\csc(x^{2})}, ecos⁡(3x)e^{\cos(3x)} and 5tan⁡x5^{\tan x}, sin⁡(x3)+sin⁡3x\sin(x^{3})+\sin^{3}x, arctan⁡(x)\arctan(\sqrt{x}), each with a value to find at the given point. By the end of the set, you can rebuild the table of the six trigonometric derivatives from two of them (exercise 1), differentiate axa^{x} without forgetting the factor ln⁡a\ln a (exercise 2), simplify ln⁡(sec⁡x+tan⁡x)\ln(\sec x+\tan x), whose derivative is sec⁡x\sec x (exercise 3), differentiate xxx^{x} or xsin⁡xx^{\sin x} by logarithmic differentiation and know when it is pointless (exercise 5), prove the derivatives of the inverse functions by implicit differentiation and handle the domain of arccos⁡(3x)\arccos\left(\frac{3}{x}\right) (exercise 6), and describe a hanging cable, the catenary, with the hyperbolic cosine (exercise 7).

Real models arrive in Part B: carbon-14 dating and a cooling coffee, where the relative rate stays constant but the rate of change itself does not (exercise 9), then the logistic spread of a rumour and its maximum rate of spread, found without the second derivative (exercise 10). Part C adds the tangent lines to the exponential and the logarithm that pass through the origin (exercise 11), identities of the inverse functions such as arcsin⁡x+arccos⁡x\arcsin x+\arccos x (exercise 12), the extrema of exsin⁡xe^{x}\sin x (exercise 13), the tide and the rule of twelfths (exercise 14), and the best viewing angle of a screen (exercise 15).

The thread of the set fits in one sentence: the chain rule never changes, it is the starting TABLE that changes, and each family brings its own trap. The base logarithm factor for the exponential, the derivative of the inside for the logarithm, the choice of the sign of the square root for the inverse functions, the absence of any minus sign for the hyperbolic functions. The solutions deal with the mistakes that cost the most: cos⁡(x2)\cos(x^{2}) confused with cos⁡2x\cos^{2}x (exercise 1), the two minus signs of csc⁡3(1−4t)\csc^{3}(1-4t) only one of which gets kept at random (exercise 1), (ex)2(e^{x})^{2} taken for ex2e^{x^{2}} (exercise 2) and the derivative of ln⁡(x2)\ln(x^{2}) reduced to 1x2\frac{1}{x^{2}} (exercise 8).

16 corrected exercises • 160 points • 225 minutes

This chapter is part of 201-SN2-RE, every chapter with its set and sheet Before you start Revision sheet: the traps and the method for this chapter

Self-checking set Type 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.

Course recap

  • • TRIGONOMETRY: (sin⁡x)′=cos⁡x(\sin x)'=\cos x, (cos⁡x)′=−sin⁡x(\cos x)'=-\sin x, (tan⁡x)′=sec⁡2x=1+tan⁡2x(\tan x)'=\sec^{2}x=1+\tan^{2}x, (cot⁡x)′=−csc⁡2x(\cot x)'=-\csc^{2}x, (sec⁡x)′=sec⁡xtan⁡x(\sec x)'=\sec x\tan x, (csc⁡x)′=−csc⁡xcot⁡x(\csc x)'=-\csc x\cot x. The “co” functions carry the minus sign.
  • • EXPONENTIAL AND LOGARITHM: (eu)′=u′eu(e^{u})'=u'e^{u}, (ax)′=axln⁡a(a^{x})'=a^{x}\ln a, (ln⁡u)′=u′u(\ln u)'=\frac{u'}{u}, (log⁡ax)′=1xln⁡a(\log_{a}x)'=\frac{1}{x\ln a}, (ln⁡∣x∣)′=1x(\ln|x|)'=\frac{1}{x} for x≠0x\neq 0. The number ee is the only base whose tangent line at 0 has slope 1.
  • • LOGARITHMIC DIFFERENTIATION: for y=uvy=u^{v}, or for a product or quotient of at least three factors with exponents, take ln⁡\ln of both sides, differentiate, then multiply by yy. The quantity y′y\frac{y'}{y} is the RELATIVE rate of change.
  • • INVERSE FUNCTIONS: (arcsin⁡x)′=11−x2(\arcsin x)'=\frac{1}{\sqrt{1-x^{2}}} for ∣x∣<1|x|<1, (arccos⁡x)′=−11−x2(\arccos x)'=\frac{-1}{\sqrt{1-x^{2}}}, (arctan⁡x)′=11+x2(\arctan x)'=\frac{1}{1+x^{2}} on R\mathbb{R}. The domain of the derivative is the INTERIOR of the domain of the function: at the endpoints, the tangent line is vertical.
  • • HYPERBOLIC FUNCTIONS: cosh⁡x=ex+e−x2\cosh x=\frac{e^{x}+e^{-x}}{2}, sinh⁡x=ex−e−x2\sinh x=\frac{e^{x}-e^{-x}}{2}, (cosh⁡x)′=sinh⁡x(\cosh x)'=\sinh x, (sinh⁡x)′=cosh⁡x(\sinh x)'=\cosh x, WITHOUT a minus sign, and cosh⁡2x−sinh⁡2x=1\cosh^{2}x-\sinh^{2}x=1. Consequence: cos⁡\cos and sin⁡\sin satisfy y′′=−yy''=-y and oscillate, cosh⁡\cosh and sinh⁡\sinh satisfy y′′=yy''=y and grow.
  • • MODELS: y′=kyy'=ky gives the exponential y=y0ekty=y_{0}e^{kt}, with a constant RELATIVE rate y′y\frac{y'}{y}. y′=k(y−a)y'=k(y-a) gives Newton's law of cooling, where it is the GAP from aa that drives the change. y′=rKy(K−y)y'=\frac{r}{K}y(K-y) gives the logistic model, with maximum growth rate rK4\frac{rK}{4} reached at y=K2y=\frac{K}{2}.

Calculations: derivatives to compute

Exercise 0.1: Twelve derivatives to compute directly

For each function, compute the derivative with the differentiation formulas, without simplifying more than necessary, then evaluate it at the given point. The notations tg, cotg and cosec found in some French-language textbooks stand for tan⁡\tan, cot⁡\cot and csc⁡\csc.

One method for all twelve lines: spot the LAST operation (sum, product, quotient, power or composite function), apply the rule that goes with it, then multiply by the derivative of the inside at every layer. The calculator, in radian mode, is only for the final value.

  • a) f(x)=cot⁡(5x2+x)f(x)=\cot(5x^{2}+x), then f′(0.3)f'(0.3).
  • b) g(t)=t4csc⁡(2t)g(t)=t^{4}\csc(2t), then g′(1)g'(1).
  • c) h(x)=sec⁡2(3x)h(x)=\sec^{2}(3x), then h′(0.2)h'(0.2).
  • d) y=csc⁡(x2)y=\sqrt{\csc(x^{2})}, then y′(1)y'(1).
  • e) f(x)=ecos⁡(3x)f(x)=e^{\cos(3x)}, then f′(0.5)f'(0.5).
  • f) g(x)=5tan⁡xg(x)=5^{\tan x}, then g′(π4)g'\left(\frac{\pi}{4}\right).
  • g) h(t)=ln⁡(csc⁡t)+csc⁡(ln⁡t)h(t)=\ln(\csc t)+\csc(\ln t), then h′(2)h'(2).
  • h) s(x)=sin⁡(x3)+sin⁡3xs(x)=\sin(x^{3})+\sin^{3}x, then s′(1)s'(1).
  • i) f(x)=arctan⁡(x)f(x)=\arctan(\sqrt{x}), then f′(3)f'(3).
  • j) g(x)=arccos⁡(2x)x2+1g(x)=\frac{\arccos(2x)}{x^{2}+1}, then g′(0.25)g'(0.25).
  • k) h(x)=e2xcos⁡(x2)h(x)=e^{2x}\cos(x^{2}), then h′(1)h'(1).
  • l) y=1+sec⁡(2x)1−tan⁡(2x)y=\frac{1+\sec(2x)}{1-\tan(2x)}, then y′(0.1)y'(0.1).

Type your answers, the page tells you right or wrong 0/12

a)
b)
c)
d)
e)
f)
g)
h)
i)
j)
k)
l)
Show the solution

Answers

  • a) f′(x)=−(10x+1)csc⁡2(5x2+x)f'(x)=-(10x+1)\csc^{2}(5x^{2}+x); f′(0.3)≈−8.609f'(0.3)\approx-8.609
  • b) g′(t)=2t3csc⁡(2t)(2−tcot⁡(2t))g'(t)=2t^{3}\csc(2t)\left(2-t\cot(2t)\right); g′(1)≈5.406g'(1)\approx 5.406
  • c) h′(x)=6sec⁡2(3x)tan⁡(3x)h'(x)=6\sec^{2}(3x)\tan(3x); h′(0.2)≈6.026h'(0.2)\approx 6.026
  • d) y′=−xcot⁡(x2)csc⁡(x2)y'=-x\cot(x^{2})\sqrt{\csc(x^{2})}; y′(1)≈−0.700y'(1)\approx-0.700
  • e) f′(x)=−3sin⁡(3x) ecos⁡(3x)f'(x)=-3\sin(3x)\,e^{\cos(3x)}; f′(0.5)≈−3.212f'(0.5)\approx-3.212
  • f) g′(x)=5tan⁡xln⁡5sec⁡2xg'(x)=5^{\tan x}\ln 5\sec^{2}x; g′(π4)=10ln⁡5≈16.094g'\left(\frac{\pi}{4}\right)=10\ln 5\approx 16.094
  • g) h′(t)=−cot⁡t−csc⁡(ln⁡t)cot⁡(ln⁡t)th'(t)=-\cot t-\frac{\csc(\ln t)\cot(\ln t)}{t}; h′(2)≈−0.484h'(2)\approx-0.484
  • h) s′(x)=3x2cos⁡(x3)+3sin⁡2xcos⁡xs'(x)=3x^{2}\cos(x^{3})+3\sin^{2}x\cos x; s′(1)≈2.769s'(1)\approx 2.769
  • i) f′(x)=12x(1+x)f'(x)=\frac{1}{2\sqrt{x}(1+x)}; f′(3)=324≈0.0722f'(3)=\frac{\sqrt{3}}{24}\approx 0.0722
  • j) g′(x)=−2(x2+1)1−4x2−2xarccos⁡(2x)(x2+1)2g'(x)=\frac{\frac{-2(x^{2}+1)}{\sqrt{1-4x^{2}}}-2x\arccos(2x)}{(x^{2}+1)^{2}}; g′(0.25)≈−2.637g'(0.25)\approx-2.637
  • k) h′(x)=2e2x(cos⁡(x2)−xsin⁡(x2))h'(x)=2e^{2x}\left(\cos(x^{2})-x\sin(x^{2})\right); h′(1)≈−4.451h'(1)\approx-4.451
  • l) y′=2sec⁡(2x)tan⁡(2x)(1−tan⁡(2x))+2sec⁡2(2x)(1+sec⁡(2x))(1−tan⁡(2x))2y'=\frac{2\sec(2x)\tan(2x)(1-\tan(2x))+2\sec^{2}(2x)(1+\sec(2x))}{(1-\tan(2x))^{2}}; y′(0.1)≈7.137y'(0.1)\approx 7.137

The move that decides everything happens BEFORE differentiating: name the last operation. t4csc⁡(2t)t^{4}\csc(2t) is a product, sec⁡2(3x)\sec^{2}(3x) a power, ecos⁡(3x)e^{\cos(3x)} a composite exponential, arccos⁡(2x)x2+1\frac{\arccos(2x)}{x^{2}+1} a quotient. Once the rule is chosen, count the layers: each one leaves its inner derivative as a factor.

a)

Composite function, outer function cot⁡\cot: f′(x)=−csc⁡2(5x2+x)⋅(10x+1)f'(x)=-\csc^{2}(5x^{2}+x)\cdot(10x+1). At x=0.3x=0.3, the argument equals 0.750.75 and 10x+1=410x+1=4, so f′(0.3)=−4sin⁡2(0.75)≈−8.609f'(0.3)=-\frac{4}{\sin^{2}(0.75)}\approx-8.609. Pitfall: forgetting the minus sign of the cotangent, which gives +8.609+8.609.

b)

Product first: g′(t)=4t3csc⁡(2t)+t4⋅(−csc⁡(2t)cot⁡(2t))⋅2=2t3csc⁡(2t)(2−tcot⁡(2t))g'(t)=4t^{3}\csc(2t)+t^{4}\cdot\left(-\csc(2t)\cot(2t)\right)\cdot 2=2t^{3}\csc(2t)\left(2-t\cot(2t)\right). At t=1t=1: g′(1)=2csc⁡2 (2−cot⁡2)≈2×1.0998×2.4577≈5.406g'(1)=2\csc 2\,(2-\cot 2)\approx 2\times 1.0998\times 2.4577\approx 5.406. Careful, cot⁡2\cot 2 is NEGATIVE (22 radians is more than π2\frac{\pi}{2}), hence the 2.45772.4577.

c)

Power first, h(x)=(sec⁡(3x))2h(x)=\left(\sec(3x)\right)^{2}: h′(x)=2sec⁡(3x)⋅sec⁡(3x)tan⁡(3x)⋅3=6sec⁡2(3x)tan⁡(3x)h'(x)=2\sec(3x)\cdot\sec(3x)\tan(3x)\cdot 3=6\sec^{2}(3x)\tan(3x). At x=0.2x=0.2: h′(0.2)=6tan⁡(0.6)cos⁡2(0.6)≈6.026h'(0.2)=\frac{6\tan(0.6)}{\cos^{2}(0.6)}\approx 6.026. Three layers, three factors: the 22 of the power, the derivative of the secant, the 33.

d)

The radical becomes a power: y=(csc⁡(x2))12y=\left(\csc(x^{2})\right)^{\frac{1}{2}}. So y′=12(csc⁡(x2))−12⋅(−csc⁡(x2)cot⁡(x2))⋅2x=−xcot⁡(x2)csc⁡(x2)y'=\frac{1}{2}\left(\csc(x^{2})\right)^{-\frac{1}{2}}\cdot\left(-\csc(x^{2})\cot(x^{2})\right)\cdot 2x=-x\cot(x^{2})\sqrt{\csc(x^{2})}, since csc⁡ucsc⁡u=csc⁡u\frac{\csc u}{\sqrt{\csc u}}=\sqrt{\csc u}. At x=1x=1: y′(1)=−cot⁡1csc⁡1≈−0.700y'(1)=-\cot 1\sqrt{\csc 1}\approx-0.700.

e)

f′(x)=ecos⁡(3x)⋅(−sin⁡(3x))⋅3=−3sin⁡(3x) ecos⁡(3x)f'(x)=e^{\cos(3x)}\cdot\left(-\sin(3x)\right)\cdot 3=-3\sin(3x)\,e^{\cos(3x)}. The exponential is copied as it is; it is its exponent that gets differentiated. At x=0.5x=0.5: f′(0.5)=−3sin⁡(1.5) ecos⁡(1.5)≈−3.212f'(0.5)=-3\sin(1.5)\,e^{\cos(1.5)}\approx-3.212.

f)

Constant base, variable exponent: (au)′=auln⁡a⋅u′(a^{u})'=a^{u}\ln a\cdot u', so g′(x)=5tan⁡xln⁡5⋅sec⁡2xg'(x)=5^{\tan x}\ln 5\cdot\sec^{2}x. At x=π4x=\frac{\pi}{4}, tan⁡x=1\tan x=1 and sec⁡2x=2\sec^{2}x=2: g′(π4)=10ln⁡5≈16.094g'\left(\frac{\pi}{4}\right)=10\ln 5\approx 16.094. Pitfall: the power rule, tan⁡x⋅5tan⁡x−1\tan x\cdot 5^{\tan x-1}, which only applies when the exponent is constant.

g)

Two terms, two compositions, in opposite orders. (ln⁡(csc⁡t))′=−csc⁡tcot⁡tcsc⁡t=−cot⁡t\left(\ln(\csc t)\right)'=\frac{-\csc t\cot t}{\csc t}=-\cot t: the logarithm first, then the simplification. (csc⁡(ln⁡t))′=−csc⁡(ln⁡t)cot⁡(ln⁡t)⋅1t\left(\csc(\ln t)\right)'=-\csc(\ln t)\cot(\ln t)\cdot\frac{1}{t}: the cosecant first, then the derivative of ln⁡t\ln t. So h′(t)=−cot⁡t−csc⁡(ln⁡t)cot⁡(ln⁡t)th'(t)=-\cot t-\frac{\csc(\ln t)\cot(\ln t)}{t}, and h′(2)≈0.458−0.942=−0.484h'(2)\approx 0.458-0.942=-0.484.

h)

sin⁡(x3)\sin(x^{3}): the power is INSIDE, hence cos⁡(x3)⋅3x2\cos(x^{3})\cdot 3x^{2}. sin⁡3x=(sin⁡x)3\sin^{3}x=(\sin x)^{3}: the power is OUTSIDE, hence 3sin⁡2xcos⁡x3\sin^{2}x\cos x. So s′(x)=3x2cos⁡(x3)+3sin⁡2xcos⁡xs'(x)=3x^{2}\cos(x^{3})+3\sin^{2}x\cos x, and s′(1)=3cos⁡1(1+sin⁡21)≈2.769s'(1)=3\cos 1\left(1+\sin^{2}1\right)\approx 2.769.

i)

f′(x)=11+(x)2⋅12x=12x(1+x)f'(x)=\frac{1}{1+(\sqrt{x})^{2}}\cdot\frac{1}{2\sqrt{x}}=\frac{1}{2\sqrt{x}(1+x)}, for x>0x>0. At x=3x=3: f′(3)=183=324≈0.0722f'(3)=\frac{1}{8\sqrt{3}}=\frac{\sqrt{3}}{24}\approx 0.0722. The square of x\sqrt{x} is xx: that is what makes the final form so short.

j)

Quotient, with u=arccos⁡(2x)u=\arccos(2x), u′=−21−4x2u'=\frac{-2}{\sqrt{1-4x^{2}}}, v=x2+1v=x^{2}+1, v′=2xv'=2x: g′(x)=−21−4x2(x2+1)−2xarccos⁡(2x)(x2+1)2g'(x)=\frac{\frac{-2}{\sqrt{1-4x^{2}}}(x^{2}+1)-2x\arccos(2x)}{(x^{2}+1)^{2}}, for ∣x∣<12|x|<\frac{1}{2}. At x=0.25x=0.25: u′=−20.75≈−2.309u'=\frac{-2}{\sqrt{0.75}}\approx-2.309 and arccos⁡(0.5)=π3\arccos(0.5)=\frac{\pi}{3}, so g′(0.25)=−2.309×1.0625−0.5×π31.06252≈−2.637g'(0.25)=\frac{-2.309\times 1.0625-0.5\times\frac{\pi}{3}}{1.0625^{2}}\approx-2.637. Pitfall: the minus sign in the derivative of arccos⁡\arccos.

k)

Product: h′(x)=2e2xcos⁡(x2)+e2x⋅(−sin⁡(x2))⋅2x=2e2x(cos⁡(x2)−xsin⁡(x2))h'(x)=2e^{2x}\cos(x^{2})+e^{2x}\cdot\left(-\sin(x^{2})\right)\cdot 2x=2e^{2x}\left(\cos(x^{2})-x\sin(x^{2})\right). At x=1x=1: h′(1)=2e2(cos⁡1−sin⁡1)≈−4.451h'(1)=2e^{2}(\cos 1-\sin 1)\approx-4.451.

l)

Quotient: the numerator 1+sec⁡(2x)1+\sec(2x) has derivative 2sec⁡(2x)tan⁡(2x)2\sec(2x)\tan(2x), the denominator 1−tan⁡(2x)1-\tan(2x) has derivative −2sec⁡2(2x)-2\sec^{2}(2x). So y′=2sec⁡(2x)tan⁡(2x)(1−tan⁡(2x))+2sec⁡2(2x)(1+sec⁡(2x))(1−tan⁡(2x))2y'=\frac{2\sec(2x)\tan(2x)\left(1-\tan(2x)\right)+2\sec^{2}(2x)\left(1+\sec(2x)\right)}{\left(1-\tan(2x)\right)^{2}}, the minus of the quotient rule and the minus of v′v' giving a PLUS. At x=0.1x=0.1: y′(0.1)≈7.137y'(0.1)\approx 7.137.

Overall check: for each line, the rate f(a+0.001)−f(a−0.001)0.002\frac{f(a+0.001)-f(a-0.001)}{0.002} computed on the calculator lands within a thousandth of the value found. If it does not, look first for a forgotten inner factor, then for the minus sign of a “co” function.

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.

Part A: the basics (/50)

Exercise 1: The table of the six trigonometric derivatives, rebuilt

You do not learn six derivatives by heart: you learn TWO, those of sine and cosine, and the other four come out of the quotient rule in three lines. This is the thread of the whole set: the chain rule never changes, only the starting table changes.

-3-2-1123-5-4-3-2-112345y = tan x
  • a) State (sin⁡x)′(\sin x)' and (cos⁡x)′(\cos x)'. Writing tan⁡x=sin⁡xcos⁡x\tan x=\frac{\sin x}{\cos x}, prove that (tan⁡x)′=sec⁡2x(\tan x)'=\sec^{2}x, then show that this expression also equals 1+tan⁡2x1+\tan^{2}x.
  • b) In the same way, establish (sec⁡x)′=sec⁡xtan⁡x(\sec x)'=\sec x\tan x and (cot⁡x)′=−csc⁡2x(\cot x)'=-\csc^{2}x.
  • c) Differentiate f(x)=sin⁡(3x)f(x)=\sin(3x), g(x)=cos⁡2xg(x)=\cos^{2}x and h(x)=tan⁡(x2+1)h(x)=\tan(x^{2}+1).
  • d) The figure shows y=tan⁡xy=\tan x. At which points is the function undefined, and why can its derivative NEVER be zero?
  • e) In several French-language Quebec textbooks, tan⁡\tan is written tg, cot⁡\cot is written cotg and csc⁡\csc is written cosec: they are the same functions. Differentiate f(x)=cotg(x2−2x)f(x)=\mathrm{cotg}(x^{2}-2x) and g(t)=t3sec⁡(2t)g(t)=t^{3}\sec(2t), then compute g′(π8)g'\left(\frac{\pi}{8}\right).
  • f) Differentiate h(t)=csc⁡3(1−4t)h(t)=\csc^{3}(1-4t), counting the minus signs. Then rewrite k(θ)=tan⁡2(5θ)3k(\theta)=\sqrt[3]{\tan^{2}(5\theta)} as a power before differentiating it, compute k′(π20)k'\left(\frac{\pi}{20}\right), and say what happens at θ=0\theta=0.
  • g) Differentiate s(x)=cos⁡(x2)+cos⁡2xs(x)=\cos(x^{2})+\cos^{2}x and compute s′(1)s'(1). What distinguishes the way the two terms are written?

Type your answers, the page tells you right or wrong 0/16

a)
b)
c)
d)
e)
f)
g)
Show the solution

Answers

  • a) (tan⁡x)′=sec⁡2x=1+tan⁡2x(\tan x)'=\sec^{2}x=1+\tan^{2}x
  • b) (sec⁡x)′=sec⁡xtan⁡x(\sec x)'=\sec x\tan x; (cot⁡x)′=−csc⁡2x(\cot x)'=-\csc^{2}x
  • c) 3cos⁡(3x)3\cos(3x); −sin⁡(2x)-\sin(2x); 2xsec⁡2(x2+1)2x\sec^{2}(x^{2}+1)
  • d) Undefined at π2+kπ\frac{\pi}{2}+k\pi; derivative ≥1\ge 1
  • e) f′(x)=−(2x−2)csc⁡2(x2−2x)f'(x)=-(2x-2)\csc^{2}(x^{2}-2x); g′(t)=t2sec⁡(2t)(3+2ttan⁡(2t))g'(t)=t^{2}\sec(2t)\left(3+2t\tan(2t)\right), g′(π8)≈0.826g'\left(\frac{\pi}{8}\right)\approx 0.826
  • f) h′(t)=12csc⁡3(1−4t)cot⁡(1−4t)h'(t)=12\csc^{3}(1-4t)\cot(1-4t); k′(π20)=203k'\left(\frac{\pi}{20}\right)=\frac{20}{3}; cusp at θ=0\theta=0
  • g) s′(x)=−2xsin⁡(x2)−sin⁡(2x)s'(x)=-2x\sin(x^{2})-\sin(2x); s′(1)≈−2.592s'(1)\approx-2.592
a)

(sin⁡x)′=cos⁡x(\sin x)'=\cos x and (cos⁡x)′=−sin⁡x(\cos x)'=-\sin x: the minus sign is on the cosine, and that is the only thing to remember. Quotient rule: (tan⁡x)′=cos⁡x⋅cos⁡x−sin⁡x⋅(−sin⁡x)cos⁡2x=cos⁡2x+sin⁡2xcos⁡2x=1cos⁡2x=sec⁡2x(\tan x)'=\frac{\cos x\cdot\cos x-\sin x\cdot(-\sin x)}{\cos^{2}x}=\frac{\cos^{2}x+\sin^{2}x}{\cos^{2}x}=\frac{1}{\cos^{2}x}=\sec^{2}x. Splitting the fraction term by term instead: cos⁡2x+sin⁡2xcos⁡2x=1+tan⁡2x\frac{\cos^{2}x+\sin^{2}x}{\cos^{2}x}=1+\tan^{2}x. The two forms are the same thing, and you choose the one that simplifies what follows: sec⁡2x\sec^{2}x when integrating, 1+tan⁡2x1+\tan^{2}x when the problem already gives tan⁡x\tan x. Numerical check at x=0.6x=0.6: sec⁡2(0.6)≈1.4718\sec^{2}(0.6)\approx 1.4718 and 1+tan⁡2(0.6)≈1+0.4718=1.47181+\tan^{2}(0.6)\approx 1+0.4718=1.4718 ✓.

b)

sec⁡x=1cos⁡x\sec x=\frac{1}{\cos x}, so by the quotient rule (sec⁡x)′=0⋅cos⁡x−1⋅(−sin⁡x)cos⁡2x=sin⁡xcos⁡2x=1cos⁡x⋅sin⁡xcos⁡x=sec⁡xtan⁡x(\sec x)'=\frac{0\cdot\cos x-1\cdot(-\sin x)}{\cos^{2}x}=\frac{\sin x}{\cos^{2}x}=\frac{1}{\cos x}\cdot\frac{\sin x}{\cos x}=\sec x\tan x. Likewise cot⁡x=cos⁡xsin⁡x\cot x=\frac{\cos x}{\sin x} gives (cot⁡x)′=−sin⁡x⋅sin⁡x−cos⁡xcos⁡xsin⁡2x=−1sin⁡2x=−csc⁡2x(\cot x)'=\frac{-\sin x\cdot\sin x-\cos x\cos x}{\sin^{2}x}=\frac{-1}{\sin^{2}x}=-\csc^{2}x. Remember the symmetry: the three “co” functions (cosine, cotangent, cosecant) all carry a minus sign.

c)

f′(x)=3cos⁡(3x)f'(x)=3\cos(3x): chain rule, the inside 3x3x has derivative 3. g(x)=(cos⁡x)2g(x)=(\cos x)^{2} so g′(x)=2cos⁡x⋅(−sin⁡x)=−2sin⁡xcos⁡x=−sin⁡(2x)g'(x)=2\cos x\cdot(-\sin x)=-2\sin x\cos x=-\sin(2x): the compact form is shorter, and it confirms the computation through the double-angle identity. h′(x)=sec⁡2(x2+1)⋅2x=2xcos⁡2(x2+1)h'(x)=\sec^{2}(x^{2}+1)\cdot 2x=\frac{2x}{\cos^{2}(x^{2}+1)}. The classic trap is to write g′(x)=−2sin⁡xg'(x)=-2\sin x, forgetting the chain rule: check at x=1x=1, g′(1)=−sin⁡2≈−0.909g'(1)=-\sin 2\approx-0.909, whereas −2sin⁡1≈−1.683-2\sin 1\approx-1.683, almost twice as large.

d)

tan⁡x\tan x is undefined where cos⁡x=0\cos x=0, that is, at x=π2+kπx=\frac{\pi}{2}+k\pi: these are the two vertical asymptotes in the figure. Its derivative sec⁡2x=1cos⁡2x\sec^{2}x=\frac{1}{\cos^{2}x} has a square in the denominator, so it is strictly positive wherever it exists, and it is at least 1 (reached when cos⁡2x=1\cos^{2}x=1). Consequence read on the graph: the tangent function is strictly increasing on each branch and has NO horizontal tangent line, hence no local extremum. A function can be increasing on each piece without being increasing on its whole domain: between two branches, it drops from +∞+\infty to −∞-\infty.

e)

The notation changes nothing: cotg is cot⁡\cot, and its derivative carries the minus sign of the “co” functions. Chain rule: f′(x)=−csc⁡2(x2−2x)⋅(2x−2)=−(2x−2)csc⁡2(x2−2x)f'(x)=-\csc^{2}(x^{2}-2x)\cdot(2x-2)=-(2x-2)\csc^{2}(x^{2}-2x), wherever sin⁡(x2−2x)≠0\sin(x^{2}-2x)\neq 0. For gg, it is a PRODUCT before being a composition: g′(t)=3t2sec⁡(2t)+t3⋅sec⁡(2t)tan⁡(2t)⋅2=t2sec⁡(2t)(3+2ttan⁡(2t))g'(t)=3t^{2}\sec(2t)+t^{3}\cdot\sec(2t)\tan(2t)\cdot 2=t^{2}\sec(2t)\left(3+2t\tan(2t)\right). At t=π8t=\frac{\pi}{8}, 2t=π42t=\frac{\pi}{4}, so sec⁡(2t)=2\sec(2t)=\sqrt{2} and tan⁡(2t)=1\tan(2t)=1: g′(π8)=π2642(3+π4)≈0.826g'\left(\frac{\pi}{8}\right)=\frac{\pi^{2}}{64}\sqrt{2}\left(3+\frac{\pi}{4}\right)\approx 0.826. The trap is to differentiate each factor and multiply, 3t2⋅2sec⁡(2t)tan⁡(2t)3t^{2}\cdot 2\sec(2t)\tan(2t), which gives about 1.3091.309: the derivative of a product is not the product of the derivatives.

f)

Three layers, the power, the cosecant and 1−4t1-4t: h′(t)=3csc⁡2(1−4t)⋅(−csc⁡(1−4t)cot⁡(1−4t))⋅(−4)=12csc⁡3(1−4t)cot⁡(1−4t)h'(t)=3\csc^{2}(1-4t)\cdot\left(-\csc(1-4t)\cot(1-4t)\right)\cdot(-4)=12\csc^{3}(1-4t)\cot(1-4t). Two minus signs, that of the cosecant and that of u′=−4u'=-4, which cancel: write them both down before simplifying, otherwise you keep one of them at random. For kk, the radical becomes an exponent: k(θ)=(tan⁡(5θ))23k(\theta)=\left(\tan(5\theta)\right)^{\frac{2}{3}}, so k′(θ)=23(tan⁡(5θ))−13⋅sec⁡2(5θ)⋅5=10sec⁡2(5θ)3tan⁡(5θ)3k'(\theta)=\frac{2}{3}\left(\tan(5\theta)\right)^{-\frac{1}{3}}\cdot\sec^{2}(5\theta)\cdot 5=\frac{10\sec^{2}(5\theta)}{3\sqrt[3]{\tan(5\theta)}}. At θ=π20\theta=\frac{\pi}{20}, 5θ=π45\theta=\frac{\pi}{4}, tan⁡(5θ)=1\tan(5\theta)=1 and sec⁡2(5θ)=2\sec^{2}(5\theta)=2: k′(π20)=203≈6.667k'\left(\frac{\pi}{20}\right)=\frac{20}{3}\approx 6.667. The negative exponent −13-\frac{1}{3} says what the radical was hiding: at θ=0\theta=0, tan⁡(5θ)=0\tan(5\theta)=0 and the derivative does not exist. The function equals 0 there and is continuous, but its graph has a CUSP there, the same one as x23\sqrt[3]{x^{2}} at the origin.

g)

The two terms look alike and have nothing in common. In cos⁡(x2)\cos(x^{2}), the square is INSIDE: the outer function is the cosine, hence −sin⁡(x2)⋅2x-\sin(x^{2})\cdot 2x. In cos⁡2x=(cos⁡x)2\cos^{2}x=(\cos x)^{2}, the square is OUTSIDE: the outer function is the power, hence 2cos⁡x⋅(−sin⁡x)=−sin⁡(2x)2\cos x\cdot(-\sin x)=-\sin(2x). So s′(x)=−2xsin⁡(x2)−sin⁡(2x)s'(x)=-2x\sin(x^{2})-\sin(2x) and s′(1)=−2sin⁡1−sin⁡2≈−1.683−0.909=−2.592s'(1)=-2\sin 1-\sin 2\approx-1.683-0.909=-2.592. The reflex: before differentiating, rewrite cos⁡2x\cos^{2}x as (cos⁡x)2(\cos x)^{2}; the parentheses say which function is outside.

Exercise 2: The exponential: what the base changes, and what it does not

All exponentials look alike: they pass through the point (0, 1) and never take the value zero. Only one is its own derivative, and that is what defines the number ee.

-2-112123456common point (0, 1)different slopes3ˣeˣ2ˣ
  • a) Differentiate f(x)=e3xf(x)=e^{3x}, g(x)=e−x2g(x)=e^{-x^{2}} and h(x)=xe−xh(x)=xe^{-x}. For hh, find the point where the tangent line is horizontal.
  • b) The general formula is (ax)′=axln⁡a(a^{x})'=a^{x}\ln a. Differentiate u(x)=2xu(x)=2^{x} and v(x)=3x2v(x)=3^{x^{2}}. What does the formula become for a=ea=e?
  • c) The figure shows the three curves y=2xy=2^{x}, y=exy=e^{x} and y=3xy=3^{x} with their tangent lines at 0. Give the slope of each and say in one sentence what characterizes the number ee.
  • d) Many students write (ex)2=ex2(e^{x})^{2}=e^{x^{2}}. Differentiate both expressions and show, by a computation at x=2x=2, that they have nothing in common.

Type your answers, the page tells you right or wrong 0/10

a)
b)
c)
d)
Show the solution

Answers

  • a) 3e3x3e^{3x}; −2xe−x2-2xe^{-x^{2}}; e−x(1−x)e^{-x}(1-x), horizontal tangent at (1,e−1)(1, e^{-1})
  • b) 2xln⁡22^{x}\ln 2; 2xln⁡3⋅3x22x\ln 3\cdot 3^{x^{2}}; for a=ea=e, (ex)′=ex(e^{x})'=e^{x}
  • c) ln⁡2\ln 2, 11, ln⁡3\ln 3: ee is the base with slope 1 at 0
  • d) 2e2x2e^{2x} against 2xex22xe^{x^{2}}: 109.2109.2 against 218.4218.4 at 22
a)

f′(x)=3e3xf'(x)=3e^{3x}: chain rule, the derivative of the inside 3x3x comes out as a factor. g′(x)=e−x2⋅(−2x)=−2xe−x2g'(x)=e^{-x^{2}}\cdot(-2x)=-2xe^{-x^{2}}: this is the Gaussian, and its derivative is zero at x=0x=0, which is its maximum. h′(x)=1⋅e−x+x⋅(−e−x)=e−x(1−x)h'(x)=1\cdot e^{-x}+x\cdot(-e^{-x})=e^{-x}(1-x): product rule, then factor out the exponential, always. Since e−x>0e^{-x}>0, the derivative is zero only at x=1x=1, where h(1)=e−1≈0.368h(1)=e^{-1}\approx 0.368. It is a maximum: h′>0h'>0 before, h′<0h'<0 after.

b)

u′(x)=2xln⁡2≈0.693⋅2xu'(x)=2^{x}\ln 2\approx 0.693\cdot 2^{x}: the factor ln⁡2\ln 2 is the one most often forgotten, and it costs the whole slope computation. v(x)=3x2v(x)=3^{x^{2}} needs the chain rule ON TOP of the formula: v′(x)=3x2ln⁡3⋅2x=2xln⁡3⋅3x2v'(x)=3^{x^{2}}\ln 3\cdot 2x=2x\ln 3\cdot 3^{x^{2}}. For a=ea=e, ln⁡e=1\ln e=1 and the formula gives back (ex)′=ex(e^{x})'=e^{x}: the general formula contains the special case, it is not one more rule to memorize.

c)

The slopes at 0 are ln⁡2≈0.693\ln 2\approx 0.693, ln⁡e=1\ln e=1 and ln⁡3≈1.099\ln 3\approx 1.099. This is exactly the characterization of ee: it is the ONLY base whose exponential curve crosses the point (0, 1) with slope 1, hence the only exponential equal to its own derivative. The other bases are horizontally stretched or compressed versions of that one, since ax=exln⁡aa^{x}=e^{x\ln a}.

d)

(ex)2=e2x(e^{x})^{2}=e^{2x} by the laws of exponents, so its derivative is 2e2x2e^{2x}. On the other hand, ex2e^{x^{2}} has derivative 2xex22xe^{x^{2}}. At x=2x=2: 2e4≈109.22e^{4}\approx 109.2 against 4e4≈218.44e^{4}\approx 218.4; and the functions themselves are e4≈54.6e^{4}\approx 54.6 against e4≈54.6e^{4}\approx 54.6, equal at x=0x=0 and at x=2x=2 only, the two solutions of 2x=x22x=x^{2}. At x=3x=3 they equal e6≈403e^{6}\approx 403 and e9≈8103e^{9}\approx 8103: a factor of 20. Remember the reading: the exponent of a power of a power gets MULTIPLIED, not composed.

Exercise 3: The logarithm, and the domain that comes before the derivative

The formula (ln⁡u)′=u′u(\ln u)'=\frac{u'}{u} fits in five characters, and yet this is the chapter where the most marks are lost: because people differentiate before looking at where the function exists.

  • a) Differentiate f(x)=ln⁡(3x)f(x)=\ln(3x), g(x)=ln⁡(x2+1)g(x)=\ln(x^{2}+1) and h(x)=ln⁡(cos⁡x)h(x)=\ln(\cos x), giving the domain each time.
  • b) You should find the same derivative for ln⁡(3x)\ln(3x) and for ln⁡x\ln x. Explain why this is not a mistake.
  • c) Differentiate u(x)=log⁡2xu(x)=\log_{2}x and v(x)=xln⁡x−xv(x)=x\ln x-x. What do you notice about v′v'?
  • d) Prove that (ln⁡∣x∣)′=1x(\ln|x|)'=\frac{1}{x} for every x≠0x\neq 0, treating the case x<0x<0 separately.
  • e) Give the domain of w(x)=ln⁡(x2−4)w(x)=\ln(x^{2}-4), then its derivative. Why does the expression obtained only make sense on that domain, even though it can be computed formally elsewhere?
  • f) Differentiate F(x)=ln⁡(sec⁡x+tan⁡x)F(x)=\ln(\sec x+\tan x), G(x)=ln⁡(csc⁡x−cot⁡x)G(x)=\ln(\csc x-\cot x) and L(x)=ln⁡(sin⁡x)L(x)=\ln(\sin x) on (0,π2)\left(0,\frac{\pi}{2}\right), simplifying all the way. What do you notice?

Type your answers, the page tells you right or wrong 0/12

a)
b)
c)
d)
e)
Left interval ,
Right interval ,
f)
Show the solution

Answers

  • a) 1x\frac{1}{x} for x>0x>0; 2xx2+1\frac{2x}{x^{2}+1} on R\mathbb{R}; −tan⁡x-\tan x where cos⁡x>0\cos x>0
  • b) ln⁡(3x)=ln⁡3+ln⁡x\ln(3x)=\ln 3+\ln x: same derivative
  • c) 1xln⁡2\frac{1}{x\ln 2}; v′(x)=ln⁡xv'(x)=\ln x
  • d) (ln⁡(−x))′=1x(\ln(-x))'=\frac{1}{x} for x<0x<0
  • e) (−∞,−2)∪(2,+∞)(-\infty,-2)\cup(2,+\infty); w′(x)=2xx2−4w'(x)=\frac{2x}{x^{2}-4}
  • f) F′(x)=sec⁡xF'(x)=\sec x; G′(x)=csc⁡xG'(x)=\csc x; L′(x)=cot⁡xL'(x)=\cot x
a)

f′(x)=33x=1xf'(x)=\frac{3}{3x}=\frac{1}{x}, for x>0x>0. g′(x)=2xx2+1g'(x)=\frac{2x}{x^{2}+1}, defined on all of R\mathbb{R} since x2+1>0x^{2}+1>0 always: this is the rare logarithm with no restriction. h′(x)=−sin⁡xcos⁡x=−tan⁡xh'(x)=\frac{-\sin x}{\cos x}=-\tan x, on the intervals where cos⁡x>0\cos x>0, for example (−π2,π2)\left(-\frac{\pi}{2},\frac{\pi}{2}\right). The logarithm requires a STRICTLY positive argument, not merely a nonzero one.

b)

Because ln⁡(3x)=ln⁡3+ln⁡x\ln(3x)=\ln 3+\ln x: the two functions differ by a CONSTANT, and two functions that differ by a constant have the same derivative. Geometrically, the graph of ln⁡(3x)\ln(3x) is that of ln⁡x\ln x shifted vertically by ln⁡3≈1.099\ln 3\approx 1.099: a vertical shift changes no slope. This is the first useful example of the theorem “zero derivative, so constant function”, which will come back in the chapter on antiderivatives.

c)

u(x)=log⁡2x=ln⁡xln⁡2u(x)=\log_{2}x=\frac{\ln x}{\ln 2}, so u′(x)=1xln⁡2≈1.4427xu'(x)=\frac{1}{x\ln 2}\approx\frac{1.4427}{x}: changing the base only multiplies by a constant. v′(x)=(1⋅ln⁡x+x⋅1x)−1=ln⁡x+1−1=ln⁡xv'(x)=\left(1\cdot\ln x+x\cdot\frac{1}{x}\right)-1=\ln x+1-1=\ln x. In other words, vv is an antiderivative of ln⁡\ln, a result that will be proved again by integration by parts in 201-SN3-RE, integral calculus: it is useful to know right now that it can be checked in one line by differentiation.

d)

For x>0x>0, ∣x∣=x|x|=x and (ln⁡x)′=1x(\ln x)'=\frac{1}{x}. For x<0x<0, ∣x∣=−x|x|=-x, so ln⁡∣x∣=ln⁡(−x)\ln|x|=\ln(-x) and by the chain rule (ln⁡(−x))′=−1−x=1x(\ln(-x))'=\frac{-1}{-x}=\frac{1}{x}: the minus sign of the inside and that of the denominator cancel. Both pieces give the same formula, hence the result for every x≠0x\neq 0. Check at x=−2x=-2: ln⁡∣−2+0.000001∣−ln⁡∣−2∣0.000001≈−0.5=1−2\frac{\ln|-2+0.000001|-\ln|-2|}{0.000001}\approx-0.5=\frac{1}{-2} ✓. This is why the antiderivative of 1x\frac{1}{x} is written ln⁡∣x∣\ln|x| and not ln⁡x\ln x.

e)

x2−4>0x^{2}-4>0 is equivalent to ∣x∣>2|x|>2, so the domain is (−∞,−2)∪(2,+∞)(-\infty,-2)\cup(2,+\infty). The derivative is w′(x)=2xx2−4w'(x)=\frac{2x}{x^{2}-4}. This fraction CAN be computed at x=0x=0 and gives 0, but w(0)=ln⁡(−4)w(0)=\ln(-4) does not exist: an algebraic expression knows nothing about the domain of the function it came from. The derivative of a function is never defined outside the domain of the function, even when the formula itself accepts the number. This is the mistake that leads to an extremum at x=0x=0 on an interval where the function does not exist.

f)

(ln⁡u)′=u′u(\ln u)'=\frac{u'}{u} each time, then look for the common factor. F′(x)=sec⁡xtan⁡x+sec⁡2xsec⁡x+tan⁡x=sec⁡x(tan⁡x+sec⁡x)sec⁡x+tan⁡x=sec⁡xF'(x)=\frac{\sec x\tan x+\sec^{2}x}{\sec x+\tan x}=\frac{\sec x\left(\tan x+\sec x\right)}{\sec x+\tan x}=\sec x. G′(x)=−csc⁡xcot⁡x+csc⁡2xcsc⁡x−cot⁡x=csc⁡x(csc⁡x−cot⁡x)csc⁡x−cot⁡x=csc⁡xG'(x)=\frac{-\csc x\cot x+\csc^{2}x}{\csc x-\cot x}=\frac{\csc x\left(\csc x-\cot x\right)}{\csc x-\cot x}=\csc x: the cotangent and the cosecant each carry a minus sign, and they offset each other in the numerator. L′(x)=cos⁡xsin⁡x=cot⁡xL'(x)=\frac{\cos x}{\sin x}=\cot x, the twin of the −tan⁡x-\tan x of question a. What to notice: the derivative of the inside is a multiple of the inside, so the fraction SIMPLIFIES, and an answer left as a quotient loses the simplification mark. Read backwards, these three lines give the antiderivatives of sec⁡x\sec x, of csc⁡x\csc x and of cot⁡x\cot x, which will come back in 201-SN3-RE. Domain: on (0,π2)\left(0,\frac{\pi}{2}\right), sec⁡x+tan⁡x>0\sec x+\tan x>0, sin⁡x>0\sin x>0 and csc⁡x−cot⁡x=1−cos⁡xsin⁡x>0\csc x-\cot x=\frac{1-\cos x}{\sin x}>0, so all three logarithms exist.

Exercise 4: Nesting three layers without losing a factor

The chain rule never gets tired: you can apply it as many times as there are nested functions. The only discipline is to name the layers from the outside in BEFORE writing anything.

  • a) Differentiate f(x)=esin⁡(2x)f(x)=e^{\sin(2x)}, naming the three layers.
  • b) Differentiate g(x)=ln⁡(tan⁡x)g(x)=\ln(\tan x) and simplify the result as far as possible.
  • c) Differentiate h(x)=1+e2xh(x)=\sqrt{1+e^{2x}}.
  • d) Differentiate k(x)=sin⁡(ln⁡(x2+1))k(x)=\sin\left(\ln(x^{2}+1)\right).
  • e) Differentiate p(x)=(ln⁡x)3p(x)=(\ln x)^{3} and q(x)=ln⁡(x3)q(x)=\ln(x^{3}). Compare the two results at x=ex=e and say what the notation hides.

Type your answers, the page tells you right or wrong 0/6

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) 2cos⁡(2x)esin⁡(2x)2\cos(2x)e^{\sin(2x)}
  • b) 2sin⁡(2x)\frac{2}{\sin(2x)}
  • c) e2x1+e2x\frac{e^{2x}}{\sqrt{1+e^{2x}}}
  • d) cos⁡(ln⁡(x2+1))⋅2xx2+1\cos\left(\ln(x^{2}+1)\right)\cdot\frac{2x}{x^{2}+1}
  • e) 3(ln⁡x)2x\frac{3(\ln x)^{2}}{x} and 3x\frac{3}{x}: equal at ee, not at e2e^{2}
a)

Three layers: the exponential, then the sine, then 2x2x. f′(x)=esin⁡(2x)⋅cos⁡(2x)⋅2=2cos⁡(2x)esin⁡(2x)f'(x)=e^{\sin(2x)}\cdot\cos(2x)\cdot 2=2\cos(2x)e^{\sin(2x)}. The factor 2 from the innermost layer is the one that gets forgotten: check at x=0x=0, f′(0)=2⋅1⋅e0=2f'(0)=2\cdot 1\cdot e^{0}=2, whereas without the factor you would find 1, a slope half as large as it should be.

b)

g′(x)=(tan⁡x)′tan⁡x=sec⁡2xtan⁡x=1cos⁡2x⋅cos⁡xsin⁡x=1sin⁡xcos⁡xg'(x)=\frac{(\tan x)'}{\tan x}=\frac{\sec^{2}x}{\tan x}=\frac{1}{\cos^{2}x}\cdot\frac{\cos x}{\sin x}=\frac{1}{\sin x\cos x}. You can go further: sin⁡xcos⁡x=12sin⁡(2x)\sin x\cos x=\frac{1}{2}\sin(2x), so g′(x)=2sin⁡(2x)g'(x)=\frac{2}{\sin(2x)}. Check at x=0.7x=0.7: 2sin⁡1.4≈20.9854≈2.0296\frac{2}{\sin 1.4}\approx\frac{2}{0.9854}\approx 2.0296, and the numerical rate of change gives the same value. Simplifying is not cosmetic: this is the form that will make the integral recognizable later.

c)

h(x)=(1+e2x)1/2h(x)=(1+e^{2x})^{1/2}, so h′(x)=12(1+e2x)−1/2⋅2e2x=e2x1+e2xh'(x)=\frac{1}{2}(1+e^{2x})^{-1/2}\cdot 2e^{2x}=\frac{e^{2x}}{\sqrt{1+e^{2x}}}. The two factors of 2 cancel, which is exactly why you must write both: making them vanish in your head is the best way to lose one.

d)

k′(x)=cos⁡(ln⁡(x2+1))⋅2xx2+1k'(x)=\cos\left(\ln(x^{2}+1)\right)\cdot\frac{2x}{x^{2}+1}. Three layers again, and no simplification possible: the answer stays a product, which is normal and does not signal a mistake.

e)

p(x)=(ln⁡x)3p(x)=(\ln x)^{3} is a POWER whose inside is a logarithm: p′(x)=3(ln⁡x)2⋅1x=3(ln⁡x)2xp'(x)=3(\ln x)^{2}\cdot\frac{1}{x}=\frac{3(\ln x)^{2}}{x}. On the other hand, q(x)=ln⁡(x3)=3ln⁡xq(x)=\ln(x^{3})=3\ln x is a logarithm whose inside is a power, and the property of the logarithm reduces it to a constant times ln⁡x\ln x: q′(x)=3xq'(x)=\frac{3}{x}. At x=ex=e: p′(e)=3e≈1.104p'(e)=\frac{3}{e}\approx 1.104 and q′(e)=3e≈1.104q'(e)=\frac{3}{e}\approx 1.104, equal by accident since (ln⁡e)2=1(\ln e)^{2}=1. At x=e2x=e^{2}: p′=3⋅4e2≈1.624p'=\frac{3\cdot 4}{e^{2}}\approx 1.624 against q′=3e2≈0.406q'=\frac{3}{e^{2}}\approx 0.406, a factor of 4. The notation ln⁡3x\ln^{3}x, which you sometimes meet, means pp and not qq: exactly the kind of ambiguity that justifies writing the parentheses.

Exercise 5: Logarithmic differentiation: when no rule applies

There is one family that neither the power rule nor the exponential rule can handle: the one where the base AND the exponent contain the variable. The workaround is always the same, take the logarithm of both sides before differentiating.

  • a) Why do neither (xn)′=nxn−1(x^{n})'=nx^{n-1} nor (ax)′=axln⁡a(a^{x})'=a^{x}\ln a apply to y=xxy=x^{x}? Compute y′y' by logarithmic differentiation.
  • b) Same method for y=xsin⁡xy=x^{\sin x}, for x>0x>0.
  • c) Differentiate y=(x2+1)3x−1(2x+5)4y=\frac{(x^{2}+1)^{3}\sqrt{x-1}}{(2x+5)^{4}} by logarithmic differentiation, and say in one sentence why the direct method would be painful.
  • d) Give an example where logarithmic differentiation is possible but pointless, and explain the criterion for choosing.

Type your answers, the page tells you right or wrong 0/6

a)
b)
c)
d)
Show the solution

Answers

  • a) y′=xx(ln⁡x+1)y'=x^{x}(\ln x+1)
  • b) y′=xsin⁡x(cos⁡xln⁡x+sin⁡xx)y'=x^{\sin x}\left(\cos x\ln x+\frac{\sin x}{x}\right)
  • c) y′=y(6xx2+1+12(x−1)−82x+5)y'=y\left(\frac{6x}{x^{2}+1}+\frac{1}{2(x-1)}-\frac{8}{2x+5}\right)
  • d) x5exx^{5}e^{x}: the product rule is enough
a)

The power rule requires a CONSTANT exponent; the exponential rule requires a CONSTANT base. Here both vary, so this is a new case. Set y=xxy=x^{x} with x>0x>0 and take the logarithm: ln⁡y=xln⁡x\ln y=x\ln x. Differentiate both sides with respect to xx, the left side by the chain rule since yy depends on xx: y′y=ln⁡x+x⋅1x=ln⁡x+1\frac{y'}{y}=\ln x+x\cdot\frac{1}{x}=\ln x+1. So y′=xx(ln⁡x+1)y'=x^{x}(\ln x+1). Check at x=2x=2: y′=4(ln⁡2+1)≈4⋅1.693=6.773y'=4(\ln 2+1)\approx 4\cdot 1.693=6.773, and the numerical rate 2.0000012.000001−220.000001≈6.773\frac{2.000001^{2.000001}-2^{2}}{0.000001}\approx 6.773 ✓. Along the way, y′=0y'=0 gives ln⁡x=−1\ln x=-1, that is x=e−1≈0.368x=e^{-1}\approx 0.368: this is the minimum of xxx^{x}, where the function equals about 0.69220.6922.

b)

ln⁡y=sin⁡x⋅ln⁡x\ln y=\sin x\cdot\ln x, then y′y=cos⁡xln⁡x+sin⁡xx\frac{y'}{y}=\cos x\ln x+\frac{\sin x}{x}, so y′=xsin⁡x(cos⁡xln⁡x+sin⁡xx)y'=x^{\sin x}\left(\cos x\ln x+\frac{\sin x}{x}\right). The structure is always the same: the result is the function itself multiplied by the derivative of its logarithm. This is what is called the LOGARITHMIC derivative, y′y\frac{y'}{y}, and it measures a RELATIVE rate of change, in percent per unit, not in units per unit.

c)

ln⁡y=3ln⁡(x2+1)+12ln⁡(x−1)−4ln⁡(2x+5)\ln y=3\ln(x^{2}+1)+\frac{1}{2}\ln(x-1)-4\ln(2x+5): the logarithm turns the product into a sum, the quotient into a difference and the exponents into coefficients, which replaces a quotient rule nested inside a product rule by three independent derivatives. Differentiate: y′y=6xx2+1+12(x−1)−82x+5\frac{y'}{y}=\frac{6x}{x^{2}+1}+\frac{1}{2(x-1)}-\frac{8}{2x+5}, then multiply by yy. Check at x=2x=2: y=125⋅194=1256561≈0.019052y=\frac{125\cdot 1}{9^{4}}=\frac{125}{6561}\approx 0.019052 and y′y=125+12−89≈2.0111\frac{y'}{y}=\frac{12}{5}+\frac{1}{2}-\frac{8}{9}\approx 2.0111, so y′≈0.038318y'\approx 0.038318. The numerical rate confirms it. Directly, you would have to differentiate a quotient whose numerator is itself a product of two composite functions: three pages, and a sign error almost guaranteed.

d)

For y=x5exy=x^{5}e^{x}, logarithmic differentiation works and gives y′y=5x+1\frac{y'}{y}=\frac{5}{x}+1, so y′=x5ex(5x+1)=x4ex(5+x)y'=x^{5}e^{x}\left(\frac{5}{x}+1\right)=x^{4}e^{x}(5+x); but the product rule gives the same result in one line. The criterion: switch to the logarithm when the function is a PRODUCT OR QUOTIENT OF AT LEAST THREE FACTORS with exponents, or when the variable is in the exponent. Two simple factors do not justify it, and the detour then costs more than the direct rule.

Part B: problems and reasoning (/50)

Exercise 6: The inverse trigonometric functions and their algebraic derivatives

The remarkable fact of this section: the inverses of trigonometric functions have ALGEBRAIC derivatives, with no trigonometry at all. This is why they will show up as antiderivatives of fractions in 201-SN3-RE, integral calculus.

-6-5-4-3-2-1123456-2-112y = π/2y = −π/2y = arctan x
  • a) Let y=arcsin⁡xy=\arcsin x, so sin⁡y=x\sin y=x with y∈[−π2,π2]y\in\left[-\frac{\pi}{2},\frac{\pi}{2}\right]. By differentiating both sides with respect to xx, prove that (arcsin⁡x)′=11−x2(\arcsin x)'=\frac{1}{\sqrt{1-x^{2}}}, justifying the choice of sign of the square root.
  • b) By the same method, prove that (arctan⁡x)′=11+x2(\arctan x)'=\frac{1}{1+x^{2}}.
  • c) Differentiate f(x)=arctan⁡(2x)f(x)=\arctan(2x), g(x)=arcsin⁡(x3)g(x)=\arcsin\left(\frac{x}{3}\right) and h(x)=xarctan⁡x−12ln⁡(1+x2)h(x)=x\arctan x-\frac{1}{2}\ln(1+x^{2}). Comment on the result obtained for hh.
  • d) The figure shows y=arctan⁡xy=\arctan x. Read from it how the derivative behaves as x→±∞x\to\pm\infty, then explain why (arcsin⁡x)′(\arcsin x)' blows up at x=±1x=\pm 1 even though arcsin⁡\arcsin is perfectly well defined there.
  • e) Give the domain of p(x)=arccos⁡(3x)p(x)=\arccos\left(\frac{3}{x}\right), compute p′(x)p'(x) while simplifying the square root, then p′(−5)p'(-5). Same work for q(x)=arcsin⁡(x)q(x)=\arcsin(\sqrt{x}): domain, derivative, then q′(14)q'\left(\frac{1}{4}\right).

Type your answers, the page tells you right or wrong 0/11

a)
b)
c)
d)
e)
Domain of qq ,
Show the solution

Answers

  • a) (arcsin⁡x)′=11−x2(\arcsin x)'=\frac{1}{\sqrt{1-x^{2}}}, positive root because cos⁡y≥0\cos y\ge 0
  • b) (arctan⁡x)′=11+x2(\arctan x)'=\frac{1}{1+x^{2}}
  • c) 21+4x2\frac{2}{1+4x^{2}}; 19−x2\frac{1}{\sqrt{9-x^{2}}}; h′(x)=arctan⁡xh'(x)=\arctan x
  • d) Slope tends to 00 at infinity; vertical tangent line of arcsin⁡\arcsin at ±1\pm 1
  • e) Domain ∣x∣≥3|x|\ge 3; p′(x)=3∣x∣x2−9p'(x)=\frac{3}{|x|\sqrt{x^{2}-9}}, p′(−5)=0.15p'(-5)=0.15; q′(x)=12x(1−x)q'(x)=\frac{1}{2\sqrt{x(1-x)}} on (0,1)(0,1), q′(14)≈1.155q'\left(\frac{1}{4}\right)\approx 1.155
a)

Differentiate sin⁡y=x\sin y=x with respect to xx, treating yy as a function of xx: cos⁡y⋅y′=1\cos y\cdot y'=1, so y′=1cos⁡yy'=\frac{1}{\cos y}. It remains to express cos⁡y\cos y in terms of xx: from sin⁡2y+cos⁡2y=1\sin^{2}y+\cos^{2}y=1 we get cos⁡y=±1−x2\cos y=\pm\sqrt{1-x^{2}}. The sign is decided by the chosen DOMAIN: on [−π2,π2]\left[-\frac{\pi}{2},\frac{\pi}{2}\right], the cosine is positive or zero, so we keep the positive root and (arcsin⁡x)′=11−x2(\arcsin x)'=\frac{1}{\sqrt{1-x^{2}}} for ∣x∣<1|x|<1. The sign never comes from the computation; it comes from the interval on which the function was made invertible.

b)

From tan⁡y=x\tan y=x we get sec⁡2y⋅y′=1\sec^{2}y\cdot y'=1, so y′=cos⁡2yy'=\cos^{2}y. Now sec⁡2y=1+tan⁡2y=1+x2\sec^{2}y=1+\tan^{2}y=1+x^{2}, hence y′=11+x2y'=\frac{1}{1+x^{2}}. This time there is no choice of sign: the result is the reciprocal of a square, positive everywhere, which confirms that arctan⁡\arctan is strictly increasing on all of R\mathbb{R}.

c)

f′(x)=21+4x2f'(x)=\frac{2}{1+4x^{2}} by the chain rule. g′(x)=1/31−x29=19−x2g'(x)=\frac{1/3}{\sqrt{1-\frac{x^{2}}{9}}}=\frac{1}{\sqrt{9-x^{2}}} after bringing the 3 under the root, a much more usable form. For hh: h′(x)=arctan⁡x+x1+x2−12⋅2x1+x2=arctan⁡xh'(x)=\arctan x+\frac{x}{1+x^{2}}-\frac{1}{2}\cdot\frac{2x}{1+x^{2}}=\arctan x. The last two terms cancel exactly: hh is therefore an antiderivative of arctan⁡\arctan, which will be proved again by integration by parts in the next course. Check at x=1x=1: h′(1)=arctan⁡1=π4≈0.7854h'(1)=\arctan 1=\frac{\pi}{4}\approx 0.7854, and the numerical rate of change gives the same value.

d)

In the figure, the curve flattens out towards its two asymptotes y=±π2y=\pm\frac{\pi}{2}: the slope tends to 0, as confirmed by 11+x2→0\frac{1}{1+x^{2}}\to 0. The maximum slope is 1 and it is reached at x=0x=0. For arcsin⁡\arcsin, as x→1−x\to 1^{-} the denominator 1−x2→0+\sqrt{1-x^{2}}\to 0^{+}, so the derivative tends to +∞+\infty: the graph of arcsin⁡\arcsin has a VERTICAL TANGENT LINE at x=1x=1. There is no contradiction; a function can be continuous and defined at a point without being differentiable there. This is the mirror reading of the graph of sin⁡\sin, whose tangent line is horizontal at π2\frac{\pi}{2}: swapping the axes swaps horizontal and vertical.

e)

arccos⁡\arccos requires an argument in [−1,1][-1,1]: ∣3x∣≤1\left|\frac{3}{x}\right|\le 1 is equivalent to ∣x∣≥3|x|\ge 3, so the domain is (−∞,−3]∪[3,+∞)\left(-\infty,-3\right]\cup\left[3,+\infty\right), and the derivative exists for ∣x∣>3|x|>3. Chain rule: p′(x)=−11−9x2⋅(−3x2)=3x21−9x2p'(x)=\frac{-1}{\sqrt{1-\frac{9}{x^{2}}}}\cdot\left(-\frac{3}{x^{2}}\right)=\frac{3}{x^{2}\sqrt{1-\frac{9}{x^{2}}}}. To simplify, 1−9x2=x2−9x2=x2−9∣x∣\sqrt{1-\frac{9}{x^{2}}}=\frac{\sqrt{x^{2}-9}}{\sqrt{x^{2}}}=\frac{\sqrt{x^{2}-9}}{|x|}, and this is where everything is decided: x2=∣x∣\sqrt{x^{2}}=|x|, not xx. Hence p′(x)=3∣x∣x2x2−9=3∣x∣x2−9p'(x)=\frac{3|x|}{x^{2}\sqrt{x^{2}-9}}=\frac{3}{|x|\sqrt{x^{2}-9}}, and p′(−5)=35⋅4=0.15p'(-5)=\frac{3}{5\cdot 4}=0.15. The wrong form 3xx2−9\frac{3}{x\sqrt{x^{2}-9}} would give −0.15-0.15, a negative slope for an INCREASING function: for x<−3x<-3, 3x\frac{3}{x} decreases and arccos⁡\arccos decreases, so pp increases. For qq, you need x≥0x\ge 0 for the root and x≤1\sqrt{x}\le 1 for arcsin⁡\arcsin: the domain is [0,1][0,1]. Then q′(x)=11−x⋅12x=12x(1−x)q'(x)=\frac{1}{\sqrt{1-x}}\cdot\frac{1}{2\sqrt{x}}=\frac{1}{2\sqrt{x(1-x)}}, defined on (0,1)(0,1) only: vertical tangent lines at both endpoints. Finally q′(14)=12316=23≈1.155q'\left(\frac{1}{4}\right)=\frac{1}{2\sqrt{\frac{3}{16}}}=\frac{2}{\sqrt{3}}\approx 1.155.

Exercise 7: The hyperbolic functions and the shape of the catenary

The hyperbolic cosine is not a parabola, even though it looks like one: it is the shape taken by a cable hanging under its own weight. Its derivatives are those of trigonometry, but without the minus sign.

-3-2-1123123456y = cosh xy = 1 + x²/2
  • a) Let cosh⁡x=ex+e−x2\cosh x=\frac{e^{x}+e^{-x}}{2} and sinh⁡x=ex−e−x2\sinh x=\frac{e^{x}-e^{-x}}{2}. Compute their derivatives and compare with the pair (cos⁡,sin⁡)(\cos,\sin).
  • b) Prove the identity cosh⁡2x−sinh⁡2x=1\cosh^{2}x-\sinh^{2}x=1, then differentiate the left side to check that it is consistent with question a.
  • c) Let tanh⁡x=sinh⁡xcosh⁡x\tanh x=\frac{\sinh x}{\cosh x}. Show that (tanh⁡x)′=1cosh⁡2x=1−tanh⁡2x(\tanh x)'=\frac{1}{\cosh^{2}x}=1-\tanh^{2}x, and deduce whether tanh⁡\tanh is increasing or decreasing.
  • d) A cable hanging between two poles takes the shape y=acosh⁡(xa)y=a\cosh\left(\frac{x}{a}\right) with a=20a=20 metres, the origin being at the lowest point of the cable. Give the slope of the cable at x=20x=20 metres, and how high it has risen above its lowest point.
  • e) The figure overlays y=cosh⁡xy=\cosh x and the parabola y=1+x22y=1+\frac{x^{2}}{2}. What do they share at x=0x=0, and where do they part ways?

Type your answers, the page tells you right or wrong 0/8

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) (cosh⁡x)′=sinh⁡x(\cosh x)'=\sinh x, (sinh⁡x)′=cosh⁡x(\sinh x)'=\cosh x
  • b) cosh⁡2x−sinh⁡2x=1\cosh^{2}x-\sinh^{2}x=1
  • c) (tanh⁡x)′=1cosh⁡2x>0(\tanh x)'=\frac{1}{\cosh^{2}x}>0: increasing
  • d) Slope sinh⁡1≈1.175\sinh 1\approx 1.175; 10.8610.86 m
  • e) Same value, slope and concavity at 00; gap of 0.7620.762 at 22
a)

(cosh⁡x)′=ex−e−x2=sinh⁡x(\cosh x)'=\frac{e^{x}-e^{-x}}{2}=\sinh x and (sinh⁡x)′=ex+e−x2=cosh⁡x(\sinh x)'=\frac{e^{x}+e^{-x}}{2}=\cosh x: each one is the derivative of the other, WITHOUT a minus sign. That is the whole difference from the trigonometric pair, where (cos⁡x)′=−sin⁡x(\cos x)'=-\sin x. The consequence is structural: cos⁡\cos and sin⁡\sin satisfy y′′=−yy''=-y and oscillate, while cosh⁡\cosh and sinh⁡\sinh satisfy y′′=+yy''=+y and blow up. A single sign separates a wave from exponential growth.

b)

cosh⁡2x−sinh⁡2x=(ex+e−x)2−(ex−e−x)24=4exe−x4=1\cosh^{2}x-\sinh^{2}x=\frac{(e^{x}+e^{-x})^{2}-(e^{x}-e^{-x})^{2}}{4}=\frac{4e^{x}e^{-x}}{4}=1, using the identity A2−B2=(A−B)(A+B)A^{2}-B^{2}=(A-B)(A+B) or simply expanding. Check by differentiation: (cosh⁡2x−sinh⁡2x)′=2cosh⁡xsinh⁡x−2sinh⁡xcosh⁡x=0\left(\cosh^{2}x-\sinh^{2}x\right)'=2\cosh x\sinh x-2\sinh x\cosh x=0, and a function with zero derivative on R\mathbb{R} is constant; its value at 0 is 1−0=11-0=1. The two routes agree, which validates the derivatives of question a.

c)

By the quotient rule: (tanh⁡x)′=cosh⁡xcosh⁡x−sinh⁡xsinh⁡xcosh⁡2x=cosh⁡2x−sinh⁡2xcosh⁡2x=1cosh⁡2x(\tanh x)'=\frac{\cosh x\cosh x-\sinh x\sinh x}{\cosh^{2}x}=\frac{\cosh^{2}x-\sinh^{2}x}{\cosh^{2}x}=\frac{1}{\cosh^{2}x}. Splitting the fraction term by term instead, you get 1−tanh⁡2x1-\tanh^{2}x: this is the exact copy of what was done for tan⁡\tan in exercise 1. Since cosh⁡x≥1>0\cosh x\geq 1>0, the derivative is strictly positive: tanh⁡\tanh is strictly increasing on R\mathbb{R}, and its maximum slope is 1, at x=0x=0. It is bounded between −1-1 and 1, which makes it the classic activation function of neural networks.

d)

y′=asinh⁡(xa)⋅1a=sinh⁡(xa)y'=a\sinh\left(\frac{x}{a}\right)\cdot\frac{1}{a}=\sinh\left(\frac{x}{a}\right): the parameter aa disappears from the slope, a neat result. At x=20x=20 with a=20a=20: y′=sinh⁡(1)=e−e−12≈1.1752y'=\sinh(1)=\frac{e-e^{-1}}{2}\approx 1.1752, an inclination of about arctan⁡(1.1752)≈49.6\arctan(1.1752)\approx 49.6 degrees. The height is y(20)−y(0)=20cosh⁡(1)−20cosh⁡(0)=20(1.5431−1)≈10.86y(20)-y(0)=20\cosh(1)-20\cosh(0)=20(1.5431-1)\approx 10.86 metres. So the cable has risen a little less than 11 metres at 20 metres from the lowest point.

e)

At x=0x=0 the two curves both equal 1, have zero slope and the SAME concavity: (cosh⁡)′′(0)=cosh⁡0=1(\cosh)''(0)=\cosh 0=1 and (1+x22)′′=1\left(1+\frac{x^{2}}{2}\right)''=1. They therefore agree up to order 2, which explains why the catenary was long believed to be a parabola. They part ways at order 4: cosh⁡x=1+x22+x424+…\cosh x=1+\frac{x^{2}}{2}+\frac{x^{4}}{24}+\dots, so the gap is about x424\frac{x^{4}}{24}, negligible near 0 but visible from x=2x=2 on, where cosh⁡2≈3.762\cosh 2\approx 3.762 against 33 for the parabola.

Exercise 8: Five statements to correct

Each of the following five statements is FALSE. Say why and give the correct statement.

  • 1) “Since (xn)′=nxn−1(x^{n})'=nx^{n-1}, we have (2x)′=x 2x−1(2^{x})'=x\,2^{x-1}.”
  • 2) “(ln⁡(x2))′=1x2\left(\ln(x^{2})\right)'=\frac{1}{x^{2}}, since the derivative of the logarithm is the reciprocal of its argument.”
  • 3) “(sin⁡2x)′=cos⁡2x\left(\sin^{2}x\right)'=\cos^{2}x, by analogy with the derivative of sine.”
  • 4) “(arcsin⁡x)′=11−x2(\arcsin x)'=\frac{1}{\sqrt{1-x^{2}}} is defined for every real xx, since it is a fraction.”
  • 5) “cosh⁡\cosh and cos⁡\cos differ only by a sign, so (cosh⁡x)′=−sinh⁡x(\cosh x)'=-\sinh x.”

Type your answers, the page tells you right or wrong 0/5

1)
2)
3)
4)
Domain of (arcsin⁡x)′(\arcsin x)' ,
5)
Show the solution

Answers

  • 1) (2x)′=2xln⁡2(2^{x})'=2^{x}\ln 2
  • 2) (ln⁡(x2))′=2x\left(\ln(x^{2})\right)'=\frac{2}{x}
  • 3) (sin⁡2x)′=sin⁡(2x)\left(\sin^{2}x\right)'=\sin(2x)
  • 4) Defined on (−1,1)(-1,1) only
  • 5) (cosh⁡x)′=sinh⁡x(\cosh x)'=\sinh x
1)

FALSE. The power rule requires a CONSTANT exponent and a variable base; here it is the other way around, the base 2 is constant and the exponent varies. The correct formula is (ax)′=axln⁡a(a^{x})'=a^{x}\ln a, so (2x)′=2xln⁡2(2^{x})'=2^{x}\ln 2. The numerical check settles it: at x=3x=3, the false formula gives 3⋅22=123\cdot 2^{2}=12, the true one gives 8ln⁡2≈5.5458\ln 2\approx 5.545, and the numerical rate of change is indeed about 5.5455.545.

2)

FALSE. The derivative of the logarithm is not “the reciprocal of the argument”, it is u′u\frac{u'}{u}: the derivative of the inside is missing. Here u=x2u=x^{2} so u′=2xu'=2x and (ln⁡(x2))′=2xx2=2x\left(\ln(x^{2})\right)'=\frac{2x}{x^{2}}=\frac{2}{x}. You can get it without computation by simplifying first: ln⁡(x2)=2ln⁡∣x∣\ln(x^{2})=2\ln|x|, with derivative 2x\frac{2}{x}. At x=2x=2: 22=1\frac{2}{2}=1, whereas the statement claims 0.250.25.

3)

FALSE. sin⁡2x\sin^{2}x means (sin⁡x)2(\sin x)^{2}, a POWER, so it is the chain rule that applies: (sin⁡2x)′=2sin⁡xcos⁡x=sin⁡(2x)\left(\sin^{2}x\right)'=2\sin x\cos x=\sin(2x). The analogy invoked has no basis; you do not differentiate a square by differentiating what is inside it. At x=1x=1: sin⁡2≈0.909\sin 2\approx 0.909, whereas cos⁡21≈0.292\cos^{2}1\approx 0.292.

4)

FALSE. A fraction is defined only if its denominator is nonzero AND if the square root exists: you need 1−x2>01-x^{2}>0, so −1<x<1-1<x<1. The derivative is defined only on the INTERIOR of the domain of arcsin⁡\arcsin, which is [−1,1][-1,1]: at both endpoints, arcsin⁡\arcsin exists but has a vertical tangent line. Confusing the domain of a function with that of its derivative is the most costly mistake of the chapter.

5)

FALSE. The minus sign appears with cos⁡\cos, not with cosh⁡\cosh: (cos⁡x)′=−sin⁡x(\cos x)'=-\sin x but (cosh⁡x)′=+sinh⁡x(\cosh x)'=+\sinh x. You can check it directly on the definition, (ex+e−x2)′=ex−e−x2=sinh⁡x\left(\frac{e^{x}+e^{-x}}{2}\right)'=\frac{e^{x}-e^{-x}}{2}=\sinh x. This sign is not a detail of notation: it is what makes cos⁡\cos oscillate and cosh⁡\cosh grow without bound.

Exercise 9: Problem: radioactive decay and the relative rate

A quantity that decreases in proportion to itself follows an exponential. All of carbon-14 dating fits in that sentence and in one derivative.

  • a) The mass of a carbon-14 sample follows N(t)=N0e−λtN(t)=N_{0}e^{-\lambda t}, where tt is in years. Compute N′(t)N'(t) and show that the ratio N′(t)N(t)\frac{N'(t)}{N(t)} is CONSTANT. What does this constant mean physically?
  • b) The half-life of carbon 14 is 5730 years. Find λ\lambda, then the percentage lost per year.
  • c) A bone fragment contains 22 percent of the carbon 14 of a living bone of the same mass. Date it.
  • d) A coffee at 85 degrees Celsius in a room at 20 degrees cools according to Newton's law T(t)=20+65e−ktT(t)=20+65e^{-kt}, with tt in minutes. It is at 60 degrees after 10 minutes. Compute kk, then the cooling rate at that moment, in degrees per minute.
  • e) Why is the cooling rate of the coffee NOT constant even though the relative rate of question a) is? Compare the two situations.

Type your answers, the page tells you right or wrong 0/8

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) N′N=−λ\frac{N'}{N}=-\lambda
  • b) λ≈1.2097×10−4\lambda\approx 1.2097\times 10^{-4} per year; 0.01210.0121 % per year
  • c) About 12 50012\,500 years
  • d) k≈0.0486k\approx 0.0486 per minute; T′(10)≈−1.94T'(10)\approx-1.94 °C/min
  • e) It is T′T−20\frac{T'}{T-20} that is constant
a)

N′(t)=N0e−λt⋅(−λ)=−λN(t)N'(t)=N_{0}e^{-\lambda t}\cdot(-\lambda)=-\lambda N(t), so N′(t)N(t)=−λ\frac{N'(t)}{N(t)}=-\lambda, constant and independent of tt as well as of N0N_{0}. Physically: every year, the same PROPORTION of nuclei decays, whatever the quantity left. The quantity lost, on the other hand, decreases with time, which is exactly why exponential decay never reaches zero. This is the very definition of the exponential model: the derivative is proportional to the function.

b)

The half-life TT satisfies e−λT=12e^{-\lambda T}=\frac{1}{2}, so λ=ln⁡2T=0.69315730≈1.2097×10−4\lambda=\frac{\ln 2}{T}=\frac{0.6931}{5730}\approx 1.2097\times 10^{-4} per year. The relative loss per year is therefore about 0.01210.0121 percent, roughly one nucleus in 8,267 per year. Notice that the computation does not depend on any initial quantity, which is what makes dating possible without knowing the original mass.

c)

Solve e−λt=0.22e^{-\lambda t}=0.22, that is t=−ln⁡(0.22)λ=1.51411.2097×10−4≈12 516t=\frac{-\ln(0.22)}{\lambda}=\frac{1.5141}{1.2097\times 10^{-4}}\approx 12\,516 years. Round honestly to about 12,500 years: the precision of the 22 percent measurement does not justify four significant figures. An error of one percentage point in the measurement moves the date by more than 350 years, an order of magnitude that a serious solution should point out.

d)

T(10)=60T(10)=60 gives 20+65e−10k=6020+65e^{-10k}=60, that is e−10k=4065=0.61538e^{-10k}=\frac{40}{65}=0.61538, hence k=−ln⁡(0.61538)10≈0.048551k=\frac{-\ln(0.61538)}{10}\approx 0.048551 per minute. The rate is T′(t)=−65k e−kt=−k (T(t)−20)T'(t)=-65k\,e^{-kt}=-k\,(T(t)-20); at t=10t=10, T′(10)=−0.048551×40≈−1.942T'(10)=-0.048551\times 40\approx-1.942 degrees per minute. The negative sign says that the temperature is dropping, and writing T′=−k(T−20)T'=-k(T-20) shows that it is the GAP from the room temperature, not the temperature itself, that drives the cooling.

e)

In both cases the derivative is proportional to something, but not to the same quantity. For carbon, N′=−λNN'=-\lambda N: the reference is zero, so the relative rate N′N\frac{N'}{N} is constant. For the coffee, T′=−k(T−20)T'=-k(T-20): the reference is 20 degrees, not zero, so it is T′T−20\frac{T'}{T-20} that is constant, while T′T\frac{T'}{T} varies. At the start the coffee loses −0.048551×65≈−3.16-0.048551\times 65\approx-3.16 degrees per minute, against −1.94-1.94 after ten minutes: the cooling rate has dropped by nearly 40 percent while the rate relative to the gap has not moved. The lesson goes beyond these two models: before saying “the rate is constant”, say relative to WHAT.

Exercise 10: Problem: the logistic model and its maximum rate of spread

An exponential grows without bound, which no real population does. The logistic model fixes this, and the derivative answers the real question: when is the spread fastest?

246810121416182020040060080010001200P(t)
  • a) A rumour spreads on a campus of 1000 people according to P(t)=10001+99e−0.6tP(t)=\frac{1000}{1+99e^{-0.6t}}, where tt is in days. How many people know it at t=0t=0, and what value does PP approach?
  • b) Compute P′(t)P'(t) by the chain rule applied to the power −1-1, and check that P′>0P'>0 for every tt.
  • c) Show that P′P' can be written P′=0.61000P(1000−P)P'=\frac{0.6}{1000}P(1000-P), then deduce from it, WITHOUT computing P′′P'', the moment when the spread is fastest and the rate of spread reached that day.
  • d) Confirm with P′′P'' that this moment is indeed an inflection point, and read it on the figure.
  • e) After how many days does 90 percent of the campus know the rumour? Compare with the pure exponential model Q(t)=10e0.6tQ(t)=10e^{0.6t}, which has the same starting value and almost the same starting slope.

Type your answers, the page tells you right or wrong 0/9

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) 1010 people; limit 10001000
  • b) P′(t)=59 400 e−0.6t(1+99e−0.6t)2>0P'(t)=\frac{59\,400\,e^{-0.6t}}{\left(1+99e^{-0.6t}\right)^{2}}>0
  • c) Maximum at P=500P=500: 150150 people per day, t≈7.66t\approx 7.66 days
  • d) P′′P'' changes sign at P=500P=500: inflection point
  • e) 11.3211.32 days against 7.507.50 days
a)

P(0)=10001+99=10P(0)=\frac{1000}{1+99}=10 people. As t→+∞t\to+\infty, e−0.6t→0e^{-0.6t}\to 0 so P→1000P\to 1000: this is the plateau, the capacity of the campus, drawn dashed in the figure below. The model is built so that the population never exceeds 1000, unlike an exponential.

b)

Write P(t)=1000(1+99e−0.6t)−1P(t)=1000\left(1+99e^{-0.6t}\right)^{-1}. Chain rule: P′(t)=−1000(1+99e−0.6t)−2⋅99e−0.6t⋅(−0.6)P'(t)=-1000\left(1+99e^{-0.6t}\right)^{-2}\cdot 99e^{-0.6t}\cdot(-0.6), that is P′(t)=59 400 e−0.6t(1+99e−0.6t)2P'(t)=\frac{59\,400\,e^{-0.6t}}{\left(1+99e^{-0.6t}\right)^{2}}. The two minus signs cancel, so P′>0P'>0 for every tt: the rumour never retreats, which is exactly what the model should say.

c)

Setting u=99e−0.6tu=99e^{-0.6t}, we have P=10001+uP=\frac{1000}{1+u} and 1000−P=1000u1+u1000-P=\frac{1000u}{1+u}, so 0.61000P(1000−P)=0.61000⋅10001+u⋅1000u1+u=600u(1+u)2\frac{0.6}{1000}P(1000-P)=\frac{0.6}{1000}\cdot\frac{1000}{1+u}\cdot\frac{1000u}{1+u}=\frac{600u}{(1+u)^{2}}, which is exactly the expression of P′P' obtained in b. This form says much more: P′P' is a product of two factors whose SUM is constant, equal to 1000. Such a product is largest when the two factors are equal, so when P=500P=500, half the campus. The rate of spread is then 0.61000×500×500=150\frac{0.6}{1000}\times 500\times 500=150 people per day. The moment is found by solving 99e−0.6t=199e^{-0.6t}=1, that is t=ln⁡990.6≈7.659t=\frac{\ln 99}{0.6}\approx 7.659 days.

d)

Differentiating P′=0.61000(1000P−P2)P'=\frac{0.6}{1000}\left(1000P-P^{2}\right) gives P′′=0.61000(1000−2P)P′P''=\frac{0.6}{1000}\left(1000-2P\right)P'. Since P′>0P'>0, the sign of P′′P'' is that of 1000−2P1000-2P: positive as long as P<500P<500, negative afterwards, with a zero and a CHANGE OF SIGN at P=500P=500. It is therefore indeed an inflection point, and the curve of the figure switches there from concave up to concave down: this is the point where it stops accelerating, marked by the dot where the two dashed lines meet in the figure below.

e)

P(t)=900P(t)=900 gives 1+99e−0.6t=10009001+99e^{-0.6t}=\frac{1000}{900}, that is 99e−0.6t=1999e^{-0.6t}=\frac{1}{9} and t=ln⁡8910.6≈11.32t=\frac{\ln 891}{0.6}\approx 11.32 days. The exponential model Q(t)=10e0.6tQ(t)=10e^{0.6t} reaches 900 at t=ln⁡900.6≈7.50t=\frac{\ln 90}{0.6}\approx 7.50 days, and it would reach 1000 at 7.687.68 days and then 10,000 at 11.511.5 days, which makes no sense on a campus of 1000 people. Yet the two models are practically indistinguishable at the very beginning, with the same starting value and almost the same starting slope: at t=1t=1, P≈18.0P\approx 18.0 and Q≈18.2Q\approx 18.2. That is the lesson of the problem: two models can fit the same initial data and diverge completely, and it is the derivative, not the value, that reveals which of the two assumptions was really made.

246810121416182020040060080010001200P(t)plateau P = 1000inflection point

Part C: the classics (/50)

Exercise 11: The tangent lines to the exponential and the logarithm that pass through the origin

The figure shows y=exy=e^{x} and y=ln⁡xy=\ln x, with two dashed tangent lines that pass through the origin.

-2-11234-2-112345y = eˣy = ln x
  • a) Write the equation of the tangent line to the curve y=exy=e^{x} at the point with xx-coordinate aa.
  • b) For which value of aa does this tangent line pass through the origin? Give its equation.
  • c) Same question for the curve y=ln⁡xy=\ln x at the point with xx-coordinate bb.
  • d) Prove that ln⁡x≤xe\ln x\le\frac{x}{e} for every x>0x>0, then deduce that πe<eπ\pi^{e}<e^{\pi}.
  • e) At which point is the tangent line to y=e2xy=e^{2x} parallel to the line y=4x+1y=4x+1? Give its yy-intercept.

Type your answers, the page tells you right or wrong 0/9

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) y=eax+ea(1−a)y=e^{a}x+e^{a}(1-a)
  • b) a=1a=1: y=exy=ex
  • c) b=eb=e: y=xey=\frac{x}{e}
  • d) ln⁡x≤xe\ln x\le\frac{x}{e}; πe≈22.46<eπ≈23.14\pi^{e}\approx 22.46<e^{\pi}\approx 23.14
  • e) x=ln⁡22≈0.347x=\frac{\ln 2}{2}\approx 0.347; y=4x+2−2ln⁡2y=4x+2-2\ln 2
a)

(ex)′=ex(e^{x})'=e^{x}, so the slope at aa is eae^{a} and the tangent line is y=ea(x−a)+ea=eax+ea(1−a)y=e^{a}(x-a)+e^{a}=e^{a}x+e^{a}(1-a).

b)

The tangent line passes through the origin when its yy-intercept is zero: ea(1−a)=0e^{a}(1-a)=0. An exponential is never zero, so a=1a=1. The tangent line is y=exy=ex, with slope e≈2.718e\approx 2.718, and it touches the curve at (1,e)(1, e). This is another way of meeting ee: the only line through the origin that just touches the exponential curve has slope ee.

c)

The slope at bb is 1b\frac{1}{b}: y=1b(x−b)+ln⁡b=xb−1+ln⁡by=\frac{1}{b}(x-b)+\ln b=\frac{x}{b}-1+\ln b. The yy-intercept is zero when ln⁡b=1\ln b=1, that is b=eb=e. The tangent line is y=xey=\frac{x}{e}, with slope 1e≈0.368\frac{1}{e}\approx 0.368, and it touches the curve at (e,1)(e, 1). The two points of contact (1,e)(1, e) and (e,1)(e, 1) are symmetric about the line y=xy=x, like the two curves, which are inverses of each other, and the two slopes are reciprocals.

d)

Let g(x)=xe−ln⁡xg(x)=\frac{x}{e}-\ln x on (0,+∞)(0, +\infty). g′(x)=1e−1xg'(x)=\frac{1}{e}-\frac{1}{x} is negative for x<ex<e and positive for x>ex>e: gg decreases then increases, and its minimum is g(e)=1−1=0g(e)=1-1=0. So g(x)≥0g(x)\ge 0, that is ln⁡x≤xe\ln x\le\frac{x}{e}, with equality only at x=ex=e: the graph of the logarithm lies below its tangent line through the origin, the one from c). At x=πx=\pi: ln⁡π<πe\ln\pi<\frac{\pi}{e}, that is eln⁡π<πe\ln\pi<\pi, and since the exponential is increasing, πe<eπ\pi^{e}<e^{\pi}. Numerically, πe≈22.46\pi^{e}\approx 22.46 and eπ≈23.14e^{\pi}\approx 23.14.

e)

The slope of y=e2xy=e^{2x} is 2e2x2e^{2x}. Parallel to y=4x+1y=4x+1: 2e2x=42e^{2x}=4, that is e2x=2e^{2x}=2 and x=ln⁡22≈0.347x=\frac{\ln 2}{2}\approx 0.347. The point of contact is (ln⁡22,2)\left(\frac{\ln 2}{2}, 2\right) and the tangent line is y=4(x−ln⁡22)+2=4x+2−2ln⁡2y=4\left(x-\frac{\ln 2}{2}\right)+2=4x+2-2\ln 2, with yy-intercept 2−2ln⁡2≈0.6142-2\ln 2\approx 0.614. Forgetting the factor 22 from the chain rule would give e2x=4e^{2x}=4 and a wrong point.

Exercise 12: Zero derivative, constant function: two identities of the inverse functions

Assume that a differentiable function whose derivative is zero on an INTERVAL is constant there. Recall (arcsin⁡x)′=11−x2(\arcsin x)'=\frac{1}{\sqrt{1-x^{2}}} and (arctan⁡x)′=11+x2(\arctan x)'=\frac{1}{1+x^{2}}.

  • a) By differentiating cos⁡y=x\cos y=x with y∈[0,π]y\in[0,\pi], establish (arccos⁡x)′(\arccos x)' on (−1,1)(-1,1).
  • b) Deduce that arcsin⁡x+arccos⁡x\arcsin x+\arccos x is constant on [−1,1][-1,1] and give its value.
  • c) Let L(x)=arcsin⁡x−arctan⁡(x1−x2)L(x)=\arcsin x-\arctan\left(\frac{x}{\sqrt{1-x^{2}}}\right) on (−1,1)(-1, 1). Compute L′(x)L'(x), then L(0)L(0). What do you conclude?
  • d) Let H(x)=arctan⁡(1+x1−x)−arctan⁡xH(x)=\arctan\left(\frac{1+x}{1-x}\right)-\arctan x for x≠1x\neq 1. Show that H′(x)=0H'(x)=0 and give the value of HH on each of the two intervals of its domain.
  • e) Check the identity of c) at x=12x=\frac{1}{2}, then explain why it does not extend to x=1x=1.

Type your answers, the page tells you right or wrong 0/8

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) (arccos⁡x)′=−11−x2(\arccos x)'=-\frac{1}{\sqrt{1-x^{2}}}
  • b) arcsin⁡x+arccos⁡x=π2\arcsin x+\arccos x=\frac{\pi}{2}
  • c) L′=0L'=0 and L(0)=0L(0)=0: arcsin⁡x=arctan⁡x1−x2\arcsin x=\arctan\frac{x}{\sqrt{1-x^{2}}}
  • d) H=π4H=\frac{\pi}{4} on (−∞,1)(-\infty,1), −3π4-\frac{3\pi}{4} on (1,+∞)(1,+\infty)
  • e) π6\frac{\pi}{6} on both sides; meaningless at x=1x=1
a)

cos⁡y=x\cos y=x gives −sin⁡y⋅y′=1-\sin y\cdot y'=1, so y′=−1sin⁡yy'=-\frac{1}{\sin y}. On [0,π][0,\pi], sin⁡y≥0\sin y\ge 0, so sin⁡y=1−cos⁡2y=1−x2\sin y=\sqrt{1-\cos^{2}y}=\sqrt{1-x^{2}} and (arccos⁡x)′=−11−x2(\arccos x)'=-\frac{1}{\sqrt{1-x^{2}}} on (−1,1)(-1,1). The sign of the square root is chosen by the interval [0,π][0,\pi], just as that of arcsin⁡\arcsin was chosen by [−π2,π2]\left[-\frac{\pi}{2},\frac{\pi}{2}\right].

b)

(arcsin⁡x+arccos⁡x)′=11−x2−11−x2=0(\arcsin x+\arccos x)'=\frac{1}{\sqrt{1-x^{2}}}-\frac{1}{\sqrt{1-x^{2}}}=0 on the interval (−1,1)(-1,1): the sum is constant there, and by continuity at the endpoints as well. At x=0x=0: arcsin⁡0+arccos⁡0=0+π2\arcsin 0+\arccos 0=0+\frac{\pi}{2}. So arcsin⁡x+arccos⁡x=π2\arcsin x+\arccos x=\frac{\pi}{2} for every x∈[−1,1]x\in[-1,1]: the two angles are complementary, like the two acute angles of a right triangle.

c)

Let u=x1−x2=x(1−x2)−1/2u=\frac{x}{\sqrt{1-x^{2}}}=x(1-x^{2})^{-1/2}: u′=(1−x2)−1/2+x2(1−x2)−3/2=(1−x2)−3/2u'=(1-x^{2})^{-1/2}+x^{2}(1-x^{2})^{-3/2}=(1-x^{2})^{-3/2} and 1+u2=11−x21+u^{2}=\frac{1}{1-x^{2}}. So (arctan⁡u)′=u′1+u2=11−x2\left(\arctan u\right)'=\frac{u'}{1+u^{2}}=\frac{1}{\sqrt{1-x^{2}}}, which is also the derivative of arcsin⁡\arcsin: L′(x)=0L'(x)=0. On the interval (−1,1)(-1, 1), LL is constant, equal to L(0)=0L(0)=0. So arcsin⁡x=arctan⁡x1−x2\arcsin x=\arctan\frac{x}{\sqrt{1-x^{2}}}: this is the reading of a right triangle with hypotenuse 11 and opposite side xx, whose adjacent side is 1−x2\sqrt{1-x^{2}}.

d)

Let u=1+x1−xu=\frac{1+x}{1-x}: u′=(1−x)+(1+x)(1−x)2=2(1−x)2u'=\frac{(1-x)+(1+x)}{(1-x)^{2}}=\frac{2}{(1-x)^{2}} and 1+u2=(1−x)2+(1+x)2(1−x)2=2+2x2(1−x)21+u^{2}=\frac{(1-x)^{2}+(1+x)^{2}}{(1-x)^{2}}=\frac{2+2x^{2}}{(1-x)^{2}}. So (arctan⁡u)′=u′1+u2=22+2x2=11+x2\left(\arctan u\right)'=\frac{u'}{1+u^{2}}=\frac{2}{2+2x^{2}}=\frac{1}{1+x^{2}}, and H′(x)=0H'(x)=0. On (−∞,1)(-\infty,1): H(0)=arctan⁡1−0=π4H(0)=\arctan 1-0=\frac{\pi}{4}. On (1,+∞)(1,+\infty): at x=3x=\sqrt{3}, 1+31−3=−(2+3)\frac{1+\sqrt{3}}{1-\sqrt{3}}=-(2+\sqrt{3}), whose arctangent is −5π12-\frac{5\pi}{12}, and H(3)=−5π12−π3=−3π4H(\sqrt{3})=-\frac{5\pi}{12}-\frac{\pi}{3}=-\frac{3\pi}{4}. In other words, arctan⁡1+x1−x=arctan⁡x+π4\arctan\frac{1+x}{1-x}=\arctan x+\frac{\pi}{4} for x<1x<1, and this equality loses π\pi when crossing x=1x=1.

e)

arcsin⁡12=π6\arcsin\frac{1}{2}=\frac{\pi}{6}, and 1/23/2=13\frac{1/2}{\sqrt{3}/2}=\frac{1}{\sqrt{3}}, whose arctangent is also π6\frac{\pi}{6} ✓. At x=1x=1, arcsin⁡1=π2\arcsin 1=\frac{\pi}{2} exists, but x1−x2\frac{x}{\sqrt{1-x^{2}}} is no longer defined: the right side makes no sense, like the tangent of a right angle. The identity holds on the open interval where the zero derivative was established, and nowhere else.

Exercise 13: The horizontal tangent lines of e to the x times sin x

Let f(x)=exsin⁡xf(x)=e^{x}\sin x, studied on [0,2π][0,2\pi].

  • a) Compute f′(x)f'(x) and factor out the exponential.
  • b) Solve f′(x)=0f'(x)=0 on [0,2π][0,2\pi].
  • c) Give the nature and the value of each extremum, to one decimal place.
  • d) Show that f(x+π)=−eπf(x)f(x+\pi)=-e^{\pi}f(x). What does this tell you about the ratio of two consecutive extrema?
  • e) Compute f′′(x)f''(x) and give the xx-coordinates of the inflection points on [0,2π][0,2\pi].

Type your answers, the page tells you right or wrong 0/8

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) f′(x)=ex(sin⁡x+cos⁡x)f'(x)=e^{x}(\sin x+\cos x)
  • b) 3π4\frac{3\pi}{4} and 7π4\frac{7\pi}{4}
  • c) Maximum ≈7.5\approx 7.5, minimum ≈−172.6\approx-172.6
  • d) Ratio −eπ≈−23.14-e^{\pi}\approx-23.14
  • e) f′′(x)=2excos⁡xf''(x)=2e^{x}\cos x; inflection points at π2\frac{\pi}{2} and 3π2\frac{3\pi}{2}
a)

Product rule: f′(x)=exsin⁡x+excos⁡x=ex(sin⁡x+cos⁡x)f'(x)=e^{x}\sin x+e^{x}\cos x=e^{x}(\sin x+\cos x). Factoring out the exponential is the move that makes the equation f′(x)=0f'(x)=0 readable.

b)

ex>0e^{x}>0, so f′(x)=0f'(x)=0 is equivalent to sin⁡x+cos⁡x=0\sin x+\cos x=0, that is tan⁡x=−1\tan x=-1 where cos⁡x≠0\cos x\neq 0. On [0,2π][0,2\pi]: x=3π4x=\frac{3\pi}{4} and x=7π4x=\frac{7\pi}{4}. No solution is lost by dividing by cos⁡x\cos x: if cos⁡x=0\cos x=0, then sin⁡x=±1\sin x=\pm 1 and the sum is not zero.

c)

sin⁡x+cos⁡x=2sin⁡(x+π4)\sin x+\cos x=\sqrt{2}\sin\left(x+\frac{\pi}{4}\right) is positive on [0,3π4)\left[0,\frac{3\pi}{4}\right), negative on (3π4,7π4)\left(\frac{3\pi}{4},\frac{7\pi}{4}\right), positive afterwards. Maximum at 3π4\frac{3\pi}{4}: f=e3π/4⋅22≈10.55×0.7071≈7.5f=e^{3\pi/4}\cdot\frac{\sqrt{2}}{2}\approx 10.55\times 0.7071\approx 7.5. Minimum at 7π4\frac{7\pi}{4}: f=−e7π/4⋅22≈−244.2×0.7071≈−172.6f=-e^{7\pi/4}\cdot\frac{\sqrt{2}}{2}\approx-244.2\times 0.7071\approx-172.6.

d)

f(x+π)=ex+πsin⁡(x+π)=eπex⋅(−sin⁡x)=−eπf(x)f(x+\pi)=e^{x+\pi}\sin(x+\pi)=e^{\pi}e^{x}\cdot(-\sin x)=-e^{\pi}f(x). Now f′(x+π)=−eπf′(x)f'(x+\pi)=-e^{\pi}f'(x) as well: the critical points repeat every π\pi, and each extremum is the opposite of the previous one multiplied by eπ≈23.14e^{\pi}\approx 23.14. Check: −172.67.46≈−23.1\frac{-172.6}{7.46}\approx-23.1. The curve oscillates like a sine, but with an amplitude that is multiplied by 2323 every half-period.

e)

f′′(x)=ex(sin⁡x+cos⁡x)+ex(cos⁡x−sin⁡x)=2excos⁡xf''(x)=e^{x}(\sin x+\cos x)+e^{x}(\cos x-\sin x)=2e^{x}\cos x. It is zero and changes sign at x=π2x=\frac{\pi}{2} and x=3π2x=\frac{3\pi}{2}: two inflection points. They are NOT halfway between the extrema, unlike what happens for sin⁡x\sin x: each one comes exactly π4\frac{\pi}{4} before the next extremum (π2\frac{\pi}{2} before 3π4\frac{3\pi}{4}, 3π2\frac{3\pi}{2} before 7π4\frac{7\pi}{4}). Compared with sin⁡x\sin x, the extrema move right by π4\frac{\pi}{4} and the inflection points by π2\frac{\pi}{2}: the growth of exe^{x} delays the moment when the slope changes sign.

Exercise 14: Problem: the tide and the rule of twelfths

In a harbour, the water depth, in metres, is modelled by h(t)=3+2.5cos⁡(πt6.2)h(t)=3+2.5\cos\left(\frac{\pi t}{6.2}\right), where tt is the time in hours since high tide. The figure shows a complete cycle, with the 22 m level dashed.

123456789101112131234562 m of waterh(t)
  • a) Give the depth at high tide, at low tide, and the tidal range, the difference between the two.
  • b) Compute h′(t)h'(t), then the rate of change of the water level one hour after high tide.
  • c) At what moment does the water fall fastest, and at what rate?
  • d) Sailors split the fall into six equal intervals and remember that it loses 112\frac{1}{12}, 212\frac{2}{12}, 312\frac{3}{12}, 312\frac{3}{12}, 212\frac{2}{12} then 112\frac{1}{12} of the tidal range. Compare with the model on the first and the third interval.
  • e) A boat needs 22 m of water. Until what time after high tide can it leave, and how fast is the level falling at that moment?

Type your answers, the page tells you right or wrong 0/10

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) 5.55.5 m, 0.50.5 m, tidal range 55 m
  • b) h′(t)=−2.5π6.2sin⁡(πt6.2)h'(t)=-\frac{2.5\pi}{6.2}\sin\left(\frac{\pi t}{6.2}\right); h′(1)≈−0.61h'(1)\approx-0.61 m/h
  • c) t=3.1t=3.1 h: 1.271.27 m/h
  • d) 0.330.33 m against 0.420.42; exactly 1.251.25 m
  • e) t≈3.91t\approx 3.91 h; falling at 1.161.16 m/h
a)

High tide: h(0)=3+2.5=5.5h(0)=3+2.5=5.5 m. Low tide when the cosine equals −1-1, at t=6.2t=6.2 h: h=0.5h=0.5 m. Tidal range: 55 m.

b)

By the chain rule, h′(t)=−2.5⋅π6.2sin⁡(πt6.2)≈−1.267sin⁡(πt6.2)h'(t)=-2.5\cdot\frac{\pi}{6.2}\sin\left(\frac{\pi t}{6.2}\right)\approx-1.267\sin\left(\frac{\pi t}{6.2}\right). At t=1t=1: h′(1)≈−1.267×0.485≈−0.61h'(1)\approx-1.267\times 0.485\approx-0.61 m/h. The factor π6.2\frac{\pi}{6.2}, which comes from the chain rule, is the one that gets forgotten: without it, you would announce a rate of 1.211.21 m/h, about twice the true value.

c)

∣h′(t)∣|h'(t)| is largest when sin⁡(πt6.2)=1\sin\left(\frac{\pi t}{6.2}\right)=1, that is t=3.1t=3.1 h, at mid-tide. The level is then falling at 2.5π6.2≈1.27\frac{2.5\pi}{6.2}\approx 1.27 m/h. At the same moment, h′′(t)h''(t), proportional to −cos⁡-\cos, is zero: this is an inflection point of the curve.

d)

Each interval lasts 6.26\frac{6.2}{6} h, and the argument of the cosine advances by π6\frac{\pi}{6}. First interval: 2.5(1−cos⁡π6)≈0.332.5\left(1-\cos\frac{\pi}{6}\right)\approx 0.33 m, against 512≈0.42\frac{5}{12}\approx 0.42 m for the rule. Third: 2.5(cos⁡π3−cos⁡π2)=1.252.5\left(\cos\frac{\pi}{3}-\cos\frac{\pi}{2}\right)=1.25 m, exactly 312×5\frac{3}{12}\times 5. The rule is a rough approximation at the ends, where the level moves slowly, and accurate in the middle, where the derivative is largest. It translates into whole fractions what the derivative says: slow, fast, slow.

e)

h(t)=2h(t)=2 gives cos⁡(πt6.2)=−0.4\cos\left(\frac{\pi t}{6.2}\right)=-0.4, that is πt6.2=arccos⁡(−0.4)≈1.982\frac{\pi t}{6.2}=\arccos(-0.4)\approx 1.982 and t≈3.91t\approx 3.91 h, about 33 h 5555 min after high tide. At that moment, h′≈−1.267×sin⁡(1.982)≈−1.267×0.917≈−1.16h'\approx-1.267\times\sin(1.982)\approx-1.267\times 0.917\approx-1.16 m/h: the level is still falling at almost full speed, and a few minutes of delay cost more than ten centimetres of water.

Exercise 15: Problem: the best viewing angle of a screen

A vertical screen has its bottom edge 11 m above the eyes of a viewer and its top edge 44 m above. The viewer stands at a horizontal distance x>0x>0 from the wall, in metres. Let θ(x)\theta(x) be the angle under which the viewer sees the screen, in radians. The figure below shows the situation; it is not to scale.

θx1 m4 meyescreen
  • a) Express θ(x)\theta(x) using two arctangents.
  • b) Compute θ′(x)\theta'(x) and put it over a common denominator.
  • c) At what distance is the viewing angle largest? Justify with the sign of θ′\theta' and with the limits at 0+0^{+} and at +∞+\infty.
  • d) Give this maximum angle in degrees, then check that its tangent is exactly 34\frac{3}{4}.
  • e) Show that, for a screen between heights aa and bb above the eyes, the best distance is ab\sqrt{ab}. Apply this to a painting whose edges are 0.50.5 m and 22 m above the eyes.

Type your answers, the page tells you right or wrong 0/6

a)
b)
c)
d)
e)
Show the solution

Answers

  • a) θ(x)=arctan⁡4x−arctan⁡1x\theta(x)=\arctan\frac{4}{x}-\arctan\frac{1}{x}
  • b) θ′(x)=12−3x2(x2+16)(x2+1)\theta'(x)=\frac{12-3x^{2}}{(x^{2}+16)(x^{2}+1)}
  • c) x=2x=2 m
  • d) 36.87°36.87°, tan⁡θ=34\tan\theta=\frac{3}{4}
  • e) x=abx=\sqrt{ab}; 11 m for the painting
a)

The line of sight to the top of the screen makes with the horizontal an angle whose tangent is 4x\frac{4}{x}, the line of sight to the bottom an angle whose tangent is 1x\frac{1}{x}. The screen is seen under their difference: θ(x)=arctan⁡4x−arctan⁡1x\theta(x)=\arctan\frac{4}{x}-\arctan\frac{1}{x}.

b)

(arctan⁡kx)′=−kx21+k2x2=−kx2+k2\left(\arctan\frac{k}{x}\right)'=\frac{-\frac{k}{x^{2}}}{1+\frac{k^{2}}{x^{2}}}=\frac{-k}{x^{2}+k^{2}}. So θ′(x)=−4x2+16+1x2+1=−4(x2+1)+(x2+16)(x2+16)(x2+1)=12−3x2(x2+16)(x2+1)\theta'(x)=\frac{-4}{x^{2}+16}+\frac{1}{x^{2}+1}=\frac{-4(x^{2}+1)+(x^{2}+16)}{(x^{2}+16)(x^{2}+1)}=\frac{12-3x^{2}}{(x^{2}+16)(x^{2}+1)}.

c)

The denominator is positive: θ′\theta' has the sign of 12−3x212-3x^{2}, positive for x<2x<2, negative for x>2x>2. The angle increases then decreases: maximum at x=2x=2 m. At the ends, θ→π2−π2=0\theta\to\frac{\pi}{2}-\frac{\pi}{2}=0 as x→0+x\to 0^{+}, since standing against the wall you see the screen edge-on, and θ→0−0=0\theta\to 0-0=0 as x→+∞x\to+\infty. The maximum is therefore absolute.

d)

θ(2)=arctan⁡2−arctan⁡12≈63.43°−26.57°=36.87°\theta(2)=\arctan 2-\arctan\frac{1}{2}\approx 63.43°-26.57°=36.87°. By the formula tan⁡(A−B)=tan⁡A−tan⁡B1+tan⁡Atan⁡B\tan(A-B)=\frac{\tan A-\tan B}{1+\tan A\tan B}: 2−0.51+2×0.5=1.52=34\frac{2-0.5}{1+2\times 0.5}=\frac{1.5}{2}=\frac{3}{4}. The optimal angle is the one of the 33, 44, 55 triangle.

e)

The same computation gives θ′(x)=−bx2+b2+ax2+a2\theta'(x)=\frac{-b}{x^{2}+b^{2}}+\frac{a}{x^{2}+a^{2}}, which is zero when b(x2+a2)=a(x2+b2)b(x^{2}+a^{2})=a(x^{2}+b^{2}), that is (b−a)x2=ab(b−a)(b-a)x^{2}=ab(b-a) and x2=abx^{2}=ab: the best distance is the GEOMETRIC MEAN of the two heights. This gives back 1×4=2\sqrt{1\times 4}=2. For the painting: 0.5×2=1\sqrt{0.5\times 2}=1 m. It is not the arithmetic mean 1.251.25: stepping back to the average of the heights would be too far.

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.

What is it about?
I read every request myself, nobody else sees it. No need to write your name or your school.
When your request is dealt with, you hear about it by email and in your space.

By sending, you agree that I read your request to answer it. It is kept with your account, erased with it, and never published: privacy policy.

Previous chapter The derivative and its rules, 201-SN2-RE Next chapter Implicit differentiation, 201-SN2-RE

© Ahmed Squalli Houssaini. Exercise set published at www.letuteurscientifique.ca/en/exercices/cegep-calc1-transcendental-derivatives. Free for personal and classroom use; republishing it elsewhere requires written permission (legal notice).

See also

Looking for a differential calculus tutor in Montreal?

Get in touch for a first session. We work on the derivatives of trigonometric, exponential and logarithmic functions at the level of real CEGEP tests, from the trigonometric table all the way to logarithmic differentiation and exponential models.

Site by Studio Squalli