MATH 203 Calculus I • Concordia University, Montreal

Corrected exercises: the chain rule (MATH 203)

This is the corrected exercise set for the chain rule in MATH 203, Differential and Integral Calculus I, at Concordia University, section 3.6 of Thomas' Calculus. The chain rule is the rule that every later chapter of the course leans on: implicit differentiation, the derivatives of logarithms and inverse functions, related rates and optimization are all chain rules in disguise. Each solution names the inside function before differentiating, the sentence that earns the method mark.

The thread running through the whole set: the rule itself is short, and the marks are lost in the ALGEBRA around it. Before differentiating, rewrite roots and reciprocals as powers and compute the new exponent with care; after differentiating, factor each bracket at its LOWEST power, the most negative one, so that the derivative can be set to zero, signed or evaluated. The department runs algebra tutorials for this course for exactly this reason.

The traps named in the solutions: the exponent −3−1-3 - 1 written −2-2, the reciprocal differentiated as 1v′\frac{1}{v'}, the innermost factor forgotten because it is only a number, a table read in the row of xx instead of the row of g(x)g(x), the power rule applied to 2−x2^{-x}, a factorisation taken at the higher power, the product rule skipped in a second derivative, the factor −1-1 of the inside function −x-x, a chain rule applied at a corner, and a derivative evaluated at a number of hours instead of a wind speed.

10 corrected exercises • 100 points • 150 minutes

Revision sheet for this chapter → Every MATH 203 chapter →

Self-checking set Type your answers under each 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.

Course recap

  • • Chain Rule (Outside-Inside Rule): (f∘g)′(x)=f′(g(x)) g′(x)(f \circ g)'(x) = f'(g(x))\, g'(x), or dydx=dydu⋅dudx\frac{dy}{dx} = \frac{dy}{du} \cdot \frac{du}{dx} with u=g(x)u = g(x). Hypotheses: gg differentiable at xx, ff differentiable at g(x)g(x).
  • • Power Chain Rule: ddxun=nun−1u′\frac{d}{dx}u^n = nu^{n-1}u' for every real nn; rewrite first: cuk=cu−k\frac{c}{u^k} = cu^{-k}, umn=um/n\sqrt[n]{u^m} = u^{m/n}.
  • • Repeated use: (f(g(h(x))))′=f′(g(h(x)))⋅g′(h(x))⋅h′(x)\left(f(g(h(x)))\right)' = f'(g(h(x))) \cdot g'(h(x)) \cdot h'(x), one factor per layer.
  • • ax=exln⁡aa^x = e^{x\ln a} with ln⁡a\ln a a constant, so ddxau=auln⁡a⋅u′\frac{d}{dx}a^u = a^u\ln a \cdot u'; laws of exponents first: 1ax=a−x\frac{1}{a^x} = a^{-x}, ax=ax/2\sqrt{a^x} = a^{x/2}.
  • • Factor the lowest power: umu^{m} with mm the smallest exponent present; a term in unu^{n} keeps un−mu^{n - m}.
  • • Tangent line at aa: y=F(a)+F′(a)(x−a)y = F(a) + F'(a)(x - a); a horizontal tangent is a zero of F′F', from the outside factor OR the inside factor.

Part A: the basics (/50)

Exercise 1: Powers of a function: rewrite the exponent, then name u

The Chain Rule (Thomas 3.6): if f(u)f(u) is differentiable at u=g(x)u = g(x) and gg is differentiable at xx, then (f∘g)′(x)=f′(g(x)) g′(x)(f \circ g)'(x) = f'(g(x))\, g'(x). Thomas reads it as the Outside-Inside Rule: differentiate the outside function, evaluate it at the inside function left alone, then multiply by the derivative of the inside. Its most used case is the Power Chain Rule, ddxun=nun−1dudx\frac{d}{dx}u^n = n u^{n-1}\frac{du}{dx}, valid for every real exponent nn.

In MATH 203 the marks on this exercise are not lost in the rule but in the algebra around it. Before differentiating, REWRITE: a root is a fractional exponent, a constant over a power is a negative exponent. Then compute the new exponent n−1n - 1 with care (−3−1=−4-3 - 1 = -4, 23−1=−13\frac{2}{3} - 1 = -\frac{1}{3}). Begin every part with the sentence that names the inside function.

  • a) Differentiate y=6(2x2+1)3y = \frac{6}{(2x^2 + 1)^3} and give y′(1)y'(1).
  • b) Differentiate y=(x2−2x)23y = \sqrt[3]{(x^2 - 2x)^2} and give y′(4)y'(4). Where is y′y' undefined?
  • c) Differentiate y=(x−2x)3y = \left(\sqrt{x} - \frac{2}{x}\right)^3, rewriting the INSIDE as powers of xx first, and give y′(4)y'(4).
  • d) Differentiate y=4sin⁡xy = \frac{4}{\sqrt{\sin x}} and give the exact value of y′(π6)y'\left(\frac{\pi}{6}\right).
  • e) A student writes ddx1(3x−1)2=12(3x−1)⋅3\frac{d}{dx}\frac{1}{(3x - 1)^2} = \frac{1}{2(3x - 1) \cdot 3}. Explain the error, give the correct derivative and its value at x=1x = 1.

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

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

Answers

  • a) y′=−72x(2x2+1)4y' = -\frac{72x}{(2x^2 + 1)^4}, y′(1)=−89y'(1) = -\frac{8}{9}
  • b) y′=4(x−1)3x2−2x3y' = \frac{4(x - 1)}{3\sqrt[3]{x^2 - 2x}}, y′(4)=2y'(4) = 2; undefined at x=0x = 0 and x=2x = 2
  • c) y′=3(x−2x)2(12x+2x2)y' = 3\left(\sqrt x - \frac{2}{x}\right)^2\left(\frac{1}{2\sqrt x} + \frac{2}{x^2}\right), y′(4)=8132y'(4) = \frac{81}{32}
  • d) y′=−2cos⁡x(sin⁡x)3/2y' = -\frac{2\cos x}{(\sin x)^{3/2}}, y′(π6)=−26≈−4.8990y'\left(\frac{\pi}{6}\right) = -2\sqrt 6 \approx -4.8990
  • e) ddx(3x−1)−2=−6(3x−1)3\frac{d}{dx}(3x - 1)^{-2} = -\frac{6}{(3x - 1)^3}, which is −34-\frac{3}{4} at x=1x = 1

a) Rewrite first: y=6(2x2+1)−3y = 6(2x^2 + 1)^{-3}, the 66 staying upstairs. Let u=2x2+1u = 2x^2 + 1, so y=6u−3y = 6u^{-3}, dydu=−18u−4\frac{dy}{du} = -18u^{-4}, and u′=4xu' = 4x. By the Power Chain Rule, y′=−18(2x2+1)−4⋅4x=−72x(2x2+1)4y' = -18(2x^2 + 1)^{-4} \cdot 4x = -\frac{72x}{(2x^2 + 1)^4}. The exponent is −3−1=−4-3 - 1 = -4: moving TOWARD zero, to −2-2, is the classic slip with a negative exponent. At x=1x = 1: u=3u = 3, 34=813^4 = 81, so y′(1)=−7281=−89y'(1) = -\frac{72}{81} = -\frac{8}{9}. The quotient rule gives the same result, but with a chain rule hidden in the derivative of the denominator and a fraction to simplify at the end.

b) The index of the root goes in the denominator of the exponent: (x2−2x)23=(x2−2x)2/3\sqrt[3]{(x^2 - 2x)^2} = (x^2 - 2x)^{2/3}. Let u=x2−2xu = x^2 - 2x, u′=2x−2u' = 2x - 2. Then y′=23u−1/3u′=2(2x−2)3(x2−2x)1/3=4(x−1)3x2−2x3y' = \frac{2}{3}u^{-1/3} u' = \frac{2(2x - 2)}{3(x^2 - 2x)^{1/3}} = \frac{4(x - 1)}{3\sqrt[3]{x^2 - 2x}}, since 23−1=−13\frac{2}{3} - 1 = -\frac{1}{3} and a negative exponent sends the cube root to the denominator. At x=4x = 4: u=8u = 8, 83=2\sqrt[3]{8} = 2, so y′(4)=4⋅33⋅2=2y'(4) = \frac{4 \cdot 3}{3 \cdot 2} = 2. The function is defined everywhere, but y′y' is undefined where u=0u = 0, at x=0x = 0 and x=2x = 2: the rewriting u−1/3u^{-1/3} is what shows it.

c) Rewrite the inside: u=x1/2−2x−1u = x^{1/2} - 2x^{-1}, so u′=12x−1/2+2x−2=12x+2x2u' = \frac{1}{2}x^{-1/2} + 2x^{-2} = \frac{1}{2\sqrt x} + \frac{2}{x^2}. The minus sign of −2x−1-2x^{-1} meets the exponent −1-1 and becomes a PLUS: that sign is where the marks go. The outside function is the cube: y′=3u2u′=3(x−2x)2(12x+2x2)y' = 3u^2 u' = 3\left(\sqrt x - \frac{2}{x}\right)^2\left(\frac{1}{2\sqrt x} + \frac{2}{x^2}\right). At x=4x = 4: u=2−12=32u = 2 - \frac{1}{2} = \frac{3}{2} and u′=14+18=38u' = \frac{1}{4} + \frac{1}{8} = \frac{3}{8}, so y′(4)=3⋅94⋅38=8132y'(4) = 3 \cdot \frac{9}{4} \cdot \frac{3}{8} = \frac{81}{32}. Expanding the cube first would produce four terms with half-integer exponents: four chances to lose a sign for no gain.

d) Rewrite: y=4(sin⁡x)−1/2y = 4(\sin x)^{-1/2}. The outside function is 4u−1/24u^{-1/2}, of derivative −2u−3/2-2u^{-3/2} (since −12−1=−32-\frac{1}{2} - 1 = -\frac{3}{2}); the inside is u=sin⁡xu = \sin x, with u′=cos⁡xu' = \cos x. So y′=−2(sin⁡x)−3/2cos⁡x=−2cos⁡x(sin⁡x)3/2y' = -2(\sin x)^{-3/2}\cos x = -\frac{2\cos x}{(\sin x)^{3/2}}. At π6\frac{\pi}{6}: sin⁡π6=12\sin\frac{\pi}{6} = \frac{1}{2} and (12)−3/2=23/2=22\left(\frac{1}{2}\right)^{-3/2} = 2^{3/2} = 2\sqrt 2, a negative exponent flipping the fraction; cos⁡π6=32\cos\frac{\pi}{6} = \frac{\sqrt 3}{2}. So y′(π6)=−2⋅22⋅32=−26≈−4.8990y'\left(\frac{\pi}{6}\right) = -2 \cdot 2\sqrt 2 \cdot \frac{\sqrt 3}{2} = -2\sqrt 6 \approx -4.8990. The calculator confirms the decimal; the exact value is the expected answer.

e) The student took the derivative of 1v\frac{1}{v} to be 1v′\frac{1}{v'}, which is no rule at all, and then differentiated the square as if it were the function. Rewrite: y=(3x−1)−2y = (3x - 1)^{-2}, with u=3x−1u = 3x - 1 and u′=3u' = 3. Then y′=−2(3x−1)−3⋅3=−6(3x−1)3y' = -2(3x - 1)^{-3} \cdot 3 = -\frac{6}{(3x - 1)^3}. At x=1x = 1: −68=−34-\frac{6}{8} = -\frac{3}{4}, while the student's expression gives 112\frac{1}{12}, a POSITIVE slope. A sign check exposes it: for x>13x > \frac{1}{3} the denominator (3x−1)2(3x - 1)^2 grows, so yy decreases and its derivative must be negative. A constant over a power is a negative power; rewriting it removes the temptation.

Tick the exercises you have done or want to review: a free account, no password, keeps your ticks from one visit to the next and tells you which chapter to tackle next. Create your space, an email is enough.

Exercise 2: Three and four layers: one factor per layer

When a function is built in several layers, the chain rule is applied once per layer, and the derivative is a PRODUCT with one factor per layer. Thomas calls this repeated use of the Chain Rule: ddxf(g(h(x)))=f′(g(h(x)))⋅g′(h(x))⋅h′(x)\frac{d}{dx}f(g(h(x))) = f'(g(h(x))) \cdot g'(h(x)) \cdot h'(x).

The safe method: list the layers from the outside in BEFORE writing anything, then write one factor per layer, each outside derivative keeping everything inside it untouched, then count the factors against the layers. A calculator helps with the final decimal only if it is in RADIAN mode.

  • a) Differentiate y=1+xy = \sqrt{1 + \sqrt x} and give y′(9)y'(9).
  • b) Differentiate y=11+cos⁡2xy = \frac{1}{\sqrt{1 + \cos 2x}} and give the exact value of y′(π3)y'\left(\frac{\pi}{3}\right).
  • c) Differentiate y=cos⁡(x2+1)y = \cos\left(\sqrt{x^2 + 1}\right) and give y′(3)y'(\sqrt 3) to four decimals.
  • d) Let h(t)=esin⁡2(πt)h(t) = e^{\sin^2(\pi t)}. List its four layers, find h′(t)h'(t) and the exact value of h′(14)h'\left(\frac{1}{4}\right).
  • e) A student writes ddx(3+sin⁡3x)−2=−2(3+sin⁡3x)−3cos⁡3x\frac{d}{dx}(3 + \sin 3x)^{-2} = -2(3 + \sin 3x)^{-3}\cos 3x. Count the layers, correct the answer and give its value at x=0x = 0.

Type your answers, the page tells you right or wrong 0/7

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

Answers

  • a) y′=14x 1+xy' = \frac{1}{4\sqrt x\,\sqrt{1 + \sqrt x}}, y′(9)=124y'(9) = \frac{1}{24}
  • b) y′=sin⁡2x(1+cos⁡2x)3/2y' = \frac{\sin 2x}{(1 + \cos 2x)^{3/2}}, y′(π3)=6y'\left(\frac{\pi}{3}\right) = \sqrt 6
  • c) y′=−xsin⁡x2+1x2+1y' = -\frac{x\sin\sqrt{x^2 + 1}}{\sqrt{x^2 + 1}}, y′(3)=−32sin⁡2≈−0.7875y'(\sqrt 3) = -\frac{\sqrt 3}{2}\sin 2 \approx -0.7875
  • d) Layers e(⋅)e^{(\cdot)}, square, sine, πt\pi t; h′(t)=2πsin⁡(πt)cos⁡(πt) esin⁡2(πt)h'(t) = 2\pi\sin(\pi t)\cos(\pi t)\, e^{\sin^2(\pi t)}, h′(14)=πeh'\left(\frac{1}{4}\right) = \pi\sqrt e
  • e) Three layers; y′=−6cos⁡3x(3+sin⁡3x)3y' = -\frac{6\cos 3x}{(3 + \sin 3x)^3}, which is −29-\frac{2}{9} at x=0x = 0

a) Layers: the outer square root, then 1+w1 + w, then w=xw = \sqrt x. Factors: 121+x\frac{1}{2\sqrt{1 + \sqrt x}} for the outer root, 11 for 1+w1 + w, and 12x\frac{1}{2\sqrt x} for the inner root. So y′=121+x⋅12x=14x 1+xy' = \frac{1}{2\sqrt{1 + \sqrt x}} \cdot \frac{1}{2\sqrt x} = \frac{1}{4\sqrt x\,\sqrt{1 + \sqrt x}}. At x=9x = 9: 9=3\sqrt 9 = 3 and 4=2\sqrt 4 = 2, so y′(9)=14⋅3⋅2=124y'(9) = \frac{1}{4 \cdot 3 \cdot 2} = \frac{1}{24}. The two roots are two different layers, each with its own factor: the frequent error writes 121+x\frac{1}{2\sqrt{1 + x}}, merging them.

b) Rewrite: y=(1+cos⁡2x)−1/2y = (1 + \cos 2x)^{-1/2}. Layers: the power −12-\frac{1}{2}, then 1+cos⁡w1 + \cos w, then w=2xw = 2x. Factors: −12(1+cos⁡2x)−3/2-\frac{1}{2}(1 + \cos 2x)^{-3/2}, then −sin⁡2x-\sin 2x, then 22. The product: y′=−12⋅(−sin⁡2x)⋅2⋅(1+cos⁡2x)−3/2=sin⁡2x(1+cos⁡2x)3/2y' = -\frac{1}{2} \cdot (-\sin 2x) \cdot 2 \cdot (1 + \cos 2x)^{-3/2} = \frac{\sin 2x}{(1 + \cos 2x)^{3/2}}. Two minus signs meet and cancel, one from the exponent and one from the cosine: losing either one flips every slope. At π3\frac{\pi}{3}: sin⁡2π3=32\sin\frac{2\pi}{3} = \frac{\sqrt 3}{2} and 1+cos⁡2π3=121 + \cos\frac{2\pi}{3} = \frac{1}{2}, with (12)3/2=122\left(\frac{1}{2}\right)^{3/2} = \frac{1}{2\sqrt 2}, so y′=32⋅22=6y' = \frac{\sqrt 3}{2} \cdot 2\sqrt 2 = \sqrt 6. Check by an identity: 1+cos⁡2x=2cos⁡2x1 + \cos 2x = 2\cos^2 x, so on (0,π2)\left(0, \frac{\pi}{2}\right), y=12cos⁡xy = \frac{1}{\sqrt 2\cos x} and y′=sin⁡x2cos⁡2xy' = \frac{\sin x}{\sqrt 2\cos^2 x}, which is 3/22/4=6\frac{\sqrt 3/2}{\sqrt 2/4} = \sqrt 6 at π3\frac{\pi}{3} as well.

c) Layers: cos⁡\cos, then the square root, then x2+1x^2 + 1. Factors: −sin⁡x2+1-\sin\sqrt{x^2 + 1}, then 12x2+1\frac{1}{2\sqrt{x^2 + 1}}, then 2x2x. So y′=−xsin⁡x2+1x2+1y' = -\frac{x\sin\sqrt{x^2 + 1}}{\sqrt{x^2 + 1}}, the 22 cancelling. At x=3x = \sqrt 3: x2+1=2\sqrt{x^2 + 1} = 2, so y′(3)=−32sin⁡2≈−0.7875y'(\sqrt 3) = -\frac{\sqrt 3}{2}\sin 2 \approx -0.7875, with sin⁡2\sin 2 meaning the sine of 22 RADIANS. A calculator left in degree mode gives sin⁡2∘\sin 2^\circ and the answer −0.0302-0.0302: every formula of the course assumes radians.

d) Layers, from the outside in: e(⋅)e^{(\cdot)}, the square, the sine, and πt\pi t. Four layers, four factors: h′(t)=esin⁡2(πt)⋅2sin⁡(πt)⋅cos⁡(πt)⋅πh'(t) = e^{\sin^2(\pi t)} \cdot 2\sin(\pi t) \cdot \cos(\pi t) \cdot \pi. The exponential comes back unchanged, with its whole exponent. At t=14t = \frac{1}{4}: sin⁡π4=cos⁡π4=22\sin\frac{\pi}{4} = \cos\frac{\pi}{4} = \frac{\sqrt 2}{2}, so sin⁡2=12\sin^2 = \frac{1}{2} and 2sin⁡cos⁡=12\sin\cos = 1; hence h′(14)=e1/2⋅1⋅π=πe≈5.1797h'\left(\frac{1}{4}\right) = e^{1/2} \cdot 1 \cdot \pi = \pi\sqrt e \approx 5.1797. The double-angle identity of Thomas 1.3 gives a shorter form, h′(t)=πsin⁡(2πt) esin⁡2(πt)h'(t) = \pi\sin(2\pi t)\, e^{\sin^2(\pi t)}.

e) Layers: the power −2-2, then 3+sin⁡w3 + \sin w, then w=3xw = 3x. Three layers, and the student wrote two factors: the inner 33 is missing. Correct: y′=−2(3+sin⁡3x)−3⋅cos⁡3x⋅3=−6cos⁡3x(3+sin⁡3x)3y' = -2(3 + \sin 3x)^{-3} \cdot \cos 3x \cdot 3 = -\frac{6\cos 3x}{(3 + \sin 3x)^3}. At x=0x = 0: −627=−29-\frac{6}{27} = -\frac{2}{9}, three times the student's −227-\frac{2}{27}. The innermost factor is forgotten precisely because it is only a number; counting layers against factors catches it every time.

Exercise 3: The chain rule from a table of values

On a MATH 203 exam and in WeBWorK, the chain rule is often tested on functions known only through a table. The rule is then pure bookkeeping: (f∘g)′(a)=f′(g(a)) g′(a)(f \circ g)'(a) = f'(g(a))\, g'(a). Compute the inside value g(a)g(a) FIRST, write it down as a number, and only then go to the row of that number to read f′f'.

The table gives values of two differentiable functions ff and gg and of their derivatives. Some of the questions below cannot be answered from it: saying so, with the reason, is the answer.

xxf(x)f(x)f′(x)f'(x)g(x)g(x)g′(x)g'(x)
0022−1-11133
11443344−2-2
2211550012\frac{1}{2}
4433−2-22266
  • a) Let h(x)=f(g(x))h(x) = f(g(x)). Find h′(0)h'(0) and h′(2)h'(2). A student answers h′(0)=−3h'(0) = -3: what did the student compute?
  • b) Let k(x)=g(f(x))k(x) = g(f(x)). Find k′(2)k'(2).
  • c) Let P(x)=f(x)P(x) = \sqrt{f(x)} and Q(x)=g(x)Q(x) = g(\sqrt x). Find P′(1)P'(1) and Q′(4)Q'(4).
  • d) Let R(x)=f(g(f(x)))R(x) = f(g(f(x))) and S(x)=2g(x)S(x) = 2^{g(x)}. Find R′(0)R'(0), then the exact value of S′(1)S'(1).
  • e) Among (f∘g)′(4)(f \circ g)'(4), (g∘f)′(4)(g \circ f)'(4) and (f∘f)′(0)(f \circ f)'(0), compute those the table allows, and explain why one of them cannot be found.

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

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

Answers

  • a) h′(0)=f′(1) g′(0)=9h'(0) = f'(1)\,g'(0) = 9, h′(2)=f′(0) g′(2)=−12h'(2) = f'(0)\,g'(2) = -\frac{1}{2}; −3=f′(0) g′(0)-3 = f'(0)\,g'(0) is the wrong row.
  • b) k′(2)=g′(f(2)) f′(2)=g′(1)⋅5=−10k'(2) = g'(f(2))\,f'(2) = g'(1) \cdot 5 = -10
  • c) P′(1)=f′(1)2f(1)=34P'(1) = \frac{f'(1)}{2\sqrt{f(1)}} = \frac{3}{4}; Q′(4)=g′(2)⋅14=18Q'(4) = g'(2) \cdot \frac{1}{4} = \frac{1}{8}
  • d) R′(0)=f′(0) g′(2) f′(0)=12R'(0) = f'(0)\,g'(2)\,f'(0) = \frac{1}{2}; S′(1)=24ln⁡2⋅(−2)=−32ln⁡2S'(1) = 2^{4}\ln 2 \cdot (-2) = -32\ln 2
  • e) (f∘g)′(4)=30(f \circ g)'(4) = 30, (f∘f)′(0)=−5(f \circ f)'(0) = -5; (g∘f)′(4)(g \circ f)'(4) needs g′(3)g'(3), not in the table.

a) h′(0)=f′(g(0))⋅g′(0)h'(0) = f'(g(0)) \cdot g'(0). Read g(0)=1g(0) = 1 first, then f′f' in the row x=1x = 1: f′(1)=3f'(1) = 3. With g′(0)=3g'(0) = 3, h′(0)=3⋅3=9h'(0) = 3 \cdot 3 = 9. The student's −3-3 is f′(0)⋅g′(0)=(−1)⋅3f'(0) \cdot g'(0) = (-1) \cdot 3: the outside derivative was read at x=0x = 0 instead of at g(0)=1g(0) = 1. The outside function never sees xx, only the value that gg hands it. Similarly h′(2)=f′(g(2))⋅g′(2)=f′(0)⋅12=−12h'(2) = f'(g(2)) \cdot g'(2) = f'(0) \cdot \frac{1}{2} = -\frac{1}{2}, since g(2)=0g(2) = 0.

b) k′(2)=g′(f(2))⋅f′(2)k'(2) = g'(f(2)) \cdot f'(2). First f(2)=1f(2) = 1, then g′(1)=−2g'(1) = -2; with f′(2)=5f'(2) = 5, k′(2)=−2⋅5=−10k'(2) = -2 \cdot 5 = -10. The order of composition decides which function is read where: in g∘fg \circ f, the table of g′g' is read at a value of ff.

c) P=uP = \sqrt u with u=f(x)u = f(x), so P′(x)=f′(x)2f(x)P'(x) = \frac{f'(x)}{2\sqrt{f(x)}} and P′(1)=324=34P'(1) = \frac{3}{2\sqrt 4} = \frac{3}{4}. For Q(x)=g(x)Q(x) = g(\sqrt x), the inside is x\sqrt x, of derivative 12x\frac{1}{2\sqrt x}: Q′(x)=g′(x)⋅12xQ'(x) = g'(\sqrt x) \cdot \frac{1}{2\sqrt x}. At x=4x = 4: 4=2\sqrt 4 = 2, g′(2)=12g'(2) = \frac{1}{2} and 124=14\frac{1}{2\sqrt 4} = \frac{1}{4}, so Q′(4)=18Q'(4) = \frac{1}{8}. Reading g′(4)=6g'(4) = 6 because the question says x=4x = 4 gives 32\frac{3}{2}, twelve times too large.

d) Three layers, three factors, each read at its own point. Follow the values from the inside: f(0)=2f(0) = 2, then g(2)=0g(2) = 0. So R′(0)=f′(g(f(0)))⋅g′(f(0))⋅f′(0)=f′(0)⋅g′(2)⋅f′(0)=(−1)⋅12⋅(−1)=12R'(0) = f'(g(f(0))) \cdot g'(f(0)) \cdot f'(0) = f'(0) \cdot g'(2) \cdot f'(0) = (-1) \cdot \frac{1}{2} \cdot (-1) = \frac{1}{2}. The two factors f′(0)f'(0) are read at different stages that happen to share the value 00. For S(x)=2g(x)=eg(x)ln⁡2S(x) = 2^{g(x)} = e^{g(x)\ln 2}, with ln⁡2\ln 2 a constant: S′(x)=2g(x)ln⁡2⋅g′(x)S'(x) = 2^{g(x)}\ln 2 \cdot g'(x). At x=1x = 1: g(1)=4g(1) = 4, so S′(1)=24ln⁡2⋅(−2)=−32ln⁡2≈−22.1807S'(1) = 2^4 \ln 2 \cdot (-2) = -32\ln 2 \approx -22.1807.

e) (f∘g)′(4)=f′(g(4))⋅g′(4)=f′(2)⋅6=5⋅6=30(f \circ g)'(4) = f'(g(4)) \cdot g'(4) = f'(2) \cdot 6 = 5 \cdot 6 = 30. (f∘f)′(0)=f′(f(0))⋅f′(0)=f′(2)⋅(−1)=−5(f \circ f)'(0) = f'(f(0)) \cdot f'(0) = f'(2) \cdot (-1) = -5. But (g∘f)′(4)=g′(f(4))⋅f′(4)=g′(3)⋅(−2)(g \circ f)'(4) = g'(f(4)) \cdot f'(4) = g'(3) \cdot (-2), and the table has no row x=3x = 3: the value cannot be found. Taking g′(4)=6g'(4) = 6 instead would be the wrong-row error again, disguised as an answer.

Exercise 4: Exponentials in base a: the laws of exponents first

For a>0a > 0, ax=exln⁡aa^x = e^{x\ln a}, where ln⁡a\ln a is a CONSTANT. With the inside function u=xln⁡au = x\ln a, the chain rule gives ddxax=axln⁡a\frac{d}{dx}a^x = a^x\ln a, and more generally ddxau=auln⁡a⋅u′\frac{d}{dx}a^u = a^u \ln a \cdot u'. Nothing about logarithms is differentiated here.

Most of the work, and most of the lost marks, come BEFORE the derivative: rewrite with the laws of exponents so that a single exponential remains. 1ax=a−x\frac{1}{a^x} = a^{-x}, ax=ax/2\sqrt{a^x} = a^{x/2}, (ax)k=akx(a^x)^k = a^{kx}. The figure shows the curve of part d).

-11234560.10.20.30.40.50.6y = x²·3⁻ˣx
  • a) Differentiate y=52xy = \frac{5}{2^x} and give y′(1)y'(1).
  • b) Differentiate y=3xy = \sqrt{3^x} and give y′(2)y'(2).
  • c) Differentiate y=2x2−2xy = 2^{x^2 - 2x}. Find the point where its tangent is horizontal, and give y′(2)y'(2).
  • d) Differentiate y=x2 3−xy = x^2\,3^{-x}, write y′y' as a product of factors, and find the xx-values where the tangent is horizontal.
  • e) Find the base a>0a > 0 for which y=axy = a^x satisfies y′=3yy' = 3y for all xx. Then find aa so that y=a−2xy = a^{-2x} satisfies y′=−yy' = -y.

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

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

Answers

  • a) y′=−5ln⁡2⋅2−xy' = -5\ln 2 \cdot 2^{-x}, y′(1)=−52ln⁡2≈−1.7329y'(1) = -\frac{5}{2}\ln 2 \approx -1.7329
  • b) y′=ln⁡32 3x/2y' = \frac{\ln 3}{2}\,3^{x/2}, y′(2)=32ln⁡3≈1.6479y'(2) = \frac{3}{2}\ln 3 \approx 1.6479
  • c) y′=(2x−2)ln⁡2⋅2x2−2xy' = (2x - 2)\ln 2 \cdot 2^{x^2 - 2x}; horizontal at (1,12)\left(1, \frac{1}{2}\right); y′(2)=2ln⁡2≈1.3863y'(2) = 2\ln 2 \approx 1.3863
  • d) y′=x 3−x(2−xln⁡3)y' = x\,3^{-x}(2 - x\ln 3); horizontal at x=0x = 0 and x=2ln⁡3≈1.8205x = \frac{2}{\ln 3} \approx 1.8205
  • e) a=e3≈20.0855a = e^3 \approx 20.0855; a=e≈1.6487a = \sqrt e \approx 1.6487

a) Rewrite: y=5⋅2−x=5e−xln⁡2y = 5 \cdot 2^{-x} = 5e^{-x\ln 2}. The inside function is u=−xln⁡2u = -x\ln 2, with u′=−ln⁡2u' = -\ln 2. So y′=5⋅2−x⋅(−ln⁡2)=−5ln⁡2⋅2−xy' = 5 \cdot 2^{-x} \cdot (-\ln 2) = -5\ln 2 \cdot 2^{-x}. At x=1x = 1: y′(1)=−5ln⁡22≈−1.7329y'(1) = -\frac{5\ln 2}{2} \approx -1.7329. A student who applies the power rule to 2−x2^{-x} and writes −x 2−x−1-x\,2^{-x-1} confuses a variable exponent with a variable base: the power rule needs a constant exponent.

b) A power of a power multiplies the exponents: 3x=(3x)1/2=3x/2\sqrt{3^x} = (3^x)^{1/2} = 3^{x/2}. With u=x2u = \frac{x}{2}: y′=3x/2ln⁡3⋅12y' = 3^{x/2}\ln 3 \cdot \frac{1}{2}. At x=2x = 2: y′(2)=3ln⁡32≈1.6479y'(2) = \frac{3\ln 3}{2} \approx 1.6479. The Power Chain Rule on the square root reaches the same place with more writing: 12(3x)−1/2⋅3xln⁡3=ln⁡32 3x/2\frac{1}{2}(3^x)^{-1/2} \cdot 3^x\ln 3 = \frac{\ln 3}{2}\,3^{x/2}, after using 3x⋅3−x/2=3x/23^x \cdot 3^{-x/2} = 3^{x/2}, another law of exponents.

c) The outside function is 2u2^u, the inside u=x2−2xu = x^2 - 2x: y′=2x2−2xln⁡2⋅(2x−2)y' = 2^{x^2 - 2x}\ln 2 \cdot (2x - 2). The factor 2x2−2x2^{x^2 - 2x} is never zero and ln⁡2≠0\ln 2 \ne 0, so y′=0y' = 0 only when 2x−2=02x - 2 = 0: x=1x = 1, where y=2−1=12y = 2^{-1} = \frac{1}{2}. The horizontal tangent is at (1,12)\left(1, \frac{1}{2}\right). At x=2x = 2: u=0u = 0, so y′(2)=1⋅ln⁡2⋅2=2ln⁡2≈1.3863y'(2) = 1 \cdot \ln 2 \cdot 2 = 2\ln 2 \approx 1.3863. Saying WHY the exponential factor cannot vanish is part of the answer.

d) The last operation is a product: product rule, with the chain rule on 3−x3^{-x}, whose derivative is 3−xln⁡3⋅(−1)3^{-x}\ln 3 \cdot (-1). So y′=2x 3−x−x2ln⁡3⋅3−xy' = 2x\,3^{-x} - x^2\ln 3 \cdot 3^{-x}. Factor the common factors, xx (its lowest power) and 3−x3^{-x}: y′=x 3−x(2−xln⁡3)y' = x\,3^{-x}(2 - x\ln 3). Since 3−x>03^{-x} > 0, y′=0y' = 0 exactly when x=0x = 0 or x=2ln⁡3≈1.8205x = \frac{2}{\ln 3} \approx 1.8205: two horizontal tangents, as the figure shows. At the second one, 3−2/ln⁡3=e−23^{-2/\ln 3} = e^{-2} (because 3−x=e−xln⁡33^{-x} = e^{-x\ln 3}), so y=4(ln⁡3)2e−2≈0.4485y = \frac{4}{(\ln 3)^2}e^{-2} \approx 0.4485. The factored form makes the equation solvable in one line; the expanded form hides it.

e) By the chain rule, (ax)′=axln⁡a=(ln⁡a) y(a^x)' = a^x\ln a = (\ln a)\, y. So y′=3yy' = 3y for all xx exactly when ln⁡a=3\ln a = 3, that is a=e3≈20.0855a = e^3 \approx 20.0855. For y=a−2xy = a^{-2x}: y′=a−2xln⁡a⋅(−2)=−2ln⁡a⋅yy' = a^{-2x}\ln a \cdot (-2) = -2\ln a \cdot y, and −2ln⁡a=−1-2\ln a = -1 gives ln⁡a=12\ln a = \frac{1}{2}, a=e1/2=e≈1.6487a = e^{1/2} = \sqrt e \approx 1.6487. Every exponential axa^x is ekxe^{kx} with k=ln⁡ak = \ln a, and the inside derivative kk is the factor that the derivative carries.

Exercise 5: Factor the lowest power: the algebra after the chain rule

When the product or quotient rule meets the chain rule, the derivative comes out as a SUM of terms containing the same brackets at different powers. It is not finished: a sum cannot be set equal to zero or signed. Factor each bracket at its LOWEST power, the most negative exponent (−12-\frac{1}{2} is lower than 12\frac{1}{2}, and −3-3 is lower than −2-2). What is left inside the square bracket then has only non-negative powers and simplifies to a polynomial.

The rule for the leftover exponents: after taking out umu^{m}, a term containing unu^{n} keeps un−mu^{n - m}. Rewriting a quotient as a product with a negative exponent, before differentiating, usually saves the quotient rule altogether. The figure shows the curve of parts d) and e).

-3-2-112-3-2-1123456y = (x + 2)·∛(x² − 1)x
  • a) Differentiate y=x1−2xy = x\sqrt{1 - 2x}, write y′y' as a single fraction, give y′(−4)y'(-4) and the xx-value of the horizontal tangent.
  • b) Differentiate y=(x+2)3(2x−1)2y = \frac{(x + 2)^3}{(2x - 1)^2} by rewriting it as a product. Give y′y' factored, y′(1)y'(1), and the xx-values of the horizontal tangents.
  • c) Differentiate y=x(x2+4)3/2y = \frac{x}{(x^2 + 4)^{3/2}} and show that y′=4−2x2(x2+4)5/2y' = \frac{4 - 2x^2}{(x^2 + 4)^{5/2}}. Give y′(0)y'(0) and the positive zero of y′y'.
  • d) Differentiate y=(x+2)x2−13y = (x + 2)\sqrt[3]{x^2 - 1}, factor, and give y′(3)y'(3).
  • e) Using d), find the xx-values where the curve has a horizontal tangent, and those where y′y' does not exist. Compare with the figure.

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

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

Answers

  • a) y′=1−3x1−2xy' = \frac{1 - 3x}{\sqrt{1 - 2x}}, y′(−4)=133y'(-4) = \frac{13}{3}, horizontal at x=13x = \frac{1}{3}
  • b) y′=(x+2)2(2x−11)(2x−1)3y' = \frac{(x + 2)^2(2x - 11)}{(2x - 1)^3}, y′(1)=−81y'(1) = -81, horizontal at x=−2x = -2 and x=112x = \frac{11}{2}
  • c) y′=4−2x2(x2+4)5/2y' = \frac{4 - 2x^2}{(x^2 + 4)^{5/2}}, y′(0)=18y'(0) = \frac{1}{8}, positive zero x=2x = \sqrt 2
  • d) y′=5x2+4x−33(x2−1)2/3y' = \frac{5x^2 + 4x - 3}{3(x^2 - 1)^{2/3}}, y′(3)=92y'(3) = \frac{9}{2}
  • e) Horizontal at x=−2±195x = \frac{-2 \pm \sqrt{19}}{5} (≈−1.2718\approx -1.2718 and 0.47180.4718); y′y' undefined at x=±1x = \pm 1 (vertical tangents)

a) y=x(1−2x)1/2y = x(1 - 2x)^{1/2}, a product. With u=1−2xu = 1 - 2x, u′=−2u' = -2: y′=(1−2x)1/2+x⋅12(1−2x)−1/2(−2)=(1−2x)1/2−x(1−2x)−1/2y' = (1 - 2x)^{1/2} + x \cdot \frac{1}{2}(1 - 2x)^{-1/2}(-2) = (1 - 2x)^{1/2} - x(1 - 2x)^{-1/2}. The lowest power of the bracket is −12-\frac{1}{2}; taking it out of (1−2x)1/2(1 - 2x)^{1/2} leaves (1−2x)1/2−(−1/2)=(1−2x)1(1 - 2x)^{1/2 - (-1/2)} = (1 - 2x)^1. So y′=(1−2x)−1/2[(1−2x)−x]=1−3x1−2xy' = (1 - 2x)^{-1/2}\left[(1 - 2x) - x\right] = \frac{1 - 3x}{\sqrt{1 - 2x}}, for x<12x < \frac{1}{2}. At x=−4x = -4: 139=133\frac{13}{\sqrt 9} = \frac{13}{3}. Horizontal tangent where the numerator vanishes: x=13x = \frac{1}{3}.

b) Rewrite: y=(x+2)3(2x−1)−2y = (x + 2)^3(2x - 1)^{-2}. Product rule, chain rule on each factor: y′=3(x+2)2(2x−1)−2+(x+2)3⋅(−2)(2x−1)−3⋅2=3(x+2)2(2x−1)−2−4(x+2)3(2x−1)−3y' = 3(x + 2)^2(2x - 1)^{-2} + (x + 2)^3 \cdot (-2)(2x - 1)^{-3} \cdot 2 = 3(x + 2)^2(2x - 1)^{-2} - 4(x + 2)^3(2x - 1)^{-3}. Lowest powers: (x+2)2(x + 2)^2 and (2x−1)−3(2x - 1)^{-3}, NOT (2x−1)−2(2x - 1)^{-2}. The leftover: y′=(x+2)2(2x−1)−3[3(2x−1)−4(x+2)]=(x+2)2(2x−11)(2x−1)3y' = (x + 2)^2(2x - 1)^{-3}\left[3(2x - 1) - 4(x + 2)\right] = \frac{(x + 2)^2(2x - 11)}{(2x - 1)^3}. At x=1x = 1: 9⋅(−9)1=−81\frac{9 \cdot (-9)}{1} = -81. Horizontal tangents at x=−2x = -2 and x=112x = \frac{11}{2}. Factoring (2x−1)−2(2x - 1)^{-2} instead would leave (2x−1)−1(2x - 1)^{-1} inside the bracket, a fraction inside the factorisation: the sign that the wrong power was taken out.

c) Rewrite: y=x(x2+4)−3/2y = x(x^2 + 4)^{-3/2}. Then y′=(x2+4)−3/2+x⋅(−32)(x2+4)−5/2⋅2x=(x2+4)−3/2−3x2(x2+4)−5/2y' = (x^2 + 4)^{-3/2} + x \cdot \left(-\frac{3}{2}\right)(x^2 + 4)^{-5/2} \cdot 2x = (x^2 + 4)^{-3/2} - 3x^2(x^2 + 4)^{-5/2}. The lowest power is −52-\frac{5}{2}, and −32−(−52)=1-\frac{3}{2} - \left(-\frac{5}{2}\right) = 1: y′=(x2+4)−5/2[(x2+4)−3x2]=4−2x2(x2+4)5/2y' = (x^2 + 4)^{-5/2}\left[(x^2 + 4) - 3x^2\right] = \frac{4 - 2x^2}{(x^2 + 4)^{5/2}}. At x=0x = 0: 445/2=432=18\frac{4}{4^{5/2}} = \frac{4}{32} = \frac{1}{8}. Zeros: x2=2x^2 = 2, the positive one is 2≈1.4142\sqrt 2 \approx 1.4142. The quotient rule gives (x2+4)3/2−3x2(x2+4)1/2(x2+4)3\frac{(x^2 + 4)^{3/2} - 3x^2(x^2 + 4)^{1/2}}{(x^2 + 4)^3}, a complex fraction that needs this very factorisation to be finished.

d) y=(x+2)(x2−1)1/3y = (x + 2)(x^2 - 1)^{1/3}. With u=x2−1u = x^2 - 1: y′=(x2−1)1/3+(x+2)⋅13(x2−1)−2/3⋅2xy' = (x^2 - 1)^{1/3} + (x + 2) \cdot \frac{1}{3}(x^2 - 1)^{-2/3} \cdot 2x. Lowest power −23-\frac{2}{3}, and 13−(−23)=1\frac{1}{3} - \left(-\frac{2}{3}\right) = 1: y′=(x2−1)−2/3[(x2−1)+2x(x+2)3]=3x2−3+2x2+4x3(x2−1)2/3=5x2+4x−33(x2−1)2/3y' = (x^2 - 1)^{-2/3}\left[(x^2 - 1) + \frac{2x(x + 2)}{3}\right] = \frac{3x^2 - 3 + 2x^2 + 4x}{3(x^2 - 1)^{2/3}} = \frac{5x^2 + 4x - 3}{3(x^2 - 1)^{2/3}}. At x=3x = 3: x2−1=8x^2 - 1 = 8, 82/3=48^{2/3} = 4, so y′(3)=45+12−312=5412=92y'(3) = \frac{45 + 12 - 3}{12} = \frac{54}{12} = \frac{9}{2}.

e) y′=0y' = 0 when 5x2+4x−3=05x^2 + 4x - 3 = 0: discriminant 16+60=7616 + 60 = 76, so x=−4±7610=−2±195x = \frac{-4 \pm \sqrt{76}}{10} = \frac{-2 \pm \sqrt{19}}{5}, about −1.2718-1.2718 and 0.47180.4718. The denominator vanishes at x=±1x = \pm 1, where y′y' does not exist: the numerator is 66 at x=1x = 1 and −2-2 at x=−1x = -1, not zero, so ∣y′∣|y'| grows without bound and the curve has a VERTICAL tangent there, as the figure shows at the two marked points. The factored form answers both questions at once; the unfactored sum answers neither.

Part B: problems and reasoning (/50)

Exercise 6: Second and higher derivatives: the chain inside the product

The first derivative of a composite is a PRODUCT, f′(g(x))⋅g′(x)f'(g(x)) \cdot g'(x), so the second derivative needs the product rule, and the chain rule again inside the first factor. Nothing new is required, only the discipline of seeing the product that the first derivative created.

Write y′y' with negative exponents rather than as a quotient before differentiating again, and finish y′′y'' by factoring the lowest power: that is where the marks of this exercise are decided.

  • a) Let y=11+x2y = \frac{1}{1 + x^2}. Show that y′′=6x2−2(1+x2)3y'' = \frac{6x^2 - 2}{(1 + x^2)^3}, give y′′(1)y''(1) and the positive zero of y′′y''.
  • b) Let y=tan⁡(2x)y = \tan(2x). Find y′′y'' and its value at x=π8x = \frac{\pi}{8}.
  • c) Let y=(4−x2)3/2y = (4 - x^2)^{3/2}. Find y′′y'' as a single fraction and give y′′(1)y''(1) exactly.
  • d) Let g(x)=f(x2)g(x) = f(x^2), where ff is twice differentiable with f′(4)=−1f'(4) = -1 and f′′(4)=3f''(4) = 3. Express g′′(x)g''(x) in terms of f′f' and f′′f'' and find g′′(2)g''(2). A student answers 4848: what was forgotten?
  • e) Let y=(1−2x)−1y = (1 - 2x)^{-1}. Find y′y', y′′y'', y′′′y''', guess the formula for the nn-th derivative, and give y′′′(0)y'''(0) and y(5)(0)y^{(5)}(0).

Type your answers, the page tells you right or wrong 0/7

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

Answers

  • a) y′′=6x2−2(1+x2)3y'' = \frac{6x^2 - 2}{(1 + x^2)^3}, y′′(1)=12y''(1) = \frac{1}{2}, positive zero x=13x = \frac{1}{\sqrt 3}
  • b) y′′=8sec⁡2(2x)tan⁡(2x)y'' = 8\sec^2(2x)\tan(2x), y′′(π8)=16y''\left(\frac{\pi}{8}\right) = 16
  • c) y′′=6x2−124−x2y'' = \frac{6x^2 - 12}{\sqrt{4 - x^2}}, y′′(1)=−23y''(1) = -2\sqrt 3
  • d) g′′(x)=2f′(x2)+4x2f′′(x2)g''(x) = 2f'(x^2) + 4x^2 f''(x^2), g′′(2)=46g''(2) = 46; 4848 forgets the term 2f′(x2)2f'(x^2).
  • e) y(n)=2n n! (1−2x)−(n+1)y^{(n)} = 2^n\, n!\,(1 - 2x)^{-(n+1)}; y′′′(0)=48y'''(0) = 48, y(5)(0)=3840y^{(5)}(0) = 3840

a) y=(1+x2)−1y = (1 + x^2)^{-1}, so y′=−(1+x2)−2⋅2x=−2x(1+x2)−2y' = -(1 + x^2)^{-2} \cdot 2x = -2x(1 + x^2)^{-2}, a product. Product rule, chain rule on the second factor: y′′=−2(1+x2)−2+(−2x)⋅(−2)(1+x2)−3⋅2x=−2(1+x2)−2+8x2(1+x2)−3y'' = -2(1 + x^2)^{-2} + (-2x) \cdot (-2)(1 + x^2)^{-3} \cdot 2x = -2(1 + x^2)^{-2} + 8x^2(1 + x^2)^{-3}. Lowest power −3-3: y′′=(1+x2)−3[−2(1+x2)+8x2]=6x2−2(1+x2)3y'' = (1 + x^2)^{-3}\left[-2(1 + x^2) + 8x^2\right] = \frac{6x^2 - 2}{(1 + x^2)^3}. At x=1x = 1: 48=12\frac{4}{8} = \frac{1}{2}. Zeros: x2=13x^2 = \frac{1}{3}, the positive one 13≈0.5774\frac{1}{\sqrt 3} \approx 0.5774. Differentiating −2x(1+x2)2\frac{-2x}{(1 + x^2)^2} by the quotient rule works too, but produces a factor (1+x2)(1 + x^2) to cancel against (1+x2)4(1 + x^2)^4: one more place to slip.

b) y′=sec⁡2(2x)⋅2=2(sec⁡2x)2y' = \sec^2(2x) \cdot 2 = 2(\sec 2x)^2. For the second derivative, sec⁡2(2x)\sec^2(2x) is itself three layers: the square, the secant, 2x2x. So y′′=2⋅2sec⁡(2x)⋅sec⁡(2x)tan⁡(2x)⋅2=8sec⁡2(2x)tan⁡(2x)y'' = 2 \cdot 2\sec(2x) \cdot \sec(2x)\tan(2x) \cdot 2 = 8\sec^2(2x)\tan(2x). At π8\frac{\pi}{8}: 2x=π42x = \frac{\pi}{4}, sec⁡2π4=2\sec^2\frac{\pi}{4} = 2 and tan⁡π4=1\tan\frac{\pi}{4} = 1, so y′′(π8)=16y''\left(\frac{\pi}{8}\right) = 16. The usual loss is the last factor 22, which gives 88: in a second derivative the inside derivative of 2x2x appears TWICE, once per differentiation.

c) y′=32(4−x2)1/2⋅(−2x)=−3x(4−x2)1/2y' = \frac{3}{2}(4 - x^2)^{1/2} \cdot (-2x) = -3x(4 - x^2)^{1/2}. Product rule: y′′=−3(4−x2)1/2+(−3x)⋅12(4−x2)−1/2(−2x)=−3(4−x2)1/2+3x2(4−x2)−1/2y'' = -3(4 - x^2)^{1/2} + (-3x) \cdot \frac{1}{2}(4 - x^2)^{-1/2}(-2x) = -3(4 - x^2)^{1/2} + 3x^2(4 - x^2)^{-1/2}. Lowest power −12-\frac{1}{2}: y′′=(4−x2)−1/2[−3(4−x2)+3x2]=6x2−124−x2y'' = (4 - x^2)^{-1/2}\left[-3(4 - x^2) + 3x^2\right] = \frac{6x^2 - 12}{\sqrt{4 - x^2}}. At x=1x = 1: −63=−23≈−3.4641\frac{-6}{\sqrt 3} = -2\sqrt 3 \approx -3.4641, rationalised by 63=633\frac{6}{\sqrt 3} = \frac{6\sqrt 3}{3}.

d) First derivative: g′(x)=f′(x2)⋅2xg'(x) = f'(x^2) \cdot 2x, a product. Product rule: g′′(x)=[f′(x2)]′⋅2x+f′(x2)⋅2g''(x) = \left[f'(x^2)\right]' \cdot 2x + f'(x^2) \cdot 2, and [f′(x2)]′=f′′(x2)⋅2x\left[f'(x^2)\right]' = f''(x^2) \cdot 2x by the chain rule once more. So g′′(x)=4x2f′′(x2)+2f′(x2)g''(x) = 4x^2 f''(x^2) + 2f'(x^2). At x=2x = 2: x2=4x^2 = 4, so g′′(2)=16⋅3+2⋅(−1)=46g''(2) = 16 \cdot 3 + 2 \cdot (-1) = 46. The student's 48=16⋅348 = 16 \cdot 3 is the first term alone: the product rule was skipped and the derivative of the factor 2x2x forgotten. Note that f′f' and f′′f'' are read at x2=4x^2 = 4, never at 22.

e) y′=−1⋅(1−2x)−2⋅(−2)=2(1−2x)−2y' = -1 \cdot (1 - 2x)^{-2} \cdot (-2) = 2(1 - 2x)^{-2}. y′′=2⋅(−2)(1−2x)−3⋅(−2)=8(1−2x)−3y'' = 2 \cdot (-2)(1 - 2x)^{-3} \cdot (-2) = 8(1 - 2x)^{-3}. y′′′=8⋅(−3)(1−2x)−4⋅(−2)=48(1−2x)−4y''' = 8 \cdot (-3)(1 - 2x)^{-4} \cdot (-2) = 48(1 - 2x)^{-4}. Each differentiation brings down the exponent, −(k+1)-(k + 1), and the inside derivative −2-2: their product is 2(k+1)2(k + 1), positive. Hence y(n)=2n n! (1−2x)−(n+1)y^{(n)} = 2^n\, n!\,(1 - 2x)^{-(n+1)}, which matches 2,8,482, 8, 48. At x=0x = 0: y′′′(0)=48y'''(0) = 48 and y(5)(0)=25⋅120=3840y^{(5)}(0) = 2^5 \cdot 120 = 3840. Forgetting the inside factor gives n!n! with alternating signs, the pattern of 11+x\frac{1}{1 + x} instead.

Exercise 7: Tangent lines to a logistic curve

The figure shows the curve y=41+e−xy = \frac{4}{1 + e^{-x}}, which rises from 00 toward the dashed line y=4y = 4. It is a composite built in layers: a reciprocal, then 1+w1 + w, then w=e−xw = e^{-x}, itself an exponential of −x-x.

The tangent line at x=ax = a is y=y(a)+y′(a)(x−a)y = y(a) + y'(a)(x - a). On this curve, the derivative is easy to get and awkward to use until it is REWRITTEN with the laws of exponents; that rewriting is the heart of the exercise.

-6-5-4-3-2-112345612345y = 4/(1 + e⁻ˣ)y = 4x
  • a) Write yy as a power, differentiate, and show that y′=4e−x(1+e−x)2=4ex(ex+1)2y' = \frac{4e^{-x}}{(1 + e^{-x})^2} = \frac{4e^{x}}{(e^{x} + 1)^2}. Give y′(ln⁡2)y'(\ln 2).
  • b) Find the tangent line at x=0x = 0. Explain why the curve has no horizontal tangent.
  • c) Find the points of the curve where the tangent has slope 34\frac{3}{4}. (Hint: set w=e−xw = e^{-x}.)
  • d) Give the equations of the two tangent lines of c), with their yy-intercepts to four decimals.
  • e) Show that y(x)+y(−x)=4y(x) + y(-x) = 4 for all xx. Differentiate this identity with the chain rule to compare y′(−x)y'(-x) and y′(x)y'(x), and explain the two points of c).

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

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

Answers

  • a) y′=4e−x(1+e−x)2=4ex(ex+1)2y' = \frac{4e^{-x}}{(1 + e^{-x})^2} = \frac{4e^{x}}{(e^{x} + 1)^2}, y′(ln⁡2)=89y'(\ln 2) = \frac{8}{9}
  • b) y=2+xy = 2 + x; y′>0y' > 0 everywhere, so no horizontal tangent.
  • c) (ln⁡3,3)(\ln 3, 3) and (−ln⁡3,1)(-\ln 3, 1)
  • d) y=34x+3−34ln⁡3y = \frac{3}{4}x + 3 - \frac{3}{4}\ln 3 (intercept ≈2.1760\approx 2.1760) and y=34x+1+34ln⁡3y = \frac{3}{4}x + 1 + \frac{3}{4}\ln 3 (intercept ≈1.8240\approx 1.8240)
  • e) Differentiating gives y′(x)−y′(−x)=0y'(x) - y'(-x) = 0: y′(−x)=y′(x)y'(-x) = y'(x), so the points of c) are symmetric about (0,2)(0, 2).

a) y=4(1+e−x)−1y = 4(1 + e^{-x})^{-1}. Outside: 4u−14u^{-1}, derivative −4u−2-4u^{-2}; inside: u=1+e−xu = 1 + e^{-x}, and u′=e−x⋅(−1)=−e−xu' = e^{-x} \cdot (-1) = -e^{-x}, itself a chain rule with inside −x-x. So y′=−4(1+e−x)−2⋅(−e−x)=4e−x(1+e−x)2y' = -4(1 + e^{-x})^{-2} \cdot (-e^{-x}) = \frac{4e^{-x}}{(1 + e^{-x})^2}. Two minus signs cancel, and the result is positive, as the rising curve requires. Multiply top and bottom by e2xe^{2x}: 4e−xe2x=4ex4e^{-x}e^{2x} = 4e^{x} and (1+e−x)2e2x=(ex(1+e−x))2=(ex+1)2(1 + e^{-x})^2 e^{2x} = \left(e^{x}(1 + e^{-x})\right)^2 = (e^{x} + 1)^2, so y′=4ex(ex+1)2y' = \frac{4e^{x}}{(e^{x} + 1)^2}. At x=ln⁡2x = \ln 2: eln⁡2=2e^{\ln 2} = 2, so y′(ln⁡2)=89y'(\ln 2) = \frac{8}{9}.

b) At x=0x = 0: y(0)=42=2y(0) = \frac{4}{2} = 2 and y′(0)=44=1y'(0) = \frac{4}{4} = 1, so the tangent is y=2+xy = 2 + x. For every xx, ex>0e^{x} > 0 and (ex+1)2>0(e^{x} + 1)^2 > 0, so y′>0y' > 0: the numerator is never zero and the curve has no horizontal tangent. Its slope tends to 00 at both ends without ever reaching it.

c) Solve 4e−x(1+e−x)2=34\frac{4e^{-x}}{(1 + e^{-x})^2} = \frac{3}{4}. With w=e−x>0w = e^{-x} > 0: 16w=3(1+w)216w = 3(1 + w)^2, so 3w2−10w+3=03w^2 - 10w + 3 = 0, that is (3w−1)(w−3)=0(3w - 1)(w - 3) = 0. Then w=3w = 3 gives e−x=3e^{-x} = 3, x=−ln⁡3x = -\ln 3, and y=41+3=1y = \frac{4}{1 + 3} = 1; w=13w = \frac{1}{3} gives x=ln⁡3x = \ln 3 and y=41+1/3=3y = \frac{4}{1 + 1/3} = 3. The points are (ln⁡3,3)(\ln 3, 3) and (−ln⁡3,1)(-\ln 3, 1). The substitution turns an exponential equation into a quadratic, and both roots are kept because both are positive.

d) At (ln⁡3,3)(\ln 3, 3): y=3+34(x−ln⁡3)=34x+3−34ln⁡3y = 3 + \frac{3}{4}(x - \ln 3) = \frac{3}{4}x + 3 - \frac{3}{4}\ln 3, intercept ≈2.1760\approx 2.1760. At (−ln⁡3,1)(-\ln 3, 1): y=1+34(x+ln⁡3)=34x+1+34ln⁡3y = 1 + \frac{3}{4}(x + \ln 3) = \frac{3}{4}x + 1 + \frac{3}{4}\ln 3, intercept ≈1.8240\approx 1.8240. The two lines are parallel, and their intercepts add up to 44: their midpoint is on the line y=2y = 2, which part e) explains. The solution figure shows the three tangents.

e) y(−x)=41+exy(-x) = \frac{4}{1 + e^{x}} and y(x)=41+e−x=4exex+1y(x) = \frac{4}{1 + e^{-x}} = \frac{4e^{x}}{e^{x} + 1}, so y(x)+y(−x)=4ex+4ex+1=4y(x) + y(-x) = \frac{4e^{x} + 4}{e^{x} + 1} = 4. Differentiate both sides. The term y(−x)y(-x) is a composite with inside function −x-x, of derivative −1-1: ddxy(−x)=−y′(−x)\frac{d}{dx}y(-x) = -y'(-x). So y′(x)−y′(−x)=0y'(x) - y'(-x) = 0, that is y′(−x)=y′(x)y'(-x) = y'(x). The curve is symmetric about the point (0,2)(0, 2), and symmetric points carry equal slopes: x=ln⁡3x = \ln 3 and x=−ln⁡3x = -\ln 3 both have slope 34\frac{3}{4}, with heights 33 and 11 symmetric about 22. Forgetting the factor −1-1 from the inside function would give y′(−x)=−y′(x)y'(-x) = -y'(x), impossible for a curve whose slope is always positive.

-6-5-4-3-2-112345612345y = 4/(1 + e⁻ˣ)y = 4x

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 with numbers, and write the correct statement. Three of the five errors are errors of algebra, not of calculus.

  • a) ddx(3x−1)−1/2=−32(3x−1)1/2\frac{d}{dx}(3x - 1)^{-1/2} = -\frac{3}{2}(3x - 1)^{1/2}.
  • b) The derivative of a square is the square of the derivative, so ddxcos⁡2x=(−sin⁡x)2=sin⁡2x\frac{d}{dx}\cos^2 x = (-\sin x)^2 = \sin^2 x.
  • c) ddx(ex)3=3e2x\frac{d}{dx}\left(e^{x}\right)^3 = 3e^{2x}.
  • d) The derivative of x32x+1x^3\sqrt{2x + 1} simplifies to x22x+1 (x+3)x^2\sqrt{2x + 1}\,(x + 3).
  • e) If h=f∘gh = f \circ g and h′(a)=0h'(a) = 0, then g′(a)=0g'(a) = 0.

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

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

Answers

  • a) False: the exponent is −32-\frac{3}{2}; the derivative is −32(3x−1)−3/2-\frac{3}{2}(3x - 1)^{-3/2}, ≈−0.5303\approx -0.5303 at x=1x = 1.
  • b) False: ddxcos⁡2x=−2sin⁡xcos⁡x\frac{d}{dx}\cos^2 x = -2\sin x\cos x; at x=π4x = \frac{\pi}{4}, −1-1 and not 12\frac{1}{2}.
  • c) False: (ex)3=e3x(e^x)^3 = e^{3x}, derivative 3e3x3e^{3x}; at x=1x = 1, 3e33e^3 and not 3e23e^2.
  • d) False: the derivative is x2(7x+3)2x+1\frac{x^2(7x + 3)}{\sqrt{2x + 1}}; at x=4x = 4, 4963\frac{496}{3} and not 336336.
  • e) False: h′(a)=f′(g(a)) g′(a)h'(a) = f'(g(a))\,g'(a) can vanish through f′(g(a))f'(g(a)); f(u)=u2f(u) = u^2, g(x)=x−1g(x) = x - 1, a=1a = 1.

a) FALSE. The inside factor 33 is there, but the new exponent is wrong: −12−1=−32-\frac{1}{2} - 1 = -\frac{3}{2}, not +12+\frac{1}{2}. The student subtracted 11 in the wrong direction, a slip of arithmetic on fractions. Correct: ddx(3x−1)−1/2=−12(3x−1)−3/2⋅3=−32(3x−1)3/2\frac{d}{dx}(3x - 1)^{-1/2} = -\frac{1}{2}(3x - 1)^{-3/2} \cdot 3 = -\frac{3}{2(3x - 1)^{3/2}}. At x=1x = 1: −32⋅2−3/2≈−0.5303-\frac{3}{2} \cdot 2^{-3/2} \approx -0.5303, while the claim gives −322≈−2.1213-\frac{3}{2}\sqrt 2 \approx -2.1213. Check of size: the function 13x−1\frac{1}{\sqrt{3x - 1}} flattens as xx grows, so its slope must shrink toward 00, never grow like a square root.

b) FALSE. cos⁡2x=(cos⁡x)2\cos^2 x = (\cos x)^2 is a composite: the outside function is the square, the inside is u=cos⁡xu = \cos x. The outside derivative is 2u2u, evaluated at the inside left alone, then multiplied by u′=−sin⁡xu' = -\sin x: ddxcos⁡2x=2cos⁡x⋅(−sin⁡x)=−2sin⁡xcos⁡x=−sin⁡2x\frac{d}{dx}\cos^2 x = 2\cos x \cdot (-\sin x) = -2\sin x\cos x = -\sin 2x. Counterexample at x=π4x = \frac{\pi}{4}: the true slope is −2⋅22⋅22=−1-2 \cdot \frac{\sqrt 2}{2} \cdot \frac{\sqrt 2}{2} = -1, while the claim gives 12\frac{1}{2}, and even a POSITIVE slope, for a function that decreases on (0,π2)\left(0, \frac{\pi}{2}\right). Squaring the derivative of the inside is not a rule; the square is differentiated as a square, 2u2u, and the inside derivative multiplies.

c) FALSE. By the laws of exponents, (ex)3=e3x(e^x)^3 = e^{3x}, whose derivative is 3e3x3e^{3x} (inside function 3x3x). The student applied the power rule to the cube and stopped: 3(ex)2=3e2x3(e^x)^2 = 3e^{2x}, without the inside derivative exe^x. Done correctly, the Power Chain Rule gives 3(ex)2⋅ex=3e3x3(e^x)^2 \cdot e^x = 3e^{3x}, the same answer. At x=1x = 1: 3e3≈60.263e^3 \approx 60.26 against 3e2≈22.173e^2 \approx 22.17. At x=0x = 0 both give 33, which is why a check at 00 would not catch it.

d) FALSE. By the product and chain rules, ddxx3(2x+1)1/2=3x2(2x+1)1/2+x3⋅12(2x+1)−1/2⋅2=3x2(2x+1)1/2+x3(2x+1)−1/2\frac{d}{dx}x^3(2x + 1)^{1/2} = 3x^2(2x + 1)^{1/2} + x^3 \cdot \frac{1}{2}(2x + 1)^{-1/2} \cdot 2 = 3x^2(2x + 1)^{1/2} + x^3(2x + 1)^{-1/2}. The student factored out (2x+1)1/2(2x + 1)^{1/2}, the HIGHER power, and left x3x^3 in the second term, when it should have left x3(2x+1)−1x^3(2x + 1)^{-1}. The lowest power is −12-\frac{1}{2}: x2(2x+1)−1/2[3(2x+1)+x]=x2(7x+3)2x+1x^2(2x + 1)^{-1/2}\left[3(2x + 1) + x\right] = \frac{x^2(7x + 3)}{\sqrt{2x + 1}}. At x=4x = 4: the unfactored sum is 144+643=4963144 + \frac{64}{3} = \frac{496}{3}, the correct form gives 16⋅313=4963\frac{16 \cdot 31}{3} = \frac{496}{3}, and the claim gives 16⋅3⋅7=33616 \cdot 3 \cdot 7 = 336. Evaluating both forms at one convenient point is the cheapest check of any factorisation.

e) FALSE. By the chain rule, h′(a)=f′(g(a))⋅g′(a)h'(a) = f'(g(a)) \cdot g'(a), a product: it vanishes when EITHER factor does. Take f(u)=u2f(u) = u^2, g(x)=x−1g(x) = x - 1 and a=1a = 1: h(x)=(x−1)2h(x) = (x - 1)^2 and h′(1)=0h'(1) = 0, yet g′(1)=1g'(1) = 1. Here the zero comes from f′(g(1))=f′(0)=0f'(g(1)) = f'(0) = 0. Correct statement: for differentiable ff and gg, h′(a)=0h'(a) = 0 if and only if f′(g(a))=0f'(g(a)) = 0 or g′(a)=0g'(a) = 0. This is why the horizontal tangents of a composite come from the zeros of the outside factor AND of the inside factor.

Exercise 9: A radiosonde in cascade: degrees per kilometre, kilometres per minute

A weather balloon carrying a radiosonde is released at t=0t = 0 and rises at 55 m/s, so its altitude after tt minutes is h=0.3th = 0.3t kilometres (0≤t≤1000 \le t \le 100). The air temperature TT, in degrees Celsius, depends on the altitude as in the International Standard Atmosphere, drawn in the figure: T=15−6.5hT = 15 - 6.5h for 0≤h≤110 \le h \le 11, T=−56.5T = -56.5 for 11≤h≤2011 \le h \le 20, and T=−56.5+(h−20)T = -56.5 + (h - 20) for 20≤h≤3220 \le h \le 32. The air pressure is modelled by P=100e−h/7P = 100e^{-h/7} kilopascals.

The temperature the radiosonde reads depends on the altitude, which depends on the time: dTdt=dTdh⋅dhdt\frac{dT}{dt} = \frac{dT}{dh} \cdot \frac{dh}{dt}, each derivative read at the current value of its OWN variable, and the units cancel like fractions. The calculator is used for the final decimals only.

48121620242832-60-50-40-30-20-101020slope −6.5 °C/kmslope 0slope +1 °C/kmh (km)T (°C)
  • a) Give the units of dTdh\frac{dT}{dh}, dhdt\frac{dh}{dt} and dTdt\frac{dT}{dt}, then compute dTdt\frac{dT}{dt} at t=20t = 20 minutes, with a sentence of interpretation.
  • b) Compute dTdt\frac{dT}{dt} at t=50t = 50 and at t=80t = 80 minutes. What does the sign at t=80t = 80 say?
  • c) At what time does the balloon reach 1111 km? Does dTdt\frac{dT}{dt} exist at that instant? Give the one-sided rates.
  • d) Find dPdh\frac{dP}{dh}, then dPdt\frac{dP}{dt} at t=35t = 35 minutes, in kPa per minute to four decimals.
  • e) In a simplified gas model the balloon's volume is V=T+273PV = \frac{T + 273}{P} cubic metres (2.882.88 m³ at launch). Rewrite VV as a function of hh without a fraction in ee, find dVdh\frac{dV}{dh} in factored form, and dVdt\frac{dV}{dt} at t=20t = 20 minutes.

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

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

Answers

  • a) °C per km, km per min, °C per min; dTdt=−6.5⋅0.3=−1.95\frac{dT}{dt} = -6.5 \cdot 0.3 = -1.95 °C/min at t=20t = 20
  • b) 00 °C/min at t=50t = 50; +0.3+0.3 °C/min at t=80t = 80 (the air warms above 2020 km)
  • c) t=1103≈36.67t = \frac{110}{3} \approx 36.67 min; dTdt\frac{dT}{dt} does not exist: −1.95-1.95 on the left, 00 on the right
  • d) dPdh=−1007e−h/7\frac{dP}{dh} = -\frac{100}{7}e^{-h/7}; dPdt=−307e−1.5≈−0.9563\frac{dP}{dt} = -\frac{30}{7}e^{-1.5} \approx -0.9563 kPa/min
  • e) V=(T+273)eh/7100V = \frac{(T + 273)e^{h/7}}{100}, dVdh=eh/7100[T′(h)+T+2737]\frac{dV}{dh} = \frac{e^{h/7}}{100}\left[T'(h) + \frac{T + 273}{7}\right]; dVdt≈0.2055\frac{dV}{dt} \approx 0.2055 m³/min

a) dTdh\frac{dT}{dh} is a change of degrees per change of kilometres: °C per km. dhdt\frac{dh}{dt}: km per minute. Their product: °Ckm⋅kmmin=°Cmin\frac{\text{°C}}{\text{km}} \cdot \frac{\text{km}}{\text{min}} = \frac{\text{°C}}{\text{min}}, the unit of dTdt\frac{dT}{dt}. At t=20t = 20: h=6h = 6 km, in the first piece, where dTdh=−6.5\frac{dT}{dh} = -6.5; and dhdt=0.3\frac{dh}{dt} = 0.3. So dTdt=−6.5⋅0.3=−1.95\frac{dT}{dt} = -6.5 \cdot 0.3 = -1.95 °C/min: twenty minutes after launch, the radiosonde reads a temperature falling by about 22 degrees each minute. The slope of TT is read at h=6h = 6, the current altitude, not at 2020, which is a number of minutes.

b) At t=50t = 50: h=15h = 15 km, on the flat piece, so dTdh=0\frac{dT}{dh} = 0 and dTdt=0⋅0.3=0\frac{dT}{dt} = 0 \cdot 0.3 = 0 °C/min, whatever the speed of the balloon. At t=80t = 80: h=24h = 24 km, where dTdh=1\frac{dT}{dh} = 1, so dTdt=1⋅0.3=0.3\frac{dT}{dt} = 1 \cdot 0.3 = 0.3 °C/min. The positive sign says the radiosonde reads a WARMING air while still climbing: above 2020 km the ozone layer heats the stratosphere. A product of rates changes sign when one link does.

c) 0.3t=110.3t = 11 gives t=1103≈36.67t = \frac{110}{3} \approx 36.67 min. At h=11h = 11 the graph of TT has a corner: slope −6.5-6.5 on the left, 00 on the right, so TT is not differentiable at h(t)h(t), and the hypothesis of the chain rule fails. Look at each side: just before, dTdt=−6.5⋅0.3=−1.95\frac{dT}{dt} = -6.5 \cdot 0.3 = -1.95 °C/min; just after, 00. The one-sided rates differ, so dTdt\frac{dT}{dt} does not exist at that instant. Checking that the outside function is differentiable at the INSIDE value is part of applying the chain rule.

d) P=100euP = 100e^{u} with inside u=−h7u = -\frac{h}{7} and u′=−17u' = -\frac{1}{7}, so dPdh=−1007e−h/7\frac{dP}{dh} = -\frac{100}{7}e^{-h/7} kPa per km, which is also −P7-\frac{P}{7}. At t=35t = 35: h=10.5h = 10.5, e−10.5/7=e−1.5e^{-10.5/7} = e^{-1.5}, and dPdt=dPdh⋅dhdt=−1007e−1.5⋅0.3=−307e−1.5≈−0.9563\frac{dP}{dt} = \frac{dP}{dh} \cdot \frac{dh}{dt} = -\frac{100}{7}e^{-1.5} \cdot 0.3 = -\frac{30}{7}e^{-1.5} \approx -0.9563 kPa/min. Units: kPakm⋅kmmin=kPamin\frac{\text{kPa}}{\text{km}} \cdot \frac{\text{km}}{\text{min}} = \frac{\text{kPa}}{\text{min}}.

e) Dividing by P=100e−h/7P = 100e^{-h/7} is multiplying by eh/7100\frac{e^{h/7}}{100}, a law of exponents: V=1100(T+273)eh/7V = \frac{1}{100}(T + 273)e^{h/7}, a product with no fraction left. Product rule, chain rule on eh/7e^{h/7} (inside h7\frac{h}{7}, derivative 17\frac{1}{7}): dVdh=1100[T′(h)eh/7+(T+273)17eh/7]=eh/7100[T′(h)+T+2737]\frac{dV}{dh} = \frac{1}{100}\left[T'(h)e^{h/7} + (T + 273)\frac{1}{7}e^{h/7}\right] = \frac{e^{h/7}}{100}\left[T'(h) + \frac{T + 273}{7}\right], the common factor eh/7e^{h/7} taken out. At t=20t = 20: h=6h = 6, T=15−39=−24T = 15 - 39 = -24, T+273=249T + 273 = 249, T′(6)=−6.5T'(6) = -6.5, so the bracket is 2497−6.5=203.57≈29.07\frac{249}{7} - 6.5 = \frac{203.5}{7} \approx 29.07 and dVdh=e6/7100⋅203.57≈0.6850\frac{dV}{dh} = \frac{e^{6/7}}{100} \cdot \frac{203.5}{7} \approx 0.6850 m³ per km. Then dVdt=0.6850⋅0.3≈0.2055\frac{dV}{dt} = 0.6850 \cdot 0.3 \approx 0.2055 m³/min. The cooling alone would shrink the balloon; the bracket shows that the falling pressure wins, by 29.0729.07 against 6.56.5.

Exercise 10: Wind chill during a storm: a cascade with a real exponent

The wind chill index used by Environment and Climate Change Canada is W=13.12+0.6215T−11.37V0.16+0.3965 T V0.16W = 13.12 + 0.6215T - 11.37V^{0.16} + 0.3965\,T\,V^{0.16}, where TT is the air temperature in °C and VV the wind speed in km/h; WW is read like a temperature, in °C. On a day when the air stays at T=−20T = -20 °C, a storm sets in: tt hours after noon, the wind blows at V=20+6tV = 20 + 6t km/h (0≤t≤50 \le t \le 5).

The wind chill depends on the wind, which depends on the time. The exponent 0.160.16 is a real number: the Power Rule still applies, and the exponent of the derivative is 0.16−1=−0.840.16 - 1 = -0.84, a NEGATIVE exponent that moves VV to the denominator. The figure shows WW as a function of VV at T=−20T = -20 °C.

102030405060-40-35-30-25-20wind chill at T = −20 °CV (km/h)W (°C)
  • a) Show that at T=−20T = -20 °C, W=0.69−19.3V0.16W = 0.69 - 19.3V^{0.16}. Compute WW at V=32V = 32 km/h to two decimals.
  • b) Find dWdV\frac{dW}{dV} and its value at V=32V = 32 km/h, with its unit, to four decimals.
  • c) Find dWdt\frac{dW}{dt} at t=2t = 2 hours, in °C per hour to two decimals. A student computes dWdV\frac{dW}{dV} at V=2V = 2: what does that give, and why is it absurd?
  • d) Write WW directly as a function of tt, differentiate with the chain rule, and check c).
  • e) Compute dWdt\frac{dW}{dt} at t=0t = 0 and at t=5t = 5. The wind rises at a constant rate, yet the wind chill drops more and more slowly: which link of the chain explains it?

Type your answers, the page tells you right or wrong 0/7

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

Answers

  • a) 13.12−12.43=0.6913.12 - 12.43 = 0.69 and −11.37−7.93=−19.3-11.37 - 7.93 = -19.3; W(32)=0.69−19.3⋅20.8≈−32.91W(32) = 0.69 - 19.3 \cdot 2^{0.8} \approx -32.91 °C
  • b) dWdV=−3.088V−0.84=−3.088V0.84\frac{dW}{dV} = -3.088V^{-0.84} = -\frac{3.088}{V^{0.84}}; ≈−0.1680\approx -0.1680 °C per km/h at V=32V = 32
  • c) dWdt≈−0.1680⋅6≈−1.01\frac{dW}{dt} \approx -0.1680 \cdot 6 \approx -1.01 °C/h; dWdV\frac{dW}{dV} at V=2V = 2 reads the wind at a number of hours.
  • d) W(t)=0.69−19.3(20+6t)0.16W(t) = 0.69 - 19.3(20 + 6t)^{0.16}, W′(t)=−18.528(20+6t)−0.84W'(t) = -18.528(20 + 6t)^{-0.84}, W′(2)≈−1.01W'(2) \approx -1.01 °C/h
  • e) ≈−1.50\approx -1.50 °C/h at t=0t = 0, ≈−0.69\approx -0.69 °C/h at t=5t = 5; the link dWdV\frac{dW}{dV} shrinks as VV grows.

a) At T=−20T = -20: 0.6215⋅(−20)=−12.430.6215 \cdot (-20) = -12.43, so the constant part is 13.12−12.43=0.6913.12 - 12.43 = 0.69; and 0.3965⋅(−20)=−7.930.3965 \cdot (-20) = -7.93, so the coefficient of V0.16V^{0.16} is −11.37−7.93=−19.3-11.37 - 7.93 = -19.3. Hence W=0.69−19.3V0.16W = 0.69 - 19.3V^{0.16}. At V=32=25V = 32 = 2^5: 320.16=20.8≈1.741132^{0.16} = 2^{0.8} \approx 1.7411, so W≈0.69−33.60=−32.91W \approx 0.69 - 33.60 = -32.91 °C. A power of a power multiplies exponents, 5⋅0.16=0.85 \cdot 0.16 = 0.8: the calculator is not even needed until the last step.

b) Power Rule: dWdV=−19.3⋅0.16 V0.16−1=−3.088V−0.84=−3.088V0.84\frac{dW}{dV} = -19.3 \cdot 0.16\,V^{0.16 - 1} = -3.088V^{-0.84} = -\frac{3.088}{V^{0.84}}. The exponent 0.16−1=−0.840.16 - 1 = -0.84 is where students lose the mark, writing 1.161.16 or −1.16-1.16. At V=32V = 32: 32−0.84=2−4.2≈0.0544132^{-0.84} = 2^{-4.2} \approx 0.05441, so dWdV≈−0.1680\frac{dW}{dV} \approx -0.1680 °C per km/h: near 3232 km/h, each extra km/h of wind makes it feel about 0.170.17 degree colder.

c) Chain rule: dWdt=dWdV⋅dVdt\frac{dW}{dt} = \frac{dW}{dV} \cdot \frac{dV}{dt}. At t=2t = 2: V=32V = 32 km/h, dWdV≈−0.1680\frac{dW}{dV} \approx -0.1680, and dVdt=6\frac{dV}{dt} = 6 km/h per hour. So dWdt≈−1.008\frac{dW}{dt} \approx -1.008, about −1.01-1.01 °C per hour. Units: °Ckm/h⋅km/hh=°Ch\frac{\text{°C}}{\text{km/h}} \cdot \frac{\text{km/h}}{\text{h}} = \frac{\text{°C}}{\text{h}}. The student who evaluates dWdV\frac{dW}{dV} at V=2V = 2 gets −3.088⋅2−0.84≈−1.73-3.088 \cdot 2^{-0.84} \approx -1.73, then −10.35-10.35 °C/h after multiplying by 66: a wind chill falling ten degrees an hour. The 22 is a number of HOURS fed to a function whose input is a wind speed; the outside derivative must be read at the inside VALUE, V(2)=32V(2) = 32.

d) W(t)=0.69−19.3(20+6t)0.16W(t) = 0.69 - 19.3(20 + 6t)^{0.16}. With the inside u=20+6tu = 20 + 6t and u′=6u' = 6: W′(t)=−19.3⋅0.16(20+6t)−0.84⋅6=−18.528(20+6t)−0.84W'(t) = -19.3 \cdot 0.16(20 + 6t)^{-0.84} \cdot 6 = -18.528(20 + 6t)^{-0.84}. At t=2t = 2: −18.528⋅32−0.84≈−1.008-18.528 \cdot 32^{-0.84} \approx -1.008 °C/h, as in c). Composing first and differentiating once, or differentiating link by link: both roads must meet, and checking that they do is the cheapest verification there is.

e) At t=0t = 0: V=20V = 20, W′(0)=−18.528⋅20−0.84≈−1.50W'(0) = -18.528 \cdot 20^{-0.84} \approx -1.50 °C/h. At t=5t = 5: V=50V = 50, W′(5)=−18.528⋅50−0.84≈−0.69W'(5) = -18.528 \cdot 50^{-0.84} \approx -0.69 °C/h. The link dVdt=6\frac{dV}{dt} = 6 is constant; the link dWdV=−3.088V0.84\frac{dW}{dV} = -\frac{3.088}{V^{0.84}} shrinks in size as VV grows, because of the negative exponent. The figure shows it: the curve is steep at low wind and flattens. The first few km/h of wind cost the most degrees, which is why a light breeze on a cold day feels so much worse than calm air.

102030405060-40-35-30-25-20wind chill at T = −20 °Cslope −0.168 at V = 32V (km/h)W (°C)

© Ahmed Squalli Houssaini. Exercise set published at www.letuteurscientifique.ca/en/exercices/math203-chain-rule. Free for personal and classroom use; republishing it elsewhere requires written permission (legal notice).

See also

Struggling with MATH 203?

I tutor first-year calculus at Concordia and McGill, in English or in French, in Montreal or online, with the algebra behind every rule. Get in touch for a first session.

Site by Studio Squalli