MATH 203 Calculus I • Concordia University, Montreal

Corrected exercises: concavity and curve sketching (MATH 203)

This is the corrected exercise set for concavity and curve sketching in MATH 203, Differential and Integral Calculus I, at Concordia University, section 4.4 of Thomas' Calculus. The second derivative gives the concavity, the points of inflection and a second test for local extrema, and the chapter ends with the complete sketch: domain, intercepts, symmetry, asymptotes, increase, concavity, then one table and the curve. A scientific calculator is allowed, so decimals appear where a context asks for them, but every derivative is computed and FACTORED by hand.

The thread running through the whole set: the concavity is read on the sign of f′′f'', and that sign can only be read on a factored form. In MATH 203 the marks are lost between the rule and the sign table, in the algebra: a quotient rule applied twice and then expanded instead of pulling out the common factor x2+cx^2 + c and cancelling one power of it; a fractional power differentiated without factoring out the SMALLEST power; an exponential left inside a sum instead of pulled out as a factor that is never zero. Each solution names the gesture where it pays.

The traps named in the solutions: a zero of f′′f'' taken for a point of inflection, a point where f′′f'' does not exist forgotten as a candidate, a zero of f′′f'' outside the domain, a factor xx divided out and lost, the silent Second Derivative Test read as no extremum, a cusp taken for a point of inflection, concave up confused with increasing, the tallest curve taken for ff, a calculator left in degree mode, and a table of values trusted to locate an inflection point exactly.

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

  • • Concavity Test: f′′>0f'' > 0 on II gives concave up on II, f′′<0f'' < 0 gives concave down. Concave up means f′f' increasing, NOT ff increasing.
  • • Point of inflection: a point (c,f(c))(c, f(c)) of the graph, with a tangent line (possibly vertical), where the concavity changes.
  • • Candidates: f′′(c)=0f''(c) = 0 or f′′(c)f''(c) undefined, with cc in the domain. Keep a candidate only if f′′f'' changes sign.
  • • Second Derivative Test: f′(c)=0f'(c) = 0 and f′′(c)<0f''(c) < 0 give a local max, f′′(c)>0f''(c) > 0 a local min; f′′(c)=0f''(c) = 0 gives NO information.
  • • Algebra: (uvn)′\left(\frac{u}{v^n}\right)' has the factor vn−1v^{n-1} in its numerator, cancel one power; x1/3=x−5/3x2x^{1/3} = x^{-5/3}x^2, factor out the smallest power; e−x>0e^{-x} > 0 never changes a sign.
  • • Curve sketching: domain, intercepts, symmetry, asymptotes, f′f' factored, f′′f'' factored, one table, then the curve.

Part A: the basics (/50)

Exercise 1: Inflection points: a zero of the second derivative is neither necessary nor sufficient

Concavity Test (Thomas 4.4): if f′′>0f'' > 0 on an interval II, the graph of ff is concave up on II; if f′′<0f'' < 0, concave down. A point (c,f(c))(c, f(c)) where the graph has a tangent line (possibly vertical) and where the concavity CHANGES is a point of inflection.

The candidates are therefore the numbers where f′′(c)=0f''(c) = 0 AND the numbers of the domain where f′′(c)f''(c) does not exist; a candidate is kept only if the sign of f′′f'' changes there. The figure shows the graph of the function hh of part c).

-3-2-1123-8-6-4-22468y = h(x)
  • a) Let f(x)=x4+2x3−12x2+3f(x) = x^4 + 2x^3 - 12x^2 + 3. Find the intervals on which ff is concave up or concave down, and the points of inflection.
  • b) Let g(x)=(x−2)4+3xg(x) = (x - 2)^4 + 3x. Show that g′′(2)=0g''(2) = 0, then decide whether (2,g(2))(2, g(2)) is a point of inflection. Do not expand (x−2)4(x - 2)^4.
  • c) Let h(x)=x7/3−7x1/3h(x) = x^{7/3} - 7x^{1/3}. Show that h′′(x)=14(2x2+1)9x5/3h''(x) = \frac{14(2x^2 + 1)}{9x^{5/3}} by factoring out the smallest power of xx, and check that the equation h′′(x)=0h''(x) = 0 has no solution. Does hh have a point of inflection? Describe the tangent line there.
  • d) A student's rule reads: the points of inflection are exactly the solutions of f′′(x)=0f''(x) = 0. Use b) and c) to show that the rule fails in BOTH directions.

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

a)
Concave down on ,
b)
c)
Show the solution

Answers

  • a) Concave up on (−∞,−2)(-\infty, -2) and (1,∞)(1, \infty), concave down on (−2,1)(-2, 1); inflection points (−2,−45)(-2, -45) and (1,−6)(1, -6).
  • b) g′′(x)=12(x−2)2≥0g''(x) = 12(x - 2)^2 \ge 0: no sign change, so no inflection point; gg is concave up on R\mathbb{R}.
  • c) h′′h'' never vanishes, but changes sign at 00 where h′′h'' is undefined: inflection point (0,0)(0, 0) with a VERTICAL tangent.
  • d) g′′(2)=0g''(2) = 0 without inflection (not sufficient); hh has an inflection where h′′h'' does not exist (not necessary).

a) f′(x)=4x3+6x2−24xf'(x) = 4x^3 + 6x^2 - 24x and f′′(x)=12x2+12x−24=12(x2+x−2)=12(x+2)(x−1)f''(x) = 12x^2 + 12x - 24 = 12(x^2 + x - 2) = 12(x + 2)(x - 1). Factor BEFORE looking at signs: a product of two linear factors has the sign ++ outside its roots and −- between them. So f′′>0f'' > 0 on (−∞,−2)(-\infty, -2) and (1,∞)(1, \infty), where ff is concave up, and f′′<0f'' < 0 on (−2,1)(-2, 1), where it is concave down. The sign changes at −2-2 and at 11, both in the domain, and a polynomial has a tangent line everywhere: two points of inflection, f(−2)=16−16−48+3=−45f(-2) = 16 - 16 - 48 + 3 = -45 and f(1)=1+2−12+3=−6f(1) = 1 + 2 - 12 + 3 = -6. A point of inflection is a POINT: writing only x=−2x = -2 and x=1x = 1 loses half a mark on most MATH 203 finals.

b) Chain rule with inner function x−2x - 2: g′(x)=4(x−2)3+3g'(x) = 4(x - 2)^3 + 3 and g′′(x)=12(x−2)2g''(x) = 12(x - 2)^2, so g′′(2)=0g''(2) = 0. But (x−2)2≥0(x - 2)^2 \ge 0 on both sides of 22: g′′g'' does NOT change sign, and gg is concave up on (−∞,2)(-\infty, 2) and on (2,∞)(2, \infty), hence on all of R\mathbb{R}. The candidate is rejected: no inflection point. This is why the question says not to expand: (x−2)4=x4−8x3+24x2−32x+16(x - 2)^4 = x^4 - 8x^3 + 24x^2 - 32x + 16 gives g′′(x)=12x2−48x+48g''(x) = 12x^2 - 48x + 48, and the student who expands must now FACTOR it back into 12(x−2)212(x - 2)^2 to read its sign. The factored form was there from the start; expanding it costs time on an exam and often a sign.

c) Differentiate term by term with the power rule: h′(x)=73x4/3−73x−2/3h'(x) = \frac{7}{3}x^{4/3} - \frac{7}{3}x^{-2/3} and h′′(x)=289x1/3+149x−5/3h''(x) = \frac{28}{9}x^{1/3} + \frac{14}{9}x^{-5/3}. Now the algebraic gesture of the chapter: factor out the SMALLEST power, x−5/3x^{-5/3}. Since x1/3=x−5/3⋅x2x^{1/3} = x^{-5/3} \cdot x^{2}, h′′(x)=149x−5/3(2x2+1)=14(2x2+1)9x5/3h''(x) = \frac{14}{9}x^{-5/3}(2x^2 + 1) = \frac{14(2x^2 + 1)}{9x^{5/3}}. The factor 2x2+1≥12x^2 + 1 \ge 1 never vanishes, so h′′(x)=0h''(x) = 0 has NO solution. Yet x5/3=(x3)5x^{5/3} = \left(\sqrt[3]{x}\right)^5 has the sign of xx: h′′<0h'' < 0 for x<0x < 0 (concave down) and h′′>0h'' > 0 for x>0x > 0 (concave up). The candidate is x=0x = 0, where h′′h'' does not exist but h(0)=0h(0) = 0 does. The tangent line: h′(x)=7(x2−1)3x2/3h'(x) = \frac{7(x^2 - 1)}{3x^{2/3}}, and as x→0x \to 0 the numerator tends to −7-7 while 3x2/3→0+3x^{2/3} \to 0^+ from both sides, so h′(x)→−∞h'(x) \to -\infty: the graph has a VERTICAL tangent at the origin, which the figure shows. With a tangent line and a change of concavity, (0,0)(0, 0) is a point of inflection. Note that h′′h'' written as the SUM 289x1/3+149x−5/3\frac{28}{9}x^{1/3} + \frac{14}{9}x^{-5/3} hides everything: its sign is unreadable until the smallest power is out.

d) Not sufficient: in b), g′′(2)=0g''(2) = 0 and (2,g(2))(2, g(2)) is not a point of inflection, because the sign of g′′g'' does not change. Not necessary: in c), h′′h'' never vanishes and (0,0)(0, 0) is a point of inflection, because the candidate there is a point where h′′h'' does not exist. The correct rule: list the zeros of f′′f'' AND the points of the domain where f′′f'' is undefined, then keep a candidate only if f′′f'' changes sign there and the graph has a tangent line.

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: The Second Derivative Test, and what to do when it is silent

Second Derivative Test for Local Extrema (Thomas 4.4): suppose f′′f'' is continuous on an open interval containing cc and f′(c)=0f'(c) = 0. If f′′(c)<0f''(c) < 0, ff has a local maximum at cc; if f′′(c)>0f''(c) > 0, a local minimum. If f′′(c)=0f''(c) = 0, the test FAILS: ff may have a local maximum, a local minimum, or neither, and the First Derivative Test must decide.

The figure shows the graph of g(x)=x+2cos⁡xg(x) = x + 2\cos x on [0,2π][0, 2\pi], used in part b). A scientific calculator is allowed; give decimal values to four decimal places.

1234567123456789y = x + 2 cos x
  • a) Let f(x)=x3−6x2−15x+4f(x) = x^3 - 6x^2 - 15x + 4. Find the critical numbers of ff and classify them with the Second Derivative Test. Give the local extreme values.
  • b) Let g(x)=x+2cos⁡xg(x) = x + 2\cos x on (0,2π)(0, 2\pi). Find the critical numbers and classify them with the Second Derivative Test. Give the exact local extreme values, then their decimal values.
  • c) Let p(x)=x4+4x3p(x) = x^4 + 4x^3. Show that pp has two critical numbers. At which one is the Second Derivative Test silent? Conclude anyway, and say what the graph does at that point.
  • d) Let k(x)=x4−4x3+6x2−4x+3k(x) = x^4 - 4x^3 + 6x^2 - 4x + 3. Show that k′(1)=k′′(1)=0k'(1) = k''(1) = 0. Recognize a binomial in k(x)k(x), and conclude about x=1x = 1 without any sign table.

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

a)
b)
c)
d)
Show the solution

Answers

  • a) Critical numbers −1-1 and 55; local max f(−1)=12f(-1) = 12 (f′′(−1)=−18f''(-1) = -18), local min f(5)=−96f(5) = -96 (f′′(5)=18f''(5) = 18).
  • b) Local max g(π6)=π6+3≈2.2556g\left(\frac{\pi}{6}\right) = \frac{\pi}{6} + \sqrt 3 \approx 2.2556; local min g(5π6)=5π6−3≈0.8859g\left(\frac{5\pi}{6}\right) = \frac{5\pi}{6} - \sqrt 3 \approx 0.8859.
  • c) Local min p(−3)=−27p(-3) = -27; the test is silent at 00, where pp has NO extremum: (0,0)(0, 0) is an inflection point with a horizontal tangent.
  • d) k(x)=(x−1)4+2k(x) = (x - 1)^4 + 2: absolute minimum k(1)=2k(1) = 2, although the test is silent.

a) f′(x)=3x2−12x−15=3(x2−4x−5)=3(x+1)(x−5)f'(x) = 3x^2 - 12x - 15 = 3(x^2 - 4x - 5) = 3(x + 1)(x - 5), so the critical numbers are −1-1 and 55. f′′(x)=6x−12f''(x) = 6x - 12. At −1-1: f′′(−1)=−18<0f''(-1) = -18 < 0, local maximum f(−1)=−1−6+15+4=12f(-1) = -1 - 6 + 15 + 4 = 12. At 55: f′′(5)=18>0f''(5) = 18 > 0, local minimum f(5)=125−150−75+4=−96f(5) = 125 - 150 - 75 + 4 = -96. The test is faster than a sign table here because f′′f'' is a one-line computation and never vanishes at a critical number. The classic slip is the sign of f′′f'' itself: f′′(−1)f''(-1) NEGATIVE means concave down, the shape of a cap, and a cap on a horizontal tangent is a maximum.

b) g′(x)=1−2sin⁡x=0  ⟺  sin⁡x=12g'(x) = 1 - 2\sin x = 0 \iff \sin x = \frac{1}{2}, which on (0,2π)(0, 2\pi) gives x=π6x = \frac{\pi}{6} and x=5π6x = \frac{5\pi}{6} (the second one comes from sin⁡(π−θ)=sin⁡θ\sin(\pi - \theta) = \sin\theta, and forgetting it is the usual loss). g′′(x)=−2cos⁡xg''(x) = -2\cos x. At π6\frac{\pi}{6}: g′′=−2⋅32=−3<0g'' = -2 \cdot \frac{\sqrt 3}{2} = -\sqrt 3 < 0, local maximum g(π6)=π6+2⋅32=π6+3≈2.2556g\left(\frac{\pi}{6}\right) = \frac{\pi}{6} + 2 \cdot \frac{\sqrt 3}{2} = \frac{\pi}{6} + \sqrt 3 \approx 2.2556. At 5π6\frac{5\pi}{6}: g′′=−2⋅(−32)=3>0g'' = -2 \cdot \left(-\frac{\sqrt 3}{2}\right) = \sqrt 3 > 0, local minimum g(5π6)=5π6−3≈0.8859g\left(\frac{5\pi}{6}\right) = \frac{5\pi}{6} - \sqrt 3 \approx 0.8859. Calculator in RADIAN mode: in degree mode, cos⁡π6\cos\frac{\pi}{6} is computed as the cosine of 0.520.52 degrees and every decimal is wrong. The figure agrees: a small bump then a small dip, and the line y=xy = x wins after that.

c) p′(x)=4x3+12x2=4x2(x+3)p'(x) = 4x^3 + 12x^2 = 4x^2(x + 3): factor out the common 4x24x^2 instead of dividing by xx, which would lose the root 00. Critical numbers 00 and −3-3. p′′(x)=12x2+24x=12x(x+2)p''(x) = 12x^2 + 24x = 12x(x + 2). At −3-3: p′′(−3)=36>0p''(-3) = 36 > 0, local minimum p(−3)=81−108=−27p(-3) = 81 - 108 = -27. At 00: p′′(0)=0p''(0) = 0, the test is SILENT. It does not say no extremum; it says nothing. The First Derivative Test decides: 4x2≥04x^2 \ge 0, so near 00 the sign of p′p' is the sign of x+3x + 3, positive on both sides of 00. No sign change, no extremum. Moreover p′′=12x(x+2)p'' = 12x(x + 2) changes sign at 00 (from −- to ++): the origin is a point of inflection where the tangent line is HORIZONTAL, a flat step in the graph.

d) k′(x)=4x3−12x2+12x−4k'(x) = 4x^3 - 12x^2 + 12x - 4, so k′(1)=4−12+12−4=0k'(1) = 4 - 12 + 12 - 4 = 0, and k′′(x)=12x2−24x+12k''(x) = 12x^2 - 24x + 12, so k′′(1)=0k''(1) = 0: the test is silent again. Now look at the coefficients 1,−4,6,−41, -4, 6, -4: they are those of (x−1)4=x4−4x3+6x2−4x+1(x - 1)^4 = x^4 - 4x^3 + 6x^2 - 4x + 1. So k(x)=(x−1)4+2k(x) = (x - 1)^4 + 2. Since (x−1)4≥0(x - 1)^4 \ge 0 with equality only at x=1x = 1, k(x)≥2=k(1)k(x) \ge 2 = k(1) for every xx: kk has an ABSOLUTE minimum 22 at x=1x = 1. Compare with c): at both points f′=f′′=0f' = f'' = 0, and the conclusions are opposite (no extremum in c), a minimum in d)). That is exactly what the word fails means. The algebra did the work: recognizing the binomial replaced a whole sign table, and k′(x)=4(x−1)3k'(x) = 4(x - 1)^3 is then read at a glance.

Exercise 3: The quotient rule twice: pull out the common factor before you expand

Sketch the graph of f(x)=x2−9x2+27f(x) = \frac{x^2 - 9}{x^2 + 27} by the procedure of Thomas 4.4: domain, intercepts, symmetry, asymptotes, f′f' and its sign, f′′f'' and its sign, then one table and the curve.

The second derivative of a quotient is where most MATH 203 copies go wrong: the numerator given by the quotient rule contains a factor of the denominator, and it must be pulled out and cancelled BEFORE anything is expanded.

  • a) Find the domain, the intercepts and the symmetry of ff. Show that f(x)=1−36x2+27f(x) = 1 - \frac{36}{x^2 + 27} and deduce the horizontal asymptote and the position of the graph relative to it.
  • b) Show that f′(x)=72x(x2+27)2f'(x) = \frac{72x}{(x^2 + 27)^2}. Find the intervals of increase and decrease and the extreme value of ff.
  • c) Apply the quotient rule to f′f'. Show that the numerator has the factor x2+27x^2 + 27, cancel it, and obtain f′′(x)=216(9−x2)(x2+27)3f''(x) = \frac{216(9 - x^2)}{(x^2 + 27)^3}. Compute f′′(0)f''(0).
  • d) Study the concavity and give the points of inflection. What is the slope of the graph at each of them?
  • e) Sketch the graph and give the range of ff.

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

a)
b)
Decreasing on ,
c)
d)
Concave up on ,
e)
Range ,
Show the solution

Answers

  • a) Domain R\mathbb{R}; intercepts (±3,0)(\pm 3, 0) and (0,−13)\left(0, -\frac{1}{3}\right); ff is even; asymptote y=1y = 1 at both ends, graph always BELOW it.
  • b) Decreasing on (−∞,0](-\infty, 0], increasing on [0,∞)[0, \infty); absolute minimum f(0)=−13f(0) = -\frac{1}{3}.
  • c) f′′(x)=216(9−x2)(x2+27)3f''(x) = \frac{216(9 - x^2)}{(x^2 + 27)^3}, f′′(0)=881f''(0) = \frac{8}{81}.
  • d) Concave up on (−3,3)(-3, 3), down on (−∞,−3)(-\infty, -3) and (3,∞)(3, \infty); inflection points (−3,0)(-3, 0) and (3,0)(3, 0), slopes −16-\frac{1}{6} and 16\frac{1}{6}.
  • e) Range [−13,1)\left[-\frac{1}{3}, 1\right).

a) x2+27≥27>0x^2 + 27 \ge 27 > 0, so the domain is R\mathbb{R} and there is no vertical asymptote. f(0)=−927=−13f(0) = -\frac{9}{27} = -\frac{1}{3}; f(x)=0  ⟺  x2=9  ⟺  x=±3f(x) = 0 \iff x^2 = 9 \iff x = \pm 3. Only x2x^2 appears, so f(−x)=f(x)f(-x) = f(x): ff is even. Add and subtract 3636 in the numerator: x2−9x2+27=(x2+27)−36x2+27=1−36x2+27\frac{x^2 - 9}{x^2 + 27} = \frac{(x^2 + 27) - 36}{x^2 + 27} = 1 - \frac{36}{x^2 + 27}. As x→±∞x \to \pm\infty, 36x2+27→0\frac{36}{x^2 + 27} \to 0, so y=1y = 1 is the horizontal asymptote at both ends; and since 36x2+27>0\frac{36}{x^2 + 27} > 0, f(x)<1f(x) < 1 everywhere: the graph stays BELOW its asymptote and never reaches it.

b) Quotient rule: f′(x)=2x(x2+27)−(x2−9)(2x)(x2+27)2f'(x) = \frac{2x(x^2 + 27) - (x^2 - 9)(2x)}{(x^2 + 27)^2}. Factor the common 2x2x out of the numerator before expanding anything: 2x[(x2+27)−(x2−9)]=2x⋅36=72x2x\left[(x^2 + 27) - (x^2 - 9)\right] = 2x \cdot 36 = 72x. So f′(x)=72x(x2+27)2f'(x) = \frac{72x}{(x^2 + 27)^2}, which has the sign of xx (the denominator is a positive square). ff decreases on (−∞,0](-\infty, 0], increases on [0,∞)[0, \infty), and since it decreases then increases on the WHOLE line, f(0)=−13f(0) = -\frac{1}{3} is the absolute minimum. The form 1−36x2+271 - \frac{36}{x^2 + 27} gives the same derivative in one line by the chain rule: ddx[−36(x2+27)−1]=36(x2+27)−2(2x)\frac{d}{dx}\left[-36(x^2 + 27)^{-1}\right] = 36(x^2 + 27)^{-2}(2x).

c) Quotient rule on 72x(x2+27)2\frac{72x}{(x^2 + 27)^2}, the derivative of the denominator by the chain rule being 2(x2+27)(2x)2(x^2 + 27)(2x): f′′(x)=72(x2+27)2−72x⋅2(x2+27)(2x)(x2+27)4f''(x) = \frac{72(x^2 + 27)^2 - 72x \cdot 2(x^2 + 27)(2x)}{(x^2 + 27)^4}. Both terms of the numerator contain x2+27x^2 + 27: pull it out, 72(x2+27)[(x2+27)−4x2]=72(x2+27)(27−3x2)72(x^2 + 27)\left[(x^2 + 27) - 4x^2\right] = 72(x^2 + 27)(27 - 3x^2), and cancel ONE power against the denominator: f′′(x)=72(27−3x2)(x2+27)3=216(9−x2)(x2+27)3f''(x) = \frac{72(27 - 3x^2)}{(x^2 + 27)^3} = \frac{216(9 - x^2)}{(x^2 + 27)^3}. The expanding route gives the numerator −216x4−3888x2+52 488-216x^4 - 3888x^2 + 52\,488 over (x2+27)4(x^2 + 27)^4: correct, but a degree 4 polynomial with large coefficients whose sign must now be found by factoring it back, which is where the marks go. f′′(0)=216×9273=194419 683=881f''(0) = \frac{216 \times 9}{27^3} = \frac{1944}{19\,683} = \frac{8}{81}.

d) The denominator (x2+27)3(x^2 + 27)^3 is positive, so f′′f'' has the sign of 9−x2=(3−x)(3+x)9 - x^2 = (3 - x)(3 + x): positive on (−3,3)(-3, 3), where ff is concave up, negative for ∣x∣>3|x| > 3, where ff is concave down. The sign changes at ±3\pm 3, which are in the domain: two points of inflection, (−3,0)(-3, 0) and (3,0)(3, 0). They are also the xx-intercepts: the graph crosses the axis exactly where it changes its bending. The slopes there: f′(3)=216362=2161296=16f'(3) = \frac{216}{36^2} = \frac{216}{1296} = \frac{1}{6} and, by symmetry, f′(−3)=−16f'(-3) = -\frac{1}{6}. Consistency with the asymptote: an increasing curve that levels off below y=1y = 1 must end concave DOWN, which is what f′′<0f'' < 0 for x>3x > 3 says.

e) The table from left to right: coming from just below y=1y = 1 on the far left, decreasing and concave down; inflection at (−3,0)(-3, 0); concave up down to the minimum (0,−13)\left(0, -\frac{1}{3}\right); up again, inflection at (3,0)(3, 0); then concave down toward y=1y = 1 from below. Range: the minimum −13-\frac{1}{3} is reached, the value 11 never is, and every value between is taken by continuity: [−13,1)\left[-\frac{1}{3}, 1\right).

-12-10-8-6-4-224681012-1-0.50.511.5y = 1min (0, −1/3)(3, 0)(−3, 0)

Exercise 4: A square root: factor out the smallest power, and reject a zero outside the domain

Sketch the graph of f(x)=(x−2)x+1f(x) = (x - 2)\sqrt{x + 1}. A square root brings a one-sided domain, an endpoint where the derivative may blow up, and derivatives made of powers 12\frac{1}{2}, −12-\frac{1}{2}, −32-\frac{3}{2} whose sign is unreadable until they are FACTORED.

A scientific calculator is allowed, but every value asked for here is exact.

  • a) Find the domain, the intercepts and the sign of ff, and lim⁡x→∞f(x)\lim_{x\to\infty} f(x). Does the graph have an asymptote?
  • b) Show that f′(x)=3x2x+1f'(x) = \frac{3x}{2\sqrt{x + 1}} by writing f′f' over the common denominator 2x+12\sqrt{x + 1}. Find the critical number and describe the tangent line at the endpoint (−1,0)(-1, 0).
  • c) Write f′(x)=32x(x+1)−1/2f'(x) = \frac{3}{2}x(x + 1)^{-1/2}, differentiate, factor out the smallest power (x+1)−3/2(x + 1)^{-3/2} and show that f′′(x)=3(x+2)4(x+1)3/2f''(x) = \frac{3(x + 2)}{4(x + 1)^{3/2}}. The equation f′′(x)=0f''(x) = 0 gives x=−2x = -2: is there a point of inflection? Study the concavity.
  • d) Classify the critical number with the Second Derivative Test, compute f′′(0)f''(0), and explain why the extremum is absolute. Give the range of ff.
  • e) Compute the slope of the graph where it crosses the positive xx-axis, then sketch the graph.

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

a)
b)
c)
Concave up on ,
d)
Range ,
e)
Show the solution

Answers

  • a) Domain [−1,∞)[-1, \infty); intercepts (−1,0)(-1, 0), (2,0)(2, 0), (0,−2)(0, -2); f<0f < 0 on (−1,2)(-1, 2), f>0f > 0 for x>2x > 2; f→∞f \to \infty, no asymptote.
  • b) Critical number 00; vertical tangent at (−1,0)(-1, 0) since f′(x)→−∞f'(x) \to -\infty as x→−1+x \to -1^+.
  • c) x=−2x = -2 is NOT in the domain: no inflection point; ff is concave up on (−1,∞)(-1, \infty).
  • d) f′′(0)=32>0f''(0) = \frac{3}{2} > 0: minimum f(0)=−2f(0) = -2, absolute; range [−2,∞)[-2, \infty).
  • e) f′(2)=3f'(2) = \sqrt 3.

a) x+1\sqrt{x + 1} requires x+1≥0x + 1 \ge 0: the domain is [−1,∞)[-1, \infty). f(x)=0  ⟺  x=2f(x) = 0 \iff x = 2 or x=−1x = -1, and f(0)=−2⋅1=−2f(0) = -2 \cdot 1 = -2. Since x+1≥0\sqrt{x + 1} \ge 0, ff has the sign of x−2x - 2: negative on (−1,2)(-1, 2), positive after 22. As x→∞x \to \infty both factors tend to ∞\infty, so f→∞f \to \infty. No vertical asymptote (ff is continuous on its closed domain), no horizontal one, and no slant one either, because f(x)x=(1−2x)x+1→∞\frac{f(x)}{x} = \left(1 - \frac{2}{x}\right)\sqrt{x + 1} \to \infty. There is no symmetry to look for: the domain itself is not symmetric about 00.

b) Product rule, then chain rule on the root (inner function x+1x + 1): f′(x)=x+1+x−22x+1f'(x) = \sqrt{x + 1} + \frac{x - 2}{2\sqrt{x + 1}}. As a SUM, its sign is unreadable. Over the common denominator: x+1=2(x+1)2x+1\sqrt{x + 1} = \frac{2(x + 1)}{2\sqrt{x + 1}}, so f′(x)=2(x+1)+(x−2)2x+1=3x2x+1f'(x) = \frac{2(x + 1) + (x - 2)}{2\sqrt{x + 1}} = \frac{3x}{2\sqrt{x + 1}} for x>−1x > -1. The denominator is positive: f′f' has the sign of xx. Critical number 00 (f′(0)=0f'(0) = 0). At the endpoint, as x→−1+x \to -1^+ the numerator tends to −3-3 and the denominator to 0+0^+: f′(x)→−∞f'(x) \to -\infty. The graph leaves (−1,0)(-1, 0) with a VERTICAL tangent, going down.

c) Product rule on 32x(x+1)−1/2\frac{3}{2}x(x + 1)^{-1/2}: f′′(x)=32[(x+1)−1/2−12x(x+1)−3/2]f''(x) = \frac{3}{2}\left[(x + 1)^{-1/2} - \frac{1}{2}x(x + 1)^{-3/2}\right]. The two powers are −12-\frac{1}{2} and −32-\frac{3}{2}; the SMALLEST is −32-\frac{3}{2}, and (x+1)−1/2=(x+1)−3/2(x+1)(x + 1)^{-1/2} = (x + 1)^{-3/2}(x + 1). So f′′(x)=32(x+1)−3/2[(x+1)−x2]=32(x+1)−3/2⋅x+22=3(x+2)4(x+1)3/2f''(x) = \frac{3}{2}(x + 1)^{-3/2}\left[(x + 1) - \frac{x}{2}\right] = \frac{3}{2}(x + 1)^{-3/2} \cdot \frac{x + 2}{2} = \frac{3(x + 2)}{4(x + 1)^{3/2}}. The frequent error is to factor out (x+1)−1/2(x + 1)^{-1/2}, the power that looks smaller because 12<32\frac{1}{2} < \frac{3}{2}: the bracket then keeps a negative power inside, and the sign is still hidden. Now f′′(x)=0f''(x) = 0 gives x=−2x = -2, but −2<−1-2 < -1 is not in the domain: it is NOT a candidate, and there is no point of inflection. On the whole domain x+2>1>0x + 2 > 1 > 0, so f′′>0f'' > 0: ff is concave up on (−1,∞)(-1, \infty).

d) f′′(0)=3⋅24⋅1=32>0f''(0) = \frac{3 \cdot 2}{4 \cdot 1} = \frac{3}{2} > 0 and f′(0)=0f'(0) = 0: by the Second Derivative Test, ff has a local minimum f(0)=−2f(0) = -2. It is absolute: f′<0f' < 0 on (−1,0)(-1, 0) and f′>0f' > 0 on (0,∞)(0, \infty), so ff decreases on all of [−1,0][-1, 0] and increases on all of [0,∞)[0, \infty). The range: from its minimum −2-2 the function increases without bound, and it is continuous, so the range is [−2,∞)[-2, \infty). The value at the endpoint, f(−1)=0f(-1) = 0, is a local maximum in Thomas' sense for an endpoint, but not the largest value.

e) The positive xx-intercept is x=2x = 2: f′(2)=3⋅223=33=3f'(2) = \frac{3 \cdot 2}{2\sqrt 3} = \frac{3}{\sqrt 3} = \sqrt 3 (rationalize: 33=333\frac{3}{\sqrt 3} = \frac{3\sqrt 3}{3}). The sketch: start at (−1,0)(-1, 0) with a vertical tangent, go down to (0,−2)(0, -2) where the tangent is horizontal, come back up through (2,0)(2, 0) with slope 3\sqrt 3, and rise without bound, bending upward all the way. A drawing with an S shape near x=−2x = -2 would be drawing a point that does not exist.

-3-2-11234-3-2-112345min (0, −2)vertical tangent

Exercise 5: Three curves to match: f, its derivative and its second derivative

The figure shows, on [−3,3][-3, 3], the graphs of a function ff, of f′f' and of f′′f'', labelled A, B and C in no particular order. The function is f(x)=xe−x2f(x) = xe^{-x^2}, but part a) must be answered from the figure alone.

The rule behind every matching question: where one curve has a horizontal tangent, the NEXT one (its derivative) crosses the axis; where a curve rises, the next one is above the axis.

-3-2-1123-2-1.5-1-0.50.511.52ABC
  • a) From the figure alone, decide which curve is ff, which is f′f' and which is f′′f''. Give two independent reasons.
  • b) Compute f′(x)f'(x) and f′′(x)f''(x), pulling out the factor e−x2e^{-x^2} at each step. Check on the figure the values f′(0)f'(0) and f′′(1)f''(1).
  • c) Find the local extrema of ff and its points of inflection, with exact coordinates.
  • d) At which xx is the slope of ff largest, and what is that slope? Explain why this point is a point of inflection of ff.
  • e) Show that ff is odd. What does that say about the symmetry of f′f' and of f′′f''? Check it on the figure.

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

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

Answers

  • a) C is ff, A is f′f', B is f′′f''.
  • b) f′(x)=(1−2x2)e−x2f'(x) = (1 - 2x^2)e^{-x^2}, f′′(x)=2x(2x2−3)e−x2f''(x) = 2x(2x^2 - 3)e^{-x^2}; f′(0)=1f'(0) = 1, f′′(1)=−2e≈−0.7358f''(1) = -\frac{2}{e} \approx -0.7358.
  • c) Local max (12,12e)\left(\frac{1}{\sqrt 2}, \frac{1}{\sqrt{2e}}\right), local min (−12,−12e)\left(-\frac{1}{\sqrt 2}, -\frac{1}{\sqrt{2e}}\right); inflection points (0,0)(0, 0) and (±32,±32 e−3/2)\left(\pm\sqrt{\frac{3}{2}}, \pm\sqrt{\frac{3}{2}}\, e^{-3/2}\right).
  • d) At x=0x = 0, slope 11: the maximum of f′f', where f′′f'' changes sign.
  • e) ff odd, so f′f' is even and f′′f'' is odd.

a) Curve C has horizontal tangents near x=±0.7x = \pm 0.7, and curve A crosses the axis exactly there: A is the derivative of C. Curve A has horizontal tangents at 00 and near ±1.2\pm 1.2, and curve B crosses the axis at those three places: B is the derivative of A. So C is ff, A is f′f', B is f′′f''. Second, independent reason, by signs: between the two horizontal tangents of C the curve rises, and curve A is above the axis there; outside, C falls and A is below. Third check: C is odd (symmetric about the origin), A is even (symmetric about the yy-axis), B is odd again, and derivatives alternate between odd and even, as e) proves. The TALLEST curve is not ff: here f′′f'' reaches about 1.951.95 while ff never exceeds 0.430.43.

b) Product rule and chain rule (inner function −x2-x^2, derivative −2x-2x): f′(x)=e−x2+x⋅(−2x)e−x2f'(x) = e^{-x^2} + x \cdot (-2x)e^{-x^2}. Pull out the exponential, which is a factor of BOTH terms: f′(x)=(1−2x2)e−x2f'(x) = (1 - 2x^2)e^{-x^2}. Again: f′′(x)=−4xe−x2+(1−2x2)(−2x)e−x2=[−4x−2x+4x3]e−x2=2x(2x2−3)e−x2f''(x) = -4xe^{-x^2} + (1 - 2x^2)(-2x)e^{-x^2} = \left[-4x - 2x + 4x^3\right]e^{-x^2} = 2x(2x^2 - 3)e^{-x^2}. The gain of factoring is immediate: e−x2>0e^{-x^2} > 0 for every xx, so the sign of f′f' is the sign of 1−2x21 - 2x^2 and the sign of f′′f'' is the sign of 2x(2x2−3)2x(2x^2 - 3), two polynomials read at a glance. Checks: f′(0)=1f'(0) = 1, the top of curve A; f′′(1)=2(2−3)e−1=−2e≈−0.7358f''(1) = 2(2 - 3)e^{-1} = -\frac{2}{e} \approx -0.7358, and curve B is indeed near −0.74-0.74 at x=1x = 1.

c) f′f' changes from −- to ++ at −12-\frac{1}{\sqrt 2} and from ++ to −- at 12\frac{1}{\sqrt 2}: local minimum and local maximum, f(12)=12e−1/2=12e≈0.4289f\left(\frac{1}{\sqrt 2}\right) = \frac{1}{\sqrt 2}e^{-1/2} = \frac{1}{\sqrt{2e}} \approx 0.4289, and its opposite at −12-\frac{1}{\sqrt 2}. For the concavity, 2x(2x2−3)2x(2x^2 - 3) vanishes at 00 and at x2=32x^2 = \frac{3}{2}, x=±32=±62≈±1.2247x = \pm\sqrt{\frac{3}{2}} = \pm\frac{\sqrt 6}{2} \approx \pm 1.2247, and changes sign at each: f′′<0f'' < 0 on (−∞,−62)\left(-\infty, -\frac{\sqrt 6}{2}\right), >0> 0 on (−62,0)\left(-\frac{\sqrt 6}{2}, 0\right), <0< 0 on (0,62)\left(0, \frac{\sqrt 6}{2}\right), >0> 0 after. Three points of inflection: (0,0)(0, 0) and (±62,±62e−3/2)\left(\pm\frac{\sqrt 6}{2}, \pm\frac{\sqrt 6}{2}e^{-3/2}\right), about (±1.2247,±0.2733)(\pm 1.2247, \pm 0.2733). A copy that stops at 2x2=32x^2 = 3 after dividing by xx loses the inflection at the origin: never divide by a factor that can vanish, factor it out.

d) The slope of ff is f′(x)f'(x), whose largest value, read on curve A and confirmed by f′(x)=(1−2x2)e−x2≤1⋅1f'(x) = (1 - 2x^2)e^{-x^2} \le 1 \cdot 1 (both factors are at most 11 where f′≥0f' \ge 0), is f′(0)=1f'(0) = 1. A maximum of f′f' is a place where f′f' stops increasing and starts decreasing, which is exactly a change of sign of (f′)′=f′′(f')' = f'': the steepest point of ff is a point of inflection. That is the geometric meaning of an inflection point, and the reason it matters in applications: the RATE is extreme there.

e) f(−x)=(−x)e−(−x)2=−xe−x2=−f(x)f(-x) = (-x)e^{-(-x)^2} = -xe^{-x^2} = -f(x): ff is odd. Differentiating f(−x)=−f(x)f(-x) = -f(x) with the chain rule gives −f′(−x)=−f′(x)-f'(-x) = -f'(x), so f′(−x)=f′(x)f'(-x) = f'(x): f′f' is even. Differentiating once more, −f′′(−x)=f′′(x)-f''(-x) = f''(x): f′′f'' is odd. On the figure, A is symmetric about the yy-axis and B about the origin, which was the third reason of a).

Part B: problems and reasoning (/50)

Exercise 6: A fractional power of a quadratic: two cusps, and inflection points far from them

Sketch the graph of f(x)=(x2−2x)2/3f(x) = (x^2 - 2x)^{2/3}, defined for every real xx since (x2−2x)2/3=(x2−2x3)2(x^2 - 2x)^{2/3} = \left(\sqrt[3]{x^2 - 2x}\right)^2. The derivative does not exist at two points of the domain, and the second derivative shows the algebraic gesture of the chapter in its purest form: two powers of x2−2xx^2 - 2x, and the smaller one must come out.

A scientific calculator is allowed for decimal values, to four decimal places.

  • a) Find the domain, the intercepts and the sign of ff, and its limits at ±∞\pm\infty. Show that f(2−x)=f(x)f(2 - x) = f(x): what symmetry of the graph does this give?
  • b) Show that f′(x)=4(x−1)3(x2−2x)1/3f'(x) = \frac{4(x - 1)}{3(x^2 - 2x)^{1/3}} and list ALL the critical numbers. Classify them with a sign table of f′f' and describe the graph at x=0x = 0 and x=2x = 2.
  • c) Differentiate f′(x)=43(x−1)(x2−2x)−1/3f'(x) = \frac{4}{3}(x - 1)(x^2 - 2x)^{-1/3} with the product rule, factor out (x2−2x)−4/3(x^2 - 2x)^{-4/3} and show that f′′(x)=4(x2−2x−2)9(x2−2x)4/3f''(x) = \frac{4(x^2 - 2x - 2)}{9(x^2 - 2x)^{4/3}}. Compute f′′(1)f''(1) and confirm the nature of the critical number 11.
  • d) Study the concavity and find the points of inflection. Are the cusps points of inflection?
  • e) Sketch the graph. The function is concave down on (1−3,0)(1 - \sqrt 3, 0), on (0,2)(0, 2) and on (2,1+3)(2, 1 + \sqrt 3): is it concave down on the whole interval (1−3,1+3)(1 - \sqrt 3, 1 + \sqrt 3)?

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

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

Answers

  • a) Domain R\mathbb{R}, f≥0f \ge 0; intercepts (0,0)(0, 0) and (2,0)(2, 0); f→∞f \to \infty at both ends; graph symmetric about the line x=1x = 1.
  • b) Critical numbers 00, 11, 22; minima f(0)=f(2)=0f(0) = f(2) = 0 at two cusps, local max f(1)=1f(1) = 1.
  • c) f′′(1)=−43<0f''(1) = -\frac{4}{3} < 0: local maximum at 11 confirmed.
  • d) Concave up for x<1−3x < 1 - \sqrt 3 and x>1+3x > 1 + \sqrt 3, down between (except at 00 and 22); inflection points (1±3,43)(1 \pm \sqrt 3, \sqrt[3]{4}); the cusps are not.
  • e) No: the cusps break the concavity; ff is concave down on each of the three pieces only.

a) The cube root is defined for every real number, so the domain is R\mathbb{R} and ff is continuous everywhere. As a square, f≥0f \ge 0, with f=0  ⟺  x2−2x=x(x−2)=0  ⟺  x=0f = 0 \iff x^2 - 2x = x(x - 2) = 0 \iff x = 0 or x=2x = 2; the yy-intercept is the origin. As x→±∞x \to \pm\infty, x2−2x→∞x^2 - 2x \to \infty and f→∞f \to \infty: no horizontal asymptote, and no slant one since f(x)f(x) grows like ∣x∣4/3|x|^{4/3}, faster than any line. Symmetry: (2−x)2−2(2−x)=4−4x+x2−4+2x=x2−2x(2 - x)^2 - 2(2 - x) = 4 - 4x + x^2 - 4 + 2x = x^2 - 2x, so f(2−x)=f(x)f(2 - x) = f(x). The points xx and 2−x2 - x are mirror images across the vertical line x=1x = 1: the graph is symmetric about x=1x = 1. It is neither even nor odd, and looking only for those two symmetries misses it. Completing the square says the same thing: x2−2x=(x−1)2−1x^2 - 2x = (x - 1)^2 - 1.

b) Chain rule with inner function x2−2xx^2 - 2x: f′(x)=23(x2−2x)−1/3(2x−2)=4(x−1)3(x2−2x)1/3f'(x) = \frac{2}{3}(x^2 - 2x)^{-1/3}(2x - 2) = \frac{4(x - 1)}{3(x^2 - 2x)^{1/3}}. Critical numbers: f′(1)=0f'(1) = 0, and f′f' does not exist at x=0x = 0 and x=2x = 2, which ARE in the domain. Three critical numbers. The cube root has the sign of its argument x(x−2)x(x - 2), so the table reads: for x<0x < 0, (−)(+)<0\frac{(-)}{(+)} < 0; on (0,1)(0, 1), (−)(−)>0\frac{(-)}{(-)} > 0; on (1,2)(1, 2), (+)(−)<0\frac{(+)}{(-)} < 0; for x>2x > 2, f′>0f' > 0. First Derivative Test: local minima f(0)=f(2)=0f(0) = f(2) = 0, local maximum f(1)=(−1)2/3=(−13)2=1f(1) = (-1)^{2/3} = \left(\sqrt[3]{-1}\right)^2 = 1. At x=2x = 2: f′(x)→−∞f'(x) \to -\infty as x→2−x \to 2^- and +∞+\infty as x→2+x \to 2^+ (numerator near 43\frac{4}{3}, denominator →0∓\to 0^{\mp}). The tangent turns vertical from both sides while the graph goes down then up: a CUSP, and by symmetry another one at 00.

c) Product rule: f′′(x)=43(x2−2x)−1/3+43(x−1)⋅(−13)(x2−2x)−4/3(2x−2)f''(x) = \frac{4}{3}(x^2 - 2x)^{-1/3} + \frac{4}{3}(x - 1) \cdot \left(-\frac{1}{3}\right)(x^2 - 2x)^{-4/3}(2x - 2). The powers are −13-\frac{1}{3} and −43-\frac{4}{3}; the smallest is −43-\frac{4}{3}, and (x2−2x)−1/3=(x2−2x)−4/3(x2−2x)(x^2 - 2x)^{-1/3} = (x^2 - 2x)^{-4/3}(x^2 - 2x). So f′′(x)=49(x2−2x)−4/3[3(x2−2x)−2(x−1)2]f''(x) = \frac{4}{9}(x^2 - 2x)^{-4/3}\left[3(x^2 - 2x) - 2(x - 1)^2\right], and the bracket is 3x2−6x−2x2+4x−2=x2−2x−23x^2 - 6x - 2x^2 + 4x - 2 = x^2 - 2x - 2: f′′(x)=4(x2−2x−2)9(x2−2x)4/3f''(x) = \frac{4(x^2 - 2x - 2)}{9(x^2 - 2x)^{4/3}}. The denominator (x2−2x)4/3=(x2−2x3)4(x^2 - 2x)^{4/3} = \left(\sqrt[3]{x^2 - 2x}\right)^4 is POSITIVE for x≠0,2x \ne 0, 2: an even power of a cube root, whatever the sign inside. f′′(1)=4(1−2−2)9⋅(−1)4/3=4(−3)9⋅1=−43<0f''(1) = \frac{4(1 - 2 - 2)}{9 \cdot (-1)^{4/3}} = \frac{4(-3)}{9 \cdot 1} = -\frac{4}{3} < 0, with f′(1)=0f'(1) = 0: the Second Derivative Test confirms the local maximum at 11. At 00 and 22 the test cannot even be applied, since f′f' does not exist there.

d) f′′f'' has the sign of x2−2x−2x^2 - 2x - 2, whose roots are x=2±4+82=1±3x = \frac{2 \pm \sqrt{4 + 8}}{2} = 1 \pm \sqrt 3, about −0.7321-0.7321 and 2.73212.7321: positive outside them, concave up; negative between them (except at 00 and 22, where f′′f'' is undefined), concave down. The sign changes at 1±31 \pm \sqrt 3, points of the domain where the graph has a tangent: points of inflection. There x2−2x=(x−1)2−1=3−1=2x^2 - 2x = (x - 1)^2 - 1 = 3 - 1 = 2, so f=22/3=43≈1.5874f = 2^{2/3} = \sqrt[3]{4} \approx 1.5874: the points (1±3,43)(1 \pm \sqrt 3, \sqrt[3]{4}). At the cusps, f′′f'' is undefined but NEGATIVE on both sides: no change of concavity, and in any case no tangent line. The cusps are not points of inflection.

e) The graph: from the far left, coming down, concave up; inflection at (1−3,43)(1 - \sqrt 3, \sqrt[3]{4}); concave down into the cusp (0,0)(0, 0); up, concave down, to the rounded top (1,1)(1, 1); down, concave down, into the cusp (2,0)(2, 0); up again, concave down until (1+3,43)(1 + \sqrt 3, \sqrt[3]{4}), then concave up; all of it symmetric about x=1x = 1. On (1−3,1+3)(1 - \sqrt 3, 1 + \sqrt 3) as a whole, ff is NOT concave down. Thomas defines concave down on an interval as f′f' decreasing on that interval, and f′f' jumps from −∞-\infty to +∞+\infty across 22: it is not decreasing across the cusp. Geometrically, the chord from (2−h,f(2−h))(2 - h, f(2 - h)) to (2+h,f(2+h))(2 + h, f(2 + h)) lies ABOVE the cusp point, which a concave down curve never allows. Concavity is decided piece by piece, and a point where f′f' does not exist cuts the pieces.

-2-11234-112345max (1, 1)cusp (0, 0)cusp (2, 0)1 + √31 − √3

Exercise 7: An exponential study: pull out e to the minus x, and meet a flat inflection point

Sketch the graph of f(x)=x3e−xf(x) = x^3e^{-x}. Every derivative of a product by e−xe^{-x} is again a polynomial times e−xe^{-x}, provided the exponential is pulled out as a common factor at each step; then its sign is decided by the polynomial alone, since e−x>0e^{-x} > 0.

A scientific calculator is allowed; give decimal values to four decimal places. L'Hôpital's Rule may be used as a tool.

  • a) Find the domain, the intercept and the sign of ff. Find lim⁡x→−∞f(x)\lim_{x\to-\infty} f(x) and lim⁡x→∞f(x)\lim_{x\to\infty} f(x), naming the form of each limit before computing it. Give the horizontal asymptote.
  • b) Show that f′(x)=x2(3−x)e−xf'(x) = x^2(3 - x)e^{-x}. Find the critical numbers, and classify them.
  • c) Show that f′′(x)=x(x2−6x+6)e−xf''(x) = x(x^2 - 6x + 6)e^{-x}. Find the intervals of concavity and the three points of inflection, with exact xx-coordinates and decimal yy-coordinates.
  • d) Compute f′′(3)f''(3) and f′′(0)f''(0). What does the Second Derivative Test say at each critical number, and what does the origin look like on the graph?
  • e) Gather the results in one table, sketch the graph and give the range of ff.

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

a)
b)
c)
d)
Show the solution

Answers

  • a) Domain R\mathbb{R}; intercept (0,0)(0, 0); ff has the sign of xx; f→−∞f \to -\infty at −∞-\infty; f→0f \to 0 at +∞+\infty (asymptote y=0y = 0 on the right only).
  • b) Critical numbers 00 and 33; no extremum at 00; absolute maximum f(3)=27e−3≈1.3443f(3) = 27e^{-3} \approx 1.3443.
  • c) Concave down on (−∞,0)(-\infty, 0) and (3−3,3+3)(3 - \sqrt 3, 3 + \sqrt 3), up on (0,3−3)(0, 3 - \sqrt 3) and (3+3,∞)(3 + \sqrt 3, \infty); inflections at x=0x = 0, 3−33 - \sqrt 3 (y≈0.5736y \approx 0.5736), 3+33 + \sqrt 3 (y≈0.9334y \approx 0.9334).
  • d) f′′(3)=−9e−3≈−0.4481f''(3) = -9e^{-3} \approx -0.4481: maximum; f′′(0)=0f''(0) = 0: silent, and (0,0)(0, 0) is a flat inflection point.
  • e) Range (−∞,27e−3](-\infty, 27e^{-3}].

a) ff is defined on R\mathbb{R}; f(0)=0f(0) = 0 is the only intercept, since e−x>0e^{-x} > 0; and ff has the sign of x3x^3, that of xx. As x→−∞x \to -\infty: x3→−∞x^3 \to -\infty and e−x→+∞e^{-x} \to +\infty, a form (−∞)(+∞)(-\infty)(+\infty) which is NOT indeterminate: f→−∞f \to -\infty. As x→∞x \to \infty: x3→∞x^3 \to \infty and e−x→0e^{-x} \to 0, the form ∞⋅0\infty \cdot 0, indeterminate. Rewrite f(x)=x3exf(x) = \frac{x^3}{e^x}, form ∞∞\frac{\infty}{\infty}, and apply L'Hôpital's Rule three times, checking the form each time: 3x2ex\frac{3x^2}{e^x}, 6xex\frac{6x}{e^x}, 6ex→0\frac{6}{e^x} \to 0. So y=0y = 0 is a horizontal asymptote on the RIGHT only; on the left the graph plunges.

b) Product rule: f′(x)=3x2e−x+x3(−e−x)f'(x) = 3x^2e^{-x} + x^3(-e^{-x}). Pull out every common factor, not only the exponential: f′(x)=x2e−x(3−x)f'(x) = x^2e^{-x}(3 - x). Critical numbers: 00 and 33. Signs: x2≥0x^2 \ge 0 and e−x>0e^{-x} > 0, so f′f' has the sign of 3−x3 - x, positive on both sides of 00 and changing from ++ to −- at 33. So 00 is NOT an extremum (ff increases through it), and f(3)=27e−3≈1.3443f(3) = 27e^{-3} \approx 1.3443 is a local maximum, even the absolute maximum since ff increases on all of (−∞,3](-\infty, 3] and decreases on all of [3,∞)[3, \infty).

c) Write f′(x)=(3x2−x3)e−xf'(x) = (3x^2 - x^3)e^{-x} and differentiate once more: f′′(x)=(6x−3x2)e−x−(3x2−x3)e−x=(x3−6x2+6x)e−x=x(x2−6x+6)e−xf''(x) = (6x - 3x^2)e^{-x} - (3x^2 - x^3)e^{-x} = (x^3 - 6x^2 + 6x)e^{-x} = x(x^2 - 6x + 6)e^{-x}. The quadratic formula gives the roots of x2−6x+6x^2 - 6x + 6: x=6±36−242=3±3x = \frac{6 \pm \sqrt{36 - 24}}{2} = 3 \pm \sqrt 3, about 1.26791.2679 and 4.73214.7321. So f′′f'' has the sign of x(x−(3−3))(x−(3+3))x(x - (3 - \sqrt 3))(x - (3 + \sqrt 3)): −- on (−∞,0)(-\infty, 0), ++ on (0,3−3)(0, 3 - \sqrt 3), −- on (3−3,3+3)(3 - \sqrt 3, 3 + \sqrt 3), ++ on (3+3,∞)(3 + \sqrt 3, \infty). Three changes of sign, three points of inflection: (0,0)(0, 0), (3−3,f(3−3))≈(1.2679,0.5736)\left(3 - \sqrt 3, f(3 - \sqrt 3)\right) \approx (1.2679, 0.5736) and (3+3,f(3+3))≈(4.7321,0.9334)\left(3 + \sqrt 3, f(3 + \sqrt 3)\right) \approx (4.7321, 0.9334). The trap: dividing x3−6x2+6x=0x^3 - 6x^2 + 6x = 0 by xx and keeping only 3±33 \pm \sqrt 3, which loses the most important one.

d) f′′(3)=3(9−18+6)e−3=−9e−3≈−0.4481<0f''(3) = 3(9 - 18 + 6)e^{-3} = -9e^{-3} \approx -0.4481 < 0: the Second Derivative Test confirms the maximum at 33. f′′(0)=0f''(0) = 0: the test is silent at 00, and b) already decided with the First Derivative Test, no extremum. What the origin looks like: f′(0)=0f'(0) = 0, a HORIZONTAL tangent, and f′′f'' changes sign there, from concave down to concave up. The graph flattens, crosses its own tangent line y=0y = 0, and keeps rising: a flat point of inflection, like the one of x3x^3 at the origin.

e) The table, left to right: from −∞-\infty the curve rises, concave down; flat inflection at (0,0)(0, 0); rising and concave up until x=3−3x = 3 - \sqrt 3; rising, now concave down, to the maximum (3,27e−3)(3, 27e^{-3}); falling, still concave down, to x=3+3x = 3 + \sqrt 3; then falling and concave up, settling on y=0y = 0 from above, as a curve that decreases toward an asymptote from above must. Range: every value from −∞-\infty up to the maximum is taken by continuity: (−∞,27e−3](-\infty, 27e^{-3}].

-112345678910-3-2.5-2-1.5-1-0.50.511.52max (3, 27/e³)flat inflection (0, 0)x = 3 − √3x = 3 + √3

Exercise 8: Five statements to correct

Each statement below was written by a student in a MATH 203 assignment on concavity, and each one is false. Say what is wrong, give a counterexample, and write a correct statement. The answer boxes ask which of the proposed functions is a counterexample.

  • a) If f′′(x)>0f''(x) > 0 for every xx in an interval II, then ff is increasing on II.
  • b) If f′′(c)f''(c) does not exist, then (c,f(c))(c, f(c)) cannot be a point of inflection.
  • c) If f′(c)=0f'(c) = 0 and f′′(c)=0f''(c) = 0, then ff has no local extremum at cc.
  • d) If f′′(x)>0f''(x) > 0 for every real xx, then ff has an absolute minimum.
  • e) If ff is continuous on (a,c)(a, c) and concave down on (a,b)(a, b) and on (b,c)(b, c), then ff is concave down on (a,c)(a, c).

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

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

Answers

  • a) False: x2x^2 on (−∞,0)(-\infty, 0) is concave up and decreasing. Concave up means f′f' increasing, not ff.
  • b) False: x1/3x^{1/3}, or hh of Exercise 1, has an inflection point at 00 where the second derivative does not exist.
  • c) False: (x−1)4+2(x - 1)^4 + 2 has a minimum at 11 with f′(1)=f′′(1)=0f'(1) = f''(1) = 0. The test is silent, it does not say no.
  • d) False: exe^x has f′′>0f'' > 0 and no minimum. True if moreover ff has a critical number.
  • e) False: x2/3x^{2/3} with b=0b = 0 on (−1,1)(-1, 1). True if moreover ff is differentiable at bb.

a) FALSE. f(x)=x2f(x) = x^2 has f′′(x)=2>0f''(x) = 2 > 0 everywhere, and it is DECREASING on (−∞,0)(-\infty, 0). The second derivative is the rate of change of f′f', not of ff: f′′>0f'' > 0 says that f′f' increases, that the slopes grow, and growing slopes can all be negative (−4-4, −2-2, −1-1...). Correct statement (Thomas' definition read the right way): if f′′>0f'' > 0 on II, then f′f' is increasing on II and the graph of ff is concave up on II. The sign of f′f', not of f′′f'', decides the direction.

b) FALSE. y=x1/3y = x^{1/3} has y′′=−29x−5/3y'' = -\frac{2}{9}x^{-5/3}, undefined at 00, positive for x<0x < 0 and negative for x>0x > 0; the graph has a vertical tangent at the origin, so (0,0)(0, 0) is a point of inflection. The function h(x)=x7/3−7x1/3h(x) = x^{7/3} - 7x^{1/3} of Exercise 1 does the same. Correct statement: the candidates for inflection are the points of the domain where f′′=0f'' = 0 OR where f′′f'' does not exist; each one is decided by the sign of f′′f'' on both sides.

c) FALSE. k(x)=(x−1)4+2k(x) = (x - 1)^4 + 2, the function of Exercise 2 d), has k′(1)=k′′(1)=0k'(1) = k''(1) = 0 and an absolute minimum at 11. When f′′(c)=0f''(c) = 0, the Second Derivative Test gives NO information: a local minimum (Exercise 2 d)), a local maximum (3−(x−1)43 - (x - 1)^4) or neither (Exercise 2 c) at 00) are all possible. Correct statement: if f′(c)=0f'(c) = 0 and f′′(c)=0f''(c) = 0, use the First Derivative Test.

d) FALSE. f(x)=exf(x) = e^x has f′′(x)=ex>0f''(x) = e^x > 0 for every xx, and no minimum: its values come as close to 00 as we like without ever reaching it. Being concave up everywhere does not force the graph to turn around. Correct statement: if f′′>0f'' > 0 on R\mathbb{R} AND f′(c)=0f'(c) = 0 for some cc, then f(c)f(c) is the absolute minimum, because f′f' is increasing, negative before cc and positive after. For instance x2+exx^2 + e^x has f′′=2+ex>0f'' = 2 + e^x > 0 and a critical number between −1-1 and 00, where f′=2x+exf' = 2x + e^x changes sign.

e) FALSE. f(x)=x2/3f(x) = x^{2/3} is continuous, with f′′(x)=−29x−4/3<0f''(x) = -\frac{2}{9}x^{-4/3} < 0 on (−1,0)(-1, 0) and on (0,1)(0, 1): concave down on each piece. But on (−1,1)(-1, 1) the chord from (−1,1)(-1, 1) to (1,1)(1, 1) lies ABOVE the point (0,0)(0, 0) of the graph, as the figure shows, which never happens for a concave down function; and f′f' jumps from −∞-\infty to +∞+\infty at 00, so it is not decreasing across 00. The same thing happens at the cusps of Exercise 6. Correct statement: if moreover ff is differentiable at bb, then f′f' is decreasing on the whole of (a,c)(a, c) and ff is concave down on (a,c)(a, c).

-1.5-1-0.50.511.5-0.50.511.5chord ABOVE the curvey = ∛(x²)

Exercise 9: Fertilizer and wheat: locating the point of diminishing returns

An agronomist models the extra yield of a wheat field, YY in kilograms per hectare, as a function of the amount of nitrogen fertilizer applied, xx in kilograms per hectare: Y(x)=900x2x2+30 000Y(x) = \frac{900x^2}{x^2 + 30\,000} for 0≤x≤4000 \le x \le 400. The figure shows the graph of YY.

The derivative Y′(x)Y'(x) is the MARGINAL yield: about how many extra kilograms of wheat one more kilogram of fertilizer brings. The point of diminishing returns is the point of inflection where YY passes from concave up to concave down: beyond it, each extra kilogram of fertilizer still raises the yield, but by less than the previous one. A scientific calculator is allowed.

501001502002503003504001002003004005006007008009001000Y = 900Y(x)x (kg/ha)Y (kg/ha)
  • a) Show that Y′(x)=54 000 000 x(x2+30 000)2Y'(x) = \frac{54\,000\,000\,x}{(x^2 + 30\,000)^2} and that YY is increasing on [0,400][0, 400]. Find lim⁡x→∞Y(x)\lim_{x\to\infty} Y(x) and interpret it for the field.
  • b) Show that Y′′(x)=162 000 000 (10 000−x2)(x2+30 000)3Y''(x) = \frac{162\,000\,000\,(10\,000 - x^2)}{(x^2 + 30\,000)^3}, pulling out the common factor before expanding. Find the point of diminishing returns exactly.
  • c) Show that the marginal yield is largest at that point. Give its value, with units, and interpret it.
  • d) Compute Y(101)−Y(100)Y(101) - Y(100) and Y(301)−Y(300)Y(301) - Y(300) to the hundredth, compare them with Y′(100)Y'(100) and Y′(300)Y'(300), and explain what the two numbers show about the fertilizer.
  • e) On a test plot the agronomist records the yield every 5050 kg per hectare: Y(0)=0Y(0) = 0, Y(50)≈69.23Y(50) \approx 69.23, Y(100)=225Y(100) = 225, Y(150)≈385.71Y(150) \approx 385.71, Y(200)≈514.29Y(200) \approx 514.29, Y(250)≈608.11Y(250) \approx 608.11. Which 5050 kg step brings the largest gain? Explain why it is not the step that ENDS at x=100x = 100.

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

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

Answers

  • a) Y′>0Y' > 0 for x>0x > 0: increasing; Y→900Y \to 900: the field can never give more than 900900 extra kg per hectare.
  • b) Point of diminishing returns (100,225)(100, 225): 100100 kg per hectare of fertilizer, 225225 kg per hectare of extra wheat.
  • c) Y′(100)=3.375Y'(100) = 3.375 kg of wheat per kg of fertilizer, the largest marginal yield.
  • d) Y(101)−Y(100)≈3.37Y(101) - Y(100) \approx 3.37 close to Y′(100)=3.375Y'(100) = 3.375; Y(301)−Y(300)≈1.12Y(301) - Y(300) \approx 1.12 close to Y′(300)=1.125Y'(300) = 1.125: the same kilogram brings three times less.
  • e) The step from 100100 to 150150 (+160.71+160.71): its average marginal yield beats the step from 5050 to 100100 (+155.77+155.77), because Y′Y' decreases slowly after its peak.

a) Write Y(x)=900x2(x2+30 000)−1Y(x) = 900x^2(x^2 + 30\,000)^{-1}, or use the quotient rule: Y′(x)=1800x(x2+30 000)−900x2⋅2x(x2+30 000)2Y'(x) = \frac{1800x(x^2 + 30\,000) - 900x^2 \cdot 2x}{(x^2 + 30\,000)^2}. Pull out 1800x1800x: 1800x[(x2+30 000)−x2]=1800x⋅30 000=54 000 000 x1800x\left[(x^2 + 30\,000) - x^2\right] = 1800x \cdot 30\,000 = 54\,000\,000\,x. So Y′(x)=54 000 000 x(x2+30 000)2>0Y'(x) = \frac{54\,000\,000\,x}{(x^2 + 30\,000)^2} > 0 for 0<x≤4000 < x \le 400: YY is increasing, more fertilizer always means more wheat in this model. Dividing by x2x^2, Y(x)=9001+30 000/x2→900Y(x) = \frac{900}{1 + 30\,000/x^2} \to 900 as x→∞x \to \infty: whatever the dose, the extra yield stays below 900900 kg per hectare. The field saturates.

b) Quotient rule on Y′Y', with ddx(x2+30 000)2=2(x2+30 000)(2x)\frac{d}{dx}(x^2 + 30\,000)^2 = 2(x^2 + 30\,000)(2x): Y′′(x)=54×106[(x2+30 000)2−x⋅2(x2+30 000)(2x)](x2+30 000)4Y''(x) = \frac{54 \times 10^6\left[(x^2 + 30\,000)^2 - x \cdot 2(x^2 + 30\,000)(2x)\right]}{(x^2 + 30\,000)^4}. Pull out x2+30 000x^2 + 30\,000 and cancel one power: Y′′(x)=54×106[(x2+30 000)−4x2](x2+30 000)3=54×106 (30 000−3x2)(x2+30 000)3=162×106 (10 000−x2)(x2+30 000)3Y''(x) = \frac{54 \times 10^6\left[(x^2 + 30\,000) - 4x^2\right]}{(x^2 + 30\,000)^3} = \frac{54 \times 10^6\,(30\,000 - 3x^2)}{(x^2 + 30\,000)^3} = \frac{162 \times 10^6\,(10\,000 - x^2)}{(x^2 + 30\,000)^3}. On [0,400][0, 400], 10 000−x2=(100−x)(100+x)10\,000 - x^2 = (100 - x)(100 + x) has the sign of 100−x100 - x: YY is concave up on [0,100)[0, 100) and concave down on (100,400](100, 400]. The concavity changes from up to down at x=100x = 100, and Y(100)=900×10 00040 000=225Y(100) = \frac{900 \times 10\,000}{40\,000} = 225. The point of diminishing returns is (100,225)(100, 225). Expanding (x2+30 000)2(x^2 + 30\,000)^2 at the first step leads to a degree 4 numerator with coefficients in the billions, the typical place where a copy stops.

c) The sign of Y′′Y'', which is the derivative of Y′Y', says that Y′Y' increases on [0,100][0, 100] and decreases on [100,400][100, 400]: the marginal yield is largest at x=100x = 100. Its value: Y′(100)=54 000 000×100(40 000)2=5.4×1091.6×109=3.375Y'(100) = \frac{54\,000\,000 \times 100}{(40\,000)^2} = \frac{5.4 \times 10^9}{1.6 \times 10^9} = 3.375 kg of wheat per kg of fertilizer. Around 100100 kg per hectare, each extra kilogram of nitrogen brings about 3.43.4 kg of wheat, the best rate of the whole curve; beyond it, the rate falls. An inflection point of YY is the maximum of the RATE Y′Y', which is exactly why it is the point the agronomist cares about.

d) With the calculator, Y(101)−Y(100)≈228.3749−225≈3.37Y(101) - Y(100) \approx 228.3749 - 225 \approx 3.37, very close to Y′(100)=3.375Y'(100) = 3.375: the derivative is the gain of one more unit, approximately. Y′(300)=54×106×300(120 000)2=1.62×10101.44×1010=1.125Y'(300) = \frac{54 \times 10^6 \times 300}{(120\,000)^2} = \frac{1.62 \times 10^{10}}{1.44 \times 10^{10}} = 1.125, and Y(301)−Y(300)≈1.12Y(301) - Y(300) \approx 1.12. The kilogram of fertilizer added at 300300 kg per hectare still increases the yield, but brings three times less wheat than the one added at 100100: that is what diminishing returns means, and it is the concave down part of the curve. Whether the 301301st kilogram is worth buying is a question of prices, not of this chapter.

e) The successive gains are 69.2369.23, 155.77155.77, 160.71160.71, 128.57128.57 and 93.8293.82 kg per hectare. The largest one is the step from 100100 to 150150, not the step from 5050 to 100100 that ends at the inflection point. A 5050 kg step measures the AVERAGE marginal yield over the step, and Y′Y' is not symmetric about its peak: it climbs steeply before 100100 and comes down slowly after it (Y′(50)≈2.56Y'(50) \approx 2.56 while Y′(150)≈2.94Y'(150) \approx 2.94). A table with a coarse step therefore only brackets the point of diminishing returns, somewhere between 5050 and 150150; the second derivative locates it exactly at 100100.

Exercise 10: An epidemic curve: the day the outbreak stops accelerating, and what it predicts

During an influenza outbreak, the total number of cases reported tt days after the first report is modelled by the logistic function N(t)=30001+49e−t/4N(t) = \frac{3000}{1 + 49e^{-t/4}}. The figure shows its S shape. The derivative N′(t)N'(t) is the number of NEW cases per day.

The epidemic is said to be accelerating while the number of new cases per day increases, that is while NN is concave up. A scientific calculator is allowed.

510152025303540500100015002000250030003500N = 3000N(t)t (days)N (cases)
  • a) Multiply the numerator and the denominator by et/4e^{t/4} to show that N(t)=3000et/4et/4+49N(t) = \frac{3000e^{t/4}}{e^{t/4} + 49}. Find N(0)N(0) and lim⁡t→∞N(t)\lim_{t\to\infty} N(t).
  • b) Using the second form, show that N′(t)=36 750 et/4(et/4+49)2N'(t) = \frac{36\,750\,e^{t/4}}{(e^{t/4} + 49)^2}. Deduce that NN is increasing, and compute N′(0)N'(0).
  • c) Show that N′′(t)=36 750 et/4(49−et/4)4(et/4+49)3N''(t) = \frac{36\,750\,e^{t/4}(49 - e^{t/4})}{4(e^{t/4} + 49)^3}. Find the time t∗t^* of the point of inflection exactly, then to the hundredth, and show that N(t∗)=1500N(t^*) = 1500.
  • d) Find the largest number of new cases per day. During which days is the epidemic accelerating?
  • e) Show that for every logistic function N(t)=A1+Be−ktN(t) = \frac{A}{1 + Be^{-kt}}, with AA, BB, kk positive and B>1B > 1, the point of inflection is reached when N=A2N = \frac{A}{2}. In another city, the daily number of new cases peaked on day 2121, when the total reached 26002600 cases. Predict the final size of that outbreak.

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

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

Answers

  • a) N(0)=60N(0) = 60; N(t)→3000N(t) \to 3000.
  • b) N′>0N' > 0: increasing; N′(0)=14.7N'(0) = 14.7 new cases per day.
  • c) t∗=4ln⁡49=8ln⁡7≈15.57t^* = 4\ln 49 = 8\ln 7 \approx 15.57 days, N(t∗)=1500N(t^*) = 1500.
  • d) At most N′(t∗)=187.5N'(t^*) = 187.5 new cases per day; accelerating for 0≤t<t∗≈15.570 \le t < t^* \approx 15.57.
  • e) Inflection when Be−kt=1Be^{-kt} = 1, so N=A2N = \frac{A}{2}; final size 2×2600=52002 \times 2600 = 5200 cases.

a) 30001+49e−t/4⋅et/4et/4=3000et/4et/4+49e−t/4et/4=3000et/4et/4+49\frac{3000}{1 + 49e^{-t/4}} \cdot \frac{e^{t/4}}{e^{t/4}} = \frac{3000e^{t/4}}{e^{t/4} + 49e^{-t/4}e^{t/4}} = \frac{3000e^{t/4}}{e^{t/4} + 49}, since e−t/4et/4=e0=1e^{-t/4}e^{t/4} = e^0 = 1 (a law of exponents, and the step most often skipped). N(0)=30001+49=60N(0) = \frac{3000}{1 + 49} = 60. As t→∞t \to \infty, e−t/4→0e^{-t/4} \to 0 in the first form, so N(t)→3000N(t) \to 3000: the model predicts 30003000 cases in total. The second form has no negative exponent, which makes the quotient rule cleaner.

b) Quotient rule, with ddtet/4=14et/4\frac{d}{dt}e^{t/4} = \frac{1}{4}e^{t/4} by the chain rule: N′(t)=750et/4(et/4+49)−3000et/4⋅14et/4(et/4+49)2N'(t) = \frac{750e^{t/4}(e^{t/4} + 49) - 3000e^{t/4} \cdot \frac{1}{4}e^{t/4}}{(e^{t/4} + 49)^2}. Pull out 750et/4750e^{t/4}: 750et/4[(et/4+49)−et/4]=750×49 et/4=36 750 et/4750e^{t/4}\left[(e^{t/4} + 49) - e^{t/4}\right] = 750 \times 49\,e^{t/4} = 36\,750\,e^{t/4}. So N′(t)=36 750 et/4(et/4+49)2>0N'(t) = \frac{36\,750\,e^{t/4}}{(e^{t/4} + 49)^2} > 0: the total keeps increasing, as a total must. N′(0)=36 750502=14.7N'(0) = \frac{36\,750}{50^2} = 14.7 new cases per day at the start.

c) Write E=et/4E = e^{t/4}, so E′=E4E' = \frac{E}{4}. Quotient rule on 36 750E(E+49)2\frac{36\,750E}{(E + 49)^2}: the numerator is 36 750[E4(E+49)2−E⋅2(E+49)E4]=36 750E4(E+49)[(E+49)−2E]36\,750\left[\frac{E}{4}(E + 49)^2 - E \cdot 2(E + 49)\frac{E}{4}\right] = \frac{36\,750E}{4}(E + 49)\left[(E + 49) - 2E\right]. Cancel one factor E+49E + 49: N′′(t)=36 750E(49−E)4(E+49)3N''(t) = \frac{36\,750E(49 - E)}{4(E + 49)^3}. Everything is positive except 49−E49 - E: N′′>0N'' > 0 while et/4<49e^{t/4} < 49 and N′′<0N'' < 0 after. The concavity changes when et/4=49e^{t/4} = 49, that is t4=ln⁡49\frac{t}{4} = \ln 49, t∗=4ln⁡49=8ln⁡7≈15.57t^* = 4\ln 49 = 8\ln 7 \approx 15.57 days. There, N(t∗)=3000×4949+49=1500N(t^*) = \frac{3000 \times 49}{49 + 49} = 1500.

d) N′′N'' is the derivative of N′N', so N′N' increases before t∗t^* and decreases after: the daily count peaks at t∗t^*, with N′(t∗)=36 750×49982=36 750196=187.5N'(t^*) = \frac{36\,750 \times 49}{98^2} = \frac{36\,750}{196} = 187.5 new cases per day. The epidemic is accelerating while NN is concave up, from the first report until day 15.5715.57, so during days 00 to 1515; from about the middle of day 1515 on, cases still rise but the daily count falls. A report that says the epidemic is still accelerating because the total is still rising confuses N′>0N' > 0 with N′′>0N'' > 0.

e) Multiply by ekte^{kt}: N(t)=Aektekt+BN(t) = \frac{Ae^{kt}}{e^{kt} + B}. The same computation as in b) and c), with AA, BB, kk in place of 30003000, 4949, 14\frac{1}{4}, gives N′(t)=ABk ekt(ekt+B)2N'(t) = \frac{ABk\,e^{kt}}{(e^{kt} + B)^2} and N′′(t)=ABk2ekt(B−ekt)(ekt+B)3N''(t) = \frac{ABk^2e^{kt}(B - e^{kt})}{(e^{kt} + B)^3}, which changes sign exactly when ekt=Be^{kt} = B, a time that exists and is positive since B>1B > 1. At that time N=ABB+B=A2N = \frac{AB}{B + B} = \frac{A}{2}. The epidemic stops accelerating when HALF of its final size is reached. In the other city, the peak of new cases is the inflection point, so A2=2600\frac{A}{2} = 2600 and the model predicts A=5200A = 5200 cases in total. This is how the concavity of an S curve is used in practice: the peak of the daily curve, observed in real time, gives the final size before it happens.

© Ahmed Squalli Houssaini. Exercise set published at www.letuteurscientifique.ca/en/exercices/math203-concavity-curve-sketching. 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, and the algebra comes with it. Get in touch for a first session.

Site by Studio Squalli