MATH 203 Calculus I • Concordia University, Montreal

Revision sheet: concavity and curve sketching (MATH 203)

This sheet is not a summary of section 4.4 of Thomas' Calculus: you already have the course notes and the procedure for graphing. It answers one question only, what makes students lose marks on concavity and curve sketching in MATH 203 at Concordia University, and which precise gesture avoids each loss.

The rules of the chapter fit in four lines. The marks are lost in the algebra between the rule and the sign table: a numerator expanded to degree four when a common factor was waiting to be cancelled, a fractional power whose sign is misread, a factor divided out and lost. The sheet puts that algebra first, then the candidates, then the complete sketch.

Mark this sheet as read or add it to your favourites: a free account, no password, keeps your read sheets and favourites from one visit to the next and tells you which chapter to tackle next. Create your space, an email is enough.

The thread of the chapter

The concavity is read on the SIGN of f′′f'', and that sign can only be read on a factored form: pull out the common factor of a quotient rule, the smallest power of a fractional exponent, the exponential that never vanishes. Then every zero or undefined point of f′′f'' is only a candidate, kept if the sign changes at a point of the domain.

This chapter is part of MATH 203, Calculus I (Concordia)

The essentials

Factor first: the sign of f double prime is read on a product

  • • Quotient rule on uvn\frac{u}{v^n}: the numerator of the derivative contains vn−1v^{n-1}. Pull it out and cancel ONE power before expanding anything: ddx12−x2(x2+12)2\frac{d}{dx}\frac{12 - x^2}{(x^2 + 12)^2} gives 2x(x2−36)(x2+12)3\frac{2x(x^2 - 36)}{(x^2 + 12)^3}.
  • • Fractional powers: factor out the SMALLEST power, the most negative exponent. 289x1/3+149x−5/3=149x−5/3(2x2+1)\frac{28}{9}x^{1/3} + \frac{14}{9}x^{-5/3} = \frac{14}{9}x^{-5/3}(2x^2 + 1), because x1/3=x−5/3⋅x2x^{1/3} = x^{-5/3} \cdot x^2.
  • • Exponentials: pull e−xe^{-x} or e−x2/2e^{-x^2/2} out of every term; it is positive and never zero, so the sign is that of the polynomial left.
  • • Never divide by a factor that can vanish: x3−6x2+6x=x(x2−6x+6)x^3 - 6x^2 + 6x = x(x^2 - 6x + 6) has THREE roots, not two.
  • • Signs of the usual factors: (x−a)2≥0(x - a)^2 \ge 0 never changes sign; x5/3x^{5/3} has the sign of xx; x4/3=(x3)4≥0x^{4/3} = \left(\sqrt[3]{x}\right)^4 \ge 0; x+3>0\sqrt{x + 3} > 0 on its domain.
-12-10-8-6-4-224681012-0.2-0.15-0.1-0.050.050.10.150.2(6, 1/8)(−6, −1/8)updowndownup
xx2+12\frac{x}{x^2 + 12} has f′′=2x(x2−36)(x2+12)3f'' = \frac{2x(x^2 - 36)}{(x^2 + 12)^3}: three sign changes, three points of inflection at −6-6, 00 and 66, and the concavity alternates between them.

Write f′′f'' in factored form on its own line before the table: a marker gives the concavity marks only when the table can be checked against that line.

Candidates, then decisions

  • • Candidates for inflection: every cc of the DOMAIN where f′′(c)=0f''(c) = 0 or f′′(c)f''(c) does not exist.
  • • Decision: (c,f(c))(c, f(c)) is a point of inflection if the sign of f′′f'' changes at cc and the graph has a tangent line there (a vertical one counts).
  • • Second Derivative Test: f′(c)=0f'(c) = 0 with f′′(c)<0f''(c) < 0 gives a local max, with f′′(c)>0f''(c) > 0 a local min, with f′′(c)=0f''(c) = 0 NOTHING: use the First Derivative Test.
  • • Procedure for graphing: domain, intercepts, symmetry, asymptotes, f′f' factored and its sign, f′′f'' factored and its sign, one table, the curve.

The rules in table form

Each row reads from left to right: the assumptions, then the result. A red cell is not an answer, it is the finding that the form settles nothing and the instruction to rewrite it. Every case is followed by a worked example.

The sign of a factor in f double prime

Read a line as: this factor, on this set, has this sign. The red lines are the signs students write and that are false.

FactorOnSign
e−xe^{-x} R\mathbb{R} ++

Example: e−5≈0.0067>0e^{-5} \approx 0.0067 > 0 and e5≈148>0e^{5} \approx 148 > 0: (x−3)e−x(x - 3)e^{-x} has the sign of x−3x - 3.

e−xe^{-x} x>0x > 0 −- false sign

Example: e−2≈0.1353e^{-2} \approx 0.1353: a small positive number, not a negative one.

What to do: An exponential is always positive; the minus sign in the exponent makes it SMALL, never negative.

(x−2)2(x - 2)^2 x≠2x \ne 2 ++

Example: 12(x−2)212(x - 2)^2 at 1.91.9 and 2.12.1: 0.120.12 both times, no sign change, no inflection.

x5/3x^{5/3} x≠0x \ne 0 sign of xx

Example: (−8)5/3=(−2)5=−32(-8)^{5/3} = (-2)^5 = -32 and 85/3=328^{5/3} = 32.

x4/3x^{4/3} x<0x < 0 −- false sign

Example: (−8)4/3=(−2)4=16>0(-8)^{4/3} = (-2)^4 = 16 > 0.

What to do: Write xp/3=(x3)px^{p/3} = \left(\sqrt[3]{x}\right)^p: an even pp gives a positive factor, an odd pp the sign of xx.

(x2+12)3(x^2 + 12)^3 R\mathbb{R} ++

Example: x2+12≥12x^2 + 12 \ge 12, so (x2+12)3≥1728>0(x^2 + 12)^3 \ge 1728 > 0: only the numerator decides.

Once each factor has its line, the sign of f′′f'' is a product of signs, and the table writes itself.

The mistakes that cost marks

These are the errors I correct most often in session. Each one costs marks on a paper, even when the reasoning behind it is right.

1. Expanding the quotient rule numerator instead of cancelling the common factor

2 to 3 marks: the whole concavity study stops there

What not to write

“f′′=−2x(x2+12)2−(12−x2)⋅4x(x2+12)(x2+12)4=2x5−48x3−864x(x2+12)4f'' = \frac{-2x(x^2 + 12)^2 - (12 - x^2) \cdot 4x(x^2 + 12)}{(x^2 + 12)^4} = \frac{2x^5 - 48x^3 - 864x}{(x^2 + 12)^4}, and I cannot factor the numerator.”

What to write

“Pull out −2x(x2+12)-2x(x^2 + 12): f′′(x)=−2x[(x2+12)+2(12−x2)](x2+12)3=2x(x2−36)(x2+12)3f''(x) = \frac{-2x\left[(x^2 + 12) + 2(12 - x^2)\right]}{(x^2 + 12)^3} = \frac{2x(x^2 - 36)}{(x^2 + 12)^3}.”

Why: Both terms of the numerator contain x2+12x^2 + 12; cancelling one power leaves a cubic that factors at sight. Expanding creates a quintic whose roots must be rediscovered.

2. Declaring a point of inflection where f double prime vanishes

1 mark, and an S drawn where the curve is a cup

What not to write

“f(x)=x4+4xf(x) = x^4 + 4x has f′′(x)=12x2f''(x) = 12x^2, so f′′(0)=0f''(0) = 0 and (0,0)(0, 0) is a point of inflection.”

What to write

“f′′(x)=12x2≥0f''(x) = 12x^2 \ge 0 on both sides of 00: no change of sign, so no point of inflection; ff is concave up on R\mathbb{R}.”

Why: A zero of f′′f'' only NOMINATES a point. What makes an inflection is a change of concavity, and a square does not change sign.

3. Concluding no inflection point because f double prime never vanishes

1 to 2 marks

What not to write

“f(x)=x+x3f(x) = x + \sqrt[3]{x} has f′′(x)=−29x5/3f''(x) = -\frac{2}{9x^{5/3}}, which is never 00: no point of inflection.”

What to write

“f′′f'' is undefined at 00, a point of the domain: it is a candidate. f′′>0f'' > 0 for x<0x < 0 and <0< 0 for x>0x > 0, and the tangent at the origin is vertical (f′(x)=1+13x2/3→∞f'(x) = 1 + \frac{1}{3x^{2/3}} \to \infty): (0,0)(0, 0) is a point of inflection.”

-2-1.5-1-0.50.511.52-4-3-2-11234inflection (0, 0)
x+x3x + \sqrt[3]{x} bends up on the left, down on the right, and passes through the origin with a vertical tangent: an inflection point where f′′f'' does not exist.

Why: The candidates are the zeros of f′′f'' AND the points of the domain where f′′f'' does not exist. Forgetting the second list loses every inflection with a vertical tangent.

4. Misreading the sign of a fractional power, and calling a cusp an inflection point

1 mark for the false point, and a wrong sketch near the origin

What not to write

“f(x)=x+3x2/3f(x) = x + 3x^{2/3} has f′′(x)=−23x4/3f''(x) = -\frac{2}{3x^{4/3}}, and x4/3x^{4/3} changes sign at 00: (0,0)(0, 0) is a point of inflection.”

What to write

“x4/3=(x3)4>0x^{4/3} = \left(\sqrt[3]{x}\right)^4 > 0 for x≠0x \ne 0, so f′′<0f'' < 0 on both sides of 00: concave down on (−∞,0)(-\infty, 0) and on (0,∞)(0, \infty), no point of inflection. At 00, f′(x)=1+2x−1/3f'(x) = 1 + 2x^{-1/3} jumps from −∞-\infty to +∞+\infty: a cusp, and a local minimum.”

-12-10-8-6-4-224-4-224681012cusp (0, 0)max (−8, 4)
x+3x2/3x + 3x^{2/3} bends down on both sides of the origin: a cusp and a local minimum, not an inflection point; the top (−8,4)(-8, 4) is a local maximum.

Why: An even power of a cube root is never negative. The origin is a critical number where f′f' jumps from −∞-\infty to +∞+\infty, not a change of concavity.

5. Reading a silent Second Derivative Test as no extremum

1 mark, and often the extremum the question was about

What not to write

“g(x)=3−x6g(x) = 3 - x^6: g′(0)=0g'(0) = 0 and g′′(0)=0g''(0) = 0, so by the Second Derivative Test gg has no extremum at 00.”

What to write

“The test fails at 00. First Derivative Test: g′(x)=−6x5g'(x) = -6x^5 goes from ++ to −- at 00, so g(0)=3g(0) = 3 is a local (in fact absolute) maximum.”

Why: When f′′(c)=0f''(c) = 0 the test gives NO information: a max, a min or neither are all possible. Silent is not negative.

6. Giving an exponential factor a sign

the whole concavity row of the table

What not to write

“f(x)=(x−1)e−xf(x) = (x - 1)e^{-x} has f′′(x)=(x−3)e−xf''(x) = (x - 3)e^{-x}; for x>0x > 0 the exponent −x-x is negative, so e−x<0e^{-x} < 0 and f′′>0f'' > 0 on (0,3)(0, 3).”

What to write

“e−x>0e^{-x} > 0 for every xx, so f′′f'' has the sign of x−3x - 3: concave down on (−∞,3)(-\infty, 3), up on (3,∞)(3, \infty), point of inflection (3,2e−3)(3, 2e^{-3}).”

Why: A negative exponent makes the power small, never negative: e−3=1e3≈0.0498e^{-3} = \frac{1}{e^3} \approx 0.0498. The exponential is pulled out and forgotten.

7. Dividing by x and losing a point of inflection

1 mark, and usually the most visible point of the graph

What not to write

“f′′(x)=(x3−6x2+6x)e−x=0f''(x) = (x^3 - 6x^2 + 6x)e^{-x} = 0, divide by xx: x2−6x+6=0x^2 - 6x + 6 = 0, so the inflection points are at 3±33 \pm \sqrt 3.”

What to write

“f′′(x)=x(x2−6x+6)e−xf''(x) = x(x^2 - 6x + 6)e^{-x}: the candidates are 00 and 3±33 \pm \sqrt 3, and f′′f'' changes sign at all three.”

Why: Dividing an equation by xx assumes x≠0x \ne 0 and silently throws away the root 00. Factor it out instead.

8. Placing the point of diminishing returns at the top of the curve

the whole application question

What not to write

“The yield stops increasing at the point of diminishing returns, so I look for Y′(x)=0Y'(x) = 0.”

What to write

“The point of diminishing returns is where Y′′Y'' changes from ++ to −-: the yield still increases after it, but its RATE Y′Y' is largest there and decreases afterwards.”

Why: Diminishing returns is a statement about the marginal output Y′Y', not about YY. An inflection point of YY is an extremum of Y′Y'.

Which method to choose

Which algebra makes the sign of f double prime readable, by the FORM of f prime

Look at the formula of the first derivative before differentiating again

  • If a quotient uvn\frac{u}{v^n} with n≥2n \ge 2 → quotient rule, then pull vn−1v^{n-1} out of the numerator and cancel it

    Example: 72x(x2+27)2\frac{72x}{(x^2 + 27)^2} gives f′′=216(9−x2)(x2+27)3f'' = \frac{216(9 - x^2)}{(x^2 + 27)^3}

  • If a sum of powers xpx^{p} with fractional exponents → factor out the smallest (most negative) power, then read the sign of each factor

    Example: 49x−2/3−89x−5/3=49x−5/3(x−2)\frac{4}{9}x^{-2/3} - \frac{8}{9}x^{-5/3} = \frac{4}{9}x^{-5/3}(x - 2)

  • If a power of a composite, (x−1)(x2−2x)−1/3(x - 1)(x^2 - 2x)^{-1/3} → product and chain rules, then factor out (x2−2x)−4/3(x^2 - 2x)^{-4/3}

    Example: f′′=4(x2−2x−2)9(x2−2x)4/3f'' = \frac{4(x^2 - 2x - 2)}{9(x^2 - 2x)^{4/3}} for f=(x2−2x)2/3f = (x^2 - 2x)^{2/3}

  • If a polynomial times e±xe^{\pm x} or e−x2/2e^{-x^2/2} → pull the exponential out of every term, it never vanishes; study the polynomial

    Example: (x3−6x2+6x)e−x=x(x2−6x+6)e−x(x^3 - 6x^2 + 6x)e^{-x} = x(x^2 - 6x + 6)e^{-x}

  • If a logistic A1+Be−kt\frac{A}{1 + Be^{-kt}} → multiply by ekte^{kt} to get Aektekt+B\frac{Ae^{kt}}{e^{kt} + B}, then quotient rule and cancel

    Example: inflection when ekt=Be^{kt} = B, at N=A2N = \frac{A}{2}

If no branch applies, the numerator is a polynomial: factor it (common factor, then roots). Never expand a power of the denominator.

How the answer is expected to be written

A marker ticks steps. Here they are in order, with the concluding sentence expected word for word.

Concavity and points of inflection

When to use it: “Find the intervals of concavity and the points of inflection of ff”

  1. 1 State the domain of ff.
  2. 2 Compute f′′f'' and write it FACTORED, on its own line.
  3. 3 List the candidates: zeros of f′′f'' and points of the domain where f′′f'' does not exist; reject those outside the domain.
  4. 4 Sign of each factor on each interval, then the sign of f′′f'', in a table.
  5. 5 Conclude interval by interval, then give each point of inflection as a POINT (c,f(c))(c, f(c)), checking that the graph has a tangent there.

Concluding sentence

“Since f′′f'' changes from negative to positive at x=−6x = -6, which is in the domain, the graph is concave down on (−∞,−6)(-\infty, -6), concave up on (−6,0)(-6, 0), and (−6,−18)\left(-6, -\frac{1}{8}\right) is a point of inflection.”

The trap: Stopping at the x-values: the question asks for points, with both coordinates.

Marking: Typically 1 mark for the factored second derivative, 1 for the candidates, 1 for the sign table, 1 for the intervals and points.

Check before you hand in

Five minutes of checking recover more marks than one more problem started in a hurry.

The typical problem, taken apart

A logarithm squared: a one-sided domain, L'Hôpital twice, two extrema and one inflection

Sketch the graph of f(x)=x(ln⁡x)2f(x) = x(\ln x)^2, giving its extrema and its point of inflection exactly.

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

Step 1

Domain (0,∞)(0, \infty). f≥0f \ge 0, and f(x)=0f(x) = 0 only at x=1x = 1. At 0+0^+ the form is 0⋅∞0 \cdot \infty: write (ln⁡x)21/x\frac{(\ln x)^2}{1/x}, form ∞∞\frac{\infty}{\infty}; L'Hôpital gives 2ln⁡x⋅1x−1/x2=−2xln⁡x\frac{2\ln x \cdot \frac{1}{x}}{-1/x^2} = -2x\ln x, again 0⋅∞0 \cdot \infty; write −2ln⁡x1/x\frac{-2\ln x}{1/x} and apply it once more: −2/x−1/x2=2x→0\frac{-2/x}{-1/x^2} = 2x \to 0. At ∞\infty, f→∞f \to \infty.

Why

The limit at 0+0^+ tells where the curve starts: at the origin, as a hole, since 00 is not in the domain. The form is named before EACH use of the rule, which is where the marks are.

Step 2

f′(x)=(ln⁡x)2+x⋅2ln⁡x⋅1x=(ln⁡x)2+2ln⁡x=ln⁡x (ln⁡x+2)f'(x) = (\ln x)^2 + x \cdot 2\ln x \cdot \frac{1}{x} = (\ln x)^2 + 2\ln x = \ln x\,(\ln x + 2). Zeros: x=1x = 1 and x=e−2x = e^{-2}. With u=ln⁡xu = \ln x, the sign is that of u(u+2)u(u + 2): positive on (0,e−2)(0, e^{-2}), negative on (e−2,1)(e^{-2}, 1), positive after 11.

Why

Factoring out ln⁡x\ln x turns the sign study into that of a quadratic in ln⁡x\ln x, read at a glance because ln⁡\ln is increasing.

Step 3

Local maximum f(e−2)=e−2(−2)2=4e2≈0.5413f(e^{-2}) = e^{-2}(-2)^2 = \frac{4}{e^2} \approx 0.5413; local minimum f(1)=0f(1) = 0, absolute since f≥0f \ge 0 everywhere.

Why

The laws of logarithms do the evaluation: ln⁡e−2=−2\ln e^{-2} = -2. The local maximum is NOT absolute: f→∞f \to \infty as x→∞x \to \infty.

Step 4

f′′(x)=2ln⁡x⋅1x+2x=2(ln⁡x+1)xf''(x) = 2\ln x \cdot \frac{1}{x} + \frac{2}{x} = \frac{2(\ln x + 1)}{x}. On the domain x>0x > 0, so f′′f'' has the sign of ln⁡x+1\ln x + 1: negative on (0,e−1)(0, e^{-1}) (concave down), positive after (concave up). Point of inflection (1e,1e)≈(0.3679,0.3679)\left(\frac{1}{e}, \frac{1}{e}\right) \approx (0.3679, 0.3679), since f(1e)=1e(−1)2f\left(\frac{1}{e}\right) = \frac{1}{e}(-1)^2.

Why

The candidate is in the domain and f′′f'' changes sign there: both conditions are written, not assumed. The positive factor 1x\frac{1}{x} is set aside before reading the sign.

Step 5

Table and sketch: from the hole at the origin the curve rises, concave down, to (e−2,4e2)\left(e^{-2}, \frac{4}{e^2}\right), falls, bends at (1e,1e)\left(\frac{1}{e}, \frac{1}{e}\right), touches the axis at its minimum (1,0)(1, 0) and rises without bound, concave up. Range [0,∞)[0, \infty).

0.20.40.60.811.21.41.61.82-0.20.20.40.60.811.2maxinflectionmin (1, 0)

Why

The inflection point sits between the maximum and the minimum: a curve that turns from a cap into a cup must change its bending in between, which the table confirms.

The conclusion, written out

“ff has a local maximum 4e2\frac{4}{e^2} at x=e−2x = e^{-2}, an absolute minimum 00 at x=1x = 1, is concave down on (0,1e)\left(0, \frac{1}{e}\right) and concave up on (1e,∞)\left(\frac{1}{e}, \infty\right), with point of inflection (1e,1e)\left(\frac{1}{e}, \frac{1}{e}\right).”

The classic mistake on this problem: Writing f′(x)=2ln⁡xf'(x) = 2\ln x (differentiating only one factor, or treating (ln⁡x)2(\ln x)^2 as ln⁡(x2)\ln(x^2)), or calling 4e2\frac{4}{e^2} the absolute maximum of a function that tends to infinity.

Learn by heart

  • • Concave up: f′′>0f'' > 0, f′f' increasing. It does NOT mean ff increasing.
  • • Candidates for inflection: f′′=0f'' = 0 OR f′′f'' undefined, at a point of the DOMAIN; kept only if f′′f'' changes sign.
  • • A point of inflection is a POINT (c,f(c))(c, f(c)), with a tangent line, possibly vertical.
  • • Second Derivative Test: f′′(c)<0f''(c) < 0 max, f′′(c)>0f''(c) > 0 min, f′′(c)=0f''(c) = 0 no information.
  • • Quotient rule twice: pull out the common factor, cancel one power, never expand the denominator.
  • • Fractional powers: factor out the smallest power; x4/3≥0x^{4/3} \ge 0, x5/3x^{5/3} has the sign of xx.
  • • An inflection point of ff is an extremum of f′f': steepest point, point of diminishing returns, peak of new cases.

Frequently asked questions

How do you find the points of inflection of a function?

Compute the second derivative and factor it completely. List the candidates, the zeros of the second derivative and the points of the domain where it does not exist. Study the sign of each factor on each interval. A candidate is a point of inflection only if the sign changes there and the graph has a tangent line. Give the answer as a point, with both coordinates.

Is every zero of the second derivative a point of inflection?

No. A zero of the second derivative is only a candidate. For x to the fourth plus 4x, the second derivative 12x squared vanishes at 0 but stays positive on both sides, so the graph is a cup with no inflection. Conversely, x plus the cube root of x has an inflection at the origin where the second derivative does not even exist.

What does it mean when the second derivative test is inconclusive?

When the first and second derivatives are both zero at a critical number, the second derivative test gives no information: the function may have a maximum, a minimum or neither there. It does not mean there is no extremum. Go back to the first derivative test and read the sign of the first derivative on both sides of the point.

How do I simplify the second derivative of a quotient?

After the quotient rule, the numerator contains a factor of the denominator, because the derivative of a squared or cubed denominator keeps one copy of its base. Factor that base out of the whole numerator and cancel one power before expanding anything. For example the second derivative of x over x squared plus 12 becomes 2x times x squared minus 36, over x squared plus 12 cubed.

Practise it

Corrected exercises: Concavity and curve sketching, MATH 203 at Concordia

A method is proved on a paper, not on a sheet. The set for the same chapter takes each of these traps into a problem, with the solution written out step by step.

  • 10 corrected exercises
  • 100 points
  • 150 minutes
Do the exercises
Previous sheet Monotonic functions and the First Derivative Test Next sheet Applied optimization

© Ahmed Squalli Houssaini. Revision sheet published at www.letuteurscientifique.ca/en/fiches/math203-concavity-curve-sketching. Free for personal and classroom use; republishing it elsewhere requires written permission (legal notice).

See also

Looking for a MATH 203 tutor in Montreal?

Get in touch for a first session. Curve sketching is the long question of the MATH 203 final, and most of its marks go to the algebra that makes the sign table readable.

Site by Studio Squalli