AP Calculus BC

Lesson Set 4: Concavity and the Second Derivative

Concavity and the Second Derivative

Concavity (Mathematical Definition): For f a function differentiable on an open interval I, f is

concave upward on I if f ' is increasing on I and concave downward on I if f ' is decreasing. It is
useful to read this definition thinking of acceleration and deceleration.

Concavity (Graphical Definition): Start by drawing a few graphs: 1) y = x2, 2) y = x3, and 3) y =
x3 + 6x2 + 8x. The terms concave up and concave down are visually obvious. However, look at
what is happening to the y values in each instance for increasing values of x. This is best seen
with a table.

Concavity (Table Definition):

x

f (x)

f (x)

g (x)

g (x)

h (x)

h (x)

--

--

--

12

+7

12

+7

12

+7

21

+9

15

+3

19

+7

Looking at the -values tells us that f (x) is concave up, g (x) is concave down, and h (x)
is linear and neither concave up nor down.

Testing for Concavity

Let f be a function whose second derivative exists on an open interval I.

1) If f ''(x) > 0 for all x on I, f is concave up.
2) If f ''(x) < 0 for all x on I, f is concave down.

Finding Intervals of Concavity

1) Find f ''(x).
2) Solve f ''(x) = 0 for x.
3) Find any points of x such that f ''(x) is undefined.
4) All points from 2) and 3) are the boundaries of our intervals.
5) Pick a point on the middle of each interval.
If f ''(x) > 0, then f (x) concave op on I. If f ''(x) < 0, then f (x) concave down on I.

Examples:
1) f (x) = 6 (x2 + 3) -1.
g x
2)

x2 1
x2 4

Pick a few additional examples for students to try.

Points of Inflection

Just like turning points for first derivatives are the points where the direction changes,
points of inflection are where concavity changes direction.
We define a point of inflection as a value of x such that f (x) is defined and either f ''(x) =
0 or f ''(x) is undefined. Concavity may change at a vertical asymptote or hole, but they are not
considered points of inflection.
As with testing critical points for local extrema, we will test values on each of our
intervals to see what the concavity is for each interval.

Example: f (x) = x4 4x3.

Create a few other examples for students to try.

Second Derivative Test

Let f be a function with f '(c) = 0 and the second derivative exists on an interval containing c.
1) If f ''(c) > 0, then (c, f (c)) is a relative minimum.
2) If f ''(c) < 0, then (c, f (c)) is a relative maximum.
If f ''(c) = 0, then this test does not apply. Use the first derivative test already learned.

To illustrate why this works, draw a few pictures.

Examples: 1) y = x2, 2) y = -x2, and 3) y = x3 + 6x2 + 8x. y = x3 illustrates an exception.

Example: f (x) = -3x5 + 5x3.

Drawing Derivatives of Graphs

Now that we have learned all the characteristics of graphs and their derivatives, let's try
some sketching and matching exercises.

Optimization Problems

Optimization problems are the life blood for many calculus exams (including the AP
exam). They can both check for good differentiation and check for understanding of how to use
the derivative to solve many problems. Some of the examples below are classic exam
problems.
Despite the stigma attached to word problems and the usual blank stares, these problems
are all relatively easy. The best advice I can give is to read the problem carefully, note exactly for
what the question is asking, and note what information is being provided. A few problems
provide extraneous information and a few assume you know some trick or formula from basic
algebra or geometry. However, all give you exactly what you need. Thus, there is a clear path to
the answer.

Common Optimization Problems

Example 1:

A manufacturer wants to design an open box (no top) having a square base and a surface
area of 108 square inches. What dimensions will produce a maximum volume?
V = l w h = x2 h.

h
SA = 108 = x2 + 4xh. 108 x2 = 4xh

V = x2

108 x 2
4x

108 x x 3
4

27 x
=

x3
4

108 x 2
4x

To maximize:

dV
3x 2
27
dx
4

27
0=

3x 2
27
4

3x 2
4

36 = x2 x = 6,-6. We want 6 because negative length makes no sense.

x = 6 108 = 36 + 24h h = 3.

Dimensions: 6 x 6 x 3.

Example 2:

Which points on the graph of y = 4 x2 are closest to the point (0, 2)?

x2 x1

Distance Formula:

x2 0

y2 y1

. Note x1 = 0 and y1 = 2.
2

y2 0

.
d x2 4 x2 2

d=

d x 2 4 4 x 2 x 4 x 4 3x 2 4

d'

1
1 4
x 3x 2 4 2 4 x3 6 x
2

1
1 4
x 3 x 2 4 2 4 x3 6 x
2

0 = 4x3 6x = 2x (2x2 3)

3
2

x = 0 or

Interval

x < -(3/2)

-(3/2) < x < 0

0 < x < (3/2)

x > (3/2)

-2

-1

d (c)

4(-2)3 6 (-2) =
-20

4(-1)3 6 (-1) =
2

4 (1) 6 (1) =
-2

4 (8) 6 (2) =
20

Incr/Decr

Decr

Incr

Decr

Incr

3
2

x
and

are the x-coordinates where the function is closest to (0,2).

The points are:

3
2

3 5
,
2 2

and

3 5
,
2 2

Example 3:

A rectangular page is to have 24 square inches of print. The margins at the top of the page
are to be 1.5 inches, and the margins at the left and right are to be 1 inch. What should the
dimensions of the page be to minimize the amount of paper?

Inner Area: A = 24 = xy y = 24/x.

Outer Area: A = (x + 2)(y + 3)

A = (x + 2)(24/x + 3) = 24 + 3x + 48/x + 6 = 30 + 3x + 48/x
Minimize area: A = 0.
A = 0 = 3 48x-2.
3 = 48x-3 x2 = 16 x = 4.
y = 24/x = 24/4 = 6. We want a 6x4 area for print and a 6x9 sheet of paper.

Example 4:

Find two numbers whose product is 144 and whose sum is a minimum. These are typical
SAT problems, but the proper solution requires calculus.

xy = 144 y = 144/x.
x + y = S x + 144/x = S.
S = 1 144x-2 0 = 1 144x-2 1 = 144x-2 x2 = 144 x = 12. y = 12.
(For any two numbers, we will accept both 12 and -12. However, we should look carefully at the
wording as sometimes we only want positive numbers.)

Example 5:

Find two numbers who sum is 80 and whose product is a maximum.

x + y = 80. y = 80 x.
xy = P. P = 80x x2.
P = 0 80 2x = 0 x = 40. y = 40.

Example 6:

Folding a square piece of material (of 30 inch sides) with sides of length x, find the
length of x that maximizes the volume. (Draw a picture.)

V = (30 x)(30 x)x = (900 60x + x2)x = x3 60x2 + 900x.

V = 0 3x2 120x + 900 = 0 x2 40x + 300 = 0 (x 10)(x 30) = 0
x = 10 or x = 30. We want x = 10 because 30 x gives us a side of length 0.

Part Two

Linear Approximations

Class Activity:
Graph f (x) = 1 + sin x. (Set window for -5 to 5)
Note: f (0) = 1. This will be our starting point.

Though we have learned to use our calculators to find the value at a specific point for a
given function, it is very useful to be able to approximate this value using tangent line
approximations from known points. After all this is how many of our predecessors found them.

Before we spend time discussing tangent line approximations and finding values at
ugly points, lets zoom in on our graph above.
a) Reset the window for -1 to 1 for x and 0 to 2 for y.
b) Reset the window for -.5 to .5 for x and .5 to 1.5 for y.

Notice that for each of these steps the function looks more linear, especially the second
view.

1) Guess the slope of this line.

2) Find f (x) and f (0).
3) We will use this information to estimate the equation of this line.
Remember that y = mx + b is equivalent to y = y0 + f (x0)(x x0).
4) Plug in for x0, y0, and f (x0).
y 1 = 1 (x 0)
y = x + 1.
Now, we can use this line to approximate f (.01) and f (-.1).
Tangent Approx.: f (.01) 1.01 and f (-.1) .9.
Calculator: f (.01) = 1.0099998 and f (-.1) = .9002

Examples:
Find tangent line approximations for:
1.

y = 2x at x = 2. Estimate y (1.98).

2. y = 2ex at x = -1. Estimate y (-.9).

3. y = tan 2x at x = /8. Estimate y (.41).
4. y = sin x cos x at x = /2. Estimate y (1.5).

Differentials

Draw a picture of a curve and tangent line like on p. 222.

Tangent line: y = f (c) + f (c)(x c).

y = f (c + x) f (c).
f (c + x) f (c) + f (c) [(c + x) c] = f (c) + f (c)(x).
y f (c) + f (c)(x) f (c) = f (c)(x).
We know: dy = f (x)dx from derivatives. Letting dx = x, means that
dy y y f (c)dx.
The main idea is that y/x (slope) dy/dx.
Comparing dy and y

Let y = x2. Find dy when x = 1 and dx = x = .01. Compare this with y.

f (x) = x2 f (x) = 2x and f (1) = 2.
dy = f (x) dx = 2 (.01) = .02.
y = f (x + x) f (x) = f (1.01) f (1) = 1.0201 1 = .0201.
Obviously, dy is a good estimate for y. This means I can use the differential as a key to
making a good approximation for a value of a function at an ugly value.

Example 1: Without the aid of a calculator, estimate the value of f (x) = sin x at x = 8/7. Let c =
, and
x = /7.
f (x) = sin x f (x) = cos x f () = cos = -1.
dy = f () x = -1 (/7) y -/7 -.4488.
f (8/7) = f () + y sin () + -.4488 = -.4488.
Per calculator: sin (8/7) -.4339.

Error Propagation

One important application to differentials is error propagation, which is common in

physics.
Error propagation uses the formula f (x + x) f (x) = y, where f (x + x) is the exact value, f
(x) is the measured value, and y is the resulting error.
We will use dy = f (x)dx, dx is the error in the calculation or measurement of x and f (x)
is the rate of change of y at the measured value of x.

Example 2:
The radius of a ball bearing is measured to be .7 inch. If the measurement has a possible
error of .01 in., estimate the possible propagated error of the calculated volume of the ball
bearing.

V = (4/3)r3.
V dV = 4r2 dr (dr - possible error)
V 4(.7)2 (.01) = .062 in3.

Percentage Error

If we wish to find the percentage error in a problem, we do so by dividing

The percentage error in the previous problem would have been:

dV 4 r 2 dr 3 dr 3 .01
=
=
=
.0429
V
4 3
r
.7
r
3
Or 4.29%

f ' ( x)
f (x) .