Math 191

Supplement: Linear Approximation

Linear Approximation

Introduction

By now we have seen many examples in which we determined the tangent line to the graph of a
function f (x) at a point x = a. A linear approximation (or tangent line approximation) is
the simple idea of using the equation of the tangent line to approximate values of f (x) for x near
x = a.

A picture really tells the whole story here. Take a look at the figure below in which the graph of a
function f (x) is plotted along with its tangent line at x = a. Notice how, near the point of contact
(a, f (a)), the tangent line nearly coincides with the graph of f (x), while the distance between the
tangent line and graph grows as x moves away from a.
y

(a, f (a))

PSfrag replacements

x
a

Figure 1: Graph of f (x) with tangent line at x = a

In other words, for a given value of x close to a, the difference between the corresponding y value
on the graph of f (x) and the y value on the tangent line is very small.

c Glen Pugh 2005

1
Math 191
Supplement: Linear Approximation

The Linear Approximation Formula

Translating our observations about graphs into practical formulas is easy. The tangent line in
Figure 1 has slope f 0 (a) and passes through the point (a, f (a)), and so using the point-slope formula
y − y0 = m(x − x0 ), the equation of the tangent line can be expressed

y − f (a) = f 0 (a)(x − a),

or equivalently, isolating y,
y = f (a) + f 0 (a)(x − a) .
(Observe how this last equation gives us a new simple and efficient formula for the equation of the
tangent line.) Again, the idea in linear approximation is to approximate the y values on the graph
y = f (x) with the y values of the tangent line y = f (a) + f 0 (a)(x − a), so long as x is not too far
away from a. That is,

for x near a, f (x) ≈ f (a) + f 0 (a)(x − a) . (1)

Equation (1) is called the linear approximation (or tangent line approximation) of f (x) at x = a.
(Instead of “at”, some books use “about”, or “near”, but it means the same thing.)

Notice how we use “≈” instead of “=” to indicate that f (x) is being approximated. Also notice
that if we set x = a in Equation (1) we get true equality, which makes sense since the graphs of
f (x) and the tangent line coincide at x = a.

A Simple Example
√
Let’s look at a simple example: consider the function f (x) = x. The tangent line to f (x) at x = 1
is y = x/2 + 1/2 (so here a = 1 is the x value at which we are finding the tangent line.) This is
actually the function and tangent line plotted in Figure 1. So here, for x near x = 1,
√ x 1
x≈ + .
2 2

√
To see how well the approximation works, let’s approximate 1.1:
√ 1.1 1
1.1 ≈ +
2 2
= 1.05
√ .
Using a calculator, we find 1.1 = 1.0488 to four decimal places. So our approximation has an error
of about 0.1%; not bad considering the simplicity of the calculation in the linear approximation!

c Glen Pugh 2005

2
Math 191
Supplement: Linear Approximation

On the other hand, if we try to use the same linear approximation

√ for an x value far from x = 1,
the results are not so good. For example, let’s approximate 0.25:
√ 0.25 1
0.25 ≈ +
2 2
= 0.625
√
The exact value is 0.25 = 0.5, so our approximation has an error of 25%, a pretty poor approxi-
mation.

More Examples

Example 1: Find the linear approximation of f (x) = x sin (πx2 ) about x = 2. Use the approxi-
mation to estimate f (1.99).

Solution: Here a = 2 so we need f (2) and f 0 (2):

f (2) = 2 sin (4π) = 0,

while
f 0 (x) = sin (πx2 ) + x cos (πx2 ) 2πx ,
so that
f 0 (2) = sin (4π) + 8π cos (4π) = 8π .
Therefore the linear approximation is

f (x) ≈ f (2) + f 0 (2)(x − 2) ,

i.e.
for x near 2, x sin (πx2 ) ≈ 8π(x − 2) .

Using this to estimate f (1.99), we find

.
f (1.99) ≈ 8π(1.99 − 2) = −0.08π = −0.251
.
to three decimals. (Checking with a calculator we find f (1.99) = −0.248 to three decimals.)

√
3
Example 2: Use a tangent line approximation to estimate 28 to 4 decimal places.

Solution: In this example we must come up with the appropriate √ function and point at which to
find the equation of the tangent line. Since we wish to estimate 3 28, f (x) = x1/3 . For the a-value

c Glen Pugh 2005

3
Math 191
Supplement: Linear Approximation

in Equation (1) we ask: at what value of x near 28 do we know f (x) exactly? Answer: x = 27,
which is a perfect cube.

Thus, using f (x) = x1/3 we find f 0 (x) = (1/3)x−2/3 , so that f (27) = 3 and f 0 (27) = 1/27. The
linear approximation formula is then
f (x) ≈ f (27) + f 0 (27)(x − 27) ,
i.e., for x near 27,
1
x1/3 ≈ 3 + (x − 27) .
27
√
3
Using this to approximate 28 we find
√
3 1
28 ≈ 3 + (28 − 27)
27
82
=
27
.
= 3.0370

√
3 .
A calculator check gives 28 = 3.0366 to 4 decimals.

Example 3: Consider the implicit function defined by

3(x2 + y 2 )2 = 100xy .

Use a tangent line approximation at the point (3, 1) to estimate the value of y when x = 3.1.

Solution: Even though y is defined implicitly as a function of x here, the tangent line to the graph
of 3(x2 + y 2 )2 = 100xy at (3, 1) can easily be found and used to estimate y for x near 3.

First, find y 0 . Differentiating both sides of 3(x2 + y 2 )2 = 100xy with respect to x gives
6(x2 + y 2 )(2x + 2yy 0) = 100y + 100xy 0 .
Now substitute (x, y) = (3, 1):
6(9 + 1)(6 + 2y 0 ) = 100 + 300y 0
which yields y 0 = 13/9. Thus the equation of the tangent line is
13
y−1= (x − 3), or
9
13 30
y = x− .
9 9

c Glen Pugh 2005

4
Math 191
Supplement: Linear Approximation

Thus, for points (x, y) on the graph of 3(x2 + y 2 )2 = 100xy with x near 3,
13 30
y≈ x− .
9 9
.
Setting x = 3.1 in this last equation gives y ≈ 103/90 = 1.14 to two decimals.

Exercises
1. Physicists often use the approximation sin x ≈ x for small x. Convince yourself that this is
valid by finding the linear approximation of f (x) = sin x at x = 0.
Solution For x near 0, f (x) ≈ f (0) + f 0 (0)(x − 0). Using f (x) = sin x, f (0) = sin (0) = 0 and
f 0 (0) = cos (0) = 1 we find sin x ≈ x.

2. Find the linear approximation of f (x) = x3 − x about x = 1 and use it to estimate f (0.9).
Solution For x near 1, f (x) ≈ f (1) + f 0 (1)(x − 1). Using f (x) = x3 − x, f (1) = 0 and f 0 (1) = 2 we
find f (x) ≈ 2(x − 1), so f (0.9) ≈ 2(0.9 − 1) = −0.2.

3. Use a linear approximation to estimate cos 62◦ to three decimal places. Check your estimate
using your calculator. For this problem recall the trig value of the special angles:
θ sin θ cos θ tan
√ √θ
π/3 3/2
√ 1/2
√ 3
π/4 1/ 2 1/
√ 2 1√
π/6 1/2 3/2 1/ 3
Solution Here 62◦ is close to 60◦ which is π/3 radians, and we know cos (π/3) = 1/2. Letting
f (x) = cos x, for x near π/3, f (x) ≈ f (π/3) + f 0 (π/3)(x − π/3). Since 62◦ = 62π/180 radians and
f 0 (x) = − sin x, this gives

cos 62◦ ≈ 1/2 − sin (π/3)(62π/180 − π/3)

√
= 1/2 − ( 3/2)(π/90)
.
= 0.470

√
4
4. Use a tangent line approximation to estimate 15 to three decimal places.
√ √
Solution 15 is near 16 where we know 4 16 = 2 exactly. Letting f (x) = 4 x, we have for x near 16,
√
f (x) ≈ f (16) + f 0 (16)(x − 16). That is, 4 x ≈ 2 + (1/32)(x − 16). Thus
√
4
x ≈ 2 + (1/32)(15 − 16)
= 63/32
.
= 1.969 .

c Glen Pugh 2005

5
Math 191
Supplement: Linear Approximation

5. Define y implicitly as a function of x via x2/3 + y 2/3 = 5. Use a tangent line approximation
at (8, 1) to estimate the value of y when x = 7.
Solution First find the equation of the tangent line to the curve at (8, 1) and then substitute x = 7.
Differentiating implicity with respect to x we find
2 −1/3 2 −1/3 0
x + y y =0
3 3
and substituting (x, y) = (8, 1) yields y 0 = −1/2. Thus the equation of the tangent line is
1
y = 1 − (x − 8)
2
and substituting x = 7 we find y = 3/2. That is, (7, 3/2) is the point on the tangent line. Thus the
point on the curve with x coordinate x = 7 has corresponding y coordinate y ≈ 3/2.

6. Suppose f (x) is a differentiable function whose graph passes through the points (−1, 4) and
(1, 7). The estimate f (−0.8) ≈ 5 is obtained
 using a linear approximation about x = −1.
d
(f (x))2 

Using this information, find .
dx x=−1
Solution This problem was made more difficult by adding extra information which is not needed
for the solution: the point (1, 7) plays no part. First, note that since (−1, 4) is on the graph of f (x),
f (−1) = 4. For x near −1, f (x) ≈ f (−1) + f 0 (−1)(x + 1). Using this linear approximation, the
estimate f (−0.8) ≈ 5 was made; that is

5 = 4 + f 0 (−1)(−0.8 + 1)

So that f 0 (−1) = 5. Now do the derivative, remembering the chain rule:


d 2
 
(f (x))  = 2f (x)f 0 (x)x=−1
dx x=−1
= 2(4)(5)
= 40 .

7. The profit P (q) from producing q units of goods is given by

P (q) = 396q − 2.2q 2 + k

for some constant k. Using a linear approximation about q = 80 we find P (81) ≈ 17244.
What is k?
Solution For q near 80, P (q) ≈ P (80) + P 0 (80)(q − 80). Using this approximation, P (81) ≈ 17244,
so that

17244 = P (80) + P 0 (80)(q − 80)

17244 = [396(80) − 2.2(80)2 + k] + [396 − 4.4(80)](1)

where in this last equation the first expression in square brackets is P (80) and the second expression
in square brackets is P 0 (80). Solving this last equation for k gives k = −400 (note the original
answers had k = 400 which is incorrect).

c Glen Pugh 2005

6