MATH 104 Small Class week #10 Week of November 22, 2021

1 Asymptotes.
1. True/False. State whether each statement is correct and explain why.

a. A function f (x) has a slant asymptote if and only if f 0 (x) has a horizontal asymptote.
b. A function never crosses its asymptotes.
5xn +1
c. The function f (x) = 3xm +1
has a horizontal asymptote if and only if n ≤ m.
5xn +1
d. The function f (x) = 3xm +1
has a slant asymptote if and only if n = m + 1.
e. If y = 1 is a horizontal asymptote for f (x) then limx→∞ f (x) = 1 and limx→−∞ f (x) = 1.

Solution:

a. True. If limx→+∞ f (x) = a, then the slope of tangent lines of f (x) go to a as x goes to
+∞. Therefore, f (x) will become very close to a straight line of slope a.
cos(x)
b. False. The function x
has asymptote y = 0 but crosses it infinitely many times.
5xn +1 5xn
c. True. At infinity, we have 3xm +1
≈ 3xm
= 5xn−m , which converges to a number only if
n − m ≤ 0.

d. True. Like the previous point, we have f (x) ≈ 5xn−m , so it has a slant asymptote
exactly when n − m = 1.

e. False. The function ex has a horizontal asymptote y = 0 at −∞ but does not have an
asymptote at +∞.

2. Determine the slant asymptotes at +∞ for the following functions.

6x3 + 2x2 − 1
a. f (x) = .
3x2 − 17x + 4
cos(x)
b. f (x) = 1 + x + .
x
4x17 + 6x10 + 4x7 − 13x6 + x3 + x − 4
c. f (x) = .
x16 + x15 + 3x9 + 9x3 − 17x + 120

Solution:

a. y = 2x + 12.

b. y = x + 1.

c. y = 4x − 4.
 2 Approximation
Recall that, to each differentiable function f (x), we can associate the degree n Taylor polyno-
mial of f (x) centered on the point a, namely
f 00 (a) f (n) (a)
Tn (x) = Tn,f,a (x) = f (a) + f 0 (a)(x − a) + (x − a)2 + · · · + (x − a)n ,
2! n!
which is the best approximation to the function f (x) near x = a by a polynomial of degree
≤ n.
• n = 0 : we get f (x) ≈ f (a), this is constant approximation.
• n = 1 : we get f (x) ≈ f (a) + f 0 (a)(x − a), this is linear approximation.
00 (a)
• n = 2 : we get f (x) ≈ f (a) + f 0 (a)(x − a) + f 2
(x − a)2 , this is quadratic approximation.
3. Explain how the Mean Value Theorem can be interpreted as a tool to evaluate the error in
the constant approximation.

Solution: The degree 0 Taylor polynomial centered at a is T0 (x) = f (a). So the error
is E0 (x) = f (x) − T0 (x) = f (x) − f (a). The Mean Value Theorem states that there is c
between x and a such that f (x)−f
x−a
(a)
= f 0 (c), hence E0 (x) = f 0 (c)(x − a), which is exactly
the formula for the error.
√
4. a. Find the best linear approximation to the function f (x) = 1 + x2 near x = 1.
√
b. Use your answer to part a. to estimate the value of 3 (Hint: think about which value of
x to plug into your approximation from part a.).
√
c. What is the difference between 3 and the approximate value from part b.? By what
percentage does this approximation differ from the true value?

Solution:

a. The best linear approximation is the best degree one approximation; by definition,
this is the degree one Taylor polynomial of f (x) centered on 1. Our formula for this
polynomial is
T1 (x) = f (a) + f 0 (a)(x − a),
and because f (x) is given, we need only compute f 0 (x). We have

d 1 x
f 0 (x) = (1 + x2 )1/2 = (1 + x2 )−1/2 (2x) = √ ,
dx 2 1 + x2
and so plugging in a = 1 gives
√
√ √ 0 1 1 2
f (1) = 1 + 12 = 2, f (1) = √ =√ = ,
1+1 2 2 2

and so the best linear approximation to f (x) near x = 1 is

√ √ √
√ 2 2 2
T1 (x) = 2 + (x − 1) = x+ .
2 2 2
 b. We observe that
√
r
√ 2
√ √
3= 1+2= 1+ 2 = f ( 2),
√ √
and so we will approximate the value of 3 = f ( 2) using the polynomial T1 (x) from
part a. We have
√ √ √
√ √ √ 2 √ 2 2
3 = f ( 2) ≈ T1 ( 2) = ( 2) + =1+ .
2 2 2
According to a calculator, this is 1.707106 . . ..

c. The exact value of the difference is, of course,

√ ! √
√ √ √ 2 √ 2
f ( 2) − T1 ( 2) = 3 − 1+ = 3− − 1,
2 2

which cannot be simplified; according to a calculator, its value is 0.024944 . . .. The fact
that this value
√ is positive indicates that the approximation is an underestimate for the
true value of 3.
In terms of percentage difference from the true value, we have the formula
√ √
(true value) − (approximate value) 3 − 22 − 1
× 100% = √ × 100% ≈ 1.44%,
(true value) 3
so this is a reasonably (though not terribly) accurate approximation. For completeness,
we give the first ten decimal places of each value to compare:
√ √
f ( 2) = 1.7320508075 . . . , T1 ( 2) = 1.7071067811 . . .

5. For each of the following functions f (x) and points a:

(i) Find the linear approximating polynomial centered at a,

(ii) Find the quadratic approximating polynomial centered at a.
(iii) Use the previous answers to approximate the given quantities.

a. f (x) = 8x3/2 , a = 1, approximate 8 · 1.13/2 .

b. f (x) = e−x , a = 0, approximate e−0.3 .
c. f (x) = (1 + x)−1 , a = 0, approximate 1
1.05
.

Solution: The question is asking us to compute (i)T1 (x), (ii)T2 (x), and (iii) evaluate
T1 (x) and T2 (x) at given points. We use the formula given at the top of the problem
sheets, namely
T1 (x) = f (a) + f 0 (a)(x − a)
and
f 00 (a)
T2 (x) = f (a) + f 0 (a)(x − a) + (x − a)2
2!
f 00 (a)
= T1 (x) + (x − a)2 .
2

a. (i) We compute √
f 0 (x) = (3/2)(8)x1/2 = 12 x,
and so √
f (1) = 8(1)3/2 = 8, f 0 (1) = 12 1 = 12,
thus
T1 (x) = 8 + 12(x − 1) = 12x − 4.
(ii) By our formula and our work in part (i), it suffices to compute f 00 (x). We have
d 6
f 00 (x) = 12x1/2 = 6x−1/2 = √ ,
dx x

so f 00 (1) = √6
1
= 6, and we have

6
T2 (x) = T1 (x) + (x − 1)2 = 8 + 12(x − 1) + 3(x − 1)2
2
= 3x2 + 6x − 1.

(iii) We observe that

8 · 1.13/2 = f (1.1),
and so we will approximate 8 · 1.13/2 by the values T1 (1.1) and T2 (1.1). We have

T1 (1.1) = 12(1.1) − 4 = 9.2

and
T2 (1.1) = 3(1.1)2 + 6(1.1) − 1 = 9.23,
and we are done. For comparison, the decimal expansion of 8 · 1.13/2 is 9.22952 . . .

b. (i) We proceed as in part a.:

f 0 (x) = −e−x , f (0) = e−0 = 1, f 0 (0) = −e−0 = −1,

and so
T1 (x) = 1 + (−1)(x − 0) = 1 − x.
(ii) We have
d
f 00 (x) = (−e−x ) = e−x = f (x),
dx
so
f 00 (0) = f (0) = 1
and thus
1 1
T2 (x) = T1 (x) + (x − 0)2 = 1 − x + x2 .
2 2
 (iii) Clearly e−0.3 = f (0.3), and so we compute

T1 (0.3) = 1 − (0.3) = 0.7

and
(0.3)2
T2 (0.3) = 1 − (0.3) += 0.745,
2
and compare them to the decimal expansion e−0.3 = 0.740818 . . .

c. (i) We proceed as in parts a. and b.:

−1 −1
f 0 (x) = −(1 + x)−2 = , f (0) = (1 + 0)−1 = 1, f 0 (0) = = −1,
(1 + x)2 (1 + 0)2

and so
T1 (x) = 1 + (−1)(x − 0) = 1 − x.
We observe that this is the same as the best linear approximation to e−x near
x = 0 from part b.; this is solely because the value of these functions and their
derivatives agree at x = 0.
(ii) We compute

d 2
f 00 (x) = −(1 + x)−2 = 2(1 + x)−3 =


,
dx (1 + x)3
so
2
f 00 (0) = = 2,
(1 + 0)3
hence
f 00 (0)
T2 (x) = T1 (x) + (x − 0)2 = 1 − x + x2 .
2
(iii) We have
1 1
= = f (0.05),
1.05 1 + 0.05
so we compute
T1 (0.05) = 1 − (0.05) = 0.95,
and
1 − (0.05) + (0.05)2 = 0.9525,
1
and compare them to the (repeating) decimal expansion 1.05
= 0.952380.

6. For each of the linear approximations done in the previous question, state whether you over-
approximated or under-approximated, and estimate the error of the approximation. Give an
interval as small as possible containing the real value. Use a calculator to compare with the
exact value.

Solution:
 a. We have f 00 (x) = 6x−1/2 > 0 so we are under-approximating. The maximum of f 00 (x)
on [1, 1.1] is f 00 (1) = 6, hence the error is less than 3(0.1)2 . This yields

9.2 = T1 (1.1) ≤ 8(1.1)3/2 ≤ T1 (1.1) + 0.03 = 9.23.

b. We have f 00 (x) = e−x > 0 so we are under-approximating. The maximum of f 00 (x) on

[−0.3, 0] is f 00 (0) = 1, hence the error is less than 21 (−0.3)2 = 0.045. This yields

0.7 = T1 (−0.3) ≤ e−0.3 ≤ T1 (−0.3) + 0.045 = 0.745.

c. We have f 00 (x) = 2(1 + x)−3 > 0 so we are under-approximating. The maximum of

f 00 (x) on [0, 0.05] is f 00 (0) = 2, hence the error is less than (0.05)2 = 0.0025. This yields

0.95 = T1 (0.05) ≤ (1.05)−1 ≤ T1 (0.05) + 0.0025 = 0.9525.

dy
7. Compute dx for the curve given by y 3 + 3xy = 7 at the point (2, 1), and approximate the
y-coordinate of the point where x = 2.1.

Solution: We use implicit differentiation to get 3yy 0 + 3y + 3xy 0 = 0. At the point (2, 1)
we get 3y 0 + 3 + 6y 0 = 0, so y 0 = −1
3
, so T1 (x) = 1 − 13 (x − 2). This yields T2 (2.1) = 1 − 30
1
=
0.96666....

8. For each of the following functions, find the degree n Taylor polynomial centered at 0 (for
n = 0, 1, 2), and then graph the Taylor polynomials and the function on the same axis.

a. f (x) = cos x.
b. f (x) = ln(1 − x).

Solution: Because a = 0 has been given, we will use the formulae

T0 (x) = f (0),
T1 (x) = f (0) + f 0 (0)x,
f 00 (0) 2
T2 (x) = f (0) + f 0 (0)x + x.
2!

a. We compute the first two derivatives:

f (x) = cos x, f 0 (x) = − sin x, f 00 (x) = − cos x,

and then evaluate them at the point a = 0:

f (0) = cos 0 = 1, f 0 (0) = − sin 0 = 0, f 00 (0) = − cos 0 = −1.

 It now follows immediately that

T0 (x) = 1,
T1 (x) = 1 + 0 · x = 1,
2 x2
T2 (x) = 1 + 0 · x + (−1)x = 1 − .
2

Observe that T0 = T1 .

The graphs are given below, with T0 (x) = T1 (x) in red, T2 (x) in blue, and f (x) in black.

−3 −2 −1 1 2 3
−1

−2

−3

It is clear from the picture that the approximations are only any good in a small area
around x = 0 (for instance, the value of T2 (2) is far from the value of cos(2)).

b. We proceed as in part a.:

1 1 −1
f (x) = ln(1 − x), f 0 (x) = (−1) = , f 00 (x) = −(x − 1)−2 = ,
1−x x−1 (x − 1)2

and so
1 −1
f (0) = ln 1 = 0, f 0 (0) = = −1, f 00 (0) = = −1,
0−1 (0 − 1)2

and we have

T0 (x) = 0,
T1 (x) = 0 + (−1)x = −x,
(−1) 2 x2
T2 (x) = 0 + (−1)x + x = −x − .
2 2
 We give the graphs below, with T0 in red, T1 in blue, T2 in green, and f in black.

0.5

−0.6 −0.4 −0.2 0.2 0.4 0.6

−0.5

−1

9. A bank owns a portfolio of bonds whose value P (r) depends on the interest rate r (measured
in percent, for example, r = 5 means a 5% interest rate). The bank’s quantitative analyst
determines that

P (5) = 1000 000, P 0 (5) = −400 000, P 00 (5) = 500 000.

In finance, this second derivative is called bond convexity. Find the degree 2 Taylor polynomial
of P (r) centered at r = 5, and use it to approximate the value of the portfolio if the interest
rate moves to r = 5.5%.

Solution:
T2 (r) = 1000 000 − 400 000(r − 5) + 250 000(r − 5)2 .
We get that P (5.5) ≈ T2 (5.5) = 1000 000 − 200 000 + 50 000 = 850 000.