8 Infinite Series, Improper Integrals, and Taylor Series: Mastered Material Check

8
Infinite series, improper integrals, and Taylor series
This chapter has several goals. The first is to understand how concepts that
were discussed for finite series and integrals can be meaningfully extended
to infinite series and improper integrals - i.e. integrals of functions over an
infinite domain. In this part of the discussion, we find that the notions of
convergence and divergence are critical. Mastered Material Check
A second goal is to approximate functions using power series, also called 1. Give an example of of both a finite
Taylor series. Such series can be described informally as infinite polynomi- and infinite polynomial.
als (i.e. polynomials containing infinitely many terms). 2. Circle the point of tangency
between f (x) and LA in Figure 8.1.
Understanding when these objects are meaningful is also related to the
issue of convergence, so we use the background assembled in the first part of
the chapter to address such concepts arising in the second part.
y
Figure 8.1: The function y = f (x) (solid
heavy curve) is shown together with its
LA
linear approximation (LA, dashed line)
HOA at the point x0 , and a better “higher order”
approximation (HOA, thin solid curve).
y=f(x) Notice that this better approximation stays
closer to the graph of the function near x0 .
In this chapter, we discuss how such better
approximations can be obtained.
x
x0
The theme of approximation has appeared before in our calculus courses.

In Math 154, we discussed a linear approximation to a function. The idea
was to approximate the value of the function close to a point on its graph
using a straight line (the tangent line). The approximation was good only
close to the point of tangency. Further away, the graph of the functions curves
away from that straight line. This leads naturally to the question:
• Can we make a better approximation if we include other terms to describe
this “curving away”?
Here we extend such linear approximation methods. Our goal is to increase
the accuracy of the linear approximation by including higher order terms
206 I N T E G R A L C A L C U L U S W I T H A P P L I C AT I O N S F O R L I F E S C I E N C E S
(quadratic, cubic, etc), i.e. to find a polynomial that approximates the given
function.
8.1 Convergence and divergence of series
Section 8.1 Learning goals
1. Define an infinite series that is convergent; define an infinite series that is

divergent.
2. Define the partial sum of an infinite series.
3. Give the conditions under which an infinite geometric series converges;

give the sum of that infinite geometric series.
We begin with a review. Recall the geometric series discussed in Sec-

tion 1.4.
The sum of a finite geometric series,
N
1 − r N +1
SN = 1 + r + r2 + · · · + rN = ∑ rk , is SN =
1−r
. (8.1)
k =0
Mastered Material Check
Recall the following definitions, first discussed in Section 1.5 3. Practice finding the sum of a
geometric series; simplify (compute
Definition 8.1 (Convergence of infinite series) An infinite series that has a if possible) the following:
finite sum is said to be convergent. Otherwise it is divergent. (a) ∑12 k
k =0 2 ;
1n
Definition 8.2 (Partial sums and convergence) Suppose that S is an (in- (b) ∑15
n=0 2 ;
finite) series whose terms are ak . Then the partial sums, SN , of this series (c) ∑20
k =0 ak .
are
N
SN = ∑ ak .
k =0
We say that the sum of the infinite series is S, and write
∞ N
S= ∑ ak , provided that S = lim SN = lim
N→∞ N→∞
∑ ak .
k =0 k =0
That is, we consider the infinite series as the limit of partial sums SN as the
number of terms n is increased. If this limit exists, we say that the infinite se-
ries converges to S. If the limit does not exist, we say that the series diverges.
This leads to the following conclusion:

The sum of an infinite geometric series,
∞
1
S = 1 + r + r2 + · · · + rk + · · · = ∑ rk = 1 − r , provided |r| < 1. (8.2)
k =0
If this inequality is not satisfied, then we say that this sum does not exist
(meaning that it is not finite).
I N F I N I T E S E R I E S , I M P RO P E R I N T E G R A L S , A N D TAY L O R S E R I E S 207
Remember that an infinite series, i.e. a sum with infinitely many terms
added up, can exhibit either of these two very different behaviours. It may
converge in some cases, as the first example shows, or diverge (fail to con-
verge) in other cases. We see examples of each of these trends again. It is
essential to be able to distinguish the two. Divergent series (or series that
diverge under certain conditions) must be handled with particular care, for
otherwise we may find contradictions or “seemingly reasonable” calculations
that have meaningless results.
We can think of convergence or divergence of series using a geometric
analogy. If we start on the number line at the origin and take a sequence
of steps {a0 , a1 , a2 , . . . , ak , . . . }, we can think of S = ∑∞
k=0 ak as the total
distance we have travelled. S converges if that distance remains finite and
if we approach some fixed number. In order for the sum of ‘infinitely many
“convergence” Figure 8.2: An informal schematic illus-

trating the concept of convergence and
divergence of infinite series. Here we show
only a few terms of the infinite series: from
left to right, each step is a term in the series.
In the top example, the sum of the steps gets
“divergence” closer and closer to some (finite) value. In
the bottom example, the steps lead to an ever
increasing total sum.
things’ to add up to a finite number, the terms have to get smaller. But just
getting smaller is not, in itself, enough to guarantee convergence. (We show
this later with the harmonic series.) There are rigorous mathematical tests
which help determine whether a series converges or not. We discuss some of
these tests in Appendix B.7.
8.2 Improper integrals
1. Define an improper integral of the first kind.
2. Define convergence and divergence in the context of improper integrals.
3. Explain why integrals of the form

∞
x−p dx
1
converge or diverge.
4. Use the integral comparison test to show that a given integral converges or
diverges.
There is a close connection between certain infinite series and improper

integrals, i.e. integrals over an infinite domain. We already encountered
an example of an improper integral in Section 3.6. Recall the following
definition:
Definition 8.3 An improper integral is an integral performed over an

infinite domain, e.g. ∞
f (x) dx.
a
The value of such an integral is understood to be a limit, as given in the
following definition:
∞ b
f (x) dx = lim f (x) dx
a b→∞ a
i.e. we evaluate an improper integral by first computing a definite integral
over a finite domain a ≤ x ≤ b, and then taking a limit as the endpoint b moves
off to larger and larger values. The definite integral can be interpreted as an
area under the graph of the function. The essential question being addressed
here is whether that area remains bounded when we include the “infinite
tail” of the function (i.e. as the endpoint b moves to larger values.) For some
functions (whose values get small enough fast enough) the answer is “yes”. Mastered Material Check
4. If f (x) = 0, does the improper
Definition 8.4 When the limit
integral a∞ f (x)dx converge or
b diverge? What about f (x) = 1?
lim f (x)dx
b→∞ a
exists, we say that the improper integral converges. Otherwise we say that
the improper integral diverges.
With this in mind, we compute a number of classic integrals:
Example 8.1 (A decaying exponential: convegent improper integral)

Recall that the definite integral of a decaying exponential function over a y
closed interval converges. Recapitulate this important result in the context of
our discussion of improper integrals.
Solution. Suppose that r > 0 and let

y= 1
∞ b x
I= e−rt dt ≡ lim e−rt dt.

0 b→∞ 0
Then note that b > 0 so that y= 1

x2
b x
1 −rt 1 1 1
I = lim − e lim (e−rb − e0 ) = − ( lim e−rb −1) = .
= − r b→∞
1
b→∞ r r b→∞ r
0
0 Figure 8.3: In Sections 8.2 and 8.3, we

consider two functions whose values
We have used the fact that limb→∞ e−rb
= 0 since (for r, b > 0) the exponential decrease along the x axis, f (x) = 1/x and
function is decreasing with increasing b. Thus the limit exists (is finite) and f (x) = 1/x2 . We show that one, but not the
other encloses a finite (bounded) area over
the integral converges. In fact it converges to the value I = 1/r. the interval (1, ∞). To do so, we compute
We now consider a classic and counter-intuitive result, and one of the most an improper integral for each one. The
important results in this chapter. heavy arrow is meant to remind us that we
are considering areas over an unbounded
domain.
Example 8.2 (The improper integral of 1/x diverges) Consider the func-
tion
1
y = f (x) = .
x
Show that the improper integral of this function diverges.
Solution. Examining the graph of this function for positive x, e.g. for the
interval (1, ∞), we know that values decrease to zero as x increases. The
function is not only itself bounded, but also falls to arbitrarily small values Mastered Material Check
as x increases. Nevertheless, this is insufficient to guarantee that the enclosed 5. What characteristic of the function
area remains finite! at x = 0 causes us to choose the
interval (0, ∞) instead of [0, ∞)?
We show this in the following calculation.
∞ b b
1 1
I = dx = lim dx = lim ln(x) = lim (ln(b) − ln(1))
1 x b→∞ a x b→∞ 1 b→∞
= lim ln(b) = ∞.
b→∞
The fact that we get an infinite value for this integral follows from the ob-
servation that ln(b) increases without bound as b increases, that is the limit
does not exist (is not finite). Thus the area under the curve f (x) = 1/x over
the interval 1 ≤ x < ∞ is infinite. We say that the improper integral of 1/x
diverges (or does not converge).
Example 8.3 Now consider the related function

∞
1 1
y = f (x) = , and the corresponding integral I= dx.
x2 1 x2
Does this integral converge of diverge?
Solution. Then
b b
1
I = lim x −2
dx = lim (−x −1
) = − lim − 1 = 1.
b→∞ 1 b→∞ 1 b→∞ b
Thus, the limit exists, and, in fact, I = 1, so, in contrast to Example 8.2, this
integral converges.
We observe that the behaviours of the improper integrals of the functions
1/x and 1/x2 are very different. The former diverges, while the latter con-
verges. The only difference between these functions is the power of x. As
shown in Figure 8.3, that power affects how rapidly the graph “falls off” to
zero as x increases. The function 1/x2 decreases much faster than 1/x. Con-
sequently 1/x2 has a sufficiently “slim” infinite “tail”, that the area under
its graph does not become infinite - not an easy concept to digest! This ob-
servations leads us to wonder what power p is needed to make the improper
integral of a function 1/x p converge. We answer this question below.
When does the integral of 1/x p converge?

We consider an arbitrary power, p, that can be any real number. We ask when
the corresponding improper integral converges or diverges. Let Mastered Material Check
∞ 6. Give an example of a real number
1
I= dx. and a number that is not real.
1 xp
7. Give an example of a true “if.. then”
For p = 1 we have already established that this integral diverges (Example 8.2), statement that fails when reversed.
and for p = 2 we have seen that it is convergent (Example 8.3). By a similar
calculation, we find that
b
x1−p 1 1−p
I = lim = lim b −1 .
b→∞ (1 − p) 1 b→∞ 1 − p
Thus this integral converges provided that the term b1−p does not “blow up”
as b increases. For this to be true, we require that the exponent (1 − p) should
be negative, i.e. 1 − p < 0 or p > 1. In this case, we have
1
I= .
p−1
To summarize our result,
∞
1
dx converges if p > 1, diverges if p ≤ 1.
1 xp
Example 8.4 (Does the improper integral of 1/x p converge?) Determine
whether the following improper integral converge or diverge:
∞
1
(a) √ dx,
1 x
∞
(b) x−1.01 dx.
1
Solution.
√ 1
(a) The integral diverges. We see this from the following argument: x = x2 ,
so p = 12 < 1. Thus, by the general result, this integral diverges.
(b) The integral converges. Here p = 1.01 > 1, so the result implies conver-
gence of the integral.
Integral comparison tests

We have computed some improper integral explicitly. These can be used
to make comparisons that helps us identify when other improper integrals
converge or diverge.
Example 8.5 (Comparison of improper integrals) Determine whether the

following integral converges:
∞
x
dx.
1 1 + x3
The integral comparison test Table 8.1: The integral comparison test.
These statements must be carefully noted.
Suppose we are given two functions, f (x) and g(x), both continuous on some
What is assumed and concluded works
infinite interval [a, ∞). Suppose, moreover, that, at all points on this interval “one way”. That is the order “if..then” is
important. Reversing that order leads to a
0 ≤ f (x) ≤ g(x). common error.
Then the following conclusions can be made:

∞ ∞
1. f (x) dx ≤ g(x) dx. (This means that the area under f (x) is smaller
a a
than or equal to the area under g(x).)
∞ ∞
2. If g(x) dx converges, then f (x) dx converges. (If the larger area is
a a
finite, so is the smaller one.)
∞ ∞
3. If f (x) dx diverges, then g(x) dx diverges. (If the smaller area is
a a
infinite, so is the larger one.)
Mastered Material Check

Solution. Note that for all x > 0
8. Using the integral comparison test
(and perhaps the results of
x x 1
0≤ ≤ 3 = 2. Example 8.5), determine whether
1 + x3 x x the following integrals converge:
∞
x
Thus, using the integral comparison test, (a) dx;
1 1 + x4
∞
x
∞ ∞ (b) dx.
x 1 1 2 + x3
dx ≤ dx.
1 1 + x3 1 x2
Since the larger integral on the right is known to converge, so does the
smaller integral on the left.
8.3 Comparing integrals and series
1. Describe why the harmonic series diverges.
2. Identify when the p-series converges.
The convergence of infinite series was discussed earlier, in Section 1.5

and here again in Section 8.1. Many tests for convergence are provided in
the Appendix B.7, and are not discussed in detail. However, an interesting
connection exists between convergence of series and integrals. We can use
the convergence/divergence of an integral/series to determine the behaviour
of the other. This is the topic we now examine.
The harmonic series

The harmonic series is a sum of terms of the form 1/k where k = 1, 2, . . . .
At first appearance, this series might seem to have the desired qualities of
a convergent series, simply because the successive terms being added are
getting smaller, but this is deceptive and actually wrong. Mastered Material Check
The harmonic series 9. We have already noticed a similar

surprise in connection with the
∞
1 1 1 1 1 improper integral of 1/x. These two
∑ k = 1 + 2 + 3 + 4 + ... + k + ... diverges “surprises” are closely related, as we
k =1 show here using a comparison of the
series and the integral.
We establish that the harmonic series diverges by comparing it to the 10. Give justification for why the
improper integral of the related function improper integral of
function f (x) = 1x is related to the
1 harmonic series.
y = f (x) = .
x
1
In Figure 8.4 we show on one graph a comparison of the area under this
curve, and a staircase area representing the first few terms in the harmonic 0.8
the function y = 1
x
series. For the area of the staircase, we note that the width of each step is 1,
0.6
and the heights form the sequence
0.4
1 1 1
{1, , , , . . . }. The harmonic serie
2 3 4 0.2
Thus the area of (infinitely many) of these steps can be expressed as the
(infinite) harmonic series, 2 4 6 8 10
∞
1 1 1 1 1 1 1
+ 1· + 1· + ··· = 1 + + + + ... = ∑ .
Figure 8.4: The harmonic series is a sum
A = 1·1+1· that corresponds to the area under the
2 3 4 2 3 4 k =1 k
staircase shown above. Note that we have
On the other hand, the area under the graph of the function y = f (x) = 1/x for purposely shown the stairs arranged so that
they are higher than the function. This is
0 ≤ x < ∞ is given by the improper integral essential in drawing the conclusion that the
∞ sum of the series is infinite: It is larger than
1 an area under 1/x that we already know to
dx.
1 x be infinite, by Section 8.2.
We saw in Section 8.2 that this integral diverges!

From Figure 8.4 we see that the areas under the function, A f and under the
staircase, As , satisfy
0 < A f < As .
Thus, since the smaller of the two (the improper integral) is infinite, so is
the larger (the sum of the harmonic series). We have established, using this
comparison, that the the sum of the harmonic series cannot be finite, so that
this series diverges.
The p-series
More generally, we can compare series of the form
∞ ∞
1 1
∑ kp to the integral
1 xp
dx
k =1
in precisely the same way. This leads to the conclusion that Mastered Material Check
The p-series, 11. Can you relate the p-series to
infinite geometric series as seen in
∞
1 Section 8.1?
∑ kp converges if p > 1, diverges if p ≤ 1.
k =1
Example 8.6 Does the following series converge?
∞
1 1 1 1
∑ k2 = 1 + 4 + 9 + 16 + . . .
k =1
Solution. Since p = 2 > 1, this series converges.

Notice, the comparison does not give us a value to which the sum con-
verges. It merely indicates that the series does converge.
Even though the relationship between series and improper integrals only
indicates convergence/divergence of one or the other, we can exploit it to
determine upper and lower bounds for one or the other, in the case that they
do converge.
Example 8.7 In the above convergent series we have
∞ ∞
1 1
S= ∑ k2 = 1 + ∑ k2 .
k =1 k =2
Use this to determine bounds for the series.
Solution. Interpreting the series ∑∞ 2

k=2 1/k as a Riemann sum, we find that it
∞
approximates the improper integral 1 1/x2 . More precisely, if the Riemann
sum uses the right endpoints of the subintervals to approximate the area under
the curve 1/x2 , then
∞ ∞
1 1
∑ k2 < 1 x2
dx
k =2
must hold. The improper integral easily evaluates to 1. Hence the integral Mastered Material Check
∞
provides an upper bound for S and we find that S < 2 must hold. Similarly, 12. Verify that
1
dx evaluates to 1.
2
1 x
∞ ∞ ∞ 13. Sketch a plot or use a computer to

1 1 1
∑ k2 > 1 (x + 1)2
dx =
2 x2
dx, verify that
∞ ∞
k =2 1 1
dx = dx.
1 (x + 1)2 2 x2
again assuming that the Riemann sum uses the right endpoints of the subin-
tervals to approximate the area. This time the improper integral evaluates
to 1/2, which provides a lower bound for the limit of the series and hence
S > 3/2. Thus 3/2 < S < 2 must hold. Indeed, the series actually converges
to S = π 2 /6 ≈ 1.645.
8.4 From geometric series to Taylor polynomials
1. Illustrate how a higher order approximation of a function does a better job

of approximating a function than a linear approximation.
2. Explain why applying arithmetic to series must be exercised with caution.
3. Justify why we use Taylor polynomials.
Among the most basic functions are polynomials, i.e. functions such as
p(x) = x5 + 2x2 + 3x + 2.
Our introduction to differential calculus started with such functions for

a reason: they are convenient and simple to handle. We found that it is
straightforward to compute derivatives of polynominals. The same can be
said for integrals; one of our first examples in Section 3.4 was the integral
of a polynomial. We only needed to use a power rule to integrate each term. Mastered Material Check
An additional convenience of polynomials is that “evaluating” the function, 14. What is the derivative of
(i.e. plugging in an x value and determining the corresponding y value) can be p(x) = 1 + a1 x + a2 x2 + . . . an xn ?
done by simple multiplications and additions, i.e. by basic operations easily
handled by an ordinary calculator. This is not the case for, say, trigonometric
functions, exponential functions, or for that matter, most other functions we
considered.
Note: to find the decimal value of sin(2.5) we would need a scientific cal-
culator. These days the distinction is blurred, since powerful hand-held
calculators are ubiquitous. Before such devices were available, the ease of
evaluating polynomials made them even more important.
For this reason, being able to approximate a function by a polynomial is
an attractive proposition. This forms our main concern in the sections that
follow.
We determine connections between several functions and their polyno-
mial approximations by exploiting the geometric series. We use both the
results for convergence of the geometric series (from Section 8.1) and the
formula for the sum of that series to derive a number of results. Recall from
Sections 1.5 and 8.1 that the sum of an infinite geometric series is
∞
1
S = 1 + r + r2 + · · · + rk + · · · = ∑ rk = 1 − r , provided |r| < 1. (8.3)
k =0
To connect this result to a statement about a function, we need a “variable”.

Let us consider the behaviour of this series when we vary the quantity r.
To emphasize that now r is our variable, it is helpful to change notation by
substituting r = x into the above equation, while remembering that the formula
in Eqn (8.3) holds only provided |r| = |x| < 1.
Example 8.8 (A straightforward expansion) Substitute the variable x = r

into the series (8.3). How does this help with approximating the function,
1
f (x) = ?
1−x
Solution. Formally, rewriting the above with this substitution, leads to the
conclusion that
1
= 1 + x + x2 + . . . (8.4)
1−x
We can think of this result as follows: let
1
f (x) = . (8.5)
1−x
Then for every x in −1 < x < 1, it is true that f (x) can be approximated by
terms in the polynomial
p ( x ) = 1 + x + x2 + . . . (8.6)
In other words, by Eqn. (8.3), for |x| < 1 the two expressions “are the same”, Mastered Material Check
in the sense that the polynomial converges to the value of the function. 15. Why only for −1 < x < 1 is this
We refer to this p(x) as an (infinite) Taylor polynomial or simply a Taylor true?
series for the function f (x). We illustrate a Taylor series’ utility in Exam- 16. What is the difference between a
Taylor polynomial and a Taylor
ple 8.9. series?
Example 8.9 Compute the value of the function f (x) given by Eqn. (8.5) for
x = 0.1 without using a calculator.
Solution: Plugging in the value x = 0.1 into the function directly leads to
1/(1 − 0.1) = 1/0.9, whose evaluation with no calculator requires long
division.
Note: this example is slightly “trivial”, in the sense that evaluating the
function itself is not very difficult. However, in other cases, we find that the
polynomial expansion is the only way to find the desired value.
Using the polynomial representation, we have a simpler method:
p(0.1) = 1 + 0.1 + 0.12 + · · · = 1 + 0.1 + 0.01 + · · · = 1.11 . . .
We provide a few other examples based on substitutions of various sorts

using the geometric series as a starting point.
Example 8.10 (Another substitution) Use the substitution r = −t to

determine a series expansion for the function
1
f (t ) = .
1+t
Solution. We make the substitution r = −t, then |r| < 1 means that | − t| =
|t| < 1, so that the formula Eqn. (8.3) for the sum of a geometric series implies
that:
1
= 1 + (−t ) + (−t )2 + (−t )3 + . . .
1 − (−t )
1
= 1 − t + t2 − t3 + t4 + . . . provided |t| < 1.
1+t
This is a series expansion for the function 1/(1 + t ). Mastered Material Check
We can go farther with this example by a new manipulation, whereby we 17. Give a cubic approximation to the
integrate both sides to arrive at a new function and its expansion, shown next. function
1
g(r ) = .
Example 8.11 (An expansion for the logarithm) Use the results of Ex- 1+r
ample 8.10, and integrate both sides. to arrive at an expansion for the
18. Practice sigma notation by
logarithm. expanding the sigma notation on the
right hand side of Eqn. (8.7).
Solution. On the left, we integrate the function f (t ) = 1/(1 + t ) (to arrive at
a logarithm type integral) and on the right we integrate the power terms of the
expansion. We are permitted to integrate the series term by term provided that
the series converges. This is an important restriction that we emphasize:
Important: Manipulation of infinite series by integration, differentiation,
addition, multiplication, or any other “term by term” computation makes
sense only so long as the original series converges.
Provided |t| < 1, we have that
x x
1
dt = (1 − t + t 2 − t 3 + t 4 − . . .) dt,
0 1+t 0
x2 x3 x4
ln(1 + x) = x− + − + ...
2 3 4
This procedure allowed us to find a series representation for a new function,
ln(1 + x).
∞
x2 x3 x4 xk
ln(1 + x) = x − + − + . . . = ∑ (−1)k+1 . (8.7)
2 3 4 k =1 k
The sigma notation appended on the right is just a compact notation that
represents the pattern of the terms (see Chapter 1).
An expansion for the logarithm is definitely useful, in the sense that
(without a scientific calculator or log tables) it is not possible to easily
calculate the value of this function at a given point.
Example 8.12 (Evaluating the logarithm for x = 0.25) When x = 0.25, we

cannot find ln(1 + 0.25) = ln(1.25) using simple operations. Use a three term
(cubic) series representation of ln(1 + x) to find an approximation of ln(1.25).
How does it compare to the true value?
Solution. The terms of the series are computable by only multiplication,

division, and addition:
0.252 0.253
0.25 − + ≈ 0.2239
2 3
A scientific calculator gives ln(1.25) ≈ 0.2231, so this approximation pro-

duced by the series is relatively good.
Example 8.13 When is the series for ln(1 + x) in Eqn. (8.7) expected to
converge?
Solution. Retracing our steps from the beginning of Example 8.10, we note
that the value of t is not permitted to leave the interval |t| < 1 so we need also
|x| < 1 in the integration step.
Note: strictly speaking, we should have ensured that we are inside this
interval of convergence before we computed the last example.
We certainly cannot expect the series for ln(1 + x) to converge when |x| > 1.
Indeed, for x = −1, we have ln(1 + x) = ln(0) which is undefined. Also note
that for x = −1 the right hand side of Eqn. (8.7) becomes

1 1 1
− 1 + + + + ... .
2 3 4
This is the recognizable harmonic series (multiplied by -1). But we already

know from Section 8.3 that the harmonic series diverges. Thus, we must
avoid x = −1, since the expansion will not converge there, and neither is
the function defined. This example illustrates that outside the interval of
convergence, the series and the function become “meaningless”.
Example 8.14 Find a series expansion for arctan.
Solution. Suppose we make the substitution r = −t 2 into the geometric series

formula, and recall that we need |r| < 1 for convergence. Then
1
= 1 + (−t 2 ) + (−t 2 )2 + (−t 2 )3 + . . .
1 − (−t 2 )
∞
1
1 + t2
= 1 − t2 + t4 − t6 + t8 + . . . = ∑ (−1)nt 2n .
k =0
This series will converge provided |t| < 1. Mastered Material Check
Now integrate both sides, and recall that the antiderivative of the function 19. Consider Example 8.14; find a linear
1/(1 + t 2 ) is arctan(t ). Then approximation of arctan(0.5). How
good is it?
x x
1 20. Consider Example 8.14; find a cubic
dt = (1 − t 2 + t 4 − t 6 + t 8 + . . .) dt approximation of arctan(0.5). How
0 1 + t2 0
good is it?
∞
x3 x5 x7 x(2k−1)
arctan(x) = x− + − + . . . = ∑ (−1)k+1 . (8.8)
3 5 7 k =1 (2k − 1)
Example 8.15 (An expansion of π) Use the expansion of arctan(x) to

determine an expansion of π.
Solution. Consider substituting in x = 1 into Eqn. (8.8). Then
1 1 1
arctan(1) = 1 − + − + ...
3 5 7
But arctan(1) = π/4. Thus we have found a way of computing the Mastered Material Check
irrational number π, namely 21. Draw a triangle to illustrate
that arctan(1) = π/4.

∞ 22. Illustrate the expansion of π given in
1 1 1 1
π = 4 1 − + − + . . . = 4 ∑ (−1) k +1
. Example 8.15 by adding up the
3 5 7 k =1 (2k − 1) first 100 or 1000 terms of this series
with a spreadsheet.

While it is true that this series converges, the convergence is slow. This
means that it is not practical to use such a series as an approximation for π.
(There are other series that converge to π very rapidly that are used in any
practical application.)
8.5 Taylor Series: a systematic approach
1. Define the nth degree Taylor polynomial and the Taylor series of a func-
tion f (x) centred at a point x = a.
2. Describe the usefulness of Taylor series.
3. Give the power series expansions for 1

1−x , ln(1 − x), ex , sin(x), cos(x), and
arctan(x).
In Section 8.4, we found a variety of Taylor series expansions directly

from the formula for a geometric series. Here we ask how such Taylor series
can be constructed more systematically, if we are given a function that we
want to approximate 1 . 1
The development of this section was
motivated by online notes by David Austin.
Suppose we have a function which we want to represent by a series,
∞
f (x) = a0 + a1 x + a2 x2 + a3 x3 + . . . = ∑ ak x k .
k =0
Here we use the function to directly determine the coefficients ak . To deter-

mine a0 , let x = 0 and note that
f (0) = a0 + a1 0 + a2 02 + a3 03 + . . . = a0 .
We conclude that
a0 = f (0).
By differentiating both sides we find the following:
f (x) = a1 + 2a2 x + 3a3 x2 + . . . + kak xk−1 + . . .
f (x) = 2a2 + 2 · 3a3 x + . . . + (k − 1)kak xk−2 + . . .
f (x) = 2 · 3a3 + . . . + (k − 2)(k − 1)kak xk−3 + . . .
f (k ) ( x ) = 1 · 2 · 3 · 4 . . . kak + . . .
Here we have used the notation f (k) (x) to denote the k’th derivative of the
function. Now evaluate each of the above derivatives at x = 0. Then
f ( 0 ) = a1 , ⇒ a1 = f (0),
f (0)
f (0) = 2a2 , ⇒ a2 = 2 ,
f (0)
f (0) = 2 · 3a3 , ⇒ a3 = 2·3 ,
f (k ) (0)
f (k) (0) = k!ak , ⇒ ak = k! .
This gives us a technique for calculating all coefficients ak . This means that
if we can compute all the derivatives of the function f (x), then we know the
coefficients of the Taylor series as well. Because we have evaluated all the
coefficients by the substitution x = 0, we say that the resulting power series is
the Taylor series of the function about x = 0 (sometimes called the Maclaurin
series).
Taylor series for the exponential function, ex

Consider the function f (x) = ex . All derivatives of this function are equal to
ex . In particular,
f (k) (x) = ex ⇒ f (k) (0) = 1.
Thus the coefficients of the Taylor series are
f (k ) ( 0 ) 1
ak = = .
k! k!
Therefore the Taylor series for ex about x = 0 is
∞ k
x2 x3 x4 xk x
1+x+ + + + ... + + ... = ∑ .
2 6 24 k! k=0 k!
This is a very interesting series. We state here without proof that this series
converges for all values of x. Further, the function defined by the series is in
fact equal to ex that is,
∞ k
x2 x3 x
ex = 1 + x + + + ... = ∑ .
2 6 k=0 k!
The implication is that the function ex is completely determined (for all

x values) by its behaviour (i.e. derivatives of all orders) at x = 0. In other
words, the value of the function at x = 1, 000, 000 is determined by the
behaviour of the function around x = 0. This means that ex is a very special
function with superior “predictable features”. If a function f (x) agrees with
its Taylor polynomial on a region (−a, a), as was the case here, we say that f
is analytic on this region. It is known that ex is analytic for all x.
Note: we can use the results of this Section 8.5 to establish the fact that the
exponential function grows “faster” than any power function xn . That is the
same as saying that the ratio of ex to xn (for any power n) increases with x. Mastered Material Check
We can also obtain a Taylor series expansion for functions related to ex , 23. Use the Taylor series expansion
without assembling the derivatives. We start with the result that given to find a quadratic
2
approximation to e0.1 . How close
∞ are you?
u2 u3 uk
eu = 1 + u + + + ... = ∑ .
2 6 k=0 k!
Then, for example, the substitution u = x2 leads to
( x2 ) 2 ( x2 ) 3 ∞
( x2 ) k 2
+ ... = ∑
2
ex = 1 + x2 + + . T1 T5
2 6 k=0 k! 1
Taylor series of sin(x)

1 2 3 4 5 6 7
In this example we determine the Taylor series for the sine function, f (x) = −1 sin(x)
sin(x). T3 T7
The function and its derivatives are −2
f (x) = sin x, f (x) = cos x, f (x) = − sin x, f (x) = − cos x, f (4) (x) = sin x, . . . Figure 8.5: An approximation of the
function y = sin(x) by successive Taylor
polynomials, T1 , T3 , T5 , T7 . The higher
After this, the cycle repeats. This means that
Taylor polynomials do a better job of
approximating the function on a larger
f (0) = 0, f (0) = 1, f (0) = 0, f (0) = −1, . . . interval about x = 0.
and so on in a cyclic fashion. In other words,

1 1
a0 = 0, a1 = 1, a2 = 0, a3 = − , a4 = 0, a5 = , . . .
3! 5!
Thus,
∞
x3 x5 x7 x2n+1
sin x = x − + − + . . . = ∑ (−1)n .
3! 5! 7! n=0 (2n + 1)!
We state here without proof that the function sin(x) is analytic, so that the
expansion converges to the function for all x.
Taylor series of cos(x)

The Taylor series for cos(x) could be found by a similar sequence of steps.
But in this case, this is unnecessary: we already know the expansion for
sin(x), so we can find the Taylor series for cos(x) through differentiation
term by term. Since sin(x) is analytic, this is permitted for all x. We leave as
an exercise for the reader to show that
∞
x2 x4 x6 x2n
cos(x) = 1 − + − + . . . = ∑ (−1)n . (8.9)
2 4! 6! n=0 (2n)!
Since cos(x) has symmetry properties of an even function, we find that its Mastered Material Check
Taylor series is composed of even powers of x only. 24. Verify the Taylor series expansion
of cos(x) given in Eqn. (8.9).
8.6 Taylor polynomials as approximations
1. Given two Taylor approximations, identify which is better.
2. Describe why adding terms to a Taylor series expansion leads to an

improved approximation.
We demonstrate how successive terms in a Taylor series expansion lead to

approximations that improve.
Here we illustrate this with the expansion for sin(x) obtained above.
Consider the sequence of polynomials
T1 (x) = x,
x3
T3 (x) = x− ,
3!
x3 x5
T5 (x) = x− + ,
3! 5!
x3 x5 x7
T7 (x) = x− + − ,
3! 5! 7!
where the index n indicates the order of the polynomial Tn (x) (highest power
of x). These polynomials provide a better and better approximation to the
function sin(x) close to x = 0. The first of these is just a linear (or tangent
line) approximation. The second improves this with a quadratic (in this case
cubic!) approximation, etc. Figure 8.5 illustrates how the first few Taylor
polynomials approximate the function sin(x) near x = 0. Observe that as we
keep more terms, n, in the polynomial Tn (x), the approximating curve “hugs”
the graph of sin(x) over a longer and longer range.
Example 8.16 (The error in successive approximations) For a given value

of x close to the base point (at x = 0), the error in the approximation between
the polynomials and the function is the vertical distance between the graphs
of the polynomial and the function sin(x) (shown in black). For example, at
x = 2 radians sin(2) = 0.9093 (as found on a scientific calculator). Find, and
comment on approximations T1 (2), T3 (2), T5 (2) and T7 (2).
Solution. The approximations are: T1 (2) = 2, which is very inaccurate;

T3 (2) = 2 − 23 /3! ≈ 0.667 which is better, but too small; T5 (2) ≈ 0.9333
is much closer and T7 (2) ≈ .9079 that is closer still. In general, we can
approximate the size of the error using the next term that would occur in the
polynomial if we kept a higher order expansion. The details of estimating
such errors is omitted from our discussion. Other courses
Taylor series are further examined in
MACM 316: Numerical Analysis.
8.7 Application of Taylor series
1. Use Taylor series to evaluate integrals.
2. Use Taylor series to evaluate differential equations.
In this section we illustrate some of the applications of Taylor series to

problems that may be difficult to solve using other conventional methods.
Some functions do not have an antiderivative that can be expressed in terms
of other simple functions. Integrating these functions can be a problem, as
we cannot use the Fundamental Theorem of Calculus specifies. However, In
some cases, we can approximate the value of the definite integral using a Tay-
lor series. Another application of Taylor series is to compute an approximate
solution to a differential equation.
Example 8.17 (Using a Taylor series to evaluate an integral) Evaluate the

definite integral
1
sin(x2 ) dx.
0
Solution. A simple substitution (e.g. u = x2 ) does not work here, and we Mastered Material Check
cannot find an antiderivative. Here is how we might use Taylor series: we 25. Consider Example 8.17; why does
know that the series expansion for sin(t ) is the simple substitution u = x2 not
work?
t3 t5 t7
sint = t − + − + ...
3! 5! 7!
Substituting t = x2 , we have
x6 x10 x14
sin(x2 ) = x2 − + − + ...
3! 5! 7!
In spite of the fact that we cannot antidifferentiate the function, we can

antidifferentiate the Taylor series, just as we would a polynomial:
1 1
x6 x10 x14
sin(x2 ) dx = ( x2 − + − + . . .) dx
0 0 3! 5! 7!
1
x3 x7 x11 x15
= − + − + . . .
3 7 · 3! 11 · 5! 15 · 7! 0
1 1 1 1
= − + − + ...
3 7 · 3! 11 · 5! 15 · 7!
This is an alternating series with terms decreasing in magnitude which is

enough to determine that it converges. Adding up the first four terms, the
pattern becomes clear: the series converges to 0.31026.
Example 8.18 (Series solution of a differential equation) We are already

familiar with the differential equation and initial condition that describes
unlimited exponential growth:
dy
= y, y(0) = 1.
dx
From Chapter 7, we know that the solution of this differential equation and
initial condition is y(x) = ex . Let us pretend we do not know this fact in
illustrating the usefulness of Taylor series.
Solution. In some cases, where separation of variables does not work, the
Taylor series option has great practical value. Let us express the “unknown”
solution to the differential equation as
y = a0 + a1 x + a2 x2 + a3 x3 + a4 x4 + . . .
Our task is to determine values for the coefficients ai .

Since this function satisfies the condition y(0) = 1, we must have y(0) =
a0 = 1. Differentiating this power series leads to
dy
= a1 + 2a2 x + 3a3 x2 + 4a4 x3 + . . .
dx
dy
But according to the differential equation, dx = y. Thus, it must be true that
the two Taylor series match, i.e.
a0 + a1 x + a2 x2 + a3 x3 + a4 x4 + . . . = a1 + 2a2 x + 3a3 x2 + 4a4 x3 + . . .
This equality holds for all values of x. This can only happen if the coefficients
of like terms are the same, i.e. if the constant terms on either side of the
equation are equal, if the terms of the form Cx2 on either side are equal, and
so on for all powers of x. Equating coefficients, we obtain:
a0 = a1 = 1, ⇒ a1 = 1,
a1 = 2a2 , ⇒ a2 = a21 = 12 ,
a2
a2 = 3a3 , ⇒ a3 = 3 = 1
2·3 ,
a3
a3 = 4a4 , ⇒ a4 = 4 = 1
2·3·4 ,
an−1
an−1 = nan , ⇒ an = n = 1
1·2·3...n = 1
n! .
This means that
x2 x3 xn
y = 1+x+ + + . . . + + . . . = ex ,
2! 3! n!
which, as we have seen, is the expansion for the exponential function. This
agrees with the solution we expected. In this example, we would hardly need
to use series to arrive at the right conclusion. However, another example
discussed in the optional material in Appendix B.9, where we would not find
it as easy to discover the solution by other techniques discussed previously.
Summary
The main points of this chapter can be summarized as follows:
1. We reviewed the definition of an improper integral

∞ b
f (x) dx = lim f (x) dx.
a b→∞ a
2. We computed some examples of improper integrals and discussed their

convergence or divergence. We recalled (from earlier chapters) that
∞
I= e−rt dt converges,
0
whereas ∞
1
I= dx diverges.
1 x
3. More generally, we showed that
∞
1
dx converges if p > 1, diverges if p ≤ 1.
1 xp
4. Using a comparison between integrals and series we showed that the
harmonic series,
∞
1 1 1 1 1
∑ k = 1 + 2 + 3 + 4 + ... + k + ... diverges.
k =1
5. More generally, our results led to the conclusion that the “p” series,
∞
1
∑ kp converges if p > 1, diverges if p ≤ 1.
k =1
6. We studied Taylor series and showed that some can be found using the
formula for convergent geometric series. Two examples of Taylor series
that were obtained in this way are
x2 x3 x4
ln(1 + x) = x − + − + ... for |x| < 1,
2 3 4
and
x3 x5 x7
arctan(x) = x − + − + ... for |x| < 1.
3 5 7
7. In discussing Taylor series, we considered some of the following ques-

tions: (a) For what range of values of x can we expect the series to con-
verge? (b) Suppose we approximate the function on the right by a finite
number of terms on the left. How good is that approximation? Another
interesting question is: (c) If we include more and more such terms, does
that approximation get better and better? (i.e., does the series converge to
the function?) (d) Is the convergence rate rapid? Some of these questions
occupy the attention of career mathematicians, and are beyond the scope
of this introductory calculus course.
8. More generally, we showed that the Taylor series for a function about x = 0
∞
f (x) = a0 + a1 x + a2 x2 + a3 x3 + . . . = ∑ ak x k
k =0
can be found by computing the coefficients
f (k ) ( 0 )
ak = .
k!
9. We discussed some of the applications of Taylor series. We used Taylor

series to approximate a function, to find an approximation for a definite
integral of a function, and to solve a differential equation.
Quick Concept Checks
1. Match plots (i), (ii) and (ii) with the infinite geometric series (a), (b) and (c) below;
3 Sn
20 Sn 60 Sn ∞
15
(a) ∑ 1k
2 k =0
40
∞
10 1k
1 20
(b) ∑2
5 k =0
n ∞
n n
5 10 15 20 5 10 15 20 5 10 15 20 (c) ∑ 1.1k
(i) (ii) (iii) k =0
2. Using the integral comparison test, determine whether or not the following integral converges:
∞
1
dx.
1 x−1
3. Using the Taylor series found in this chapter, which of the following can be approximated? Consider convergence.
(a) ln(0.75);
(b) arctan(−0.2);
(c) sin(9.5);
(d) e−5 ;
(e) ln(2.4);
4. Determine the Taylor series for cos(x) + sin(x2 ).
Exercises
8.1. Convergence of sequences. Determine which of the following

sequences converge or diverge
n
2
(a) {en } (b) {2−n } (c) {ne−2n } (d) (e) (f) {ln(n)}.
n 2
8.2. Convergence of series. Which of the following series converge and

which diverge? Give a reason.
∞ ∞ ∞
3
(a) ∑ 1−n (b) ∑ 3−n (c) ∑n
n=0 n=0 n=1
∞ ∞ ∞ 2
3
(d) ∑ 0.1 n
(e) ∑ 0.1 −n
(f) ∑ n
.
n=0 n=0 n=1
8.3. Comparing integral and series. Using the spreadsheet or a hand-

drawn sketch, compare the areas represented by the following integral
and series. ∞ ∞
1 1
1 x
2
dx, ∑ 2
.
k =1 k
Draw a sketch such that the convergence of the integral implies

convergence of the series. (Hint: what kind of comparison should be
set up? This will determine how you sketch the bar graph representing
the series alongside the function y = f (x) = 1/x2 .)
8.4. Alternating harmonic series. Use the spreadsheet to show the first
100 terms in the alternating harmonic series,
100
1 1 1 1
S= ∑ (−1)k+1 k = 1 − 2 + 3 − 4 + . . . .
k =1
(You can use a bar graph to represent these terms.) On the same graph,
show the first 100 “partial sums”, i.e. the graph of
n
1
Sn = ∑ (−1)k+1 k .
k =1
To what value does this alternating harmonic series converge?

8.5. Converging to π. Use the spreadsheet to show the first 100 terms in
the series,
100
1 1 1 1
S= ∑ (−1)k+1 2k − 1 = 1 − 3 + 5 − 7 + . . . .
k =1
(You can use a bar graph to represent these terms.) On the same graph,
show the first 100 “partial sums” multiplied by 4, i.e. the graph of

N
1
4 SN = 4 ∑ (−1)k+1 .
k =1 2k − 1
Show that this converges to π. Find how many correct decimal places
are achieved by adding up the first one-thousand terms of this series.
8.6. Growing tree trunk. The trunk of a tree gets wider by adding a
new growth every year. (In cross section, these show up as “rings”,
one ring for each year). The first year the trunk has radius 1 inch. In
the second year the radius increases by 1/2 inch, in the third year by
1/4 inch, and so on every year.
(a) What is the trunk radius for a 20 year old tree?
(b) For a 50 year old tree?
(c) How big can the radius of the trunk get if the tree keeps growing
this way indefinitely?
(d) What will the cross-sectional area of the trunk be after a long time?
8.7. Snowflakes. A symmetric flat snowflake starts out as a square of

size 1 millimeter. Every time step a new branch is added on the
furthest edge of each of the previous branches. Each branch is a little
square whose area is 1/ f times the area of its “parent branch”. (See
Figure 8.6.)
(a) What is the area of the snowflake after a very long time?
(b) For what values of f is this final area finite?
(c) What is the final area in the case that f = 4?
8.8. Comparing integrals. Consider the integral

D Figure 8.6: A symmetrical snowflake, for
1
A= p
dx. problem 8.7.
1 x
(a) Sketch a region in the plane whose area represents this if (i) p > 1
and (ii) p < 1.
(b) Evaluate the integral for p = 1.
(c) How does the area A depend on the value of D in each of the cases
(i) and (ii). Does the area increase without bound as D increases?
Or does the area approach some constant?
(d) With this in mind, how might we try to understand an integral of
the form ∞
1
p
dx.
1 x
8.9. Convergence of integrals. Which of the following improper integrals

converge? Give a reason in each case.
∞ ∞ ∞
1
(a) 1.001
dx (b) x dx (c) x−3 dx
1 x
∞ 1∞ 1∞
(d) ex dx (e) e−2x dx (f) x e−x dx.
0 0 0
8.10. Gravitational force. The gravitational force between two objects

of mass m1 and m2 is F = Gm1 m2 /r2 where r is the distance of
separation. Initially the objects are a distance D apart. The work done
in moving an object from position D to position x against a force F is
defined as x
W= F (r ) dr.
D
Find the total work needed to move one of these objects infinitely far
away.
8.11. Gabriel’s Horn. “Gabriel’s Horn” is the surface of revolution formed
by rotating the graph of the function y = f (x) = 1/x about the x axis
for 1 ≤ x < ∞.
(a) Find the volume of air inside this shape and show that it is finite.
(b) When we cut a cross-section of this horn along the x − y-plane, we

see a flat area which is wedged between the curves y = 1/x and
y = −1/x. Show that this “cross-sectional area” is infinite.
(c) The surface area of a surface of revolution generated by revolving
the function y = f (x) for a ≤ x ≤ b about the x-axis is given by
b
S= 2π f (x) 1 + ( f (x))2 dx.
a
Write down an integral that would represent the surface area of

“Gabriel’s Horn”.
(d) The integral in part (c) is not easy to evaluate explicitly – i.e.
we cannot find an anti-derivative. However, we can show that it
diverges. Set up a comparison that shows that the surface area of
Gabriel’s horn is infinite.
8.12. Taylor series of cos(x). The Taylor series for the function sin(x) is
given by
x3 x5 x7
sin(x) = x − + − + . . . .
3! 5! 7!
Find the Taylor series of
y = cos(x)
by differentiating this function. Using the spreadsheet, plot the follow-

ing functions (on the same graph):
y = cos(x), y = T2 (x), y = T4 (x), y = T6 (x), y = T8 (x),
where Tk is the polynomial made up of the first k (non-zero) terms in

the Taylor series for cos(x).
8.13. Estimating e. An expansion for the function ex is given by
∞ k
x2 x3 x
ex = 1 + x + + + ... = ∑ .
2 6 n=0 k!
Use the first seven terms of this series to estimate the value of the base
of natural logarithms, i.e. of e1 .
8.14. Taylor series. Find the Taylor series for each of the following func-
tions about x = 0:
(a) f (x) = x cos(x),
(b) e−x ,
2
sin(x)
(c) ,
x
(d) Use your results to find a Taylor series representation for the
following integrals:
x x
e−x dx.
2
(i) x cos(x) dx (ii)
0 0
8.15. Taylor series. Find the Taylor series at x = 0 of the following func-
tions:
f (x) = e−x
2 /2
(a) (b) f (x) = ln(1 + x)
√ √ ex + e−x
(c) f (x) = x sin x (d) f (x) = .
2
8.16. Taylor series to approximate. Using the Taylor series of f (x) near
a point x∗ , the value of a smooth function f (x) in the vicinity of x∗
can be approximated by f (x∗ ) and f (k) (x∗ ) (where f (k) (x) is the kth
derivative of f (x)), and the distance between x and x∗ , Δx = x − x∗ :
f (x∗ ) f (x∗ ) f ( k ) ( x∗ )

f (x) = f (x∗ ) + f (x∗ )Δx + (Δx)2 + (Δx)3 + · · · + (Δx)k + · · ·
2! 3! k!
where k! = k · (k − 1) · (k − 2) · · · 3 · 2 · 1 is called the factorial of the
integer k.
If |Δx| 1 (i.e., x is very close to x∗ ), (Δx)k becomes negligibly small.
When keeping only the first k terms of the series, the error arising from
throwing away all the terms subsequent to the kth term typically has
a magnitude of the same order as |Δx|k+1 . This error can be made
arbitrarily small by keeping more terms in the series. Thus, Taylor
series is often used in calculating the approximate value of a function
that is not a polynomial.
√
For example, knowing that 25 = 5, we can calculate an approximate
√
value of 26 by using the Taylor series. Note that,

√ √ 1 1
26 = 25 + 1 = 25(1 + ) = 5 1 + .
25 25
Now let x∗ = 1 and x = 1 + 25

1
, thus Δx = x − x∗ = 251
= 0.04 which is
small. Now using the Taylor series and keeping the first three terms:
√
√ √ √ [( x) ]|x=x∗
∗
x ≈ x + [( x) ]|x=x Δx +∗ (Δx)2
2!
which yields

1 1 1
1+ ≈ 1 + 0.04 − 0.042 = 1.0198.
25 2 8
Thus,
√ 1
26 = 5 1+ ≈ 5 · 1.0198 = 5.099.
25
√
Using a calculator, we find 26 = 5.0990195 . . . , which is identical for
the first 4 digits.
Use the first 3 terms of Taylor series to estimate the values of the
following functions. Calculate by hand and compare the result with
that obtained by a scientific calculator!

√ √
(a) 110 (Hint: 110 = 100 + 10 and 100 = 10),
0.1 0
(b) e (Hint: 0.1 = 0 + 0.1 and e = 1),
(c) cos(3) (Hint: 3 = π − (π − 3) ≈ π − 0.14 and cos(π ) = −1),
(d) arctan(1.1) (Hint: arctan(1) = π/4, (arctan(x)) = 1/(1 + x2 )).
8.17. Taylor series. Find the Taylor series at x = 0, up to and including the
xn term, for the following functions:
(a) f (x) = tan x, n = 3 (b) f (x) = (x + 1)ex , n = 4

2 −2x
(c) f (x) = ex , n=3 (d) f (x) = (sin x)2 , n = 6.
8.18. Taylor series.

(a) Find a Taylor series for the function f (x) = 1/(1 + x3 ) about x = 0.
Show that this can be done by making the substitution r = −x3 into
1
the sum of a geometric series (S = ∑ rk = 1−r ).
(b) Use the same idea to find the Taylor series for the function f (t ) =
1
.
1 + t2
(c) Use your result in part (b) to find a Taylor
xseries for the function
1
arctan(x). (Hint: recall that arctan(x) = 2
dt.)
0 1+t
8.19. Taylor series. Let E (x) be the function defined by
x −t
e −1
E (x) = dt.
0 t
(a) Use the Taylor series about t = 0 for the function e−t to write down
−t
the analogous Taylor series about t = 0 for the function e t−1 .
(b) Use your result in part (a) to determine the Taylor series about x = 0
for the integral E (x).
8.20. Growth of the exponential function. Consider the exponential
function y = f (x) = ex . Write down a Taylor Series series expansion
for this function. Divide both sides by xn . Now examine the terms for
the quantity ex /xn . Use this expansion to argue that the exponential
function grows faster than any power function.
8.21. Taylor series. Find a Taylor series about x = 0 for
x
F (x) = ln(1 + t 2 ) dt.
0
8.22. Differential equations and Taylor series. Use a Taylor series repre-
sentation to find the function y(t ) that satisfies the differential equation
y (t ) = y + bt with the initial condition y(0) = 1. This type of equation
is called a non-homogeneous differential equation. Show that when
b = 0 your answer agrees with the known exponential solution of the
equation y (t ) = y.
8.23. Differential equations and Taylor series. Use Taylor series to find
the function that satisfies the following second order differential
equation and initial conditions:
d2y
+ y = 0, y(0) = 0, y (0) = 1.
dt 2
A second order differential equation is one in which a second deriva-
tive appears. Notice that this type of differential equation comes with
two initial conditions. Your answer should display y(t ) as a Taylor se-
ries expansion of the desired function. (You may be able to then guess
what elementary function has this expansion as its Taylor series.)

8 Infinite Series, Improper Integrals, and Taylor Series: Mastered Material Check

Încărcat de

Informații document

Descriere originală:

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

8 Infinite Series, Improper Integrals, and Taylor Series: Mastered Material Check

Încărcat de

Drepturi de autor:

Formate disponibile

8

Inﬁnite series, improper integrals, and Taylor series

The theme of approximation has appeared before in our calculus courses.

8.1 Convergence and divergence of series

Section 8.1 Learning goals

1. Define an infinite series that is convergent; define an infinite series that is

2. Deﬁne the partial sum of an inﬁnite series.

3. Give the conditions under which an inﬁnite geometric series converges;

We begin with a review. Recall the geometric series discussed in Sec-

This leads to the following conclusion:

“convergence” Figure 8.2: An informal schematic illus-

8.2 Improper integrals

Section 8.2 Learning goals

1. Deﬁne an improper integral of the ﬁrst kind.

2. Deﬁne convergence and divergence in the context of improper integrals.

3. Explain why integrals of the form

There is a close connection between certain inﬁnite series and improper

Deﬁnition 8.3 An improper integral is an integral performed over an

With this in mind, we compute a number of classic integrals:

Example 8.1 (A decaying exponential: convegent improper integral)

Solution. Suppose that r > 0 and let

I= e−rt dt ≡ lim e−rt dt.

Then note that b > 0 so that y= 1

0 Figure 8.3: In Sections 8.2 and 8.3, we

Example 8.3 Now consider the related function

Does this integral converge of diverge?

When does the integral of 1/x p converge?

Integral comparison tests

Example 8.5 (Comparison of improper integrals) Determine whether the

Then the following conclusions can be made:

Mastered Material Check

8.3 Comparing integrals and series

Section 8.3 Learning goals

1. Describe why the harmonic series diverges.

2. Identify when the p-series converges.

The convergence of inﬁnite series was discussed earlier, in Section 1.5

The harmonic series

The harmonic series 9. We have already noticed a similar

We saw in Section 8.2 that this integral diverges!

Example 8.6 Does the following series converge?

Solution. Since p = 2 > 1, this series converges.

Example 8.7 In the above convergent series we have

Use this to determine bounds for the series.

Solution. Interpreting the series ∑∞ 2

∞ ∞ ∞ 13. Sketch a plot or use a computer to

8.4 From geometric series to Taylor polynomials

Section 8.4 Learning goals

1. Illustrate how a higher order approximation of a function does a better job

2. Explain why applying arithmetic to series must be exercised with caution.

3. Justify why we use Taylor polynomials.

Our introduction to differential calculus started with such functions for

To connect this result to a statement about a function, we need a “variable”.

Example 8.8 (A straightforward expansion) Substitute the variable x = r

p(0.1) = 1 + 0.1 + 0.12 + · · · = 1 + 0.1 + 0.01 + · · · = 1.11 . . .

We provide a few other examples based on substitutions of various sorts

Example 8.10 (Another substitution) Use the substitution r = −t to

Example 8.12 (Evaluating the logarithm for x = 0.25) When x = 0.25, we

Solution. The terms of the series are computable by only multiplication,

A scientiﬁc calculator gives ln(1.25) ≈ 0.2231, so this approximation pro-

This is the recognizable harmonic series (multiplied by -1). But we already

Example 8.14 Find a series expansion for arctan.

Solution. Suppose we make the substitution r = −t 2 into the geometric series

f (x) = a1 + 2a2 x + 3a3 x2 + . . . + kak xk−1 + . . .

f (x) = 2a2 + 2 · 3a3 x + . . . + (k − 1)kak xk−2 + . . .

f (x) = 2 · 3a3 + . . . + (k − 2)(k − 1)kak xk−3 + . . .

f (x∗ ) f (x∗ ) f ( k ) ( x∗ )