pg442 [V] G2 5-36058 / HCG / Cannon & Elich kr 11-20-95 QC1

442 Chapter 8 Discrete Mathematics: Functions on the Set of Natural Numbers

(a) Take several odd, positive integers for a1 and write Exercises 79–80 Large Numbers
out enough terms of the 3N 2 1 sequence to reach 79. The factorial sequence $n!% increases very rapidly. For
a repeating loop. instance,
(b) Show that not every positive integer reaches the
same loop (as appears to be the case for the 3N 1 1 10! < 3.6 3 10 6 20! < 2.4 3 10 18
sequence). How many different loops can you find? 50! < 3.0 3 10 64 70! < 1.2 3 10 100
76. Programs for computers or programmable calculators To get some idea of how large these numbers are,
often use an IF . . . THEN . . . instruction that termi- look at 20!. Simple computation gives 20! 5
nates the programs if certain conditions hold but branch 2,432,902,008,176,690,000. Now suppose a computer
to other instructions otherwise. Describe the kind of printer that operates at 100 characters per second were
difficulties that we could conceivably encounter if we to print out a manuscript with 20! characters. How long
programmed a computer to run the 3N 1 1 sequence would it take the printer to do the job?
and print out the terms until reaching 1.
80. For the manuscript described in Exercise 79, suppose
Exercises 77–78 Compare Sequences each page contains about 4000 characters. How thick
would the manuscript be? The thickness of a ream of
77. Evaluate the first five terms of $an% and $bn% where
paper (500 pages) is approximately 2 inches. For com-
n 4 2 6n 3 1 23n 2 2 18n 1 24 parison, the distance from the earth to the sun is 93
an 5 2n21, bn 5 .
24 million miles.
Are the sequences identical? Exercises 81–82 Fibonacci Sequence Use the definition
78. (a) Evaluate the first six terms of $an% where on page 436.
an 5 Ïn 1 Ïn 1 9 2 6Ïn. 81. Show that fn11yfn 2 1 5 fn21yfn for n $ 2.
82. Explore Write out the first twelve terms of the Fi-
Use your calculator to simplify each term. Is an 5 3
bonacci sequence. Make a guess as to which terms are
for every n $ 1? Explain.
divisible by 2, by 3.
(b) If bn 5 Ïn 1 _ Ïn 2 3 _ , are the sequences $an%
and $bn% identical? Explain.

8.2 GRAPHS AND CONVERGENCE

...the vast scope of modern mathematics. I have in mind an expanse swarming
with beaut@y#,... worthy of being surveyed from one end to the other and
It just came to me that I studied even in its smallest details: its valleys, streams, rocks, woods and
flowers.
could use this technique,
this theorem, in connection Arthur Cayley
with these curves in Hilbert
space that I was dealing
with—and get the answer!
It just came to me out of
the blue one day. It has
Calculus is based on the study of limits. At this point, we use only intuitive ideas
always struck me as so of limits, but we can use graphs and their end behavior to get strong feelings about
amazing. One half of me the existence or nonexistence of limits of certain sequences. Without making a
had been bouncing around precise definition, when a sequence $an% has a limit L, we say that $an% converges
with this theorem a lot and to L and we write lim an 5 L.
the other half had been nA`
doing this problem, and Since a sequence is a function, we can draw a graph; but we are only interested
they had never gotten in those points for which the x-pixel values are positive integers. In general, our
together. graphs will be drawn with an x-range of @0, c#, where c is the number of pixel
Andrew Gleason
columns on our calculator, and in almost all graphs we will want to use dot mode.
pg443 [R] G1 5-36058 / HCG / Cannon & Elich cr 11-20-95 QC2

8.2 Graphs and Convergence 443

TECHNOLOGY TIP r Pixel columns

The number of pixel columns, c, for several calculators is as follows:
Model acols Model acols
TI-81 95 HP-38,48 130
TI-82 94 Casio fx-7700 94
TI-85 126 Casio fx-9700 126

To illustrate convergence, we look at the end behavior for several sequences in

the first example.

cEXAMPLE 1 Convergence Use a graph to make a reasonable determina-

tion of the end behavior for the sequence, and then support your conclusion alge-
braically.
n21 n
(a) an 5 (b) bn 5 3n 2 1 (c) cn 5 ~21!n
n12 n13
Solution
(a) Graphical Draw a graph of y 5 xx 2 1 2 using @0, c# 3 @0, 2# (see Technology Tip
1

above). See Figure 1. The right portion of the graph appears to be horizontal,
[0, c] by [0, 2] but we know that the calculator has only so many pixels available. Tracing
along the curve indicates that the y-coordinate is approaching 1 as x increases.
FIGURE 1 That is, it appears that lim an 5 1; we say that the sequence $an% converges
nA`
to 1.
Algebraic If we divide the numerator and denominator of an by n, we get
1
12
n21 n
an 5 5 .
n12 2
11
n
In this form, it should be clear that an A 1 as n A `.
(b) Graphical and Algebraic The graph of y 5 3x 2 1 is a line with slope 3. If we
use an x-range of @0, c#, we need a correspondingly large y-range or the graph
goes off scale almost immediately, but even without a graph for this particular
function we know the end behavior of a line. The values of bn continue to
increase without bound and do not approach any number. We say that the
sequence $bn% diverges. In this case, as in Chapter 3 when working with
rational functions, we write lim bn 5 `.
nA`
(c) Graphical Graphing y 5 ~21!x xy~x 1 3! in @0, c# 3 @23, 3# gives very dif-
ferent looking graphs in connected or dot modes. Whichever you choose, make
sure that each x-pixel coordinate is an integer and that you know how to
interpret what the graph shows. Tracing (in either mode) shows that as x
increases, the y-values jump back and forth, with the positive values approach-
ing 1 and the negative values approaching 21. We conclude that $cn% diverges
because the cn values do not approach a single number as n A `.
pg444 [V] G2 5-36058 / HCG / Cannon & Elich cr 11-30-95 MP1

444 Chapter 8 Discrete Mathematics: Functions on the Set of Natural Numbers

Algebraic Disregarding the ~21!n, the expression n 1n 3 does have a limit,

n 1
lim 5 lim 5 1.
nA` n 1 3 nA` 3
11
n
It follows that cn A ~21!n, so that when n is even, cn A 1 and when n is odd,
cn A 21. Again, $cn% diverges because the cn values do not approach a single
number as n A `. b

Sequences Defined Recursively

For sequences defined recursively we cannot enter functions for graphing as we did
in Example 1. Nevertheless, functionality built into our graphing calculators makes
it possible to investigate limits of some such sequences quite easily. Consider the
sequence $an% defined by

3
a1 5 2, an11 5 1 1 for n $ 1. (1)
an
We want to calculate a number of terms of the sequence without having to go
through all the steps of the recursive definition for each term. We describe some
options in the following Technology Tip, and then look at additional examples.

TECHNOLOGY TIP r Calculating recursively defined sequences

For algebraic operation calculators (TI, Casio, and HP–38), essentially all of
the steps for evaluating the sequence in Equation (1) can be handled on the
home screen by making use of the machine capacity to store values.
Begin by storing the initial value 2 in the x-register: 2 A X , ENT .
Then compute and store the next value: 1 1 3yXAX , ENT .
The calculator displays the computed value, 2.5, which has been stored.
When we press ENTER again, the same computation is repeated with the new
x-value and displayed value, 2.2, is our a3 . As we ENTER repeatedly, the terms of
the sequence are displayed. It soon becomes clear that the terms are
approaching a limit, <2.3027756.
On the HP–48, we can write a simple program to accomplish the same
thing.
Press ,, .. (above the subtract key) to begin a program. Then type
A X91 1 3X9 A Num (above 5 7 ) and ENTER . What the program does is to take the
EVAL

number on the stack, call it x, compute 1 1 3yx symbolically, convert the

symbolic computation to a number. The result is displayed on the stack so
that the process can be repeated.
To use the program, we need to store it as a variable, so we type a name,
say ’RECR’ for “recursive”. Then ENTER and 5 7 . The name RECR should appear on
STO

your 5 7 menu. Now enter 2 on the stack, press the soft key beneath 5
VAR 7,
RECR

and the new value appears. By repeatedly pressing 5 7 , the value continues
RECR

to change, approaching 2.3027756.

pg445 [R] G1 5-36058 / HCG / Cannon & Elich cr 11-30-95 MP1

8.2 Graphs and Convergence 445

Strategy: For (c), if the cEXAMPLE 2 Nested square roots Sequence $an% is defined by
sequence converges to a
number c, both an and an11 a1 5 Ï3, an11 5 Ï3 1 an for n $ 1.
approach c, leading to the
(a) Write out the first three terms in exact form.
equation c 5 Ï3 1 c. Then
solve for c. (b) Use the Technology Tip (page 444) to approximate the first few terms of the
sequence and find the apparent limit of the sequence (six decimal places).
(c) Justify your conclusion in (b) algebraically.
Solution
(a) a1 5 Ï3, a2 5 Ï3 1 a1 5 Ï3 1 Ï3,
a3 5 Ï3 1 a2 5 Ï3 1 Ï3 1 Ï3 .
(b) Following the Technology Tip, for all machines except the HP-48, we storeÏ3
in the x-register, and then enter Ï(3 1 X) A X. Repeating the computation gives a
sequence beginning 1.732051, 2.175328, 2.274935, 2.296723, . . . . After sev-
eral more terms, the sequence settles on a number approximately equal to
2.302776. On the HP-48, we must change the recursive part of the definition
in our program RECR by pressing the tick-mark key and the soft key under RECR .
With ’RECR’ on the stack, we press 57 and go into the program, replacing ’1 1 3yX’
EDIT

by the recursive part of our new sequence, ’Ï~3 1 X)’. With the new program, we
enter Ï3 and then repeat the soft key under RECR , getting the same sequence of
terms.
(c) Following the strategy, we want to solve the equation c 5 Ï3 1 c for c.
Squaring both sides, we get the equation c 2 5 3 1 c, or c 2 2 c 2 3 5 0. By
the quadratic formula, taking the positive sign (why not 6?), we get c 5
~1 1 Ï13!y2 < 2.302775638, obviously the number we were approximating
in part (b). b
cEXAMPLE 3 Nested cube roots Repeat Example 2 for the sequence $cn%
defined by
3 3
c1 5 Ï2, cn11 5 Ï2 1 cn for n $ 1.
Solution
3 3 3
(a) c1 5 Ï2, c2 5 Ï2 1 c1 5 Ï2 1 Ï2
3
,
c3 5 Ï2 1 c2 5 Ï2 1 Ï2 1 Ï2
3 3 3 3
.
(b) On algebraic operation machines, we store 2^(1y3) in the x-register, and then
enter (2 1 X)^(1y3) A X. The sequence begins 1.259921, 1.482754, 1.515797,
1.520575, 1.521264, . . . . The sequence settles on a number approximately
equal to 1.5213797. On the HP-48, we enter ’(2 1 X)^(1y3)’ as the recursive part of
3
the definition in RECR . After entering Ï2 on the stack, repeating the soft key
under RECR gives the same sequence of terms.
(c) Since it appears that the sequence converges to a number c, both cn and cn11
must approach the same number c, so the recursive portion of the definition
gives an equation which can be cubed:
3
c 5 Ï2 1 c, or c 3 2 c 2 2 5 0.
The cubic equation is not one we can solve in exact form conveniently, but by
graphical methods, or by using a solve routine, or by using Newton’s Method
from Chapter 3, the one real zero is approximately 1.5213797. b
pg446 [V] G2 5-36058 / HCG / Cannon & Elich cr 11-30-95 MP1

446 Chapter 8 Discrete Mathematics: Functions on the Set of Natural Numbers

Continued Fractions

Continued fractions is a topic studied in number theory courses that has applica-
tions in many areas, including the programming of routines for computers and
graphing calculators. In the next example we illustrate the continued fraction,
1
11 , as a recursively defined sequence.
1
11
1
11
11...

cEXAMPLE 4 Continued fractions Sequence $an% is defined by

a1 5 1, an11 5 1 1 1yan for n $ 1.
(a) Write out the first four terms, first without simplifying, and then as a simple
fraction.
(b) Approximate the first few terms of the sequence and find the apparent limit of
the sequence.
(c) Justify your conclusion in (b) algebraically.
Solution
1 2 1 1 3
(a) a1 5 1, a2 5 1 1 5 , a3 5 1 1 5 1 1 5 ,
1 1 a2 2 2
1 1 1 2 5
a4 5 1 1 5 1 1 511 3 511 5 .
a3 1 1 12 2 3 3
The numbers in the numerator and denominator of the fractions remind us of
the Fibonacci sequence $ fn% from page 436: 1, 1, 2, 3, 5, 8, 13, . . . . That is,
a1 5 f2yf1 , a2 5 f3yf2 , a3 5 f4yf3 , a4 5 f5yf4 , and a reasonable guess is that
an 5 fn11yfn .
(b) For decimal approximations, we use the Technology Tip, beginning with 1 and
using 1 1 1yX A X for the recursion. The sequence appears to settle down on a
number c < 1.618034.
(c) If both an11 and an approach c, then in the limit the recursion relation becomes
1
c511 .
c
Multiplying through by c leads to the quadratic equation c 2 2 c 2 1 5 0,
whose positive root is given by c 5 1 12Ï5 < 1.618034. b

The limit number of the sequence, 1 12Ï5 , is called the Golden Ratio, reflecting
some aesthetic considerations of the ancient Greeks. It is a number that turns up in
many diverse applications. See Exercise 27.
In the next example, we see another instance of a sequence that diverges even
though parts of the sequence, called subsequences, converge. We had one such
sequence in Example 1, given by cn 5 ~21!n n 1n 3 . From the graph in dot mode,
we saw that the sequence consisting of the even-numbered terms $c2 , c4 , c6 , . . .%
converges to 1; the odd-numbered terms form a subsequence that converges to 21.
The same kind of behavior is possible with a recursively defined sequence.
pg447 [R] G1 5-36058 / HCG / Cannon & Elich cr 11-30-95 MP1

8.2 Graphs and Convergence 447

cEXAMPLE 5 Subsequences Sequence $an% is defined by

2an
a1 5 3, an11 5 for n $ 1.
~an 2 2!
Write out the first few terms. Does the sequence have a limit? Describe some
convergent subsequences of the sequence.

Solution
Either by using the Technology Tip (page 444) or by direct computation, it is clear
that the sequence begins 3, 6, 3, 6, 3, 6, 3, 6, . . . . The sequence has no limit
because the terms are not getting close to any number as n A `. The subsequence
of odd-numbered terms contains only 3, $a2n21% 5 $3, 3, 3, 3, . . .%, which obvi-
ously converges to 3. Similarly, the subsequence consisting of even-numbered
terms $a2n% 5 $6, 6, 6, . . .% converges to 6. b

TECHNOLOGY TIP r Calculating with two-step recursions

Any graphing calculator can be programmed to calculate more involved

recursively defined sequences, but general programming is not our focus in
this text. To learn about programming on your calculator, consult your
instruction manual. The Texas Instrument TI-82 and TI-85, HP–38, and the
Casio fx7700 and fx9700, allow us to handle two-step recursively defined
sequences, such as the Fibonacci sequence, directly on the home screen, as
described below.
The Fibonacci sequence (page 436) is defined by
f1 5 1, f2 5 1, fn 5 fn22 1 fn21 , n . 2.
On the home screen, we store the initial values and their sum:
1 A A: 1 A B: A 1 B A C and ENTER

(The colon is located above the decimal point on TI and HP–38, and on the
PRGM menu on the Casio.) The display shows f3 as 2. We must reassign values

for the next step:

B A A:C A B: A 1 B A C .

Now ENTER and we see 3 as f4 , and we can repeat for as many terms as desired.
We revisit this problem in matrix form in Exercise 70 of Section 9.6.

Equations of the Form F (x) 5 x

When looking for the roots of an equation it is sometimes possible to isolate an x,

writing the equation in the form F~x! 5 x. Under certain circumstances it is
possible to use an iterative process to approximate a root of such an equation to
great accuracy. Basically, the solution is found as the limit of a recursive sequence.
We begin with an initial approximation a1 and define an11 5 F~an!. Determining
the conditions under which such an iteration converges requires calculus, but we
illustrate the procedure in Example 6.
pg448 [V] G2 5-36058 / HCG / Cannon & Elich jb 12-1-95 MP3

448 Chapter 8 Discrete Mathematics: Functions on the Set of Natural Numbers

cEXAMPLE 6 Solving an equation

(a) Approximate the root of the equation 2x 2 cos x 5 0 from a graph.
(b) Write the equation in the form f ~x! 5 x and use the approximation from part
(a) as a1 and let an11 5 f ~an!. Iterate to approximate the limit L of the sequence to
8 decimal places and verify that L satisfies the original equation.
Solution
(a) From a graph, we can see that there is a root of the equation near 0.5.
cos x cos x
(b) The equation is equivalent to x 5 , so we take f ~x! 5 and define
2 2
the sequence by
cos an
a1 5 0.5, an11 5 .
2
Following the Technology Tip for recursive sequences, we store 0.5 A X , ENT.
Then follow with (COS X)y2 A X , ENT and iterate, getting a sequence beginning
0.43879128, 0.45263292, . . . . The sequence settles quickly on the number
L 5 0.45018361, and when we substitute L for x in the expression 2x 2 cos
x, we get a number very near 0, as desired. b

EXERCISES 8.2
Check Your Understanding
It will be helpful to use the Technology Tip (page 444) to get 8. The sum of the first 5 terms is .
the first several terms of $an%.
Exercises 9–10 Sequence $bn% is defined by b1 5 1 and
Exercises 1–5 True or False. Give reasons. Use sequence bn11 5 Ï1 1 b 2n for n $ 1.
$an% defined by a1 5 1 and an11 5 1 1 .
1 9. b5 5 .
an 10. The smallest prime number that is greater than b5 is
1. Every term of $an% is less than or equal to 2. .
2. The sequence is decreasing; that is an11 , an for ev-
ery n. Develop Mastery
3. The even-numbered terms are greater than the odd- Exercises 1–10 Does it Converge? Use a graph to help
numbered terms. you determine whether or not the sequence appears to con-
4. The subsequence consisting of the odd-numbered verge. Explain.
terms, $a1 , a3 , a5 , . . .%, is decreasing. ~21!n~n 1 1!
1. an 5 2n 2 5 2. an 5
5. The subsequence consisting of the even-numbered n
terms, $a2 , a4 , a6 , . . .%, is increasing. 2n 1 3 n11
3. an 5 4. an 5 2
n11 n 11

S D
Exercises 6–10 Fill in the blank so that the resulting
statement is true. 1 n
5. an 5 3 2 22n 6. an 5 1 1
n

S D S D
Exercises 6–8 Sequence $an% is defined by a1 5 1 and
an11 5 an 1 4 for n . 1. 1 n 2 n
7. an 5 2 1 1 2 8. an 5 1 1
n n
6. The smallest integer greater than a5 is .
7. The number of terms of $an% between 8 and 20 9. an 5 Ïn 1 Ïn 1 64 2 16Ïn
is . 10. an 5 Ïn 1 _ Ïn 2 5 _
pg449 [R] G1 5-36058 / HCG / Cannon & Elich kr 11-20-95 QC1

8.2 Graphs and Convergence 449

Exercises 11–16 Sequences Defined Recursively Use 1

the Technology Tip (page 444). (a) Give the first three terms 26. (a) a1 5 1, an11 5 2an 2
an
of $an%. (b) The sequence converges to a number c. Use your 1 1
calculator to get a six-decimal-place approximation for c. (b) a1 5 , an11 5 5an 2
2 an
(c) Justify your answer algebraically. See Examples 2
and 3. 1 1
(c) a1 5 , an11 5 10an 2
11. a1 5 Ï3, an11 5 Ï3 2 an for n . 1. 3 an
1 1
12. a1 5 Ï5, an11 5 Ï5 1 an for n . 1. (d) a1 5 , an11 5 17an 2
4 an
13. a1 5 Ï7, an11 5 Ï7 2 an for n . 1.
3 3 27. Fibonacci Related For a1 5 3, an11 5 3 2 a1n ,
14. a1 5 Ï5, an11 5 Ï5 2 an for n . 1.
3 3
(a) Write the first six terms as simple fractions.
15. a1 5 Ï6, an11 5 Ï6 1 an for n . 1. (b) Guess a relationship between $an% and the
4 4
16. a1 5 Ï5, an11 5 Ï5 1 an for n . 1. Fibonacci sequence.
Exercises 17–22 Continued Fractions (a) Find the 28. For a1 5 Ïln 2, an11 5 Ïln 2 1 an ,
first three terms of $cn%. (b) Find a 6-decimal-place ap- (a) use the Technology Tip to find a six-decimal-place
proximation for the limit to which $cn% appears to converge. approximation to the limit c to which $an% con-
(c) Justify algebraically. See Example 4. verges.
2
1 (b) Show that c is a root of e x 2x 2 2 5 0.
17. c1 5 3, cn11 5 3 1 for n . 1.
cn
Exercises 29–30 Golden Ratio, 1 12Ï5 Given that $an%
1
18. c1 5 5, cn11 5 5 1 for n . 1. converges to the number c, use an algebraic approach to
cn verify that c is the number given. Then use the Technology
1 Tip to get a calculator check.
19. c1 5 3, cn11 5 3 2 for n . 1.
cn
1
20. c1 5 1, cn11 5 1 1 2 for n . 1.
29. a1 5 2, an11 5 Î 11
1
an 2
; c is the square root of the
cn golden ratio.
1 1
21. c1 5 3, cn11 5 3 1 2 for n . 1. 30. a1 5 3, an11 5 2 1; c is the negative of the golden
cn an
1 ratio.
22. c1 5 2, cn11 5 2 2 2 for n . 1.
cn
Exercises 23–24 Repeating Terms Sequence $an% is Exercises 31–32 Explore The recursive formula for
defined recursively. (a) Give the first six terms. Does $an% an11 is given along with different values of a1 . In each case
converge? You may wish to use the Technology Tip. use the Technology Tip to get the first three terms and the
(b) Determine a60 and the sum of the first sixty terms. limit (six decimal places) to which the sequence converges.
Try other values of a1 . Describe the role that a1 plays.

Î
1
23. a1 5 2, an11 5 1 2 . 1
an 31. an11 5 11 ,
2 a 2n
24. a1 5 2, an11 5 1 1 . (a) a1 5 1 (b) a1 5 8 (c) a1 5 24
an
Exercises 25–26 Recognizing a Pattern Find the first 1
32. an11 5 1 1 ,
four terms of $an%. Make a generalization and justify alge- an
braically. (a) a1 5 1 (b) a1 5 3

25. (a) a1 5 1, S
an11 5 0.5 an 1
1
an
D (c) a1 5 25

S D
Exercises 33–36 Subsequences (a) Does $an% con-
4
(b) a1 5 2, an11 5 0.5 an 1 verge? (b) Find subsequences of $an% that converge. See
an

S D
Example 5.
9 ~21!n n ~21!~2n!
(c) a1 5 3, an11 5 0.5 an 1 33. an 5 34. an 5
an n11 n11
(d) a1 5 4, an11 S
5 0.5 an 1
16
an
D 35. an 5 sin SD
np
2
36. an 5 cos~np !
pg450 [V] G2 5-36058 / HCG / Cannon & Elich cr 11-30-95 MP1

450 Chapter 8 Discrete Mathematics: Functions on the Set of Natural Numbers

Exercises 37–40 Your Choice Give a sequence $an% of would lead to 0. For example, solving a1n 2 1 5 0, we get
your choice that meets the given conditions. an 5 1. If we were to try a1 5 1, we would get a2 5 0 and
37. an . 0 if n is odd, an , 0 if n is even. 1
then a3 would be undefined. (a) Solve 2 1 5 1 and
38. The sequence $an% does not converge, but the subse- an
quence of odd-numbered terms (all of which are greater find another inadmissible value for a1 . (b) Find a se-
than 2) converges to 2, while the subsequence of even- quence in exact form of inadmissible initial value num-
numbered terms (all of which are less than 22) con- bers. (c) Show that a1 5 3y5 is inadmissible by comput-
verges to 22. ing the first few terms in exact form. (d) Try a1 5 0.6 and
compute the first few terms by using the Technology Tip
39. For every n, 0 , an , 1, and $an% converges to (page 444). Explain the difference in results from part (c).
(a) 1; (b) 0.5.
54. The sequence $an% is given by an 5 x n 1 x2n, where
40. For all odd n, 0 , an , 1, and for all even n,
21 1 Ï3 i
1 , an , 2, and $an% converges to 1. x5 .
2
(a) Use DeMoivre’s theorem to evaluate x n 1 x2n,
then show that an 5 2 cos(n · 1208).
Exercises 41–43 Repeating Terms (a) Give the first four
(b) Write out the first six terms of the sequence and find
terms of $an%. (b) What is a47 ? a72 ? (c) Find the sum of
their sum.
the first twenty terms. (d) Explain the repeating behavior
(c) What is the sum of the first 100 terms?
of $an%. (Hint for Exercise 41: Consider f ~x! 5 2xy~x 2 2!,
and show that f 21~x! 5 f ~x!. What is f 21~ f ~x!!?)
2an
41. a1 5 6, an11 5 Exercises 55–60 Sequence $an% converges to a number L.
an 2 2
(a) Use the Technology Tip to approximate L. (b) Use
23an
42. a1 5 1, an11 5 algebra to find the exact value of L.

S D
2an 1 3
1 4
3an 55. a1 5 1 an 5 a 1 for n $ 2
43. a1 5 4, an11 5 2 n21 an21

S D
2an 2 3
1 3
56. a1 5 2 an 5 a 1 for n $ 2
2 n21 an21
Exercises 44–50 Roots of f~x! 5 x Follow the instruc-
tions for Example 6 for the given equation.
57. a1 5 1 an 5
1
2
S an21 1
2
an21
D for n $ 2

44. cos x 5 x 45. 5x 2 2 cos x 5 0 58. a1 5 Ï2 an 5 Ï2 1 an21 for n $ 2

2x
46. x 5 e 47. x 5 cos~xy4! 59. a1 5 Ï6 an 5 Ï6 1 an21 for n $ 2

48. x 5 cos S D
41x
4
49. x 5 ln~4 1 x! 60. a1 5 1 an 5
4
an21
2 Ï2 for n $ 2

50. x 1 4 5 e x

Exercises 61–62 Sequences $an% and $bn% converge; se-

Exercises 51–52 Roots of 2x 5 x 10 Find the limit L of quence $cn% diverges. (a) Find approximations (6 decimal
$an% to eight decimal places. Show that L is a root of the places) for the limits of $an% and $bn%. (b) Find subsequences
equation 2x 5 x 10. of $cn% that converge and approximate their limits.
10 ln an 1 1
51. a1 5 50, an11 5 52. a1 5 1, an11 5 20.1an 61. a1 5 1, an11 5 4 1 ; b1 5 1, bn11 5 4 2
ln 2 an bn
53. Explore In the recursive formula for Exercise 30, ~21!n
c1 5 1, cn11 5 4 1
an11 5 a1n 2 1, many different initial values give se- cn
quences that converge to the same value. There are 3 3
62. a1 5 4, an11 5 4 1 ; b1 5 4, bn11 5 4 2
initial values that do not work, however. We obviously an bn
cannot use a1 5 0 because a2 would then be undefined, ~21!n 3
c1 5 4, cn11 5 4 1
and we cannot use a number as an initial value that cn