This action might not be possible to undo. Are you sure you want to continue?
https://www.scribd.com/doc/94098511/IntroductiontoAbstractMathematics
11/04/2013
text
original
Introduction to Abstract Mathematics
(Second part of MA103)
Amol Sasane
ii
Contents
1 Analysis 1
1.1 The real numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Sequences and limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.1 Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.2 Limit of a convergent sequence . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.3 Bounded and monotone sequences . . . . . . . . . . . . . . . . . . . . . . . 14
1.2.4 Algebra of limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.2.5 Sandwich theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.2.6 Subsequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.3 Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.3.1 Deﬁnition of continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.3.2 Continuous functions preserve convergent sequences . . . . . . . . . . . . . 30
1.3.3 Extreme value theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.3.4 Intermediate value theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2 Algebra 39
2.1 Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.1.1 Deﬁnition of a Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.1.2 Subgroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.1.3 Homomorphisms and isomorphisms . . . . . . . . . . . . . . . . . . . . . . . 49
2.1.4 Cosets and Lagrange’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.2 Vector spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.2.1 Deﬁnition of a vector space . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.2.2 Subspaces and linear combinations . . . . . . . . . . . . . . . . . . . . . . . 59
iii
iv Contents
2.2.3 Basis of a vector space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.2.4 Linear transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Solutions 69
Bibliography 127
Index 129
Chapter 1
Analysis
Analysis is the theory behind real numbers, sequences, and functions. The word ‘theory’ is im
portant. You might, for example, have a good idea of what we mean by a ‘limit’ of a convergent
sequence or the notion of a ‘continuous’ function, but in this part of the course we try to formalize
such notions.
1.1 The real numbers
The rational number system is inadequate for many purposes. For instance, there is no rational
number q such that q
2
= 2, and the set
S = ¦q ∈ Q [ q
2
≤ 2¦
does not have a largest element in Q. So we see that the rational number system has certain holes.
The real number system R ﬁlls these gaps. Thus the set
S = ¦q ∈ R [ q
2
≤ 2¦
has a largest element. This is a consequence of a very important property of the real numbers,
called the least upper bound property. But before we state this property of R, we need a few
deﬁnitions.
Deﬁnitions. Let S be a subset of R.
1. An element u ∈ R is said to be an upper bound of S if for all x ∈ S, x ≤ u. If the set of all
upper bounds of S is not empty, then S is said to be bounded above.
2. An element l ∈ R is said to be a lower bound of S if for all x ∈ S, l ≤ x. If the set of all
lower bounds of S is not empty, then S is said to be bounded below.
3. The set S is said to be bounded if it is bounded above and it is bounded below.
Examples.
1. The set S = ¦x ∈ R [ 0 ≤ x < 1¦ is bounded. Any real number y satisfying 1 ≤ y (for
instance 1, π, 100) is an upper bound of S, and any real number z satisfying z ≤ 0 (for
instance 0, −1) is a lower bound of S.
1
2 Chapter 1. Analysis
2. The set S = ¦n [ n ∈ N¦ is not bounded. Although it is bounded below (any real number
x ≤ 1 serves as a lower bound), it has no upper bound, and so it is not bounded above.
3. The set
1
S = ¦(−1)
n
[ n ∈ N¦ is bounded. It is bounded above by 1 and bounded below by
−1.
4. The set S =
_
1
n
¸
¸
n ∈ N
_
is bounded. Any real number x satisfying 1 ≤ x is an upper
bound, and 0 is a lower bound.
5. The sets Z and R are neither bounded above nor bounded below. Indeed, this is a conse
quence of the inequality z < z + 1.
6. The set ∅ is bounded. (Why?) ♦
We now introduce the notions of a least upper bound (also called supremum) and a greatest
lower bound (also called inﬁmum) of a subset S of R.
Deﬁnitions. Let S be a subset of R.
1. An element u
∗
∈ R is said to be a least upper bound of S (or a supremum of S) if
(a) u
∗
is an upper bound of S, and
(b) if u is an upper bound of S, then u
∗
≤ u.
2. An element l
∗
∈ R is said to be a greatest lower bound of S (or an inﬁmum of S) if
(a) l
∗
is a lower bound of S, and
(b) if l is a lower bound of S, then l ≤ l
∗
.
Example. If S = ¦x ∈ R [ 0 ≤ x < 1¦, then the supremum of S is 1 and the inﬁmum of S is 0.
Clearly 1 is an upper bound of S.
Now we show that if u is another upper bound, then 1 ≤ u. Suppose not, that is, u < 1. Then
we have
0 ≤ u <
u + 1
2
< 1, (1.1)
where the ﬁrst inequality is a consequence of the facts that u is an upper bound of S and 0 ∈ S,
while the last two inequalities follow using u < 1. From (1.1), it follows that the number
u+1
2
satisﬁes 0 <
u+1
2
< 1, and so it belongs to S. The middle inequality in (1.1) above then shows
that u cannot be an upper bound for S, a contradiction. Hence 1 is a supremum.
Next we show that this is the only supremum, since if u
∗
is another supremum, then in
particular u
∗
is also an upper bound, and the above argument shows that 1 ≤ u
∗
. But 1 < u
∗
is
not possible as 1 is an upper bound, and as u
∗
is a supremum, u
∗
must be less than or equal to
1. So it follows that u
∗
= 1.
Similarly one can show that the inﬁmum of S is 0. ♦
In the above example, there was a unique supremum and inﬁmum of the set S. In fact, this
is always the case and we have the following result.
1
Note that this set is simply the ﬁnite set (that is, the set has ﬁnite cardinality) {−1, 1}. More generally, any
ﬁnite set S is bounded.
1.1. The real numbers 3
Theorem 1.1.1 If the least upper bound of a subset S of R exists, then it is unique.
Proof Suppose that u
∗
and u
∗
are two least upper bounds of S. Then in particular u
∗
and u
∗
are also upper bounds of S. Now since u
∗
is a least upper bound of S and u
∗
is an upper bound
of S, it follows that
u
∗
≤ u
∗
. (1.2)
Furthermore, since u
∗
is a least upper bound of S and u
∗
is an upper bound of S, it follows that
u
∗
≤ u
∗
. (1.3)
From (1.2) and (1.3), we obtain u
∗
= u
∗
.
Thus it makes sense to talk about the least upper bound of a set. The least upper bound of a
set S (if it exists) is denoted by
sup S
(the abbreviation of ‘supremum of S’). Similarly, the inﬁmum of a set S (if it exists) is also unique,
and is denoted by
inf S.
When the supremum and the inﬁmum of a set belong to the set, then we give them special names:
Deﬁnitions.
1. If supS ∈ S, then sup S is called a maximum of S, denoted by max S.
2. If inf S ∈ S, then inf S is called a minimum of S, denoted by min S.
Examples.
1. If S = ¦x ∈ R [ 0 ≤ x < 1¦, then supS = 1 ,∈ S and so max S does not exist. But
inf S = 0 ∈ S, and so min S = 0.
2. If S = ¦n [ n ∈ N¦, then sup S does not exist, inf S = 1, max S does not exist, and min S = 1.
3. If S = ¦(−1)
n
[ n ∈ N¦, then sup S = 1, inf S = −1, max S = 1, min S = −1.
4. If S =
_
1
n
¸
¸
n ∈ N
_
, then sup S = 1 and max S = 1. We show below (after Theorem 1.1.2)
that inf S = 0. So min S does not exist.
5. For the sets Z and R, sup, inf, max, min do not exist.
6. For the set ∅, sup, inf, max, min do not exist. ♦
In the above examples, we note that if S is nonempty and bounded above, then its supremum
exists. In fact this is a fundamental property of the real numbers, called the least upper bound
property of the real numbers, which we state below:
If S is a nonempty subset of R having an upper bound, then sup S exists.
4 Chapter 1. Analysis
Remarks(∗).
1. Note that the set of rational numbers do not possess this property. For instance, the set
S = ¦q ∈ Q [ q
2
≤ 2¦
has an upper bound, say 2 (indeed, if q > 2, then q
2
> 4 > 2, and so q ,∈ S), but we now
show that it does not have a supremum in Q. Assume on the contrary that u
∗
∈ Q is a
supremum for S. Deﬁne
r = u
∗
−
u
2
∗
−2
u
∗
+ 2
. (1.4)
Then we can check that
r
2
−2 =
2(u
2
∗
−2)
(u
∗
+ 2)
2
(1.5)
We have the following cases:
1
◦
Suppose that u
2
∗
< 2. From (1.5) we obtain r ∈ S, and from (1.4) it follows that r > u
∗
,
which contradicts the fact that u
∗
is an upper bound of S.
2
◦
Suppose that u
2
∗
= 2. This is impossible, since u
∗
is a rational number.
3
◦
Suppose that u
2
∗
> 2. We see that r is an upper bound of S (indeed, if q > r, then
2
q
2
> r
2
> 2, and so q ,∈ S). But (1.4) implies that r < u
∗
, contradicting the fact that u
∗
is
the supremum.
2. In Exercise 6 on page 7 below, given a nonempty set S of R, we deﬁne −S = ¦−x [ x ∈ S¦.
One can show that if a nonempty subset S of R is bounded below, then −S is bounded above
and so sup(−S) exists, by the least upper bound property. The negative of this supremum,
namely −sup(−S), can then be shown to serve as the greatest lower bound of S (this is
precisely the content of Exercise 6). Thus the real numbers also have the ‘greatest lower
bound property’: If S is a nonempty subset of R having an lower bound, then inf S exists.
We now prove the following theorem, which is called the Archimedean property of the real
numbers.
Theorem 1.1.2 (Archimedean property.) If x, y ∈ R and x > 0, then there exists an n ∈ N such
that y < nx.
Proof If not, then the nonempty set S = ¦nx [ n ∈ N¦ has an upper bound y, and so by the least
upper bound property of the reals, it has a least upper bound u
∗
. But then u
∗
− x < u
∗
(since
x is positive) and so it follows that u
∗
− x cannot be an upper bound of S. Hence there exists a
natural number m such that u
∗
− x < mx, that is, u
∗
< (m + 1)x ∈ S. This contradicts the fact
that u
∗
is an upper bound of S.
Example. If S =
_
1
n
¸
¸
n ∈ N
_
, then inf S = 0. We know that 0 is a lower bound of S. Suppose
that l is a lower bound of S such that l > 0. By the Archimedean property (with the real numbers
x and y taken as x = 1 (> 0) and y =
1
l
), there exists a n ∈ N such that
1
l
= y < nx = n 1 = n,
and so
1
n
< l, contradicting the fact that l is a lower bound of S. Thus any lower bound of S
must be less than or equal to 0. Hence 0 is the inﬁmum of S. ♦
2
the last inequality follows from (1.5)
1.1. The real numbers 5
Deﬁnition. An interval is a set consisting of all the real numbers between two given real numbers,
or of all the real numbers on one side or the other of a given number. So an interval is a set of
any of the following forms, where a, b ∈ R:
PSfrag replacements
a
a
a
a
a
a
b
b
b
b
b
b
(a, b) = ¦x ∈ R [ a < x < b¦
[a, b] = ¦x ∈ R [ a ≤ x ≤ b¦
(a, b] = ¦x ∈ R [ a < x ≤ b¦
[a, b) = ¦x ∈ R [ a ≤ x < b¦
(a, ∞) = ¦x ∈ R [ a < x¦
[a, ∞) = ¦x ∈ R [ a ≤ x¦
(−∞, b) = ¦x ∈ R [ x < b¦
(−∞, b] = ¦x ∈ R [ x ≤ b¦
(−∞, ∞) = R
In the above notation for intervals, a parenthesis ‘(’ or ‘)’ means that the respective endpoint
is not included, and a square bracket ‘[’ or ‘]’ means that the endpoint is included. Thus [0, 1)
means the set of all real numbers x such that 0 ≤ x < 1. (Note that the use of the symbol ∞ in
the notation for intervals is simply a matter of convenience and is not be taken as suggesting that
there is a number ∞.)
In analysis, in order to talk about notions such as convergence and continuity, we will need a
notion of ‘closeness’ between real numbers. This is provided by the absolute value [ [, and the
distance between real numbers x and y is [x −y[. We give the deﬁnitions below.
Deﬁnitions.
1. The absolute value of a real number x is denoted by [x[, and it is deﬁned as follows:
[x[ =
_
x if x ≥ 0,
−x if x < 0.
2. The distance between two real numbers x and y is the absolute value [x−y[ of their diﬀerence.
Thus [1[ = 1, [0[ = 0, [ −1[ = 1, and the distance between the real numbers −1 and 1 is equal
to [ −1 −1[ = [ −2[ = 2. The distance gives a notion of closeness of two points, which is crucial in
the formalization of the notions of analysis. We can now specify regions comprising points close
to a certain point x
0
∈ R in terms of inequalities in absolute values, that is, by demanding that
the distance of the points of the region, to the point x
0
, is less than a certain positive number δ,
say δ = 0.01 or δ = 0.0000001, and so on. See Exercise 9 on page 8 and Figure 1.1.
PSfrag replacements
x
0
−δ x
0
+δ
I
x
0
x
Figure 1.1: The interval I = (x
0
− δ, x
0
+δ) = ¦x ∈ R [ [x − x
0
[ < δ¦ is the set of all points in R
whose distance to the point x
0
is strictly less than δ (> 0).
6 Chapter 1. Analysis
The following properties of the absolute value will be useful in the sequel.
Theorem 1.1.3 If x, y are real numbers, then
[xy[ = [x[ [y[ and (1.6)
[x +y[ ≤ [x[ +[y[. (1.7)
Proof We prove (1.6) by exhausting all possible cases:
1
◦
x = 0 or y = 0. Then [x[ = 0 or [y[ = 0, and so [x[ [y[ = 0. On the other hand, as x = 0 or
y = 0, it follows that xy = 0 and so [xy[ = 0.
2
◦
x > 0 and y > 0. Then [x[ = x and [y[ = y, and so [x[ [y[ = xy. On the other hand, as x > 0
and y > 0, it follows that xy > 0 and so [xy[ = xy.
3
◦
x > 0 and y < 0. Then [x[ = x and [y[ = −y, and so [x[ [y[ = x(−y) = −xy. On the other
hand, as x > 0 and y < 0, it follows that xy < 0 and so [xy[ = −xy.
4
◦
x < 0 and y > 0. This follows from 3
◦
above by interchanging x and y.
5
◦
x < 0 and y < 0. Then [x[ = −x and [y[ = −y, and so [x[ [y[ = (−x)(−y) = xy. On the
other hand, as x < 0 and y < 0, it follows that xy > 0 and so [xy[ = xy.
This proves (1.6).
Next we prove (1.7). First observe that from the deﬁnition of [ [, it follows that for any real
x ∈ R, [x[ ≥ x: indeed if x ≥ 0, then [x[ = x, while if x < 0, then −x > 0, and so [x[ = −x > 0 > x.
From (1.6), we also have [ −x[ = [ −1 x[ = [ −1[[x[ = 1[x[ = [x[, for all x ∈ R, and so it follows
that [x[ = [ −x[ ≥ −x for all x ∈ R. We have the following cases:
1
◦
x+y ≥ 0. Then [x+y[ = x+y. As [x[ ≥ x and [y[ ≥ y, we obtain [x[ +[y[ ≥ x+y = [x+y[.
2
◦
x + y < 0. Then [x + y[ = −(x + y). Since [x[ ≥ −x and [y[ ≥ −y, it follows that
[x[ +[y[ ≥ −x + (−y) = −(x +y) = [x +y[.
This proves (1.7).
It is easy to check that the distance satisﬁes the following properties:
D1. (Positive deﬁniteness.) For all x, y ∈ R, [x − y[ ≥ 0. If [x −y[ = 0 then x = y.
D2. (Symmetry.) For all x, y ∈ R, [x −y[ = [y −x[.
D3. (Triangle inequality.) For all x, y, z ∈ R, [x −z[ ≤ [x −y[ +[y −z[.
Exercises.
1. Provide the following information about the set S
An
upper
bound
A
lower
bound
Is S
bounded?
sup S inf S
If sup S exists,
then is sup S
in S?
If inf S exists,
then is inf S
in S?
max S min S
where S is given by:
1.1. The real numbers 7
(a) (0, 1]
(b) [0, 1]
(c) (0, 1)
(d)
_
1
n
¸
¸
n ∈ Z ¸ ¦0¦
_
(e)
_
−
1
n
¸
¸
n ∈ N
_
(f)
_
n
n+1
¸
¸
¸ n ∈ N
_
(g) ¦x ∈ R [ x
2
≤ 2¦
(h) ¦0, 2, 5, 2005¦
(i)
_
(−1)
n
_
1 +
1
n
_ ¸
¸
n ∈ N
_
(j) ¦x
2
[ x ∈ R¦
(k) ¦
x
2
1+x
2
[ x ∈ R¦ .
2. Determine whether the following statements are TRUE or FALSE.
(a) If u is an upper bound of a subset S of R, and u
< u, then u
is not an upper bound
for S.
(b) If u
∗
is the least upper bound of a subset S of R, and is any positive real number,
then u
∗
− is not an upper bound of S.
(c) Every subset of R has a maximum.
(d) Every subset of R has a supremum.
(e) Every bounded subset of R has a maximum.
(f) Every bounded subset of R has a supremum.
(g) Every bounded nonempty subset of R has a supremum.
(h) Every set that has a supremum is bounded above.
(i) For every set that has a maximum, the maximum belongs to the set.
(j) For every set that has a supremum, the supremum belongs to the set.
(k) For every set S that is bounded above, [S[ deﬁned by ¦[x[ [ x ∈ S¦ is bounded.
(l) For every set S that is bounded, [S[ deﬁned by ¦[x[ [ x ∈ S¦ is bounded.
(m) For every bounded set S, if inf S < x < sup S, then x ∈ S.
3. For any nonempty bounded set S, prove that inf S ≤ supS, and that the equality holds iﬀ
3
S is a singleton set (that is a set with cardinality 1).
4. Let A and B be nonempty subsets of R that are bounded above and such that A ⊂ B. Prove
that sup A ≤ sup B.
5. Let A and B be nonempty subsets of R that are bounded above and deﬁne
A+B = ¦x +y [ x ∈ A and y ∈ B¦.
Prove that sup(A+B) exists and that sup(A+B) ≤ supA+ sup B.
6. Let S be a nonempty subset of real numbers which is bounded below. Let −S denote the set
of all real numbers −x, where x belongs to S. Prove that inf S exists and inf S = −sup(−S).
7. Let S be a nonempty set of positive real numbers, and deﬁne S
−1
=
_
1
x
¸
¸
x ∈ S
_
. Show that
S
−1
is bounded above iﬀ inf S > 0. Furthermore, in case inf S > 0, show that supS
−1
=
1
inf S
.
3
The abbreviation ‘iﬀ’ is standard in Mathematics, and it stands for ‘if and only if’.
8 Chapter 1. Analysis
8. Let A
n
, n ∈ N, be a collection of sets.
The notation
n∈N
A
n
denotes the intersection of the sets A
n
, n ∈ N, that is,
n∈N
A
n
= ¦x [ ∀n ∈ N, x ∈ A
n
¦,
and we use
_
n∈N
A
n
to denote the union of the sets A
n
, n ∈ N, that is,
_
n∈N
A
n
= ¦x [ ∃n ∈ N such that x ∈ A
n
¦.
Prove that
(a) ∅ =
n∈N
_
0,
1
n
_
.
(b) ¦0¦ =
n∈N
_
0,
1
n
_
.
(c) (0, 1) =
_
n∈N
_
1
n + 2
, 1 −
1
n + 2
_
.
(d) [0, 1] =
n∈N
_
−
1
n
, 1 +
1
n
_
.
9. Let x
0
∈ R and δ > 0. Prove that (x
0
−δ, x
0
+δ) = ¦x ∈ R [ [x −x
0
[ < δ¦.
10. Prove that if x, y are real numbers, then [[x[ −[y[[ ≤ [x −y[.
11. Show that a subset S of R is bounded iﬀ there exists a M ∈ R such that for all x ∈ S,
[x[ ≤ M.
1.2 Sequences and limits
1.2.1 Sequences
The notion of a sequence occurs in ordinary conversation. An example is the phrase “an unfor
tunate sequence of events”. In this case, we envision one event causing another, which in turn
causes another event and so on. We can identify a ﬁrst event, a second event, etcetera.
A sequence of real numbers is a list
a
1
, a
2
, a
3
, . . .
of real numbers, where there is the ﬁrst number (namely a
1
), the second number (namely a
2
), and
so on. For example,
1,
1
2
,
1
3
, . . .
is a sequence. The ﬁrst number is 1, the second number is
1
2
and so on. (There may not be a
connection between the numbers appearing in a sequence.) If we think of a
1
as f(1), a
2
as f(2),
and so on, then it becomes clear that a sequence is a special type of function, namely one with
domain N and codomain R.
1.2. Sequences and limits 9
Deﬁnition. A sequence is a function f : N →R.
Only the notation is somewhat unusual. Instead of writing f(n) for the value of f at a natural
number n, we write a
n
. The entire sequence is then written in any one of the following ways:
(a
n
)
n∈N
, (a
n
)
∞
n=1
, (a
n
)
n≥1
, (a
n
).
In (a
n
)
∞
n=1
, the ∞ symbol indicates that the assignment process 1 → a
1
, 2 → a
2
, . . . continues
indeﬁnitely. The nth term a
n
of a sequence may be deﬁned explicitly by a formula involving n, as
in the example given above:
a
n
=
1
n
, n ∈ N.
It might also sometimes be deﬁned recursively. For example,
a
1
= 1, a
n+1
=
n
n + 1
a
n
for n ∈ N.
(Write down the ﬁrst few terms of this sequence.)
Examples.
1.
_
1
n
_
n∈N
is a sequence with the nth term given by
1
n
, for n ∈ N. This is the sequence
1,
1
2
,
1
3
, . . . .
2.
_
1 +
1
n
_
n∈N
is a sequence with the nth term given by 1 +
1
n
, for n ∈ N. This is the sequence
2,
3
2
,
4
3
,
5
4
,
6
5
,
7
6
, . . . .
3.
_
(−1)
n
_
1 +
1
n
__
n∈N
is a sequence with the nth term given by (−1)
n
_
1 +
1
n
_
, for n ∈ N. This
is the sequence
−2,
3
2
, −
4
3
,
5
4
, −
6
5
,
7
6
, . . . .
4. ((−1)
n
)
n∈N
is a sequence with the nth term given by (−1)
n
, for n ∈ N. This sequence is
simply
−1, 1, −1, 1, −1, 1, . . .
with the nth term equal to −1 if n is odd, and 1 if n is even.
5. (1)
n∈N
is a sequence with the nth term given by 1, for n ∈ N. This is the constant sequence
1, 1, 1, . . . .
6. (n)
n∈N
is a sequence with the nth term given by n, for n ∈ N. This is the increasing sequence
1, 2, 3, . . . .
7.
_
1
1
1
+
1
2
2
+
1
3
3
+ +
1
n
n
_
n∈N
is a sequence with the nth term given by
1
1
1
+
1
2
2
+
1
3
3
+ +
1
n
n
,
for n ∈ N. This is the sequence of ‘partial sums’
1
1
1
,
1
1
1
+
1
2
2
,
1
1
1
+
1
2
2
+
1
3
3
, . . . .
♦
10 Chapter 1. Analysis
1.2.2 Limit of a convergent sequence
A sequence can be graphed. For instance, the ﬁrst 4 points of the graph of the sequence
_
1
n
_
n∈N
are displayed in Figure 1.2. This portion of the graph suggests that the terms of the sequence
PSfrag replacements
1
2 3 4
1
Figure 1.2: First four points of the graph of the sequence
_
1
n
_
n∈N
.
_
1
n
_
n∈N
tend toward 0 as n increases. This is consistent with the idea of convergence that you
might have encountered before: a sequence (a
n
)
n∈N
converges to some real number L, if the terms
a
n
get “closer and closer” to L as n “increases without bound”. Symbolically, this is represented
using the notation
lim
n→∞
a
n
= L,
where L denotes the limit of the sequence. If there is no such ﬁnite number L to which the terms
of the sequence get arbitrarily close, then the sequence is said to diverge.
Th problem with this characterization is its imprecision. Exactly what does it mean for the
terms of a sequence to get “closer and closer”, or “as close as we like”, or “arbitrarily close” to some
number L? Even if we accept this apparent ambiguity, how would one use the deﬁnition given
in the preceding paragraph to prove theorems that involve sequences? Since sequences are used
throughout analysis, the concepts of their convergence and divergence must be carefully deﬁned.
PSfrag replacements
1
2
2 3 4
1
Figure 1.3: First four points of the graph of the sequence
_
1 +
1
n
_
n∈N
.
For example, the terms of
_
1 +
1
n
_
n∈N
get “closer and closer” to 0 (indeed the distance to 0
keeps decreasing), but its limit is 1. See Figure 1.3.
The terms of
_
(−1)
n
_
1 +
1
n
__
n∈N
get “as close as we like” or “arbitrarily close” to 1, but the
sequence has no limit. See Figure 1.4.
Deﬁnition. The sequence (a
n
)
n∈N
is said to converge to L if for every > 0, there exists
4
an
N ∈ N such that for all n > N,
[a
n
−L[ < .
4
depending on
1.2. Sequences and limits 11
PSfrag replacements
1
1
2
2
3 4 5 6 7 8
−1
−2
Figure 1.4: First eight points of the graph of the sequence
_
(−1)
n
_
1 +
1
n
__
n∈N
.
Then we say that (a
n
)
n∈N
is convergent (with limit L) and write
lim
n→∞
a
n
= L.
If there does not exist a number L such that lim
n→∞
a
n
= L, then the sequence (a
n
)
n∈N
is said to
be divergent.
Note that [a
n
− L[ < iﬀ a
n
∈ (L − , L + ). Hence pictorially, for a convergent sequence
with limit L, this deﬁnition means the following, as illustrated in Figure 1.5: Pick any > 0, and
consider the shaded strip of width around the horizontal line passing through L. Then one can
ﬁnd a N ∈ N, large enough, such that all the terms a
n
of the sequence, for n > N lie in the shaded
strip.
PSfrag replacements
L +
L +
L
L
L
L −
L −
N N + 1 N + 2 N + 3
Figure 1.5: Convergence of a sequence with limit L.
Examples.
1.
_
1
n
_
n∈N
is a convergent sequence with limit 0.
12 Chapter 1. Analysis
Given > 0, we need to ﬁnd a N such that for all n > N,
[a
n
−L[ =
¸
¸
¸
¸
1
n
−0
¸
¸
¸
¸
=
1
n
< .
If we choose N ∈ N such that N >
1
(such a N exists by the Archimedean property!), then
for n > N (⇔
1
n
<
1
N
), we have
[a
n
−L[ =
¸
¸
¸
¸
1
n
−0
¸
¸
¸
¸
=
1
n
<
1
N
< .
Hence lim
n→∞
1
n
= 0.
2.
_
1 +
1
n
_
n∈N
is a convergent sequence with limit 1.
Given > 0, we need to ﬁnd a N such that for all n > N,
[a
n
−L[ =
¸
¸
¸
¸
1 +
1
n
−1
¸
¸
¸
¸
=
1
n
< .
Again we choose a N ∈ N such that N >
1
and so for n > N we have
[a
n
−L[ =
1
n
<
1
N
< .
Hence lim
n→∞
_
1 +
1
n
_
= 1.
3.
_
(−1)
n
_
1 +
1
n
__
n∈N
is a divergent sequence.
Let a
n
= (−1)
n
_
1 +
1
n
_
for n ∈ N. In order to prove that
(a
n
)
n∈N
is divergent,
we have to show that
[(a
n
)
n∈N
is convergent] ,
that is,
[∃L ∈ R such that ∀ > 0 ∃N ∈ N such that ∀n > N, [a
n
−L[ < ] ,
that is,
∀L ∈ R ∃ > 0 such that ∀N ∈ N ∃n > N such that [a
n
−L[ ≥ .
Let L ∈ R. Now we need to prove
∃ > 0 such that ∀N ∈ N ∃n > N such that [a
n
−L[ ≥ .
Let = 1. (It is not obvious that = 1 would work, but it has been found by trial and
error.) Now we will show that
∀N ∈ N ∃n > N such that [a
n
−L[ ≥ .
So let N ∈ N. If L ≥ 0, then choose n to be any odd number > N. Then we have
[a
n
−L[ =
¸
¸
¸
¸
(−1)
n
_
1 +
1
n
_
−L
¸
¸
¸
¸
=
¸
¸
¸
¸
−1 −
1
n
−L
¸
¸
¸
¸
= 1 +
1
n
+L > 1 = .
If L < 0, then choose n to be any even number > N. Then we have
[a
n
−L[ =
¸
¸
¸
¸
(−1)
n
_
1 +
1
n
_
−L
¸
¸
¸
¸
=
¸
¸
¸
¸
1 +
1
n
−L
¸
¸
¸
¸
= 1 +
1
n
−L > 1 = .
Thus we have shown that for all L ∈ R, there exists a > 0 (namely = 1) such that for all
N ∈ N, there exists a n > N (namely any odd number > N if L ≥ 0, and any even number
> N if L < 0) such that [a
n
−L[ ≥ . Thus the sequence is divergent. ♦
1.2. Sequences and limits 13
The notation lim
n→∞
a
n
suggests that the limit of a convergent sequence is unique. Indeed this
is the case, and we prove this below.
Theorem 1.2.1 A convergent sequence has a unique limit.
Proof Consider a convergent sequence (a
n
)
n∈N
and suppose that it has distinct limits L
1
and
L
2
. Let
=
[L
1
−L
2
[
2
> 0,
where the positivity of the deﬁned above follows from the fact that L
1
,= L
2
. Since L
1
is a limit,
∃N
1
∈ N such that for all n > N
1
,
[a
n
−L
1
[ < .
Since L
2
is a limit, ∃N
2
∈ N such that for all n > N
2
,
[a
n
−L
2
[ < .
Consequently for n > max¦N
1
, N
2
¦,
2 = [L
1
−L
2
[ = [L
1
−a
n
+a
n
−L
2
[ ≤ [L
1
−a
n
[ +[a
n
−L
2
[ = [a
n
−L
1
[ +[a
n
−L
2
[ < + = 2,
a contradiction.
Exercises.
1. (a) Prove that the constant sequence (1)
n∈N
is convergent.
(b) Can the limit of a convergent sequence be one of the terms of the sequence?
(c) If none of the terms of a convergent sequence equal its limit, then prove that the terms
of the sequence cannot consist of a ﬁnite number of distinct values.
(d) Prove that the sequence ((−1)
n
)
n∈N
is divergent.
2. Prove that lim
n→∞
1
n
,= 1.
3. In each of the cases listed below, give an example of a divergent sequence (a
n
)
n∈N
that
satisﬁes the given conditions. Suppose that L = 1.
(a) For every > 0, there exists an N such that for inﬁnitely many n > N, [a
n
−L[ < .
(b) There exists an > 0 and a N ∈ N such that for all n > N, [a
n
−L[ < .
4. Let S be a nonempty subset of R that is bounded above. Show that there exists a sequence
(a
n
)
n∈N
contained in S (that is, a
n
∈ S for all n ∈ N) and which is convergent with limit
equal to sup S.
5. Let (a
n
)
n∈N
be a sequence such that for all n ∈ N, a
n
≥ 0. Prove that if (a
n
)
n∈N
is convergent
with limit L, then L ≥ 0.
6. A sequence (a
n
)
n∈N
is said to be Cauchy if for every > 0, there exists a N ∈ N such that
for all n, m > N, [a
n
−a
m
[ < .
Show that every convergent sequence is Cauchy.
Hint: [a
n
−a
m
[ = [a
n
−L +L −a
m
[ ≤ [a
n
−L[ +[a
m
−L[.
14 Chapter 1. Analysis
1.2.3 Bounded and monotone sequences
It is cumbersome to check from the deﬁnition if a sequence is convergent or not. In this section,
we will study a condition under which we can conclude that a sequence is convergent even without
knowing its limit! We will prove that if a sequence is both ‘bounded’ as well as ‘monotone’, then
it is always convergent.
Deﬁnition. A sequence (a
n
)
n∈N
is said to be bounded if there exists a M > 0 such that
for all n ∈ N, [a
n
[ ≤ M. (1.8)
Note that a sequence is bounded iﬀ the set S = ¦a
n
[ n ∈ N¦ is bounded. (See Exercise 11 on
page 8).
Examples.
1. (1)
n∈N
is bounded, since [1[ = 1 ≤ 1 for all n ∈ N.
2.
_
1
n
_
n∈N
is bounded, since
¸
¸
1
n
¸
¸
=
1
n
≤ 1 for all n ∈ N.
3.
_
1 +
1
n
_
n∈N
is bounded, since
¸
¸
1 +
1
n
¸
¸
= 1 +
1
n
≤ 2 for all n ∈ N.
4.
_
(−1)
n
_
1 +
1
n
__
n∈N
is bounded, since
¸
¸
(−1)
n
_
1 +
1
n
_¸
¸
= 1 +
1
n
≤ 2 for all n ∈ N.
5. The sequence (a
n
)
n∈N
deﬁned by
a
n
=
1
1
1
+
1
2
2
+
1
3
3
+ +
1
n
n
, n ∈ N
is bounded. Indeed this can be seen as follows:
[a
n
[ = a
n
=
1
1
1
+
1
2
2
+
1
3
3
+ +
1
n
n
<
1
1
1
+
1
2
2
+
1
2
3
+ +
1
2
n
=
1
1
1
+
1
2
_
1 −
1
2
_
+
1
2
2
_
1 −
1
2
_
+ +
1
2
n−1
_
1 −
1
2
_
= 1 +
1
2
−
1
2
2
+
1
2
2
−
1
2
3
+− +
1
2
n−1
−
1
2
n
= 1 +
1
2
−
1
2
n
<
3
2
.
(Write down a detailed proof using induction on n.) So all the terms are bounded by
3
2
, and
so the sequence is bounded.
6. The sequence (a
n
)
n∈N
given by a
n
= n for n ∈ N, is not bounded. Indeed, given any M > 0,
there exists an N ∈ N such that M < N (Archimedean property with y = M and x = 1).
Thus
[∃M > 0 such that for all n ∈ N, [a
n
[ = [n[ = n ≤ M],
and so (n)
n∈N
is not bounded. ♦
1.2. Sequences and limits 15
The sequences (1)
n∈N
,
_
1
n
_
n∈N
,
_
1 +
1
n
_
n∈N
are all convergent, and we have shown above that
these are also bounded. This is not a coincidence, and in the next theorem we show that the set
of all convergent sequences is contained in the set of all bounded sequences. See Figure 1.6.
convergent bounded
sequences sequences
PSfrag replacements
L +
Figure 1.6: All convergent sequences are bounded.
Theorem 1.2.2 If a sequence is convergent, then it is bounded.
Proof Let (a
n
)
n∈N
be a convergent sequence with limit L. Let := 1 > 0. Then ∃N ∈ N such
that for all n > N
[a
n
−L[ < = 1.
Hence for all n > N,
[a
n
[ = [a
n
−L +L[ ≤ [a
n
−L[ +[L[ < 1 +[L[.
Let M = max¦[a
1
[, . . . , [a
N
[, 1 +[L[¦. Then for all n ∈ N
[a
n
[ ≤ M
and so (a
n
)
n∈N
is bounded.
Deﬁnitions. A sequence (a
n
)
n∈N
is said to be increasing if for all n ∈ N, a
n
≤ a
n+1
. Thus
(a
n
)
n∈N
is increasing if
a
1
≤ a
2
≤ a
3
≤ . . . .
A sequence (a
n
)
n∈N
is said to be decreasing if for all n ∈ N, a
n
≥ a
n+1
. Thus (a
n
)
n∈N
is decreasing
if
a
1
≥ a
2
≥ a
3
≥ . . . .
A sequence is said to be monotone if it is increasing or decreasing.
Examples.
Sequence Is it Is it Is it
increasing? decreasing? monotone?
_
1
n
_
n∈N
No Yes Yes
_
1 +
1
n
_
n∈N
No Yes Yes
_
(−1)
n
_
1 +
1
n
__
n∈N
No No No
(1)
n∈N
Yes Yes Yes
(n)
n∈N
Yes No Yes
_
1
1
1
+
1
2
2
+
1
3
3
+ +
1
n
n
_
n∈N
Yes No Yes
♦
The following theorem can be useful in showing that sequences converge when one does not
know the limit beforehand.
16 Chapter 1. Analysis
Theorem 1.2.3 If a sequence is monotone and bounded, then it is convergent.
Proof
1
◦
Let (a
n
)
n∈N
be an increasing sequence. Since (a
n
)
n∈N
is bounded, it follows that the set
S = ¦a
n
[ n ∈ N¦
has an upper bound and so sup S exists. We show that in fact (a
n
)
n∈N
converges to sup S. Indeed
given > 0, then since supS − < supS, it follows that sup S − is not an upper bound for S
and so ∃a
N
∈ S such that sup S − < a
N
, that is
sup S −a
N
< .
Since (a
n
)
n∈N
is an increasing sequence, for n > N, we have a
N
≤ a
n
. Since supS is an upper
bound for S, a
n
≤ sup S and so [a
n
−sup S[ = sup S −a
n
, Thus for n > N we obtain
[a
n
−sup S[ = sup S −a
n
≤ sup S −a
N
< .
2
◦
If (a
n
)
n∈N
is a decreasing sequence, then clearly (−a
n
)
n∈N
is an increasing sequence. Further
more if (a
n
)
n∈N
is bounded, then (−a
n
)
n∈N
is bounded as well ([ − a
n
[ = [a
n
[ ≤ M). Hence by
the case considered above, it follows that (−a
n
)
n∈N
is a convergent sequence with limit
sup¦−a
n
[ n ∈ N¦ = −inf¦a
n
[ n ∈ N¦ = −inf S,
where S = ¦a
n
[ n ∈ N¦ (see Exercise 6 on page 7). So given > 0, ∃N ∈ N such that for all
n > N, [ − a
n
− (−inf S)[ < , that is, [a
n
− inf S[ < . Thus (a
n
)
n∈N
is convergent with limit
inf S.
Examples.
1. We have shown that the sequence (a
n
)
n∈N
deﬁned by
a
n
=
1
1
1
+
1
2
2
+
1
3
3
+ +
1
n
n
, n ∈ N
is monotone (indeed, it is increasing since a
n+1
− a
n
=
1
(n+1)
(n+1)
> 0 for all n ∈ N) and
bounded (see item 5 on page 14). Thus it follows from Theorem 1.2.3 that this sequence
5
is
convergent.
2. The following table gives a summary of the valid implications, and gives counterexamples to
implications which are not true.
Question Answer Reason/Counterexample
Is every convergent sequence bounded? Yes Theorem 1.2.2
Is every bounded sequence convergent? No ((−1)
n
)
n∈N
is bounded,
but not convergent.
Is every convergent sequence monotone? No
_
(−1)
n
n
_
n∈N
is convergent,
but not monotone: −1 <
1
2
> −
1
3
.
Is every monotone sequence convergent? No (n)
n∈N
is not convergent.
Is every bounded AND monotone Yes Theorem 1.2.3
sequence convergent?
♦
5
Although it is known that this sequence is convergent to some limit L ∈ R, it is so far not even known if the
limit L is rational or irrational, and this is still an open problem in mathematics! Also associated with this sequence
is the interesting identity
∞
X
n=1
1
n
n
=
Z
1
0
1
x
x
dx, the proof of which is beyond the scope of this course.
1.2. Sequences and limits 17
Exercises.
1. Let (a
n
)
n∈N
be a sequence deﬁned by
a
1
= 1 and a
n
=
2n + 1
3n
a
n−1
for n ≥ 2.
Prove that (a
n
)
n∈N
is convergent.
2. If (b
n
)
n∈N
is a bounded sequence, then prove that
_
bn
n
_
n∈N
is a convergent sequence with
limit 0.
3. (a) (∗) If (a
n
)
n∈N
is a convergent sequence with limit L, then prove that the sequence
(s
n
)
n∈N
, where
s
n
=
a
1
+ +a
n
n
, n ∈ N,
is also convergent with limit L.
(b) Give an example of a sequence (a
n
)
n∈N
such that (s
n
)
n∈N
is convergent but (a
n
)
n∈N
is
divergent.
4. (∗) Given a bounded sequence (a
n
)
n∈N
, deﬁne
l
k
= inf¦a
n
[ n ≥ k¦ and u
k
= sup¦a
n
[ n ≥ k¦, k ∈ N.
Show that the sequences (l
n
)
n∈N
, (u
n
)
n∈N
are bounded and monotone, and conclude that
they are convergent. (Their respective limits are denoted by liminf
n→∞
a
n
and limsup
n→∞
a
n
.)
5. (∗) Recall the deﬁnition of a Cauchy sequence from Exercise 6 on page 13. Prove that every
Cauchy sequence is bounded.
1.2.4 Algebra of limits
In this section we will learn that if we ‘algebraically’ combine the terms of convergent sequences,
then the new sequence which is obtained, is again convergent, and moreover the limit of this
sequence is the same algebraic combination of the limits. In this manner we can sometimes prove
the convergence of complicated sequences by breaking them down and writing them as an algebraic
combination of simple sequences. Thus, we conveniently apply arithmetic rules to compute the
limits of sequences if the terms are the sum, product, quotient of terms of simpler sequences with
a known limit. For instance, using the formal deﬁnition of a limit, one can show that the sequence
(a
n
)
n∈N
deﬁned by
a
n
=
4n
2
+ 9
3n
2
+ 7n + 11
converges to
4
3
. However, it is simpler to observe that
a
n
=
n
2
_
4 +
9
n
2
_
n
2
_
3 +
7
n
+
11
n
2
_ =
4 +
9
n
2
3 +
7
n
+
11
n
2
,
where the terms
9
n
2
,
7
n
,
11
n
2
all have limit 0, and by a repeated application of Theorem 1.2.4 given
below, we obtain
lim
n→∞
a
n
=
lim
n→∞
_
4 +
9
n
2
_
lim
n→∞
_
3 +
7
n
+
11
n
2
_ =
lim
n→∞
4 + lim
n→∞
9
n
2
lim
n→∞
3 + lim
n→∞
7
n
+ lim
n→∞
11
n
2
=
4 + 0
3 + 0 + 0
=
4
3
.
18 Chapter 1. Analysis
Theorem 1.2.4 If (a
n
)
n∈N
and (b
n
)
n∈N
are convergent sequences, then the following hold:
1. For all α ∈ R, (αa
n
)
n∈N
is a convergent sequence and lim
n→∞
αa
n
= α lim
n→∞
a
n
.
2. ([a
n
[)
n∈N
is a convergent sequence and lim
n→∞
[a
n
[ =
¸
¸
¸ lim
n→∞
a
n
¸
¸
¸.
3. (a
n
+b
n
)
n∈N
is a convergent sequence and lim
n→∞
(a
n
+b
n
) = lim
n→∞
a
n
+ lim
n→∞
b
n
.
4. (a
n
b
n
)
n∈N
is a convergent sequence and lim
n→∞
a
n
b
n
=
_
lim
n→∞
a
n
__
lim
n→∞
b
n
_
.
5. For all k ∈ N, (a
k
n
)
n∈N
is a convergent sequence and lim
n→∞
a
k
n
=
_
lim
n→∞
a
n
_
k
.
6. If for all n ∈ N, b
n
,= 0 and lim
n→∞
b
n
,= 0, then
_
1
b
n
_
n∈N
is convergent and moreover,
lim
n→∞
1
b
n
=
1
lim
n→∞
b
n
.
Proof Let (a
n
)
n∈N
and (b
n
)
n∈N
converge to L
a
and L
b
, respectively.
1. If α = 0, then αa
n
= 0 for all n ∈ N and clearly (0)
n∈N
is a convergent sequence with limit
0. Thus
lim
n→∞
αa
n
= 0 = 0L
a
= α lim
n→∞
a
n
.
If α ,= 0, then given > 0, let N ∈ N be such that for all n > N,
[a
n
−L
a
[ <
[α[
,
that is
[αa
n
−αL
a
[ = [α[ [a
n
−L
a
[ < [α[
[α[
= .
Hence (αa
n
)
n∈N
is convergent with limit αL
a
, that is,
lim
n→∞
αa
n
= αL
a
= α lim
n→∞
a
n
.
2. Given > 0, let N ∈ N be such that for all n > N,
[a
n
−L
a
[ < .
Then we have for all n > N:
[[a
n
[ −[L
a
[[ ≤ [a
n
−L
a
[ < .
Hence ([a
n
[)
n∈N
is convergent with limit [L
a
[, that is,
lim
n→∞
[a
n
[ = [L
a
[ =
¸
¸
¸ lim
n→∞
a
n
¸
¸
¸ .
3. Given > 0, let N
1
∈ N be such that for all n > N
1
,
[a
n
−L
a
[ <
2
.
1.2. Sequences and limits 19
Let N
2
∈ N be such that for all n > N
2
,
[b
n
−L
b
[ <
2
.
Then for all n > N := max¦N
1
, N
2
¦, we have
[a
n
+b
n
−(L
a
+L
b
)[ = [a
n
−L
a
+b
n
−L
b
[ ≤ [a
n
−L
a
[ +[b
n
−L
b
[ <
2
+
2
= .
Hence (a
n
+b
n
)
n∈N
is convergent with limit L
a
+L
b
, that is,
lim
n→∞
(a
n
+b
n
) = L
a
+L
b
= lim
n→∞
a
n
+ lim
n→∞
b
n
.
4. Note that
[a
n
b
n
−L
a
L
b
[ = [a
n
b
n
−L
a
b
n
+L
a
b
n
−L
a
L
b
[
≤ [a
n
b
n
−L
a
b
n
[ +[L
a
b
n
−L
a
L
b
[
= [a
n
−L
a
[ [b
n
[ +[L
a
[ [b
n
−L
b
[. (1.9)
Given > 0, we need to ﬁnd a N such that for all n > N,
[a
n
b
n
−L
a
L
b
[ < .
This can be achieved by ﬁnding a N such that each of the summands in (1.9) is less than
2
for n > N. This can be done as follows.
Step 1. Since (b
n
)
n∈N
is convergent, by Theorem 1.2.2 it follows that it is bounded: ∃M > 0
such that for all n ∈ N, [b
n
[ ≤ M. Let N
1
∈ N be such that for all n > N
1
,
[a
n
−L
a
[ <
2M
.
Step 2. Let N
2
∈ N be such that for all n > N
2
,
[b
n
−L
b
[ <
2([L
a
[ + 1)
.
Thus for n > N := max¦N
1
, N
2
¦, we have
[a
n
b
n
−L
a
L
b
[ ≤ [a
n
−L
a
[ [b
n
[ +[L
a
[ [b
n
−L
b
[
<
2M
M +[L
a
[
2([L
a
[ + 1)
=
2
+
2
= .
So (a
n
b
n
)
n∈N
is a convergent sequence with limit L
a
L
b
, that is,
lim
n→∞
a
n
b
n
= L
a
L
b
=
_
lim
n→∞
a
n
__
lim
n→∞
b
n
_
.
5. This can be shown by using induction on k and from part 4 above. It is trivially true with
k = 1. Suppose that it holds for some k, then (a
k
n
)
n∈N
is convergent and
lim
n→∞
a
k
n
=
_
lim
n→∞
a
n
_
k
.
Hence by part 4 above applied to the sequences (a
n
)
n∈N
and (a
k
n
)
n∈N
, we obtain that the
sequence (a
n
a
k
n
)
n∈N
is convergent and
lim
n→∞
a
n
a
k
n
=
_
lim
n→∞
a
n
__
lim
n→∞
a
k
n
_
=
_
lim
n→∞
a
n
__
lim
n→∞
a
n
_
k
=
_
lim
n→∞
a
n
_
k+1
.
20 Chapter 1. Analysis
Thus (a
k+1
n
)
n∈N
is convergent and
lim
n→∞
a
k+1
n
=
_
lim
n→∞
a
n
_
k+1
.
6. Let N
1
∈ N be such that for all n > N
1
,
[b
n
−L
b
[ <
[L
b
[
2
.
Thus for all n > N
1
,
[L
b
[ −[b
n
[ ≤ [[L
b
[ −[b
n
[[ ≤ [b
n
−L
b
[ <
[L
b
[
2
,
and so [b
n
[ ≥
L
b

2
. Let > 0, and let N
2
∈ N be such that for all n > N
2
,
[b
n
−L
b
[ <
[L
b
[
2
2
.
Hence for n > N := max¦N
1
, N
2
¦, we have
¸
¸
¸
¸
1
b
n
−
1
L
b
¸
¸
¸
¸
=
[b
n
−L
b
[
[b
n
[ [L
b
[
<
[L
b
[
2
2
2
[L
b
[
1
[L
b
[
= .
So
_
1
b
n
_
n∈N
is convergent and lim
n→∞
1
b
n
=
1
L
b
=
1
lim
n→∞
b
n
.
Exercises.
1. Recall the convergent sequence (a
n
)
n∈N
from Exercise 1 on page 17 deﬁned by
a
1
= 1 and a
n
=
2n + 1
3n
a
n−1
for n ≥ 2.
What is its limit?
Hint: If (a
n
)
n∈N
is a convergent sequence with limit L, then (a
n+1
)
n∈N
is also a convergent
sequence with limit L.
2. Suppose that the sequence (a
n
)
n∈N
is convergent, and assume that the sequence (b
n
)
n∈N
is
bounded. Prove that the sequence (c
n
)
n∈N
deﬁned by
c
n
=
a
n
b
n
+ 5n
a
2
n
+n
is convergent, and ﬁnd its limit.
3. (a) Let (a
n
)
n∈N
be a convergent sequence with limit L and suppose that a
n
≥ 0 for all
n ∈ N. Prove that the sequence (
√
a
n
)
n∈N
is also convergent, with limit
√
L.
Hint: First show that L ≥ 0. Let > 0. If L = 0, then choose N ∈ N large enough so
that for n > N, [a
n
− L[ = a
n
<
2
. If L > 0, then choose N ∈ N large enough so that
for n > N, [
√
a
n
−
√
L[[
√
a
n
+
√
L[ = [a
n
−L[ <
√
L.
(b) Show that (
_
n
2
+n −n)
n∈N
is a convergent sequence and ﬁnd its limit.
Hint: ‘Rationalize the numerator’ by using
√
n
2
+n +n.
4. Prove that if (a
n
)
n∈N
and (b
n
)
n∈N
are convergent sequences such that for all n ∈ N, a
n
≤ b
n
,
then
lim
n→∞
a
n
≤ lim
n→∞
b
n
.
Hint: Use Exercise 5 on page 13.
1.2. Sequences and limits 21
1.2.5 Sandwich theorem
Another useful theorem that is useful in proving that sequences are convergent and in determining
their limits is the socalled sandwich theorem. Roughly speaking, it says that if a sequence is
sandwiched between two convergent limits with the same limit, then the sandwiched sequence is
also convergent with the same limit.
Theorem 1.2.5 (Sandwich theorem.) Let (a
n
)
n∈N
, (b
n
)
n∈N
be convergent sequences with the
same limit, that is,
lim
n→∞
a
n
= lim
n→∞
b
n
.
If (c
n
)
n∈N
is a third sequence such that
for all n ∈ N, a
n
≤ c
n
≤ b
n
,
then (c
n
)
n∈N
is also convergent with the same limit, that is,
lim
n→∞
a
n
= lim
n→∞
c
n
= lim
n→∞
b
n
.
Proof Let L denote the common limit of (a
n
)
n∈N
and (b
n
)
n∈N
:
lim
n→∞
a
n
= L = lim
n→∞
b
n
.
Given > 0, let N
1
∈ N be such that for all n > N
1
, [a
n
−L[ < . Hence for n > N
1
,
L −a
n
≤ [L −a
n
[ = [a
n
−L[ < ,
and so L −a
n
< , that is,
L − < a
n
.
Let N
2
∈ N be such that for all n > N
2
, [b
n
−L[ < . So for n > N
2
, b
n
−L < , that is,
b
n
< L +.
Thus for n > N := max¦N
1
, N
2
¦, we have
L − < a
n
≤ c
n
≤ b
n
< L +,
and so L − < c
n
< L + . Consequently, c
n
−L < and −(c
n
−L) < , and so
[c
n
−L[ < .
This proves that (c
n
)
n∈N
is convergent with limit L.
Examples.
1. lim
n→∞
n
10
n
= 0.
It can be shown by induction that for all n ∈ N, n
2
< 10
n
.
Consequently, we have
0 ≤
n
10
n
≤
n
n
2
=
1
n
.
Since lim
n→∞
0 = 0 = lim
n→∞
1
n
, from the Sandwich theorem it follows that the sequence
_
n
10
n
_
n∈N
is convergent and
lim
n→∞
n
10
n
= 0.
Thus the sequence
1
10
,
2
100
,
3
1000
,
4
10000
, . . . is convergent with limit 0.
22 Chapter 1. Analysis
2. lim
n→∞
2
1
n
= 1.
2 > 1 and so 2
1
n
> 1 (for otherwise 2 = (2
1
n
)
n
≤ 1, a contradiction). Let a
n
:= 2
1
n
− 1 ≥ 0.
Then 2 = (1 +a
n
)
n
≥ 1 +na
n
(It can be shown using induction that for all real x ≥ −1 and
for all n ∈ N, (1 +x)
n
≥ 1 +nx.) Hence
0 ≤ a
n
≤
1
n
and so using the Sandwich theorem it follows that lim
n→∞
a
n
= 0. Consequently, lim
n→∞
2
1
n
= 1,
that is, the sequence 2,
√
2,
3
√
2,
4
√
2,
5
√
2, . . . is convergent with limit 1.
3. For any a, b ∈ R, lim
n→∞
([a[
n
+[b[
n
)
1
n
= max¦[a[, [b[¦.
Clearly,
(max¦[a[, [b[¦)
n
≤ [a[
n
+[b[
n
≤ (max¦[a[, [b[¦)
n
+ (max¦[a[, [b[¦)
n
and so
max¦[a[, [b[¦ ≤ ([a[
n
+[b[
n
)
1
n
≤ 2
1
n
max¦[a[, [b[¦.
So using the Sandwich theorem, it follows that lim
n→∞
([a[
n
+[b[
n
)
1
n
= max¦[a[, [b[¦.
In particular, with a = 24 and b = 2005, we have that lim
n→∞
(24
n
+ 2005
n
)
1
n
= 2005, that is,
the sequence 2028, 2005.1436, 2005.001146260873, ... is convergent with limit 2005.
4. If a ∈ (0, 1), then lim
n→∞
a
n
= 0.
Since 0 < a < 1, it follows that 1 <
1
a
and so h :=
1
a
−1 > 0. Then we have
1
a
n
= (1 +h)
n
≥ 1 +nh ≥ nh
and so
0 ≤ a
n
≤
1
nh
.
Hence from the Sandwich theorem, it follows that lim
n→∞
a
n
= 0.
5. lim
n→∞
_
1
n
2
+ 1
+
1
n
2
+ 2
+ +
1
n
2
+n
_
= 0.
For all n ∈ N, we have
n
n
2
+n
≤
1
n
2
+ 1
+
1
n
2
+ 2
+ +
1
n
2
+n
≤
n
n
2
+ 1
,
and since
lim
n→∞
n
n
2
+n
= 0 = lim
n→∞
n
n
2
+ 1
.
it follows from the Sandwich theorem that lim
n→∞
_
1
n
2
+ 1
+
1
n
2
+ 2
+ +
1
n
2
+n
_
= 0. ♦
Exercises.
1. Prove that the sequence
_
n!
n
n
_
n∈N
is convergent and lim
n→∞
n!
n
n
= 0.
Hint: Observe that 0 ≤
n!
n
n
=
1
n
2
n
n
n
≤
1
n
1 1 ≤
1
n
.
1.2. Sequences and limits 23
2. Prove that for all k ∈ N, the sequence
_
1
k
+ 2
k
+ 3
k
+ +n
k
n
k+2
_
n∈N
is convergent and
lim
n→∞
1
k
+ 2
k
+ 3
k
+ +n
k
n
k+2
= 0.
3. (a) Using induction, prove that if x ≥ −1 and n ∈ N, then
(1 +x)
n
≥ 1 +nx. (1.10)
(b) Show that for all n ∈ N,
1 ≤ n
1
n
< (1 +
√
n)
2
n
≤
_
1 +
1
√
n
_
2
.
Hint: Take x =
1
√
n
in the inequality (1.10).
(c) Prove that (n
1
n
)
n∈N
is convergent and ﬁnd its limit.
4. Let (a
n
)
n∈N
be a sequence contained in the interval (a, b) (that is, for all n ∈ N, a < a
n
< b).
If (a
n
)
n∈N
is convergent with limit L, then prove that L ∈ [a, b].
Hint: Use Exercise 5 on page 13.
Give an example to show that L needn’t belong to (a, b).
5. Let (a
n
)
n∈N
be a convergent sequence, and let (b
n
)
n∈N
satisfy [b
n
− a
n
[ <
1
n
for all n ∈ N.
Show that (b
n
)
n∈N
is also convergent. What is its limit?
Hint: Observe that −
1
n
+a
n
< b
n
< a
n
+
1
n
for all n ∈ N.
1.2.6 Subsequences
In this section we prove an important result in analysis, known as the BolzanoWeierstrass theorem,
which says that every bounded sequence has a convergent ‘subsequence’. We begin this section
by deﬁning what we mean by a subsequence of a sequence.
Deﬁnition. Let (a
n
)
n∈N
be a sequence and let (n
k
)
k∈N
be a sequence of natural numbers such
that n
1
< n
2
< n
3
< . . . . Then (a
n
k
)
k∈N
is called a subsequence of (a
n
)
n∈N
.
Examples.
1.
_
1
2n
_
n∈N
,
_
1
n
2
_
n∈N
,
_
1
n!
_
n∈N
and
_
1
n
n
_
n∈N
are all subsequences of
_
1
n
_
n∈N
. Also the sequence
(a
n
)
n∈N
deﬁned by
a
n
=
1
p
, where p is the nth prime in the inﬁnite sequence of increasing primes 2, 3, 5, 7, 11, . . .
is a subsequence of
_
1
n
_
n∈N
. The sequence
1
2
, 1,
1
3
,
1
4
, . . .
is not a subsequence of
_
1
n
_
n∈N
.
24 Chapter 1. Analysis
2. The sequences ((−1)
2n
)
n∈N
(that is, the constant sequence 1, 1, 1, . . . ) and ((−1)
2n−1
)
n∈N
(that is the constant sequence −1, −1, −1, . . . ) are subsequences of ((−1)
n
)
n∈N
. ♦
Exercise(∗). Beginning with 2 and 7, the sequence 2, 7, 1, 4, 7, 4, 2, 8, 2, 8, . . . is constructed by
multiplying successive pairs of its terms and adjoining the result as the next one or two members
of the sequence depending on whether the product is a one or twodigit number. Thus we start
with 2 and 7, giving the product 14, and so the next two terms are 1, 4. Proceeding in this manner,
we get subsequent terms as follows:
2, 7
2, 7, 1, 4
2, 7, 1, 4
2, 7, 1, 4, 7
2, 7, 1, 4, 7
2, 7, 1, 4, 7, 4
2, 7, 1, 4, 7, 4
2, 7, 1, 4, 7, 4, 2, 8
2, 7, 1, 4, 7, 4, 2, 8
2, 7, 1, 4, 7, 4, 2, 8, 2, 8
.
.
.
Prove that this sequence has the constant subsequence 6, 6, 6, . . . .
Hint: Show that 6 appears an inﬁnite number of times as follows. Since the terms 2, 8, 2, 8 are
adjacent, they give rise to the adjacent terms 1, 6, 1, 6 at some point, which in turn give rise to
the adjacent terms 6, 6, 6 eventually, and so on. Proceeding in this way, ﬁnd out if you get a loop
containing the term 6.
Theorem 1.2.6 If (a
n
)
n∈N
is a convergent sequence with limit L, then any subsequence of (a
n
)
n∈N
is also convergent with the limit L.
Proof Let (a
n
k
)
k∈N
be a subsequence of (a
n
)
n∈N
. Given > 0, let N ∈ N be such that for all
n > N, [a
n
−L[ < . Since the sequence n
1
< n
2
< n
3
< . . . , it follows that there exists a K ∈ N
such that n
K
> N. Then for all k > K, n
k
> n
K
> N. Hence for k > K, [a
n
k
− L[ < , and so
(a
n
k
)
k∈N
is convergent with limit L.
Examples.
1.
_
1
2n
_
n∈N
,
_
1
n
2
_
n∈N
,
_
1
n!
_
n∈N
and
_
1
n
n
_
n∈N
are convergent sequences with limit 0.
2. The sequence ((−1)
n
)
n∈N
is divergent since the subsequence 1, 1, 1, . . . has limit 1, while the
subsequence −1, −1, −1, . . . has limit −1. ♦
Theorem 1.2.7 Every sequence has a monotone subsequence.
1.2. Sequences and limits 25
We ﬁrst give an illustration of the idea behind this proof. Assume that (a
n
)
n∈N
is the given
sequence. Imagine that a
n
is the height of the hotel with number n, which is followed by hotel
n + 1, and so on, along an inﬁnite line, where at inﬁnity there is the sea. A hotel is said to have
the seaview property if it is higher than all hotels following it. See Figure 1.7. Now there are only
PSfrag replacements
1 2 3 4 5 6
. . .
. . .
Figure 1.7: The seaview property.
two possibilities:
1
◦
There are inﬁnitely many hotels with the seaview property. Then their heights form a decreasing
subsequence.
2
◦
There is only a ﬁnite number of hotels with the seaview property. Then after the last hotel
with the seaview property, one can start with any hotel and then always ﬁnd one that is at least
as high, which is taken as the next hotel, and then ﬁnding yet another that is at least as high as
that one, and so on. The heights of these hotels form an increasing subsequence.
Proof Let
S = ¦m ∈ N [ for all n > m, a
n
< a
m
¦.
Then we have the following two cases.
1
◦
S is inﬁnite. Arrange the elements of S in increasing order: n
1
< n
2
< n
3
< . . . . Then (a
n
k
)
k∈N
is a decreasing subsequence of (a
n
)
n∈N
.
2
◦
S is ﬁnite. If S empty, then deﬁne n
1
= 1, and otherwise let n
1
= max S+1. Deﬁne inductively
n
k+1
= min¦m ∈ N [ m > n
k
and a
m
≥ a
n
k
¦.
(The minimum exists since the set ¦m ∈ N [ m > n
k
and a
m
≥ a
n
k
¦ is a nonempty subset of N:
indeed otherwise if it were empty, then n
k
∈ S, and this is not possible if S was empty, and also
impossible if S was not empty, since n
k
> max S.) Then (a
n
k
)
k∈N
is an increasing subsequence of
(a
n
)
n∈N
.
An important consequence of the above theorem is the following result.
Theorem 1.2.8 (BolzanoWeierstrass theorem.) Every bounded sequence has a convergent sub
sequence.
Proof Let (a
n
)
n∈N
be a bounded sequence. Then there exists a M > 0 such that for all n ∈ N,
[a
n
[ ≤ M. From Theorem 1.2.7 above, it follows that the sequence (a
n
)
n∈N
has a monotone
subsequence (a
n
k
)
k∈N
. Then clearly for all k ∈ N, [a
n
k
[ ≤ M and so the sequence (a
n
k
)
k∈N
is
also bounded. Since (a
n
k
)
k∈N
is monotone and bounded, it follows from Theorem 1.2.3 that it is
convergent.
26 Chapter 1. Analysis
Example. Consider the sequence (a
n
)
n∈N
of fractional parts of integral multiples of
√
2, deﬁned
by
a
n
= n
√
2 −¸n
√
2, for n ∈ N,
where for x ∈ R, ¸x denotes the greatest integer part of x, which is deﬁned as the largest integer
less than or equal to x:
¸x = min¦z ∈ Z [ x < z¦ −1.
(By the Archimedean property, there exist natural numbers n
1
and n
2
such that x < n
1
and
−x < n
2
, that is, −n
2
< x < n
1
. Hence the set ¦z ∈ Z [ x < z¦ is a nonempty subset of Z
(indeed, n
1
belongs to it!), and it is bounded below (by −n
2
∈ Z), and so the minimum of the set
¦z ∈ Z [ x < z¦ exists in Z. Consequently ¸ is a welldeﬁned function.)
The terms of the sequence (a
n
)
n∈N
are as follows:
√
2 = 1.414213 . . . a
1
= 0.414213 . . .
2
√
2 = 2.828427 . . . a
2
= 0.828427 . . .
3
√
2 = 4.242640 . . . a
3
= 0.242640 . . .
4
√
2 = 5.656854 . . . a
4
= 0.656854 . . .
5
√
2 = 7.071067 . . . a
5
= 0.071067 . . .
6
√
2 = 8.485281 . . . a
6
= 0.485281 . . .
.
.
.
The sequence (a
n
)
n∈N
is bounded: indeed, 0 ≤ a
n
< 1. So by the BolzanoWeierstrass theorem it
has a convergent subsequence
6
. ♦
Exercise. (∗) Recall the deﬁnition of a Cauchy sequence from Exercise 6 on page 13, where we
had already seen that every convergent sequence is Cauchy. Using BolzanoWeierstrass theorem,
we can prove the converse. Show that if a sequence is Cauchy, then it is convergent.
Hint: Proceed as follows. Let (a
n
)
n∈N
be a Cauchy sequence. From Exercise 5 on page 17, it
follows that (a
n
)
n∈N
is bounded. By the Bolzano Weierstrass theorem, it follows that (a
n
)
n∈N
has a convergent subsequence, say (a
n
k
)
k∈N
with limit L. Prove (using the fact that (a
n
)
n∈N
is
Cauchy), that then (a
n
)
n∈N
is itself convergent with limit L.
1.3 Continuity
A function f : R → R is a rule of correspondence that assigns to each real number a unique
real number. Many bizarre functions make appearances in analysis, and in order to avoid falling
into pitfalls with simplistic thinking, we need deﬁnitions and hypothesis of theorems to be stated
carefully and clearly.
Within the huge collection of functions, there is an important subset: the continuous functions.
Continuous functions play a prominent role in analysis since they possess some useful properties.
In this section we give the formal deﬁnition of continuous functions and prove two of the most
important properties: the extreme value theorem and the intermediate value theorem.
6
In fact, it can be shown that these fractional parts an are “dense” in (0, 1). Thus given any number L ∈ (0, 1),
there exists a subsequence of the sequence (an)
n∈N
above that converges to L.
1.3. Continuity 27
1.3.1 Deﬁnition of continuity
In everyday speech, a ‘continuous’ process is one that proceeds without gaps of interruptions or
sudden changes. What does it mean for a function f : R → R to be continuous? The common
informal deﬁnition of this concept states that a function f is continuous if one can sketch its graph
without lifting the pencil. In other words, the graph of f has no breaks in it. If a break does
occur in the graph, then this break will occur at some point. Thus (based on this visual view of
continuity), we ﬁrst give the formal deﬁnition of the continuity of a function at a point below.
Next, if a function is continuous at each point, then it will be called continuous.
If a function has a break at a point, say c, then even if points x are close to c, the points f(x)
do not get close to f(c). See Figure 1.8.
PSfrag replacements
c
f(c)
Figure 1.8: A function with a break at c. If x lies to the left of c, then f(x) is not close to f(c),
no matter how close x comes to c.
This motivates the following deﬁnition of continuity, which guarantees that if a function is
continuous at a point c, then we can make f(x) as close as we like to f(c), by choosing x suﬃciently
close to c. See Figure 1.9.
PSfrag replacements
f(c) +
f(c)
f(c) +
f(x)
c −δ c c +δ x
Figure 1.9: The deﬁnition of the continuity of a function at point c. If the function is continuous
at c, then given any > 0 (which determines a strip around the line y = f(c) of width 2), there
exists a δ > 0 (which determines an interval of width 2δ around the point c) such that whenever
x lies in this width (so that x satisﬁes c − δ < x < c + δ, that is, [x − c[ < δ), then f(x) satisﬁes
f(c) − < f(x) < f(c) +, that is, [f(x) −f(c)[ < .
Deﬁnitions. Let I be an interval in R and let c ∈ I. A function f : I → R is continuous at c if
for every > 0, there exists a δ > 0 such that for all x ∈ I satisfying [x −c[ < δ, [f(x) −f(c)[ < .
A function f : I →R is continuous (on I) if for every c ∈ I, f is continuous at c.
Examples.
1. f : R →R given by f(x) = 1 for all x ∈ R is continuous.
28 Chapter 1. Analysis
Let c ∈ R = (−∞, ∞). Given > 0, let δ be an arbitrary positive real number; for instance,
let δ = 1. Then if x ∈ R and [x −c[ < δ = 1, we have:
[f(x) −f(c)[ = [1 −1[ = [0[ = 0 < .
So f is continuous at c. Since the choice of c ∈ R was arbitrary, it follows that f is continuous
on R. See Figure 1.10.
PSfrag replacements
1
0
x
Figure 1.10: The continuous constant function 1.
2. f : R →R given by
f(x) =
_
0 if x = 0,
1 if x ∈ R ¸ ¦0¦.
is not continuous at 0.
Suppose that it is continuous at 0. Then given =
1
2
> 0, let δ > 0 be such that if x ∈ R and
[x[ = [x−0[ < δ, then [f(x) −f(0)[ = [f(x) −0[ = [f(x)[ < =
1
2
. But now take x =
δ
2
∈ R,
and so [x[ =
¸
¸
δ
2
¸
¸
=
δ
2
< δ. Thus [f(x)[ =
¸
¸
f
_
δ
2
_¸
¸
= [1[ = 1 >
1
2
= , a contradiction. So f
cannot be continuous at 0.
However, for all c ∈ R ¸ ¦0¦, f is continuous at c. This can be seen as follows. Given > 0,
let δ =
c
2
> 0. Then if x ∈ R and [x −c[ < δ, we have
[c[ −[x[ ≤ [[c[ −[x[[ ≤ [c −x[ = [x −c[ < δ =
[c[
2
and so
[x[ >
[c[
2
> 0.
Thus x ,= 0 and so f(x) = 1. Hence if x ∈ R and [x −c[ < δ, we obtain
[f(x) −f(c)[ = [1 −1[ = [0[ = 0 < .
Consequently f is continuous at c. See Figure 1.11.
PSfrag replacements
1
0
x
Figure 1.11: A function continuous everywhere except at 0.
3. f : R →R given by f(x) = x for all x ∈ R is continuous.
Let c ∈ R. Given > 0, let δ = . Then if x ∈ R and [x −c[ < δ, we have:
[f(x) −f(c)[ = [x −c[ < δ = .
So f is continuous at c. Since the choice of c ∈ R was arbitrary, it follows that f is continuous
on R.
1.3. Continuity 29
4. f : (0, 1) →R given by f(x) =
1
x
for all x ∈ R is continuous (on (0, 1)).
Let c ∈ (0, 1). Given > 0, let δ = min
_
c
2
,
c
2
2
_
(> 0). Then if x ∈ (0, 1) and [x − c[ < δ,
we have
[c[ −[x[ ≤ [[c[ −[x[[ ≤ [c −x[ = [x −c[ < δ ≤
[c[
2
and so
[c[
2
< [x[, that is,
1
[x[
<
2
[c[
.
Consequently, if x ∈ (0, 1) and [x −c[ < δ,
¸
¸
¸
¸
1
x
−
1
c
¸
¸
¸
¸
=
[c − x[
[x[ [c[
=
[x −c[
[x[ [c[
< δ
2
[c[
1
[c[
=
2δ
c
2
≤ .
So f is continuous at c. Since the choice of c ∈ (0, 1) was arbitrary, it follows that f is
continuous (on (0, 1)). ♦
Exercises.
1. Let the function f : R →R be given by f(x) = x
2
.
(a) Prove that f is continuous at 0.
(b) (∗) Suppose that c is a nonzero real number. Prove that f is continuous at c.
In Exercise 1 on page 31, we will give a slick proof of the fact that f is continuous on R.
2. Let f : R →R be a function that satisﬁes f(x +y) = f(x) +f(y) for all x, y ∈ R.
(a) Suppose that f is continuous at some real number c. Prove that f is continuous on R.
Hint: Since f is continuous at c, given > 0, ∃δ > 0 such that for all x ∈ R satisfying
[x − c[ < δ, [f(x) − f(c)[ < . Show that given any other point c
∈ R, the function f
is continuous at c
by showing that the same δ works (for this ).
(b) Give an example of such a function.
3. Suppose that f : R → R and there exists a M > 0 such that for all x ∈ R, [f(x)[ ≤ M[x[.
Prove that f is continuous at 0.
Hint: Find f(0).
4. Let f : R →R be deﬁned by
f(x) =
_
0 if x is rational,
1 if x is irrational.
Prove that for every c ∈ R, f is not continuous at c.
Hint: Use the fact that there are irrational numbers arbitrarily close to any rational number
and rational numbers arbitrarily close to any irrational number.
5. Let f : R →R be a continuous function. Prove that if for some c ∈ R, f(c) > 0, then there
exists a δ > 0 such that for all x ∈ (c −δ, c +δ), f(x) > 0.
If f and g are functions on R, then the composition of g with f, denoted by g ◦ f, is deﬁned
by
(g ◦ f)(x) = g(f(x)), x ∈ R.
The following theorem implies that the composition of continuous functions is continuous.
30 Chapter 1. Analysis
Theorem 1.3.1 Let c ∈ R. If f : R →R is continuous at c and g : R →R is continuous at f(c),
then g ◦ f is continuous at c.
Proof Since g is continuous at f(c), it follows that given > 0, ∃δ > 0 such that for all y ∈ R
satisfying [y −f(c)[ < δ,
[g(y) −g(f(c))[ < .
Since f is continuous at c, ∃δ
1
> 0 such that for all x ∈ R satisfying [x −c[ < δ
1
,
[f(x) −f(c)[ < δ.
Consequently, for all x ∈ R satisfying [x −c[ < δ
1
, we have [f(x) −f(c)[ < δ, and so
[g(f(x)) −g(f(c))[ < ,
that is,
[(g ◦ f)(x) −(g ◦ f)(c)[ < .
Hence g ◦ f is continuous at c.
1.3.2 Continuous functions preserve convergent sequences
We now give an alternative characterization of continuity.
Theorem 1.3.2 Let I be an interval in R and let c ∈ I. Suppose that f : I → R is a function.
Then f is continuous at c iﬀ
for every convergent sequence (x
n
)
n∈N
contained in I with limit c,
(f(x
n
))
n∈N
is convergent and lim
n→∞
f(x
n
) = f(c).
(1.11)
Proof
Only if: Suppose that f is continuous at c ∈ I and let (x
n
)
n∈N
be a convergent sequence contained
in I with limit c.
Since f is continuous at c ∈ I, given > 0, ∃δ > 0 such that for all x ∈ I satisfying [x−c[ < δ,
[f(x) −f(c)[ < .
Since (x
n
)
n∈N
is convergent with limit c, ∃N ∈ N such that for all n > N, [x
n
−c[ < δ.
Consequently for n > N, [f(x
n
) −f(c)[ < . So (f(x
n
))
n∈N
is convergent with limit f(c).
If: Suppose that (1.11) holds. Then we need to show that f is continuous at c and we prove this
by contradiction. Assume that f is not continuous at c, that is,
[∀ > 0 ∃δ > 0 such that ∀x ∈ I such that [x −c[ < δ, [f(x) −f(c)[ < ]
that is,
∃ > 0 such that ∀δ > 0 ∃x ∈ I such that [x −c[ < δ but [f(x) −f(c)[ ≥ .
Hence if δ =
1
n
, then we can ﬁnd x
n
∈ I such that [x
n
−c[ < δ =
1
n
, but [f(x
n
) −f(c)[ ≥ .
Claim 1: The sequence (x
n
)
n∈N
is contained in I and is convergent with limit c.
1.3. Continuity 31
Indeed, we have for all n ∈ N, x
n
∈ I. Furthermore, given any ζ > 0, we can ﬁnd N ∈ N such
that
1
ζ
< N (Archimedean property), that is,
1
N
< ζ. Hence for n > N, [x
n
− c[ <
1
N
< ζ. So
(x
n
)
n∈N
is convergent with limit c.
Claim 2: The sequence (f(x
n
))
n∈N
does not converge to f(c).
Indeed for all n ∈ N, we have [f(x
n
)−f(c)[ ≥ . Thus for instance
2
> 0, but it is not possible
to ﬁnd a large enough N ∈ N such that for all n > N, [f(x
n
) −f(c)[ <
2
(for if this were possible,
then we would arrive at the contradiction ≤ [f(x
n
) −f(c)[ <
2
).
The Claims 1 and 2 show that (1.11) does not hold, a contradiction. Hence f is continuous
at c.
Exercises.
1. Recall Exercise 1 on page 29: The function f(x) : R → R is given by f(x) = x
2
. Using the
characterization of continuity provided in Theorem 1.3.2, prove that f is continuous on R.
2. Prove that
7
if f : R → R is continuous and f(x) = 0 if x is rational, then f(x) = 0 for all
x ∈ R.
Hint: Given any real number c, there exists a sequence of rational numbers (q
n
)
n∈N
that
converges to c.
3. Let f : R →R be a function that preserves divergent sequences, that is, for every divergent
sequence (x
n
)
n∈N
, (f(x
n
))
n∈N
is divergent as well. Prove that f is onetoone.
Hint: Let x
1
, x
2
be distinct real numbers, and consider the sequence x
1
, x
2
, x
1
, x
2
, . . . .
Using the above theorem, we obtain the following useful Theorem 1.3.3. But before we state
this result, we introduce some convenient notation.
Let I be an interval in R. Given functions f : I →R and g : I →R, we deﬁne the following:
1. If α ∈ R, then we deﬁne the function αf : I →R by (αf)(x) = α f(x), x ∈ I.
2. We deﬁne the absolute value of f, [f[ : I →R by [f[(x) = [f(x)[, x ∈ I.
3. The sum of f and g, f +g : I →R is deﬁned by (f +g)(x) = f(x) +g(x), x ∈ I.
4. The product of f and g, fg : I →R is deﬁned by (fg)(x) = f(x)g(x), x ∈ I.
5. If k ∈ N, then we deﬁne the kth power of f, f
k
: I →R by f
k
(x) = (f(x))
k
, x ∈ I.
6. If for all x ∈ I, g(x) ,= 0, then we deﬁne
1
g
: I →R by
_
1
g
_
(x) =
1
g(x)
, x ∈ I.
Theorem 1.3.3 Let I be an interval in R and let c ∈ I. Suppose that f : I → R and g : I → R
are continuous at c. Then:
1. For all α ∈ R, αf is continuous at c.
2. [f[ is continuous at c.
3. f +g is continuous at c.
7
See Exercise 4 on page 29.
32 Chapter 1. Analysis
4. fg is continuous at c.
5. For all k ∈ N, f
k
is continuous at c.
6. If for all x ∈ I, g(x) ,= 0, then
1
g
is continuous at c.
Proof Suppose that (x
n
)
n∈N
is a convergent sequence contained in I, with limit c. Since f and g
are continuous at c, from Theorem 1.3.2 , it follows that (f(x
n
))
n∈N
and (g(x
n
))
n∈N
are convergent
with limits f(c) and g(c), respectively. Hence from Theorem 1.2.4, it follows that:
1. (α f(x
n
))
n∈N
is convergent with limit α f(c), that is, ((αf)(x
n
))
n∈N
is convergent with
limit (αf)(c). So from Theorem 1.3.2 , it follows that αf is continuous at c.
2. ([f(x
n
)[)
n∈N
is convergent with limit [f(c)[, that is, ([f[(x
n
))
n∈N
is convergent with limit
[f[(c). So from Theorem 1.3.2 , it follows that [f[ is continuous at c.
3. (f(x
n
)+g(x
n
))
n∈N
is convergent with limit f(c)+g(c), that is, ((f +g)(x
n
))
n∈N
is convergent
with limit (f +g)(c). So from Theorem 1.3.2 , it follows that f +g is continuous at c.
4. (f(x
n
)g(x
n
))
n∈N
is convergent with limit f(c)g(c), that is, ((fg)(x
n
))
n∈N
is convergent with
limit (fg)(c). So from Theorem 1.3.2 , it follows that fg is continuous at c.
5. ((f(x
n
))
k
)
n∈N
is convergent with limit (f(c))
k
, that is, (f
k
(x
n
))
n∈N
is convergent with limit
f
k
(c). So from Theorem 1.3.2 , it follows that f
k
is continuous at c.
6.
_
1
g(xn)
_
n∈N
is convergent with limit
1
g(c)
(since for all x ∈ I, g(x) ,= 0, in particular g(x
n
) ,= 0
and g(c) ,= 0), that is,
__
1
g
_
(x
n
)
_
n∈N
is convergent with limit
_
1
g
_
(c). So from Theorem
1.3.2 , it follows that
1
g
is continuous at c.
Example. Since f : R → R given by f(x) = x for x ∈ R is continuous (see Example 3 on page
28), it follows that for all k ∈ N, x
k
is continuous. Thus given arbitrary scalars a
0
, a
1
, . . . , a
N
in R,
it follows that the functions a
0
1, a
1
x, . . . , a
N
x
N
are continuous. Consequently the polynomial
function p : R →R deﬁned by
p(x) = a
0
+a
1
x + +a
N
x
N
, x ∈ R
is continuous. ♦
Exercise. Show that the rational function f : R →R deﬁned by
f(x) =
x
2
1 +x
2
, x ∈ R,
is continuous on R.
1.3.3 Extreme value theorem
Below we show that a continuous function on an interval [a, b] attains its maximum and minimum
values. See Figure 1.12.
1.3. Continuity 33
PSfrag replacements
a
b
c
d
Figure 1.12: Extreme value theorem.
Theorem 1.3.4 (Extreme value theorem). If f : [a, b] → R is continuous, then there exist
c, d ∈ [a, b] such that
f(c) = sup¦f(x) [ x ∈ [a, b]¦, and (1.12)
f(d) = inf¦f(x) [ x ∈ [a, b]¦. (1.13)
Note that since c, d ∈ [a, b], the supremum and inﬁmum in (1.12) and (1.13) are in fact the
maximum and minimum, respectively.
Proof
Step 1. We ﬁrst show that f is bounded, that is, the set
S = ¦f(x) [ x ∈ [a, b]¦
is bounded. Suppose that S is not bounded. Then given n ∈ N, ∃x
n
∈ [a, b] such that [f(x
n
)[ > n
(for if this fails for some N ∈ N, then [f(x)[ ≤ N for all x ∈ [a, b], and so S would be bounded, a
contradiction). In this way, we get a sequence (x
n
)
n∈N
. Since
a ≤ x
n
≤ b for all n ∈ N, (1.14)
(x
n
)
n∈N
is bounded, and so by the BolzanoWeierstrass theorem (Theorem 1.2.8), it follows that
it has a convergent subsequence (x
n
k
)
k∈N
that converges to some limit L. Now we show that
L ∈ [a, b]. If not, then either L > b or L < a. If L > b, then L −b > 0, and so ∃K ∈ N such that
for all k > K, [x
n
k
−L[ < L −b. But then for all k > K, we obtain
L −x
n
k
≤ [L −x
n
k
[ = [x
n
k
−L[ < L −b,
that is, x
n
k
> b, which contradicts (1.14). Similarly, if L < a, then a − L > 0, and so ∃K ∈ N
such that for all k > K, [x
n
k
−L[ < a −L. But then for all k > K, we obtain
x
n
k
−L ≤ [x
n
k
−L[ < a −L,
that is, x
n
k
< a, which contradicts (1.14). Thus L ∈ [a, b]. But by Theorem 1.2.2 it follows
that (f(x
n
k
))
k∈N
cannot be convergent, since it is not bounded (indeed, [f(x
n
k
)[ > n
k
!). From
Theorem 1.3.2, we see that this contradicts the continuity of f. So S must be bounded.
Step 2. Let M := sup¦f(x) [ x ∈ [a, b]¦. We prove that there exists a c ∈ [a, b] such that
f(c) = M. For each n ∈ N, M −
1
n
< M, and so M −
1
n
cannot be an upper bound for
¦f(x) [ x ∈ [a, b]¦. So ∃x
n
∈ [a, b] such that
M −
1
n
< f(x
n
) ≤ M. (1.15)
By the BolzanoWeierstrass theorem, (x
n
)
n∈N
has a convergent subsequence (x
n
k
)
k∈N
with limit
c ∈ [a, b]. Since f is continuous (f(x
n
k
))
k∈N
is convergent with limit f(c). From (1.15), it follows
34 Chapter 1. Analysis
from the Sandwich theorem that f(c) = M. This proves the existence of c ∈ [a, b] such that (1.12)
holds.
Step 3. Let m := inf¦f(x) [ x ∈ [a, b]¦. The proof that there exists a d ∈ [a, b] such that f(d) = m
is similar to Step 2 above, and is left as an exercise to the reader.
Examples.
1. The function f : [0, 1] → R given by f(x) = 1 for all x ∈ [0, 1] is continuous and for any
c, d ∈ [0, 1] we have
f(c) = 1 = sup¦f(x) [ x ∈ [0, 1]¦ = inf¦f(x) [ x ∈ [0, 1]¦ = 1 = f(d).
2. The function f : (0, 1) →R given by f(x) = x is continuous on (0, 1). If
S = ¦x [ x ∈ (0, 1)¦,
then sup S = 1 and inf S = 0, but these values are not attained. Thus the statement of
Theorem 1.3.4 does not hold if [a, b] is replaced by (a, b). ♦
Exercises.
1. Complete the details in Step 3 of the proof of Theorem 1.3.4.
2. Give an example of a function f : [0, 1] → R that is not continuous on [0, 1], but f satisﬁes
the conclusion of Theorem 1.3.4.
3. A function f : R →R is periodic if there exists a T > 0 such that for all x ∈ R, f(x + T) =
f(x). If f : R →R is continuous and periodic, then prove that f is bounded, that is, the set
S = ¦f(x) [ x ∈ R¦ is bounded.
4. Let f : [a, b] →R be continuous on [a, b], and deﬁne f
∗
as follows:
f
∗
(x) =
_
f(a) if x = a,
max¦f(y) [ y ∈ [a, x]¦ if x ∈ (a, b].
(a) Show that f
∗
is a welldeﬁned function.
(b) (∗) Prove that f
∗
is continuous on [a, b].
(c) If f : [−1, 1] →R is given by f(x) = x −x
2
, then ﬁnd f
∗
.
1.3.4 Intermediate value theorem
We now prove one of the most fundamental (and obvious!) theorems on continuous functions:
a continuous function cannot “hop over” intermediate values. For instance, if the height of a
mountain is 1976 meters above sea level, then given any number between 0 and 1976, say 399,
there must exist a point on the mountain that is exactly 399 meters above sea level. See Figure
1.13.
Theorem 1.3.5 (Intermediate value theorem). If f : [a, b] →R is continuous and y is such that
f(a) ≤ y ≤ f(b), then there exists a c ∈ [a, b] such that f(c) = y.
1.3. Continuity 35
PSfrag replacements
a b c
f(a)
f(b)
y
Figure 1.13: Intermediate value theorem.
Note that in the above statement of the theorem, the order f(a) ≤ y ≤ f(b) can be reversed
without changing the conclusion. Indeed, if y ∈ R is such that f(b) ≤ y ≤ f(a), then we have
(−f)(a) ≤ −y ≤ (−f)(b),
and applying the above Theorem 1.3.5 to the continuous function −f (Theorem 1.3.3.1 with
α = −1!), we get the existence of c ∈ [a, b] such that (−f)(c) = −y, that is, f(c) = y.
Proof Suppose that f(a) ≤ y ≤ f(b), and deﬁne
S = ¦x ∈ [a, b] [ y ≤ f(x)¦.
S is not empty (since b ∈ S) and S is bounded below by a (since S ⊂ [a, b]). Thus S has an
inﬁmum, say c.
We claim that f(c) = y. Suppose that our claim is false. Then either f(c) < y or f(c) > y.
1
◦
Suppose that f(c) < y. Thus := y − f(c) > 0, and since f is continuous at c, ∃δ > 0 such
that for all x ∈ [a, b] such that [x −c[ < δ, [f(x) −f(c)[ < y − f(c). Since c is the inﬁmum of S,
∃x ∈ S such that c ≤ x < c +δ, that is, x ∈ [a, b], 0 ≤ x −c = [x −c[ < δ and y ≤ f(x). Thus we
have
y −f(c) ≤ f(x) −f(c) ≤ [f(x) −f(c)[ < y −f(c),
a contradiction.
2
◦
Suppose that f(c) > y. Thus := f(c) − y > 0, and since f is continuous at c, ∃δ > 0 and
moreover satisfying
8
δ ≤ c − a , such that for all x ∈ [a, b] such that [x − c[ < δ, [f(x) − f(c)[ <
y −f(c). Let x := c −
δ
2
. Then x ∈ [a, c] ⊂ [a, b]: indeed,
a ≤
c +a
2
≤ c −
δ
2
≤ c ≤ b.
Furthermore,
[x −c[ = c −x = c −
_
c −
δ
2
_
=
δ
2
< δ.
Hence f(c) − f(x) ≤ [f(c) − f(x)[ = [f(x) − f(c)[ < f(c) − y, and so f(x) ≥ y. Consequently,
x ∈ S and since c = inf S, we have c ≤ x. Thus we obtain
c ≤ x = c −
δ
2
< c,
a contradiction.
8
If not, then replace δ by
δ
N
with N ∈ N large enough to guarantee
δ
c−a
< N.
36 Chapter 1. Analysis
So f(c) = y, and this completes the proof.
Examples.
1. Every odd polynomial with real coeﬃcients has at least one real root.
Suppose that p is a polynomial with degree 2m+ 1, m ∈ N ∪ ¦0¦. Let
p(x) = a
0
+a
1
x + +a
2m
x
2m
+a
2m+1
x
2m+1
,
where a
2m+1
,= 0.
In order to show that p has a real root, we will choose a large enough N ∈ N such that
p(N) and p(−N) have opposite signs, and then restrict our attention to an interval [−N, N].
Then appealing to the intermediate value theorem, we can conclude that p must vanish at
some point in this interval, that is, for some real c ∈ [−N, N], p(c) = 0. The proof is long,
and so we have divided it into a sequence of steps.
Step 1. For large positive n, p(n) has the same sign as a
2m+1
.
Since
lim
n→∞
_
a
0
n
2m+1
+
a
1
n
2m
+ +
a
2m
n
_
= 0,
it follows that there exists N
1
∈ N such that for all n > N
1
,
¸
¸
¸
¸
p(n)
n
2m+1
−a
2m+1
¸
¸
¸
¸
<
[a
2m+1
[
2
.
Now we show that for n > N
1
, p(n) has the same sign as a
2m+1
. Since a
2m+1
,= 0, either
a
2m+1
> 0 or a
2m+1
< 0. If a
2m+1
> 0, then for all n > N
1
a
2m+1
−
p(n)
n
2m+1
≤
¸
¸
¸
¸
a
2m+1
−
p(n)
n
2m+1
¸
¸
¸
¸
=
¸
¸
¸
¸
p(n)
n
2m+1
−a
2m+1
¸
¸
¸
¸
<
[a
2m+1
[
2
=
a
2m+1
2
,
and so
0 <
a
2m+1
2
<
p(n)
n
2m+1
.
Thus p(n) is also positive for all n > N
1
. Similarly, if a
2m+1
< 0, then for all n > N
1
p(n)
n
2m+1
−a
2m+1
≤
¸
¸
¸
¸
p(n)
n
2m+1
−a
2m+1
¸
¸
¸
¸
<
[a
2m+1
[
2
= −
a
2m+1
2
,
and so
p(n)
n
2m+1
<
a
2m+1
2
< 0.
Thus p(n) is also negative for all n > N
1
.
So it follows that for n > N
1
, p(n) has the same sign as a
2m+1
.
Step 2. For large positive n, p(−n) has the same sign as −a
2m+1
.
Similarly, using the fact that
lim
n→∞
_
a
0
(−n)
2m+1
+
a
1
(−n)
2m
+ +
a
2m
(−n)
_
= 0,
we now show that there exists N
2
∈ N such that for all n > N
2
, p(−n) has the same sign as
−a
2m+1
. Let N
2
∈ N be such that for all n > N
2
,
¸
¸
¸
¸
p(−n)
n
2m+1
+a
2m+1
¸
¸
¸
¸
=
¸
¸
¸
¸
−
p(−n)
n
2m+1
−a
2m+1
¸
¸
¸
¸
<
[a
2m+1
[
2
.
1.3. Continuity 37
If a
2m+1
> 0, then for all n > N
2
p(−n)
n
2m+1
+a
2m+1
≤
¸
¸
¸
¸
p(−n)
n
2m+1
+a
2m+1
¸
¸
¸
¸
<
[a
2m+1
[
2
=
a
2m+1
2
,
and so
p(−n)
n
2m+1
< −
a
2m+1
2
< 0.
Thus p(−n) is negative for all n > N
2
. Similarly, if a
2m+1
< 0, then for all n > N
2
−
p(−n)
n
2m+1
−a
2m+1
≤
¸
¸
¸
¸
p(−n)
n
2m+1
+a
2m+1
¸
¸
¸
¸
<
[a
2m+1
[
2
= −
a
2m+1
2
,
and so
p(−n)
n
2m+1
> −
a
2m+1
2
> 0.
Thus p(−n) is positive for all n > N
2
.
So it follows that for n > N
2
, p(−n) has the same sign as −a
2m+1
.
Step 3. Application of the intermediate value theorem and the conclusion.
Hence if N := max¦N
1
, N
2
¦ + 1, then p(N) has the same sign as a
2m+1
, while p(−N) has
the same sign as −a
2m+1
. Thus the continuous function p must vanish a some point in the
interval [−N, N].
The polynomial p(x) = x
2005
−x
1976
+
1
399
has a real root in [−1, 1]: indeed p(1) =
1
399
> 0
and p(−1) = −2 +
1
399
< 0 and so ∃c ∈ [−1, 1] such that p(c) = 0.
2. At any given time, there exists a pair of diametrically opposite points on the equator which
have the same temperature.
Let T(Θ) denote the surface temperature at the point at longitude Θ. See Figure 1.14.
(Note that Θ(0) = Θ(2π).) Assuming that Θ is a continuous function of Θ, it follows that
PSfrag replacements
Θ
Figure 1.14: The point on the equator with longitude Θ.
the function f : [0, π] → R deﬁned by f(Θ) = T(Θ) − T(Θ + π) is continuous as well. If
f(0) = 0, then it follows that the temperatures at 0 and 180
◦
longitude are the same. If
f(0) ,= 0, then since f(0) and f(π) = −f(0) have opposite signs, by the intermediate value
theorem, it follows that f must vanish at some point, and so the claim follows. ♦
Exercises.
1. Suppose that f : [0, 1] →R is a continuous function such that for all x ∈ [0, 1], 0 ≤ f(x) ≤ 1.
Prove that there exists at least one c ∈ [0, 1] such that f(c) = c.
Hint: Consider the continuous function g(x) = f(x) − x, and use the intermediate value
theorem.
38 Chapter 1. Analysis
2. At 8:00 a.m. on Saturday, a hiker begins walking up the side of a mountain to his weekend
campsite. On Sunday morning at 8:00 a.m., he walks back down the mountain along the
same trail. It takes him one hour to walk up, but only half an hour to walk down. At some
point on his way down, he realizes that he was at the same spot at exactly the same time
on Saturday. Prove that he is right.
Hint: Let u(t) and d(t) be the position functions for the walks up and down, and apply the
intermediate value theorem to f(t) = u(t) −d(t).
3. Show that the polynomial function p(x) = 2x
3
−5x
2
−10x+5 has a real root in the interval
[−1, 2].
4. Let f : R →R be continuous. If S := ¦f(x) [ x ∈ R¦ is neither bounded above nor bounded
below, prove that S = R.
Hint: If y ∈ R, then since S is neither bounded above nor bounded below, there exist
x
0
, x
1
∈ R such that f(x
0
) < y < f(x
1
).
5. (a) (∗) Show that given any continuous function f : R → R, there exists a x
0
∈ [0, 1] and
a m ∈ Z such that f(x
0
) = mx
0
. In other words, the graph of f intersects some line
y = mx at some point x
0
in [0, 1].
Hint: If f(0) = 0, take x
0
= 0 and any m ∈ Z. If f(0) > 0, then choose N ∈ N
satisfying N > f(1), and apply the intermediate value theorem to the continuous
function g(x) = f(x) −Nx on the interval [0, 1]. If f(0) < 0, then ﬁrst choose a N ∈ N
such that N > −f(1), and consider the function g(x) = f(x) + Nx, and proceed in a
similar manner.
(b) (∗) Prove that there does not exist a continuous function f : R →R such that assumes
rational values at irrational numbers, and irrational values at rational numbers, that
is,
f(Q) ⊂ R ¸ Q and f(R ¸ Q) ⊂ Q.
Hint: Note that for every m ∈ Z, there does not exist a x
0
∈ R such that f(x
0
) = mx
0
.
Chapter 2
Algebra
In this part of the course, we study two important algebraic objects: groups and vector spaces.
We begin with a discussion about groups.
2.1 Groups
In this section, we study one of the most basic algebraic objects, namely a group. A group is a set
on which a law of composition is deﬁned, such that certain properties hold. The precise deﬁnition
is given in the next subsection.
2.1.1 Deﬁnition of a Group
By a law of composition on a set S, we mean a rule for combining pairs a, b of S to get another
element, say c, of S. We denote the set of all ordered pairs of elements from a set S by S S,
that is, S S = ¦(a, b) [ a, b ∈ S¦.
Deﬁnition. A law of composition on a set S is a function f : S S → S.
The functional notation c = f(a, b) is not very convenient for what is going to follow, and so
instead, the element obtained by applying the law of composition to a pair (a, b) is usually denoted
using a notation resembling that used for addition or multiplication:
c = a ∗ b, or ab, or a ◦ b, or a +b and so on,
with a ﬁxed choice being made for the particular law in question.
Examples.
1. The addition of integers is a law of composition on Z. Indeed, the sum of two integers is yet
another integer, and addition is the function from the set Z Z to the set Z that assigns
a +b to the pair (a, b), denoted by (a, b) → a +b.
2. The multiplication of real numbers is a law of composition on R.
39
40 Chapter 2. Algebra
3. If a, b are rational numbers, then let a ∗ b = a +b −ab. The function from QQ to Q given
by (a, b) → a ∗ b is a law of composition on Q.
4. If a, b are real numbers, then deﬁne a ∗ b =
√
a
2
+b
2
. Then (a, b) → a ∗ b is not a law of
composition on Q, since 1 ∈ Q, but 1 ∗ 1 =
√
2 ,∈ Q. However, (a, b) → a ∗ b is a law of
composition on R.
5. Let n ∈ N, and let S denote the set of all matrices of size n n with real entries. Then
matrix multiplication is a law of composition on S.
6. Let n ∈ N, and let GL(n, R) denote the set of all invertible matrices of size n n with
real entries. Then matrix multiplication is a law of composition on GL(n, R). Indeed, if
A, B ∈ GL(n, R), then the matrix AB is again a matrix of size n n with real entries, and
moreover, since A and B are invertible, it follows that AB is also invertible.
7. Let a, b be real numbers such that a < b. Let C[a, b] denote the set of all continuous functions
on the interval [a, b]. Let addition of functions be deﬁned as follows: if f, g belong to C[a, b],
then
(f +g)(x) = f(x) +g(x), x ∈ [a, b].
Then addition of functions is a law of composition on C[a, b], since the sum of continuous
functions is again continuous; see Theorem 1.3.3.
8. If a, b ∈ N, then let
a ∗ b =
a
b
.
∗ is not a law of composition on N, since 1 ∗ 2 =
1
2
,∈ N. ♦
On a set there may be several diﬀerent laws of compositions that can be deﬁned. Some laws
of compositions are nicer than others, that is, they possess some desirable properties. A group is
a set G together with a law of composition on G that has three such desirable properties, and we
give the deﬁnition below.
Deﬁnition. A group is a set G together with a law of composition (a, b) → a ∗ b : G G → G,
which has the following properties:
G1. (Associativity.) For all a, b, c ∈ G, (a ∗ b) ∗ c = a ∗ (b ∗ c).
G2. (Identity.) There exists an element
1
e ∈ G such that for all a ∈ G, a ∗ e = a = e ∗ a.
G3. (Inverses.) For every a ∈ G, there exists an element
2
a
−1
∈ G such that a∗a
−1
= e = a
−1
∗a.
G1, G2, G3 are called group axioms. Sometimes we use the notation (G, ∗) for the group.
Remarks. Note that hidden in the deﬁnition of a group, is the following axiom G0:
G0. For all a, b ∈ G, a ∗ b ∈ G. (That is, ∗ is actually a law of composition on G.)
Hence when checking that a certain set G is a group with respect to a certain operation ∗ that
combines pairs of elements from G, we have to check, ﬁrst of all, that for every a, b ∈ G, a ∗ b
belongs to G.
1
Such an element e is called an identity element of the group G.
2
depending on a, and such an element a
−1
is called an inverse of the element a in the group G.
2.1. Groups 41
Examples.
1. Z with addition is a group. Indeed, the addition of integers is a law of composition on Z
that satisﬁes the group axioms:
G1. For all a, b, c ∈ Z, (a +b) +c = a + (b +c).
G2. 0 serves as an identity element: for all a ∈ Z, a + 0 = a = 0 +a.
G3. If a ∈ Z, then −a ∈ Z and a + (−a) = 0 = −a +a.
2. R with multiplication is a not a group. Indeed, although the multiplication of real numbers
is a law of composition on R that satisﬁes G1 and G2, but the group axiom G3 does not
hold:
G1. For all a, b, c ∈ R, (ab)c = a(bc).
G2. (If e is an identity element, then we must have ae = a = ea for all a ∈ R, and in
particular, with a = 1, we should have 1e = 1, and so, e = 1. And so if e is an identity
element, then it must be equal to 1!) 1 serves as an identity element: for all a ∈ R,
a1 = a = 1a.
G3. Does not hold, since 0 ∈ R, but for all a
−1
∈ R, 0a
−1
= a
−1
0 = 0 ,= 1 = e. So there is
no inverse of the element 0 ∈ R.
However, the set of nonzero real numbers, or the set of positive real numbers are both groups
with multiplication, since in addition to G1 and G2, now in each case G3 also holds.
3. Let n ∈ N, and let GL(n, R) denote the set of all invertible matrices of size n n with real
entries. Then GL(n, R) is group with matrix multiplication. Indeed, matrix multiplication
is a law of composition on GL(n, R) that satisﬁes the group axioms:
G1. For all A, B, C ∈ GL(n, R), (AB)C = A(BC), since matrix multiplication is associative.
G2. The identity matrix
I
n
=
_
¸
¸
¸
_
1 0 . . . 0
0 1 . . . 0
.
.
.
.
.
.
.
.
.
.
.
.
0 0 . . . 1
_
¸
¸
¸
_
serves as an identity element. I
n
is an invertible matrix of size n n with real entries,
and so it belongs to GL(n, R), and moreover, for all A ∈ GL(n, R), AI
n
= A = I
n
A.
G3. If A ∈ GL(n, R), then A is an invertible matrix, and so there exists a matrix A
−1
such
that AA
−1
= I
n
= A
−1
A. The matrix A
−1
∈ GL(n, R) and serves as an inverse of A.
This group is called the general linear group.
4. Let a, b be real numbers such that a < b. Then C[a, b] is a group with addition of functions.
Indeed, the addition of functions is a law of composition on C[a, b] that satisﬁes the group
axioms:
G1. For all f, g, h ∈ C[a, b], and any x ∈ [a, b] we have
(f + (g +h))(x) = f(x) + (g +h)(x)
= f(x) + (g(x) +h(x))
= (f(x) +g(x)) +h(x) (since addition is associative in R!)
= (f +g)(x) +h(x)
= ((f +g) +h)(x).
Hence f + (g +h) = (f + g) +h.
42 Chapter 2. Algebra
G2. The constant function 0 deﬁned by 0(x) = 0 for all x ∈ [a, b], serves as an identity
element. 0 is a continuous function on [a, b] and so 0 ∈ C[a, b]. Moreover, for all
f ∈ C[a, b], we have for all x ∈ [a, b]:
(f +0)(x) = f(x) +0(x)
= f(x) + 0
= f(x) (since 0 is an identity element for addition in R!)
= 0 +f(x)
= 0(x) +f(x)
= (0 +f)(x).
Hence f +0 = f = 0 +f.
G3. If f ∈ C[a, b], then deﬁne −f by (−f)(x) = −f(x), for x ∈ [a, b]. Given f ∈ C[a, b], we
have for all x ∈ [a, b]:
(f + (−f))(x) = f(x) + (−f)(x)
= f(x) + (−f(x))
= 0 = 0(x) = 0
= −f(x) +f(x)
= (−f)(x) +f(x)
= (−f +f)(x).
Hence f + (−f) = 0 = −f +f. ♦
We note that the groups in the above Examples 1 and 4, also satisfy
G4. (Commutativity.) For all a, b ∈ G, a ∗ b = b ∗ a ,
while the group in Example 3 (with n = 2) does not satisfy G4:
_
1 1
0 1
_ _
1 0
1 1
_
=
_
2 1
1 1
_
,=
_
1 1
1 2
_
=
_
1 0
1 1
_ _
1 1
0 1
_
.
This gives rise to the following natural deﬁnition.
Deﬁnition. A group (G, ∗) is said to be abelian
3
if for all a, b ∈ G, a ∗ b = b ∗ a.
Examples.
1. The set of integers Z with addition is an abelian group.
2. The set of positive real numbers with multiplication is an abelian group.
3. Let n, m ∈ N. The set matrices R
n×m
of size n m with entries in R with matrix addition
is an abelian group.
4. Let n ∈ N. The set GL(n, R) with matrix multiplication is a group, but it is not an abelian
group if n > 1. ♦
3
after the Norwegian mathematician Abel
2.1. Groups 43
We now prove a few elementary theorems concerning groups.
Theorem 2.1.1 There is a unique identity element in a group.
Proof Let e and e
be identity elements in (G, ∗). Since e ∈ G and e
is an identity, we obtain
e = e ∗ e
.
Moreover, since e
∈ G and e is an identity, we also have
e ∗ e
= e
.
Consequently, e = e
.
Theorem 2.1.2 Let (G, ∗) be a group and let a ∈ G. Then a has a unique inverse.
Proof Let the group have the identity e. If a
1
and a
2
are inverses of a, then we have
a
1
= a
1
∗ e (since a
1
∈ G and e is the identity)
= a
1
∗ (a ∗ a
2
) (since a
2
is an inverse of a)
= (a
1
∗ a) ∗ a
2
(associativity)
= e ∗ a
2
(since a
1
is an inverse of a)
= a
2
(since a
2
∈ G and e is the identity).
Example. If A, B are matrices of size n n with real entries and I
n
− AB is invertible, then
I
n
− BA is also invertible, and (I
n
− BA)
−1
= I
n
+ B(I
n
− AB)
−1
A. Indeed, it is easy to check
that
(I
n
−BA)(I
n
+B(I
n
−AB)
−1
A) = I
n
= (I
n
+B(I
n
−AB)
−1
A)(I
n
−BA),
and so it follows that I
n
− BA is invertible. By the uniqueness of the inverse of the element
I
n
−BA in the group GL(n, R), it follows that (I
n
−BA)
−1
= I
n
+B(I
n
− AB)
−1
A. ♦
Deﬁnitions. A group (G, ∗) is said to be ﬁnite, if the set G has ﬁnite cardinality. The order of
a ﬁnite group (G, ∗) is the cardinality of G. A group is said to be inﬁnite if it is not ﬁnite.
Examples.
1. The set ¦−1, 1¦ with multiplication is a ﬁnite group of order 2.
2. The set Z with addition is an inﬁnite group. ♦
A ﬁnite group can be completely described by writing its group table. This is a table that
displays the law of composition as follows: the elements of the group are listed in the ﬁrst row and
the ﬁrst column, and then given a, b ∈ G, the element a ∗ b is entered in the row corresponding to
a and the column corresponding to b, as shown below.
44 Chapter 2. Algebra
∗ . . . b . . .
.
.
.
a a ∗ b
.
.
.
We clarify this further by considering a simple example.
Example. The ﬁnite group ¦−1, 1¦ with multiplication can be described by the group table given
below.
1 −1
1 1 −1
−1 −1 1
The table completely describes the law of composition: 1 1 = 1, 1 (−1) = −1, −1 1 = −1
and −1 (−1) = 1. ♦
Exercises.
1. (a) Show that the set Z
6
of integers modulo 6, with addition modulo 6 is a group. Write
down its group table.
(b) Is Z
6
with multiplication modulo 6 also a group? If, instead, we consider the set Z
∗
6
of
nonzero integers modulo 6, then is Z
∗
6
a group with multiplication modulo 6?
(c) (∗) Let m be an integer such that m ≥ 2, and let Z
∗
m
denote the set of nonzero integers
modulo m. Prove that Z
∗
m
is a group with multiplication modulo m iﬀ m is a prime
number.
Hint: If m is a prime number, then any r ∈ ¦1, . . . , m− 1¦ is coprime to m, and so
there exist integers s and t such that sm +tr = 1.
2. (a) Given n ∈ N, let S
n
be the set of all bijections from the set ¦1, 2, 3, . . . , n¦ onto itself.
Prove that the set S
n
with the composition of functions is a group. This is called the
symmetric group S
n
.
Hint: Composition of bijections is again a bijection, and composition of functions
on a set is also associative. The identity element is simply the identity map ι :
¦1, 2, 3, . . . , n¦ → ¦1, 2, 3, . . . , n¦ given by ι(m) = m for all m ∈ ¦1, 2, 3, . . . , n¦, while
the inverse of an element f ∈ S
n
is simply the inverse of that bijection.
(b) What is the order of S
n
?
(c) If n = 3, give examples of bijections f and g on ¦1, 2, 3¦ such that f ◦ g ,= g ◦ f.
(d) (∗) Show that S
n
is abelian iﬀ n ≤ 2.
3. Consider the set
S =
__
a b
−b a
_ ¸
¸
¸
¸
a, b ∈ R and a
2
+b
2
,= 0
_
.
Show that S with the operation of matrix multiplication forms a group.
4. Let (G, ∗) be a group.
(a) Show that if a, b, c ∈ G are such that a ∗ b = a ∗ c, then b = c.
(b) Show that if a, b ∈ G, then the equation a ∗ x = b has a unique solution.
(c) Show that if a, b ∈ G, then (a ∗ b)
−1
= b
−1
∗ a
−1
.
2.1. Groups 45
2.1.2 Subgroups
Deﬁnition. A subset H of a group G is called a subgroup of G, if it has the following properties:
H1. (Closure) If a, b ∈ H, then a ∗ b ∈ H.
H2. (Identity) e ∈ H.
H3. (Inverses) If a ∈ H, then a
−1
∈ H.
These conditions ensure that H is itself a group which is contained in the group G, and this
is explained as follows:
H1. This condition tells us that the law of composition ∗ on the group G can be used to deﬁne
a law of composition on H, namely, the function from H H to H given by (a, b) → a ∗ b,
which is called the induced law of composition. Since ∗ is associative, it follows that the
induced law of composition is associative as well: for all a, b, c ∈ H, a ∗ (b ∗ c) = (a ∗ b) ∗ c.
H2. This shows that H, with the induced law of composition, has an identity element. Indeed
the identity element from G (which also belongs to H) serves as the identity element also in
H: for every a ∈ H, a ∗ e = a = e ∗ a.
H3. Finally this shows that every element in H possesses an inverse element in H. Of course,
since G is a group, we already knew that a possesses an inverse element a
−1
∈ G. But
now H3 says that this inverse element is in H. Thus for all a ∈ H, ∃a
−1
∈ H such that
a ∗ a
−1
= e = a
−1
∗ a.
Thus the conditions H1, H2, H3 imply that the subset H, with the induced law of composition, is
a group. Thus, a subgroup is itself a group which sits in a larger group.
Examples.
1. The subset of even integers ¦2m [ m ∈ Z¦ is a subgroup of the group of integers Z with
addition. Indeed, the sum of even numbers is even (and so H1 holds), 0 is even (and so H2
holds), and ﬁnally, given the even number 2m, −2m = 2(−m) is even as well (and so H3
holds).
2. The group of integers with addition (Z, +) is a subgroup of the group of rational numbers
with addition (Q, +), which in turn is a subgroup of the group of real numbers with addition
(R, +).
3. If G is a group with identity e, then ¦e¦ and G are both subgroups of G.
4. The subset of symmetric matrices of size 2 2, namely
__
a b
b d
_ ¸
¸
¸
¸
a, b, d ∈ R
_
with real entries is a subgroup of the set of all 2 2 matrices having real entries with matrix
addition. Indeed, given any two symmetric matrices
_
a b
b d
_
and
_
a
b
b
d
_
,
46 Chapter 2. Algebra
their sum
_
a b
b d
_
+
_
a
b
b
d
_
=
_
a +a
b +b
b +b
d +d
_
is also symmetric, and so H1 holds. Clearly the identity element
_
0 0
0 0
_
is symmetric, and so H2 also holds. Finally, the inverse (with respect to matrix addition) of
any symmetric matrix
_
a b
b d
_
,
is the element
_
−a −b
−b −d
_
which is also symmetric, and so H3 holds.
5. The subset of upper triangular invertible matrices,
UT(2, R) =
__
a b
0 d
_ ¸
¸
¸
¸
a, b, d ∈ R and ad ,= 0
_
is a subgroup of the group GL(2, R) with matrix multiplication. Indeed, if
_
a b
0 d
_
,
_
a
b
0 d
_
∈ UT(2, R),
then
_
a b
0 d
_ _
a
b
0 d
_
=
_
aa
ab
+bd
0 dd
_
∈ UT(2, R),
and so H1 holds. Clearly
_
1 0
0 1
_
∈ UT(2, R),
and so H2 also holds. Finally,
if
_
a b
0 d
_
∈ UT(2, R), then
_
a b
0 d
_
−1
=
_
1
a
−
b
ad
0
1
d
_
∈ UT(2, R).
♦
Exercises.
1. Determine if the following statements are TRUE or FALSE.
(a) The nonnegative integers form a subgroup of Z with addition.
(b) The odd integers form a subgroup of Z with addition.
(c) If G is abelian and H is a subgroup of G, then H is abelian.
2. Is there an inﬁnite group with a ﬁnite subgroup?
3. (a) Consider the group of integers Z with addition. Suppose that H is the subgroup of Z
comprising multiples of 4, and let K be the subgroup of Z comprising multiples of 6.
What is H ∩ K?
2.1. Groups 47
(b) If H and K are subgroups of a group G, then show that H∩K is also a subgroup of G.
4. Let C[0, 1] denote the group comprising the set of continuous functions on the interval [0, 1]
with addition of functions deﬁned in the usual way: if f, g belong to C[0, 1], then for all
x ∈ [0, 1], (f +g)(x) = f(x) +g(x).
(a) Let H
1
= ¦f ∈ C[0, 1] [ f
_
1
2
_
= 0¦. Prove that H
1
is a subgroup of C[0, 1].
(b) Let H
2
denote the set of all polynomial functions, that is, the set of all functions
p : [0, 1] →R such that there exists an n ∈ N ∪ ¦0¦ and real numbers a
0
, a
1
, a
2
, . . . , a
n
such that p(x) = a
0
+ a
1
x + a
2
x
2
+ + a
n
x
n
, for all x ∈ [0, 1]. Show that H
2
is a
subgroup of C[0, 1].
5. Let G be a group. The center of G is the set Z(G) = ¦z ∈ G [ ∀a ∈ G, z ∗ a = a ∗ z¦.
(a) Show that Z(G) is not empty.
(b) If G is abelian, then determine Z(G).
(c) Show that Z(G) is a subgroup of G.
(d) (∗) If G is the group GL(2, R) with matrix multiplication, then determine Z(G).
Hint: Take a =
_
1 1
0 1
_
,
_
1 0
1 1
_
.
Let G be a group and let a ∈ G. We deﬁne
a
0
= e and a
n
= a
n−1
∗ a for n ∈ N.
Moreover, if n ∈ N, then we deﬁne
a
−n
= (a
n
)
−1
.
It can be shown that the usual laws of exponents hold: for all m, n ∈ Z,
a
m
∗ a
n
= a
m+n
and a
mn
= (a
m
)
n
.
(Exercise!)
Deﬁnitions. Let G be a group and suppose that a ∈ G.
1. If there exists an m ∈ N such that a
m
= e, then a is said to have ﬁnite order.
2. If a has ﬁnite order, then the order of a, denoted by ord(a), is
ord(a) = min¦m ∈ N [ a
m
= e¦.
3. If a does not have ﬁnite order, then a is said to have inﬁnite order.
Examples.
1. The element −1 has order 2 in the group of nonzero real numbers with multiplication.
2. The element 2 has inﬁnite order in the group of integers with addition.
3. The element
_
_
0 0 1
1 0 0
0 1 0
_
_
is an element of order 3 in the group GL(3, R). ♦
48 Chapter 2. Algebra
We now prove that given any element from a group, the set of its powers form a subgroup of
the group.
Theorem 2.1.3 Suppose that G is a group and a ∈ G. Let ¸a) = ¦a
n
[ n ∈ Z¦. Then:
1. ¸a) is a subgroup of G.
2. If a is an element with ﬁnite order m, then ¸a) = ¦e, a, a
2
, a
3
, . . . , a
m−1
¦.
Proof
1. We prove that H1, H2, H3 hold.
H1. Given m, n ∈ Z, clearly m+n ∈ Z and so a
m
∗ a
n
= a
m+n
∈ ¸a).
H2. e = a
0
∈ ¸a).
H3. For any m ∈ Z, −m ∈ Z and so (a
m
)
−1
= a
−1·m
= a
−m
∈ ¸a).
So ¸a) is a subgroup of G.
2. Clearly ¦e, a, a
2
, a
3
, . . . , a
m−1
¦ ⊂ ¸a). Conversely, if n ∈ Z, then there exist integers q and r,
such that 0 ≤ r ≤ m−1, and n = q m +r. So
a
n
= a
q·m+r
= a
q·m
∗ a
r
= (a
m
)
q
∗ a
r
= (e)
q
∗ a
r
= e ∗ a
r
= a
r
∈ ¦e, a, a
2
, a
3
, . . . , a
m−1
¦.
The element [1] in the group Z
5
with addition modulo 5 has ﬁnite order, since
[1] ⊕[1] ⊕[1] ⊕[1] ⊕[1] = [0],
and the subgroup ¸[1]) = ¦[0], [1], [2], [3], [4]¦ is in fact the whole group. This example motivates
the following deﬁnition.
Deﬁnition. A group G is said to be cyclic if there exists an element a ∈ G such that G = ¸a).
Such an element a is then called a generator of the group G.
Exercises.
1. If G is a ﬁnite group, then show that every element in the group has a ﬁnite order, which is
at most equal to [G[.
Hint: If a ∈ G, then consider the set S = ¦e, a, a
2
, a
3
, . . . , a
G
¦, and use the pigeonhole
principle.
2. If G is a group and a ∈ G, then show that for all m, n ∈ Z, a
m
∗a
n
= a
m+n
and (a
m
)
n
= a
mn
.
3. Determine the order of
_
1 1
−1 0
_
in the group GL(2, R).
4. Prove that in any group (G, ∗), and for any a, b in G, the orders of a ∗ b and b ∗ a are the
same.
5. Is the group of integers with addition cyclic? What is a generator of this group? Is it unique?
2.1. Groups 49
2.1.3 Homomorphisms and isomorphisms
Deﬁnition. Let (G, ∗) and (G
, ∗
) be groups. A homomorphism ϕ : G → G
is a function such
that
for all a, b ∈ G, ϕ(a ∗ b) = ϕ(a) ∗
ϕ(b).
Examples.
1. Let G be R with addition, and G
be the set of positive reals with multiplication. The
exponential function from G to G
, given by x → 2
x
, is a homomorphism. Indeed, we have
2
x+y
= 2
x
2
y
for all real x, y.
2. Let G be the group GL(2, R) with matrix multiplication, and G
be the group of nonzero
real numbers with multiplication. Then the determinant function det : G → G
is a homo
morphism, since for all 2 2 real matrices A, B we have det(AB) = det(A) det(B).
3. Let G be the group C[0, 1] of continuous functions on the interval [0, 1] with addition, and
G
be the group R with addition. Then the function f → f
_
1
2
_
is a homomorphism. Indeed,
if f, g are continuous functions on the interval [0, 1], then (f +g)
_
1
2
_
= f
_
1
2
_
+g
_
1
2
_
, by the
deﬁnition of f +g.
4. If G is a group, then the identity map ι : G → G deﬁned by ι(a) = a for all a ∈ G, and the
trivial map ζ : G → G deﬁned by ζ(a) = e for all a ∈ G are homomorphisms. ♦
Thus a homomorphism is a function between two groups that respects the law of composition.
In the next theorem we show that it preserves the identity element and the inverses of elements
as well.
Theorem 2.1.4 Let G be a group with identity e and G
be a group with identity e
. If ϕ : G → G
is a homomorphism, then:
1. ϕ(e) = e
.
2. If a ∈ G, then (ϕ(a))
−1
= ϕ(a
−1
).
Proof We have
e
∗
ϕ(e) = ϕ(e) = ϕ(e ∗ e) = ϕ(e) ∗
ϕ(e),
and so canceling ϕ(e) on both sides, we obtain e
= ϕ(e). Next,
ϕ(a
−1
) ∗
ϕ(a) = ϕ(a
−1
∗ a) = ϕ(e) = e
and similarly,
e
= ϕ(e) = ϕ(a ∗ a
−1
) = ϕ(a) ∗
ϕ(a
−1
).
Thus ϕ(a
−1
) ∗
ϕ(a) = e
= ϕ(a) ∗
ϕ(a
−1
), and by the uniqueness of the inverse of ϕ(a) in G
, we
obtain (ϕ(a))
−1
= ϕ(a
−1
).
Every group homomorphism ϕ determines two important subgroups: its image and its kernel.
Deﬁnitions. Let G, G
be groups and let ϕ : G → G
be a group homomorphism.
50 Chapter 2. Algebra
1. The kernel of ϕ is the set ker(ϕ) = ¦a ∈ G [ ϕ(a) = e
¦.
2. The image of ϕ is the set im(ϕ) = ¦a
∈ G
[ ∃a ∈ G such that ϕ(a) = a
¦.
Using Theorem 2.1.4, we now prove the following result.
Theorem 2.1.5 Let G, G
be groups and let ϕ : G → G
be a group homomorphism. Then:
1. ker(ϕ) is a subgroup of G.
2. im(ϕ) is a subgroup of G
.
Proof It is easy to check that ker(ϕ) is a subgroup of G. Indeed if a, b belong to ker(ϕ), then
ϕ(a ∗ b) = ϕ(a) ∗
ϕ(b) = e
∗
e
= e
, and so H1 holds. Moreover, as ϕ(e) = e
, e ∈ ker(ϕ) and so
H2 holds too. Finally, H3 holds, since if a ∈ ker(ϕ), then ϕ(a
−1
) = (ϕ(a))
−1
= (e
)
−1
= e
, and
so a
−1
∈ ker(ϕ). Hence ker(ϕ) is a subgroup of G.
We now also check that im(ϕ) is a subgroup of G
. If a
, b
belong to im(ϕ), then there exist
elements a, b in G such that ϕ(a) = a
and ϕ(b) = b
. Consequently, ϕ(a∗b) = ϕ(a)∗
ϕ(b) = a
∗
b
,
and so there exists an element in G, namely a∗b, such that ϕ(a∗b) = a
∗
b
, that is, a
∗
b
∈ im(ϕ).
Thus H1 holds. Since ϕ(e) = e
, it follows that e
∈ im(ϕ). Finally, if a
∈ im(ϕ), then there exists
an a ∈ G such that ϕ(a) = a
, and so a
−1
= (ϕ(a))
−1
= ϕ(a
−1
). Hence a
−1
∈ im(ϕ), and H3
holds. So im(ϕ) is a subgroup of G
.
Examples.
1. Let G be R with addition, and G
be the set of positive reals with multiplication. The
exponential function from G to G
, given by x → 2
x
, is a homomorphism with kernel the
trivial subgroup comprising the element 0, and the image is the whole group of positive reals
with multiplication, since it can be shown that given any y > 0, there exists a unique real
number (called the logarithm of y to the base 2, denoted by log
2
y) such that y = 2
log
2
y
.
2. Let G be the group GL(2, R) with matrix multiplication, and G
be the group of nonzero
real numbers with multiplication. Then the determinant function det : G → G
is a homo
morphism. Its kernel is the set of all invertible matrices with determinant equal to 1, and
we denote this subgroup by SL(2, R), and it is called the special linear group:
SL(2, R) = ¦A ∈ GL(2, R) [ det(A) = 1¦.
The image of this homomorphism is the whole group of nonzero reals: indeed, given any real
number a not equal to zero, we have that
A :=
_
1 0
0 a
_
∈ GL(2, R),
and det(A) = 1 a −0 0 = a.
3. Let G be the group C[0, 1] of continuous functions on the interval [0, 1] with addition, and
G
be the group R with addition. Then the function f → f
_
1
2
_
is a homomorphism, and
its kernel is the set of all continuous functions on the interval [0, 1] that have a root at
1
2
(for instance the straight line f(x) = x −
1
2
belongs to the kernel). The image is the set of
all real numbers, since given any a ∈ R, the constant function f(x) = a for all x ∈ [0, 1] is
continuous, and f
_
1
2
_
= a. ♦
2.1. Groups 51
In Example 1 above, the homomorphism between the two groups was also bijective. We give
a special name to such homomorphisms.
Deﬁnition. Let G, G
be groups. A homomorphism ϕ : G → G
is said to be an isomorphism if
it is bijective.
Examples.
1. Let G be R with addition, and G
be the set of positive reals with multiplication. The
exponential function from G to G
, given by x → 2
x
, is an isomorphism.
2. If G is a group, then the identity function ι : G → G deﬁned by ι(a) = a for all a ∈ G is an
isomorphism.
3. Let G be the subgroup of GL(2, R) comprising all matrices of the form
_
1 x
0 1
_
, where
x ∈ R. Let G
be the group R with addition. Then the function from G to G
, given by
_
1 x
0 1
_
→ x,
is an isomorphism. ♦
Isomorphisms are important because their existence between two groups means that the two
groups are essentially the “same”, in the sense that as far as algebraic properties go, there is no
real diﬀerence between them.
Exercises.
1. (a) Let G, G
, G
be groups and let ϕ : G → G
and ψ : G
→ G
be group homomorphisms.
Prove that the composition ψ ◦ ϕ : G → G
deﬁned by (ψ ◦ ϕ)(a) = ψ(ϕ(a)), a ∈ G is
a group homomorphism.
(b) Describe the kernel of ψ ◦ ϕ.
2. A subgroup N of a group G is called a normal subgroup if for every a ∈ N and every b ∈ G,
then b ∗ a ∗ b
−1
∈ N. Prove that if G, G
are groups and ϕ : G → G
is a homomorphism,
then ker(ϕ) is a normal subgroup of G.
3. If G is an abelian group, then show that the function from G to G, given by a → a
−1
is an
isomorphism.
4. Let G, G
be groups and let ϕ : G → G
be an isomorphism. Prove that the inverse function
ϕ
−1
: G
→ G is also an isomorphism.
5. Let G be a group and let a be an element of G.
(a) Prove that the function from Z to ¸a), given by m → a
m
, is a homomorphism from the
group of integers Z with addition to the subgroup ¸a).
(b) If a has inﬁnite order, then prove that it is an isomorphism.
52 Chapter 2. Algebra
2.1.4 Cosets and Lagrange’s theorem
Given a subgroup H of a group G, deﬁne the relation R on G by
aRb if b = a ∗ h for some h ∈ H. (2.1)
Then R is an equivalence relation:
E1. (Reﬂexivity) For all a ∈ G, aRa since e ∈ H and a = a ∗ e.
E2. (Symmetry) For all a, b in G, if aRb, then there exists a h ∈ H such that b = a ∗ h and so
b ∗ h
−1
= a. Since H is a subgroup, and h ∈ H, it follows that h
−1
∈ H. Thus bRa.
E3. (Transitivity) For all a, b, c in G, if aRb and bRc, then there exist elements h
1
, h
2
in H such
that b = a ∗ h
1
and c = b ∗ h
2
. Hence we obtain c = b ∗ h
2
= (a ∗ h
1
) ∗ h
2
= a ∗ (h
1
∗ h
2
).
Since h
1
, h
2
∈ H and H is a subgroup, it follows that h
1
∗ h
2
∈ H, and so aRc.
Deﬁnition. Let H be a subgroup of a group G, and let R be the equivalence relation given by
(2.1). If a ∈ G, then the equivalence class of a, namely the set
¦b ∈ G [ aRb¦ = ¦b ∈ G [ ∃h ∈ H such that b = a ∗ h¦ = ¦a ∗ h [ h ∈ H¦,
is called a left coset of H, and is denoted by a ∗ H.
We know that the equivalence classes of an equivalence relation partition the set. (Recall that
a by partition of a set S, we mean a subdivision of the set S into nonoverlapping subsets:
S = union of disjoint, nonempty subsets of S.
For example, the sets ¦1, 3¦, ¦2, 5¦, ¦4¦ form a partition of the ¦1, 2, 3, 4, 5¦. The two sets, of even
integers, and of odd integers, form a partition of the set Z of all integers.)
Hence we obtain the following result:
PSfrag replacements
G
a ∗ H = a
∗ H
(aRa
)
b ∗ H
c ∗ H
Figure 2.1: Distinct cosets partition the group.
Corollary 2.1.6 Let H be a subgroup of a group G. Then the left cosets of H partition the group
G.
2.1. Groups 53
Remarks.
1. The notation a∗H denotes a certain subset of G. As with any equivalence relation, diﬀerent
notations may represent the same subset. In fact, we know that a ∗ H is the unique coset
containing a, and so
a ∗ H = b ∗ H iﬀ aRb.
Corollary 2.1.6 says that if a ∗ H and b ∗ H have an element in common then they are equal.
2. One can also deﬁne the relation R
on G by aR
b if b = h∗a for some h ∈ H. The associated
equivalence classes are called right cosets.
Examples.
1. Consider the group G of the integers modulo 6, Z
6
, with addition modulo 6, and let H be
the subgroup ¸[2]) = ¦[0], [2], [4]¦. The left cosets, which we now denote by a ⊕H are
[0] ⊕H = [2] ⊕H = [4] ⊕H = ¦[0], [2], [4]¦, and
[1] ⊕H = [3] ⊕H = 5 ⊕H = ¦[1], [3], [5]¦.
Note that the cosets ¦[0], [2], [4]¦ and ¦[1], [3], [5]¦ do form a partition of G:
G = ¦[0], [1], [2], [3], [4], [5]¦ = ¦[0], [2], [4]¦∪¦[1], [3], [5]¦, and ¦[0], [2], [4]¦∩¦[1], [3], [5]¦ = ∅.
2. Consider the group G of the integers Z, with addition, and let H be the subgroup of even
numbers ¦2m [ m ∈ Z¦. The left cosets, which we now denote by a +H are
= −2 +H = 0 +H = ¦2m [ m ∈ Z¦ = ¦. . . , −4, −2, 0, 2, 4, . . . ¦ = 2 +H = . . . ,
= −1 +H = 1 +H = ¦2m+ 1 [ m ∈ Z¦ = ¦. . . , −3, −1, 1, 3, . . . ¦ = 3 +H = . . . .
Note that the cosets ¦. . . , −4, −2, 0, 2, 4, . . . ¦ and ¦. . . , −3, −1, 1, 3, . . . ¦ do indeed partition
the set of all integers. ♦
Note that in Example 1 above, there are only two distinct cosets, and
[G[ = 6 = 3 2 = [H[ (number of cosets of H).
In particular the order of H (namely 3) divides the order of G (namely 6). This is not a coincidence.
We now prove an important result concerning the order of a group G and the number of cosets of
a subgroup H, due to Lagrange.
Theorem 2.1.7 (Lagrange’s theorem) Let H be a subgroup of a ﬁnite group G. Then the order
of H divides the order of G.
Proof Note that there is a bijective function from the subgroup H to the coset a ∗ H, given by
h → a∗ h, for h ∈ H. Consequently, each coset a∗ H has the same number of elements as H does.
Since G is the union of the cosets of H, and since these cosets do not overlap, we obtain the
counting formula
[G[ = [H[ (number of cosets of H).
In particular, [H[ divides [G[.
54 Chapter 2. Algebra
Corollary 2.1.8 Let G be a ﬁnite group and let a ∈ G. Then the order
4
of a divides the order of
G. In particular a
G
= e.
Proof Let a have order m. ¸a) is a subgroup of G, and from Theorem 2.1.3, it follows that
[¸a)[ = m divides [G[, and so [G[ = m k for some k ∈ N. Thus a
G
= a
m·k
= (a
m
)
k
= e
k
= e.
The following theorem characterizes all groups whose order is a prime number.
Corollary 2.1.9 If G is a group with prime order p, then G is cyclic, and G = ¸a) for every
a ∈ G¸ ¦e¦.
Proof If a ,= e, then a has order > 1, say m. Since m divides p, and p is prime, it follows that
m = p. As G itself has order p, it now follows that G = ¸a), and so G is cyclic.
Exercises.
1. Determine if the following statements are TRUE or FALSE.
(a) If H is a subgroup of G and a, b ∈ G are such that a ,= b, then (a ∗ H) ∩ (b ∗ H) = ∅.
(b) If H is a subgroup of G and a ∈ G is such that a ∗ H has 4 elements, then H has 4
elements.
(c) If H is a subgroup of a ﬁnite group G, then for any a ∈ G, the left coset a ∗ H has the
same number of elements as the right coset H ∗ a.
2. (a) Verify that
G =
__
x y
0 1
_ ¸
¸
¸
¸
x, y ∈ R and x > 0
_
is a group with matrix multiplication.
(b) Show that
H =
__
x 0
0 1
_ ¸
¸
¸
¸
x ∈ R and x > 0
_
is a subgroup of G.
(c) Any element of G can be represented by a point in the (x, y)plane. Draw the partitions
of the plane into left and into right cosets of H.
3. Let H and K be subgroups of a group G of orders 3 and 5 respectively. Prove that H∩K =
¦e¦.
Hint: H ∩ K is a subgroup of H as well as K.
4. Prove or disprove that the group S
4
has an element of order 16.
5. (a) Let p be a prime, and let Z
∗
p
denote the group of nonzero integers modulo p with
multiplication modulo p. Show that if a is an integer such that a is not divisible by p,
then [a]
p−1
= [1].
(b) (∗) Prove Fermat’s little theorem: for any integer a, a
p
≡ a(mod p).
Hint: If a ∈ Z is not divisible by p, then [a] ,= [0], and so by part (5a) above,
[a]
p−1
= [1]. Hence p[(a
p−1
−1), and so p[(a
p
−a).
(c) (∗) Show that 7 divides 2222
5555
+ 5555
2222
.
Hint: Note that 2222 = 7 317+3, so that in Z
7
, [2222
5555
] = [3]
5555
. Now use the fact
that [3]
6
= [1] to conclude that [2222
5555
] = [3
5
]. Proceeding in a similar manner, show
that [5555
2222
] = [3
2
]. Hence we obtain [2222
5555
+5555
2222
] = [3
5
+3
2
] = [3
2
28] = [0].
4
By Exercise 1 on page 48, it follows that every element in a ﬁnite group has a ﬁnite order.
2.2. Vector spaces 55
2.2 Vector spaces
In this section we introduce a very important algebraic object, called a vector space. Roughly
speaking it is a set of elements, called “vectors”. Any two vectors can be “added”, resulting in a
new vector, and any vector can be multiplied by an element from R so as to give a new vector.
The precise deﬁnition is given in the next subsection.
2.2.1 Deﬁnition of a vector space
Deﬁnition. A vector space V is a set together with two functions, + : V V → V , called vector
addition, and : R V → V , called scalar multiplication, such that (V, +) is an abelian group,
and the following hold:
V1. For all v ∈ V , 1 v = v.
V2. For all α, β ∈ R and all v ∈ V , α (β v) = (αβ) v.
V3. For all α, β ∈ R and all v ∈ V , (α +β) v = α v +β v.
V4. For all α ∈ R and all v
1
, v
2
∈ V , α (v
1
+v
2
) = α v
1
+α v
2
.
V3 and V4 are called the distributive laws. The elements of a vector space are called vectors.
We observe that since (V, +) is an abelian group, a vector space also has the following prop
erties:
1. For all v
1
, v
2
, v
3
∈ V , v
1
+ (v
2
+v
3
) = (v
1
+v
2
) +v
3
.
2. There exists an element 0 ∈ V (called the
5
zero vector) such that for all v ∈ V , v +0 = v =
0 +v.
3. For every v ∈ V , there exists a unique
6
element in V , denoted by −v, such that v + (−v) =
0 = −v +v.
4. For all v
1
, v
2
∈ V , v
1
+v
2
= v
2
+v
1
.
Examples.
1. Let n, m ∈ N. The set R
n×m
of n m matrices having real entries with matrix addition is
an abelian group. Deﬁne scalar multiplication as follows: if α ∈ R and
if α ∈ R and A=
_
¸
_
a
11
. . . a
1n
.
.
.
.
.
.
a
m1
. . . a
mn
_
¸
_ ∈ R
n×m
, then αA=
_
¸
_
αa
11
. . . αa
1n
.
.
.
.
.
.
αa
m1
. . . αa
mn
_
¸
_. (2.2)
Then α A ∈ R
n×m
, and moreover V1, V2, V3, V4 are satisﬁed:
5
Since there is a unique identity element in a group, the zero vector is unique!
6
In a group, every element has a unique inverse!
56 Chapter 2. Algebra
V1. If A ∈ R
n×m
, then clearly
1 A = 1
_
¸
_
a
11
. . . a
1n
.
.
.
.
.
.
a
m1
. . . a
mn
_
¸
_ =
_
¸
_
1a
11
. . . 1a
1n
.
.
.
.
.
.
1a
m1
. . . 1a
mn
_
¸
_ =
_
¸
_
a
11
. . . a
1n
.
.
.
.
.
.
a
m1
. . . a
mn
_
¸
_ = A.
V2. For all α, β ∈ R and all A ∈ R
n×m
,
α (β A) = α
_
_
_β
_
¸
_
a
11
. . . a
1n
.
.
.
.
.
.
a
m1
. . . a
mn
_
¸
_
_
_
_
= α
_
¸
_
βa
11
. . . βa
1n
.
.
.
.
.
.
βa
m1
. . . βa
mn
_
¸
_
=
_
¸
_
α(βa
11
) . . . α(βa
1n
)
.
.
.
.
.
.
α(βa
m1
) . . . α(βa
mn
)
_
¸
_
=
_
¸
_
(αβ)a
11
. . . (αβ)a
1n
.
.
.
.
.
.
(αβ)a
m1
. . . (αβ)a
mn
_
¸
_
= (αβ)
_
¸
_
a
11
. . . a
1n
.
.
.
.
.
.
a
m1
. . . a
mn
_
¸
_
= (αβ) A.
V3. For all α, β ∈ R and all A ∈ R
n×m
,
(α +β) A = (α +β)
_
¸
_
a
11
. . . a
1n
.
.
.
.
.
.
a
m1
. . . a
mn
_
¸
_
=
_
¸
_
(α +β)a
11
. . . (α +β)a
1n
.
.
.
.
.
.
(α +β)a
m1
. . . (α +β)a
mn
_
¸
_
=
_
¸
_
αa
11
+βa
11
. . . αa
1n
+βa
1n
.
.
.
.
.
.
αa
m1
+βa
m1
. . . αa
mn
+βa
mn
_
¸
_
=
_
¸
_
αa
11
. . . αa
1n
.
.
.
.
.
.
αa
m1
. . . αa
mn
_
¸
_ +
_
¸
_
βa
11
. . . βa
1n
.
.
.
.
.
.
βa
m1
. . . βa
mn
_
¸
_
= α
_
¸
_
a
11
. . . a
1n
.
.
.
.
.
.
a
m1
. . . a
mn
_
¸
_ +β
_
¸
_
a
11
. . . a
1n
.
.
.
.
.
.
a
m1
. . . a
mn
_
¸
_
= α A+β A.
2.2. Vector spaces 57
V4. For all α ∈ R and all A, B ∈ R
n×m
,
α (A+B) = α
_
_
_
_
¸
_
a
11
. . . a
1n
.
.
.
.
.
.
a
m1
. . . a
mn
_
¸
_ +
_
¸
_
b
11
. . . b
1n
.
.
.
.
.
.
b
m1
. . . b
mn
_
¸
_
_
_
_
= α
_
¸
_
a
11
+b
11
. . . a
1n
+b
1n
.
.
.
.
.
.
a
m1
+b
m1
. . . a
mn
+b
mn
_
¸
_
=
_
¸
_
α(a
11
+b
11
) . . . α(a
1n
+b
1n
)
.
.
.
.
.
.
α(a
m1
+b
m1
) . . . α(a
mn
+b
mn
)
_
¸
_
=
_
¸
_
αa
11
+αb
11
. . . αa
1n
+αb
1n
.
.
.
.
.
.
αa
m1
+αb
m1
. . . αa
mn
+αb
mn
_
¸
_
=
_
¸
_
αa
11
. . . αa
1n
.
.
.
.
.
.
αa
m1
. . . αa
mn
_
¸
_ +
_
¸
_
αb
11
. . . αb
1n
.
.
.
.
.
.
αb
m1
. . . αb
mn
_
¸
_
= α
_
¸
_
a
11
. . . a
1n
.
.
.
.
.
.
a
m1
. . . a
mn
_
¸
_ +β
_
¸
_
b
11
. . . b
1n
.
.
.
.
.
.
b
m1
. . . b
mn
_
¸
_
= α A+β B.
Hence R
m×n
is a vector space with matrix addition and with scalar multiplication deﬁned
by (2.2). If n = 1, then we denote the vector space of column vectors R
m×1
by R
m
.
2. Let a, b be real numbers with a < b. The set C[a, b] of continuous functions on the interval
[a, b] with addition of functions is an abelian group. Let scalar multiplication be deﬁned as
follows:
if α ∈ R and f ∈ C[a, b], then (α f)(x) = αf(x), x ∈ [a, b]. (2.3)
Then α f ∈ C[a, b], and moreover V1, V2, V3, V4 are satisﬁed:
V1. Let f ∈ C[a, b]. For all x ∈ [a, b], we have
(1 f)(x) = 1f(x) = f(x),
and so 1 f = f.
V2. Let α, β ∈ R and f ∈ C[a, b]. For all x ∈ [a, b], we have
(α (β f))(x) = α(β f)(x)
= α(βf(x))
= (αβ)f(x)
= ((αβ) f)(x),
and so (α (β f) = (αβ) f.
V3. Let α, β ∈ R and f ∈ C[a, b]. For all x ∈ [a, b], we have
((α +β) f)(x) = (α +β)f(x)
= αf(x) +βf(x)
= (α f)(x) + (β f)(x)
= (α f +β f)(x),
58 Chapter 2. Algebra
and so (α +β) f = α f +β f.
V4. Let α ∈ R and f, g ∈ C[a, b]. For all x ∈ [a, b], we have
(α (f +g))(x) = α(f +g)(x)
= α(f(x) +g(x))
= αf(x) +αg(x)
= (α f)(x) + (α g)(x)
= (α f +α g)(x),
and so α (f +g) = α f +α g.
Hence C[a, b] with addition and scalar multiplication is a vector space. ♦
We now prove a few elementary properties of vector spaces.
Theorem 2.2.1 Let V be a vector space. Then the following hold:
1. For all v ∈ V , 0 v = 0.
2. For all α ∈ R, α 0 = 0.
3. If v ∈ V , then (−1) v = −v.
Proof Please go through the entire proof carefully, noting which symbols refer to the number
0 ∈ R, and which refer to the zero vector 0 ∈ V .
1. To see this, we use the distributive law to write
0 v + 0 v = (0 + 0) v = 0 v.
Now add −(0 v) on both sides, to obtain 0 v = 0.
2. Similarly, α 0 +α 0 = α (0 +0) = α 0, and hence α 0 = 0.
3. Finally, we have
v + (−1) v = 1 v + (−1) v (since 1 v = v)
= (1 + (−1)) v (distributive law)
= 0 v (since 1 + (−1) = 0)
= 0 (since 0 v = 0),
and so v + (−1) v = 0 = (−1) v + v. Hence (−1) v is an inverse of v. But the inverse of an
element from a group is unique, and so (−1) v = −v.
Exercises.
1. Let n ∈ N.
(a) Is the set of invertible nn matrices having real entries with matrix addition and with
scalar multiplication deﬁned by (2.2) a vector space?
2.2. Vector spaces 59
(b) Is the set of invertible nn matrices having real entries with matrix multiplication and
with scalar multiplication deﬁned by (2.2) a vector space?
2. Let V be a vector space. Prove that if α ∈ R and v ∈ V are such that α v = 0, then either
α = 0 or v = 0.
Hint: If α ,= 0, then α
−1
∈ R. Premultiply both sides of α v = 0 by α
−1
.
3. (∗) Consider the set R
∞
of all sequences with addition deﬁned as follows:
if (a
n
)
n∈N
, (b
n
)
n∈N
∈ R
∞
, then (a
n
)
n∈N
+ (b
n
)
n∈N
= (a
n
+b
n
)
n∈N
, (2.4)
and scalar multiplication deﬁned as follows:
if α ∈ R and (a
n
)
n∈N
∈ R
∞
, then α (a
n
)
n∈N
= (αa
n
)
n∈N
. (2.5)
Prove that R
∞
is a vector space with the above addition and scalar multiplication.
2.2.2 Subspaces and linear combinations
Deﬁnition. Let V be a vector space. A subset U is called a subspace of V if
S1. 0 ∈ U.
S2. If v
1
, v
2
∈ U, then v
1
+v
2
∈ U.
S3. If v ∈ U and α ∈ R, then α v ∈ U.
Subspaces of a vector space are just like subgroups of a group, that is, a subspace of a vector
space is itself a vector space with the same addition and scalar multiplication as with V (this is
easy to check). So a subspace is really a smaller vector space sitting inside a larger vector space.
Examples.
1. If V is a vector space, then the subset U comprising only the zero vector, namely U = ¦0¦,
is a subspace of V .
Also, the entire vector space, that is U = V , is a subspace of V .
If a subspace U of V is neither ¦0¦ nor V , then it is called a proper subspace of V .
2. Consider the vector space R
2×2
with matrix addition and scalar multiplication deﬁned by
(2.2). Then the set of upper triangular matrices
U
1
=
__
a b
0 d
_ ¸
¸
¸
¸
a, b, d ∈ R
_
is a subspace of R
2×2
.
Also, the set of symmetric matrices
U
2
=
__
a b
b d
_ ¸
¸
¸
¸
a, b, d ∈ R
_
is a subspace of R
2×2
.
60 Chapter 2. Algebra
3. Let a, b ∈ R and a < b. Consider the set of all polynomial functions
P[a, b] =
_
p : [a, b] →R
¸
¸
¸
¸
∃n ∈ N ∪ ¦0¦ and a
0
, a
1
, a
2
, . . . , a
n
∈ R such that
p(x) = a
0
+a
1
x +a
2
x
2
+ +a
n
x
n
for all x ∈ [a, b]
_
Then U = P[a, b] is a subspace of the vector space C[0, 1] with addition of functions and
scalar multiplication deﬁned by (2.3). ♦
Deﬁnitions. Let V be a vector space.
1. If v
1
, . . . , v
n
are vectors in V and α
1
, . . . , α
n
belong to R, then the vector α
1
v
1
+ +α
n
v
n
is called a linear combination of the vectors v
1
, . . . , v
n
.
2. Let S be a nonempty subset of a vector space V . The span of S, denoted by span(S), is
deﬁned as the set of all possible linear combinations
7
of vectors from S:
span(S) = ¦α
1
v
1
+ +α
n
v
n
[ n ∈ N, v
1
, . . . , v
n
∈ S, α
1
, . . . , α
n
∈ R¦.
Examples.
1. Let m ∈ N. Any vector in the vector space R
m
is a linear combination of the vectors
v
1
=
_
¸
¸
¸
_
1
0
.
.
.
0
_
¸
¸
¸
_
, . . . , v
m
=
_
¸
¸
¸
_
0
0
.
.
.
1
_
¸
¸
¸
_
.
Hence span(¦v
1
, . . . , v
m
¦) = R
m
.
2. Let a, b ∈ R with a < b. Any polynomial p on the interval [a, b] is a linear combination of
the functions from the set S = ¦1, x, x
2
, . . . ¦. Hence span(S) = P[a, b] in the vector space
C[a, b]. ♦
The span of a set of vectors turns out to be a special subspace of the vector space.
Theorem 2.2.2 Let V be a vector space and S be a nonempty subset of V . Then span(S) is the
smallest subspace of V that contains S.
Proof We ﬁrst show that span(S) is a subspace of V .
Let v ∈ S. Then 0 = 0 v ∈ span(S). If u, v ∈ span(S), then we know that
u = α
1
u
1
+ +α
n
u
n
and v = β
1
v
1
+ +β
m
v
m
for some vectors u
1
, . . . , u
n
, v
1
, . . . , v
m
∈ S and scalars α
1
, . . . , α
n
, β
1
, . . . , β
m
∈ R. Consequently,
u +v = α
1
u
1
+ +α
n
u
n
+β
1
v
1
+ +β
m
v
m
∈ span(S).
7
Note that although S might be inﬁnite, a linear combination, by deﬁnition, is always a linear combination of a
ﬁnite set of vectors from S.
2.2. Vector spaces 61
Finally, if v ∈ span(S), then we know that v = α
1
v
1
+ + α
n
v
n
for some v
1
, . . . , v
n
∈ S and
scalars α
1
, . . . , α
n
∈ R, and so for any β ∈ R, we have
βv = β(α
1
v
1
+ +α
n
v
n
) = β(α
1
v
1
)+ +β(α
n
v
n
) = (βα
1
)v
1
+ +(βα
n
)v
n
∈ span(S).
So span(S) satisﬁes S1,S2,S3, and so it is a subspace of V . Moreover, if v ∈ S, then v = 1 v ∈
span(S), and so S ⊂ span(S). Thus span(S) contains S.
If U is another subspace that contains the vectors from S, then from S2 and S3 it follows that
it certainly contains all linear combinations of vectors from S belong to U, and so span(S) ⊂ U.
Hence span(S) is the smallest subspace of V containing S.
Deﬁnitions. Let V be a vector space, and suppose that v
1
, . . . , v
n
are vectors that belong to V .
1. The vectors v
1
, . . . , v
n
are called linearly independent if the following condition holds:
if ∃α
1
, . . . , α
n
∈ R such that α
1
v
1
+ +α
n
v
n
= 0, then α
1
= = α
n
= 0.
An arbitrary subset S of vectors from V is said to be linearly independent if every nonempty
ﬁnite set of vectors from S is an independent set of vectors.
2. The vectors v
1
, . . . , v
n
are called linearly dependent if they are not linearly independent, that
is,
∃α
1
, . . . , α
n
∈ R, not all zeros, such that α
1
v
1
+ +α
n
v
n
= 0.
An arbitrary subset S of vectors from V is said to be a linearly dependent if there exists a
nonempty ﬁnite set of dependent vectors from S.
Examples.
1. Let V be a vector space. Then any ﬁnite set of vectors from V containing the zero vector is
linearly dependent. Indeed if v
1
, . . . , v
n
∈ V and v
k
= 0, then
0 v
1
+ + 0 v
k−1
+ 1 v
k
+ 0 v
k+1
+ + 0 v
n
= 0.
2. The vectors
v
1
=
_
¸
¸
¸
_
1
0
.
.
.
0
_
¸
¸
¸
_
, . . . , v
m
=
_
¸
¸
¸
_
0
.
.
.
0
1
_
¸
¸
¸
_
are linearly independent in R
m
. Indeed if α
1
v
1
+ +α
m
v
m
= 0, then
_
¸
_
α
1
.
.
.
α
m
_
¸
_ = α
1
_
¸
¸
¸
_
1
0
.
.
.
0
_
¸
¸
¸
_
+ +α
m
_
¸
¸
¸
_
0
.
.
.
0
1
_
¸
¸
¸
_
=
_
¸
_
0
.
.
.
0
_
¸
_,
and so α
1
= = α
m
= 0.
3. Let a, b ∈ R with a < b. The functions 1, x on the interval [a, b] are linearly independent.
Indeed, if for all x ∈ [a, b], α 1 +β x = 0(x), then in particular, we have
α +βa = 0 and α +βb = 0,
and since a ,= b, it follows that α = β = 0. ♦
62 Chapter 2. Algebra
Exercises.
1. Determine if the following statements are TRUE or FALSE:
(a) The union of two subspaces of a vector space V is a subspace of V .
(b) The intersection of two subspaces of a vector space V is a subspace of V .
(c)
_
_
1
2
3
_
_
∈ span
_
_
_
_
_
_
_
1
3
4
_
_
,
_
_
0
1
1
_
_
_
_
_
_
_
in the vector space R
3
.
(d) The vectors
_
_
1
0
1
_
_
,
_
_
1
2
1
_
_
,
_
_
0
3
0
_
_
are linearly independent in the vector space R
3
.
(e) If v
1
, v
2
, v
3
, v
4
are linearly independent, then v
1
, v
2
, v
3
are linearly independent.
(f) If v
1
, v
2
, v
3
, v
4
are linearly dependent, then v
1
, v
2
, v
3
are linearly dependent.
2. (a) Prove that if t
1
and t
2
are distinct real numbers in R, then
span
___
1
t
1
_
,
_
1
t
2
___
= R
2
in the vector space R
2
.
(b) (∗) Prove that R
2
is not the union of a ﬁnite number of proper subspaces.
Hint: Consider the inﬁnite subset S =
__
1
t
_ ¸
¸
¸
¸
t ∈ R
_
.
3. (∗) Let R
∞
be the vector space of all sequences, with addition and scalar multiplication
deﬁned by (2.4) and (2.5), respectively. We deﬁne the following subsets of R
∞
:
(a)
∞
is the set of all bounded sequences.
(b) c is the set of all convergent sequences.
(c) c
0
is the set of all convergent sequences with limit 0.
(d) c
00
= ¦(a
n
)
n∈N
∈ R
∞
[ ∃N ∈ N such that ∀n > N, a
n
= 0¦, is the set of all sequences
that are eventually zero.
Prove that c
00
⊂ c
0
⊂ c ⊂
∞
⊂ R
∞
, and that each is a subspace of the next one.
4. Consider the vector space C[0, 1] with addition of functions and scalar multiplication deﬁned
by (2.3). Let S(y
1
, y
2
) = ¦f ∈ C[0, 1] [ f(0) = y
1
and f(1) = y
2
¦. Show that S(y
1
, y
2
) is a
subspace of C[0, 1] iﬀ y
1
= 0 = y
2
.
2.2.3 Basis of a vector space
Deﬁnition. Let V be a vector space. Then a set of vectors B is said to be a basis of V if
B1. span(B) = V , and
B2. B is linearly independent.
2.2. Vector spaces 63
Example. The vectors
v
1
=
_
¸
¸
¸
_
1
0
.
.
.
0
_
¸
¸
¸
_
, . . . , v
m
=
_
¸
¸
¸
_
0
0
.
.
.
1
_
¸
¸
¸
_
form a basis of R
m
. ♦
Theorem 2.2.3 Let V be a vector space and B be a basis of V such that B has n elements. Then
any linearly independent set S of n vectors is also a basis of V .
Proof We simply prove that span(S) = V . This is proved as follows: we claim that for each
k ∈ ¦0, 1, . . . , n¦, there is a set S
k
with n elements such that it has n − k elements from B and
the other k elements belong to S, and such that span(S
k
) = V . We prove this claim by induction
on k.
For k = 0, we simply deﬁne S
0
= B. Then S
0
is a set with n elements such that it has
n − 0 = n elements from B and the other 0 elements belong to S, and since B is a basis,
V = span(B) = span(S
0
).
Suppose that the claim is true for some k. Thus S
k
is a set with n elements such that it has
n − k elements v
1
, . . . , v
n−k
from B and the other k elements, say u
1
, . . . , u
k
, belong to S, and
such that span(S
k
) = V . Now suppose that u
r
is an element from S such that it does not belong
to ¦u
1
, . . . , u
k
¦. Since span(S
k
) = V , there exist numbers α
1
, . . . , α
n−k
, β
1
, . . . , β
k
∈ R such that
u
r
= α
1
v
1
+ +α
n−k
v
n−k
+β
1
u
1
+ +β
k
u
k
.
Then we have the following two cases:
1
◦
If α
1
= = α
n−k
= 0, then we get that 1 u
r
−β
1
u
1
− − β
k
u
k
= 0, which contradicts
the linear independence of S. So this case is not possible.
2
◦
Hence there must exist an α
j
,= 0. So the spans of the sets
S
k
= ¦v
1
, . . . , v
j−1
, v
j
, v
j+1
, . . . , v
n−k
, u
1
, . . . , u
k
¦, and
S
k+1
:= ¦v
1
, . . . , v
j−1
, u
r
, v
j+1
, . . . , v
n−k
, u
1
, . . . , u
k
¦
are the same. (Why?) Moreover, S
k+1
has n elements, n − k − 1 of these belong to B, and the
other elements (namely, u
r
, u
1
, . . . , u
k
) belong to S.
Hence by induction, for k = n, we obtain that S
n
has n elements, such that 0 of these are in
B, and the other n − 0 elements are in S, and such that span(S
n
) = V . But S has n elements,
and so S
n
= S. Thus span(S) = V .
Given a vector space, there are of course many bases. However, the next result says that the
cardinality of the basis is unique for any given vector space.
Corollary 2.2.4 If a vector space V has a basis with n elements, then every basis of V has the
same number of elements.
Proof Suppose that B is a basis of V with n elements and suppose that B
is another basis of V .
64 Chapter 2. Algebra
Let B
have > n elements. Then take any n distinct elements v
1
, . . . , v
n
from B
. From the
previous theorem, it follows that these span V , and so if v ∈ B
¸ ¦v
1
, . . . , v
n
¦, it can be written
as a linear combination of ¦v
1
, . . . , v
n
¦, which contradicts the independence of B
.
If B
has < n elements, then by interchanging the roles of B and B
and proceeding as above,
we once again arrive at a contradiction.
The above result motivates the following natural deﬁnitions.
Deﬁnition. Let V be a vector space.
1. If there exists a basis B of V such that B has n elements, then n is called the dimension of
V , and it is denoted by dim(V ).
2. If a vector space has a basis with a ﬁnite number of elements, then it is called a ﬁnite
dimensional vector space. .
3. If a vector space is not ﬁnite dimensional, then it is called an inﬁnite dimensional vector
space.
Exercises.
1. Prove or disprove that
B =
_
_
_
_
_
1
0
0
_
_
,
_
_
1
1
0
_
_
,
_
_
1
1
1
_
_
_
_
_
is a basis for R
3
.
2. Prove that if B is a basis of a ﬁnite dimensional vector space V , then every element v ∈ V
can be written as a unique linear combination of the vectors from B.
3. (∗) Show that C[0, 1] is inﬁnite dimensional.
Hint: One can prove this by contradiction. Let C[0, 1] be a ﬁnite dimensional vector space
with dimension d, say. First show that the set B = ¦x, x
2
, . . . , x
d
¦ is linearly independent.
Then by Theorem 2.2.3, B is a basis for C[0, 1], and so the constant function 1 should be a
linear combination of the functions from B. Derive a contradiction.
4. (a) (∗) For k ∈ N, let e
k
denote the sequence with the kth term equal to 1, and all other
terms equal to zero:
e
k
= (a
n
)
n∈N
, where a
n
=
_
1 if n = k,
0 if n ,= k,
and let B = ¦e
k
[ k ∈ N¦. Prove that B is a basis for the vector space c
00
, comprising
all sequences that are eventually zero.
(b) (∗) Is B = ¦e
k
[ k ∈ N¦ also a basis for the vector space R
∞
?
Hint: Consider the constant sequence (1)
n∈N
.
2.2. Vector spaces 65
2.2.4 Linear transformations
Deﬁnition. Let U, V be two vector spaces. A function T : U → V is called a linear transformation
if it satisﬁes the following two properties:
L1. For all u
1
, u
2
∈ U, T(u
1
+u
2
) = T(u
1
) +T(u
2
).
L2. For all u ∈ U, and all α ∈ R, T(α u) = α T(u).
Thus just like group homomorphisms that are functions that respect group operations, linear
transformations are functions that respect vector space operations.
Examples.
1. Let m, n ∈ N. Let
A =
_
¸
_
a
11
. . . a
1n
.
.
.
.
.
.
a
m1
. . . a
mn
_
¸
_ ∈ R
m×n
.
Then the function T
A
: R
n
→R
m
deﬁned by
T
A
_
¸
_
x
1
.
.
.
x
n
_
¸
_ =
_
¸
_
a
11
x
1
+ +a
1n
x
n
.
.
.
a
m1
x
1
+ +a
mn
x
n
_
¸
_ =
_
¸
¸
¸
¸
¸
¸
¸
_
n
k=1
a
1k
x
k
.
.
.
n
k=1
a
mk
x
k
_
¸
¸
¸
¸
¸
¸
¸
_
for all
_
¸
_
x
1
.
.
.
x
n
_
¸
_ ∈ R
n
, (2.6)
is a linear transformation from the vector space R
n
to the vector space R
m
. Indeed, we have
T
A
_
_
_
_
¸
_
x
1
.
.
.
x
n
_
¸
_ +
_
¸
_
y
1
.
.
.
y
n
_
¸
_
_
_
_ = T
A
_
¸
_
x
1
+y
1
.
.
.
x
n
+y
n
_
¸
_ =
_
¸
¸
¸
¸
¸
¸
¸
_
n
k=1
a
1k
(x
k
+y
k
)
.
.
.
n
k=1
a
mk
(x
k
+y
k
)
_
¸
¸
¸
¸
¸
¸
¸
_
=
_
¸
¸
¸
¸
¸
¸
¸
_
n
k=1
(a
1k
x
k
+a
1k
y
k
)
.
.
.
n
k=1
(a
mk
x
k
+a
mk
y
k
)
_
¸
¸
¸
¸
¸
¸
¸
_
=
_
¸
¸
¸
¸
¸
¸
¸
_
n
k=1
a
1k
x
k
+
n
k=1
a
1k
y
k
.
.
.
n
k=1
a
mk
x
k
+
n
k=1
a
mk
y
k
_
¸
¸
¸
¸
¸
¸
¸
_
=
_
¸
¸
¸
¸
¸
¸
¸
_
n
k=1
a
1k
x
k
.
.
.
n
k=1
a
mk
x
k
_
¸
¸
¸
¸
¸
¸
¸
_
+
_
¸
¸
¸
¸
¸
¸
¸
_
n
k=1
a
1k
y
k
.
.
.
n
k=1
a
mk
y
k
_
¸
¸
¸
¸
¸
¸
¸
_
= T
A
_
¸
_
x
1
.
.
.
x
n
_
¸
_ +T
A
_
¸
_
y
1
.
.
.
y
n
_
¸
_,
66 Chapter 2. Algebra
for all
_
¸
_
x
1
.
.
.
x
n
_
¸
_,
_
¸
_
y
1
.
.
.
y
n
_
¸
_ ∈ R
n
,
and so L1 holds. Moreover,
T
A
_
_
_α
_
¸
_
x
1
.
.
.
x
n
_
¸
_
_
_
_ = T
A
_
¸
_
αx
1
.
.
.
αx
n
_
¸
_ =
_
¸
¸
¸
¸
¸
¸
¸
_
n
k=1
a
1k
αx
k
.
.
.
n
k=1
a
mk
αx
k
_
¸
¸
¸
¸
¸
¸
¸
_
=
_
¸
¸
¸
¸
¸
¸
¸
_
α
n
k=1
a
1k
x
k
.
.
.
α
n
k=1
a
mk
x
k
_
¸
¸
¸
¸
¸
¸
¸
_
= α
_
¸
¸
¸
¸
¸
¸
¸
_
n
k=1
a
1k
x
k
.
.
.
n
k=1
a
mk
x
k
_
¸
¸
¸
¸
¸
¸
¸
_
= α T
A
_
¸
_
x
1
.
.
.
x
n
_
¸
_,
for all α ∈ R and all vectors
_
¸
_
x
1
.
.
.
x
n
_
¸
_ ∈ R
n
,
and so L2 holds as well. Hence T
A
is a linear transformation.
2. The function T : C[0, 1] →R given by
Tf = f
_
1
2
_
for all f ∈ C[0, 1],
is a linear transformation from the vector space C[0, 1] to the vector space R. Indeed, we
have
T(f +g) = (f +g)
_
1
2
_
= f
_
1
2
_
+g
_
1
2
_
= T(f) +T(g), for all f, g ∈ C[0, 1],
and so L1 holds. Furthermore
T(α f) = (α f)
_
1
2
_
= αf
_
1
2
_
= αT(f), for all α ∈ R and all f ∈ C[0, 1],
and so L2 holds too. Thus T is a linear transformation. ♦
Just as in the case of homomorphisms between groups, there are two important subsets asso
ciated with a linear transformation between vector spaces, whose deﬁnitions are given below.
Deﬁnitions. Let U, V be vector spaces and T : U → V a linear transformation.
1. The kernel of T is deﬁned to be the set ker(T) = ¦u ∈ U [ T(u) = 0
V
¦, where 0
V
denotes
the zero vector in V .
2. The image of T is deﬁned to be the set im(T) = ¦v ∈ V [ ∃u ∈ U such that T(u) = v¦.
2.2. Vector spaces 67
Examples.
1. Let A ∈ R
m×n
, and let T
A
: R
n
→ R
m
be the linear transformation deﬁned by (2.6). The
kernel of the linear transformation is the set of all vectors
x =
_
¸
_
x
1
.
.
.
x
n
_
¸
_ ∈ R
n
such that the system of linear equations
a
11
x
1
+ +a
1n
x
n
= 0
.
.
.
a
m1
x
1
+ +a
mn
x
n
= 0
is simultaneously satisﬁed.
The range of T
A
is the set of all vectors
y =
_
¸
_
y
1
.
.
.
y
m
_
¸
_ ∈ R
m
such that there exists a vector
x =
_
¸
_
x
1
.
.
.
x
n
_
¸
_ ∈ R
n
such that
a
11
x
1
+ +a
1n
x
n
= y
1
.
.
.
a
m1
x
1
+ + a
mn
x
n
= y
m
.
2. The function T : C[0, 1] → R given by Tf = f
_
1
2
_
for all f ∈ C[0, 1], is a linear transfor
mation from the vector space C[0, 1] to the vector space R, with kernel equal to the set of
continuous functions on the interval [0, 1] that vanish at the point
1
2
. The range of T is the
whole vector space R. ♦
Analogous to Theorem 2.1.5 for homomorphisms between groups, we now prove the following
result.
Theorem 2.2.5 Let U, V be vector spaces and T : U → V a linear transformation. Then:
1. ker(T) is a subspace of U.
2. im(T) is a subspace of V .
Proof In each of the cases, we check that S1, S2, S3 hold.
68 Chapter 2. Algebra
Let 0
U
, 0
V
denote the zero vectors in the vector spaces U, V , respectively. It is easy to
check that ker(T) is a subspace of U. Indeed, as T(0
U
) = T(0
U
+ 0
U
) = T(0
U
) + T(0
U
), it
follows that T(0
U
) = 0
V
, and so 0
U
∈ ker(T). Thus S1 holds. If u
1
, u
2
belong to ker(T), then
T(u
1
+u
2
) = T(u
1
) +T(u
2
) = 0
V
+0
V
= 0
V
, and so S2 holds as well. Finally, S3 holds, since if
α ∈ R and u ∈ ker(T), then T(α u) = α T(u) = α 0
V
= 0
V
. Hence ker(T) is a subspace of U.
We now also check that im(T) is a subspace of V . Since T(0
U
) = 0
V
, it follows that 0
V
∈
im(T), and so S1 holds. If v
1
, v
2
belong to im(T), then there exist elements u
1
, u
2
in U such that
T(u
1
) = v
1
and T(u
2
) = v
2
. Consequently, T(u
1
+ u
2
) = T(u
1
) + T(u
2
) = v
1
+ v
2
, and so there
exists an element in U, namely u
1
+u
2
, such that T(u
1
+u
2
) = v
1
+v
2
, that is, v
1
+v
2
∈ im(T).
Thus S2 holds. Finally, if α ∈ R and v ∈ im(T), then there exists a u ∈ U such that T(u) = v,
and so α v = α T(u) = T(α u). Hence α v ∈ im(T), and S3 holds. So im(T) is a subspace of
V .
Exercises.
1. Find the kernel and image of the linear transformations T
A
: R
2
→ R
2
, where A ∈ R
2×2
is
given by:
(a) A =
_
1 1
0 1
_
.
(b) A =
_
1 1
0 0
_
.
(c) A =
_
0 0
0 0
_
.
Each vector
_
x
y
_
in R
2
can be represented by a point in the (x, y)plane. In each of the
cases, draw a picture of the subspaces ker(T
A
) and im(T
A
) in the plane.
2. Consider the vector space R
2
with matrix addition and the usual scalar multiplication deﬁned
by (2.2). Deﬁne the function T : R
2
→R
2
as follows:
if
_
x
1
x
2
_
∈ R
2
, then T
_
x
1
x
2
_
=
_
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
_
_
x
2
1
x2
x
2
_
if x
1
x
2
,= 0,
_
x
1
x
2
_
if x
1
x
2
= 0.
Show that T satisﬁes L2, but not L1, and hence it is not a linear transformation.
3. (∗) Let c denote the vector space of all convergent sequences. Consider the function T : c →R
given by
T((a
n
)
n∈N
) = lim
n→∞
a
n
, for all sequences (a
n
)
n∈N
∈ c.
(a) Prove that T is a linear transformation from the vector space c to the vector space R.
(b) What is the kernel of T?
(c) Show that im(T) = R.
Solutions
Analysis
Solutions to the exercises on page 6
1. (a) S = (0, 1].
——————————————————————————————————————
An upper bound of S. 1 is an upper bound, since for all x ∈ S = (0, 1], we have x ≤ 1.
In fact any real number u ≥ 1 is an upper bound.
——————————————————————————————————————
A lower bound of S. 0 is a lower bound, since for all x ∈ S = (0, 1], we have 0 < x. In
fact any real number l ≤ 0 is a lower bound.
——————————————————————————————————————
Is S bounded? Yes. S is bounded above, since 1 is an upper bound of S. S is also
bounded below, since 0 is a lower bound. Since S is bounded above as well as bounded
below, it is bounded.
——————————————————————————————————————
Supremum of S. sup S = 1. Indeed, 1 is an upper bound, and moreover, if u is also an
upper bound, then 1 ≤ u (since 1 ∈ S).
——————————————————————————————————————
Inﬁmum of S. inf S = 0. First of all, 0 is a lower bound. Let l be a lower bound of
S. We prove that l ≤ 0. (We do this by supposing that l > 0, and arriving at a
contradiction. The contradiction is obtained as follows: if l > 0, then we will see that
the average of 0 and l, namely
l
2
, is an element in S that is less than the lower bound
l, which is a contradiction to the deﬁnition of a lower bound!) If l > 0, then 0 <
l
2
.
Moreover, since l ≤ 1 (l is a lower bound of S and 1 ∈ S) it follows that
l
2
≤
1
2
≤ 1.
Thus
l
2
∈ S. But since l > 0, it follows that
l
2
< l, a contradiction. Hence l ≤ 0.
——————————————————————————————————————
If sup S exists, then is sup S ∈ S? Yes, sup S = 1 ∈ (0, 1] = S.
——————————————————————————————————————
If inf S exists, then is inf S ∈ S? No, inf S = 0 ,∈ (0, 1] = S.
——————————————————————————————————————
Maximum of S. max S = 1, since sup S = 1 ∈ S.
——————————————————————————————————————
Minimum of S. min S does not exist since inf S = 0 ,∈ S.
——————————————————————————————————————
69
70
(b) S = [0, 1].
——————————————————————————————————————
An upper bound of S. 1 is an upper bound, since for all x ∈ S = [0, 1], we have x ≤ 1.
——————————————————————————————————————
A lower bound of S. 0 is a lower bound, since for all x ∈ S = (0, 1], we have 0 ≤ x.
——————————————————————————————————————
Is S bounded? Yes, since S is bounded above (1 is an upper bound) and it is bounded
below (0 is a lower bound).
——————————————————————————————————————
Supremum of S. sup S = 1. Indeed, 1 is an upper bound, and moreover, if u is also an
upper bound, then 1 ≤ u (since 1 ∈ S).
——————————————————————————————————————
Inﬁmum of S. inf S = 0. Indeed, 0 is an lower bound, and moreover, if l is also an
lower bound, then l ≤ 0 (since 0 ∈ S).
——————————————————————————————————————
If sup S exists, then is sup S ∈ S? Yes, sup S = 1 ∈ [0, 1] = S.
——————————————————————————————————————
If inf S exists, then is inf S ∈ S? Yes, inf S = 0 ∈ [0, 1] = S.
——————————————————————————————————————
Maximum of S. max S = 1, since sup S = 1 ∈ S.
——————————————————————————————————————
Minimum of S. min S = 0, since inf S = 0 ∈ S.
——————————————————————————————————————
(c) S = (0, 1).
——————————————————————————————————————
An upper bound of S. 1 is an upper bound, since for all x ∈ S = (0, 1), x < 1.
——————————————————————————————————————
A lower bound of S. 0 is a lower bound, since for all x ∈ S = (0, 1), 0 < x.
——————————————————————————————————————
Is S bounded? Yes, since S is bounded above (1 is an upper bound) and it is bounded
below (0 is a lower bound).
——————————————————————————————————————
Supremum of S. supS = 1.
First of all, 1 is a upper bound. Let u be an upper bound of S. We prove that 1 ≤ u.
(We do this by supposing that u < 1 and arriving at a contradiction. The contradiction
is obtained as follows: if u < 1, then we will see that the average of u and 1, namely
u+1
2
, is an element in S that is larger than the upper bound u, which is a contradiction
to the deﬁnition of an upper bound!) Since u is an upper bound and since
1
2
∈ S, it
follows that 0 < u (since 0 <
1
2
≤ u). So if u < 1, then 0 < u =
u+u
2
<
u+1
2
<
1+1
2
= 1.
Hence
u+1
2
∈ S. But u <
u+1
2
contradicts the fact that u is an upper bound of S.
——————————————————————————————————————
Inﬁmum of S. inf S = 0.
First of all, 0 is a lower bound. Let l be a lower bound of S. We prove that l ≤ 0. If
l > 0, then 0 <
l
2
. Moreover, since
1
2
∈ S and l is a lower bound of S, it follows that
l ≤
1
2
. Thus we have 0 <
l
2
< l ≤
1
2
< 1, and so
l
2
∈ S. But
l
2
< l contradicts the fact
that l is a lower bound of S.
——————————————————————————————————————
If sup S exists, then is sup S ∈ S? No, supS = 1 ,∈ (0, 1) = S.
——————————————————————————————————————
If inf S exists, then is inf S ∈ S? No, inf S = 0 ,∈ (0, 1) = S.
71
——————————————————————————————————————
Maximum of S. max S does not exist, since sup S = 1 ,∈ S.
——————————————————————————————————————
Minimum of S. min S does not exist, since inf S = 0 ,∈ S.
——————————————————————————————————————
(d) S =
_
1
n
[ n ∈ Z ¸ ¦0¦
_
=
_
1
n
[ n ∈ N
_
∪
_
−
1
n
[ n ∈ N
_
.
——————————————————————————————————————
An upper bound of S. 1 is an upper bound, since for all n ∈ N,
1
n
≤ 1 and −
1
n
≤ 0 ≤ 1.
——————————————————————————————————————
A lower bound of S. −1 is a lower bound, since for all n ∈ N, −1 ≤ 0 ≤
1
n
and −1 ≤ −
1
n
.
——————————————————————————————————————
Is S bounded? Yes, since S is bounded above (1 is an upper bound) and it is bounded
below (−1 is a lower bound).
——————————————————————————————————————
Supremum of S. supS = 1. 1 is an upper bound. Moreover, if u is also an upper bound,
then since 1 =
1
1
∈ S, 1 ≤ u.
——————————————————————————————————————
Inﬁmum of S. inf S = −1. −1 is a lower bound. Moreover, if l is also a lower bound,
then since −1 =
1
−1
∈ S, it follows that l ≤ −1.
——————————————————————————————————————
If sup S exists, then is sup S ∈ S? Yes, sup S = 1 ∈ S.
——————————————————————————————————————
If inf S exists, then is inf S ∈ S? Yes, inf S = −1 ∈ S.
——————————————————————————————————————
Maximum of S. max S = 1, since sup S = 1 ∈ S.
——————————————————————————————————————
Minimum of S. min S = −1, since inf S = −1 ∈ S.
——————————————————————————————————————
(e) S =
_
−
1
n
[ n ∈ N
_
.
——————————————————————————————————————
An upper bound of S. 0 is an upper bound, since for all n ∈ N, −
1
n
< 0.
——————————————————————————————————————
A lower bound of S. −1 is a lower bound, since for all n ∈ N, −1 ≤ −
1
n
.
——————————————————————————————————————
Is S bounded? Yes, since S is bounded above (0 is an upper bound) and it is bounded
below (−1 is a lower bound).
——————————————————————————————————————
Supremum of S. sup S = 0. 0 is an upper bound. Moreover, if u is an upper bound and
if u < 0, then let N ∈ N be such that
1
−u
< N (Archimedean principle with y =
1
−u
and x = 1!), and so we obtain u < −
1
N
∈ S, a contradiction to the fact that u is an
upper bound of S. Thus if u is an upper bound, then 0 ≤ u.
——————————————————————————————————————
Inﬁmum of S. inf S = −1. −1 is a lower bound, and if l is another lower bound, then
since −1 = −
1
1
∈ S, it follows that l ≤ −1.
——————————————————————————————————————
If sup S exists, then is sup S ∈ S? No, since sup S = 0 ,∈ S.
——————————————————————————————————————
If inf S exists, then is inf S ∈ S? Yes, inf S = −1 ∈ S.
72
——————————————————————————————————————
Maximum of S. max S does not exist since sup S = 0 ,∈ S.
——————————————————————————————————————
Minimum of S. min S = −1 since inf S = −1 ∈ S.
——————————————————————————————————————
(f) S =
_
n
n+1
[ n ∈ N
_
.
——————————————————————————————————————
An upper bound of S. 1 is an upper bound, since for all n ∈ N,
n
n+1
< 1.
——————————————————————————————————————
A lower bound of S.
1
2
is a lower bound because for all n ∈ N,
1
2
≤
n
n+1
(since n+1 ≤ 2n,
that is, 1 ≤ n).
——————————————————————————————————————
Is S bounded? Yes, since S is bounded above (1 is an upper bound) and it is bounded
below (
1
2
is a lower bound).
——————————————————————————————————————
Supremum of S. supS = 1. 1 is an upper bound of S. If u < 1 is an upper bound,
then let N ∈ N be such that
u
u−1
< N (Archimedean property). Then u <
N
N+1
,
contradicting the fact that u is an upper bound.
——————————————————————————————————————
Inﬁmum of S. inf S =
1
2
.
1
2
is a lower bound, and if l is a lower bound, then since
1
2
=
1
1+1
∈ S, it follows that l ≤
1
2
.
——————————————————————————————————————
If sup S exists, then is sup S ∈ S? No, since sup S = 1 ,∈ S.
——————————————————————————————————————
If inf S exists, then is inf S ∈ S? Yes, since inf S =
1
2
∈ S.
——————————————————————————————————————
Maximum of S. max S does not exist since sup S = 1 ,∈ S.
——————————————————————————————————————
Minimum of S. min S exists since inf S =
1
2
∈ S.
——————————————————————————————————————
(g) S = ¦x ∈ R [ x
2
≤ 2¦.
——————————————————————————————————————
An upper bound of S.
√
2 is an upper bound. (x >
√
2 implies x
2
> 2, and so x ,∈ S.
In other words, if x ∈ S, then x ≤
√
2, that is,
√
2 is an upper bound of S.)
——————————————————————————————————————
A lower bound of S. −
√
2 is a lower bound. (x < −
√
2 implies that −x >
√
2 > 1 > 0.
So using the fact that if a > b and c > 0, then ac > bc, we get x
2
= (−x)(−x) >
(−x)
√
2 >
√
2
√
2 = 2, and so, x ,∈ S. In other words, if x ∈ S, then x ≥ −
√
2, and so
−
√
2 is a lower bound.)
——————————————————————————————————————
Is S bounded? Yes, since S is bounded above (
√
2 is an upper bound) and it is bounded
below (−
√
2 is a lower bound).
——————————————————————————————————————
Supremum of S. sup S =
√
2. First of all,
√
2 is an upper bound of S. Let u be an
upper bound such that u <
√
2. Then
u+
√
2
2
∈ S. (As 0 ∈ S and u is an upper bound
of S, 0 ≤ u. As u <
√
2, u <
u+
√
2
2
<
√
2. Hence we have 0 <
u+
√
2
2
<
√
2, and
so
_
u+
√
2
2
_
2
=
_
u+
√
2
2
__
u+
√
2
2
_
<
_
u+
√
2
2
_
√
2 <
√
2
√
2 = 2. Thus
u+
√
2
2
∈ S.) But
73
u <
u+
√
2
2
contradicts the fact that u is an upper bound. So if u is an upper bound of
S, then
√
2 ≤ u.
——————————————————————————————————————
Inﬁmum of S. inf S = −
√
2. −
√
2 is a lower bound. Let l be a lower bound such that
−
√
2 < l. Then we have −l <
√
2, and since sup S =
√
2, it follows that −l cannot be
an upper bound of S. So there exists an x ∈ S such that −l < x. But since x ∈ S,
we have (−x)
2
= x
2
≤ 2. So it follows that −x ∈ S. As l is a lower bound, l ≤ −x,
that is, x ≤ −l. From −l < x and x ≤ −l, we arrive at the contradiction that −l < −l.
Consequently, if l is a lower bound, then l ≤ −
√
2.
——————————————————————————————————————
If sup S exists, then is sup S ∈ S? Yes, sup S =
√
2 ∈ S.
——————————————————————————————————————
If inf S exists, then is inf S ∈ S? Yes, inf S = −
√
2 ∈ S.
——————————————————————————————————————
Maximum of S. max S =
√
2.
——————————————————————————————————————
Minimum of S. min S = −
√
2.
——————————————————————————————————————
(h) S = ¦0, 2, 5, 2005¦.
——————————————————————————————————————
An upper bound of S. 2005 is an upper bound, since 0 ≤ 2005, 2 ≤ 2005, 5 ≤ 2005 and
2005 ≤ 2005.
——————————————————————————————————————
A lower bound of S. 0 is a lower bound, since 0 ≤ 0, 0 ≤ 2, 0 ≤ 5 and 0 ≤ 2005.
——————————————————————————————————————
Is S bounded? Yes, since S is bounded above (2005 is an upper bound) and it is bounded
below (0 is a lower bound).
——————————————————————————————————————
Supremum of S. sup S = 2005. 2005 is an upper bound, and if u is also an upper bound,
then since 2005 ∈ S, it follows that 2005 ≤ u.
——————————————————————————————————————
Inﬁmum of S. inf S = 0. 0 is an lower bound, and if l is also an lower bound, then
since 0 ∈ S, it follows that l ≤ 0.
——————————————————————————————————————
If sup S exists, then is sup S ∈ S? Yes, sup S = 2005 ∈ ¦0, 2, 5, 2005¦ = S.
——————————————————————————————————————
If inf S exists, then is inf S ∈ S? Yes, inf S = 0 ∈ ¦0, 2, 5, 2005¦ = S.
——————————————————————————————————————
Maximum of S. max S = 2005 since supS = 2005 ∈ S.
——————————————————————————————————————
Minimum of S. min S = 0 since inf S = 0 ∈ S.
——————————————————————————————————————
(i) S =
_
(−1)
n
_
1 +
1
n
_
[ n ∈ N
_
.
(This set has the elements −
2
1
,
3
2
, −
4
3
,
5
4
, . . . .)
——————————————————————————————————————
An upper bound of S.
3
2
is an upper bound.
If n ∈ N and n is even, then (−1)
n
_
1 +
1
n
_
= 1 +
1
n
≤ 1 +
1
2
=
3
2
.
74
If n ∈ N and n is odd, then (−1)
n
_
1 +
1
n
_
= −1 −
1
n
< 0 <
3
2
.
——————————————————————————————————————
A lower bound of S. −2 is a lower bound.
If n ∈ N and n is even, then (−1)
n
_
1 +
1
n
_
= 1 +
1
n
> 0 > −2.
If n ∈ N and n is odd, then (−1)
n
_
1 +
1
n
_
= −1 −
1
n
≥ −1 −1 = −2.
——————————————————————————————————————
Is S bounded? Yes, since S is bounded above (
3
2
is an upper bound) and it is bounded
below (−2 is a lower bound).
——————————————————————————————————————
Supremum of S. sup S =
3
2
.
3
2
is an upper bound, and if u is also an upper bound,
then since
3
2
= (−1)
2
_
1 +
1
2
_
∈ S, it follows that
3
2
≤ u.
——————————————————————————————————————
Inﬁmum of S. inf S = −2. −2 is a lower bound, and if l is also a lower bound, then
since −2 = (−1)
1
_
1 +
1
1
_
∈ S, it follows that l ≤ −2.
——————————————————————————————————————
If sup S exists, then is sup S ∈ S? Yes, sup S =
3
2
∈ S.
——————————————————————————————————————
Maximum of S. max S =
3
2
.
——————————————————————————————————————
Minimum of S. min S = −2.
——————————————————————————————————————
(j) S = ¦x
2
[ x ∈ R¦.
——————————————————————————————————————
An upper bound of S. There does not exist an upper bound of S. Let u ∈ R be an
upper bound of S, then u +
1
2
∈ R, and so u
2
+ u +
1
4
=
_
u +
1
2
_
2
∈ S. But since
u
2
+
1
4
> 0, it follows that u < u
2
+u +
1
4
=
_
u +
1
2
_
2
∈ S, contradicting the fact that
u is an upper bound of S. Thus S does not have an upper bound.
——————————————————————————————————————
A lower bound of S. 0 is a lower bound of S, since for all x ∈ R, 0 ≤ x
2
.
——————————————————————————————————————
Is S bounded? No, since the set is not bounded above.
——————————————————————————————————————
Supremum of S. sup S does not exist, since the set does not have an upper bound, and
so it cannot have a least upper bound. (Recall that every least upper bound is an upper
bound).
——————————————————————————————————————
Inﬁmum of S. inf S = 0. 0 is a lower bound, and if l is also a lower bound, then since
0 = 0
2
∈ S, it follows that l ≤ 0.
——————————————————————————————————————
If sup S exists, then is sup S ∈ S? sup S does not exist.
——————————————————————————————————————
If inf S exists, then is inf S ∈ S? Yes, inf S = 0 ∈ S.
——————————————————————————————————————
Maximum of S. max S does not exist, since sup S does not exist.
——————————————————————————————————————
Minimum of S. min S = 0, since inf S = 0 ∈ S.
——————————————————————————————————————
75
(k) S =
_
x
2
1+x
2
[ x ∈ R
_
.
——————————————————————————————————————
An upper bound of S. 1 is an upper bound of S, since for all x ∈ R,
x
2
1+x
2
< 1.
——————————————————————————————————————
A lower bound of S. 0 is a lower bound of S, since for all x ∈ R, 0 ≤
x
2
1+x
2
.
——————————————————————————————————————
Is S bounded? Yes, since S is bounded above (1 is an upper bound) and it is bounded
below (0 is a lower bound).
——————————————————————————————————————
Supremum of S. sup S = 1. 1 is an upper bound. Let u be an upper bound and let
u < 1. Since 0 =
0
2
1+0
2
∈ S, 0 ≤ u. Let N ∈ N be such that
_
u
1−u
< N. Then we have
u <
N
2
1+N
2
∈ S, a contradiction to the fact that u is an upper bound of S. Hence if u is
an upper bound of S, then 1 ≤ u.
——————————————————————————————————————
Inﬁmum of S. inf S = 0. 0 is a lower bound, and if l is also a lower bound, then since
0 =
0
2
1+0
2
∈ S, it follows that l ≤ 0.
——————————————————————————————————————
If sup S exists, then is sup S ∈ S? No, supS = 1 ,∈ S.
——————————————————————————————————————
If inf S exists, then is inf S ∈ S? Yes, inf S = 0 ∈ S.
——————————————————————————————————————
Maximum of S. max S does not exist, since sup S ,∈ S.
——————————————————————————————————————
Minimum of S. min S = 0, since inf S = 0 ∈ S.
——————————————————————————————————————
2. (a) FALSE (if S = ¦1¦, then u = 3 is an upper bound of S, and although u
= 2 < 3 = u,
u
(= 2) is an upper bound of ¦1¦ = S).
(b) TRUE (if > 0, then u
∗
− < u
∗
, and so u
∗
− cannot be an upper bound of S).
(c) FALSE (N has no maximum).
(d) FALSE (N has no supremum).
(e) FALSE ([0, 1) is bounded, but it does not have a maximum).
(f) FALSE (∅ is bounded, but it does not have a supremum).
(g) TRUE (least upper bound property of R).
(h) TRUE (the supremum itself is an upper bound).
(i) TRUE (deﬁnition of maximum).
(j) FALSE (the set [0, 1) has supremum 1, but 1 ,∈ [0, 1)).
(k) FALSE (−N is bounded above since 0 is an upper bound, but [−N[ = N is not bounded).
(l) TRUE (l ≤ x ≤ u implies x ≤ u and −x ≤ −l, and so we have
x ≤ u ≤ max¦−l, u¦ and −x ≤ −l ≤ max¦−l, u¦.
Thus [x[ ≤ max¦−l, u¦ and so max¦−l, u¦ is an upper bound of [S[. Moreover, for
every y ∈ [S[, we have y = [x[ for some x ∈ S, and so y = [x[ ≥ 0. Thus 0 is a lower
bound of [S[.)
(m) FALSE (if S = ¦0, 1¦, then inf S = 0 <
1
2
< 1 = supS, but
1
2
,∈ S).
76
3. Since S is bounded, in particular, it is bounded above, and furthermore, since it is nonempty,
supS exists, by the least upper bound property of R.
Since S is bounded, in particular it is bounded below, and furthermore, since it is nonempty,
inf S exists, by the greatest lower bound property of R.
Let x ∈ S. Since inf S is a lower bound of S,
inf S ≤ x. (2.7)
Moreover, since sup S is an upper bound of S,
x ≤ sup S. (2.8)
From (2.7) and (2.8), we obtain inf S ≤ supS.
Let inf S = supS. If x ∈ S, then we have
inf S ≤ x ≤ sup S, (2.9)
and so inf S = x(= sup S) (for if inf S < x, then from (2.9), inf S < supS, a contradiction).
Thus S is a singleton set. Conversely, if S = ¦x¦, then clearly x is an upper bound. If u < x
is an upper bound, then x ≤ u < x gives x < x, a contradiction. So sup S = x. Clearly
x is also a lower bound. If l > x is also a lower bound, then x > l ≥ x gives x > x, a
contradiction. So inf S = x = sup S.
4. Since sup B is an upper bound of B, we have x ≤ sup B for all x ∈ B. Since A ⊂ B, in
particular we obtain x ≤ sup B for all x ∈ A. So sup B is an upper bound of A, and so by
the deﬁnition of the least upper bound of A, we obtain sup A ≤ sup B.
5. Since A is bounded above, ∃M
A
∈ R such that ∀x ∈ A, x ≤ M
A
.
Since B is bounded above, ∃M
B
∈ R such that ∀y ∈ B, y ≤ M
B
.
Consequently if x ∈ A and y ∈ B, x +y ≤ M
A
+M
B
.
So ∀z ∈ A+B, z ≤ M
A
+M
B
, and so M
A
+M
B
is an upper bound for A+B. So A+B is
bounded above.
Clearly A + B is nonempty. Indeed, A is nonempty implies that ∃x ∈ A; B is nonempty
implies that ∃y ∈ B; thus x +y ∈ A+B, and so A+B is not empty.
Since A+B is bounded above and it is not empty, by the least upper bound property of R,
it follows that sup(A+B) exists.
Since sup A is an upper bound of A, ∀x ∈ A, x ≤ sup A.
Since sup B is an upper bound of B, ∀y ∈ B, y ≤ sup B.
So for all x ∈ A and y ∈ B, x +y ≤ sup A+ sup B.
Since every z ∈ A + B is such that z = x + y with x ∈ A and y ∈ B, it follows that for
all z ∈ A + B, z ≤ sup A + sup B. So sup A + sup B is an upper bound of A + B, and
consequently, by the deﬁnition of the least upper bound of A+B,
sup(A+B) ≤ sup A+ supB.
6. Let l be a lower bound of S: ∀x ∈ S, l ≤ x. So ∀x ∈ S, −x ≤ −l, in other words,
∀y ∈ −S, y ≤ −l.
Thus −S is bounded above because −l is an upper bound of −S.
Since S is nonempty, it follows that ∃x ∈ S, and so we obtain that −x ∈ −S. Hence −S is
nonempty.
77
As −S is nonempty and bounded above, it follows that sup(−S) exists, by the least upper
bound property of R.
Since sup(−S) is an upper bound of −S, we have:
∀y ∈ −S, y ≤ sup(−S),
that is,
∀x ∈ S, −x ≤ sup(−S),
that is,
∀x ∈ S, −sup(−S) ≤ x.
So −sup(−S) is a lower bound of S.
Next we prove that −sup(−S) is the greatest lower bound of S. Suppose that l
is a lower
bound of S such that −sup(−S) < l
. Then we have
∀x ∈ S, −sup(−S) < l
≤ x,
that is,
∀x ∈ S, −x ≤ −l
< sup(−S),
that is,
∀y ∈ −S, y ≤ −l
< sup(−S).
So −l
is an upper bound of −S, and −l
< sup(−S), which contradicts the fact that sup(−S)
is the least upper bound of −S. Hence l
≤ −sup(−S).
Consequently, inf S exists and inf S = −sup(−S).
7. (S is nonempty and bounded below (0 is a lower bound), and so by the greatest lower bound
property of R, inf S exists.)
If: Let inf S > 0. Since inf S is a lower bound, it follows that
∀x ∈ S, inf S ≤ x,
that is,
∀x ∈ S,
1
x
≤
1
inf S
,
that is,
∀y ∈ S
−1
, y ≤
1
inf S
.
So
1
inf S
is an upper bound of S
−1
. Thus S
−1
is bounded above.
Only if: Suppose S
−1
is bounded above (with an upper bound u, say). Then
∀y ∈ S
−1
y ≤ u,
that is,
∀x ∈ S,
1
x
≤ u. (2.10)
Since S is not empty, ∃x
∗
∈ S and so
1
x∗
≤ u. But x
∗
∈ S implies that x
∗
> 0, and so
1
x∗
> 0. Consequently u > 0. Hence from (2.10), we have
∀x ∈ S,
1
u
≤ x.
Thus
1
u
is a lower bound of S, and so
1
u
≤ inf S.
78
But u > 0 implies
1
u
> 0, and consequently inf S
_
≥
1
u
_
> 0.
If inf S > 0, then as in the If part, we see that
1
inf S
is an upper bound of S
−1
and so we
obtain
sup S
−1
≤
1
inf S
. (2.11)
Furthermore, since u := sup S
−1
is an upper bound of S
−1
, as in the Only if part, we see
that
1
u
=
1
sup S
−1
is a lower bound of S, and so
1
supS
−1
≤ inf S,
that is,
1
inf S
≤ sup S
−1
. (2.12)
From (2.11) and (2.12), it follows that
sup S
−1
=
1
inf S
.
8. (a) If x ∈
n∈N
_
0,
1
n
_
, then
for all n ∈ N, 0 < x <
1
n
. (2.13)
Let N ∈ N be such that
1
x
< N (Archimedean property). Thus
1
N
< x, which contra
dicts (2.13). So
_
∃x ∈
n∈N
_
0,
1
n
_
_
, that is,
n∈N
_
0,
1
n
_
= ∅.
(b) Clearly ¦0¦ ⊂
_
0,
1
n
¸
for all n ∈ N and so
¦0¦ ⊂
n∈N
_
0,
1
n
_
. (2.14)
Let x ∈
n∈N
_
0,
1
n
_
. Then x ∈ [0, 1] and so x ≥ 0. If x > 0, then let N ∈ N be
such that
1
x
< N (Archimedean property), that is,
1
N
< x. So x ,∈
_
0,
1
N
¸
, and hence
x ,∈
n∈N
_
0,
1
n
_
. Consequently, if x ∈
n∈N
_
0,
1
n
_
, then x = 0, that is,
n∈N
_
0,
1
n
_
⊂ ¦0¦. (2.15)
From (2.14) and (2.15),
n∈N
_
0,
1
n
_
= ¦0¦.
(c) Let n ∈ N. If x ∈
_
1
n
, 1 −
1
n
¸
, then 0 <
1
n
≤ x ≤ 1 −
1
n
< 1, and so x ∈ (0, 1). Hence
_
n∈N
_
1
n
, 1 −
1
n
_
⊂ (0, 1). (2.16)
79
If x ∈ (0, 1), then 0 < x < 1. Let N
1
∈ N be such that
1
x
< N
1
(Archimedean property),
that is,
1
N1
< x. Let N
2
∈ N be such that
1
1−x
< N
2
(Archimedean property), that is,
x < 1 −
1
N2
. Thus with N := max¦N
1
, N
2
¦, we have
1
N
≤
1
N
1
< x < 1 −
1
N
2
≤ 1 −
1
N
,
that is, x ∈
_
1
N
, 1 −
1
N
_
⊂
_
n∈N
_
1
n
, 1 −
1
n
_
. So we have
(0, 1) ⊂
_
n∈N
_
1
n
, 1 −
1
n
_
. (2.17)
From (2.16) and (2.17), we obtain (0, 1) =
_
n∈N
_
1
n
, 1 −
1
n
_
.
(d) If x ∈ [0, 1], then for any n ∈ N,
−
1
n
< 0 ≤ x ≤ 1 < 1 +
1
n
,
and so x ∈
_
−
1
n
, 1 +
1
n
_
. Hence
[0, 1] ⊂
n∈N
_
−
1
n
, 1 +
1
n
_
. (2.18)
Let x ∈
n∈N
_
−
1
n
, 1 +
1
n
_
. Then
−
1
n
≤ x ≤ 1 +
1
n
for all n ∈ N. (2.19)
We prove that this implies that 0 ≤ x ≤ 1. For if x < 0, then let N
1
∈ N be such that
−
1
x
< N
1
, that is, x < −
1
N1
, a contradiction to (2.19). Similarly, if x > 1, then let
N
2
∈ N be such that
1
x−1
< N
2
, that is, x > 1 +
1
N2
, a contradiction to (2.18). Hence
we see that neither x < 0 nor x > 1 are possible, and hence x ∈ [0, 1]. Thus
n∈N
_
−
1
n
, 1 +
1
n
_
⊂ [0, 1]. (2.20)
(2.18) and (2.20) imply
n∈N
_
−
1
n
, 1 +
1
n
_
= [0, 1].
9. If x ∈ (x
0
−δ, x
0
+δ), then x
0
− δ < x < x
0
+δ. Adding −x
0
, we obtain −δ < x − x
0
< δ.
So x − x
0
< δ, and −(x − x
0
) < δ. Thus [x − x
0
[ < δ, and x ∈ ¦x ∈ R [ [x − x
0
[ < δ¦.
Consequently (x
0
−δ, x
0
+δ) ⊂ ¦x ∈ R [ [x −x
0
[ < δ¦.
If x ∈ ¦x ∈ R [ [x − x
0
[ < δ¦, then [x − x
0
[ < δ. Since for any real number r, [r[ ≥ r
and [r[ ≥ −r, we obtain that x − x
0
≤ [x − x
0
[ < δ and −(x − x
0
) ≤ [x − x
0
[ < δ. Hence
−δ < x − x
0
< δ. Adding x
0
, this yields x
0
− δ < x < x
0
+ δ, that is, x ∈ (x
0
− δ, x
0
+ δ).
So ¦x ∈ R [ [x −x
0
[ < δ¦ ⊂ (x
0
−δ, x
0
+δ).
Thus (x
0
−δ, x
0
+δ) = ¦x ∈ R [ [x −x
0
[ < δ¦.
80
10. From the inequality (1.7) on page 6, we have that [x[ = [x −y +y[ ≤ [x −y[ +[y[, that is,
[x[ −[y[ ≤ [x −y[ (2.21)
for all x, y ∈ R. Interchanging x and y in (2.21), we obtain
[y[ −[x[ ≤ [y −x[ = [ −(x −y)[ = [ −1[[x −y[ = 1 [x −y[ = [x −y[,
and so,
−([x[ −[y[) ≤ [x −y[ (2.22)
for all x, y ∈ R. From (2.21) and (2.22), we obtain [[x[ −[y[[ ≤ [x −y[ for all x, y ∈ R.
11. If S is bounded, then it is bounded above and it is bounded below. Thus S has an upper
bound, say u, and a lower bound, say l. So for all x ∈ S, l ≤ x ≤ u, that is, x ≤ u and
−x ≤ −l, and so we have
x ≤ u ≤ max¦−l, u¦ and −x ≤ −l ≤ max¦−l, u¦.
Thus [x[ ≤ max¦−l, u¦ =: M.
Conversely, if there exists a M such that for all x ∈ S, [x[ ≤ M, we have −M ≤ x ≤ M. So
−M is a lower bound of S and M is an upper bound of S. Thus S is bounded.
81
Solutions to the exercises on page 13
1. (a) We prove that (1)
n∈N
is convergent sequence with limit 1. Given > 0, let N ∈ N, say
N = 1. Then for all n > N = 1, we have
[a
n
−L[ = [1 −1[ = [0[ < .
(b) Yes. For instance, the constant sequence (1)
n∈N
converges to 1.
(c) Suppose that the terms of the convergent sequence (a
n
)
n∈N
(with limit, say, L) lie in
the ﬁnite set ¦v
1
, . . . , v
m
¦. If L ,∈ ¦v
1
, . . . , v
m
¦, then with
:= min¦[v
1
−L[, . . . , [v
m
−L[¦ > 0,
let N ∈ N be such that for all n > N, [a
n
−L[ < . In particular, with n = N+1 > N, we
have [a
N+1
−L[ < . But a
N+1
∈ ¦v
1
, . . . , v
m
¦. Let a
N+1
= v
k
for some k ∈ ¦1, . . . , m¦.
Then we have
[v
k
−L[ = [a
N+1
−L[ < = min¦[v
1
−L[, . . . , [v
m
−L[¦ ≤ [v
k
−L[,
a contradiction. So L ∈ ¦v
1
, . . . , v
m
¦, that is, L must be one of the terms. Thus we
have shown that
terms of the sequence
take ﬁnitely many values
=⇒
L must be one
of the terms
,
that is,
L is not equal to
any of the terms
=⇒
terms of the sequence cannot
consist of ﬁnitely many values
.
(d) Suppose that ((−1)
n
)
n∈N
is a convergent sequence with limit L. Then from part 1c
above, it follows that L = 1 or L = −1: indeed the terms of the sequence take ﬁnitely
many values, namely 1 and −1, and so L must be one of these terms. So we have the
following two cases:
1
◦
If the limit is 1, then given = 1 > 0, let N ∈ N be such that for all n > N,
[(−1)
n
− 1[ < = 1. Let n be any odd number > N. Then we have 2 = [ − 2[ =
[ −1 −1[ = [(−1)
n
−1[ < = 1, a contradiction.
2
◦
If the limit is −1, then given = 1 > 0, let N ∈ N be such that for all n > N,
[(−1)
n
− (−1)[ < = 1. Let n be any even number > N. Then we have 2 = [2[ =
[1 + 1[ = [(−1)
n
−(−1)[ < = 1, a contradiction.
So ((−1)
n
)
n∈N
is divergent.
2. If lim
n→∞
1
n
= 1, then given any > 0, there exists an N ∈ N such that for all n ∈ N such that
n > N,
¸
¸
¸
¸
1
n
−1
¸
¸
¸
¸
< .
Let =
1
2
. (This choice is motivated by Figure 2.2, from which we see that, for instance
= 1 won’t give us a contradiction, but =
1
2
would.) Then there exists a N
∗
∈ N such that
for all n > N
∗
,
¸
¸
¸
¸
1
n
−1
¸
¸
¸
¸
< =
1
2
,
that is,
1 −
1
n
=
¸
¸
¸
¸
1
n
−1
¸
¸
¸
¸
<
1
2
.
82
PSfrag replacements
1
Figure 2.2:
_
1
n
_
n∈N
.
Consequently, for all n > N
∗
,
1 −
1
n
<
1
2
,
that is, n < 2. In particular, since n = N
∗
+1 > N
∗
, we obtain N
∗
+1 < 2, that is, N
∗
< 1.
But there does not exist any natural number N
∗
that is less than 1. Hence we arrive at a
contradiction, and so lim
n→∞
1
n
,= 1.
3. (a) We have seen that the sequence
_
(−1)
n
_
1 +
1
n
__
n∈N
is divergent. Given any > 0, let
N ∈ N be such that
1
< N. Then for all even n > N (there are obviously inﬁnitely
many such n), we have
[a
n
−L[ =
¸
¸
¸
¸
(−1)
n
_
1 +
1
n
_
−1
¸
¸
¸
¸
=
¸
¸
¸
¸
1 +
1
n
−1
¸
¸
¸
¸
=
1
n
<
1
N
< .
(b) Again, for the divergent sequence
_
(−1)
n
_
1 +
1
n
__
n∈N
, with = 4 > 0, for all N ∈ N
and all n > N, we have
[a
n
−L[ ≤ [a
n
[ +[L[ ≤ 2 + 1 = 3 < 4 = .
4. S is nonempty and bounded above, and so by the least upper bound property of R, it follows
that sup S exists.
Given n ∈ N, we have
1
n
> 0, and so sup S −
1
n
< supS. Thus S −
1
n
is not an upper
bound of S. Hence there must exist an element in S, which we denote by a
n
, such that
[a
n
≤ sup S −
1
n
], that is, a
n
> sup S −
1
n
. In this way we construct the sequence (a
n
)
n∈N
.
As sup S is an upper bound of S and so we also have a
n
≤ sup S for all n ∈ N. Consequently,
∀n ∈ N, sup S −
1
n
< a
n
≤ sup S < sup S +
1
n
,
that is,
∀n ∈ N, −
1
n
< a
n
−sup S <
1
n
, that is, [a
n
−sup S[ <
1
n
.
Given > 0, let N ∈ N be such that N >
1
. Then for all n > N, we have
[a
n
−sup S[ <
1
n
<
1
N
< .
Hence (a
n
)
n∈N
is convergent with limit equal to sup S.
5. Suppose L < 0. Then := −
L
2
> 0, and so there exists a N ∈ N such that for all n > N,
[a
n
−L[ < = −
L
2
. Hence with n = N + 1 (> N), we obtain
a
N+1
−L ≤ [a
N+1
−L[ < −
L
2
,
that is, a
N+1
<
L
2
< 0, a contradiction.
83
6. Let (a
n
)
n∈N
be a convergent sequence with limit L¿ Given > 0, there exists a N ∈ N such
that
for all n > N, [a
n
−L[ <
2
. (2.23)
Hence if n, m > N, then
[a
n
−a
m
[ = [a
n
−L +L −a
m
[
≤ [a
n
−L[ +[L −a
m
[ (triangle inequality)
= [a
n
−L[ +[a
m
−L[ (using [r[ = [ −r[ for all real r)
<
2
+
2
(using (2.23)).
Consequently (a
n
)
n∈N
is a Cauchy sequence.
Solutions to the exercises on page 17
1. We prove that the sequence is monotone and bounded, and hence it must be convergent.
We prove that [a
n
[ ≤ 1 for all n ∈ N. We prove this using induction. We have [a
1
[ = [1[ = 1.
If k ∈ N is such that [a
k
[ ≤ 1, then
[a
k+1
[ =
¸
¸
¸
¸
2k + 3
3k + 3
a
k
¸
¸
¸
¸
=
¸
¸
¸
¸
2k + 3
3k + 3
¸
¸
¸
¸
[a
k
[ =
_
2k + 3
3k + 3
_
[a
k
[ ≤ 1 1 = 1,
and so the claim follows from induction. So the sequence is bounded.
Since n ≥ 1, it follows that 2n + 1 ≤ 3n, and so
2n+1
3n
≤ 1 for all n ∈ N. Furthermore, note
that for all n ∈ N, a
n
≥ 0 (induction!). Hence for all n ≥ 2, a
n
=
2n+1
3n
a
n−1
≤ 1a
n−1
= a
n−1
.
So (a
n
)
n∈N
is decreasing.
As the sequence is bounded and monotone, it is convergent.
2. Let M > 0 be such that for all n ∈ N, [b
n
[ ≤ M. Given > 0, let N ∈ N be such that
M
< N, that is,
1
N
<
M
. Then for all n > N,
¸
¸
¸
¸
b
n
n
−0
¸
¸
¸
¸
=
[b
n
[
n
≤
M
n
<
M
N
< M
M
= .
Hence
_
bn
n
_
n∈N
is convergent with limit 0.
3. (a) Given > 0, ∃N
1
∈ N such that ∀n > N
1
, [a
n
− L[ <
2
. Since (a
n
)
n∈N
is convergent,
it is bounded: ∃M > 0 such that ∀n ∈ N, [a
n
[ ≤ M. Choose
8
N ∈ N such that
max
_
N
1
,
N
1
(M +[L[)
2
_
< N,
and so,
N > N
1
and
N
1
(M +[L[)
N
<
2
.
8
This is arrived at by working backwards; we wish to make
˛
˛
˛
a
1
+···+an
n
− L
˛
˛
˛ less than for all n > N, so we
manipulate this (as shown in the chain of inequalities that follow) to see if can indeed achieve this by choosing the
N large enough.
84
Then for n > N, we have:
¸
¸
¸
¸
a
1
+ +a
N1
+a
N1+1
+ +a
n
n
− L
¸
¸
¸
¸
=
¸
¸
¸
¸
a
1
+ +a
N1
+a
N1+1
+ +a
n
−nL
n
¸
¸
¸
¸
=
[a
1
+ +a
N1
+ a
N1+1
+ +a
n
−nL[
n
≤
[a
1
−L[ + +[a
N1
−L[ +[a
N1+1
−L[ + +[a
n
−L[
n
≤
([a
1
[ +[L[ + +[a
N1
[ +[L[) +
2
+ +
2
n
≤
N
1
(M +[L[) + (n −N
1
)
2
n
≤
N
1
(M +[L[)
n
+
_
1 −
N
1
n
_
2
<
N
1
(M +[L[)
N
+ 1
2
<
2
+
2
= .
So
_
a1+···+an
n
_
n∈N
is a convergent sequence with limit L.
(b) If a
n
= (−1)
n
, n ∈ N, then (a
n
)
n∈N
is divergent, but the sequence with nth term
a
1
+ +a
n
n
=
(−1)
1
+ (−1)
2
+ + (−1)
n
n
=
_
0 if n is even
−
1
n
if n is odd
,
is convergent with limit equal to 0. Indeed, given > 0, let N ∈ N be such that
1
< N.
Then for n > N, we have
[a
n
−0[ = [a
n
[ =
_
0 if n is even
1
n
if n is odd
_
≤
1
n
<
1
N
< .
So
_
a1+···+an
n
_
n∈N
is convergent with limit 0.
4. Since (a
n
)
n∈N
is bounded, it follows that there exists a M such that for all n ∈ N, [a
n
[ ≤ M,
that is, −M ≤ a
n
≤ M. If k ∈ N, then in particular, for all n ≥ k, −M ≤ a
n
≤ M, and so
the set ¦a
n
[ n ≥ k¦ is bounded. By the least upper bound property of R, it then follows that
inf¦a
n
[ n ≥ k¦ and sup¦a
n
[ n ≥ k¦ exist, that is, l
k
and u
k
are welldeﬁned. Furthermore,
for each k,
−M ≤ inf¦a
n
[ n ≥ k¦ ≤ sup¦a
n
[ n ≥ k¦ ≤ M,
and so we see that the sequences (l
k
)
k∈N
and (u
k
)
k∈N
are bounded.
Clearly ¦a
n
[ n ≥ k + 1¦ ⊂ ¦a
n
[ n ≥ k¦, and by Exercise 4 on page 7, we then obtain that
u
k+1
= sup¦a
n
[ n ≥ k + 1¦ ≤ sup¦a
n
[ n ≥ k¦ = u
k
,
and so (u
k
)
k∈N
is a decreasing sequence.
Similarly, ¦−a
n
[ n ≥ k + 1¦ ⊂ ¦−a
n
[ n ≥ k¦, and so we have
sup¦−a
n
[ n ≥ k + 1¦ ≤ sup¦−a
n
[ n ≥ k¦.
Using Exercise 6 on page 7, we obtain
inf¦a
n
[ n ≥ k + 1¦ = −sup¦−a
n
[ n ≥ k + 1¦ ≥ −sup¦−a
n
[ n ≥ k¦ = inf¦a
n
[ n ≥ k¦,
85
that is, l
k+1
≥ l
k
. Consequently, (l
k
)
k∈N
is a increasing sequence.
As the sequences (u
k
)
k∈N
, (l
k
)
k∈N
are both bounded and monotone, by Theorem 1.2.3, it
follows that they are convergent.
5. Let (a
n
)
n∈N
be a Cauchy sequence. Then given := 1 > 0, there exists a N ∈ N such that
for all n, m > N, [a
n
−a
m
[ < = 1. In particular, with m := N +1 (> N), [a
n
−a
N+1
[ < 1
for all n > N, that is,
[a
n
[ = [a
N+1
+a
n
−a
N+1
[ ≤ [a
N+1
[ +[a
n
−a
N+1
[ < [a
N+1
[ + 1 for all n > N.
Deﬁning M = max¦[a
1
[, . . . , [a
N
[, [a
N+1
[ + 1¦, we see that [a
n
[ ≤ M for all n ∈ N, and so
(a
n
)
n∈N
is bounded.
86
Solutions to the exercises on page 20
1. First we prove the
Claim: (a
n
)
n∈N
is convergent with limit L, then (a
n+1
)
n∈N
is also convergent with limit L.
Proof Let > 0. Then ∃N ∈ N such that for all n > N, [a
n
−L[ < . Thus for all n > N,
we have n + 1 > N + 1 > N, and so [a
n+1
− L[ < . Hence (a
n+1
)
n∈N
is convergent with
limit L.
Alternately, observe that (a
n+1
)
n∈N
is a subsequence of (a
n
)
n∈N
, and use Theorem 1.2.6.
We now apply this result to our sequence (a
n
)
n∈N
, which satisﬁes:
a
n+1
=
2(n + 1) + 1
3(n + 1)
a
n
=
2 +
3
n
3 +
3
n
a
n
,
for all n ∈ N. Since the sequence
_
1
n
_
n∈N
is convergent with limit 0, by the theorem on
algebra of limits, it follows that the sequence
_
2+
3
n
3+
3
n
_
n∈N
is convergent with limit
2+3·0
3+3·0
=
2
3
.
Again applying the theorem on algebra of limits, we obtain
L = lim
n→∞
a
n+1
= lim
n→∞
_
2 +
3
n
3 +
3
n
a
n
_
= lim
n→∞
_
2 +
3
n
3 +
3
n
_
lim
n→∞
a
n
=
2
3
L.
Hence
1
3
L = 0, that is, L = 0. So lim
n→∞
a
n
= 0.
2. We have
c
n
=
a
n
b
n
+ 5n
a
2
n
+n
=
a
n
bn
n
+ 5
a
n
a
n
1
n
+ 1
for all n ∈ N.
N. The sequence
_
a
n
bn
n
+ 5
_
n∈N
is convergent.
The sequence (a
n
)
n∈N
is convergent with limit, say L. Since (b
n
)
n∈N
is bounded,
the sequence
_
bn
n
_
n∈N
is convergent with limit 0 (see Exercise 2 on page 17). Hence
_
a
n
bn
n
_
n∈N
is convergent with limit L 0 = 0. The sequence (5)
n∈N
is convergent with
limit 5. So the sequence
_
a
n
bn
n
+ 5
_
n∈N
is convergent with limit 0 + 5 = 5.
D. The sequence
_
a
2
n
n
+ 1
_
n∈N
has nonzero terms for all n ∈ N and it is convergent with
the nonzero limit 1.
We have
a
2
n
n
+ 1 ≥ 1 for all n ∈ N, and so
a
2
n
n
+ 1 ,= 0 for all n ∈ N.
Since the sequence (a
n
)
n∈N
is convergent with limit L, it follows that the sequence
(a
2
n
)
n∈N
is convergent with limit L
2
. Since
_
1
n
_
n∈N
is convergent with limit 0, it follows
that
_
a
2
n
1
n
_
n∈N
is convergent with limit L
2
0 = 0. Finally, as (1)
n∈N
is convergent with
limit 1, it follows that the sequence
_
a
2
n
n
+ 1
_
n∈N
is convergent with limit 0+1 = 1(,= 0).
87
From the parts N and D above and the theorem on the algebra of limits, it follows that
(c
n
)
n∈N
is convergent with limit
5
1
= 5.
3. (a) We begin by showing that L ≥ 0. If L < 0, then := −
L
2
> 0. Let N ∈ N be such that
for all n > N, [a
n
−L[ < = −
L
2
. Then we have
a
n
−L ≤ [a
n
−L[ < −
L
2
,
and so a
n
<
L
2
< 0 for all n > N, a contradiction to the fact that
a
n
≥ 0 for all n ∈ N.
So L ≥ 0.
Now we show that (
√
a
n
)
n∈N
is convergent with limit
√
L. Let >). We consider the
only two possible cases, namely L = 0 and L > 0:
1
◦
If L = 0, then let N ∈ N be such that for all n > N,
[a
n
−L[ = [a
n
−0[ = [a
n
[ = a
n
<
2
.
Then for n > N, we have
√
a
n
< , that is,
[
√
a
n
−
√
L[ = [
√
a
n
−
√
0[ = [
√
a
n
[ =
√
a
n
< .
So (
√
a
n
)
n∈N
is convergent with limit
√
L.
2
◦
If L > 0, then let N ∈ N be such that for n > N, [a
n
− L[ <
√
L. Then for all
n > N, we obtain
> [a
n
−L[
= [(
√
a
n
−
√
L)(
√
a
n
+
√
L)[
= [
√
a
n
−
√
L[[
√
a
n
+
√
L)[
= [
√
a
n
−
√
L[(
√
a
n
+
√
L)
and so
[
√
a
n
−
√
L[ <
√
L
√
a
n
+
√
L
≤
√
L
√
L
= .
Hence (
√
a
n
)
n∈N
is convergent with limit
√
L.
(b) For all n ∈ N, we have
_
n
2
+n −n = (
_
n
2
+n −n)
√
n
2
+n +n
√
n
2
+n +n
=
n
2
+n −n
2
√
n
2
+n +n
=
n
√
n
2
+n +n
=
n(1)
n
_
1
n
√
n
2
+n + 1
_
=
1
_
n
2
+n
n
2
+ 1
=
1
_
1 +
1
n
+ 1
.
88
As (1)
n∈N
is convergent with limit 1 and
_
1
n
_
n∈N
is convergent with limit 0, it follows
that
_
1 +
1
n
_
n∈N
is convergent with limit 1. Furthermore 1 +
1
n
≥ 0 for all n ∈ N,
and so by the previous part, it follows that
__
1 +
1
n
_
n∈N
is convergent with limit
√
1 = 1. Hence
__
1 +
1
n
+ 1
_
n∈N
is convergent with limit 1 + 1 = 2(,= 0). Also
_
1 +
1
n
+ 1 > 1 > 0. Thus by the theorem on algebra of limits, we conclude that
the sequence
_
1
√
1+
1
n
+1
_
n∈N
is convergent with limit
1
2
, that is, (
√
n
2
+n − n)
n∈N
is
convergent with limit
1
2
.
4. Consider the sequence (b
n
− a
n
)
n∈N
. As a
n
≤ b
n
, it follows that b
n
− a
n
≥ 0 for all
n ∈ N. From the theorem on algebra of limits, it follows that the sequence (b
n
− a
n
)
n∈N
is
convergent (being the sum of the convergent sequence (b
n
)
n∈N
and the convergent sequence
(−a
n
)
n∈N
). Moreover, its limit is lim
n→∞
b
n
− lim
n→∞
a
n
. From Exercise 5 on page 13, we obtain
lim
n→∞
b
n
− lim
n→∞
a
n
≥ 0, that is lim
n→∞
b
n
≥ lim
n→∞
a
n
.
Solutions to the exercises on page 22
1. For all n ∈ N, we have
0 ≤
n!
n
n
=
1
n
2
n
n −1
n
n
n
≤
1
n
1 1 1 =
1
n
.
Since (0)
n∈N
and
_
1
n
_
n∈N
are both convergent with the same limit 0, from the Sandwich
theorem, it follows that
_
n!
n
n
_
n∈N
is convergent with the limit 0.
2. Let k ∈ N. For all n ∈ N, we have
0 ≤
1
k
+ 2
k
+ 3
k
+ +n
k
n
k+2
≤
n
k
+n
k
+n
k
+ +n
k
n
k+2
≤
n n
k
n
k+2
=
1
n
.
Thus
0 ≤
1
k
+ 2
k
+ 3
k
+ +n
k
n
k+2
≤
1
n
for all n ∈ N. Since the sequences (0)
n∈N
and
_
1
n
_
n∈N
are both convergent with limit 0, from
the Sandwich theorem, we obtain that
lim
n→∞
1
k
+ 2
k
+ 3
k
+ +n
k
n
k+2
= 0.
3. (a) We prove the claim using induction. Let x ≥ −1. Clearly (1 + x)
1
= 1 +x = 1 + 1 x.
If for some k ∈ N, (1 +x)
k
≥ 1 +kx, then we have
(1 +x)
k+1
= (1 +x)
k
(1 +x)
≥ (1 +kx)(1 +x) (by the induction hypothesis and since 1 +x ≥ 0)
= 1 +kx +x +x
2
= 1 + (k + 1)x +x
2
≥ 1 + (k + 1)x.
Hence by induction, the result follows.
89
(b) For all n ∈ N,
n
1
n
≥ 1 (2.24)
(for if n
1
n
< 1, then n = (n
1
n
)
n
< 1
n
= 1, a contradiction!). Clearly for all n ∈ N,
n
1
n
= (
√
n
2
)
1
n
=
√
n
2
n
< (1 +
√
n)
2
n
. (2.25)
Finally,
_
1 +
1
√
n
_
n
≥ 1 +n
1
√
n
= 1 +
√
n,
and so
_
1 +
1
√
n
_
2
=
__
1 +
1
√
n
_
n
_
2
n
≥ (1 +
√
n)
2
n
. (2.26)
Combining (2.24), (2.25), and (2.26), we obtain
1 ≤ n
1
n
< (1 +
√
n)
2
n
≤
_
1 +
1
√
n
_
2
(2.27)
for all n ∈ N.
(c) Since lim
n→∞
1
n
= 0, using Exercise 3a on page 20, it follows that lim
n→∞
1
√
n
= 0. Hence
lim
n→∞
_
1 +
1
√
n
_
2
= (1 + 0)
2
= 1 = lim
n→∞
1.
Using (2.27), it follows from the Sandwich theorem that (n
1
n
)
n∈N
is convergent and
lim
n→∞
n
1
n
= 1.
4. Consider the sequence (a
n
− a)
n∈N
. As a
n
∈ (a, b) for all n ∈ N, we have a
n
− a ≥ 0.
From the theorem on algebra of limits, it follows that the sequence (a
n
−a)
n∈N
is convergent
(being the sum of the convergent sequence (a
n
)
n∈N
and the convergent sequence (−a)
n∈N
).
Moreover, its limit is L−a. From Exercise 5 on page 13, we obtain L−a ≥ 0, that is a ≤ L.
Next consider the sequence (b − a
n
)
n∈N
. As a
n
∈ (a, b) for all n ∈ N, we have b − a
n
≥ 0.
From the theorem on algebra of limits, it follows that the sequence (b −a
n
)
n∈N
is convergent
(being the sum of the convergent sequence (−a
n
)
n∈N
and the convergent sequence (b)
n∈N
).
Moreover, its limit is −L + b. From Exercise 5 on page 13, we obtain −L + b ≥ 0, that is
L ≤ b.
Consequently a ≤ L ≤ b, that is, L ∈ [a, b].
Consider the sequence (
1
n
)
n∈N
contained in (0, 1). It is convergent with limit 0 ,∈ (0, 1).
5. For all n ∈ N, we have −
1
n
< b
n
− a
n
<
1
n
, and so by adding a
n
, we have −
1
n
+ a
n
< b
n
<
1
n
+a
n
. By the theorem on algebra of limits, we know that
lim
n→∞
_
−
1
n
+a
n
_
= lim
n→∞
−
1
n
+ lim
n→∞
a
n
= 0 + lim
n→∞
a
n
= lim
n→∞
a
n
lim
n→∞
_
1
n
+a
n
_
= lim
n→∞
1
n
+ lim
n→∞
a
n
= 0 + lim
n→∞
a
n
.
So by the sandwich theorem, it follows that (b
n
)
n∈N
is convergent with the limit lim
n→∞
a
n
.
90
Solution to the exercise on page 24
Observe that
the terms 2, 8, 2, 8 appear adjacently
and so the terms 1, 6, 1, 6 appear adjacently
and so the terms 6, 6, 6 appear adjacently
and so the terms 3, 6, 3, 6 appear adjacently
and so the terms 1, 8, 1, 8 appear adjacently
and so the terms 8, 8, 8 appear adjacently
and so the terms 6, 4, 6, 4 appear adjacently
and so the terms 2, 4, 2, 4 appear adjacently
and so the terms 8, 8, 8 appear adjacently
.
.
.
Hence we get the loop
. . . , 8, 8, 8, → . . . , 6, 4, 6, 4, → . . . , 2, 4, 2, 4, → . . . , 8, 8, 8, →,
which contains 6, and so 6 appears inﬁnite number of times. Thus we can choose indices
n
1
< n
2
< n
3
< . . .
such that for all k ∈ N, a
n
k
= 6. So (6)
k∈N
is a subsequence of the given sequence.
Solution to the exercise on page 26
Let (a
n
)
n∈N
be a Cauchy sequence. From Exercise 5 on page 17, it follows that (a
n
)
n∈N
is bounded.
By the BolzanoWeierstrass theorem, it follows that (a
n
)
n∈N
has a convergent subsequence, say
(a
n
k
))
k∈N
with limit L. Using the fact that (a
n
)
n∈N
is Cauchy, we now prove that (a
n
)
n∈N
is itself
convergent with limit L.
Let > 0. Since (a
n
)
n∈N
is Cauchy, there exists a N ∈ N such that for all n, m > N,
[a
n
−a
m
[ <
2
.
As (a
n
k
)
k∈N
is convergent with limit L, there exists a K ∈ N such that for all n
K
> N and
[a
nK
−L[ <
2
. Then for all n > N, we have
[a
n
−L[ = [a
n
−a
nK
+ a
nK
−L[
≤ [a
n
−a
nK
[ +[a
nK
−L[ (triangle inequality)
<
2
+
2
= .
Consequently, (a
n
)
n∈N
is convergent with limit L.
91
Solutions to the exercises on page 29
1. (a) Let > 0. If δ :=
√
, then we note that δ > 0. Moreover, if x ∈ R and [x −0[ = [x[ <
δ =
√
, then we obtain
[x
2
−0
2
[ = [x
2
[ = [x[ [x[ < δ δ =
√
√
= .
So f is continuous at 0.
(b) Let c ∈ R and suppose that c ,= 0. Let > 0. If x ∈ R, then
[x
2
−c
2
[ = [(x −c)(x +c)[ = [x −c[ [x +c[.
If x ∈ R is such that [x −c[ < δ, then
x < c +δ ≤ [c +δ[ ≤ [c[ +[δ[ = [c[ +δ, and
−x < δ −c ≤ [δ −c[ ≤ [δ[ +[ −c[ = δ +[c[.
Thus if x ∈ R satisﬁes [x −c[ < δ, then [x[ < δ +[c[, and so
[x +c[ ≤ [x[ +[c[ < δ +[c[ +[c[ = δ + 2[c[.
So if [x − c[ < δ, we have [x
2
− c
2
[ = [x − c[ [x + c[ < δ (δ + 2[c[). Thus in order to
make [x
2
−c
2
[ less than , we choose δ such that δ(δ + 2[c[) < : indeed, let
δ := min
_
2[c[ + 1
, 1
_
.
Since > 0, it follows that δ is positive. Furthermore, if x ∈ R satisﬁes [x − c[ < δ,
then we obtain
[x
2
−c
2
[ < δ(δ + 2[c[) ≤
2[c[ + 1
(1 + 2[c[) = .
2. (a) Let c
∈ R and > 0. Since f is continuous at c, there exists a δ > 0 such that for all
x ∈ R satisfying [x−c[ < δ, [f(x) −f(c)[ < . Then for all x ∈ R satisfying [x−c
[ < δ,
we have
9
[f(x) −f(c
)[ = [f(x) −f(c
) +f(c) −f(c)[ = [f(x −c
+c) −f(c)[ < ,
since [(x−c
+c) −c[ = [x−c
[ < δ. So f is continuous at c
. Since the choice of c
∈ R
was arbitrary, it follows that f is continuous on R.
(b) Let α ∈ R, and let f : R →R be given by f(x) = αx, for all x ∈ R. Then
f(x +y) = α(x +y) = αx +αy = f(x) +f(y).
3. We observe that as [f(0)[ ≤ M[0[ = M 0 = 0, it follows that [f(0)[ = 0, that is, f(0) = 0.
Given > 0, we deﬁne δ =
M
. Then for all x ∈ R satisfying [x[ = [x −0[ < δ, we have
[f(x) −f(0)[ = [f(x) −0[ = [f(x)[ ≤ M[x[ = M[x −0[ < Mδ = M
M
= .
Hence f is continuous at 0.
4. Let c ∈ R. Suppose that f is continuous at c. Consider =
1
2
> 0. Then ∃δ > 0 such that
for all x ∈ R satisfying [x −c[ < δ, [f(x) −f(c)[ < =
1
2
. We have the following two cases:
9
Here we use the fact that f is ‘additive’. First of all, f(0) = f(0 + 0) = f(0) + f(0), and so f(0) = 0. Hence it
follows that −f(c
) = f(−c
), since 0 = f(0) = f(c
− c
) = f(c
) + f(−c
). Finally we have f(x) − f(c
) + f(c) =
f(x) + f(−c
) + f(c) = f(x − c
) + f(c) = f(x − c
+ c).
92
1
◦
c ∈ Q. Then there exists x ∈ R ¸ Q such that [x −c[ < δ.
But [f(x) −f(c)[ = [1 −0[ = [1[ = 1 >
1
2
, a contradiction.
2
◦
c ∈ R ¸ Q. Then there exists x ∈ Q such that [x −c[ < δ.
But [f(x) −f(c)[ = [0 −1[ = [ −1[ = 1 >
1
2
, a contradiction.
Hence f is not continuous at c.
5. Since
f(c)
2
> 0, there exists a δ > 0 such that for all x ∈ R satisfying [x −c[ < δ,
[f(x) −f(c)[ <
f(c)
2
.
Thus for all x ∈ R satisfying [x −c[ < δ (that is, c −δ < x < c +δ), we have
f(c) −f(x) ≤ [f(c) −f(x)[ = [f(x) −f(c)[ <
f(c)
2
,
and so f(x) >
f(c)
2
> 0.
Solutions to the exercises on page 31
1. Let (x
n
)
n∈N
be a convergent sequence with limit c. Then (x
2
n
)
n∈N
is also convergent with limit
c
2
. Thus (f(x
n
))
n∈N
is convergent with limit f(c). So by Theorem 1.3.2, f is continuous.
2. If n ∈ N, then there exists q
n
∈ Q such that [q
n
−c[ <
1
n
.
Claim: (q
n
)
n∈N
is convergent with limit c.
Proof Let > 0. By the Archimedean principle, there exists an N ∈ N such that
1
< N.
Thus for n > N,
[q
n
−c[ <
1
n
<
1
N
< .
This proves the claim.
Since f is continuous at c, by Theorem 1.3.2, we have f(c) = lim
n→∞
f(q
n
) = lim
n→∞
0 = 0.
3. If x
1
,= x
2
, then the sequence x
1
, x
2
, x
1
, x
2
, . . . is divergent. (Indeed, the subsequence
x
1
, x
1
, x
1
, . . . converges to x
1
, while the subsequence x
2
, x
2
, x
2
, . . . converges to x
2
, and so
by Theorem 1.2.6, it follows that the sequence x
1
, x
2
, x
1
, x
2
, . . . is divergent.)
Thus the sequence f(x
1
), f(x
2
), f(x
1
), f(x
2
), . . . is divergent. Consequently f(x
1
) ,= f(x
2
):
for otherwise if f(x
1
) = f(x
2
), then the sequence
f(x
1
), f(x
2
), f(x
1
), f(x
2
) . . .
f(x
1
), f(x
1
), f(x
1
), f(x
1
) . . .
is a constant sequence, and so it is convergent with limit f(x
1
) (= f(x
2
)).
So we have shown that if x
1
,= x
2
, then f(x
1
) ,= f(x
2
), that is, the function f is onetoone.
93
Solution to the exercise on page 32
We apply Theorem 1.3.3 several times in order to prove this.
Since the function x → x is continuous on R, it follows that the function x → x
2
is continuous
on R as well. Moreover the function x → 1 is continuous on R, and so we obtain that the function
x → 1 + x
2
is continuous on R. As 1 + x
2
≥ 1 > 0 for all real x, we conclude that the function
x →
1
1+x
2
is continuous on R. Hence the function x → x
2
1
1+x
2
=
x
2
1+x
2
is continuous on R, that
is, f is continuous on R.
94
Solutions to the exercises on page 34
1. Let m := inf¦f(x) [ x ∈ [a, b]¦. We prove that there exists a d ∈ [a, b] such that f(d) = m.
For each n ∈ N, m < m +
1
n
, and so m+
1
n
cannot be a lower bound for ¦f(x) [ x ∈ [a, b]¦.
So there exists an x
n
∈ [a, b] such that
m ≤ f(x
n
) < m +
1
n
. (2.28)
By the BolzanoWeierstrass theorem, (x
n
)
n∈N
has a convergent subsequence (x
n
k
)
k∈N
with
limit d ∈ [a, b]. Since f is continuous at d, it follows that (f(x
n
k
))
k∈N
is convergent with
limit f(d). From (2.28), using the Sandwich theorem, we conclude that f(d) = m.
2. Let f : [0, 1] →R be deﬁned by
f(x) =
_
0 for all 0 ≤ x <
1
2
,
1 for all
1
2
≤ x ≤ 1.
Then f is not continuous at
1
2
: indeed the sequence
_
1
2
−
1
n+1
_
n∈N
is contained in [0, 1],
PSfrag replacements
1
1
0 1
2
Figure 2.3: A discontinuous function on [0, 1] that attains its extreme values.
and it is convergent with limit
1
2
−0 =
1
2
. However,
lim
n→∞
f
_
1
2
−
1
n + 1
_
= lim
n→∞
0 = 0 ,= 1 = f
_
1
2
_
.
But ¦f(x) [ x ∈ [0, 1]¦ = ¦0, 1¦, and so
sup¦f(x) [ x ∈ [0, 1]¦ = sup¦0, 1¦ = 1 = f(1), and
inf¦f(x) [ x ∈ [0, 1]¦ = inf¦0, 1¦ = 0 = f(0).
3. Consider the function g : [0, T] →R, given by
g(x) = f(x) for all x ∈ [0, T].
Then g is continuous on [0, T].
(Indeed, the continuity of g on [0, T] is a trivial consequence of the continuity of f on R: If
c ∈ [0, T], and > 0, then there exists a positive δ such that for all x ∈ R satisfying [x−c[ < δ,
[f(x)−f(c)[ < . Hence for all x ∈ [0, T] satisfying [x−c[ < δ, [g(x)−g(c)[ = [f(x)−f(c)[ < .
So g is continuous at c. But the choice of c was arbitrary, and so g is continuous on [0, T].)
Applying the extreme value theorem to g, we conclude that there exist c, d ∈ [0, T] such that
g(c) = max¦g(x) [ x ∈ [0, T]¦ and
g(d) = min¦g(x) [ x ∈ [0, T]¦.
So for all x ∈ [0, T], g(d) ≤ g(x) ≤ g(c), that is, f(d) ≤ f(x) ≤ f(c). So far we have proved
95
PSfrag replacements
0
g = f[
[0,T]
T
c
d
f(c)
f(d)
−T
2T 3T 4T
. . . . . .
Figure 2.4: A continuous periodic function is bounded.
the fact that f is bounded on [0, T]. We now prove that f is bounded on R by using the
periodicity of f. See Figure 2.4.
Now if x is any real number, there exists a n ∈ Z such that x = nT +r, where r ∈ R is such
that r ∈ [0, T). (Indeed, we have
x
T
=
_
x
T
_
+ Θ
where Θ ∈ [0, 1). Consequently, we obtain x = nT +r, where n :=
_
x
T
_
∈ Z and r := T Θ ∈
[0, T).) Thus f(x) = f(nT +r) = f(r). As f(d) ≤ f(r) ≤ f(c), it follows that f(d) ≤ f(x) ≤
f(c). Since the choice of x ∈ R was arbitrary, it follows that f(d) ≤ f(x) ≤ f(c) for all
x ∈ R. So f(c) and f(d) are upper and lower bounds, respectively, of the set ¦f(x) [ x ∈ R¦,
and so it is bounded.
4. (a) If x ∈ (a, b], then let f[
[x,b]
denote the restriction of f to [x, b], deﬁned by f[
[x,b]
(y) =
f(y) for all y ∈ [x, b]. We note that f[
[x,b]
is a continuous function. Applying the
intermediate value theorem to f[
[x,b]
, we see that
max¦f[
[x,b]
(y) [ y ∈ [a, x]¦ = max¦f(y) [ y ∈ [a, x]¦
exists, and so f
∗
is welldeﬁned.
(b) First we observe that f
∗
is an increasing function, that is, if x < y then, f
∗
(x) ≤ f
∗
(y).
Furthermore, since sup(A ∪ B) = max¦supA, sup B¦ (why?), it follows that if x < y,
then
f
∗
(y) ≤ max¦f
∗
(x), max¦f(z) [ z ∈ [x, y]¦.
Let c ∈ [a, b], and let > 0. Choose δ > 0 such that for all x ∈ [a, b] such that [x−c[ < δ,
[f(x) − f(c)[ < . Let x ∈ [a, b] be such that [x − c[ < δ. Then we have the following
two cases:
1
◦
Let c < x < c +δ. Then we have
f
∗
(x) = max¦f
∗
(c), max¦f(z) [ z ∈ [c, x]¦¦
≤ max¦f
∗
(c), f(c) +¦
≤ max¦f
∗
(c), f
∗
(c) +¦
= f
∗
(c) +,
and so [f
∗
(x) −f
∗
(c)[ = f
∗
(x) −f
∗
(c) ≤ .
2
◦
Let c −δ < x ≤ c. If x = c, then [f
∗
(x) −f
∗
(c)[ = 0 < trivially. If c −δ < x < c,
then for any z ∈ (c −δ, c), we have
[f(z) −f(x)[ = [f(z) −f(c) +f(c) −f(x)[ ≤ [f(z) −f(c)[ +[f(c) −f(x)[ = 2,
and so f(z) ≤ f(x) + 2. Thus
f
∗
(c) = max¦f
∗
(x), max¦f(z) [ z ∈ [x, c]¦¦
≤ max¦f
∗
(x), f(x) + 2¦
≤ max¦f
∗
(x), f
∗
(x) + 2¦
= f
∗
(x) + 2,
and so [f
∗
(x) −f
∗
(c)[ = f
∗
(c) −f
∗
(x) ≤ 2.
96
Hence for all x ∈ [a, b] satisfying [x−c[ < δ, we have [f
∗
(x) −f
∗
(c)[ < 2. Consequently
f
∗
is continuous at c. Since the choice of c was arbitrary, it follows that f
∗
is continuous
on [a, b].
(c) f
∗
is given by
f
∗
(x) =
_
x −x
2
if 0 ≤ x ≤
1
2
,
1
4
if
1
2
< x ≤ 1.
See Figure 2.5.
PSfrag replacements
0
0
1
1
1
4
1
4
1
2
1
2
f
f
∗
f(d)
Figure 2.5: f and f
∗
.
Solutions to the exercises on page 37
1. Since the function f : [0, 1] →R and the function x → x from [0, 1] to R are both continuous
on the interval [0, 1], it follows that also the function g : [0, 1] →R deﬁned by
g(x) = f(x) −x, for all x ∈ [0, 1]
is continuous on [0, 1]. Since 0 ≤ f(x) ≤ 1 for all x ∈ [0, 1], we have
g(0) = f(0) −0 = f(0) ≥ 0, and
g(1) = f(1) −1 ≤ 0.
So by the intermediate value theorem, there exists a c ∈ [0, 1] such that g(c) = 0, that is,
f(c) = c.
2. Let the weekend campsite be at altitude H. Let u : [0, 1] → R be the position function for
the walk up, and d :
_
0,
1
2
¸
→ R be the position function for the walk down. (We assume
that these are continuous functions.) Consider the function f :
_
0,
1
2
¸
→R given by
f(t) = u(t) −d(t), t ∈
_
0,
1
2
_
.
Then f is also continuous, and moreover
f(0) = u(0) −d(0) = 0 −H = −H < 0, while
f
_
1
2
_
= u
_
1
2
_
−d
_
1
2
_
= u
_
1
2
_
−0 = u
_
1
2
_
> 0.
See Figure 2.6. Hence by the intermediate value theorem, it follows that there exists a
c ∈
_
0,
1
2
¸
such that f(c) = 0, that is, u(c) = d(c). So at time c past 8:00, the hiker was
exactly at the same spot on Saturday and Sunday.
97
PSfrag replacements
0
H
u d
1
2
1 c
u(c) = d(c)
Figure 2.6: The walks up and down.
3. The polynomial function p : [−1, 2] →R is continuous on the interval [−1, 2]. Moreover,
p(−1) = 2 (−1)
3
−5 (−1)
2
−10 (−1) + 5 = −2 −5 + 10 + 5 = 8 > 0, and
p(2) = 2 (2)
3
−5 (2)
2
−10 (2) + 5 = 16 −20 −20 + 5 = −19 < 0.
Since p(−1) > 0 > p(2), from the intermediate value theorem applied to the continuous
function p on the interval [−1, 2], we conclude that there must exist a c ∈ [−1, 2] such that
p(c) = 0. So p has a real root in the interval [−1, 2].
4. Obviously S ⊂ R. We now show the reverse inclusion. Let y ∈ R.
As S is not bounded above, y is not an upper bound of S, that is, there exists a x
0
∈ R such
that f(x
0
) < y.
Similarly, since S is not bounded below, y is not a lower bound of S, and so there exists a
x
1
∈ R such that f(x
1
) > y.
Now consider the restriction of f to the interval with endpoints x
0
and x
1
with the end
points included in the interval. Applying the intermediate value theorem to this continuous
function, it follows that there exists a real number c such that f(c) = y.
This shows that S = R.
5. (a) The following three cases are possible:
1
◦
f(0) = 0. Then let x
0
= 0 and let m ∈ Z. Clearly f(x
0
) = f(0) = 0 = m0 = mx
0
.
2
◦
f(0) > 0. Choose N ∈ N satisfying N > f(1) (that such a N exists follows
from the Archimedean property). Consider the function g : [0, 1] → R deﬁned by
g(x) = f(x) −Nx, x ∈ [0, 1]. As the functions f and x → Nx are continuous, so is
g. Note that g(0) = f(0) − N0 = f(0) > 0, while g(1) = f(1) − N < 0. Applying
the intermediate value theorem to g (with y = 0), it follows that there exists a
x
0
∈ [0, 1] such that g(x
0
) = 0, that is, f(x
0
) = Nx
0
.
3
◦
f(0) < 0. Choose a N ∈ N such that N > −f(1) (again the Archimedean property
guarantees the existence of such a N), and consider the continuous function g :
[0, 1] → R deﬁned by g(x) = f(x) + Nx. We observe that g(0) = f(0) < 0, and
g(1) = f(1) + N > 0, and so by the intermediate value theorem, it follows that
there exists a x
0
∈ [0, 1] such that g(x
0
) = 0, that is, f(x
0
) = −Nx
0
.
This completes the proof.
(b) Suppose that such a continuous function exists. From the part above, it follows that
there exists a x
0
∈ R and a m ∈ Z such that f(x
0
) = mx
0
. We have the following two
possible cases:
1
◦
x
0
∈ Q. But then f(x
0
) is irrational, while mx
0
is rational, a contradiction.
2
◦
x
0
,∈ Q. But then f(x
0
) is rational, whereas mx
0
is irrational, a contradiction.
So f cannot be continuous.
98
Algebra
Solutions to the exercises on page 44
1. (a) Addition modulo 6 is a law of composition on Z
6
, since if [a], [b] ∈ Z
6
, then [a] ⊕[b] =
[a +b] ∈ Z
6
. Moreover, we verify below that the group axioms are satisﬁed.
G1. Addition modulo 6 is associative: if [a], [b], [c] ∈ Z
6
, then
([a] ⊕[b]) ⊕[c] = [a +b] ⊕[c]
= [(a +b) +c]
= [a + (b +c)]
= [a] ⊕[b +c]
= [a] ⊕([b] ⊕[c]).
G2. [0] is the identity element: for all [a] ∈ Z
6
, we have
[0] ⊕[a] = [0 +a] = [a] = [a + 0] = [a] ⊕[0].
G3. If [a] ∈ Z
6
, then [−a] ∈ Z
6
, and
[a] ⊕[−a] = [a + (−a)] = [0] = [−a +a] = [−a] ⊕[a].
So Z
6
with addition modulo 6 forms a group.
We give its group table below:
⊕ [0] [1] [2] [3] [4] [5]
[0] [0] [1] [2] [3] [4] [5]
[1] [1] [2] [3] [4] [5] [0]
[2] [2] [3] [4] [5] [0] [1]
[3] [3] [4] [5] [0] [1] [2]
[4] [4] [5] [0] [1] [2] [3]
[5] [5] [0] [1] [2] [3] [4]
(b) No; multiplication modulo 6 is a law of composition on Z
6
that satisﬁes G1 and G2,
but G3 is not true:
G1. Multiplication modulo 6 is associative: if [a], [b], [c] ∈ Z
6
, then
([a] ⊗[b]) ⊗[c] = [ab] ⊗[c]
= [(ab)c]
= [a(bc)]
= [a] ⊗[bc]
= [a] ⊗([b] ⊗[c]).
G2. (If e = [k] is the identity element for some integer k, then for all [a] ∈ Z
6
, [a] ⊗e =
[a] = e⊗[a]. In particular, with [a] = [1] ∈ Z
6
, we obtain e = [k] = [1k] = [1]⊗[k] =
[1] ⊗ e = [1]. So if there is an identity element e, then it must be [1].) Indeed [1]
serves as an identity element, since for all [a] ∈ Z
6
,
[a] ⊗[1] = [a 1] = [a] = [1 a] = [a] ⊗[a].
G3 is not satisﬁed. For instance, [0] ∈ Z
6
, but for all [a] ∈ Z
6
, [0] ⊗ [a] = [a] ⊗ [0] =
[0] ,= [1]. So there does not exist an inverse of the element [0] in Z
6
.
99
Hence Z
6
with multiplication modulo 6 is not a group.
Z
∗
6
with multiplication modulo 6 is also not a group, since multiplication modulo 6 is
not a law of composition on the set Z
∗
6
: indeed, [2], [3] ∈ Z
∗
6
, but [2] ⊗ [3] = [2 3] =
[6] = [0] ,∈ Z
∗
6
.
(c) If: Let m be a prime number.
First we show that multiplication modulo m is a law of composition on Z
∗
m
. Let
[a], [b] ∈ Z
∗
m
. Then a and b are not divisible by m, and since m is prime, it follows that
ab is also not divisible by m. Thus [a] ⊗[b] = [ab] ,= [0], and so [a] ⊗[b] ∈ Z
∗
m
.
Next we show that the group axioms are satisﬁed:
G1. If [a], [b], [c] ∈ Z
∗
m
, then
([a] ⊗[b]) ⊗[c] = [ab] ⊗[c]
= [(ab)c]
= [a(bc)]
= [a] ⊗[bc]
= [a] ⊗([b] ⊗[c]).
G2. [1] ,= [0] and so [1] ∈ Z
∗
m
. Furthermore, For all [a] ∈ Z
∗
m
, we have
[a] ⊗[1] = [a 1] = [a] = [1 a] = [1] ⊗[a].
G3. If [a] ∈ Z
∗
m
, then m does not divide a (for otherwise [a] = [0]). As m is prime, it
follows that gcd(a, m) = 1. (Indeed, gcd(a, m) divides m, and since m is prime,
the only factors of m are 1, m, −1, −m. As gcd(a, m) ≥ 0, the only possible values
are m and 1. Also, gcd(a, m) divides a, and since a is not divisible by m, it follows
that gcd(a, m) ,= m. So the only possible value of gcd(a, m) is 1.) So there exist
s, t ∈ Z such that
as +mt = 1.
Consequently
[s] ⊗[a] = [sa] = [as] = [a] ⊗[s]
= [as]
= [1 −mt]
= [1] ⊕[m(−t)]
= [1] ⊕[0]
= [1].
Furthermore [s] ,= [0], since otherwise m[s, and so m[as + mt, that is m[1, a
contradiction. Hence [s] ∈ Z
∗
m
, and so [s] is the inverse of [a].
Hence Z
∗
m
with multiplication modulo m is a group.
Only if: Suppose that m = p q, where p, q ∈ ¦2, 3, . . . , m− 1¦. Clearly, as m does
not divide p or q, it follows that [p] ,= [0] and [q] ,= [0]. So [p], [q] ∈ Z
∗
m
. However,
[p] ⊗[q] = [p q] = [m] = [0] ,∈ Z
∗
m
,
and so multiplication modulo m is not a law of composition on Z
∗
m
. Consequently, Z
∗
m
with multiplication modulo m is not a group.
2. (a) Composition of functions gives a law of composition on the set S
n
of all bijections from
¦1, 2, 3, . . . , n¦ onto itself. Indeed if f, g are bijections from ¦1, 2, 3, . . . , n¦ onto itself,
then f ◦ g is again a bijection from ¦1, 2, 3, . . . , n¦ onto itself:
100
i. f ◦ g is onetoone. Let k
1
, k
2
∈ ¦1, 2, 3, . . . , n¦. If (f ◦ g)(k
1
) = (f ◦ g)(k
2
), then
f(g(k
1
)) = f(g(k
2
)), and since f is onetoone, we get g(k
1
) = g(k
2
). Furthermore,
since g is onetoone, k
1
= k
2
.
ii. f ◦ g is onto. If k ∈ ¦1, 2, 3, . . . , n¦, then since f is onto, there exists a k
∈
¦1, 2, 3, . . . , n¦ such that f(k
) = k. Moreover, since g is onto, there exists a
k
∈ ¦1, 2, 3, . . . , n¦ such that g(k
) = k
. Thus (f ◦g)(k
) = f(g(k
)) = f(k
) = k.
So f ◦ g is onto.
Next we show that the group axioms are satisﬁed:
G1. If f, g, h ∈ S
n
, then for all k ∈ ¦1, 2, 3, . . . , n¦,
((f ◦ g) ◦ h)(k) = (f ◦ g)(h(k)) = f(g(h(k))) = f((g ◦ h)(k)) = (f ◦ (g ◦ h))(k).
So (f ◦ g) ◦ h = f ◦ (g ◦ h).
G2. The identity function ι : ¦1, 2, 3, . . . , n¦ → ¦1, 2, 3, . . . , n¦ deﬁned by ι(k) = k for
all k ∈ ¦1, 2, 3, . . . , n¦, serves as the identity element in S
n
. Indeed, for all f ∈ S
n
,
we have
(f ◦ ι)(k) = f(ι(k)) = f(k) = ι(f(k)) = (ι ◦ f)(k), ∀k ∈ ¦1, 2, 3, . . . , n¦.
Thus f ◦ ι = f = ι ◦ f.
G3. If f ∈ S
n
, then since f is a bijection, it follows that for all k ∈ ¦1, 2, 3, . . . , n¦, there
exists a unique element k
such that f(k
) = k. Deﬁne f
−1
: ¦1, 2, 3, . . . , n¦ →
¦1, 2, 3, . . . , n¦ as follows: if k ∈ ¦1, 2, 3, . . . , n¦, then f
−1
(k) = k
, where k
∈
¦1, 2, 3, . . . , n¦ is such that f(k
) = k. Then for all k ∈ ¦1, 2, 3, . . . , n¦, we have
(f ◦ f
−1
)(k) = f(f
−1
(k)) = k = ι(k) = k = f
−1
(f(k)) = (f
−1
◦ f)(k),
and so f ◦ f
−1
= ι = f
−1
◦ f.
Hence S
n
with the composition of functions is a group.
(b) The number of bijections from a set with n elements onto itself is n!, and so the order
of S
n
is n!.
Indeed, in order to specify the bijection f, one needs to specify f(1), . . . , f(n). The
number of ways of specifying f(1) is n (as it can be any one of the numbers 1, . . . , n). As
f is onetoone, f(2) can be only be any one of the elements of the set ¦1, 2, 3, . . . , n¦ ¸
¦f(1)¦ (which has n − 1 elements). Proceeding in this way, the number of distinct
bijections we get are n (n −1) 2 1 = n!.
(c) Let f : ¦1, 2, 3¦ → ¦1, 2, 3¦ be the function
f(1) = 2
f(2) = 1
f(3) = 3,
and g : ¦1, 2, 3¦ → ¦1, 2, 3¦ be the function
g(1) = 1
g(2) = 3
g(3) = 2.
Then
(f ◦ g)(1) = f(g(1)) = f(1) = 2 ,= 3 = g(2) = g(f(1)) = (g ◦ f)(1),
and so f ◦ g ,= g ◦ f.
101
(d) If: S
1
has the only element ι : ¦1¦ → ¦1¦, and so ι ◦ ι = ι = ι ◦ ι. Thus S
1
is abelian.
S
2
has the two elements ι : ¦1, 2¦ → ¦1, 2¦, and r : ¦1, 2¦ → ¦1, 2¦ given by
r(1) = 2
r(2) = 1.
Since ι ◦ r = r ◦ ι we see that S
2
is abelian.
So if n ≤ 2, then S
n
is abelian.
Only if: If n > 2, then consider the bijections f : ¦1, 2, 3, . . . , n¦ → ¦1, 2, 3, . . . , n¦
and g : ¦1, 2, 3, . . . , n¦ → ¦1, 2, 3, . . . , n¦ given by
f(1) = 2
f(2) = 1
f(n) = n ∀n ∈ ¦3, . . . , n¦
and
g(1) = 1
g(2) = 3
g(3) = 2
g(n) = n ∀n ∈ ¦1, 2, 3, . . . , n¦ ¸ ¦1, 2, 3¦
.
Then
(f ◦ g)(1) = f(g(1)) = f(1) = 2 ,= 3 = g(2) = g(f(1)) = (g ◦ f)(1),
and so f ◦ g ,= g ◦ f. Thus S
n
is not abelian.
Hence S
n
is abelian iﬀ n ≥ 2.
3. We note that if
_
a b
−b a
_
,
_
a
b
−b
a
_
∈ S,
then
_
a b
−b a
_ _
a
b
−b
a
_
=
_
aa
−bb
ab
+ ba
−(ab
+ba
) aa
−bb
_
,
and
(aa
−bb
)
2
+ (ab
+ba
)
2
= a
2
a
2
−2aa
bb
+b
2
b
2
+a
2
b
2
−2ab
ba
+b
2
a
2
= a
2
(a
2
+b
2
) +b
2
(b
2
+a
2
)
= (a
2
+b
2
)(a
2
+b
2
)
,= 0.
Hence matrix multiplication is a law of composition on S. Next we verify that the group
axioms are satisﬁed:
G1. Matrix multiplication is associative, and so for all A, B, C ∈ S, (AB)C = A(BC).
G2. Let I denote the identity matrix. Then we have
I =
_
1 0
0 1
_
=
_
1 0
−0 1
_
and moreover, 1
2
+ 0
2
= 1 ,= 0. So I belongs to S. Furthermore, for every A ∈ S,
AI = A = IA.
G3. Let
A =
_
a b
−b a
_
∈ S.
Then a
2
+b
2
,= 0. If
B :=
_
a
a
2
+b
2
b
a
2
+b
2
−b
a
2
+b
2
a
a
2
+b
2
_
,
102
then
_
a
a
2
+b
2
_
2
+
_
b
a
2
+b
2
_
2
=
a
2
+b
2
(a
2
+b
2
)
2
=
1
a
2
+b
2
,= 0.
So B ∈ S, and furthermore, AB = I = BA.
So S is a group with matrix multiplication.
4. (a) Let a, b, c ∈ G, and suppose that a ∗ b = a ∗ c. Since a ∈ G, there exists a
−1
∈ G such
that a ∗ a
−1
= e = a
−1
∗ a, where e denotes the identity element in G. Then we have
b = e ∗ b (since e is the identity element)
= (a
−1
∗ a) ∗ b (since a
−1
is the inverse of a)
= a
−1
∗ (a ∗ b) (associativity)
= a
−1
∗ (a ∗ c) (since a ∗ b = a ∗ c)
= (a
−1
∗ a) ∗ c (associativity)
= e ∗ c (since a
−1
is the inverse of a)
= c (since e is the identity element).
(b) Clearly
a ∗ (a
−1
∗ b) = (a ∗ a
−1
) ∗ b = e ∗ b = b,
and so x = a
−1
∗ b is a solution to a ∗ x = b.
If there are two solutions, say x
1
and x
2
, then
a ∗ x
1
= b = a ∗ x
2
,
and by part 4a above, it follows that x
1
= x
2
. Hence a ∗ x = b has the unique solution
x = a
−1
∗ b.
(c) Since a, b ∈ G, there exist elements a
−1
, b
−1
∈ G such that
a ∗ a
−1
= e = a
−1
∗ a and b ∗ b
−1
= e = b
−1
∗ b.
We have
(a ∗ b) ∗ (b
−1
∗ a
−1
) = a ∗ (b ∗ (b
−1
∗ a
−1
)) (associativity)
= a ∗ ((b ∗ b
−1
) ∗ a
−1
) (associativity)
= a ∗ (e ∗ a
−1
) (since b
−1
is the inverse of b)
= a ∗ a
−1
(since e is the identity element)
= e (since a
−1
is the inverse of a),
and
(b
−1
∗ a
−1
) ∗ (a ∗ b) = ((b
−1
∗ a
−1
) ∗ a) ∗ b (associativity)
= (b
−1
∗ (a
−1
∗ a)) ∗ b (associativity)
= (b
−1
∗ e) ∗ b (since a
−1
is the inverse of a)
= b
−1
∗ b (since e is the identity element)
= e (since b
−1
is the inverse of b).
Thus b
−1
∗ a
−1
is an inverse of a ∗ b. Since the inverse of a ∗ b is unique, it follows that
(a ∗ b)
−1
= b
−1
∗ a
−1
.
103
Solutions to the exercises on page 46
1. (a) FALSE. Although H1 and H2 hold, H3 does not hold. Indeed, 1 is a nonnegative integer,
but its inverse −1 is not a nonnegative integer.
(b) FALSE. H1 and H2 are not true. Indeed, the sum of the odd integers 1 and −1 is 0,
which is not an odd integer. So H1 and H2 do not hold. But H3 is true, since if m is
an odd integer, then so is −m.
(c) TRUE. For all a, b ∈ H, we have that a, b also belong to G, and as G is abelian, we
know that a ∗ b = b ∗ a.
2. Yes. For instance, given any inﬁnite group G, the set ¦e¦ comprising the identity element e
is a subgroup of the group G. Thus ¦0¦ is a ﬁnite subgroup of the inﬁnite group Z of integers
with addition. The subgroup ¦−1, 1¦ of the set of nonzero real numbers with multiplication
is another example.
3. (a) We note that
H = ¦4m [ m ∈ Z¦ = ¦. . . , −12, −8, −4, 0, 4, 8, 12, . . . ¦, and
K = ¦6m [ m ∈ Z¦ = ¦. . . , −18, −12, −6, 0, 6, 12, 18, . . . ¦.
Claim: H ∩ K = ¦12m[ m ∈ Z¦ = ¦. . . , −12, 0, 12, . . . ¦.
Proof If k ∈ H∩K, then k ∈ H and k ∈ K. Hence 4[k and 6[k. So there exist integers
k
1
and k
2
such that k = 4k
1
and k = 6k
2
. Thus 4k
1
= 6k
2
and so 2k
1
= 3k
2
. Since
integers possess a unique factorization into primes, it follows that 2[k
2
. Consequently,
k
2
= 2k
2
, and so k = 6k
2
= 6(2k
2
) = 12k
2
. Hence k ∈ ¦12m [ m ∈ Z¦. So we have
shown that H ∩ K ⊂ ¦12m [ m ∈ Z¦.
Conversely, if k ∈ ¦12m [ m ∈ Z¦, then k = 12m for some m ∈ Z. Thus k = 4(3m) =
6(2m), and so k is a multiple of 4 and 6. Consequently k belongs to H as well as K.
Hence ¦12m[ m ∈ Z¦ ⊂ H ∩ K.
(b) We check that H1, H2, H3 hold:
H1. If a, b ∈ H ∩ K, then a, b belong to both H and K. As H is a subgroup and
a, b ∈ H, it follows that a ∗ b ∈ H. Similarly, as K is a subgroup and a, b ∈ K, it
follows that a ∗ b ∈ K as well. Hence a ∗ b ∈ H ∩ K.
H2. As H is a subgroup, e ∈ H. Also, as K is a subgroup, it follows that e ∈ K as well.
Thus e ∈ H ∩ K.
H3. If a ∈ H ∩K, then a ∈ H and a ∈ K. Since H is a subgroup and a ∈ H, it follows
that a
−1
∈ H. Also, since K is a subgroup and a ∈ K, it follows that a
−1
∈ K.
Consequently a
−1
∈ H ∩ K.
So H ∩ K is a subgroup of G.
4. (a) We check that H1, H2, H3 hold:
H1. If f, g ∈ H
1
, then f and g are continuous on the interval [0, 1], and moreover,
f
_
1
2
_
= 0 and g
_
1
2
_
= 0. As f, g are continuous on the interval [0, 1], it follows
that f +g is also continuous on [0, 1]. Furthermore,
(f +g)
_
1
2
_
= f
_
1
2
_
+g
_
1
2
_
= 0 + 0 = 0.
Consequently, f +g ∈ H
1
.
104
H2. The constant function 0 : [0, 1] → R, deﬁned by 0(x) = 0 for all x ∈ [0, 1], is the
identity element in the group C[0, 1] with addition of functions. Moreover,
0
_
1
2
_
= 0,
and so 0 ∈ H
1
.
H3. If f ∈ H
1
, then f
_
1
2
_
= 0. The inverse of f in C[0, 1] is the function −f : [0, 1] →R,
deﬁned by (−f)(x) = −f(x) for all x ∈ [0, 1]. As
(−f)
_
1
2
_
= −f
_
1
2
_
= −0 = 0,
it follows that −f ∈ H
1
.
So H
1
is a subgroup of the group C[0, 1] with addition of functions.
(b) We check that H1, H2, H3 hold:
H1. If p, q ∈ H
2
, then p and q are continuous on the interval [0, 1], and moreover, there
exist n, m ∈ N ∪ ¦0¦ and real numbers a
0
, a
1
, a
2
, . . . , a
n
, b
0
, b
1
, . . . , b
m
, such that
p(x) = a
0
+a
1
x +a
2
x
2
+ +a
n
x
n
, for all x ∈ [0, 1], and
q(x) = b
0
+b
1
x +b
2
x
2
+ +b
m
x
m
, for all x ∈ [0, 1].
If n ≤ m, then for all x ∈ [0, 1],
(p +q)(x) = p(x) +q(x)
= a
0
+a
1
x +a
2
x
2
+ +a
n
x
n
+b
0
+b
1
x +b
2
x
2
+ +b
m
x
m
= a
0
+b
0
+ (a
1
+b
1
)x + (a
2
+b
2
)x
2
+ + (a
n
+b
n
)x
n
+b
n+1
x
n+1
+ +b
m
x
m
,
and so p +q ∈ H
2
.
On the other hand, if n > m, then for all x ∈ [0, 1],
(p +q)(x) = p(x) +q(x)
= a
0
+a
1
x +a
2
x
2
+ +a
n
x
n
+b
0
+b
1
x +b
2
x
2
+ +b
m
x
m
= a
0
+b
0
+ (a
1
+b
1
)x + (a
2
+b
2
)x
2
+ + (a
m
+b
m
)x
m
+a
m+1
x
m+1
+ +a
n
x
n
,
and so p +q ∈ H
2
.
H2. The constant function 0 : [0, 1] → R, deﬁned by 0(x) = 0 for all x ∈ [0, 1], is the
identity element in the group C[0, 1] with addition of functions. Clearly, 0 ∈ H
2
(with n = 0 and a
0
= 0).
H3. If p ∈ H
2
, then there exists an n ∈ N ∪ ¦0¦ and real numbers a
0
, a
1
, a
2
, . . . , a
n
,
such that p(x) = a
0
+a
1
x+a
2
x
2
+ +a
n
x
n
, for all x ∈ [0, 1]. The inverse of p in
C[0, 1] is the function −p : [0, 1] →R, deﬁned by (−p)(x) = −p(x) for all x ∈ [0, 1].
As
(−p)(x) = −p(x)
= −(a
0
+a
1
x +a
2
x
2
+ +a
n
x
n
)
= −a
0
+ (−a
1
)x + (−a
2
)x
2
+ + (−a
n
)x
n
,
for all x ∈ [0, 1], it follows that −p ∈ H
2
.
So H
2
is a subgroup of the group C[0, 1] with addition of functions.
5. Let e denote the identity in the group G.
105
(a) Since e ∈ G, and for every a ∈ G, a ∗ e = a = e ∗ a, it follows that e ∈ Z(G). Thus
Z(G) is not empty.
(b) Clearly Z(G) ⊂ G.
Let z ∈ G. For every a ∈ G, z ∗ a = a ∗ z, since G is abelian. So z ∈ Z(G). Hence
G ⊂ Z(G).
Thus Z(G) = G.
(c) We check that H1, H2, H3 hold:
H1. If z
1
, z
2
∈ Z(G), then for every a ∈ G, we have
(z
1
∗ z
2
) ∗ a = z
1
∗ (z
2
∗ a) (associativity)
= z
1
∗ (a ∗ z
2
) (since z
2
∈ Z(G))
= (z
1
∗ a) ∗ z
2
(associativity)
= (a ∗ z
1
) ∗ z
2
(since z
1
∈ Z(G))
= a ∗ (z
1
∗ z
2
) (associativity).
Thus z
1
∗ z
2
∈ Z(G).
H2. e ∈ Z(G) since for every a ∈ G, e ∗ a = a = a ∗ e.
H3. Let z ∈ Z(G). Then for every a ∈ G, we have a
−1
∈ G, and so
z ∗ a
−1
= a
−1
∗ z.
Consequently (z ∗ a
−1
)
−1
= (a
−1
∗ z)
−1
, that is, (a
−1
)
−1
∗ z
−1
= z
−1
∗ (a
−1
)
−1
.
But a ∗ a
−1
= e = a
−1
∗ a, and from the uniqueness of inverses, it follows that
(a
−1
)
−1
= a. Thus we obtain a ∗ z
−1
= (a
−1
)
−1
∗ z
−1
= z
−1
∗ (a
−1
)
−1
= z
−1
∗ a.
Hence z
−1
∈ Z(G).
So Z(G) is a subgroup of the group G.
(d) If z =
_
α β
γ δ
_
∈ Z(GL(2, R)), then since
a =
_
1 1
0 1
_
∈ GL(2, R),
it follows that
_
α β
γ δ
_ _
1 1
0 1
_
=
_
1 1
0 1
_ _
α β
γ δ
_
.
Thus
_
α α +β
γ γ +δ
_
=
_
α +γ β +δ
γ δ
_
,
and so it follows that γ = 0 and α = δ. Also, since
b =
_
1 0
1 1
_
∈ GL(2, R),
it follows that
_
α β
0 α
_ _
1 0
1 1
_
=
_
1 0
1 1
_ _
α β
0 α
_
.
Thus
_
α +β β
α α
_
=
_
α β
α β +α
_
,
and so it follows that β = 0. Hence we have shown that
Z(GL(2, R)) ⊂
_
α I =
_
α 0
0 α
_ ¸
¸
¸
¸
α ∈ R ¸ ¦0¦
_
. (2.29)
106
Let α ∈ R ¸ ¦0¦. Then α I ∈ GL(2, R), and for all
_
p q
r s
_
∈ GL(2, R),
we have
_
α 0
0 α
_ _
p q
r s
_
=
_
αp αq
αr αs
_
=
_
pα qα
rα sα
_
=
_
p q
r s
_ _
α 0
0 α
_
.
Hence α I ∈ Z(GL(2, R). Thus
__
α 0
0 α
_ ¸
¸
¸
¸
α ∈ R ¸ ¦0¦
_
⊂ Z(GL(2, R). (2.30)
From (2.29) and (2.30), we conclude that
Z(GL(2, R)) =
__
α 0
0 α
_ ¸
¸
¸
¸
α ∈ R ¸ ¦0¦
_
.
Solutions to the exercises on page 48
1. We know that S = ¦e, a, a
2
, a
3
, . . . , a
G
¦ ⊂ G, and that S has [G[ + 1 elements, while
G has [G[ elements. So from the pigeonhole principle, it follows that there exist distinct
k
1
, k
2
∈ ¦0, 1, 2, 3, . . . , [G[¦ such that a
k1
= a
k2
. We may assume that k
2
> k
1
(otherwise we
can interchange them), and so a
k2−k1
= e. As k
2
−k
1
∈ N, it follows that a has ﬁnite order.
Furthermore, as k
2
≤ [G[ and k
1
≥ 0, we obtain k
2
−k
1
≤ [G[. Thus
ord(a) = min¦m ∈ N [ a
m
= e¦ ≤ k
2
−k
1
≤ [G[.
So a has order at most equal to [G[.
2. Claim: For all m, n ∈ N, a
m
∗ a
n
= a
m+n
.
Proof We prove this in several steps.
Step 1. Let m ≥ 0. We show that a
m
∗ a
n
= a
m+n
for all n ≥ 0 by induction on n. Clearly
a
m
∗ a
0
= a
m
∗ e = a
m
= a
m+0
,
and so the result is true for n = 0. If a
m
∗ a
k
= a
m+k
for some k ≥ 0, then we have
a
m
∗ a
k+1
= a
m
∗ (a
k
∗ a) (deﬁnition of a
k+1
)
= (a
m
∗ a
k
) ∗ a (associativity)
= a
m+k
∗ a (induction hypothesis)
= a
(m+k)+1
(deﬁnition of a
(m+k)+1
)
= a
m+(k+1)
.
So by induction, we have a
m
∗ a
n
= a
m+n
for all n ≥ 0. But the choice of m ≥ 0 was
arbitrary, and so we have shown that
for all m ≥ 0, and all n ≥ 0, a
m
∗ a
n
= a
m+n
. (2.31)
Step 2. Let m < 0 and n ≥ 0. Then we have the following two cases:
107
1
◦
If n + m > 0, then from (2.31), we have a
−m
∗ a
n+m
= a
−m+(n+m)
= a
n
, and so
a
−m
∗ a
n+m
= a
n
. We observe that a
m
∗ a
−m
= (a
−m
)
−1
∗ a
−m
= e. So we obtain
a
n+m
= e ∗ a
n+m
= (a
m
∗ a
−m
) ∗ a
n+m
= a
m
∗ (a
−m
∗ a
n+m
) = a
m
∗ a
n
.
2
◦
If n + m ≤ 0, then from (2.31), we have a
n
∗ a
−n−m
= a
n+(−n−m)
= a
−m
, and so
a
n
∗ a
−n−m
= a
−m
. So by premultiplying by a
m
= (a
−m
)
−1
and postmultiplying by
a
n+m
= (a
−n−m
)
−1
, we obtain a
m
∗ a
n
= a
n+m
= a
m+n
.
Hence we have shown that for all m < 0, and all n ≥ 0, a
m
∗ a
n
= a
m+n
. Combining this
result with the result from Step 1, we obtain
for all m ∈ Z, and all n ≥ 0, a
m
∗ a
n
= a
m+n
. (2.32)
Step 3. Let m ∈ Z and n < 0. Then from Step 2, we have a
m+n
∗ a
−n
= a
(m+n)−n
= a
m
.
By postmultiplying both sides by a
n
= (a
−n
)
−1
, we then obtain a
m+n
= a
m
∗ a
n
. Hence we
have shown that for all m ∈ Z and all n < 0, a
m
∗a
n
= a
m+n
. Combining this result with the
result from Step 2, namely (2.32), we obtain that for all integers m and n, a
m
∗a
n
= a
m+n
.
Next we prove the following claim.
Claim: For all m, n ∈ N, (a
m
)
n
= a
mn
.
Proof First we show that for all m ∈ Z and all n ≥ 0, (a
m
)
n
= a
mn
, by induction on n.
Let m ∈ Z. We have (a
m
)
0
= e = a
0
= a
m0
, and so the statement holds with n = 0. If for
some k ≥ 0 there holds that (a
m
)
k
= a
mk
, then we have
(a
m
)
k+1
= (a
m
)
k
∗ a
m
= a
mk
∗ a
m
= a
mk+m
= a
m(k+1)
.
So by induction, we have that (a
m
)
n
= a
mn
for all n ≥ 0. But the choice of m ∈ Z was
arbitrary, and so it follows that for all m ∈ Z and all n ≥ 0, (a
m
)
n
= a
mn
.
If n < 0, then we have (a
m
)
n
=
_
(a
m
)
(−n)
_
−1
= (a
m(−n)
)
−1
= (a
−mn
)
−1
= a
mn
, where
the last equality follows from the following fact: for all k ∈ Z, (a
−k
)
−1
= a
k
(Indeed, for
k < 0, we know that a
k
= (a
−k
)
−1
, by the deﬁnition of a
k
for a negative k. If k ≥ 0,
then −k ≥ 0, and so a
−k
= (a
−(−k)
)
−1
= (a
k
)
−1
. Hence by taking inverses, we obtain
(a
−k
)
−1
= ((a
k
)
−1
)
−1
= a
k
.)
3. We have
_
1 1
−1 0
_
1
=
_
1 1
−1 0
_
,
_
1 1
−1 0
_
2
=
_
1 1
−1 0
_
1
_
1 1
−1 0
_
=
_
1 1
−1 0
_ _
1 1
−1 0
_
=
_
0 1
−1 −1
_
,
_
1 1
−1 0
_
3
=
_
1 1
−1 0
_
2
_
1 1
−1 0
_
=
_
0 1
−1 −1
_ _
1 1
−1 0
_
=
_
−1 0
0 −1
_
,
_
1 1
−1 0
_
4
=
_
1 1
−1 0
_
3
_
1 1
−1 0
_
=
_
−1 0
0 −1
_ _
1 1
−1 0
_
=
_
−1 −1
1 0
_
,
_
1 1
−1 0
_
5
=
_
1 1
−1 0
_
4
_
1 1
−1 0
_
=
_
−1 −1
1 0
_ _
1 1
−1 0
_
=
_
0 −1
1 1
_
,
_
1 1
−1 0
_
6
=
_
1 1
−1 0
_
5
_
1 1
−1 0
_
=
_
0 −1
1 1
_ _
1 1
−1 0
_
=
_
1 0
0 1
_
.
108
Thus
ord
__
1 1
−1 0
__
= min
_
m ∈ N
¸
¸
¸
¸
_
1 1
−1 0
_
m
=
_
1 0
0 1
__
= 6.
4. If a ∗ b has ﬁnite order, say m, then (a ∗ b)
m
= e, that is,
(a ∗ b) ∗ ∗ (a ∗ b)
. ¸¸ .
m times
= e,
and so
a ∗ (b ∗ a) ∗ ∗ (b ∗ a)
. ¸¸ .
(m−1) times
∗b = e,
that is,
a ∗ (b ∗ a)
m−1
∗ b = e. (2.33)
Hence by premultiplying (2.33) by b, and postmultiplying by b
−1
, we obtain
(b ∗ a) ∗ (b ∗ a)
m−1
= e,
that is, (b∗a)
m
= e. So we see that b∗a has ﬁnite order, and ord(b∗a) ≤ m. By interchanging
the roles of a and b, we obtain that m ≤ ord(b ∗ a), and so ord(b ∗ a) = m.
So we have shown that
if a ∗ b has ﬁnite order m, then b ∗ a has ﬁnite order m.
By interchanging a and b, we obtain
a ∗ b has ﬁnite order m iﬀ b ∗ a has ﬁnite order m.
Hence if a ∗ b has inﬁnite order, then b ∗ a has inﬁnite order as well.
So the orders of a ∗ b and b ∗ a are the same.
5. Yes, since ¸1) = Z.
1 is a generator of Z with addition.
As ¸−1) = Z, we see that −1 is also a generator of Z with addition. So the cyclic group Z
with addition does not have a unique generator.
Solutions to the exercises on page 51
1. (a) For all a, b in G, we have
(ψ ◦ ϕ)(a ∗ b) = ψ(ϕ(a ∗ b)) (deﬁnition of ψ ◦ ϕ)
= ψ(ϕ(a) ∗
ϕ(b)) (since ϕ is a homomorphism)
= ψ(ϕ(a)) ∗
ψ(ϕ(b)) (since ψ is a homomorphism)
= (ψ ◦ ϕ)(a) ∗
(ψ ◦ ϕ)(b) (deﬁnition of ψ ◦ ϕ),
and so ψ ◦ ϕ : G → G
is a homomorphism.
(b) We have
ker(ψ ◦ ϕ) = ¦a ∈ G [ (ψ ◦ ϕ)(a) = e
¦
= ¦a ∈ G [ ψ(ϕ(a)) = e
¦
= ¦a ∈ G [ ϕ(a) ∈ ker(ψ)¦
= ϕ
−1
(ker(ψ)).
109
2. If a ∈ ker(ϕ), then ϕ(a) = e
. Thus for all b ∈ G,
ϕ(b∗a∗b
−1
) = ϕ(b)∗
ϕ(a)∗
ϕ(b
−1
) = ϕ(b)∗
e
∗
ϕ(b
−1
) = ϕ(b)∗
ϕ(b
−1
) = ϕ(b)∗
(ϕ(b))
−1
= e
,
and so b ∗ a ∗ b
−1
∈ ker(ϕ). Since the choice of a ∈ ker(ϕ) was arbitrary, it follows that
∀a ∈ ker(ϕ) and ∀b ∈ G, b ∗ a ∗ b
−1
∈ ker(ϕ),
and so ker(ϕ) is a normal subgroup of G.
3. For all a, b ∈ G, we have
ϕ(a ∗ b) = (a ∗ b)
−1
(deﬁnition of ϕ)
= b
−1
∗ a
−1
(Exercise 4c from the exercises on page 44)
= a
−1
∗ b
−1
(since G is abelian)
= ϕ(a) ∗ ϕ(b) (deﬁnition of ϕ),
and so ϕ : G → G is a homomorphism.
Let a, b ∈ G and a ,= b. Then a
−1
,= b
−1
(for otherwise, a = (a
−1
)
−1
= (b
−1
)
−1
= b). Hence
ϕ(a) = a
−1
,= b
−1
= ϕ(b), and so ϕ : G → G is onetoone. For all a ∈ G, we have a
−1
∈ G
and ϕ(a
−1
) = (a
−1
)
−1
= a, and so ϕ : G → G is onto. As ϕ : G → G is onetoone and
onto, it is a bijection.
Consequently ϕ : G → G is an isomorphism.
4. First we show that ϕ
−1
: G
→ G is a homomorphism.
Let a
, b
∈ G
. Since ϕ : G → G
is a bijection, it follows that there exist unique a, b ∈ G
such that ϕ(a) = a
and ϕ(b) = b
. We have
ϕ(a ∗ b) = ϕ(a) ∗
ϕ(b) = a
∗
b
,
and so ϕ
−1
(a
∗
b
) = a ∗ b = ϕ
−1
(a
) ∗ ϕ
−1
(b
). So ϕ
−1
is a homomorphism.
If a
,= b
, then clearly ϕ
−1
(a
) ,= ϕ
−1
(b
) (for otherwise a
= ϕ(ϕ
−1
(a
)) = ϕ(ϕ
−1
(b
)) = b
).
Thus ϕ
−1
: G
→ G is onetoone. For all a ∈ G, we have ϕ
−1
(ϕ(a)) = a, and so ϕ
−1
: G
→
G is onto. Hence ϕ
−1
: G
→ G is a bijection.
Consequently ϕ
−1
: G
→ G is an isomorphism.
5. Let us denote the homomorphism m → a
m
from Z to ¸a) by ϕ.
(a) For all m, n ∈ Z, we have ϕ(m+n) = a
m+n
= a
m
∗a
n
= ϕ(m)∗ϕ(n), and so ϕ : Z → ¸a)
is a homomorphism.
(b) If m ,= n, then ϕ(m) = a
m
,= a
n
= ϕ(n) (for otherwise, if m > n, then a
m−n
= e, and
if m < n, then a
n−m
= e; in either case we get a contradiction to the fact that a has
inﬁnite order). Hence ϕ is onetoone.
Furthermore, if b ∈ ¸a), then b = a
m
for some m ∈ Z. We have ϕ(m) = a
m
= b, and so
ϕ is onto.
So ϕ is a bijection and consequently, ϕ : Z → ¸a) is an isomorphism.
110
Solutions to the exercises on page 54
1. (a) FALSE.
For example, consider the group Z of integers with addition, and let H be the subgroup
of even integers. Then 0 ,= 2, but 0 +H = 2 +H (both these cosets are equal to H).
(b) TRUE.
Consider the function f : H → a ∗ H given by h → a ∗ h, for all h ∈ H. Then it
can be seen that f is a bijection as follows. If a ∈ G and h
1
, h
2
∈ H are such that
a ∗ h
1
= a ∗ h
2
, then by premultiplication with a
−1
, we obtain h
1
= h
2
. Thus f is
onetoone. Furthermore, if b ∈ a ∗ H, then there exists a h ∈ H such that b = a ∗ h.
Consequently f(h) = a ∗ h = b. Hence f is onto. So f is a bijection, and it follows that
the cardinalities of H and a ∗ H are the same.
(c) TRUE.
As in part 1b above, one can show that the function g : H → H∗a, given by h → h∗a,
for all h ∈ H, is a bijection. Thus a ∗ H, H and H ∗ a all have the same cardinalities.
2. (a) Matrix multiplication is a law of composition on G:
if A =
_
x y
0 1
_
and B =
_
x
y
0 1
_
belong to G, then x > 0 and x
> 0.
We have
AB =
_
x y
0 1
_ _
x
y
0 1
_
=
_
xx
xy
+y
0 1
_
and xx
> 0,
and so AB ∈ G. Moreover the group axioms are satisﬁed:
G1. As matrix multiplication is associative, it follows that in particular for elements
A, B, C from G, there holds (AB)C = A(BC).
G2. The identity matrix
I =
_
1 0
0 1
_
∈ G (x := 1 > 0, y := 0),
and furthermore, for all A ∈ G, clearly we have AI = A = IA.
G3. If
_
x y
0 1
_
∈ G,
then x > 0. Thus
1
x
> 0, and
_
1
x
−
y
x
0 1
_
∈ G.
Moreover,
_
x y
0 1
_ _
1
x
−
y
x
0 1
_
=
_
1 0
0 1
_
=
_
1
x
−
y
x
0 1
_ _
x y
0 1
_
.
So G is a group with matrix multiplication.
(b) Clearly H ⊂ G. We now check that H1, H2, H3 are satisﬁed:
H1. If
_
x 0
0 1
_
,
_
x
0
0 1
_
∈ H,
111
then x > 0 and x
> 0. We have
_
x 0
0 1
_ _
x
0
0 1
_
=
_
xx
0
0 1
_
and xx
> 0. So
_
x 0
0 1
_ _
x
0
0 1
_
∈ H.
H2. Clearly the identity element from G, namely the identity matrix
I =
_
1 0
0 1
_
,
also belongs to H.
H3. If
_
x 0
0 1
_
∈ H,
then x > 0. So
1
x
> 0, and the inverse of
_
x 0
0 1
_
in the group G, namely the matrix
_
1
x
0
0 1
_
,
also belongs to H.
Thus H is a subgroup of the group G.
(c) Let
a =
_
x
0
y
0
0 1
_
∈ G.
Then the corresponding left coset of H is
aH =
__
x
0
y
0
0 1
_ _
x 0
0 1
_ ¸
¸
¸
¸
x ∈ R and x > 0
_
=
__
xx
0
y
0
0 1
_ ¸
¸
¸
¸
x ∈ R and x > 0
_
.
Thus the left coset aH is a straight line parallel to the xaxis, and it passes through
the point (x
0
, y
0
). See Figure 2.7. The right coset of H corresponding to the element
a =
_
x
0
y
0
0 1
_
∈ G
is
Ha =
__
x 0
0 1
_ _
x
0
y
0
0 1
_ ¸
¸
¸
¸
x ∈ R and x > 0
_
=
__
xx
0
xy
0
0 1
_ ¸
¸
¸
¸
x ∈ R and x > 0
_
.
Thus the right coset Ha is a straight line passing through the origin and the point
(x
0
, y
0
). See Figure 2.8.
112
0
PSfrag replacements
a
x
y
aH
H
Figure 2.7: Partition of the group G (represented by the right half plane: x > 0, y ∈ R) by left
cosets aH, a ∈ G (represented by lines parallel to the xaxis).
0
PSfrag replacements
a
x
y
aH
H
Figure 2.8: Partition of the group G by right cosets Ha, a ∈ G.
3. H ∩ K is a subset of the group H. Furthermore, the following hold:
H1. If a, b ∈ H∩K, then a, b ∈ H and a, b ∈ K. As H and K are subgroups of G, it follows
that a ∗ b ∈ H and a ∗ b ∈ K. Hence a ∗ b ∈ H ∩ K.
H2. The identity element in the group H is the identity element e from G. As K is a
subgroup of G, it follows that e also belongs to K. Thus e ∈ H ∩ K.
H3. If a ∈ H ∩ K, then a ∈ H and a ∈ K. As H and K are subgroups, a
−1
∈ H and
a
−1
∈ K. Consequently a
−1
∈ H ∩ K.
So H ∩ K is a subgroup of H.
By interchanging the roles of H and K, it follows that H ∩ K is also a subgroup of K.
Thus by Lagrange’s theorem, we have that
[H ∩ K[ divides [H[ = 3, and [H ∩ K[ divides [K[ = 5.
As gcd(3, 5) = 1, it follows that [H ∩ K[ divides 1. So [H ∩ K[ = 1, and so it comprises just
one element. As e ∈ H ∩ K, it follows that H ∩ K = ¦e¦.
4. The cardinality of the group S
4
is 4! = 24. As 16 does not divide 24, it follows from Corollary
2.1.8 that S
4
cannot have an element of order 16.
5. (a) If p does not divide a, then [a] ,= [0], and so [a] ∈ Z
∗
p
. The set Z
∗
p
= ¦[1], . . . , [p − 1]¦
has cardinality equal to p − 1. As [1] is the identity element in the group Z
∗
p
with
multiplication modulo p, from Corollary 2.1.8 it follows that [a]
p−1
= [1].
113
(b) If p divides a, then p also divides a
p
, and hence it divides a
p
− a as well, that is,
a
p
≡ a(mod p).
If p does not divide a, then [a] ,= [0], and so by part 5a above, [a]
p−1
= [1]. Hence
[a
p−1
] = [1], and so [a
p−1
−1] = [0]. Thus p divides a
p−1
−1, and consequently, it also
divides a(a
p−1
−1) = a
p
−a. So a
p
≡ a(mod p).
(c) Note that 2222 = 317 7 + 3, so that in Z
7
,
[2222
5555
] = [2222]
5555
= [3]
5555
.
As [3] ,= [0] in Z
7
, from part (5b) above, it follows that [3]
7−1
= [3]
6
= [1]. Thus we
obtain
[2222
5555
] = [3]
5555
= [3]
925·6+5
= ([3]
6
)
925
[3]
5
= [1]
925
[3]
5
= [1][3]
5
= [3]
5
.
Proceeding in a similar manner, we can show that [5555
2222
] = [3
2
]. Indeed, 5555 =
793 7 + 4, so that
[5555
2222
] = [5555]
2222
= [4]
2222
.
Again, [4] ,= [0] in Z
7
, and so by part (5b), [4]
7−1
= [4]
6
= [1]. Hence
[5555
2222
] = [4]
2222
= [4]
370·6+2
] = ([4]
6
)
370
[4]
2
= [1][4]
2
= [4]
2
= [−3]
2
= [(−3)
2
] = [3
2
].
Finally,
[2222
5555
+ 5555
2222
] = [3
5
+ 3
2
] = [3
2
28] = [0],
which proves that 7[2222
5555
+ 5555
2222
.
114
Solutions to the exercises on page 58
1. (a) NO.
For instance the sum of two invertible matrices may not be invertible. The identity
matrix I is invertible, and so is its additive inverse, −I. However, I + (−I) = 0, which
is not invertible. So matrix addition, (A, B) → A + B is not a law of composition on
the set of invertible matrices, and so it does not form an abelian group. Consequently
it is not a vector space.
Alternately, we can observe that the scalar multiplication of the real number 0 with an
invertible matrix is the zero matrix, which is not invertible.
(b) NO.
The scalar multiplication of the real number 0 with an invertible matrix is the zero
matrix, which is not invertible.
2. Let α ∈ R and v ∈ V be such that
α v = 0. (2.34)
Then if α ,= 0, then we obtain
0 = α
−1
0 (Theorem 2.2.1)
= α
−1
(α v) (using (2.34))
= (α
−1
α) v (using V2)
= (1) v (since α
−1
α = 1)
= v (using V1).
This proves the claim.
3. First we check that R
∞
, with addition deﬁned by (2.4), is an abelian group. Clearly (2.4)
gives a law of composition on R
∞
. Moreover, we have:
G1. For all (a
n
)
n∈N
, (b
n
)
n∈N
, (c
n
)
n∈N
in R
∞
, we have
((a
n
)
n∈N
+ (b
n
)
n∈N
) + (c
n
)
n∈N
= (a
n
+b
n
)
n∈N
+ (c
n
)
n∈N
= ((a
n
+b
n
) +c
n
)
n∈N
= (a
n
+ (b
n
+c
n
))
n∈N
= (a
n
)
n∈N
+ (b
n
+c
n
)
n∈N
= (a
n
)
n∈N
+ ((b
n
)
n∈N
+ (c
n
)
n∈N
).
G2. The sequence (0)
n∈N
serves as the identity element: for all (a
n
)
n∈N
∈ R
∞
, we have
(a
n
)
n∈N
+ (0)
n∈N
= (a
n
+ 0)
n∈N
= (a
n
)
n∈N
= (0 +a
n
)
n∈N
= (0)
n∈N
+ (a
n
)
n∈N
.
G3. If (a
n
)
n∈N
∈ R
∞
, then (−a
n
)
n∈N
∈ R
∞
, and furthermore,
(a
n
)
n∈N
+(−a
n
)
n∈N
= (a
n
+−a
n
)
n∈N
= (0)
n∈N
= (−a
n
+a
n
)
n∈N
= (−a
n
)
n∈N
+(a
n
)
n∈N
.
G4. Finally, commutativity holds, since for all (a
n
)
n∈N
, (b
n
)
n∈N
∈ R
∞
, we have
(a
n
)
n∈N
+ (b
n
)
n∈N
= (a
n
+b
n
)
n∈N
= (b
n
+a
n
)
n∈N
= (b
n
)
n∈N
+ (a
n
)
n∈N
.
Next we check that V1, V2, V3, V4 are also satisﬁed:
V1. For all (a
n
)
n∈N
∈ R
∞
, 1 (a
n
)
n∈N
= (1a
n
)
n∈N
= (a
n
)
n∈N
.
115
V2. For all α, β ∈ R and all (a
n
)
n∈N
∈ R
∞
,
α (β (a
n
)
n∈N
) = α (βa
n
)
n∈N
= (α(βa
n
))
n∈N
= ((αβ)a
n
)
n∈N
= (αβ) (a
n
)
n∈N
.
V3. For all α, β ∈ R and all (a
n
)
n∈N
∈ R
∞
,
(α +β) (a
n
)
n∈N
= ((α +β)a
n
)
n∈N
= (αa
n
+βa
n
)
n∈N
= (αa
n
)
n∈N
+ (βa
n
)
n∈N
= α (a
n
)
n∈N
+β (a
n
)
n∈N
.
V4. For all α ∈ R and all (a
n
)
n∈N
, (b
n
)
n∈N
∈ R
∞
,
α ((a
n
)
n∈N
+ (b
n
)
n∈N
) = α (a
n
+b
n
)
n∈N
= (α(a
n
+b
n
))
n∈N
= (αa
n
+αb
n
)
n∈N
= (αa
n
)
n∈N
+ (αb
n
)
n∈N
= α (a
n
)
n∈N
+α (b
n
)
n∈N
.
So R
∞
is a vector space with addition deﬁned by (2.4) and scalar multiplication deﬁned by
(2.5).
Solutions to the exercises on page 62
1. (a) FALSE.
Consider the subspaces
U
1
= span
__
1
0
__
and U
2
= span
__
0
1
__
of R
2
. Then
v
1
:=
_
1
0
_
∈ U
1
⊂ U
1
∪ U
2
and v
2
:=
_
0
1
_
∈ U
2
⊂ U
1
∪ U
2
,
but
v
1
+v
2
=
_
1
1
_
,∈ U
1
∪ U
2
,
since
_
1
1
_
,∈ span
__
1
0
__
and
_
1
1
_
,∈ span
__
0
1
__
.
Thus S2 is not satisﬁed, and so U
1
∪ U
2
is not a vector space.
(Each vector
_
x
y
_
in R
2
can be represented by a point in the (x, y)plane, and the
above can be seen pictorially in Figure 2.9).
(b) TRUE.
Let U
1
, U
2
be susbpaces of the vector space V . We check that S1, S2, S3 hold for
U
1
∩ U
2
:
S1. As U
1
, U
2
are susbpaces, 0 ∈ U
1
and 0 ∈ U
2
. Thus 0 ∈ U
1
∩ U
2
.
S2. If v
1
, v
2
belong to U
1
∩U
2
, then v
1
, v
2
belong to U
1
and v
1
, v
2
belong to U
2
. As U
1
is a subspace, v
1
+ v
2
∈ U
1
. Moreover, as U
2
is a subspace, v
1
+ v
2
∈ U
2
. Hence
v
1
+v
2
∈ U
1
∩ U
2
.
S3. Let α ∈ R and v ∈ U
1
∩U
2
. Thus v ∈ U
1
and v ∈ U
2
. As U
1
is a subspace, it follows
that α v ∈ U
1
. Also, as U
2
is a subspace, it follows that α v ∈ U
2
. Consequently
α v ∈ U
1
∩ U
2
.
So U
1
∩ U
2
is a susbpaces of the vector space V .
116
PSfrag replacements
U
1
U
2
x
y
R
2
v
1
v
2
v
1
+v
2
Figure 2.9: The union U
1
∪ U
2
of subspaces U
1
and U
2
is not a subspace.
(c) TRUE.
Indeed,
_
_
1
2
3
_
_
= 1
_
_
1
3
4
_
_
+ (−1)
_
_
0
1
1
_
_
.
(d) FALSE.
We have
(3)
_
_
1
0
1
_
_
+ (−3)
_
_
1
2
1
_
_
+ (2)
_
_
0
3
0
_
_
=
_
_
0
0
0
_
_
.
(e) TRUE.
Suppose that there exist scalars such that α
1
v
1
+α
2
v
2
+α
3
v
3
= 0. Then we have
α
1
v
1
+α
2
v
2
+α
3
v
3
+0 v
4
= α
1
v
1
+α
2
v
2
+α
3
v
3
+0 = α
1
v
1
+α
2
v
2
+α
3
v
3
= 0,
and by the independence of v
1
, v
2
, v
3
, v
4
, it follows that α
1
= α
2
= α
3
= 0. So v
1
, v
2
, v
3
are linearly independent.
(f) FALSE.
For instance, in R
3
, the vectors
v
1
:=
_
_
1
0
0
_
_
, v
2
:=
_
_
0
1
0
_
_
, v
3
:=
_
_
0
0
1
_
_
, v
4
:=
_
_
1
1
1
_
_
,
are linearly dependent, since 1 v
1
+ 1 v
2
+ 1 v
3
+ (−1) v
4
= 0, while the vectors
v
1
, v
2
, v
3
are linearly independent.
2. (a) Clearly,
span
___
1
t
1
_
,
_
1
t
2
___
⊂ R
2
.
(In order to prove the reverse inclusion, observe that if
_
x
y
_
= α
_
1
t
1
_
+β
_
1
t
2
_
,
then we obtain the linear equations
α +β = x,
αt
1
+βt
2
= y,
117
which have the solution α =
y−xt2
t1−t2
and β =
xt1−y
t1−t2
.) We now complete the proof as
follows: suppose that
_
x
y
_
∈ R
2
.
As t
1
−t
2
,= 0, we have
_
x
y
_
=
y −xt
2
t
1
−t
2
_
1
t
1
_
+
xt
1
−y
t
1
−t
2
_
1
t
2
_
∈ span
___
1
t
1
_
,
_
1
t
2
___
.
So
R
2
⊂ span
___
1
t
1
_
,
_
1
t
2
___
.
Hence the claim follows.
(b) Suppose that U
1
, . . . , U
n
are subspaces of R
2
, such that U
1
∪ ∪ U
n
= R
2
. Since the
inﬁnite set
S :=
__
1
t
_ ¸
¸
¸
¸
t ∈ R
_
is contained in R
2
= U
1
∪ ∪ U
n
, it follows that inﬁnitely many elements from S
belong to one of the subspaces, say U
k
for some k ∈ ¦1, . . . , n¦. Thus there exist real
numbers t
1
, t
2
such that
_
1
t
1
_
,
_
1
t
2
_
∈ U
k
.
As U
k
is a subspace of R
2
, it follows that
span
___
1
t
1
_
,
_
1
t
2
___
⊂ U
k
.
From the result of the part above, we obtain R
2
⊂ U
k
, and so U
k
= R
2
, that is U
k
is
not a proper subspace.
3.
∞
is a subspace of R
∞
. We have:
S1. The sequence (0)
n∈N
is the zero vector in R
∞
, and it belongs to
∞
, since it is bounded:
indeed for all n ∈ N, the nth term of the sequence (0)
n∈N
is equal to 0, and [0[ = 0 ≤ 1.
S2. If (a
n
)
n∈N
, (b
n
)
n∈N
belong to
∞
, then there exist M
1
, M
2
> 0 such that for all n ∈ N,
[a
n
[ ≤ M
1
and [a
n
[ ≤ M
2
. Hence for all n ∈ N, [a
n
+b
n
[ ≤ [a
n
[ +[b
n
[ ≤ M
1
+M
2
, and
so the sequence (a
n
+b
n
)
n∈N
is bounded, that is, (a
n
)
n∈N
+ (b
n
)
n∈N
belongs to
∞
.
S3. Let α ∈ R and (a
n
)
n∈N
∈
∞
. Then there exists an M > 0 such that for all n ∈ N,
[a
n
[ ≤ M. Thus for all n ∈ N, [αa
n
[ = [α[[a
n
[ ≤ ([α[ + 1)M, and so the sequence
(αa
n
)
n∈N
is bounded, that is, α (a
n
)
n∈N
belongs to
∞
.
So
∞
is a subspace of R
∞
.
c is a subspace of
∞
. Indeed, there holds:
S1. The sequence (0)
n∈N
is the zero vector in
∞
, and it belongs to c, since it is convergent.
S2. If (a
n
)
n∈N
, (b
n
)
n∈N
belong to c, then by Theorem 1.2.4, it follows that the sequence
(a
n
+b
n
)
n∈N
is also convergent, and so (a
n
)
n∈N
+ (b
n
)
n∈N
belongs to c.
S3. Let α ∈ R and (a
n
)
n∈N
∈ c. Then by Theorem 1.2.4, the sequence (αa
n
)
n∈N
is also
convergent, that is, α (a
n
)
n∈N
belongs to c.
Thus c is a subspace of
∞
.
c
0
is a subspace of c. We have:
118
S1. The sequence (0)
n∈N
is the zero vector in c, and it belongs to c
0
, since it is a convergent
sequence with limit equal to 0.
S2. If (a
n
)
n∈N
, (b
n
)
n∈N
belong to c
0
, then by Theorem 1.2.4, it follows that the sequence
(a
n
+b
n
)
n∈N
is also convergent and
lim
n→∞
(a
n
+b
n
) = lim
n→∞
a
n
+ lim
n→∞
b
n
= 0 + 0 = 0,
and so (a
n
)
n∈N
+ (b
n
)
n∈N
belongs to c
0
.
S3. Let α ∈ R and (a
n
)
n∈N
∈ c
0
. Then by Theorem 1.2.4, the sequence (αa
n
)
n∈N
is also
convergent and
lim
n→∞
αa
n
= α lim
n→∞
a
n
= α0 = 0,
that is, α (a
n
)
n∈N
belongs to c
0
.
Hence c
0
is a subspace of c.
c
00
is a subspace of c
0
. Indeed, there holds:
S1. The sequence (0)
n∈N
is the zero vector in c
0
, and it belongs to c
00
, since for all n > 1,
a
n
= 0.
S2. If (a
n
)
n∈N
, (b
n
)
n∈N
belong to c
00
, then there exist N
1
, N
2
∈ N such that
∀n > N
1
, a
n
= 0 and ∀n > N
2
, b
n
= 0.
Thus with N := max¦N
1
, N
2
¦ ∈ N, for all n > N, we obtain a
n
+ b
n
= 0 + 0 = 0.
Hence (a
n
)
n∈N
+ (b
n
)
n∈N
= (a
n
+b
n
)
n∈N
belongs to c
00
.
S3. Let α ∈ R and (a
n
)
n∈N
∈ c
00
. Then there exists an N ∈ N such that for all n > N,
a
n
= 0, and so αa
n
= α0 = 0. Thus (αa
n
)
n∈N
= α (a
n
)
n∈N
belongs to c
00
.
Hence c
00
is a subspace of c
0
.
(The examples
_
1
n
_
n∈N
∈ c
0
, (1)
n∈N
∈ c, ((−1)
n
)
n∈N
∈
∞
, (n)
n∈N
∈ R
∞
,
show that each of the inclusions are strict.)
4. If: Let y
1
= 0 = y
2
. Then we have:
S1. The zero function 0 ∈ S(0, 0), since 0 is continuous on [0, 1] and moreover, 0(0) = 0
and 0(1) = 0.
S2. If f, g ∈ S(0, 0), then f +g ∈ S(0, 0). Indeed, as f, g are continuous on [0, 1], so is f +g,
and also (f +g)(0) = f(0) +g(0) = 0 +0 = 0 and (f +g)(1) = f(1) +g(1) = 0 +0 = 0.
S3. Let f ∈ S(0, 0) and α ∈ R. Then α f is continuous on [0, 1], and furthermore,
(α f)(0) = αf(0) = α0 = 0 and (α f)(1) = αf(1) = α0 = 0.
Hence S(0, 0) is a subspace of C[0, 1].
Only if: Suppose that S(y
1
, y
2
) is a subspace of C[0, 1]. If f ∈ S(y
1
, y
2
), then 2 f ∈
S(y
1
, y
2
). Thus (2f)(0) = y
1
and (2f)(1) = y
2
, and so 2y
1
= y
1
and 2y
2
= y
2
. Consequently
y
1
= 0 = y
2
.
119
Solutions to the exercises on page 64
1. We have:
B1. If
_
_
x
y
z
_
_
∈ R
3
,
then
_
_
x
y
z
_
_
= (x−y)
_
_
1
0
0
_
_
+(y−z)
_
_
1
1
0
_
_
+z
_
_
1
1
1
_
_
∈ span
_
_
_
_
_
_
_
1
0
0
_
_
,
_
_
1
1
0
_
_
,
_
_
1
1
1
_
_
_
_
_
_
_
.
Thus
span
_
_
_
_
_
_
_
1
0
0
_
_
,
_
_
1
1
0
_
_
,
_
_
1
1
1
_
_
_
_
_
_
_
= R
3
.
B2. If α, β, γ are scalars such that
α
_
_
1
0
0
_
_
+β
_
_
1
1
0
_
_
+γ
_
_
1
1
1
_
_
=
_
_
0
0
0
_
_
,
then we obtain
α +β +γ = 0,
β +γ = 0,
γ = 0,
and so it follows that α = β = γ = 0. So B is linearly independent.
Since B1, B2 hold it follows that B is a basis.
2. Suppose that B = ¦v
1
, . . . , v
n
¦ is a basis of the vector space V , and let v ∈ V . As v ∈ V =
span(B), it follows that there exist scalars α
1
, . . . , α
n
such that v = α
1
v
1
+ + α
n
v
n
.
So v is a linear combination of vectors from V . In order to prove uniquness, suppose that
there exist scalars α
1
, . . . , α
n
such that
α
1
v
1
+ +α
n
v
n
= v = α
1
v
1
+ +α
n
v
n
.
So we obtain (α
1
−α
1
) v
1
+ +(α
n
−α
n
) v
n
= 0, and by the independence of v
1
, . . . , v
n
,
it then follows that α
1
−α
1
= = α
n
−α
n
= 0, that is α
1
= α
1
, . . . , α
n
= α
n
.
3. We prove this by contradiction. Suppose that C[0, 1] is a ﬁnite dimensional vector space
with dimension d, say.
Consider the d functions f
1
, . . . , f
d
deﬁned by
f
1
(x) = x
f
2
(x) = x
2
f
3
(x) = x
3
.
.
.
f
d
(x) = x
d
_
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
_
for all x ∈ [0, 1]. (2.35)
The functions f
1
, . . . , f
d
are all polynomials and so they are all continuous on [0, 1], that is,
f
1
, . . . , f
d
∈ C[0, 1]. Now we prove the following.
120
Claim: The functions f
1
, . . . , f
d
given by (2.35) are linearly independent in C[0, 1].
Proof Suppose that there exist scalars α
1
, . . . , α
d
, not all zeros, such that
α
1
f
1
+ +α
d
f
d
= 0.
Then for all x ∈ [0, 1], (α
1
f
1
+ +α
d
f
d
)(x) = 0(x), that is,
for all x ∈ [0, 1], α
1
x + +α
d
x
d
= 0.
Let k ∈ ¦1, 2, . . . , d¦ be the smallest number such that α
k
,= 0. Then we obtain
for all x ∈ [0, 1], α
k
x
k
+α
k+1
x
k+1
+ +α
d
x
d
= 0,
and so for all x ∈ [0, 1], x
k
(α
k
+ α
k+1
x + α
k+2
x
2
+ + α
d
x
d−k
) = 0. For all n ∈ N,
0 <
1
n
< 1, and so we have
for all n ∈ N,
_
1
n
_
k
_
α
k
+α
k+1
_
1
n
_
+α
k+2
_
1
n
_
2
+ +α
d
_
1
n
_
d−k
_
= 0,
that is,
for all n ∈ N, α
k
+α
k+1
_
1
n
_
+α
k+2
_
1
n
_
2
+ +α
d
_
1
n
_
d−k
= 0.
Hence
0 = lim
n→∞
0
= lim
n→∞
_
α
k
+α
k+1
_
1
n
_
+ α
k+2
_
1
n
_
2
+ +α
d
_
1
n
_
d−k
_
= α
k
+ 0 + + 0
= α
k
,
a contradiction. So f
1
, . . . , f
d
are linearly independent in C[0, 1].
Consequently, by Theorem 2.2.3, it follows that ¦f
1
, . . . , f
d
¦ is a basis of C[0, 1], and in
particular, they span the whole space C[0, 1].
The constant function 1 : [0, 1] →R, deﬁned by 1(x) = 1 for all x ∈ [0, 1] clearly belongs to
C[0, 1], and so, from the above, there must exist scalars β
1
, . . . , β
d
such that
1 = β
1
f
1
+ +β
d
f
d
.
Thus for all x ∈ [0, 1], 1(x) = (β
1
f
1
+ +β
d
f
d
)(x), that is,
for all x ∈ [0, 1], 1 = β
1
x + +β
d
x
d
.
In particular, with x = 0, we obtain 1 = 0, a contradiction. So C[0, 1] is not a ﬁnite
dimensional vector space.
4. (a) We have
B1. Let (a
n
)
n∈N
∈ c
00
. Then there exists an N ∈ N such that for all n > N, a
n
= 0.
Clearly
(a
n
)
n∈N
= a
1
e
1
+ +a
n
e
n
∈ span(¦e
k
[ k ∈ N¦).
So c
00
⊂ span(B).
121
On the other hand, suppose that k
1
, . . . , k
m
∈ N, and α
1
, . . . , α
m
be scalars. If
N := max¦k
1
, . . . , k
m
¦, then for all n > N, the nth term of the sequence
α
1
e
k1
+ +α
m
e
km
is α
1
0 + + α
m
0 = 0. Thus α
1
e
k1
+ + α
m
e
km
∈ c
00
. Consequently
span(B) ⊂ c
00
.
So span(B) = c
00
.
B2. Suppose that k
1
< < k
m
are natural numbers, and α
1
, . . . , α
m
be scalars, not
all zeros, such that
α
1
e
k1
+ +α
m
e
km
= (0)
n∈N
.
Let l ∈ ¦1, . . . , m¦ be the largest number such that α
l
,= 0. Then equating the
k
l
th term on both sides, we obtain that α
l
= 0, a contradiction. So B is linearly
independent.
(b) Indeed, if B were a basis for R
∞
, then the sequence (1)
n∈N
would be a linear combination
of elements from B. But as seen above, any linear combination of elements from B is
in c
00
, that is, it is a sequence that is eventually zero. Clearly (1)
n∈N
,∈ c
00
, and so it
is not a linear combination of elements from c
00
.
Solutions to the exercises on page 68
1. (a) We have
ker(T
A
) =
__
x
y
_ ¸
¸
¸
¸
T
A
_
x
y
_
=
_
0
0
__
=
__
x
y
_ ¸
¸
¸
¸
_
x +y
y
_
=
_
0
0
__
=
__
x
y
_ ¸
¸
¸
¸
x = y = 0
_
=
__
0
0
__
.
We depict this set in the (x, y)plane in Figure 2.10.
PSfrag replacements
R
2
x
y
ker(T
A
)
Figure 2.10: Kernel of T
A
.
Clearly im(T
A
) ⊂ R
2
. On the other hand, if
_
x
y
_
∈ R
2
, then
_
x
y
_
= T
A
_
x −y
y
_
,
and so R
2
⊂ im(T
A
). Hence im(T
A
) = R
2
. We depict this set in the (x, y)plane in
Figure 2.11.
122
PSfrag replacements
im(T
A
) = R
2
x
y
Figure 2.11: Image of T
A
.
(b) We have
ker(T
A
) =
__
x
y
_ ¸
¸
¸
¸
T
A
_
x
y
_
=
_
0
0
__
=
__
x
y
_ ¸
¸
¸
¸
_
x +y
0
_
=
_
0
0
__
=
__
x
y
_ ¸
¸
¸
¸
x +y = 0
_
.
We depict this set in the (x, y)plane in Figure 2.12.
PSfrag replacements
R
2
x
y
ker(T
A
)
Figure 2.12: Kernel of T
A
.
We have
im(T
A
) ⊂ S :=
__
x
0
_
∈ R
2
¸
¸
¸
¸
x ∈ R
_
,
since
for all
_
x
y
_
∈ R
2
, T
A
_
x
y
_
=
_
x +y
0
_
∈ S.
Conversely, since
_
x
0
_
= T
A
_
x
0
_
,
it follows that S ⊂ im(T
A
). Hence im(T
A
) = S. We depict this set in the (x, y)plane
in Figure 2.13.
123
PSfrag replacements
R
2
x
y
im(T
A
)
Figure 2.13: Image of T
A
.
(c) We have
ker(T
A
) =
__
x
y
_ ¸
¸
¸
¸
T
A
_
x
y
_
=
_
0
0
__
=
__
x
y
_ ¸
¸
¸
¸
x ∈ R and y ∈ R
_
= R
2
.
We depict this set in the (x, y)plane in Figure 2.14.
PSfrag replacements
ker(T
A
) = R
2
x
y
Figure 2.14: Kernel of T
A
.
Clearly
im(T
A
) =
_
0
0
_
.
We depict this set in the (x, y)plane in Figure 2.15.
2. L1 is not satisﬁed, since
T
__
−1
0
_
+
_
0
1
__
= T
__
−1
1
__
=
_
(−1)
2
1
1
_
=
_
1
1
_
,
while
T
_
−1
0
_
+T
_
0
1
_
=
_
−1
0
_
+
_
0
1
_
=
_
−1
1
_
.
124
PSfrag replacements
R
2
x
y
im(T
A
)
Figure 2.15: Image of T
A
.
If α ∈ R ¸ ¦0¦ and
_
x
1
x
2
_
∈ R
2
, then we have
T
_
α
_
x
1
x
2
__
= T
_
αx
1
αx
2
_
=
_
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
_
_
α
2
x
2
1
αx2
αx
2
_
if x
1
x
2
,= 0
_
αx
1
αx
2
_
if x
1
x
2
= 0
_
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
_
=
_
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
_
α
_
x
2
1
x2
x
2
_
if x
1
x
2
,= 0
α
_
x
1
x
2
_
if x
1
x
2
= 0
_
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
_
= α
_
T
_
x
1
x
2
__
.
If α = 0, and
_
x
1
x
2
_
∈ R
2
, we have
T
_
α
_
x
1
x
2
__
= T
_
0
0
_
=
_
0
0
_
= 0
_
T
_
x
1
x
2
__
.
Thus for all α ∈ R, and all
_
x
1
x
2
_
∈ R
2
,
T
_
α
_
x
1
x
2
__
= α
_
T
_
x
1
x
2
__
,
and so L2 holds.
3. (a) We verify that L1 and L2 hold:
125
L1. For all (a
n
)
n∈N
and (b
n
)
n∈N
in c, we have
T((a
n
)
n∈N
+ (b
n
)
n∈N
) = T((a
n
+b
n
)
n∈N
)
= lim
n→∞
(a
n
+b
n
)
= lim
n→∞
a
n
+ lim
n→∞
b
n
(Theorem 1.2.4)
= T((a
n
)
n∈N
) +T((b
n
)
n∈N
).
L2. For all α ∈ R, and all (a
n
)
n∈N
∈ c, we have
T(α (a
n
)
n∈N
) = T((αa
n
)
n∈N
)
= lim
n→∞
αa
n
= α lim
n→∞
a
n
(Theorem 1.2.4)
= α T((a
n
)
n∈N
).
So T is a linear transformation from c to R.
(b) We have
ker(T) = ¦(a
n
)
n∈N
∈ c [ T((a
n
)
n∈N
) = 0¦ = ¦(a
n
)
n∈N
∈ c [ lim
n→∞
a
n
= 0¦,
and so ker(T) is the set comprising all convergent sequences with limit 0, that is, the
subspace c
0
.
(c) Clearly, im(T) ⊂ R. Moreover, if L ∈ R, then the constant sequence L, L, L, . . . is
convergent with limit L. Thus T((L)
n∈N
) = L, and so L ∈ im(T). So R ⊂ im(T).
Consequently, im(T) = R.
126
Bibliography
[1] N.L. Biggs. Discrete Mathematics. Clarendon Press, 2002.
(Chapter 20 on Groups.)
[2] K.G. Binmore. Mathematical Analysis: A Straightforward Approach. Cambridge University
Press, 1982.
[3] V. Bryant. Yet Another Introduction to Analysis. Cambridge University Press, 1990.
[4] P.R. Halmos. Finite Dimensional Vector Spaces. Springer, 1996.
[5] W. Rudin. Principles of Mathematical Analysis. McGrawHill, 3rd Edition, 1976.
127
128 Bibliography
Index
kth power of a function, 31
abelian group, 42
absolute value, 5
absolute value of a function, 31
Archimedean property, 4
associativity, 40
basis of a vector space, 63
BolzanoWeierstrass theorem, 25
bounded above, set, 1
bounded below, set, 1
bounded function, 33, 34
bounded sequence, 14
bounded set, 1
Cauchy sequence, 13
center of a group, 47
commutativity, 42
continuity of a function at a point, 27
continuous function, 27
convergent sequence, 10
counting formula, 53
cyclic, 48
decreasing sequence, 15
dimension of a vector space, 64
distance, 5
distributive laws, 55
divergent sequence, 11
eventually zero sequence, 62
extreme value theorem, 33
Fermat’s little theorem, 54
ﬁnite dimensional vector space, 64
ﬁnite group, 43
ﬁnite order, group element with, 47
fractional part of a real number, 26
general linear group, 41
generator of a group, 48
greatest integer part of a real number, 26
greatest upper bound, 2
group, 40
group axioms, 40
group table, 43
identity element in a group, 40
iﬀ, 7
image of a homomorphism, 50
image of a linear transformation, 66
increasing sequence, 15
induced law of composition, 45
inﬁmum, 2
inﬁnite dimensional of a vector space, 64
inﬁnite group, 43
inﬁnite order, group element with, 47
intermediate value theorem, 34
interval, 5
inverse of an element in a group, 40
isomorphism, 51
kernel of a homomorphism, 50
kernel of a linear transformation, 66
Lagrange’s theorem, 53
law of composition on a set, 39
least upper bound, 2
least upper bound property, 3
linear combination of vectors, 60
linear transformation, 65
linearly dependent vectors, 61
linearly independent vectors, 61
linearly dependent set, 61
linearly independent set, 61
lower bound, 1
maximum, 3
minimum, 3
monotone sequence, 15
normal subgroup, 51
order of a group, 43
order of an element in a group, 47
partition, 52
periodic function, 34
polynomial function, 32
positive deﬁniteness, 6
product of functions, 31
129
130 Index
proper subspace, 59
rational function, 32
right cosets, 53
Sandwich theorem, 21
scalar multiplication, 55
seaview property, 25
sequence, 9
span of a subset of a vector space, 60
special linear group, 50
subgroup, 45
subsequence, 23
subspace of a vector space, 59
sum of functions, 31
supremum, 2
symmetric group, 44
symmetric matrix, 45, 59
symmetry, 6
triangle inequality, 6
upper bound, 1
upper triangular matrix, 46, 59
vector, 55
vector addition, 55
vector space, 55
zero vector, 55
ii
Contents
1 Analysis 1.1 The real numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Sequences and limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2.1 1.2.2 1.2.3 1.2.4 1.2.5 1.2.6 Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Limit of a convergent sequence . . . . . . . . . . . . . . . . . . . . . . . . . Bounded and monotone sequences . . . . . . . . . . . . . . . . . . . . . . . Algebra of limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sandwich theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Subsequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1 8 8 10 14 17 21 23 26 27 30 32 34 39 39 39 45 49 52 55 55 59
1.3 Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.1 1.3.2 1.3.3 1.3.4 2 Algebra 2.1 Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1 2.1.2 2.1.3 2.1.4 Deﬁnition of a Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Subgroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Homomorphisms and isomorphisms . . . . . . . . . . . . . . . . . . . . . . . Cosets and Lagrange’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . Deﬁnition of continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Continuous functions preserve convergent sequences . . . . . . . . . . . . . Extreme value theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Intermediate value theorem . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Vector spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 2.2.2 Deﬁnition of a vector space . . . . . . . . . . . . . . . . . . . . . . . . . . . Subspaces and linear combinations . . . . . . . . . . . . . . . . . . . . . . . iii
. . . . . . . . . . . . Linear transformations .4 Solutions Bibliography Index Contents Basis of a vector space . . . . .3 2. . . . . . . . . . . .2. . . .iv 2. . . . 62 65 69 127 129 . . . . . . . .2. . . . . . . . . . . . . . . . . . .
for example. If the set of all lower bounds of S is not empty. then S is said to be bounded below. and any real number z satisfying z ≤ 0 (for instance 0. Thus the set S = {q ∈ R  q 2 ≤ 2} has a largest element. and the set S = {q ∈ Q  q 2 ≤ 2} does not have a largest element in Q. 1. The real number system R ﬁlls these gaps. Examples. but in this part of the course we try to formalize such notions. An element u ∈ R is said to be an upper bound of S if for all x ∈ S. called the least upper bound property. Deﬁnitions. there is no rational number q such that q 2 = 2. An element l ∈ R is said to be a lower bound of S if for all x ∈ S. 1. we need a few deﬁnitions. The set S is said to be bounded if it is bounded above and it is bounded below. π. You might. 2. x ≤ u. The set S = {x ∈ R  0 ≤ x < 1} is bounded. 1. But before we state this property of R. have a good idea of what we mean by a ‘limit’ of a convergent sequence or the notion of a ‘continuous’ function.Chapter 1 Analysis Analysis is the theory behind real numbers. sequences. 1 . So we see that the rational number system has certain holes. 100) is an upper bound of S. Any real number y satisfying 1 ≤ y (for instance 1. This is a consequence of a very important property of the real numbers.1 The real numbers The rational number system is inadequate for many purposes. then S is said to be bounded above. l ≤ x. The word ‘theory’ is important. If the set of all upper bounds of S is not empty. −1) is a lower bound of S. and functions. 3. For instance. Let S be a subset of R.
Similarly one can show that the inﬁmum of S is 0. Hence 1 is a supremum. Then we have u+1 < 1. any ﬁnite set S is bounded. The set1 S = {(−1)n  n ∈ N} is bounded. . So it follows that u∗ = 1. 1 Note that this set is simply the ﬁnite set (that is. it has no upper bound. and so it belongs to S. 3. Although it is bounded below (any real number x ≤ 1 serves as a lower bound). The sets Z and R are neither bounded above nor bounded below. and so it is not bounded above.1) above then shows 2 that u cannot be an upper bound for S. and In the above example. The set S = n n ∈ N is bounded. From (1. while the last two inequalities follow using u < 1. and the above argument shows that 1 ≤ u∗ . 5.1) 0≤u< 2 where the ﬁrst inequality is a consequence of the facts that u is an upper bound of S and 0 ∈ S. Indeed. Example. this is a consequence of the inequality z < z + 1. Clearly 1 is an upper bound of S. this is always the case and we have the following result.1). then the supremum of S is 1 and the inﬁmum of S is 0. In fact. The set S = {n  n ∈ N} is not bounded. An element u∗ ∈ R is said to be a least upper bound of S (or a supremum of S) if (b) if u is an upper bound of S. there was a unique supremum and inﬁmum of the set S. An element l∗ ∈ R is said to be a greatest lower bound of S (or an inﬁmum of S) if (b) if l is a lower bound of S. Suppose not. 2. The set ∅ is bounded. ♦ (a) l∗ is a lower bound of S. (1. Deﬁnitions. More generally. u∗ must be less than or equal to 1. Analysis 2. Any real number x satisfying 1 ≤ x is an upper bound. Next we show that this is the only supremum. that is. the set has ﬁnite cardinality) {−1. If S = {x ∈ R  0 ≤ x < 1}. (Why?) ♦ We now introduce the notions of a least upper bound (also called supremum) and a greatest lower bound (also called inﬁmum) of a subset S of R. 1}. and as u∗ is a supremum.2 Chapter 1. a contradiction. since if u∗ is another supremum. then in particular u∗ is also an upper bound. Let S be a subset of R. It is bounded above by 1 and bounded below by −1. then u∗ ≤ u. then l ≤ l∗ . then 1 ≤ u. The middle inequality in (1. But 1 < u∗ is not possible as 1 is an upper bound. Now we show that if u is another upper bound. 1 4. 1. it follows that the number u+1 2 satisﬁes 0 < u+1 < 1. and (a) u∗ is an upper bound of S. and 0 is a lower bound. 6. u < 1.
When the supremum and the inﬁmum of a set belong to the set. max. denoted by max S. ♦ In the above examples. For the sets Z and R. the inﬁmum of a set S (if it exists) is also unique. and min S = 1. The real numbers 3 Theorem 1. Then in particular u∗ and u∗ are also upper bounds of S. we obtain u∗ = u∗ . min do not exist. then sup S is called a maximum of S.1.3) 5. which we state below: If S is a nonempty subset of R having an upper bound. In fact this is a fundamental property of the real numbers.2) Furthermore.2) that inf S = 0. it follows that u∗ ≤ u ∗ . 1. max S = 1. If inf S ∈ S. The least upper bound of a set S (if it exists) is denoted by sup S (the abbreviation of ‘supremum of S’). If sup S ∈ S. If S = {n  n ∈ N}.1. sup. then it is unique.3). max S does not exist. 6. From (1. 1 4. we note that if S is nonempty and bounded above. min do not exist. If S = {x ∈ R  0 ≤ x < 1}. min S = −1.2) and (1. max. then sup S = 1. (1. 1. For the set ∅. We show below (after Theorem 1. inf S = 1. Now since u∗ is a least upper bound of S and u∗ is an upper bound of S. then its supremum exists. and is denoted by inf S. (1. 3.1. So min S does not exist. then inf S is called a minimum of S. inf S = −1. But inf S = 0 ∈ S. inf. denoted by min S. Thus it makes sense to talk about the least upper bound of a set. then we give them special names: Deﬁnitions. sup. 2. since u∗ is a least upper bound of S and u∗ is an upper bound of S. and so min S = 0. inf.1. then sup S does not exist. called the least upper bound property of the real numbers.1 If the least upper bound of a subset S of R exists. it follows that u∗ ≤ u ∗ . then sup S exists. Proof Suppose that u∗ and u∗ are two least upper bounds of S. then sup S = 1 ∈ S and so max S does not exist. If S = {(−1)n  n ∈ N}. 2. Examples. If S = n n ∈ N . Similarly. . then sup S = 1 and max S = 1.
Hence 0 is the inﬁmum of S. if q > 2. The negative of this supremum. there exists a n ∈ N such that = y < nx = n · 1 = n. Analysis 1.4) it follows that r > u∗ . contradicting the fact that u∗ is the supremum. Hence there exists a natural number m such that u∗ − x < mx. 1 Example. Thus any lower bound of S n must be less than or equal to 0. u∗ < (m + 1)x ∈ S. This is impossible. which is called the Archimedean property of the real numbers. then inf S = 0.4) implies that r < u∗ . y ∈ R and x > 0. Proof If not. From (1. Assume on the contrary that u∗ ∈ Q is a supremum for S. given a nonempty set S of R. it has a least upper bound u∗ . We know that 0 is a lower bound of S. Suppose that l is a lower bound of S such that l > 0.4 Remarks(∗). One can show that if a nonempty subset S of R is bounded below. namely − sup(−S). l l 1 and so < l. the set S = {q ∈ Q  q 2 ≤ 2} has an upper bound. 3◦ Suppose that u2 > 2.1. This contradicts the fact that u∗ is an upper bound of S. Theorem 1. and so q ∈ S). and so by the least upper bound property of the reals. Deﬁne u2 − 2 r = u∗ − ∗ . But then u∗ − x < u∗ (since x is positive) and so it follows that u∗ − x cannot be an upper bound of S. by the least upper bound property. then q 2 > 4 > 2. ♦ 2 the last inequality follows from (1.4) u∗ + 2 Then we can check that r2 − 2 = We have the following cases: 1◦ Suppose that u2 < 2. For instance. Chapter 1.5) we obtain r ∈ S. but we now show that it does not have a supremum in Q. (1. Thus the real numbers also have the ‘greatest lower bound property’: If S is a nonempty subset of R having an lower bound. ∗ which contradicts the fact that u∗ is an upper bound of S. ∗ 2.5) . then −S is bounded above and so sup(−S) exists. and from (1. if q > r. If S = n n ∈ N .2 (Archimedean property. In Exercise 6 on page 7 below. Note that the set of rational numbers do not possess this property.5) 2◦ Suppose that u2 = 2. We see that r is an upper bound of S (indeed. By the Archimedean property (with the real numbers 1 x and y taken as x = 1 (> 0) and y = 1 ). that is. contradicting the fact that l is a lower bound of S. then2 ∗ q 2 > r2 > 2. can then be shown to serve as the greatest lower bound of S (this is precisely the content of Exercise 6). But (1. We now prove the following theorem. 2(u2 − 2) ∗ (u∗ + 2)2 (1. say 2 (indeed. we deﬁne −S = {−x  x ∈ S}. since u∗ is a rational number. then inf S exists. then the nonempty set S = {nx  n ∈ N} has an upper bound y. and so q ∈ S).) If x. then there exists an n ∈ N such that y < nx.
The distance between two real numbers x and y is the absolute value x−y of their diﬀerence.1. ∞) = R In the above notation for intervals. Thus 1 = 1. ∞) = {x ∈ R  a < x} [a. Thus [0. So an interval is a set of any of the following forms. and so on. 0 = 0.  − 1 = 1. where a. An interval is a set consisting of all the real numbers between two given real numbers. to the point x0 . b) = {x ∈ R  a < x < b} [a. The distance gives a notion of closeness of two points. or of all the real numbers on one side or the other of a given number. which is crucial in the formalization of the notions of analysis. 1. and the distance between real numbers x and y is x − y. . and it is deﬁned as follows: x = x if x ≥ 0. b] = {x ∈ R  a < x ≤ b} replacements [a. b] = {x ∈ R  x ≤ b} (−∞. that is. −x if x < 0. ∞) = {x ∈ R  a ≤ x} (−∞. Deﬁnitions. The absolute value of a real number x is denoted by x. I x0 x x0 + δ x0 − δ Figure 1. and a square bracket ‘[’ or ‘]’ means that the endpoint is included. We can now specify regions comprising points close to a certain point x0 ∈ R in terms of inequalities in absolute values. x0 + δ) = {x ∈ R  x − x0  < δ} is the set of all points in R whose distance to the point x0 is strictly less than δ (> 0). (Note that the use of the symbol ∞ in the notation for intervals is simply a matter of convenience and is not be taken as suggesting that there is a number ∞. The real numbers 5 Deﬁnition. in order to talk about notions such as convergence and continuity. b ∈ R: (a.01 or δ = 0. 1) means the set of all real numbers x such that 0 ≤ x < 1. a parenthesis ‘(’ or ‘)’ means that the respective endpoint is not included. is less than a certain positive number δ.1. b] = {x ∈ R  a ≤ x ≤ b} PSfrag (a.) In analysis.1. and the distance between the real numbers −1 and 1 is equal to  − 1 − 1 =  − 2 = 2. We give the deﬁnitions below. This is provided by the absolute value  · . by demanding that PSfrag points of the the distance of the replacements region. See Exercise 9 on page 8 and Figure 1. we will need a notion of ‘closeness’ between real numbers.0000001. b) = {x ∈ R  a ≤ x < b} (a. b) = {x ∈ R  x < b} (−∞.1: The interval I = (x0 − δ. say δ = 0. a a a a a a b b b b b b 2.
Since x ≥ −x and y ≥ −y.) For all x. Theorem 1. This proves (1. it follows that xy < 0 and so xy = −xy. and so x = −x > 0 > x. it follows that for any real x ∈ R. On the other hand. as x = 0 or y = 0. as x > 0 and y < 0. (Triangle inequality. Then x + y = −(x + y). and so x y = (−x)(−y) = xy.) For all x. Proof We prove (1. Analysis (1.) For all x. as x < 0 and y < 0. First observe that from the deﬁnition of  · . as x > 0 and y > 0. x − y = y − x. then x = x. If x − y = 0 then x = y. Then x = x and y = −y. Then x + y = x + y.6 The following properties of the absolute value will be useful in the sequel. for all x ∈ R. We have the following cases: 1◦ x + y ≥ 0. it follows that xy > 0 and so xy = xy. 4◦ x < 0 and y > 0.6). it follows that xy = 0 and so xy = 0. then xy = x y and x + y ≤ x + y. then −x > 0. (Symmetry.7) 1◦ x = 0 or y = 0. Then x = −x and y = −y. z ∈ R. D3. then is inf S in S? max S min S where S is given by: . and so x y = 0. As x ≥ x and y ≥ y. and so x y = x(−y) = −xy.6) by exhausting all possible cases: Chapter 1. x − z ≤ x − y + y − z.6) (1. 5◦ x < 0 and y < 0.3 If x. it follows that xy > 0 and so xy = xy. 3◦ x > 0 and y < 0. y ∈ R. while if x < 0. 2◦ x > 0 and y > 0. we also have  − x =  − 1 · x =  − 1x = 1x = x. Provide the following information about the set S An upper bound A lower bound Is S bounded? sup S inf S If sup S exists. we obtain x + y ≥ x + y = x + y. and so it follows that x =  − x ≥ −x for all x ∈ R. On the other hand. 1. it follows that x + y ≥ −x + (−y) = −(x + y) = x + y. x − y ≥ 0. and so x y = xy. then is sup S in S? If inf S exists. On the other hand. y ∈ R. (Positive deﬁniteness. It is easy to check that the distance satisﬁes the following properties: D1. Next we prove (1. y. Exercises.6). y are real numbers.1. D2. Then x = x and y = y. On the other hand.7). This proves (1. x ≥ x: indeed if x ≥ 0. Then x = 0 or y = 0. This follows from 3◦ above by interchanging x and y. From (1.7). 2◦ x + y < 0.
2005} (i) (−1)n 1 + 2 2 (g) {x ∈ R  x2 ≤ 2} 1 n x (k) { 1+x2  x ∈ R} . and deﬁne S −1 = x x ∈ S . (l) For every set S that is bounded. (b) If u∗ is the least upper bound of a subset S of R. (m) For every bounded set S. 3 The abbreviation ‘iﬀ’ is standard in Mathematics. 5.1. (h) Every set that has a supremum is bounded above. is any positive real number. show that sup S −1 = 1 inf S . (g) Every bounded nonempty subset of R has a supremum. (i) For every set that has a maximum. (k) For every set S that is bounded above. in case inf S > 0. Let A and B be nonempty subsets of R that are bounded above and such that A ⊂ B. then u is not an upper bound for S. (d) Every subset of R has a supremum. Prove that sup(A + B) exists and that sup(A + B) ≤ sup A + sup B. 1] (b) [0. (j) {x  x ∈ R} n∈N 2. and then u∗ − is not an upper bound of S. 6. For any nonempty bounded set S. the maximum belongs to the set. Prove that sup A ≤ sup B. Let A and B be nonempty subsets of R that are bounded above and deﬁne A + B = {x + y  x ∈ A and y ∈ B}. and it stands for ‘if and only if’. Furthermore. Let −S denote the set of all real numbers −x. (e) Every bounded subset of R has a maximum. 1] (c) (0. (f) Every bounded subset of R has a supremum. 2. the supremum belongs to the set. where x belongs to S. . The real numbers 7 (a) (0. and that the equality holds iﬀ3 S is a singleton set (that is a set with cardinality 1). Let S be a nonempty set of positive real numbers. 1) (d) (e) (f) 1 n 1 −n n n+1 n ∈ Z \ {0} n∈N n∈N (h) {0. 1 7.1. Let S be a nonempty subset of real numbers which is bounded below. Prove that inf S exists and inf S = − sup(−S). S deﬁned by {x  x ∈ S} is bounded. then x ∈ S. (a) If u is an upper bound of a subset S of R. and u < u. Determine whether the following statements are TRUE or FALSE. S deﬁned by {x  x ∈ S} is bounded. 3. 5. Show that −1 S is bounded above iﬀ inf S > 0. 4. if inf S < x < sup S. (c) Every subset of R has a maximum. prove that inf S ≤ sup S. (j) For every set that has a supremum.
11. . − .2 1. Let x0 ∈ R and δ > 0. that is. the second number (namely a2 ).1− n+2 n+2 1 1 . Prove that if x. and so on. a2 . 1] = n∈N 9. . An example is the phrase “an unfortunate sequence of events”. For example. We can identify a ﬁrst event. . a2 as f (2). . The ﬁrst number is 1. n∈N 1 . n 1 1 . An = {x  ∃n ∈ N such that x ∈ An }. where there is the ﬁrst number (namely a1 ). A sequence of real numbers is a list a1 .1 Sequences and limits Sequences The notion of a sequence occurs in ordinary conversation. y are real numbers.2. 1) = n∈N (d) [0. . etcetera. 1. The notation n∈N Chapter 1. 1 1 1. 1 . Let An . Analysis An denotes the intersection of the sets An . a second event. we envision one event causing another. then x − y ≤ x − y. then it becomes clear that a sequence is a special type of function. a3 . 2 3 1 is a sequence. n ∈ N. In this case. An = {x  ∀n ∈ N. x ≤ M . (There may not be a connection between the numbers appearing in a sequence. x ∈ An }. namely one with domain N and codomain R. x0 + δ) = {x ∈ R  x − x0  < δ}. that is. n∈N Prove that (a) ∅ = 0. be a collection of sets.8 8. n ∈ N.1+ n n (b) {0} = n∈N (c) (0.) If we think of a1 as f (1). Show that a subset S of R is bounded iﬀ there exists a M ∈ R such that for all x ∈ S. Prove that (x0 − δ. the second number is 2 and so on. n∈N and we use n∈N An to denote the union of the sets An . which in turn causes another event and so on. 10. and so on. of real numbers. . . n 0. . . . n ∈ N.
for n ∈ N. . . . . n=1 In (an )∞ . . 1 n n∈N is a sequence with the nth term given by 1 n. . −1. 1. This is the sequence 3 4 5 6 7 2. Sequences and limits 9 Deﬁnition. with the nth term equal to −1 if n is odd. 1. . The entire sequence is then written in any one of the following ways: (an )n∈N . 3. 7. 2. . n+1 (Write down the ﬁrst few terms of this sequence. as in the example given above: 1 an = . . 2 3 4 5 6 1 3. . . . the ∞ symbol indicates that the assignment process 1 → a1 . . for n ∈ N. . Only the notation is somewhat unusual. . This is the sequence 1 1 1.1. . we write an . 2 3 2. . . (1)n∈N is a sequence with the nth term given by 1. and 1 if n is even. (−1)n 1 + n is the sequence n∈N is a sequence with the nth term given by (−1)n 1 + 3 4 5 6 7 −2.. . . for n ∈ N. . . . This sequence is simply −1. (an ). 5. ((−1)n )n∈N is a sequence with the nth term given by (−1)n . . . . an+1 = n an for n ∈ N. 6. . . for n ∈ N. − .. . for n ∈ N. a1 = 1. . 1 + 1 n n∈N 1 is a sequence with the nth term given by 1 + n . . The nth term an of a sequence may be deﬁned explicitly by a formula involving n. (n)n∈N is a sequence with the nth term given by n.. continues n=1 indeﬁnitely. Instead of writing f (n) for the value of f at a natural number n. . . − . (an )∞ . n ∈ N. 1 + 21 + 31 + · · · + nn n∈N is a sequence with the nth term given by 2 3 for n ∈ N. . . n 1 1 1 1 1 1 . n It might also sometimes be deﬁned recursively. + 2. 1 + 2 + 3. 1. A sequence is a function f : N → R. This 4. −1. This is the constant sequence 1. 1.) Examples. . 1..2. for n ∈ N. This is the sequence of ‘partial sums’ 1 11 1 11 1 1 + 22 + 33 +· · ·+ n1 . 2 → a2 . For example. (an )n≥1 . 2 3 4 5 6 1 n . This is the increasing sequence 1. 1. 11 11 2 1 2 3 ♦ .
then the sequence is said to diverge.2: First four points of the graph of the sequence 1 n n∈N tend toward 0 as n increases. the terms of 1 + n n∈N get “closer and closer” to 0 (indeed the distance to 0 keeps decreasing). but the sequence has no limit.2. Figure 1. 2 PSfrag replacements 1 1 2 3 4 1 n n∈N . 4 depending > 0. If there is no such ﬁnite number L to which the terms of the sequence get arbitrarily close. Deﬁnition.2. Figure 1. but its limit is 1. n→∞ where L denotes the limit of the sequence. This portion of the graph suggests that the terms of the sequence PSfrag replacements 1 1 2 3 4 1 n n∈N .2 Limit of a convergent sequence 1 A sequence can be graphed. or “as close as we like”. if the terms an get “closer and closer” to L as n “increases without bound”. Analysis 1. This is consistent with the idea of convergence that you might have encountered before: a sequence (an )n∈N converges to some real number L.4. the ﬁrst 4 points of the graph of the sequence n n∈N are displayed in Figure 1. the concepts of their convergence and divergence must be carefully deﬁned.10 Chapter 1. or “arbitrarily close” to some number L? Even if we accept this apparent ambiguity. For instance. Exactly what does it mean for the terms of a sequence to get “closer and closer”.3: First four points of the graph of the sequence 1 + 1 For example.3. 1 The terms of (−1)n 1 + n n∈N get “as close as we like” or “arbitrarily close” to 1. Symbolically. there exists4 an on . this is represented using the notation lim an = L. The sequence (an )n∈N is said to converge to L if for every N ∈ N such that for all n > N . Th problem with this characterization is its imprecision. See Figure 1. how would one use the deﬁnition given in the preceding paragraph to prove theorems that involve sequences? Since sequences are used throughout analysis. an − L < . See Figure 1.
for a convergent sequence with limit L. 1. Then one can ﬁnd a N ∈ N. this deﬁnition means the following. L + ).2. Hence pictorially. L L+ L L− PSfrag replacements L+ L L− N N +1 N +2 N +3 Figure 1. Examples. for n > N lie in the shaded strip.1. and consider the shaded strip of width around the horizontal line passing through L. large enough. then the sequence (an )n∈N is said to n→∞ be divergent. Sequences and limits PSfrag replacements 11 2 1 1 2 3 4 5 6 7 8 −1 −2 Figure 1. . Then we say that (an )n∈N is convergent (with limit L) and write n→∞ lim an = L. Note that an − L < iﬀ an ∈ (L − .4: First eight points of the graph of the sequence (−1)n 1 + 1 n n∈N . such that all the terms an of the sequence.5: Pick any > 0. 1 n n∈N is a convergent sequence with limit 0. If there does not exist a number L such that lim an = L.5: Convergence of a sequence with limit L. as illustrated in Figure 1.
n n (such a N exists by the Archimedean property!). Let L ∈ R. Analysis 1 1 −0 = < . (It is not obvious that error. Now we need to prove ∀L ∈ R ∃ > 0 such that ∀N ∈ N ∃n > N such that an − L ≥ . (−1)n 1 + n∈N Let an = (−1) 1+ for n ∈ N. then 1 1 1 −0 = < < . 1 n 3. is a divergent sequence. that is. So let N ∈ N. then choose n to be any odd number > N . n n Thus we have shown that for all L ∈ R. we have 1 Chapter 1. In order to prove that (an )n∈N is divergent. an − L = If we choose N ∈ N such that N > 1 1 for n > N (⇔ n < N ). Thus the sequence is divergent. then choose n to be any even number > N . and any even number > N if L < 0) such that an − L ≥ . If L ≥ 0. an − L < ] . ¬ [(an )n∈N is convergent] . we need to ﬁnd a N such that for all n > N .) Now we will show that = 1 would work. 1 1 −1 = < . we need to ﬁnd a N such that for all n > N . n N an − L = Hence lim n→∞ 1+ 1 n n 1 n = 1.12 Given > 0. there exists a n > N (namely any odd number > N if L ≥ 0. we have to show that that is. Then we have an − L = (−1)n 1 + 1 n − L = −1 − 1 1 −L =1+ +L>1= . 1 + n→∞ 1 = 0. Then we have an − L = (−1)n 1 + 1 n −L = 1+ 1 1 −L =1+ −L>1= . ∃ > 0 such that ∀N ∈ N ∃n > N such that an − L ≥ . ♦ . n n 1 n n∈N Given > 0. but it has been found by trial and ∀N ∈ N ∃n > N such that an − L ≥ . ¬ [∃L ∈ R such that ∀ > 0 ∃N ∈ N such that ∀n > N. n n N an − L = Hence lim 2. there exists a > 0 (namely = 1) such that for all N ∈ N. n is a convergent sequence with limit 1. an − L = 1 + Again we choose a N ∈ N such that N > 1 and so for n > N we have 1 1 < < . Let = 1. n n If L < 0.
2 where the positivity of the deﬁned above follows from the fact that L1 = L2 . then L ≥ 0.1 A convergent sequence has a unique limit. 2 = L1 − L2  = L1 − an + an − L2  ≤ L1 − an  + an − L2  = an − L1  + an − L2  < + = 2 . then prove that the terms of the sequence cannot consist of a ﬁnite number of distinct values. > 0 and a N ∈ N such that for all n > N . Indeed this n→∞ is the case. Let S be a nonempty subset of R that is bounded above. ∃N2 ∈ N such that for all n > N2 . Show that there exists a sequence (an )n∈N contained in S (that is. Prove that lim n→∞ 1 = 1. an − L < . A sequence (an )n∈N is said to be Cauchy if for every for all n. there exists an N such that for inﬁnitely many n > N . n 3.1. (b) There exists an 4. Proof Consider a convergent sequence (an )n∈N and suppose that it has distinct limits L1 and L2 . ∃N1 ∈ N such that for all n > N1 . (d) Prove that the sequence ((−1)n )n∈N is divergent. Let L1 − L2  = > 0.2. an ∈ S for all n ∈ N) and which is convergent with limit equal to sup S. Since L1 is a limit. Show that every convergent sequence is Cauchy. Consequently for n > max{N1 . Prove that if (an )n∈N is convergent with limit L. 6. (a) For every > 0. there exists a N ∈ N such that Hint: an − am  = an − L + L − am  ≤ an − L + am − L. an − L2  < . 2. Sequences and limits 13 The notation lim an suggests that the limit of a convergent sequence is unique. m > N . and we prove this below. an ≥ 0. 5. . an − L1  < . Suppose that L = 1.2. an − am  < . give an example of a divergent sequence (a n )n∈N that satisﬁes the given conditions. Theorem 1. an − L < . (b) Can the limit of a convergent sequence be one of the terms of the sequence? (c) If none of the terms of a convergent sequence equal its limit. 1. (a) Prove that the constant sequence (1)n∈N is convergent. Exercises. In each of the cases listed below. Since L2 is a limit. Let (an )n∈N be a sequence such that for all n ∈ N. > 0. N2 }. a contradiction.
and so the sequence is bounded.3 Bounded and monotone sequences It is cumbersome to check from the deﬁnition if a sequence is convergent or not. Indeed.2. 6. (1)n∈N is bounded. < 2 1− 1 2 3 (Write down a detailed proof using induction on n. n ∈ N 11 2 3 n is bounded.14 Chapter 1. an  = n = n ≤ M ]. since 1 = 1 ≤ 1 for all n ∈ N. Deﬁnition. (1. 1. The sequence (an )n∈N deﬁned by an = 1 1 1 1 + 2 + 3 +···+ n. The sequence (an )n∈N given by an = n for n ∈ N. Thus ¬[∃M > 0 such that for all n ∈ N. 5. 1 n n∈N is bounded. Analysis 1. an  ≤ M. (See Exercise 11 on page 8). then it is always convergent. and so (n)n∈N is not bounded. 2. is not bounded. Indeed this can be seen as follows: an  = an 1 1 1 1 + 2 + 3 +··· + n = 11 2 3 n 1 1 1 1 + 2 + 3 +··· + n < 11 2 2 2 1 1 1 1 1 1 = 1− + 2 1− + · · · + n−1 + 11 2 2 2 2 2 1 1 1 1 1 1 = 1 + − 2 + 2 − 3 + − · · · + n−1 − n 2 2 2 2 2 2 1 1 = 1+ − n 2 2 3 . given any M > 0. since (−1)n 1 + =1+ 1 n ≤ 2 for all n ∈ N. 1 n 4. ♦ . In this section. there exists an N ∈ N such that M < N (Archimedean property with y = M and x = 1). we will study a condition under which we can conclude that a sequence is convergent even without knowing its limit! We will prove that if a sequence is both ‘bounded’ as well as ‘monotone’. since 1 n = 1 n ≤ 1 for all n ∈ N. Examples. 1 n 3. A sequence (an )n∈N is said to be bounded if there exists a M > 0 such that for all n ∈ N. (−1)n 1 + is bounded. 1 + 1 n n∈N is bounded.) So all the terms are bounded by 2 . since 1 + 1 n n∈N =1+ 1 n ≤ 2 for all n ∈ N.8) Note that a sequence is bounded iﬀ the set S = {an  n ∈ N} is bounded.
.6: All convergent sequences are bounded. . . . convergent sequences bounded sequences PSfrag replacements L+ Figure 1. . 1 11 + 1 22 (1)n∈N (n)n∈N + 31 + · · · + 3 1 nn n∈N Is it increasing? No No No Yes Yes Yes Is it decreasing? Yes Yes No Yes No No Is it monotone? Yes Yes No Yes Yes Yes ♦ The following theorem can be useful in showing that sequences converge when one does not know the limit beforehand. A sequence (an )n∈N is said to be increasing if for all n ∈ N. .6. .2. . and we have shown above that these are also bounded.2 If a sequence is convergent. Deﬁnitions. n n∈N . . Thus (an )n∈N is increasing if a1 ≤ a 2 ≤ a 3 ≤ . This is not a coincidence. Hence for all n > N . Sequence 1 n n∈N 1 1 + n n∈N 1 (−1)n 1 + n n∈N A sequence (an )n∈N is said to be decreasing if for all n ∈ N. aN . Sequences and limits 15 1 1 The sequences (1)n∈N . Then ∃N ∈ N such Let M = max{a1 . Examples. Let that for all n > N an − L < = 1. an ≤ an+1 .2. and in the next theorem we show that the set of all convergent sequences is contained in the set of all bounded sequences. then it is bounded. A sequence is said to be monotone if it is increasing or decreasing. an  = an − L + L ≤ an − L + L < 1 + L. 1 + n n∈N are all convergent. := 1 > 0. . Then for all n ∈ N an  ≤ M and so (an )n∈N is bounded. 1 + L}. Theorem 1. Proof Let (an )n∈N be a convergent sequence with limit L.1. See Figure 1. . Thus (an )n∈N is decreasing if a1 ≥ a 2 ≥ a 3 ≥ . an ≥ an+1 .
it follows that sup S − is not an upper bound for S and so ∃aN ∈ S such that sup S − < aN . Since (an )n∈N is bounded. Theorem 1.3 that this sequence 5 is convergent. ♦ it is known that this sequence is convergent to some limit L ∈ R. and gives counterexamples to implications which are not true.2. We have shown that the sequence (an )n∈N deﬁned by an = 1 1 1 1 + 2 + 3 +···+ n. where S = {an  n ∈ N} (see Exercise 6 on page 7). it follows that (−an )n∈N is a convergent sequence with limit sup{−an  n ∈ N} = − inf{an  n ∈ N} = − inf S. we have aN ≤ an . the proof of which is beyond the scope of this course. for n > N . that is. n ∈ N 1 1 2 3 n 1 is monotone (indeed.2 Is every bounded sequence convergent? No ((−1)n )n∈N is bounded. it is so far not even known if the limit L is rational or irrational. n x n 0 x n=1 . it is increasing since an+1 − an = (n+1)(n+1) > 0 for all n ∈ N) and bounded (see item 5 on page 14). then (−an )n∈N is bounded as well ( − an  = an  ≤ M ). We show that in fact (an )n∈N converges to sup S. that is sup S − aN < .3 n∈N is convergent.2. Since (an )n∈N is an increasing sequence. ∃N ∈ N such that for all n > N . Since sup S is an upper bound for S. Proof 1◦ Let (an )n∈N be an increasing sequence. 2◦ If (an )n∈N is a decreasing sequence. an ≤ sup S and so an − sup S = sup S − an . then clearly (−an )n∈N is an increasing sequence. Thus for n > N we obtain an − sup S = sup S − an ≤ sup S − aN < . Analysis Theorem 1. it follows that the set S = {an  n ∈ N} has an upper bound and so sup S exists.16 Chapter 1. Is every convergent sequence monotone? Is every monotone sequence convergent? Is every bounded AND monotone sequence convergent? 5 Although No No Yes (−1)n n 1 1 but not monotone: −1 < 2 > − 3 .3 If a sequence is monotone and bounded.2. Thus (an )n∈N is convergent with limit inf S. Thus it follows from Theorem 1. The following table gives a summary of the valid implications. and this is still an open problem in mathematics! Also associated with this sequence Z 1 ∞ X 1 1 is the interesting identity = dx. Indeed given > 0. an − inf S < . then since sup S − < sup S. 1. but not convergent. So given > 0. then it is convergent. Question Answer Reason/Counterexample Is every convergent sequence bounded? Yes Theorem 1.2. Furthermore if (an )n∈N is bounded. Hence by the case considered above. (n)n∈N is not convergent. Examples. 2.  − an − (− inf S) < .
bn n n∈N 2n + 1 an−1 for n ≥ 2. we obtain 7 11 n . 1. 3+0+0 3 . 4. (∗) Given a bounded sequence (an )n∈N . then the new sequence which is obtained. Sequences and limits 17 Exercises. (∗) Recall the deﬁnition of a Cauchy sequence from Exercise 6 on page 13. Thus. Prove that every Cauchy sequence is bounded. using the formal deﬁnition of a limit. one can show that the sequence (an )n∈N deﬁned by 4n2 + 9 an = 2 3n + 7n + 11 4 converges to 3 . below. product. Show that the sequences (ln )n∈N . it is simpler to observe that 9 n2 4 + n 2 7 11 n2 3 + n + n 2 9 4 + n2 7 11 .2. For instance. 2. 3n is a convergent sequence with 3.) n→∞ 5. deﬁne lk = inf{an  n ≥ k} and uk = sup{an  n ≥ k}. where a1 + · · · + a n . and conclude that they are convergent. (un )n∈N are bounded and monotone. n2 = all have limit 0.2. is again convergent. sn = n is also convergent with limit L.4 given 9 n2 9 n2 n→∞ lim an = n→∞ lim 4+ n→∞ lim 3+ 11 7 + n n2 = n→∞ lim 4 + lim n→∞ 7 11 lim 3 + lim + lim 2 n→∞ n→∞ n n→∞ n = 4 4+0 = . 3 + n + n2 an = 9 where the terms n2 . Let (an )n∈N be a sequence deﬁned by a1 = 1 and an = Prove that (an )n∈N is convergent. (b) Give an example of a sequence (an )n∈N such that (sn )n∈N is convergent but (an )n∈N is divergent. n→∞ k ∈ N. (a) (∗) If (an )n∈N is a convergent sequence with limit L. then prove that limit 0. and moreover the limit of this sequence is the same algebraic combination of the limits.2. If (bn )n∈N is a bounded sequence. quotient of terms of simpler sequences with a known limit.4 Algebra of limits In this section we will learn that if we ‘algebraically’ combine the terms of convergent sequences. then prove that the sequence (sn )n∈N . In this manner we can sometimes prove the convergence of complicated sequences by breaking them down and writing them as an algebraic combination of simple sequences. we conveniently apply arithmetic rules to compute the limits of sequences if the terms are the sum. n ∈ N. and by a repeated application of Theorem 1. 1.1. However. (Their respective limits are denoted by lim inf an and lim sup an .
n→∞ bn lim bn lim n→∞ Proof Let (an )n∈N and (bn )n∈N converge to La and Lb .18 Chapter 1. n∈N 1 1 = . Analysis Theorem 1. then the following hold: 1. that is. then given > 0. (αan )n∈N is a convergent sequence and lim αan = α lim an . If for all n ∈ N. respectively. n→∞ α = . then αan = 0 for all n ∈ N and clearly (0)n∈N is a convergent sequence with limit 0. Then we have for all n > N : an  − La  ≤ an − La  < . (an bn )n∈N is a convergent sequence and lim an bn = n→∞ n→∞ lim an n→∞ lim bn . For all k ∈ N. n→∞ 3. 6. an − La  < . (an + bn )n∈N is a convergent sequence and lim (an + bn ) = lim an + lim bn . that is. Given > 0. n→∞ n→∞ 2. . (ak )n∈N is a convergent sequence and lim ak = n n n→∞ n→∞ lim an . k 5.4 If (an )n∈N and (bn )n∈N are convergent sequences. If α = 0. let N1 ∈ N be such that for all n > N1 . n→∞ lim an  = La  = lim an . n→∞ 2. (an )n∈N is a convergent sequence and lim an  = lim an . let N ∈ N be such that for all n > N .2. lim αan = αLa = α lim an . let N ∈ N be such that for all n > N . Given > 0. an − La  < α . n→∞ n→∞ 3. then n→∞ 1 bn is convergent and moreover. 1. Hence (an )n∈N is convergent with limit La . Thus lim αan = 0 = 0La = α lim an . that is αan − αLa  = α an − La  < α Hence (αan )n∈N is convergent with limit αLa . n→∞ n→∞ If α = 0. For all α ∈ R. n→∞ n→∞ n→∞ 4. bn = 0 and lim bn = 0. an − La  < 2 .
Suppose that it holds for some k. This can be shown by using induction on k and from part 4 above.9) Step 1. Sequences and limits 19 Let N2 ∈ N be such that for all n > N2 . by Theorem 1. that is. Since (bn )n∈N is convergent. So (an bn )n∈N is a convergent sequence with limit La Lb . n→∞ lim an bn = La Lb = n→∞ lim an n→∞ lim bn . . lim (an + bn ) = La + Lb = lim an + lim bn .1. n→∞ n→∞ 4. an bn − La Lb  < . Let N1 ∈ N be such that for all n > N1 . This can be achieved by ﬁnding a N such that each of the summands in (1. 2 (1. bn  ≤ M . we have an bn − La Lb  ≤ an − La  bn  + La  bn − Lb  < = = 2M 2 . 2 + 2 = . that is. + M + La  2 2(La  + 1) 2(La  + 1) . n→∞ 2 .2. 2M . we obtain that the n sequence (an · ak )n∈N is convergent and n n→∞ lim an ak = n n→∞ lim an n→∞ lim ak = n k n→∞ k+1 lim an n→∞ lim an = n→∞ lim an . N2 }. we have an + bn − (La + Lb ) = an − La + bn − Lb  ≤ an − La  + bn − Lb  < Hence (an + bn )n∈N is convergent with limit La + Lb . Note that an bn − La Lb  = an bn − La bn + La bn − La Lb  ≤ an bn − La bn  + La bn − La Lb  = an − La  bn  + La  bn − Lb . then (ak )n∈N is convergent and n n→∞ lim ak = n k n→∞ lim an .9) is less than for n > N . Let N2 ∈ N be such that for all n > N2 . we need to ﬁnd a N such that for all n > N . It is trivially true with k = 1.2. an − La  < Step 2.2 it follows that it is bounded: ∃M > 0 such that for all n ∈ N. bn − Lb  < Then for all n > N := max{N1 . Given > 0. This can be done as follows. 5. Hence by part 4 above applied to the sequences (an )n∈N and (ak )n∈N . N2 }. bn − Lb  < Thus for n > N := max{N1 .
2 Hence for n > N := max{N1 . 2 Lb  − bn  ≤ Lb  − bn  ≤ bn − Lb  < and so bn  ≥ Lb  2 . Recall the convergent sequence (an )n∈N from Exercise 1 on page 17 deﬁned by a1 = 1 and an = What is its limit? Hint: If (an )n∈N is a convergent sequence with limit L. Prove that the sequence (cn )n∈N deﬁned by cn = is convergent. (b) Show that ( n2 + n − n)n∈N is a convergent sequence and ﬁnd its limit. Prove that if (an )n∈N and (bn )n∈N are convergent sequences such that for all n ∈ N. bn − Lb  < Thus for all n > N1 . Suppose that the sequence (an )n∈N is convergent.  an − L an + L = an − L < L. If L > 0. then√ − choose N ∈ N large enough so that √ √ √ for n > N . Let > 0. then lim an ≤ lim bn . √ Hint: ‘Rationalize the numerator’ by using n2 + n + n. 6. then (an+1 )n∈N is also a convergent sequence with limit L. an ≤ bn . Lb  . bn − Lb  < Lb 2 . 2 Lb  . Analysis lim ak+1 = n n→∞ lim an k+1 . . bn Lb lim bn n→∞ Exercises. 3. 1. N2 }. and let N2 ∈ N be such that for all n > N2 . Prove that the sequence ( an )n∈N is also convergent. we have So 1 bn 1 Lb 2 2 1 bn − Lb  1 − < = . Hint: First show that L ≥ 0. If L = 0. = bn Lb bn  Lb  2 Lb  Lb  n→∞ is convergent and lim n∈N 1 1 1 = = . (a) Let (an )n∈N be a convergent sequence with limit L and suppose that an ≥ 0 for all √ √ n ∈ N. an bn + 5n a2 + n n 2n + 1 an−1 for n ≥ 2.20 Thus (ak+1 )n∈N is convergent and n n→∞ Chapter 1. n→∞ n→∞ Hint: Use Exercise 5 on page 13. with limit L. Let > 0. and assume that the sequence (bn )n∈N is bounded. then choose N ∈ N large enough so that for n > N . 3n 4. 2. Let N1 ∈ N be such that for all n > N1 . an√ L = an < 2 . and ﬁnd its limit.
n→∞ Given > 0. 1000 . lim n = 0. Let N2 ∈ N be such that for all n > N2 . n2 < 10n . and −(cn − L) < . from the Sandwich theorem it follows that the sequence n→∞ n→∞ n is convergent and n lim = 0. that is. n→∞ n→∞ Proof Let L denote the common limit of (an )n∈N and (bn )n∈N : n→∞ lim an = L = lim bn .5 (Sandwich theorem.5 Sandwich theorem Another useful theorem that is useful in proving that sequences are convergent and in determining their limits is the socalled sandwich theorem. bn − L < . that is. n→∞ 1 Since lim 0 = 0 = lim . Sequences and limits 21 1. Theorem 1. then the sandwiched sequence is also convergent with the same limit. let N1 ∈ N be such that for all n > N1 . 10000 . L − an ≤ L − an  = an − L < . that is. is convergent with limit 0. n 10 n n n 10n n∈N L − < an . . N2 }.2. bn − L < . 1. that is. n→∞ 10n 2 3 4 1 Thus the sequence 10 . Consequently. an − L < . we have 0≤ n n 1 ≤ 2 = . cn − L < This proves that (cn )n∈N is convergent with limit L.) Let (an )n∈N . then (cn )n∈N is also convergent with the same limit. 10n It can be shown by induction that for all n ∈ N. So for n > N2 . Examples. and so cn − L < . and so L − < cn < L + . n→∞ n→∞ If (cn )n∈N is a third sequence such that for all n ∈ N. n→∞ lim an = lim cn = lim bn .1. Consequently. an ≤ cn ≤ bn . . bn < L + . (bn )n∈N be convergent sequences with the same limit. Roughly speaking. Thus for n > N := max{N1 . . 100 . Hence for n > N1 . .2. lim an = lim bn .2. we have L − < a n ≤ cn ≤ bn < L + . it says that if a sequence is sandwiched between two convergent limits with the same limit. and so L − an < .
n 1 1 1 n ≤ 2 + 2 +···+ 2 ≤ 2 . the sequence 2. If a ∈ (0.001146260873. nh Hence from the Sandwich theorem. a contradiction). it follows that lim (an + bn ) n = max{a. we have that lim (24n + 2005n ) n = 2005. 4. n→∞ 1 Since 0 < a < 1. is convergent with limit 2005. . 4 2. b}. n→∞ 1 Clearly. 2. 2005. Analysis 2 > 1 and so 2 n > 1 (for otherwise 2 = (2 n )n ≤ 1.22 2. n→∞ n→∞ √ √ √ √ that is. 0 ≤ an ≤ n→∞ and so 5. +n n +1 n +2 n +n n +1 lim n n = 0 = lim 2 . that is. . b}. b}. we have n2 n2 and since n→∞ = 0. . then lim an = 0. Prove that the sequence n! nn is convergent and lim n∈N n! = 0. Let an := 2 n − 1 ≥ 0. 3. n→∞ 1 Chapter 1. 1). lim (an + bn ) n = max{a. n→∞ n + 1 n2 + n n→∞ it follows from the Sandwich theorem that lim 1 1 1 + +···+ 2 n2 + 1 n2 + 2 n +n = 0. For any a. Then we have 1 = (1 + h)n ≥ 1 + nh ≥ nh an 1 . (1 + x)n ≥ 1 + nx. with a = 24 and b = 2005. it follows that lim an = 0. b ∈ R. b}) max{a. n→∞ the sequence 2028.1436. it follows that 1 < 1 a and so h := 1 a − 1 > 0.) Hence 0 ≤ an ≤ 1 n 1 1 1 1 and so using the Sandwich theorem it follows that lim an = 0. . lim 2 n = 1. lim 1 1 1 + 2 +···+ 2 n→∞ +1 n +2 n +n For all n ∈ N. n→∞ nn 1 2 n 1 1 n! Hint: Observe that 0 ≤ n = · · · · · · ≤ · 1 · · · · · 1 ≤ . is convergent with limit 1. b}) + (max{a. ♦ Exercises. b} ≤ (an + bn ) n ≤ 2 n max{a. 3 2. 2005... n n n n n n . b}) ≤ an + bn ≤ (max{a. Then 2 = (1 + an )n ≥ 1 + nan (It can be shown using induction that for all real x ≥ −1 and for all n ∈ N. 5 2. lim 2 n = 1. 1 1 n n n So using the Sandwich theorem. n→∞ 1 In particular. 1. and so (max{a. Consequently.
and let (bn )n∈N satisfy bn − an  < Show that (bn )n∈N is also convergent.10). . 5. the sequence 1k + 2 k + 3 k + · · · + n k nk+2 is convergent and 1k + 2 k + 3 k + · · · + n k = 0. . .10) √ n) n ≤ 2 1 1+ √ n 2 . Deﬁnition. n2 n∈N . (an )n∈N deﬁned by 1 n! n∈N and 1 nn n∈N are all subsequences of 1 n n∈N . 1 √ n in the inequality (1.2. . . 1 ≤ n n < (1 + Hint: Take x = 1 n 1 n∈N (1. 1. 7. (c) Prove that (n )n∈N is convergent and ﬁnd its limit. (b) Show that for all n ∈ N. 4. .1. known as the BolzanoWeierstrass theorem. We begin this section by deﬁning what we mean by a subsequence of a sequence. 1 1 2n n∈N . a < an < b). Let (an )n∈N be a sequence contained in the interval (a. then (1 + x)n ≥ 1 + nx. where p is the nth prime in the inﬁnite sequence of increasing primes 2. 3. What is its limit? 1 Hint: Observe that − n + an < bn < an + 1 n 1 n for all n ∈ N. which says that every bounded sequence has a convergent ‘subsequence’. . Let (an )n∈N be a convergent sequence. . Then (ank )k∈N is called a subsequence of (an )n∈N . Hint: Use Exercise 5 on page 13. b) (that is. . Prove that for all k ∈ N. Examples. prove that if x ≥ −1 and n ∈ N. is a subsequence of The sequence 1 1 1 . 1. b). b]. If (an )n∈N is convergent with limit L. 2 3 4 is not a subsequence of 1 n n∈N . p 1 n n∈N . Also the sequence an = 1 . (a) Using induction.2. 5. for all n ∈ N. n→∞ nk+2 lim 3. 11. . Give an example to show that L needn’t belong to (a.6 Subsequences In this section we prove an important result in analysis. . 1. . for all n ∈ N. then prove that L ∈ [a. Let (an )n∈N be a sequence and let (nk )k∈N be a sequence of natural numbers such that n1 < n2 < n3 < . Sequences and limits 23 2.
8 . 4 2. Theorem 1. . . . 7 2. Thus we start with 2 and 7. we get subsequent terms as follows: 2. 6 eventually. 7. . 8. . ♦ Theorem 1. 4. 6. 4. the sequence 2. . Since the terms 2. 1. Hence for k > K. 1. 1. it follows that there exists a K ∈ N such that nK > N . . Hint: Show that 6 appears an inﬁnite number of times as follows. 7. . 8 2. . Given > 0.7 Every sequence has a monotone subsequence. 1. Proof Let (ank )k∈N be a subsequence of (an )n∈N . 7. −1. 2. nk > nK > N . 1. 4. .or twodigit number. 1. 4. . 2. 7. 1.6 If (an )n∈N is a convergent sequence with limit L. let N ∈ N be such that for all n > N . 7. The sequence ((−1)n )n∈N is divergent since the subsequence 1. Then for all k > K. 1. 4 2. has limit −1. and so on. 4. an − L < . 1. 7. . . 7. 1. 1 2n n∈N . 2. 8. Examples. 8 2. . −1. . and so the next two terms are 1. . 4. Proceeding in this way. . Analysis 2.2. . has limit 1. 2. and so (ank )k∈N is convergent with limit L. 1. Beginning with 2 and 7. is constructed by multiplying successive pairs of its terms and adjoining the result as the next one or two members of the sequence depending on whether the product is a one. . 7. then any subsequence of (an )n∈N is also convergent with the limit L. 2. 4. 2. . −1. 4. while the subsequence −1. 7 2. Proceeding in this manner. . . 4 2. 4. 8 are adjacent. 7.2. The sequences ((−1)2n )n∈N (that is. they give rise to the adjacent terms 1. Since the sequence n1 < n2 < n3 < . 4. ank − L < . 7. ) and ((−1)2n−1 )n∈N (that is the constant sequence −1. 4. ♦ Exercise(∗). 1. 4.24 Chapter 1. 6. 6. 1. 7. 2. 1 n2 n∈N . 1. 7 2. ) are subsequences of ((−1)n )n∈N . −1. 1. 7. 6. 7. 6 at some point. 8. 2. the constant sequence 1. 1 n! n∈N and 1 nn n∈N are convergent sequences with limit 0. 4. which in turn give rise to the adjacent terms 6. . 1. 4 2. 7. 7. 7. ﬁnd out if you get a loop containing the term 6. giving the product 14. . 8. Prove that this sequence has the constant subsequence 6. .
which is followed by hotel n + 1. it follows from Theorem 1. An important consequence of the above theorem is the following result. Then (ank )k∈N is a decreasing subsequence of (an )n∈N . it follows that the sequence (an )n∈N has a monotone subsequence (ank )k∈N .2.7: The seaview property. an  ≤ M .3 that it is convergent. .. . Proof Let (an )n∈N be a bounded sequence. ank  ≤ M and so the sequence (ank )k∈N is also bounded. Then we have the following two cases. 1◦ S is inﬁnite. .7 above. Then there exists a M > 0 such that for all n ∈ N. since nk > max S. which is taken as the next hotel.. and also impossible if S was not empty. Assume that (an )n∈N is the given sequence. and this is not possible if S was empty. Now there are only . Figure 1. 1 2 3 4 5 6 . Then after the last hotel with the seaview property.2. Deﬁne inductively nk+1 = min{m ∈ N  m > nk and am ≥ ank }. along an inﬁnite line. Then clearly for all k ∈ N. Since (ank )k∈N is monotone and bounded. Arrange the elements of S in increasing order: n1 < n2 < n3 < . and then ﬁnding yet another that is at least as high as that one.8 (BolzanoWeierstrass theorem. A hotel is said to have PSfrag replacements the seaview property if it is higher than all hotels following it. an < am }. where at inﬁnity there is the sea. then deﬁne n1 = 1. See Figure 1. and otherwise let n1 = max S + 1. 2◦ S is ﬁnite. Then their heights form a decreasing subsequence.) Every bounded sequence has a convergent subsequence.1.. Sequences and limits 25 We ﬁrst give an illustration of the idea behind this proof. then nk ∈ S. one can start with any hotel and then always ﬁnd one that is at least as high. two possibilities: 1◦ There are inﬁnitely many hotels with the seaview property. 2◦ There is only a ﬁnite number of hotels with the seaview property.) Then (ank )k∈N is an increasing subsequence of (an )n∈N . . (The minimum exists since the set {m ∈ N  m > nk and am ≥ ank } is a nonempty subset of N: indeed otherwise if it were empty. From Theorem 1. If S empty. The heights of these hotels form an increasing subsequence.2.7. Theorem 1..2. Imagine that an is the height of the hotel with number n. and so on. Proof Let S = {m ∈ N  for all n > m. and so on.
. a4 = 0.26 Chapter 1. Prove (using the fact that (an )n∈N is Cauchy).) The terms of the sequence (an )n∈N are as follows: √ 2 = 1. From Exercise 5 on page 17. . √ 2 2 = 2. Hint: Proceed as follows. Hence the set {z ∈ Z  x < z} is a nonempty subset of Z (indeed. (∗) Recall the deﬁnition of a Cauchy sequence from Exercise 6 on page 13.242640 . . and so the minimum of the set {z ∈ Z  x < z} exists in Z. and in order to avoid falling into pitfalls with simplistic thinking. Within the huge collection of functions. say (ank )k∈N with limit L. √ 3 2 = 4. that then (an )n∈N is itself convergent with limit L. . . √ 2. a2 = 0. −n2 < x < n1 . The sequence (an )n∈N is bounded: indeed. . . . a1 = 0. . . that is. Let (an )n∈N be a Cauchy sequence. a6 = 0. Analysis Example. So by the BolzanoWeierstrass theorem it has a convergent subsequence6 . where we had already seen that every convergent sequence is Cauchy.414213 . x denotes the greatest integer part of x. Continuous functions play a prominent role in analysis since they possess some useful properties. 1. By the Bolzano Weierstrass theorem.071067 . there is an important subset: the continuous functions. 0 ≤ an < 1.3 Continuity A function f : R → R is a rule of correspondence that assigns to each real number a unique real number.414213 .656854 . .071067 . deﬁned where for x ∈ R.656854 . . .485281 . .828427 . .828427 . we need deﬁnitions and hypothesis of theorems to be stated carefully and clearly. Consequently · is a welldeﬁned function. . . n there exists a subsequence of the sequence (an )n∈N above that converges to L. there exist natural numbers n1 and n2 such that x < n1 and −x < n2 . √ 4 2 = 5. Thus given any number L ∈ (0. . √ 6 2 = 8. . which is deﬁned as the largest integer less than or equal to x: x = min{z ∈ Z  x < z} − 1. . 6 In fact. and it is bounded below (by −n2 ∈ Z). Show that if a sequence is Cauchy. it follows that (an )n∈N has a convergent subsequence. it follows that (an )n∈N is bounded. . 1). a3 = 0. . Using BolzanoWeierstrass theorem. then it is convergent. ♦ Exercise. 1). (By the Archimedean property. n1 belongs to it!). In this section we give the formal deﬁnition of continuous functions and prove two of the most important properties: the extreme value theorem and the intermediate value theorem. .485281 . √ 5 2 = 7. Many bizarre functions make appearances in analysis.242640 . we can prove the converse. for n ∈ N. Consider the sequence (an )n∈N of fractional parts of integral multiples of by √ √ an = n 2 − n 2 . . . a5 = 0. . . it can be shown that these fractional parts a are “dense” in (0. .
3. What does it mean for a function f : R → R to be continuous? The common informal deﬁnition of this concept states that a function f is continuous if one can sketch its graph without lifting the pencil. that is.9. we ﬁrst give the formal deﬁnition of the continuity of a function at a point below.3.1 Deﬁnition of continuity In everyday speech. then given any > 0 (which determines a strip around the line y = f (c) of width 2 ). then this break will occur at some point. Thus (based on this visual view of continuity). If a function has a break at a point.8: A function with a break at c. PSfrag replacements f (c) + f (c) f (x) f (c) + c−δ x c c+δ Figure 1. A function f : I → R is continuous at c if for every > 0. say c.8. by choosing x suﬃciently close to c. then even if points x are close to c. then f (x) is not close to f (c). Deﬁnitions. Let I be an interval in R and let c ∈ I. f (x) − f (c) < . no matter how close x comes to c. x − c < δ). If the function is continuous at c. 1. that is. See Figure 1. Next. f : R → R given by f (x) = 1 for all x ∈ R is continuous. which guarantees that if a function is continuous at a point c. the graph of f has no breaks in it. a ‘continuous’ process is one that proceeds without gaps of interruptions or sudden changes.9: The deﬁnition of the continuity of a function at point c. then it will be called continuous. See Figure 1. If x lies to the left of c. f (c) PSfrag replacements c Figure 1. there exists a δ > 0 such that for all x ∈ I satisfying x − c < δ. then we can make f (x) as close as we like to f (c). then f (x) satisﬁes f (c) − < f (x) < f (c) + . f is continuous at c. the points f (x) do not get close to f (c). A function f : I → R is continuous (on I) if for every c ∈ I. if a function is continuous at each point.1. In other words. there exists a δ > 0 (which determines an interval of width 2δ around the point c) such that whenever x lies in this width (so that x satisﬁes c − δ < x < c + δ. . Examples. f (x) − f (c) < . Continuity 27 1. If a break does occur in the graph. This motivates the following deﬁnition of continuity.
∞). Given let δ = c > 0. So f cannot be continuous at 0. Thus f (x) = f 2 = 1 = 1 > 2 = . let δ be an arbitrary positive real number. for instance. Hence if x ∈ R and x − c < δ. But now take x = 2 ∈ R.10: The continuous constant function 1. Then if x ∈ R and x − c < δ. However. it follows that f is continuous on R. See Figure 1. let δ > 0 be such that if x ∈ R and 1 δ x = x − 0 < δ. for all c ∈ R \ {0}. Since the choice of c ∈ R was arbitrary. Let c ∈ R. 1 PSfrag replacements 0 x Figure 1. Analysis Let c ∈ R = (−∞. See Figure 1. 2 Thus x = 0 and so f (x) = 1. f : R → R given by f (x) = x for all x ∈ R is continuous. a contradiction. Since the choice of c ∈ R was arbitrary. Given > 0. f : R → R given by f (x) = is not continuous at 0. Then if x ∈ R and x − c < δ = 1. f is continuous at c. . c > 0. it follows that f is continuous on R. Then if x ∈ R and x − c < δ.11. So f is continuous at c. Given > 0. So f is continuous at c. we obtain f (x) − f (c) = 1 − 1 = 0 = 0 < .11: A function continuous everywhere except at 0. 2. 1 PSfrag replacements 0 x Figure 1. we have: f (x) − f (c) = x − c < δ = . Consequently f is continuous at c. 1 Suppose that it is continuous at 0. 0 if x = 0. Then given = 2 > 0. δ δ δ 1 and so x = 2 = 2 < δ. 1 if x ∈ R \ {0}. we have: f (x) − f (c) = 1 − 1 = 0 = 0 < . we have 2 c − x ≤ c − x ≤ c − x = x − c < δ = and so x > c 2 > 0. let δ = .10.28 Chapter 1. 3. then f (x) − f (0) = f (x) − 0 = f (x) < = 2 . let δ = 1. This can be seen as follows.
the function f is continuous at c by showing that the same δ works (for this ). 4. 5. 1) and x − c < δ. ♦ Exercises. ∃δ > 0 such that for all x ∈ R satisfying x − c < δ. f (c) > 0. Suppose that f : R → R and there exists a M > 0 such that for all x ∈ R. let δ = min (> 0). Let f : R → R be deﬁned by f (x) = 0 if x is rational. In Exercise 1 on page 31. f is not continuous at c. Let f : R → R be a continuous function. Hint: Find f (0). Show that given any other point c ∈ R. f : (0. x c x c x c c c c So f is continuous at c. (a) Prove that f is continuous at 0. 1) was arbitrary. 1) and x − c < δ. 2 4. Let the function f : R → R be given by f (x) = x2 . 1 if x is irrational. 1. Given we have and so > 0. If f and g are functions on R. c − x 1 1 x − c 2 1 2δ = − = <δ· · = 2 ≤ . < . denoted by g ◦ f . c + δ). Prove that f is continuous at 0. Prove that f is continuous on R. (b) Give an example of such a function. Continuity 1 x 29 for all x ∈ R is continuous (on (0. Prove that if for some c ∈ R. 1)). then there exists a δ > 0 such that for all x ∈ (c − δ. Then if x ∈ (0. by The following theorem implies that the composition of continuous functions is continuous. f (x) − f (c) < . given > 0. c 2 c − x ≤ c − x ≤ c − x = x − c < δ ≤ c 1 2 < x. Hint: Use the fact that there are irrational numbers arbitrarily close to any rational number and rational numbers arbitrarily close to any irrational number. Hint: Since f is continuous at c. . f (x) ≤ M x. if x ∈ (0. 1)). then the composition of g with f . 1). 3. f (x) > 0. x ∈ R. is deﬁned (g ◦ f )(x) = g(f (x)). it follows that f is continuous (on (0. Prove that for every c ∈ R. (b) (∗) Suppose that c is a nonzero real number.3. we will give a slick proof of the fact that f is continuous on R. that is. 1) → R given by f (x) = Let c ∈ (0. (a) Suppose that f is continuous at some real number c. c c2 2. 2. 2 x c Consequently. Prove that f is continuous at c. y ∈ R. Let f : R → R be a function that satisﬁes f (x + y) = f (x) + f (y) for all x. Since the choice of c ∈ (0.1.
for all x ∈ R satisfying x − c < δ1 . 1. . Assume that f is not continuous at c.3. If: Suppose that (1. we have f (x) − f (c) < δ. (g ◦ f )(x) − (g ◦ f )(c) < . ∃ > 0 such that ∀δ > 0 ∃x ∈ I such that x − c < δ but f (x) − f (c) ≥ . Hence g ◦ f is continuous at c. (f (xn ))n∈N is convergent and lim f (xn ) = f (c).1 Let c ∈ R. ¬ [∀ > 0 ∃δ > 0 such that ∀x ∈ I such that x − c < δ. > 0. Analysis Theorem 1. ∃N ∈ N such that for all n > N .2 Let I be an interval in R and let c ∈ I. n→∞ (1. f (xn ) − f (c) < . If f : R → R is continuous at c and g : R → R is continuous at f (c). Then we need to show that f is continuous at c and we prove this by contradiction.11) holds. Theorem 1.30 Chapter 1.3. then we can ﬁnd xn ∈ I such that xn − c < δ = 1 n. xn − c < δ. ∃δ > 0 such that for all x ∈ I satisfying x − c < δ.3. Since (xn )n∈N is convergent with limit c. then g ◦ f is continuous at c. Consequently for n > N . ∃δ1 > 0 such that for all x ∈ R satisfying x − c < δ1 . given > 0. that is. f (x) − f (c) < ] that is. Hence if δ = 1 n. that is. Proof Since g is continuous at f (c). it follows that given satisfying y − f (c) < δ. ∃δ > 0 such that for all y ∈ R Since f is continuous at c. So (f (xn ))n∈N is convergent with limit f (c). f (x) − f (c) < . but f (xn ) − f (c) ≥ . g(y) − g(f (c)) < . f (x) − f (c) < δ. Consequently.2 Continuous functions preserve convergent sequences We now give an alternative characterization of continuity. Claim 1: The sequence (xn )n∈N is contained in I and is convergent with limit c.11) Proof Only if: Suppose that f is continuous at c ∈ I and let (xn )n∈N be a convergent sequence contained in I with limit c. Since f is continuous at c ∈ I. Suppose that f : I → R is a function. Then f is continuous at c iﬀ for every convergent sequence (xn )n∈N contained in I with limit c. and so g(f (x)) − g(f (c)) < .
. 7 See Exercise 4 on page 29. f  : I → R by f (x) = f (x).11) does not hold. x ∈ I. 3. Exercises. Continuity 31 Indeed. If for all x ∈ I. Prove that f is onetoone.3. 2. x2 be distinct real numbers. we have f (xn ) − f (c) ≥ . . So (xn )n∈N is convergent with limit c. xn ∈ I. but it is not possible to ﬁnd a large enough N ∈ N such that for all n > N . 3. Prove that7 if f : R → R is continuous and f (x) = 0 if x is rational. N < ζ. x2 . Suppose that f : I → R and g : I → R are continuous at c. f (xn ) − f (c) < 2 (for if this were possible. (f (xn ))n∈N is divergent as well. a contradiction. we deﬁne the following: 1. Theorem 1. xn − c < N < ζ.3. x1 . Using the characterization of continuity provided in Theorem 1. . there exists a sequence of rational numbers (qn )n∈N that converges to c. f + g is continuous at c. 1. 3. then we deﬁne the kth power of f . 2. we can ﬁnd N ∈ N such 1 1 1 that ζ < N (Archimedean property). f k : I → R by f k (x) = (f (x))k . Recall Exercise 1 on page 29: The function f (x) : R → R is given by f (x) = x2 . given any ζ > 0. Hint: Given any real number c. Using the above theorem. that is. g(x) = 0. prove that f is continuous on R. and consider the sequence x1 . then we deﬁne 1 g : I → R by 1 g (x) = 1 g(x) . Furthermore.3. x ∈ I. Given functions f : I → R and g : I → R. 2. αf is continuous at c. Hence for n > N . we have for all n ∈ N. 6. f + g : I → R is deﬁned by (f + g)(x) = f (x) + g(x).1.2. Hint: Let x1 . If α ∈ R. then f (x) = 0 for all x ∈ R. f  is continuous at c. The Claims 1 and 2 show that (1. The sum of f and g. we introduce some convenient notation.3. The product of f and g. Let I be an interval in R. then we would arrive at the contradiction ≤ f (xn ) − f (c) < 2 ). Indeed for all n ∈ N. we obtain the following useful Theorem 1. then we deﬁne the function αf : I → R by (αf )(x) = α · f (x). Then: 1. Let f : R → R be a function that preserves divergent sequences. x ∈ I.3. We deﬁne the absolute value of f . x ∈ I. f g : I → R is deﬁned by (f g)(x) = f (x)g(x). But before we state this result. . Thus for instance 2 > 0. Claim 2: The sequence (f (xn ))n∈N does not converge to f (c). If k ∈ N. 5. For all α ∈ R. x2 .3 Let I be an interval in R and let c ∈ I. . 4. x ∈ I. Hence f is continuous at c. that is. for every divergent sequence (xn )n∈N . x ∈ I.
g(x) = 0. Thus given arbitrary scalars a0 . If for all x ∈ I. in particular g(xn ) = 0 is convergent with limit 1 g and g(c) = 0).2 . that is. it follows that f k is continuous at c. xk is continuous. it follows that (f (xn ))n∈N and (g(xn ))n∈N are convergent with limits f (c) and g(c). Since f : R → R given by f (x) = x for x ∈ R is continuous (see Example 3 on page 28). So from Theorem 1. ((f g)(xn ))n∈N is convergent with limit (f g)(c).3. (f k (xn ))n∈N is convergent with limit f k (c). 6. Show that the rational function f : R → R deﬁned by f (x) = is continuous on R. 6. x2 . . 1 g(xn ) n∈N is convergent with limit 1 g 1 g(c) n∈N (since for all x ∈ I.2 . So from Theorem 1. Hence from Theorem 1. x ∈ R.3. respectively.2 .12. aN in R. that is. Analysis is continuous at c. (f (xn ))n∈N is convergent with limit f (c). aN · xN are continuous. . Since f and g are continuous at c. . ((f +g)(xn ))n∈N is convergent with limit (f + g)(c). So from Theorem is continuous at c. 5. . 3.32 4. that is. b] attains its maximum and minimum values. it follows that αf is continuous at c. 2. that is. with limit c. g(x) = 0.2. ((f (xn ))k )n∈N is convergent with limit (f (c))k . then 1 g Chapter 1. 4. f g is continuous at c. f k is continuous at c.2 . (f (xn )g(xn ))n∈N is convergent with limit f (c)g(c).2 . So from Theorem 1.3 Extreme value theorem Below we show that a continuous function on an interval [a. (f (xn )+g(xn ))n∈N is convergent with limit f (c)+g(c). For all k ∈ N. So from Theorem 1. from Theorem 1. See Figure 1. it follows that f  is continuous at c.3. x ∈ R is continuous.3. 1. ((αf )(xn ))n∈N is convergent with limit (αf )(c). it follows that 1 g (xn ) (c). . a1 . it follows that f + g is continuous at c. So from Theorem 1. a1 · x.4. that is. 1 + x2 ♦ 1. it follows that: 1. it follows that f g is continuous at c. Exercise. . (α · f (xn ))n∈N is convergent with limit α · f (c).2 . Example. . 5. it follows that the functions a0 · 1. (f (xn ))n∈N is convergent with limit f (c).3.2 .3.3. .3. that is. Proof Suppose that (xn )n∈N is a convergent sequence contained in I. . Consequently the polynomial function p : R → R deﬁned by p(x) = a0 + a1 x + · · · + aN xN . it follows that for all k ∈ N.
then there exist c.3. M − n < M . Similarly. b]}. So ∃xn ∈ [a.12) and (1. From (1.3.15). b].2 it follows that (f (xnk ))k∈N cannot be convergent. then f (x) ≤ N for all x ∈ [a. xnk − L < a − L. we obtain L − xnk ≤ L − xnk  = xnk − L < L − b. b]} is bounded. But then for all k > K. ∃xn ∈ [a. (1. We ﬁrst show that f is bounded. Step 2. xnk > b. a contradiction). n (1. and so ∃K ∈ N such that for all k > K. b]. and so S would be bounded. which contradicts (1. that is.14) (xn )n∈N is bounded. f (xnk ) > nk !). xnk − L < L − b. We prove that there exists a c ∈ [a. b].13) are in fact the maximum and minimum. But then for all k > K.12) (1. b] → R is continuous. d ∈ [a. Let M := sup{f (x)  x ∈ [a. Proof Step 1. b] such that 1 1 f (c) = M . and so by the BolzanoWeierstrass theorem (Theorem 1. then L − b > 0. b]}. Thus L ∈ [a. Continuity 33 PSfrag replacements d a c b Figure 1.1. respectively. that is. which contradicts (1. and so ∃K ∈ N such that for all k > K.12: Extreme value theorem.15) By the BolzanoWeierstrass theorem.14). b] such that M− 1 < f (xn ) ≤ M. if L < a. b]. Theorem 1. If f : [a.14). b]. (1.2.3. Suppose that S is not bounded. From Theorem 1. then either L > b or L < a. d ∈ [a.4 (Extreme value theorem). Since f is continuous (f (xnk ))k∈N is convergent with limit f (c). that is. we obtain xnk − L ≤ xnk − L < a − L. xnk < a. In this way. then a − L > 0. So S must be bounded. b] such that f (d) f (c) = sup{f (x)  x ∈ [a. since it is not bounded (indeed. (xn )n∈N has a convergent subsequence (xnk )k∈N with limit c ∈ [a. the set S = {f (x)  x ∈ [a. Then given n ∈ N. b]}. we see that this contradicts the continuity of f . and so M − n cannot be an upper bound for {f (x)  x ∈ [a. we get a sequence (xn )n∈N . it follows .8). If not. b]}. it follows that it has a convergent subsequence (xnk )k∈N that converges to some limit L. But by Theorem 1. Now we show that L ∈ [a. If L > b. and = inf{f (x)  x ∈ [a. Since a ≤ xn ≤ b for all n ∈ N. the supremum and inﬁmum in (1. b] such that f (xn ) > n (for if this fails for some N ∈ N. For each n ∈ N.2.13) Note that since c.2.
then given any number between 0 and 1976. but f satisﬁes the conclusion of Theorem 1. b] such that (1. then there exists a c ∈ [a. b] such that f (d) = m is similar to Step 2 above. Give an example of a function f : [0. Let f : [a. If f : R → R is continuous and periodic. b] is replaced by (a. This proves the existence of c ∈ [a. Examples. 1] → R is given by f (x) = x − x2 . f (x + T ) = f (x).34 Chapter 1. Step 3. See Figure 1. 1] is continuous and for any c.4 does not hold if [a. 3. 1. 1] we have f (c) = 1 = sup{f (x)  x ∈ [0. The proof that there exists a d ∈ [a. (b) (∗) Prove that f∗ is continuous on [a. b). 1] → R that is not continuous on [0. 1. 1]} = 1 = f (d). b]. 4. 1)}. b] such that f (c) = y. b]}. . Complete the details in Step 3 of the proof of Theorem 1. ♦ Exercises. The function f : (0. there must exist a point on the mountain that is exactly 399 meters above sea level. then sup S = 1 and inf S = 0. 1) → R given by f (x) = x is continuous on (0. (c) If f : [−1. If f : [a. then prove that f is bounded. A function f : R → R is periodic if there exists a T > 0 such that for all x ∈ R.3.4. Thus the statement of Theorem 1. say 399.3. and deﬁne f∗ as follows: f∗ (x) = f (a) if x = a. d ∈ [0. but these values are not attained. 1). The function f : [0.3. 1. 1]} = inf{f (x)  x ∈ [0. if the height of a mountain is 1976 meters above sea level. Analysis from the Sandwich theorem that f (c) = M .3.12) holds. x]} if x ∈ (a. (a) Show that f∗ is a welldeﬁned function. b]. Let m := inf{f (x)  x ∈ [a.4. Theorem 1. If S = {x  x ∈ (0. 1] → R given by f (x) = 1 for all x ∈ [0. b] → R is continuous and y is such that f (a) ≤ y ≤ f (b). max{f (y)  y ∈ [a.5 (Intermediate value theorem). b]. For instance.13.3. 2. and is left as an exercise to the reader.4 Intermediate value theorem We now prove one of the most fundamental (and obvious!) theorems on continuous functions: a continuous function cannot “hop over” intermediate values. that is. b] → R be continuous on [a. the set S = {f (x)  x ∈ R} is bounded. 1]. then ﬁnd f∗ . 2.
2 not. then replace δ by δ N with N ∈ N large enough to guarantee δ c−a < N. b]: indeed. b] such that x − c < δ. 0 ≤ x − c = x − c < δ and y ≤ f (x). b] such that (−f )(c) = −y. f (c) = y. S is not empty (since b ∈ S) and S is bounded below by a (since S ⊂ [a. then we have (−f )(a) ≤ −y ≤ (−f )(b). x ∈ S and since c = inf S. say c. Thus := y − f (c) > 0. Proof Suppose that f (a) ≤ y ≤ f (b). Consequently. and since f is continuous at c. x − c = c − x = c − c − c+a δ ≤ c − ≤ c ≤ b. if y ∈ R is such that f (b) ≤ y ≤ f (a). Note that in the above statement of the theorem. a contradiction. ∃δ > 0 such that for all x ∈ [a. 2 Hence f (c) − f (x) ≤ f (c) − f (x) = f (x) − f (c) < f (c) − y. Then either f (c) < y or f (c) > y.1 with α = −1!). b]).3. Indeed. Thus S has an inﬁmum.1. that is.3. 2 2 δ 2 = δ < δ. Thus := f (c) − y > 0. f (x) − f (c) < y − f (c). Continuity 35 f PSfrag replacements(b) y f (a) a c b Figure 1. Let x := c − δ . Thus we obtain c≤x =c− a contradiction. and deﬁne S = {x ∈ [a. Since c is the inﬁmum of S. we get the existence of c ∈ [a. 1◦ Suppose that f (c) < y. Suppose that our claim is false.3. and applying the above Theorem 1. b]  y ≤ f (x)}. 2 a≤ Furthermore. b]. ∃δ > 0 and moreover satisfying8 δ ≤ c − a . such that for all x ∈ [a. the order f (a) ≤ y ≤ f (b) can be reversed without changing the conclusion. c] ⊂ [a.3.13: Intermediate value theorem.5 to the continuous function −f (Theorem 1. Then x ∈ [a. 8 If δ < c. . b] such that x − c < δ. ∃x ∈ S such that c ≤ x < c + δ. that is. We claim that f (c) = y. Thus we have y − f (c) ≤ f (x) − f (c) ≤ f (x) − f (c) < y − f (c). and so f (x) ≥ y. and since f is continuous at c. x ∈ [a. 2◦ Suppose that f (c) > y. f (x) − f (c) < y − f (c). we have c ≤ x.
n→∞ Since lim a0 2m+1 n + a2m+1  p(n) − a2m+1 < . n2m+1 2 Thus p(n) is also negative for all n > N1 . 2m+1 n n n 2 2 0< p(n) a2m+1 < 2m+1 . p(c) = 0. Every odd polynomial with real coeﬃcients has at least one real root. Since a2m+1 = 0. p(−n) has the same sign as −a2m+1 . N ]. N ]. then for all n > N1 a2m+1 − and so p(n) a2m+1  p(n) a2m+1 p(n) ≤ a2m+1 − 2m+1 = 2m+1 − a2m+1 < = . if a2m+1 < 0. we can conclude that p must vanish at some point in this interval. that is. Suppose that p is a polynomial with degree 2m + 1. p(−n) has the same sign as −a2m+1 . p(n) has the same sign as a2m+1 . then for all n > N1 a2m+1  p(n) a2m+1 p(n) − a2m+1 ≤ 2m+1 − a2m+1 < =− . Examples. Chapter 1. The proof is long. So it follows that for n > N1 . For large positive n. Let N2 ∈ N be such that for all n > N2 . Similarly. p(−n) p(−n) a2m+1  + a2m+1 = − 2m+1 − a2m+1 < . where a2m+1 = 0. for some real c ∈ [−N. Step 1. m ∈ N ∪ {0}. we will choose a large enough N ∈ N such that p(N ) and p(−N ) have opposite signs. Then appealing to the intermediate value theorem. and this completes the proof. using the fact that n→∞ we now show that there exists N2 ∈ N such that for all n > N2 . and so we have divided it into a sequence of steps. lim a1 a2m a0 + +··· + 2m+1 2m (−n) (−n) (−n) = 0. 1. If a2m+1 > 0. a1 a2m = 0. Analysis In order to show that p has a real root. and so Similarly. Step 2. 2m+1 n 2 Now we show that for n > N1 . n2m+1 n 2 .36 So f (c) = y. For large positive n. p(n) has the same sign as a2m+1 . p(n) has the same sign as a2m+1 . +···+ 2m n n it follows that there exists N1 ∈ N such that for all n > N1 . either a2m+1 > 0 or a2m+1 < 0. n2m+1 n 2 2 a2m+1 p(n) < < 0. and then restrict our attention to an interval [−N. 2 n Thus p(n) is also positive for all n > N1 . Let p(x) = a0 + a1 x + · · · + a2m x2m + a2m+1 x2m+1 .
while p(−N ) has the same sign as −a2m+1 . If f (0) = 0. N ]. by the intermediate value theorem. Thus the continuous function p must vanish a some point in the interval [−N. n2m+1 n 2 2 p(−n) a2m+1 <− < 0. Let T (Θ) denote the surface temperature at the point at longitude Θ.14: The point on the equator with longitude Θ. Step 3. If f (0) = 0. 1]. See Figure 1. So it follows that for n > N2 .1. Similarly. 0 ≤ f (x) ≤ 1. 1] such that f (c) = c. Suppose that f : [0. Continuity 37 If a2m+1 > 0. N2 } + 1. 1 399 >0 2. 1. 2m+1 n 2 Thus p(−n) is positive for all n > N2 . and so the claim follows. 1] → R is a continuous function such that for all x ∈ [0. p(−n) has the same sign as −a2m+1 . then since f (0) and f (π) = −f (0) have opposite signs. n2m+1 2 Thus p(−n) is negative for all n > N2 . (Note that Θ(0) = Θ(2π).) Assuming that Θ is a continuous function of Θ. 1 The polynomial p(x) = x2005 − x1976 + 399 has a real root in [−1. Application of the intermediate value theorem and the conclusion. 1]: indeed p(1) = 1 and p(−1) = −2 + 399 < 0 and so ∃c ∈ [−1. it follows that Θ PSfrag replacements Figure 1. then for all n > N2 − and so p(−n) p(−n) a2m+1 a2m+1  − a2m+1 ≤ 2m+1 + a2m+1 < =− . there exists a pair of diametrically opposite points on the equator which have the same temperature. At any given time. . Hence if N := max{N1 . it follows that f must vanish at some point. ♦ Exercises. Hint: Consider the continuous function g(x) = f (x) − x.3. and use the intermediate value theorem.14. then for all n > N2 a2m+1  p(−n) a2m+1 p(−n) + a2m+1 ≤ 2m+1 + a2m+1 < = . then it follows that the temperatures at 0 and 180◦ longitude are the same. 2m+1 n n 2 2 and so a2m+1 p(−n) >− > 0. then p(N ) has the same sign as a2m+1 . 1] such that p(c) = 0. π] → R deﬁned by f (Θ) = T (Θ) − T (Θ + π) is continuous as well. if a2m+1 < 0. Prove that there exists at least one c ∈ [0. the function f : [0.
38
Chapter 1. Analysis
2. At 8:00 a.m. on Saturday, a hiker begins walking up the side of a mountain to his weekend campsite. On Sunday morning at 8:00 a.m., he walks back down the mountain along the same trail. It takes him one hour to walk up, but only half an hour to walk down. At some point on his way down, he realizes that he was at the same spot at exactly the same time on Saturday. Prove that he is right. Hint: Let u(t) and d(t) be the position functions for the walks up and down, and apply the intermediate value theorem to f (t) = u(t) − d(t). 3. Show that the polynomial function p(x) = 2x3 − 5x2 − 10x + 5 has a real root in the interval [−1, 2]. 4. Let f : R → R be continuous. If S := {f (x)  x ∈ R} is neither bounded above nor bounded below, prove that S = R. Hint: If y ∈ R, then since S is neither bounded above nor bounded below, there exist x0 , x1 ∈ R such that f (x0 ) < y < f (x1 ). 5. (a) (∗) Show that given any continuous function f : R → R, there exists a x0 ∈ [0, 1] and a m ∈ Z such that f (x0 ) = mx0 . In other words, the graph of f intersects some line y = mx at some point x0 in [0, 1]. Hint: If f (0) = 0, take x0 = 0 and any m ∈ Z. If f (0) > 0, then choose N ∈ N satisfying N > f (1), and apply the intermediate value theorem to the continuous function g(x) = f (x) − N x on the interval [0, 1]. If f (0) < 0, then ﬁrst choose a N ∈ N such that N > −f (1), and consider the function g(x) = f (x) + N x, and proceed in a similar manner. (b) (∗) Prove that there does not exist a continuous function f : R → R such that assumes rational values at irrational numbers, and irrational values at rational numbers, that is, f (Q) ⊂ R \ Q and f (R \ Q) ⊂ Q. Hint: Note that for every m ∈ Z, there does not exist a x0 ∈ R such that f (x0 ) = mx0 .
Chapter 2
Algebra
In this part of the course, we study two important algebraic objects: groups and vector spaces. We begin with a discussion about groups.
2.1
Groups
In this section, we study one of the most basic algebraic objects, namely a group. A group is a set on which a law of composition is deﬁned, such that certain properties hold. The precise deﬁnition is given in the next subsection.
2.1.1
Deﬁnition of a Group
By a law of composition on a set S, we mean a rule for combining pairs a, b of S to get another element, say c, of S. We denote the set of all ordered pairs of elements from a set S by S × S, that is, S × S = {(a, b)  a, b ∈ S}. Deﬁnition. A law of composition on a set S is a function f : S × S → S. The functional notation c = f (a, b) is not very convenient for what is going to follow, and so instead, the element obtained by applying the law of composition to a pair (a, b) is usually denoted using a notation resembling that used for addition or multiplication: c = a ∗ b, or ab, or a ◦ b, or a + b and so on,
with a ﬁxed choice being made for the particular law in question. Examples. 1. The addition of integers is a law of composition on Z. Indeed, the sum of two integers is yet another integer, and addition is the function from the set Z × Z to the set Z that assigns a + b to the pair (a, b), denoted by (a, b) → a + b. 2. The multiplication of real numbers is a law of composition on R. 39
40
Chapter 2. Algebra
3. If a, b are rational numbers, then let a ∗ b = a + b − ab. The function from Q × Q to Q given by (a, b) → a ∗ b is a law of composition on Q. √ a 4. If a, b are real numbers, then deﬁne a ∗ b = √ 2 + b2 . Then (a, b) → a ∗ b is not a law of composition on Q, since 1 ∈ Q, but 1 ∗ 1 = 2 ∈ Q. However, (a, b) → a ∗ b is a law of composition on R. 5. Let n ∈ N, and let S denote the set of all matrices of size n × n with real entries. Then matrix multiplication is a law of composition on S. 6. Let n ∈ N, and let GL(n, R) denote the set of all invertible matrices of size n × n with real entries. Then matrix multiplication is a law of composition on GL(n, R). Indeed, if A, B ∈ GL(n, R), then the matrix AB is again a matrix of size n × n with real entries, and moreover, since A and B are invertible, it follows that AB is also invertible. 7. Let a, b be real numbers such that a < b. Let C[a, b] denote the set of all continuous functions on the interval [a, b]. Let addition of functions be deﬁned as follows: if f, g belong to C[a, b], then (f + g)(x) = f (x) + g(x), x ∈ [a, b]. Then addition of functions is a law of composition on C[a, b], since the sum of continuous functions is again continuous; see Theorem 1.3.3. 8. If a, b ∈ N, then let a∗b= a . b
1 2
∗ is not a law of composition on N, since 1 ∗ 2 =
∈ N.
♦
On a set there may be several diﬀerent laws of compositions that can be deﬁned. Some laws of compositions are nicer than others, that is, they possess some desirable properties. A group is a set G together with a law of composition on G that has three such desirable properties, and we give the deﬁnition below. Deﬁnition. A group is a set G together with a law of composition (a, b) → a ∗ b : G × G → G, which has the following properties: G1. (Associativity.) For all a, b, c ∈ G, (a ∗ b) ∗ c = a ∗ (b ∗ c). G2. (Identity.) There exists an element1 e ∈ G such that for all a ∈ G, a ∗ e = a = e ∗ a. G3. (Inverses.) For every a ∈ G, there exists an element2 a−1 ∈ G such that a∗a−1 = e = a−1 ∗a. G1, G2, G3 are called group axioms. Sometimes we use the notation (G, ∗) for the group. Remarks. Note that hidden in the deﬁnition of a group, is the following axiom G0: G0. For all a, b ∈ G, a ∗ b ∈ G. (That is, ∗ is actually a law of composition on G.) Hence when checking that a certain set G is a group with respect to a certain operation ∗ that combines pairs of elements from G, we have to check, ﬁrst of all, that for every a, b ∈ G, a ∗ b belongs to G.
1 Such 2 depending
an element e is called an identity element of the group G. on a, and such an element a−1 is called an inverse of the element a in the group G.
. R). b] we have (f + (g + h))(x) = f (x) + (g + h)(x) = f (x) + (g(x) + h(x)) = (f (x) + g(x)) + h(x) (since addition is associative in R!) = (f + g)(x) + h(x) = ((f + g) + h)(x).. b] is a group with addition of functions. serves as an identity element. the set of nonzero real numbers. B. e = 1. but the group axiom G3 does not hold: G1. since matrix multiplication is associative. 0a−1 = a−1 0 = 0 = 1 = e. G2.. or the set of positive real numbers are both groups with multiplication. 0 0 . The identity matrix G1. For all a. and let GL(n. we should have 1e = 1. 1.2. b.. the addition of functions is a law of composition on C[a. R). 0 0 . R) that satisﬁes the group axioms: G2. Indeed. b be real numbers such that a < b. (AB)C = A(BC). This group is called the general linear group. and so it belongs to GL(n. Groups 41 Examples. In = 1 0 . a + 0 = a = 0 + a. g. R). Then GL(n. h ∈ C[a. and so. . . and so there exists a matrix A−1 such that AA−1 = In = A−1 A. R with multiplication is a not a group. The matrix A−1 ∈ GL(n. . R). b] that satisﬁes the group axioms: G1. 2. since 0 ∈ R. Let n ∈ N. For all f. (ab)c = a(bc). 3. (If e is an identity element. . Hence f + (g + h) = (f + g) + h. with a = 1. 1 G3. for all A ∈ GL(n. C ∈ GL(n. matrix multiplication is a law of composition on GL(n. For all A. c ∈ R. since in addition to G1 and G2. If a ∈ Z. However. So there is no inverse of the element 0 ∈ R. AIn = A = In A. G3. R) denote the set of all invertible matrices of size n × n with real entries. . now in each case G3 also holds.. and any x ∈ [a. (a + b) + c = a + (b + c). . G2. . Then C[a. Z with addition is a group. 4.1. then we must have ae = a = ea for all a ∈ R. For all a. b]. although the multiplication of real numbers is a law of composition on R that satisﬁes G1 and G2. Let a. If A ∈ GL(n. Indeed. R) and serves as an inverse of A. c ∈ Z. Indeed. 0 1 . b. then −a ∈ Z and a + (−a) = 0 = −a + a. the addition of integers is a law of composition on Z that satisﬁes the group axioms: G1. then it must be equal to 1!) 1 serves as an identity element: for all a ∈ R. . and moreover. and in particular. 0 serves as an identity element: for all a ∈ Z.. but for all a−1 ∈ R. G3. . R) is group with matrix multiplication. a1 = a = 1a. And so if e is an identity element. Does not hold. .. then A is an invertible matrix. In is an invertible matrix of size n × n with real entries.. Indeed.
Given f ∈ C[a.42 Chapter 2. for x ∈ [a. ∗) is said to be abelian3 if for all a. b]. We note that the groups in the above Examples 1 and 4. Let n. m ∈ N. 1. a ∗ b = b ∗ a . then deﬁne −f by (−f )(x) = −f (x). A group (G. The set of integers Z with addition is an abelian group. while the group in Example 3 (with n = 2) does not satisfy G4: 1 1 0 1 1 0 1 1 = 2 1 1 1 = 1 1 1 2 = 1 0 1 1 1 1 0 1 . The constant function 0 deﬁned by 0(x) = 0 for all x ∈ [a. Moreover. b]. If f ∈ C[a. b] and so 0 ∈ C[a. b]: (f + (−f ))(x) = f (x) + (−f )(x) = f (x) + (−f (x)) = 0 = 0(x) = 0 = −f (x) + f (x) = (−f )(x) + f (x) = (−f + f )(x). we have for all x ∈ [a. Hence f + 0 = f = 0 + f .) For all a. b]. G3. b]: (f + 0)(x) = f (x) + 0(x) = f (x) + 0 = f (x) (since 0 is an identity element for addition in R!) = 0 + f (x) = 0(x) + f (x) = (0 + f )(x). Let n ∈ N. Examples. 0 is a continuous function on [a. 4. b]. ♦ 3 after the Norwegian mathematician Abel . Deﬁnition. 3. b]. The set of positive real numbers with multiplication is an abelian group. Algebra G2. The set GL(n. serves as an identity element. but it is not an abelian group if n > 1. b ∈ G. Hence f + (−f ) = 0 = −f + f . b ∈ G. R) with matrix multiplication is a group. 2. a ∗ b = b ∗ a. (Commutativity. also satisfy G4. for all f ∈ C[a. we have for all x ∈ [a. ♦ This gives rise to the following natural deﬁnition. b]. The set matrices Rn×m of size n × m with entries in R with matrix addition is an abelian group.
If a1 and a2 are inverses of a.2 Let (G. Proof Let e and e be identity elements in (G. it is easy to check that (In − BA)(In + B(In − AB)−1 A) = In = (In + B(In − AB)−1 A)(In − BA). B are matrices of size n × n with real entries and In − AB is invertible. R). 1} with multiplication is a ﬁnite group of order 2. ∗) is said to be ﬁnite.2. 2.1 There is a unique identity element in a group. it follows that (In − BA)−1 = In + B(In − AB)−1 A. Indeed. ∗) be a group and let a ∈ G. if the set G has ﬁnite cardinality. and then given a. e = e . ∗). Examples. The set Z with addition is an inﬁnite group. since e ∈ G and e is an identity. The order of a ﬁnite group (G.1.1. Groups 43 We now prove a few elementary theorems concerning groups. we also have e∗e =e. The set {−1. b ∈ G. Example. A group is said to be inﬁnite if it is not ﬁnite. we obtain e=e∗e. Theorem 2. then we have a1 = a1 ∗ e (since a1 ∈ G and e is the identity) = a1 ∗ (a ∗ a2 ) (since a2 is an inverse of a) = (a1 ∗ a) ∗ a2 (associativity) = e ∗ a2 (since a1 is an inverse of a) = a2 (since a2 ∈ G and e is the identity). 1. the element a ∗ b is entered in the row corresponding to a and the column corresponding to b. This is a table that displays the law of composition as follows: the elements of the group are listed in the ﬁrst row and the ﬁrst column. Then a has a unique inverse. then In − BA is also invertible. and (In − BA)−1 = In + B(In − AB)−1 A. ♦ Deﬁnitions. and so it follows that In − BA is invertible. If A. Moreover. as shown below. ∗) is the cardinality of G. By the uniqueness of the inverse of the element In − BA in the group GL(n. Proof Let the group have the identity e. Since e ∈ G and e is an identity. Theorem 2. A group (G.1. ♦ A ﬁnite group can be completely described by writing its group table. . Consequently.
. n}. . . (b) What is the order of Sn ? (c) If n = 3. . . then the equation a ∗ x = b has a unique solution. 1 · (−1) = −1. (a) Given n ∈ N. Show that S with the operation of matrix multiplication forms a group. . ∗) be a group. .. . (a) Show that the set Z6 of integers modulo 6. . 2. Hint: Composition of bijections is again a bijection. n} onto itself. Example. give examples of bijections f and g on {1. (d) (∗) Show that Sn is abelian iﬀ n ≤ 2. Prove that Z∗ is a group with multiplication modulo m iﬀ m is a prime m number. . 1. while the inverse of an element f ∈ Sn is simply the inverse of that bijection. m − 1} is coprime to m. 2. ♦ Exercises. Consider the set S= a b −b a a. . The identity element is simply the identity map ι : {1. ... we consider the set Z∗ of 6 nonzero integers modulo 6. 3. 3. then any r ∈ {1. (b) Is Z6 with multiplication modulo 6 also a group? If. . . and let Z∗ denote the set of nonzero integers m modulo m. then is Z∗ a group with multiplication modulo 6? 6 (c) (∗) Let m be an integer such that m ≥ 2. 3. · 1 −1 1 1 −1 −1 −1 1 The table completely describes the law of composition: 1 · 1 = 1. (b) Show that if a. and so there exist integers s and t such that sm + tr = 1. . 1} with multiplication can be described by the group table given below. . Write down its group table. . −1 · 1 = −1 and −1 · (−1) = 1. Hint: If m is a prime number. Algebra We clarify this further by considering a simple example. . Let (G. and composition of functions on a set is also associative. 4. 3} such that f ◦ g = g ◦ f . b a∗b . . instead. . 2. . . . with addition modulo 6 is a group. b ∈ R and a2 + b2 = 0 .44 ∗ . .. Chapter 2. 3. n} given by ι(m) = m for all m ∈ {1. . a . 3. b. n} → {1. let Sn be the set of all bijections from the set {1. b ∈ G. (c) Show that if a. b ∈ G. then b = c. Prove that the set Sn with the composition of functions is a group. (a) Show that if a. The ﬁnite group {−1. . 2. then (a ∗ b)−1 = b−1 ∗ a−1 . c ∈ G are such that a ∗ b = a ∗ c. 2. This is called the symmetric group Sn . 2.
given the even number 2m. ∃a−1 ∈ H such that a ∗ a−1 = e = a−1 ∗ a. Indeed the identity element from G (which also belongs to H) serves as the identity element also in H: for every a ∈ H. The group of integers with addition (Z. −2m = 2(−m) is even as well (and so H3 holds). H2. it follows that the induced law of composition is associative as well: for all a. Since ∗ is associative. is a group. then a ∗ b ∈ H. Thus. then a−1 ∈ H. which is called the induced law of composition. Thus the conditions H1. the function from H × H to H given by (a. b. with the induced law of composition. Thus for all a ∈ H. with the induced law of composition.1. Examples. and this is explained as follows: H1. given any two symmetric matrices a b b d and a b b d . then {e} and G are both subgroups of G. A subset H of a group G is called a subgroup of G. (Closure) If a. (Inverses) If a ∈ H. namely.2. H3. a subgroup is itself a group which sits in a larger group. H2. since G is a group. a ∗ e = a = e ∗ a. If G is a group with identity e. which in turn is a subgroup of the group of real numbers with addition (R. and ﬁnally. +) is a subgroup of the group of rational numbers with addition (Q. 4. b) → a ∗ b. namely a b b d a. +). H3. H3 imply that the subset H. 0 is even (and so H2 holds). Indeed. Finally this shows that every element in H possesses an inverse element in H. b ∈ H. The subset of even integers {2m  m ∈ Z} is a subgroup of the group of integers Z with addition. H2.2 Subgroups Deﬁnition. But now H3 says that this inverse element is in H. c ∈ H. b. if it has the following properties: H1. has an identity element. Indeed. (Identity) e ∈ H. These conditions ensure that H is itself a group which is contained in the group G. This shows that H. +). Of course. the sum of even numbers is even (and so H1 holds). a ∗ (b ∗ c) = (a ∗ b) ∗ c. we already knew that a possesses an inverse element a −1 ∈ G. 1. The subset of symmetric matrices of size 2 × 2. d ∈ R with real entries is a subgroup of the set of all 2 × 2 matrices having real entries with matrix addition. Groups 45 2. 3. .1. 2. This condition tells us that the law of composition ∗ on the group G can be used to deﬁne a law of composition on H.
(b) The odd integers form a subgroup of Z with addition. Is there an inﬁnite group with a ﬁnite subgroup? 3.46 their sum Chapter 2. (c) If G is abelian and H is a subgroup of G. if a b 0 d then a b 0 d a 0 b d . then H is abelian. Clearly ∈ U T (2. Algebra a b b d + a b b d = a+a b+b b+b d+d is also symmetric. b d is the element −a −b −b −d which is also symmetric. Indeed. 1. and so H2 also holds. the inverse (with respect to matrix addition) of any symmetric matrix a b . b. and so H3 holds. (a) The nonnegative integers form a subgroup of Z with addition. Suppose that H is the subgroup of Z comprising multiples of 4. R) = a b 0 d a. ab + bd dd = 1 0 0 1 ∈ U T (2. R). What is H ∩ K? . U T (2. a 0 b d aa 0 ∈ U T (2. and so H1 holds. R). Finally. d ∈ R and ad = 0 is a subgroup of the group GL(2. 2. Finally. Determine if the following statements are TRUE or FALSE. and so H2 also holds. 5. then a b 0 d −1 = 0 1 a b − ad 1 d ∈ U T (2. R). The subset of upper triangular invertible matrices. Clearly the identity element 0 0 0 0 is symmetric. R). (a) Consider the group of integers Z with addition. if a b 0 d ∈ U T (2. and so H1 holds. ♦ Exercises. R). and let K be the subgroup of Z comprising multiples of 6. R) with matrix multiplication.
denoted by ord(a). for all x ∈ [0. 5. Groups 47 (b) If H and K are subgroups of a group G. The element −1 has order 2 in the group of nonzero real numbers with multiplication. (f + g)(x) = f (x) + g(x). If a has ﬁnite order. R). (b) Let H2 denote the set of all polynomial functions. Let G be a group. 1]  f 1 2 = 0}. Let C[0. R) with matrix multiplication. 1]. . 1. 1]. Show that H2 is a subgroup of C[0.2. 1] → R such that there exists an n ∈ N ∪ {0} and real numbers a0 . 1. . Moreover. We deﬁne a0 = e and an = an−1 ∗ a for n ∈ N. . then a is said to have inﬁnite order. 3. 0 1 0 0 is an element of order 3 in the group GL(3. (Exercise!) Deﬁnitions. then we deﬁne a−n = (an )−1 . a1 . Let G be a group and let a ∈ G. g belong to C[0. 1]. then the order of a. the set of all functions p : [0. 1 0 ♦ . 1] with addition of functions deﬁned in the usual way: if f. . 1]. It can be shown that the usual laws of exponents hold: for all m. (b) If G is abelian. Examples. z ∗ a = a ∗ z}. (c) Show that Z(G) is a subgroup of G. The center of G is the set Z(G) = {z ∈ G  ∀a ∈ G. (a) Show that Z(G) is not empty. n ∈ Z. then determine Z(G). then for all x ∈ [0. The element 1 0 inﬁnite order in the group of integers with addition. If there exists an m ∈ N such that am = e.1. Let G be a group and suppose that a ∈ G. 2. that is. (d) (∗) If G is the group GL(2. then a is said to have ﬁnite order. The element 2 has 0 3. If a does not have ﬁnite order. . is ord(a) = min{m ∈ N  am = e}. 1] denote the group comprising the set of continuous functions on the interval [0. am ∗ an = am+n and amn = (am )n . 1 1 1 0 Hint: Take a = . 1]. if n ∈ N. a2 . an such that p(x) = a0 + a1 x + a2 x2 + · · · + an xn . then determine Z(G). 4. 2. then show that H ∩ K is also a subgroup of G. Prove that H1 is a subgroup of C[0. 0 1 1 1 (a) Let H1 = {f ∈ C[0.
a. Let a = {an  n ∈ Z}. . . 4. 2. a. am ∗an = am+n and (am )n = amn . Is the group of integers with addition cyclic? What is a generator of this group? Is it unique? . . which is at most equal to G.3 Suppose that G is a group and a ∈ G. H1. am−1 }. 5. a. . a2 . Determine the order of 1 1 −1 0 in the group GL(2. e = a0 ∈ a . For any m ∈ Z. . H2. clearly m + n ∈ Z and so am ∗ an = am+n ∈ a . n ∈ Z. Proof 1. then there exist integers q and r. and for any a. Prove that in any group (G. So a is a subgroup of G. a3 . a3 . . a. . H2. . am−1 }. −m ∈ Z and so (am )−1 = a−1·m = a−m ∈ a . . Conversely. R). . Algebra We now prove that given any element from a group. and n = q · m + r. 2. . This example motivates the following deﬁnition. Given m. Clearly {e.48 Chapter 2. . Then: 1. a2 . n ∈ Z. Theorem 2. the set of its powers form a subgroup of the group. Such an element a is then called a generator of the group G. a2 . aG }. [1]. and use the pigeonhole principle. . ∗). Exercises. then show that every element in the group has a ﬁnite order. 1. then a = {e. Hint: If a ∈ G. We prove that H1. . a3 . such that 0 ≤ r ≤ m − 1. . . if n ∈ Z. and the subgroup [1] = {[0]. Deﬁnition. If a is an element with ﬁnite order m. a is a subgroup of G. then show that for all m. [4]} is in fact the whole group. [3]. since [1] ⊕ [1] ⊕ [1] ⊕ [1] ⊕ [1] = [0]. H3. If G is a ﬁnite group. b in G. a3 . If G is a group and a ∈ G. H3 hold. then consider the set S = {e. am−1 } ⊂ a . a2 .1. 2. So an = aq·m+r = aq·m ∗ ar = (am )q ∗ ar = (e)q ∗ ar = e ∗ ar = ar ∈ {e. The element [1] in the group Z5 with addition modulo 5 has ﬁnite order. A group G is said to be cyclic if there exists an element a ∈ G such that G = a . 3. [2]. the orders of a ∗ b and b ∗ a are the same.
Let (G. g are continuous functions on the interval [0. 1] with addition.1. Let G be the group GL(2. and so canceling ϕ(e) on both sides. Examples. 3. Deﬁnitions. Groups 49 2. Then the determinant function det : G → G is a homomorphism. and G be the set of positive reals with multiplication. and G be the group of nonzero real numbers with multiplication. Let G. Then the function f → f 2 is a homomorphism. b ∈ G. by the 2 2 2 deﬁnition of f + g. B we have det(AB) = det(A) det(B). ♦ Thus a homomorphism is a function between two groups that respects the law of composition. then: 1. since for all 2 × 2 real matrices A. then the identity map ι : G → G deﬁned by ι(a) = a for all a ∈ G. then (ϕ(a))−1 = ϕ(a−1 ). ϕ(a ∗ b) = ϕ(a) ∗ ϕ(b). y. Next. then (f + g) 1 = f 1 + g 1 . In the next theorem we show that it preserves the identity element and the inverses of elements as well. Every group homomorphism ϕ determines two important subgroups: its image and its kernel.1. and 1 G be the group R with addition. given by x → 2x . .3 Homomorphisms and isomorphisms Deﬁnition. is a homomorphism. and by the uniqueness of the inverse of ϕ(a) in G . ∗) and (G . R) with matrix multiplication. we obtain (ϕ(a))−1 = ϕ(a−1 ). ϕ(a−1 ) ∗ ϕ(a) = ϕ(a−1 ∗ a) = ϕ(e) = e and similarly. Let G be the group C[0. If ϕ : G → G is a homomorphism. Theorem 2. 4. Let G be R with addition. If a ∈ G. ∗ ) be groups.1. 1] of continuous functions on the interval [0. we obtain e = ϕ(e). and the trivial map ζ : G → G deﬁned by ζ(a) = e for all a ∈ G are homomorphisms. The exponential function from G to G . 1. e = ϕ(e) = ϕ(a ∗ a−1 ) = ϕ(a) ∗ ϕ(a−1 ). A homomorphism ϕ : G → G is a function such that for all a. Thus ϕ(a−1 ) ∗ ϕ(a) = e = ϕ(a) ∗ ϕ(a−1 ). ϕ(e) = e .2. we have 2x+y = 2x 2y for all real x. G be groups and let ϕ : G → G be a group homomorphism. if f.4 Let G be a group with identity e and G be a group with identity e . 2. Indeed. 1]. Indeed. 2. Proof We have e ∗ ϕ(e) = ϕ(e) = ϕ(e ∗ e) = ϕ(e) ∗ ϕ(e). If G is a group.
Consequently. The image of this homomorphism is the whole group of nonzero reals: indeed. and G be the set of positive reals with multiplication.50 1. then ϕ(a ∗ b) = ϕ(a) ∗ ϕ(b) = e ∗ e = e . and so H1 holds. Then the function f → f 2 is a homomorphism. 3. is a homomorphism with kernel the trivial subgroup comprising the element 0. since if a ∈ ker(ϕ). im(ϕ) is a subgroup of G . Chapter 2. Its kernel is the set of all invertible matrices with determinant equal to 1. if a ∈ im(ϕ). and the image is the whole group of positive reals with multiplication. and f 1 = a. 1] with addition. Proof It is easy to check that ker(ϕ) is a subgroup of G. and G be the group of nonzero real numbers with multiplication. such that ϕ(a∗b) = a ∗ b . Hence ker(ϕ) is a subgroup of G. R) with matrix multiplication. Let G be the group GL(2. and its kernel is the set of all continuous functions on the interval [0. 1] of continuous functions on the interval [0.4. namely a∗b. since given any a ∈ R. b belong to ker(ϕ). we now prove the following result.1. a ∗ b ∈ im(ϕ). Moreover. b in G such that ϕ(a) = a and ϕ(b) = b . G be groups and let ϕ : G → G be a group homomorphism. ϕ(a∗b) = ϕ(a)∗ ϕ(b) = a ∗ b . b belong to im(ϕ). Using Theorem 2. we have that A := and det(A) = 1 · a − 0 · 0 = a. Examples. and so there exists an element in G. it follows that e ∈ im(ϕ). Hence a −1 ∈ im(ϕ). Then the determinant function det : G → G is a homomorphism. The kernel of ϕ is the set ker(ϕ) = {a ∈ G  ϕ(a) = e }. e ∈ ker(ϕ) and so H2 holds too. Then: 1. R) = {A ∈ GL(2. Let G be the group C[0. and so a −1 = (ϕ(a))−1 = ϕ(a−1 ). Let G be R with addition. and H3 holds. and we denote this subgroup by SL(2. and it is called the special linear group: SL(2. R)  det(A) = 1}. Finally. there exists a unique real number (called the logarithm of y to the base 2. ♦ 2 1 0 0 a ∈ GL(2. The exponential function from G to G . then ϕ(a−1 ) = (ϕ(a))−1 = (e )−1 = e .1. H3 holds. 2. . If a . that is.5 Let G. R). given by x → 2x . and so a−1 ∈ ker(ϕ). ker(ϕ) is a subgroup of G. Finally. denoted by log2 y) such that y = 2log2 y . 1] is continuous. Theorem 2. 1] that have a root at 1 2 (for instance the straight line f (x) = x − 1 belongs to the kernel). So im(ϕ) is a subgroup of G . 2. as ϕ(e) = e . Algebra 2. then there exists an a ∈ G such that ϕ(a) = a . the constant function f (x) = a for all x ∈ [0. and 1 G be the group R with addition. The image is the set of 2 all real numbers. given any real number a not equal to zero. The image of ϕ is the set im(ϕ) = {a ∈ G  ∃a ∈ G such that ϕ(a) = a }. Since ϕ(e) = e . We now also check that im(ϕ) is a subgroup of G . Indeed if a. Thus H1 holds. 1. since it can be shown that given any y > 0. R). then there exist elements a.
G be groups and let ϕ : G → G be an isomorphism. 4. Prove that the composition ψ ◦ ϕ : G → G deﬁned by (ψ ◦ ϕ)(a) = ψ(ϕ(a)). given by a → a−1 is an isomorphism. G are groups and ϕ : G → G is a homomorphism. then show that the function from G to G. If G is a group. 1. there is no real diﬀerence between them. Let G be the group R with addition. given by m → am . Then the function from G to G . . Deﬁnition. (a) Prove that the function from Z to a . given by x → 2x .1. (b) If a has inﬁnite order. Examples. G be groups and let ϕ : G → G and ψ : G → G be group homomorphisms. then b ∗ a ∗ b−1 ∈ N . (b) Describe the kernel of ψ ◦ ϕ. The exponential function from G to G . 1. A homomorphism ϕ : G → G is said to be an isomorphism if it is bijective. Let G be a group and let a be an element of G. Prove that if G. is an isomorphism. then the identity function ι : G → G deﬁned by ι(a) = a for all a ∈ G is an isomorphism. (a) Let G. Let G. and G be the set of positive reals with multiplication. 2. Let G. Let G be the subgroup of GL(2.2. 3. R) comprising all matrices of the form 1 x . then ker(ϕ) is a normal subgroup of G. G be groups. is a homomorphism from the group of integers Z with addition to the subgroup a . 2. Groups 51 In Example 1 above. A subgroup N of a group G is called a normal subgroup if for every a ∈ N and every b ∈ G. in the sense that as far as algebraic properties go. a ∈ G is a group homomorphism. Exercises. 3. → x. If G is an abelian group. ♦ Isomorphisms are important because their existence between two groups means that the two groups are essentially the “same”. then prove that it is an isomorphism. 5. G . Let G be R with addition. where 0 1 x ∈ R. We give a special name to such homomorphisms. the homomorphism between the two groups was also bijective. given by 1 x 0 1 is an isomorphism. Prove that the inverse function ϕ−1 : G → G is also an isomorphism.
) Hence we obtain the following result: G b∗H (2.1). (Transitivity) For all a. The two sets. the sets {1.52 Chapter 2. and is denoted by a ∗ H. it follows that h−1 ∈ H. and let R be the equivalence relation given by (2. Then the left cosets of H partition the group G. 2. is called a left coset of H. then there exists a h ∈ H such that b = a ∗ h and so b ∗ h−1 = a. and of odd integers. {2. 3. b. it follows that h1 ∗ h2 ∈ H. deﬁne the relation R on G by aRb if b = a ∗ h for some h ∈ H. form a partition of the set Z of all integers. aRa since e ∈ H and a = a ∗ e. then the equivalence class of a.1. we mean a subdivision of the set S into nonoverlapping subsets: S = union of disjoint.1) c∗H a∗H =a ∗H PSfrag replacements (aRa ) Figure 2. Thus bRa.1: Distinct cosets partition the group. (Reﬂexivity) For all a ∈ G. and h ∈ H. . (Symmetry) For all a. if aRb. For example. Corollary 2. 5}. of even integers.4 Cosets and Lagrange’s theorem Given a subgroup H of a group G.1. 5}. then there exist elements h1 . c in G.6 Let H be a subgroup of a group G. Let H be a subgroup of a group G. 4. {4} form a partition of the {1. h2 ∈ H and H is a subgroup. Deﬁnition. We know that the equivalence classes of an equivalence relation partition the set. E2. Algebra 2. Since H is a subgroup. Hence we obtain c = b ∗ h2 = (a ∗ h1 ) ∗ h2 = a ∗ (h1 ∗ h2 ). namely the set {b ∈ G  aRb} = {b ∈ G  ∃h ∈ H such that b = a ∗ h} = {a ∗ h  h ∈ H}. nonempty subsets of S. If a ∈ G. h2 in H such that b = a ∗ h1 and c = b ∗ h2 . if aRb and bRc. Then R is an equivalence relation: E1. (Recall that a by partition of a set S. E3. and so aRc. b in G. Since h1 . 3}.
0. . Consider the group G of the integers modulo 6. and [1] ⊕ H = [3] ⊕ H = 5 ⊕ H = {[1]. 3. Consequently. [5]} = {[0]. −3. . . We now prove an important result concerning the order of a group G and the number of cosets of a subgroup H. and let H be the subgroup of even numbers {2m  m ∈ Z}. each coset a ∗ H has the same number of elements as H does. Proof Note that there is a bijective function from the subgroup H to the coset a ∗ H. for h ∈ H. given by h → a ∗ h. . [3]. 1.6 says that if a ∗ H and b ∗ H have an element in common then they are equal. [4]} ∩ {[1]. and let H be the subgroup [2] = {[0]. . we obtain the counting formula G = H · (number of cosets of H). 2. 4. Then the order of H divides the order of G. . . [2]. Since G is the union of the cosets of H.7 (Lagrange’s theorem) Let H be a subgroup of a ﬁnite group G. −4. which we now denote by a ⊕ H are [0] ⊕ H = [2] ⊕ H = [4] ⊕ H = {[0]. . 0. . [1]. −1. . ♦ · · · = −2 + H = 0 + H = {2m  m ∈ Z} = {. . . . and since these cosets do not overlap. . This is not a coincidence. The notation a ∗ H denotes a certain subset of G. [2]. [4]} and {[1]. [2]. . 2.1. [5]}. . . and G = 6 = 3 · 2 = H · (number of cosets of H). −2. . [5]}. In particular the order of H (namely 3) divides the order of G (namely 6). and so a ∗ H = b ∗ H iﬀ aRb. [2]. . } = 3 + H = . [4]. As with any equivalence relation. which we now denote by a + H are · · · = −1 + H = 1 + H = {2m + 1  m ∈ Z} = {. 3. . 4. [4]}. [3]. . [5]} do form a partition of G: G = {[0].2. Groups 53 Remarks. [2]. . . Corollary 2. [4]}. Examples. [2]. The associated equivalence classes are called right cosets. −4. 1. One can also deﬁne the relation R on G by aR b if b = h ∗ a for some h ∈ H. 1. . 1. The left cosets. Z6 . . Note that the cosets {[0]. [3]. [3]. Consider the group G of the integers Z. with addition. The left cosets. due to Lagrange. H divides G. [4]} ∪ {[1]. In particular. . Note that the cosets {. −3. diﬀerent notations may represent the same subset.1. −2. [5]} = ∅. . 2.1. [3]. with addition modulo 6. −1. and {[0]. . Theorem 2. there are only two distinct cosets. } do indeed partition the set of all integers. In fact. . } and {. } = 2 + H = . 2. . . we know that a ∗ H is the unique coset containing a. Note that in Example 1 above.
Proof If a = e. 4 By . [a]p−1 = [1]. a is a subgroup of G. Since m divides p. Proceeding in a similar manner. and from Theorem 2.8 Let G be a ﬁnite group and let a ∈ G.1. The following theorem characterizes all groups whose order is a prime number. Corollary 2. so that in Z7 . 2. b ∈ G are such that a = b. (c) Any element of G can be represented by a point in the (x.1. show that [55552222 ] = [32 ]. Proof Let a have order m. and so p(ap − a). Algebra Corollary 2. then (a ∗ H) ∩ (b ∗ H) = ∅. 4. Let H and K be subgroups of a group G of orders 3 and 5 respectively. 1. Exercise 1 on page 48. G= 3. Exercises.1. Prove or disprove that the group S4 has an element of order 16. Then the order4 of a divides the order of G. Prove that H ∩ K = {e}. it follows that  a  = m divides G. then a has order > 1. it now follows that G = a . Hence p(ap−1 − 1). Thus aG = am·k = (am )k = ek = e. the left coset a ∗ H has the same number of elements as the right coset H ∗ a. say m. and let Z∗ denote the group of nonzero integers modulo p with p multiplication modulo p. y ∈ R and x > 0 0 1 is a group with matrix multiplication. (b) If H is a subgroup of G and a ∈ G is such that a ∗ H has 4 elements. (c) If H is a subgroup of a ﬁnite group G. Draw the partitions of the plane into left and into right cosets of H. then for any a ∈ G. and p is prime. Hint: If a ∈ Z is not divisible by p. (a) Let p be a prime. and so G is cyclic. then [a]p−1 = [1]. In particular aG = e. then G is cyclic. and G = a for every a ∈ G \ {e}. ap ≡ a(mod p). (c) (∗) Show that 7 divides 22225555 + 55552222. Hint: Note that 2222 = 7 · 317 + 3. (b) Show that x 0 H= x ∈ R and x > 0 0 1 is a subgroup of G. As G itself has order p. (a) Verify that x y x. (a) If H is a subgroup of G and a. then H has 4 elements. y)plane. Show that if a is an integer such that a is not divisible by p. and so G = m · k for some k ∈ N. it follows that every element in a ﬁnite group has a ﬁnite order. Hence we obtain [22225555 +55552222] = [35 +32 ] = [32 ·28] = [0]. Determine if the following statements are TRUE or FALSE. (b) (∗) Prove Fermat’s little theorem: for any integer a.9 If G is a group with prime order p. and so by part (5a) above. [22225555] = [3]5555 .3.54 Chapter 2. then [a] = [0]. it follows that m = p. Now use the fact that [3]6 = [1] to conclude that [22225555 ] = [35 ]. Hint: H ∩ K is a subgroup of H as well as K. 5.
called vector addition. V4. For all α ∈ R and all v1 . α · (v1 + v2 ) = α · v1 + α · v2 . and · : R × V → V . 2. called “vectors”.2. every element has a unique inverse! 5 Since 6 In . The set Rn×m of n × m matrices having real entries with matrix addition is an abelian group. a1n αa11 . V3. (α + β) · v = α · v + β · v. . α · (β · v) = (αβ) · v. there exists a unique6 element in V . Roughly speaking it is a set of elements. For all v1 . A vector space V is a set together with two functions. αamn Then α · A ∈ Rn×m . am1 . . Vector spaces 55 2.2. V3 and V4 are called the distributive laws. For all v ∈ V . v3 ∈ V . denoted by −v. β ∈ R and all v ∈ V . V2. v1 + v2 = v2 + v1 . such that v + (−v) = 0 = −v + v. Let n.2 Vector spaces In this section we introduce a very important algebraic object. 1 · v = v. . the zero vector is unique! a group. . The elements of a vector space are called vectors. . . ∈ Rn×m . V4 are satisﬁed: there is a unique identity element in a group. . v + 0 = v = 0 + v. +) is an abelian group. v2 ∈ V .2) . resulting in a new vector. 4. called scalar multiplication. V3. 2. called a vector space. v2 . and the following hold: V1. m ∈ N. The precise deﬁnition is given in the next subsection. . . then α·A = . αa1n . 3. V2. . and any vector can be multiplied by an element from R so as to give a new vector. For every v ∈ V . . . There exists an element 0 ∈ V (called the5 zero vector) such that for all v ∈ V . a vector space also has the following properties: 1. . . and moreover V1. . . β ∈ R and all v ∈ V . For all α. (2.2. We observe that since (V. +) is an abelian group. if α ∈ R and A = . Deﬁne scalar multiplication as follows: if α ∈ R and a11 . 1. v2 ∈ V . + : V × V → V . amn αam1 . For all α. . such that (V. Examples. For all v1 . Any two vectors can be “added”.1 Deﬁnition of a vector space Deﬁnition. v1 + (v2 + v3 ) = (v1 + v2 ) + v3 .
. . = . . am1 . . αam1 + βam1 . . αa1n βa11 . . . . . . . am1 . . . . . a1n 1a11 . . . 1·A = 1· . amn 1am1 . (α + β)amn αa11 + βa11 . . . βamn α(βa11 ) . then clearly a11 . .... βa1n . (αβ)am1 . . . . . . . .. . . amn = α · A + β · A. . β ∈ R and all A ∈ Rn×m . . = α· . βam1 . . . . amn (α + β)a11 . . = . . . . a1n .. . am1 . .. = . (α + β) · A = (α + β) · . . . . . . . (α + β)am1 ... . = α · β · . a1n . am1 . . . 1amn am1 Chapter 2. . . . . α · (β · A) a11 . . . (αβ)a1n . αa1n + βa1n . For all α. 1a1n a11 . If A ∈ Rn×m . . . . . . a1n . amn = (αβ) · A. . amn . α(βam1 ) . +β· .. . . .. . . . = α· . βa1n . . .. = . . αamn a11 . . . α(βamn ) (αβ)a11 . αamn + βamn αa11 . . . βam1 . a1n . . = . = A. . . . . . . (α + β)a1n . = (αβ) · . α(βa1n ) . = . . . . βamn a11 . For all α. .56 V1.. Algebra . = + . . a1n . .. . . . . amn V2. . . . . a1n . αam1 . . . (αβ)amn a11 . . . am1 . amn βa11 .. . . . V3. am1 . . . . . β ∈ R and all A ∈ Rn×m . . . . . a11 .
b]. . .. . . For all x ∈ [a. . . and so 1 · f = f . V2. b]. we have (1 · f )(x) = 1f (x) = f (x). b]. V3. Hence Rm×n is a vector space with matrix addition and with scalar multiplication deﬁned by (2. If n = 1. amn + bmn b1n . 2. . αbmn a11 . αa11 + αb11 . and so (α · (β · f ) = (αβ) · f . .. b].. . β ∈ R and f ∈ C[a. . . +β· . bmn = α · A + β · B. . .. V2. α · (A + B) = α · . . . . α(a1n + b1n ) . αa1n αb1n . .. we have ((α + β) · f )(x) = (α + β)f (x) = αf (x) + βf (x) = (α · f )(x) + (β · f )(x) = (α · f + β · f )(x). . B ∈ Rn×m . . ... . . Let α. b be real numbers with a < b. am1 . . α(am1 + bm1 ) . . . + . . b] of continuous functions on the interval [a. and moreover V1. . . . . Let α. Let a. am1 a11 + b11 . αamn αbm1 . . we have (α · (β · f ))(x) = α(β · f )(x) = α(βf (x)) = (αβ)f (x) = ((αβ) · f )(x). . (2. bm1 amn . then we denote the vector space of column vectors Rm×1 by Rm . = α· . . V4 are satisﬁed: V1. a1n + b1n . x ∈ [a. . . αam1 . For all x ∈ [a. = . αam1 + αbm1 . .. . b]. . Vector spaces 57 V4. For all α ∈ R and all A. . . b]. . a11 . β ∈ R and f ∈ C[a. amn bm1 . a1n b11 . . b] with addition of functions is an abelian group.2. . then (α · f )(x) = αf (x). am1 + bm1 αa11 . b]. = .. + . . . bmn αamn + αbmn αb11 . . b]. b1n . For all x ∈ [a. . .2.2). The set C[a. . Let f ∈ C[a. b].3) Then α · f ∈ C[a.. . . . . α(amn + bmn ) αa1n + αb1n . Let scalar multiplication be deﬁned as follows: if α ∈ R and f ∈ C[a. . . . = . V3. b11 a1n . . . .. ... . . = α· . . α(a11 + b11 ) . ..
3. α · 0 + α · 0 = α · (0 + 0) = α · 0. Exercises. Now add −(0 · v) on both sides. b]. and hence α · 0 = 0. we use the distributive law to write 0 · v + 0 · v = (0 + 0) · v = 0 · v. 2. 0 · v = 0. Let α ∈ R and f. Let n ∈ N. and so α · (f + g) = α · f + α · g. then (−1) · v = −v.2. But the inverse of an element from a group is unique. ♦ We now prove a few elementary properties of vector spaces. Proof Please go through the entire proof carefully. b] with addition and scalar multiplication is a vector space. Hence C[a. For all x ∈ [a. Hence (−1) · v is an inverse of v. Finally. 1. and so v + (−1) · v = 0 = (−1) · v + v. 1. g ∈ C[a. Chapter 2. b]. (a) Is the set of invertible n × n matrices having real entries with matrix addition and with scalar multiplication deﬁned by (2. Algebra V4. and which refer to the zero vector 0 ∈ V . Similarly. noting which symbols refer to the number 0 ∈ R. we have v + (−1) · v = 1 · v + (−1) · v (since 1 · v = v) = (1 + (−1)) · v (distributive law) = 0 · v (since 1 + (−1) = 0) = 0 (since 0 · v = 0). Then the following hold: 1.1 Let V be a vector space. 2. If v ∈ V . To see this. For all α ∈ R. and so (−1) · v = −v. to obtain 0 · v = 0.58 and so (α + β) · f = α · f + β · f . 3. we have (α · (f + g))(x) = α(f + g)(x) = α(f (x) + g(x)) = αf (x) + αg(x) = (α · f )(x) + (α · g)(x) = (α · f + α · g)(x). α · 0 = 0. For all v ∈ V .2) a vector space? . Theorem 2.
2. the set of symmetric matrices U2 = is a subspace of R2×2 .4) 2.2). If a subspace U of V is neither {0} nor V . then the subset U comprising only the zero vector. Subspaces of a vector space are just like subgroups of a group. S3. b. 2. d ∈ R .2 Subspaces and linear combinations Deﬁnition. Vector spaces 59 (b) Is the set of invertible n × n matrices having real entries with matrix multiplication and with scalar multiplication deﬁned by (2. then it is called a proper subspace of V . then α · (an )n∈N = (αan )n∈N . If v1 . 3. If v ∈ U and α ∈ R.2. (bn )n∈N ∈ R∞ . Hint: If α = 0. So a subspace is really a smaller vector space sitting inside a larger vector space. Premultiply both sides of α · v = 0 by α−1 . a b b d a. Then the set of upper triangular matrices U1 = is a subspace of R2×2 . (∗) Consider the set R∞ of all sequences with addition deﬁned as follows: if (an )n∈N . Prove that R∞ is a vector space with the above addition and scalar multiplication. then α · v ∈ U . v2 ∈ U . Also. Prove that if α ∈ R and v ∈ V are such that α · v = 0. the entire vector space.2. then α−1 ∈ R. and scalar multiplication deﬁned as follows: if α ∈ R and (an )n∈N ∈ R∞ . Let V be a vector space. then either α = 0 or v = 0. If V is a vector space. then v1 + v2 ∈ U . namely U = {0}. 1. is a subspace of V . Let V be a vector space.5) (2. a subspace of a vector space is itself a vector space with the same addition and scalar multiplication as with V (this is easy to check). then (an )n∈N + (bn )n∈N = (an + bn )n∈N . Examples. 0 ∈ U . that is U = V . d ∈ R a b 0 d a.2) a vector space? 2. Consider the vector space R2×2 with matrix addition and scalar multiplication deﬁned by (2. A subset U is called a subspace of V if S1. Also. (2. S2. that is. b. is a subspace of V .
. Let a. vm = . denoted by span(S). b ∈ R and a < b. b ∈ R with a < b. 1] with addition of functions and scalar multiplication deﬁned by (2. b] is a subspace of the vector space C[0. α1 . . is deﬁned as the set of all possible linear combinations7 of vectors from S: span(S) = {α1 · v1 + · · · + αn · vn  n ∈ N. . }. . . 7 Note that although S might be inﬁnite. . Let S be a nonempty subset of a vector space V . . . . . Hence span(S) = P [a. . . . . . b] Then U = P [a. . 2. . βm ∈ R. . Consider the set of all polynomial functions P [a. . . αn . . vn ∈ S. Any polynomial p on the interval [a. . . un . . . . 1. Any vector in the vector space Rm is a linear combination of the vectors 0 1 0 0 v1 = . ♦ Deﬁnitions. . an ∈ R such that p(x) = a0 + a1 x + a2 x2 + · · · + an xn for all x ∈ [a. b] = p : [a. by deﬁnition. . . . b] → R Chapter 2. β1 . Then 0 = 0 · v ∈ span(S). . αn ∈ R}. If v1 . then we know that u = α1 · u1 + · · · + αn · un and v = β1 · v1 + · · · + βm · vm for some vectors u1 . b] in the vector space C[a. . v ∈ span(S). . . .3). ♦ The span of a set of vectors turns out to be a special subspace of the vector space. . a1 . Let m ∈ N.2 Let V be a vector space and S be a nonempty subset of V . Theorem 2. . vn are vectors in V and α1 . . vn . Then span(S) is the smallest subspace of V that contains S. . 2. u + v = α1 · u1 + · · · + αn · un + β1 · v1 + · · · + βm · vm ∈ span(S). b] is a linear combination of the functions from the set S = {1. Let V be a vector space. 1. . b]. . 1 0 Hence span({v1 . v1 . a2 . vm ∈ S and scalars α1 . Let a. Consequently. x. vm }) = Rm . . is always a linear combination of a ﬁnite set of vectors from S. . The span of S. If u. . . . . . αn belong to R.60 3. . a linear combination. . . . Let v ∈ S. Examples. . Proof We ﬁrst show that span(S) is a subspace of V . . then the vector α1 · v1 + · · ·+ αn · vn is called a linear combination of the vectors v1 . . v1 . . Algebra ∃n ∈ N ∪ {0} and a0 . . . x2 .2.
. . .S3. . if v ∈ S. vn are called linearly dependent if they are not linearly independent. . and so span(S) ⊂ U . . . αn ∈ R. An arbitrary subset S of vectors from V is said to be a linearly dependent if there exists a nonempty ﬁnite set of dependent vectors from S. . and since a = b. . = . . . . An arbitrary subset S of vectors from V is said to be linearly independent if every nonempty ﬁnite set of vectors from S is an independent set of vectors. Examples. + · · · + αm · . . The vectors v1 . = α1 · . and so S ⊂ span(S). . then in particular. Indeed. we have β ·v = β ·(α1 ·v1 +· · ·+αn ·vn ) = β ·(α1 ·v1 )+· · ·+β ·(αn ·vn ) = (βα1 )·v1 +· · ·+(βαn )·vn ∈ span(S). 0 1 are linearly independent in R . . . . Vector spaces 61 Finally. b] are linearly independent. 0 0 1 . The vectors 1 0 . 1. . ∃α1 . 0 0 . Deﬁnitions. αn ∈ R. b]. . if v ∈ span(S). . . we have α + βa = 0 and α + βb = 0. then v = 1 · v ∈ span(S). that is. then α1 = · · · = αn = 0. The functions 1. Indeed if v1 . m v1 = if α1 · v1 + · · · + αm · vm = 0. such that α1 · v1 + · · · + αn · vn = 0. . Thus span(S) contains S. . . Let V be a vector space. . . and suppose that v1 . then we know that v = α1 · v1 + · · · + αn · vn for some v1 . If U is another subspace that contains the vectors from S.S2. . . x on the interval [a. . . . Then any ﬁnite set of vectors from V containing the zero vector is linearly dependent. Let V be a vector space. vn are called linearly independent if the following condition holds: if ∃α1 . . . Hence span(S) is the smallest subspace of V containing S. . . and so it is a subspace of V . . . 1. . . = . it follows that α = β = 0. . αm 0 and so α1 = · · · = αm = 0. 2. not all zeros. . then 0 0 . . if for all x ∈ [a. 2. . . . ♦ . Let a. Indeed 1 α1 0 . Moreover. α · 1 + β · x = 0(x). vn ∈ S and scalars α1 . vn are vectors that belong to V . then 0 · v1 + · · · + 0 · vk−1 + 1 · vk + 0 · vk+1 + · · · + 0 · vn = 0. . The vectors v1 . . vn ∈ V and vk = 0.2. So span(S) satisﬁes S1. then from S2 and S3 it follows that it certainly contains all linear combinations of vectors from S belong to U . b ∈ R with a < b. αn ∈ R such that α1 · v1 + · · · + αn · vn = 0.2. vm 3. and so for any β ∈ R.
1 in the vector space R3 . 3 4 1 0 1 1 (d) The vectors 0 .2. B is linearly independent. (c) c0 is the set of all convergent sequences with limit 0. span(B) = V .62 Exercises. 1. y2 ) = {f ∈ C[0. v4 are linearly dependent. v3 are linearly dependent. v2 .5). v3 . (b) (∗) Prove that R2 is not the union of a ﬁnite number of proper subspaces. and B2. Determine if the following statements are TRUE or FALSE: (a) The union of two subspaces of a vector space V is a subspace of V .3 Basis of a vector space Deﬁnition. 1] iﬀ y1 = 0 = y2 . Let V be a vector space. (d) c00 = {(an )n∈N ∈ R∞  ∃N ∈ N such that ∀n > N. Let S(y1 . y2 ) is a subspace of C[0. then v1 . then span in the vector space R2 . 1] with addition of functions and scalar multiplication deﬁned by (2. 1 t2 = R2 is the set of all bounded sequences. 2. 1]  f (0) = y1 and f (1) = y2 }. (∗) Let R∞ be the vector space of all sequences.4) and (2. t 3. then v1 . v3 are linearly independent.3). We deﬁne the following subsets of R∞ : (a) ∞ 1 t1 . (b) c is the set of all convergent sequences. v3 . respectively. an = 0}. . (f) If v1 . 0 1 1 (e) If v1 . Prove that c00 ⊂ c0 ⊂ c ⊂ ∞ ⊂ R∞ . Algebra (b) The intersection of two subspaces of a vector space V is a subspace of V . 4. 2 . Then a set of vectors B is said to be a basis of V if B1. 1 0 1 (c) 2 ∈ span 3 . Consider the vector space C[0. 1 Hint: Consider the inﬁnite subset S = t∈R . with addition and scalar multiplication deﬁned by (2. 3 are linearly independent in the vector space R3 . (a) Prove that if t1 and t2 are distinct real numbers in R. v4 are linearly independent. is the set of all sequences that are eventually zero. 2. Show that S(y1 . Chapter 2. and that each is a subspace of the next one. v2 . v2 . v2 .
and such that span(Sn ) = V . . . and the other n − 0 elements are in S.2. Sk+1 has n elements. . . (Why?) Moreover. . uk } are the same. n}. we obtain that Sn has n elements. . . .2. there exist numbers α1 . . . 2◦ Hence there must exist an αj = 0. This is proved as follows: we claim that for each k ∈ {0. . . u1 . and since B is a basis. . ur . . αn−k . . . We prove this claim by induction on k. . . vj+1 . the next result says that the cardinality of the basis is unique for any given vector space. . So the spans of the sets Sk Sk+1 = {v1 . vj . . . Hence by induction. . . and such that span(Sk ) = V . The vectors form a basis of Rm . . . vn−k . Thus Sk is a set with n elements such that it has n − k elements v1 . say u1 . . Given a vector space. .4 If a vector space V has a basis with n elements. βk ∈ R such that ur = α1 · v1 + · · · + αn−k · vn−k + β1 · u1 + · · · + βk · uk . . vn−k . . . So this case is not possible. Proof Suppose that B is a basis of V with n elements and suppose that B is another basis of V . . 0 . . u1 . . . uk }.3 Let V be a vector space and B be a basis of V such that B has n elements. we simply deﬁne S0 = B. Then we have the following two cases: 1◦ If α1 = · · · = αn−k = 0. . Now suppose that ur is an element from S such that it does not belong to {u1 . . . . . and the other elements (namely. and so Sn = S. . . . . . V = span(B) = span(S0 ). vj−1 . . But S has n elements. such that 0 of these are in B. u1 . 1 ♦ Theorem 2. . Vector spaces 63 Example. Corollary 2. which contradicts the linear independence of S.2. . . β1 . v1 = 1 0 . uk ) belong to S. Then any linearly independent set S of n vectors is also a basis of V . Proof We simply prove that span(S) = V . . . and such that span(Sk ) = V . there are of course many bases. . ur . then every basis of V has the same number of elements. vn−k from B and the other k elements.2. Then S0 is a set with n elements such that it has n − 0 = n elements from B and the other 0 elements belong to S. 1. for k = n. vm = 0 0 . uk . n − k − 1 of these belong to B. . belong to S. . For k = 0. . However. there is a set Sk with n elements such that it has n − k elements from B and the other k elements belong to S. . vj−1 . . and := {v1 . uk }. Suppose that the claim is true for some k. . then we get that 1 · ur − β1 · u1 − · · · − βk · uk = 0. . . Thus span(S) = V . Since span(Sk ) = V . . vj+1 . .
and it is denoted by dim(V ). If B has < n elements. 1. First show that the set B = {x. . Let V be a vector space. . .2. 1]. vn }. 1] is inﬁnite dimensional. let ek denote the sequence with the kth term equal to 1. Prove that if B is a basis of a ﬁnite dimensional vector space V . If there exists a basis B of V such that B has n elements. . 3. Deﬁnition. 0 if n = k. . 4. Exercises. and all other terms equal to zero: ek = (an )n∈N .3. . 1] be a ﬁnite dimensional vector space with dimension d. Let C[0. Prove that B is a basis for the vector space c00 . (b) (∗) Is B = {ek  k ∈ N} also a basis for the vector space R∞ ? Hint: Consider the constant sequence (1)n∈N . . Derive a contradiction. 1 . . vn }. Then take any n distinct elements v1 . 1. it can be written as a linear combination of {v1 . and let B = {ek  k ∈ N}. then every element v ∈ V can be written as a unique linear combination of the vectors from B. . it follows that these span V . B is a basis for C[0. 2. If a vector space has a basis with a ﬁnite number of elements. say. (∗) Show that C[0. 2. . where an = 1 if n = k. . . we once again arrive at a contradiction. . 3. (a) (∗) For k ∈ N. and so the constant function 1 should be a linear combination of the functions from B. . comprising all sequences that are eventually zero. then it is called an inﬁnite dimensional vector space. vn from B . The above result motivates the following natural deﬁnitions. . then n is called the dimension of V . . x2 . and so if v ∈ B \ {v1 . Then by Theorem 2.64 Chapter 2. If a vector space is not ﬁnite dimensional. Prove or disprove that 1 1 1 B = 0 . . then by interchanging the roles of B and B and proceeding as above. Algebra Let B have > n elements. xd } is linearly independent. 1 1 0 0 is a basis for R3 . Hint: One can prove this by contradiction. which contradicts the independence of B . then it is called a ﬁnite dimensional vector space. . From the previous theorem.
6) k=1 = k=1 n (a1k xk + a1k yk ) . TA . Thus just like group homomorphisms that are functions that respect group operations. for all . For all u ∈ U . . u2 ∈ U . V be two vector spaces. am1 x1 + · · · + amn xn xn k=1 n k=1 a1k xk x1 . xn + y n = n k=1 n k=1 n a11 x1 + · · · + a1n xn x1 . . xn amk xk (2. Examples. . .. . + . . + . ∈ Rm×n . yn x1 . . (amk xk + amk yk ) a1k xk . . Let Then the function TA : Rn → Rm deﬁned by a11 . and all α ∈ R. k=1 n amk xk + k=1 k=1 k=1 n k=1 k=1 a1k yk . amk yk a1k yk amk yk y1 . A= . . xn y1 . . = . . 1. linear transformations are functions that respect vector space operations. . yn x1 + y 1 .2. .. = TA = . TA .. A ..4 Linear transformations Deﬁnition. Let m. A function T : U → V is called a linear transformation if it satisﬁes the following two properties: L1. a1n . amk xk n k=1 n a1k (xk + yk ) . we have n x1 . . amk (xk + yk ) n k=1 n n a1k xk + . Let U. xn . . = = . n ∈ N. Indeed. am1 . For all u1 . amn n is a linear transformation from the vector space Rn to the vector space Rm . +T .2. . . L2. . . Vector spaces 65 2. ∈ Rn . T (α · u) = α · T (u). .2. . T (u1 + u2 ) = T (u1 ) + T (u2 ). . = TA . . .
The image of T is deﬁned to be the set im(T ) = {v ∈ V  ∃u ∈ U such that T (u) = v}. xn αxn = α· n k=1 n k=1 n k=1 k=1 for all α ∈ R and all vectors a1k xk . Hence TA is a linear transformation. 1]. . Moreover. and so L1 holds. for all f. x1 . n amk xk amk αxk α k=1 n x1 . Just as in the case of homomorphisms between groups. for all α ∈ R and all f ∈ C[0. . = . . . . where 0V denotes the zero vector in V . . yn n x1 αx1 . ♦ and so L2 holds too. . g ∈ C[0. . 1] → R given by Tf = f 1 2 for all f ∈ C[0. Indeed. . TA α · . Furthermore T (α · f ) = (α · f ) 1 2 = αf 1 2 = αT (f ). . n . xn y1 . 2. . Deﬁnitions. amk xk a1k xk a1k αxk α k=1 . xn and so L2 holds as well. V be vector spaces and T : U → V a linear transformation. = . . we have T (f + g) = (f + g) 1 2 =f 1 2 +g 1 2 = T (f ) + T (g). . ∈R . The function T : C[0. Let U. The kernel of T is deﬁned to be the set ker(T ) = {u ∈ U  T (u) = 0V }. x1 . Thus T is a linear transformation. . = T A . 1]. 1]. there are two important subsets associated with a linear transformation between vector spaces. 1] to the vector space R. . . 1. . Algebra and so L1 holds. whose deﬁnitions are given below.66 for all Chapter 2. ∈ Rn . = α · TA . 2. xn is a linear transformation from the vector space C[0.
with kernel equal to the set of continuous functions on the interval [0. ♦ Analogous to Theorem 2. The function T : C[0.1. The kernel of the linear transformation is the set of all vectors x1 . . 1] that vanish at the point 1 .6).2. V be vector spaces and T : U → V a linear transformation. we now prove the following result. ∈ Rn . 1]. . ym x1 . The range of TA is the set of all vectors y1 . ∈ Rn . xn a11 x1 + · · · + a1n xn am1 x1 + · · · + amn xn = y1 . Let A ∈ Rm×n . ker(T ) is a subspace of U .2. 2.5 Let U. The range of T is the 2 whole vector space R. S2. 1] to the vector space R. = 0 . we check that S1. is a linear transfor2 mation from the vector space C[0.5 for homomorphisms between groups. S3 hold. 1] → R given by T f = f 1 for all f ∈ C[0.2. = 0 such that there exists a vector such that 2. im(T ) is a subspace of V . . . . = ym . Theorem 2. 1. and let TA : Rn → Rm be the linear transformation deﬁned by (2. x = . Proof In each of the cases. ∈ Rm . y = . x = . Then: 1. Vector spaces 67 Examples. xn such that the system of linear equations a11 x1 + · · · + a1n xn am1 x1 + · · · + amn xn is simultaneously satisﬁed.
v2 belong to im(T ). Consider the vector space R2 with matrix addition and the usual scalar multiplication deﬁned by (2. We now also check that im(T ) is a subspace of V . Thus S2 holds. it follows that 0V ∈ im(T ). Exercises. T (u1 + u2 ) = T (u1 ) + T (u2 ) = v1 + v2 . 0V denote the zero vectors in the vector spaces U. but not L1. Hence α · v ∈ im(T ). In each of the y cases. respectively. v1 + v2 ∈ im(T ). and so 0U ∈ ker(T ). x in R2 can be represented by a point in the (x. . Thus S1 holds. So im(T ) is a subspace of V. where A ∈ R2×2 is given by: (a) A = (b) A = (c) A = Each vector 1 1 0 1 1 1 0 0 0 0 0 0 . Indeed. (∗) Let c denote the vector space of all convergent sequences.2). y)plane. 2. and so there exists an element in U . and so S2 holds as well. Finally. u2 belong to ker(T ). . S3 holds. it follows that T (0U ) = 0V . then T = x2 x2 x1 if x1 x2 = 0. for all sequences (an )n∈N ∈ c. as T (0U ) = T (0U + 0U ) = T (0U ) + T (0U ). then T (α · u) = α · T (u) = α · 0V = 0V . namely u1 + u2 . 3. then T (u1 + u2 ) = T (u1 ) + T (u2 ) = 0V + 0V = 0V . x2 Show that T satisﬁes L2. u2 in U such that T (u1 ) = v1 and T (u2 ) = v2 .68 Chapter 2. if α ∈ R and v ∈ im(T ). Algebra Let 0U . then there exists a u ∈ U such that T (u) = v. If u1 . Find the kernel and image of the linear transformations TA : R2 → R2 . and S3 holds. Finally. such that T (u1 + u2 ) = v1 + v2 . It is easy to check that ker(T ) is a subspace of U . since if α ∈ R and u ∈ ker(T ). and hence it is not a linear transformation. Since T (0U ) = 0V . n→∞ (a) Prove that T is a linear transformation from the vector space c to the vector space R. draw a picture of the subspaces ker(TA ) and im(TA ) in the plane. Consequently. . If v1 . Consider the function T : c → R given by T ((an )n∈N ) = lim an . (b) What is the kernel of T ? (c) Show that im(T ) = R. V . that is. Hence ker(T ) is a subspace of U . Deﬁne the function T : R2 → R2 as follows: x2 1 x2 if x1 x2 = 0. x2 x1 x1 2 if ∈ R . and so S1 holds. then there exist elements u1 . and so α · v = α · T (u) = T (α · u). 1.
In fact any real number l ≤ 0 is a lower bound. then 0 < 2 . 1] = S. then is inf S ∈ S? No. S is bounded above. it is bounded. First of all. is an element in S that is less than the lower bound l l. Since S is bounded above as well as bounded below. —————————————————————————————————————— Minimum of S. We prove that l ≤ 0.Solutions Analysis Solutions to the exercises on page 6 1. we have 0 < x. —————————————————————————————————————— Supremum of S. Indeed. 0 is a lower bound. Let l be a lower bound of S. 1] = S. it follows that 2 < l. —————————————————————————————————————— If inf S exists. since for all x ∈ S = (0. (a) S = (0. In fact any real number u ≥ 1 is an upper bound. 1]. The contradiction is obtained as follows: if l > 0. if u is also an upper bound. —————————————————————————————————————— Inﬁmum of S. since for all x ∈ S = (0. and moreover. —————————————————————————————————————— Is S bounded? Yes. —————————————————————————————————————— An upper bound of S. (We do this by supposing that l > 0. inf S = 0 ∈ (0. —————————————————————————————————————— If sup S exists. then 1 ≤ u (since 1 ∈ S). 1]. then we will see that l the average of 0 and l. —————————————————————————————————————— Maximum of S. and arriving at a contradiction. then is sup S ∈ S? Yes. max S = 1. sup S = 1 ∈ (0. since 0 is a lower bound. —————————————————————————————————————— 69 . But since l > 0. which is a contradiction to the deﬁnition of a lower bound!) If l > 0. 0 is a lower bound. l l Thus 2 ∈ S. since 1 is an upper bound of S. since sup S = 1 ∈ S. sup S = 1. 1 is an upper bound. Hence l ≤ 0. 1 is an upper bound. 1]. inf S = 0. min S does not exist since inf S = 0 ∈ S. since l ≤ 1 (l is a lower bound of S and 1 ∈ S) it follows that 2 ≤ 2 ≤ 1. l 1 Moreover. a contradiction. we have x ≤ 1. —————————————————————————————————————— A lower bound of S. namely 2 . S is also bounded below.
—————————————————————————————————————— Minimum of S. 0 is a lower bound. 1) = S. First of all. then is inf S ∈ S? Yes. x < 1. it 2 follows that 0 < u (since 0 < 1 ≤ u). if u is also an upper bound. sup S = 1 ∈ [0. So if u < 1. then 0 < 2 . then 1 ≤ u (since 1 ∈ S). —————————————————————————————————————— Supremum of S. sup S = 1. —————————————————————————————————————— Is S bounded? Yes. then 0 < u = u+u < u+1 < 1+1 = 1. 1] = S. —————————————————————————————————————— If inf S exists. —————————————————————————————————————— If sup S exists. namely u+1 2 . since S is bounded above (1 is an upper bound) and it is bounded below (0 is a lower bound). 2 2 —————————————————————————————————————— Inﬁmum of S. 0 is an lower bound. Indeed. if l is also an lower bound. inf S = 0. —————————————————————————————————————— An upper bound of S. —————————————————————————————————————— An upper bound of S. If 1 l l > 0. —————————————————————————————————————— Supremum of S. since sup S = 1 ∈ S. —————————————————————————————————————— A lower bound of S. 1 is a upper bound. But u < u+1 contradicts the fact that u is an upper bound of S. Thus we have 0 < 2 < l ≤ 2 < 1. —————————————————————————————————————— Inﬁmum of S. 1]. —————————————————————————————————————— If inf S exists. —————————————————————————————————————— (c) S = (0. 2 2 2 2 Hence u+1 ∈ S. sup S = 1 ∈ (0. we have x ≤ 1. then we will see that the average of u and 1. Let l be a lower bound of S. 1]. 1 is an upper bound. then is inf S ∈ S? No. sup S = 1. —————————————————————————————————————— If sup S exists. 1).70 (b) S = [0. since 2 ∈ S and l is a lower bound of S. Indeed. We prove that 1 ≤ u. 1]. since S is bounded above (1 is an upper bound) and it is bounded below (0 is a lower bound). We prove that l ≤ 0. inf S = 0 ∈ [0. and moreover. 0 < x. The contradiction is obtained as follows: if u < 1. 1). 1). since for all x ∈ S = (0. —————————————————————————————————————— Is S bounded? Yes. 0 is a lower bound. Let u be an upper bound of S. then l ≤ 0 (since 0 ∈ S). it follows that 1 l 1 l l l ≤ 2 . and so 2 ∈ S. 1) = S. and moreover. since inf S = 0 ∈ S. Moreover. which is a contradiction to the deﬁnition of an upper bound!) Since u is an upper bound and since 1 ∈ S. —————————————————————————————————————— Maximum of S. But 2 < l contradicts the fact that l is a lower bound of S. —————————————————————————————————————— A lower bound of S. First of all. max S = 1. . 0 is a lower bound. 1 is an upper bound. inf S = 0. 1] = S. is an element in S that is larger than the upper bound u. since for all x ∈ S = (0. (We do this by supposing that u < 1 and arriving at a contradiction. then is sup S ∈ S? Yes. we have 0 ≤ x. min S = 0. since for all x ∈ S = (0. 1 is an upper bound. then is sup S ∈ S? No. inf S = 0 ∈ (0. since for all x ∈ S = [0.
Moreover. since inf S = 0 ∈ S. −1 ≤ − n . —————————————————————————————————————— Supremum of S. if l is also a lower bound. a contradiction to the fact that u is an upper bound of S. then is inf S ∈ S? Yes. since inf S = −1 ∈ S. max S does not exist. since sup S = 1 ∈ S. since sup S = 1 ∈ S. —————————————————————————————————————— 1 1 1 (d) S = n  n ∈ Z \ {0} = n  n ∈ N ∪ − n  n ∈ N . —————————————————————————————————————— 1 1 An upper bound of S. —————————————————————————————————————— If sup S exists. since S is bounded above (0 is an upper bound) and it is bounded below (−1 is a lower bound). then is sup S ∈ S? No. —————————————————————————————————————— Supremum of S. sup S = 1. max S = 1. since for all n ∈ N. inf S = −1. —————————————————————————————————————— If inf S exists. since for all n ∈ N. 1 is an upper bound. —————————————————————————————————————— 1 A lower bound of S. −1 ≤ 0 ≤ n and −1 ≤ − n . if u is an upper bound and 1 1 if u < 0. —————————————————————————————————————— Is S bounded? Yes. —————————————————————————————————————— Minimum of S. —————————————————————————————————————— 1 An upper bound of S. −1 is a lower bound. since for all n ∈ N. —————————————————————————————————————— Is S bounded? Yes. 1 is an upper bound. 1 then since −1 = −1 ∈ S. if u is also an upper bound. min S = −1. −1 is a lower bound. and so we obtain u < − N ∈ S. it follows that l ≤ −1. —————————————————————————————————————— Inﬁmum of S. then let N ∈ N be such that −u < N (Archimedean principle with y = −u 1 and x = 1!). inf S = −1. n ≤ 1 and − n ≤ 0 ≤ 1. Thus if u is an upper bound. 1 —————————————————————————————————————— Inﬁmum of S. Moreover. then is inf S ∈ S? Yes. since S is bounded above (1 is an upper bound) and it is bounded below (−1 is a lower bound). then 0 ≤ u. . −1 is a lower bound. sup S = 1 ∈ S. sup S = 0. —————————————————————————————————————— If inf S exists. then 1 since −1 = − 1 ∈ S. —————————————————————————————————————— 1 1 A lower bound of S. min S does not exist. since sup S = 0 ∈ S. —————————————————————————————————————— Minimum of S. —————————————————————————————————————— 1 (e) S = − n  n ∈ N . it follows that l ≤ −1. —————————————————————————————————————— Maximum of S. then is sup S ∈ S? Yes. then since 1 = 1 ∈ S. since for all n ∈ N. 0 is an upper bound. 1 ≤ u.71 —————————————————————————————————————— Maximum of S. −1 is a lower bound. inf S = −1 ∈ S. − n < 0. 0 is an upper bound. inf S = −1 ∈ S. Moreover. and if l is another lower bound. —————————————————————————————————————— If sup S exists.
1 ≤ n). n+1 < 1. u < u+ 2 < 2. 1 is a lower bound. max S does not exist since sup S = 0 ∈ S. —————————————————————————————————————— √ √ Supremum of S. —————————————————————————————————————— n An upper bound of S. —————————————————————————————————————— Inﬁmum of S. if x ∈ S. sup S = 2. then ac > bc. Let u be an √ √ upper bound such that u < 2. then x ≥ − 2. If u < 1 is an upper bound.) But 2 2 2 2 2 . u then let N ∈ N be such that u−1 < N (Archimedean property). that is. and if l is a lower bound. since inf S = 2 ∈ S.) —————————————————————————————————————— √ Is S bounded? Yes. Then u+ 2 ∈ S. —————————————————————————————————————— (g) S = {x ∈ R  x2 ≤ 2}. 1 ≤ n+1 (since n+1 ≤ 2n. √ In other words. and so x ∈ S. 1 is an upper bound. 2 —————————————————————————————————————— Supremum of S. then is inf S ∈ S? Yes. 2 2 that is. —————————————————————————————————————— Maximum of S. 2 is an upper bound. − 2 is a lower bound. sup S = 1. and so √ − 2 is a lower bound. since for all n ∈ N. it follows that l ≤ 2 . 1 is an upper bound of S. As u < 2.) —————————————————————————————————————— √ √ √ A lower bound of S. inf S = 1 . +1 contradicting the fact that u is an upper bound. √ > 2 implies x2 > 2. 2 is an upper bound of S. then is sup S ∈ S? No. if x ∈ S. since S is bounded above (1 is an upper bound) and it is bounded below ( 1 is a lower bound). —————————————————————————————————————— 1 If inf S exists. (As 0 ∈ S and u is an upper bound 2√ √ √ √ √ of S. max S does not exist since sup S = 1 ∈ S. In other words. —————————————————————————————————————— Minimum of S. since S is bounded above ( 2 is an upper bound) and it is bounded √ below (− 2 is a lower bound). and so. min S exists since inf S = 2 ∈ S. —————————————————————————————————————— Is S bounded? Yes. x ∈ S. we get x2 = (−x)(−x) > √ √ √ √ (−x) 2 > 2 2 = 2. Thus u+ 2 ∈ S. then x ≤ 2. 0 ≤ u. min S = −1 since inf S = −1 ∈ S. —————————————————————————————————————— √ √ (x An upper bound of S. then since 2 2 1 1 1 2 = 1+1 ∈ S. —————————————————————————————————————— 1 Minimum of S. —————————————————————————————————————— If sup S exists. First of all. Then u < NN . 2 is an upper bound of S. and 2 2 √ √ √ √ √ 2 √ √ √ u+ 2 = u+ 2 < u+ 2 so u+ 2 2 < 2 2 = 2. since sup S = 1 ∈ S.72 —————————————————————————————————————— Maximum of S. (x < − 2 implies that −x > 2 > 1 > 0. —————————————————————————————————————— n (f) S = n+1  n ∈ N . So using the fact that if a > b and c > 0. Hence we have 0 < u+ 2 < 2. —————————————————————————————————————— n A lower bound of S. 1 is a lower bound because for all n ∈ N.
As l is a lower bound. 2 ≤ 2005. it follows that −l cannot be an upper bound of S. then is sup S ∈ S? Yes. inf S = 0 ∈ {0. 5. since 0 ≤ 0.73 u < u+ 2 contradicts the fact that u is an upper bound. inf S = − 2. √ Consequently. 2. 2005 is an upper bound. x ≤ −l. since S is bounded above (2005 is an upper bound) and it is bounded below (0 is a lower bound). —————————————————————————————————————— √ If inf S exists. . But since x ∈ S. that is. 2 1 1 If n ∈ N and n is even. . 0 ≤ 5 and 0 ≤ 2005. . 2. since 0 ≤ 2005. and if u is also an upper bound. 5 . sup S = 2005. —————————————————————————————————————— Minimum of S. So it follows that −x ∈ S. 0 is a lower bound. 0 is an lower bound. Let l be a lower bound such that √ √ √ − 2 < l. it follows that 2005 ≤ u. —————————————————————————————————————— Inﬁmum of S. then is inf S ∈ S? Yes. then since 2005 ∈ S. inf S = − 2 ∈ S. 2. 5. − 3 . From −l < x and x ≤ −l. —————————————————————————————————————— If inf S exists. —————————————————————————————————————— An upper bound of S. 2 2 √ . —————————————————————————————————————— Is S bounded? Yes. − 2 is a lower bound. sup S = 2 ∈ S. then 2 ≤ u. 2005 is an upper bound. 2005}. we arrive at the contradiction that −l < −l. max S = 2005 since sup S = 2005 ∈ S. 2 . —————————————————————————————————————— If sup S exists. So if u is an upper bound of 2√ S. then l ≤ − 2. So there exists an x ∈ S such that −l < x. then is inf S ∈ S? Yes.) 1 4 —————————————————————————————————————— An upper bound of S. max S = 2. we have (−x)2 = x2 ≤ 2. —————————————————————————————————————— A lower bound of S. 2005} = S. —————————————————————————————————————— Supremum of S. 5. and since sup S = 2. —————————————————————————————————————— √ Maximum of S. Then we have −l < 2. —————————————————————————————————————— √ If sup S exists. then (−1)n 1 + n = 1 + n ≤ 1 + 1 = 3 . 2005} = S. —————————————————————————————————————— √ √ Inﬁmum of S. —————————————————————————————————————— Maximum of S. —————————————————————————————————————— (h) S = {0. then is sup S ∈ S? Yes. if l is a lower bound. then since 0 ∈ S. 0 ≤ 2. 5 ≤ 2005 and 2005 ≤ 2005. —————————————————————————————————————— 1 (i) S = (−1)n 1 + n  n ∈ N . it follows that l ≤ 0. 3 4 (This set has the elements − 2 . and if l is also an lower bound. inf S = 0. min S = 0 since inf S = 0 ∈ S. l ≤ −x. —————————————————————————————————————— √ Minimum of S. sup S = 2005 ∈ {0. 3 is an upper bound. . min S = − 2.
sup S does not exist. since the set does not have an upper bound. then since 0 = 02 ∈ S. —————————————————————————————————————— Minimum of S. since inf S = 0 ∈ S. 2 3 1 3 2 then since 2 = (−1) 1 + 2 ∈ S. —————————————————————————————————————— 2 . since the set is not bounded above. and if l is also a lower bound. −2 is a lower bound. then (−1)n 1 + n = −1 − n < 0 < 2 . it follows that l ≤ 0. contradicting the fact that 2 u is an upper bound of S. sup S = 3 . 1 1 If n ∈ N and n is odd. —————————————————————————————————————— Is S bounded? No. —————————————————————————————————————— 3 Supremum of S. —————————————————————————————————————— Inﬁmum of S. then since −2 = (−1)1 1 + 1 ∈ S. —————————————————————————————————————— An upper bound of S. But since 2 2 1 1 u2 + 4 > 0. 0 ≤ x2 . 0 is a lower bound. inf S = 0. Let u ∈ R be an 2 1 upper bound of S. and so it cannot have a least upper bound. and so u2 + u + 4 = u + 1 ∈ S. min S = −2. since for all x ∈ R. —————————————————————————————————————— If sup S exists. then is inf S ∈ S? Yes. then (−1)n 1 + n = 1 + n > 0 > −2. then is sup S ∈ S? sup S does not exist. 0 is a lower bound of S. −2 is a lower bound. —————————————————————————————————————— A lower bound of S. —————————————————————————————————————— Is S bounded? Yes. —————————————————————————————————————— (j) S = {x2  x ∈ R}. then (−1)n 1 + n = −1 − n ≥ −1 − 1 = −2. min S = 0. —————————————————————————————————————— Minimum of S. since S is bounded above ( 3 is an upper bound) and it is bounded 2 below (−2 is a lower bound). 2 is an upper bound.74 1 3 1 If n ∈ N and n is odd. it follows that u < u2 + u + 4 = u + 1 ∈ S. and if l is also a lower bound. since sup S does not exist. (Recall that every least upper bound is an upper bound). it follows that 2 ≤ u. and if u is also an upper bound. inf S = −2. —————————————————————————————————————— A lower bound of S. sup S = 3 ∈ S. Thus S does not have an upper bound. —————————————————————————————————————— Inﬁmum of S. 1 —————————————————————————————————————— If sup S exists. then u + 1 ∈ R. inf S = 0 ∈ S. 2 —————————————————————————————————————— 3 Maximum of S. it follows that l ≤ −2. max S does not exist. then is sup S ∈ S? Yes. —————————————————————————————————————— Supremum of S. —————————————————————————————————————— Maximum of S. max S = 2 . —————————————————————————————————————— If inf S exists. There does not exist an upper bound of S. 1 1 If n ∈ N and n is even.
—————————————————————————————————————— x2 A lower bound of S. (k) FALSE (−N is bounded above since 0 is an upper bound. u}. Let u be an upper bound and let 02 u u < 1. and if l is also a lower bound. u} and so max{−l. —————————————————————————————————————— Inﬁmum of S.) 1 2 (j) FALSE (the set [0. but it does not have a maximum). (i) TRUE (deﬁnition of maximum). Then we have 2 2 N u < 1+N 2 ∈ S. (e) FALSE ([0. 1+x2 < 1. 0 ≤ u. for every y ∈ S. since for all x ∈ R. 1}. (l) TRUE (l ≤ x ≤ u implies x ≤ u and −x ≤ −l. then since 02 0 = 1+02 ∈ S. —————————————————————————————————————— Supremum of S. —————————————————————————————————————— 2. (h) TRUE (the supremum itself is an upper bound). u} is an upper bound of S. 1 is an upper bound of S. then is sup S ∈ S? No. since sup S ∈ S. Thus 0 is a lower bound of S. then is inf S ∈ S? Yes. we have y = x for some x ∈ S. min S = 0. and so we have x ≤ u ≤ max{−l. since for all x ∈ R. 1 is an upper bound. since inf S = 0 ∈ S. but 1 2 ∈ S). 0 is a lower bound. but −N = N is not bounded). > 0. Moreover. a contradiction to the fact that u is an upper bound of S. —————————————————————————————————————— x2 An upper bound of S. Hence if u is an upper bound of S. —————————————————————————————————————— If sup S exists. (m) FALSE (if S = {0. inf S = 0 ∈ S. (b) TRUE (if (c) FALSE (N has no maximum). but it does not have a supremum). max S does not exist. —————————————————————————————————————— Minimum of S. —————————————————————————————————————— If inf S exists. it follows that l ≤ 0. 1) is bounded. since S is bounded above (1 is an upper bound) and it is bounded below (0 is a lower bound). and so y = x ≥ 0.75 x (k) S = 1+x2  x ∈ R . (g) TRUE (least upper bound property of R). Since 0 = 1+02 ∈ S. then u = 3 is an upper bound of S. then u∗ − < u∗ . 1)). u} and − x ≤ −l ≤ max{−l. then inf S = 0 < < 1 = sup S. inf S = 0. . then 1 ≤ u. Let N ∈ N be such that 1−u < N . —————————————————————————————————————— Is S bounded? Yes. and so u∗ − cannot be an upper bound of S). 0 is a lower bound of S. 0 ≤ 1+x2 . 1) has supremum 1. but 1 ∈ [0. u (= 2) is an upper bound of {1} = S). (a) FALSE (if S = {1}. (d) FALSE (N has no supremum). —————————————————————————————————————— Maximum of S. (f) FALSE (∅ is bounded. Thus x ≤ max{−l. and although u = 2 < 3 = u. sup S = 1. sup S = 1 ∈ S.
in particular we obtain x ≤ sup B for all x ∈ A.9) (2. So inf S = x = sup S. inf S ≤ x. If u < x is an upper bound. and so we obtain that −x ∈ −S. x ≤ MA . Since A is bounded above. sup S exists. since it is nonempty. sup(A + B) ≤ sup A + sup B. Since A ⊂ B. Thus −S is bounded above because −l is an upper bound of −S. a contradiction.7) and so inf S = x(= sup S) (for if inf S < x. by the least upper bound property of R. Consequently if x ∈ A and y ∈ B. it is bounded above. and so by the deﬁnition of the least upper bound of A. l ≤ x. inf S exists. in particular. a contradiction. Since every z ∈ A + B is such that z = x + y with x ∈ A and y ∈ B. So sup B is an upper bound of A. x + y ≤ sup A + sup B. ∀x ∈ A. then clearly x is an upper bound. x + y ≤ MA + MB . by the deﬁnition of the least upper bound of A + B. Since inf S is a lower bound of S. Conversely. Clearly x is also a lower bound. since sup S is an upper bound of S. in other words. So sup S = x. Hence −S is nonempty. Thus S is a singleton set. a contradiction). If x ∈ S. we obtain inf S ≤ sup S. and consequently. Let l be a lower bound of S: ∀x ∈ S. y ≤ sup B. since it is nonempty. If l > x is also a lower bound.7) and (2. Since sup B is an upper bound of B. Let x ∈ S. then x ≤ u < x gives x < x. we obtain sup A ≤ sup B. inf S < sup S. it follows that sup(A + B) exists. x ≤ sup S. x ≤ sup A. So A + B is bounded above. Indeed. y ≤ MB . (2. ∀y ∈ −S. So for all x ∈ A and y ∈ B. by the least upper bound property of R. ∃MA ∈ R such that ∀x ∈ A. A is nonempty implies that ∃x ∈ A. Clearly A + B is nonempty. y ≤ −l. ∃MB ∈ R such that ∀y ∈ B.8) (2. it follows that ∃x ∈ S. −x ≤ −l. Since sup A is an upper bound of A. 6. and so A + B is not empty. it follows that for all z ∈ A + B. B is nonempty implies that ∃y ∈ B. Moreover. z ≤ MA + MB . So ∀z ∈ A + B. Since S is nonempty. Since sup B is an upper bound of B. So sup A + sup B is an upper bound of A + B. Since A + B is bounded above and it is not empty. thus x + y ∈ A + B. ∀y ∈ B. if S = {x}. From (2. Let inf S = sup S. and furthermore. . and so MA + MB is an upper bound for A + B. z ≤ sup A + sup B. 4. and furthermore. we have x ≤ sup B for all x ∈ B. 5.8). then from (2. then x > l ≥ x gives x > x. Since B is bounded above. by the greatest lower bound property of R. Since S is bounded. in particular it is bounded below. then we have inf S ≤ x ≤ sup S.76 3. Since S is bounded.9). So ∀x ∈ S.
) If: Let inf S > 0. − sup(−S) < l ≤ x. Thus 1 u that is. inf S exists. which contradicts the fact that sup(−S) is the least upper bound of −S. ∃x∗ ∈ S and so x∗ ≤ u. Next we prove that − sup(−S) is the greatest lower bound of S. y ≤ So 1 inf S is an upper bound of S −1 . Hence l ≤ − sup(−S). ∀x ∈ S. inf S is bounded above. Since inf S is a lower bound. u is a lower bound of S. that is. 7. Since sup(−S) is an upper bound of −S. Then ∀y ∈ S −1 y ≤ u. y ≤ −l < sup(−S). (2. and −l < sup(−S).10) x 1 Since S is not empty. ∀x ∈ S. that is. Consequently u > 0. 1 ≤ x. ∀y ∈ S −1 . Hence from (2. that is. Consequently. So − sup(−S) is a lower bound of S. that is. and so 1 x∗ > 0. 1 1 ≤ . u . inf S exists and inf S = − sup(−S). by the least upper bound property of R. it follows that sup(−S) exists. we have: ∀y ∈ −S. x inf S 1 . it follows that ∀x ∈ S. −x ≤ −l < sup(−S). ∀x ∈ S. that is. y ≤ sup(−S). −x ≤ sup(−S). that is. But x∗ ∈ S implies that x∗ > 0. (S is nonempty and bounded below (0 is a lower bound). So −l is an upper bound of −S. and so by the greatest lower bound property of R. Thus S −1 Only if: Suppose S −1 is bounded above (with an upper bound u. say).77 As −S is nonempty and bounded above. − sup(−S) ≤ x. ∀x ∈ S. ∀y ∈ −S. and so 1 ≤ inf S. we have ∀x ∈ S. Then we have ∀x ∈ S. Suppose that l is a lower bound of S such that − sup(−S) < l .10). inf S ≤ x. ∀x ∈ S. 1 ≤ u.
and consequently inf S ≥ 1 u > 0. 1). Then x ∈ [0.13) < x. and hence 1 1 0. n n n∈N 0. . inf S 8. n (2. 1 (b) Clearly {0} ⊂ 0. 1] and so x ≥ 0. if x ∈ 0.11) and (2. then 0 < ≤x≤1− 1 n < 1. (a) If x ∈ 0.16) n∈N 1 1 ⊂ (0.12) 1 . n∈N 1 . Hence (2. Thus ¬ ∃x ∈ 0. inf S Furthermore. 1 n 1 = {0}. is an upper bound of S −1 and so we (2.14) 1 . and so S 1 ≤ inf S. .15). 1 ≤ sup S −1 . that is. we see 1 that u = sup1 −1 is a lower bound of S. that is. n∈N 1 x < 0. 0 < x < Let N ∈ N be such that dicts (2. n∈N 1 . 1). 1 n = ∅. n∈N 0.14) and (2. N .1− n n . Consequently. . n 1 N (2. as in the Only if part. n for all n ∈ N and so {0} ⊂ Let x ∈ x∈ 0. then n for all n ∈ N. N < x. n∈N 1 ⊂ {0}. then x = 0. and so x ∈ (0. If x ∈ 1 n. n∈N 1 n . that is. inf S From (2.12). since u := sup S −1 is an upper bound of S −1 .13). n 1 n (c) Let n ∈ N. then let N ∈ N be n such that n∈N 1 1 N (Archimedean property). If x > 0. 1 − . So x ∈ 0. which contra < N (Archimedean property). n (2. So 1 x (2. we see that obtain sup S −1 ≤ > 0. sup S −1 that is.11) 1 inf S 1 . n∈N 0. then as in the If part.78 But u > 0 implies 1 u If inf S > 0.15) From (2. it follows that sup S −1 = 1 .
that is. Thus with N := max{N1 . 1) ⊂ n∈N From (2. Adding x0 . we have 1 1 1 1 ≤ <x < 1− ≤1− . that is. So {x ∈ R  x − x0  < δ} ⊂ (x0 − δ. Then n n − 1 1 ≤ x ≤ 1 + for all n ∈ N.19) We prove that this implies that 0 ≤ x ≤ 1. Let N1 ∈ N be such that x < N1 (Archimedean property). 1) = n∈N (d) If x ∈ [0. 1].1+ .1− n n 1 1 . − 1 and so x ∈ − n . then x − x0  < δ. Adding −x0 . if x > 1. Thus (x0 − δ. 1].17) n∈N (0. this yields x0 − δ < x < x0 + δ. we obtain −δ < x − x0 < δ. Thus 1 1 − .18). x > 1 + N2 . .79 1 If x ∈ (0. then x0 − δ < x < x0 + δ. then 0 < x < 1. 1]. So x − x0 < δ. (2. then let N1 ∈ N be such that 1 1 − x < N1 . a contradiction to (2. n n . that is. n n (2. 1 x < 1 − N2 . 1] ⊂ 1 1 − . Hence −δ < x − x0 < δ.1+ n n . then let 1 1 N2 ∈ N be such that x−1 < N2 . So we have . If x ∈ (x0 − δ. x ∈ (x0 − δ. and −(x − x0 ) < δ.20) imply n∈N 1 1 − . For if x < 0.17). and hence x ∈ [0. N2 }. Consequently (x0 − δ. a contradiction to (2. 9. x0 + δ). then for any n ∈ N. .1+ n n = [0. Hence we see that neither x < 0 nor x > 1 are possible. x ∈ 1 1 . N N1 N2 N that is.16) and (2. Thus x − x0  < δ. x0 + δ). N1 < x.1− N N ⊂ 1 1 . we obtain that x − x0 ≤ x − x0  < δ and −(x − x0 ) ≤ x − x0  < δ. (2. 1]. r ≥ r and r ≥ −r.1− n n 1 1 . 1).1− n n (2. If x ∈ {x ∈ R  x − x0  < δ}. x0 + δ) = {x ∈ R  x − x0  < δ}.1+ n n ⊂ [0. Similarly. Hence [0.19).18) n∈N Let x ∈ n∈N 1 1 − . x0 + δ). x0 + δ) ⊂ {x ∈ R  x − x0  < δ}. x < − N1 . Since for any real number r. that is.20) n∈N (2.18) and (2. we obtain (0. and x ∈ {x ∈ R  x − x0  < δ}. 1 + 1 n 1 1 <0≤x ≤1<1+ . 1 1 that is. . Let N2 ∈ N be such that 1−x < N2 (Archimedean property).
If S is bounded. y ∈ R. that is. Conversely. and so. Interchanging x and y in (2. From (2. that is. we have −M ≤ x ≤ M .22) for all x. u} and − x ≤ −l ≤ max{−l. u}. say u. x ≤ u and −x ≤ −l. From the inequality (1. if there exists a M such that for all x ∈ S. y ∈ R. we obtain y − x ≤ y − x =  − (x − y) =  − 1x − y = 1 · x − y = x − y.21) and (2. Thus S is bounded. Thus x ≤ max{−l. y ∈ R. we have that x = x − y + y ≤ x − y + y. So for all x ∈ S.7) on page 6. So −M is a lower bound of S and M is an upper bound of S.21). and so we have x ≤ u ≤ max{−l. say l. 11.21) . Thus S has an upper bound. then it is bounded above and it is bounded below. and a lower bound. −(x − y) ≤ x − y (2. u} =: M .80 10. x ≤ M . (2. l ≤ x ≤ u. we obtain x − y ≤ x − y for all x. x − y ≤ x − y for all x.22).
a contradiction. vm }. . . Then we have 2 = 2 = 1 + 1 = (−1)n − (−1) < = 1. . n Let = 1 . Then for all n > N = 1. let N ∈ N be such that for all n > N . vm − L} > 0. (−1)n − (−1) < = 1. vm − L} ≤ vk − L. an −L < . ◦ 2 If the limit is −1. say (d) Suppose that ((−1)n )n∈N is a convergent sequence with limit L.) Then there exists a N∗ ∈ N such that 2 for all n > N∗ . =⇒ L must be one of the terms . Let aN +1 = vk for some k ∈ {1. let N ∈ N be such that for all n > N . Then we have vk − L = aN +1 − L < = min{v1 − L. . . . a contradiction. 1− 1 1 1 = −1 < . and so L must be one of these terms. (This choice is motivated by Figure 2. . . In particular. . then given = 1 > 0. . . (−1)n − 1 < = 1. . . . . For instance. So ((−1)n )n∈N is divergent. we have aN +1 −L < . So L ∈ {v1 . the constant sequence (1)n∈N converges to 1. . So we have the following two cases: 1◦ If the limit is 1. . L is not equal to any of the terms =⇒ terms of the sequence cannot consist of ﬁnitely many values . . namely 1 and −1. vm }. from which we see that. a contradiction.81 Solutions to the exercises on page 13 1. vm }. say. then with := min{v1 − L. (a) We prove that (1)n∈N is convergent sequence with limit 1. n 2 that is. we have an − L = 1 − 1 = 0 < . then given = 1 > 0. but = 1 would. If lim = 1. > 0. (b) Yes. (c) Suppose that the terms of the convergent sequence (an )n∈N (with limit. Let n be any odd number > N . 1 −1 < . . . If L ∈ {v1 . for instance 2 = 1 won’t give us a contradiction. . . . then given any > 0. let N ∈ N be such that for all n > N .2. vm }. n n 2 . 1 2. Then we have 2 =  − 2 =  − 1 − 1 = (−1)n − 1 < = 1. that is. with n = N +1 > N . Given N = 1. Thus we have shown that terms of the sequence take ﬁnitely many values that is. . L) lie in the ﬁnite set {v1 . Let n be any even number > N . 1 1 −1 < = . . there exists an N ∈ N such that for all n ∈ N such that n→∞ n n > N. m}. it follows that L = 1 or L = −1: indeed the terms of the sequence take ﬁnitely many values. L must be one of the terms. Then from part 1c above. let N ∈ N. . But aN +1 ∈ {v1 . . .
Then := − L > 0. . let 1 N ∈ N be such that < N . Hence with n = N + 1 (> N ). In particular. that is. which we denote by an . and so sup S − n < sup S. we obtain 2 L aN +1 − L ≤ aN +1 − L < − . n n n > 0. Thus S − n is not an upper bound of S. it follows that sup S exists. for all n > N∗ . n < 2. let N ∈ N be such that N > 1 . and so there exists a N ∈ N such that for all n > N . and so by the least upper bound property of R. Then for all n > N . Then for all even n > N (there are obviously inﬁnitely many such n). an > sup S − n . that is. such that 1 1 ¬[an ≤ sup S − n ]. Hence we arrive at a 1 = 1. we have an − L = (−1)n 1 + 1 n −1 = 1+ 1 n 1 1 1 −1 = < < . since n = N∗ + 1 > N∗ . ∀n ∈ N. n n N . But there does not exist any natural number N∗ that is less than 1. for all N ∈ N (b) Again. n 2 that is. In this way we construct the sequence (an )n∈N . contradiction. we have an − sup S < 1 1 < < . 2 that is. 1 n n∈N . 2 an − L < = − L . with = 4 > 0. Hence there must exist an element in S. aN +1 < L 2 < 0. we have n > 0.82 1 PSfrag replacements Figure 2. Given any > 0. we have n∈N an − L ≤ an  + L ≤ 2 + 1 = 3 < 4 = . that is. 1 1 1 Given n ∈ N. we obtain N∗ + 1 < 2. ∀n ∈ N. (a) We have seen that the sequence (−1)n 1 + n n∈N is divergent. for the divergent sequence (−1)n 1 + and all n > N . Suppose L < 0.2: Consequently. n N Hence (an )n∈N is convergent with limit equal to sup S. a contradiction. − sup S − 1 1 < an ≤ sup S < sup S + . N∗ < 1. n n Given 1 1 1 < an − sup S < . that is. and so lim n→∞ n 1− 1 3. S is nonempty and bounded above. As sup S is an upper bound of S and so we also have an ≤ sup S for all n ∈ N. 4. 1 1 < . an − sup S < . 5. Consequently.
Then for all n > N . then ak+1  = 2k + 3 2k + 3 ak = ak  = 3k + 3 3k + 3 2k + 3 3k + 3 ak  ≤ 1 · 1 = 1. ∃N1 ∈ N such that ∀n > N1 . N1 (M + L) < . an ≥ 0 (induction!). m > N . that is. = an − L + am − L (using r =  − r for all real r) Consequently (an )n∈N is a Cauchy sequence. N 2 ˛ ˛ ˛ ˛ is arrived at by working backwards. it is convergent. So the sequence is bounded. we wish to make ˛ a1 +···+an − L˛ less than for all n > N . Since (an )n∈N is convergent. Solutions to the exercises on page 17 1. 3. We prove this using induction. and so. and hence it must be convergent. n n n N M Hence bn n n∈N is convergent with limit 0. Let M > 0 be such that for all n ∈ N. and so the claim follows from induction. so we n manipulate this (as shown in the chain of inequalities that follow) to see if can indeed achieve this by choosing the N large enough.83 6.23) for all n > N. 3n So (an )n∈N is decreasing. We have a1  = 1 = 1. Let (an )n∈N be a convergent sequence with limit L¿ Given > 0. an  ≤ M . and so 2n+1 ≤ 1 for all n ∈ N. As the sequence is bounded and monotone. 2. it follows that 2n + 1 ≤ 3n. an = 2n+1 an−1 ≤ 1·an−1 = an−1 . Furthermore. We prove that an  ≤ 1 for all n ∈ N. > 0. note 3n that for all n ∈ N. Choose8 N ∈ N such that max N1 . bn  ≤ M . let N ∈ N be such that bn bn  M M −0 = ≤ < <M· = . We prove that the sequence is monotone and bounded. it is bounded: ∃M > 0 such that ∀n ∈ N. Given 1 M < N . there exists a N ∈ N such that (2. Since n ≥ 1. N < M . an − L < . an − L < 2 . N > N1 and 8 This N1 (M + L) 2 < N. then an − am  = an − L + L − am  ≤ an − L + L − am  (triangle inequality) < 2 + 2 (using (2. Hence for all n ≥ 2. 2 Hence if n.23)). If k ∈ N is such that ak  ≤ 1. . (a) Given > 0.
Indeed. 1 − n if n is odd 1 = = ≤ ≤ ≤ ≤ < < = So (b) If a1 +···+an is a convergent sequence with n n∈N an = (−1)n . and so (uk )k∈N is a decreasing sequence. lk and uk are welldeﬁned. but the sequence with nth term 0 if n is even . then in particular. Since (an )n∈N is bounded. it then follows that inf{an  n ≥ k} and sup{an  n ≥ k} exist. Clearly {an  n ≥ k + 1} ⊂ {an  n ≥ k}. we have an − 0 = an  = So a1 +···+an n n∈N 1 n < N. By the least upper bound property of R. 0 if n is even if n is odd ≤ 1 1 < < . −M ≤ an ≤ M . −M ≤ inf{an  n ≥ k} ≤ sup{an  n ≥ k} ≤ M. 4. Furthermore. and so the set {an  n ≥ k} is bounded. n ∈ N. and so we see that the sequences (lk )k∈N and (uk )k∈N are bounded. −M ≤ an ≤ M . it follows that there exists a M such that for all n ∈ N. we have: a1 + · · · + aN1 + aN1 +1 + · · · + an −L n a1 + · · · + aN1 + aN1 +1 + · · · + an − nL n a1 + · · · + aN1 + aN1 +1 + · · · + an − nL n a1 − L + · · · + aN1 − L + aN1 +1 − L + · · · + an − L n (a1  + L + · · · + aN1  + L) + 2 + · · · + 2 n N1 (M + L) + (n − N1 ) 2 n N1 (M + L) N1 · + 1− n n 2 N1 (M + L) +1· N 2 2 . let N ∈ N be such that Then for n > N . If k ∈ N. and so we have sup{−an  n ≥ k + 1} ≤ sup{−an  n ≥ k}. for all n ≥ k. that is. that is. . for each k. we then obtain that uk+1 = sup{an  n ≥ k + 1} ≤ sup{an  n ≥ k} = uk . n N is convergent with limit 0.84 Then for n > N . then (an )n∈N is divergent. given > 0. Similarly. + 2 limit L. and by Exercise 4 on page 7. we obtain inf{an  n ≥ k + 1} = − sup{−an  n ≥ k + 1} ≥ − sup{−an  n ≥ k} = inf{an  n ≥ k}. an  ≤ M . Using Exercise 6 on page 7. a1 + · · · + a n (−1)1 + (−1)2 + · · · + (−1)n = = n n is convergent with limit equal to 0. {−an  n ≥ k + 1} ⊂ {−an  n ≥ k}.
we see that an  ≤ M for all n ∈ N. 5. aN . and so (an )n∈N is bounded.3. As the sequences (uk )k∈N .85 that is. an − aN +1  < 1 for all n > N . . (lk )k∈N is a increasing sequence. In particular. Consequently. . Let (an )n∈N be a Cauchy sequence. aN +1  + 1}. an − am  < = 1. there exists a N ∈ N such that for all n. it follows that they are convergent.2. that is. m > N . lk+1 ≥ lk . (lk )k∈N are both bounded and monotone. Then given := 1 > 0. . by Theorem 1. with m := N + 1 (> N ). Deﬁning M = max{a1 . . . an  = aN +1 + an − aN +1  ≤ aN +1  + an − aN +1  < aN +1  + 1 for all n > N.
Proof Let > 0. The sequence +1 the nonzero limit 1.2. Hence (an+1 )n∈N is convergent with limit L. Since n n∈N is convergent with limit 0. Finally. it follows n 1 that a2 · n n∈N is convergent with limit L2 ·0 = 0. a2 a2 n n We have n + 1 ≥ 1 for all n ∈ N. Since (bn )n∈N is bounded. Alternately. and so an+1 − L < . and use Theorem 1. which satisﬁes: an+1 = = for all n ∈ N. that is. we have n + 1 > N + 1 > N . by the theorem on 3 2+ n 3 3+ n algebra of limits. So the sequence an · bn + 5 n∈N is convergent with limit 0 + 5 = 5.6. 3 n→∞ 2. . Then ∃N ∈ N such that for all n > N . Hence n an · bn n∈N is convergent with limit L · 0 = 0. 3+ n is convergent with limit 0. The sequence (an )n∈N is convergent with limit. L = 0. a2 n n D. 3 lim an+1 lim 2+ 3+ 2+ 3+ 3 n 3 n 3 n 3 n n→∞ an lim an n→∞ lim n→∞ 2 L. an − L < . and so n + 1 = 0 for all n ∈ N. Thus for all n > N . 3 Hence 1 L = 0. We have cn = N. = 1 a2 + n an · an · n + 1 n +5 n∈N is convergent. it follows that the sequence a2 n n n∈N has nonzero terms for all n ∈ N and it is convergent with +1 n∈N is convergent with limit 0+1 = 1(= 0). it follows that the sequence Again applying the theorem on algebra of limits. First we prove the Claim: (an )n∈N is convergent with limit L. n the sequence bn n∈N is convergent with limit 0 (see Exercise 2 on page 17). Since the sequence 1 n n∈N 2(n + 1) + 1 an 3(n + 1) 3 2+ n 3 an . The sequence (5)n∈N is convergent with n limit 5. So lim an = 0. say L. observe that (an+1 )n∈N is a subsequence of (an )n∈N . The sequence an · bn n n an · bn + 5 an bn + 5n for all n ∈ N. Since the sequence (an )n∈N is convergent with limit L. it follows that the sequence 1 (a2 )n∈N is convergent with limit L2 . We now apply this result to our sequence (an )n∈N .86 Solutions to the exercises on page 20 1. we obtain L = = = = n→∞ n∈N is convergent with limit 2+3·0 3+3·0 = 2. then (an+1 )n∈N is also convergent with limit L. as (1)n∈N is convergent with n limit 1.
√ √ √ √ √ √  an − L =  an − 0 =  an  = an < . √ √ So ( an )n∈N is convergent with limit L. √ √ Now we show that ( an )n∈N is convergent with limit L.87 From the parts N and D above and the theorem on the algebra of limits. >). we have n2 +n−n = = = = = = +1 . √ 2◦ If L > 0. an − L < = − L . 3. We consider the an − L = an − 0 = an  = an < 2 . Let N ∈ N be such that 2 for all n > N . then := − L > 0. So L ≥ 0. we have an < . +1 1 1+ 1 n . Then we have 2 L an − L ≤ an − L < − . an − L < L. that is. Let only two possible cases. then let N ∈ N be such that for n > N . it follows that 5 (cn )n∈N is convergent with limit 1 = 5. (a) We begin by showing that L ≥ 0. √ √  an − L < √ √ n2 + n + n + n − n) · √ ( n2 + n + n 2 2 n +n−n √ n2 + n + n n √ 2+n+n n n(1) √ 1 n n n2 + n + 1 1 n2 n2 +n n2 √ √ Hence ( an )n∈N (b) For all n ∈ N. namely L = 0 and L > 0: 1◦ If L = 0. √ Then for n > N . 2 and so an < L 2 < 0 for all n > N . we obtain > an − L √ √ √ √ = ( an − L)( an + L) √ √ √ √ =  an − L an + L) √ √ √ √ =  an − L( an + L) and so √ L L √ ≤ √ = . If L < 0. a contradiction to the fact that an ≥ 0 for all n ∈ N. then let N ∈ N be such that for all n > N . an + L L √ is convergent with limit L. Then for all n > N .
that is lim bn ≥ lim an . Moreover. nn n n n n n n 1 Since (0)n∈N and n n∈N are both convergent with the same limit 0. from 1k + 2 k + 3 k + · · · + n k = 0. its limit is lim bn − lim an . n∈N 1+ 1 n 4. we obtain that lim are both convergent with limit 0. Thus by the theorem on algebra of limits. it follows that bn − an ≥ 0 for all n ∈ N. Since the sequences (0)n∈N and the Sandwich theorem. n→∞ n→∞ n→∞ n→∞ Solutions to the exercises on page 22 1. Also n∈N + 1 > 1 > 0. from the Sandwich n! theorem. If for some k ∈ N. we obtain n→∞ n→∞ lim bn − lim an ≥ 0. As an ≤ bn . . it follows that n∈N √ 1 1 = 1. From Exercise 5 on page 13. From the theorem on algebra of limits. For all n ∈ N. n→∞ nk+2 3. 1 is convergent with limit 1+ n and so by the previous part. Clearly (1 + x)1 = 1 + x = 1 + 1 · x. that is. then we have (1 + x)k+1 = (1 + x)k (1 + x) ≥ (1 + kx)(1 + x) (by the induction hypothesis and since 1 + x ≥ 0) = 1 + kx + x + x2 = 1 + (k + 1)x + x2 ≥ 1 + (k + 1)x. For all n ∈ N. Let k ∈ N.88 1 As (1)n∈N is convergent with limit 1 and n n∈N is convergent with limit 0. Hence 1+ n +1 is convergent with limit 1 + 1 = 2(= 0). we have 0≤ 1 2 n−1 n 1 1 n! = · · ···· · ≤ ·1· ···· 1·1 = . (1 + x)k ≥ 1 + kx. nk+2 nk+2 n n 1 1k + 2 k + 3 k + · · · + n k ≤ nk+2 n 1 n n∈N for all n ∈ N. it follows that the sequence (bn − an )n∈N is convergent (being the sum of the convergent sequence (bn )n∈N and the convergent sequence (−an )n∈N ). Hence by induction. 2. it follows that nn n∈N is convergent with the limit 0. Furthermore 1 + n ≥ 0 for all n ∈ N. the result follows. Let x ≥ −1. we conclude that √ is convergent with limit 1 . ( n2 + n − n)n∈N is the sequence √ 1 1 2 1+ n +1 1 convergent with limit 2 . we have 0≤ Thus 0≤ 1k + 2 k + 3 k + · · · + n k nk + n k + n k + · · · + n k n · nk 1 ≤ ≤ k+2 = . it follows 1 1 that 1 + n n∈N is convergent with limit 1. Consider the sequence (bn − an )n∈N . (a) We prove the claim using induction.
n→∞ .89 (b) For all n ∈ N. it follows that the sequence (an − a)n∈N is convergent (being the sum of the convergent sequence (an )n∈N and the convergent sequence (−a)n∈N ). b) for all n ∈ N. Consequently a ≤ L ≤ b. (2.25) Finally.25). we know that n→∞ lim − 1 + an n 1 + an n = lim − n→∞ 1 + lim an n n→∞ 1 + lim an n n→∞ = 0 + lim an n→∞ = n→∞ n→∞ lim an n→∞ lim = lim n→∞ = 0 + lim an . 1 1+ √ n and so 1 1+ √ n 2 n √ 1 ≥ 1 + n · √ = 1 + n. n 1 1+ √ n n 2 n = ≥ (1 + √ n) n .27). using Exercise 3a on page 20.24) (for if n n < 1. 1). Clearly for all n ∈ N. Hence (c) Since lim n→∞ n→∞ n n lim 1 1+ √ n 2 n→∞ = (1 + 0)2 = 1 = lim 1.24). its limit is L − a. 1 1 1 5. we have − n + an < bn < 1 n + an . 4. then n = (n n )n < 1n = 1. b]. For all n ∈ N. So by the sandwich theorem. 1 (2. 1 1 nn ≥ 1 √ 2 √ 2 1 √ 2 1 n n = ( n ) n = n n < (1 + n) n . Moreover. and (2. we obtain L − a ≥ 0. As an ∈ (a. From Exercise 5 on page 13. it follows that lim √ = 0. that is L ≤ b. Consider the sequence (an − a)n∈N . 2 (2. a contradiction!). 1). we have − n < bn − an < n . From Exercise 5 on page 13. and so by adding an . 1 1 = 0. we obtain 1 ≤ n n < (1 + 1 √ n) n ≤ 2 1 1+ √ n 2 (2. its limit is −L + b. 1 Consider the sequence ( n )n∈N contained in (0. it follows that the sequence (b − an )n∈N is convergent (being the sum of the convergent sequence (−an )n∈N and the convergent sequence (b)n∈N ).27) for all n ∈ N. we have an − a ≥ 0. we have b − an ≥ 0. Moreover. Next consider the sequence (b − an )n∈N . that is a ≤ L. it follows that (bn )n∈N is convergent with the limit lim an . b) for all n ∈ N. that is. As an ∈ (a. n→∞ 1 n→∞ Using (2. L ∈ [a.26). (2. it follows from the Sandwich theorem that (n n )n∈N is convergent and 1 lim n n = 1. we obtain −L + b ≥ 0.26) Combining (2. By the theorem on algebra of limits. From the theorem on algebra of limits. It is convergent with limit 0 ∈ (0. From the theorem on algebra of limits.
Let > 0. . 4 appear adjacently 8. it follows that (an )n∈N has a convergent subsequence. 1. 6. 2. . 6. Thus we can choose indices n1 < n 2 < n 3 < . 4. Using the fact that (an )n∈N is Cauchy. 4. 8. 8. 4. . 8. Solution to the exercise on page 26 Let (an )n∈N be a Cauchy sequence. 6. . we have an − L = an − anK + anK − L ≤ an − anK  + anK − L (triangle inequality) < 2 + 2 = .90 Solution to the exercise on page 24 Observe that the terms 2. 8 appear adjacently . So (6)k∈N is a subsequence of the given sequence. such that for all k ∈ N. 4. 8. . 8 appear adjacently and and and and so so so so the the the the terms terms terms terms 8. . 8. 3. . 8. Consequently. . Then for all n > N . 8. 6 appear adjacently and so the terms 3. 2. 2. ank = 6. By the BolzanoWeierstrass theorem. there exists a K ∈ N such that for all nK > N and anK − L < 2 . 1. . 6. . 8. . From Exercise 5 on page 17. 4 appear adjacently 2. 8. it follows that (an )n∈N is bounded. 8 appear adjacently and so the terms 1. 6. . which contains 6. 6 appear adjacently and so the terms 6. 4. . 6. · · · → . 6 appear adjacently and so the terms 1. Since (an )n∈N is Cauchy. there exists a N ∈ N such that for all n. and so 6 appears inﬁnite number of times. say (ank ))k∈N with limit L. · · · →. an − am  < 2 . · · · → . . . Hence we get the loop . (an )n∈N is convergent with limit L. we now prove that (an )n∈N is itself convergent with limit L. m > N . . . 4. 2. 8. As (ank )k∈N is convergent with limit L. · · · → . 8 appear adjacently 6.
Furthermore. f (x) − f (c) < = 2 . Thus in order to make x2 − c2  less than . then we note that δ > 0. Consider = 1 > 0. Then for all x ∈ R satisfying x − c  < δ. We observe that as f (0) ≤ M 0 = M · 0 = 0. f (0) = 0. If δ := δ= . Since f is continuous at c. and let f : R → R be given by f (x) = αx. it follows that f (0) = 0. let δ := min 2c + 1 . 3. Let > 0. Suppose that f is continuous at c. Then f (x + y) = α(x + y) = αx + αy = f (x) + f (y).1 . f (0) = f (0 + 0) = f (0) + f (0). Since > 0. So if x − c < δ. then we obtain √ √ x2 − 02  = x2  = x · x < δ · δ = · = . we choose δ such that δ(δ + 2c) < : indeed. (b) Let c ∈ R and suppose that c = 0. f (x) − f (c) < . First of all.91 Solutions to the exercises on page 29 √ . we deﬁne δ = M . that is. Since the choice of c ∈ R was arbitrary. then x2 − c2  = (x − c)(x + c) = x − c · x + c. if x ∈ R satisﬁes x − c < δ. (a) Let √> 0. So f is continuous at 0. Moreover. M = . Finally we have f (x) − f (c ) + f (c) = f (x) + f (−c ) + f (c) = f (x − c ) + f (c) = f (x − c + c). Hence it follows that −f (c ) = f (−c ). We have the following two cases: 9 Here we use the fact that f is ‘additive’. 4. it follows that δ is positive. we have x2 − c2  = x − c · x + c < δ · (δ + 2c). for all x ∈ R. Then for all x ∈ R satisfying x = x − 0 < δ. Given > 0. So f is continuous at c . (b) Let α ∈ R. there exists a δ > 0 such that for all x ∈ R satisfying x − c < δ. Let c ∈ R. and −x < δ − c ≤ δ − c ≤ δ +  − c = δ + c. (a) Let c ∈ R and > 0. if x ∈ R and x − 0 = x < 1. it follows that f is continuous on R. then we obtain x2 − c2  < δ(δ + 2c) ≤ (1 + 2c) = . If x ∈ R is such that x − c < δ. Thus if x ∈ R satisﬁes x − c < δ. we have f (x) − f (0) = f (x) − 0 = f (x) ≤ M x = M x − 0 < M δ = M Hence f is continuous at 0. we have9 f (x) − f (c ) = f (x) − f (c ) + f (c) − f (c) = f (x − c + c) − f (c) < . then x < c + δ ≤ c + δ ≤ c + δ = c + δ. . 2c + 1 2. since (x − c + c) − c = x − c  < δ. Then ∃δ > 0 such that 2 1 for all x ∈ R satisfying x − c < δ. If x ∈ R. and so f (0) = 0. since 0 = f (0) = f (c − c ) = f (c ) + f (−c ). then x < δ + c. and so x + c ≤ x + c < δ + c + c = δ + 2c.
If n ∈ N. Then (x2 )n∈N is also convergent with limit n c2 . .) Thus the sequence f (x1 ). (Indeed. then there exists qn ∈ Q such that qn − c < Claim: (qn )n∈N is convergent with limit c. Then there exists x ∈ Q such that x − c < δ.3. f is continuous. x2 . . . . f (x1 ). Hence f is not continuous at c. a contradiction. f (x1 ). . . then the sequence x1 . f (x1 ).92 1◦ c ∈ Q. by Theorem 1. 3. Since f (c) 2 But f (x) − f (c) = 1 − 0 = 1 = 1 > 1 . then f (x1 ) = f (x2 ). 2 > 0. x2 . . Thus (f (xn ))n∈N is convergent with limit f (c). If x1 = x2 . x2 . . x2 . f (x1 ). and so it is convergent with limit f (x1 ) (= f (x2 )). then the sequence f (x1 ). . 5. qn − c < < n N This proves the claim. . is divergent. < N.2. Then there exists x ∈ R \ Q such that x − c < δ. Solutions to the exercises on page 31 1. that is. . x2 . is divergent. Let (xn )n∈N be a convergent sequence with limit c. converges to x2 . and so by Theorem 1. By the Archimedean principle. f (x2 ). f (x1 ). there exists an N ∈ N such that Thus for n > N .2. 2◦ c ∈ R \ Q. Consequently f (x1 ) = f (x2 ): for otherwise if f (x1 ) = f (x2 ). 2 But f (x) − f (c) = 0 − 1 =  − 1 = 1 > 1 . f (x2 ). f (x2 ). x1 . a contradiction. 2.3. we have f (c) = lim f (qn ) = lim 0 = 0.2. . f (x1 ) . x1 . is divergent. f (x2 ) . x2 . while the subsequence x2 . 2 > 0. the function f is onetoone. . Proof Let > 0. x1 . . .6. c − δ < x < c + δ). . . Since f is continuous at c. f (x) − f (c) < f (c) . converges to x1 . it follows that the sequence x1 . the subsequence x1 . So by Theorem 1. we have f (c) − f (x) ≤ f (c) − f (x) = f (x) − f (c) < and so f (x) > f (c) 2 f (c) . there exists a δ > 0 such that for all x ∈ R satisfying x − c < δ. . 1 1 < . is a constant sequence. . . 2 Thus for all x ∈ R satisfying x − c < δ (that is. So we have shown that if x1 = x2 . x1 . n→∞ n→∞ 1 1 n.
3. that is.93 Solution to the exercise on page 32 We apply Theorem 1.3 several times in order to prove this. Hence the function x → x2 · 1+x2 = 1+x2 is continuous on R. and so we obtain that the function x → 1 + x2 is continuous on R. . we conclude that the function 1 x2 1 x → 1+x2 is continuous on R. Since the function x → x is continuous on R. As 1 + x2 ≥ 1 > 0 for all real x. it follows that the function x → x2 is continuous on R as well. f is continuous on R. Moreover the function x → 1 is continuous on R.
From (2. T ] such that g(d) = min{g(x)  x ∈ [0. 1]. g(d) ≤ g(x) ≤ g(c). f (x)−f (c) < . 1}. n (2. T ] → R. g(c) = max{g(x)  x ∈ [0.3: A discontinuous function on [0. T ]} and So for all x ∈ [0. using the Sandwich theorem. 1] → R be deﬁned by f (x) = Then f is not continuous at 1 2: 0 1 for all 0 ≤ x < 1 . So g is continuous at c. 1] that attains its extreme values. and > 0. b] such that m ≤ f (xn ) < m + 1 . 2 1 2 indeed the sequence − 1 n+1 n∈N is contained in [0. (xn )n∈N has a convergent subsequence (xnk )k∈N with limit d ∈ [a. the continuity of g on [0. g(x)−g(c) = f (x)−f (c) < . m < m + n . then there exists a positive δ such that for all x ∈ R satisfying x−c < δ. 2 = lim 0 = 0 = 1 = f n→∞ n→∞ 1 1 − 2 n+1 1 2 . 3. d ∈ [0. 1 1 For each n ∈ N. given by g(x) = f (x) for all x ∈ [0. b]}. T ] is a trivial consequence of the continuity of f on R: If c ∈ [0. We prove that there exists a d ∈ [a.94 Solutions to the exercises on page 34 1. Let f : [0. But the choice of c was arbitrary.) Applying the extreme value theorem to g. 2 for all 1 ≤ x ≤ 1. and so m + n cannot be a lower bound for {f (x)  x ∈ [a. T ]. (Indeed. 2. 1} = 0 = f (0). that is. However. 1 PSfrag replacements 0 1 2 1 Figure 2. 1]} = inf{0. T ]. So there exists an xn ∈ [a. T ]. we conclude that f (d) = m. and so sup{f (x)  x ∈ [0. and so g is continuous on [0. Since f is continuous at d. 1]} = {0. 1} = 1 = f (1). f (d) ≤ f (x) ≤ f (c). So far we have proved . and it is convergent with limit lim f 1 2 − 0 = 1 . But {f (x)  x ∈ [0.28) By the BolzanoWeierstrass theorem. T ] satisfying x−c < δ. Consider the function g : [0. it follows that (f (xnk ))k∈N is convergent with limit f (d). b]. and inf{f (x)  x ∈ [0. b]}. T ]}. Hence for all x ∈ [0. T ]. Then g is continuous on [0. 1]} = sup{0.28). b] such that f (d) = m. Let m := inf{f (x)  x ∈ [a. T ]. we conclude that there exist c.
b] . and so f∗ is welldeﬁned. (a) If x ∈ (a. then f∗ (x) − f∗ (c) = 0 < then for any z ∈ (c − δ. 2◦ Let c − δ < x ≤ c. b]. b].T ] . f∗ (x) + 2 } = f∗ (x) + 2 . Consequently. b]. Furthermore. 1).b] (y) = f (y) for all y ∈ [x. c). Applying the intermediate value theorem to f [x.) Thus f (x) = f (nT + r) = f (r). it follows that f (d) ≤ f (x) ≤ f (c). respectively.f (c) f (d) g = f [0. So f (c) and f (d) are upper and lower bounds.. Then we have the following two cases: 1◦ Let c < x < c + δ. Let x ∈ [a.b] is a continuous function. then f∗ (y) ≤ max{f∗ (x). f∗ (c) + } = f∗ (c) + . max{f (z)  z ∈ [c. (Indeed.b](y)  y ∈ [a. c]}} ≤ max{f∗ (x). b] be such that x − c < δ. We now prove that f is bounded on R by using the periodicity of f . y]}.. Then we have f∗ (x) = max{f∗ (c). As f (d) ≤ f (r) ≤ f (c). . and so f∗ (x) − f∗ (c) = f∗ (x) − f∗ (c) ≤ . we obtain x = nT + r. Thus trivially. x]} = max{f (y)  y ∈ [a. x]}} ≤ max{f∗ (c). we see that max{f [x.b] denote the restriction of f to [x.4: A continuous periodic function is bounded. we have and so f (z) ≤ f (x) + 2 . where n := T ∈ Z and r := T · Θ ∈ [0. f (c) + } ≤ max{f∗ (c). Since the choice of x ∈ R was arbitrary. the fact that f is bounded on [0. sup B} (why?). 95 Figure 2. b]. f∗ (c) = max{f∗ (x). of the set {f (x)  x ∈ R}. max{f (z)  z ∈ [x. If c − δ < x < c. and so it is bounded.. that is.4. since sup(A ∪ B) = max{sup A. deﬁned by f [x. then let f [x. f (z) − f (x) = f (z) − f (c) + f (c) − f (x) ≤ f (z) − f (c) + f (c) − f (x) = 2 . Let c ∈ [a. Choose δ > 0 such that for all x ∈ [a. if x < y then.. and so f∗ (x) − f∗ (c) = f∗ (c) − f∗ (x) ≤ 2 . Now if x is any real number. max{f (z)  z ∈ [x. (b) First we observe that f∗ is an increasing function. b] such that x−c < δ. we have x x +Θ = T T x where Θ ∈ [0. x]} exists. and let > 0. it follows that if x < y. where r ∈ R is such that r ∈ [0. See Figure 2. f (x) + 2 } ≤ max{f∗ (x). f∗ (x) ≤ f∗ (y). there exists a n ∈ Z such that x = nT + r. We note that f [x. T ). f (x) − f (c) < . it follows that f (d) ≤ f (x) ≤ f (c) for all x ∈ R. −T 0 c d T 2T 3T 4T . T ]. 4. T ). If x = c.
Consequently f∗ is continuous at c. and moreover f (0) = u(0) − d(0) = 0 − H = −H < 0. Let the weekend campsite be at altitude H. Since the choice of c was arbitrary. f (c) = c. 1] → R be the position function for 1 the walk up. 2 > 0. u(c) = d(c).5. b]. Hence by the intermediate value theorem. there exists a c ∈ [0. 1] is continuous on [0. 2 → R given by f (t) = u(t) − d(t). the hiker was 2 exactly at the same spot on Saturday and Sunday. 1] → R and the function x → x from [0. it follows that also the function g : [0. 1] → R deﬁned by g(x) = f (x) − x. See Figure 2. and d : 0. we have f∗ (x) − f∗ (c) < 2 . 1 4 f 0 PSfrag replacements 1 2 1 1 4 f∗ f (d) 0 1 2 1 Figure 2. Since the function f : [0. 1]. 1 such that f (c) = 0. Since 0 ≤ f (x) ≤ 1 for all x ∈ [0. 2. that is. it follows that there exists a c ∈ 0. and g(1) = f (1) − 1 ≤ 0.5: f and f∗ . for all x ∈ [0. 4 See Figure 2. 1]. (We assume 1 that these are continuous functions. Let u : [0. 1] to R are both continuous on the interval [0. 1] such that g(c) = 0. Then f is also continuous. we have g(0) = f (0) − 0 = f (0) ≥ 0. t ∈ 0. (c) f∗ is given by 1 x − x2 if 0 ≤ x ≤ 2 . while 1 1 1 1 1 f =u −d =u −0=u 2 2 2 2 2 1 . So by the intermediate value theorem. that is. .) Consider the function f : 0.6. 2 → R be the position function for the walk down.96 Hence for all x ∈ [a. f∗ (x) = 1 1 if 2 < x ≤ 1. b] satisfying x − c < δ. Solutions to the exercises on page 37 1. 1]. it follows that f∗ is continuous on [a. So at time c past 8:00.
The polynomial function p : [−1. y is not a lower bound of S. 4. Choose a N ∈ N such that N > −f (1) (again the Archimedean property guarantees the existence of such a N ). and so there exists a x1 ∈ R such that f (x1 ) > y. that is. and consider the continuous function g : [0. 1] such that g(x0 ) = 0. that is. 2] → R is continuous on the interval [−1. Applying the intermediate value theorem to g (with y = 0). p(−1) = 2 · (−1)3 − 5 · (−1)2 − 10 · (−1) + 5 = −2 − 5 + 10 + 5 = 8 > 0. But then f (x0 ) is rational. from the intermediate value theorem applied to the continuous function p on the interval [−1. Similarly. since S is not bounded below. that is. But then f (x0 ) is irrational. This shows that S = R. Consider the function g : [0. This completes the proof. 3. Moreover. while g(1) = f (1) − N < 0. ◦ 3 f (0) < 0. it follows that there exists a real number c such that f (c) = y. whereas mx0 is irrational. 1] → R deﬁned by g(x) = f (x) − N x.97 PSfrag replacements H d u(c) = d(c) u 0 c 1 2 1 Figure 2. Note that g(0) = f (0) − N 0 = f (0) > 0. . 2]. we conclude that there must exist a c ∈ [−1. 5. Now consider the restriction of f to the interval with endpoints x0 and x1 with the endpoints included in the interval. 2◦ x0 ∈ Q. As S is not bounded above. Then let x0 = 0 and let m ∈ Z. Since p(−1) > 0 > p(2). 2]. y is not an upper bound of S. so is g. it follows that there exists a x0 ∈ [0. Obviously S ⊂ R. So f cannot be continuous. We observe that g(0) = f (0) < 0. and p(2) = 2 · (2)3 − 5 · (2)2 − 10 · (2) + 5 = 16 − 20 − 20 + 5 = −19 < 0. x ∈ [0.6: The walks up and down. From the part above. it follows that there exists a x0 ∈ R and a m ∈ Z such that f (x0 ) = mx0 . there exists a x0 ∈ R such that f (x0 ) < y. 2] such that p(c) = 0. f (x0 ) = N x0 . 2]. Let y ∈ R. and so by the intermediate value theorem. So p has a real root in the interval [−1. 1]. 2◦ f (0) > 0. (a) The following three cases are possible: 1◦ f (0) = 0. a contradiction. Choose N ∈ N satisfying N > f (1) (that such a N exists follows from the Archimedean property). 1] → R deﬁned by g(x) = f (x) + N x. We now show the reverse inclusion. 1] such that g(x0 ) = 0. f (x0 ) = −N x0 . As the functions f and x → N x are continuous. a contradiction. Clearly f (x0 ) = f (0) = 0 = m0 = mx0 . while mx0 is rational. and g(1) = f (1) + N > 0. We have the following two possible cases: 1◦ x0 ∈ Q. it follows that there exists a x0 ∈ [0. Applying the intermediate value theorem to this continuous function. (b) Suppose that such a continuous function exists.
then it must be [1]. [c] ∈ Z6 . Moreover. [a] ⊗ e = [a] = e⊗[a]. multiplication modulo 6 is a law of composition on Z6 that satisﬁes G1 and G2. then ([a] ⊗ [b]) ⊗ [c] = [ab] ⊗ [c] = [(ab)c] = [a(bc)] = [a] ⊗ [bc] = [a] ⊗ ([b] ⊗ [c]). G3. and [a] ⊕ [−a] = [a + (−a)] = [0] = [−a + a] = [−a] ⊕ [a].) Indeed [1] serves as an identity element. . [c] ∈ Z6 . Multiplication modulo 6 is associative: if [a]. since for all [a] ∈ Z6 . [b]. since if [a]. (a) Addition modulo 6 is a law of composition on Z6 . we verify below that the group axioms are satisﬁed. So Z6 with addition modulo 6 forms a group. G1. G2. then for all [a] ∈ Z6 . [b] ∈ Z6 . If [a] ∈ Z6 . then [−a] ∈ Z6 . (If e = [k] is the identity element for some integer k.98 Algebra Solutions to the exercises on page 44 1. G3 is not satisﬁed. G2. Addition modulo 6 is associative: if [a]. [a] ⊗ [1] = [a · 1] = [a] = [1 · a] = [a] ⊗ [a]. [0] is the identity element: for all [a] ∈ Z6 . We give its group table below: ⊕ [0] [1] [2] [3] [4] [5] [0] [0] [1] [2] [3] [4] [5] [1] [1] [2] [3] [4] [5] [0] [2] [2] [3] [4] [5] [0] [1] [3] [3] [4] [5] [0] [1] [2] [4] [4] [5] [0] [1] [2] [3] [5] [5] [0] [1] [2] [3] [4] (b) No. but G3 is not true: G1. So there does not exist an inverse of the element [0] in Z6 . we obtain e = [k] = [1k] = [1]⊗[k] = [1] ⊗ e = [1]. [0] ∈ Z6 . then [a] ⊕ [b] = [a + b] ∈ Z6 . [b]. then ([a] ⊕ [b]) ⊕ [c] = [a + b] ⊕ [c] = [(a + b) + c] = [a + (b + c)] = [a] ⊕ [b + c] = [a] ⊕ ([b] ⊕ [c]). In particular. [0] ⊗ [a] = [a] ⊗ [0] = [0] = [1]. with [a] = [1] ∈ Z6 . For instance. So if there is an identity element e. we have [0] ⊕ [a] = [0 + a] = [a] = [a + 0] = [a] ⊕ [0]. but for all [a] ∈ Z6 .
99 Hence Z6 with multiplication modulo 6 is not a group. Z∗ with multiplication modulo 6 is also not a group, since multiplication modulo 6 is 6 not a law of composition on the set Z∗ : indeed, [2], [3] ∈ Z∗ , but [2] ⊗ [3] = [2 · 3] = 6 6 [6] = [0] ∈ Z∗ . 6
(c) If: Let m be a prime number. First we show that multiplication modulo m is a law of composition on Z∗ . Let m [a], [b] ∈ Z∗ . Then a and b are not divisible by m, and since m is prime, it follows that m ab is also not divisible by m. Thus [a] ⊗ [b] = [ab] = [0], and so [a] ⊗ [b] ∈ Z∗ . m Next we show that the group axioms are satisﬁed: G1. If [a], [b], [c] ∈ Z∗ , then m ([a] ⊗ [b]) ⊗ [c] = [ab] ⊗ [c] = [(ab)c] = [a(bc)]
= [a] ⊗ [bc] = [a] ⊗ ([b] ⊗ [c]). G2. [1] = [0] and so [1] ∈ Z∗ . Furthermore, For all [a] ∈ Z∗ , we have m m [a] ⊗ [1] = [a · 1] = [a] = [1 · a] = [1] ⊗ [a]. G3. If [a] ∈ Z∗ , then m does not divide a (for otherwise [a] = [0]). As m is prime, it m follows that gcd(a, m) = 1. (Indeed, gcd(a, m) divides m, and since m is prime, the only factors of m are 1, m, −1, −m. As gcd(a, m) ≥ 0, the only possible values are m and 1. Also, gcd(a, m) divides a, and since a is not divisible by m, it follows that gcd(a, m) = m. So the only possible value of gcd(a, m) is 1.) So there exist s, t ∈ Z such that as + mt = 1. Consequently [s] ⊗ [a] = [sa] = [as] = [a] ⊗ [s] = [as] = [1 − mt] = [1] ⊕ [m(−t)] = [1] ⊕ [0] = [1].
Hence Z∗ with multiplication modulo m is a group. m
Furthermore [s] = [0], since otherwise ms, and so mas + mt, that is m1, a contradiction. Hence [s] ∈ Z∗ , and so [s] is the inverse of [a]. m
Only if: Suppose that m = p · q, where p, q ∈ {2, 3, . . . , m − 1}. Clearly, as m does not divide p or q, it follows that [p] = [0] and [q] = [0]. So [p], [q] ∈ Z∗ . However, m [p] ⊗ [q] = [p · q] = [m] = [0] ∈ Z∗ , m and so multiplication modulo m is not a law of composition on Z∗ . Consequently, Z∗ m m with multiplication modulo m is not a group. 2. (a) Composition of functions gives a law of composition on the set Sn of all bijections from {1, 2, 3, . . . , n} onto itself. Indeed if f, g are bijections from {1, 2, 3, . . . , n} onto itself, then f ◦ g is again a bijection from {1, 2, 3, . . . , n} onto itself:
100 i. f ◦ g is onetoone. Let k1 , k2 ∈ {1, 2, 3, . . . , n}. If (f ◦ g)(k1 ) = (f ◦ g)(k2 ), then f (g(k1 )) = f (g(k2 )), and since f is onetoone, we get g(k1 ) = g(k2 ). Furthermore, since g is onetoone, k1 = k2 . ii. f ◦ g is onto. If k ∈ {1, 2, 3, . . . , n}, then since f is onto, there exists a k ∈ {1, 2, 3, . . . , n} such that f (k ) = k. Moreover, since g is onto, there exists a k ∈ {1, 2, 3, . . . , n} such that g(k ) = k . Thus (f ◦g)(k ) = f (g(k )) = f (k ) = k. So f ◦ g is onto.
Next we show that the group axioms are satisﬁed: G1. If f, g, h ∈ Sn , then for all k ∈ {1, 2, 3, . . . , n},
((f ◦ g) ◦ h)(k) = (f ◦ g)(h(k)) = f (g(h(k))) = f ((g ◦ h)(k)) = (f ◦ (g ◦ h))(k). So (f ◦ g) ◦ h = f ◦ (g ◦ h). G2. The identity function ι : {1, 2, 3, . . . , n} → {1, 2, 3, . . . , n} deﬁned by ι(k) = k for all k ∈ {1, 2, 3, . . . , n}, serves as the identity element in Sn . Indeed, for all f ∈ Sn , we have (f ◦ ι)(k) = f (ι(k)) = f (k) = ι(f (k)) = (ι ◦ f )(k), ∀k ∈ {1, 2, 3, . . . , n}. Thus f ◦ ι = f = ι ◦ f . G3. If f ∈ Sn , then since f is a bijection, it follows that for all k ∈ {1, 2, 3, . . . , n}, there exists a unique element k such that f (k ) = k. Deﬁne f −1 : {1, 2, 3, . . . , n} → {1, 2, 3, . . . , n} as follows: if k ∈ {1, 2, 3, . . . , n}, then f −1 (k) = k , where k ∈ {1, 2, 3, . . . , n} is such that f (k ) = k. Then for all k ∈ {1, 2, 3, . . . , n}, we have (f ◦ f −1 )(k) = f (f −1 (k)) = k = ι(k) = k = f −1 (f (k)) = (f −1 ◦ f )(k), Hence Sn with the composition of functions is a group. (b) The number of bijections from a set with n elements onto itself is n!, and so the order of Sn is n!. Indeed, in order to specify the bijection f , one needs to specify f (1), . . . , f (n). The number of ways of specifying f (1) is n (as it can be any one of the numbers 1, . . . , n). As f is onetoone, f (2) can be only be any one of the elements of the set {1, 2, 3, . . . , n} \ {f (1)} (which has n − 1 elements). Proceeding in this way, the number of distinct bijections we get are n · (n − 1) · · · · · 2 · 1 = n!. (c) Let f : {1, 2, 3} → {1, 2, 3} be the function f (1) = 2 f (2) = 1 f (3) = 3, and g : {1, 2, 3} → {1, 2, 3} be the function g(1) = 1 g(2) = 3 g(3) = 2. Then (f ◦ g)(1) = f (g(1)) = f (1) = 2 = 3 = g(2) = g(f (1)) = (g ◦ f )(1), and so f ◦ g = g ◦ f . and so f ◦ f −1 = ι = f −1 ◦ f .
101 (d) If: S1 has the only element ι : {1} → {1}, and so ι ◦ ι = ι = ι ◦ ι. Thus S1 is abelian. S2 has the two elements ι : {1, 2} → {1, 2}, and r : {1, 2} → {1, 2} given by r(1) r(2) Since ι ◦ r = r ◦ ι we see that S2 is abelian. So if n ≤ 2, then Sn is abelian. Only if: If n > 2, then consider the bijections f : {1, 2, 3, . . . , n} → {1, 2, 3, . . . , n} and g : {1, 2, 3, . . . , n} → {1, 2, 3, . . . , n} given by f (1) = 2 f (2) = 1 f (n) = n ∀n ∈ {3, . . . , n} Then (f ◦ g)(1) = f (g(1)) = f (1) = 2 = 3 = g(2) = g(f (1)) = (g ◦ f )(1), and so f ◦ g = g ◦ f . Thus Sn is not abelian. Hence Sn is abelian iﬀ n ≥ 2. 3. We note that if a b −b a a b −b a (aa − bb )2 + (ab + ba )2 a −b b a , a −b b a ∈ S, , g(1) = 1 g(2) = 3 g(3) = 2 g(n) = n ∀n ∈ {1, 2, 3, . . . , n} \ {1, 2, 3} = 2 = 1.
and
.
then
=
aa − bb ab + ba −(ab + ba ) aa − bb
and = a2 a 2 − 2aa bb + b2 b 2 + a2 b 2 − 2ab ba + b2 a 2 = a2 (a 2 + b 2 ) + b2 (b 2 + a 2 ) = (a2 + b2 )(a 2 + b 2 ) = 0. Hence matrix multiplication is a law of composition on S. Next we verify that the group axioms are satisﬁed: G1. Matrix multiplication is associative, and so for all A, B, C ∈ S, (AB)C = A(BC). G2. Let I denote the identity matrix. Then we have I= 1 0 0 1 = 1 0 −0 1
and moreover, 12 + 02 = 1 = 0. So I belongs to S. Furthermore, for every A ∈ S, AI = A = IA. G3. Let A= Then a2 + b2 = 0. If B := a b −b a
a a2 +b2 −b a2 +b2
∈ S.
b a2 +b2 a a2 +b2
,
(a) Let a. Hence a ∗ x = b has the unique solution x = a−1 ∗ b. it follows that (a ∗ b)−1 = b−1 ∗ a−1 . Since a ∈ G. and furthermore. where e denotes the identity element in G. say x1 and x2 . then a ∗ (a−1 ∗ b) = (a ∗ a−1 ) ∗ b = e ∗ b = b. If there are two solutions. 4. Then we have b = e ∗ b (since e is the identity element) = (a−1 ∗ a) ∗ b (since a−1 is the inverse of a) = a−1 ∗ (a ∗ b) (associativity) = a−1 ∗ (a ∗ c) (since a ∗ b = a ∗ c) = (a−1 ∗ a) ∗ c (associativity) = e ∗ c (since a−1 is the inverse of a) = c (since e is the identity element). (b) Clearly and so x = a−1 ∗ b is a solution to a ∗ x = b. Since the inverse of a ∗ b is unique. and by part 4a above. Thus b−1 ∗ a−1 is an inverse of a ∗ b. a ∗ a−1 = e = a−1 ∗ a and b ∗ b−1 = e = b−1 ∗ b. So S is a group with matrix multiplication. b ∈ G. We have (a ∗ b) ∗ (b−1 ∗ a−1 ) = a ∗ (b ∗ (b−1 ∗ a−1 )) (associativity) = a ∗ ((b ∗ b−1 ) ∗ a−1 ) (associativity) (c) Since a.102 then a 2 + b2 a 2 + b 2 + b2 a 2 = a2 + b 2 1 = 2 = 0. there exists a−1 ∈ G such that a ∗ a−1 = e = a−1 ∗ a. and = a ∗ (e ∗ a−1 ) (since b−1 is the inverse of b) = a ∗ a−1 (since e is the identity element) (b−1 ∗ a−1 ) ∗ (a ∗ b) = ((b−1 ∗ a−1 ) ∗ a) ∗ b (associativity) = (b−1 ∗ (a−1 ∗ a)) ∗ b (associativity) = (b−1 ∗ e) ∗ b (since a−1 is the inverse of a) = b−1 ∗ b (since e is the identity element) = e (since b−1 is the inverse of b). and suppose that a ∗ b = a ∗ c. 2 + b 2 )2 (a a + b2 So B ∈ S. b. AB = I = BA. . it follows that x1 = x2 . b−1 ∈ G such that = e (since a−1 is the inverse of a). a ∗ x1 = b = a ∗ x 2 . there exist elements a−1 . c ∈ G.
3. As H is a subgroup and a. H3 hold: H1. it follows that a ∗ b ∈ K as well. The subgroup {−1. b also belong to G. the sum of the odd integers 1 and −1 is 0. Also. Consequently k belongs to H as well as K. . 1]. So there exist integers k1 and k2 such that k = 4k1 and k = 6k2 . then k ∈ H and k ∈ K. Claim: H ∩ K = {12m  m ∈ Z} = {. Hence {12m  m ∈ Z} ⊂ H ∩ K. we know that a ∗ b = b ∗ a. . 1 is a nonnegative integer. as K is a subgroup. (f + g) Consequently. 12. (b) We check that H1. . it follows that a ∗ b ∈ H. g ∈ H1 . Conversely. it follows that a−1 ∈ K. b ∈ H ∩ K.}. 8. . Also. (a) FALSE. Thus 4k1 = 6k2 and so 2k1 = 3k2 . Hence k ∈ {12m  m ∈ Z}. since if m is an odd integer. 4. . e ∈ H. Thus {0} is a ﬁnite subgroup of the inﬁnite group Z of integers with addition. . since K is a subgroup and a ∈ K. b ∈ H. b ∈ H. −12. . Thus e ∈ H ∩ K. then k = 12m for some m ∈ Z. Hence 4k and 6k. . which is not an odd integer. −12. (c) TRUE. . 0. then f and g are continuous on the interval [0. then a. Although H1 and H2 hold. −12. . H3 hold: H1. Since H is a subgroup and a ∈ H. But H3 is true. Proof If k ∈ H ∩K. then a ∈ H and a ∈ K.103 Solutions to the exercises on page 46 1. So H1 and H2 do not hold.}. . . 1 f 1 = 0 and g 2 = 0. Similarly. −6. 4. b belong to both H and K. For instance. 12. it follows 2 that f + g is also continuous on [0. 0. it follows that e ∈ K as well. H1 and H2 are not true. As H is a subgroup. If a. Furthermore. 0. it follows that 2k2 . Consequently. then so is −m. 1} of the set of nonzero real numbers with multiplication is another example. As f. and so k is a multiple of 4 and 6. . 1]. H2. 18. we have that a. . f + g ∈ H1 . it follows that a−1 ∈ H. and so k = 6k2 = 6(2k2 ) = 12k2 . . Since integers possess a unique factorization into primes. Thus k = 4(3m) = 6(2m). . the set {e} comprising the identity element e is a subgroup of the group G. 6. Yes. −18. H3. and moreover. as K is a subgroup and a. Consequently a−1 ∈ H ∩ K. H3 does not hold. Hence a ∗ b ∈ H ∩ K. H2. −4. 1]. (a) We check that H1. 12. If f. and K = {6m  m ∈ Z} = {. So H ∩ K is a subgroup of G. }. Indeed. (b) FALSE. g are continuous on the interval [0. 1 2 =f 1 2 +g 1 2 = 0 + 0 = 0. . and as G is abelian. given any inﬁnite group G. . So we have shown that H ∩ K ⊂ {12m  m ∈ Z}. b ∈ K. (a) We note that H = {4m  m ∈ Z} = {. If a ∈ H ∩ K. k2 = 2k2 . . −8. H2. Indeed. 2. if k ∈ {12m  m ∈ Z}. For all a. but its inverse −1 is not a nonnegative integer.
H2.104 H2. deﬁned by 0(x) = 0 for all x ∈ [0. for all x ∈ [0. 1]. deﬁned by (−p)(x) = −p(x) for all x ∈ [0. = −a0 + (−a1 )x + (−a2 )x2 + · · · + (−an )xn . . (b) We check that H1. then for all x ∈ [0. an . 1]. 1]. On the other hand. then p and q are continuous on the interval [0. If f ∈ H1 . As (−p)(x) = −p(x) = −(a0 + a1 x + a2 x2 + · · · + an xn ) p(x) = a0 + a1 x + a2 x2 + · · · + an xn . (p + q)(x) = p(x) + q(x) = a 0 + a 1 x + a 2 x2 + · · · + a n xn + b 0 + b 1 x + b 2 x2 + · · · + b m xm = a0 + b0 + (a1 + b1 )x + (a2 + b2 )x2 + · · · + (am + bm )xm + am+1 xm+1 + · · · + an xn . 1 2 = −f 1 2 = −0 = 0. 1]. 1] with addition of functions. 1]. H3 hold: H1. If p. . So H1 is a subgroup of the group C[0. (p + q)(x) = p(x) + q(x) = a 0 + a 1 x + a 2 x2 + · · · + a n xn + b 0 + b 1 x + b 2 x2 + · · · + b m xm and so p + q ∈ H2 . 1] with addition of functions. 1] with addition of functions. then f 1 = 0. and so p + q ∈ H2 . The constant function 0 : [0. deﬁned by 0(x) = 0 for all x ∈ [0. for all x ∈ [0. The inverse of p in C[0. 1] → R. and moreover. q ∈ H2 . 1] → R. . is the identity element in the group C[0. b1 . m ∈ N ∪ {0} and real numbers a0 . is the identity element in the group C[0. an . 0 ∈ H2 (with n = 0 and a0 = 0). If p ∈ H2 . . Moreover. 1] is the function −p : [0. it follows that −p ∈ H2 . and so 0 ∈ H1 . 0 1 2 = 0. As (−f ) it follows that −f ∈ H1 . 5. a2 . b0 . 2 deﬁned by (−f )(x) = −f (x) for all x ∈ [0. 1]. bm . The constant function 0 : [0. then there exists an n ∈ N ∪ {0} and real numbers a0 . such that p(x) = a0 + a1 x + a2 x2 + · · · + an xn . a1 . 1]. H2. . then for all x ∈ [0. a1 . and = a0 + b0 + (a1 + b1 )x + (a2 + b2 )x2 + · · · + (an + bn )xn + bn+1 xn+1 + · · · + bm xm . 1] → R. for all x ∈ [0. H3. The inverse of f in C[0. Clearly. . 1] is the function −f : [0. . 1]. . . for all x ∈ [0. 1] → R. 1]. there exist n. if n > m. . a2 . . such that q(x) = b0 + b1 x + b2 x2 + · · · + bm xm . 1]. If n ≤ m. So H2 is a subgroup of the group C[0. . 1]. 1] with addition of functions. . Let e denote the identity in the group G. H3.
So z ∈ Z(G). Hence z −1 ∈ Z(G). (a−1 )−1 ∗ z −1 = z −1 ∗ (a−1 )−1 . So Z(G) is a subgroup of the group G. Thus Z(G) is not empty.29) . and from the uniqueness of inverses. we have a−1 ∈ G. z ∗ a = a ∗ z. (c) We check that H1. since G is abelian. e ∗ a = a = a ∗ e.105 (a) Since e ∈ G. Let z ∈ Z(G). Then for every a ∈ G. a ∗ e = a = e ∗ a. β α . . and for every a ∈ G. But a ∗ a−1 = e = a−1 ∗ a. Hence G ⊂ Z(G). R)). (2. (b) Clearly Z(G) ⊂ G. R). α 0 β α Thus = and so it follows that β = 0. it follows that (a−1 )−1 = a. then for every a ∈ G. Let z ∈ G. and so z ∗ a−1 = a−1 ∗ z. z2 ∈ Z(G). Hence we have shown that Z(GL(2. e ∈ Z(G) since for every a ∈ G. it follows that e ∈ Z(G). H2. H3. (d) If z = α γ β δ Thus α α+β γ γ+δ and so it follows that γ = 0 and α = δ. H2. R)) ⊂ α·I = α 0 0 α α ∈ R \ {0} . H3 hold: H1. If z1 . α β γ δ ∈ Z(GL(2. Also. = 1 1 0 1 α+γ γ α γ β+δ δ β δ . we have (z1 ∗ z2 ) ∗ a = z1 ∗ (z2 ∗ a) = z1 ∗ (a ∗ z2 ) = (z1 ∗ a) ∗ z2 = (a ∗ z1 ) ∗ z2 (associativity) (since z2 ∈ Z(G)) = a ∗ (z1 ∗ z2 ) (associativity). For every a ∈ G. then since a= it follows that 1 1 0 1 1 1 0 1 = ∈ GL(2. Thus we obtain a ∗ z −1 = (a−1 )−1 ∗ z −1 = z −1 ∗ (a−1 )−1 = z −1 ∗ a. since b= it follows that 1 0 1 1 1 0 1 1 α+β α β α ∈ GL(2. (associativity) (since z1 ∈ Z(G)) Thus z1 ∗ z2 ∈ Z(G). R). Thus Z(G) = G. that is. = 1 0 1 1 α β α β+α α 0 . Consequently (z ∗ a−1 )−1 = (a−1 ∗ z)−1 .
R). Then α · I ∈ GL(2. . Let m ≥ 0. We know that S = {e. am ∗ an = am+n . So by induction. . Step 1. a. . and so the result is true for n = 0. If am ∗ ak = am+k for some k ≥ 0. as k2 ≤ G and k1 ≥ 0. a2 . . it follows that a has ﬁnite order. ∈ GL(2. 2. But the choice of m ≥ 0 was arbitrary. R). 3. .30). . Clearly am ∗ a0 = am ∗ e = am = am+0 . We show that am ∗ an = am+n for all n ≥ 0 by induction on n. Thus ord(a) = min{m ∈ N  am = e} ≤ k2 − k1 ≤ G. Proof We prove this in several steps. Thus α 0 0 α α ∈ R \ {0} ⊂ Z(GL(2. then we have am ∗ ak+1 = am ∗ (ak ∗ a) = (am ∗ ak ) ∗ a = am+k ∗ a = a(m+k)+1 (deﬁnition of ak+1 ) (associativity) (induction hypothesis) (deﬁnition of a(m+k)+1 ) = am+(k+1) . and so we have shown that for all m ≥ 0. we obtain k2 − k1 ≤ G. (2. Hence α · I ∈ Z(GL(2. am ∗ an = am+n .106 Let α ∈ R \ {0}. R)) = α 0 0 α α ∈ R \ {0} . and all n ≥ 0. we conclude that Z(GL(2. it follows that there exist distinct k1 .31) . . Let m < 0 and n ≥ 0. G} such that ak1 = ak2 . we have am ∗ an = am+n for all n ≥ 0. R). As k2 − k1 ∈ N. and so ak2 −k1 = e. and for all p q r s we have α 0 0 α p q r s = αp αq αr αs = pα qα rα sα = p q r s α 0 0 α . 2. Step 2. a3 . aG } ⊂ G. Claim: For all m.29) and (2. 1. We may assume that k2 > k1 (otherwise we can interchange them). So from the pigeonhole principle. Then we have the following two cases: (2. k2 ∈ {0. and that S has G + 1 elements. while G has G elements. R). n ∈ N. Solutions to the exercises on page 48 1. Furthermore. . So a has order at most equal to G.30) From (2.
where the last equality follows from the following fact: for all k ∈ Z. am ∗ an = am+n . Claim: For all m.31). (am )n = amn . Then from Step 2. We observe that am ∗ a−m = (a−m )−1 ∗ a−m = e. If n < 0. we then obtain am+n = am ∗ an . Hence we have shown that for all m ∈ Z and all n < 0. we obtain am ∗ an = an+m = am+n . n ∈ N. we know that ak = (a−k )−1 . then we have (am )k+1 = (am )k ∗ am = amk ∗ am = amk+m = am(k+1) . and so the statement holds with n = 0. we have am+n ∗ a−n = a(m+n)−n = am . then −k ≥ 0. and all n ≥ 0. by induction on n. namely (2.31). we obtain for all m ∈ Z. By postmultiplying both sides by an = (a−n )−1 . we have an ∗ a−n−m = an+(−n−m) = a−m . we obtain that for all integers m and n. 2◦ If n + m ≤ 0. then from (2. Let m ∈ Z and n < 0. Combining this result with the result from Step 2. am ∗ an = am+n . Hence by taking inverses. Next we prove the following claim. and so a−m ∗ an+m = an . So by premultiplying by am = (a−m )−1 and postmultiplying by an+m = (a−n−m )−1 . Combining this result with the result from Step 1. We have 1 1 −1 0 1 1 −1 0 1 1 −1 0 1 1 −1 0 1 1 −1 0 1 1 −1 0 1 −1 = 2 1 1 −1 0 1 1 −1 0 1 1 −1 0 1 1 −1 0 1 1 −1 0 1 1 −1 0 . Let m ∈ Z. = 4 3 = 5 4 = 6 5 = . and so it follows that for all m ∈ Z and all n ≥ 0. then we have (am )n = (am )(−n) = (am(−n) )−1 = (a−mn )−1 = amn . If for some k ≥ 0 there holds that (am )k = amk . we obtain (a−k )−1 = ((ak )−1 )−1 = ak . But the choice of m ∈ Z was arbitrary. and so a−k = (a−(−k) )−1 = (ak )−1 . (2. . am ∗an = am+n .107 1◦ If n + m > 0.32). Hence we have shown that for all m < 0. So by induction.32) Step 3. we have that (am )n = amn for all n ≥ 0. Proof First we show that for all m ∈ Z and all n ≥ 0. am ∗ an = am+n . by the deﬁnition of ak for a negative k. If k ≥ 0. .) 3. We have (am )0 = e = a0 = am0 . for k < 0. (am )n = amn . (am )n = amn . and all n ≥ 0. . we have a−m ∗ an+m = a−m+(n+m) = an . and so an ∗ a−n−m = a−m . So we obtain an+m = e ∗ an+m = (am ∗ a−m ) ∗ an+m = am ∗ (a−m ∗ an+m ) = am ∗ an . 1 = 3 2 1 1 −1 0 1 1 −1 0 1 1 −1 0 1 1 −1 0 1 1 −1 0 = = = = = 1 1 −1 0 0 1 −1 −1 −1 0 0 −1 −1 −1 1 0 0 −1 1 1 1 1 −1 0 1 1 −1 0 1 1 −1 0 1 1 −1 0 1 1 −1 0 = = = = = 0 1 −1 −1 −1 0 0 −1 −1 −1 1 0 0 −1 1 1 1 0 0 1 . (a−k )−1 = ak (Indeed. . then from (2.
1 is a generator of Z with addition. then b ∗ a has inﬁnite order as well. As −1 = Z. Hence if a ∗ b has inﬁnite order. that is. a ∗ (b ∗ a)m−1 ∗ b = e. and ord(b∗a) ≤ m. we have (ψ ◦ ϕ)(a ∗ b) = ψ(ϕ(a ∗ b)) (deﬁnition of ψ ◦ ϕ) = ψ(ϕ(a) ∗ ϕ(b)) (since ϕ is a homomorphism) = ψ(ϕ(a)) ∗ ψ(ϕ(b)) (since ψ is a homomorphism) = (ψ ◦ ϕ)(a) ∗ (ψ ◦ ϕ)(b) (deﬁnition of ψ ◦ ϕ). then b ∗ a has ﬁnite order m. we obtain a ∗ b has ﬁnite order m iﬀ b ∗ a has ﬁnite order m. So we see that b∗a has ﬁnite order. (a ∗ b) ∗ · · · ∗ (a ∗ b) = e. Yes.33) by b. (2. we obtain Solutions to the exercises on page 51 1. we obtain that m ≤ ord(b ∗ a). So the orders of a ∗ b and b ∗ a are the same. and postmultiplying by b (b ∗ a) ∗ (b ∗ a)m−1 = e. (m−1) times that is. b in G. (b) We have and so ψ ◦ ϕ : G → G is a homomorphism. ker(ψ ◦ ϕ) = {a ∈ G  (ψ ◦ ϕ)(a) = e } = {a ∈ G  ψ(ϕ(a)) = e } = {a ∈ G  ϕ(a) ∈ ker(ψ)} = ϕ−1 (ker(ψ)). 4. . 5. By interchanging a and b. then (a ∗ b)m = e.33) −1 . So the cyclic group Z with addition does not have a unique generator. that is.108 Thus ord 1 1 −1 0 = min m ∈ N 1 1 −1 0 m = 1 0 0 1 = 6. m times and so a ∗ (b ∗ a) ∗ · · · ∗ (b ∗ a) ∗b = e. So we have shown that if a ∗ b has ﬁnite order m. we see that −1 is also a generator of Z with addition. say m. (b∗a)m = e. (a) For all a. If a ∗ b has ﬁnite order. Hence by premultiplying (2. since 1 = Z. By interchanging the roles of a and b. and so ord(b ∗ a) = m.
we have ϕ−1 (ϕ(a)) = a. then ϕ(a) = e . For all a ∈ G. (b) If m = n. b ∈ G such that ϕ(a) = a and ϕ(b) = b . Thus ϕ−1 : G → G is onetoone. ϕ : Z → a is an isomorphism. and if m < n. then clearly ϕ−1 (a ) = ϕ−1 (b ) (for otherwise a = ϕ(ϕ−1 (a )) = ϕ(ϕ−1 (b )) = b ). if m > n. For all a. If a ∈ ker(ϕ). if b ∈ a . and so ϕ : Z → a is a homomorphism. Let a. = a−1 ∗ b−1 = ϕ(a) ∗ ϕ(b) and so ϕ : G → G is a homomorphism. Hence ϕ is onetoone. 4. First we show that ϕ−1 : G → G is a homomorphism. it is a bijection. So ϕ−1 is a homomorphism. and so ϕ−1 (a ∗ b ) = a ∗ b = ϕ−1 (a ) ∗ ϕ−1 (b ). n ∈ Z. a = (a−1 )−1 = (b−1 )−1 = b). we have a−1 ∈ G and ϕ(a−1 ) = (a−1 )−1 = a. We have ϕ(a ∗ b) = ϕ(a) ∗ ϕ(b) = a ∗ b . and so ϕ : G → G is onetoone. b ∈ G . Then a−1 = b−1 (for otherwise. we have ϕ(a ∗ b) = (a ∗ b)−1 = b−1 ∗ a−1 (deﬁnition of ϕ) (Exercise 4c from the exercises on page 44) (since G is abelian) (deﬁnition of ϕ). b ∗ a ∗ b−1 ∈ ker(ϕ). As ϕ : G → G is onetoone and onto. Let us denote the homomorphism m → am from Z to a by ϕ. . and so ϕ−1 : G → G is onto. we have ϕ(m+n) = am+n = am ∗an = ϕ(m)∗ϕ(n). If a = b . then an−m = e. Consequently ϕ : G → G is an isomorphism. Hence ϕ−1 : G → G is a bijection. ϕ(b∗a∗b−1 ) = ϕ(b)∗ ϕ(a)∗ ϕ(b−1 ) = ϕ(b)∗ e ∗ ϕ(b−1 ) = ϕ(b)∗ ϕ(b−1 ) = ϕ(b)∗ (ϕ(b))−1 = e . b ∈ G. 3. Let a . For all a ∈ G. Thus for all b ∈ G. 5. and so ϕ : G → G is onto. We have ϕ(m) = am = b. Hence ϕ(a) = a−1 = b−1 = ϕ(b). in either case we get a contradiction to the fact that a has inﬁnite order). it follows that ∀a ∈ ker(ϕ) and ∀b ∈ G. Consequently ϕ−1 : G → G is an isomorphism. Since the choice of a ∈ ker(ϕ) was arbitrary. then b = am for some m ∈ Z.109 2. and so ϕ is onto. b ∈ G and a = b. then ϕ(m) = am = an = ϕ(n) (for otherwise. Since ϕ : G → G is a bijection. and so b ∗ a ∗ b−1 ∈ ker(ϕ). then am−n = e. it follows that there exist unique a. Furthermore. and so ker(ϕ) is a normal subgroup of G. (a) For all m. So ϕ is a bijection and consequently.
consider the group Z of integers with addition. G2. (b) TRUE. given by h → h ∗ a. for all A ∈ G. it follows that in particular for elements A. G1. (b) Clearly H ⊂ G. H3 are satisﬁed: H1. Moreover the group axioms are satisﬁed: and furthermore. If x y ∈ G. For example. y := 0). for all h ∈ H. is a bijection. (a) Matrix multiplication is a law of composition on G: if A = We have AB = x y 0 1 x 0 y 1 = xx 0 xy + y 1 and xx > 0. The identity matrix I= 1 0 0 1 ∈ G (x := 1 > 0. Thus a ∗ H. there holds (AB)C = A(BC). We now check that H1. As in part 1b above. H2. (a) FALSE. if b ∈ a ∗ H. and 1 x y −x 1 0 Moreover. we obtain h1 = h2 . C from G. x 0 0 1 ∈ H. and it follows that the cardinalities of H and a ∗ H are the same. then there exists a h ∈ H such that b = a ∗ h. and let H be the subgroup of even integers. So f is a bijection. Thus 1 x > 0. x y 0 1 0 1 x y −x 1 ∈ G. Hence f is onto. 2. then x > 0 and x > 0. one can show that the function g : H → H ∗ a. (c) TRUE. Consider the function f : H → a ∗ H given by h → a ∗ h. Then 0 = 2. If a ∈ G and h1 . Furthermore. . H and H ∗ a all have the same cardinalities.110 Solutions to the exercises on page 54 1. 0 1 then x > 0. Consequently f (h) = a ∗ h = b. So G is a group with matrix multiplication. x y 0 1 and B = x 0 y 1 belong to G. Thus f is onetoone. for all h ∈ H. If x 0 0 1 . B. G3. h2 ∈ H are such that a ∗ h1 = a ∗ h2 . then by premultiplication with a−1 . but 0 + H = 2 + H (both these cosets are equal to H). = 1 0 0 1 = 0 1 x y −x 1 x 0 y 1 . and so AB ∈ G. Then it can be seen that f is a bijection as follows. clearly we have AI = A = IA. As matrix multiplication is associative.
See Figure 2. The right coset of H corresponding to the element a= is Ha = = x 0 xx0 0 0 1 xy0 1 x0 0 y0 1 x ∈ R and x > 0 x0 0 y0 1 ∈G x ∈ R and x > 0 . H3. 1 x 0 1 . Thus the right coset Ha is a straight line passing through the origin and the point (x0 . See Figure 2. H2. and the inverse of x 0 0 1 in the group G. Thus H is a subgroup of the group G. So 1 x > 0. If 1 0 0 1 . y0 ). and it passes through the point (x0 .8. x 0 0 1 ∈ H.7.111 then x > 0 and x > 0. Clearly the identity element from G. So 0 1 x 0 0 1 x 0 0 1 x 0 = xx 0 0 1 0 1 ∈ H. Then the corresponding left coset of H is aH = = x0 0 xx0 0 y0 1 y0 1 x 0 0 1 x ∈ R and x > 0 x ∈ R and x > 0 . y0 ). namely the identity matrix I= also belongs to H. then x > 0. namely the matrix 0 also belongs to H. Thus the left coset aH is a straight line parallel to the xaxis. (c) Let a= x0 0 y0 1 ∈ G. . We have x 0 and xx > 0.
it follows from Corollary 2.1. a ∈ G. 4. a ∈ G (represented by lines parallel to the xaxis).7: Partition of the group G (represented by the right half plane: x > 0. b ∈ K. (a) If p does not divide a. The identity element in the group H is the identity element e from G. . [p − 1]} p p has cardinality equal to p − 1. If a. 3. The set Z∗ = {[1]. b ∈ H ∩ K. As gcd(3. b ∈ H and a. So H ∩ K is a subgroup of H. H3. Thus by Lagrange’s theorem. it follows that H ∩ K is also a subgroup of K. y aH a PSfrag replacements0 H x Figure 2.8 it follows that [a]p−1 = [1]. it follows that H ∩ K = {e}. Hence a ∗ b ∈ H ∩ K. As [1] is the identity element in the group Z∗ with p multiplication modulo p. then [a] = [0]. As H and K are subgroups of G. and so it comprises just one element. Thus e ∈ H ∩ K. By interchanging the roles of H and K. As e ∈ H ∩ K. then a ∈ H and a ∈ K. Furthermore. y ∈ R) by left cosets aH. . and H ∩ K divides K = 5. . a−1 ∈ H and a−1 ∈ K. it follows that H ∩ K divides 1. If a ∈ H ∩ K.8 that S4 cannot have an element of order 16. As K is a subgroup of G. it follows that e also belongs to K. the following hold: H1. then a. Consequently a−1 ∈ H ∩ K. 5. . we have that H ∩ K divides H = 3. 5) = 1. The cardinality of the group S4 is 4! = 24. So H ∩ K = 1. from Corollary 2. As H and K are subgroups. .8: Partition of the group G by right cosets Ha.112 y a PSfrag replacements0 aH H x Figure 2.1. and so [a] ∈ Z∗ . H ∩ K is a subset of the group H. As 16 does not divide 24. H2. it follows that a ∗ b ∈ H and a ∗ b ∈ K.
Proceeding in a similar manner. it also divides a(ap−1 − 1) = ap − a. [4] = [0] in Z7 . then p also divides ap . and consequently. so that in Z7 . [22225555] = [2222]5555 = [3]5555 . it follows that [3]7−1 = [3]6 = [1]. Indeed. Hence [55552222 ] = [4]2222 = [4]370·6+2 ] = ([4]6 )370 [4]2 = [1][4]2 = [4]2 = [−3]2 = [(−3)2 ] = [32 ]. that is. Thus p divides ap−1 − 1. 5555 = 793 · 7 + 4. Finally. So ap ≡ a(mod p). and hence it divides ap − a as well. [a]p−1 = [1]. [4]7−1 = [4]6 = [1]. from part (5b) above. As [3] = [0] in Z7 . Thus we obtain [22225555 ] = [3]5555 = [3]925·6+5 = ([3]6 )925 [3]5 = [1]925 [3]5 = [1][3]5 = [3]5 . . then [a] = [0]. Hence [ap−1 ] = [1]. ap ≡ a(mod p). and so by part (5b). we can show that [55552222 ] = [32 ]. and so by part 5a above. (c) Note that 2222 = 317 · 7 + 3. and so [ap−1 − 1] = [0]. If p does not divide a. Again. so that [55552222] = [5555]2222 = [4]2222 . [22225555 + 55552222 ] = [35 + 32 ] = [32 · 28] = [0].113 (b) If p divides a. which proves that 722225555 + 55552222.
we have ((an )n∈N + (bn )n∈N ) + (cn )n∈N = (an + bn )n∈N + (cn )n∈N = ((an + bn ) + cn )n∈N = (an + (bn + cn ))n∈N = (an )n∈N + (bn + cn )n∈N = (an )n∈N + ((bn )n∈N + (cn )n∈N ). which is not invertible. 2. For all (an )n∈N .34) This proves the claim. Moreover. (cn )n∈N in R∞ . (2. G4. (bn )n∈N ∈ R∞ . α · v = 0. and furthermore. G2. V4 are also satisﬁed: V1. So matrix addition. we have (an )n∈N + (0)n∈N = (an + 0)n∈N = (an )n∈N = (0 + an )n∈N = (0)n∈N + (an )n∈N . First we check that R∞ . V2. Finally. . Consequently it is not a vector space.1) = α−1 · (α · v) (using (2.114 Solutions to the exercises on page 58 1. (A. then (−an )n∈N ∈ R∞ .34)) = (α−1 α) · v (using V2) = (1) · v (since α−1 α = 1) = v (using V1). For instance the sum of two invertible matrices may not be invertible. I + (−I) = 0. we have (an )n∈N + (bn )n∈N = (an + bn )n∈N = (bn + an )n∈N = (bn )n∈N + (an )n∈N . (b) NO. For all (an )n∈N ∈ R∞ . is an abelian group. and so it does not form an abelian group.4). V3. (bn )n∈N . −I. Clearly (2. The scalar multiplication of the real number 0 with an invertible matrix is the zero matrix. Next we check that V1. we can observe that the scalar multiplication of the real number 0 with an invertible matrix is the zero matrix. However. with addition deﬁned by (2. B) → A + B is not a law of composition on the set of invertible matrices. G3. then we obtain 0 = α−1 · 0 (Theorem 2.4) gives a law of composition on R∞ . 3. since for all (an )n∈N . we have: G1. which is not invertible. commutativity holds. The sequence (0)n∈N serves as the identity element: for all (an )n∈N ∈ R∞ . which is not invertible. (an )n∈N +(−an )n∈N = (an +−an )n∈N = (0)n∈N = (−an +an )n∈N = (−an )n∈N +(an )n∈N . (a) NO. The identity matrix I is invertible. If (an )n∈N ∈ R∞ . 1 · (an )n∈N = (1an )n∈N = (an )n∈N .2. Let α ∈ R and v ∈ V be such that Then if α = 0. Alternately. and so is its additive inverse.
as U2 is a subspace. V4. Consequently α · v ∈ U 1 ∩ U2 . . If v1 . (α + β) · (an )n∈N = ((α + β)an )n∈N = (αan + βan )n∈N = (αan )n∈N + (βan )n∈N = α · (an )n∈N + β · (an )n∈N . S2. v2 belong to U2 . Thus 0 ∈ U1 ∩ U2 . α · (β · (an )n∈N ) = α · (βan )n∈N = (α(βan ))n∈N = ((αβ)an )n∈N = (αβ) · (an )n∈N . Consider the subspaces U1 = span of R2 . 0 ∈ U1 and 0 ∈ U2 . v2 belong to U1 ∩ U2 . (a) FALSE. Solutions to the exercises on page 62 1. x in R2 can be represented by a point in the (x. U2 be susbpaces of the vector space V . Hence v1 + v 2 ∈ U 1 ∩ U 2 . as U2 is a subspace. As U1 is a subspace. For all α. S1.5). v1 + v2 ∈ U2 . then v1 . y)plane.4) and scalar multiplication deﬁned by (2. 1 0 and U2 = span 0 1 ∈ U1 ∪ U2 . v1 + v2 ∈ U1 . S3 hold for U1 ∩ U 2 : So U1 ∩ U2 is a susbpaces of the vector space V . (b) TRUE. (bn )n∈N ∈ R∞ . it follows that α · v ∈ U2 . S2. and so U1 ∪ U2 is not a vector space. 1 1 ∈ span 0 1 . α · ((an )n∈N + (bn )n∈N ) = α · (an + bn )n∈N = (α(an + bn ))n∈N = (αan + αbn )n∈N = (αan )n∈N + (αbn )n∈N = α · (an )n∈N + α · (bn )n∈N .115 V2. U2 are susbpaces. So R∞ is a vector space with addition deﬁned by (2. As U1 is a subspace. Moreover. and the (Each vector y above can be seen pictorially in Figure 2. Let U1 .9). Also. We check that S1. it follows that α · v ∈ U1 . v2 belong to U1 and v1 . As U1 . For all α. β ∈ R and all (an )n∈N ∈ R∞ . β ∈ R and all (an )n∈N ∈ R∞ . ∈ span Thus S2 is not satisﬁed. Let α ∈ R and v ∈ U1 ∩U2 . V3. For all α ∈ R and all (an )n∈N . Thus v ∈ U1 and v ∈ U2 . Then v1 := but v1 + v 2 = since 1 1 1 0 1 0 ∈ U1 ⊂ U1 ∪ U2 and v2 := 1 1 and 0 1 ∈ U2 ⊂ U 1 ∪ U2 . S3.
v3 are linearly independent.116 y R2 PSfrag replacements U1 x v1 v2 v1 + v 2 U2 Figure 2. Then we have α1 · v1 + α2 · v2 + α3 · v3 + 0 · v4 = α1 · v1 + α2 · v2 + α3 · v3 + 0 = α1 · v1 + α2 · v2 + α3 · v3 = 0. since 1 · v1 + 1 · v2 + 1 · v3 + (−1) · v4 = 0. 1 2. v4 . it follows that α1 = α2 = α3 = 0. (d) FALSE. 3 4 1 1 1 0 0 (3) · 0 + (−3) · 2 + (2) · 3 = 0 . v2 . Indeed. (a) Clearly. in R3 . 1 t1 1 t2 ⊂ R2 . 0 0 are linearly dependent. (In order to prove the reverse inclusion. 1 1 0 0 (e) TRUE. while the vectors v1 . We have 1 1 0 2 = 1 · 3 + (−1) · 1 . v3 . 1 1 v4 := 1 . So v1 .9: The union U1 ∪ U2 of subspaces U1 and U2 is not a subspace. v2 . the vectors 0 1 v1 := 0 . and by the independence of v1 . span . = y. (c) TRUE. v2 . v2 := 1 . then we obtain the linear equations α+β αt1 + βt2 = x. . 1 t2 0 v3 := 0 . v3 are linearly independent. Suppose that there exist scalars such that α1 · v1 + α2 · v2 + α3 · v3 = 0. For instance. observe that if x y =α· 1 t1 +β· . (f) FALSE.
(bn )n∈N belong to ∞ . = y − xt2 · t1 − t 2 1 t1 + xt1 − y · t1 − t 2 1 t1 1 t2 ∈ span 1 t2 1 t1 .) We now complete the proof as ∈ R2 . (b) Suppose that U1 . We have: . that is. it follows that inﬁnitely many elements from S belong to one of the subspaces. an  ≤ M . n}. We have: S2. . t1 t2 As Uk is a subspace of R2 . t2 such that 1 1 . it follows that span 1 t1 . αan  = αan  ≤ (α + 1)M . The sequence (0)n∈N is the zero vector in R∞ . As t1 − t2 = 0. . .4. Thus c is a subspace of ∞ . Thus for all n ∈ N. 3. an + bn  ≤ an  + bn  ≤ M1 + M2 . From the result of the part above. The sequence (0)n∈N is the zero vector in . . is a subspace of R∞ . S3. . Indeed. and it belongs to ∞ . ∞ is a subspace of R∞ . Let α ∈ R and (an )n∈N ∈ ∞ . and so Uk = R2 . and so the sequence (αan )n∈N is bounded. Un are subspaces of R2 . Thus there exist real numbers t1 . . the sequence (αan )n∈N is also convergent.2. c0 is a subspace of c. . since it is convergent. M2 > 0 such that for all n ∈ N. Then there exists an M > 0 such that for all n ∈ N. (an )n∈N + (bn )n∈N belongs to ∞ . α · (an )n∈N belongs to c. 1 t2 ⊂ Uk . α · (an )n∈N belongs to ∞ . and so the sequence (an + bn )n∈N is bounded. Hence for all n ∈ N.2. So ∞ S1. an  ≤ M1 and an  ≤ M2 . S2. Then by Theorem 1. we have x y So R2 ⊂ span Hence the claim follows. ∞ c is a subspace of . then there exist M1 .4. it follows that the sequence (an + bn )n∈N is also convergent. we obtain R2 ⊂ Uk . If (an )n∈N . . that is Uk is not a proper subspace. the nth term of the sequence (0)n∈N is equal to 0. (bn )n∈N belong to c.117 which have the solution α = follows: suppose that y−xt2 t1 −t2 and β = x y xt1 −y t1 −t2 . 1 t2 . and it belongs to c. . and so (an )n∈N + (bn )n∈N belongs to c. say Uk for some k ∈ {1. If (an )n∈N . that is. Since the inﬁnite set 1 S := t∈R t is contained in R2 = U1 ∪ · · · ∪ Un . since it is bounded: indeed for all n ∈ N. . that is. there holds: ∞ S1. Let α ∈ R and (an )n∈N ∈ c. and 0 = 0 ≤ 1. then by Theorem 1. ∈ Uk . such that U1 ∪ · · · ∪ Un = R2 . S3.
N2 ∈ N such that ∀n > N1 .2. 0). Then by Theorem 1. 0(0) = 0 and 0(1) = 0. since 0 is continuous on [0. Consequently y1 = 0 = y 2 . Let f ∈ S(0. y2 ). S2. an = 0. then by Theorem 1. S3. since for all n > 1. S3. Indeed. and it belongs to c00 . The sequence (0)n∈N is the zero vector in c. (1)n∈N ∈ c. there holds: S1. Let α ∈ R and (an )n∈N ∈ c00 . (bn )n∈N belong to c00 . 1]. Thus (2·f )(0) = y1 and (2·f )(1) = y2 . y2 ) is a subspace of C[0. and so 2y1 = y1 and 2y2 = y2 . If (an )n∈N . Then we have: S1. 0) and α ∈ R.) 4. Hence (an )n∈N + (bn )n∈N = (an + bn )n∈N belongs to c00 . The zero function 0 ∈ S(0. and it belongs to c0 . The sequence (0)n∈N is the zero vector in c0 . Then α · f is continuous on [0. If f. n→∞ n→∞ and so (an )n∈N + (bn )n∈N belongs to c0 .2. 1]. If: Let y1 = 0 = y2 .4. y2 ). then there exist N1 . the sequence (αan )n∈N is also convergent and lim αan = α lim an = α0 = 0. Let α ∈ R and (an )n∈N ∈ c0 . (bn )n∈N belong to c0 . n→∞ n→∞ that is. then f + g ∈ S(0. g ∈ S(0.4. .118 S1. Hence c00 is a subspace of c0 . an = 0 and ∀n > N2 . α · (an )n∈N belongs to c0 . (α · f )(0) = αf (0) = α0 = 0 and (α · f )(1) = αf (1) = α0 = 0. since it is a convergent sequence with limit equal to 0. as f. n∈N (n)n∈N ∈ R∞ . Only if: Suppose that S(y1 . Indeed. N2 } ∈ N. Thus with N := max{N1 . Then there exists an N ∈ N such that for all n > N . 0). 1]. (The examples 1 n ∈ c0 . S2. S3. 0) is a subspace of C[0. c00 is a subspace of c0 . we obtain an + bn = 0 + 0 = 0. 1] and moreover. ((−1)n )n∈N ∈ ∞ . so is f + g. Hence c0 is a subspace of c. Thus (αan )n∈N = α · (an )n∈N belongs to c00 . it follows that the sequence (an + bn )n∈N is also convergent and n→∞ lim (an + bn ) = lim an + lim bn = 0 + 0 = 0. bn = 0. 0). Hence S(0. S2. 1]. If (an )n∈N . show that each of the inclusions are strict. an = 0. g are continuous on [0. then 2 · f ∈ S(y1 . If f ∈ S(y1 . and so αan = α0 = 0. and furthermore. for all n > N . and also (f + g)(0) = f (0) + g(0) = 0 + 0 = 0 and (f + g)(1) = f (1) + g(1) = 0 + 0 = 0.
. . . 1 . 1 . fd (x) = xd (2. We prove this by contradiction. that is. . . 1 0 0 B2. . f1 . As v ∈ V = span(B). . . So v is a linear combination of vectors from V . . So we obtain (α1 − α1 ) · v1 + · · · + (αn − αn ) · vn = 0. . If α. We have: B1. Since B1. . . . 1]. . 3. . Suppose that C[0. . . . 1 . it follows that there exist scalars α1 . Suppose that B = {v1 . So B is linearly independent. Now we prove the following. . . β. . suppose that there exist scalars α1 . fd ∈ C[0. . 1 0 0 1 0 0 z Thus 1 1 1 span 0 . and by the independence of v1 . 1]. = 0. z then 1 1 1 1 1 x 1 y = (x−y)· 0 +(y−z)· 1 +z· 1 ∈ span 0 . . vn . . . .35) The functions f1 . and let v ∈ V . B2 hold it follows that B is a basis. . . If x y ∈ R3 . = 0. fd are all polynomials and so they are all continuous on [0. and so it follows that α = β = γ = 0. that is α1 = α1 . . say. . αn such that v = α1 · v1 + · · · + αn · vn . 2. it then follows that α1 − α1 = · · · = αn − αn = 0. . αn such that α1 · v 1 + · · · + α n · v n = v = α 1 · v 1 + · · · + α n · v n . Consider the d functions f1 . . 0 1 0 0 then we obtain α+β+γ β+γ γ = 0. 1 = R3 .119 Solutions to the exercises on page 64 1. In order to prove uniquness. . . 1]. . αn = αn . . fd deﬁned by f1 (x) = x f2 (x) = x2 3 f3 (x) = x for all x ∈ [0. 1] is a ﬁnite dimensional vector space with dimension d. γ are scalars such that 0 1 1 1 α · 0 +β · 1 +γ · 1 = 0 . vn } is a basis of the vector space V .
120 Claim: The functions f1 , . . . , fd given by (2.35) are linearly independent in C[0, 1]. Proof Suppose that there exist scalars α1 , . . . , αd , not all zeros, such that α1 · f1 + · · · + αd · fd = 0. Then for all x ∈ [0, 1], (α1 · f1 + · · · + αd · fd )(x) = 0(x), that is, for all x ∈ [0, 1], α1 x + · · · + αd xd = 0.
Let k ∈ {1, 2, . . . , d} be the smallest number such that αk = 0. Then we obtain for all x ∈ [0, 1], αk xk + αk+1 xk+1 + · · · + αd xd = 0,
and so for all x ∈ [0, 1], xk (αk + αk+1 x + αk+2 x2 + · · · + αd xd−k ) = 0. For all n ∈ N, 1 0 < n < 1, and so we have for all n ∈ N, that is, for all n ∈ N, Hence 0 = =
n→∞
1 n
k
αk + αk+1
1 n
+ αk+2
1 n
2
+ · · · + αd
1 n
d−k
= 0,
αk + αk+1
1 n
+ αk+2
1 n
2
+ · · · + αd
1 n
d−k
= 0.
lim 0 lim αk + αk+1 1 n + αk+2 1 n
2
n→∞
+ · · · + αd
1 n
d−k
= αk ,
= αk + 0 + · · · + 0
a contradiction. So f1 , . . . , fd are linearly independent in C[0, 1]. Consequently, by Theorem 2.2.3, it follows that {f1 , . . . , fd } is a basis of C[0, 1], and in particular, they span the whole space C[0, 1]. The constant function 1 : [0, 1] → R, deﬁned by 1(x) = 1 for all x ∈ [0, 1] clearly belongs to C[0, 1], and so, from the above, there must exist scalars β1 , . . . , βd such that 1 = β 1 · f1 + · · · + β d · fd . Thus for all x ∈ [0, 1], 1(x) = (β1 · f1 + · · · + βd · fd )(x), that is, for all x ∈ [0, 1], 1 = β 1 x + · · · + β d xd .
In particular, with x = 0, we obtain 1 = 0, a contradiction. So C[0, 1] is not a ﬁnite dimensional vector space. 4. (a) We have B1. Let (an )n∈N ∈ c00 . Then there exists an N ∈ N such that for all n > N , an = 0. Clearly (an )n∈N = a1 · e1 + · · · + an · en ∈ span({ek  k ∈ N}). So c00 ⊂ span(B).
121 On the other hand, suppose that k1 , . . . , km ∈ N, and α1 , . . . , αm be scalars. If N := max{k1 , . . . , km }, then for all n > N , the nth term of the sequence α 1 · e k1 + · · · + α m · e km is α1 0 + · · · + αm 0 = 0. Thus α1 · ek1 + · · · + αm · ekm ∈ c00 . Consequently span(B) ⊂ c00 . So span(B) = c00 . B2. Suppose that k1 < · · · < km are natural numbers, and α1 , . . . , αm be scalars, not all zeros, such that α1 · ek1 + · · · + αm · ekm = (0)n∈N .
Let l ∈ {1, . . . , m} be the largest number such that αl = 0. Then equating the kl th term on both sides, we obtain that αl = 0, a contradiction. So B is linearly independent. (b) Indeed, if B were a basis for R∞ , then the sequence (1)n∈N would be a linear combination of elements from B. But as seen above, any linear combination of elements from B is in c00 , that is, it is a sequence that is eventually zero. Clearly (1)n∈N ∈ c00 , and so it is not a linear combination of elements from c00 .
Solutions to the exercises on page 68
1. (a) We have ker(TA ) = = = = x y x y x y 0 0 . TA x y = = 0 0 0 0
x+y y x=y=0
We depict this set in the (x, y)plane in Figure 2.10. y R2
x PSfrag replacements ker(TA )
Figure 2.10: Kernel of TA . x x−y x = TA , ∈ R2 , then y y y and so R2 ⊂ im(TA ). Hence im(TA ) = R2 . We depict this set in the (x, y)plane in Figure 2.11. Clearly im(TA ) ⊂ R2 . On the other hand, if
122 y
im(TA ) = R2
x PSfrag replacements
Figure 2.11: Image of TA . (b) We have ker(TA ) = = = x y x y x y TA x y = = 0 0 0 0
x+y 0
x+y =0 .
We depict this set in the (x, y)plane in Figure 2.12. y R2
x PSfrag replacements ker(TA )
Figure 2.12: Kernel of TA . We have im(TA ) ⊂ S := since for all Conversely, since x 0 = TA x 0 , x y ∈ R2 , x 0 TA
∈ R2 x y =
x∈R , x+y 0 ∈ S.
it follows that S ⊂ im(TA ). Hence im(TA ) = S. We depict this set in the (x, y)plane in Figure 2.13.
y)plane in Figure 2.123 y R2 im(TA ) x PSfrag replacements Figure 2. y ker(TA ) = R2 x y x y TA x y = 0 0 x ∈ R and y ∈ R x PSfrag replacements Figure 2. −1 0 +T 0 1 = + 0 1 = −1 1 . 2. since T while T −1 0 + 0 1 =T −1 1 −1 0 = (−1)2 1 1 = 1 1 . (c) We have ker(TA ) = = = R2 . y)plane in Figure 2. L1 is not satisﬁed. We depict this set in the (x. Clearly im(TA ) = 0 0 .13: Image of TA .14. . We depict this set in the (x.15.14: Kernel of TA .
and so L2 holds.15: Image of TA .124 y R2 x PSfrag replacements im(TA ) Figure 2. if x1 x2 = 0 x1 x2 If α = 0. and x1 x2 T ∈ R2 . 3. (a) We verify that L1 and L2 hold: . we have α· x1 x2 x1 x2 T α· =T 0 0 = 0 0 =0· T x1 x2 . and all ∈ R2 . If α ∈ R \ {0} and T x1 x2 α· ∈ R2 . x1 x2 =α· T x1 x2 . then we have x1 x2 αx1 = T αx2 α2 x2 1 αx2 αx2 = αx1 αx2 x2 1 α · x2 x2 = α · x1 x2 = α· T if x1 x2 = 0 if x1 x2 = 0 if x1 x2 = 0 . Thus for all α ∈ R.
and so L ∈ im(T ). . L. . L2. im(T ) ⊂ R. (b) We have ker(T ) = {(an )n∈N ∈ c  T ((an )n∈N ) = 0} = {(an )n∈N ∈ c  lim an = 0}. Thus T ((L)n∈N ) = L. L. For all (an )n∈N and (bn )n∈N in c.4) = α · T ((an )n∈N ).2. . For all α ∈ R. n→∞ and so ker(T ) is the set comprising all convergent sequences with limit 0. then the constant sequence L. if L ∈ R.2. is convergent with limit L. the subspace c0 . . So T is a linear transformation from c to R. and all (an )n∈N ∈ c. that is. So R ⊂ im(T ). Moreover. we have T ((an )n∈N + (bn )n∈N ) = T ((an + bn )n∈N ) = = n→∞ lim (an + bn ) lim an + lim bn n→∞ n→∞ (Theorem 1. (c) Clearly. im(T ) = R. Consequently.125 L1. we have T (α · (an )n∈N ) = T ((αan )n∈N ) = n→∞ lim αan n→∞ = α lim an (Theorem 1.4) = T ((an )n∈N ) + T ((bn )n∈N ).
126 .
Binmore. [4] P. [5] W. 1982.R. [3] V. Principles of Mathematical Analysis. Discrete Mathematics. 3rd Edition.G. 1996. 2002.L. Clarendon Press. Cambridge University Press.Bibliography [1] N. Cambridge University Press. Rudin. Yet Another Introduction to Analysis. Biggs. Finite Dimensional Vector Spaces. 1990. 1976. (Chapter 20 on Groups. Mathematical Analysis: A Straightforward Approach.) [2] K. Bryant. Halmos. Springer. McGrawHill. 127 .
128 Bibliography .
43 ﬁnite order. 3 minimum. 52 periodic function. 55 divergent sequence. 34 interval. 3 linear combination of vectors. 31 abelian group. 64 inﬁnite group. 43 order of an element in a group. 1 maximum. 66 Lagrange’s theorem. 42 absolute value. 48 decreasing sequence. 45 inﬁmum. 15 normal subgroup. 39 least upper bound. 53 law of composition on a set. 5 absolute value of a function. 61 lower bound. 41 generator of a group. 4 associativity. group element with. 10 counting formula. 5 inverse of an element in a group. 40 isomorphism. 26 greatest upper bound. 60 linear transformation. 2 inﬁnite dimensional of a vector space. 26 general linear group. group element with. 63 BolzanoWeierstrass theorem. set. 33. 61 linearly independent vectors. 6 product of functions. 50 kernel of a linear transformation. 15 induced law of composition. 40 group axioms. 40 129 group table. 13 center of a group. 31 . 1 bounded below. 1 Cauchy sequence. 40 iﬀ. 5 distributive laws. 51 kernel of a homomorphism. 65 linearly dependent vectors. 42 continuity of a function at a point. 66 increasing sequence. 14 bounded set. 62 extreme value theorem. 51 order of a group. 47 fractional part of a real number. 15 dimension of a vector space. 50 image of a linear transformation. 40 basis of a vector space. 43 inﬁnite order. 64 distance. 47 intermediate value theorem. 61 linearly dependent set. 54 ﬁnite dimensional vector space. set. 33 Fermat’s little theorem. 7 image of a homomorphism. 2 group. 2 least upper bound property. 47 partition. 47 commutativity. 61 linearly independent set. 48 greatest integer part of a real number. 25 bounded above. 34 bounded sequence. 31 Archimedean property. 27 convergent sequence. 1 bounded function. 11 eventually zero sequence. 3 monotone sequence. 32 positive deﬁniteness. 27 continuous function. 34 polynomial function. 43 identity element in a group.Index kth power of a function. 64 ﬁnite group. 53 cyclic.
55 vector addition. 44 symmetric matrix. 9 span of a subset of a vector space. 46.130 proper subspace. 55 zero vector. 25 sequence. 55 Index . 32 right cosets. 59 symmetry. 23 subspace of a vector space. 50 subgroup. 1 upper triangular matrix. 6 triangle inequality. 60 special linear group. 2 symmetric group. 21 scalar multiplication. 53 Sandwich theorem. 45 subsequence. 55 vector space. 6 upper bound. 45. 59 rational function. 59 vector. 59 sum of functions. 55 seaview property. 31 supremum.
This action might not be possible to undo. Are you sure you want to continue?