The Real Numbers: Terminology
The Real Numbers: Terminology
The Real Numbers: Terminology
(a) x : a x a ) is called an
neighborhood of a
Definition A point x is an accumulation point for G R if every neighborhood of x
meets G in a point other than x.
Example Accumulation Points
(a) If a
n
is a sequence converging to limit L, then L is an accumulation point for the set
A a
n
: n N)
(b) We have already mentioned that supA need not belong to A R. It is an exercise to
show that if does not belong to A, then is an accumulation point for A. A similar remark
applies to inf A, A R.
Theorem 1.9 (Bolzano-Weierstrass theorem) Every bounded, infinite subset of must have at
least one accumulation point.
Solved Problems
Representation of the reals
Problem 1.1 Convert each of the following rational numbers to a decimal representation;
1.
1
8
,
11
80
,
111
800
2.
1
7
,
1
27
,
1
271
3.
1
6
,
17
66
,
2493
9900
Solution: By carrying out the division (e.g., using a calculator) we find
1.
1
8
. 12500,
11
80
. 137500,
111
800
. 1387500
2.
1
7
0. 142857142857,
1
27
. 037037,
1
271
. 0036900369
3.
1
6
. 166,
17
66
: . 2575757,
2493
9900
: . 251818
8
The decimal representations in group 1 are all examples of what are called terminating
decimal representations; they consist of all zeroes from some point onwards. The second
group are all examples of periodic decimal representations; they consist of the same block
of digits, repeated over and over. The length of the block of digits that is repeated is called
the period of the representation. The rational numbers in the third group are examples of
eventually periodic representations. They become periodic after some point in the
expansion. The first two numbers in this group begin to repeat after one digit while the third
number begins to repeat after two digits.
Using a description of the real line as being composed of a union of intervals of unit
length, each of which can be decomposed into intervals of length
1
10
, which can, in turn, be
decomposed further into intervals of length
1
100
etc, etc ... we can devise an algorithm for
associating any decimal expansion with a sequence of nested intervals. It then follows from
the nested interval theorem that every decimal expansion can be identified with a unique
point on the real line (we have to agree to disqualify decimal expansions that eventually
become all 9s in order for the association to be one to one). The decimal expansions that
are terminating, periodic or eventually periodic correspond to rational numbers and all
others correspond to irrational numbers.
Problem 1.2 Convert the following decimal representations to a rational number:
1. a. x . 0027100271 y . 12343434
b. x . 5000 y . 4999
Solution:
1. a. x . 0027100271 (5 digit repeating pattern )
10
5
x 271. 0027100271
(10
5
1)x 271 i. e. , x
271
10
5
1
271
99999
y . 12343434
100y 12. 3434 (2 digit repeating pattern)
10
4
y 1234. 3434
(10
4
10
2
)y 1222 y
1222
10
4
10
2
1222
9900
1. b x . 5000 y . 4999
10x 5. 000 10y 4. 999
x
1
2
100y 49. 999
90y 45 so y
1
2
This last example illustrates why we have to disqualify representations terminating in a
string of 9s if we want the association between real numbers and decimal representations
to be one to one.
Problem 1.3 Prove that the square of any odd integer is necessarily odd.
Solution: An odd integer has the form N 2n 1 for some integer, n. Then
N
2
(2n 1)
2
4n
2
4n 1
2(2n
2
2n) 1 2m 1
9
where m 2n
2
2n is an integer. Thus N
2
is odd whenever N is odd. Similarly, if N is even
then N 2n and we can show in the same way that every even integer has an even square.
Problem 1.4 Prove that 2 is irrational.
Solution: Suppose 2
p
q
where the fraction has been reduced to lowest terms; i.e., p
and q have no factors in common except 1. Then this assumption implies that p
2
2q
2
,
which means that p
2
is an even integer. Then p must be an even integer by the result in the
previous problem, and we can write p 2m for some integer, m. But this leads to 4m
2
2q
2
or q
2
2m
2
which implies that q must also be even. Since our original assumption implies
that p and q have no factors in common, this is a contradiction to our assumption that
2
p
q
. Then there are no integers p, q such that 2
p
q
.
Mathematical Induction
Problem 1.5 Use mathematical induction to prove that for every positive integer M,
k1
M
k
2
M(M 1)(2M 1)
6
(1)
Solution: Let the set of integers M for which (1) is valid be denoted by A. Then 1 A since
1
2
1 - 2 - 3
6
. Now suppose that A contains all the integers from 1 up to the positive integer
m. Then it follows from (1) that
k1
m
k
2
(m 1)
2
m(m 1)(2m 1)
6
(m 1)
2
m(m 1)(2m 1)
6
6(m
2
2m 1)
6
2m
3
9m
2
13m 6
6
Now
(m 1)(m 2)(2m 3) 2m
3
9m
2
13m 6
hence
k1
m1
k
2
(m 1)(m 2)(2m 3)
6
and since this is just (1) with M replaced by m 1, we have proved that m 1 A whenever
m A. Then A is an inductive set and by the principle of mathematical induction, A N.
Problem 1.6 Use mathematical induction to prove that for x 1,
(1 x)
n
1 nx n N (1)
Solution: Let the set of integers n for which (1) is valid be denoted by A. Clearly 1 A.
Now suppose that A contains all the integers from 1 up to the positive integer m. Then it
follows from multiplying both sides of (1) (with n m ) by the positive number (1 x), that
(1 x)
m1
(1 mx)(1 x) 1 (m 1)x mx
2
1 (m 1)x
i.e.,
(1 x)
m1
1 (m 1)x.
10
Since this is just (1) with m replaced by m 1, we have proved that m 1 A whenever
m A. Then A is an inductive set and by the principle of mathematical induction, A N.
Properties of and
Problem 1.7 Prove that the set N of natural numbers has no upper bound.
Solution:We make use of the fact that N is an inductive set. In fact, we sometimes define
N by saying that 1 belongs to N and if n is any element of N, then n 1 is also an element of
N. Suppose there is an upper bound for N. Since N is a nonempty subset of R, then N has a
least upper bound, say . Then 1 is not an upper bound for N, which is to say, there
exists n N such that 1 n. But in this case n 1 and since N is an inductive set,
n 1 N. This contradiction to the definition of means there can be no upper bound for
N.
Problem 1.8 Prove that the every nonempty subset of, N, the natural numbers, has a first
element
Solution:Let T denote a nonempty subset of N, and suppose k T. Let S denote the
intersection of T with the following finite subset of N, 1, 2, , k); i.e., S T 1, 2, , k).
Since S is finite, it contains a first (smallest) element, p. Now for any t T, t p, we must
have p t. For if t S, then t k p, so t p, and if t is an element of S, then t p, since
p is the smallest element of S and t p. We have proved t p for all t T, so p is the first
element in T. This property of the natural numbers is referred to as the well ordering
property.
Problem 1.9 Prove that every nonempty set of reals having a lower bound has a GLB.
Solution:Let S denote a nonempty set of real numbers having a lower bound . Let
T t s for s S ; i.e., T is the reflection of S through the origin. Then every lower
bound for S is mapped onto an upper bound for T.
Since s for all s in S, and since s is equivalent to s , it follows that t
for all t T. That is, is an upper bound for T. Now the completeness axiom asserts the
existence of a least upper bound for T, call it
. . Then
is
the least upper bound for T.
Problem 1.10 Prove the Archimedean property; i.e., for every z R, there exists n N
such that z n. This asserts that there are no infinitely large real numbers.
Solution:Let z R and define a set A n N : n z) as the set of all natural numbers
less than or equal to z. If A is empty, then the result follows immediately. If A is not empty,
then A is bounded above by z and hence by the completeness axiom, A has a least upper
bound, say . Since is the least upper bound, it follows that 1 is not an upper bound
for A so there exists some m A with 1 m. But in that case, m 1, which is to say
m 1 is a natural number not belonging to A (a natural number larger than z ). This proves
the result. Notice that the proof of the Archimedean property is quite similar to the proof in
problem 1.7. This is because if there were some real number z for which the Archimedean
property failed, then this z would be an upper bound for the natural numbers. Then the
11
nonexistence of infinite reals is equivalent to the nonexistence of an upper bound for the
natural numbers.
Problem 1.11 Prove the following statement of the Archimedean property :
0 For all positive real numbers, z there exists a unique n N such that n 1 z n.
This assertion is equivalent to the Archimedean property
Solution:In the solution of problem 1.10, we showed that for any positive real number, z,
the set of natural numbers, n, such that n z is not empty. Then by the well ordering of N,
the set has a first element, n
0
. Since n
0
is the first element in the set, it follows that n
0
1 is
not in the set; i.e.,n
0
1 z n
0
Problem 1.12 Prove the following statements :
1. For all positive real numbers, y, z there exists n N such that n - y z
2. For all positive real numbers, y there exists n N such that 0
1
n
y
Statements 1 and 2 could be interpreted as asserting that there are no infinitesimally small
reals. These assertions are each equivalent to the Archimedean property
Solution: Let y, z denote positive real numbers. Then 0
z
y
R, and by the Archimedean
property, there exists some n N such that
z
y
n. But this is just the result, 1. If we
choose z 1 in 1, then we get 2. This result asserts that there is no real number that is
infinitesimal in the sense that it is closer to zero than
1
n
for every n N.
Problem 1.13 Prove that the rationals and irrationals are everywhere dense in the reals.
Solution:We have to show that between any two reals there is a rational number, and an
irrational number.
Let x and y be real numbers with 0 x y. Then y x 0 and statement 2 of the previous
problem (the Archimedean property) asserts
0
1
m
y x for some m N.
Using the version of the Archimedean property stated in problem 1.11, the set of natural
numbers k such that m - y k is not empty, and by the well ordering of N, this set contains a
first element, n. Then
n 1
m
y
n
m
,
and
x y (y x)
n
m
1
m
n 1
m
;
i.e.,
x
n 1
m
y
This proves that between any two positive reals there is a rational number. If x y 0,
there is a natural number k such that x k and then 0 k x k y. Using the result just
proved, there exists a rational number r between k x and k y, and r k is a rational
number between x and y.
To show there is an irrational number between x and y, use the previous result to find a
12
rational number, r, between the real numbers,
x
2
and
y
2
. Then r 2 is an irrational
number between x and y.
Sequences
Problem 1.14 Prove the monotone sequence theorem.
Solution: Let x
n
) be an increasing sequence that is bounded above. Then the subset of
the reals consisting of all the values of x
n
is a set with an upper bound, hence it has a least
upper bound, say , by the completeness axiom. Then x
n
, for every n, and since is
the least upper bound, it follows that for every 0, is not an upper bound for the
values x
n
. This means that for every 0 there is some M such that x
M
. Since
the sequence x
n
) is increasing, it is clear that x
n
, for every n M. But then by
the definition of convergence, it follows that the sequence converges to . The proof when
the sequence is decreasing is similar.
Problem 1.15 Prove the nested interval theorem.
Solution:Let I
1
|a
1
, b
1
], I
2
|a
2
, b
2
], denote a sequence of nested intervals; i.e., I
n1
is
contained, together with its endpoints, in I
n
. Suppose also that the length of I
n
decreases to
zero with increasing n. Then
a
n
a
n1
b
n1
b
n
for n 1, 2,
and
b
n
a
n
0 as n .
Then the sequence a
n
) is increasing and is bounded above by b
1
. Similarly, b
n
) is
decreasing and is bounded below by a
1
. Then, by the monotone sequence theorem, each
sequence converges to a limit. Let
lim
n
a
n
a and lim
n
b
n
b.
Since b
n
a
n
tends to zero, it follows that a b. Then a b is simultaneously the least
upper bound for the sequence a
n
) and the greatest lower bound for the sequence b
n
).
Then a must belong to every one of the intervals I
n
|a
n
, b
n
]. Moreover, it is evident that
there cannot be two such points, separated by a positive distance, say h 0. For if there
were, then for n sufficiently large that b
n
a
n
h, we could not have both points in the
interval I
n
. Thus a is the unique point common to all the intervals. In the case that the length
of I
n
is not assumed to tend to zero with increasing n, we can conclude only that the interval
|a, b] is contained in I
n
for every n. Since a
n
b
n
for every n, it follows that a
n
b for every
n and this implies that a b.
Problem 1.16 For m , let S
m
) denote the sequence whose terms are given by,
S
m
n1
m
r
n1
;
i.e. S
1
1, S
2
1 r, S
3
1 r r
2
,
Prove that for 0 r 1, the sequence converges to the limit S
1
1 r
. This is the well
known formula for the sum of a geometric series.
Solution: Since r 0, it is clear that the sequence S
m
) is increasing. In fact, for any fixed
13
m, we have
S
m
1 r r
2
r
m1
and rS
m
r r
2
r
m1
r
m
.
Then S
m
rS
m
1 r
m
,
hence
S
m
1 r
m
1 r
.
From this last expression, it follows that S
m
) is bounded above by S
1
1 r
. In fact, S is
the least upper bound for the sequence. To see this, note that for any 0, we have
S S
m
S, provided m is sufficiently large that m
log((1 r))
log(1 r)
. Then by the monotone
convergence theorem, the sequence is convergent to S. This sequence is the so called
sequence of partial sums for the geometric series.
Problem 1.17 Let supA and suppose does not belong to A. Then show that is an
accumulation point for A.
Solution: To say supA means that x for all x A and for every 0, is not an
upper bound for A. That is, for every 0, there exists and x in A such that x .
But that is equivalent to saying that every neighborhood, ( , ) of contains points
of A; i.e., is an accumulation point of A.
Problem 1.18 Prove the Bolzano-Weierstrass theorem, "Every bounded, infinite subset of
must have at least one accumulation point."
Solution: Let A denote a bounded, infinite subset of R; i.e., A is contained in a bounded
interval, I
1
|a, b] and A contains infinitely many points. Now write I
1
|a, b] as the union of
two equal parts, I
1
a,
1
2
(a b)
1
2
(a b), b . At least one of the two parts must
contain infinitely many points since if this were not true, their union which equals I
1
would
contain finitely many points (which contradicts what we known about I
1
). Let I
2
denote one
of these two intervals containing infinitely many points and write I
2
as the union of two
half-intervals. Again, at least one of these two half-intervals must contain an infinite number
of points and we denote by I
3
one of the half-intervals with infinitely many points. Continuing
in this way, we generate a sequence of nested intervals I
1
I
2
I
3
I
n
Now it
is evident that for each n, I
n
contains I
n1
together with its endpoints and the length of I
n
(which equals (b a) - 2
n
) tends to zero as n tends to infinity. Then the nested interval
theorem asserts the existence of a point, p, common to all intervals.
To see that p is an accumulation point for A, consider N
(p) (p , p ) for
0. Choose an n N sufficiently large that (b a) - 2
n
. Since p belongs to every I
n
,
and since is greater than the length of I
n1
, it follows that I
n1
is contained in N
(p). By
construction, I
n1
contains infinitely many points of A which means that N