Notes-Ma131 ch6 PDF
Notes-Ma131 ch6 PDF
Notes-Ma131 ch6 PDF
Completeness II
55
56 CHAPTER 6. COMPLETENESS II
a5 b5
a4 b4
a3 b3
a2 b2
a1 b1
2. Look back at your proof that every convergent sequence is bounded (Work-
book 3). Is it true that every bounded sequence is convergent?
Theorem Bolzano-Weierstrass
Every bounded sequence has a convergent subsequence.
Example The weird, oscillating sequence (sin n) is far from being convergent.
But, since 1 sin n 1, we are guaranteed that it has a convergent subse-
quence.
Proof. Recall the theorem of Section 3.4: every sequence has a monotonic
subsequence. If the sequence is bounded, the subsequence is also bounded, and
it converges by the theorem of Section 5.4.
Exercise 5 Explain why (an ) and (bn ) converge to a limit L. Explain why it
is possible to find a subsequence (xni ) so that xnk [ak , bk ] and show that this
subsequence is convergent.
Exercise 6 Have you proved both the if and the only if parts of this
test?
Is there a similar test that works for general non-monotonic sequences?
Exercise 7 Cleverclogs Test says that a sequence converges if and only if
an+1 an 0. Give an example to show that Cleverclogs test is completely
false (alas).
There is a test for convergence of a general sequence, which does not involve
the limit, which we shall discover in this section.
Definition
A sequence (an ) has the Cauchy property if, for each > 0 there exists a natural
number N such that |an am | < for all n, m > N .
We use the shorthand a Cauchy sequence for a sequence with the Cauchy
property. In words, the Cauchy property means that for any positive , no
matter how small, we can find a point in the sequence beyond which any two
of the terms are less than apart. So the terms are getting more and more
clustered "or crowded.
Example n1 is a Cauchy sequence. Fix > 0. We have to find a natural
number N such that if n, m > N then
1
1 < .
n m
But
1
1 1 + 1 = 1 + 1 .
n m n m n m
1 1
2. Find the decimal representation of 17 . Is your answer exacly 17 ?
The easiest decimal representations are the finite ones - the ones that have
only a finite number of decimal places, like 342.5017. A positive finite decimal
has the form d0 .d1 d2 . . . dn where d0 is a non-negative integer and each of the
d1 , d2 , . . . , dn is one of the integers 0, 1, 2, . . . , 9. Then d0 .d1 d2 . . . dn is defined
to be the number:
d1 d2 dn
d0 + + + + n
10 102 10
Pn
or, written more compactly, j=0 dj 10j .
Exercise 14 What changes are needed when defining a negative finite deci-
mal?
j
d1 = max{j : d0 + x}
10
d1 j
d2 = max{j : d0 + + x}
10 102
...
d1 d2 dn1 j
dn = max{j : d0 + + + + n1 + n x}
10 102 10 10
It is easy to check that each digit is in {0, 1, 2, . . . , 9}. Moreover, after N digits
PN
we must have x 101N < n=0 dn 10n x so that x is the limit of the sequence
of sums by the Sandwich Rule.
Up and Down
Notice that every non-negative
* Consequences of Completeness for Decimals * infinite decimal is the limit of
increasing finite decimals, be-
At the moment, whenever we talk about a decimal expansion, d0 .d1 d2 d3 . . . , we cause you are always adding an
need to show that the sequence of sums converges. What would be useful is a additional non-negative term
theorem to state that this sequence always converges. as you go.
However, every negative infi-
Theorem Infinite decimal sequences version nite decimal is the limit of de-
Every infinite decimal d0 .d1 d2 d3 . . . represents a real number. creasing finite decimals.
d0 .d1 d2 d3 d4 . . .
This is result 6 from earlier in the workbook. With this result, our analysis
of completeness is complete. we succeed in defining a real
number.
* Is 0.999 . . . Equal to 1? *
Although we have finished our examination of completeness, there are still some
things we can do with decimals.
Example What is 0.121 2?
For this decimal, we have d0 = 0 and the sequence (dn ) is defined by d2n = 2
and d2n+1 = 1. Then the sequence of sums is:
n
X
dj 10j
j=0
We know that this converges and thus to find the limit it is sufficient to find
the limit of a subsequence of the sequence of sums. The subsequence we choose
62 CHAPTER 6. COMPLETENESS II
2n
X n
X "
dj 10j = d2k1 102k+1 + d2k 102k
j=0 k=1
Xn
= (1 10 + 2) 102k
k=1
n
X 12
=
100k
k=1
n1
12 X 1
=
100 100k
k=0
" 1 n !
12 1 100
= 1
100 1 100
n
12 1
= 1
99 100
This last exercise already shows one of the annoying features of decimals.
You can have two different decimal representations for the same number. In-
deed, any number with a finite decimal representation also has a representation
as a decimal with recurring 9s, for example 2.15 is the same as 2.1499999. . . .
Theorem
Suppose a positive real number has two different representations as an infinite
decimal. Then one of these is finite and the other ends with a recurring string
of nines.
n
X
x = lim ak 10k
n
k=0
N
X n
X
k
= ak 10 + lim ak 10k
n
k=0 k=N +1
N
X n
X
k
ak 10 + lim 9 10k
n
k=0 k=N +1
N " 1
X
k 9 1 10nN
= ak 10 + lim 1
10N +1 n 1 10
k=0
N
X 1
= ak 10k +
10N
k=0
N
X 1
= ak 10k + (aN + 1)10N
k=0
N
X
bk 10k as aN < bN
k=0
n
X
lim bk 10k as the sequence is increasing
n
k=0
=x
Since we started and ended with the number x, the above inequalities must Recurring Nines
all be equalities. So it is possible to have two decimal representations for x The problem of non-
provided all the decimal digits an are 9 when n > N and all the digits bn are uniqueness of decimal
0 when n > N and bN = aN + 1. If even a single one of these digits fails to representations is annoy-
be a 9 (respectively a 0) then the above chain of inequalities becomes a strict ing but not too bad. In
inequality and we reach the contradiction x < x. many problems we can just
agree to use one of the two
* Classifying Decimals * representations - for instance
by banning any representation
We now classify decimals into three types.
that has recurring nines.
Definition
An infinite decimal d0 .d1 d2 d3 d4 . . . is
Examples
532.89764 is terminating.
0.3333 is recurring.
You can see that a terminating decimal is really just a finite decimal in
disguise. It is also an example of a recurring decimal, since it ends with a string
of repeated zeros.
* Terminating Decimals *
Exercise 17 Suppose x = p/q for integers p, q where the only prime factors of
q are 2s and 5s. Show that x has a terminating decimal representation. [Hint:
show that x = p /10n for some integer p and some n 0.]
Together the last two exercises have shown the following theorem:
Theorem Characterisation of terminating decimals
A number x can be represented by a terminating decimal if and only if x = p/q
for integers p, q where the only prime factors of q are 2s and 5s.
* Recurring Decimals *
Exercise 19 6 as a fraction.
Express the recurring decimal 1.2345
Exercise 21
333
1. Express 22 as a recurring decimal.
2. Use long division to express 71 as a recurring decimal. Write out the long
division sum explicitly (dont use a calculator). In your long division circle
the remainders after each subtraction step. Are all the possible remainders
0,1,2,3,4,5,6 involved? How long is the repeating block?
1
3. Use long division to express 13 as a recurring decimal. In your long division
circle the remainders after each subtraction step. Are all the possible
remainders 0, 1, 2, . . . , 11, 12 involved? How long is the repeating block?
The exercise above should convince you of the following result: The decimal system
The earliest evidence of the
decimal system goes back to
Theorem China, in the 13th century
Every rational number can be represented by a recurring infinite decimal or a B.C. The zero makes a furtive
terminating infinite decimal. appearance in the 4th century
B.C. in China where it is rep-
resented by a space, and in
the 2nd century B.C. in Baby-
lon where it is represented
* Complete Classification * by two small wedges. At
about the same time, a proper
symbol appears in India that
We now have a complete understanding of recurring decimals. Recurring deci-
is called sunya in Sanskrit.
mals represent rationals and rationals always have recurring decimal representa- This word will travel through
tions. What about non-recurring decimals? Since every number has a decimal the Perso-Arabic World as
representation, it follows that any irrational number must have a non-recurring sifr, which will give both
infinite decimal representation. the words zero and cipher.
Fibonacci (11701250) intro-
duced the decimal system in
Theorem Europe.
Every real number has a decimal representation and every decimal represents a Notice that the Greek and Ro-
real number. man civilisations did not know
The rationals are the set of terminating or recurring decimals. about the zero. The Baby-
The irrationals are the set of non-recurring decimals. lonians did not use the dec-
If a number has two distinct representations then one will terminate and the imal system, choosing 60 in-
other will end with a recurring string of nines. stead; neither did Precolom-
bians civilisations like the
Mayas, choosing 20.
66 CHAPTER 6. COMPLETENESS II