Sie sind auf Seite 1von 24

Introduction to Number Theory

Elementary Diophantine Approximation


B.O. Stratmann
Contents.
1. Quick Review on Continued Fractions
2. Elementary Diophantine Approximations
2.1 Hurwitzs Theorem
2.2 Lagrange and Markov spectra
2.3 Badly approximable numbers
3. Metrical Diophantine Approximations
3.1 The Borel-Cantelli Lemma
3.2 Metrical Diophantine approximations
4. A rst Trip through the Zoo of Prime Numbers
1
1 Quick Review on Continued Fractions
Every irrational number can be approximated by a sequence of rationals p
n
/q
n
which are good approximations in the sense that there exists a constant c > 0 such
that


p
n
q
n

<
c
q
2
n
for all n N.
The rationals p
n
/q
n
are called convergents (or approximants). For
= [a
0
; a
1
, a
2
, . . .] = a
0
+
1
a
1
+
1
a
2
+ . . .
,
they are given by
p
n
/q
n
= [a
0
; a
1
, a
2
, . . . , a
n
]
(we shall always assume that a
0
0 and a
i+1
1 for all i N). There are useful
formulas which relate these quantities.
For n N we have (with p
1
= q
0
= 1, q
1
= 0 and p
0
= a
0
)
p
n+1
= a
n+1
p
n
+ p
n1
;
q
n+1
= a
n+1
q
n
+ q
n1
;
q
n
p
n1
p
n
q
n1
= (1)
n
.
Denition 1.1 For = [a
0
; a
1
, a
2
, . . .] and n N, let r
n
and s
n
be dened as
follows.
r
n
:= [a
n
; a
n+1
, a
n+2
, . . .] and s
n
:=
q
n1
q
n
.
For these quantities the following holds (for n N).
r
n
= a
n
+
1
r
n+1
;
Since q
n+1
= a
n+1
q
n
+ q
n1
, we have for the ratio
q
n+1
q
n
= a
n+1
+
1
q
n
/q
n1
.
Clearly, this process may be continued until q
1
/q
0
= a
1
is reached. Therefore,
s
n+1
=
1
[a
n+1
; a
n
, . . . , a
1
]
.
Theorem 1.2 For = [a
0
; a
1
, a
2
, . . .] and n N, we have
=
p
n
r
n+1
+ p
n1
q
n
r
n+1
+ q
n1
.
2
Proof: (by induction) For n = 0 we have
p
0
r
1
+ p
1
q
0
r
1
+ q
1
=
a
0
r
1
+ 1
r
1
= a
0
+
1
r
1
= .
Now assume that the statement is true for n. Then
=
p
n
r
n+1
+ p
n1
q
n
r
n+1
+ q
n1
=
p
n
(a
n+1
+
1
r
n+2
) + p
n1
q
n
(a
n+1
+
1
r
n+2
) + q
n1
=
p
n
a
n+1
r
n+2
+ p
n
+ p
n1
r
n+2
q
n
a
n+1
r
n+2
+ q
n
+ q
n1
r
n+2
=
(p
n
a
n+1
+ p
n1
)r
n+2
+ p
n
(q
n
a
n+1
+ q
n1
)r
n+2
+ q
n
=
p
n+1
r
n+2
+ p
n
q
n+1
r
n+2
+ q
n
.
2
Corollary 1.3


p
n
q
n

=
1
q
2
n
(r
n+1
+ s
n
)
for all n N.
Proof:


p
n
q
n

p
n
r
n+1
+ p
n1
q
n
r
n+1
+ q
n1

p
n
q
n

p
n
q
n
r
n+1
+ p
n1
q
n
p
n
q
n
r
n+1
p
n
q
n1
(q
n
r
n+1
+ q
n1
)q
n

q
n
p
n1
p
n
q
n1
q
2
n
(r
n+1
+ s
n
)

=
1
q
2
n
(r
n+1
+ s
n
)
.
2
2 Elementary Diophantine Approximations
2.1 Hurwitzs Theorem
Theorem 2.1 For all irrationals = [a
0
; a
1
, a
2
, . . .] and for all n N, we have that


p
i
q
i

1
2q
2
i
is fullled for at least one element i {n, n + 1}.
Proof: By way of contradiction, assume that the statement in the theorem is false.
This means that


p
i
q
i

>
1
2q
2
i
holds simultaneously for i = n and i = n + 1. Since


p
i
q
i

=
1
q
2
i
(r
i+1
+s
i
)
, this is
equivalent to
r
i+1
+ s
i
< 2 for i = n, n + 1.
3
(a) For i = n we get 2 > r
n+1
+ s
n
= a
n+1
+
1
r
n+2
+ s
n
, and hence,
1
r
n+2
< 2 (a
n+1
+ s
n
) = 2
1
s
n+1
.
(b) For i = n + 1 we get
r
n+2
< 2 s
n+1
.
Combining (a) and (b), we derive 1 < 4 2s
n+1
2s
1
n+1
+ 1,
and hence 0 < 2 s
n+1
s
1
n+1
, implying
0 > (s
n+1
1)
2
,
and hence we derive a contradiction. 2
Theorem 2.2 For all irrationals = [a
0
; a
1
, a
2
, . . .] and for all n N, we have that


p
i
q
i

5 q
2
i
is fullled for at least one element i {n, n + 1, n + 2}.
Note, the number
1

5
is called the Hurwitz number.
Proof: As in the proof of the previous theorem (with 2 replaced by

5), assume
by way of contradiction that for each i {n, n + 1, n + 2} we have
r
i+1
+ s
i
<

5.
Proceeding for i = n and i = n+1 as in (a) and (b) in the previous proof, we derive
s
2
n+1

5s
n+1
+ 1 < 0. (1)
Analogously, for i = n + 1 and i = n + 2, we get
s
2
n+2

5s
n+2
+ 1 < 0. (2)
By the quadratic formula, (1) and (2) give (with :=

5+1
2
and

:=

51
2
)

< s
i
< for i = n + 1, n + 2. (3)
Using this, we get
s
n+2
=
1
a
n+2
+ s
n+1

1
1 + s
n+1
<
1
1 +

,
which contradicts (3). 2
4
Theorem 2.3 (Hurwitzs theorem) For the golden mean :=

5+1
2
= [1; 1, 1, 1, . . .]
we have that


p
n
q
n

C
q
2
n
is satised for at most nitely many reduced p
n
/q
n
if and only if C <
1

5
.
Proof: First note that r
n
= for all n N. Secondly, note that
s
1
n
= [a
n;
a
n1
, . . . , a
1
] = + ([a
n;
a
n1
, . . . , a
1
] [a
n;
a
n1
, . . .]) = +
n
,
where for
n
we have that lim
n

n
= 0. Hence, it follows
s
n
=
1
+
n
=
1

+
1
+
n

=
1

+

n

2
+
n
=
1

+
n
,
where for
n
we have that lim
n

n
= 0. These two observations then give
r
n+1
+ s
n
= +
1

+
n
=

5 +
n

5 ( for n ).
Now, if C <
1

5
is given, say C =
1

5+
for some xed > 0, then


p
n
q
n

=
1
q
2
n
(r
n+1
+ s
n
)
=
1
q
2
n
(

5 +
n
)

1
q
2
n
(

5 + )
,
where the latter inequality can be fullled only for nitely many n (due to the fact
that

5 + <

5 +
n
can be satised for at most nitely many n). 2
Corollary 2.4 For each irrational number , the inequality


p
n
q
n

K
q
2
n
is fullled for innitely many reduced p
n
/q
n
as long as K
1

5
.
2.2 Lagrange and Markov Spectra
Denition 2.5 Let c denote some positive real number. An irrational is
called c-approximable if an only if


p
n
q
n

<
c
q
2
n
is satised for innitely many reduced p
n
/q
n
.
To each irrational number we associate a non-negative real number (),
dened by
() := inf{c > 0 : is c-approximable }.
5
Two irrational numbers , are called equivalent (and we write ) if and
only if there exist k, l N such that r
k
() = r
l
() (i.e. eventually the continued
fraction expansions of and coincide).
Lemma 2.6 Let , be irrational. If , then () = ().
Proof: Let , be irrational such that . Then there exist k, l N such that
r
k+i
() = r
l+i
() for all i N. Without loss of generality, assume that l k. Then
and must be of the form
= [a
0
; a
1
, . . . , a
k
, c
1
, c
2
, c
3
, . . .] and = [b
0
; b
1
, . . . , b
k
, b
k+1
, . . . , b
l
, c
1
, c
2
, c
3
, . . .].
In order to prove the assertion of the lemma, it is sucient to show that

1
r
k+n
() + s
k+n1
()

1
r
l+n
() + s
l+n1
()

0 for n .
For this it is sucient to show that
|r
k+n
() + s
k+n1
() (r
l+n
() + s
l+n1
())| 0 for n .
But this follows, since
|r
k+n
() + s
k+n1
() (r
l+n
() + s
l+n1
())| = |s
k+n1
() s
l+n1
()|
=

1
[c
n1
; . . . , c
1
, a
k
, . . . , a
0
]

1
[c
n1
; . . . , c
1
, b
l
, . . . , b
0
]

0 ( for n ).
2
Denition 2.7 An irrational is called noble number (i.e. the continued
fraction expansion of a noble number has from some stage onward exclusively 1s as
its entries).
Corollary 2.8 For each irrational number we have that ()
1

5
.
A number is a noble number if and only if () =
1

5
.
Theorem 2.9 Let N be some xed positive integer. If = [a
0
; a
1
, a
2
, . . .] is irra-
tional such that for some n N we have that


p
i
q
i

>
1
q
2
i

N
2
+ 4
is fullled for all i {n, n + 1, n + 2}, then it follows that a
n+2
< N.
6
Proof: We proceed as in the proof of the rst two theorem of the section (with 2,
resp.

5, now replaced by

N
2
+ 4). In this way, considering i = n and i = n + 1,
we derive
s
2
n+1

_
N
2
+ 4 s
n+1
+ 1 < 0.
And also, by considering i = n + 1 and i = n + 2, we derive along the same lines
s
2
n+2

_
N
2
+ 4 s
n+2
+ 1 < 0.
Then, using the quadratic formula, we obtain

N
2
+ 4 N
2
< s
i
, s
1
i
<

N
2
+ 4 + N
2
for i = n + 1, n + 2.
Using this, we then have
a
n+2
= s
n+1
+ a
n+2
s
n+1
= s
1
n+2
s
n+1
<

N
2
+ 4 + N
2

N
2
+ 4 N
2
= N.
2
Corollary 2.10 For each irrational number and for every N N, exactly one of
the following two alternatives occurs.
Either:


p
n
q
n

1
q
2
n

N
2
+ 4
is fullled for innitely many p
n
/q
n
(or with other words, () 1/

N
2
+ 4),
Or: There exists a number n
0
> 0 (depending on N and ) such that
a
n
< N for all n n
0
(or with other words, B
N
(see Denition 2.18).).
Corollary 2.11 For each non-noble irrational number we have that


p
n
q
n

1
2

2 q
2
n
is fullled for innitely many reduced p
n
/q
n
. (Or with other words, for each non-
noble number we have () 1/(2

2).)
In fact, by means of similar ideas as in the proof of Hurwitzs theorem (theorem
2.3), one derives that
() =
1
2

2
if and only if

2 (= [1; 2, 2, 2, . . .]).
Proof: This follows immediately, since if = [a
0
; a
1
, . . .] is non-noble then we have
a
n
2, for innitely many n. Hence, by theorem 2.9, we have


p
q

8 q
2
n
for innitely many reduced p/q, which implies that () 1/(2

2). 2
7
Lemma 2.12 Let = [a
0
; a
1
, a
2
, . . .] be an irrational number such that () is nei-
ther equal to
1

5
nor to
1
2

2
, but such that [b
0
; b
1
, b
2
, . . .] with b
i
2 for all i N.
It then follows that
()
6
17
.
Proof: Without loss of generality we can assume that there are innitely many 1s
and 2s in [b
0
; b
1
, b
2
, . . .] (since otherwise would be equivalent to either 1/

5 or
1/(2

2)). Hence there are innitely many values n such that a


n
= 1 and a
n+1
= 2.
For these n, we have
r
n+1
+ s
n
= [a
n+1
; a
n+2
, . . .] +
1
[a
n
; . . . , a
1
]
2 +
1
2 +
1
1
+
1
1 +
1
1
=
7
3
+
1
2
=
17
6
.
It follows that ()
6
17
. 2
Lemma 2.13 If = [a
0
; a
1
, . . .] is irrational such that a
n
3 for innitely many
n, then ()
1

13
.
Proof: By Theorem 2.9 we have that if a
n
3 for innitely many n, then


p
n2
q
n2

3
2
+ 4 q
2
n2
_
=
1

13 q
2
n2
_
must hold for innitely many n. Hence, ()
1

13
. 2
Denition 2.14 The set of numbers
L := {() : is irrational }
is called Lagrange spectrum.
The set of numbers
M := L
_
1
3
,
1

5
_
is called Markov spectrum.
Note, in some books one nds a slightly dierent use of the term Markov spectrum.
Also note that since
1
3
>
1

13
, we have by Lemma 2.13 that irrational numbers in
the Markov spectrum, that is with () > 1/3, must have the property that they
are equivalent to irrational numbers whose continued fraction expansion contain
exclusively 1s and 2s.
As an immediate consequence of Hurwitzs Theorem (Theorem 2.3), we obtain the
following theorem.
8
Theorem 2.15
L
_
0,
1

5
_
.
Proposition 2.16 For an irrational number we have that () M if and only
if [a
0
; a
1
, a
2
, . . .], for [a
0
; a
1
, a
2
, . . .] such that a
n
2 for all n N.
One can say much more about the structure of the Markov spectrum. It has the
following very interesting properties. The proof of this theorem is slightly more
involved and will be omitted.
Theorem 2.17 The Markov spectrum M consists of a countable set of numbers in
(1/3; 1/

5], and these numbers accumulate only at the value


1
3
.
In fact, much more can be said about the Markov and the Lagrange spectrum.
Nevertheless, there are still plenty of fascinating open problems concerning these
spectra. We now list a few known results about them. Some of these we have
already obtained.
Each number in the Markov spectrum is of the form 1/
_
9
4
m
2
, where m is
a positive integer solution of the equation m
2
+ k
2
+ l
2
= 3mkl, for k and l
some positive integers. It is known that there are innitely many solutions m
of this equation. The rst numbers in the Markov spectrum are
1

5
_
_
=
1
_
9
4
1
2
_
_
,
1
2

2
_
_
=
1
_
9
4
2
2
_
_
,
5

221
_
_
=
1
_
9
4
5
2
_
_
,
1
_
9
4
13
2
,
1
_
9
4
29
2
,
1
_
9
4
34
2
,
1
_
9
4
89
2
,
1
_
9
4
194
2
,
1
_
9
4
433
2
, . . . .
Note that since 1/
_
9
4
m
2
accumulates at 1/3 (for m tending to innity), it
is clear that the Markov spectrum accumulates at 1/3.
We have that (x)
1

12
if and only if x is equivalent to a number whose
continued fraction expansion contains exclusively 1s and 2s.
In the interval
_
1

13
,
1

12
_
the Lagrange spectrum is empty. That is
L
_
1

13
,
1

12
_
= .
Let f be the so called Freimann number which is given by
f :=
491993569
2221564096 + 283748

462
.
One then knows that in the interval [0, f ) the Lagrange spectrum is continuous.
This means that for every c [0, f ) there exists an irrational number x such
that (x) = c.
9
2.3 Badly Approximable Numbers
Denition 2.18 For N N dene
B
N
:= { = [a
0
; a
1
, a
2
, . . .] irrational : n
0
> 0 such that a
n
< N n n
0
}.
The set of badly approximable numbers B is then dened as
B :=
_
N>0
B
N
= { irrational : N > 0 such that B
N
}.
With other words, B
N
if and only if , for some = [b
0
; b
1
, . . .] with b
i
< N
for all i N. Furthermore, B if and only if there exists M N such that B
M
.
The following corollary claries why the elements in B are called badly approx-
imable.
Lemma 2.19 If is an irrational number such that / B
N
for some
N N, then


p
n
q
n

1
q
2
n

N
2
+ 4
is fullled for innitely many reduced p
n
/q
n
(i.e. () 1/

N
2
+ 4).
For each B there exists a constant C > 0 such that for all n N we have


p
n
q
n

>
C
q
2
n
.
Proof: The rst part is an immediate consequence of theorem 2.9. For the second
part, consider = [a
0
; a
1
, . . .] B. Then there exist numbers M and m
0
such that
a
n
< M for all n m
0
. Using this, we derive r
n+1
+ s
n
< M + 1 + 1 = M + 2, and
hence


p
n
q
n

>
1
(M + 2) q
2
n
for all n m
0
.
For n < m
0
we have that there exists a number c
n
> 0 such that


p
n
q
n

>
c
n
q
2
n
.
If we dene C := min{1/(M + 2), c
0
, c
1
, . . . , c
m
0
1
} (i.e. C is the smallest number
in this nite set of numbers), then the result follows. 2
10
3 Metrical Diophantine Approximations
In this section we restrict the investigations to the unit interval I := [0, 1).
3.1 The Borel-Cantelli Lemma
Denition 3.1 A set of subsets of I is called a -algebra of I if the following
conditions are satised.
I ;
If A I, then A
c
I (where A
c
:= I \ A denotes the complement of A in I);


nN
A
n
I for all sequences (A
n
) with A
n
I (for all n N).
Denition 3.2 The Borel--algebra
0
of I is the smallest -algebra of I which
contains all possible intervals of I of the form [x, y) (for 0 x < y < 1).
The elements of
0
are called Borel sets.
Denition 3.3 Each element in
0
can be measured by the Lebesgue measure in
I. In particular, if A is an interval (i.e. A = [x, y) for some 0 x < y < 1), then
(A) is just the length of that interval (i.e. (A) = ([x, y)) = y x).
Properties:
(I) = 1;
(A) 0 for all A
0
;
(

nN
A
n
) =

nN
(A
n
) for every sequence (A
n
) of pairwise disjoint ele-
ments A
n

0
(i.e. A
i
A
j
= i = j).
For A
0
we have:
(A) = 0 if and only if for all > 0 there exists a sequence (A
n
) of elements
A
n

0
such that
A
_
nN
A
n
and

nN
(A
n
) < .
Note, every countable set in I is of zero -measure.
More general, in order to nd out if a given Borel set is of zero -measure, the
following theorem is often helpful.
Theorem 3.4 (Borel-Cantelli lemma)
If (A
n
) is a sequence of elements A
n

0
such that

nN
(A
n
) < , then we have
(A

) = 0,
where the limsup-set A

is dened by
A

:= { I : A
n
for innitely many n}.
11
Proof: The convergence of

nN
(A
n
) implies that for each > 0 there exists an
integer n
0
such that

nn
0
(A
n
) < .
Now note that by denition of A

, we have that
A


_
nn
0
A
n
.
Hence, it follows that
(A

)
_
_
_
nn
0
A
n
_
_

nn
0
(A
n
) < .
2
3.2 Metrical Diophantine Approximations
Denition 3.5 Let a
1
, . . . , a
n
N\{0} be given. The n-cylinder I(a
1
, . . . , a
n
) (also
called fundamental interval of order n) is dened by (here we use the common
notation [x
1
, x
2
, . . .] := [0; x
1
, x
2
, . . .])
I(a
1
, . . . , a
n
) := { = [x
1
, x
2
, x
3
, . . .] I irrational : x
i
= a
i
for all 1 i n}.
Properties:
For every I(a
1
, . . . , a
n
) we have
=
p
n
r
n+1
() + p
n1
q
n
r
n+1
() + q
n1
,
where p
n
, p
n1
, q
n
, q
n1
are xed (depending only on a
1
, . . . , a
n
).

I(a
1
, . . . , a
n
) =
_
_
_
_
p
n
q
n
,
p
n
+p
n1
q
n
+q
n1
_
for n even
_
p
n
+p
n1
q
n
+q
n1
,
p
n
q
n
_
for n odd.

(I(a
1
, . . . , a
n
)) =
1
q
2
n
(1 + s
n
)
.
Proof: These properties are immediate consequences of the following.
By theorem 1.2, we have
=
p
n
r
n+1
() + p
n1
q
n
r
n+1
() + q
n1
=
p
n
+ p
n1
/r
n+1
()
q
n
+ q
n1
/r
n+1
()
.
12
Since 1 r
n+1
() and since r
n+1
() can get arbitrary large if varies, we see that
(I(a
1
, . . . , a
n
)) =

p
n
+ p
n1
q
n
+ q
n1

p
n
q
n

p
n
q
n
+ p
n1
q
n
p
n
q
n
q
n1
p
n
q
2
n
(1 + s
n
)

p
n1
q
n
q
n1
p
n
q
2
n
(1 + s
n
)

=
1
q
2
n
(1 + s
n
)
.
Furthermore, observe that
p
n
q
n
<
p
n
+ p
n1
q
n
+ q
n1
if and only if p
n
q
n1
q
n
p
n1
< 0.
But we know (since q
n
p
n1
p
n
q
n1
= (1)
n
) that the left hand side of the latter
inequality is equal to (1) if and only if n is even. 2
For the next theorem, recall the denition of the set of badly approximable irrational
numbers (Denition 2.18).
Theorem 3.6 For B

:= B I we have
(B

) = 0.
Proof: For n, N N we dene the sets
A
N
:= { = [a
1
, a
2
, . . .] I irrational : a
i
< N i N}, A :=
_
NN
A
N
,
A
(n)
N
:= { = [a
1
, a
2
, . . .] I irrational : a
i
< N i {1, . . . , n}}.
We want to show that (A) = 0. For this, since A
N
A
(n)
N
, it is sucient to show
that lim
n
(A
(n)
N
) = 0, and this is what we are now going to prove.
Note that A
(n+1)
N
A
(n)
N
, and that each A
(n+1)
N
can be written as a union of disjoint
fundamental intervals as follows
A
(n+1)
N
=
_
(a
1
,...,a
n+1
):
a
i
<N,i=1,...,n+1
I(a
1
, . . . , a
n
, a
n+1
) =
_
(a
1
,...,a
n
)
a
i
<N,i=1,...,n
_
k:
k<N
I(a
1
, . . . , a
n
, k).
For xed (a
1
, . . . , a
n
), we now calculate the Lebesgue measure of

k:k<N
I(a
1
, . . . , a
n
, k)
as follows.

_
_
_
1k<N
I(a
1
, . . . , a
n
, k)
_
_
=

p
n
+ p
n1
q
n
+ q
n1

p
n
N + p
n1
q
n
N + q
n1

= . . .
=
N 1
q
2
n
(1 + s
n
)(N + s
n
)
<
N 1
q
2
n
N(1 + s
n
)
=
_
1
1
N
_
(I(a
1
, . . . , a
n
)).
13
Using the latter estimate, we get
(A
(n+1)
N
) =
_
_
_
_
_
(a
1
,...,a
n
):
a
i
<N,i=1,...,n
_
k:
k<N
I(a
1
, . . . , a
n
, k)
_
_
_
_
=

(a
1
,...,a
n
):
a
i
<N,i=1,...,n

_
_
_
_
k:
k<N
I(a
1
, . . . , a
n
, k)
_
_
_

(a
1
,...,a
n
):
a
i
<N,i=1,...,n
(I(a
1
, . . . , a
n
))
_
1
1
N
_
=
_
1
1
N
_
(A
(n)
N
).
Applying this estimate n times, we derive
(A
(n+1)
N
)
_
1
1
N
_
(A
(n)
N
)
_
1
1
N
_
2
(A
(n1)
N
) . . .
_
1
1
N
_
n
(A
(1)
N
),
which then implies
(A
(n+1)
N
) 0 for n .
From this we obtain that (since A
N
A
(n+1)
N
)
(A
N
) = 0 for all N N,
and hence, since
(A) =
_
_
NN
A
N
_

NN
(A
N
) = 0,
we obtain the desired result
(A) = 0.
Finally, observe that B

if and only if A, from which we derive


(B

) = 0.
2
By inspection of the proof of the previous theorem, we nd that in there we in fact
proved slightly more than we actually formulated in the theorem. Namely, we have
seen that the following is true.
Corollary 3.7 For B

N
:= B
N
I we have
(B

N
) = 0 for all N N.
Also, combining the previous theorem and corollary 2.19, we immediately obtain
the following result.
Corollary 3.8

__
I irrational : C > 0 such that


p
q

>
C
q
2
for all
p
q
__
= 0.
14
We have now seen that the set of badly approximable numbers does not contribute
to sets of irrational numbers of positive Lebesgue measure. Hence, if we want to
investigate sets of positive measure, then we have to look for irrationals which are
more rapidly approximated by their approximants than it is the case for badly
approximable irrationals. The contra-positive of the following theorem gives a rst
indication of how an irrational number has to look like in order to have a chance to
contribute to positive Lebesgue measure. In particular, the theorem species how
fast the a
n
() have to increase at least such that has a chance to contribute to
positive Lebesgue measure.
Theorem 3.9 If : N R
+
is a function such that

n=1
1/(n) diverges, then
(B

) = 0,
where B

:= { = [a
1
, a
2
, . . .] I irrational : a
n
< (n) n N}.
Note: A good choice for would be (n) = nlog(n) (recall that

n=1
1
nlog(n)
diverges).
Proof: The proof is basically the same as the proof of the previous theorem. As
before, we obtain that

_
_
_
_
k:
k<(n+1)
I(a
1
, . . . , a
n
, k)
_
_
_ <
_
1
1
(n + 1)
_
(I(a
1
, . . . , a
n
)).
Hence, with B
(n)

:= { = [a
1
, a
2
, . . .] I irrational : a
i
< (i) i {1, . . . , n}}, we
get
(B
(n+1)

) <
_
1
1
(n + 1)
_
(B
(n)

) < . . . <
n

k=1
_
1
1
(k + 1)
_
(B
(1)

).
Using the fact that 1 x < e
x
for each 0 < x < 1, we can continue as follows.
(B
(n+1)

) < e

n
k=1
1
(k+1)
(B
(1)

),
which implies (since

n
k=1
1/(k + 1) gets arbitrary large, due to the divergence
condition in the theorem)
(B
(n+1)

) 0 for n ,
and hence (since B

B
(n+1)

for all n),


(B

) = 0.
2
15
Note that with the special choice of , that is (n) = nlog(n), an immediate con-
sequence of the previous theorem is (for this essentially consider the complement of
B

in I)
({ = [a
1
, a
2
, . . .] I irrational : a
n
nlog(n) for innitely many n N}) = 1.
In contrast to the previous theorem, we now investigate how fast the a
n
() can in-
crease at most such that has a chance to contribute to positive Lebesgue measure.
Theorem 3.10 If : N R
+
is a function such that

n=1
1/(n) converges, then
(W

) = 0,
where W

:= { = [a
1
, a
2
, . . .] I irrational : a
n
> (n) for innitely many n}.
Note: A good choice for would be (n) = n(log(n))
1+
, for any xed > 0 (recall
that

n=1
1
n(log(n))
1+
converges, for every > 0).
Proof: We have that

_
_
_
_
k:
k(n+1)
I(a
1
, . . . , a
n
, k)
_
_
_ =

p
n
(n + 1) + p
n1
q
n
(n + 1) + q
n1

p
n
q
n

= . . .
=
1
q
2
n
(1 + s
n
)
1 + s
n
(n + 1) + s
n
<
2
(n + 1)
(I(a
1
, . . . , a
n
)).
Hence, with W
(n)

:= { = [a
1
, a
2
, . . .] I irrational : a
n
> (n)}, we get
(W
(n+1)

) =
_
_
_
_
(a
1
,...,a
n
)
_
k:
k(n+1)
I(a
1
, . . . , a
n
, k)
_
_
_
<
2
(n + 1)

(a
1
,...,a
n
)
(I(a
1
, . . . , a
n
))
2
(n + 1)
.
Now, an application of the Borel-Cantelli lemma (theorem 3.4) nishes the proof. 2
Note that with the special choice of , that is (n) = n(log(n))
1+
, an immediate
consequence of the previous theorem is (for this essentially consider the complement
of W

in I) that for each > 0,

_
{ = [a
1
, a
2
, . . .] I irrational : n
0
such that a
n
< n(log(n))
1+
n n
0
}
_
= 1.
Combining this with the remark after Theorem 3.9, we hence have that the continued
fraction expansion of an irrational number = [a
1
, a
2
, . . .] which contributes to a
set of full Lebesgue measure has the property that for each > 0 we have
a
n
> nlog(n) for innitely many n, whereas a
n
< n(log(n))
1+
eventually.
16
Finally, we mention the following important theorem (without proof). In this the-
orem we use the notion of a (, )-Khintchine function, by which we mean the
following.
A (, )-Khintchine function : R
+
R
+
is a non-increasing function which
is not decreasing too rapidly, in the sense that there exist positive numbers
< 1 and 1 such that for all x R
+
we have that (x) (x).
Theorem 3.11 (Khintchines theorem)
For a (, )-Khintchine function let
K

:= { I :


p
n
q
n

<
(q
n
)
q
2
n
is fullled for innitely many n}.
Then the following holds.
(i) (K

) = 0 if and only if

nN
(
n
) converges.
(ii) (K

) = 1 if and only if

nN
(
n
) diverges.
Remark: In case (i), a good choice for the function would be (x) = (log(x))
(1+)
(for any > 0). And in case (ii), a good choice for the function would be
(x) = (log(x))
1
.
With these choices, we then obtain that for from a set of full -measure we have
that the two inequalities
1
q
2
n
(log(q
n
))
1+
<


p
n
q
n

<
1
q
2
n
log(q
n
)
,
are fullled simultaneously for innitely many p
n
/q
n
(more precisely, the left-hand
inequality is fullled even for all p
n
/q
n
apart from nitely many exceptions).
17
4 A rst Trip through the Zoo of Prime Numbers
Denition 4.1 A positive integer p = 1 is called a prime number if p is divisible
only by 1 and p.
Theorem 4.2 (Unique prime factorisation theorem) Every positive integer
N = 1 is either a prime number or can be written uniquely as a product of prime
numbers.
Proof: This is left as an exercise (use complete mathematical induction). 2
Its a very old fact (Euclid 325-265 B.C., in Book IX of the Elements) that the set of
primes is innite. Presumably, one of the rst rigorous proofs of this fact was given
by Euler.
Theorem 4.3 (Euler) There are innitely many prime numbers.
Proof: Assume (by way of contradiction) that there are only nitely many prime
numbers. Then let P := {p
1
, p
2
, . . . , p
n
} be the set of all these primes, ordered such
that p
i
< p
i+1
for all i = 1, . . . , n 1. Consider
q := p
1
p
2
... p
n
+ 1.
Since q > p
n
, it follows that q / P, and hence q is not a prime number. Since every
number has a unique prime factorisation, it follows that there exist q
1
, . . . , q
k
P
such that
q := q
1
q
2
... q
k
.
Combining this with the denition of q, it follows that
p
1
p
2
... p
n
+ 1 = q
1
q
2
... q
k
.
Since in the product p
1
p
2
... p
n
every prime number appears exactly once as a
factor, we must have that one of these, say p
j
, is equal to q
1
. That is, q
1
= p
j
for
some j {1, . . . , n}. By dividing the above equality by q
1
, we obtain
p
1
p
2
...p
j1
p
j+1
... p
n
+
1
q
1
= q
2
... q
k
.
Since in here the right hand side is an element of N whereas the left hand side is
not, this gives a contradiction. 2
The Sieve of Erathosthenes.
This sieve represents a method for how to nd all prime numbers less than some
given number N N. The method is as follows.
1. Write down a list consisting of all numbers from 2 up to N.
2. p
1
(= 2) is the rst prime number, and hence stays on the list. Then remove
all multiples of 2 from the list.
18
3. p
2
(= 3) is the next number (following p
1
), and hence stays on the list. Then
remove all multiples of p
2
from the list.
4. p
3
(= 5) is the next number (following p
2
), and hence stays on the list. Then
remove all multiples of p
3
from the list.
5. p
4
is the next number (following p
3
) (of course p
4
= 7), and hence stays on
the list. Then remove all multiples of p
4
from the list.
. . . . . . . . . . . . . . . . . .
The sieve ends once we have reached for the rst time a number, say p
n
, for which
p
2
n
> N. In this way we have obtained all prime numbers p
1
, p
2
, . . . , p
n
which are
between 2 and N.
Lemma 4.4 There are arbitrarily large gaps in the sequence of prime numbers. Or
with other words, for each arbitrarily large number N N there exists a number
n N such that there are no prime numbers between n and n + N.
Proof: Let N N be given. Then dene n := (N + 1)!, and observe the following
n + 1 might be a prime number;
n + 2 = (N + 1)! + 2 is divisible by 2, and hence not a prime number;
n + 3 = (N + 1)! + 3 is divisible by 3, and hence not a prime number;
n + 4 = (N + 1)! + 4 is divisible by 4, and hence not a prime number;
.
.
.
n + (N + 1) = (N + 1)! + (N + 1) is divisible by (N + 1), and hence not a
prime number.
Therefore, we now have that all the N numbers between n +2 and n +(N +1) are
not prime numbers. 2
Denition 4.5 The prime number counting function is dened for each N N
by
(N) := {p : p is a prime number, and p N}.
As a rst little estimate we obtain the following. Note that an immediate conse-
quence of this lemma is that there are innitely many primes, and hence the lemma
gives an alternative proof of Eulers theorem.
Lemma 4.6
(N) >
log N
2 log 2
for all N N \ {1}.
19
Proof: Let us rst remark the following. By the unique prime factorisation theorem
we have that every positive integer n = 1 can be written as the product of a square
number and a product of distinct prime numbers. To see this, note that for each
n = 1 there have to be prime numbers p
1
, . . . , p
k
, p
k+1
, . . . , p
k+l
, as well as k odd
numbers m
1
, . . . , m
k
and l even numbers m
k+1
, . . . , m
k+l
, such that (note, since the
m
1
, . . . , m
k
are odd, we have that
m
i
1
2
is either equal to zero or a positive integer
(for i = 1, . . . , k))
n =
k+l

i=1
p
m
i
i
=
k

i=1
p
m
i
i

k+l

i=k+1
p
m
i
i
= p
1
. . . p
k

k

i=1
p
m
i
1
i

k+l

i=k+1
p
m
i
i
= p
1
. . . p
k

_
_
k

i=1
p
m
i
1
2
i

k+l

i=k+1
p
m
i
2
i
_
_
2
.
Now, let N N \ {1} be given, and consider all integers between 2 and N. We give
an upper bound for the number of ways one can possibly write the numbers between
2 and N in form of a product of a square number and a product of distinct prime
numbers.
Question: How many distinct squares are there among the numbers between
2 and N?
Answer: Less than

N.
Question: How many distinct products of distinct prime numbers are there at
most among the numbers between 2 and N?
Answer: Less than
_
(N)
1
_
+
_
(N)
2
_
+
_
(N)
3
_
+ . . . +
_
(N)
(N) 1
_
+ 1 = 2
(N)
1 < 2
(N)
.
(Recall that the binomial coecients
_
n
k
_
are dened by
_
n
k
_
=
n!
(nk)!k!
).
Combining these two estimates, we obtain
N <

N 2
(N)
.
Solving this for (N) then gives the result. 2
Already Euclid knew that the set of primes is innite, and a much more recent and
famous result (by Jacques Hadamard (1865-1963) and C.-J. de la Vallee-Poussin
(1866-1962)) shows that the density of primes is ruled by the following law. This
law had already been conjectured before by Gauss. Since the proof of this law is
rather involved (and strictly speaking this result is not a number theoretical result,
since most proofs make heavy use of probability theory), here we can only state this
very important law.
20
Theorem 4.7
lim
N
(N)
N
log N
= 1.
Getting a more exact gure for the function is presumably one of the most impor-
tant problems in mathematics. Here, a very big step forward would be to verify the
Riemann Hypothesis (given that it is true). In here we use li to denote the function
which is given by
li(N) =
_
N
2
1
log x
d(x).
Note that lim
N
li(N)/(N/ log(N)) = 1.
Riemann Hypothesis:
There exists a constant C > 0 such that for all N suciently large
li(N) CN
1
2
+
(N) li(N) + CN
1
2
+
for all > 0.
The Riemann Hypothesis is Problem 8 on Hilberts famous list of 23 problems (Paris,
1900). In the meanwhile, mathematicians found numerous ways to state the Rie-
mann Hypothesis in equivalent forms which come in completely dierent disguises.
Let us give one of these.
Denition 4.8 For n N let
F
n
:=
_
p
q
[0, 1] : 0 p < q n, and p, q are coprime
_
.
If we assume that the set F
n
= {f
1
, . . . , f
k
n
} is ordered such that f
1
< f
2
< . . . < f
k
n
,
then F
n
is called the Farey sequence of order n. Note that in here k
n
denotes the
number of elements in F
n
.
In the 1920s Franel and Landau found the following elementary way of formulating
the Riemann Hypothesis.
Riemann Hypothesis:
There exists a constant C
0
> 0 such that for all n suciently large
n
1
2
+
C
0

k
n

i=1

f
i

i 1
k
n

n
1
2
+
+ C
0
for all > 0.
Finally, note that the original statement of the Riemann Hypothesis was formulated
in terms of the Riemann zeta-function (z) which is given by
(z) :=

n=1
1
n
z
for z C.
21
For real values z this series is called the harmonic series (s), that is
(s) :=

n=1
1
n
s
for s R.
Of course, this series converges if and only if s > 1 (note for complex values the
situation is by no means as simple as this!). Nevertheless, to compute the actual
values of this series for particular s > 1 is usually a problem. It can be done for
instance for even numbers s. Here we have the following result of Euler.
Theorem 4.9 (Euler)
(2n) = (1)
n1
(2)
2n
2(2n)!
B
2n
for all n N,
where the B
k
are the Bernoulli numbers given by
z
e
z
1
=

k=0
B
k
k!
z
k
.
For instance, we have
(2) =

2
6
, (4) =

4
90
, (6) =

6
945
.
For odd numbers s things are far less well-understood. It was a mathematical
sensation, when in 1978 Apery proved
(3) is irrational.
Also, for instance one knows (this is a result of Zudilin)
One of (5), (7), (9), (11) has to be an irrational number.
Furthermore, it is know that (this is a result of Rioval)
(2n + 1) is irrational for innitely many n N.
Twin Primes.
Denition 4.10 A couple of primes (p, q) are said to be twin primes if q = p + 2.
Except for the couple (2, 3), this is clearly the smallest possible distance between two
primes.
For example (3, 5), (5, 7), (11, 13), (17, 19), (29, 31), ..., (419, 421), ... are twin primes.
So far the following is not known.
22
Conjecture: There are innitely many twin primes.
Based on heuristic considerations, G. H. Hardy (1877-1947) and J. E. Littlewood
(1885-1977) developed a law (the twin prime conjecture (1922)) to estimate the
density of twin primes.
The prime number theorem says that the probability that a number N is prime is
about 1/ log(N). Therefore, if the probability that N+2 is prime would be indepen-
dent of the probability that N is prime, we should have the approximation for the
twin prime counting function
2
(N) := {(p, q) : (p, q) are twin prime number, and
p, q N},
lim
N

2
(N)
N
(log(N))
2
= 1.
A more careful analysis shows that this is too simple. In fact we have the following
more accurate conjecture.
Twin prime conjecture:
lim
N

2
(N)
N
(log(N))
2
= 2C
2
,
where C
2
= 0.6601618151468695739278121100145 . . . is the twin prime con-
stant.
The following rather remarkable result from 1919 is due to the Norwegian mathe-
matician V. Brun (1885-1978). Although it is not known if there are innitely many
twin prime numbers, Brun showed that the sum of the inverses of all twin primes
converges to a constant.
Theorem 4.11

(p,q) twin primes


_
1
p
+
1
q
_
= B
2
,
where B
2
= 1.902160583104... is the Brun constant.
A few more Conjectures.
There are still loads of other (old and new) unsolved problems concerning prime
numbers. Here is just a very tiny list of some of them.
1. Are there innitely many primes of the form N
2
+ 1 ? (Dirichlet proved that
every arithmetic progression (a + bn)
nN
with a, b coprime contains innitely
many primes.)
2. Is there always a prime between N
2
and (N + 1)
2
? (The fact that there is
always a prime between N and 2N was proved by Chebyshev.)
23
3. N
2
N + 41 is prime for 0 N 40. Are there innitely many primes of
this form? The same question applies to N
2
79N + 1601 which is prime for
0 N 79.
4. Are there innitely many primes of the form
_
_
_

p prime
pN
p
_
_
_ + 1 ?
5. Are there innitely many primes of the form
_
_
_

p prime
pN
p
_
_
_ 1 ?
6. Are there innitely many primes of the form N! + 1?
7. Are there innitely many primes of the form N! 1?
8. If p is a prime number, is 2p 1 then always not divisible by the square of a
prime number.
9. Does the Fibonacci sequence contain innitely many prime numbers?
24

Das könnte Ihnen auch gefallen