Anda di halaman 1dari 8

1 Integers and primes

What can be simpler than addition and multiplication in the natural numbers,
N, the integers, Z, and the rational numbers Q?

N = { 0, 1, 2, . . . }
Z = { . . . , 2, 1, 0, 1, 2, . . . }
Q = {a/b such that a, b Z, b 6= 0}.

1.1 Divisibility and primes


We distinguish four classes of integers:

1. Zero
2. Units
The integers 1 and 1. They are the integers that have an integer multi-
plicative inverse. Their only positive factor is 1.
3. Primes
A prime number is a nonzero integer p such that there is no factorization
p = mn with |m| < p and |n| < p. In other words, p can not be constructed
by multiplication from smaller integers. It has exactly two positive factors,
1 and |p|. Often prime numbers are assumed to be positive in which case
its two positive factors are 1 and itself.
Examples: 2, 3, 5, 7, 11, 144169, -2, -3. The number 1 is NOT a prime
number.
The prime counting function (n) is given by

(n) = The number of primes from 1 to n.

4. Composites
A composite number is a nonzero integer with 3 or more positive factors.
Examples: 4, 6, 8, 9, 10, 1010

How many primes less than 30? Less than 100? What fraction of the integers
from 1 to n are prime ((n)/n)?
Fundamental theorem of arithmetic Every integer greater than 1 is a
prime or a product of primes in one and only one way.
Proof of existence of a prime factorization: This is the easy half of the
theorem, proved by induction. Check out 2, 3, 4, 5 ,6 ,7, 8 by trial and error.
If all integers from 2 to N are a product of primes, what about N + 1? It is
either a prime OR a product of of two integers both between 2 and N which we
already know are each products of primes so N + 1 itself is a product of primes.
QED

1
Proof of uniqueness of prime factorizations is harder. There is some subtlety.
We require
Euclids Lemma. If a prime divides a product of two integers, then it must
divide one of the factors.
Try to prove Euclids Lemma on your own! Not so easy. The proof takes a
detour through greatest common divisors and will come later.
Proof of uniqueness of prime factorizations using Euclids Lemma.
Step one is to generalize Euclids lemma to products of more than two factors.
If a prime divides any product of integers, then it must divide one of the factors.
So suppose p is prime and p|a1 a2 a3 aN . We must show that p|ai for some
i. The case N = 1 is trivial, the case N = 2 is Euclids Lemma. The case
N > 2 follows by induction on the number of factors N in the product. Since
p|a1 a2 a3 aN then p|(a1 a2 a3 aN 1 )aN , so by Euclid p|a1 a2 a3 aN 1 (and
hence one of the ai for 1 i N 1) or p|aN .
Step two: Proceed by induction. By trial and error we see that the first
few integers, 2, 3, 4, 5, 6, 7, have unique factorizations into products of primes:
Suppose we have verified that all integers a with 1 a < c have unique factor-
izations into a product of primes. Suppose
c = p1 p2 pM = q1 q2 qN
where the ps and qs are primes. We must show that M = N and that the
ps and qs are the same. Assume that integers a with 1 a < c have unique
factorizations. Then by step 1, one of the ps must divide one of the qs, and
because they are both primes, they must be equal. By numbering suitably, we
can assume that pM = qN . Canceling the factor, we have
c0 = p1 p2 pM 1 = q1 q2 qN 1 < c.
We already know that c0 has a unique factorization, so M 1 = N 1 and the
ps in the factorization of c0 are the same as the qs in the factorization of c0 .
QED
Difficulty of primality testing: Theres a polynomial time algorithm, but it
is not very fast.
Difficulty of factoring: no polynomial time algorithm known. The inability
of computers to factor the product of two 300 digits primes is the key to some
currently unbreakable encryption algorithms

1.2 Euclidean algorithm


Theorem (Division Algorithm) Let a, b Z with b 6= 0. Then we can divide
a by b getting an integer quotient q and integer remainder r where r is smaller
than the divisor b. More precisely, There exist q, r Z such
a = qb + r and |r| < |b|.
Moreover, we can require that 0 r < |b| and then q and r are uniquely
determined.

2
Proof.
Let q be the largest integer that is less than or equal to the rational number
a/b. Let r = a qb. Note that r is an integer because a, q, and b are. From the
inequalities q a/b < q + 1 we see that |a/b q| < 1. Hence |r| = |b(a/b q)| =
|b| |a/b q| < |b|.
If r < 0, then a = q 0 b + r0 with q 0 = q 1 and 0 < r0 = r + b < |b|.
For uniqueness, suppose a = q1 b + r1 = q2 b + r2 where 0 r1 < |b| and
0 r2 < |b|. Then (q1 q2 )b = r2 r1 . Evidently b is a factor of r2 r1 , and
since b < r2 r1 < b we must have r2 r1 = 0. Hence also q1 = q2 = 0.
QED
If you list all the integers that divide both of two nonzero integers a and
b, you get a finite list. Among them is a largest, called the greatest common
divisor.
Example:
divisors of 30: 1, 2, 3, 4, 5, 6, 10, 30
divisors of 42: 1, 2, 3, 6, 7, 14, 21, 42.
common divisors of 30 and 42: 1, 2, 3, 6.
gcd(30, 42) = 6.
Notice that the set of common divisors of 30 and 42 is actually the set of all
divisors of their gcd. If you know the gcd of two integers, you can list all their
common divisors by listing all divisors of a special single number, their gcd. Its
pretty neat, really. The set of common divisors of two numbers is the same as
the set of all divisors of a single special number. Proof coming soon.
More examples: gcd(5, 7) = 1, gcd(12, 24) = 12, gcd(60, 84) = 12, gcd(0, 411) =
411.
The key to efficiently computing gcds is that many pairs of integers have
the same sets of common divisors.
Useful Lemma: Suppose

a = qb + r

where a, b, q, and r are four integers, with b 6= 0.

1. The set of common divisors of a and b is the same as the set of common
divisors of b and r. In particular,

gcd(a, b) = gcd(b, r).

2. The two equations

ax + by = d and bx + ry = d

possess integer solutions (x, y) for exactly the same integers d.

Proof:

3
1. If c is a common divisor of a and b, with a = cm and b = cn, then
r = c(m qn) so c is also a divisor of r.
If c is a common divisor of b and r, with b = cm and r = cn, then
a = c(qm + n), so c is also a divisor of a.
2. If ax + by = d, then bx1 + ry1 = d with x1 = qx + y and y1 = x.
If bx + ry = d, then ax1 + by1 = d with x1 = y and y1 = x qy.

Theorem (Euclidean algorithm). Using the division algorithm a finite


number of times gives a good way (i.e., not requiring factorization of integers,
which is hard) to find a greatest common divisor d for each pair of integers a
and b not both zero. Moreover, the set of common divisors of a and b is the
same as the set of common divisors of 0 and d, namely it is the set of divisors
of the single number d.
Proof. Suppose a b > 0. Do repeated divisions with remainder, always
taking nonnegative remainders to produce a sequence

a, b, r1 , r2 , r3 , . . .

where each ri is a remainder upon division with the preceding two elements:

a = q1 b + r1
b = q2 r1 + r2
r1 = q3 r2 + r3
r2 = q4 r3 + r4

rm1 = qm+1 rm + rm+1 .

Since b > r1 > r2 > 0 the sequence must terminate with rM +1 = 0 for
some M .
Furthermore, by the Useful Lemma, every pair of consecutive elements has
the same common divisors and the same gcd:

gcd(a, b) = gcd(b, r1 ) = gcd(r1 , r2 ) = = gcd(rM , 0) = rM .

QED
Example: To find gcd(84, 30), produce the remainder sequence (84, 30, 24, 6, 0)
so the gcd is 6.
A little proof analysis shows more, one of the most useful results in all of
number theory.
Master Theorem of linear diophantine equations: There exists an
integer solution (x, y) of the equation

ax + by = d

4
if and only if d is a multiple of gcd(a, b). In particular, there is a solution of the
equation
ax + by = gcd(a, b).
Proof Use the second part of the useful lemma. QED
Proof of Euclids Lemma, using the Master Theorem. Suppose that p is
prime and p|ab. If p does not divide a, then gcd(a, p) = 1, so there exist
integers x and y such that ax + py = 1. Multiplying by b gives (ab)x + pby = b.
Since p divides both terms on the left side of the equality it must divide their
sum. So p|b. QED
Proposition. Let m and n be positive integers. If gcd(m, n) = 1 and a is a
multiple of both m and n, then a is also a multiple of the product mn.
Proof. By the Master Theorem, there are integers r and s such that rm +
sn = 1. Hence
a = arm + asn.
Since a is a multiple of n, we see that arm is a multiple of mn. Since a is a
multiple of m we see that asn is also a multiple of mn. Therefore a is the sum
of two multiples of mn and thus is itself also a multiple of mn. QED

1.3 Distribution of the primes.

Theorem (Euclid) There are infinitely many primes.


Proof: Factor p1 p2 pN + 1.
Theorem (Euler) The sum of the reciprocals of primes diverges:
X1 1 1 1 1
= + + + + =
p 2 3 5 7

Proof: (1737) First appearance of the Riemann zeta function.


Chebychev: There are constants A and B such that
n n
A < (n) < B .
ln n ln n
Proof. Chebychev (1848-1850) found an elementary proof based on an anal-
ysis of binomial coefficients.
Prime number theorem: There are approximately n/ ln n primes less than
n. More precisely,
(n)
lim = 1.
n n/ ln n

Sometimes people say that the probability that an integer between 1 and n is
prime is about 1/ ln n. This is a good intuitive guide to the meaning of the prime
number theorem, but should not be taken literally. It is not strictly speaking
true.

5
Proof: Not easy at all. Outline of proof given by Riemann in 1859. Details
completed by Hadamard and de la Vallee-Poussin in 1896.The most natural
proofs call upon the theorems of complex analysis. The elementary proofs
are combinatorially intricate.
It can be deduced from the prime number theorem that the nth prime pn is
about n ln n. More precisely

n ln n + n ln ln n n < pn < n ln n + n ln ln n for n 6.

Sieve of Eratosthenes
Theorem Primes in arithmetic progressions. If gcdd(a, b) = 1, then there is
an infinite number of integer values of x such that ax + b is prime.
Proof. Dirichlet 1837 One of the great theorems of all time. No elementary
proof known.
Examples: 10x + 1, 10x + 3, 10x + 7, 10x + 9.
Example. We can do the case 4x + 3 very easily by imitating Euclid. Factor

4p1 p2 p3 pN 1.

Primes represented by nonlinear polynomials such as x2 + 1. Not known -


but guessed that x2 + 1 is prime for infinitely many integers x.

1.4 Prime lore

Prime gaps - can be as big as you please. Consider the n 1 consecutive


composites. n! + 2, n! + 3, , n! + n P 1
Twin primes: Are there infinitely many? Known that twins < .
Current closest: There exist an infinite number of pairs p1 < p2 of primes
where p2 p1 246. The breakthrough was made in 2013 by Yitang Zhang.
Bertrands theorem: for every integer n > 1 there is a prime p with n < p <
2n. (Theres a beautiful elementary proof, said to be from the book.)
Erdos proved in 1934 that for any positive integer k, there are at least k
primes between n and 2n for all n past a certain point N that depends on k.
Legendre conjecture (not settled) states that for every n there is a prime
number between n2 and (n + 1)2
Mersenne primes are of the form 2p 1 where p is prime. It is conjectured
(but not known) that there are infinitely many Mersenne primes. The largest
known prime is the Mersenne prime with p = 74, 207, 281 with 22,338,618 digits,
discovered Jan 7, 2016.
Goldbach Conjecture (not settled): Every even integer greater than 2 is the
sum of two primes.
Green- Tao theorem(2004). There are arbitrarily long arithmetic progres-
sions of primes. The current known record is a 26 prime arithmetic progression
beginning with 43,142,746,595,714,191.

6
Erdos-Turan
P conjecture (not settled) : Every set A of positive integers such
that nA 1/n = contains arbitrarily long arithmetic progressions.
Conjecture (unsettled) There are arbitrarily long arithmetic progressions
made from consecutive primes. Current record is 10 consecutive primes in arith-
metic progression.

1.5 The abc conjecture


The abc conjecture concerns the abc equation, a + b = c. The conjecture is the
most important unsolved problem in diophantine equations today. Its about
primitive solutions of the abc equation, those with gcd(a, b) = 1.
Definition. The radical of a nonzero integer is the product of its prime
divisors.
Examples: rad(24 35 52 17 101) = 2 3 5 17 101 = 51510, rad(6) = 6,
rad(12) = 6, rad(1) = 1.
The abc conjecture For every k > 1 there is only a finite number (possibly
zero) of triples of positive integers a, b, c such that
1. a + b = c

2. gcd(a, b) = 1
3. c > rad(abc)k
The abc conjecture was proposed by Oesterle and Masser in 1985. A long
proof was announced by Mochizuki in 2012. Mathematicians have been working
hard but so far have not understood enough of the proof to decide whether
it is correct. The conjecture is important because if true it can be used to
give relatively simple solutions of many other great problems in mathematics,
including Fermats Last Theorem.
The abc conjecture implies that there is some K > 1 such that c < rad(abc)K
for all primitive solutions of the equation a + b = c. Some have suggested that
K = 2 has this property.
Example: Assuming that c < rad(abc)2 for all primitive solutions of the abc
equation, find all nonnegative integer solutions (m, n) of the equation

1 + 2m = 3n .

SOLUTION. We take a = 1, b = 2m , c = 3n , so rad(abc) = 6. Therefore

3n = c < rad(abc)2 = 36.

Hence the only possibilities for n are 0, 1, 2, 3. Checking these four cases one
by one we find only one solution to the equation:

1 + 23 = 32 .

7
Example: Assuming that c < rad(abc)2 for all primitive solutions of the abc
equation, find all positive primitive solutions (x, y, z) Fermats equation

xn + y n = z n

for integers n 3.
SOLUTION We have

zn < rad(xn y n z n )2 = rad(xyz)2 (xyz)2 z 6 .

Hence n < 6. We only need to investigate Fermats equation for n = 3, 4, 5.


These are classic cases for which elementary proofs show that there are no
primitive solutions. QED
Definition. The quality of a primitive solution of the abc equation is given
by
ln c
quality(a, b, c) = .
ln rad(abc)
Since
quality(a, b, c) > k if and only if c > rad(abc)k
we have a new formulation.
The abc conjecture For every k > 1 there is only a finite number (possibly
zero) of primitive solutions of a + b = c such that

quality(a, b, c) > k.

Since large quality is both interesting and rare, extensive searches have been
made for large quality triples. As of today, March 5, 2017, only 239 triples are
known with quality greater than 1.4. The three with quality greater than 1.6
are
quality a b c
10
1.6299 2 3 109 235
2 2 6 3
1.6260 11 3 5 7 221 23
1.6235 19 1307 7 292 318 28 322 54
If the abc conjecture is true, there can only be a finite number of triples with
higher quality. If there is a higher quality triple, computer searches show that
it must have c > 1020 . Notice that high quality corresponds to products of high
powers of small numbers of relatively small primes.

Anda mungkin juga menyukai