Sei sulla pagina 1di 4

A principal ideal domain that is not an Euclidean

Domain
Edgar Elizeche
St. Stephens College
March 15, 2015

Basics

Definition 1.1. We define =

1+ 19
2

and let A = Z[] = {a + b|a, b Z}

In order to prove that A is a PID we need to prove first that A is a ring.


For this we need the fact that is a root of x2 x + 5 = 0
Since complex roots come in conjugate pairs, we get that is also a root
of the polynomial.
From Cardanno-Viettes equations we get that
= 5

(1.1)

+ =1

(1.2)

Also we get the neat fact that


2 = 5

(1.3)

With this weaponry in our arsenal we are ready to prove that A is a ring.
We can consider A as a subring of C. Therefore we can apply the subring
test.
Let x, y A i.e. x = a + b, y = c + d for some a, b, c, d Z Then:
x y = (a + b) (c + d) = (a c) + (b d)
xy = (a + b) (c + d) = ac + (bc + ad) + bd2
xy = ac + (bc + ad) + bd( 5)
xy = (ac 5bd) + (bc + ad + bd)
Therefore A is a ring.
1

(by (1.3))

A is not an Euclidean Domain

Definition 2.1. We say that an Integral Domain is an Euclidean Domain


if | | : A Z such that:
1. | |(a) = |a| > 0 for all a A; |a| = 0 a = 0
2. |ab| = |a||b|a, b A.
3. (Euclidean algorithm) Given a, b A, b 6= 0, q, r A such that
a = bq + r with |r| < |b|.
We will weaken condition 2 from definition 2.1 to:
|a| 6 |b| if a divides b(b 6= 0)

(2.1)

Now, we assumed that there is such a function | | (i.e. A is an Euclidean


Domain). We will reach a contradicition. First we define one more set
Definition 2.2. We define U to be the set of nonzero elements with minimal
norm | |
Now from equation (2.1) we have that every unit is in U (as every unit
divides all the elements, in particular those of U ). Also we have that every
element of minimal order is a unit. We prove this as follows:
Let u U . By definition 2.1 part (3) for a = 1, b = u we get a q, r A
such that 1 = uq + r with |r| < |u|. But u, by definition, has minimal
nonzero norm. Therefore r = 0, and u divides 1, and therefore it is a unit.
We now will show that U = {1, 1}
To attain this feat we will need the help of our dear friend, the complex
norm.
Definition 2.3. We define N : C R as:
N (z) = z z
We have also some properties of the complex norm that are useful. In
particular:
N (a + b) = (a + b)(a + b) = a2 + ab + 5b2
And also:
N (xy) = N (x)N (y) for all x, y A
N (x) > 0 for all x A and N (x) = 0 x = 0
2

Now, again, since the units in U divide 1, therefore the complex norm has
to be 1 (by the first property of complex norm). Also is u = a + b U then
N (u) = N (a + b) = a2 + ab + 5b2 = 1 So if ab > 0 then b = 0 and a = 1.
Similarly, if ab 6 0 then N (u) = N (u) = (a + b)2 ab + 4b2 = 1. So again
b = 0 and a = 1. Therefore U = {1, 1}
Now we assume that m is of minimal form among the elements of A/(U
{0}). Condition (3) from definition (2.1) tells us that q, r A such that
2 = qm + r, with |r| < |m|; therefore r is either 1, 1 or 0. Therefore
m divides 2 or 3. We claim that m has to be one of 2 or 3. This is
a consequence from the fact that 2 and 3 are irreducibles in A Assume if
possible 2 = (a + b)(c + d). Then 4 = N (2) = N (a + b)N (c + d). Since
a + b and c + d/theta are non-units, their norms are not 1. Therefore
2 = N (a + b) = a2 + ab + 5b2 = N (a + b) = (a + b)2 ab + 4b2
Therefore, again considering the cases ab > 0 and ab 6 0 we get b and d
must equal zero. Therefore 2 = (a + b)(c + d) = ac and this is normal
factorization in integers. Therefore either a = 1 or c = 1. Therefore 2 is
irreducible in A. Similarly 3 is also an irreducible in A.
Now, again using definition 2.1 part 3, is congruent to 0, 1 or 1
modulo m. Hence, , + 1 or 1 is divisible by either 2 or 3. But this
is not possible as N () = N () = N (1) = N ( 1) = 5 and N ( +1) = 7,
but N (2) = 4 and N (3) = 9.
Therefore A is not an Euclidean Domain.

A is a PID

In this section we will show that A is a Principal Ideal Domain. Now, for
this we will need the following identity:
x = a + b; a, b Q = x = 5b + (a + b)

(3.1)

This is easy to prove:


x = a + b
x = a + b2
x = a + b( 5)
x = 5b + (a + b)
We will prove the fact that A is almost Euclidean; meaning that given
any elements , A, 6= 0 if does not divide and N () > N () then
3

there exist , A such that:


0 < N ( ) < N ()

(3.2)

We call this property almost Euclidean because if = 1 for all , A


then A is Euclidean.
To prove (3.2) take , A; 6= 0. If does not divide and N () >
N () write

= p + q

where p and q are rational numbers and at least one of them is not an integer
(otherwise divides ). This is possible since we can see A Q[] which is
a subring of C.
Now, we split the proof into seven cases, which will lead to elements
, A such that


0<N
< 1 therefore N ( ) < N ()

Case 1: q Z, therefore p
/ Z. Then we can take = 1 and = [p]+q
(where [x] denotes the integer closest to x i.e. [x] = bx + 12 c where bxc
denotes the floor function). Then:


0<N
= N (p [p])

 1 2
1
6
because |r [r]| 6
2
2
1
= <1
4
Case 2a:p Z but 5q
/ Z. Then, as
= , = [p + 5q] p. We have:
0<N

<

= p + 5q p, we may take

1
4

as before

Case 2b: p Z and 5q Z. Then we take = 1 and = p + [q].


1
Now, as 5q Z therefore q = m
as |q [q]|
5 . Therefore, 
 < 2 we have that
or |q [q]| = 25 . Therefore, N
 2
5N (q [q]) 6 5 52 = 45 < 1
Case 3a:

|q [q]| =

1
5

= N ((q [q])) =

Potrebbero piacerti anche