Definition 1An integral domain Rmathend000# endowed with a function
d : R { - }
mathend000#
is an Euclidean domain if the following two conditions hold
for all a, bRmathend000# with a 0
mathend000# and b 0
mathend000# we have
d (ab) d (a)
mathend000#,
for all a, bRmathend000# with b 0
mathend000#
there exist q, rRmathend000# such that
a = bq + randd (r) < d (b).
(1)
The elements qmathend000# and rmathend000# are called the quotient and the
remainder of amathend000# w.r.t. bmathend000#
(although qmathend000# and rmathend000# may not be unique).
The function dmathend000# is called the Euclidean size.
Example 1Here are some classical examples.
R = mathend000# with
d (a) = | a |
mathend000# for
amathend000#.
Here the quotient qmathend000# and the remainder rmathend000# of amathend000# w.r.t. bmathend000#
(with b 0
mathend000#) can be made unique by requiring r 0
mathend000#
(hence we have
0 r < bmathend000#).
R = [x]
mathend000# where mathend000# is a field
with
d (a) = deg(a)
mathend000# the degree of amathend000# for
aR, a 0
mathend000#
and
d (0) = - mathend000#.
Uniqueness of the quotient and the remainder is easy to show
in that case. Indeed
a = bq1 + r1 = bq2 + r2with deg(r1) < deg(b) and deg(r2) < deg(b)
(2)
implies
r1 - r2 = b (q1 - q2) with deg(r1 - r2) < deg(b)
(3)
Hence we must have
q1 - q2 = 0
mathend000# and thus
r1 - r2 = 0
mathend000#.
R = mathend000# is a field with d (a) = 1
mathend000# for
a, a 0
mathend000#
and d (0) = 0
mathend000#.
In this case the quotient qmathend000# and the remainder rmathend000# of amathend000# w.r.t. bmathend000#
are a/bmathend000# and 0
mathend000# respectively.
Let Rmathend000# be the ring of the complex numbers whose real and imaginary
parts are integer numbers. Hence
R = {x + iy | x, y}
(4)
Consider as a map dmathend000# from Rmathend000# to
{ - }
mathend000#
the norm of an element.
Hence
d (x + iy) = x2 + y2mathend000# with
x, ymathend000#.
It is easy to check that for every a, bRmathend000# with
a, b 0
mathend000#
we have
d (ab) d (a)
mathend000#.
Indeed for
x, y, z, tmathend000# we have
d ((x + iy)(z + it))
=
d (xz - ty + yz + txi)
=
(xz - ty)2 + (yz + tx)2
=
x2z2 + t2y2 -2xzty + y2z2 + t2x2 +2xzty
=
x2 (z2 + t2) + y2 (z2 + t2)
=
y2 + x2z2 + t2
=
d (x + iy) d (z + it)
(5)
Moreover for every a 0
mathend000# we have
d (a) 1
mathend000#.
Therefore we have proved that
d (ab) d (a)
mathend000# holds
for every a, bRmathend000# with
a, b 0
mathend000#.
Now given a, bRmathend000# with b 0
mathend000# we are looking for
a quotient and a remainder of amathend000# w.r.t. bmathend000#.
Hence we are looking for qmathend000# such that
d (a - bq) < d (b)
mathend000#.
Such a qmathend000# can be constructed as follows.
Let q'mathend000# be such that
a - q'b = 0
mathend000#
that is
q' = - a/b = - a/d (b)
mathend000#
where
mathend000# is the conjugate of bmathend000#.
Hence q'mathend000# writes
x' + iy'mathend000# with
x', y'mathend000#.
Let
x, ymathend000# be such that
| x - x' | 1/2
mathend000# and
| y - y' | 1/2
mathend000#.
Then
d (a - bq)
=
d (a - bq + bq' - bq')
=
d (b(q - q'))
=
d (b) | x - x' + | y - y'
d (b)/2
<
d (b).
(6)
It turns out that several qmathend000# can be chosen.
For instance with
a = 1 + imathend000# and
b = 2 - 2 imathend000#
we have
a - bq = - 1 - imathend000# with q = imathend000#
and
a - bq = 1 + imathend000# with q = 0
mathend000#.
In both cases
d (a - bq) = 2 < 4 = d (b)
mathend000#.
Finally this shows that a quotient and a remainder of amathend000# w.r.t. bmathend000#
may not be uniquely defined in Rmathend000#.