Principal Ideal Domain
For elements a1,a2,…,an of the ring R, we define
(a1,a2,…,an)=Ra1+Ra2+⋯+Ran={i=1∑nriai∣ri∈R}.
Then, (a1,a2,…,an) is an ideal.
If an ideal I is equal to (a1,a2,…,an) for some elements ai∈I, we say that I is finitely generated. If I=(a) for some a∈I, we say that I is a principal ideal.
Principal Ideal Domain
If every ideal of R is principal, then we say that R is a principal ideal domain (PID).
Every Euclidean domain is a PID, but not every PID is an Euclidean domain, i.e. PID ⊃ Euclidean domain.
However finding counterexamples to PID = Euclidean domain is somewhat hard.
In practice, it is useful to show that a given ring is an Euclidean domain, which gives us the added bonus of showing that it is a PID.
Terminology
Division
If a,b∈R,b=0, then we say that b divides a if ∃c∈R,a=bc and write b∣a.
In terms of ideals, a∣b⟺(b)⊆(a).
Associate
a,b∈R are associates if a=bu for some unit u.
A unit is defined as u∈R such that (u)=R.
And in terms of ideals, a and b are associates iff (a)=(b).
Irreducibility
An element p∈R is irreductible if a∣p implies that a is either a unit or an associate of p.
Prime
A nonunit p∈R∖{0} is said to be prime if p∣ab implies that p∣a or p∣b.
In terms of ideals, p prime iff ab∈(p)⟹a∈(p)∨b∈(p).
Greatest Common Divisor
Greatest Common Divisor
d∈R is said to be a (gcd) of two elements a,b∈R if
- d∣a and d∣b,
- d′∣a∧d′∣b⟹d′∣d.
If both d and d′ are gcds of a and b, then d is associate to d′.
Let R be a PID and a,b∈R. Then a and b have a gcd and (a,b)=(d).
Form the ideal (a,b).
Since R is a PID there exists d such that (a,b)=(d).
Since (a)⊆(d) and (b)⊆(d) we have that d∣a and d∣b.
If d′∣a and d′∣b, then (a)⊆(d′) and (b)⊆(d′).
Then, (d)=(a,b)⊆(d′) and d′∣d.
Therefore (a,b)=(d).
Coprimality
Coprime
a and b are coprime iff their common divisors are units.
Corollary 1
If R is a PID and a,b∈R are coprime, then (a,b)=R.
Corollary 2
If R is a PID, and p∈R is irreductible, then p prime.
Sps. that p∣ab and that p∤a.
Since p∤a it follows that they only share units as common divisors.
By corollary 1 (a,p)=R, so (ab,pb)=(b).
Since ab∈(p) and pb∈(p) we have (b)⊆(p).
Therefore p∣b and in a PID, prime elements ⟺ irreductible elements.
In a PID we will use prime and irreductible interchangeably.
Order
Let (a1)⊆(a2)⊆(a3)⊆… be an ascending chain of ideals.
Then there exists k∈N such that (ak)=(ak+l) for l=0,1,2,….
That is, the chain breaks in finitely many steps.
Let I=⋃i=1∞(ai), obviously I is an ideal so I=(a) for some a∈R.
But a∈⋃i=1∞(ai) implies that a∈(ak) for some k.
This shows that I=(a)⊆(ak).
It then follows that I=(ak)=(ak+l)=…
Proposition 5
Every nonzero nonunit of R is a product of irreductibles.
Let a∈R∖{0}, where a is a nonunit, then we want to prove the following cases.
-
a is divisible by irreductibles
In the case that a itself is irreductible, we are done.
Otherwise, a=a1b1 where a1,b1 are nonunits.
If a1 irreductible, done.
Otherwise a1=a2b2 where a2,b2 are nonunits.
If a2 irreductible, done.
Otherwise, continue as before.
Notice that (a)⊂(a1)⊂(a2)⊂… and by lemma 5 we have that this is a finite chain, so there must exist some k such that ak is irreductible.
-
a is the product of irreductibles
If a is irreductible, done.
Otherwise let p1 be some irreductible such that p1∣a.
Then a=p1c1.
If c1 is a unit, done.
Otherwise, continue as before.
Again, notice that (a)⊂(c1)⊂(c2)⊂…, which is a finite chain by lemma 5.
There must exist some k for which a=p1p2…pkck, where ck a unit and pkck is irreductible.
We use the following lemma to construct the ord function akin to those for Z and k[x].
Let p be prime and a=0. Then there exists n∈Z such that pn∣a but pn+1∤a.
Sps. that there does not exist an n satisfying the lemma, then for each integer m>0 there would be an element bm such that a=pmbm.
Then pbm+1=bm so that (b1)⊂(b2)⊂(b3)⊂ which would be an infinite chain which contradicts lemma 5.
Then the integer n defined in lemma 6 is uniquely determined by p and a, and we set n=ordpa.
Unique Factorization Proof
If a,b∈R∖{0}, then ordp(ab)=ordpa+ordpb.
Let α=ordpa and β=ordpb.
Then a=pαc and b=pβd with p∤c and p∤d.
Thus ab=pα+βcd.
Since p prime, p∤cd.
Therefore ordpab=α+β=ordpa+ordpb
Let S⊂R be a set of primes with the properties
- ∀p∈R∃s∈R, such that p and s are associates.
- ∄a,b∈S such that a and b are associates.
To generate the set S we can choose a representative prime from each class of associate primes.
In Z we can construct the set S by choosing all positive primes.
In k[x] we chose S to be the set of monic irreductible polynomials.
There does not exist a general way to construct S for an arbitrary PID.
Let R be a PID and S a set of primes with the aforementioned properties.
Then if a∈R∖{0}, then we can write
a=up∏pe(p),(†)
where u is a unit and the product is over all p∈S.
The unit u and the exponents e(p) are uniquely determined by a.
In fact, e(p)=ordpq.
The existence of such a decomposition follows from proposition 5.
To show the uniqueness, let q∈S and apply ordq to both sides of eq. (†).
Using lemma 7, we get
ordqa=ordqu+p∑e(p)ordqp.
By definition of ordq we get that ordqu=0 and that ordqp=0 if q=p and 1 otherwise.
So ordqa=e(q). Since exponents e(q) are uniquely determined, so is the unit u.