ProofComplete

Proof of Unique Factorization in PIDs

We prove that every principal ideal domain is a unique factorization domain, establishing the key link in the hierarchy ED \subset PID \subset UFD.


Statement

Theorem6.9PID implies UFD

Every principal ideal domain is a unique factorization domain. That is, every nonzero non-unit element of a PID can be written as a product of irreducible elements, uniquely up to order and associates.


Proof

Proof

Let RR be a PID. We prove existence and uniqueness of factorization separately.

Part 1: Existence of factorization. We show every nonzero non-unit has a factorization into irreducibles using the ascending chain condition.

Claim: RR satisfies the ACC on principal ideals. If (a1)(a2)(a_1) \subseteq (a_2) \subseteq \cdots, then I=(an)I = \bigcup (a_n) is an ideal. Since RR is a PID, I=(d)I = (d) for some dd. Then d(aN)d \in (a_N) for some NN, so I=(d)(aN)I = (d) \subseteq (a_N), giving (an)=(aN)(a_n) = (a_N) for all nNn \geq N.

Now suppose some nonzero non-unit aa has no factorization into irreducibles. Then aa is not irreducible, so a=b1c1a = b_1 c_1 with neither b1b_1 nor c1c_1 a unit. Then (a)(b1)(a) \subsetneq (b_1) and (a)(c1)(a) \subsetneq (c_1). At least one of b1,c1b_1, c_1 (say b1b_1) also has no irreducible factorization. Repeat: b1=b2c2b_1 = b_2 c_2 with (b1)(b2)(b_1) \subsetneq (b_2), and b2b_2 has no factorization. This produces a strictly ascending chain (a)(b1)(b2)(a) \subsetneq (b_1) \subsetneq (b_2) \subsetneq \cdots, contradicting the ACC. So every nonzero non-unit factors into irreducibles.

Part 2: Irreducible implies prime in a PID.

Let pp be irreducible. Since RR is a PID, (p)(p) is an ideal. We show (p)(p) is prime.

Suppose pabp \mid ab, i.e., ab(p)ab \in (p). Consider d=gcd(p,a)d = \gcd(p, a). In a PID, (d)=(p,a)=(p)+(a)(d) = (p, a) = (p) + (a). Since dpd \mid p and pp is irreducible: either dd is a unit or dd is an associate of pp.

If dd is an associate of pp: pap \mid a, done. If dd is a unit: then (p,a)=R(p, a) = R, so 1=sp+ta1 = sp + ta for some s,tRs, t \in R. Then b=spb+tabb = spb + tab. Since pabp \mid ab: ptabp \mid tab. And pspbp \mid spb. So pbp \mid b.

Part 3: Uniqueness of factorization.

Suppose p1pm=q1qnp_1 \cdots p_m = q_1 \cdots q_n with all factors irreducible (hence prime). Since p1q1qnp_1 \mid q_1 \cdots q_n and p1p_1 is prime, p1qjp_1 \mid q_j for some jj. Since qjq_j is irreducible, p1p_1 and qjq_j are associates. After reordering, p1q1p_1 \sim q_1. Cancel (valid in an integral domain): p2pmq2qnp_2 \cdots p_m \sim q_2 \cdots q_n. By induction, m=nm = n and piqip_i \sim q_i after reordering. \blacksquare


Consequences

ExampleFactorization in specific PIDs

In Z[i]\mathbb{Z}[i]: 5=(2+i)(2i)5 = (2+i)(2-i) and N(2+i)=5N(2+i) = 5 is prime in Z\mathbb{Z}, so 2+i2+i and 2i2-i are irreducible (and prime) in Z[i]\mathbb{Z}[i].

The rational prime pp factors in Z[i]\mathbb{Z}[i] as:

  • p=ππˉp = \pi \bar{\pi} with π≁πˉ\pi \not\sim \bar{\pi} if p1(mod4)p \equiv 1 \pmod{4} (e.g., 5=(2+i)(2i)5 = (2+i)(2-i)).
  • pp remains prime if p3(mod4)p \equiv 3 \pmod{4} (e.g., 33 is prime in Z[i]\mathbb{Z}[i]).
  • 2=i(1+i)22 = -i(1+i)^2 (ramified).
RemarkBeyond PIDs

The proof uses two PID-specific properties: ACC on principal ideals (for existence) and the Bezout property gcd(a,b)(a,b)\gcd(a,b) \in (a,b) (for irreducible     \implies prime). In general UFDs, existence needs the ACC, while irreducible     \implies prime is given as an axiom.