Skip to content

Euclidean Domains, PIDs, and UFDs

An integral domain RR is a Euclidean domain if there exists a function δ:R{0}N0\delta : R \setminus \{0\} \to \mathbb{N}_0 such that for all a,bRa, b \in R with b0b \neq 0:

  1. There exist q,rRq, r \in R with a=bq+ra = bq + r and either r=0r = 0 or δ(r)<δ(b)\delta(r) \lt \delta(b).

Example. Z\mathbb{Z} is a Euclidean domain with δ(a)=a\delta(a) = |a|.

Example. F[x]F[x] is a Euclidean domain with δ(f)=deg(f)\delta(f) = \deg(f).

Example. Z[i]\mathbb{Z}[i] is a Euclidean domain with δ(a+bi)=a2+b2\delta(a + bi) = a^2 + b^2.

Example. Z[ω]\mathbb{Z}[\omega] (Eisenstein integers, ω=e2πi/3\omega = e^{2\pi i/3}) is a Euclidean domain with δ(a+bω)=a2ab+b2\delta(a + b\omega) = a^2 - ab + b^2.

An integral domain RR is a principal ideal domain (PID) if every ideal of RR is principal (generated by a single element).

Theorem 11.1. Every Euclidean domain is a PID.

Proof. Let II be a non-zero ideal of the Euclidean domain RR. Choose dI{0}d \in I \setminus \{0\} Minimising δ(d)\delta(d). We claim I=(d)I = (d). For any aIa \in IWrite a=qd+ra = qd + r with r=0r = 0 or δ(r)<δ(d)\delta(r) \lt \delta(d). Since r=aqdIr = a - qd \in IMinimality of δ(d)\delta(d) Forces r=0r = 0So a=qd(d)a = qd \in (d). \blacksquare

Corollary 11.2. Z\mathbb{Z}, F[x]F[x]And Z[i]\mathbb{Z}[i] are PIDs.

An integral domain RR is a unique factorization domain (UFD) if:

  1. Every non-zero, non-unit element factors into irreducibles.
  2. The factorization is unique up to ordering and associates.

Theorem 11.3. Every PID is a UFD.

The chain of implications is:

Euclidean domainPIDUFD\mathrm{Euclidean\ domain} \Rightarrow \mathrm{PID} \Rightarrow \mathrm{UFD}

None of the reverse implications hold .

Example. Z[x]\mathbb{Z}[x] is a UFD but not a PID. (The ideal (2,x)(2, x) is not principal.)

Example. Z[5]\mathbb{Z}[\sqrt{-5}] is not a UFD: 6=23=(1+5)(15)6 = 2 \cdot 3 = (1 + \sqrt{-5})(1 - \sqrt{-5}) Gives two distinct factorizations into irreducibles.

Problem. Show that Z[5]\mathbb{Z}[\sqrt{-5}] is not a UFD.

Solution

Solution. We show that 2,3,1+5,152, 3, 1 + \sqrt{-5}, 1 - \sqrt{-5} are all irreducible in Z[5]\mathbb{Z}[\sqrt{-5}] and that 2,32, 3 are not associates of 1±51 \pm \sqrt{-5}.

The norm is N(a+b5)=a2+5b2N(a + b\sqrt{-5}) = a^2 + 5b^2. Note N(αβ)=N(α)N(β)N(\alpha\beta) = N(\alpha)N(\beta) and α\alpha is a unit iff N(α)=1N(\alpha) = 1.

N(2)=4N(2) = 4. If 2=αβ2 = \alpha\beta with neither a unit, then N(α)N(β)=4N(\alpha)N(\beta) = 4So N(α)=N(β)=2N(\alpha) = N(\beta) = 2. But a2+5b2=2a^2 + 5b^2 = 2 has no integer solutions. So 22 is irreducible.

Similarly, N(3)=9N(3) = 9And a2+5b2=3a^2 + 5b^2 = 3 has no solutions, so 33 is irreducible.

N(1±5)=6N(1 \pm \sqrt{-5}) = 6. If 1+5=αβ1 + \sqrt{-5} = \alpha\beta with neither a unit, then N(α),N(β){2,3}N(\alpha), N(\beta) \in \{2, 3\}. But a2+5b2=2a^2 + 5b^2 = 2 and a2+5b2=3a^2 + 5b^2 = 3 have no solutions. So 1±51 \pm \sqrt{-5} are irreducible.

Now 23=6=(1+5)(15)2 \cdot 3 = 6 = (1 + \sqrt{-5})(1 - \sqrt{-5}) gives two distinct factorizations into Irreducibles (the factors are not associates since their norms are different: 4,94, 9 vs 6,66, 6). Therefore Z[5]\mathbb{Z}[\sqrt{-5}] is not a UFD. \blacksquare

TypeDefinitionExample
Euclidean domainHas division algorithm with δ\deltaZ\mathbb{Z}, F[x]F[x], Z[i]\mathbb{Z}[i]
PIDEvery ideal is principalZ\mathbb{Z}, Z[i]\mathbb{Z}[i]
UFDUnique factorisation into irreduciblesZ[x]\mathbb{Z}[x]
Integral domainNo zero divisorsAll of the above

FieldsEuclidean domainsPIDsUFDsIntegral domains\mathrm{Fields} \subset \mathrm{Euclidean\ domains} \subset \mathrm{PIDs} \subset \mathrm{UFDs} \subset \mathrm{Integral\ domains}

  • Assuming every PID is Euclidean. Z[1+192]\mathbb{Z}[\frac{1+\sqrt{-19}}{2}] is a PID but not Euclidean, showing the inclusion is strict.
  • Thinking Z[x]\mathbb{Z}[x] is a PID. The ideal (2,x)(2, x) is not principal because any generator would have to divide both 22 and xx, forcing it to be ±1\pm 1, which generates the whole ring.
  • Confusing irreducibles with primes in non-UFDs. In Z[5]\mathbb{Z}[\sqrt{-5}], 22 is irreducible but not prime: 2(1+5)(15)2 \mid (1+\sqrt{-5})(1-\sqrt{-5}) but 2(1±5)2 \nmid (1\pm\sqrt{-5}).
  • Forgetting that units and associates do not affect uniqueness of factorisation. The factorisation 6=23=(2)(3)6 = 2 \cdot 3 = (-2) \cdot (-3) is considered the same up to associates.
  • Number theory: The Euclidean algorithm in Z\mathbb{Z} computes gcds and is the foundation of modular arithmetic and RSA cryptography.
  • Coding theory: Polynomial rings over finite fields Fq[x]F_q[x] are Euclidean domains, enabling BCH and Reed-Solomon error-correcting codes.
  • Computer algebra: The Euclidean algorithm in Z[i]\mathbb{Z}[i] solves Diophantine equations and factors Gaussian integers, used in lattice-based cryptography.
  • Algebraic geometry: UFDs guarantee that irreducible algebraic varieties have well-defined coordinate rings, enabling unique factorisation of polynomial ideals.

Problem 1. Determine whether Z[2]\mathbb{Z}[\sqrt{2}] is a Euclidean domain.

Solution. The norm is N(a+b2)=a22b2N(a + b\sqrt{2}) = |a^2 - 2b^2|. For any α,βZ[2]\alpha, \beta \in \mathbb{Z}[\sqrt{2}] with β0\beta \neq 0, write α/β=u+v2\alpha/\beta = u + v\sqrt{2} with u,vQu,v \in \mathbb{Q}. Choose integers p,qp,q such that up1/2|u - p| \leq 1/2 and vq1/2|v - q| \leq 1/2. Then r=αβ(p+q2)r = \alpha - \beta(p + q\sqrt{2}) satisfies N(r)=(up)22(vq)2N(β)(1/4+2/4)N(β)=(3/4)N(β)<N(β)N(r) = |(u-p)^2 - 2(v-q)^2| \cdot N(\beta) \leq (1/4 + 2/4)N(\beta) = (3/4)N(\beta) < N(\beta). Thus Z[2]\mathbb{Z}[\sqrt{2}] is a Euclidean domain. \blacksquare

Problem 2. Is Z[x]\mathbb{Z}[x] a PID? Justify.

Solution. No. Consider the ideal I=(2,x)={2f(x)+xg(x):f,gZ[x]}I = (2, x) = \{2f(x) + xg(x) : f,g \in \mathbb{Z}[x]\}. If I=(h(x))I = (h(x)), then h(x)2h(x) \mid 2 and h(x)xh(x) \mid x, so h(x)h(x) divides any combination. Since h2h \mid 2, hh is constant: ±1\pm 1 or ±2\pm 2. If h=±2h = \pm 2, then x(2)x \in (2) is impossible since xx has coefficient 11 for xx. If h=±1h = \pm 1, then (h)=Z[x](h) = \mathbb{Z}[x], but II contains only polynomials with even constant term, so 1I1 \notin I. Thus II is not principal. \blacksquare