math.wiki
Abstract algebra › Rings in depth › Unique factorization domains
Sign in

Unique factorization domains

The integers break into primes in only one way, and so do the polynomials over a field. In which other rings does every element factor into irreducibles in essentially one way, and what goes wrong where it does not?

Where you are 4 in Rings in depth

Why this exists

The fundamental theorem of arithmetic says that every integer above 1 is a product of primes in only one way, up to order. Euclidean domains and principal ideal domains ended with the fact behind it: in a principal ideal domain, an irreducible element that divides a product divides one of the factors. Nothing forces this in general. In \displaystyle \mathbb{Z}[\sqrt{-5}],

\begin{aligned} 21 &= 3 \cdot 7 = (1 + 2\sqrt{-5})(1 - 2\sqrt{-5}) \\ &= (4 + \sqrt{-5})(4 - \sqrt{-5}), \end{aligned}

and none of the six factors splits further (Example 1). This lesson names the rings where factorization is unique, proves that every principal ideal domain is one, and goes past them: \mathbb{Z}[x] and F[x, y] are not principal ideal domains, yet they factor uniquely. The bridge is Gauss’s lemma, which compares D[x] with F[x] for F the field of fractions of D.

Definition

Throughout, D is an integral domain. Divisibility, associates and irreducible elements are as in Euclidean domains and principal ideal domains: p is irreducible if it is nonzero, not a unit, and every factorization p = bc has b or c a unit.

An element p \in D is prime if it is nonzero, not a unit, and whenever p \mid ab with a, b \in D, then p \mid a or p \mid b. Equivalently, \langle p \rangle is a nonzero prime ideal (Prime and maximal ideals).

D is a unique factorization domain if

  • (i) every nonzero element that is not a unit is a product p_1 p_2 \cdots p_r of irreducible elements, with r \ge 1 (a product of one is allowed), and
  • (ii) whenever p_1 \cdots p_r = q_1 \cdots q_s with every p_i and q_j irreducible, then r = s and there is a permutation \sigma of \{1, \dots, r\} with p_i and q_{\sigma(i)} associates for every i.

So factorization is unique up to order and units: 2 \cdot 3 and (-3)(-2) count as one factorization of 6. In a unique factorization domain the number r of irreducible factors of a depends only on a; call it the length of a, and give units length 0.

The field of fractions \operatorname{Frac}(D) is built from the pairs (a, b) with a, b \in D and b \ne 0, where (a, b) \sim (c, d) means ad = bc. The class of (a, b) is written \displaystyle \frac{a}{b} or a/b, and

\begin{aligned} \frac{a}{b} + \frac{c}{d} &= \frac{ad + bc}{bd}, \\ \frac{a}{b} \cdot \frac{c}{d} &= \frac{ac}{bd}. \end{aligned}

So \operatorname{Frac}(\mathbb{Z}) = \mathbb{Q}, and \operatorname{Frac}(F[x]) is the field F(x) of rational functions f/g with g \ne 0.

Let D be a unique factorization domain. A nonzero f \in D[x] is primitive if no irreducible element of D divides every coefficient of f. In \mathbb{Z}[x] this says the coefficients have greatest common divisor 1.

Intuition

Factoring an element is a walk down its divisors until only units lie below. In \mathbb{Z} every walk down from 21 passes through 3 and 7. In \displaystyle \mathbb{Z}[\sqrt{-5}], 21 has six irreducible divisors and three incompatible walks.

21
1
3
7
21
1
3
7
\displaystyle 1 + 2\sqrt{-5}
\displaystyle 1 - 2\sqrt{-5}
\displaystyle 4 + \sqrt{-5}
\displaystyle 4 - \sqrt{-5}
\mathbb{Z}
\displaystyle \mathbb{Z}[\sqrt{-5}]
Figure 1The divisors of 21, up to sign, in \mathbb{Z} and in \displaystyle \mathbb{Z}[\sqrt{-5}], with lines for divisibility. In \displaystyle \mathbb{Z}[\sqrt{-5}] all six middle elements are irreducible and each pair multiplies to 21, so 21 has three different factorizations (Example 1).

One property separates the pictures. In \displaystyle \mathbb{Z}[\sqrt{-5}], 3 divides \displaystyle 21 = (1 + 2\sqrt{-5})(1 - 2\sqrt{-5}) but neither factor: it is irreducible without being prime. Theorem 2 says this is the whole story: once factorizations exist, they are unique exactly when every irreducible element is prime.

For polynomials, 2x + 4 = 2(x + 2) is reducible in \mathbb{Z}[x] but not in \mathbb{Q}[x], where 2 is a unit. Gauss’s lemma says that, constants aside, the two rings agree.

Results

Throughout, D is an integral domain.

Theorem 1. Every prime element of D is irreducible.

Theorem 2 (the prime criterion). Suppose every nonzero nonunit of D is a product of irreducible elements. Then D is a unique factorization domain if and only if every irreducible element of D is prime.

Lemma 3 (chains of ideals). If D is a principal ideal domain and I_1 \subseteq I_2 \subseteq I_3 \subseteq \cdots are ideals of D, then there is an N with I_n = I_N for every n \ge N.

Theorem 4. Every principal ideal domain is a unique factorization domain.

Corollary 5. Every Euclidean domain is a unique factorization domain; in particular \mathbb{Z}, \mathbb{Z}[i], and F[x] for a field F are.

Theorem 6 (the field of fractions). The relation \sim is an equivalence relation, the sum and product above are well defined, and they make \operatorname{Frac}(D) a field into which a \mapsto a/1 is an injective ring homomorphism.

From here on a and a/1 are identified, so D \subseteq \operatorname{Frac}(D).

Theorem 7 (Gauss’s lemma). (a) Let p be a prime element of D and g, h \in D[x]. If p divides every coefficient of gh, then p divides every coefficient of g or every coefficient of h. (b) If D is a unique factorization domain, the product of two primitive polynomials in D[x] is primitive.

From here on D is a unique factorization domain and F = \operatorname{Frac}(D), so D[x] \subseteq F[x].

Lemma 8 (content). (a) Every nonzero f \in F[x] is f = c f_0 with c \in F nonzero and f_0 \in D[x] primitive. (b) If c f_0 = c' f_0' are two such, then c' = uc and f_0 = u f_0' for a unit u of D. (c) If f \in D[x], then c \in D.

The element c, fixed up to a unit of D, is the content of f.

Theorem 9 (from F[x] to D[x]). Let f \in D[x] have degree at least 1. (a) If f = gh with g, h \in F[x], there is a nonzero a \in F with ag and a^{-1}h in D[x], so f = (ag)(a^{-1}h) is a factorization in D[x] with the same degrees. (b) If f is primitive, then f is irreducible in D[x] if and only if it is irreducible in F[x].

Theorem 10. If D is a unique factorization domain, so is D[x]. Its irreducible elements are the irreducible elements of D and the primitive polynomials of positive degree that are irreducible in F[x].

Corollary 11. \mathbb{Z}[x] is a unique factorization domain, and so is D[x_1, \dots, x_n] for every unique factorization domain D and positive integer n; in particular F[x, y] is, for a field F.

Why it is true

Proof of Theorem 1. Let p be prime and p = bc; say p \mid b, so b = pt. Then p = ptc, and cancelling p \ne 0 gives tc = 1: c is a unit. \square

Proof of Theorem 2. Suppose every irreducible is prime; by induction, a prime that divides a product divides a factor. A product of s \ge 2 irreducibles is not irreducible, since in q_1 \cdot (q_2 \cdots q_s) neither factor is a unit (q_2 would divide 1). We prove (ii) by induction on r. If r = 1, then s = 1. If r \ge 2, then s \ge 2, and p_1 \mid q_j for some j, say q_j = p_1 v; as q_j is irreducible and p_1 is not a unit, v is a unit. Cancel p_1 and absorb v into another q_k, still irreducible; the induction hypothesis matches what is left, with r - 1 = s - 1.

Conversely, let D be a unique factorization domain, p irreducible, and ab = pc. If a = 0, p \mid a; if a is a unit, b = a^{-1}pc; likewise for b. Otherwise, factoring a, b and c (keeping pc whole if c is a unit) gives two factorizations of ab, and by (ii) p is an associate of an irreducible factor of a or of b. \square

Proof of Lemma 3. Let I be the union of the I_n. Two elements of I lie in a common I_n, the ideals being nested, so their difference is in I; and I absorbs multiplication because each I_n does. So I is an ideal, I = \langle a \rangle, and a \in I_N for some N. For n \ge N, \langle a \rangle \subseteq I_N \subseteq I_n \subseteq I = \langle a \rangle. \square

Proof of Theorem 4. Existence. Suppose a nonzero nonunit a_0 is not a product of irreducibles. Then a_0 = a_1 b_1 with neither factor a unit, and one factor, say a_1, is not a product of irreducibles either. If a_1 = u a_0, cancelling in a_0 = u a_0 b_1 makes b_1 a unit; so a_0, a_1 are not associates and \langle a_0 \rangle \subsetneq \langle a_1 \rangle (Euclidean domains and principal ideal domains, Lemma 3). Repeating gives a strictly increasing chain of ideals, which Lemma 3 forbids. Uniqueness. In a principal ideal domain every irreducible is prime (Euclidean domains and principal ideal domains, Theorem 6), so Theorem 2 applies. \square

Proof of Corollary 5. Every Euclidean domain is a principal ideal domain (Euclidean domains and principal ideal domains, Theorem 2), and the three rings are Euclidean (there, Theorem 1). \square

Proof of Theorem 6. Reflexivity and symmetry are clear. If ad = bc and cf = de, then adf = bcf = bde, so d(af - be) = 0 and af = be. If ab' = a'b, then

(ad + bc)b'd = a'bd^2 + bb'cd = (a'd + b'c)bd

and (ac)(b'd) = (a'c)(bd), so neither operation depends on the pair chosen for the first fraction, nor, by symmetry, for the second; and bd \ne 0 as D is a domain. As (ea)/(eb) = a/b for e \ne 0, finitely many fractions can be put over one denominator, and then each ring axiom is the same axiom in D. The zero is 0/1, the one is 1/1 \ne 0/1, and a nonzero a/b has inverse b/a. Finally a \mapsto a/1 respects sums, products and 1, and a/1 = b/1 forces a = b. \square

Proof of Theorem 7. (a) As p is prime, \langle p \rangle is a prime ideal, so D/\langle p \rangle is an integral domain (Prime and maximal ideals, Theorem 1), and so is (D/\langle p \rangle)[x] (Polynomial rings and the division algorithm, Theorem 2). Reducing each coefficient is a ring homomorphism \pi : D[x] \to (D/\langle p \rangle)[x], since sums and products of polynomials are given by the same formulas in both rings, and \pi(g) = 0 says that p divides every coefficient of g. If \pi(g)\pi(h) = \pi(gh) = 0, then \pi(g) = 0 or \pi(h) = 0. (b) If gh is not primitive, an irreducible p divides all its coefficients; p is prime by Theorem 2, so by (a) g or h is not primitive. \square

Proof of Lemma 8. (a) If b is the product of the denominators of the coefficients of f, then g = bf \in D[x]. Fix a nonzero coefficient a of g. While g is not primitive, an irreducible divides every coefficient and can be divided out, lowering the length of the coefficient in the place of a; so after finitely many steps g = d g_0 with d \in D and g_0 primitive, and f = (d/b) g_0.

(b) Clearing denominators turns c f_0 = c' f_0' into a f_0 = a' f_0' with a, a' \in D nonzero and a'/a = c'/c. An irreducible p \mid a divides a' times each coefficient of f_0'; it is prime (Theorem 2) and does not divide them all, so p \mid a'. Cancel p from both and repeat until a is a unit. Then an irreducible dividing a' would divide every coefficient of f_0, so a' is a unit too, and u = a'/a works.

(c) By (a) with b = 1, f = d g_0 with d \in D, and by (b), c is a unit times d. \square

Proof of Theorem 9. (a) By Lemma 8, g = c_1 g_0 and h = c_2 h_0 with g_0, h_0 primitive, so f = (c_1 c_2) g_0 h_0 with g_0 h_0 primitive (Theorem 7(b)), and c_1 c_2 \in D by Lemma 8(c). Take a = c_2: then ag = c_1 c_2 g_0 and a^{-1}h = h_0 lie in D[x].

(b) The units of D[x] are those of D (Polynomial rings and the division algorithm, Theorem 2); those of F[x] are the nonzero constants. Let f be irreducible in F[x] and f = gh in D[x]. Degrees add and cannot both be positive, so one factor is a constant d \in D dividing every coefficient of f; by primitivity d has no irreducible factor, so it is a unit. Conversely, if f is irreducible in D[x] and f = gh in F[x] with both degrees positive, (a) gives a factorization in D[x] into two polynomials of positive degree, neither a unit. \square

Proof of Theorem 10. An irreducible of D stays irreducible in D[x], as in a factorization of a constant both degrees are 0. A nonconstant irreducible f of D[x] is primitive, since f = p f_1 with p irreducible in D would split it into nonunits. With Theorem 9(b), the irreducibles of D[x] are the ones listed.

Existence. A nonzero nonunit constant factors in D. For \deg f \ge 1, write f = d f_0 (Lemma 8(c)) and f_0 = g_1 \cdots g_k with each g_i irreducible in F[x] (Corollary 5), and g_i = c_i h_i with h_i primitive, so irreducible in D[x] by Theorem 9(b). As h_1 \cdots h_k is primitive (Theorem 7(b)), Lemma 8(b) makes u = c_1 \cdots c_k a unit of D, and f = (du) h_1 \cdots h_k, with du factored in D or, if a unit, absorbed into h_1.

Uniqueness, by Theorem 2. Let f be irreducible in D[x] with f \mid gh. If f is a constant, it is prime in D (Theorem 2), and Theorem 7(a) gives f \mid g or f \mid h. Otherwise f is irreducible, so prime, in F[x] (Euclidean domains and principal ideal domains, Theorem 6); say g = fk with k \in F[x], and g \ne 0. With k = c k_0 (Lemma 8), g = c(f k_0) where f k_0 is primitive (Theorem 7(b)), so c \in D by Lemma 8(c), k \in D[x], and f \mid g in D[x]. \square

Proof of Corollary 11. \mathbb{Z} is a unique factorization domain (Corollary 5), so Theorem 10 applies. For n \ge 2, D[x_1, \dots, x_n] = D[x_1, \dots, x_{n-1}][x_n], and induction on n applies Theorem 10 once per variable. \square

Worked examples

Example 1

Three factorizations of 21

Show that the six factors 3, 7, \displaystyle 1 \pm 2\sqrt{-5}, \displaystyle 4 \pm \sqrt{-5} are irreducible and pairwise not associates in \displaystyle \mathbb{Z}[\sqrt{-5}], and that 3 is not prime.

Solution

Write \displaystyle w = \sqrt{-5} and N(a + bw) = a^2 + 5b^2, which is multiplicative (Euclidean domains and principal ideal domains). If uv = 1 then N(u)N(v) = 1, and a^2 + 5b^2 = 1 only for a = \pm 1, b = 0: the units are \pm 1. The norms are

\begin{aligned} N(3) &= 9, \quad N(7) = 49, \\ N(1 \pm 2w) &= N(4 \pm w) = 21. \end{aligned}

Splitting any of these into two nonunits would need a factor of norm 3 or 7. But a^2 + 5b^2 \in \{3, 7\} has no integer solution: b = 0 needs a^2 \in \{3, 7\}, b = \pm 1 needs a^2 \in \{-2, 2\}, and |b| \ge 2 gives 5b^2 \ge 20. So all six are irreducible. Associates differ by a sign, and no two of the six do, so the three factorizations are different. Finally 3 divides 21 = (1 + 2w)(1 - 2w) but neither factor, since \displaystyle \frac13 \pm \frac23 w is not in the ring: 3 is irreducible and not prime, as Theorem 2 requires. Norms 63 and 147 would leave a cofactor of norm 7 or 3, so testing the elements of norm 9 (\pm 3, \pm 2 \pm w) and 49 (\pm 7, \pm 2 \pm 3w) shows the figure lists every divisor of 21 up to sign.

Example 2

Isabel’s lockbox

Isabel’s lockbox opens when the solver enters a factorization into irreducibles of \mathbb{Z}[i] of the Gaussian integer z on its dial. As 5 = (2 + i)(2 - i) = (1 + 2i)(1 - 2i), the lock needs a fixed form that every z has in exactly one way. Find one, and the entry for z = 4 + 22i.

Solution

Call a + bi normalized if a > 0 and b \ge 0. Multiplying by i sends a + bi to -b + ai, a quarter turn carrying each of \{a > 0, b \ge 0\}, \{a \le 0, b > 0\}, \{a < 0, b \le 0\}, \{a \ge 0, b < 0\} onto the next. These split the nonzero Gaussian integers, so exactly one of z, iz, -z, -iz is normalized. By Corollary 5 a nonzero nonunit z is u\,p_1 \cdots p_r with u \in \{\pm 1, \pm i\} and each p_k a normalized irreducible; by (ii) the list is fixed up to order, normalized associates being equal, and then u is fixed by cancellation.

For z = 4 + 22i, N(z) = 500 = 2^2 \cdot 5^3, and an element of prime norm is irreducible. First 4 + 22i = 2(2 + 11i) with 2 = -i(1 + i)^2. The normalized elements of norm 5 are 2 + i and 1 + 2i, and

\frac{2 + 11i}{2 + i} = \frac{(2 + 11i)(2 - i)}{5} = 3 + 4i,

with (3 + 4i)/(2 + i) = 2 + i likewise, so

4 + 22i = -i\,(1 + i)^2 (2 + i)^3.

The entry is u = -i with 1 + i twice and 2 + i three times.

Example 3

Factoring in \mathbb{Z}[x] and in \mathbb{Q}[x]

Factor f = 12x^3 + 10x^2 + 6x - 12 into irreducibles in \mathbb{Z}[x] and in \mathbb{Q}[x].

Solution

Among the candidates of the The rational root theorem, \displaystyle x = \frac23 works: \displaystyle f(\frac23) = \frac{32}{9} + \frac{40}{9} + 4 - 12 = 0. So in \mathbb{Q}[x], by the The remainder and factor theorems,

f = \left(x - \tfrac23\right)(12x^2 + 18x + 18).

Theorem 9(a) with a = 3 clears the denominator: f = (3x - 2)(4x^2 + 6x + 6) in \mathbb{Z}[x]. The coefficients 4, 6, 6 share the irreducible 2 of \mathbb{Z}, so

f = 2\,(3x - 2)(2x^2 + 3x + 3).

Each factor is irreducible in \mathbb{Z}[x] (Theorem 10): 2 is irreducible in \mathbb{Z}; 3x - 2 is primitive of degree 1; and 2x^2 + 3x + 3 is primitive with discriminant 9 - 24 = -15 < 0, so it has no rational root and, being quadratic, is irreducible in \mathbb{Q}[x]. In \mathbb{Q}[x] the content 2 is a unit, leaving the two irreducible factors 3x - 2 and 2x^2 + 3x + 3.

Example 4

An irreducible polynomial in two variables

Prove that x^2 - y^3 is irreducible in \mathbb{Q}[x, y].

Solution

Proof. View \mathbb{Q}[x, y] as D[x] with D = \mathbb{Q}[y], a unique factorization domain (Corollary 5) whose field of fractions is \mathbb{Q}(y). As a polynomial in x, f = x^2 - y^3 has leading coefficient 1, so it is primitive, and by Theorem 9(b) it is enough that f is irreducible in \mathbb{Q}(y)[x]. Of degree 2, it factors there only with a linear factor, that is, only if r^2 = y^3 for some r \in \mathbb{Q}(y) (The remainder and factor theorems). Write r = p/q with p, q \in \mathbb{Q}[y], q \ne 0, and cancel common irreducible factors until none is left. Then p^2 = y^3 q^2. An irreducible factor of q would divide p^2, hence p, being prime (Theorem 2); so q is a nonzero constant, and comparing degrees gives 2\deg p = 3, which is impossible. \square

Boundaries

  • Strictly weaker than principal. \mathbb{Z}[x] and F[x, y] factor uniquely (Corollary 11) but are not principal ideal domains (Euclidean domains and principal ideal domains, Theorem 7).
  • Existence can fail. In R = \{f \in \mathbb{Q}[x] : f(0) \in \mathbb{Z}\}, whose units are \pm 1, every cx equals \displaystyle 2 \cdot \frac{c}{2}x, so no cx is irreducible, and as degrees add, x is never a product of irreducibles; the principal ideals generated by \displaystyle x, \frac{x}{2}, \frac{x}{4}, \dots increase forever. In \displaystyle \mathbb{Z}[\sqrt{-5}] factorizations exist (Exercise 13) but are not unique.
  • Subrings do not inherit it. \mathbb{Z}[2i] lies inside \mathbb{Z}[i] and does not factor uniquely (Exercise 5).

Where this goes next

Irreducible polynomials and Eisenstein's criterion leans on Theorem 9: to show a polynomial with integer coefficients is irreducible over \mathbb{Q}, it is enough to rule out factorizations in \mathbb{Z}[x], which reduction mod a prime and Eisenstein’s criterion do. An irreducible f \in F[x] then gives the field F[x]/\langle f \rangle of Field extensions and degree.

Exercises

Factoring in \mathbb{Z}[i]

  1. Andre sets Isabel’s lockbox to 7 + 17i. Give the entry it accepts, in the form of Example 2.
  2. Factor 65 into normalized irreducibles of \mathbb{Z}[i], with a unit in front.
  3. How many divisors does 65 have in \mathbb{Z}[i], counted up to associates? List the normalized divisors of norm 65.

Rings that do not factor uniquely

  1. In \displaystyle \mathbb{Z}[\sqrt{-5}], show that \displaystyle 14 = 2 \cdot 7 = (3 + \sqrt{-5})(3 - \sqrt{-5}) are two different factorizations into irreducibles, and that 2 is not prime.
  2. Let \mathbb{Z}[2i] = \{a + 2bi : a, b \in \mathbb{Z}\}. Show that 4 = 2 \cdot 2 = (2i)(-2i) are two different factorizations into irreducibles of \mathbb{Z}[2i].

Polynomials and fractions

  1. Factor 6x^3 + 3x^2 - 18x - 9 into irreducibles in \mathbb{Z}[x] and in \mathbb{Q}[x].
  2. Is 3x^2 + 2 irreducible in \mathbb{Z}[x]? In \mathbb{Q}[x]? And 4x^2 + 2?
  3. Write \displaystyle f = \frac34 x^2 - \frac12 x + \frac56 as c f_0 with c \in \mathbb{Q} and f_0 \in \mathbb{Z}[x] primitive. How many such pairs (c, f_0) are there?

Proofs

  1. Prove that x^2 + y^2 - 1 is irreducible in \mathbb{Q}[x, y].
  2. Let D be a unique factorization domain. Prove that any two nonzero a, b \in D have a greatest common divisor.
  3. Let D be a unique factorization domain, F = \operatorname{Frac}(D) and a \in D. Prove that if a = r^2 with r \in F, then r \in D.
  4. Prove that every nonzero prime ideal of a unique factorization domain contains a prime element.
  5. Let N : D \setminus \{0\} \to \mathbb{Z}_{\ge 1} satisfy N(ab) = N(a)N(b), with N(u) = 1 exactly when u is a unit. Prove that every nonzero nonunit of D is a product of irreducibles, and deduce it for \displaystyle \mathbb{Z}[\sqrt{-5}].
Answers
  1. N(7 + 17i) = 338 = 2 \cdot 13^2, and (7 + 17i)(1 - i)/2 = 12 + 5i. The normalized elements of norm 13 are 3 + 2i and 2 + 3i, and (2 + 3i)^2 = -5 + 12i, so 12 + 5i = -i(2 + 3i)^2. So 7 + 17i = -i\,(1 + i)(2 + 3i)^2.
  2. 65 = (2 + i)(2 - i)(3 + 2i)(3 - 2i), all of prime norm. As 2 - i = -i(1 + 2i) and 3 - 2i = -i(2 + 3i), 65 = -(1 + 2i)(2 + i)(2 + 3i)(3 + 2i).
  3. If de = 65, factorizations of d and e together give one of 65, so by uniqueness d is a unit times a product of some of the four irreducibles of Exercise 2: 2^4 = 16 divisors up to associates. Norm 65 needs one factor of norm 5 and one of norm 13; the four products, normalized, are 1 + 8i, 4 + 7i, 7 + 4i, 8 + i.
  4. The norms are 4, 49, 14, 14, so a split into nonunits needs a factor of norm 2 or 7, and a^2 + 5b^2 \in \{2, 7\} has no integer solution: all four are irreducible, and no two differ by a sign. And 2 divides \displaystyle (3 + \sqrt{-5})(3 - \sqrt{-5}) but neither factor, as \displaystyle \frac32 \pm \frac12\sqrt{-5} is not in the ring.
  5. N(a + 2bi) = a^2 + 4b^2 is multiplicative, so the units have norm 1 and are \pm 1. N(2) = N(\pm 2i) = 4 and a^2 + 4b^2 = 2 has no solution, so 2 and \pm 2i are irreducible, and (2i)(-2i) = 4. As \pm 2i \ne \pm 2, the factorizations differ.
  6. The content is 3: f = 3(2x + 1)(x^2 - 3). The candidates \pm 1, \pm 3 give x^2 - 3 = -2 or 6, so x^2 - 3 has no rational root, is irreducible in \mathbb{Q}[x], and, being primitive, in \mathbb{Z}[x]. In \mathbb{Z}[x]: 3 \cdot (2x + 1)(x^2 - 3). In \mathbb{Q}[x]: (2x + 1)(x^2 - 3), the 3 being a unit.
  7. 3x^2 + 2 is positive on \mathbb{R}, so a quadratic with no rational root: irreducible in \mathbb{Q}[x], and, being primitive, in \mathbb{Z}[x] (Theorem 9(b)). 4x^2 + 2 is irreducible in \mathbb{Q}[x] for the same reason, but 4x^2 + 2 = 2(2x^2 + 1) is a product of two nonunits of \mathbb{Z}[x].
  8. 12f = 9x^2 - 6x + 10, with coefficients of greatest common divisor 1, so \displaystyle c = \frac{1}{12} and f_0 = 9x^2 - 6x + 10. By Lemma 8(b) the only other pair is \displaystyle c = -\frac{1}{12}, f_0 = -9x^2 + 6x - 10: two pairs.
  9. Proof. As in Example 4, with D = \mathbb{Q}[y] and f = x^2 - (1 - y^2) monic in x, it suffices that p^2 = (1 - y^2)q^2 has no solution with p/q in lowest terms; as there, q is a nonzero constant. Then p^2 = q^2(1 - y)(1 + y): the irreducible 1 - y occurs an even number of times on the left and once on the right, against uniqueness in \mathbb{Q}[y]. \square
  10. Proof. Choose pairwise nonassociate irreducibles p_1, \dots, p_k such that every irreducible factor of a or b is an associate of one of them, and write a = u \prod p_i^{e_i}, b = v \prod p_i^{f_i} with u, v units. Then d = \prod p_i^{\min(e_i, f_i)} divides a and b. If c \mid a and c \mid b, uniqueness applied to c \cdot (a/c) = a gives c = w \prod p_i^{g_i} with w a unit and g_i \le e_i; likewise g_i \le f_i, so c \mid d. \square
  11. Proof. Write r = p/q with p, q \in D, q \ne 0, and cancel common irreducible factors until none is left. Then p^2 = aq^2. If q were not a unit, an irreducible \pi \mid q would divide p^2, hence p, as \pi is prime (Theorem 2). So q is a unit and r = pq^{-1} \in D. \square
  12. Proof. Let P be a nonzero prime ideal and a \in P nonzero. As P is proper, a is not a unit, so a = p_1 \cdots p_r with each p_i irreducible. As P is prime, some p_i \in P, and p_i is prime by Theorem 2. \square
  13. Proof. A nonzero nonunit a has N(a) \ge 2; use Strong induction and well-ordering on N(a). If a is not irreducible, a = bc with b, c nonunits, so N(b), N(c) \ge 2 and both are less than N(a) = N(b)N(c); by the hypothesis b and c are products of irreducibles, hence so is a. On \displaystyle \mathbb{Z}[\sqrt{-5}], \displaystyle N(a + b\sqrt{-5}) = a^2 + 5b^2 is multiplicative, positive on nonzero elements, and 1 exactly at the units \pm 1 (Example 1). \square
math.wiki · updated 2026-11-17 · CC-BY-SA-4.0