Factorization in Polynomial Rings

These notes are a summary of some of the important points on divisibility in polynomial rings from 17 and 18 of Gallians Contemporary Abstract Algebra. Most of the important results about the structure of F [X] follow in one way oranother from one key fact. Theorem 1 (Division Algorithm) If F is a eld, f, g F [X] and f = 0, there are q, r F [X] such that g = qf + r and either r = 0 or degr < degf.

PIDs
Denition 2 A principal ideal domain (PID) is an integral domain D in which every ideal has the form a = {ra : r D} for some a D. For example, Z is a PID, since every ideal is of the form nZ. Theorem 3 If F is a eld, then F [X] is a PID. Proof We know that F [X] is an integral domain. Let I be an ideal. If I = {0}, then I = 0 . Suppose I = {0}. Let g I be a nonzero polynomial of minimal degree. We claim that I = g . Suppose f I. By the division algorithm, there are nonzero polynomials q and r such that f = qg + r and either r = 0 or deg(r) < degg. Since f, g I, r = f qg I. Since g is of minimal degree in I, we must have r = 0. Thus f = qg g .

If I = {0} and f I is of minimal degree, then f is a minimal polynomial of I and I = f . Denition 4 Let D be an integral domain. If a D is nonzero and not a unit, we say that a is irreducible if whenever b, c D and a = bc then b is a unit or c is a unit. 1

Otherwise we say that a is reducible. For example, in Z, n is irreducible if and only if n is prime. Suppose F is a eld and f F [X]. If f has degree 1, then f is irreducible. If f has degree 2 or 3, then f is irreducible if f has no zero in F . [If f = gh where neither g nor h is a unit, then one of g or h has degree 1 and has a root.] Here are some examples X 2 2 is irreducible in Q[X] but reducible in R[X] since X 2 2 = (X 2)(X + 2). X 2 + 1 is irreducible in R[X], but reducible in C[X] since X 2 + 1 = (X + i)(X i). 2X + 2 is irreducible in R[X], but reducible in Z[X] since (2X + 2) = 2(X + 2) and 2 is a unit in R, but a nonunit in Z. This example shows we have to be more careful in D[X] when D is not a eld. We recall some basic denitions from 14.

Denition 5 Let R be a commutative ring and let I = R be an ideal of R. I is a prime ideal if whenever a, b R and ab I, then a I or b I. I is a maximal ideal if whenever J is an ideal and I J R, then J = I or J = R.

If R is a commutative ring with unity, then every maximal ideal is prime, but prime ideals need not be maximal. For example, in R[X, Y ]. The ideal X is prime (since R[X, Y ]/ X R[Y ] an integral domain), but not max= imal since X X, Y R[X, Y ]. We remind you of one key fact about prime and maximal ideals. Theorem 6 If R is a commutative ring with unity and I is an ideal then: i) I is prime if and only if R/I is an integral domain; ii) I is maximal if and only if R/I is a eld. Proposition 7 If D is an integral domain, and a is prime, then a is irredcuible. Proof Suppose a = bc. We must show that either b or c is a unit. Since bc a and a is a prime ideal, either b a or c a . Suppose b a . Then b = ad for some d D. Thus a = bc = adc. Since D is an integral domain, 1 = dc. Thus c is a unit. A similar argument shows that if c a , then b is a unit. Thus a is irreducible. 2

The converse is true in F [X] for F a eld. Indeed, if a is irreducible, then a is maximal. The proof works just as well for all PIDs. Theorem 8 Let D be a PID and a D. The following are equivalent: i) a is irreducible; ii) a is maximal; iii) a is prime. Proof ii) iii) In any commuative ring with unity, every maximal ideal is prime. iii) i) This is Proposition 7 i) ii) Suppose a is irreducible. Since a is not a unit a = D. Let J be an ideal such that a J D. We must show that J = I or J = D. Since D is a PID, there is b D such that J = b . Since a J, a = bc for some c D. Since a is irreducible, either b or c is a unit. case 1: b is a unit. Then b has an inverse b1 D. If d D, then d = db1 b J. Thus J = D. case 2: c is a unit. Since a = bc, b = c1 a a . Since b a , J = a . Corollary 9 If F is a eld and p F [X] is irreducible, then F [X]/ p is a eld. Proof Since p is irreducible, p is maximal and F [X]/ p > is a eld.

UFDs
Suppose F is a eld and f F [X] is reducible. Then we can factor f = gh where f and g both have lower degree. If either g or h is reducible, then we can factor again. For example if f = 2X 4 7X 3 + 8X 2 3X we see that f = X(2X 3 7X 2 + 8X 3) = X(X 1)(2X 2 5X + 3) = X(X 1)(X 1)(2X 3) 3

Proposition 10 If F is a eld and f F [X] is nonzero and not a unit, then for some n there are irreducible polynomials g1 , . . . , gn F [X] such that f = g 1 g2 g n . This is similar to the fact that in the natural numbers N we can factor every element as a product of primes. In N the prime factorization is unique. Is this true in F [X]? Not quite. For example (X 2 1) = (X 1)(X + 1) = Of course even in Z we have 6 = 2(3) = (2)(3). Indeed if we have one irreducible factorization, then by multiplying by suitable units we can always get another. The next denition gives us the right way to state uniqueness of factorization. Denition 11 If D is a domain, we say that a and b are associates if there is a unit u D such that a = ub. X 1 2 2 (2X + 2).

Note that if u is a unit and a = ub, then b = u1 a. Thus being associates is a symmetric relation. Denition 12 If D is a domain, we say that D is a Unique Factorization Domain (or UFD) if: i) if f D is nonzero and not a unit, then there are irreducible elements g1 , . . . , gn D such that f = g1 g2 gn , and ii) if p1 , . . . , pn , q1 , . . . , qm D are irreducible, and p1 pn = q1 qm , then n = m and there is Sn such that pi is an associate of q(i) for i = 1, . . . , n. In other words if f = p1 pn = q1 qm are two factorizations of f into irreducible factors, then n = m and we can renumber the qs so that pi and qi are associates for all i. Suppose F is a eld and f = p 1 p n = q1 q m are irreducible factorizations of f in F [X]. Since p1 is irreducible, p1 is a prime ideal. We need one easy lemma. 4

Lemma 13 If D is an integral domain, I D is a prime ideal, a1 , . . . , an D and a1 a2 an I, then some ai I. Proof We prove this by induction. If n = 2 this is the denition of a prime ideal. If n > 2 and a1 (a2 an ) I, then either a1 I and we are done, or (a2 an ) I. In the later case, by induction aj I for some j = 2, . . . , n. Since q1 qm = p1 (p2 pn ), there is an i such that qi p1 . Thus qi = up1 for some i, since qi is irreducible, u must be a unit. Thus p1 pn = q1 qm = q1 qi1 (up1 )qi+1 qm . Since D is an integral domain, p2 pn = uq1 qi1 qi+1 qm . We have gotten rid of one irreducible from each side, but at the cost of introducing a unit. This leads us to the following lemma which gives the right induction. Lemma 14 Suppose F is a eld, p1 , . . . , pn , q1 , . . . , qm F [X] are irreducible, u is a unit, and p1 pn = uq1 qm . Then n = m and we can renumber the qs so that pi and qi are associates for all i. Proof We prove this by induction on n. Suppose n = 1. Then p, q1 , . . . , qm are irreducible, u is a unit and p = uq1 qm . Since uq1 . . . qm p and p is a prime ideal, there is qi such that qi p . Then there is w p such that qi = wp. Since qi is irreducible and p is not a unit, w is a unit. Thus p = uq1 qi1 (wp)qi+1 qm and, since F [X] is an integral domain. 1 = uwq1 qi1 qi+1 qm . Since no qi is a unit, we must have m = 1 and p = uq1 . Thus p and q are associates. Suppose n > 1. The begining of the argument is similar. Since uq1 qm p1 , there is a unit w and a qi such that qi = wp1 . Then p1 pn = uq1 qi1 (wp)qi+1 qm 5

and, since F [X] is an integral domain. p2 pn = uwq1 qi1 qi+1 qm . By induction, n 1 = m 1 and we can renumber the qs as so that pi and qi are associates. Putting together Lemma 10 Lemma 14 we prove that polynomial rings are UFDs. Theorem 15 If F is a eld, then F [X] is a unique factorization domain.

Two Important Theorems

f (X, Y ) =
i=0 j=0

ai,j X i Y j D[X, Y ].

For j = 0, . . . , m let gj (X) D[X] be the polynomial

gj (X) =
i=0

ai,j X i .

Then f (X, Y ) =

j=1

gj (X)Y j D[X][Y ].

Similarly if f D[X][Y ], by multiplying out we get a polynomial in D[X, Y ]. Similarly we can identify D[X1 , . . . , Xn ] = D[X1 ] . . . [Xn ]. This allows us to inductively apply Theorem 17. 6

Corollary 18 i) The polynomial ring Z[X1 , . . . , Xn ] is a unique factorization domain. ii) If F is a eld, then the polynomial ring F [X1 , . . . , Xn ] is a unique factorization domain. Proof Since Z and F [X1 ] are unique factorization domains, Theorem 17 and induction tell us that Z[X1 , . . . , Xn ] and F [X1 , . . . , Xn ].