7

Factorisation in Q[X]

In this section we deal with some very concrete matters concerning the factorisation of polynomials with rational coecients. Any such polynomial is a rational multiple of a polynomial with integer coecients, and so we may concentrate on the integral associate. For such polynomials we have various techniques to help us, in particular we can work modulo p for various primes; or we can use other arithmetic tricks. We describe some of these, and we prove the important theorem which reconciles factorisation of polynomials over Q and over Z.

7.1

Factorisation in Z[X] and Zp [X]

We begin with a general lemma. Lemma 7.1.1. Let R and S be rings, and let f : R S be a homomorphism, f : a a. Then : R[X] S[X] given by : N ak X k N ak X k is a homomorphism with k=0 k=0 image f (R)[X] and kernel (ker f )[X]. Proof. In view of how we add and multiply polynomials all we need to check is that ak + bk = ak + bk and
r+s=n

ar bs =
r+s=n

ar bs .

The remarks on kernel and image are clear. We can use what we have learned about rings and homomorphisms to give a useful practical test for irreducibility1 in the ring Z[X]. Proposition 7.1.2 (Eisensteins criterion). Let f (X) = N ck X k be a polynomial in Z[X], k=0 and let p Z be a prime. Suppose that p|ck for all k = 0, 1, . . . , N 1, p |cN and p2 |c0 . Then f (X) has no factor of smaller degree in Z[X]. Proof. Suppose that we had that f (X) = g(X)h(X), with g(X) = N1 ak X k and h(X) = k=0 N2 bk X k and N1 , N2 > 1. Extending the natural homomorphism Z Zp to the polyk=0 nomial rings we would then have that (g(X)(h(X)) = (f (X)) = cN X N = 0, so that (g(X)) = aX N1 and h(X) = bX N2 by the unique factorisation in the Euclidean domain Zp [X]. Hence we have that a0 = b0 = 0; that is p|a0 and p|b0 , yielding p2 |a0 b0 = c0 contrary to hypothesis. There are many examples of this. For instance, (X 2 +125)2 (X 3 +25)4 +5 has only factors of degree 16 in Z[X] and so, being monic, is irreducible. p More famous are the cyclotomic polynomials or prime order. Let p (X) := X 1 ; then X1 p has no factor of degree k with 1 < k < (p 1). We cant use Eisenstein directly; but instead we apply the result to p (X + 1); the details are an exercise.
1

Well, almost irreducibility: see later.

27

7.2

Factorization in Z[X] and Q[X]

Although we often work with polynomials with integral coecients, we are really interested in whether these factorise in the ring Q[X]. It is clear that in a trivial way a polynomial may be a product of two irreducibles in Z[X] but of only one irreducible in Q[X]take for example f (X) = 2X. We can easily check that in the large ring Q[X] this is irreducible; but in the small ring it is the product of two irreducibles 2 and X. What we want to see is that essentially nothing more complicated than this is possible. Theorem 7.2.1. Let 0 = h Z[X]. Suppose that h = f g with f, g Q[X]. Then there exist f , g Z[X] with deg f = deg f and deg g = deg g , such that h = f g . Proof. Write f = M ak X k , g = N bk X k and let ck := r+s=k ar bs be the coecients k=0 k=0 of h. We may multiply f by the least common multiple of the denominators of its coecients, and then divide the resulting polynomial by the highest common factor of its (integral!) coecients. In that way we get f = f , where Q and f is a polynomial of the same degree as f , and has integral coecients whose highest common factor is 1. We can nd corresponding Q and g Z[X] with g = . g We can nd corresponding Zthere are no denominators to clear so is integral and h Z[X] with h = h. Then we have, after re-arranging the rationals, n h(X) = m f (X) g (X), for some m, n Z.

(The Xs are inserted into the polynomials for clarity.) How are m and n related? Well n is the highest common factor of the coecients of the polynomial on the left hand side. But the highest common factor of the coecients of the polynomial on the right hand side is m times the highest common factor of the coecients of f (X) g (X). In a moment we will prove that the latter highest common factor is 1. So m = n and we have that h(X) = f (X) g (X), which we multiply by Z to get h(X) = h(X) = f (X) g (X). Dene f := f and g := g and we are done. The missing lemma is so important it deserves a separate subsection.

7.3

Gausss Lemma

We begin with some pieces of notation. Denition 1. Let 0 = f Z[X]. Then by the content of f , denoted by c(f ), we mean the highest common factor of the coecients of f . If the content of f is 1 then we say f is primitive. Proposition 7.3.1 (Gausss Lemma). Let f, g Z[X] and suppose that c(f ) = c(g) = 1. Then c(f g) = 1. 28

Proof. If false let the prime p divide all the coecients of the product. Extend the natural homomorphism Z Zp to a homomorphism of the polynomial rings in the standard way. Then 0 = (f g) = (f ) (g) and so, since Zp [X] is an integral domain, either (f ) = 0 or (g) = 0. But that implies p divides every coecient of f or every coecient of g, a contradiction. Corollary 7.3.2. Let f, g Z[X]; then c(f g) = c(f )c(g). Proof. Trivial.

7.4

Z[X] is a UFD

We can use Gausss Lemma and the techniques of the previous sections to prove that in the ring Z[X] every polynomial factorises uniquely into a product of primes in Z times a product of primitive polynomials each irreducible in Q[X].

7.5

An algorithm for factorisation in Q[X]

How can we factorise a polynomial f Q[X]? As weve seen we may assume that f Z[X] and carry out our factorisations over Z. Here are a couple of trivial starting points: Lemma 7.5.1. Suppose that n Z. If f (n) = 0 then X n|f . Proof. Remainder Theorem and Gausss Lemma. Lemma 7.5.2. Suppose that g|f in Z[X]; and that n Z. Then g(n)|f (n). Proof. Trivial. With these in mind we search for factors of f in Z[X] of degree d = 1, 2, . . . , deg f 1. Of course whenever we nd a factor we divide it out and then look recursively at two factors we have found. The following lemma controls the possible values of a factor of degree d may take. Lemma 7.5.3. Let a0 , a1 , a2 , . . . , ad Z be distinct integers, and suppose that f (ak ) = 0 for each k. Then there are only nitely many (d + 1)-tuples (b0 , b1 , . . . , bd ) such that bk |f (ak ) for each k. Proof. This is essentially the fact that Z is a UFD. The following lemma allows us to reconstruct a polynomial of degree d from its values at (d + 1) places. Lemma 7.5.4. Let a0 , a1 , a2 , . . . , ad Z be distinct integers, and let b0 , b1 , . . . , bd Q. There is exactly one polynomial g Q[X] of degree at most d such that g(ak ) = bk for all k.

29

Proof. This is just the Chinese Remainder Theorem. To be absolutely specic we could, as Lagrange did, take d X aj bk g(X) = . ak aj
k=0 j=k

By the previous two lemmas we can now construct a nite list of all those polynomials g of degree d in Q[X] which can possibly be divisors of f . be more precise if m = deg f for any d m 1 we calculate, f (0), f (1), ..., f (d + m). Since f has at most m roots, at least d + 1, among those are = 0. Lets say that for {a0 , a1 , ..., ad } {0, 1, ..., d + m} we have f (ai ) = 0. Using lemmas 7.5.3, 7.5.4 we produce a nite list of polynomials of degree d which are possible divisors of f . If g is such a polynomial we check whether in fact g Z[X], and then, by long division, whether g|f . If g|f for some g then we write f = gh and we repeat the procedure for g, h. This gives us an algorithm to decompose any polynomial f Q[X] as a product of irreducibles. Of course if no g from our nite list divides f we conclude that f is irreducible.

7.6

Factorization in R[X] and C[X]

According to the Fundamental Theorem of Algebra any f C[X] has a root in C. In particular a polynomial f C[X] is irreducible if and only if deg f = 1. Let f R[X] be a polynomial with deg f > 2. If f has areal root clearly it is not irreducible. If a non-real a C is a root of f then a is also a root. So (X a)(X a) divides 2 2aX + |a|2 R[X]. So again f is not irreducible. Hence f . But (X a)(X a) = X the irreducible polynomials in R[X] are the polynomials of degree 1 and the polynomials of degree 2 with no real roots.

30

Which of the following are true? 1. If the polynomial f (x) Z[X] is irreducible in Z[X] then f (X) is also irreducible in Z3 [X]. 2. If the polynomial f (x) Z7 [X] is irreducible in Z7 [X] then f (X) is also irreducible in Z[X]. 3. If K is a eld, a polynomial of degree 3, f (X) K[X] is irreducible if and only if it has no roots in K. 4. There is a monic polynomial f (X) Z[X] such that f ( 2008 ) = 0. 2009

31