Unless otherwise stated, in this section we denote by K an arbitrary field and by K[X] the ring of polynomials in one indeterminate X and with coefficients from K. Since K[X] is a PID, it enjoys many properties similar to those of . To start with, we take a look at these properties. Then we introduce the concept of algebraic elements and discuss how irreducible polynomials can be used to construct (algebraic) extensions of fields. When no confusions are likely, we denote a polynomial by f only.
Since K[X] is a PID and hence a UFD, every polynomial in K[X] can be written essentially uniquely as a product of prime polynomials. Conventionally prime polynomials are more commonly referred to as irreducible polynomials. Similar to the case of the ring K[X] contains an infinite number of irreducible elements, for if K is infinite, then is an infinite set of irreducible polynomials of K[X], and if K is finite, then as we will see later, there is an irreducible polynomial of degree d in K[X] for every .
It is important to note here that the concept of irreducibility of a polynomial is very much dependent on the field K. If K ⊆ L is a field extension, then a polynomial in K[X] is naturally an element of L[X] also. A polynomial which is irreducible over K need not continue to remain so over L. For example, the polynomial x2 – 2 is irreducible over , but reducible over , since , being a real number but not a rational number. As a second example, the polynomial x2 + 1 is irreducible over both and but not over . In fact, we will show shortly that an irreducible polynomial in K[X] of degree > 1 becomes reducible over a suitable extension of K.
For polynomials f(X), with g(X) ≠ 0, there exist unique polynomials q(X) and r(X) in K[X] such that f(X) = q(X)g(X) + r(X) with r(X) = 0 or deg r(X) < deg g(X). The polynomials q(X) and r(X) are respectively called the quotient and remainder of polynomial division of f(X) by g(X) and can be obtained by the so-called long division procedure. We use the notations: q(X) = f(X) quot g(X) and r(X) = f(X) rem g(X).
Whenever we talk about the gcd of two non-zero polynomials, we usually refer to the monic gcd, that is, a polynomial with leading coefficient 1. This makes the gcd of two polynomials unique. We have gcd(f(X), g(X)) = gcd(g(X), r(X)), where r(X) = f(X) rem g(X). This gives rise to an algorithm (similar to the Euclidean gcd algorithm for integers) for computing the gcd of two polynomials. Bézout relations also hold for polynomials. More specifically:
Let f(X), , not both zero, and d(X) the (monic) gcd of f(X) and g(X). Then there are polynomials u(X), such that d(X) = u(X)f(X) + v(X)g(X). (Such an equality is called a Bézout relation.) Furthermore, if f(X) and g(X) are non-zero and not both constant, then u(X) and v(X) can be so chosen that deg u(X) < deg g(X) and deg v(X) < deg f(X).[6]
Proof Similar to the proof of Proposition 2.16. |
The concept of congruence can be extended to polynomials, namely, if , then two polynomials g(X), are said to be congruent modulo f(X), denoted g(X) ≡ h(X) (mod f(X)), if f(X)|(g(X) – h(X)), that is, if there exists with g(X) – h(X) = u(X)f(X), or equivalently, if g(X) rem f(X) = h(X) rem f(X).
The principal ideals 〈f(X)〉 of K[X] play an important role (as do the ideals 〈n〉 of ). Let us investigate the structure of the quotient ring R := K[X]/〈f(X)〉 for a non-constant polynomial . If r(X) denotes the remainder of division of by f(X), then it is clear that the residue classes of g(X) and r(X) are the same in R. On the other hand, two polynomials g(X), with deg g(X) < deg f(X) and deg h(X) < deg f(X) represent the same residue class in R if and only if g(X) = h(X). Thus elements of R are uniquely representable as polynomials of degrees < deg f(X). In other words, we may represent the ring R as the set together with addition and multiplication modulo the polynomial f(X). The ring R contains all the constant polynomials , that is, the field K is canonically embedded in R. In general, R is not a field. The next theorem gives the criterion for R to be a field.
For a non-constant polynomial , the ring K[X]/〈f(X)〉 is a field if and only if f(X) is irreducible in K[X]. Proof If f(X) is reducible over K, then we can write f(X) = g(X)h(X) for some polynomials g(X), with 1 ≤ deg g < deg f and 1 ≤ deg h < deg f. Then both g and h represent non-zero elements in K[X]/〈f(X)〉, whose product is 0, that is, K[X]/〈f(X)〉 has non-zero zero divisors. Conversely, if f(X) is irreducible over K and if g(X) is a non-zero polynomial of degree < deg f(X), then gcd(f(X), g(X)) = 1, so that by Proposition 2.23 there exist polynomials u(X), with u(X)f(X) + v(X)g(X) = 1 and deg v(X) < deg f(X). Thus we see that v(X)g(X) ≡ 1 (mod f(X)), that is, g(X) has a multiplicative inverse modulo f(X). |
Let L := K[X]/〈f(X)〉 with f(X) irreducible over K. Then K ⊆ L is a field extension. If deg f(X) = 1, then L is isomorphic to K. If deg f(X) ≥ 2, then L is a proper extension of K. This gives us a useful and important way of representing the extension field L, given a representation for K. (For example, see Section 2.9.)
The study of the roots of a polynomial is the central objective in algebra. We now derive some elementary properties of roots of polynomials.
Let . An element is said to be a root of f, if f(a) = 0. |
Let and . Then f(X) = (X – a)q(X) + f(a) for some . In particular, a is a root of f(X) if and only if X – a divides f(X). Proof Polynomial division of f(X) by X – a gives f(X) = (X – a)q(X) + r(X) with deg r(X) < deg(X – a) = 1. Thus r(X) is a constant polynomial. Let us denote r(X) by . Substituting X = a gives f(a) = r. |
A non-zero polynomial with d := deg f can have at most d roots in K. Proof We proceed by induction on d. The result clearly holds for d = 0. So assume that d ≥ 1 and that the result holds for all polynomials of degree d – 1. If f has no roots in K, we are done. So assume that f has a root, say, . By Proposition 2.24, we have f(X) = (X – a)g(X) for some . Clearly, deg g = d – 1 and so by the induction hypothesis g has at most d – 1 roots. Since K is a field (and hence does not contain non-zero zero divisors), it follows that the roots of f are precisely a and the roots of g. This establishes the induction step. |
In the last proof, the only result we have used to exploit the fact that K is a field is that K contains no non-zero zero divisors. This is, however, true for every integral domain. Thus Proposition 2.25 continues to hold if K is any integral domain (not necessarily a field). However, if K is not an integral domain, the proposition is not necessarily true. For example, if ab = 0 with a, , a ≠ b, then the polynomial X2 + (b – a)X has at least three roots: 0, a and a – b.
For a field extension K ⊆ L and for a polynomial , we may think of the roots of f in L, since too. Clearly, all the roots of f in K are also roots of f in L. However, the converse is not true in general. For example, the only roots of X4 – 1 in are ±1, whereas the roots of the same polynomial in are ±1, ±i. Indeed we have the following important result.
For any non-constant polynomial , there exists a field extension K′ of K such that f has a root in K′. Proof If f has a root in K, taking K′ = K proves the proposition. So we assume that f has no root in K (which implies that deg f ≥ 2). In principle, we do not require f to be irreducible. But if we consider a non-constant factor g of f, irreducible over K, we see that the roots of g in any extension L of K are roots of f in L too. Thus we may replace f by g and assume, without loss of generality, that f is irreducible. We construct the field extension K′ := K[X]/〈f〉 of K and denote the equivalence class of X in K′ by α. (One also writes x, X or [X] to denote this equivalence class.) It is clear that , that is, α is a root of f(X) in K′. |
We say that the field K′ in the proof of the last proposition is obtained by adjoining the root α of f and denote this as K′ = K(α). We can write f(X) = (X – α)f1(X), where and deg f1 = (deg f) – 1. Now there is a field extension K″ of K′, where f1 has a root. Proceeding in this way we prove the following result.
A non-constant polynomial f in K[X] with deg f = d has d roots (not necessarily all distinct) in some field extension L of K. |
If a polynomial of degree d ≥ 1 has all its roots α1, . . . , αd in L, then f(X) = a(X – α1) · · · (X – αd) for some (actually ). In this case, we say that f splits (completely or into linear factors) over L.
Let be a non-constant polynomial. A minimal (with respect to inclusion) field extension of K, over which f splits completely is called a splitting field of f over K.[7] This is a minimal field which contains K and all the roots of f.
|
Every non-constant polynomial has a splitting field L over K. Quite importantly, this field L is unique in some sense. This allows us to call the splitting field of f instead of a splitting field of f. We discuss these topics further in Section 2.8.
Let f be a non-constant polynomial in K[X] and let α be a root of f (in some extension of K). The largest natural number n for which (X –α)n|f(X) is called the multiplicity of the root α (in f). If n = 1 (resp. n > 1), then α is called a simple (resp. multiple) root of f. If all the roots of f are simple, then we call f a square-free polynomial. It is easy to see that f is square-free, only if f is not divisible by the square of a non-constant polynomial in K[X]. The reverse implication also holds, if char K = 0 or if K is a finite field (or, more generally, if K is a perfect field—see Exercise 2.76). |
The notion of multiplicity can be extended to a non-root β of f by setting the multiplicity of β to zero.
Here we assume, unless otherwise stated, that K ⊆ L is a field extension.
|
Let be algebraic over K. A non-constant polynomial of least positive degree with f(α) = 0 is called a minimal polynomial of α over K. |
Let be algebraic over K. A minimal polynomial f of α over K is irreducible over K. If is a polynomial with h(α) = 0, then f|h. In particular, any two minimal polynomials f and g of α satisfy g(X) = cf(X) for some . Proof Let f = f1f2 for some non-constant polynomials f1, . Since K is a field and 0 = f(α) = f1(α)f2(α), we have f1(α) = 0 or f2(α) = 0. But deg f1 < deg f and deg f2 < deg f, a contradiction to the choice of f. Using polynomial division one can write h(X) = q(X)f(X) + r(X) for some polynomials q, . Now h(α) = 0 implies r(α) = 0. Since deg r < deg f, by the choice of f we must then have r(X) = 0, that is, f|h. Finally, if f and g are two minimal polynomials of α over K, then f|g and g|f and it follows that g(X) = cf(X) for some unit c of K[X]. But the only units of K[X] are the non-zero elements of K. |
By Proposition 2.28, a monic minimal polynomial f of α over K is uniquely determined by α and K. It is, therefore, customary to define the minimal polynomial of α over K to be this (unique) monic polynomial. Unless otherwise stated, we will stick to this revised definition and write f(X) = minpolyα, K(X).
For a field K, the following conditions are equivalent.
Proof [(a)⇒(b)] Consider a non-constant irreducible polynomial and the field extension L = K[X]/〈f〉 of K. We have seen that L contains a root of f. We will prove in Section 2.8 that such an extension is algebraic (Corollary 2.11). Hence (a) implies that L = K, that is, K contains a root of f. [(b)⇒(c)] Let be a non-constant polynomial. By (b), f has a root, say, . Thus f(X) = (X – α1)f1(X) for some with deg f1 = (deg f) – 1. If f1 is a constant polynomial, we are done. Otherwise, we find as above and with f1(X) = (X – α2)f2(X) and with deg f2 = (deg f) – 2. Proceeding in this way proves (c). [(c)⇒(d)] Obvious. [(d)⇒(a)] Let be algebraic over K and let . Since f is irreducible, by (d) deg f = 1, that is, f(X) = X – α, that is, . |
A field K satisfying the equivalent conditions of Proposition 2.29 is called an algebraically closed field. For an arbitrary field K, a minimal algebraically closed field containing K is called an algebraic closure of K. |
We will see in Section 2.8 that an algebraic closure of every field exists and is unique in some sense. The algebraic closure of an algebraically closed field K is K itself. We end this section with the following well-known theorem. We will not prove the theorem in this book, because every known proof of it uses some kind of complex analysis which this book does not deal with.
The field is algebraically closed. is not algebraically closed, since the proper extension is algebraic (See Example 2.10). Indeed, is the algebraic closure of . |
2.51 | Let R be a ring and f, . Show that:
|
2.52 | Let f, , where R is an integral domain. Show that if f(ai) = g(ai) for i = 1, . . . , n, where n > max(deg f, deg g) and where a1, . . . , an are distinct elements of R, then f = g. In particular, if f(a) = g(a) for an infinite number of , then f = g. |
2.53 | Lagrange’s interpolation formula Let K be a field and let a0, . . . , an be distinct elements of K. Show that for (not necessarily all distinct), there exists a unique polynomial of degree ≤ n such that f(ai) = bi for all i = 0, . . . , n. [H] |
2.54 | Polynomials over a UFD Let R be a UFD. For a non-zero polynomial , a gcd of the coefficients of f is called a content of f and is denoted by cont f. One can then write f = (cont f)f1, where with cont . f1 is called a primitive part of f and is often denoted as pp f. It is clear that cont f and pp f are unique up to multiplication by units of R. If for a non-zero polynomial the content cont (or, equivalently, if f and pp f are associates), then f is called a primitive polynomial. Show that for two non-zero polynomials f, the elements cont(f g) and (cont f)(cont g) are associates in R. In particular, the product of two primitive polynomials is again primitive. |
2.55 | Let R be a UFD. Show that a non-constant polynomial is irreducible over R if and only if f is irreducible over Q(R), where Q(R) denotes the quotient field of R (see Exercise 2.34). |
2.56 |
|
2.57 | Let K ⊆ L be a field extension and f1, . . . , fn non-constant polynomials in K[X]. Show that each fi, i = 1, . . . , n, splits over L if and only if the product f1 · · · fn splits over L. |
2.58 | Show that the irreducible polynomials in have degrees ≤ 2. [H] |
2.59 | Show that a finite field (that is, a field with finite cardinality) is not algebraically closed. In particular, the algebraic closure of a finite field is infinite. |
2.60 | A complex number z is called an algebraic number, if z is algebraic over . An algebraic number z is called an algebraic integer, if z is a root of a monic polynomial in . Show that:
|
2.61 | Let K be a field and . The formal derivative f′ of f is defined to be the polynomial . Show that:
|
2.62 | Let be a non-constant polynomial of degree d and let α1, . . . , αd be the roots of f (in some extension field of K). The quantity is called the discriminant of f. Prove the following assertions:
|
13.58.121.8