You are on page 1of 12

Chapter 4, Arithmetic in F [x]

Polynomial arithmetic and the division algorithm.

We begin by defining the ring of polynomials with coefficients in a ring R. After some
preliminary results, we specialize to the case where R is a field, denoted F . In this chapter
and the next, we will see that much of what works for the ring of integers also works
for polynomials over a field including a division algorithm, factorization into irreducible
elements, and even a form of modular arithmetic.

Theorem 4.1. Let R be a ring. There exists a ring R[x] that contains an element x not
in R and has the properties:
(1) R is a subring of R[x].
(2) xr = rx for all r ∈ R (i.e., x lies in the center of R[x]).
(3) Every nonzero element of R[x] can be uniquely written in the form
a0 + a1 x + · · · + an xn , for some n ≥ 0, ai ∈ R, and an 6= 0.

Definitions. The elements of R[x] are called polynomials with coefficients in R. The
elements ai are called coefficients and x is called an indeterminate. The element an is
called the leading coefficient of the polynomial and n is called the degree. Elements
of R are called constant polynomials, and have degree 0 if they are nonzero. The zero
polynomial does not have a defined degree.

The choice of the letter x is, of course, irrelevant. The construction of the ring R[x] is
done in appendix G, pp. 542+, and has no x. See page 543. Here x = (0, 1, 0, 0, 0, . . . ). If
R has an identity, then it also works as the identity of R[x] and has the formal description
(1, 0, 0, 0, . . . ). More generally, the element r ∈ R corresponds to the element (r, 0, 0, 0, . . . ).

From a more advanced viewpoint, there is a way to characterize the polynomial ring in
terms of homomorphisms. R[x] is the unique ring containing R with the property that for
any ring homomorphism f : R → S and any element s in the center of S, there exists a
unique homomorphism φ : R[x] → S such that φ(r) = f (r) for all r ∈ R and φ(x) = s.

This is the way to think of polynomials as functions, as it gives a meaning


√ to evaluation
of the polynomial
√ at x = s. For example, take R = S = R, s = 2. φ(p(x)) is the
element p( 2) ∈ R. For the next few sections, we ignore the idea of a polynomial as a
function and just think of it as an element of the ring R[x]. Addition and multiplication
are determined by the associative and distributive laws, so are exactly as you have always
done them (unless you lack commutativity for multiplication, in which case you must be
careful). If R is commutative, so is R[x] because x commutes with everything. If R has
an identity, it also works as the identity in R[x].
1
2

Theorem 4.2 (extended). Let f (x), g(x) be nonzero polynomials over a ring R. If
f (x)g(x) 6= 0, then
deg[f (x)g(x)] ≤ deg f (x) + deg g(x).
If R has no zero divisors, then we always have

deg[f (x)g(x)] = deg f (x) + deg g(x).

In particular, this applies if R is an integral domain.

Proof. Write f (x) = a0 + · · · + an xn and g(x) = b0 + · · · + bm xm , with an , bm 6= 0. Then

f (x)g(x) = a0 b0 + (a0 b1 + a1 b0 )x + · · · + an bm xn+m .

Thus the largest power of x that can occur is xn+m , so deg[f (x)g(x)] ≤ deg f (x)+deg g(x).
If R has no zero divisors, then an bm 6= 0 and n+m is the degree, giving us deg[f (x)g(x)] =
deg f (x) + deg g(x). 

Corollary 4.3. If R is an integral domain, so is R[x].

Proof. The theorem shows that R[x] has no zero divisors. R provides the identity. 

The converse of the corollary also holds. If R has a zero divisor r, say with rs = 0,
r 6= 0, s 6= 0, then this still holds in R[x]. The corollary can also be iterated using induction
to show that R[x, y], meaning (R[x])[y], is an integral domain when R is, and so is the n
variable ring R[x1 , x2 , . . . xn ].

Discuss degrees of sums of polynomials.

We have seen a division algorithm in Z and from high school work with polynomials,
you expect one for polynomials with coefficients in a field. In Chapter 9 we will generalize
both of these to a Euclidean domain, where we add an axiom that says there is a division
algorithm (and hence a Euclidean algorithm for gcd’s).

Theorem 4.4 (Division Algorithm). Let F be a field, f (x), g(x) ∈ F [x] with g(x) 6= 0.
Then there exist unique polynomials q(x) and r(x) such that

f (x) = g(x)q(x) + r(x) and either r(x) = 0 or deg r(x) < deg g(x).

When we give an axiom for a division algorithm, we will need a special function to say
r is in some sense smaller than g. For the integers we have their natural ordering and
3

for polynomials we have degrees. There is also a version of Theorem 4.4 when F is only
a commutative ring with 1, but you end up with a power of the leading coefficient of g
multiplying f since you can’t divide by it. The book does a very careful discussion showing
how the proof is similar to the long division algorithm for polynomials. If you have trouble
understanding the proof, read the book’s discussion.

Proof. We begin with existence of q and r. The trivial cases: if f = 0 or deg f < deg g, then
q = 0, r = f works. So we can assume that f 6= 0 and deg f ≥ deg g. We use induction
on n = deg f (x). The induction hypothesis is that the theorem works for any polynomial
of smaller degree (not just degree n − 1); this is an equivalent form of mathematical
induction. The starting case is n = 0. In this case, f and g are constant polynomials, so
lie in F . A solution is q = f /g, r = 0. In general, write f (x) = an xn + · · · a1 x + a0 and
g(x) = bm xm + · · · b1 x + b0 , where n ≥ m, an bm 6= 0. The first term in the quotient using
an n−m
the high school algorithm is x , so we form
bm
an n−m an
f1 (x) = f (x) − x g(x) = (an−1 − bm−1 )xn−1 + · · · ,
bm bm
a polynomial of lower degree than f (x) (or possibly 0). Apply the induction hypothesis to
f1 and g, to obtain polynomials q1 (x) and r(x) satisfying f1 (x) = g(x)q1 (x) + r(x), where
r(x) = 0 or deg r(x) < deg g(x). Putting these together and solving for f gives
 
an n−m
f (x) = g(x) x + q1 (x) + r(x)
bm
an n−m
as desired, the q(x) being x + q1 (x).
bm
To show uniqueness, we assume there are other polynomials q2 (x), r2 (x) satisfying the
same conditions. Then g(x)q(x) + r(x) = f (x) = g(x)q2 (x) + r2 (x), so that g(x)[q(x) −
q2 (x)] = r2 (x) − r(x). The left hand side is either zero or has degree at least equal to deg g
by Theorem 4.2. The right hand side has degree less than deg g if it is nonzero. Therefore
both must be zero, hence q = q2 , r = r2 . 

Example: page 88, exercise 5(c), 14(b).

Definition, p. 90. The definitions for divides and factor are the same as in the integers,
and will be repeated later for any ring.

Some things for F [x] are slightly different than for the integers. For example, (3x + 6) |
(x + 2) because x + 2 = 13 (3x + 6). In general, multiplication by a nonzero constant does
not affect divisibility; this is because they are units in F [x]. This is different than working
in Z[x], where there is no element 31 !
4

By Theorem 4.2, if f (x) | g(x), then deg f (x) ≤ deg g(x). To make greatest common
divisors unique, we need a new condition (analogous to assuming they were positive in Z):

Definition p. 91. Let f (x), g(x) ∈ F [x], not both 0. The greatest common divisor
of f (x) and g(x) is the monic polynomial of highest degree that divides them both. The
book gives no notation; I will continue to use gcd(f (x), g(x)). As before, we say f (x) and
g(x) are relatively prime if their gcd is 1.

The next theorem is analogous to Theorem 1.3 for the integers. It can be proved as
we did there, by first checking that the Euclidean algorithm still works in F [x]. (In using
it, the last nonzero remainder has to be multiplied by a constant to make it monic.) For
variety, we shall follow the book’s proof this time.

Theorem 4.5. Let F be a field, f (x), g(x) ∈ F [x], not both 0. Then there exists a unique
gcd d(x) of f (x) and g(x). Furthermore, there exist polynomials u(x) and v(x) such that
d(x) = f (x)u(x) + g(x)v(x).

Proof. Let t(x) be a (soon to be shown unique) monic polynomial of lowest degree in the set
{ f (x)m(x) + g(x)n(x) | m(x), n(x) ∈ F [x] }. So by definition, t(x) = f (x)u(x) + g(x)v(x)
for some polynomials u(x), v(x). Use the division algorithm to find q, r such that f (x) =
t(x)q(x) + r(x), with r = 0 or deg r < deg t. Then r(x) = f − tq = f − (f u + gv)q =
f (1 − uq) + g(−vq) ∈ S. By the minimality of the degree of t, we must have r = 0 and so
t | f . Similarly, t | g so is a common divisor. For any other common divisor s(x), write
f (x) = s(x)f1 (x), g(x) = s(x)g1 (x). Then t = f u + gv = sf1 u + sg1 v is divisible by s, so
deg s ≤ deg t. Therefore, t is a greatest common divisor. Now let d(x) be another one. By
the argument for s above, d | t. Reversing their roles, we also have that t | d. Thus they
differ by a factor which is an invertible element of F [x], that is, an element of F . Since
they are both monic, they must be equal. (Details are homework, p. 93.) 

Theorem 4.7. Let F be a field with f, g, h ∈ F [x]. If f (x) | g(x)h(x) and gcd(f (x), g(x)) =
1, then f (x) | h(x).

The proof is identical to that for elements of Z.


5

Irreducibles and unique factorization.

Theorem 4.8 (Units in R[x]). Let R be an integral domain. f (x) ∈ R[x] is a unit iff
f (x) is a constant polynomial that is a unit in R.

Proof. Clearly such functions are units in R[x]. On the other hand, suppose that f (x)g(x) =
1. Then deg f + deg g = deg 1 = 0, so deg f = deg g = 0. Hence they are in R, and units
because their product is 1.

If R has zero divisors, the theorem may fail. For example, over Z9 we have (3x+1)(6x+
1) = 1, so 3x + 1 and 6x + 1 are units in Z9 [x].

Corollary 4.9. If F is a field, then f (x) ∈ F [x] is a unit iff it is a nonzero constant
polynomial. 

Examples: Discuss units in Z[x], R[x], Zp [x].

Definition, p. 95. Two elements a, b in a commutative ring R are associates if there


exists a unit u ∈ R with a = ub (equivalently, b = u−1 a). [Note that it is an equivalence
relation; discuss associates in Z, F, F [x].]
A nonzero element p in an integral domain R is irreducible if it is not a unit and whenever
one can write p = bc with b, c ∈ R, then b or c is a unit. In particular, a nonconstant
polynomial p(x) with coefficients in an integral domain R is called irreducible if its only
divisors are its associates and units in R (which by 4.8 are the units in R[x]). A nonunit
which is not irreducible is called reducible. (As the book points out, the words prime
and irreducible have the same meaning in this context (cf. Theorem 4.11 and Theorem
9.15 for unique factorization domains).

Examples. Every polynomial of degree 1 over a field is irreducible (think of degrees).


From high school you might recall that every polynomial over R can be factored into linear
and irreducible quadratic factors. This is a special property of the real numbers and will
not be proved in this course. We will later see that there are irreducible polynomials of
every degree in Q[x]. For polynomials of degree larger than 3, they may be reducible even
though they have no roots in the field: x4 + 2x2 + 1 = (x2 + 1)2 over R. But x2 + 1 is
reducible as an element of C[x], so irreducibility depends on which ring you are thinking
of it being an element of.

Theorem 4.10. Let F be a field. A nonzero polynomial is reducible in F [x] iff it can be
written as a product of two polynomials of lower degree. 
6

This is not true for F an integral domain: in Z[x], 2x + 2 = 2(x + 1) and neither 2 nor
x + 1 is a unit.

Theorem 4.11. Let F be a field and p(x) ∈ F [x], p(x) ∈


/ F . TFAE:
(1) p(x) is irreducible.
(2) If b(x), c(x) ∈ F [x] with p(x) | b(x)c(x), then p(x) | b(x) or p(x) | c(x).
(3) If b(x), c(x) ∈ F [x] with p(x) = b(x)c(x), then b(x) or c(x) is a nonzero constant
polynomial.

Proof. (1) =⇒ (2) Same as Theorem 1.8 for Z: gcd(p, b) | p, so is a unit or associate of
p. If a unit, then Theorem 4.7 says p|c. Otherwise, p|b.
(2) =⇒ (3) By (2), p|b or p|c; we may assume the former, say pr = b, so p = bc = prc.
Hence rc = 1, since F [x] is an integral domain. But then c is a unit (nonzero element of
F ) as desired.
(3) =⇒ (1) If b is any divisor of p, then p = bc for some c. By (3), either b is a unit, and
c = b−1 p is an associate of p, or c is a unit, and b = c−1 p is an associate of p. 

Corollary 4.12. Let F be a field and p(x) ∈ F [x] be irreducible. If p divides a product of
polynomials, then it divides at least one of them.

This is proved by induction, the same as for the integers.

Theorem 4.13 (Unique factorization). Let F be a field. Every nonconstant polynomial


in F [x] can be written as a product of irreducible polynomials. The factorization is unique
up to order and multiplication by units.

This theorem does not actually require that F be a field, but it certainly needs F to
be a ring in which one has unique factorization. This is, in fact, all that is required. (See
exercise 36, p. 307.) In particular, one still has unique factorization for Z[x] and R[x, y],
but we shall not give a detailed proof in this course.

Proof. The proof is essentially the same as the proof of the Fundamental Theorem of
Arithmetic. I will only sketch the proof here, as a reminder.
Let S be the set of polynomials which fail and assume it is nonempty. Then it has a
polynomial of smallest degree; it is not irreducible, so it factors into polynomials of smaller
degree (so not in S). Their factorizations give one of the original polynomial, hence a
contradiction. For uniqueness, take two factorizations f (x) = p1 · · · pr = q1 · · · qs into
irreducibles. p1 divides the product of qi ’s, so divides one of them, hence is an associate
of it. Cancel and continue to see that r = s and all others pair up as associates. 
7

Examples. 1. (2x+2)(x+2) = (x+1)(2x+4) = (3x+3)( 32 x+ 43 ) are various factorizations


of 2x2 + 6x + 4 into irreducible polynomials in Q[x].

2. Exercise 23, page 99. x2 + 2 is irreducible in Z5 [x]: if it factors, each factor must be
linear, say (x + a)(x + b). But then −a, −b ∈ Z5 are roots of x2 + 2; but none of the five
elements work.
In Z5 [x], x4 − 4 = x4 + 1 has no roots (try the 5 possibilities), but does factor as (x2 +
2)(x2 − 2), so this is a factorization into irreducibles.

3. Exercise 21, p. 99. In finite fields we can using counting techniques. For Zp [x], there
are p degree 1 monic polynomials x + a. Andsoto count the degree 2 monic reducible
p p(p − 1) p(p + 1)
polynomials, we need all products of these: +p = +p = , the
2 2 2
number with distinct factors plus the number of squares. Since there are p2 monic degree
2 polynomials x2 +ax+b, this leaves the number of irreducible monic degree 2 polynomials
as p2 − p(p+1)
2 = p(p−1)
2 .
8

Polynomials as functions.

Let R be a commutative ring, f (x) = an xn + · · · + a0 ∈ R[x]. The function f : R → R


induced by f (x) is called a polynomial function and is defined by f (r) = an rn +
an−1 rn−1 + · · · + a1 r + a0 .

We must be very careful in writing f (x) to differentiate between the polynomial as an


element of R[x] (where x is an indeterminate) and the polynomial function from R to R
(where x is a variable). It is quite possible for two different polynomials to induce the same
polynomial function (though not on the real numbers). Your homework for this section
will ask you to find nonzero polynomials that induce the zero function. A simple example
is given by the polynomials x2 + 1, x4 + 1 ∈ Z3 [x]. They are different polynomials but
induce the same function 0 7→ 1, 1 7→ 2, −1 7→ 2. Indeed, there are only 33 = 27 possible
functions from Z3 to Z3 , but infinitely many polynomials.

Definition, p. 101. Let R be a commutative ring, f (x) ∈ R[x]. And element a ∈ R is


called a root of f (x) if f (a) = 0R (meaning that the induced function f : R → R maps
a 7→ 0R ).

The standard high school question of solving f (x) = x2 + 3x + 2 = 0 means finding the
roots of f (x), or equivalently, the set of all r ∈ R which are mapped to zero by the induced
function. You did not need to be so careful about the difference between the two meanings
of f (x) in high school because two polynomials in R[x] are equal iff their induced functions
are the same. To make use of the idea of roots in determining reducibility of polynomials,
we generalize a couple of high school theorems from R to arbitrary fields.

Theorem 4.14 (Remainder Theorem). Let F be a field, f (x) ∈ F [x] and a ∈ F . The
remainder when f (x) is divided by x − a is f (a).

Proof. Use the division algorithm and the fact that the remainder must be an element of
F. 

Corollary (Factor Theorem). Let F be a field, f (x) ∈ F [x] and a ∈ F . a is a root of


f (x) iff x − a is a factor of f (x) in F [x]. 

Corollary 4.16. Let F be a field and f (x) a polynomial of degree n in F [x]. Then f (x)
has at most n roots in F

Examples: x2 + 1 has no roots in R. It has 2 roots in C. It has infinitely many roots


in the real quaternions H, in particular, i, j and k. So commutativity is needed! 2(x2 + x)
has four roots in Z6 , namely 0, 2, 3, 5. So we must not have zero divisors either. In fact,
9

the theorem holds if F is only an integral domain, with a similar proof to the one we are
about to give, where one first extends the division algorithm as we mentioned earlier to
monic polynomials. A proof can also be based on Corollary 4.16 together with Theorem
9.30 to be done later, which says that any integral domain can be embedded in a field (like
Z in Q).

Proof of Corollary 4.16. Induct on n. If n = 0, f (x) is a constant in F , so has no roots.


Assume n > 0 and the theorem holds for polynomials of degree n − 1. If f (x) has no roots,
we are done. If f (x) has a root a ∈ F , then f (x) = (x − a)g(x) for some g(x) ∈ F [x] of
degree n − 1. By the induction hypothesis, g(x) has at most n − 1 roots in F . Any root
b 6= a of f (x) satisfies 0 = f (b) = (b − a)g(b), so b is also a root of g. Thus the number of
roots of f (x) is at most 1 + (n − 1) = n. 

Notice where we used the fact that F has no zero divisors.

Another corollary of the factor theorem is that if f (x) ∈ F [x] is irreducible, then it has
no roots in F . This gives us a test to see if a polynomial is reducible, but does not tell if it
is irreducible (recall x4 + 2x2 + 1 = (x2 + 1)2 ∈ R[x]). As we noted in an earlier example,
we do have the converse if the degree of f (x) is 2 or 3.

Corollary 4.18. If F is a field and f (x) ∈ F [x] has degree 2 or 3, then f (x) is irreducible
iff f (x) has no roots in F . 

We also see that what happened for finite fields with polynomial functions not being
the same as polynomials does not happen with infinite fields.

Corollary 4.19. Let F be an infinite field and f (x), g(x) ∈ F [x]. f (x) and g(x) induce
the same function on F iff f (x) = g(x) in F [x].

Proof. (⇐=) is clear.


(=⇒) Assume that f (a) = g(a) for every a ∈ F . Then every element of F is a root of the
polynomial f (x) − g(x). Since F is infinite, the only way Corollary 4.16 can be true is if
f (x) − g(x) is the zero polynomial. That is, we have f (x) = g(x). 

Examples: page 105: 11, 16.

Irreducibility in Q[x].

We shall reduce the problem of irreducibility in Q[x] to looking at Z[x]. Indeed, if


f (x) ∈ Q[x], we can multiply by an integer to clear denominators: cf (x) ∈ Z[x] for some
10

c ∈ Z. It is clear that f (x) factors into polynomials of degree at least one iff cf (x) does—
at least with rational coefficients. We shall see that only integer coefficients are needed.
There exist algorithms for completely factoring polynomials in Z[x], but they are beyond
the scope of this course. We shall look at some standard theorems regarding polynomials
over Z. The first is commonly done in high school:

Theorem 4.20 (Rational root test). Let f (x) = an xn + · · · + a1 x + a0 ∈ Z[x]. If the


rational number r/s 6= 0 is a root of f (x) and is in lowest terms, then r|a0 and s|an .

r
Proof. Since s is a root, we have
 r n  r n−1 r
an + an−1 + · · · + a1 + a0 = 0.
s s s
Multiply by sn to clear denominators and rearrange the terms to get

a0 sn = r[−an rn−1 − · · · − a1 sn−1 ] and


an rn = s[−an−1 rn−1 − · · · − a0 sn−1 ] .
r
Since s is in lowest terms, we have gcd(r, s) = 1, whence r|a0 and s|an . 

Examples. 1. f (x) = 2x4 + x3 − 2x2 − 4x − 3 has possible roots ±1, ±3, ± 21 , ± 32 .


Check and find that f (−1) = 0 = f (3/2). Factor out x + 1 and 2x − 3 to get f (x) =
(x + 1)(2x − 3)(x2 + x + 1). The last factor has no rational (or even real) roots, so we have
a complete factorization into irreducibles over Q.
2. x3 + 2x2 − 4x + 3 is irreducible over Q (but not over R!).

The main issue in reducing the factoring problem to Z is the question of whether a
polynomial with integer coefficients might factor with rational coefficients, but not with
integer coefficients. In fact, this cannot happen.

Lemma 4.21. Let f, g, h ∈ Z[x] with f = gh and assume that p ∈ Z is a prime which
divides every coefficient of f . Then either p divides every coefficient of g or p divides every
coefficient of h.

Proof. Write f (x) = ai xi , g(x) = bi xi and h(x) = ci xi . Assume the claim is false.
P P P
Then there is a first coefficient br not divisible by p and a first coefficient
Pr+t ct not divisible
by p. From f = gh, we see that the coefficient ar+t of f is just i=0 bi cr+t−i . By choice
of br and ct , p divides every term in the sum for ar+t except br ct since every other term
has a factor from b0 , . . . , br−1 or from c0 , . . . , ct−1 . Since p does not divide br ct , it cannot
divide the sum, a contradiction of the fact that p divides all coefficients of f . 
11

Theorem 4.22 (Gauss’s Lemma—see page 325). Let f (x) ∈ Z[x]. f (x) factors as a prod-
uct of polynomials of degrees m and n in Q[x] iff f (x) factors as a product of polynomials
of degrees m and n in Z[x].

Proof. (⇐=) is obvious. For ( =⇒ ), suppose f (x) = g(x)h(x) in Q[x]. Clear denominators
by finding integers c, d > 0 such that cg(x), dh(x) ∈ Z[x] (i.e., find a common denominator
and multiply by it). Then cdf (x) = [cg(x)][dh(x)] is a factorization of cdf (x) in Z[x] into
polynomials of the same degrees. We will be done if we can show that cd = 1. If not, it has
some positive prime divisor p, which then divides every coefficient of cdf (x), hence every
coefficient of cg(x) or of dh(x) by the lemma; we may assume it is cg(x). Write cd = pt and
factor out p to write cg(x) = pk(x) with k(x) ∈ Z[x]. Therefore ptf (x) = p[k(x)][dh(x)].
Cancel p. Continuing in this way, we can cancel every prime factor of the original cd,
leaving a factorization of f (x) into polynomials of the same degrees as g and h. 

Exercise 10, p. 114: If f (x) = g(x)h(x) ∈ Z[x] and f is monic, then the leading
coefficients of g and h must multiply to be 1, hence are either both +1 or both −1. In the
latter case, write f (x) = [−g(x)][−h(x)], so we get a factorization into monic polynomials
in either case.

Example, p. 110. Show x4 − 5x2 + 1 is irreducible over Q.

Theorem 4.23 (Eisenstein’s Criterion). Let f (x) = an xn +· · ·+a1 x+a0 be a nonconstant


polynomial with integer coefficients. Assume there is a prime p such that
(1) p divides a0 , a1 , . . . , an−1
(2) p does not divide an
(3) p2 does not divide a0 .
Then f is irreducible in Q[x].

Proof. Assume that f is reducible. By Gauss’s Lemma, we can factor f over Z, say
f (x) = (b0 + b1 x + · · · + br xr )(c0 + c1 x + · · · + cs xs ), with r, s ≥ 1. Since p|a0 = b0 c0 , we
know p divides one of them, say b0 . Since p2 6 | a0 , p cannot also divide c0 . Since an = br cs is
not divisible by p, neither is br , so there is some minimum index k so that bk is not divisible
by p. (So p does divide b0 , . . . , bk−1 .) Now we have ak = b0 ck + b1 ck−1 + · · · + bk−1 c1 + bk c0
in which p divides every term except the last, a contradiction. 

This is a very useful theorem, though it certainly doesn’t cover all cases. It is easy
to construct numerous irreducible polynomials, and in particular, polynomials of every
degree. One of the most interesting standard applications of the theorem is shown in
Exercise 20, p. 114: for p prime, f (x) = xp−1 + xp−2 + · · · + x + 1 is irreducible in Q[x].
If we multiply f (x) by x − 1, we get xp − 1, whose roots are the p roots of 1 on the unit
circle in the complex plane, namely e2πik/p , k = 0, 1, . . . , p − 1. Note that we can factor
12

f (x) iff we can factor f (x + c) for any c ∈ Q. To apply the Eisenstein criterion, we look at

(x + 1)p − 1
f (x + 1) =
(x + 1) − 1
(x + 1)p − 1
=
 x      
p p p−1 p 
= x + x + ··· + x+1 −1 x
p−1 1
   
p p
= xp−1 + xp−2 + · · · + ,
p−1 1

in which we see that the three conditions of Eisenstein’s criterion hold for the coefficients.
Therefore f (x) is irreducible.

There is one other very useful theorem; it can be used to show that a polynomial is
irreducible, but not directly that it is reducible.

Pn
Theorem 4.24. Let f (x) = i=0 ai xi ∈ Z[x] and assume that p is a positive prime not
dividing an . Let f¯(x) be the image of f (x) in Zp [x] under the homomorphism that reduces
all coefficients modulo p. If f¯(x) is irreducible in Zp [x], then f (x) is irreducible in Q[x].

Proof. By Gauss’s lemma, it suffices to show that f (x) is irreducible in Z[x]. Assume it
is reducible: f (x) = g(x)h(x). Modulo p we have f¯(x) = ḡ(x)h̄(x). Furthermore, since
p 6 |an , it also does not divide the leading coefficients of g(x) or h(x), so we have a genuine
factorization of f¯(x) whenever we have one of f (x) (that is, the degrees do not change).
Since f¯(x) is irreducible in Zp [x], the original polynomial f (x) is irreducible also. 

As an example, consider x2 − 2x − 1 ∈ Z[x]. Modulo 2, it equals x2 − 1 = (x + 1)2 so


tells us nothing. Modulo 3, we have x2 − 2x − 1 which is seen to be irreducible by trying
the 3 possible roots and finding that none works. In general, there are only finitely many
polynomials of a given degree in Zp [x] so all possibilities can be checked; this would be
very time consuming, but in practice, there are special algorithms that work quite well
by computer. That is why symbolic algebra programs such as Derive can easily factor
polynomials over the integers or modulo p (or modulo n also).

Example. Page 114, #11. 30xn − 91 has no roots in Q for any n > 1. By Eisenstein’s
criterion with p = 7, the polynomial is irreducible in Q[x] and therefore it has no roots by
Corollary 4.17.

We have mentioned the results of Section 4.6 on complex numbers and will do no more
with it at this time.

You might also like