Irreducibility of Polynomials
Inside the polynomial ring
F[x] some polynomials split apart and some refuse to. A polynomial that
cannot be written as a product of two polynomials of lower degree is called
irreducible — it is the atom of the polynomial ring, the analogue of a
prime number in \mathbb{Z}. Just as every integer factors uniquely
into primes, every polynomial over a field factors uniquely into irreducibles: F[x]
is a unique
factorization domain, and the irreducibles are its primes.
Why care so much about a polynomial that will not factor? Because an irreducible
p(x) is exactly the raw material you need to build a new field. The
quotient ring F[x]/(p(x)) is a field precisely when
p is irreducible — this single fact conjures the complex numbers out of
\mathbb{R}[x]/(x^2+1), every finite field
\mathrm{GF}(p^n) out of \mathbb{F}_p[x], and the
whole tower of field extensions that Galois theory is built on. Deciding whether a
polynomial is irreducible is therefore the gateway skill for the rest of the subject. This page is a
toolkit for making that decision.
Irreducibility is relative to the field
The first surprise: whether a polynomial is irreducible depends on which field you are allowed to
use. "Can it factor?" has no absolute answer — it always means "can it factor over
this field?"
Take x^2 + 1. Over the reals it never factors — there is no real number whose
square is -1 — so it is irreducible in \mathbb{R}[x].
But hand it to the complex numbers and it splits at once:
x^2 + 1 = (x - i)(x + i) \quad \text{in } \mathbb{C}[x].
The same story runs the other way. x^2 - 2 is irreducible over
\mathbb{Q} — no rational number squares to 2, since
\sqrt 2 is irrational — yet over \mathbb{R} it
happily factors as (x - \sqrt 2)(x + \sqrt 2). Enlarge the field and more
factorizations become available. The most generous field is
\mathbb{C}: by the Fundamental Theorem of Algebra every non-constant
polynomial factors completely into linear pieces there, so the only irreducibles over
\mathbb{C} are the degree-1 polynomials. All the interesting irreducibility
lives over smaller fields like \mathbb{Q} and
\mathbb{F}_p.
The root test — but only up to degree 3
A root gives a factor. If a \in F is a root of
f, then dividing by
(x - a) leaves no remainder — the Factor Theorem — so
f = (x - a)\, q(x) and f is reducible (as long as
\deg f \ge 2). This gives a clean test in low degree.
Let f \in F[x] with \deg f \in \{2, 3\}. Then:
-
f is reducible over F if and
only if it has a root in F.
-
Equivalently, f is irreducible exactly when it has
no root in F.
The reason: any factorization of a degree 2 or 3 polynomial into lower-degree factors must
include a linear factor (x - a) — you cannot split
2 = 1 + 1 or 3 = 1 + 2 without a piece of degree 1
— and a linear factor is a root.
Worked example. Is x^2 - 2 irreducible over
\mathbb{Q}? Its only possible roots are \pm\sqrt 2,
neither rational, so it has no rational root; being degree 2, it is irreducible over
\mathbb{Q}. Over \mathbb{R} the roots
are present, so there it is reducible — the field matters.
How do we hunt for rational roots efficiently? For a polynomial with integer
coefficients there is a finite shortlist to check.
If f(x) = a_n x^n + \dots + a_1 x + a_0 has integer coefficients and a
rational root p/q in lowest terms, then:
- p divides the constant term a_0;
- q divides the leading coefficient a_n.
So every candidate rational root is \pm(\text{divisor of } a_0)/(\text{divisor of }
a_n) — a short list you can test by substitution.
For the monic cubic x^3 - 3x + 3 the only candidate rational roots
are \pm 1, \pm 3 (divisors of 3). Checking:
f(1) = 1, f(-1) = 5,
f(3) = 21, f(-3) = -15 — none is zero. No rational
root, degree 3, so x^3 - 3x + 3 is irreducible over
\mathbb{Q}. (We will confirm this a second way with Eisenstein below.)
The root test has a strict sightline: it works only for degrees 2 and 3. From degree 4
onward a polynomial can factor into two quadratics, each with no root of its own — so the whole
polynomial has no roots at all yet is still reducible. "No roots" no longer implies
irreducible.
The classic trap is x^4 + 4. It is strictly positive for every real
x, so it has no real root — and no rational root either. Tempting to call it
irreducible. But the Sophie Germain identity factors it:
x^4 + 4 = (x^2 - 2x + 2)(x^2 + 2x + 2).
Each quadratic factor has discriminant 4 - 8 = -4 < 0, so neither has a
real root — which is exactly why the rootless factorization slipped past the naked eye. Moral: from
degree 4 up you must rule out factorizations into quadratics (and higher), not just linear
factors. The heavier tools below — Gauss and Eisenstein — are what you reach for instead.
Seeing "no real roots" on a graph
Real roots are where a curve crosses the x-axis. Plot two polynomials and the
difference jumps out: x^2 - 2 dips below the axis and crosses it twice (at
\pm\sqrt 2 \approx \pm 1.41), while x^4 + 4 floats
entirely above the axis and never touches it — no real roots. Remember the warning above, though: the
graph proves x^4+4 has no real roots, not that it is irreducible. It
still factors into that pair of rootless quadratics.
Gauss's Lemma: test over ℤ, conclude over ℚ
Fractions make factoring feel harder than it is. Gauss's Lemma lets us throw the denominators away and
work with integers, where divisibility and primes give us leverage. Call a polynomial in
\mathbb{Z}[x] primitive if the greatest common divisor of its
coefficients is 1 (no prime divides all of them).
-
The product of two primitive polynomials is primitive. (The content is
multiplicative: no prime can appear in every coefficient of a product of primitives.)
-
Consequently a primitive polynomial in \mathbb{Z}[x] is
irreducible over \mathbb{Q} if and only if it is
irreducible over \mathbb{Z}. A rational factorization can
always be rescaled into an integer one.
-
Consequently, if R is a UFD then so is
R[x] — and by induction R[x_1,\dots,x_n].
The payoff is practical: to decide irreducibility over \mathbb{Q} — the hard,
fraction-filled field — you may instead factor over \mathbb{Z}, where you get
to use integer primes, divisibility and reduction mod p. Every criterion below
secretly rides on Gauss's Lemma to promote its "irreducible over \mathbb{Z}"
conclusion up to "irreducible over \mathbb{Q}."
One companion tool is the reduction-mod-p criterion: if a
monic integer polynomial stays the same degree and becomes irreducible over the finite field
\mathbb{F}_p for some prime p, then it was already
irreducible over \mathbb{Q}. (A factorization over
\mathbb{Z} would reduce to one over \mathbb{F}_p, so
no factorization can exist.) Because \mathbb{F}_p is finite you can, in
principle, check it by brute force.
Eisenstein's Criterion: a prime that guarantees irreducibility
The single most useful shortcut is Eisenstein's criterion. Find one prime that lines up
with the coefficients in the right way and irreducibility is instant — no root-hunting, no trial
factoring.
Let f(x) = a_n x^n + a_{n-1}x^{n-1} + \dots + a_1 x + a_0 have integer
coefficients. Suppose there is a prime p such that:
- p \nmid a_n — it does not divide the leading coefficient;
- p \mid a_i for every i < n — it divides
every other coefficient;
- p^2 \nmid a_0 — its square does not divide the constant term.
Then f is irreducible over \mathbb{Q}
(and, if primitive, over \mathbb{Z}).
Example 1 — the cubic again. For x^3 - 3x + 3 take
p = 3: it divides -3 and 3
(the lower coefficients), it does not divide the leading 1, and
p^2 = 9 does not divide the constant term 3. All
three boxes tick, so x^3 - 3x + 3 is irreducible over
\mathbb{Q} — the same verdict the root test gave, reached in one line.
Example 2 — a quintic. Take 2x^5 - 6x^3 + 9x^2 - 15 with
p = 3. The leading coefficient is 2, not divisible
by 3; the others are 0, -6, 9, 0, -15, all divisible
by 3; and p^2 = 9 does not divide the constant
-15. Eisenstein fires: irreducible over \mathbb{Q}.
Doing this by root-hunting would have been hopeless — a degree-5 polynomial can hide its
irreducibility behind quadratic and cubic factors.
Example 3 — the whole family x^n - p. For any prime
p, the polynomial x^n - p satisfies Eisenstein with
that very prime: it divides the constant -p (and every zero in between) but
not the leading 1, and p^2 \nmid p. So
x^n - p is irreducible over \mathbb{Q} for every
n — which is why \sqrt[n]{p} is an algebraic number
of degree exactly n.
Sometimes Eisenstein does not apply as written but works after a clever shift
x \mapsto x + 1 — irreducibility is unchanged by such a substitution, since it
just renames the variable. The showcase is the cyclotomic polynomial
\Phi_p(x) = 1 + x + x^2 + \dots + x^{p-1} = \frac{x^p - 1}{x - 1},
for a prime p. As written, Eisenstein is useless — the coefficients are all
1. But substitute x \mapsto x + 1 and, using the
Binomial Theorem,
\Phi_p(x+1) = \frac{(x+1)^p - 1}{x} = x^{p-1} + \binom{p}{1}x^{p-2} + \dots + \binom{p}{p-1}.
Every middle binomial coefficient \binom{p}{k} is divisible by
p (for 0 < k < p), the leading coefficient is
1, and the constant term is \binom{p}{p-1} = p,
which p^2 does not divide. Eisenstein fires — so
\Phi_p is irreducible over \mathbb{Q}. This one fact
is the cornerstone of the Galois theory of roots of unity.
The criterion is named for Gotthold Eisenstein (1823–1852), a dazzling talent Gauss
ranked beside Newton and Archimedes, who died of tuberculosis at just 29.
Here is the payoff that makes irreducibility worth all this machinery. Pick an irreducible polynomial
p(x) of degree n over
\mathbb{F}_p. The quotient ring
\mathbb{F}_p[x]/(p(x)) is then a field with exactly
p^n elements — the finite field
\mathrm{GF}(p^n). Irreducibility is precisely the condition that makes the
quotient a field rather than a ring with zero divisors: the ideal
(p(x)) is maximal exactly when p is irreducible.
The complex numbers are the same trick over \mathbb{R}:
\mathbb{C} = \mathbb{R}[x]/(x^2 + 1), where the class of
x becomes i. Each irreducible polynomial is a
blueprint for a new field, and stacking these extensions is exactly what Galois theory studies. No
irreducibles, no field extensions — which is why we spend a whole page learning to spot them.