Motivation
A polynomial can be manipulated formally before assigning any value to its variable. That distinction matters over finite fields, where different polynomials may give the same function. Polynomial arithmetic provides the tools needed to construct new fields: an irreducible polynomial generates a maximal ideal, and the quotient turns its residue class into a root. Before studying extensions, we need reliable proofs of division, factorization, and irreducibility.
Here denotes a field and an integral domain when a more general coefficient ring is used. The earlier definitions of ideals, quotient rings, and maximality are prerequisites. Integer prime factorization and the integer Bézout theorem are used explicitly in the discussion of primitive polynomials.
Definitions
Definition 4.1.1
(Formal polynomials, degree, and divisibility)
A polynomial over is a finite formal sum . Equality means coefficientwise equality. The degree of a nonzero polynomial is the largest index of a nonzero coefficient; we leave the zero polynomial's degree undefined and treat it separately. A polynomial is monic if its leading coefficient is one. We write when for some polynomial in the stated coefficient ring.
For nonzero over a domain, their leading coefficients have nonzero product, so . Thus units of are exactly the nonzero constants. Two nonzero polynomials are associates if they differ by a unit. A nonconstant is irreducible when every factorization has a unit factor. An irreducibility statement always names the coefficient field: is irreducible over but factors over .
A greatest common divisor of nonzero divides both and is divisible by every common divisor. In we choose the unique monic gcd. This is a divisibility definition, not a comparison of numerical sizes or polynomial values.
Theorems
Theorem 4.1.2
(Division, Bézout, and principal ideals)
For with , there exist unique such that , with or . Every ideal of is principal. For nonzero , their monic gcd satisfies for some .
Theorem 4.1.3
(Irreducibles, fields, and unique factorization)
For nonconstant , irreducibility is equivalent to maximality of , and hence to being a field. Every nonzero nonunit polynomial in factors into irreducibles, uniquely up to reordering and multiplication of factors by units.
Proof sketch or proof idea
Proof
(Degree descent and a minimal-degree generator)
To divide, cancel the leading term of by a multiple of , using the inverse of 's leading coefficient. Repeat while the remainder degree is at least . Every cancellation strictly lowers the degree, so the process terminates. If two representations exist, . A nonzero left side has degree at least , while a nonzero right side has smaller degree. Therefore both are zero, giving uniqueness. When take .
For a nonzero ideal , choose a nonzero element of minimum degree, then rescale it to be monic. Dividing any by yields a remainder in of smaller degree, which must be zero. Hence ; the zero ideal is . Apply this to : its monic generator is a linear combination of and divides each. Every common divisor divides the combination, so is the gcd. Repeated division, the Euclidean algorithm, computes ; back-substitution computes its Bézout coefficients.
Proof
(Bézout inverses and Euclid’s lemma)
If is irreducible and , the gcd of is one. A Bézout identity gives an inverse to in the quotient, so the quotient is a field. Conversely, if with both factors nonunits, each has degree strictly between zero and . Their nonzero quotient classes have zero product, so the quotient is not a field. This proves the equivalence. It also proves Euclid's lemma: if is irreducible and , then or , because the quotient is a domain.
Existence of an irreducible factorization follows by induction on degree. An irreducible polynomial is already factored; a reducible nonconstant polynomial splits into two factors of smaller positive degree, to which induction applies. For uniqueness, an irreducible in one factorization divides the product in the other, hence divides one of its irreducible factors by repeated Euclid's lemma. The two factors are associates. Cancel them in the domain and repeat. Degrees ensure this process ends, leaving only a unit. Thus neither existence nor uniqueness was assumed in defining irreducibility.
Evaluation and the root bound
Theorem 4.1.4
(Factor theorem and root bound)
For and , exactly when . A nonzero polynomial of degree over a field has at most distinct roots in .
Proof
(Remainder at and induction on degree)
For , division by gives with constant remainder . Therefore exactly when . A nonzero degree- polynomial over a field has at most distinct roots: if is one root, factor off ; every different root is a root of , because can be canceled. Induction proves the bound.
Concept lensAlgebraic
The root bound as an injective evaluation map
Polynomials of degree at most form an -dimensional -space with basis . For distinct , the evaluation map has the Vandermonde matrix , whose determinant is . The map is therefore injective: a polynomial of degree at most vanishing at distinct points is zero, which is the root bound again. The argument needs the product of differences to be nonzero; in the points give , and indeed vanishes at all three.
For degrees two and three, reducibility is equivalent to having a root in , because a nontrivial factorization must contain a degree-one factor. For degree four this fails: is reducible over and has no real root. Testing roots alone is therefore not a general irreducibility algorithm.
Passing between integers and rationals
A nonzero integer polynomial is primitive if the gcd of its coefficients is one. Its content is the positive gcd of those coefficients. The product of two primitive integer polynomials is primitive. To prove this, suppose a prime divided every coefficient of the product. Reducing modulo would give a zero product in . Each factor reduces to a nonzero polynomial by primitivity, contradicting the degree formula in this domain.
Theorem 4.1.5
(Gauss's lemma and two irreducibility tests)
A primitive is reducible over exactly when it factors into two positive-degree polynomials in .
If its reduction modulo a prime has the same degree as and is irreducible in , then is irreducible over .
For , , if a prime does not divide , divides every for , and , then is irreducible over (Eisenstein's criterion).
Proof
(Content, reduction and monomials modulo )
For Gauss, any rational polynomial is a rational scalar times a primitive integer polynomial: clear denominators and remove the content. A rational factorization of primitive thus gives with primitive integer and . Their product is primitive. Write in lowest terms with . Since has integer coefficients, divides every coefficient of , so . Since is primitive, . Absorb the sign into one factor. The converse is immediate.
For reduction, an integer factorization supplied by Gauss would reduce to a nontrivial factorization modulo . Indeed the leading coefficient of is not divisible by , so neither leading coefficient of the factors is divisible by ; both degrees are preserved. This contradicts irreducibility of the reduction.
For Eisenstein, first remove content; its value is not divisible by because , so all three conditions persist. Gauss would give positive-degree integer factors . Modulo , their product is a nonzero constant times . Unique factorization in forces both reductions to be nonzero monomials, of their original positive degrees. Both constant coefficients are therefore divisible by , making , a contradiction. All three hypotheses of the criterion have now been used.
Proof X-ray
Where each Eisenstein hypothesis enters
The argument is a contradiction run through Gauss’s lemma and unique factorization modulo . The condition is used twice: it keeps the content prime to , and it makes the reduction of a nonzero constant times , so the reductions of and keep their positive degrees. The condition for is what makes that reduction a single monomial. Since is irreducible in , unique factorization then forces each reduced factor to be a nonzero constant times a positive power of . Hence both constant terms are divisible by , and is divisible by . This is the only use of , and it closes the argument.
A disciplined irreducibility workflow
For a polynomial over the rationals, first clear denominators and remove the content; these operations change only a nonzero scalar and therefore do not affect positive-degree factorizations. Record the degree and leading coefficient before reducing modulo a prime. If that coefficient becomes zero, the degree has dropped and the reduction theorem cannot be applied as stated. An unsuccessful reduction is not a conclusion; it is a reason to try another prime or another method.
For low degrees, the rational-root test supplies a finite list. If a primitive integer polynomial has a rational root in lowest terms with , multiplying the equation by shows that divides the constant coefficient and divides the leading coefficient. For the first assertion, move all terms containing to one side to see , then use ; the second follows similarly from . This proof uses integer prime factorization or Bézout to cancel coprime factors, rather than polynomial division over the integers.
For a monic integer polynomial, every rational root must consequently be an integer dividing its constant term. A cubic can therefore be proved irreducible by checking this finite list. A quartic may require reduction, Eisenstein, or an argument excluding quadratic factors. Once irreducibility has been established, the quotient-field conclusion is available; using that conclusion first would be circular.
For completeness, the claim that vanishes on follows from group theory already proved: every nonzero element lies in the multiplicative group of size , so Lagrange gives . Multiplying by gives , and zero also satisfies the equality. This supports the polynomial-versus-function counterexample without an unmentioned finite-field theorem.
Worked examples
Worked example 4.1.6
(An explicit inverse in a quadratic quotient)
Over , has no root and is irreducible. In , write , so . Every class has a unique representative . For its inverse is . The denominator is positive as a rational number, hence nonzero. The identity alone does not justify inverses; this calculation does.
Worked example 4.1.7
(A finite-field test)
For , reduction modulo two has values one at both zero and one. The cubic therefore has no root in , so it is irreducible there. Its degree is preserved and is primitive; hence is irreducible over . The quotient has exactly eight elements, since the unique remainders are with binary coefficients. This constructs one field of eight elements without asserting the full classification of finite fields.
Worked example 4.1.8
(Shifting before Eisenstein)
The polynomial does not directly satisfy Eisenstein for two, since its constant term is odd. But does. It is therefore irreducible over . Substitution is an automorphism of , with inverse , and preserves positive degrees, so irreducibility of the shifted polynomial proves irreducibility of itself.
History of an idea
An older remainder argument in a new coefficient system
In Elements VII.2, repeated subtraction finds a common measure (Joyce edition). Polynomial Euclid instead decreases degree through division. This compares remainder arguments: the ancient statement concerns numbers, and this passage makes no discovery-priority claim.
Common mistakes
Common mistake
(Division needs an invertible leading coefficient)
Division with a smaller-degree remainder relies on invertibility of the divisor's leading coefficient. In , one cannot divide by with an integer quotient and a constant remainder. Consequently is not covered by the principal-ideal proof for ; the preceding chapter gave the nonprincipal ideal .
Common mistake
(Irreducibility depends on the field)
An irreducible polynomial over one field may split over a larger field. An irreducible polynomial over need not have irreducible reduction at every prime; reduction is a sufficient test in the stated direction.
Common mistake
(A polynomial is not the function it defines)
is nonzero as a formal polynomial in , even though its value is zero at every field element. Polynomial equality and equality of functions are different assertions.
Counterexample mode
Eisenstein without the leading-coefficient condition
Drop the hypothesis and the criterion fails. Take and : the prime two divides every coefficient below the leading one, and . Yet is a factorization over into two polynomials of degree one, so is reducible. In the proof, the content is now divisible by , and removing it destroys the hypotheses: has a constant term not divisible by two. Equivalently, the reduction of modulo two is the zero polynomial rather than a nonzero constant times . With restored, reduction preserves degrees and the argument goes through.
Summary
Field coefficients permit division and Bézout identities. Those identities prove that irreducibles generate maximal ideals and satisfy Euclid's lemma, leading to unique factorization. Root bounds handle low degrees; Gauss, reduction, and Eisenstein handle further examples. Each method has a specific coefficient ring and explicit hypotheses, which must be checked before constructing a quotient field.
Exercises
Quick check
(Find a Bézout identity for and in .)
Solution · Answer
Division gives . Thus , proving the gcd is one and explicitly giving the coefficients.
Quick check
(Prove irreducible over . Explain why absence of rational roots alone would not suffice.)
Solution · Answer
Apply Eisenstein with : the leading coefficient is one, every other coefficient (including the zero coefficients) is divisible by three, and the constant three is not divisible by nine. The polynomial is irreducible. A quartic without rational roots could still factor into two quadratics, so the root test alone would leave a gap.
Quick check
(Factor over and explain why this does not contradict the worked example over .)
Solution · Answer
In characteristic two, and , so the factorization is . The reduction test only infers rational irreducibility from an irreducible degree-preserving reduction, not rational reducibility from a reducible reduction. The rational polynomial is irreducible by the shifted Eisenstein proof.
Quick check
(In , find the inverse of and the product .)
Solution · Answer
Since , characteristic two gives . Thus the inverse is . The requested product is .