Overview

  • How does absolute irreducibility differ from irreducibility over the ground field?
  • When can a polynomial be treated as a polynomial in one variable?
  • When do specialization or smoothness prove irreducibility?

Irreducibility and the ground field

Let KK be a field and let F∈K[x1,…,xr]F\in K[x_1,\ldots,x_r] be nonconstant. It is irreducible over KK if every factorization F=GHF=GH in this ring has a constant factor. It is absolutely irreducible if the same holds over an algebraic closure K‾\overline K.

The distinction matters even for quadratic polynomials. The polynomial x2+y2x^2+y^2 is irreducible over Q\mathbb Q, but over Q(i)\mathbb Q(i) it equals (x+iy)(x−iy)(x+iy)(x-iy). In contrast, y−x2y-x^2 is absolutely irreducible: its quotient ring is K‾[x,y]/(y−x2)≃K‾[x]\overline K[x,y]/(y-x^2)\simeq\overline K[x], a domain.

More generally, absolute irreducibility says that the hypersurface has an integral coordinate ring after extending the field. The geometric term is geometrically integral. This includes both irreducibility and reducedness. In characteristic pp, the polynomial xp−tx^p-t over K=Fp(t)K=\mathbb F_p(t) is irreducible but becomes (x−t1/p)p(x-t^{1/p})^p over K‾\overline K. Its zero set has one component, but its coordinate ring is not reduced.

References 2

One variable over a function field

Write F∈K[x1,…,xr−1][y]F\in K[x_1,\ldots,x_{r-1}][y]. Its content in yy is a greatest common divisor of its coefficients in K[x1,…,xr−1]K[x_1,\ldots,x_{r-1}]. If the content is a unit, FF is primitive in yy. Gauss’s lemma gives, for deg⁡yF>0\deg_yF>0,

F primitive⟹(F irreducible in K[x,y]  ⟺  F irreducible in K(x)[y]).F\text{ primitive}\quad\Longrightarrow\quad\bigl(F\text{ irreducible in }K[\mathbf x,y]\iff F\text{ irreducible in }K(\mathbf x)[y]\bigr).

This reduces many questions in several variables to familiar tests over a field. To prove absolute irreducibility, make the same argument over K‾(x)\overline K(\mathbf x). The coefficient field here is a rational function field; it is not the algebraic closure of that function field.

Example

If A,B∈K[x]A,B\in K[x], A≠0A\ne0, and gcd⁡(A,B)=1\gcd(A,B)=1, then A(x)y+B(x)A(x)y+B(x) is absolutely irreducible. The identity UA+VB=1UA+VB=1 persists over K‾\overline K, and a polynomial of degree one in yy is irreducible over K‾(x)\overline K(x). Thus (x2+1)y+x(x^2+1)y+x is absolutely irreducible over every field.

References 1

Eisenstein over a polynomial ring

The prime in Eisenstein’s criterion can be an irreducible polynomial. Let A=K[x1,…,xr−1]A=K[x_1,\ldots,x_{r-1}], let π∈A\pi\in A be irreducible, and write F=anyn+⋯+a0F=a_ny^n+\cdots+a_0, primitive in yy. If π∤an\pi\nmid a_n, π∣ai\pi\mid a_i for i<ni<n, and π2∤a0\pi^2\nmid a_0, then FF is irreducible in A[y]A[y].

Example

For n≥2n\ge2, the polynomial yn+xy+xy^n+xy+x is Eisenstein at xx over K‾[x]\overline K[x]. It is therefore absolutely irreducible in every characteristic.

References 1

Specializing a variable

Proposition

Let F∈K[x][y]F\in K[x][y] be primitive in yy, and let a∈Ka\in K. If its leading coefficient in yy does not vanish at aa and F(a,y)F(a,y) is irreducible in K[y]K[y], then FF is irreducible in K[x,y]K[x,y].

In a nontrivial factorization, primitivity forces both factors to have positive yy-degree. The product of their leading coefficients is the leading coefficient of FF, so neither leading coefficient vanishes at aa. Specialization would give two nonconstant factors of F(a,y)F(a,y), a contradiction.

Both hypotheses have a purpose. The polynomial (xy+1)(y+1)(xy+1)(y+1) specializes at x=0x=0 to the irreducible polynomial y+1y+1, because one factor loses degree. Likewise, (x+1)(y2+1)(x+1)(y^2+1) specializes to y2+1y^2+1 over Q\mathbb Q at x=0x=0, but is not primitive in yy.

For example, y2−xy^2-x is irreducible over Q[x,y]\mathbb Q[x,y] because its specialization at x=2x=2 is irreducible. Its absolute irreducibility follows instead from the linear-variable argument. Over an algebraically closed field, a specialized univariate polynomial of degree greater than one always factors.

References 1
  • Lang2002 · Chapter IV; the specialization proposition follows from Gauss’s lemmaSource

Smooth projective curves

For a bivariate polynomial FF of total degree dd, its homogenization is Fh(X,Y,Z)=ZdF(X/Z,Y/Z)F^{\mathrm h}(X,Y,Z)=Z^dF(X/Z,Y/Z). It defines the projective closure of the affine curve F=0F=0.

Proposition

If the projective plane curve Fh=0F^{\mathrm h}=0 is smooth over K‾\overline K, then FF is absolutely irreducible.

Two distinct projective plane components meet by Bézout’s theorem. At such an intersection, all partial derivatives of their product vanish. A repeated component also gives singular points. Thus a smooth projective plane curve has a single reduced component.

For char⁡K≠2\operatorname{char}K\ne2, the conic X2+Y2+Z2=0X^2+Y^2+Z^2=0 is smooth: its three partial derivatives cannot vanish at a projective point. Hence x2+y2+1x^2+y^2+1 is absolutely irreducible.

Affine smoothness alone does not suffice. The curve x(x−1)=0x(x-1)=0 consists of two disjoint smooth affine lines. Their projective closures meet at infinity. Conversely, an irreducible curve may be singular: y2−x3y^2-x^3 is irreducible over C\mathbb C and has a cusp at the origin.

References 1

Support, fibers, and factorization

Newton polytopes constrain factor supports. The fibers F−λF-\lambda connect irreducibility with polynomial composition. Factorization algorithms recover factors from coefficient equations, lifting, or differential identities.

These methods developed along different lines: elimination theory gives Noether’s coefficient criteria; Ostrowski’s product formula leads to the polytope method; the study of curves f(x)=g(y)f(x)=g(y) connects ramification and permutation groups with factorization. Modern algorithms often combine these viewpoints.

The following articles give the statements and examples: Newton polytopes, separated variables, polynomial fibers and Bertini’s theorem, and factorization in several variables.

References 3

Questions

Research directionSeveral variables

Which coefficient conditions imply absolute irreducibility when the Newton polytope decomposes?

Scope. A fixed multivariate support and a specified coefficient field

A decomposition of the polytope gives possible supports of factors, but does not determine whether those factors exist. For a specified family, one can study the reducibility locus in coefficient space or combine the support conditions with differential or lifting methods. The question concerns criteria adapted to that family, rather than a general factorization algorithm.

References 3
Discuss this question →

Related topics

References

  1. Geometrically integral algebras and schemes

    The Stacks Project Authors. The Stacks Project, 2026.

    Read the source
  2. Algebra

    Lang, Serge. Edition 3. Springer, Graduate Texts in Mathematics, vol. 211, 2002.

    Read the source
  3. Algebraic Curves: An Introduction to Algebraic Geometry

    Fulton, William. Author’s electronic edition, 2008.

    Read the source
  4. Decomposition of Polytopes and Polynomials

    Gao, Shuhong and Lauder, Alan G. B. Discrete & Computational Geometry, 26(1), 89–104, 2001.

    Read the source
  5. Effective Noether irreducibility forms and applications

    Kaltofen, Erich. Journal of Computer and System Sciences, 50(2), 274–295, 1995.

    Read the source
  6. Equations of the form f(x)=g(y)

    Davenport, Harold and Lewis, Donald J. and Schinzel, Andrzej. The Quarterly Journal of Mathematics, 12(1), 304–312, 1961.

    Read the source
  7. Factoring multivariate polynomials via partial differential equations

    Gao, Shuhong. Mathematics of Computation, 72(242), 801–822, 2003.

    Read the source