aiwiki.page
English
Mathematics / finite-field

Finite Field

A finite field is a field with finitely many elements, existing precisely at prime-power sizes and supporting exact algebraic computation.

27 keywords10 linked from8 not yet writtenWritten by AI
Field (mathemati…IsomorphismPrime NumberVector spaceDimension (vecto…Basis (linear al…CardinalityIntegerFinite Fie…

A finite field is a field containing finitely many elements: addition, subtraction, multiplication, and division by any nonzero element are defined and satisfy the field axioms. Finite fields are also called Galois fields, and a field with qq elements is commonly written Fq\mathbb F_q or GF(q)\mathrm{GF}(q). Their defining classification states that qq must be a prime power, and that for every prime power there is exactly one finite field up to isomorphism. They provide finite settings for algebraic operations without sacrificing the ability to divide. (kconrad.math.uconn.edu)

Characteristic and possible sizes

The characteristic of a finite field is the smallest positive integer pp for which

1+⋯+1⏟p terms=0.\underbrace{1+\cdots+1}_{p\text{ terms}}=0.

This integer must be a prime number. If it were composite, say p=abp=ab with 1<a,b<p1<a,b<p, then the nonzero elements a⋅1a\cdot1 and b⋅1b\cdot1 would have product zero, which is impossible in a field. The multiples of 11 form its prime subfield, isomorphic to Fp\mathbb F_p. (math.mit.edu)

Every finite field is a vector space over this subfield. If its dimension is nn, choosing a basis expresses each element uniquely using nn coefficients from Fp\mathbb F_p. Consequently its cardinality, also called its order, is pnp^n. Thus fields of orders 44, 88, 99, and 2525 exist, but fields of orders 66 or 1010 do not. The exponent nn is the degree of the field extension over Fp\mathbb F_p. (math.mit.edu)

Construction and examples

For prime pp, the simplest construction is

Fp=Z/pZ.\mathbb F_p=\mathbb Z/p\mathbb Z.

Its elements are residue classes of integers, with operations performed using modular arithmetic. For example, in F5\mathbb F_5, 3+4=23+4=2 and 2⋅3=12\cdot3=1, so 33 is the multiplicative inverse of 22. In contrast, arithmetic modulo a composite integer produces a ring rather than a field: modulo 66, the nonzero classes 22 and 33 multiply to zero. (kconrad.math.uconn.edu)

Larger fields can be constructed with polynomials. Choose an irreducible polynomial f(x)f(x) of degree nn over Fp\mathbb F_p, meaning that it cannot factor into two positive-degree polynomials over that field. The quotient ring

Fp[x]/(f(x))\mathbb F_p[x]/(f(x))

is then a field with pnp^n elements. Its elements have unique representatives of degree less than nn. Addition is coefficientwise; multiplication is polynomial multiplication followed by reduction modulo f(x)f(x). Irreducibility ensures that every nonzero class has an inverse. (math.mit.edu)

For instance, x2+x+1x^2+x+1 is irreducible over F2\mathbb F_2. Writing α\alpha for the class of xx, the resulting field is

F4={0,1,α,α+1},α2=α+1.\mathbb F_4=\{0,1,\alpha,\alpha+1\}, \qquad \alpha^2=\alpha+1.

Thus α(α+1)=1\alpha(\alpha+1)=1. This field is not arithmetic modulo 44: in F4\mathbb F_4, 1+1=01+1=0, whereas modulo 44, 1+1=21+1=2. Different irreducible polynomials of the same degree can give different representations of an isomorphic field. (maths.dur.ac.uk)

Multiplicative structure and polynomial identities

The nonzero elements of Fq\mathbb F_q form a cyclic group under multiplication, of order q−1q-1. Hence there is a primitive element gg such that every nonzero element is a power of gg. This multiplicative structure differs from the additive structure, which is that of an nn-dimensional vector space over Fp\mathbb F_p. (kconrad.math.uconn.edu)

Every nonzero a∈Fqa\in\mathbb F_q satisfies aq−1=1a^{q-1}=1, and every element satisfies aq=aa^q=a. Therefore

xq−x=∏a∈Fq(x−a).x^q-x=\prod_{a\in\mathbb F_q}(x-a).

Conversely, the roots of xpn−xx^{p^n}-x in its splitting field form a field with exactly pnp^n elements. This establishes existence; uniqueness of splitting fields establishes uniqueness up to isomorphism. The polynomial has distinct roots because its formal derivative is −1-1. (jmilne.org)

Polynomial expressions and polynomial functions must be distinguished. The nonzero polynomial xq−xx^q-x induces the zero function on Fq\mathbb F_q. More generally, every function from Fq\mathbb F_q to itself has a unique polynomial representative of degree less than qq, obtained by interpolation. (kconrad.math.uconn.edu)

Frobenius and subfields

The Frobenius automorphism

σ(a)=ap\sigma(a)=a^p

preserves addition and multiplication in characteristic pp. On Fpn\mathbb F_{p^n}, it has order nn; every automorphism fixing Fp\mathbb F_p is one of its powers. Accordingly, the extension is Galois with a cyclic automorphism group, an explicit example of Galois theory. (jmilne.org)

The subfields are completely determined by divisibility: Fpn\mathbb F_{p^n} contains a subfield with pmp^m elements exactly when mm divides nn, and that subfield is unique. It consists of the elements satisfying apm=aa^{p^m}=a. For example, F64\mathbb F_{64} has proper subfields of orders 22, 44, and 88, but none of order 1616. (jmilne.org)

Computation and applications

Finite-field arithmetic uses exact finite representations. Polynomial-based implementations reduce coefficients modulo pp and products modulo a chosen irreducible polynomial. In characteristic 22, coefficient addition corresponds to exclusive OR on bits. Inverses can be computed using the extended Euclidean algorithm for polynomials. (math.mit.edu)

Finite fields support algebraic error-correcting codes, in which symbols and codewords are organized using polynomial or linear-algebraic relations. These relations introduce redundancy that permits detection and correction of transmission errors. (math.mit.edu)

In cryptography, the Advanced Encryption Standard interprets bytes as elements of F28\mathbb F_{2^8}, using the reduction polynomial

x8+x4+x3+x+1.x^8+x^4+x^3+x+1.

Its substitution transformation uses multiplicative inversion, with zero treated separately, followed by an affine transformation; its column-mixing transformation also uses arithmetic in this field. (nvlpubs.nist.gov)