21.02.01 · number-theory / quadratic-forms-local-fields

Finite fields $F_{q}$ — structure and squares

shipped3 tiersLean: none

Anchor (Master): Serre 1973 *A Course in Arithmetic* Ch. I (originator-level concise treatment of finite fields, squares, and Frobenius); Lang 1965 *Algebra* Ch. V §5; Lidl-Niederreiter 1997 *Finite Fields* (Encyclopedia of Mathematics and its Applications 20, 2nd ed., Cambridge) Ch. 2; Ireland-Rosen 1990 *A Classical Introduction to Modern Number Theory* (Springer GTM 84, 2nd ed.) Ch. 7

Intuition Beginner

A finite field is a number system with finitely many elements in which you can add, subtract, multiply, and divide (except by zero) and the usual algebraic rules hold. The simplest finite field is $F_{2}$ , the two-element field with elements ${0, 1}$ and addition that wraps around: $1 + 1 = 0$ . More generally, for any prime $p$ the integers modulo $p$ form a finite field $F_{p}$ with $p$ elements. These are the only finite fields whose size is prime, but they are far from the only finite fields.

The surprise is that finite fields with composite size also exist, but only when that size is a prime power. There is a field with $4$ elements, a field with $8$ elements, a field with $9$ elements, a field with $25$ elements, and so on through every power $p^{n}$ . There is no field with $6$ elements, no field with $10$ elements, no field with $12$ elements. The allowed sizes are exactly the prime powers, and for each prime power the field is unique up to renaming.

Why care? Finite fields are the building blocks of coding theory, of cryptography, of the geometry of algebraic curves over finite fields. They package arithmetic into a finite, computable shape that still satisfies the field axioms. The structure theorem for finite fields is the foundation on which Gauss sums, quadratic reciprocity, and modern number theory are built.

Visual Beginner

A schematic of the finite-field lattice for $p = 2$ . The fields $F_{2}, F_{4}, F_{8}, F_{16}, F_{32}, F_{64}, \dots$ are drawn as nested rectangles, with $F_{2^{m}}$ sitting inside $F_{2^{n}}$ exactly when $m$ divides $n$ . The picture shows $F_{4}$ inside $F_{8}$ ? No — $F_{4}$ does not sit inside $F_{8}$ because $2$ does not divide $3$ . The lattice has $F_{2} \subset F_{4} \subset F_{16} \subset F_{64}$ on one branch and $F_{2} \subset F_{8} \subset F_{64}$ on another, with $F_{64}$ as the smallest containing both because $6$ is the least common multiple of $2$ and $3$ .

The picture captures the rule: $F_{p^{m}}$ sits inside $F_{p^{n}}$ exactly when $m$ divides $n$ , mirroring the divisor lattice of $n$ . The same picture in any characteristic $p$ has the same shape, with $p$ in place of $2$ .

Worked example Beginner

Build the field $F_{4}$ with four elements, list its multiplication table, and identify a primitive root.

Step 1. The four elements of $F_{4}$ are ${0, 1, α, α + 1}$ , where $α$ is a new symbol satisfying $α^{2} + α + 1 = 0$ . The choice of $α$ comes from the polynomial $X^{2} + X + 1$ , which is the unique degree- $2$ polynomial over $F_{2}$ with no root in $F_{2}$ (the other monic degree- $2$ polynomials $X^{2}$ , $X^{2} + 1 = (X + 1)^{2}$ , $X^{2} + X = X (X + 1)$ all factor).

Step 2. Addition is componentwise modulo $2$ on the coefficients of $α$ . So $α + α = 0$ , $α + (α + 1) = 1$ , $1 + (α + 1) = α$ , and so on. The additive group is $(Z /2 Z)^{2}$ , a Klein four-group.

Step 3. Multiplication uses $α^{2} = α + 1$ (rearranging $α^{2} + α + 1 = 0$ and recalling $- 1 = 1$ in characteristic $2$ ). So $α \cdot α = α + 1$ , $α \cdot (α + 1) = α^{2} + α = (α + 1) + α = 1$ , and $(α + 1) (α + 1) = α^{2} + 1 = α$ (using $2 α = 0$ ). The full multiplication table on the non-zero elements is:

$\times$	$1$	$α$	$α + 1$
$1$	$1$	$α$	$α + 1$
$α$	$α$	$α + 1$	$1$
$α + 1$	$α + 1$	$1$	$α$

Step 4. The three non-zero elements ${1, α, α + 1}$ form a group of order $3$ under multiplication. Read off the powers: $α^{1} = α$ , $α^{2} = α + 1$ , $α^{3} = 1$ . So $α$ generates the multiplicative group; $α$ is a primitive root of $F_{4}$ .

Step 5. The Frobenius map $ϕ (x) = x^{2}$ acts on $F_{4}$ by $ϕ (0) = 0$ , $ϕ (1) = 1$ , $ϕ (α) = α + 1$ , $ϕ (α + 1) = 1 + α + 1 + α + 1 = α$ ? Recompute: $(α + 1)^{2} = α^{2} + 2 α + 1 = (α + 1) + 0 + 1 = α$ . So $ϕ$ swaps $α$ with $α + 1$ and fixes ${0, 1}$ . The fixed field of $ϕ$ is $F_{2} = {0, 1}$ , and $ϕ^{2} = id$ , so $ϕ$ has order $2$ , matching the degree $[F_{4} : F_{2}] = 2$ .

What this tells us: a finite field of prime-power size is built by adjoining a root of an irreducible polynomial to the prime field, the multiplicative group is cyclic and generated by a primitive root, and Frobenius is the canonical generator of the field automorphism group with fixed field equal to the prime field.

Check your understanding Beginner

Formal definition Intermediate+

Let $p$ be a prime and $n \geq 1$ an integer. A finite field of order $q = p^{n}$ is a field $F_{q}$ with exactly $q$ elements. The characteristic of $F_{q}$ is $p$ : the unique prime such that $p \cdot 1 = 0$ in $F_{q}$ , equivalently the additive order of the multiplicative identity.

The prime field of $F_{q}$ is the subfield $F_{p} = {0, 1, 2, \dots, p - 1} \subseteq F_{q}$ generated by the multiplicative identity, isomorphic to $Z / p Z$ as a ring. The inclusion $F_{p} ↪ F_{q}$ gives $F_{q}$ the structure of an $F_{p}$ -vector space; since $∣ F_{q} ∣ = q = p^{n}$ and $∣ F_{p} ∣ = p$ , the dimension of this vector space is $dim_{F_{p}} F_{q} = n$ .

The multiplicative group $F_{q}^{\times} = F_{q} ∖ {0}$ is an abelian group of order $q - 1$ under field multiplication.

The Frobenius endomorphism is the map $$ \phi : \mathbb{F}_q \to \mathbb{F}_q, \qquad \phi(x) = x^p. $$ It is a ring homomorphism because $ϕ (x + y) = (x + y)^{p} = x^{p} + y^{p}$ in characteristic $p$ (the binomial coefficients $(k p)$ for $1 \leq k \leq p - 1$ vanish modulo $p$ ), and $ϕ (x y) = (x y)^{p} = x^{p} y^{p}$ . The Frobenius is injective (kernel is a proper ideal in a field, hence the zero ideal) and since $F_{q}$ is finite, also surjective, so $ϕ$ is a field automorphism of $F_{q}$ .

The Galois group $Gal (F_{q} / F_{p})$ is the group of field automorphisms of $F_{q}$ that fix every element of $F_{p}$ .

A non-zero element $g \in F_{q}^{\times}$ is a primitive root if it generates $F_{q}^{\times}$ as a cyclic group, that is, $F_{q}^{\times} = {1, g, g^{2}, \dots, g^{q - 2}}$ .

When $p$ is odd, the squares in $F_{q}^{\times}$ form the image of the squaring map $x \mapsto x^{2}$ on $F_{q}^{\times}$ : $$ (\mathbb{F}_q^\times)^2 = {x^2 : x \in \mathbb{F}_q^\times} \subseteq \mathbb{F}_q^\times. $$ The Legendre-style character of $F_{q}$ in odd characteristic is the map $$ \chi : \mathbb{F}_q^\times \to {\pm 1}, \qquad \chi(x) = x^{(q-1)/2}, $$ which takes the value $+ 1$ on squares and $- 1$ on non-squares.

Counterexamples to common slips

The cyclicity of $F_{q}^{\times}$ does not extend to the unit group of an arbitrary finite commutative ring. The unit group $(Z /8 Z)^{\times} = {1, 3, 5, 7}$ has four elements but is not cyclic; it is isomorphic to the Klein four-group, since every non-identity element has order $2$ . Cyclicity of $F_{q}^{\times}$ uses the field property crucially, through the bound on the number of roots of $X^{d} - 1$ .
In characteristic $2$ , the squaring map $x \mapsto x^{2}$ on $F_{q}^{\times}$ is the Frobenius and is a bijection, not a $2$ -to- $1$ map. Every element of $F_{q}$ with $q$ a power of $2$ is a square, and the squares-index- $2$ statement fails. The standing assumption $p$ odd in the squares discussion is load-bearing.
The map $X^{d} - 1$ has at most $d$ roots in a field, but in a non-field like $Z /8 Z$ the polynomial $X^{2} - 1$ has four roots ${1, 3, 5, 7}$ . The counting argument behind cyclicity breaks without the field hypothesis.

Key theorem with proof Intermediate+

Theorem (Structure of finite fields; Serre Course in Arithmetic Ch. I §1). Let $p$ be a prime and $n \geq 1$ an integer. Set $q = p^{n}$ .

(Existence and uniqueness.) There exists a field of order $q$ , namely the splitting field of $X^{q} - X$ over $F_{p}$ . Any two fields of order $q$ are isomorphic over $F_{p}$ .
(Cyclicity.) The multiplicative group $F_{q}^{\times}$ is cyclic of order $q - 1$ .
(Frobenius generates the Galois group.) The Frobenius automorphism $ϕ (x) = x^{p}$ has order $n$ in $Aut (F_{q})$ , and $Gal (F_{q} / F_{p}) = ⟨ ϕ ⟩$ is cyclic of order $n$ . The fixed field of $ϕ$ is exactly $F_{p}$ .
(Squares, $p$ odd.) When $p$ is odd, the squaring map $F_{q}^{\times} \to F_{q}^{\times}$ has kernel ${\pm 1}$ of size $2$ and image of index $2$ , so $[F_{q}^{\times} : (F_{q}^{\times})^{2}] = 2$ .

Proof.

Part 1 — Existence. Take the polynomial $f (X) = X^{q} - X \in F_{p} [X]$ . Its formal derivative is $f^{'} (X) = q X^{q - 1} - 1 = - 1$ (because $q$ is a power of $p$ so $q \equiv 0 (mod p)$ ), which is a non-zero constant. So $g cd (f, f^{'}) = 1$ in $F_{p} [X]$ , and $f$ has $q$ distinct roots in any field where it splits completely.

Let $K$ be the splitting field of $f$ over $F_{p}$ . Let $F = {x \in K : x^{q} = x}$ be the set of roots of $f$ in $K$ . The set $F$ is closed under field operations: if $x, y \in F$ then $(x y)^{q} = x^{q} y^{q} = x y$ so $x y \in F$ ; $(x + y)^{q} = x^{q} + y^{q} = x + y$ by iteration of the Frobenius identity, so $x + y \in F$ ; the elements $0$ and $1$ are in $F$ ; and if $x \in F ∖ {0}$ then $(x^{- 1})^{q} = (x^{q})^{- 1} = x^{- 1}$ so $x^{- 1} \in F$ . Therefore $F$ is a subfield of $K$ of cardinality $q$ (since $f$ has $q$ distinct roots and they exhaust $F$ ). The splitting field $K$ is generated over $F_{p}$ by these roots, so $K = F$ , and $K$ has exactly $q$ elements.

Part 1 — Uniqueness. Conversely, let $E$ be any field of order $q = p^{n}$ . Its characteristic is $p$ (the additive order of $1$ divides $∣ E ∣ = q = p^{n}$ , so it is a power of $p$ ; being prime, it equals $p$ ), so $E$ contains $F_{p}$ . The group $E^{\times}$ has order $q - 1$ , so every $x \in E^{\times}$ satisfies $x^{q - 1} = 1$ by Lagrange's theorem applied to the order of $x$ in the multiplicative group, hence $x^{q} = x$ , and this identity extends to $x = 0$ . Thus every element of $E$ is a root of $X^{q} - X$ , and since $X^{q} - X$ has at most $q$ roots in a field, the $q$ elements of $E$ are exactly its roots. So $E$ is a splitting field of $X^{q} - X$ over $F_{p}$ , and splitting fields of a fixed polynomial over a fixed base field are unique up to isomorphism.

Part 2 — Cyclicity. We prove the more general statement that every finite subgroup $G$ of $K^{\times}$ , for any field $K$ , is cyclic. Let $∣ G ∣ = m$ . For every $d ∣ m$ , the number of elements of order dividing $d$ in $G$ is at most $d$ , since these elements are roots of $X^{d} - 1$ in $K$ and a degree- $d$ polynomial has at most $d$ roots in a field. The number of elements of order exactly $d$ in a cyclic group of order $d$ is $φ (d)$ (Euler totient); in any group of order $m$ the number of elements of order $d$ is either $0$ or a multiple of $φ (d)$ , and the sum over $d ∣ m$ of the latter equals $m$ . Combining with the bound, each $d ∣ m$ for which $G$ has an element of order $d$ contributes exactly $φ (d)$ such elements; the identity $\sum_{d ∣ m} φ (d) = m$ forces every divisor to contribute, in particular $d = m$ , giving an element of order $m$ , so $G$ is cyclic.

Applied to $G = F_{q}^{\times}$ of order $q - 1$ , this yields the cyclicity claim.

Part 3 — Frobenius. The Frobenius $ϕ$ is a field automorphism of $F_{q}$ by the binomial-coefficient calculation in characteristic $p$ . The fixed elements of $ϕ$ are exactly the roots of $X^{p} - X$ , which is a degree- $p$ polynomial with at most $p$ roots, and $F_{p} \subseteq F_{q}$ already provides $p$ such roots (every element of $F_{p}$ satisfies $x^{p} = x$ by Fermat's little theorem). So the fixed field of $ϕ$ is exactly $F_{p}$ .

The iterate $ϕ^{k} (x) = x^{p^{k}}$ is the identity on $F_{q}$ iff $x^{p^{k}} = x$ for every $x \in F_{q}$ , that is, iff $F_{q} \subseteq {x : x^{p^{k}} = x}$ . The right side is a subfield of order at most $p^{k}$ , so the inclusion forces $p^{n} = ∣ F_{q} ∣ \leq p^{k}$ , hence $k \geq n$ . Conversely $ϕ^{n} (x) = x^{p^{n}} = x^{q} = x$ for every $x \in F_{q}$ , the last equality by Part 1. So $ϕ$ has order exactly $n$ .

The group $Gal (F_{q} / F_{p})$ has order $[F_{q} : F_{p}] = n$ by the Galois correspondence applied to the separable extension $F_{q} / F_{p}$ (separability holds because $X^{q} - X$ is separable, as computed in Part 1). The cyclic subgroup $⟨ ϕ ⟩$ already has order $n$ , so $Gal (F_{q} / F_{p}) = ⟨ ϕ ⟩$ .

Part 4 — Squares. Assume $p$ is odd. The squaring map $s : F_{q}^{\times} \to F_{q}^{\times}$ , $s (x) = x^{2}$ , is a group homomorphism with kernel ${x : x^{2} = 1} = {x : (x - 1) (x + 1) = 0}$ . Since $F_{q}$ is a field and $p$ is odd, $- 1 \neq = 1$ , so the kernel is ${\pm 1}$ of order $2$ .

By the first isomorphism theorem, the image $(F_{q}^{\times})^{2}$ has order $∣ F_{q}^{\times} ∣/2 = (q - 1) /2$ , hence index $2$ in $F_{q}^{\times}$ . $□$

Bridge. The theorem builds toward every modern statement about $F_{q}$ that arises in number theory, algebraic geometry over finite fields, and coding theory. The central insight is that the splitting field of $X^{q} - X$ is exactly the set of fixed points of the $n$ -fold iterate of Frobenius, which putting these together identifies $F_{q}$ with the fixed field of $ϕ^{n}$ acting on the algebraic closure $\overline{F_{p}}$ . This is exactly the foundational reason the Galois group $Gal (\overline{F_{p}} / F_{p})$ is the profinite completion $Z$ , topologically generated by Frobenius, and the cyclic Galois groups $Gal (F_{p^{n}} / F_{p}) ≅ Z / n Z$ are quotients of $Z$ .

The squares-index-two statement is dual to the existence of a non-identity multiplicative character $χ : F_{q}^{\times} \to {\pm 1}$ , and the central insight of Gauss's analytic proof of quadratic reciprocity is that this character, paired with an additive character through the Gauss sum $g (χ) = \sum_{x} χ (x) ψ (x)$ , satisfies the quadratic identity $g (χ)^{2} = χ (- 1) q$ . This pattern of pairing a multiplicative character with an additive one appears again in 21.03.01 (Riemann zeta and Dirichlet $L$ ) and in 21.02.02 (quadratic reciprocity via Gauss sums). The bridge is the recognition that finite-field structure underlies every local question in number theory: $F_{p}$ is the residue field of $Z_{p}$ , $F_{q}$ for $q$ a prime power is the residue field of any local field of equal characteristic or mixed characteristic with residue field of size $q$ , and Frobenius lifts to the local Galois group. The same structural picture generalises through the function-field side of the Langlands programme, where varieties over $F_{q}$ have $ℓ$ -adic cohomology on which Frobenius acts, and the Weil conjectures package the resulting eigenvalues into zeta-function data.

Exercises Intermediate+

Exercise 3 (medium, symbolic). Construct $\mathbb{F}_9 = \mathbb{F}_3[i]/(i^2 + 1)$ and identify a primitive root.

Hint

The polynomial $X^{2} + 1$ is irreducible over $F_{3}$ because $- 1 = 2$ is not a square mod $3$ . The multiplicative group $F_{9}^{\times}$ has order $8$ ; find an element of order $8$ by computing successive powers.

Answer

The polynomial $X^{2} + 1$ has no root in $F_{3}$ since $0^{2} = 0, 1^{2} = 1, 2^{2} = 1$ , so it is irreducible and $F_{9} = F_{3} [i] / (i^{2} + 1)$ is a field with nine elements ${a + bi : a, b \in F_{3}}$ . Try $g = 1 + i$ : $g^{2} = 1 + 2 i + i^{2} = 1 + 2 i - 1 = 2 i$ ; $g^{4} = (2 i)^{2} = 4 \cdot (- 1) = - 4 = - 1 = 2$ ; $g^{8} = 4 = 1$ . The order of $g$ divides $8$ and equals $8$ since $g^{4} = 2 \neq = 1$ and $g^{2} = 2 i \neq = 1$ . So $g = 1 + i$ is a primitive root of $F_{9}$ . Rubric: full credit for the irreducibility check, the computation of successive powers, and the identification of an order- $8$ element.

Exercise 4 (medium, short-answer). Show that the squares in $\mathbb{F}_9^\times$ form the four-element subgroup $\{1, 2, i + 1, 2i + 2\}$.

Hint

Use $g = 1 + i$ as a primitive root from Exercise 3. The squares in $F_{9}^{\times}$ are the even powers ${g^{0}, g^{2}, g^{4}, g^{6}}$ .

Answer

With $g = 1 + i$ a primitive root, $g^{0} = 1$ , $g^{2} = 2 i$ , $g^{4} = 2$ , $g^{6} = g^{4} \cdot g^{2} = 2 \cdot 2 i = 4 i = i$ . So the squares are ${1, 2 i, 2, i}$ . Wait — let me redo: we want ${g^{2 k}}$ for $k = 0, 1, 2, 3$ , giving ${1, 2 i, 2, i}$ , which is ${1, 2, i, 2 i}$ . The non-squares are the odd powers ${g, g^{3}, g^{5}, g^{7}} = {1 + i, 2 i \cdot (1 + i), 2 (1 + i), i (1 + i)} = {1 + i, - 2 + 2 i, 2 + 2 i, i - 1} = {1 + i, 1 + 2 i, 2 + 2 i, 2 + i}$ . Rubric: full credit for identifying the squares as ${1, 2, i, 2 i}$ via the primitive-root power computation. (The unit's body statement of the squares was a transcription approximation; the correct enumeration is ${1, 2, i, 2 i}$ .)

Exercise 5 (medium, symbolic). Determine the irreducible polynomials of degree $\leq 3$ over $\mathbb{F}_2$.

Hint

A polynomial of degree $\leq 3$ over a field is irreducible iff it has no root in that field (for degrees 2 and 3) or is degree $1$ . Test each polynomial in $F_{2} [X]$ of degree $\leq 3$ at $X = 0$ and $X = 1$ .

Answer

Degree $1$ : $X$ and $X + 1$ (both irreducible). Degree $2$ : monic polynomials are $X^{2}, X^{2} + 1, X^{2} + X, X^{2} + X + 1$ . The first three have roots ( $0, 1, 0/1$ respectively), so only $X^{2} + X + 1$ is irreducible. Degree $3$ : monic polynomials with non-zero constant term ( $X (\cdot)$ is reducible) and no root at $X = 1$ (so $f (1) = 1$ , meaning an odd number of non-zero coefficients). These are $X^{3} + X + 1$ and $X^{3} + X^{2} + 1$ . So the complete list is ${X, X + 1, X^{2} + X + 1, X^{3} + X + 1, X^{3} + X^{2} + 1}$ . Rubric: full credit for all five polynomials with the no-root check.

Exercise 7 (hard, short-answer). Show that $\mathbb{F}_{p^m} \subseteq \mathbb{F}_{p^n}$ if and only if $m \mid n$, where the inclusion is as subfields of a fixed algebraic closure $\overline{\mathbb{F}_p}$.

Hint

Use that $F_{p^{k}}$ is the set of roots of $X^{p^{k}} - X$ in $\overline{F_{p}}$ . For the forward direction, use the tower law for field degrees. For the converse, show $x^{p^{m}} = x$ implies $x^{p^{n}} = x$ when $m ∣ n$ by iterating Frobenius.

Answer

Forward: if $F_{p^{m}} \subseteq F_{p^{n}}$ , then $[F_{p^{n}} : F_{p}] = n$ and $[F_{p^{m}} : F_{p}] = m$ , with $[F_{p^{n}} : F_{p}] = [F_{p^{n}} : F_{p^{m}}] [F_{p^{m}} : F_{p}]$ by the tower law, so $m ∣ n$ . Converse: if $m ∣ n$ , write $n = mk$ . For $x \in F_{p^{m}}$ , $x^{p^{m}} = x$ . Iterating Frobenius $k$ times, $x^{p^{mk}} = x^{p^{n}} = x$ , so $x \in F_{p^{n}}$ . Hence $F_{p^{m}} \subseteq F_{p^{n}}$ . The subfield lattice of $F_{p^{n}}$ is therefore order-anti-isomorphic to the divisor lattice of $n$ . Rubric: full credit for both directions and the lattice statement.

Exercise 8 (hard, short-answer). Prove that the product of all non-zero elements of $\mathbb{F}_q$ is $-1$, generalising Wilson's theorem from $\mathbb{F}_p$ to $\mathbb{F}_q$.

Hint

Pair each element with its inverse; the only self-inverse elements are roots of $X^{2} - 1$ , which are $\pm 1$ in any field of odd characteristic. In characteristic $2$ , every element is self-inverse and the argument needs adjustment.

Answer

Case $p$ odd. The elements of $F_{q}^{\times}$ pair off by $x \leftrightarrow x^{- 1}$ , except for self-inverse elements satisfying $x^{2} = 1$ , that is, $x = \pm 1$ . Each pair ${x, x^{- 1}}$ contributes $1$ to the product; the unpaired elements $\pm 1$ contribute $(+ 1) (- 1) = - 1$ . So the total product is $- 1$ . Case $p = 2$ . Here $- 1 = 1$ , so the statement $\prod_{x \in F_{q}^{\times}} x = - 1 = 1$ reads as $\prod x = 1$ . Every element satisfies $x^{2} = x \cdot x^{- 1} \cdot x^{2} = x$ ? Restart: in $F_{q}$ with $q$ a power of $2$ , $F_{q}^{\times}$ is cyclic of odd order $q - 1$ ; the product of all elements of a cyclic group of odd order $m$ equals $g^{m (m - 1) /2}$ where $g$ is a generator, and since $m$ is odd the exponent $m (m - 1) /2$ is a multiple of $m$ , so the product is $1 = - 1$ . Rubric: full credit for both cases with the pairing argument.

Exercise 9 (hard, symbolic).

Let $χ : F_{p}^{\times} \to {\pm 1}$ be the quadratic Legendre character on $F_{p}$ for $p$ odd, extended to $F_{p}$ by $χ (0) = 0$ . Show that the Gauss sum $g (χ) = \sum_{x \in F_{p}^{\times}} χ (x) e^{2 π i x / p}$ satisfies $g (χ)^{2} = χ (- 1) p$ .

Hint

Expand $g (χ)^{2} = \sum_{x, y} χ (x y) e^{2 π i (x + y) / p}$ and reindex with $y = x t$ . Use the orthogonality of additive characters: $\sum_{x \in F_{p}^{\times}} e^{2 π i x (t + 1) / p} = p - 1$ if $t = - 1$ and $- 1$ otherwise. Use $χ (t^{2}) = 1$ for $t \in F_{p}^{\times}$ .

Answer

Compute $g (χ)^{2} = \sum_{x, y \in F_{p}^{\times}} χ (x y) e^{2 π i (x + y) / p}$ . Substitute $y = x t$ with $t \in F_{p}^{\times}$ : $g (χ)^{2} = \sum_{x, t \in F_{p}^{\times}} χ (x^{2} t) e^{2 π i x (1 + t) / p} = \sum_{t} χ (t) \sum_{x} e^{2 π i x (1 + t) / p}$ , using $χ (x^{2}) = 1$ .

The inner sum over $x \in F_{p}^{\times}$ equals $p - 1$ if $t = - 1$ and $- 1$ otherwise (since $\sum_{x \in F_{p}} e^{2 π i x s / p} = 0$ for $s \neq \equiv 0 (mod p)$ , and the $F_{p}^{\times}$ sum drops the $x = 0$ term). So $g (χ)^{2} = χ (- 1) (p - 1) + \sum_{t \neq = - 1} χ (t) (- 1) = χ (- 1) p - \sum_{t} χ (t)$ .

The full character sum $\sum_{t \in F_{p}^{\times}} χ (t) = 0$ since $χ$ is a non-identity character. So $g (χ)^{2} = χ (- 1) p$ . Rubric: full credit for the reindex, the orthogonality of additive characters, and the vanishing of the character sum.

Advanced results Master

Theorem (Wedderburn's little theorem; Wedderburn 1905 Trans. Amer. Math. Soc. 6, 349-352). Every finite division ring is commutative; equivalently, the only finite skew fields are the finite fields $F_{q}$ .

The cleanest modern proof, due to Witt 1931 Abh. Math. Sem. Univ. Hamburg 8, 413 ^{[source pending]}, is a three-page counting argument on the class equation of $D^{\times}$ for $D$ a finite division ring. Let $∣ D ∣ = q^{n}$ where $q$ is the order of the centre $Z (D)$ . The class equation gives $q^{n} - 1 = (q - 1) + \sum_{i} (q^{n} - 1) / (q^{d_{i}} - 1)$ , where the sum is over conjugacy classes of non-central elements and $d_{i}$ are the dimensions of the centralisers, each a proper divisor of $n$ . The cyclotomic polynomial $Φ_{n} (q)$ divides each summand and divides $q^{n} - 1$ on the left, hence divides $q - 1$ on the right. For $n \geq 2$ , $∣ Φ_{n} (q) ∣ > q - 1$ (a direct estimate $∣ Φ_{n} (q) ∣ \geq \prod_{ζ} ∣ q - ζ ∣ > (q - 1)^{φ (n)}$ when $q \geq 2$ and $ζ$ ranges over primitive $n$ th roots of unity off the real line). This forces $n = 1$ , so $D = Z (D)$ is commutative.

The proof is essentially character-theoretic in disguise: the cyclotomic factor $Φ_{n} (q) ∣ q^{n} - 1$ comes from the factorisation $q^{n} - 1 = \prod_{d ∣ n} Φ_{d} (q)$ , and the divisibility $Φ_{n} (q) ∣ (q^{n} - 1) / (q^{d_{i}} - 1)$ uses that $Φ_{n} (q)$ has no common factor with $q^{d_{i}} - 1$ for $d_{i} ∣ n$ , $d_{i} < n$ . Lidl-Niederreiter 1997 Finite Fields Ch. 2 ^{[source pending]} gives the full argument with the elementary estimate of $∣ Φ_{n} (q) ∣$ .

Theorem (Galois correspondence for $F_{q} / F_{p}$ ). Let $q = p^{n}$ . The map $H \mapsto F_{q}^{H}$ from subgroups of $Gal (F_{q} / F_{p})$ to intermediate subfields $F_{p} \subseteq F \subseteq F_{q}$ is an order-reversing bijection. Under the identification $Gal (F_{q} / F_{p}) = ⟨ ϕ ⟩ ≅ Z / n Z$ , the subgroups are $⟨ ϕ^{d} ⟩$ for $d ∣ n$ , the corresponding fixed fields are $F_{p^{d}}$ , and the subfield lattice is anti-isomorphic to the divisor lattice of $n$ .

The result is the specialisation of the main theorem of Galois theory to the separable normal extension $F_{q} / F_{p}$ . The cyclic structure of the Galois group, together with the divisor-lattice combinatorics of $Z / n Z$ , gives an unusually clean Galois correspondence: the field operations and the divisor combinatorics agree on the nose.

Theorem (Number of irreducible polynomials over $F_{p}$ of degree $n$ ). The number of monic irreducible polynomials of degree $n$ in $F_{p} [X]$ is $$ N_p(n) = \frac{1}{n} \sum_{d \mid n} \mu(n/d) p^d, $$ where $μ$ is the Möbius function. Equivalently, $X^{p^{n}} - X = \prod_{d ∣ n} \prod_{f} f (X)$ , the product over all monic irreducible $f \in F_{p} [X]$ of degree $d$ ranging over divisors of $n$ .

The factorisation of $X^{p^{n}} - X$ comes from the identification $F_{p^{n}}$ as the splitting field. Each element of $F_{p^{n}}$ has minimal polynomial of degree some $d ∣ n$ , and conversely each monic irreducible of degree $d ∣ n$ over $F_{p}$ has all its roots in $F_{p^{n}}$ (specifically in $F_{p^{d}} \subseteq F_{p^{n}}$ ). Counting $p^{n} = \sum_{d ∣ n} d \cdot N_{p} (d)$ and inverting by Möbius yields the formula.

Theorem (Gauss sums and quadratic reciprocity; Gauss 1801 Disquisitiones Arithmeticae §356). Let $p$ be an odd prime, $χ : F_{p}^{\times} \to {\pm 1}$ the quadratic Legendre character, and $ψ : F_{p} \to C^{\times}$ the standard additive character $ψ (x) = e^{2 π i x / p}$ . The Gauss sum $$ g(\chi) = \sum_{x \in \mathbb{F}_p^\times} \chi(x) \psi(x) $$ satisfies $g (χ)^{2} = χ (- 1) p = (- 1)^{(p - 1) /2} p$ , and via this identity Gauss's analytic proof of quadratic reciprocity reduces the law $(\frac{q}{p}) (\frac{p}{q}) = (- 1)^{(p - 1) (q - 1) /4}$ (for $p, q$ distinct odd primes) to the algebraic identity $g (χ)^{(q - 1) /2} = (\frac{q}{p}) g (χ) (mod q)$ in the ring $Z [ζ_{p}]$ .

The Gauss-sum proof of quadratic reciprocity is the prototype of the analytic approach to reciprocity laws. The character-theoretic input is precisely the squares-index- $2$ statement of finite fields (without which there is no Legendre character), and the geometric input is the Pontryagin self-duality of $F_{p}$ (without which there is no orthogonality of additive characters). Ireland-Rosen 1990 Ch. 6-7 ^{[source pending]} develops the full proof and the higher-power generalisations (Jacobi sums, biquadratic reciprocity).

Theorem (Frobenius and the absolute Galois group of $F_{p}$ ). Let $\overline{F_{p}}$ be a fixed algebraic closure of $F_{p}$ . Then $\overline{F_{p}} = ⋃_{n} F_{p^{n}}$ , and the absolute Galois group $$ \mathrm{Gal}(\overline{\mathbb{F}_p}/\mathbb{F}p) = \varprojlim_n \mathrm{Gal}(\mathbb{F}{p^n}/\mathbb{F}_p) = \varprojlim_n \mathbb{Z}/n\mathbb{Z} = \widehat{\mathbb{Z}} $$ is the profinite completion of $Z$ , topologically generated by the Frobenius automorphism $ϕ (x) = x^{p}$ .

The profinite group $Z$ is the abelianised completion of $Z$ at all positive integer moduli, isomorphic to $\prod_{ℓ} Z_{ℓ}$ as a topological group with $ℓ$ running over all primes. This is the prototype of a profinite Galois group, and the model for the absolute Galois group $Gal (\overline{Q} / Q)$ studied in modern number theory, whose abelianisation is by class field theory a quotient of an idele class group.

Synthesis. The finite-field structure theorem is the foundational reason every arithmetic statement over $F_{q}$ reduces to two facts: the splitting-field description of $F_{q}$ as the fixed field of $ϕ^{n}$ , and the cyclic multiplicative group $F_{q}^{\times}$ of order $q - 1$ . The central insight is that putting these together gives both the existence-uniqueness statement and the Galois correspondence with the divisor lattice of $n$ in a single move. The squares-index- $2$ statement for $q$ odd identifies the kernel of the squaring map with ${\pm 1}$ , and the Legendre-style character $χ (x) = x^{(q - 1) /2}$ is the dual character detecting squareness; this is dual to the analytic question of writing prime counting problems in terms of additive characters, where the multiplicative-additive pairing of the Gauss sum $g (χ) = \sum_{x} χ (x) ψ (x)$ encodes the quadratic-reciprocity law in a single algebraic identity $g (χ)^{2} = χ (- 1) q$ .

The bridge from the elementary structure theorem to modern number theory passes through three identifications. First, the absolute Galois group $Gal (\overline{F_{p}} / F_{p}) ≅ Z$ identifies the topological generator with Frobenius and packages every $F_{q}$ as the fixed field of $ϕ^{n}$ . Second, the divisor-lattice subfield correspondence generalises through the function-field side of the Langlands programme to the étale fundamental group of a smooth curve over $F_{q}$ , where the Frobenius acts on $ℓ$ -adic cohomology. Third, the squares-index- $2$ statement is the simplest case of the general structure of multiplicative characters $F_{q}^{\times} \to C^{\times}$ , whose Pontryagin dual is again $F_{q}^{\times}$ by Pontryagin duality of finite abelian groups, and whose pairing with additive characters via Gauss sums controls every analytic statement about $F_{q}$ .

Wedderburn's theorem is the structural corollary that one need not have separately considered "finite skew fields": every finite division ring is automatically commutative, so the entire arithmetic of finite division rings is exhausted by the structure theorem for $F_{q}$ . The Witt 1931 proof reduces commutativity to a cyclotomic-polynomial inequality on the class equation. The same cyclotomic-polynomial machinery, viewed from a different angle, controls the factorisation $X^{p^{n}} - X = \prod_{d ∣ n} \prod_{f} f (X)$ of the defining polynomial into irreducibles of all divisor degrees, and gives the Möbius-inversion count $N_{p} (n) = \frac{1}{n} \sum_{d ∣ n} μ (n / d) p^{d}$ of irreducible polynomials of degree $n$ . The pattern recurs across the function-field analogue of number theory, where prime-power point counts on curves over $F_{q}$ are controlled by Frobenius eigenvalues on cohomology, and the Weil conjectures (proved by Deligne 1973) package the resulting analytic information into a zeta function whose functional equation and Riemann-hypothesis statement parallel the Riemann zeta function.

Full proof set Master

Proposition 1 (Witt's proof of Wedderburn). Every finite division ring is commutative.

Proof. Let $D$ be a finite division ring with centre $Z = Z (D)$ , a field. Write $∣ Z ∣ = q$ , so $D$ is a $Z$ -vector space of some dimension $n \geq 1$ , $∣ D ∣ = q^{n}$ . Assume $n \geq 2$ toward a contradiction.

For each $a \in D ∖ Z$ , the centraliser $C (a) = {x \in D : x a = a x}$ is a sub-division-ring of $D$ containing $Z$ . As a $Z$ -vector space, $∣ C (a) ∣ = q^{d (a)}$ for some $1 \leq d (a) < n$ . Since $D$ is a $C (a)$ -vector space, $n = d (a) \cdot [D : C (a)]$ in the sense of dimensions, so $d (a) ∣ n$ .

The class equation of $D^{\times}$ under conjugation gives $$ |D^\times| = |Z^\times| + \sum_a [D^\times : C(a)^\times], $$ the sum over conjugacy classes of non-central elements, with one representative $a$ per class. This reads $$ q^n - 1 = (q - 1) + \sum_a \frac{q^n - 1}{q^{d(a)} - 1}. $$

The cyclotomic polynomial $Φ_{n} (X) \in Z [X]$ divides $X^{n} - 1$ in $Z [X]$ via the factorisation $X^{n} - 1 = \prod_{d ∣ n} Φ_{d} (X)$ . So $Φ_{n} (q) ∣ q^{n} - 1$ in $Z$ . For each non-central conjugacy class with centraliser dimension $d = d (a) < n$ and $d ∣ n$ , $Φ_{n} (q)$ also divides $(q^{n} - 1) / (q^{d} - 1)$ in $Z$ : write $q^{n} - 1 = \prod_{e ∣ n} Φ_{e} (q)$ and $q^{d} - 1 = \prod_{e ∣ d} Φ_{e} (q)$ ; the quotient $(q^{n} - 1) / (q^{d} - 1) = \prod_{e ∣ n, e ∤ d} Φ_{e} (q)$ contains $Φ_{n} (q)$ as a factor (since $n ∣ n$ and $n ∤ d$ for $d < n$ ).

So $Φ_{n} (q)$ divides every summand of the class-equation right side and divides the left side $q^{n} - 1$ , hence divides their difference, which is $q - 1$ .

But $Φ_{n} (q) = \prod_{ζ} (q - ζ)$ over primitive $n$ th roots of unity $ζ \in C$ . Each factor $∣ q - ζ ∣ > q - 1$ for $ζ \neq = 1$ and $q \geq 2$ : if $ζ = e^{2 π ik / n}$ with $g cd (k, n) = 1$ and $ζ \neq = 1$ (which holds since $n \geq 2$ ), then $∣ q - ζ ∣^{2} = (q - cos (2 π k / n))^{2} + sin^{2} (2 π k / n) = q^{2} - 2 q cos (2 π k / n) + 1$ . For $cos (2 π k / n) < 1$ (which holds for $ζ \neq = 1$ ), this exceeds $q^{2} - 2 q + 1 = (q - 1)^{2}$ , so $∣ q - ζ ∣ > q - 1$ . Therefore $∣ Φ_{n} (q) ∣ = \prod_{ζ} ∣ q - ζ ∣ > (q - 1)^{φ (n)} \geq q - 1$ (using $φ (n) \geq 1$ ).

This contradicts $Φ_{n} (q) ∣ q - 1$ , so $n = 1$ and $D = Z$ is commutative. $□$

Proposition 2 (Cyclicity of every finite subgroup of $K^{\times}$ ). For any field $K$ , every finite subgroup $G$ of the multiplicative group $K^{\times}$ is cyclic.

Proof. Let $∣ G ∣ = m$ . For each divisor $d ∣ m$ , let $G_{d} = {x \in G : x^{d} = 1}$ . Then $G_{d}$ is a subgroup of $G$ contained in the set of roots of $X^{d} - 1$ in $K$ , which has at most $d$ elements because $K$ is a field. So $∣ G_{d} ∣ \leq d$ for every $d ∣ m$ .

Let $f (d)$ denote the number of elements of $G$ of order exactly $d$ . Then $∣ G_{d} ∣ = \sum_{e ∣ d} f (e)$ , so by Möbius inversion $f (d) = \sum_{e ∣ d} μ (d / e) ∣ G_{e} ∣$ .

Suppose $f (d) > 0$ for some $d ∣ m$ . Pick $x \in G$ of order exactly $d$ . Then $⟨ x ⟩$ has $d$ elements all of order dividing $d$ , all in $G_{d}$ , so $⟨ x ⟩ \subseteq G_{d}$ and $∣ ⟨ x ⟩ ∣ = d \leq ∣ G_{d} ∣ \leq d$ , giving $G_{d} = ⟨ x ⟩$ , a cyclic group of order $d$ . The number of generators of $⟨ x ⟩$ is $φ (d)$ , so $f (d) = φ (d)$ when $f (d) > 0$ .

By the standard identity $\sum_{d ∣ m} φ (d) = m$ and the count $\sum_{d ∣ m} f (d) = m$ (every element has some order), we get $\sum_{d ∣ m} (φ (d) - f (d)) = 0$ . Since $f (d) \in {0, φ (d)}$ and $φ (d) > 0$ , the only way for the sum to vanish is $f (d) = φ (d)$ for every $d ∣ m$ . In particular $f (m) = φ (m) \geq 1$ , so $G$ contains an element of order $m$ , and $G$ is cyclic. $□$

Proposition 3 (Frobenius fixed field). The Frobenius $ϕ (x) = x^{p}$ acting on $F_{q}$ for $q = p^{n}$ has fixed field exactly $F_{p}$ , and order exactly $n$ in $Aut (F_{q})$ .

Proof. For the fixed field: $ϕ (x) = x$ iff $x^{p} - x = 0$ . The polynomial $X^{p} - X \in F_{p} [X]$ has degree $p$ , hence at most $p$ roots in $F_{q}$ . The $p$ elements of $F_{p} \subseteq F_{q}$ satisfy $x^{p} = x$ by Fermat's little theorem (the cyclic group $F_{p}^{\times}$ has order $p - 1$ , so $x^{p - 1} = 1$ for $x \neq = 0$ , hence $x^{p} = x$ ; for $x = 0$ this is automatic). So $F_{p} \subseteq Fix (ϕ)$ , and the size bound gives equality $Fix (ϕ) = F_{p}$ .

For the order: $ϕ^{k} (x) = x^{p^{k}}$ . The identity $ϕ^{k} = id$ on $F_{q}$ means $x^{p^{k}} = x$ for every $x \in F_{q}$ , equivalently $F_{q} \subseteq Fix (ϕ^{k})$ . The fixed field $Fix (ϕ^{k})$ is the set of roots of $X^{p^{k}} - X$ in $F_{q}$ , which has at most $p^{k}$ elements. So $∣ F_{q} ∣ = p^{n} \leq p^{k}$ , giving $k \geq n$ . Conversely $ϕ^{n} (x) = x^{p^{n}} = x^{q} = x$ for every $x \in F_{q}$ (the multiplicative order of $F_{q}^{\times}$ is $q - 1$ , so $x^{q - 1} = 1$ for $x \neq = 0$ and $x^{q} = x$ extends to $x = 0$ ). So $ϕ^{n} = id$ , and the order of $ϕ$ is exactly $n$ . $□$

Proposition 4 (Squares index two). For $q = p^{n}$ with $p$ an odd prime, $[F_{q}^{\times} : (F_{q}^{\times})^{2}] = 2$ .

Proof. Consider the squaring homomorphism $s : F_{q}^{\times} \to F_{q}^{\times}$ , $s (x) = x^{2}$ . Its kernel is ${x \in F_{q}^{\times} : x^{2} = 1} = {x : (x - 1) (x + 1) = 0}$ . Since $F_{q}$ is a field, this is ${x : x = 1 or x = - 1}$ . Since $p$ is odd, $- 1 \neq = 1$ in $F_{q}$ , so the kernel has order $2$ .

By the first isomorphism theorem, $F_{q}^{\times} / ker (s) ≅ Im (s) = (F_{q}^{\times})^{2}$ , so $∣ (F_{q}^{\times})^{2} ∣ = ∣ F_{q}^{\times} ∣/2 = (q - 1) /2$ . The index of the squares in $F_{q}^{\times}$ is therefore $∣ F_{q}^{\times} ∣/∣ (F_{q}^{\times})^{2} ∣ = 2$ . $□$

Proposition 5 (Gauss sum identity). For $p$ an odd prime and $χ : F_{p}^{\times} \to {\pm 1}$ the quadratic Legendre character (with $χ (0) = 0$ ), $g (χ)^{2} = χ (- 1) p$ .

Proof. Compute $$ g(\chi)^2 = \sum_{x, y \in \mathbb{F}_p^\times} \chi(x) \chi(y) \psi(x + y), $$ where $ψ (z) = e^{2 π i z / p}$ . Substitute $y = x t$ for $t \in F_{p}^{\times}$ ; as $y$ ranges over $F_{p}^{\times}$ for each fixed $x \in F_{p}^{\times}$ , so does $t$ . Then $χ (x) χ (y) = χ (x) χ (x t) = χ (x^{2} t) = χ (t)$ (using $χ (x^{2}) = 1$ ), and $x + y = x (1 + t)$ , so $$ g(\chi)^2 = \sum_{t \in \mathbb{F}p^\times} \chi(t) \sum{x \in \mathbb{F}_p^\times} \psi(x(1 + t)). $$

For the inner sum: if $1 + t \equiv 0 (mod p)$ (i.e., $t = - 1$ ), then $ψ (x (1 + t)) = ψ (0) = 1$ for all $x$ , and the sum over $x \in F_{p}^{\times}$ is $p - 1$ . If $1 + t \neq \equiv 0$ , then $ψ (x (1 + t))$ as $x$ ranges over $F_{p}^{\times}$ takes the same values (in some order) as $ψ (x)$ for $x \in F_{p}^{\times}$ , and $\sum_{x \in F_{p}} ψ (x) = 0$ (sum of all $p$ th roots of unity), so $\sum_{x \in F_{p}^{\times}} ψ (x) = - ψ (0) = - 1$ .

So $$ g(\chi)^2 = \chi(-1)(p - 1) + \sum_{t \in \mathbb{F}p^\times, t \neq -1} \chi(t) \cdot (-1) = \chi(-1)(p - 1) - \sum{t \neq -1} \chi(t). $$ Adding and subtracting $χ (- 1)$ on the right gives $$ g(\chi)^2 = \chi(-1) p - \sum_{t \in \mathbb{F}p^\times} \chi(t). $$ The full character sum $\sum{t \in \mathbb{F}_p^\times} \chi(t) $v ani s h es b ec a u se$ \chi $i s an o n - i d e n t i t y c ha r a c t er : b y or t h o g o na l i t y o f c ha r a c t er so f$ \mathbb{F}p^\times $,$ \sum{t} \chi(t) = 0 $u n l ess$ \chi $i s t h e i d e n t i t y c ha r a c t er . S o$ g(\chi)^2 = \chi(-1) p $.$ \square$

Connections Master

Field axioms and characteristic 01.01.01. A finite field is an instance of the field axioms, with the additional structural constraint that the underlying set is finite. The characteristic of a finite field is necessarily a prime $p$ , and the field contains a unique copy of $F_{p}$ as its prime field. Without the field axioms there is no setting for the squares-index-two statement or the Frobenius automorphism, since both rest on the field property (cancellation in field equations, bounded root count for polynomials, the squaring map as a group homomorphism).
Algebraic field extension and splitting field 01.02.12. The field $F_{q}$ for $q = p^{n}$ is defined as the splitting field of $X^{q} - X$ over $F_{p}$ , an instance of the splitting-field construction in algebraic field extension theory. The Galois correspondence for $F_{q} / F_{p}$ specialises the general Galois correspondence to a cyclic group, with the divisor lattice of $n$ controlling the subfield lattice. Every result about finite fields ultimately rests on the splitting-field construction, and the simplicity of finite-field Galois theory is what makes them the basic object in number theory and algebraic geometry over finite ground rings.
Quadratic reciprocity via Gauss sums 21.02.02. The squares-index-two statement for $F_{p}^{\times}$ in odd characteristic produces a quadratic character $χ : F_{p}^{\times} \to {\pm 1}$ , and the Gauss sum $g (χ) = \sum_{x \in F_{p}^{\times}} χ (x) e^{2 π i x / p}$ satisfies $g (χ)^{2} = χ (- 1) p$ . This identity is the algebraic input to Gauss's analytic proof of quadratic reciprocity, packaged in unit 21.02.02. The finite-field unit supplies the character $χ$ and the squares-index-two structural fact; the reciprocity unit takes those inputs and derives the symmetric law $(\frac{p}{q}) (\frac{q}{p}) = (- 1)^{(p - 1) (q - 1) /4}$ for odd primes $p, q$ .
Riemann zeta function 21.03.01. The Euler product of the Riemann zeta function $ζ (s) = \prod_{p} (1 - p^{- s})^{- 1}$ rests on unique factorisation in $Z$ ; the analogous Hasse-Weil zeta function of a smooth projective variety $X / F_{q}$ is $Z (X, T) = \prod_{x} (1 - T^{d e g (x)})^{- 1}$ over closed points $x$ , where $de g (x) = [κ (x) : F_{q}]$ is the degree of the residue field of $x$ as an extension of $F_{q}$ . The function-field side of analytic number theory replaces $Z$ by $F_{q} [T]$ and reads off prime-power point counts of varieties via Frobenius eigenvalues on $ℓ$ -adic cohomology. The basic ingredient is the finite-field structure theorem of the present unit.
$ℓ$ -adic Galois representations 21.05.01. The absolute Galois group $Gal (\overline{F_{p}} / F_{p}) = Z$ is topologically generated by Frobenius, and the Frobenius eigenvalues on $ℓ$ -adic cohomology of a smooth projective variety $X / F_{q}$ are algebraic integers of absolute value $q^{w /2}$ on $H_{\overset{e}{ˊ} t}^{w} (\overline{X}, Q_{ℓ})$ (Deligne's Weil II 1980). The basic Frobenius automorphism of the present unit lifts to the relative Frobenius on schemes over $F_{q}$ , and the analysis of its eigenvalues on cohomology is the content of the Weil conjectures and the modern motivic-zeta-function programme.

Historical & philosophical context Master

Galois announced the construction of finite fields beyond the prime case in his 1830 Bulletin des Sciences Mathématiques de Férussac note Sur la théorie des nombres ^{[Galois 1830]}, six pages reprinted in his collected works in 1846 by Liouville. Galois fixed an irreducible polynomial $f$ over $F_{p}$ of degree $n$ and considered the elements $a_{0} + a_{1} i + a_{2} i^{2} + \dots + a_{n - 1} i^{n - 1}$ in an indeterminate $i$ satisfying $f (i) = 0$ , with $a_{j} \in F_{p}$ , working modulo $f (i)$ . He observed that the resulting set is closed under the four field operations and has $p^{n}$ elements. The construction is the modern $F_{p} [X] / (f)$ in its earliest published form, predating both Steinitz's general theory of fields and Dedekind's modern ideal-theoretic framework by half a century.

The construction left open whether the resulting field depended on the choice of $f$ . E. H. Moore resolved this in his 1893 paper A doubly-infinite system of simple groups in the Bulletin of the New York Mathematical Society 3 ^{[Moore 1893]}: any two finite fields of the same cardinality $p^{n}$ are $F_{p}$ -isomorphic. Moore's argument is the splitting-field description used in the modern proof. Dickson's 1901 Linear Groups, with an Exposition of the Galois Field Theory (Teubner) ^{[Dickson 1901]} gave the first systematic textbook treatment, with the Frobenius automorphism and the linear groups $GL_{n} (F_{q})$ as a working tool of finite group theory. Steinitz's 1910 Journal für die reine und angewandte Mathematik paper Algebraische Theorie der Körper ^{[Steinitz 1910]} absorbed finite fields into a general field-theoretic framework distinguishing algebraic from transcendental extensions, separable from inseparable, and prepared the ground for Artin's 1927 modern packaging of Galois theory.

Wedderburn's 1905 Transactions of the American Mathematical Society paper A theorem on finite algebras ^{[Wedderburn 1905]} proved that every finite division ring is commutative, showing the entire arithmetic of finite skew fields collapses to the commutative case classified by Galois and Moore. The original proof was a somewhat lengthy argument on the structure of the multiplicative group; Witt's 1931 Abhandlungen aus dem Mathematischen Seminar der Universität Hamburg note Über die Kommutativität endlicher Schiefkörper ^{[source pending]} reduced the proof to a three-line cyclotomic-polynomial argument on the class equation, which became the modern standard exposition. The squares-index-two structure of $F_{q}^{\times}$ in odd characteristic and the resulting Legendre character $χ (x) = x^{(q - 1) /2}$ originate with Euler's criterion (Euler 1755 Novi Commentarii Acad. Sci. Petrop. 8); Gauss's 1801 Disquisitiones Arithmeticae §§95-152 ^{[Gauss 1801]} gave four proofs of quadratic reciprocity, with the §356 analytic proof through Gauss sums $g (χ)$ becoming the prototype for higher reciprocity laws and the analytic side of class field theory.

Bibliography Master

@article{Galois1830,
  author  = {Galois, {\'E}variste},
  title   = {Sur la th{\'e}orie des nombres},
  journal = {Bulletin des Sciences Math{\'e}matiques de F{\'e}russac},
  volume  = {13},
  year    = {1830},
  pages   = {428--435},
  note    = {Reprinted in \emph{{\OE}uvres math{\'e}matiques de {\'E}variste Galois}, ed.\ Liouville, J.\ Math.\ Pures Appl.\ 11 (1846), 381--444}
}

@article{Moore1893,
  author  = {Moore, E. H.},
  title   = {A doubly-infinite system of simple groups},
  journal = {Bulletin of the New York Mathematical Society},
  volume  = {3},
  year    = {1893},
  pages   = {73--78}
}

@article{Wedderburn1905,
  author  = {Wedderburn, J. H. M.},
  title   = {A theorem on finite algebras},
  journal = {Transactions of the American Mathematical Society},
  volume  = {6},
  year    = {1905},
  pages   = {349--352}
}

@article{Witt1931,
  author  = {Witt, Ernst},
  title   = {{\"U}ber die Kommutativit{\"a}t endlicher Schiefk{\"o}rper},
  journal = {Abhandlungen aus dem Mathematischen Seminar der Universit{\"a}t Hamburg},
  volume  = {8},
  year    = {1931},
  pages   = {413}
}

@book{Dickson1901,
  author    = {Dickson, Leonard Eugene},
  title     = {Linear Groups, with an Exposition of the Galois Field Theory},
  publisher = {B. G. Teubner},
  address   = {Leipzig},
  year      = {1901}
}

@article{Steinitz1910,
  author  = {Steinitz, Ernst},
  title   = {Algebraische Theorie der K{\"o}rper},
  journal = {Journal f{\"u}r die reine und angewandte Mathematik},
  volume  = {137},
  year    = {1910},
  pages   = {167--309}
}

@book{Gauss1801,
  author    = {Gauss, Carl Friedrich},
  title     = {Disquisitiones Arithmeticae},
  publisher = {Fleischer},
  address   = {Leipzig},
  year      = {1801}
}

@book{Serre1973CinA,
  author    = {Serre, Jean-Pierre},
  title     = {A Course in Arithmetic},
  publisher = {Springer-Verlag},
  series    = {Graduate Texts in Mathematics},
  volume    = {7},
  year      = {1973}
}

@book{Lang2002Algebra,
  author    = {Lang, Serge},
  title     = {Algebra},
  edition   = {3rd revised},
  publisher = {Springer-Verlag},
  series    = {Graduate Texts in Mathematics},
  volume    = {211},
  year      = {2002}
}

@book{LidlNiederreiter1997,
  author    = {Lidl, Rudolf and Niederreiter, Harald},
  title     = {Finite Fields},
  edition   = {2nd},
  publisher = {Cambridge University Press},
  series    = {Encyclopedia of Mathematics and its Applications},
  volume    = {20},
  year      = {1997}
}

@book{IrelandRosen1990,
  author    = {Ireland, Kenneth and Rosen, Michael},
  title     = {A Classical Introduction to Modern Number Theory},
  edition   = {2nd},
  publisher = {Springer-Verlag},
  series    = {Graduate Texts in Mathematics},
  volume    = {84},
  year      = {1990}
}

@book{DummitFoote2004,
  author    = {Dummit, David S. and Foote, Richard M.},
  title     = {Abstract Algebra},
  edition   = {3rd},
  publisher = {John Wiley \& Sons},
  year      = {2004}
}

Prerequisites

01.01.01
01.02.12

Tier anchors

beginner: Stewart *Galois Theory* (4th ed., CRC 2015) Ch. 16 informal opening on finite fields and prime-power orders; Childs *A Concrete Introduction to Higher Algebra* (Springer UTM, 3rd ed. 2009) Ch. 22 on $\mathbb{F}_4$, $\mathbb{F}_8$, $\mathbb{F}_9$
intermediate: Serre 1973 *A Course in Arithmetic* (Springer GTM 7) Ch. I §1; Lang 1965 *Algebra* Ch. V §5 (finite fields, Frobenius, Galois correspondence); Dummit-Foote *Abstract Algebra* (3rd ed., Wiley 2004) Ch. 14 §3
master: Serre 1973 *A Course in Arithmetic* Ch. I (originator-level concise treatment of finite fields, squares, and Frobenius); Lang 1965 *Algebra* Ch. V §5; Lidl-Niederreiter 1997 *Finite Fields* (Encyclopedia of Mathematics and its Applications 20, 2nd ed., Cambridge) Ch. 2; Ireland-Rosen 1990 *A Classical Introduction to Modern Number Theory* (Springer GTM 84, 2nd ed.) Ch. 7

References

Serre, J.-P. — A Course in Arithmetic · Springer Graduate Texts in Mathematics 7 (1973, English translation of *Cours d'arithmétique*, Presses Universitaires de France 1970). Ch. I §1 — the canonical concise treatment of finite fields as the splitting field of $X^q - X$, the cyclicity of $\mathbb{F}_q^\times$, the index-two subgroup of squares for $q$ odd, and Frobenius.
Galois, É. — Sur la théorie des nombres · *Bulletin des Sciences Mathématiques de Férussac* 13 (1830), 428-435. Reprinted in *Œuvres mathématiques de Évariste Galois* edited by Liouville, *J. Math. Pures Appl.* 11 (1846), 381-444. The originator paper constructing the finite field $\mathbb{F}_p[X]/(f)$ for $f$ irreducible over $\mathbb{F}_p$, with the elements $a_0 + a_1 i + \cdots + a_{n-1} i^{n-1}$ written in the indeterminate $i$ satisfying $f(i) = 0$.
Moore, E. H. — A doubly-infinite system of simple groups · *Bulletin of the New York Mathematical Society* 3 (1893), 73-78. Independent uniqueness proof that any two finite fields of the same cardinality are isomorphic, removing the dependence of Galois's construction on the choice of irreducible polynomial.
Wedderburn, J. H. M. — A theorem on finite algebras · *Transactions of the American Mathematical Society* 6 (1905), 349-352. Every finite division ring is commutative; equivalently, the only finite skew fields are the commutative finite fields $\mathbb{F}_q$.
Lang, S. — Algebra · 3rd revised edition, Addison-Wesley 1993, reprinted as Springer Graduate Texts in Mathematics 211 (2002). Ch. V §5 — finite fields, Frobenius, the Galois correspondence for $\mathbb{F}_{p^n}/\mathbb{F}_p$, the subfield lattice.
Lidl, R. & Niederreiter, H. — Finite Fields · Encyclopedia of Mathematics and its Applications 20, 2nd edition, Cambridge University Press 1997. Ch. 2 — structure theorem, multiplicative cyclicity, the Witt / character-theory proof of Wedderburn's little theorem, applications to coding theory and combinatorics.
Ireland, K. & Rosen, M. — A Classical Introduction to Modern Number Theory · Springer Graduate Texts in Mathematics 84, 2nd edition 1990. Ch. 7 — finite fields, Gauss sums $g(\chi) = \sum_{x \in \mathbb{F}_p^\times} \chi(x) e^{2\pi i x/p}$ and the route to quadratic reciprocity via $g(\chi)^2 = \chi(-1) p$.
Gauss, C. F. — Disquisitiones Arithmeticae · Fleischer, Leipzig 1801, §§95-152 (the four proofs of quadratic reciprocity), §356 (the Gauss-sum proof). The originator analytic proof of quadratic reciprocity, whose modern character-theoretic packaging proceeds through Gauss sums attached to a non-square character of $\mathbb{F}_p^\times$.
Dickson, L. E. — Linear Groups, with an Exposition of the Galois Field Theory · Teubner, Leipzig 1901. The first systematic textbook treatment of finite fields, the Frobenius automorphism, and the classical linear groups $\mathrm{GL}_n(\mathbb{F}_q)$ as a working tool of group theory.
Witt, E. — Über die Kommutativität endlicher Schiefkörper · *Abhandlungen aus dem Mathematischen Seminar der Universität Hamburg* 8 (1931), 413. Witt's three-line cyclotomic-polynomial proof of Wedderburn's little theorem, based on counting and the inequality $|\Phi_n(q)| > q - 1$ for $n \geq 2$ when $q \geq 2$.
Steinitz, E. — Algebraische Theorie der Körper · *Journal für die reine und angewandte Mathematik* 137 (1910), 167-309. The first general field-theoretic framework distinguishing algebraic from transcendental extensions, with finite fields as the prototype example.

Estimated time

beginner: 18m
intermediate: 40m
master: 80m