7.3 Fields, Finite Fields and Algebraic Extensions
Fields and their characteristic, the quotient fields $F[x]/\langle f\rangle$, the degree $[K:F]$ and the tower law, minimal polynomials and algebraic extensions, splitting fields, algebraically closed fields, and the complete structure of the finite fields $\mathbb{F}_{p^n}$ — the degree computations and counting results that GATE MA asks for.
Section 7.2 proved a remarkable fact: if $f$ is an irreducible polynomial over a field $F$, then the quotient ring $F[x]/\langle f\rangle$ is again a field, and in it the class of $x$ is a root of $f$. This is how mathematicians manufacture roots that do not yet exist — $\mathbb{R}[x]/\langle x^2+1\rangle$ is the complex numbers, $\mathbb{F}_2[x]/\langle x^2+x+1\rangle$ is a field with four elements. This section studies such bigger fields systematically. The key number is the degree $[K:F]$, the dimension of $K$ as a vector space over $F$: it multiplies along towers of fields, it equals the degree of the minimal polynomial for a field generated by one root, and it controls which numbers can live in which fields. With it we can decide quickly that $\sqrt[3]{2}$ is not in any field of degree $4$ over $\mathbb{Q}$, that the smallest field containing all roots of $x^3-2$ has degree $6$, and that a finite field has a prime-power number of elements, its nonzero elements form a cyclic group, and its subfields correspond to divisors.
GATE MA asks about this material in a few recurring forms: the degree of an extension such as $\mathbb{Q}(\sqrt2,\sqrt[3]{2},i)$ or of a splitting field (often a NAT question); "which of these quotient rings is a field"; the number of elements of a given order, of generators of $\mathbb{F}_q^\times$, of subfields of $\mathbb{F}_{p^n}$ or of monic irreducible polynomials of a given degree over $\mathbb{F}_q$; and MSQs about algebraic and algebraically closed fields ("every algebraic extension is finite", "no finite field is algebraically closed"). All of these come down to a handful of theorems, proved or carefully sketched below. Throughout, $F$, $K$, $L$ denote fields, $p$ a prime, $q=p^n$ a prime power, $\mathbb{F}_p=\mathbb{Z}_p=\mathbb{Z}/p\mathbb{Z}$, $\omega=e^{2\pi i/3}$ and $\zeta_n=e^{2\pi i/n}$.
Fields, Characteristic and Quotient Fields
- The characteristic of a field is $0$ or a prime $p$. If $\operatorname{char}F=p$, then $pa=0$ for every $a\in F$.
- The prime subfield of $F$ is isomorphic to $\mathbb{Q}$ if $\operatorname{char}F=0$ and to $\mathbb{F}_p$ if $\operatorname{char}F=p$.
- Every ring homomorphism $\varphi:F\to R$ from a field to a nonzero ring with $\varphi(1)=1$ is injective. In particular a field homomorphism $F\to K$ identifies $F$ with a subfield of $K$, and fields of different characteristic admit no homomorphisms between them.
Proof. (1) If $n\cdot1=0$ with $n=rs$, $1\lt r,s\lt n$, then $(r\cdot1)(s\cdot1)=0$, and since a field has no zero divisors, $r\cdot1=0$ or $s\cdot1=0$ — contradicting the minimality of $n$. So the least such $n$ is prime. Then $pa=(p\cdot1)a=0$. (2) The map $\mathbb{Z}\to F$, $m\mapsto m\cdot1$, is a ring homomorphism whose kernel is $p\mathbb{Z}$ or $0$. If the kernel is $p\mathbb{Z}$, its image is a copy of $\mathbb{Z}/p\mathbb{Z}=\mathbb{F}_p$, a field contained in every subfield (each subfield contains $1$). If the kernel is $0$, every subfield contains the quotients $(m\cdot1)(n\cdot1)^{-1}$, which form a copy of $\mathbb{Q}$. (3) The kernel is an ideal of $F$; the only ideals of a field are $0$ and $F$ (a nonzero ideal contains a unit), and the kernel is not $F$ because $\varphi(1)=1\ne0$. A homomorphism preserves $n\cdot1$, so characteristics must agree. ∎
The fields we meet are: the number fields inside $\mathbb{C}$ (characteristic $0$), the finite fields $\mathbb{F}_p$ and their extensions, and fields of rational functions such as $\mathbb{F}_p(t)$ — an infinite field of characteristic $p$. The standard way of making new fields is the following quotient construction (the ideal theory behind it is in §7.2: $F[x]$ is a PID, and $\langle f\rangle$ is maximal exactly when $f$ is irreducible).
- $K$ is a field $\iff$ $f$ is irreducible over $F$ $\iff$ $K$ is an integral domain.
- Every element of $K$ is uniquely $c_0+c_1\theta+\cdots+c_{n-1}\theta^{n-1}$ with $c_i\in F$; so $K$ is an $n$-dimensional vector space over $F$ (which sits inside $K$ as the constants), and $f(\theta)=0$.
- If $F=\mathbb{F}_q$ is finite and $f$ is irreducible, $K$ is a field with $q^n$ elements.
- For $\deg f=2$ or $3$: $f$ is irreducible over $F$ iff $f$ has no root in $F$.
Proof. (2) Division by $f$ gives every $g\in F[x]$ the form $g=qf+r$ with $\deg r\lt n$, and $g\equiv r$; two remainders of degree $\lt n$ that are congruent differ by a multiple of $f$ of degree $\lt n$, hence are equal. And $f(\theta)=f(x)+\langle f\rangle=0$. (1) If $f=gh$ with $1\le\deg g,\deg h\lt n$, then $g(\theta)h(\theta)=0$ with both factors nonzero (they are their own remainders): $K$ has zero divisors and is not a field. If $f$ is irreducible and $g(\theta)\ne0$, then $f\nmid g$, so $\gcd(f,g)=1$ (the only divisors of $f$ are units and associates of $f$), and Bézout's identity $uf+vg=1$ in the Euclidean domain $F[x]$ gives $v(\theta)g(\theta)=1$. A field is an integral domain, which closes the cycle. (3) There are $q$ choices for each of the $n$ coefficients in (2). (4) A factorisation of a quadratic or cubic into factors of positive degree contains a linear factor $x-a$, i.e. a root $a\in F$. ∎
Part (4) fails from degree $4$ on: $x^4+4=(x^2+2x+2)(x^2-2x+2)$ has no rational root but is reducible over $\mathbb{Q}$. For degree $\ge4$ use the tests of §7.2 (Eisenstein, reduction mod $p$, Gauss's lemma) or the degree arguments below.
Field Extensions, Degree and the Tower Law
Examples: $[\mathbb{C}:\mathbb{R}]=2$ with basis $1,i$; $[\mathbb{Q}(\sqrt2):\mathbb{Q}]=2$ with basis $1,\sqrt2$; $[K:F]=1$ iff $K=F$; $[\mathbb{R}:\mathbb{Q}]=\infty$, because a finite-dimensional $\mathbb{Q}$-vector space is countable.
Proof. Spanning. Every $w\in L$ is $w=\sum_j b_jv_j$ with $b_j\in K$, and each $b_j=\sum_ia_{ij}u_i$ with $a_{ij}\in F$; so $w=\sum_{i,j}a_{ij}u_iv_j$. Independence. If $\sum_{i,j}a_{ij}u_iv_j=0$ with $a_{ij}\in F$, group it as $\sum_j\left(\sum_ia_{ij}u_i\right)v_j=0$. The inner sums lie in $K$, and the $v_j$ are independent over $K$, so $\sum_ia_{ij}u_i=0$ for each $j$; the $u_i$ are independent over $F$, so every $a_{ij}=0$. Counting the basis gives the product formula. Conversely, if $[L:F]\lt\infty$, then $K$ is a subspace of the finite-dimensional $F$-space $L$, so $[K:F]\lt\infty$, and an $F$-spanning set of $L$ also spans $L$ over $K$, so $[L:K]\lt\infty$. ∎
The minimal polynomial of a matrix (§3.2) is the same idea for the ring of matrices instead of a field, and it behaves in the same way: it divides every polynomial that the object satisfies. The difference is that a field has no zero divisors, which makes the minimal polynomial of a field element irreducible.
- $m$ is irreducible over $F$, and for $g\in F[x]$: $g(\alpha)=0\iff m\mid g$. Conversely, a monic irreducible polynomial in $F[x]$ with root $\alpha$ is $m$.
- $F(\alpha)=F[\alpha]=\{g(\alpha):g\in F[x]\}$, and $g(x)+\langle m\rangle\mapsto g(\alpha)$ is an isomorphism $F[x]/\langle m\rangle\cong F(\alpha)$.
- $1,\alpha,\dots,\alpha^{n-1}$ is a basis of $F(\alpha)$ over $F$, so $[F(\alpha):F]=\deg_F\alpha=n$.
- If $\alpha$ is transcendental over $F$, then $F(\alpha)\cong F(x)$, the field of rational functions, and $[F(\alpha):F]=\infty$.
Proof. (1) If $m=gh$ with $g,h$ of smaller positive degree, then $g(\alpha)h(\alpha)=m(\alpha)=0$ in the field $K$, so $g(\alpha)=0$ or $h(\alpha)=0$, contradicting the minimality of $\deg m$. If $g(\alpha)=0$, divide: $g=qm+r$ with $\deg r\lt n$; then $r(\alpha)=0$, so $r=0$ by minimality. If $p$ is monic irreducible with $p(\alpha)=0$, then $m\mid p$, and irreducibility of $p$ forces $p=m$. (2) The evaluation map $\mathrm{ev}_\alpha:F[x]\to K$, $g\mapsto g(\alpha)$, is a ring homomorphism with image $F[\alpha]$ and, by (1), kernel $\langle m\rangle$. So $F[\alpha]\cong F[x]/\langle m\rangle$, which is a field by Theorem 7.40. A field containing $F$ and $\alpha$ contains $F[\alpha]$; and $F[\alpha]$ is itself such a field; hence $F(\alpha)=F[\alpha]$. (3) Transport the basis $1,\theta,\dots,\theta^{n-1}$ of Theorem 7.40(2) through the isomorphism. (4) Now $\mathrm{ev}_\alpha$ is injective, so it extends to the fraction fields: $g/h\mapsto g(\alpha)/h(\alpha)$ is an isomorphism $F(x)\to F(\alpha)$, and $1,\alpha,\alpha^2,\dots$ are independent. ∎
- If $[K:F]=n\lt\infty$, then every $\alpha\in K$ is algebraic over $F$ and $\deg_F\alpha$ divides $n$. In particular, if $[K:F]$ is prime, then $K=F(\alpha)$ for every $\alpha\in K\setminus F$, and there is no intermediate field other than $F$ and $K$.
- If $\alpha,\beta$ are algebraic over $F$ of degrees $m,n$, then $[F(\alpha,\beta):F]\le mn$, it is divisible by $\operatorname{lcm}(m,n)$, and it equals $mn$ when $\gcd(m,n)=1$.
- $[F(\alpha):F]$ does not depend on which polynomial we happen to know $\alpha$ satisfies — only on the irreducible one. If $\alpha$ is a root of an irreducible $g$ of degree $n$, then $[F(\alpha):F]=n$.
- Kronecker. Every nonconstant $f\in F[x]$ has a root in some extension $K$ of $F$ with $[K:F]\le\deg f$.
Proof. (1) The $n+1$ elements $1,\alpha,\dots,\alpha^n$ are dependent over $F$, which gives a nonzero polynomial with root $\alpha$. By the tower law $n=[K:F(\alpha)]\,[F(\alpha):F]$, and $[F(\alpha):F]=\deg_F\alpha$. If $n$ is prime and $\alpha\notin F$, then $[F(\alpha):F]\gt1$ divides $n$, so it is $n$, and $F(\alpha)=K$; an intermediate field $L$ has $[L:F]\in\{1,n\}$, i.e. $L=F$ or $L=K$. (2) $\beta$ is a root of $m_{\beta,F}$, which lies in $F(\alpha)[x]$, so $\deg_{F(\alpha)}\beta\le n$ and $[F(\alpha,\beta):F]=[F(\alpha)(\beta):F(\alpha)]\,[F(\alpha):F]\le nm$. Both $m$ and $n$ divide $[F(\alpha,\beta):F]$ by the tower law through $F(\alpha)$ and through $F(\beta)$; if they are coprime, $mn$ divides it, and with the upper bound it equals $mn$. (3) is Theorem 7.42(1),(3). (4) Let $g$ be an irreducible factor of $f$ (it exists because $F[x]$ is a UFD, §7.2). Then $K=F[x]/\langle g\rangle$ is a field of degree $\deg g\le\deg f$ over $F$ in which $\theta$ is a root of $g$, hence of $f$. ∎
- Simple extension: find a polynomial with root $\alpha$, prove it irreducible (no roots for degree $\le3$; Eisenstein, possibly after $x\mapsto x+1$; reduction mod $p$), and read off its degree.
- Several generators: climb a tower $F\subseteq F(\alpha)\subseteq F(\alpha,\beta)\subseteq\cdots$ and multiply. At each step ask: is the new element already in the field I have? If not, what is its degree over that field (often $2$: show a square root is not in the current field)?
- Use divisibility both ways: the degree of every intermediate field divides the total. If $m$ and $n$ both divide $d$ and $d\le mn$ with $\gcd(m,n)=1$, then $d=mn$. A field of degree $d$ contains no element of degree not dividing $d$.
- Real versus non-real: if the field so far lies in $\mathbb{R}$ and the new element is non-real and satisfies a quadratic over it, the step has degree exactly $2$.
- Look for a primitive element: often $K=F(\alpha+\beta)$, which reduces everything to one minimal polynomial (e.g. $\mathbb{Q}(\sqrt2,\sqrt3)=\mathbb{Q}(\sqrt2+\sqrt3)$).
Algebraic Extensions and Algebraically Closed Fields
- Every finite extension is algebraic.
- If $\alpha_1,\dots,\alpha_k$ are algebraic over $F$, then $F(\alpha_1,\dots,\alpha_k)/F$ is finite, of degree at most $\prod\deg_F\alpha_i$. So: $K/F$ is finite $\iff$ $K$ is generated over $F$ by finitely many algebraic elements.
- For any extension $K/F$, the set $\overline{F}_K=\{\alpha\in K:\alpha\text{ algebraic over }F\}$ is a subfield of $K$ — sums, differences, products and quotients of algebraic elements are algebraic. In particular the algebraic numbers $\overline{\mathbb{Q}}\subset\mathbb{C}$ form a field.
- Transitivity. If $L/K$ and $K/F$ are algebraic, then $L/F$ is algebraic.
- The converse of (1) fails: $\overline{\mathbb{Q}}/\mathbb{Q}$ is algebraic but $[\overline{\mathbb{Q}}:\mathbb{Q}]=\infty$. Moreover $\overline{\mathbb{Q}}$ is countable, so almost all real and complex numbers are transcendental.
Proof. (1) Theorem 7.43(1). (2) Induction along the tower $F\subseteq F(\alpha_1)\subseteq F(\alpha_1,\alpha_2)\subseteq\cdots$: $\alpha_{j}$ is algebraic over $F(\alpha_1,\dots,\alpha_{j-1})$ of degree at most $\deg_F\alpha_j$ (its minimal polynomial over $F$ still works), and degrees multiply. If $K/F$ is finite, a basis generates $K$ and consists of algebraic elements by (1). (3) If $\alpha,\beta$ are algebraic, $F(\alpha,\beta)$ is finite over $F$ by (2), hence algebraic by (1); it contains $\alpha\pm\beta$, $\alpha\beta$ and $\alpha/\beta$ ($\beta\ne0$). (4) Let $\gamma\in L$, with $\gamma^m+c_{m-1}\gamma^{m-1}+\cdots+c_0=0$, $c_j\in K$. Each $c_j$ is algebraic over $F$, so $M=F(c_0,\dots,c_{m-1})$ is finite over $F$, and $[M(\gamma):M]\le m$. Hence $[M(\gamma):F]\lt\infty$ and $\gamma$ is algebraic over $F$ by (1). (5) For each $n$, $x^n-2$ is irreducible (Eisenstein), so $\sqrt[n]2\in\overline{\mathbb{Q}}$ has degree $n$; by Theorem 7.43(1) a finite $[\overline{\mathbb{Q}}:\mathbb{Q}]$ would be divisible by every $n$. Countability: there are countably many polynomials with rational coefficients, each with finitely many roots. ∎
Which specific numbers are transcendental is much harder. Hermite (1873) proved that $e$ is transcendental and Lindemann (1882) that $\pi$ is; we use these as known facts. A useful consequence of (3): if $\alpha$ is transcendental and $\beta$ algebraic, then $\alpha+\beta$ and (for $\beta\ne0$) $\alpha\beta$ are transcendental — otherwise $\alpha=(\alpha+\beta)-\beta$ would be algebraic.
- For a field $F$ the following are equivalent: (i) $F$ is algebraically closed; (ii) every nonconstant $f\in F[x]$ splits into linear factors over $F$; (iii) the irreducible polynomials over $F$ are exactly those of degree $1$; (iv) $F$ has no algebraic extension other than $F$ itself; (v) $F$ has no finite extension of degree $\gt1$.
- Fundamental theorem of algebra. $\mathbb{C}$ is algebraically closed. Hence the irreducible polynomials over $\mathbb{R}$ have degree $1$ or $2$.
- Every algebraically closed field is infinite. In particular no finite field is algebraically closed.
- Every field $F$ has an algebraic closure, unique up to an isomorphism fixing $F$ (Steinitz). The algebraic closure of $\mathbb{Q}$ is the field $\overline{\mathbb{Q}}$ of algebraic numbers: it is algebraically closed, countable, and of infinite degree over $\mathbb{Q}$. $\mathbb{C}$ is algebraically closed but is not an algebraic closure of $\mathbb{Q}$.
Proof. (1) (i)$\Rightarrow$(ii): induct on the degree, splitting off a linear factor $x-a$ for a root $a$ each time. (ii)$\Rightarrow$(iii): an irreducible polynomial that splits has one factor only, of degree $1$. (iii)$\Rightarrow$(iv): if $K/F$ is algebraic and $\alpha\in K$, then $m_{\alpha,F}$ is irreducible, so it is $x-\alpha$ and $\alpha\in F$. (iv)$\Rightarrow$(v): finite extensions are algebraic. (v)$\Rightarrow$(i): a nonconstant $f$ has an irreducible factor $g$, and $F[x]/\langle g\rangle$ is a finite extension of degree $\deg g$; by (v) $\deg g=1$, say $g=c(x-a)$, and $a\in F$ is a root of $f$. (2) The fundamental theorem of algebra is proved in complex analysis via Liouville's theorem (§5.1): a root-free polynomial $f$ would make $1/f$ bounded and entire, hence constant. Over $\mathbb{R}$, the non-real roots of a real polynomial come in conjugate pairs $\lambda,\bar\lambda$, and $(x-\lambda)(x-\bar\lambda)=x^2-2(\operatorname{Re}\lambda)x+|\lambda|^2$ is real; so every real polynomial is a product of real linear and quadratic factors. (3) If $F=\{a_1,\dots,a_q\}$ is finite, the polynomial $f(x)=1+(x-a_1)(x-a_2)\cdots(x-a_q)$ satisfies $f(a_i)=1\ne0$ for all $i$: it has no root in $F$. (4) Existence and uniqueness of algebraic closures use Zorn's lemma (see the 📚 box). For $\overline{\mathbb{Q}}$: if $f\in\overline{\mathbb{Q}}[x]$ is nonconstant, it has a root $\gamma\in\mathbb{C}$ by (2); $\gamma$ is algebraic over $\overline{\mathbb{Q}}$, which is algebraic over $\mathbb{Q}$, so $\gamma$ is algebraic over $\mathbb{Q}$ by transitivity (Theorem 7.44(4)), i.e. $\gamma\in\overline{\mathbb{Q}}$. Countability and infinite degree are Theorem 7.44(5). $\mathbb{C}$ is not algebraic over $\mathbb{Q}$, as $\pi\in\mathbb{C}$. ∎
Splitting Fields
- A splitting field $K$ of $f$ over $F$ exists, and $[K:F]\le n!$. (In fact $[K:F]$ divides $n!$.)
- Any two splitting fields of $f$ over $F$ are isomorphic by an isomorphism that fixes $F$ pointwise. So "the" splitting field and its degree are well defined.
- If $f$ is irreducible of degree $n$, then $n$ divides $[K:F]$.
- The splitting field depends on the base field: $x^2+1$ has splitting field of degree $2$ over $\mathbb{Q}$ and $\mathbb{R}$, but degree $1$ over $\mathbb{C}$ and over $\mathbb{F}_5$.
Proof. (1) Induction on $n$, for all fields at once. By Kronecker (Theorem 7.43(4)) there is $F_1=F(\alpha_1)$ with $f(\alpha_1)=0$ and $[F_1:F]\le n$. Over $F_1$, $f=(x-\alpha_1)g$ with $\deg g=n-1$; by induction $g$ has a splitting field $K$ over $F_1$ with $[K:F_1]\le(n-1)!$. Then $f$ splits over $K$, $K=F_1(\alpha_2,\dots,\alpha_n)=F(\alpha_1,\dots,\alpha_n)$, and $[K:F]\le n\cdot(n-1)!=n!$. (The divisibility needs Galois theory: the automorphisms of $K/F$ permute the roots.) (2) Sketch. One proves more generally: if $\sigma:F\to F'$ is an isomorphism, $f'$ is the image of $f$, and $K,K'$ are splitting fields of $f,f'$, then $\sigma$ extends to an isomorphism $K\to K'$. Take an irreducible factor $g$ of $f$ with root $\alpha\in K$ and a root $\alpha'\in K'$ of the image $g'$; then $F(\alpha)\cong F[x]/\langle g\rangle\cong F'[x]/\langle g'\rangle\cong F'(\alpha')$ extends $\sigma$ (Theorem 7.42(2)), and induction on $[K:F]$ finishes the job. (3) $F(\alpha_1)\subseteq K$ has degree $n$ (Theorem 7.42(3)); apply the tower law. (4) In $\mathbb{F}_5$, $x^2+1=(x-2)(x-3)$. ∎
- $f$ has a repeated root in its splitting field iff $f$ and $f'$ have a common factor of positive degree, i.e. $\gcd(f,f')\ne1$ (a gcd computed in $F[x]$).
- If $\operatorname{char}F=0$, or $F$ is finite, every irreducible polynomial over $F$ has distinct roots. In characteristic $p$, an irreducible $f$ has a repeated root iff $f'=0$, i.e. $f(x)=g(x^p)$.
- For $q=p^n$, the polynomial $x^q-x$ has $q$ distinct roots in its splitting field over $\mathbb{F}_p$.
Proof. (1) If $f=(x-\alpha)^2h$, then $f'=2(x-\alpha)h+(x-\alpha)^2h'$ also vanishes at $\alpha$, so $m_{\alpha,F}$ divides both $f$ and $f'$. Conversely, if $\alpha$ is a common root, write $f=(x-\alpha)h$; then $f'=h+(x-\alpha)h'$ gives $h(\alpha)=0$, so $(x-\alpha)^2\mid f$. Common factors in $F[x]$ have common roots in the splitting field, and conversely the minimal polynomial of a common root is a common factor; the gcd does not change when the field grows, because it is computed by the Euclidean algorithm. (2) If $f$ is irreducible, $\gcd(f,f')$ is $1$ or $f$ (up to a unit); it is $f$ only if $f\mid f'$, which forces $f'=0$ since $\deg f'\lt\deg f$. In characteristic $0$, $f'=0$ means $f$ is constant. In characteristic $p$, $f'=\sum ka_kx^{k-1}=0$ means $a_k=0$ unless $p\mid k$. Over a finite field, $g(x^p)=\sum b_jx^{pj}=\left(\sum b_j^{1/p}x^j\right)^p$ (every element of $\mathbb{F}_q$ is a $p$-th power, Theorem 7.49(3)), which is not irreducible. (3) $(x^q-x)'=qx^{q-1}-1=-1$, coprime to everything. ∎
The standard non-example lives in the infinite field $F=\mathbb{F}_p(t)$: $x^p-t$ is irreducible (Eisenstein at the prime $t$ of $\mathbb{F}_p[t]$), yet $(x^p-t)'=0$, and in its splitting field $x^p-t=(x-t^{1/p})^p$ has a single root of multiplicity $p$. Such inseparable polynomials never occur over $\mathbb{Q}$ or over finite fields, which is why GATE never needs them.
- Quadratics: $x^2+bx+c$ has splitting field $\mathbb{Q}(\sqrt{b^2-4c})$, of degree $1$ or $2$.
- Irreducible cubics: if $f$ is an irreducible cubic with discriminant $D$ ($D=-4a^3-27b^2$ for $x^3+ax+b$), its splitting field has degree $3$ if $D$ is the square of a rational number and $6$ otherwise. A reducible cubic has degree $1$ or $2$.
- Roots of unity: the splitting field of $x^n-1$ is the cyclotomic field $\mathbb{Q}(\zeta_n)$, and $[\mathbb{Q}(\zeta_n):\mathbb{Q}]=\varphi(n)$ (the minimal polynomial of $\zeta_n$ is the cyclotomic polynomial $\Phi_n$, which is irreducible over $\mathbb{Q}$). For a prime $p$: $\Phi_p=x^{p-1}+\cdots+x+1$ and the degree is $p-1$.
- Pure equations: if $p$ is prime and $a\in\mathbb{Q}$ is not a $p$-th power in $\mathbb{Q}$, then $x^p-a$ is irreducible and its splitting field $\mathbb{Q}(\sqrt[p]a,\zeta_p)$ has degree $p(p-1)$. For example $x^3-2$: $6$; $x^5-2$: $20$. Also $x^4-2$ has splitting field $\mathbb{Q}(\sqrt[4]2,i)$ of degree $8$.
Proof. (1) The roots are $\tfrac12(-b\pm\sqrt{b^2-4c})$. (2) Let $\alpha_1,\alpha_2,\alpha_3$ be the roots, $K$ the splitting field. $[\mathbb{Q}(\alpha_1):\mathbb{Q}]=3$ and $f=(x-\alpha_1)g$ over $\mathbb{Q}(\alpha_1)$ with $g$ quadratic, so $[K:\mathbb{Q}]=3$ or $6$ according as $g$ splits over $\mathbb{Q}(\alpha_1)$ or not, i.e. as $\alpha_2-\alpha_3\in\mathbb{Q}(\alpha_1)$ or not (the roots of $g$ are $\alpha_2,\alpha_3=s\pm t$ with $s=\tfrac12(\alpha_2+\alpha_3)$ and $t=\tfrac12(\alpha_2-\alpha_3)$, and $\alpha_2+\alpha_3=-a_2-\alpha_1\in\mathbb{Q}(\alpha_1)$ for $f=x^3+a_2x^2+\cdots$). Now
$$\sqrt D=(\alpha_1-\alpha_2)(\alpha_1-\alpha_3)(\alpha_2-\alpha_3)$$equals $f'(\alpha_1)\,(\alpha_2-\alpha_3)$, with $f'(\alpha_1)\in\mathbb{Q}(\alpha_1)$ nonzero (distinct roots, Theorem 7.47). So $g$ splits over $\mathbb{Q}(\alpha_1)$ iff $\sqrt D\in\mathbb{Q}(\alpha_1)$. Finally $[\mathbb{Q}(\sqrt D):\mathbb{Q}]\in\{1,2\}$ divides $3$ only if it is $1$: $\sqrt D\in\mathbb{Q}(\alpha_1)$ iff $\sqrt D\in\mathbb{Q}$. (3) The roots of $x^n-1$ are the powers of $\zeta_n$, so the splitting field is $\mathbb{Q}(\zeta_n)$; the irreducibility of $\Phi_n$ is a theorem of Gauss (for prime $n$) and Dedekind (in general) — for $n=p$ prime, Eisenstein applies to $\Phi_p(x+1)$ (§7.2). (4) The roots are $\sqrt[p]a\,\zeta_p^k$, so the splitting field is $\mathbb{Q}(\sqrt[p]a,\zeta_p)$, which contains subfields of degrees $p$ and $p-1$; these are coprime, so the degree is divisible by $p(p-1)$, and it is at most $p(p-1)$ by Theorem 7.43(2). (The irreducibility of $x^p-a$ for prime $p$ is a classical lemma, proved in the 📚 box; for $a=2$ it is Eisenstein.) For $x^4-2$: the roots are $\pm\sqrt[4]2$, $\pm i\sqrt[4]2$, the splitting field is $\mathbb{Q}(\sqrt[4]2,i)$, and $[\mathbb{Q}(\sqrt[4]2):\mathbb{Q}]=4$ with a real field, to which the non-real $i$ adds degree $2$. ∎
Finite Fields
- $\operatorname{char}F=p$ is prime, $F$ is a vector space over its prime subfield $\mathbb{F}_p$, and $|F|=p^n$ with $n=[F:\mathbb{F}_p]$. There is no field with $6$, $10$, $12$, $\dots$ elements.
- Freshman's dream. In characteristic $p$: $(a+b)^p=a^p+b^p$, and hence $(a+b)^{p^k}=a^{p^k}+b^{p^k}$.
- The Frobenius map is an automorphism of $F$ fixing $\mathbb{F}_p$ pointwise; so every element of a finite field of characteristic $p$ is a $p$-th power (in characteristic $2$, every element is a square).
- The additive group of $F$ is $(\mathbb{Z}_p)^n$ — not cyclic when $n\ge2$.
Proof. (1) A finite field cannot contain $\mathbb{Q}$, so its characteristic is a prime $p$ (Theorem 7.39), and its prime subfield is $\mathbb{F}_p$. As a vector space over $\mathbb{F}_p$ of finite dimension $n$, $F\cong\mathbb{F}_p^n$ has $p^n$ elements. (2) $(a+b)^p=\sum_{k=0}^p\binom pka^kb^{p-k}$, and for $0\lt k\lt p$ the integer $\binom pk=\frac{p!}{k!(p-k)!}$ is divisible by $p$ ($p$ divides the numerator but not the denominator), so those terms vanish. Iterate for $p^k$. (3) $\phi$ preserves products trivially and sums by (2), and $\phi(1)=1$; it is injective (Theorem 7.39(3)), hence surjective on the finite set $F$. It fixes $\mathbb{F}_p$ by Fermat's little theorem $a^p=a$ — or because the set of fixed points is a subfield, which must contain the prime subfield. (4) $F\cong\mathbb{F}_p^n$ as an additive group, and every nonzero element has additive order $p$. ∎
- $\mathbb{F}_q$ has exactly $\varphi(q-1)$ primitive elements, and for each $d\mid q-1$ exactly $\varphi(d)$ elements of order $d$;
- the equation $x^k=1$ has exactly $\gcd(k,q-1)$ solutions in $\mathbb{F}_q$;
- every $a\in\mathbb{F}_q^\times$ satisfies $a^{q-1}=1$, and every $a\in\mathbb{F}_q$ satisfies $a^q=a$;
- for odd $q$, exactly $\tfrac{q-1}2$ elements of $\mathbb{F}_q^\times$ are squares, and $-1$ is a square iff $q\equiv1\pmod4$;
- the product of all elements of $\mathbb{F}_q^\times$ is $-1$ (Wilson's theorem for $q=p$), and for $q\gt2$ the sum of all elements of $\mathbb{F}_q$ is $0$.
Proof. Let $|G|=N$ and, for $d\mid N$, let $\psi(d)$ be the number of elements of order $d$ in $G$. If $\psi(d)\ge1$, take $g$ of order $d$: the $d$ distinct powers $1,g,\dots,g^{d-1}$ are roots of $x^d-1$, which has at most $d$ roots in a field, so they are all the roots, and every element of order $d$ is among them. In the cyclic group $\langle g\rangle$ of order $d$ exactly $\varphi(d)$ elements have order $d$ (§7.1). Hence $\psi(d)\in\{0,\varphi(d)\}$, so $\psi(d)\le\varphi(d)$ for all $d\mid N$. Every element has some order dividing $N$ (Lagrange), so
$$N=\sum_{d\mid N}\psi(d)\le\sum_{d\mid N}\varphi(d)=N,$$using Gauss's identity $\sum_{d\mid N}\varphi(d)=N$ (count the elements of $\mathbb{Z}_N$ by their order). Equality forces $\psi(d)=\varphi(d)$ for every $d$; in particular $\psi(N)=\varphi(N)\ge1$: $G$ has an element of order $N$ and is cyclic. (1) is what we just proved. (2) In a cyclic group of order $N$, the solutions of $x^k=1$ form the unique subgroup of order $\gcd(k,N)$. (3) Lagrange in $\mathbb{F}_q^\times$; multiply by $a$, and note $0^q=0$. (4) With a generator $g$, the squares are the even powers $g^{2j}$, half of the group; $-1$ is the unique element of order $2$, namely $g^{(q-1)/2}$, which is an even power iff $4\mid q-1$. (5) Pair each $a$ with $a^{-1}$; only $a=\pm1$ are their own inverses, so the product is $1\cdot(-1)=-1$ (in characteristic $2$, $-1=1$). The sum: $\sum_{a}a=\sum_a ca$ for any $c\ne0,1$, so $(c-1)\sum a=0$. ∎
Proof. Uniqueness. If $|F|=q$, every element is a root of $x^q-x$ (Theorem 7.50(3)), which has at most $q$ roots; so $F$ consists precisely of the $q$ roots, $x^q-x$ splits over $F$ as $\prod_{a\in F}(x-a)$, and $F$ is generated by these roots: $F$ is a splitting field of $x^q-x$ over $\mathbb{F}_p$. Splitting fields are unique up to isomorphism (Theorem 7.46(2)). Existence. Let $K$ be a splitting field of $x^q-x$ over $\mathbb{F}_p$ and $S\subseteq K$ its set of roots. By Theorem 7.47(3), $|S|=q$. $S$ is a subfield: if $a^q=a$ and $b^q=b$, then $(a\pm b)^q=a^q\pm b^q=a\pm b$ (freshman's dream), $(ab)^q=ab$, and $(a^{-1})^q=a^{-1}$. So $S$ is a field with $q$ elements. Finally, $\mathbb{F}_p[x]/\langle f\rangle$ is a field with $p^n$ elements (Theorem 7.40(3)), hence it is $\mathbb{F}_{p^n}$. ∎
Proof. If $L\subseteq\mathbb{F}_{p^n}$ has $p^m$ elements, the tower law gives $n=[\mathbb{F}_{p^n}:\mathbb{F}_p]=[\mathbb{F}_{p^n}:L]\cdot m$, so $m\mid n$; and $L$ consists of roots of $x^{p^m}-x$ (Theorem 7.50(3)), of which there are at most $p^m$, so $L=\{a:a^{p^m}=a\}$ is unique. Conversely, if $n=mk$, then $p^m-1$ divides $p^n-1=(p^m)^k-1$, so $x^{p^m-1}-1$ divides $x^{p^n-1}-1$, which splits into distinct linear factors over $\mathbb{F}_{p^n}$. Thus $x^{p^m}-x$ has $p^m$ distinct roots in $\mathbb{F}_{p^n}$, and they form a subfield by the argument in the proof of Theorem 7.51. ∎
- $x^{q^n}-x$ is the product of all monic irreducible polynomials over $\mathbb{F}_q$ whose degree divides $n$.
- Counting degrees: $\displaystyle\sum_{d\mid n}d\,N_q(d)=q^n$, and by Möbius inversion $\displaystyle N_q(n)=\frac1n\sum_{d\mid n}\mu(d)\,q^{n/d}$. For $n$ prime, $N_q(n)=\dfrac{q^n-q}{n}$.
- If $f$ is irreducible of degree $n$ over $\mathbb{F}_q$ and $\alpha$ is one root, its roots are $\alpha,\alpha^q,\alpha^{q^2},\dots,\alpha^{q^{n-1}}$, all distinct, and all lie in $\mathbb{F}_q(\alpha)=\mathbb{F}_{q^n}$: an irreducible polynomial over a finite field splits in the field generated by one root. Its splitting field has degree $n$ over $\mathbb{F}_q$.
- The elements $\alpha\in\mathbb{F}_{q^n}$ with $\mathbb{F}_q(\alpha)=\mathbb{F}_{q^n}$ number $n\,N_q(n)$: all elements minus those lying in a proper subfield $\mathbb{F}_{q^d}$, $d\mid n$, $d\lt n$.
- The automorphisms of $\mathbb{F}_{p^n}$ are $\phi^0,\phi,\dots,\phi^{n-1}$ (powers of Frobenius); they form a cyclic group of order $n$.
Proof. (1) $x^{q^n}-x$ has distinct roots (Theorem 7.47(3)), so it is a product of distinct monic irreducibles. If $f$ is monic irreducible of degree $d$ and $f\mid x^{q^n}-x$, a root $\alpha$ of $f$ lies in $\mathbb{F}_{q^n}$, and $\mathbb{F}_q(\alpha)$ is a subfield of size $q^d$, so $d\mid n$ by Theorem 7.52 (applied to $\mathbb{F}_{q^n}\supseteq\mathbb{F}_q(\alpha)$, comparing degrees over $\mathbb{F}_q$ by the tower law). Conversely, if $d\mid n$, then $K=\mathbb{F}_q[x]/\langle f\rangle$ has $q^d$ elements, so its element $\theta$ satisfies $\theta^{q^d}=\theta$, hence $\theta^{q^n}=\theta$ (apply $a\mapsto a^{q^d}$ repeatedly); thus $f=m_\theta\mid x^{q^n}-x$. (2) Compare degrees in (1); Möbius inversion of $g(n)=\sum_{d\mid n}h(d)$ is $h(n)=\sum_{d\mid n}\mu(d)g(n/d)$. For prime $n$ the divisors are $1$ and $n$: $nN_q(n)=q^n-q$. (3) $a\mapsto a^q$ fixes $\mathbb{F}_q$ and preserves sums and products, so $f(\alpha^{q^j})=f(\alpha)^{q^j}=0$. If $\alpha^{q^i}=\alpha^{q^j}$ with $0\le i\lt j\lt n$, applying $a\mapsto a^{q^{n-j}}$ gives $\alpha^{q^r}=\alpha^{q^n}=\alpha$ with $r=n-j+i$, $0\lt r\lt n$. The elements $a$ with $a^{q^r}=a$ form a subfield containing $\mathbb{F}_q$ and $\alpha$ (same argument as in Theorem 7.51) with at most $q^r\lt q^n$ elements, so it cannot contain $\mathbb{F}_q(\alpha)$, which has $q^n$ elements — a contradiction. So $f$ has the $n$ distinct roots $\alpha^{q^j}$, all in $\mathbb{F}_q(\alpha)$. (4) Each such $\alpha$ has degree $n$, so its minimal polynomial is one of the $N_q(n)$ irreducibles of degree $n$, each of which has $n$ roots in $\mathbb{F}_{q^n}$. (5) Sketch. An automorphism $\sigma$ is determined by $\sigma(g)$ for a primitive element $g$, and $\sigma(g)$ must be a root of $m_{g,\mathbb{F}_p}$, which has $n$ roots $g^{p^j}$; so there are at most $n$ automorphisms, and $\phi^j$, $0\le j\lt n$, are $n$ distinct ones ($\phi^j=\mathrm{id}$ means every element satisfies $a^{p^j}=a$, impossible for $0\lt j\lt n$). Details in the 📚 box. ∎
- Sizes: $|F|=p^n$; $\mathbb{F}_p[x]/\langle f\rangle$ has $p^{\deg f}$ elements and is a field iff $f$ is irreducible.
- Orders of elements: $\mathbb{F}_q^\times\cong\mathbb{Z}_{q-1}$, so everything is the arithmetic of a cyclic group: $\varphi(q-1)$ generators, $\varphi(d)$ elements of order $d\mid q-1$, $\gcd(k,q-1)$ solutions of $x^k=1$, $\tfrac{q-1}{\gcd(k,q-1)}$ nonzero $k$-th powers.
- Subfields: $\mathbb{F}_{p^m}\subseteq\mathbb{F}_{p^n}$ iff $m\mid n$; count with the divisors of $n$, and use inclusion–exclusion over the maximal proper subfields $\mathbb{F}_{p^{n/r}}$, $r$ a prime divisor of $n$.
- Irreducibles: $N_q(n)=\frac1n\sum_{d\mid n}\mu(d)q^{n/d}$; for small $n$ just subtract: $N_2(4)=\frac{16-4}{4}=3$.
- Arithmetic: represent $\mathbb{F}_{p^n}$ as $\mathbb{F}_p[\alpha]$ with $f(\alpha)=0$; list the powers of a primitive element ("log table") to multiply and invert quickly.
- Algebraic does not mean finite. $\overline{\mathbb{Q}}/\mathbb{Q}$ and $\mathbb{Q}(\sqrt2,\sqrt3,\sqrt5,\dots)/\mathbb{Q}$ are algebraic of infinite degree. Finite $\Rightarrow$ algebraic, not conversely.
- Degrees of compositums do not simply multiply. $[\mathbb{Q}(\sqrt2,\sqrt8):\mathbb{Q}]=2$, not $4$; $[\mathbb{Q}(\sqrt[3]2,\omega\sqrt[3]2):\mathbb{Q}]=6$, not $9$. The product $mn$ is an upper bound, attained for coprime degrees.
- The degree of $\alpha$ is the degree of its minimal polynomial. $\sqrt2$ is a root of $x^4-4$, but $\deg\sqrt2=2$; $1+i$ is a root of $x^4+4$, but has degree $2$.
- The minimal polynomial depends on the base field: $\sqrt2+\sqrt3$ has degree $4$ over $\mathbb{Q}$ and $2$ over $\mathbb{Q}(\sqrt2)$ (minimal polynomial $x^2-2\sqrt2x-1$).
- "No roots" proves irreducibility only in degrees $2$ and $3$: $x^4+4$ over $\mathbb{Q}$ and $x^4+x^2+1=(x^2+x+1)^2$ over $\mathbb{F}_2$ have no roots but factor.
- Isomorphic as vector spaces $\ne$ isomorphic as fields. $\mathbb{Q}(\sqrt2)$ and $\mathbb{Q}(\sqrt3)$ both have degree $2$, but a field isomorphism would send $\sqrt2$ to an element of $\mathbb{Q}(\sqrt3)$ with square $2$, and there is none.
- $\mathbb{F}_4\ne\mathbb{Z}_4$, and $\mathbb{F}_{p^n}\ne\mathbb{Z}_{p^n}$ for $n\ge2$. $\mathbb{Z}_{p^n}$ has zero divisors.
- $\mathbb{F}_{p^m}\subseteq\mathbb{F}_{p^n}$ needs $m\mid n$, not $m\le n$: $\mathbb{F}_4\not\subseteq\mathbb{F}_8$, $\mathbb{F}_8\not\subseteq\mathbb{F}_{16}$.
- Cyclic multiplicatively, not additively. $\mathbb{F}_q^\times\cong\mathbb{Z}_{q-1}$, but $(\mathbb{F}_{p^n},+)\cong\mathbb{Z}_p^n$.
- A root of an irreducible polynomial need not be primitive: $x^2+1$ is irreducible over $\mathbb{F}_3$, but its root $i$ in $\mathbb{F}_9$ has order $4$, not $8$; $x^4+x^3+x^2+x+1$ over $\mathbb{F}_2$ has roots of order $5$ in $\mathbb{F}_{16}$.
- Irreducible over $\mathbb{Q}$ does not mean irreducible mod $p$: $x^4+1$ factors modulo every prime.
- Characteristic $p$ does not mean finite: $\mathbb{F}_p(t)$, $\overline{\mathbb{F}}_p$. And over $\mathbb{F}_p(t)$ the irreducible $x^p-t$ has only one root (multiplicity $p$).
- Algebraically closed fields are infinite — no finite field is algebraically closed ($1+\prod_a(x-a)$ has no root) — and $\mathbb{C}$ is not the algebraic closure of $\mathbb{Q}$ ($\overline{\mathbb{Q}}$ is, and it is countable). $\mathbb{R}$ and $\overline{\mathbb{Q}}\cap\mathbb{R}$ are not algebraically closed.
- The splitting field depends on the base field and on the factors: $x^4+4$ has splitting field of degree $2$ over $\mathbb{Q}$, $x^4-2$ degree $8$, $x^4+1$ degree $4$; over $\mathbb{F}_q$ every irreducible of degree $n$ has splitting field of degree exactly $n$.
- $N_q(n)\ne\frac{q^n-q}{n}$ in general. That formula holds for prime $n$; for $n=4$, $q=2$ it gives $\tfrac{16-2}{4}$, not even an integer — Möbius gives the correct $\tfrac14(16-4)=3$.
- CSIR Unit 2 · §8.1 Field Extensions and Splitting Fields — the tower law and Kronecker's theorem in detail, extension of isomorphisms and the complete uniqueness proof for splitting fields, existence of algebraic closures, ruler-and-compass constructions, the irreducibility of $x^p-a$, separability, perfect fields and the primitive element theorem.
- CSIR Unit 2 · §8.2 Finite Fields — full proofs of existence and uniqueness of $\mathbb{F}_{p^n}$, cyclicity of $\mathbb{F}_q^\times$, the subfield lattice, factorisation of $x^{q^n}-x$, the automorphism group, cyclotomic polynomials mod $p$, and the trace map with quadratics in characteristic $2$.
- CSIR Unit 2 · §8.3 Galois Theory — automorphisms and fixed fields, normal and Galois extensions, the fundamental theorem, why $[K:F]$ divides $n!$, the discriminant criterion for cubics via $A_3$, and solvability by radicals.
Two Labs
Exercises
- 1 Show that $\mathbb{Z}_n$ is a field if and only if $n$ is prime, and that a finite integral domain is a field.
- 2 Which of these are fields? Give the number of elements of the finite ones. (a) $\mathbb{Z}_3[x]/\langle x^3+2x+1\rangle$; (b) $\mathbb{Z}_2[x]/\langle x^3+x^2+x+1\rangle$; (c) $\mathbb{Q}[x]/\langle x^2-2x-1\rangle$; (d) $\mathbb{R}[x]/\langle x^2+4\rangle$; (e) $\mathbb{Z}_5[x]/\langle x^2+x+1\rangle$.
- 3 With $a=\sqrt[3]2$, find $(1+a+a^2)^{-1}$ in the form $c_0+c_1a+c_2a^2$.
- 4 Find $[\mathbb{Q}(\sqrt2,\sqrt3,\sqrt5):\mathbb{Q}]$. (You may use that $\sqrt5\notin\mathbb{Q}(\sqrt2,\sqrt3)$; prove it if you can.)
- 5 Find $[\mathbb{Q}(\sqrt[4]2,i):\mathbb{Q}]$ and $[\mathbb{Q}(\sqrt[4]2,i):\mathbb{Q}(i)]$.
- 6 Show that $[\mathbb{C}:\mathbb{R}]=2$ and that there is no field $K$ with $\mathbb{R}\subsetneq K\subsetneq\mathbb{C}$.
- 7 Let $[K:F]=p$ be prime. Show that $K=F(\alpha)$ for every $\alpha\in K\setminus F$.
- 8 Show that $\mathbb{Q}(\sqrt2)$ and $\mathbb{Q}(\sqrt3)$ are not isomorphic as fields. Are $\mathbb{Q}(\sqrt2)$ and $\mathbb{Q}(\sqrt8)$?
- 9 Prove that $[\mathbb{Q}(\sqrt2+\sqrt[3]2):\mathbb{Q}]=6$ without computing a minimal polynomial. (Hint: show $\mathbb{Q}(\sqrt2+\sqrt[3]2)$ has degree divisible by $2$ and by $3$ — or show $\sqrt2,\sqrt[3]2$ lie in it.)
- 1 Find the minimal polynomial of $\sqrt2+\sqrt5$ over $\mathbb{Q}$.
- 2 Find the minimal polynomial of $\sqrt2+\sqrt3$ over $\mathbb{Q}(\sqrt2)$ and over $\mathbb{Q}(\sqrt6)$.
- 3 Show that $\alpha=2\cos\frac{2\pi}7$ is a root of $x^3+x^2-2x-1$ (use $\zeta=\zeta_7$, $\alpha=\zeta+\zeta^{-1}$ and $1+\zeta+\cdots+\zeta^6=0$) and deduce $[\mathbb{Q}(\cos\frac{2\pi}7):\mathbb{Q}]=3$.
- 4 Find the minimal polynomial of $\zeta_{12}=e^{i\pi/6}$ over $\mathbb{Q}$ and $[\mathbb{Q}(\zeta_{12}):\mathbb{Q}]$.
- 5 Find the minimal polynomial of $1+\sqrt[3]2$ over $\mathbb{Q}$, and prove that it is irreducible.
- 6 Prove or disprove: if $\alpha^3$ is algebraic over $\mathbb{Q}$, then $\alpha$ is; if $\alpha+\beta$ and $\alpha\beta$ are algebraic, then $\alpha$ and $\beta$ are; if $\alpha+\beta$ is algebraic, then $\alpha$ and $\beta$ are.
- 7 Can $\sqrt[3]2$ lie in a field $K$ with $[K:\mathbb{Q}]=8$? Can $\sqrt2$ lie in a field of degree $9$? Can $\sqrt[3]2$ lie in a field of degree $12$?
- 8 Show that $\cos1^\circ$ is algebraic, and that $\mathbb{Q}(\pi)$ is isomorphic to the field $\mathbb{Q}(x)$ of rational functions.
- 9 Let $\alpha$ be algebraic over $F$ of odd degree. Show that $F(\alpha)=F(\alpha^2)$.
- 1 Find the degree of the splitting field of $x^4-2$ over $\mathbb{Q}$, over $\mathbb{Q}(i)$ and over $\mathbb{R}$.
- 2 Find the degree of the splitting field of $x^4+1$ over $\mathbb{Q}$ and over $\mathbb{F}_3$.
- 3 Find the degree over $\mathbb{Q}$ of the splitting field of $x^6-1$, of $x^4+x^2+1$, and of $x^8-1$.
- 4 Find the degree over $\mathbb{Q}$ of the splitting field of $x^3-7x+7$ and of $x^3+3x+1$.
- 5 Find the degree over $\mathbb{Q}$ of the splitting field of $x^5-2$ and of $(x^2-2)(x^2-3)(x^2-6)$.
- 6 Show that $x^3-2$ is irreducible over $\mathbb{F}_7$ and find the degree of its splitting field over $\mathbb{F}_7$. Compare with Example 2(b).
- 7 Show that every algebraically closed field is infinite, and that $\mathbb{R}$ is not algebraically closed. Is $\mathbb{Q}(i)$?
- 8 Let $F$ be algebraically closed and $K/F$ algebraic. Show that $K=F$, and deduce that $\mathbb{C}$ has no finite extension of degree $\gt1$.
- 9 Prove or disprove: (a) $\overline{\mathbb{Q}}$ is countable; (b) every subfield of $\mathbb{C}$ containing $\overline{\mathbb{Q}}$ is algebraically closed; (c) the algebraic closure of $\mathbb{F}_p$ is a finite extension of $\mathbb{F}_p$.
- 1 Construct $\mathbb{F}_4=\mathbb{F}_2[\alpha]$ with $\alpha^2=\alpha+1$. Write its multiplication table and show that its additive group is not cyclic.
- 2 In $\mathbb{F}_8=\mathbb{F}_2[\alpha]$, $\alpha^3=\alpha+1$, compute $(\alpha^2+\alpha)^{-1}$ and $\alpha^{10}$.
- 3 In $\mathbb{F}_9=\mathbb{F}_3[i]$, $i^2=-1$, find the order of $i$ and all primitive elements.
- 4 Find the number of monic irreducible polynomials of degree $2$ and of degree $3$ over $\mathbb{F}_5$.
- 5 How many elements $\alpha\in\mathbb{F}_{2^{12}}$ satisfy $\mathbb{F}_2(\alpha)=\mathbb{F}_{2^{12}}$? Deduce $N_2(12)$.
- 6 Show that $\phi(a)=a^p$ is an automorphism of $\mathbb{F}_{p^n}$ of order exactly $n$, with fixed field $\mathbb{F}_p$.
- 7 How many elements of $\mathbb{F}_{125}^\times$ are squares? Is $-1$ a square in $\mathbb{F}_{125}$? In $\mathbb{F}_{49}$?
- 8 Show that the product of all nonzero elements of $\mathbb{F}_q$ is $-1$ and the sum of all elements is $0$ for $q\gt2$. Check both in $\mathbb{F}_8$ with the log table of Example 5.
- 9 Show that $x^4+x+1$ is irreducible over $\mathbb{F}_2$ and that its roots are primitive elements of $\mathbb{F}_{16}$.
- 10 Applied. A Reed–Solomon code used in a storage system works over $\mathbb{F}_{16}$ and needs an element $\beta$ of order exactly $5$ to define a short code of length $5$. How many choices of $\beta$ are there, and what is the minimal polynomial of each over $\mathbb{F}_2$?
E1. GATE MA · MCQ The degree $[\mathbb{Q}(\sqrt2,\sqrt[3]2,i):\mathbb{Q}]$ is (A) $6$; (B) $8$; (C) $12$; (D) $18$.
E2. GATE MA · MCQ Which of the following rings is a field? (A) $\mathbb{Z}_5[x]/\langle x^2+1\rangle$; (B) $\mathbb{Z}_7[x]/\langle x^2+1\rangle$; (C) $\mathbb{Q}[x]/\langle x^4+4\rangle$; (D) $\mathbb{Z}_2[x]/\langle x^4+x^2+1\rangle$.
E3. GATE MA · MSQ Let $F$ be a field with $81$ elements. Which of the following is/are true? (One or more options may be correct.) (A) $F$ has a subfield with $9$ elements. (B) $F$ has a subfield with $27$ elements. (C) The multiplicative group $F^\times$ has exactly $32$ generators. (D) The polynomial $x^2+x+1$ has no root in $F$.
E4. GATE MA · MSQ Which of the following statements is/are true? (One or more options may be correct.) (A) Every algebraically closed field is infinite. (B) The set of complex numbers that are algebraic over $\mathbb{Q}$ is a countable field. (C) Every algebraic extension of $\mathbb{Q}$ is a finite extension. (D) If $K$ is an algebraically closed field containing $\mathbb{Q}$, then $K$ is an algebraic closure of $\mathbb{Q}$.
E5. GATE MA · NAT The number of monic irreducible polynomials of degree $4$ over the field $\mathbb{F}_3$ is ______.
E6. GATE MA · NAT The degree over $\mathbb{Q}$ of the splitting field of $(x^3-2)(x^2-3)$ is ______.
Suggested plan — 3 lectures × 60 min.
- Fields and degrees. Definition 7.27, Theorem 7.39 (why the characteristic is prime), Theorem 7.40 and Example 1 as a quick-fire "field or not?" round — make the class find the factorisation of $x^4+4$. Definition 7.28, the tower law (Theorem 7.41) with the basis-product proof on the board, Definition 7.29 and Theorem 7.42. Example 2 in full, emphasising the question "is the new element already in my field?".
- Algebraic extensions and splitting fields. Theorem 7.43 (degree of an element divides the degree), Theorem 7.44 with the transitivity proof, Hermite–Lindemann as facts, Definition 7.31 and Theorem 7.45 (the one-line proof that finite fields are not algebraically closed is a favourite). Example 3 as a prove-or-disprove discussion. Then Definition 7.32, Theorems 7.46 and 7.48; project Interactive 2 on $x^3-2$, $x^4-2$ and the pair $x^4-10x^2+1$ / $(x^2-2)(x^2-3)$. Example 4, ending with GATE corner E6.
- Finite fields. Theorem 7.49 (freshman's dream), Theorem 7.50 with the counting proof of cyclicity, Theorems 7.51–7.53. Project Interactive 1: on $\mathbb{F}_{16}$ step $s=1,\dots,14$ and ask the class to predict each order before it appears; on $\mathbb{F}_9$ point out that $i$ has order $4$. Examples 5 and 6 on the board, the toolkit box, and GATE corner E3 and E5.
Discussion questions. (1) Why must the minimal polynomial of a field element be irreducible, while the minimal polynomial of a matrix (§3.2) need not be? (2) The splitting field of $x^3-2$ has degree $6$ over $\mathbb{Q}$, degree $3$ over $\mathbb{F}_7$, and degree $2$ over $\mathbb{F}_5$ (there $3\nmid4$ makes cubing a bijection, so $2=3^3$ is a cube, yet the other two roots are missing). Explain all three answers through the cube roots of unity: where do they live in each case? (3) $\mathbb{Q}(\sqrt2)\cong\mathbb{Q}(\sqrt3)$ as vector spaces but not as fields; $\mathbb{F}_p[x]/\langle f\rangle\cong\mathbb{F}_p[x]/\langle g\rangle$ as fields for any two irreducibles of the same degree. Why the difference?
Misconceptions to watch for. Taking the degree of any polynomial $\alpha$ satisfies as $\deg\alpha$; multiplying degrees of generators without checking whether one is already in the field below; "no roots $\Rightarrow$ irreducible" for quartics; confusing $\mathbb{F}_4$ with $\mathbb{Z}_4$; "$\mathbb{F}_{p^m}\subseteq\mathbb{F}_{p^n}$ iff $m\le n$"; "algebraic $\Rightarrow$ finite"; "$\mathbb{C}$ is the algebraic closure of $\mathbb{Q}$"; assuming every root of an irreducible polynomial over $\mathbb{F}_p$ is primitive; using $\frac{q^n-q}{n}$ for composite $n$.
Suggested homework. I(2, 4, 5, 7, 9), II(1, 3, 6, 7, 9), III(1, 2, 4, 6, 7), IV(2, 3, 5, 7, 9, 10); GATE corner E1, E4, E5.
Research question. The encryption standard AES, used worldwide in Wi-Fi, online banking and messaging apps, and the error-correcting codes in QR codes and data storage, all compute in a field with $256$ elements, so that one byte is one field element. Which field, which polynomial, and why does the choice of polynomial matter?
Why it matters. A byte $b_7b_6\dots b_0$ is read as the polynomial $b_7x^7+\cdots+b_0$ over $\mathbb{F}_2$; addition is XOR, and multiplication is polynomial multiplication reduced modulo a fixed polynomial of degree $8$. AES uses $m(x)=x^8+x^4+x^3+x+1$ (the hexadecimal number 0x11B) and needs inverses in this field for its "S-box"; the Reed–Solomon codes of QR codes use $x^8+x^4+x^3+x^2+1$ (0x11D) and need a primitive element to build their code. By Theorem 7.51 every irreducible choice gives "the same" field $\mathbb{F}_{256}$ — but not the same arithmetic tables.
What to do.
- Show that $m(x)=x^8+x^4+x^3+x+1$ is irreducible over $\mathbb{F}_2$: check that it has no root and is not divisible by any of the $1+2+3=6$ irreducible polynomials of degrees $2,3,4$. (Why is that enough?)
- Write a short program that multiplies bytes modulo $m(x)$ (shift-and-XOR). Compute the multiplicative order of $x$ (byte
0x02) and of $x+1$ (byte0x03). Which is a primitive element? Explain why the order of0x02must divide $255=3\cdot5\cdot17$. - Using a primitive element $g$, build the "log" and "antilog" tables ($k\mapsto g^k$). Use them to compute the inverse of the byte
0x53, and check by multiplying. - Count the primitive elements of $\mathbb{F}_{256}$ and the primitive polynomials of degree $8$ over $\mathbb{F}_2$ (Theorem 7.50, Theorem 7.53). Is
0x11Dprimitive? Is0x11B?
Tools: Python (integers with bitwise ^, <<, &), or a spreadsheet with the BITXOR function.
Hand in: the irreducibility check; the orders of 0x02 and 0x03; your log table (first $16$ entries) and the inverse of 0x53; the two counts of step 4 with a one-paragraph explanation of why QR-code designers prefer a primitive polynomial.
Going further. Read how a Reed–Solomon code of length $255$ uses the powers $g,g^2,\dots,g^{2t}$ of a primitive element to correct $t$ byte errors (as on CDs, DVDs and deep-space probes), and how AES combines the inversion $a\mapsto a^{-1}=a^{254}$ in $\mathbb{F}_{256}$ with an affine map over $\mathbb{F}_2$ to resist linear and differential attacks.
0x02 is $51$ (so $m$ is irreducible but not primitive) and 0x03 has order $255$ (primitive). The inverse of 0x53 is 0xCA. $\varphi(255)=128$ primitive elements, hence $128/8=16$ primitive polynomials of degree $8$; 0x11D is primitive ($x$ has order $255$ there), 0x11B is not.🧪 Self-check quiz
Q1. $[\mathbb{Q}(\sqrt2,\sqrt5):\mathbb{Q}]$ equals
Q2. The number of generators of the cyclic group $\mathbb{F}_{125}^\times$ is
Q3. Which of the following fields is algebraically closed?
Q4. Which of the following is not (isomorphic to) a subfield of $\mathbb{F}_{64}$?
Q5. The degree over $\mathbb{Q}$ of the splitting field of $x^3-2$ is