1.3 The characteristic of a field

One checks easily that the map

F,nn1F=def1F+1F++1F(n copies of 1F),\mathbb{Z}\rightarrow F,\quad n\mapsto n\cdot 1_{F}\overset{\textup{{def}}}{=}% 1_{F}+1_{F}+\cdots+1_{F}\quad(n\text{ copies of }1_{F}),

is a homomorphism of rings. For example,

(1F++1Fm)+(1F++1Fn)=1F++1Fm+n(\underbrace{1_{F}+\cdots+1_{F}}_{m})+(\underbrace{1_{F}+\cdots+1_{F}}_{n})=% \underbrace{1_{F}+\cdots+1_{F}}_{m+n}

because of the associativity of addition. Therefore its kernel is an ideal in \mathbb{Z}{}.

Case 1: The kernel of the map is (0)(0), so that

n1F=0(in F)n=0(in ).n\cdot 1_{F}=0\quad\text{(in }F\text{)}\implies n=0\quad\text{(in }\mathbb{Z}% \text{).}

Nonzero integers map to invertible elements of FF under nn1F:Fn\mapsto n\cdot 1_{F}\colon\mathbb{Z}{}\rightarrow F, and so this map extends to a homomorphism

mn(m1F)(n1F)1:F.\genfrac{}{}{}{}{\raisebox{-2.0pt}{$m$}}{\raisebox{2.0pt}{$n$}}\mapsto(m\cdot 1% _{F})(n\cdot 1_{F})^{-1}\colon\mathbb{Q}\hookrightarrow F.

In this case, FF contains a copy of \mathbb{Q}, and we say that it has characteristic zero.

Case 2: The kernel of the map is (0)\neq(0), so that n1F=0n\cdot 1_{F}=0 for some n0n\neq 0. The smallest positive such nn will be a prime pp (otherwise there will be two nonzero elements in FF whose product is zero), and pp generates the kernel. Thus, the map nn1F:Fn\mapsto n\cdot 1_{F}\colon\mathbb{Z}{}\rightarrow F defines an isomorphism from /p\mathbb{Z}/p\mathbb{Z}{} onto the subring

{m1Fm}\{m\cdot 1_{F}\mid m\in\mathbb{Z}\}

of FF. In this case, FF contains a copy of 𝔽p\mathbb{F}_{p}, and we say that it has characteristic pp.

A field isomorphic to one of the fields 𝔽2,𝔽,3𝔽,5,\mathbb{F}_{2},\mathbb{F}{}_{3},\mathbb{F}{}_{5},\ldots,\mathbb{Q} is called a prime field. Every field contains exactly one prime field (as a subfield).

1.4

More generally, a commutative ring RR is said to have characteristic pp (resp. 0) if it contains a prime field (as a subring) of characteristic pp (resp. 0).22 2 A commutative ring has a characteristic if and only if it contains a field as a subring. For example, neither \mathbb{Z}{} nor 𝔽×2𝔽3\mathbb{F}{}_{2}\times\mathbb{F}{}_{3} has a characteristic. Then the prime field is unique and, by definition, contains 1R1_{R}. Thus, if RR has characteristic p0p\neq 0, then 1R++1R=01_{R}+\cdots+1_{R}=0 (pp terms).

Let RR be a nonzero commutative ring. If RR has characteristic p0p\neq 0, then

pa=defa++ap terms=(1R++1R)p termsa=0a=0pa\overset{\smash{\lower 1.31395pt\hbox{{def}}}}{=}\underbrace{a+\cdots+a}_{p% \text{ terms}}=\underbrace{(1_{R}+\cdots+1_{R})}_{p\text{ terms}}a=0a=0

for all aRa\in R. Conversely, if pa=0pa=0 for all aRa\in R, then RR has characteristic pp.

Let RR be a nonzero commutative ring. The usual proof by induction shows that the binomial theorem

(a+b)m=am+(m1)am1b+(m2)am2b2++bm(a+b)^{m}=a^{m}+\tbinom{m}{1}a^{m-1}b+\tbinom{m}{2}a^{m-2}b^{2}+\cdots+b^{m}

holds in RR. If pp is prime, then it divides

(pr)=defp!r!(pr)!\begin{pmatrix}p\\ r\end{pmatrix}\overset{\smash{\lower 1.31395pt\hbox{{def}}}}{=}\frac{p!}{r!(p-% r)!}

for all rr with 1rp11\leq r\leq p-1 because it divides the numerator but not the denominator. Therefore, when RR has characteristic pp,

(a+b)p=ap+bpfor all a,bR,(a+b)^{p}=a^{p}+b^{p}\quad\text{for all }a,b\in R,

and so the map aap:RRa\mapsto a^{p}\colon R\rightarrow R is a homomorphism of rings (even of 𝔽p\mathbb{F}{}_{p}-algebras). It is called the Frobenius endomorphism of RR. The map aapn:RRa\mapsto a^{p^{n}}\colon R\rightarrow R, n1n\geq 1, is the composite of nn copies of the Frobenius endomorphism, and so it also is a homomorphism. Therefore,

(a1++am)pn=a1pn++ampn(a_{1}+\cdots+a_{m})^{p^{n}}=a_{1}^{p^{n}}+\cdots+a_{m}^{p^{n}}

for all aiRa_{i}\in R.

When FF is a field, the Frobenius endomorphism is injective, and hence an automorphism if FF is finite.

The characteristic exponent of a field FF is 11 if FF has characteristic 0, and pp if FF has characteristic p0p\neq 0. Thus, if qq is the characteristic exponent of FF and n1n\geq 1, then xxqnx\mapsto x^{q^{n}} is an isomorphism of FF onto a subfield of FF (denoted FqnF^{q^{n}}).