1.6 Extensions

Let FF be a field. A field containing FF is called an extension of FF.55 5 This is the usual definition of “extension” (Wikipedia: field extension), but “overfield” would be a better term because Bourbaki, for example, uses “extension” to mean a field EE together with a homomorphism from FF to EE. In other words, an extension is an FF-algebra whose underlying ring is a field. An extension EE of FF is, in particular, an FF-vector space, whose dimension is called the degree of EE over FF. It is denoted by [E:F][E\colon F]. An extension is said to finite if its degree is finite, and quadratic, cubic, etc. if it is of degree 22, 33, etc.

When EE and EE^{\prime} are extensions of FF, an FF-homomorphism EEE\rightarrow E^{\prime} is a homomorphism φ:EE\varphi\colon E\rightarrow E^{\prime} such that φ(c)=c\varphi(c)=c for all cFc\in F.

Example 1.19

(a) The field of complex numbers \mathbb{C} has degree 22 over \mathbb{R} (basis {1,i}).\{1,i\}).

(b) The field of real numbers \mathbb{R} has infinite degree over \mathbb{Q}: the field \mathbb{Q}{} is countable, and so every finite-dimensional \mathbb{Q}{}-vector space is also countable, but a famous argument of Cantor shows that \mathbb{R}{} is not countable.

(c) The field of Gaussian numbers

(i)=def{a+bia,b}\mathbb{Q}(i)\overset{\smash{\lower 1.31395pt\hbox{{def}}}}{=}\{a+bi\in\mathbb% {C}\mid a,b\in\mathbb{Q}\}

has degree 22 over \mathbb{Q} (basis {1,i}\{1,i\}).

(d) The field F(X)F(X) has infinite degree over FF; in fact, even its subspace F[X]F[X] has infinite dimension over FF (basis 1,X,X2,1,X,X^{2},\ldots).

Proposition 1.20 (multiplicativity of degrees)

Consider fields LEFL\supset E\supset F. Then L/FL/F is of finite degree if and only if L/EL/E and E/FE/F are both of finite degree, in which case

[L:F]=[L:E][E:F].[L\colon F]=[L\colon E][E\colon F].

Proof.

If LL is finite over FF, then it is certainly finite over EE; moreover, EE, being a subspace of a finite-dimensional FF-vector space, is also finite-dimensional.

Thus, assume that L/EL/E and E/FE/F are of finite degree, and let (ei)1im(e_{i})_{1\leq i\leq m} be a basis for EE as an FF-vector space and let (lj)1jn(l_{j})_{1\leq j\leq n} be a basis for LL as an EE-vector space. To complete the proof of the proposition, it suffices to show that (eilj)1im,1jn(e_{i}l_{j})_{1\leq i\leq m,1\leq j\leq n} is a basis for LL over FF, because then LL will be finite over FF of the predicted degree.

First, (eilj)i,j(e_{i}l_{j})_{i,j} spans LL. Let γL\gamma\in L. Then, because (lj)j(l_{j})_{j} spans LL as an EE-vector space,

γ=jαjlj,some αjE,\gamma={\textstyle\sum\nolimits_{j}}\alpha_{j}l_{j},\qquad\text{some }\alpha_{% j}\in E,

and because (ei)i(e_{i})_{i} spans EE as an FF-vector space,

αj=iaijei,some aijF.\alpha_{j}={\textstyle\sum\nolimits_{i}}a_{ij}e_{i},\qquad\text{some $a_{ij}% \in F$}.

On putting these together, we find that

γ=i,jaijeilj.\gamma={\textstyle\sum\nolimits_{i,j}}a_{ij}e_{i}l_{j}.

Second, (eilj)i,j(e_{i}l_{j})_{i,j} is linearly independent. A linear relation aijeilj=0\sum a_{ij}e_{i}l_{j}=0, aijFa_{ij}\in F, can be rewritten j(iaijei)lj=0\sum_{j}(\sum_{i}a_{ij}e_{i})l_{j}=0. The linear independence of the ljl_{j}’s now shows that iaijei=0\sum_{i}a_{ij}e_{i}=0 for each jj, and the linear independence of the eie_{i}’s shows that each aij=0a_{ij}=0.