Bourbaki

§ 15. Separable extensions

1. Characterization of the nilpotent elements of a ring

Proposition 1

— *Let A be a commutative ring and x an element of A. For x to be nilpotent it is necessary and sufficient that 1−xT1 - xT be invertible in the ring A[T]A[T].

We note that A[T]A[T] is a subring of the formal power series ring A[[T]]A[[T]] and that 1−xT1 - xT has in A[[T]]A[[T]] the inverse ∑n=0∞xnTn\sum_{n=0}^{\infty} x^n T^n (IV, p. 30, Prop. 5). For 1−xT1 - xT to be invertible in A[T]A[T] it is necessary and sufficient for ∑n=0∞xnTn\sum_{n=0}^{\infty} x^n T^n to be a polynomial, that is, for x to be nilpotent.

Proposition 2

— *Let A be a commutative ring. The set of nilpotent elements of A is an ideal of A equal to the intersection of the set of all prime ideals of A.

Let x be a nilpotent element of A and p a prime ideal of A. The residue class of xmod  px \mod p is a nilpotent element of the integral domain A/pA/p, hence is zero; thus we have x∈px \in p.

Let x be a non-nilpotent element of A. By Prop. 1 the principal ideal (1−xT)(1 - xT) of A[T]A[T] is distinct from A[T]A[T]. By Krull's theorem (I, p. 104) there exists a maximal ideal m of A[T]A[T] containing 1−xT1 - xT. Then m is a prime ideal of A[T]A[T], hence p=A∩mp = A \cap m is a prime ideal of A. We have 1∉m1 \notin m and 1−Tx∈m1 - Tx \in m, hence Tx∉mTx \notin m and a fortiori x∉px \notin p.

We have thus shown that the set n\mathfrak{n} of nilpotent elements of A is the intersection of the set of all prime ideals of A; since every intersection of ideals is an ideal, n\mathfrak{n} is an ideal.

Corollary

— For a commutative ring to be reduced (V, p. 34, Def. 2) it is necessary and sufficient for it to be isomorphic to a subring of a product of fields.

The condition is clearly sufficient.

Let A\mathbf{A} be reduced. By Prop. 2 the intersection nn of the set of prime ideals of A\mathbf{A} is reduced to 0. For every prime ideal p\mathfrak{p} of A\mathbf{A}, let k(p)k(\mathfrak{p}) be the field of fractions of A/p\mathbf{A}/\mathfrak{p} and φn\varphi_n the canonical homomorphism of A\mathbf{A} into k(p)k(\mathfrak{p}). Let φ\varphi be the homomorphism of A\mathbf{A} into ∏pk(p)\prod_{\mathfrak{p}} k(\mathfrak{p}) whose component of index p\mathfrak{p} is φp\varphi_{\mathfrak{p}}. The kernel of φp\varphi_{\mathfrak{p}} is p\mathfrak{p}, hence that of φ\varphi is n=0n = 0; therefore φ\varphi is an isomorphism of A\mathbf{A} onto a subring of ∏pk(p)\prod_{\mathfrak{p}} k(\mathfrak{p}).

2. Separable algebras

Definition 1

Let A\mathbf{A} be a commutative algebra over a field KK. Then AA is said to be separable over KK, or also a separable KK-algebra if the ring L⊗KAL \otimes_K \mathbf{A} is reduced for every extension LL of KK*.

Every separable algebra is clearly reduced. For a partial converse see Th. 3 (V, p. 125).

Examples. — 1) Let A\mathbf{A} be a polynomial algebra K[Xi]i∈IK[X_i]_{i \in I}. For every extension LL of KK the ring L⊗KAL \otimes_K \mathbf{A} is isomorphic to L[Xi]i∈IL[X_i]_{i \in I} (III, p. 449, Remark 2) and hence is an integral domain (IV, p. 9, Prop. 8). In other words, every polynomial algebra over a field KK is a separable KK-algebra.

2) Let A\mathbf{A} be a commutative algebra of finite degree over a field KK. For A\mathbf{A} to be separable it is necessary and sufficient for it to be etale (V, p. 34, Th. 4).

3) Let EE be an algebraic extension of a field KK. If LL is an extension of KK, then the ring L⊗KEL \otimes_K E is a union of the subrings L⊗KFL \otimes_K F where FF ranges over the set of all subextensions of finite degree of EE; therefore the ring L⊗KEL \otimes_K E is reduced if and only if this is true of L⊗KFL \otimes_K F for every subextension FF of EE, of finite degree over KK. Taking Example 2 into account we see that EE is a separable algebra in the sense of the above Def. 1 if and only if it is a separable algebraic extension in the sense of Def. 1 of V, p. 36.

Proposition 3

Let KK be a field*.

a) Every subalgebra of a separable KK-algebra is separable.

b) Every direct limit of separable KK-algebras is separable.

c) Every product of separable KK-algebras is separable.

d) Let A\mathbf{A} be a KK-algebra and K′K' an extension of KK. For A\mathbf{A} to be separable it is necessary and sufficient for the K′K'-algebra A(K′)\mathbf{A}_{(K')} obtained from A\mathbf{A} by extension of scalars to be separable.

Let LL be an extension of KK. Let A\mathbf{A} be a separable algebra over KK and BB a subalgebra of A\mathbf{A}; then the ring L⊗KAL \otimes_K \mathbf{A} is reduced, and L⊗KBL \otimes_K B is isomorphic to a subring of L⊗KAL \otimes_K \mathbf{A}, hence is reduced. Thus BB is separable and a) is proved. In the same way bb) may be proved, using the canonical isomorphism of L⊗Klim←⁡AiL \otimes_K \varprojlim A_i with lim←⁡L⊗KAi\varprojlim L \otimes_K A_i (II, p. 290, Prop. 7) and cc) is proved by remarking that L⊗K(∏i∈IAi)L \otimes_K \left( \prod_{i \in I} A_i \right) is isomorphic to a subring of ∏i∈I(L⊗KAi)\prod_{i \in I} (L \otimes_K A_i) (II, p. 306, Prop. 15).

We shall use the notation of dd). For every extension L′L' of K′K' the rings L′⊗K′A(K′)L' \otimes_{K'} A_{(K')} and L′⊗KAL' \otimes_K A are isomorphic (II, p. 278, Prop. 2). We conclude that if AA is a separable KK-algebra, then A(K′)A_{(K')} is a separable K′K'-algebra. Conversely suppose that A(K′)A_{(K')} is a separable K′K'-algebra ; the preceding remark shows that L′⊗KAL' \otimes_K A is reduced for every extension L′L' of KK containing K′K' as subextension. Let LL be an extension of KK; by the Scholium (V, p. 13) there exists an extension L′L' of KK containing K′K' as a subextension and a KK-homomorphism of LL into L′L'; the ring L⊗KAL \otimes_K A is thus isomorphic to a subring of L′⊗KAL' \otimes_K A and hence reduced. This proves that AA is a separable KK-algebra.

Proposition 4

— *Let AA be a separable algebra over a field KK and let BB be the total ring of fractions of AA; then the KK-algebra BB is separable.

Let SS be the set of cancellable elements of AA. We have identified AA with a subring of BB (I, p. 113); further, every element of SS is invertible in BB and every element of BB has the form as−1a s^{-1} with a∈Aa \in A and s∈Ss \in S. Let LL be an extension of KK and xx a nilpotent element of L⊗KBL \otimes_K B. Then xx may be written in the form

x=∑i=1nyi⊗aisi−1 x = \sum_{i=1}^n y_i \otimes a_i s_i^{-1}
with yi∈L,ai∈A,si∈Sy_i \in L, a_i \in A, s_i \in S for 1≤i≤n1 \leq i \leq n. If we put s=s1…sns = s_1 \ldots s_n then x(1⊗s)x(1 \otimes s) belongs to the subring L⊗KAL \otimes_K A of L⊗KBL \otimes_K B; since x(1⊗s)x(1 \otimes s) is nilpotent and AA separable over KK, we have x(1⊗s)=0x(1 \otimes s) = 0, and since ss is invertible in BB we find that x=0x = 0. This proves the ring L⊗KBL \otimes_K B to be reduced, whence the proposition.

Proposition 5

— *Let KK be a field and A,BA, B commutative KK-algebras. If AA is reduced and BB separable then A⊗KBA \otimes_K B is reduced.

By the Cor. of Prop. 2 (V, p. 119) AA is isomorphic to a subalgebra of a product ∏i∈ILi\prod_{i \in I} L_i where LiL_i is an extension of KK for i∈Ii \in I. Therefore A⊗KBA \otimes_K B is isomorphic to a subring of (∏i∈ILi)⊗KB\left( \prod_{i \in I} L_i \right) \otimes_K B and this latter ring is isomorphic to a subring of ∏i∈I(Li⊗KB)\prod_{i \in I} (L_i \otimes_K B) (II, p. 306, Prop. 15). Since BB is separable, each of the rings Li⊗KBL_i \otimes_K B is reduced, and so the same is true of ∏i∈I(Li⊗KB)\prod_{i \in I} (L_i \otimes_K B), and a fortiori of A⊗KBA \otimes_K B.

Corollary 1

— *Let KK be a field, LL a separable extension of KK and ff a polynomial in K[X]K[X]. If ff has no multiple factors in K[X]K[X], it has also no multiple factors in L[X]L[X].

If ff has no multiple factor in K[X]K[X], the quotient ring K[X]/(f)K[X]/(f) is reduced; for ff is a product of irreducible polynomials fif_i relatively prime in pairs, hence

K[X]/(f) is isomorphic, by I, p. 110, Prop. 9, to the product of fields K[X]/(f_i). By Prop. 5 the ring L[X]/(f), being isomorphic to L ⊗_K K[X]/(f ), is reduced; if g is a non-constant polynomial of L[X] such that g^2 divides f, then the residue class of fg^{-1} in L[X]/(f) is a non-zero nilpotent element; hence f has no multiple factors in L[X].

Corollary 2

— Let A and B be two commutative K-algebras. If A and B are separable, the same is true of A ⊗_K B.

Let L be an extension of K. The ring L ⊗_K A is reduced because A is separable; now Prop. 5 shows the ring (L ⊗_K A) ⊗_K B (isomorphic to L ⊗_K (A ⊗_K B)) to be reduced, whence the Corollary.

3. Separable extensions

Let K be a field. Since any extension of K is a K-algebra, the notion of separability introduced in Def. 1 (V, p. 119) applies in particular to the case of extensions of K. By example 3 (V, p. 119) that definition of separability agrees in the case of algebraic extensions with that of § 7 (V, p. 36, Def. 1).

Proposition 6

— Every pure extension of a field K is separable. This follows at once from Example 1 (V, p. 119) and Prop. 4 (V, p. 120).

Proposition 7

— Let E be a field, G a group of automorphisms of E and K the subfield of E consisting of the invariants of G. Then E is a separable extension of K.

Let L be an extension of K; there exists an algebraically closed extension Ω of L whose transcendence degree over K is at least equal to that of E over K. By Cor. 1 (V, p. 117) there exists a K-homomorphism u of E into R. We denote by v the Ω-algebra homomorphism of A = Ω ⊗_K E into Ω which maps λ ⊗ x to λ . u(x) for λ ∈ Ω and x ∈ Ω; we write a for the kernel of v.

For every s ∈ G let h_s be the automorphism Id, ⊗ s of the 0-algebra A; the kernel of the homomorphism v ∘ h_s of A into Ω is the prime ideal a_s = h_s^{-1}(a) of A. It is clear that the ideal b = ⋂_{s ∈ G} a_s of A is stable under the automorphisms h_s. Therefore (V, p. 63, Cor.) the ideal b is of the form c ⊗_K E, where c is an ideal of Ω. Now b ⊂ a ≠ A, and so c ≠ Ω; since Ω is a field, we have c = 0, hence b = 0.

The family (a_s), s ∈ G of prime ideals of A thus has zero intersection. By Prop. 2 (V, p. 118) the ring is reduced, and a fortiori the same holds of the subring L ⊗K E of A. Since L was an arbitrary extension of K, this proves that E is separable over K.

Proposition 8

— Let L be an extension of a field K. If L is separable over K, every subextension of L is separable over K. Conversely, if every finitely generated subextension of L is separable over K, then L is separable over K.

This follows at once from Prop. 3, a) and b) (V, p. 119).

Thus separability may be called a property of « finite character ».

Proposition 9

— Let L be an extension of a field K and M a commutative L-algebra (for example, an extension of L). If M is separable over L and L is separable over K, then M is separable over K.

Let K' be an extension of K. Since L is a separable extension of K, the ring K′⊗KLK' \otimes_K L is reduced; since M is a separable L-algebra, Prop. 5 (V, p. 120) shows the ring (K′⊗KL)⊗LM(K' \otimes_K L) \otimes_L M to be reduced. Now the ring K′⊗KMK' \otimes_K M is isomorphic to (K′⊗KL)⊗LM(K' \otimes_K L) \otimes_L M (II, p. 278, Prop. 2), hence is reduced. This proves that M is separable over K.

If the extension M is separable over K, it is not necessarily separable over L (cf. however V, p. 124, Cor. 3). For example, if p is a prime number, the field Fp(X)F_p(X) of rational fractions in one indeterminate X over FpF_p is separable over FpF_p (V, p. 121, Prop. 6) but it is a p-radical algebraic extension of Fp(Xp)F_p(X^p); in particular Fp(X)F_p(X) is not separable over Fp(Xp)F_p(X^p).

Later (V, p. 137 Prop. 5) we shall study the separability of composite extensions.

4. Mac Lane's separability criterion

Theorem 1

— Let K be a field of characteristic 0. Every reduced K-algebra and in particular every extension of K is separable over K.

We first show that every extension L of K is separable. Let B be a transcendence basis of L over K (V, p. 109, Th. 1) and let L1=K(B)L_1 = K(B). Then L1L_1 is separable over K (V, p. 121, Prop. 6). Further, L is an algebraic extension of L1L_1 and L1L_1 is a field of characteristic 0; therefore L is separable over L1L_1 (V, p. 37, Cor.). By Prop. 9, L is thus separable over K.

Now let A be a reduced algebra over the field K. By the Cor. of Prop. 2 (V, p. 119) there exists a family (Li)i∈I(L_i)_{i \in I} of extensions of K such that A is isomorphic to a subalgebra of ∏i∈ILi\prod_{i \in I} L_i. Each of the algebras LiL_i is separable over K by what has been said and so A has the same property, by Prop. 3 a) and c) (V, p. 119).

Theorem 2

— Let K be a field of characteristic p≠0p \neq 0, Kp−∞K^{p^{-\infty}} a perfect closure of K and A a commutative K-algebra. The following properties are equivalent: a) A is separable. b) There exists an extension K' of K such that K' is perfect and K′⊗KAK' \otimes_K A is reduced. c) The ring Kp−∞⊗KAK^{p^{-\infty}} \otimes_K A is reduced.

d) The ring K′⊗KAK' \otimes_K A is reduced for every extension K′K' of KK which is of finite degree and pp*-radical of height* ⩽1\leqslant 1.

e) For every family (ai)i∈I(a_i)_{i \in I}, of elements of AA linearly free over KK, the family (aip)i∈I(a_i^p)_{i \in I}, is linearly free over KK.

f) There exists a basis (ai)i∈I(a_i)_{i \in I}, of the vector KK*-space* AA suck that the family (aip)i∈I(a_i^p)_{i \in I}, is linearly free over KK.

If an extension K′K' of KK is a perfect field, it contains a subextension KK-isomorphic to Kp−∞K^{p^{-\infty}} (VV, p. 6, Prop. 3); moreover, every pp-radical extension of KK is isomorphic to a subextension of Kp−∞K^{p^{-\infty}} (VV, p. 26, Prop. 3). These remarks show the implications a)⇒b)⇒c)⇒d)a) \Rightarrow b) \Rightarrow c) \Rightarrow d).

Let us prove that d) implies e). Let (ai)i∈I(a_i)_{i \in I}, be a linearly free family in AA and let (Ai)i∈I(A_i)_{i \in I}, be a family of finite support in KK such that ∑λiaip=0\sum \lambda_i a_i^p = 0. Let K′K' be the subextension of Kp−∞K^{p^{-\infty}} generated by the elements λip−1\lambda_i^{p^{-1}}; it is of finite degree and height ⩽1\leqslant 1. Put x=∑i∈Iλip−1⊗aix = \sum_{i \in I} \lambda_i^{p^{-1}} \otimes a_i in K′⊗KAK' \otimes_K A; we have

xp=∑i∈Iλi⊗aip=1⊗∑i∈Iλiaip=0 x^p = \sum_{i \in I} \lambda_i \otimes a_i^p = 1 \otimes \sum_{i \in I} \lambda_i a_i^p = 0

By the hypothesis d) we have x=0x = 0, whence Ai=0A_i = 0 for all i∈Ii \in I.

Clearly e) implies f) and it remains to show that f) implies a). Thus let (ai)i∈I(a_i)_{i \in I} be a basis of AA over KK such that the family (aip)i∈I(a_i^p)_{i \in I} is linearly free over KK. Let LL be an extension of KK and let xx be an element of L⊗KAL \otimes_K A such that xp=0x^p = 0. Write x=∑i∈Iλi⊗aix = \sum_{i \in I} \lambda_i \otimes a_i with Ai∈LA_i \in L for all i∈Ii \in I. We have xp=∑i∈Iλip⊗aip=0x^p = \sum_{i \in I} \lambda_i^p \otimes a_i^p = 0 and since the family (aip)i∈I(a_i^p)_{i \in I}, is linearly free over KK, we have λip=0\lambda_i^p = 0, whence λi=0\lambda_i = 0 for all i∈Ii \in I; it follows that x=0x = 0. So we have proved that xp=0x^p = 0 implies x=0x = 0 in L⊗KAL \otimes_K A, from which it follows at once that L⊗KAL \otimes_K A is reduced.

Corollary 1 (Mac Lane)

— Let KK be a field of characteristic exponent pp, Ω\Omega a perfect extension of KK and LL a subextension of Ω\Omega. Then the following conditions are equivalent:

a) LL is separable over KK.

b) LL is linearly disjoint from Kp−∞K^{p^{-\infty}} over KK.

c) LL is linearly disjoint over KK from every pp*-radical extension of* KK contained in Ω\Omega, of finite degree and height ⩽1\leqslant 1.

The case where KK is of characteristic 0 is trivial because then LL is separable over KK (Th. 1) and Kp−∞=KK^{p^{-\infty}} = K by convention. Suppose then that p≠1p \neq 1, and let us first show that a) implies b). Assume that LL is separable over KK and let (ai)i∈I(a_i)_{i \in I}, be a basis of LL over KK. Let (λi)i∈I(\lambda_i)_{i \in I} be a family of finite support of elements of Kp−∞K^{p^{-\infty}} such that ∑i∈Iλiai=0\sum_{i \in I} \lambda_i a_i = 0; there exists an integer f⩾0f \geqslant 0 and elements μ_i of K such that λ_i = μ_i^{p^{-f}}. We have

∑i∈Iμiaipf=(∑i∈Iλiai)pf=0 \sum_{i \in I} \mu_i a_i^{p^f} = \left( \sum_{i \in I} \lambda_i a_i \right)^{p^f} = 0

and Th. 2 implies, by induction on f, that the family (aipf)i∈I(a_i^{p^f})_{i \in I} is linearly free over K. We thus have μi=0\mu_i = 0, whence λi=0\lambda_i = 0 for all i∈Ii \in I. Finally L is linearly disjoint from Kp−∞K^{p^{-\infty}} over K.

It is clear that b) implies c). Lastly suppose that c) holds and let K' be an extension of K of finite degree and p-radical of height ≤1\leq 1. The ring K′⊗KLK' \otimes_K L is isomorphic to a subring of Ω\Omega, hence is reduced. Now it follows from Th. 2 that L is separable over K.

Corollary 2

— *Let K be a field of characteristic exponent p, Kp−∞K^{p^{-\infty}} a perfect closure of K and L a separable extension of K. Then the ring L⊗KKp−∞L \otimes_K K^{p^{-\infty}} is a field. If moreover L is algebraic over K, then L⊗KKp−∞L \otimes_K K^{p^{-\infty}} is a perfect closure of L.

The case p=1p = 1 is trivial, so let us suppose p≠1p \neq 1. Let Ω\Omega be a perfect closure of L. By Cor. 1 there exists a K-algebra homomorphism of L⊗KKp−∞L \otimes_K K^{p^{-\infty}} onto L[Kp−∞]L[K^{p^{-\infty}}] which maps x⊗yx \otimes y to xyxy for x∈Lx \in L and y∈Kp−∞y \in K^{p^{-\infty}}. Since Kp−∞K^{p^{-\infty}} is algebraic over K, the ring L[Kp−∞]L[K^{p^{-\infty}}] is a subfield of Ω\Omega (V, p. 18, Cor. 1). Suppose further that L is algebraic over K, then L[Kp]L[K^p] is an algebraic extension of the perfect field Kp−∞K^{p^{-\infty}}, hence it is a perfect field (V, p. 43, Cor. 1); finally, since the field L[Kp−∞]L[K^{p^{-\infty}}] is ap-radical extension of L (V, p. 25, Cor.), it is a perfect closure of L.

Corollary 3

— *Let L be an algebraic extension of K and M a commutative L-algebra (for example an extension of L). If M is separable over K, it is separable over L.

The algebra L is K-isomorphic to a subalgebra of M, hence L is a separable extension of K. Therefore (Cor. 2) there exists an L-isomorphism of Lp−∞L^{p^{-\infty}} onto Kp−∞⊗KLK^{p^{-\infty}} \otimes_K L. The ring Lp−∞⊗LML^{p^{-\infty}} \otimes_L M is thus isomorphic to (Kp−∞⊗KL)⊗LM(K^{p^{-\infty}} \otimes_K L) \otimes_L M, and so to Kp−∞⊗KMK^{p^{-\infty}} \otimes_K M (II, p. 278, Prop. 2) and this latter ring is reduced because M is separable over K. Thus the ring Lp−∞⊗LML^{p^{-\infty}} \otimes_L M is reduced, which proves that M is separable over L (V, p. 122, Th. 2).

Remark

— Mac Lane's criterion may be formulated without introducing any extensions of K other than L. For by c) of Cor. 1, L is separable over K if and only if L and KPK^P are linearly disjoint over K. Since the mapping x↦xpx \mapsto x^p is an isomorphism of L onto the subfield LPL^P, we obtain the following criterion (cf. V, p. 177, Ex. 11 for an analogous criterion for algebras):

L is separable over K if and only if the subfields LPL^P and K of L are linearly disjoint over KPK^P.

5. Extensions of a perfect field

For ease of reference we summarize the principal properties of extensions of perfect fields:

Theorem 3

— Let K be a perfect field. a) Every algebraic extension of K is a perfect field. b) Every extension of K is separable. c) For a K-algebra to be separable it is necessary and sufficient that it should be reduced. d) Let A and B be two reduced K-algebras. Then A⊗KBA \otimes_K B is reduced. e) If E and F are two extensions of K, then the ring E⊗KFE \otimes_K F is reduced.

Assertion a) is just Cor. 1 of Prop. 11 (V, p. 43). Assertion b) follows from Th. 1 (V, p. 122) when K is of characteristic 0 and from Cor. 1 (V, p. 123) when K is of characteristic p≠0p \neq 0. Let us prove c). The case where K is of characteristic 0 follows from Th. 1 (V, p. 122). So it is enough to show that if K is perfect of characteristic p≠0p \neq 0 and A is a reduced K-algebra, then A is separable over K. But this follows from the equivalence of the conditions a) and b) of Th. 2 (V, p. 122; take K′=KK' = K in b)). Finally, d) follows from c) and Prop. 5 (V, p. 120) and e) is a particular case of d).

6. The characterization of separability by automorphisms

Theorem 4

— Let Ω\Omega be an algebraically closed extension of a field K and L a subextension of Ω\Omega. Then the following conditions are equivalent: a) L is separable over K. b) The intersection of the kernels of the 0-algebra homomorphisms of Ω⊗KL\Omega \otimes_K L into Ω\Omega is reduced to 0. c) For any elements a1,…,ana_1, \ldots, a_n of L linearly independent over K there exist K-automorphisms σ1,…,σn\sigma_1, \ldots, \sigma_n of Ω\Omega such that det⁡(σi(aj))≠0\det(\sigma_i(a_j)) \neq 0. d) Let V be a vector sub-K-space of L of finite dimension. Every K-linear mapping of V into Ω\Omega is a linear combination (with coefficients in R) of the restrictions to V of K-automorphisms of Ω\Omega.

d)⇒c)d) \Rightarrow c): Let a1,…,ana_1, \ldots, a_n be elements of L linearly independent over K and let V be the vector sub-K-space of L generated by a1,…,ana_1, \ldots, a_n. The mapping f↦(f(a1),…,f(an))f \mapsto (f(a_1), \ldots, f(a_n)) is an R-linear bijection of HomK(V,R)\mathrm{Hom}_K(V, R) onto Ωn\Omega^n. Suppose that d) holds, then there exist K-automorphisms a1,…,ana_1, \ldots, a_n of Ω\Omega such that the elements (σi(a1),…,σi(an))(\sigma_i(a_1), \ldots, \sigma_i(a_n)) of Ωn\Omega^n (for 1≤i≤n1 \leq i \leq n) form a basis of Ωn\Omega^n. We thus have det⁡(σi(aj))≠0\det(\sigma_i(a_j)) \neq 0, so d) implies c).

c) ⇒\Rightarrow b): Assume that c) holds and let x be in Ω⊗KL\Omega \otimes_K L. We write x in the form ∑j=1nxj⊗aj\sum_{j=1}^n x_j \otimes a_j, with x1,…,xnx_1, \ldots, x_n in Ω\Omega and the elements a1,…,ana_1, \ldots, a_n of L linearly independent over K. Choose K-automorphisms σ1,…,σn\sigma_1, \ldots, \sigma_n of Ω\Omega such that det⁡σi(aj)≠0\det \sigma_i(a_j) \neq 0; let χ1,…,χn\chi_1, \ldots, \chi_n be the R-homomorphisms of Ω⊗KL\Omega \otimes_K L into Ω\Omega such that χi(a⊗b)=a⋅σi(b)\chi_i(a \otimes b) = a \cdot \sigma_i(b) for a∈Ωa \in \Omega and b∈Lb \in L. Suppose that χi(x)=0\chi_i(x) = 0 for 1≤i≤n1 \leq i \leq n, in other words, that ∑j=1nxj⋅σi(aj)=0\sum_{j=1}^n x_j \cdot \sigma_i(a_j) = 0 for 1≤i≤n1 \leq i \leq n. Since we have assumed that the matrix (σi(aj))(\sigma_i(a_j)) has a non-zero determinant, we have xi=0x_i = 0 for 1≤i≤n1 \leq i \leq n and so x=0x = 0.

b) ⇒\Rightarrow a): Since every extension of a field of characteristic 0 is separable (V, p. 122, Th. 1) it suffices to examine the case where K is of characteristic p≠0p \neq 0. Let X be the set of all R-algebra homomorphisms of Ω⊗KL\Omega \otimes_K L into Ω\Omega and f the homomorphism of Ω⊗KL\Omega \otimes_K L into ΩX\Omega^X defined by f(u)=(χ(u))χ∈Xf(u) = (\chi(u))_{\chi \in X} for u∈Ω⊗KLu \in \Omega \otimes_K L. Condition b) means that f is injective and so implies that the ring Ω⊗KL\Omega \otimes_K L is reduced. Hence condition b) of Th. 2 (V, p. 122) is satisfied with K′=ΩK' = \Omega, so L is separable over K.

a) ⇒\Rightarrow d): Suppose that L is separable over K. Let V be a vector sub-K-space of finite dimension of L, V(Ω)=Ω⊗KVV_{(\Omega)} = \Omega \otimes_K V the vector Ω\Omega-space derived from V by extension of scalars and f0f_0 the linear form on V(Ω)V_{(\Omega)} such that f0(x⊗y)=xyf_0(x \otimes y) = xy for x∈Ω,y∈Vx \in \Omega, y \in V. Denote by G the group of K-automorphisms of Ω\Omega; for σ∈G\sigma \in G we put σV=σ⊗IdV\sigma_V = \sigma \otimes \mathrm{Id}_V and gσ=σ∘f0∘σV−1g_\sigma = \sigma \circ f_0 \circ \sigma_V^{-1}.

For each σ∈G\sigma \in G the mapping gσg_\sigma of V(Ω)V_{(\Omega)} into Ω\Omega is Ω\Omega-linear and maps x⊗yx \otimes y to x⋅σ(y)x \cdot \sigma(y) for x∈Ω,y∈Vx \in \Omega, y \in V. The kernel NσN_\sigma of gσg_\sigma is therefore a vector subspace of V(Ω)V_{(\Omega)}, and the same holds for N=∩σ∈GNσN = \cap_{\sigma \in G} N_\sigma. If p is the characteristic exponent of K, the field of invariants of G in Ω\Omega is equal to Kp−∞K^{p^{-\infty}} (V, p. 112, Prop. 10). We clearly have σV(N)=N\sigma_V(N) = N for all σ∈G\sigma \in G; therefore (V, p. 63, Cor.) the vector Ω\Omega-space N is generated by N0=N∩(Kp−∞⊗V)N_0 = N \cap (K^{p^{-\infty}} \otimes V). Since L is separable over K, the fields Kp−∞K^{p^{-\infty}} and L are linearly disjoint over K (V, p. 123, Cor. 1); we have Kp−∞⊗KV⊂Kp−∞⊗KLK^{p^{-\infty}} \otimes_K V \subset K^{p^{-\infty}} \otimes_K L and f0(x⊗y)=xyf_0(x \otimes y) = xy for x∈Ωx \in \Omega and y∈Vy \in V. It follows that the restriction of f0f_0 to Kp−∞⊗KVK^{p^{-\infty}} \otimes_K V is injective. Now f0=g1f_0 = g_1 is zero on N and a fortiori on N0⊂Kp−∞⊗KVN_0 \subset K^{p^{-\infty}} \otimes_K V. We thus have N0=0N_0 = 0, whence N=0N = 0. Since V is of finite dimension over K, V(Ω)V_{(\Omega)} is of finite dimension over Ω\Omega; the intersection of the kernels of the linear forms gσg_\sigma is zero, hence (II, p. 301, Th. 7) the family (gσ)σ∈G(g_\sigma)_{\sigma \in G} generates the dual of V(Ω)V_{(\Omega)}. Now let uu be a K-linear mapping of V into Ω\Omega; let u~\tilde{u} be the linear form on V(Ω)V_{(\Omega)} which maps x⊗yx \otimes y to xu(y)xu(y) for x∈Ωx \in \Omega and y∈Vy \in V. By what has been said there exist elements σ1,…,σn\sigma_1, \ldots, \sigma_n of G and λ1,…,λn\lambda_1, \ldots, \lambda_n of Ω\Omega such that u~=∑i=1nλigσi\tilde{u} = \sum_{i=1}^n \lambda_i g_{\sigma_i}, whence u(y)=∑i=1nλiσi(y)u(y) = \sum_{i=1}^n \lambda_i \sigma_i(y) for all y∈Vy \in V. So we have shown that a) implies d).

Exercises

See the exercises for § 15.

Transcribed from the printed edition, 2003, Springer. Not translated.