Bourbaki

§ 4. Relations between tensor products and homomorphism modules

1. The isomorphisms Hom⁡B(E⊗AF,G)→Hom⁡A(F,Hom⁡B(E,G))\operatorname{Hom}_B(E \otimes_A F, G) \to \operatorname{Hom}_A(F, \operatorname{Hom}_B(E, G)) and Hom⁡c(E⊗AF,G)→Hom⁡A(E,Hom⁡c(F,G))\operatorname{Hom}_c(E \otimes_A F, G) \to \operatorname{Hom}_A(E, \operatorname{Hom}_c(F, G))

Let E be a right A-module, F a left A-module, G a Z\mathbf{Z}-module and H the Z\mathbf{Z}-module of mappings f:E×F→Gf : E \times F \to G which are Z\mathbf{Z}-bilinear and satisfy

(1)

f(xλ,y)=f(x,λy)for x∈E,y∈F,λ∈A. f(x \lambda, y) = f(x, \lambda y) \quad \text{for } x \in E, y \in F, \lambda \in A.

It has been seen (§ 3, no. 1, Proposition 1) that there exists a canonical Z\mathbf{Z}-module homomorphism

(2)

H→Hom⁡Z(E⊗AF,G). H \to \operatorname{Hom}_{\mathbf{Z}}(E \otimes_A F, G).

On the other hand, a left A-module structure has been defined on Hom⁡Z(E,G)\operatorname{Hom}_{\mathbf{Z}}(E, G) and a right A-module structure on Hom⁡Z(F,G)\operatorname{Hom}_{\mathbf{Z}}(F, G) (§ 3, no. 3); we may therefore consider the Z\mathbf{Z}-modules Hom⁡A(E,Hom⁡Z(F,G))\operatorname{Hom}_A(E, \operatorname{Hom}_{\mathbf{Z}}(F, G)) and Hom⁡A(F,Hom⁡Z(E,G))\operatorname{Hom}_A(F, \operatorname{Hom}_{\mathbf{Z}}(E, G)). A mapping ff of E×FE \times F into G is canonically identified with a mapping of E into the set GFG^F of mappings of F into G (Set Theory, II, § 5, no. 2); by expressing the fact that the latter mapping belongs to Hom⁡A(E,Hom⁡Z(F,G))\operatorname{Hom}_A(E, \operatorname{Hom}_{\mathbf{Z}}(F, G)), we obtain precisely the fact that ff is biadditive and conditions (1); whence there is a canonical isomorphism

(3)

H→Hom⁡A(E,Hom⁡Z(F,G)) H \to \operatorname{Hom}_A(E, \operatorname{Hom}_{\mathbf{Z}}(F, G))
and similarly there is defined a canonical isomorphism

(4)

H→Hom⁡A(F,Hom⁡Z(E,G)). H \to \operatorname{Hom}_A(F, \operatorname{Hom}_{\mathbf{Z}}(E, G)).

Suppose now that E and G also have left (resp. right) B-module structures and that the A-module and B-module structures on E are compatible. Then E⊗AFE \otimes_A F has canonically a left (resp. right) B-module structure (§ 3, no. 4) and on the other hand Hom⁡B(E,G)\operatorname{Hom}_B(E, G) has canonically a left A-module structure (§ 1, no. 14). We may therefore consider the Z\mathbf{Z}-modules Hom⁡B(E⊗AF,G)\operatorname{Hom}_B(E \otimes_A F, G) and Hom⁡A(F,Hom⁡B(E,G))\operatorname{Hom}_A(F, \operatorname{Hom}_B(E, G)), which are submodules of Hom⁡Z(E⊗AF,G)\operatorname{Hom}_{\mathbf{Z}}(E \otimes_A F, G) and Hom⁡A(F,Hom⁡Z(E,G))\operatorname{Hom}_A(F, \operatorname{Hom}_{\mathbf{Z}}(E, G)) respectively (§ 2, no. 1, Theorem 2). We examine under what condition a mapping f∈Hf \in H has as image under the isomorphisms (2) and (4) an element of Hom⁡B(E⊗AF,G)\operatorname{Hom}_B(E \otimes_A F, G) and an element of Hom⁡A(E,Hom⁡B(F,G))\operatorname{Hom}_A(E, \operatorname{Hom}_B(F, G)) respectively; in each of the two cases we find the same condition

f(βx,y)=βf(x,y) f(\beta x, y) = \beta f(x, y)
(resp. f(xβ,y)=f(x,y)βf(x \beta, y) = f(x, y) \beta)

for x∈E,y∈F,β∈Bx \in E, y \in F, \beta \in B.

Similarly, suppose that F and G are left (resp. right) C-modules and that the A-module and C-module structures on F are compatible. Then, for a mapping f∈Hf \in H to have as image under (2) or (3) an element of HomC(E⊗AF,G)\mathrm{Hom}_C(E \otimes_A F, G) or HomA(E,HomC(F,G))\mathrm{Hom}_A(E, \mathrm{Hom}_C(F, G)) respectively, it is necessary and sufficient that it satisfy the same condition

f(x,γy)=γf(x,y) f(x, \gamma y) = \gamma f(x, y)
(resp. f(x,yγ)=f(x,y)γf(x, y \gamma) = f(x, y) \gamma)

for x∈E,y∈F,γ∈Cx \in E, y \in F, \gamma \in C.

We have therefore established the following result (in the notation introduced above):

Proposition 1

(a) *Let E be a (B, A)-bimodule, F a left A-module and G a left B-module. For every mapping g∈HomB(E⊗AF,G)g \in \mathrm{Hom}_B(E \otimes_A F, G), let g′g' be the mapping of F into HomB(E,G)\mathrm{Hom}_B(E, G) defined by (g′(y))(x)=g(x⊗y)(g'(y))(x) = g(x \otimes y) for x∈E,y∈Fx \in E, y \in F. The mapping g↦g′g \mapsto g' is an isomorphism

(5)

β:HomB(E⊗AF,G)→HomA(F,HomB(E,G)). \beta : \mathrm{Hom}_B(E \otimes_A F, G) \to \mathrm{Hom}_A(F, \mathrm{Hom}_B(E, G)).

(b) Let E be a right A-module, F an (A, C)-bimodule and G a right C-module. For every mapping h∈HomC(E⊗AF,G)h \in \mathrm{Hom}_C(E \otimes_A F, G), let h′h' be the mapping of E into HomC(F,G)\mathrm{Hom}_C(F, G) defined by (h′(x))(y)=h(x⊗y)(h'(x))(y) = h(x \otimes y) for x∈E,y∈Fx \in E, y \in F. The mapping h↦h′h \mapsto h' is an isomorphism

(6)

γ:HomC(E⊗AF,G)→HomA(E,HomC(F,G)). \gamma : \mathrm{Hom}_C(E \otimes_A F, G) \to \mathrm{Hom}_A(E, \mathrm{Hom}_C(F, G)).

In particular B and C may be taken to be a subring Γ\Gamma of the centre of the ring A; then for every Γ\Gamma-module G, the three Γ\Gamma-modules

HomΓ(E⊗AF,G),HomA(E,HomΓ(F,G)),HomA(F,HomΓ(E,G)) \mathrm{Hom}_\Gamma(E \otimes_A F, G), \quad \mathrm{Hom}_A(E, \mathrm{Hom}_\Gamma(F, G)), \quad \mathrm{Hom}_A(F, \mathrm{Hom}_\Gamma(E, G))

are canonically isomorphic to the Γ\Gamma-module of Γ\Gamma-*bilinear* mappings of E×FE \times F into G which satisfy (1). More particularly:

Corollary

If C is a commutative ring and E, F, G three C-modules, then the C-modules*

HomC(E⊗CF,G),HomC(E,HomC(F,G)), \mathrm{Hom}_C(E \otimes_C F, G), \qquad \mathrm{Hom}_C(E, \mathrm{Hom}_C(F, G)),
HomC(F,HomC(E,G)),L2(E,F;G) \mathrm{Hom}_C(F, \mathrm{Hom}_C(E, G)), \qquad \mathscr{L}_2(E, F; G)
*are canonically isomorphic.

2. The homomorphism** E∗⊗AF→HomA(E,F)E^* \otimes_A F \to \mathrm{Hom}_A(E, F)

Let A, B be two rings, E a left A-module, F a left B-module and G an (A, B)-*bimodule*. The Z\mathbf{Z}-module HomA(E,G)\mathrm{Hom}_A(E, G) has canonically a right B-*module* structure (§ 1, no. 14) such that (uβ)(x)=u(x)β(u\beta)(x) = u(x)\beta for β∈B\beta \in B, u∈HomA(E,G)u \in \mathrm{Hom}_A(E, G), x∈Ex \in E. On the other hand, G⊗BFG \otimes_B F has canonically a left A-module structure (§ 3, no. 4). We shall define a canonical Z\mathbf{Z}-homomorphism

ν:HomA(E,G)⊗BF→HomA(E,G⊗BF). \nu : \mathrm{Hom}_A(E, G) \otimes_B F \to \mathrm{Hom}_A(E, G \otimes_B F).

To this end, we consider, for all y∈Fy \in F and all u∈HomA(E,G)u \in \mathrm{Hom}_A(E, G), the mapping ν′(u,y):x↦u(x)⊗y\nu'(u, y) : x \mapsto u(x) \otimes y of EE onto G⊗BFG \otimes_B F. It is immediately verified that ν′(u,y)\nu'(u, y) is AA-linear and that ν′\nu' is a Z\mathbf{Z}-bilinear mapping of HomA(E,G)×F\mathrm{Hom}_A(E, G) \times F into HomA(E,G⊗BF)\mathrm{Hom}_A(E, G \otimes_B F); moreover, for all β∈B\beta \in B, ν′(uβ,y)\nu'(u\beta, y) and ν′(u,βy)\nu'(u, \beta y) are equal, for (u(x)β)⊗y=u(x)⊗(βy)(u(x)\beta) \otimes y = u(x) \otimes (\beta y). We conclude (§ 3, no. 1, Proposition 1) the existence of the desired homomorphism ν\nu such that ν(u⊗y)\nu(u \otimes y) is the AA-linear mapping x↦u(x)⊗yx \mapsto u(x) \otimes y.

It is immediately verified that, if EE is an (A,(Ci′);(Dj′))(A, (C'_i); (D'_j))-multimodule, FF a (B,(Ch′′);(Dk′′))(B, (C''_h); (D''_k))-multimodule and GG an (A,(C1′′′);B,(Dm′′′))(A, (C'''_1); B, (D'''_m))-multimodule, the mapping (7) is a ((Dj′),(Ch′′),(C1′′′);(Ci′),(Dk′′),(Dm′′′))((D'_j), (C''_h), (C'''_1); (C'_i), (D''_k), (D'''_m))-multimodule homomorphism.

Proposition 2

(i) When FF is a projective (resp. finitely generated projective) B-module, the canonical homomorphism (7) is injective (resp. bijective).

(ii) When EE is a finitely generated projective AA-module, the canonical homomorphism (7) is bijective.

(i) Fixing EE and GG, for every left BB-module FF, we write

T(F)=HomA(E,G)⊗BF,T′(F)=HomA(E,G⊗BF); T(F) = \mathrm{Hom}_A(E, G) \otimes_B F, \quad T'(F) = \mathrm{Hom}_A(E, G \otimes_B F);

for every left BB-module homomorphism u:F→F′u : F \to F', we write T(u)=1⊗uT(u) = 1 \otimes u (1 here denoting the identity mapping of HomA(E,G)\mathrm{Hom}_A(E, G)),

T′(u)=Hom(1E,1G⊗u); T'(u) = \mathrm{Hom}(1_E, 1_G \otimes u);

on the other hand we write νF\nu_F instead of ν\nu. Then we have the following lemmas:

Lemma 1

For every homomorphism u:F→F′u : F \to F', the diagram

T(F)→νFT′(F)↓T(u)↓T′(u)T(F′)→νF′T′(F′) \begin{array}{ccc} T(F) & \xrightarrow{\nu_F} & T'(F) \\ \downarrow T(u) & & \downarrow T'(u) \\ T(F') & \xrightarrow{\nu_{F'}} & T'(F') \end{array}

is commutative.

The verification is immediate.

Lemma 2

Let M,NM, N be two supplementary submodules in FF and i:M→F,j:N→Fi : M \to F, j : N \to F the canonical injections. The diagram

T(M)⊕T(N)→νM⊕νNT′(M)⊕T′(N)↓T(i)+T(j)↓T′(i)+T′(j)T(F)→νFT′(F) \begin{array}{ccc} T(M) \oplus T(N) & \xrightarrow{\nu_M \oplus \nu_N} & T'(M) \oplus T'(N) \\ \downarrow T(i) + T(j) & & \downarrow T'(i) + T'(j) \\ T(F) & \xrightarrow{\nu_F} & T'(F) \end{array}

is commutative and the vertical arrows are bijective.

The commutativity follows from Lemma 1, the other assertions from § 1, no. 6, Corollary 2 to Proposition 6 and § 3, no. 7, Proposition 7.

Lemma 3

Under the hypotheses of Lemma 2, for νF\nu_F to be injective (resp. surjective), it is necessary and sufficient that νM\nu_M and νN\nu_N be so.

This follows from Lemma 2 and § 1, no. 6, Corollary 1 to Proposition 6.

Then Lemma 3, together with § 2, no. 2, Proposition 4, shows that it suffices to consider the case where FF is a free module. But, if (bμ)(b_\mu) is a basis of FF, every element of HomA(E,G)⊗BF\mathrm{Hom}_A(E, G) \otimes_B F may then be written uniquely as ∑μuμ⊗bμ\sum_\mu u_\mu \otimes b_\mu, where uμ∈HomA(E,G)u_\mu \in \mathrm{Hom}_A(E, G) (§ 3, no. 7, Corollary 1 to Proposition 7); the image of this element under ν\nu is the AA-linear mapping x↦∑μuμ(x)⊗bμx \mapsto \sum_\mu u_\mu(x) \otimes b_\mu; it cannot be zero for all x∈Ex \in E unless uμ(x)=0u_\mu(x) = 0 for all x∈Ex \in E and all μ\mu, which is equivalent to saying that uμ=0u_\mu = 0 for all μ\mu; hence ν\nu is injective. When also FF admits a finite basis, Lemma 3 shows (by induction on the number of elements in the basis of FF) that to prove that ν\nu is surjective, it suffices to do so when F=BsF = B_s; but in this case the two sides of (7) are canonically identified with HomA(E,G)\mathrm{Hom}_A(E, G) (§ 3, no. 4, Proposition 4) and ν\nu becomes the identity.

(ii) To show the proposition when EE is projective and finitely generated, this time we fix FF and GG and write, for every left AA-module EE,

T(E)=HomA(E,G)⊗BF,T′(E)=HomA(E,G⊗BF) T(E) = \mathrm{Hom}_A(E, G) \otimes_B F, \qquad T'(E) = \mathrm{Hom}_A(E, G \otimes_B F)

and, for every left AA-module homomorphism v:E→E′v : E \to E',

T(v)=Hom(v,1G)⊗1F,T′(v)=Hom(v,1G⊗1F); T(v) = \mathrm{Hom}(v, 1_G) \otimes 1_F, \qquad T'(v) = \mathrm{Hom}(v, 1_G \otimes 1_F);

on the other hand, we write νE\nu_E instead of ν\nu. Then we have the two lemmas:

Lemma 4

For every homomorphism v:E→E′v : E \to E', the diagram

T(E′)→νE′T′(E′)↓T(v)↓T′(v)T(E)→νET′(E) \begin{array}{ccc} T(E') & \xrightarrow{\nu_{E'}} & T'(E') \\ \downarrow T(v) & & \downarrow T'(v) \\ T(E) & \xrightarrow{\nu_E} & T'(E) \end{array}

is commutative.

Lemma 5

Let MM and NN be two supplementary submodules in EE and p:E→M,q:E→Np : E \to M, q : E \to N the canonical projections. The diagram

T(M)⊕T(N)→νM⊕νNT′(M)⊕T′(N)↓T(p)+T(q)↓T′(p)+T′(q)T(E)→νET′(E) \begin{array}{ccc} T(M) \oplus T(N) & \xrightarrow{\nu_M \oplus \nu_N} & T'(M) \oplus T'(N) \\ \downarrow T(p) + T(q) & & \downarrow T'(p) + T'(q) \\ T(E) & \xrightarrow{\nu_E} & T'(E) \end{array}

is commutative and the vertical arrows are bijective.

They are proved as Lemmas 1 and 2, taking account of § 1, no. 6, Corollary 2 to Proposition 6, § 2, no. 1, Proposition 1 and § 3, no. 7, Proposition 7.

The remainder of the proof then proceeds as in (i) and is reduced to the case where E=AsE = A_s; the two sides of (7) are then canonically identified with G⊗BFG \otimes_B F and ν\nu becomes the identity.

In particular take B=AB = A and GG the (A,A)(A, A)-bimodule sAd{}_sA_d (§3§ 3, no. 4), so that the right AA-module HomA(E,sAd)\mathrm{Hom}_A(E, {}_sA_d) is just the dual E∗E^* of EE and (sAd)⊗AF({}_sA_d) \otimes_A F is canonically identified with FF (§3§ 3, no. 4, Proposition 4). Homomorphism (7) then becomes a canonical Z\mathbf{Z}-homomorphism

(11)(11)
θ:E∗⊗AF→HomA(E,F) \theta : E^* \otimes_A F \to \mathrm{Hom}_A(E, F)
and θ(x∗⊗y)\theta(x^* \otimes y) is the linear mapping of EE into FF
x↦⟨x,x∗⟩y. x \mapsto \langle x, x^* \rangle y.

Remark (1). The characterization of projective AA-modules given in §2§ 2, no. 6, Proposition 12, can also be expressed as follows: for a left AA-module EE to be projective, it is necessary and sufficient that the canonical homomorphism

θE:E∗⊗AE→HomA(E,E)=EndA(E) \theta_E : E^* \otimes_A E \to \mathrm{Hom}_A(E, E) = \mathrm{End}_A(E)
be such that 1E1_E belongs to the image of θE\theta_E.

Corollary

(i) When FF is a projective (resp. finitely generated projective) module, the canonical homomorphism (11) is injective (resp. bijective). (ii) When EE is a finitely generated projective module, the canonical homomorphism (11) is bijective.

Even when EE and FF are both finitely generated, θ\theta is not necessarily surjective, as is shown by the example A=Z,E=F=Z/2ZA = \mathbf{Z}, E = F = \mathbf{Z}/2\mathbf{Z}; the right hand side of (11) is non-zero but E∗=0E^* = 0. On the other hand, examples can be given where EE is free, but (11) is neither injective nor surjective (Exercise 3(b)).

When E admits a finite basis (ei)(e_i), the inverse isomorphism θ−1\theta^{-1} of θ\theta can be found explicitly as follows. Let (ei∗)(e_i^*) be the dual basis of (ei)(e_i) (§ 2, no. 6); for all u∈Hom(E,F)u \in \mathrm{Hom}(E, F) and all x=∑iξieix = \sum_i \xi_i e_i with ξi∈A\xi_i \in A,

u(x)=∑iξiu(ei)=∑i⟨x,ei∗⟩u(ei) u(x) = \sum_i \xi_i u(e_i) = \sum_i \langle x, e_i^* \rangle u(e_i)

and therefore u=∑iθ(ei∗⊗u(ei))u = \sum_i \theta(e_i^* \otimes u(e_i)), in other words

θ−1(u)=∑iei∗⊗u(ei). \theta^{-1}(u) = \sum_i e_i^* \otimes u(e_i).

In particular, if further F=EF = E, it is seen that the image under θE−1\theta_E^{-1} of the identity mapping 1E1_E is the element ∑iei∗⊗ei\sum_i e_i^* \otimes e_i, which is therefore independent of the basis (ei)(e_i) considered in E.

Note on the other hand that when E is a finitely generated projective module the ring structure on EndA(E)\mathrm{End}_A(E) can be transported by θE−1\theta_E^{-1} to E∗⊗AEE^* \otimes_A E; it is immediately verified that, for x,yx, y in E, x∗,y∗x^*, y^* in E∗E^*, in the ring EndA(E)\mathrm{End}_A(E),

θE(x∗⊗x)∘θE(y∗⊗y)=θE((y∗⟨y,x∗⟩)⊗x). \theta_E(x^* \otimes x) \circ \theta_E(y^* \otimes y) = \theta_E((y^* \langle y, x^* \rangle) \otimes x).

Remark (2). Let E be a right A-module; replacing E by E∗E^* in (11), we obtain a canonical Z\mathbf{Z}-homomorphism

E∗∗⊗AF→HomA(E∗,F). E^{**} \otimes_A F \to \mathrm{Hom}_A(E^*, F).

On the other hand, there is a canonical A-homomorphism cE:E→E∗∗c_E : E \to E^{**}, whence there is a Z\mathbf{Z}-homomorphism cE⊗1F:E⊗AF→E∗∗⊗AFc_E \otimes 1_F : E \otimes_A F \to E^{**} \otimes_A F; composing the latter with the homomorphism (14), we hence obtain a canonical Z\mathbf{Z}-homomorphism

θ′:E⊗AF→HomA(E∗,F) \theta' : E \otimes_A F \to \mathrm{Hom}_A(E^*, F)

such that θ′(x⊗y)\theta'(x \otimes y) is the linear mapping

x∗↦⟨x,x∗⟩y. x^* \mapsto \langle x, x^* \rangle y.

If E and F are projective modules, the mapping (15) is injective. For cEc_E is then injective (§ 2, no. 7, Corollary 4 to Proposition 13) and as F is projective, the Z\mathbf{Z}-homomorphism cE⊗1F:E⊗AF→E∗∗⊗AFc_E \otimes 1_F : E \otimes_A F \to E^{**} \otimes_A F is also injective (§ 3, no. 7, Corollary 6 to Proposition 7); finally, it has been seen (Proposition 2) that the homomorphism (14) is injective, whence the conclusion.

If E is projective and finitely generated, the mapping (15) is bijective for the two mappings of which it is composed are then bijective (§ 2, no. 7, Corollary 4 to Proposition 13 and Proposition 2 above).

3. Trace of an endomorphism

Let CC be a commutative ring and EE a CC-module. The mapping (x∗,x)↦⟨x,x∗⟩(x^*, x) \mapsto \langle x, x^* \rangle of E∗×EE^* \times E into CC is then CC*-bilinear*, since, for all γ∈C\gamma \in C, ⟨γx,x∗⟩=γ⟨x,x∗⟩\langle \gamma x, x^* \rangle = \gamma \langle x, x^* \rangle and ⟨x,x∗γ⟩=⟨x,x∗⟩γ\langle x, x^* \gamma \rangle = \langle x, x^* \rangle \gamma; we derive a canonical CC*-linear* mapping

τ:E∗⊗CE→C \tau : E^* \otimes_C E \to C

such that τ(x∗⊗x)=⟨x,x∗⟩\tau(x^* \otimes x) = \langle x, x^* \rangle (§ 3, no. 5). Suppose now also that EE is a finitely generated projective CC-module; the canonical isomorphism (11) of no. 2 is then a CC*-module* isomorphism and we can therefore define by transporting the structure a canonical linear form Tr=τ∘θE−1\mathrm{Tr} = \tau \circ \theta_E^{-1} on the CC-module EndC(E)\mathrm{End}_C(E). For all u∈EndC(E)u \in \mathrm{End}_C(E) the scalar Tr(u)\mathrm{Tr}(u) is called the trace of the endomorphism uu; every u∈EndC(E)u \in \mathrm{End}_C(E) can be written (in general in an infinity of ways) in the form x↦∑i⟨x,xi∗⟩yix \mapsto \sum_i \langle x, x_i^* \rangle y_i where xi∗∈E∗x_i^* \in E^* and yi∈Eiy_i \in E_i by virtue of no. 2, Corollary to Proposition 2; then

Tr(u)=∑i⟨yi,xi∗⟩(cf. § 10, no. 11). \mathrm{Tr}(u) = \sum_i \langle y_i, x_i^* \rangle \quad \text{(cf. § 10, no. 11).}

By definition,

Tr(u+v)=Tr(u)+Tr(v) \mathrm{Tr}(u + v) = \mathrm{Tr}(u) + \mathrm{Tr}(v)
Tr(γu)=γTr(u) \mathrm{Tr}(\gamma u) = \gamma \mathrm{Tr}(u)

for u,vu, v in EndC(E)\mathrm{End}_C(E) and γ∈C\gamma \in C. Moreover:

Proposition 3

Let CC be a commutative ring, E,FE, F two finitely generated projective CC*-modules* and u:E→Fu : E \to F and v:F→Ev : F \to E two linear mappings; then

Tr(v∘u)=Tr(u∘v). \mathrm{Tr}(v \circ u) = \mathrm{Tr}(u \circ v).

The two mappings (u,v)↦Tr(u∘v),(u,v)↦Tr(v∘u)(u, v) \mapsto \mathrm{Tr}(u \circ v), (u, v) \mapsto \mathrm{Tr}(v \circ u) of

HomC(E,F)×HomC(F,E) \mathrm{Hom}_C(E, F) \times \mathrm{Hom}_C(F, E)

into CC are CC*-bilinear*; it therefore suffices to verify (20) when uu is of the form x↦⟨x,a∗⟩bx \mapsto \langle x, a^* \rangle b and vv of the form y↦⟨y,b∗⟩ay \mapsto \langle y, b^* \rangle a, with a∈E,a∗∈E∗,b∈F,b∗∈F∗a \in E, a^* \in E^*, b \in F, b^* \in F^*. But then v∘uv \circ u is the mapping x↦⟨x,a∗⟩⟨b,b∗⟩ax \mapsto \langle x, a^* \rangle \langle b, b^* \rangle a and u∘vu \circ v the mapping y↦⟨y,b∗⟩⟨a,a∗⟩by \mapsto \langle y, b^* \rangle \langle a, a^* \rangle b. Formula (17) shows that the values of the two sides of (20) are equal to ⟨a,a∗⟩⟨b,b∗⟩\langle a, a^* \rangle \langle b, b^* \rangle.

Corollary

If u1,…,upu_1, \ldots, u_p are endomorphisms of EE, then

Tr(u1∘u2∘⋯∘up)=Tr(ui∘ui+1∘⋯∘up∘u1∘⋯∘ui−1) \mathrm{Tr}(u_1 \circ u_2 \circ \cdots \circ u_p) = \mathrm{Tr}(u_i \circ u_{i+1} \circ \cdots \circ u_p \circ u_1 \circ \cdots \circ u_{i-1})

for 1≤i≤p1 \leq i \leq p ("invariance of the trace under cyclic permutation").

It suffices to apply (20) to the product

(u1∘u2∘⋯∘ui−1)∘(ui∘ui+1∘⋯∘up).(u_1 \circ u_2 \circ \cdots \circ u_{i-1}) \circ (u_i \circ u_{i+1} \circ \cdots \circ u_p).

Note that on the other hand it is not necessarily true that

Tr⁡(u∘v∘w)=Tr⁡(u∘w∘v)\operatorname{Tr}(u \circ v \circ w) = \operatorname{Tr}(u \circ w \circ v)

for three endomorphisms u,v,wu, v, w of E.

4. The homomorphism Hom⁡c(E1,F1)⊗cHom⁡c(E2,F2)→Hom⁡c(E1⊗cE2,F1⊗cF2)\operatorname{Hom}_c(E_1, F_1) \otimes_c \operatorname{Hom}_c(E_2, F_2) \to \operatorname{Hom}_c(E_1 \otimes_c E_2, F_1 \otimes_c F_2)

Let C be a commutative ring and E1,E2,F1,F2E_1, E_2, F_1, F_2 four C-modules; in § 3, no. 5, formula (13) we defined a canonical C-module homomorphism

(21) λ:Hom⁡(E1,F1)⊗Hom⁡(E2,F2)→Hom⁡(E1⊗E2,F1⊗F2)\lambda : \operatorname{Hom}(E_1, F_1) \otimes \operatorname{Hom}(E_2, F_2) \to \operatorname{Hom}(E_1 \otimes E_2, F_1 \otimes F_2).

Proposition 4

When one of the ordered pairs* (E1,E2),(E1,F1),(E2,F2)(E_1, E_2), (E_1, F_1), (E_2, F_2) consists of finitely generated projective C-modules, the canonical homomorphism (21) *is bijective.

It is obviously sufficient to perform the proof for the ordered pairs (E1,F1)(E_1, F_1) and (E1,E2)(E_1, E_2).

We consider first the case of the ordered pair (E1,F1)(E_1, F_1); we fix E2,F1,F2E_2, F_1, F_2 and write for every C-module T(E)=Hom⁡(E,F1)⊗cHom⁡(E2,F2)T(E) = \operatorname{Hom}(E, F_1) \otimes_c \operatorname{Hom}(E_2, F_2) and T′(E)=Hom⁡(E⊗E2,F1⊗F2)T'(E) = \operatorname{Hom}(E \otimes E_2, F_1 \otimes F_2) and, for every C-homomorphism v:E→E′v : E \to E',

T(v)=Hom⁡(v,1F1)⊗1Hom⁡(E2,F2)T(v) = \operatorname{Hom}(v, 1_{F_1}) \otimes 1_{\operatorname{Hom}(E_2, F_2)}

and

T′(v)=Hom⁡(v⊗1E2,1F1⊗F2).T'(v) = \operatorname{Hom}(v \otimes 1_{E_2}, 1_{F_1 \otimes F_2}).

Then Lemmas 4 and 5 (no. 2) (where v is replaced by λ\lambda) are valid and are proved by completely analogous methods.

We next fix E2E_2 and F2F_2 and this time write, for every C-module F,

T(F)=Hom⁡(C,F)⊗cHom⁡(E2,F2)T(F) = \operatorname{Hom}(C, F) \otimes_c \operatorname{Hom}(E_2, F_2)
and T′(F)=Hom⁡(C⊗E2,F⊗F2)T'(F) = \operatorname{Hom}(C \otimes E_2, F \otimes F_2) and, for every C-homomorphism u:F→F′u : F \to F',

T(u)=Hom⁡(1C,u)⊗1Hom⁡(E2,F2)T(u) = \operatorname{Hom}(1_C, u) \otimes 1_{\operatorname{Hom}(E_2, F_2)}

and

T′(u)=Hom⁡(1C⊗1E2,u⊗1F2).T'(u) = \operatorname{Hom}(1_C \otimes 1_{E_2}, u \otimes 1_{F_2}).

This time it is immediately verified that Lemmas 1 and 2 (no. 2) (where λ\lambda always replaces v) are valid.

This being so, we show the proposition first when E1=CE_1 = C and F1F_1 is projective and finitely generated. The argument of no. 2 (which rests on Lemmas 1 and 2), together with the above remarks, reduces this to proving the proposition when also F1=CF_1 = C; then Hom⁡(E1,F1),E1⊗E2\operatorname{Hom}(E_1, F_1), E_1 \otimes E_2 and F1⊗F2F_1 \otimes F_2 are identified with C, E2E_2 and F2F_2 respectively (§ 3, no. 4, Proposition 4); the two sides of (21)

THE HOMOMORPHISM Hom⁡c(E1,F1)⊗cHom⁡(E2,F2)\operatorname{Hom}_c(E_1, F_1) \otimes_c \operatorname{Hom}(E_2, F_2)

are then both canonically identified with Hom⁡(E2,F2)\operatorname{Hom}(E_2, F_2) and, after these identifications, it is verified that λ\lambda becomes the identity.

We now suppose that F1F_1 is projective and finitely generated; the argument of no. 2 (depending this time on Lemmas 4 and 5) reduces the proof for E1E_1 any finitely generated projective module to the case where E1=CE_1 = C, that is to the first case dealt with.

For the ordered pair (E1,E2)(E_1, E_2), the procedure is similar, this time applying Lemmas 4 and 5 twice; we leave the details to the reader.

Note that when E1=C(I)E_1 = C^{(I)}, E2=C(J)E_2 = C^{(J)} are free (finitely generated or not), then Hom⁡(E1,F1)=F1I\operatorname{Hom}(E_1, F_1) = F_1^I, Hom⁡(E2,F2=F2J\operatorname{Hom}(E_2, F_2 = F_2^J and

Hom⁡(E1⊗E2,F1⊗F2)=(F1⊗F2)I×J \operatorname{Hom}(E_1 \otimes E_2, F_1 \otimes F_2) = (F_1 \otimes F_2)^{I \times J}
to within canonical isomorphisms and (21) is then identical with a special case of the canonical homomorphism (22) of § 3, no. 7.

When E2=CE_2 = C, the canonical homomorphism (21) gives, after identifying Hom⁡(E2,F2)\operatorname{Hom}(E_2, F_2) with F2F_2 and E1⊗E2E_1 \otimes E_2 with E1E_1, a canonical homomorphism

Hom⁡(E,F)⊗G→Hom⁡(E,F⊗G) \operatorname{Hom}(E, F) \otimes G \to \operatorname{Hom}(E, F \otimes G)
for any three CC-modules E,F,GE, F, G which is just the homomorphism (7) of no. 2 for A=B=CA = B = C.

Note that when F=CF = C the canonical homomorphism (22) again gives (11) (no. 2) for the case of a commutative ring.

Suppose now that F1=F2=CF_1 = F_2 = C; as F1⊗F2F_1 \otimes F_2 is identified with CC, there is this time a canonical homomorphism

μ:E∗⊗F∗→(E⊗F)∗ \mu : E^* \otimes F^* \to (E \otimes F)^*
for two CC-modules E,FE, F; for x∗∈E∗,y∗∈F∗x^* \in E^*, y^* \in F^*, the image of x∗⊗y∗x^* \otimes y^* under the canonical homomorphism (23) is the linear form uu on E⊗FE \otimes F such that
u(x⊗y)=⟨x,x∗⟩⟨y,y∗⟩. u(x \otimes y) = \langle x, x^* \rangle \langle y, y^* \rangle.
Moreover, if E1,E2,F1,F2E_1, E_2, F_1, F_2 are four CC-modules, f:E1→E2,g:F1→F2f : E_1 \to E_2, g : F_1 \to F_2 two linear mappings, it follows immediately from (24) that the diagram
E2∗⊗F2∗→μ(E2⊗F2)∗tf⊗tg↓↓t(f⊗g)E1∗⊗F1∗→μ(E1⊗F1)∗ \begin{array}{ccc} E_2^* \otimes F_2^* & \xrightarrow{\mu} & (E_2 \otimes F_2)^* \\ t_f \otimes t_g \downarrow & & \downarrow t_{(f \otimes g)} \\ E_1^* \otimes F_1^* & \xrightarrow{\mu} & (E_1 \otimes F_1)^* \end{array}
is commutative.

Corollary 1

If one of the modules E, F is projective and finitely generated, the canonical homomorphism (23) is bijective.

Corollary 2

Let E₁, E₂ be two finitely generated projective C-modules, u₁ an endomorphism of E₁ and u₂ an endomorphism of E₂; then

Tr(u1⊗u2)=Tr(u1)Tr(u2). \text{Tr}(u_1 \otimes u_2) = \text{Tr}(u_1)\text{Tr}(u_2).

By linearity, it suffices to consider the case where u₁ is of the form x1↦⟨x1,x1∗⟩y1x_1 \mapsto \langle x_1, x_1^* \rangle y_1 and u₂ of the form x2↦⟨x2,x2∗⟩y2x_2 \mapsto \langle x_2, x_2^* \rangle y_2; then the image of x1⊗x2x_1 \otimes x_2 under u1⊗u2u_1 \otimes u_2 is by definition

⟨x1,x1∗⟩⟨x2,x2∗⟩(y1⊗y2)=⟨x1⊗x2,x1∗⊗x2∗⟩(y1⊗y2) \langle x_1, x_1^* \rangle \langle x_2, x_2^* \rangle (y_1 \otimes y_2) = \langle x_1 \otimes x_2, x_1^* \otimes x_2^* \rangle (y_1 \otimes y_2)

x1∗⊗x2∗x_1^* \otimes x_2^* being canonically identified under μ\mu with an element of (E1⊗E2)∗(E_1 \otimes E_2)^*. As ⟨y1⊗y2,x1∗⊗x2∗⟩=⟨y1,x1∗⟩⟨y2,x2∗⟩\langle y_1 \otimes y_2, x_1^* \otimes x_2^* \rangle = \langle y_1, x_1^* \rangle \langle y_2, x_2^* \rangle, formula (26) follows in this case from (17).

Remark

If E, F, G are any three C-modules, it is immediately verified that the diagram

E∗⊗F∗⊗G∗→μ⊗1(E⊗F)∗⊗G∗1⊗μ↓↓μE∗⊗(F⊗G)∗→μ(E⊗F⊗G)∗ \begin{array}{ccc} E^* \otimes F^* \otimes G^* & \xrightarrow{\mu \otimes 1} & (E \otimes F)^* \otimes G^* \\ 1 \otimes \mu \downarrow & & \downarrow \mu \\ E^* \otimes (F \otimes G)^* & \xrightarrow{\mu} & (E \otimes F \otimes G)^* \end{array}

is commutative, by virtue of formula (24).

We note also that, without any hypothesis on the C-modules E, F, there are canonical *isomorphisms

(28)

(E⊗F)∗→Hom(E,F∗)(E \otimes F)^* \to \text{Hom}(E, F^*)
(29)
(E⊗F)∗→Hom(F,E∗)(E \otimes F)^* \to \text{Hom}(F, E^*)

which are just the isomorphism (6) and (5) of no. 1 for G = C, A = B = C.

Thus a canonical one-to-one correspondence has been defined between the bilinear forms on E × F, the homomorphisms of E into F and the homomorphisms of F into E: if u (resp. v) is a homomorphism of E into F* (resp. of F into E*), the corresponding bilinear form is given by

(x,y)↦⟨y,u(x)⟩(resp. (x,y)↦⟨x,v(y)⟩).(x, y) \mapsto \langle y, u(x) \rangle \quad (\text{resp. } (x, y) \mapsto \langle x, v(y) \rangle).

Exercises

See the exercises for § 4.

Transcribed from the printed edition, 1998, Springer. Not translated.