Bourbaki

§ 7. Vector bundles

Throughout this paragraph, the letter B denotes a manifold of class CrC^r (r⩾1r \geqslant 1) and the letter M denotes a set endowed with a mapping π\pi from M into B. We say that B is the base of M and for every b∈Bb \in B, we denote by MbM_b and call the fiber of M at bb the subset π−1(b)\pi^{-1}(b) of M.

7.1. Definition of vector bundles

7.1.1. A vector chart of M is a triple t=(U,φ,F)t = (U, \varphi, F), where U is an open set of B, where F is a Banach space and φ\varphi a bijection of π−1(U)\pi^{-1}(U) onto U×FU \times F such that π(φ−1(b,h))=b\pi(\varphi^{-1}(b, h)) = b for all b∈Bb \in B and all h∈Fh \in F. We say that U is the domain of the vector chart t and that t is a vector chart of M at b∈Bb \in B if b∈Ub \in U. For every b∈Ub \in U, we denote by tbt_b the bijection of F onto MbM_b defined by tb(h)=φ−1(b,h)t_b(h) = \varphi^{-1}(b, h) for h∈Fh \in F.

7.1.2. We say that two vector charts t=(U,φ,F)t = (U, \varphi, F) and t′=(U′,φ′,F′)t' = (U', \varphi', F') of M are CrC^r-compatible (or simply compatible) if there exists a mapping λ\lambda of class CrC^r of the manifold U∩U′U \cap U' into the Banach space L(F;F′)\mathscr{L}(F; F') such that:

tb=tb′∘λ(b)for all b∈U∩U′. t_b = t'_b \circ \lambda(b) \quad \text{for all } b \in U \cap U'.

7.1.3. We say that a set of vector charts of M is a CrC^r-vector atlas (or simply vector atlas) of M if it consists of vector charts which are pairwise CrC^r-compatible and whose domains have B as their union. We say that two vector atlases A\mathscr{A} and B\mathscr{B} of M are CrC^r-equivalent (or equivalent) if A∪B\mathscr{A} \cup \mathscr{B} is again a vector atlas of M. This relation is an equivalence relation.

7.1.4. A structure of vector bundle of class CrC^r (with base B) on M is the data of an equivalence class of vector atlases (Ens., Chap. II, § 6, No. 9). A vector chart belonging to a vector atlas of this class is called a vector chart of the vector bundle M.

Let M be a vector bundle with base B. For every b∈Bb \in B, there exists on the fiber MbM_b one and only one Banach space structure such that, for every vector chart t=(U,φ,F)t = (U, \varphi, F) of the vector bundle M at bb, the mapping tbt_b is an isomorphism of F onto MbM_b.

Let c=(U,ψ,E)c = (U, \psi, E) be a chart of the manifold B and let t=(U,φ,F)t = (U, \varphi, F) be a vector chart of the vector bundle M, with the same domain U. For x∈π−1(U)x \in \pi^{-1}(U), put:

α(x)=(ψ(π(x)),tπ(x)−1(x)). \alpha(x) = (\psi(\pi(x)), t_{\pi(x)}^{-1}(x)).

Then the triple (π−1(U),α,E×F)(\pi^{-1}(U), \alpha, E \times F) is a chart of the set MM. There exists on MM one and only one manifold structure of class C′C' (said to be underlying MM) for which all the charts thus obtained are charts of the manifold MM. The triple (M,B,π)(M, B, \pi) is then a fibration (6.1.1).

7.1.5. Let FF be a Banach space. Put M=B×FM = B \times F, the mapping being the first projection. There exists on MM a vector bundle structure (with basis BB) and only one for which (B,IdM,F)(B, \mathrm{Id}_M, F) is a vector chart. One says that B×FB \times F endowed with this structure is the trivial vector bundle with basis BB and fibre FF and one sometimes denotes it by FBF_B. The manifold structure of FBF_B is the product manifold structure and for every b∈Bb \in B, the mapping h↦(b,h)h \mapsto (b, h) is an isomorphism of Banach spaces of FF onto the fibre of FBF_B at the point b∈Bb \in B. One often denotes by 00 a trivial vector bundle whose fibre is reduced to 00.

7.1.6. Let MM be a vector bundle with basis BB. For b∈Bb \in B, one calls rank of MM at bb and denotes by rgb(M)\mathrm{rg}_b(M), the dimension (finite or +∞+\infty) of the Banach space MbM_b. One has dim⁡xM=dim⁡bB+rgbM\dim_x M = \dim_b B + \mathrm{rg}_b M for b=π(x)b = \pi(x). The function b↦rgbMb \mapsto \mathrm{rg}_b M is locally constant. One says that MM is of finite rank if rgbM<+∞\mathrm{rg}_b M < +\infty for every b∈Bb \in B.

7.2. Morphisms of vector bundles

7.2.1. Let BB and B′B' be two manifolds and let ff be a morphism from BB into B′B'. Let MM be a vector bundle with basis BB and M′M' a vector bundle with basis B′B'. One says that a mapping gg from MM into M′M' is an ff-morphism of vector bundles if the following condition is satisfied:

For every point b0∈Bb_0 \in B, there exists a vector chart t=(U,φ,F)t = (U, \varphi, F) of MM at b0b_0, a vector chart t′=(U′,φ′,F′)t' = (U', \varphi', F') of M′M' at f(b0)f(b_0) and a mapping λ\lambda of class C′C' from UU into L(F;F′)\mathscr{L}(F; F') such that f(U)⊂U′f(U) \subset U' and that gb∘tb=tf(b)′∘λ(b)g_b \circ t_b = t'_{f(b)} \circ \lambda(b) for every b∈Ub \in U, where gbg_b is the restriction of gg to MbM_b.

Under these hypotheses, gg is a morphism of varieties and, for every b∈Bb \in B, gg induces a continuous linear mapping of MbM_b into Mf(b)′M'_{f(b)}. The vector rank of gg at b∈Bb \in B is called and is denoted by rgb(g)\mathrm{rg}_b(g) the rank (finite or +∞+\infty) of the linear mapping gbg_b.

Conversely, if r⩾∞r \geqslant \infty or if MM is of finite rank, an ff-morphism of fibrations of (M,B,π)(M, B, \pi) into (M′,B′,π′)(M', B', \pi') which induces on each fibre MbM_b a linear mapping of MbM_b into Mf(b)′M'_{f(b)} (for every b∈Bb \in B), is an ff-morphism of vector bundles.

7.2.2. Let in addition f′f' be a morphism of B′B' into a variety B′′B''. If gg is an ff-morphism of MM into M′M', and if g′g' is an f′f'-morphism of M′M' into a vector bundle M′′M'' with base B′′B'', the mapping g′∘gg' \circ g is an (f′∘f)(f' \circ f)-morphism of MM into M′′M''. We have (g′∘g)b=gf(b)′∘gb(g' \circ g)_b = g'_{f(b)} \circ g_b for every bb in BB.

7.2.3. Let M and M' be two vector bundles with the same base B. A B-morphism, or simply a morphism of M into M', is called every Id_B-morphism. The composite of two morphisms is a morphism.

Let g be a morphism of vector bundles of M into M'; if g is bijective, it is an isomorphism of the variety M onto the variety M', the reciprocal mapping g−1g^{-1} is a morphism of vector bundles of M' into M and we have (g−1)b=gb−1(g^{-1})_b = g_b^{-1} for every b in B. The mapping g is then an isomorphism of vector bundles.

7.2.4. Let f be a morphism of a variety B' into B and let M be a vector bundle with base B. Put M′=B′×BMM' = B' \times_B M and denote by π′\pi' (resp. g) the restriction to M' of the projection of B′×MB' \times M onto B' (resp. M). There exists on M' one and only one structure of vector bundle with base B' (relative to π′\pi') for which g is a f-morphism. We say that M' is the vector bundle with base B' inverse image of M by f and denote it by f∗Mf^*M; the f-morphism g is called the canonical f-morphism of f∗Mf^*M into M.

The structure of variety of f∗Mf^*M is that of the fiber product of the varieties B' and M over B (5.11.2); for every b∈B′b \in B', the mapping x↦(b,x)x \mapsto (b, x) is an isomorphism of Banach spaces of Mf(b)M_{f(b)} onto (f∗M)b(f^*M)_b.

The formation of inverse images of vector bundles is transitive.

Let N' be a vector bundle with base B' and let h be a f-morphism of N' into M. There exists one and only one B'-morphism h~\tilde{h} of N' into f∗Mf^*M such that h=g∘h~h = g \circ \tilde{h}.

Let N be a vector bundle with base B and v a B-morphism of N into M. There exists one and only one B'-morphism, denoted by f∗vf^*v, of f∗Nf^*N into f∗Mf^*M such that the diagram

f∗N→f∗vf∗M↓↓N→vM \begin{array}{ccc} f^*N & \xrightarrow{f^*v} & f^*M \\ \downarrow & & \downarrow \\ N & \xrightarrow{v} & M \end{array}

is commutative.

7.2.5. Let B' be a subvariety of B and let i be the canonical injection of B' into B. If M is a vector bundle with base B, the inverse image i∗Mi^*M is called the vector bundle induced on B' by M and is denoted by M∣B′M|B'. If t=(U,φ,F)t = (U, \varphi, F) is a vector chart of M, then (U∩B′,φ∣π−1(U∩B′),F)(U \cap B', \varphi|\pi^{-1}(U \cap B'), F) is a vector chart of M∣B′M|B'. The canonical f-morphism of M∣B′M|B' into M is an isomorphism of varieties of M∣B′M|B' onto the subvariety π−1(B′)\pi^{-1}(B') of M.

7.2.6. Let f be a morphism of a variety B' into B and let M (resp. M') be a vector bundle with base B (resp. B'). A f-comorphism of M into M' is called a B'-morphism of f∗Mf^*M into M'. When B = B' and f=IdBf = \mathrm{Id}_B, the data of a ff-comorphism of MM into M′M' is equivalent to the data of a BB-morphism of MM into M′M'.

Let gg be a ff-comorphism of MM into M′M'. For every b∈B′b \in B', the mapping gb:x↦g(b,x)g_b : x \mapsto g(b, x) of Mf(b)M_{f(b)} into Mb′M'_b is a continuous linear mapping.

Suppose moreover that f′f' is a morphism of a variety B′′B'' into B′B' and that hh is a f′f'-comorphism of M′M' into a vector bundle M′′M'' with base B′′B''. The mapping h∘f∗gh \circ f^*g of f′∗(f∗M)=(f∘f′)∗M{f'}^*(f^*M) = (f \circ f')^*M into M′′M'' is then a (f∘f′)(f \circ f')-comorphism of MM into M′′M'', denoted by h∘gh \circ g. For b∈B′′b \in B'', one has

(h∘g)b=hb∘gf′(b). (h \circ g)_b = h_b \circ g_{f'(b)}.

7.3. Multilinear morphisms

7.3.1. Let M1,…,MdM_1, \ldots, M_d and NN be vector bundles with base BB, and let uu be a mapping of the set M1×B⋯×BMdM_1 \times_B \cdots \times_B M_d into NN. We say that uu is a multilinear morphism (or dd-linear) if the following condition is satisfied:

For every b0∈Bb_0 \in B, there exist an open neighbourhood UU of b0b_0 in BB, vector bundle charts tj=(U,φj,Fj)t^j = (U, \varphi^j, F^j) of MjM_j (for 1≤j≤d1 \leq j \leq d) and t=(U,φ,F)t = (U, \varphi, F) of NN, and a mapping λ\lambda of class CrC^r from UU into the Banach space L(F1,…,Fd;F)\mathscr{L}(F^1, \ldots, F^d; F) of continuous dd-linear mappings from F1×⋯×FdF^1 \times \cdots \times F^d into FF, such that:

(tb∘λ(b))(x1,…,xd)=u(tb1(x1),…,tbd(xd)) (t_b \circ \lambda(b))(x_1, \ldots, x_d) = u(t_b^1(x_1), \ldots, t_b^d(x_d))

for every b∈Ub \in U and all xj∈Fjx_j \in F^j.

Every multilinear morphism uu is a morphism of manifolds from the fibre product M1×B⋯×BMdM_1 \times_B \cdots \times_B M_d into NN and induces for every b∈Bb \in B a continuous dd-linear mapping ubu_b from (M1)b×⋯×(Md)b(M_1)_b \times \cdots \times (M_d)_b into NbN_b.

If ff is a morphism from BB into a manifold B′B', a multilinear morphism from M1×B⋯×BMdM_1 \times_B \cdots \times_B M_d into a vector bundle M′M' with base B′B' is by definition the composite of a multilinear morphism from M1×B⋯×BMdM_1 \times_B \cdots \times_B M_d into f∗M′f^*M' with the canonical ff-morphism of f∗M′f^*M' into M′M'.

A bilinear morphism is also called a pairing. For d=1d = 1, a linear morphism is a morphism in the sense of 7.2.1. For d=0d = 0, a 0-linear morphism is identified with a section of NN (7.4).

7.3.2. A bundle of algebras of base BB is called a vector bundle AA of base BB, endowed with a pairing of A×BAA \times_B A into AA. Each fibre AbA_b is then endowed with a structure of KK-algebra. If, for every b∈Bb \in B, the algebra AbA_b has a unit element, denoted by ebe_b, the mapping b↦ebb \mapsto e_b is a section of AA (cf. 7.4). A bundle of algebras AA is said to be locally trivial if, for every point b0b_0 of BB, there exists a vector chart t=(U,φ,E)t = (U, \varphi, E) of AA at the point b0b_0 and a structure of KK-algebra on EE such that tbt_b is an isomorphism of algebras of EE onto AbA_b for every b∈Ub \in U.

7.3.3. Let AA be a bundle of associative algebras with unit element of base BB. A bundle of A-modules of base BB is called a vector bundle MM of base BB endowed with a pairing m:A×BM→Mm : A \times_B M \to M such that the mapping mb:Ab×Mb→Mbm_b : A_b \times M_b \to M_b defines, for every b∈Bb \in B, a structure of AbA_b-module on the fibre MbM_b.

Let MM and M′M' be two bundles of AA-modules. An A-homomorphism of MM into M′M' is called any morphism g:M→M′g : M \to M' of vector bundles of base BB which induces for every bb in BB an AbA_b-linear mapping of MbM_b into Mb′M'_b.

Suppose AA is locally trivial. A fibré MM in AA-modules is said to be locally trivial if, for every point b0b_0 of BB, there exists a vector chart t=(U,φ,E)t = (U, \varphi, E) of AA at b0b_0 as in 7.3.2, a vector chart t′=(U,φ′,L)t' = (U, \varphi', L) of MM at b0b_0, and a module structure of EE on LL such that, for every b∈Ub \in U, tb′t'_b is a tbt_b-isomorphism of the AbA_b-module MbM_b onto the EE-module LL.

7.3.4. Let AA be a Banach algebra over KK (for example a field endowed with a finite-dimensional KK-algebra structure). The trivial fibré ABA_B is then a fibré in algebras, locally trivial. A locally trivial fibré in ABA_B-modules MM is also called a vector fibré over AA (with base BB). The fibres MbM_b are then topological AA-modules. An ff-morphism uu of MM into another vector fibré over AA is said to be AA-linear if the mappings ubu_b are AA-linear for every b∈Bb \in B.

7.4. Sections

7.4.1. Let MM be a vector fibré with base BB. For every open UU of BB, we denote by SMr(U)\mathscr{S}_M^r(U) the set of sections of class CrC^r of MM over UU, that is to say, morphisms ss of class CrC^r from UU into MM such that s(b)∈Mbs(b) \in M_b for every b∈Ub \in U. This set is endowed with a module structure over the ring Cr(U)\mathscr{C}^r(U) of morphic functions by the rules:

(1)

(s+s′)(b)=s(b)+s′(b) (s + s')(b) = s(b) + s'(b)
(2)
(φ⋅s)(b)=φ(b)⋅s(b) (\varphi \cdot s)(b) = \varphi(b) \cdot s(b)
for s,s′s, s' in SMr(U)\mathscr{S}_M^r(U) and φ\varphi in Cr(U)\mathscr{C}^r(U). When the open set UU varies, one obtains a sheaf SMr\mathscr{S}_M^r of mappings from BB into MM (cf. No. 5.4.1), called the sheaf of sections of MM.

7.4.2. Let M1,…,MdM_1, \ldots, M_d and NN be vector bundles with base MM and let uu be a multilinear morphism from M1×B…×BMdM_1 \times_B \ldots \times_B M_d into NN. For 1≤j≤d1 \leq j \leq d, let us give ourselves a section sjs_j of MjM_j over an open set UU of BB; one defines a section u(s1,…,sd)u(s_1, \ldots, s_d) of NN over UU by the formula:

u(s1,…,sd)(b)=ub(s1(b),…,sd(b))for b∈U. u(s_1, \ldots, s_d)(b) = u_b(s_1(b), \ldots, s_d(b)) \quad \text{for } b \in U.

The mapping (s1,…,sd)↦u(s1,…,sd)(s_1, \ldots, s_d) \mapsto u(s_1, \ldots, s_d) is Cr(U)\mathscr{C}^r(U)-multilinear. It is sometimes denoted S(u)\mathscr{S}(u).

7.4.3. Let ff be a morphism of a manifold B′B' into BB and let MM be a vector bundle with base BB. For every open set UU of BB and every s∈SMr(U)s \in \mathscr{S}_M^r(U), the mapping x↦(x,s(f(x)))x \mapsto (x, s(f(x))) is a section of class CrC^r of f∗Mf^*M over the open set f−1(U)f^{-1}(U), denoted f∗sf^*s and called the inverse image of ss by ff. The mapping s↦f∗ss \mapsto f^*s from SMr(U)\mathscr{S}_M^r(U) into Sf∗M(f−1(U))\mathscr{S}_{f^*M}(f^{-1}(U)) is semilinear with respect to the homomorphism g↦g∘(f∣f−1(U))g \mapsto g \circ (f|f^{-1}(U)) of Cr(U)\mathscr{C}^r(U) into Cr(f−1(U))\mathscr{C}^r(f^{-1}(U)).

If moreover NN is a vector bundle with base B′B' and gg an ff-comorphism of MM into NN, one sometimes denotes by S(g)\mathscr{S}(g) the mapping s↦g∘f∗ss \mapsto g \circ f^*s from SMr(U)\mathscr{S}_M^r(U) into SNr(f−1(U))\mathscr{S}_N^r(f^{-1}(U)).

7.4.4. Let MM be a vector bundle with base BB, of finite rank. A frame of MM over an open subset UU of BB is called a finite sequence (s1,…,sn)(s_1, \ldots, s_n) of sections of MM over UU such that (s1(b),…,sn(b))(s_1(b), \ldots, s_n(b)) is a basis of the vector space MbM_b for every b∈Bb \in B. The sequence (s1,…,sn)(s_1, \ldots, s_n) is then a basis of the Cr(U)\mathscr{C}^r(U)-module SMr(U)\mathscr{S}_M^r(U). If ff is a morphism of a manifold B′B' into BB, the sections f∗sjf^*s_j form a frame of f∗Mf^*M over f−1(U)f^{-1}(U).

7.4.5. Let LL be a field, endowed with a structure of KK-algebra of finite dimension and let (M,B,π)(M, B, \pi) be a fibration. Suppose that on each fibre MbM_b a structure of vector space over LL, of finite dimension, is given. There exists then at most one structure of vector bundle over LL with base BB on MM, compatible with the mapping π\pi, the manifold structure of MM and the LL-vector space structures on the fibres (7.3.4). For such a structure to exist, it is necessary and sufficient that the following condition be satisfied:

(FV) For every b0∈Bb_0 \in B, there exists an open neighbourhood UU of b0b_0 in BB and an isomorphism of manifolds φ\varphi from π−1(U)\pi^{-1}(U) onto the product U×FU \times F of UU by a vector space FF over LL of finite dimension, such that for every b∈Ub \in U, the bijection φb\varphi_b of MbM_b onto FF induced by φ\varphi is an isomorphism of vector spaces over the field LL.

The triples (U,φ,F)(U, \varphi, F) satisfying (FV) are then vector bundle charts of the vector bundle MM.

The condition (FV) is equivalent to: (FV') For every b0∈Bb_0 \in B, there exists an integer nn and nn sections s1,…,sns_1, \ldots, s_n of MM over an open neighbourhood UU of b0b_0 such that the mapping

(b,a1,…,an)↦a1s1(b)+⋯+ansn(b) (b, a_1, \ldots, a_n) \mapsto a_1 s_1(b) + \cdots + a_n s_n(b)
is an isomorphism of the manifold U×LnU \times L^n onto the manifold π−1(U)\pi^{-1}(U).

7.4.6. Let M1,…,MdM_1, \ldots, M_d and NN be vector bundles with base BB, the MjM_j being of finite rank. Suppose that for every open set UU of BB a mapping φU\varphi_U from SM1r(U)×⋯×SMdr(U)\mathscr{S}_{M_1}^r(U) \times \cdots \times \mathscr{S}_{M_d}^r(U) into SNr(U)\mathscr{S}_N^r(U) is given, Cr(U)\mathscr{C}^r(U)-multilinear, such that for V⊂UV \subset U one has:

φU(s1,…,sd)∣V=φV(s1∣V,…,sd∣V). \varphi_U(s_1, \ldots, s_d)|V = \varphi_V(s_1|V, \ldots, s_d|V).

There then exists one and only one multilinear morphism uu from M1×B…×BMdM_1 \times_B \ldots \times_B M_d into NN such that φU(s1,…,sd)=u(s1,…,sd)\varphi_U(s_1, \ldots, s_d) = u(s_1, \ldots, s_d) whatever the sections sjs_j of MM on the open set UU of BB.

7.4.7. Let ff be a morphism of a variety B′B' into BB and let MM (resp. M′M') be a vector bundle with base BB (resp. B′B') and of finite rank. Suppose that for every open set UU of BB a mapping φU\varphi_U from SMr(U)\mathscr{S}_M^r(U) into SM′r′(f−1(U)),Cr(U)\mathscr{S}_{M'}^{r'}(f^{-1}(U)), \mathscr{C}^r(U)-semilinear, is given, such that

φU(s)∣f−1(V)=φV(s∣V) \varphi_U(s)|f^{-1}(V) = \varphi_V(s|V)

for every open set V⊂UV \subset U. There then exists one and only one ff-comorphism gg from MM into M′M' such that φU(s)=S(g)(s)\varphi_U(s) = \mathscr{S}(g)(s) for every s∈SMr(U)s \in \mathscr{S}_M^r(U).

*7.4.8. Let F\mathscr{F} be a sheaf of modules over the sheaf of rings CBr\mathscr{C}_B^r. One says that F\mathscr{F} is locally free if for every b∈Bb \in B, there exists an open neighbourhood UU of bb and an integer nn such that F∣U\mathscr{F}|U is isomorphic (as a sheaf of CUr\mathscr{C}_U^r-modules) to the sheaf (CUr)n(\mathscr{C}_U^r)^n.

If MM is a vector bundle of base BB of finite rank, the sheaf SMr\mathscr{S}_M^r is locally free. Conversely, for every locally free sheaf F\mathscr{F} on BB, there exists a vector bundle MM and an isomorphism of sheaves from SMr\mathscr{S}_M^r onto F\mathscr{F}. If M′M' is another vector bundle of base BB of finite rank, the mapping g↦S(g)g \mapsto \mathscr{S}(g) is a bijection from the set of BB-morphisms of MM into M′M' onto the set of morphisms of sheaves of Cr\mathscr{C}^r-modules from SMr\mathscr{S}_M^r into SM′r′\mathscr{S}_{M'}^{r'}*

7.5. Vector subbundles, quotient vector bundles, exact sequences

In this number, a vector bundle means a vector bundle of base BB and a morphism of vector bundles means an IdB\mathrm{Id}_B-morphism.

7.5.1. Let MM be a vector bundle. A subset M′M' of MM is called a vector subbundle of MM if, for every point b∈Bb \in B, there exists a vector chart t=(U,φ,E)t = (U, \varphi, E) of MM at bb and a closed vector subspace FF of EE admitting a topological supplement, such that

φ(π−1(U)∩M′)=U×F. \varphi(\pi^{-1}(U) \cap M') = U \times F.

Under these conditions, there exists on M′M' one and only one structure of vector bundle for which the canonical injection of M′M' into MM is a morphism. For each b∈Bb \in B, the fibre Mb′M'_b of M′M' is the closed vector subspace M′∩MbM' \cap M_b of MbM_b; M′M' is a closed submanifold of MM and the structure of underlying manifold of the structure of vector bundle of M′M' coincides with that induced by the manifold structure of MM.

7.5.2. Let M′M' be a vector subbundle of MM. Let R{x,y}R\{x, y\} denote the following relation between points x,yx, y of MM: « there exists an element bb of BB such that x∈Mb,y∈Mbx \in M_b, y \in M_b and x−y∈Mb′x - y \in M'_b ». Then RR is a regular equivalence relation on MM (cf. n° 5.9.7). On the set M/RM/R, there exists one and only one structure of vector bundle such that the canonical mapping of MM onto M/RM/R is a morphism. We denote it by M/M′M/M' and call the vector bundle thus defined the quotient of MM by M′M'; for each point bb of BB, the bundle (M/M′)b(M/M')_b is the quotient topological vector space Mb/Mb′M_b/M'_b and the manifold structure on M/M′M/M' is the quotient of that of MM.

7.5.3. Retain the hypotheses of 7.5.2. For every point b0b_0 of BB, one can find an open neighbourhood UU of b0b_0, a Banach space FF, direct sum of two closed subspaces F′F' and F′′F'', and an isomorphism of vector bundles ι\iota from FUF_U onto M∣UM|U with the following properties: (i) The restriction ι′\iota' of ι\iota to U×F′U \times F' is an isomorphism from FU′F'_U onto M′∣UM'|U. (ii) If ρ\rho is the canonical mapping of MM onto M′′=M/M′M'' = M/M', and if ι′′\iota'' is the restriction of ι\iota to U×F′′U \times F'', the map ρ∘ι′′\rho \circ \iota'' is an isomorphism from FU′′F''_U onto M′′∣UM''|U.

7.5.4. If a morphism of vector bundles g:L→Mg : L \to M has its image contained in M′M', it is a morphism of vector bundles from LL into M′M'. Consider now a morphism of vector bundles h:M→Nh : M \to N and suppose that, for every bb in BB, the restriction of hbh_b to Mb′M'_b is zero. If ρ\rho is the canonical morphism from MM onto M/M′M/M', there exists one and only one morphism hˉ\bar{h} from M/M′M/M' into NN such that h=hˉ∘ρh = \bar{h} \circ \rho.

7.5.5. Let PP and QQ be two vector bundles, and let gg be a morphism from PP into QQ; for every point bb of BB, denote by NbN_b and IbI_b respectively the kernel and the image of the linear mapping gb:Pb→Qbg_b : P_b \to Q_b. Put N=⋃b∈BNbN = \bigcup_{b \in B} N_b and I=⋃b∈BIbI = \bigcup_{b \in B} I_b. The morphism gg is said to be locally direct if NN is a vector subbundle of PP and II a vector subbundle of QQ. The morphism gg then defines by passing to the quotient an isomorphism of P/NP/N onto II. It is said that NN is the kernel of gg and it is denoted by Ker g\mathrm{Ker}\,g. Analogously, the subbundle II is called the image of gg and is denoted by Im g\mathrm{Im}\,g. If r⩾∞r \geqslant \infty, the morphism gg is locally direct if and only if gg is a subimmersion. If PP is of finite rank, the morphism gg is locally direct if and only if the vector rank of gg is locally constant or equivalently if and only if gg is a subimmersion.

7.5.6. Let M→fM′→gM′′M \xrightarrow{f} M' \xrightarrow{g} M'' be two morphisms of vector bundles. We say that the sequence (f,g)(f, g) is locally direct exact if the two morphisms ff and gg are locally direct and if Im⁡f=Ker⁡g\operatorname{Im} f = \operatorname{Ker} g. If g∘f=0g \circ f = 0, the set DD of points b∈Bb \in B such that the sequence Mb→fbMb′→gbMb′′M_b \xrightarrow{f_b} M'_b \xrightarrow{g_b} M''_b is direct exact (that is to say, such that Ker⁡fb\operatorname{Ker} f_b and Im⁡gb\operatorname{Im} g_b admit topological complements and that Im⁡fb=Ker⁡gb\operatorname{Im} f_b = \operatorname{Ker} g_b) is open and the sequence M∣D→fM′∣D→gM′′∣DM|D \xrightarrow{f} M'|D \xrightarrow{g} M''|D is locally direct exact.

One defines analogously the locally direct exact sequences of arbitrary length. By abuse of language, one sometimes says direct exact sequence instead of locally direct exact sequence.

7.5.7. Let 0→M→fM′→gM′′→00 \to M \xrightarrow{f} M' \xrightarrow{g} M'' \to 0 be a sequence of morphisms of vector bundles. For this sequence to be locally direct exact, it is necessary and sufficient that ff be an isomorphism of MM onto a vector subbundle f(M)f(M) of M′M' and that gg define by passing to the quotient an isomorphism of the quotient vector bundle M′/f(M)M'/f(M) onto M′′M''.

7.6. Vector functors

In this no. and in the following three no.s 7.7 to 7.9, the letter I denotes a finite set, union of two disjoint subsets I+I_+ and I−I_-. We denote by V=(Vi)i∈I\mathscr{V} = (V_i)_{i \in I} (and analogously by V′,V′′,…\mathscr{V}', \mathscr{V}'', \ldots) a family of Banach spaces indexed by II. We denote by Hom⁡(V,V′)\operatorname{Hom}(\mathscr{V}, \mathscr{V}') the Banach space ∏i∈I+L(Vi;Vi′)×∏i∈I−L(Vi′;Vi)\prod_{i \in I_+} \mathscr{L}(V_i; V'_i) \times \prod_{i \in I_-} \mathscr{L}(V'_i; V_i) and by f=(fi)f = (f_i) an element of Hom⁡(V,V′)\operatorname{Hom}(\mathscr{V}, \mathscr{V}'). We denote by Id⁡V\operatorname{Id}_{\mathscr{V}} the element (Id⁡Vi)i∈I(\operatorname{Id}_{V_i})_{i \in I} of Hom⁡(V,V)\operatorname{Hom}(\mathscr{V}, \mathscr{V}). For f∈Hom⁡(V,V′)f \in \operatorname{Hom}(\mathscr{V}, \mathscr{V}') and f′∈Hom⁡(V′,V′′)f' \in \operatorname{Hom}(\mathscr{V}', \mathscr{V}''), we denote by f′∘ff' \circ f the element of Hom⁡(V,V′′)\operatorname{Hom}(\mathscr{V}, \mathscr{V}'') whose components are given by:

(f′∘f)i=fi′∘fiif i∈I+ (f' \circ f)_i = f'_i \circ f_i \quad \text{if } i \in I_+
(f′∘f)i=fi∘fi′if i∈I− (f' \circ f)_i = f_i \circ f'_i \quad \text{if } i \in I_-

7.6.1. A vector functor (resp. vector functor of finite dimension) of type I and of class CrC^r is called the data, for every family V=(Vi)i∈I\mathscr{V} = (V_i)_{i \in I} of Banach spaces (resp. of finite-dimensional vector spaces over KK), of a Banach space τ(V)\tau(\mathscr{V}) and, for every f∈Hom⁡(V,V′)f \in \operatorname{Hom}(\mathscr{V}, \mathscr{V}'), of an element τ(f)∈L(τ(V);τ(V′))\tau(f) \in \mathscr{L}(\tau(\mathscr{V}); \tau(\mathscr{V}')), these data being subject to the following two conditions: (a) One has τ(Id⁡V)=Id⁡τ(V)\tau(\operatorname{Id}_\mathscr{V}) = \operatorname{Id}_{\tau(\mathscr{V})} and τ(f′∘f)=τ(f′)∘τ(f)\tau(f' \circ f) = \tau(f') \circ \tau(f). (b) The mapping τ:Hom⁡(V,V′)→L(τ(V);τ(V′))\tau : \operatorname{Hom}(\mathscr{V}, \mathscr{V}') \to \mathscr{L}(\tau(\mathscr{V}); \tau(\mathscr{V}')) is of class CrC^r.

7.6.2. Let M=(Mi)i∈I\mathscr{M} = (M^i)_{i \in I} be a family of vector bundles with base BB. For b∈Bb \in B, put Mb=(Mbi)i∈I\mathscr{M}_b = (M^i_b)_{i \in I}. Let τ\tau be a vector functor and let τ(M)\tau(\mathscr{M}) be the sum set of the τ(Mb)\tau(\mathscr{M}_b) for b∈Bb \in B; there exists on τ(M)\tau(\mathscr{M}) one and only one structure of vector bundle (with base BB relative to the mapping π\pi of τ(M)\tau(\mathscr{M}) into BB such that, for every b∈Bb \in B, one has τ(Mb)={b}\tau(\mathscr{M}_b) = \{ b \}) possessing the following property:

Let UU be an open set of BB and, for every ii, let ti=(U,φi,Fi)t^i = (U, \varphi_i, F_i) be a vector chart of MiM^i, with domain UU; put F=(Fi)i∈I\mathscr{F} = (F_i)_{i \in I} and let ψb\psi_b be the element of Hom(Mb,F)\mathrm{Hom}(\mathscr{M}_b, \mathscr{F}) defined by (ψb)i=(tbi)−1(\psi_b)_i = (t^i_b)^{-1} for i∈I+i \in I_+ and (ψb)i=tbi(\psi_b)_i = t^i_b for i∈I−i \in I_-; for x∈π−1(U)x \in \pi^{-1}(U), put ψ(x)=(π(x),τ(ψπ(x))(x))\psi(x) = (\pi(x), \tau(\psi_{\pi(x)})(x)). Then the triplet (U,ψ,τ(F))(U, \psi, \tau(\mathscr{F})) is a vector chart of the vector bundle τ(M)\tau(\mathscr{M}).

Endowed with this structure, τ(M)\tau(\mathscr{M}) is called the vector bundle deduced from the family M\mathscr{M} by the vector functor τ\tau.

7.6.3. Let ff be a morphism from BB into a manifold B′B'. Let M=(Mi)\mathscr{M} = (M^i) (resp. M′=(M′i)\mathscr{M}' = ({M'}^i)) be a family indexed by II of vector bundles with base BB (resp. B′B'). For every i∈I+i \in I_+, let gig_i be an ff-morphism from MiM^i into M′i{M'}^i and for every i∈I−i \in I_-, let gig_i be an ff-comorphism from M′i{M'}^i into MiM^i. Put g=(gi)i∈Ig = (g_i)_{i \in I} and for b∈Bb \in B, put gb=((gi)b)i∈Ig_b = ((g_i)_b)_{i \in I} (cf. 7.2.1 and 7.2.6). There exists one and only one ff-morphism, denoted τ(g)\tau(g), from τ(M)\tau(\mathscr{M}) into τ(M′)\tau(\mathscr{M}') such that τ(g)b=τ(gb)\tau(g)_b = \tau(g_b) for every b∈Bb \in B.

If in particular Mi=f∗M′iM^i = f^*{M'}^i, the gig_i being the canonical morphisms or comorphisms, then the BB-morphism from τ(M)\tau(\mathscr{M}) into f∗τ(M′)f^*\tau(\mathscr{M}') defined by τ(g)\tau(g) (7.2.4) is an isomorphism: this fact is expressed by saying that τ\tau commutes with inverse images.

In particular let B′B' be a submanifold of BB and put M∣B′=(Mi∣B′)i∈I\mathscr{M}|B' = (M^i|B')_{i \in I}. The vector bundles τ(M)∣B′\tau(\mathscr{M})|B' and τ(M∣B′)\tau(\mathscr{M}|B') are then canonically B′B'-isomorphic.

7.6.4. Let τ,τ1,…,τd\tau, \tau_1, \ldots, \tau_d be vector functors (of type II and of class C′C'). A d-linear morphism θ\theta from (τ1,…,τd)(\tau_1, \ldots, \tau_d) into τ\tau is the data, for every family V\mathscr{V} of Banach spaces indexed by II, of a continuous dd-linear mapping θV\theta_{\mathscr{V}} from τ1(V)×⋯×τd(V)\tau_1(\mathscr{V}) \times \cdots \times \tau_d(\mathscr{V}) into τ(V)\tau(\mathscr{V}), this data satisfying the following condition: for every f∈Hom(V′,V′′)f \in \mathrm{Hom}(\mathscr{V}', \mathscr{V}'') one has

τ(f)∘θV=θV′′∘(τ1(f)×⋯×τd(f)). \tau(f) \circ \theta_{\mathscr{V}} = \theta_{\mathscr{V}''} \circ (\tau_1(f) \times \cdots \times \tau_d(f)).
For d=1d = 1, one simply says a morphism from τ1\tau_1 into τ\tau.

Let M\mathscr{M} be then a family indexed by II of vector bundles over BB. There exists one and only one B-d-linear morphism θM\theta_{\mathscr{M}} from τ1(M)×B⋯×Bτd(M)\tau_1(\mathscr{M}) \times_B \cdots \times_B \tau_d(\mathscr{M}) into τ(M)\tau(\mathscr{M}) such that (θM)b=θMb(\theta_{\mathscr{M}})_b = \theta_{\mathscr{M}_b} for every b∈Bb \in B.

With the notations of 7.6.3., one has

τ(g)∘θM=θM′∘(τ1(g)×⋯×τd(g)). \tau(g) \circ \theta_{\mathscr{M}} = \theta_{\mathscr{M}'} \circ (\tau_1(g) \times \cdots \times \tau_d(g)).
If d=1d = 1, and if θ\theta is an isomorphism (which means that θV\theta_{\mathscr{V}} is an isomorphism for every family V\mathscr{V}), then θM\theta_{\mathscr{M}} is an isomorphism.

7.6.5. The definitions and results of Nos. 7.6.2 to 7.6.4 extend to the case of finite-dimensional vector functors, provided that the vector bundles given are assumed everywhere to have finite rank.

They also extend to the following case: let L be a field endowed with a structure of finite-dimensional K-algebra; one takes for τ a vector functor over L (i.e. satisfying the hypotheses of 7.6.1 in which K has been replaced by L), and one considers only vector bundles over L in the sense of 7.3.4.

7.6.6. One calls a vector functor (resp. finite-dimensional vector functor) for isomorphisms the data, for every Banach space V (resp. every finite-dimensional vector space over K), of a Banach space τ(V) and, for every isomorphism f of V onto a Banach space V’, of an isomorphism τ(f) of τ(V) onto τ(V’), these data being subject to condition (a) of 7.6.1 and to the following condition:

(b’) The mapping f↦τ(f)f \mapsto \tau(f) from the open subset of L(V;V′)\mathscr{L}(V; V') constituted by the isomorphisms of V onto V’, into L(τ(V);τ(V′))\mathscr{L}(\tau(V); \tau(V')), is of class CrC^r.

The definitions and results of the preceding Nos. extend to the case of vector functors for isomorphisms (by taking I+={1}I_+ = \{1\} and I−=∅I_- = \varnothing), with the exception of those of the first paragraph of No. 7.6.3.

7.7. Direct sums, bundles of multilinear mappings, dual

7.7.1. Assume that I−=∅I_- = \varnothing. One defines a vector functor σ called the direct sum functor by putting σ(V)=⨁i∈IVi\sigma(\mathscr{V}) = \bigoplus_{i \in I} V_i and σ(f)=⨁i∈Ifi\sigma(\mathfrak{f}) = \bigoplus_{i \in I} f_i. If M=(Mi)i∈I\mathscr{M} = (M^i)_{i \in I} is a family of vector bundles over B, the vector bundle σ(M)\sigma(\mathscr{M}) is called the direct sum of the MiM^i and is denoted by ⨁i∈IMi\bigoplus_{i \in I} M^i. For every b∈Bb \in B, the fibre at bb of ⨁i∈IMi\bigoplus_{i \in I} M^i is the direct sum of the fibres of the MiM^i at bb.

Let U be an open subset of B and let si∈FMi(U)s_i \in \mathscr{F}_{M_i}(U) (for i∈Ii \in I). The mapping b↦∑i∈Isi(b)b \mapsto \sum_{i \in I} s_i(b) is then a section, denoted ∑isi\sum_i s_i, of class CrC^r of M=⨁i∈IMiM = \bigoplus_{i \in I} M^i and the mapping (si)i∈I↦∑isi(s_i)_{i \in I} \mapsto \sum_i s_i is an isomorphism of Cr(U)\mathscr{C}^r(U)-modules from ⨁i∈IFMi(U)\bigoplus_{i \in I} \mathscr{F}_{M_i}(U) onto FM(U)\mathscr{F}_M(U).

The underlying manifold of ⨁i∈IMi\bigoplus_{i \in I} M^i is identified with the fibre product ∏BMi\prod_B M^i.

We denote by pri\mathrm{pr}_i the morphism of vector bundles from ⨁i∈IMi\bigoplus_{i \in I} M^i into MiM^i which on each fibre ⨁i∈IMbi\bigoplus_{i \in I} M^i_b is the i-th projection. We define analogously the canonical injection jij_i of MiM^i into ⨁i∈IMi\bigoplus_{i \in I} M^i.

Let ff be a morphism from B into a manifold B′B'; let H be a second finite set and let N=(Nh)h∈H\mathscr{N} = (N^h)_{h \in H} be a family of vector bundles with basis B′B'. The mapping u→⨁i∈I(prh∘u∘ji)(h,i)∈H×Iu \to \bigoplus_{i \in I} (\mathrm{pr}_h \circ u \circ j_i)_{(h,i) \in H \times I} is a bijection from the set of ff-morphisms of ⨁i∈IMi\bigoplus_{i \in I} M^i into ⨁h∈HNh\bigoplus_{h \in H} N^h onto the set of matrices (uh,i)(h,i)∈H×I(u_{h,i})_{(h,i) \in H \times I}, where uh,iu_{h,i} is an ff-morphism from MiM^i into NhN^h.

If I={1,2}I = \{1,2\}, the sequence

0→M1→j1M1⊕M2→pr2M2→0 0 \to M^{1} \xrightarrow{j_1} M^1 \oplus M^2 \xrightarrow{\mathrm{pr}_2} M^2 \to 0
is exact direct.

Conversely, let MM be a vector bundle with basis BB and let M′M' be a vector subbundle of MM. Suppose that the manifold BB is paracompact and that one of the following two conditions is satisfied : (i) KK is different from R\mathbf{R} or C\mathbf{C}; (ii) K=R,r≠ωK = \mathbf{R}, r \neq \omega and the manifold BB admits partitions of unity of class CrC^r (5.3.6).

There then exists a vector subbundle M′′M'' of MM such that MM is identified with the direct sum M′⊕M′′M' \oplus M''.

7.7.2. Suppose that I+={0}I_+ = \{0\} and that I−={1,2,…,d}I_- = \{1,2,\ldots,d\}. We define a vector functor ηd\eta_d of type II and of class CrC^r by setting ηd(V)=L(V1,…,Vd;V0)\eta_d(\mathscr{V}) = \mathscr{L}(V_1,\ldots,V_d; V_0) and ηd(f)(u)=f0∘u∘(f1×⋯×fd)\eta_d(f)(u) = f_0 \circ u \circ (f_1 \times \cdots \times f_d) for u∈ηd(V)u \in \eta_d(\mathscr{V}). If M=(Mi)i∈I\mathscr{M} = (M_i)_{i \in I} is a family of vector bundles with basis BB, the vector bundle ηd(M)\eta_d(\mathscr{M}) is denoted by L(M1,…,Md;M0)\mathscr{L}(M_1,\ldots,M_d; M_0).

Let uu be a multilinear morphism from M1×B⋯×BMdM_1 \times_B \cdots \times_B M_d into M0M_0. The mapping u^:b↦ub\hat{u} : b \mapsto u_b is then a section of L(M1,…,Md;M0)\mathscr{L}(M_1,\ldots,M_d; M_0) and the mapping u↦u^u \mapsto \hat{u} is bijective.

7.7.3. Retain the notations of 7.7.2 and suppose moreover that d=1d = 1. The vector bundle L(M1;M0)\mathscr{L}(M_1; M_0) is then called the bundle of homomorphisms from M1M_1 into M0M_0. Its sections correspond to the BB-morphisms from M1M_1 into M0M_0.

If moreover M0M_0 is the trivial bundle KBK_B, the vector bundle L(M1;KB)\mathscr{L}(M_1; K_B) is called the dual of M=M1M = M_1 and is denoted by M′M': the fibre (M′)b(M')_b is the space of continuous linear forms on the fibre MbM_b of MM at the point b∈Bb \in B.

If ss (resp. tt) is a section of MM (resp. M′M') over an open set UU of BB, the mapping b↦(b,⟨s(b),t(b)⟩)b \mapsto (b, \langle s(b), t(b) \rangle) is a section, denoted ⟨s,t⟩\langle s, t \rangle of the trivial bundle KBK_B.1^1

7.8. Bundles of alternating multilinear mappings

In nos. 7.8.1 to 7.8.5, it is assumed that KK is of characteristic 0 or that the vector bundles considered are of finite rank. It is not known whether the definition of the vector functor αd\alpha_d given in 7.8.1 can be made without any restriction.

7.8.1. Let I+={0},I−={1}I_+ = \{0\}, I_- = \{1\} and let dd be an integer ⩾1\geqslant 1. We define a

1^1 When MM is of finite rank, M∗M^* is written instead of M′M'.

vector functor αd\alpha_d by denoting by αd(V)\alpha_d(\mathscr{V}) the Banach space of continuous alternating dd-linear mappings from V1dV_1^d into V0V_0 and by setting αd(f)(u)=f0∘u∘f1d\alpha_d(f)(u) = f_0 \circ u \circ f_1^d for u∈αd(V)u \in \alpha_d(\mathscr{V}). The vector bundle αd((M1,M0))\alpha_d((M_1, M_0)) is denoted Altd(M1;M0)\mathrm{Alt}^d(M_1; M_0) and is called the vector bundle of alternating dd-linear mappings from M1M_1 into M0M_0.

The canonical injection of Altd(M1;M0)\mathrm{Alt}^d(M_1; M_0) into L(M1,…,M1;M0)\mathscr{L}(M_1, \ldots, M_1; M_0) is a morphism of vector bundles; Altd(M1;M0)\mathrm{Alt}^d(M_1; M_0) is a vector subbundle of L(M1,…,M1;M0)\mathscr{L}(M_1, \ldots, M_1; M_0).

One has Alt1(M1;M0)=L(M1;M0)\mathrm{Alt}^1(M_1; M_0) = \mathscr{L}(M_1; M_0). We put Alt0(M1;M0)=M0\mathrm{Alt}^0(M_1; M_0) = M_0.

If ω\omega is a section 1^1 of Altd(M1;M0)\mathrm{Alt}^d(M_1; M_0) and if s1,…,sds_1, \ldots, s_d are sections of M1M_1, there exists one and only one section of M0M_0, denoted ω(s1,…,sd)\omega(s_1, \ldots, s_d), such that

ω(s1,…,sd)(b)=ω(b)(s1(b),…,sd(b))for all b∈B. \omega(s_1, \ldots, s_d)(b) = \omega(b)(s_1(b), \ldots, s_d(b)) \quad \text{for all } b \in B.

7.8.2 Let φ\varphi be a pairing of N×BN′N \times_B N' into N′′N'' (cf. No. 7.3.1); for each bb, we have a bilinear mapping φb\varphi_b of Nb×Nb′N_b \times N'_b into Nb′′N''_b which defines (cf. A, III, p. 142) a bilinear mapping of

Altd(M;N)b×Alte(M;N′)b \mathrm{Alt}^d(M; N)_b \times \mathrm{Alt}^e(M; N')_b
into Altd+e(M;N′′)b\mathrm{Alt}^{d+e}(M; N'')_b. The collection of these bilinear mappings defines a pairing uu of
Altd(M;N)×BAlte(M;N′) \mathrm{Alt}^d(M; N) \times_B \mathrm{Alt}^e(M; N')
into Altd+e(M;N′′)\mathrm{Alt}^{d+e}(M; N''); if ω\omega and ω′\omega' are respectively sections of Altd(M;N)\mathrm{Alt}^d(M; N) and Alte(M;N′)\mathrm{Alt}^e(M; N') on an open set UU, the section u(ω,ω′)u(\omega, \omega') of Altd+e(M;N′′)\mathrm{Alt}^{d+e}(M; N'') on UU will be denoted by ω∧φω′\omega \wedge_\varphi \omega', and called the exterior product of ω\omega and ω′\omega'. We have the formula: (1)
(ω∧φω′)(s1,…,sd+e)=∑σεσφ(ω(sσ(1),…,sσ(d)),ω′(sσ(d+1),…,sσ(d+e))) (\omega \wedge_\varphi \omega')(s_1, \ldots, s_{d+e}) = \sum_\sigma \varepsilon_\sigma \varphi(\omega(s_{\sigma(1)}, \ldots, s_{\sigma(d)}), \omega'(s_{\sigma(d+1)}, \ldots, s_{\sigma(d+e)}))
where the sis_i are sections of MM on UU, and where the summation is extended to the permutations σ\sigma of {1,2,…,d+e}\{1, 2, \ldots, d+e\} such that
σ(1)<⋯<σ(d)andσ(d+1)<⋯<σ(d+e). \sigma(1) < \cdots < \sigma(d) \quad \text{and} \quad \sigma(d+1) < \cdots < \sigma(d+e).

7.8.3. Let MM be a vector bundle and AA an algebra bundle, with base BB. Suppose that the fibres AbA_b of AA are associative and commutative algebras, possessing a unit element, denoted by ebe_b. For every open set UU of BB, we shall denote by Ωd(U)\Omega^d(U) the Cr(U)\mathscr{C}^r(U)-module formed by the sections of the bundle Altd(M;A)\mathrm{Alt}^d(M; A) and by Ω∗(U)\Omega^*(U) the direct sum of the Ωd(U)\Omega^d(U) for d≥0d \geq 0. The multiplications on each fibre define a pairing of A×BAA \times_B A into AA, whence (7.8.2) a graded algebra structure on Ω∗(U)\Omega^*(U), which is associative and anticommutative. The subalgebra Ω0(U)\Omega^0(U) is the algebra of

¹ The reader should take care not to confuse this use of the letter ω\omega with that defined p. 10.

sections of A. An element ω\omega of Ω1(U)\Omega^1(U) identifies with a UU-morphism of M∣UM|U into A∣UA|U (7.7.3): if s∈SM′(U)s \in \mathscr{S}'_M(U), we shall denote by ⟨ω,s⟩\langle \omega, s \rangle the section ω(s)\omega(s) of A (7.4.2). Let sj∈SM′(U)s_j \in \mathscr{S}'_M(U) and ωj∈Ω1(U)\omega_j \in \Omega^1(U) (for 1≤j≤d1 \leq j \leq d); we have:

(2)

ω(s1,…,sd)=det⁡(⟨ωi,sj⟩)for ω=ω1∧…∧ωd. \omega(s_1, \ldots, s_d) = \det(\langle \omega_i, s_j \rangle) \quad \text{for } \omega = \omega_1 \wedge \ldots \wedge \omega_d.

7.8.4. Let d≥1d \geq 1. There exists a pairing ii of M×BAltd(M;A)M \times_B \mathrm{Alt}^d(M; A) into Altd−1(M;A)\mathrm{Alt}^{d-1}(M; A) whose restriction to each fibre is given by the right interior product (cf. A, III, p. 156). If ss is a section of MM on the open set UU and if ω∈Ωd(U)\omega \in \Omega^d(U), we denote by i(s)ωi(s)\omega the section i(s,ω)i(s, \omega) of Altd−1(M;A)\mathrm{Alt}^{d-1}(M; A) on UU; we set i(s)ω=0i(s)\omega = 0 for ω\omega in Ω0(U)\Omega^0(U).

One thus associates with every section ss of MM on UU an endomorphism of the C′(U)\mathscr{C}'(U)-module Ω∗(U)\Omega^*(U). We have the following formulas:

(3)

(i(s)ω)(s1,…,sd−1)=ω(s,s1,…,sd−1)for ω∈Ωd(U),d≥1(i(s)\omega)(s_1, \ldots, s_{d-1}) = \omega(s, s_1, \ldots, s_{d-1}) \quad \text{for } \omega \in \Omega^d(U), d \geq 1
(4)
i(s)∘i(s)=0i(s) \circ i(s) = 0
(5)
i(s)ω=⟨ω,s⟩for ω∈Ω1(U)i(s)\omega = \langle \omega, s \rangle \quad \text{for } \omega \in \Omega^1(U)
(6)
i(s).(ω∧ω′)=i(s)ω∧ω′+(−1)dω∧i(s)ω′for ω∈Ωd(U)i(s) . (\omega \wedge \omega') = i(s)\omega \wedge \omega' + (-1)^d \omega \wedge i(s)\omega' \quad \text{for } \omega \in \Omega^d(U)
(7)
i(s)(ω1∧…∧ωp)=∑i=1p(−1)i+1⟨ωi,s⟩ω1∧…∧ω^i∧…∧ωd.i(s)(\omega_1 \wedge \ldots \wedge \omega_p) = \sum_{i=1}^p (-1)^{i+1} \langle \omega_i, s \rangle \omega_1 \wedge \ldots \wedge \hat{\omega}_i \wedge \ldots \wedge \omega_d.

In the last formula, the ωi\omega_i are in Ω1(U)\Omega^1(U) and the sign ^\hat{} indicates that the symbol which it surmounts is to be omitted.

All the operations described above on the sections are multilinear over the ring C′(U;K)\mathscr{C}'(U; K).

7.8.5. Let L be a Banach algebra over K. The definitions and results of Nos. 7.7 and 7.8 extend to the case of vector bundles over L: one defines in an analogous manner the bundles of L-multilinear or L-multilinear alternating mappings.

7.9. Tensor products, tensor spaces, exterior algebra

We retain the notations of 7.6. Moreover, we denote by L a commutative field endowed with a structure of finite-dimensional K-algebra and we call vector bundle a vector bundle over L, with base B and locally finite rank.

7.9.1. Suppose that I−=∅I_- = \varnothing. If V\mathscr{V} and V′\mathscr{V}' are two families indexed by I of finite-dimensional vector spaces over L, we denote by τ(V)\tau(\mathscr{V}) the tensor product of the ViV_i for i∈Ii \in I (A, II, p. 71) and if f∈Hom(V,V′)f \in \mathrm{Hom}(\mathscr{V}, \mathscr{V}'), we set τ(f)=⊗fi\tau(f) = \otimes f_i. We thus define a vector functor over L in finite dimension and if M=(Mi)i∈I\mathscr{M} = (M_i)_{i \in I} is a family of vector bundles, we denote by ⨂i∈IMi\bigotimes_{i \in I} M_i and call the tensor product (over L) of the MiM_i the vector bundle τ(M)\tau(\mathscr{M}).

If sis_i is a section of MiM_i on the open set U of B (for i∈Ii \in I), the mapping b↦⨂i∈Isi(b)b \mapsto \bigotimes_{i \in I} s_i(b) is a section of ⨂i∈IMi\bigotimes_{i \in I} M_i, denoted by ⨂i∈Isi\bigotimes_{i \in I} s_i. The mapping (si)i∈I↦⨂i∈Isi(s_i)_{i \in I} \mapsto \bigotimes_{i \in I} s_i is multilinear over the ring Cr(U;L)\mathscr{C}^r(U;L).

7.9.2. The canonical isomorphisms defined in Alg., Chap. II provide isomorphisms of vector functors. It follows from 7.6.4 that there are isomorphisms of vector bundles. For example, there are canonical isomorphisms:

(M1⊕M2)⊗M3⟶(M1⊗M3)⊕(M2⊗M3) (M_1 \oplus M_2) \otimes M_3 \longrightarrow (M_1 \otimes M_3) \oplus (M_2 \otimes M_3)
M1∗⊗M2⟶L(M1;M2) M_1^* \otimes M_2 \longrightarrow \mathscr{L}(M_1; M_2)

etc.

7.9.3. Let MM be a vector bundle and let II and JJ be two finite disjoint sets. The tensor bundle TJI(M)T^I_J(M) is defined as the tensor product ⨂α∈I∪JMα\bigotimes_{\alpha \in I \cup J} M_\alpha, where Mα=MM_\alpha = M if α∈I\alpha \in I, and Mα=M∗M_\alpha = M^* if α∈J\alpha \in J, M∗M^* denoting the dual of MM (A, III, p. 63). The fibre TJI(M)bT^I_J(M)_b of this bundle at a point bb is equal to the tensor space TJI(Mb)T^I_J(M_b) defined in Alg., loc. cit. When I={1,…,p}I = \{1, \ldots, p\} and J={p+1,…,p+q}J = \{p+1, \ldots, p+q\}, one writes Tqp(M)T^p_q(M) in place of TJI(M)T^I_J(M); one has

Tqp(M)=(⨂pM)⊗(⨂qM∗). T^p_q(M) = (\bigotimes^p M) \otimes (\bigotimes^q M^*).
The giving of a total order on II and on JJ defines a canonical isomorphism of TJI(M)T^I_J(M) onto Tqp(M)T^p_q(M).

7.9.4. If II (resp. JJ) is the union of two disjoint subsets I′I' and I′′I'' (resp. J′J' and J′′J''), TJI(M)T^I_J(M) is canonically identified with the tensor product TJ′I′(M)⊗TJ′′I′′(M)T^{I'}_{J'}(M) \otimes T^{I''}_{J''}(M). In particular, if s′s' (resp. s′′s'') is a section of TJ′I′(M)T^{I'}_{J'}(M) (resp. of TJ′′I′′(M)T^{I''}_{J''}(M)), the tensor product s′⊗s′′s' \otimes s'' is identified with a section of TJI(M)T^I_J(M).

7.9.5. The dual of TJI(M)T^I_J(M) is identified with TIJ(M)T^J_I(M).

7.9.6. Let i∈Ii \in I and j∈Jj \in J. For every b∈Bb \in B, the homomorphism of contraction of the indices ii and jj is defined (cf. Alg., loc. cit.); it is a homomorphism (cji)b:TJI(Mb)→TJ−{j}I−{i}(Mb)(c^i_j)_b : T^I_J(M_b) \to T^{I-\{i\}}_{J-\{j\}}(M_b). The collection of the (cji)b(c^i_j)_b defines a morphism of vector bundle spaces

cji:TJI(M)→TJ−{j}I−{i}(M), c^i_j : T^I_J(M) \to T^{I-\{i\}}_{J-\{j\}}(M),

also called contraction of the indices ii and jj. One defines analogously the contraction of the indices i1,…,iki_1, \ldots, i_k of II with the indices j1,…,jkj_1, \ldots, j_k of JJ.

7.9.7. Let dd be an integer ⩾0\geqslant 0; let VV and V′V' be two vector spaces of finite dimension over LL and f∈HomL(V,V′)f \in \mathrm{Hom}_L(V, V'); put λd(V)=⋀d(V)\lambda_d(V) = \bigwedge^d(V) and λd(f)=⋀d(f)\lambda_d(f) = \bigwedge^d(f). One thus defines a vector functor over LL in finite dimension and, if MM is a vector bundle, one denotes by ⋀d(M)\bigwedge^d(M) and calls dd-th exterior power of MM (over LL) the vector bundle λd(M)\lambda_d(M).

The canonical mapping of ⨂dV\bigotimes^d V onto ⋀d(V)\bigwedge^d(V) defines a morphism of vector functors, whence a canonical morphism of ⨂dM\bigotimes^d M into ⋀d(M)\bigwedge^d(M). This morphism is surjective.

The canonical isomorphisms of the space of alternating dd-multilinear mappings on the space ⋀d(V∗)\bigwedge^d(V^*) or on (⋀d(V))∗(\bigwedge^d(V))^* provide isomorphisms, called canonical, of the vector bundle Altd(M;L)\mathrm{Alt}^d(M; L) of alternating dd-multilinear mappings on LL, onto the vector bundle ⋀d(M∗)\bigwedge^d(M^*) or onto (⋀d(M))∗(\bigwedge^d(M))^*.

7.9.8. Put now λ(V)=⋀(V)\lambda(V) = \bigwedge(V) and λ(f)=⋀(f)\lambda(f) = \bigwedge(f); one thus again defines a vector functor on LL in finite dimension. The vector bundle λ(M)\lambda(M) is denoted by ⋀(M)\bigwedge(M). Its fibre ⋀(M)b\bigwedge(M)_b at b∈Bb \in B is the exterior algebra (over LL) of the fibre MbM_b. The vector bundle ⋀(M)\bigwedge(M) is a locally trivial vector bundle in algebras.

The definitions and properties of the interior products given in Alg., Chap. III, 3rd ed., § 10 extend immediately to the sections of the vector bundles ⋀(M)\bigwedge(M) and ⋀(M∗)\bigwedge(M^*) (cf. also the formulas (1) to (7) of Nos. 7.8.2 to 7.8.4).

7.9.9. Let MM be a vector bundle. For every integer nn, let BnB_n be the set (open) of the points b∈Bb \in B such that the dimension (over LL) of MbM_b is equal to nn. For b∈Bnb \in B_n, put Nb=⋀n(Mb)N_b = \bigwedge^n(M_b) and let NN be the sum set of the NbN_b for b∈Bb \in B. There exists on NN one and only one structure of vector bundle such that the obvious mapping of N∣BnN|B_n into ⋀n(M)∣Bn\bigwedge^n(M)|B_n is an isomorphism for every nn. Endowed with this structure, the vector bundle NN of rank one at each point is denoted by det⁡(M)\det(M).

7.10. Vector bundles and principal bundles

7.10.1. Let FF be a Banach space. A vector bundle MM with base BB is said to be pure of type FF if all the fibres MbM_b of MM (for b∈Bb \in B) are isomorphic (as Banach spaces) to FF.

Let MM be a vector bundle of base BB pure of type FF and let PP be the open subvariety of the vector bundle L(FB;M)\mathscr{L}(F_B; M) composed of the pairs (b,u)(b, u) where b∈Bb \in B and where uu is an isomorphism of Fb=FF_b = F onto MbM_b. The group GL(F)\mathrm{GL}(F) of automorphisms of FF operates on the right on PP by setting (b,u)⋅g=(b,u∘g)(b, u) \cdot g = (b, u \circ g) for (b,u)∈P(b, u) \in P and g∈GL(F)g \in \mathrm{GL}(F). Let πP\pi_P denote the mapping (b,u)↦b(b, u) \mapsto b from P into B. The quadruplet λ=(P,GL(F),B,πP)\lambda = (P, GL(F), B, \pi_P) (where GL(F)GL(F) is endowed with its canonical structure of group variety (5.12.2)) is a principal fibration (6.2.1): it is called the fibration of frames of M. The mapping ((b,u),h)↦u(h)((b, u), h) \mapsto u(h) from P×FP \times F into M endows M with a structure of associated bundle to λ\lambda, of fibre type F (6.5.1).

When F=KnF = K^n, one can identify an isomorphism uu of F onto MbM_b with the basis of MbM_b image by uu of the canonical basis of KnK^n. The fibred space of frames of M is then identified with the open subvariety of M×B…×BMM \times_B \ldots \times_B M formed by the bases (e1,…,en)(e_1, \ldots, e_n) of the different fibres MbM_b.

Let U be an open subset of B and let t=(U,φ,E)t = (U, \varphi, E) be a vector chart of M of domain U, with E=FE = F. The mapping b↦(b,tb)b \mapsto (b, t_b) is then a section of P, denoted t~\tilde{t}, and the mapping t↦t~t \mapsto \tilde{t} is a bijection from the set of vector charts of M of the form (U,φ,F)(U, \varphi, F) onto the set of sections of P over U.

7.10.2. Conversely, let λ=(Q,G,B,πQ)\lambda = (Q, G, B, \pi_Q) be a principal fibration and let φ\varphi be a homomorphism of group varieties of G into the group GL(F)GL(F) of automorphisms of a Banach space F. Let G operate on the left on F by setting g.h=φ(g)(h)g . h = \varphi(g)(h) (h∈F,g∈Gh \in F, g \in G). Let M be a fibred space associated with λ\lambda, of fibre type F, and let ρ:Q×F→M\rho : Q \times F \to M be its frame mapping (6.5.1). Let π\pi be the mapping from M into B defined by

π(ρ(q,h))=πQ(q)(q∈Q and h∈F). \pi(\rho(q, h)) = \pi_Q(q) \quad (q \in Q \text{ and } h \in F).
Let s be a section of Q over an open subset U of B; the mapping
s~:(b,h)↦(s(b),h) \tilde{s} : (b, h) \mapsto (s(b), h)
is then a bijection from U×FU \times F onto π−1(U)\pi^{-1}(U). There exists on the pair (M,π)(M, \pi) one and only one structure of vector bundle of base B for which the triples ts=(U,s~−1,F)t_s = (U, \tilde{s}^{-1}, F) are vector charts (for every section s of Q). The structure of underlying variety of this structure is that of the fibred space associated with λ\lambda.

Let q∈Qq \in Q; put b=πQ(q)b = \pi_Q(q) and let uu be the isomorphism of F onto MbM_b defined by u(h)=ρ(q,h)u(h) = \rho(q, h) (for h∈Fh \in F). The mapping f:q↦(b,u)f : q \mapsto (b, u) is a B-morphism of principal fibrations, compatible with φ\varphi, from (Q,G,B,πQ)(Q, G, B, \pi_Q) into the frame bundle (P,GL(F),B,πP)(P, GL(F), B, \pi_P) of the vector bundle M.

7.10.3. Let us take up again the notations of No. 7.6. For every i∈Ii \in I, let

λi=(Pi,Gi,B,πi) \lambda_i = (P_i, G_i, B, \pi_i)
be a principal fibration with base B and suppose that GiG_i operates on the left on a Banach space ViV_i by means of a homomorphism
φi:Gi→GL(Vi). \varphi_i : G_i \to GL(V_i).
Let MiM_i be a fibre space associated with λi\lambda_i of fibre type ViV_i. Put M=(Mi)i∈I\mathscr{M} = (M_i)_{i \in I} and V=(Vi)i∈I\mathscr{V} = (V_i)_{i \in I} and let λ\lambda be the product principal fibration of the λi\lambda_i over B (6.2.5). Put λ^=(P,G,B,πp)\hat{\lambda} = (\mathbf{P}, G, B, \pi_p), with G=∏i∈IGiG = \prod_{i \in I} G_i.

Let now τ\tau be a vector functor. For g=(gi)∈Gg = (g_i) \in G, let φ(g)\varphi(g) be the element of Hom(V,V)\mathrm{Hom}(\mathscr{V}, \mathscr{V}) defined by:

φ(g)i=φi(gi)si i∈I+ \varphi(g)_i = \varphi_i(g_i) \quad \text{si } i \in I_+
φ(g)i=φi(gi)−1si i∈I− \varphi(g)_i = \varphi_i(g_i)^{-1} \quad \text{si } i \in I_-

The group GG then operates on τ(V)\tau(\mathscr{V}) by means of the morphism g↦τ(φ(g))g \mapsto \tau(\varphi(g)) from GG into GL(τ(V))\mathbf{GL}(\tau(\mathscr{V})).

Let on the other hand x=(xi)x = (x_i) be a point of P\mathbf{P} and let b=πp(x)b = \pi_p(x). For each ii, the mapping θxi\theta_{x_i} defined in No. 6.5.2 is an isomorphism of ViV_i onto (Mi)b(M_i)_b. Let θx\theta_x be the element of Hom(V,((Mi)b)i∈I)\mathrm{Hom}(\mathscr{V}, ((M_i)_b)_{i \in I}) defined by:

(θx)i=θxisi i∈I+ (\theta_x)_i = \theta_{x_i} \quad \text{si } i \in I_+
(θx)i=θxi−1si i∈I− (\theta_x)_i = \theta_{x_i}^{-1} \quad \text{si } i \in I_-

Let ρ\rho be the mapping (x,h)↦(b,τ(θx)(h))(x, h) \mapsto (b, \tau(\theta_x)(h)) from P×τ(V)\mathbf{P} \times \tau(\mathscr{V}) into the vector bundle τ(M)\tau(\mathscr{M}); the mapping ρ\rho endows τ(M)\tau(\mathscr{M}) with a structure of fibre space associated with λ\lambda of fibre type τ(V)\tau(\mathscr{V}).

These considerations generalize to the case of finite-dimensional vector functors, or of vector functors over a field LL endowed with a structure of finite-dimensional K-algebra.

7.11. Change of structure

The structures and operations described in this paragraph are compatible with the changes of structure described in Nos. 5.13 and 5.14.

Machine translation of the English, by gpt-5-mini, gpt-5-6-mini, gpt-5-6, against glossary version 34. Not checked by a person. That model is a cut down version of the one the rest of this language was written by.