§9. Complements on Quasi-coherent Sheaves
9.1. Tensor product of quasi-coherent sheaves
Proposition (9.1.1). Let be a prescheme (resp. a locally Noetherian prescheme). Let and be two quasi-coherent (resp. coherent) -Modules; then is quasi-coherent (resp. coherent), and of finite type if and are of finite type. If admits a finite presentation and if is quasi-coherent (resp. coherent), then is quasi-coherent (resp. coherent).
Proof. The question being local, one may suppose affine (resp. affine Noetherian); moreover, if is coherent, one may suppose that it is the cokernel of a homomorphism . The assertions concerning quasi-coherent sheaves then follow from (1.3.12) and (1.3.9); the assertions concerning coherent sheaves follow from (1.5.1) and from the fact that, if and are modules of finite type over a Noetherian ring , then and are -modules of finite type.
Definition (9.1.2). Let , be two -preschemes, , the projections of , and (resp. ) a quasi-coherent -Module (resp. -Module). One calls tensor product of and over (or over ), and one denotes by (or ), the tensor product on the prescheme .
If () are -preschemes and a quasi-coherent -Module (), one defines in the same way the tensor product on the prescheme ; it is a quasi-coherent -Module by virtue of (9.1.1) and (0, 5.1.4); it is coherent if the are and if is locally Noetherian, by virtue of (9.1.1), (0, 5.3.11), and (6.1.1).
It will be noted that if one takes , the definition (9.1.2) gives back the tensor product of -Modules. Moreover, since (0, 4.3.4), the product is canonically identified with , and likewise is canonically identified with . More particularly, if one takes and one denotes by the structural morphism , one has : the ordinary tensor product and the inverse image thus appear as particular cases of the general tensor product.
The definition (9.1.2) entails immediately that, for and fixed, is a covariant additive bifunctor, right exact in and .
Proposition (9.1.3). Let , , be three affine schemes of rings , , respectively, and being -algebras. Let (resp. ) be a -module (resp. -module), and (resp. ) the associated quasi-coherent sheaf; then is canonically isomorphic to the sheaf associated to the -module .
Proof. Indeed, by virtue of (1.6.5), is canonically isomorphic to the sheaf associated to the -module and, by reason of the canonical isomorphisms between tensor products, this last is isomorphic to .
Proposition (9.1.4). Let , be two -morphisms, and (resp. ) a quasi-coherent -Module (resp. -Module). One then has .
Proof. If , are the projections of , the formula results in fact from the relations and (0, 3.5.5), and from the fact that an inverse image of a tensor product of algebraic sheaves is the tensor product of their inverse images (0, 4.3.3).
Corollary (9.1.5). Let , be two -morphisms, and (resp. ) a quasi-coherent -Module (resp. -Module). One then has
Proof. This results from (9.1.4) and from the fact that , and being the projections of .
Corollary (9.1.6). Let , , be three -preschemes, and (resp. , ) a quasi-coherent -Module (resp. -Module, -Module); the sheaf is the inverse image of under the canonical isomorphism of onto .
Proof. Indeed, this isomorphism is written , denoting by , , the projections of .
Likewise, the inverse image of under the canonical isomorphism of onto is .
Corollary (9.1.7). If is an -prescheme, every quasi-coherent -Module is the inverse image of under the canonical isomorphism of onto (3.3.3).
Proof. Indeed, this isomorphism is , where is the structural morphism , and the corollary results from (9.1.4) and from the fact that .
(9.1.8) Let be an -prescheme, a quasi-coherent -Module, and a morphism; one denotes by or the quasi-coherent sheaf on ; thus , where is the projection .
Proposition (9.1.9). Let be a morphism. For every quasi-coherent -Module on the -prescheme , is the inverse image of under the canonical isomorphism (3.3.9).
Proof. This results at once from the definitions and from (3.3.9), and is also written
Proposition (9.1.10). Let be an -prescheme, an -morphism. For every quasi-coherent -Module and every morphism , one has .
Proof. This results at once from the commutativity of the diagram
Corollary (9.1.11). Let , be two -preschemes, and (resp. ) a quasi-coherent -Module (resp. -Module). The inverse image of the sheaf under the canonical isomorphism (3.3.10) is equal to .
Proof. If , are the projections of , the isomorphism in question is none other than ; the corollary results from the propositions (9.1.4) and (9.1.10).
Proposition (9.1.12). With the notations of (9.1.2), let be a point of , , ; the stalk is isomorphic to .
Proof. Since one may reduce to the affine case, the proposition results from the formula (1.6.5.1).
Corollary (9.1.13). If and are of finite type, one has
Proof. Since and are of finite type over , one is reduced, by (9.1.12) and (0, 1.7.5), to the particular case where , that is, to proving the formula The same reasoning as in (0, 1.7.5) reduces this to verifying that one has, for every , (with ), which follows from the fact that the homomorphism is local by hypothesis.
We leave to the reader the task of extending to a product of an arbitrary number of factors the results proved in this number for two factors.
9.2. Direct image of a quasi-coherent sheaf
Proposition (9.2.1). Let be a morphism of preschemes. Suppose that there exists a covering of by affine opens having the following property: each of the admits a finite covering by affine opens contained in , such that each of the intersections is itself a finite union of affine opens. Under these conditions, for every quasi-coherent -Module , is a quasi-coherent -Module.
Proof. The question being local on , one may suppose equal to one of the , and thus suppress the indices .
a) Suppose first that the are themselves affine opens. Put , , and let and be the images of and respectively under the restrictions of to and ; one knows that the and are quasi-coherent (1.6.3). Put , ; and are quasi-coherent -Modules; we are going to define a homomorphism such that is the kernel of ; it will result that is quasi-coherent (1.3.9). It suffices to define as a homomorphism of presheaves; taking account of the definitions of and , it therefore suffices, for every open set , to define a homomorphism so as to satisfy the usual compatibility conditions as varies. If, for every section , one denotes by its restriction to , one will put and the compatibility conditions are evidently fulfilled. To prove that the kernel of is , let us define a homomorphism of into by making correspond to every section the family , where is the restriction of to ; the axioms and of sheaves (G, II, 1.1) entail that this homomorphism is bijective, which completes the demonstration in this case.
b) In the general case, the same reasoning applies once one has established that the are quasi-coherent. Now, by hypothesis, is a finite union of affine opens ; and since the are affine opens in a scheme, the intersection of any two of them is again an affine open (5.5.6). One is thus reduced to the first case, and (9.2.1) is therefore proved.
Corollary (9.2.2). The conclusion of (9.2.1) holds in each of the following cases:
a) is separated and quasi-compact.
b) is separated and of finite type.
c) is quasi-compact and the space underlying is locally Noetherian.
Proof. In case a), the are affine (5.5.6). Case b) is a particular case of a) (6.6.3). Finally, in case c), one may reduce to the case where is affine and the space underlying is Noetherian; then admits a finite affine open covering , and the , being quasi-compact, are finite unions of affine opens (2.1.3).
9.3. Extension of sections of quasi-coherent sheaves
Theorem (9.3.1). Let be a prescheme whose underlying space is Noetherian, or a scheme whose underlying space is quasi-compact. Let be an invertible -Module (0, 5.4.1), a section of over , the open set of those where (0, 5.5.1), and a quasi-coherent -Module.
(i) If is such that , there exists an integer such that .
(ii) For every section , there exists an integer such that extends to a section of over .
Proof. (i) Since the space underlying is quasi-compact, hence a finite union of affine opens such that is isomorphic to , one is reduced to the case where is affine and . In this case, is identified with an element of and one has ; is identified with an element of an -module , and with the corresponding element of , and the result is trivial, taking account of the definition of a module of fractions.
(ii) Again is a finite union of affine opens () such that , and for each , is identified, by the preceding isomorphism, with . One then knows (1.4.1) that there exists an integer such that for each , extends to a section of over . Let be the restriction of to ; one has by definition in . Now, if is a Noetherian space, is quasi-compact; if is a scheme, is an affine open (5.5.6), hence again quasi-compact. By virtue of (i), there thus exists an integer (independent of and ) such that . One concludes at once that there exists a section of over , inducing over each , and inducing consequently over .
The corollaries that follow give an interpretation of the theorem (9.3.1) in a more algebraic language:
Corollary (9.3.2). The hypotheses being those of (9.3.1), consider the graded ring and the graded -module (0, 5.4.6). If , where , then one has a canonical isomorphism (the subgroup of the module of fractions formed of the elements of degree ).
Corollary (9.3.3). Suppose the hypotheses of (9.3.1) verified, and suppose in addition that . Then if one puts , , the -module is canonically isomorphic to .
Proposition (9.3.4). Let be a Noetherian prescheme, a coherent -Module, and a coherent sheaf of ideals in , such that the support of is contained in that of . Then there exists an integer such that .
Proof. Since is a finite union of affine opens whose rings are Noetherian, one may suppose affine with Noetherian ring ; then , where is an -module of finite type, and , where is an ideal of (1.4.1 and 1.5.1). Since is Noetherian, admits a finite system of generators (). By hypothesis, every section of over is null in each of the ; if () are sections of generating , there thus exists an integer independent of and such that (1.4.1), hence for every . One concludes that if , one has , and consequently the corresponding -Module (1.3.13) is null.
Corollary (9.3.5). Under the hypotheses of (9.3.4), there exists a closed subprescheme of , whose underlying space is the support of , and such that, if is the canonical injection, one has .
Proof. Let us first remark that the supports of and of are the same, for if , one also has , and one has on the other hand for every . One may therefore, by virtue of (9.3.4), suppose that ; one then takes for the closed subprescheme of defined by , and since is then an -Module, the conclusion is immediate.
9.4. Extension of quasi-coherent sheaves
(9.4.1) Let be a topological space, a sheaf of sets (resp. of groups, of rings) on , an open part of , the canonical injection, and a subsheaf of . Since is left exact, is a subsheaf of ; if one considers the canonical homomorphism (0, 3.5.3), we shall denote by the subsheaf of . It results immediately from the definitions that for every open of , is formed of the sections whose restriction to is a section of over . One thus has , and is the largest subsheaf of inducing on ; we shall say that is the canonical extension of the subsheaf of to a subsheaf of .
Proposition (9.4.2). Let be a prescheme, an open part of such that the canonical injection is a quasi-compact morphism (which will be verified for every if the underlying space is locally Noetherian (6.6.4, (i))). Then:
(i) For every quasi-coherent -Module , is a quasi-coherent -Module and one has .
(ii) For every quasi-coherent -Module and every quasi-coherent sub--Module , the canonical extension of (9.4.1) is a quasi-coherent sub--Module of .
Proof. If ( being the injection of the underlying spaces), one has by definition for every -Module , and moreover for every -Module , by reason of the definition of a prescheme induced on an open. (i) is therefore a particular case of (9.2.2, a)); for the same reason, is quasi-coherent, and since is the inverse image of under the homomorphism , (ii) results from (4.1.1).
It will be noted that the hypothesis that the morphism is quasi-compact is also verified when the open is quasi-compact and a scheme: indeed, is then a finite union of affine opens , and for every affine open of , is an affine open (5.5.6), hence quasi-compact.
Corollary (9.4.3). Let be a prescheme, a quasi-compact open of such that the injection morphism is quasi-compact. Suppose in addition that every quasi-coherent -Module is the inductive limit of its quasi-coherent sub--Modules of finite type (which holds when is an affine scheme). Let then be a quasi-coherent -Module and a quasi-coherent sub--Module of finite type of . There then exists a quasi-coherent sub--Module of finite type of such that .
Proof. Indeed, one has , and is quasi-coherent by (9.4.2), hence the inductive limit of its quasi-coherent sub--Modules of finite type . Consequently is the inductive limit of the , hence equal to one of the since it is of finite type (0, 5.2.3).
Remark (9.4.4). Suppose that for every affine open the injection morphism is quasi-compact. Then if the conclusion of (9.4.3) holds for every affine open and every quasi-coherent sub--Module of finite type of , it results that is the inductive limit of its quasi-coherent sub--Modules of finite type. Indeed, for every affine open , one has , where is an -module, and since the latter is the inductive limit of its submodules of finite type, is the inductive limit of its quasi-coherent sub--Modules of finite type (1.3.9). Now, by hypothesis, each of these sub-Modules is induced on by a quasi-coherent sub--Module of finite type of . The finite sums of the are again quasi-coherent -Modules of finite type, for the question is local and the case where is affine has been treated in (1.3.10); it is clear then that is the inductive limit of these finite sums, whence our assertion.
Corollary (9.4.5). Under the hypotheses of (9.4.3), for every quasi-coherent -Module of finite type , there exists a quasi-coherent -Module of finite type such that .
Proof. Since is quasi-coherent (9.4.2) and , it suffices to apply (9.4.3) to .
Lemma (9.4.6). Let be a prescheme, a well-ordered set, a covering of by affine opens, an open of ; for every , put . Suppose that: For every , is quasi-compact; The immersion morphism is quasi-compact. Then, for every quasi-coherent -Module and every quasi-coherent sub--Module of finite type of , there exists a quasi-coherent sub--Module of finite type of such that .
Proof. Put ; one is going to define by transfinite recursion a family , where is a quasi-coherent sub--Module of finite type of , such that for and . The unique sub--Module of such that for every (0, 3.3.1) will answer the question. Suppose then the defined and having the preceding properties for ; if has no predecessor, one will take for the unique sub--Module of such that for every , which is licit since the with then form a covering of . If on the contrary , one has , and it will suffice to define a quasi-coherent sub--Module of finite type of such that one will then take for the sub--Module of such that and (0, 3.3.1). Now, since is affine, the existence of is assured by (9.4.3) as soon as one has proved that is quasi-compact; but is the union of and , which are both quasi-compact by virtue of the hypothesis.
Theorem (9.4.7). Let be a prescheme, an open of . Suppose one of the following conditions verified:
a) The space underlying is locally Noetherian.
b) is a quasi-compact scheme and a quasi-compact open.
For every quasi-coherent -Module and every quasi-coherent sub--Module of finite type of , there then exists a quasi-coherent sub--Module of finite type of such that .
Proof. Let be a covering of by affine opens, being supposed finite in case b); being endowed with a structure of well-ordered set, it suffices to verify that the conditions of the lemma (9.4.6) are satisfied. This is evident in hypothesis a), the spaces being Noetherian. In hypothesis b), the are affine (5.5.6), hence quasi-compact, and since is finite, is quasi-compact. Whence the theorem.
Corollary (9.4.8). Under the hypotheses of (9.4.7), for every quasi-coherent -Module of finite type , there exists a quasi-coherent -Module of finite type such that .
Proof. It suffices to apply (9.4.7) to , which is quasi-coherent (9.4.2) and such that .
Corollary (9.4.9). Let be a prescheme whose underlying space is locally Noetherian, or a quasi-compact scheme. Then every quasi-coherent -Module is the inductive limit of its quasi-coherent sub--Modules of finite type.
Proof. This results from (9.4.7) and from the remark (9.4.4).
Corollary (9.4.10). Under the hypotheses of (9.4.9), if a quasi-coherent -Module is such that every quasi-coherent sub--Module of finite type of is generated by its sections over , then is generated by its sections over .
Proof. Indeed, let be an affine open neighborhood of a point , and let be a section of over ; the sub--Module of generated by is quasi-coherent and of finite type, hence there exists a quasi-coherent sub--Module of finite type of such that (9.4.7). By hypothesis, there are therefore a finite number of sections of over and sections of over a neighborhood of such that , which proves the corollary.
9.5. Closed image of a prescheme; closure of a subprescheme
Proposition (9.5.1). Let be a morphism of preschemes such that is a quasi-coherent -Module (which holds if is quasi-compact and if moreover is separated or locally Noetherian (9.2.2)). Then there exists a smallest subprescheme of such that the canonical injection majorizes (or, what amounts to the same (4.4.1), such that the subprescheme of is identical to ).
More precisely:
Corollary (9.5.2). Under the conditions of (9.5.1), let and let be the (quasi-coherent) kernel of the homomorphism . Then the closed subprescheme of defined by verifies the conditions of (9.5.1).
Proof. Since the functor is exact, the canonical factorization gives (0, 3.5.4.3) a factorization ; as for every , is a local homomorphism, the same is true of ; if one denotes by the continuous map considered as a map of into , by the restriction , one sees thus that is a morphism of preschemes (2.2.1) such that . If now is a second closed subprescheme of , defined by a quasi-coherent sheaf of ideals of and such that the injection majorizes , one must first have , hence since is closed. Moreover, for every , must factor as , which by definition entails , and consequently is a closed subprescheme of (4.1.10).
Definition (9.5.3). When there exists a smallest closed subprescheme of such that the canonical injection majorizes , one says that is the closed image prescheme of under the morphism .
Proposition (9.5.4). If is a quasi-coherent -Module, the space underlying the closed image of under is the closure in .
Proof. Since the support of is contained in , one has (with the notations of (9.5.2)) for , hence the support of is contained in . Moreover, this support is closed and contains : indeed, if , the unit element of the ring is not null, being the germ at of the section since it is the image under of the unit element of , the latter does not belong to , hence ; this completes the demonstration.
Proposition (9.5.5) (transitivity of closed images). Let and be two morphisms of preschemes; suppose that the closed image of under exists, and that, if is the restriction of to , the closed image of under exists. Then the closed image of under exists and is equal to .
Proof. It suffices (9.5.1) to show that is the smallest closed subprescheme of such that the closed subprescheme of (equal to by (4.4.2)) is equal to ; it amounts to the same to say that is the smallest closed subprescheme of such that is majorized by the injection (4.4.1). Now, by virtue of the existence of the closed image , every having this property is such that majorizes , which is equivalent to saying that , denoting by the injection . By definition of , one concludes well that is the smallest closed subprescheme of verifying the preceding condition.
Corollary (9.5.6). Let be an -morphism such that is the closed image of under . Let be an -scheme; if two -morphisms , of into are such that , then .
Proof. Let ; since the diagonal is a closed subprescheme of , is a closed subprescheme of (4.4.1). Put ; one then has by definition of the product , hence ; since , one has , hence . One concludes (4.4.1) that the canonical injection majorizes , hence by hypothesis; consequently (4.4.1), factors as , where is a morphism , which entails .
Remark (9.5.7). If and are -schemes, the proposition (9.5.6) signifies that when is the closed image of under , is an epimorphism in the category of -schemes (T, 1.1). We shall prove in chap. V that, conversely, if the closed image of under exists and if is an epimorphism of -schemes, then one necessarily has .
Proposition (9.5.8). Suppose the hypotheses of (9.5.1) verified, and let be the closed image of under . For every open of , let be the restriction of ; then the closed image of under in exists and is equal to the prescheme induced by on the open of (in other words, to the subprescheme of (4.4.3)).
Proof. Put ; since the direct image of under is none other than the restriction of to , it is clear that the kernel of the homomorphism is the restriction of to , whence at once the proposition.
It will be noted that this result is interpreted by saying that the formation of the closed image commutes with an extension of the base prescheme that is an open immersion. We shall see in chap. IV that the same holds for an extension that is a flat morphism, provided that is separated and quasi-compact.
Proposition (9.5.9). Let be a morphism such that the closed image of under exists.
(i) If is reduced, so is .
(ii) If one supposes the hypotheses of (9.5.1) verified and if is irreducible (resp. integral), so is .
Proof. By hypothesis, the morphism factors as , where is the canonical injection. Since is reduced, factors as , where is the canonical injection (5.2.2), and it then results from the definition of that . If moreover the conditions of (9.5.1) are fulfilled, it results from (9.5.4) that is dense in ; if is irreducible, so therefore is (0, 2.1.5). The assertion relative to integral preschemes results from the conjunction of the two others.
Proposition (9.5.10). Let be a subprescheme of a prescheme , such that the canonical injection is a quasi-compact morphism. There then exists a smallest closed subprescheme of majorizing ; its underlying space is the closure of that of ; the latter is open in its closure, and the prescheme is induced on this open by .
Proof. It suffices to apply (9.5.1) to the injection , which is separated (5.5.1) and quasi-compact by hypothesis; (9.5.1) thus proves the existence of and (9.5.4) shows that its underlying space is the closure of in ; since is locally closed in , it is open in , and the last assertion comes from (9.5.8) applied to an open of such that is closed in .
With these notations, if the injection is quasi-compact, and if is the quasi-coherent sheaf of ideals of defining the closed subprescheme of , it results from (9.5.1) that the quasi-coherent sheaf of ideals of defining is the canonical extension (9.4.1) of , for it is evidently the largest quasi-coherent subsheaf of ideals of inducing on .
Corollary (9.5.11). Under the hypotheses of (9.5.10), every section of over an open of that is null in is null.
Proof. By virtue of (9.5.8), one may reduce to the case where . If one takes account of the fact that the sections of over correspond canonically to the -sections of (3.3.15) and of the fact that the latter is separated over , the corollary appears as a particular case of (9.5.6).
When there exists a smallest closed subprescheme of majorizing a subprescheme of , one says that is the closure of in , when no confusion results.
9.6. Quasi-coherent sheaves of algebras; change of the structure sheaf
Proposition (9.6.1). Let be a prescheme, a quasi-coherent -Algebra (0, 5.1.3). For a -Module to be quasi-coherent (on the ringed space ), it is necessary and sufficient that be a quasi-coherent -Module.
Proof. Since the question is local, one may suppose affine with ring , and then , where is an -algebra (1.4.3). If is quasi-coherent on the ringed space , one may also suppose that is the cokernel of a -homomorphism ; since this homomorphism is also an -homomorphism of -Modules, and since and are quasi-coherent -Modules (1.3.9, (ii)), is also a quasi-coherent -Module (1.3.9, (i)).
Inversely, if is a quasi-coherent -Module, one has , where is a -module (1.4.3); is isomorphic to a cokernel of a -homomorphism , hence is a -Module isomorphic to the cokernel of the corresponding homomorphism (1.3.13), which completes the demonstration.
In particular, if and are two quasi-coherent -Modules, is a quasi-coherent -Module; the same is true of when one supposes in addition that admits a finite presentation (1.3.13).
(9.6.2) Given a prescheme , we shall say that a quasi-coherent -Algebra is of finite type if for every , there exists an affine open neighborhood of such that is an algebra of finite type over . One then has , and for every , the -Algebra induced is of finite type, for it is isomorphic to , and is evidently an algebra of finite type over . Since the form a base of the topology of , one concludes that if is a quasi-coherent -Algebra of finite type, then for every open of , is a quasi-coherent -Algebra of finite type.
Proposition (9.6.3). Let be a locally Noetherian prescheme. Then every quasi-coherent -Algebra of finite type is a coherent sheaf of rings (0, 5.3.7).
Proof. One may again limit oneself to the case where is an affine scheme with Noetherian ring , and where , being an -algebra of finite type; is then a Noetherian ring. This being so, it is necessary to prove that the kernel of a -homomorphism is a -Module of finite type; now, it is isomorphic (as a -Module) to , where is the kernel of the corresponding homomorphism of -modules (1.3.13). Since is Noetherian, the submodule of is a -module of finite type, hence there exists a homomorphism of image ; the sequence being exact, the same is true of the corresponding sequence (1.3.5), and since is the image of (1.3.9, (i)), the proposition is proved.
Corollary (9.6.4). Under the hypotheses of (9.6.3), for a -Module to be coherent, it is necessary and sufficient that it be a quasi-coherent -Module and a -Module of finite type. If this is so, and if is a sub--Module or a quotient -Module of , then for to be a coherent -Module, it is necessary and sufficient that be a quasi-coherent -Module.
Proof. Taking account of (9.6.1), the conditions on are evidently necessary; let us show that they are sufficient. One may limit oneself to the case where is affine with Noetherian ring , , where is an -algebra of finite type, , where is a -module, and where there exists a surjective -homomorphism . One then has a corresponding exact sequence , hence is a -module of finite type; moreover, the kernel of the homomorphism is then a -module of finite type, since is Noetherian. One concludes (1.3.13) that is the cokernel of a -homomorphism and is therefore coherent since is a coherent sheaf of rings (0, 5.3.4). The same reasoning shows that a quasi-coherent sub--Module (resp. a quotient -Module) of is of finite type, whence the second part of the corollary.
Proposition (9.6.5). Let be a quasi-compact scheme or a prescheme whose underlying space is Noetherian. For every quasi-coherent -Algebra of finite type , there exists a quasi-coherent sub--Module of finite type of such that generates (0, 4.1.4) the -Algebra .
Proof. Indeed, there exists by hypothesis a finite covering of formed of affine opens such that is an algebra of finite type over ; let be a sub--module of finite type of generating the -algebra ; by virtue of (9.4.7), there exists a sub--Module of , quasi-coherent and of finite type, such that . It is clear that the sum of the answers the question.
Proposition (9.6.6). Let be a prescheme whose underlying space is locally Noetherian, or a quasi-compact scheme. Then every quasi-coherent -Algebra is the inductive limit of its quasi-coherent sub--Algebras of finite type.
Proof. Indeed, it results from (9.4.9) that is the inductive limit (as an -Module) of its quasi-coherent sub--Modules of finite type; these last generate quasi-coherent sub--Algebras of finite type of (1.3.14), of which is a fortiori the inductive limit.