text
stringlengths
128
2.05k
(1) [MATH] is bounded by [MATH] when viewed as a morphism [MATH] in [MATH] , and (2) for each bounded subset [MATH] , the restriction
[MATH] is bounded when viewed as a morphism [MATH] of [MATH] -filtered modules in [MATH] Proof. If [MATH] is [MATH] -controlled then for any subset [MATH] we have [MATH] . So [MATH] is bounded by [MATH] Now for a given subset [MATH] , let us define [MATH] Then [MATH]
verifying that [MATH] is bounded by [MATH] 3.2. Properties of fibred objects 3.2.1 Definition An [MATH] -filtered module [MATH] is called
lean or [MATH] lean if there is a number [MATH] and a monotone function [MATH] so that [EQUATION] for any subset [MATH] of [MATH]
split or [MATH] split if there is a number [MATH] and a monotone function [MATH] so that [EQUATION] for each pair of subsets [MATH] and [MATH] of [MATH]
lean/split or [MATH] lean/split if there is a number [MATH] and a monotone function [MATH] so that the [MATH] -filtered module [MATH] is [MATH] -lean, while
the [MATH] -filtered module [MATH] is [MATH] -split, insular or [MATH] insular if there is a number [MATH] and a monotone function
[MATH] so that [EQUATION] for each pair of subsets [MATH] and [MATH] of [MATH] 3.2.2 Proposition Suppose [MATH] is an [MATH] -filtered [MATH] -module.
(1) If [MATH] is [MATH] -lean then the corresponding [MATH] -filtered module [MATH] defined by assigning [MATH] is [MATH] -lean.
(2) Similarly, if [MATH] is [MATH] -insular then [MATH] is [MATH] -insular. (3) If [MATH] is [MATH] -lean then it is [MATH] -split and, further, [MATH] -lean/split.
(4) An [MATH] -filtered module [MATH] which is lean/split and insular can be thought of as an object [MATH] of [MATH] Proof. (1) Since [MATH] , we have
[EQUATION] (2) [MATH] (3) The split property follows directly from definitions, and so the lean/split property follows in view of part (1).
(4) follows from (2). 3.2.3 Definition There are two subcategories nested in [MATH] [MATH] is the full subcategory of [MATH] on objects [MATH] that are lean/split and insular,
[MATH] is the full subcategory of [MATH] on objects [MATH] such that [MATH] is a finitely generated submodule whenever [MATH] is bounded. Equivalently, the subcategory [MATH] is full on objects [MATH] such that all [MATH] -filtered modules [MATH] associated to bounded subsets [MATH] are locally finitely generated.
Clearly, [MATH] is a generalization of the bounded category [MATH] : if [MATH] then [MATH] is precisely [MATH] . On the other hand, if [MATH] then [MATH] is the full subcategory of [MATH] on locally finitely generated objects.
We proceed to define appropriate exact structures in these categories. 3.2.4 Definition Let the admissible monomorphisms in [MATH] be the boundedly bicontrolled homomorphisms [MATH] such that the module homomorphism [MATH] is a monomorphism. Let the admissible epimorphisms be the boundedly bicontrolled homomorphisms [M...
exact sequences consists of the sequences [EQUATION] where [MATH] is an admissible monomorphism, [MATH] is an admissible epimorphism, and [MATH]
3.2.5 Theorem [MATH] is a Quillen exact category. Proof. We will verify the axioms for exact structures due to Quillen with some simplifications due to B. Keller
, cf. section 2 of It follows from Lemma 3.1.7 that the collections of admissible monomorphisms and admissible epimorphisms are closed under composition and that any short exact sequence isomorphic to some sequence in
[MATH] is also in [MATH] Now suppose we are given an exact sequence [MATH] in [MATH] and a morphism [MATH] in [MATH] . Let [MATH] be some filtration data for [MATH] as a boundedly controlled epi and let [MATH] be some contol data for [MATH] as a boundedly controlled map. There is a base change diagram
[EQUATION] where [MATH] is the kernel of the epi [MATH] and [MATH] [MATH] The [MATH] -filtration on [MATH] is the standard filtration as a subobject of the product [MATH] The induced map [MATH] has the same kernel as
[MATH] and is bounded by [MATH] In fact, [MATH] so [MATH] and [EQUATION] This shows that [MATH] is boundedly bicontrolled with filtration data [MATH] Therefore, the class of admissible epimorphisms is closed under base change by arbitrary morphisms in [MATH] Cobase changes by admissible monomorphisms are similar.
3.2.6 Proposition The admissible monomorphisms are precisely the morphisms isomorphic in [MATH] to the filtration-wise monomorphisms and the admissible epimorphisms are those morphisms isomorphic to the filtration-wise epimorphisms. In other words, the exact structure [MATH] in [MATH] consists of sequences isomorphic t...
[EQUATION] which possess filtration-wise restrictions [EQUATION] for all subsets [MATH] , and each [MATH] is an exact sequence of [MATH] -modules.
Proof. Each of the sequences [MATH] in the statement is an exact sequence in [MATH] because the restriction [MATH] is monic and [MATH] is epic, therefore [MATH] and [MATH] are both bicontrolled of filtration [MATH]
Suppose [MATH] is a sequence isomorphic to such [MATH] , so there is a commutative diagram [EQUATION] Then [MATH] and [MATH] are compositions of two isomorphisms (which are clearly boundedly bicontrolled) which are either preceded by a boundedly bicontrolled monic or followed by a boundedly bicontrolled epi. By part (1...
Now suppose [MATH] is an exact sequence in [MATH] . Let [MATH] and [MATH] , then we obtain a commutative diagram [EQUATION] where the vertical maps are the canonical isomorphisms. By the construction of kernels and cokernels in the proof of Theorem 3.1.6 , there are exact sequences
[MATH] for all subsets [MATH] 3.2.7 Proposition In the exact category [MATH] (1) the lean/split objects are closed under extensions,
(2) the insular objects are closed under extensions. Suppose [MATH] is an exact sequence in [MATH] (3) If the object [MATH] is lean/split then [MATH] is lean/split.
(4) If [MATH] is insular then [MATH] is insular. (5) Suppose [MATH] is insular then [MATH] is insular if [MATH] is lean/split. Proof.
All parts are proved by adapting the proofs of Lemmas 2.2.3 and 2.2.4 To illustrate, suppose that in the exact sequence, as given in the statement, [MATH] is common filtration data for [MATH] and [MATH] and both [MATH] and [MATH] are
[MATH] -lean/split. For the first statement of part (2), notice that [MATH] is [MATH] -lean by part (1) of Lemma 2.2.3 , so we need to verify that split objects are closed under extensions. Consider two subsets [MATH] and [MATH] of [MATH] . Then
[EQUATION] Therefore [EQUATION] showing that [MATH] is [MATH] -lean/split. 3.2.8 Theorem [MATH] is closed under extensions in [MATH] . In turn, [MATH] is closed under extensions in [MATH] . Therefore, [MATH] is an exact category, and the inclusion
[MATH] is an exact embedding. Proof. The first statement follows from parts (2) and (3) of Proposition 3.2.7 Suppose [MATH] is an isomorphism with [MATH] and [MATH] is locally finitely generated, then [MATH] is a finitely generated submodule of [MATH] for any bounded subset [MATH] since [MATH] is Noetherian.
If [MATH] is an exact sequence in [MATH] [MATH] and [MATH] are locally finitely generated, and [MATH] is common filtration data for [MATH] and [MATH] , then [MATH] is a finitely generated submodule of [MATH] for any bounded subset [MATH] The kernel of the restriction of [MATH] to [MATH] is a finitely generated submodul...
3.2.9 Remark (1) There is certainly an exact embedding [MATH] which is given by the identity on objects. Because of the relaxation of the control conditions on homomorphisms, the morphism sets in the image of [MATH] are in general properly smaller than in [MATH] However [MATH] is also proper on objects. For example, th...
where the diameters of [MATH] and [MATH] are uniformly bounded from above. This is different from the weaker condition in [MATH]
(2) While there is no functor between the categories [MATH] and [MATH] , there is a “forgetful” function associating to objects of [MATH] some objects of [MATH] . It is defined by [MATH] with the [MATH] -filtration given by [MATH]
This relationship can be made much more fruitful if the nature of the objects in [MATH] is shifted to be functors from the category of bounded subsets of [MATH] to subobjects of [MATH] leading to a category that can be thought of as [MATH] . It turns out that on this level there is a well-defined exact functor [MATH] ....
3.3. Fibrewise restriction We begin to prepare for the development of fibred localization exact sequences and fibred bounded excision theorems. The model for localization and fibration theorems in controlled [MATH] -theory , sections 3 and 4] can be implemented here as well.
There are two complementary ways to introduce support in [MATH] (1) Let [MATH] be the full subcategory of [MATH] on objects [MATH] supported near [MATH] when viewed as objects [MATH] in [MATH] In other words, [MATH] is an object of [MATH] if
[MATH] for some number [MATH] (2) Let [MATH] be the full subcategory of [MATH] on objects [MATH] such that [EQUATION] for some number [MATH] and an order preserving function [MATH]
The first version of support is a straightforward generalization of support for geometric modules that was exploited in In this paper we are more interested in the latter, fibrewise version (2) of support.
3.3.1 Proposition Suppose [MATH] is a [MATH] -lean/split object of [MATH] The following are equivalent statements. (1) [MATH] is an object of [MATH]
(2) There is a number [MATH] and an order preserving function [MATH] such that [EQUATION] for all bounded subsets [MATH] (3) There is a number [MATH] and a monotone function [MATH]
such that [EQUATION] for all [MATH] Proof. [MATH] : If [MATH] satisfies (2) then [MATH] It suffices to define [MATH] such that [MATH] Since [MATH] and [MATH] is order preserving, one can take
[MATH] In the opposite direction, given a bounded subset [MATH] [EQUATION] when [MATH] [MATH] : If [MATH] is in [MATH] then [MATH] so
[EQUATION] If [MATH] is [MATH] -insular then [EQUATION] for [MATH] In the opposite direction, we have [EQUATION] for an object [MATH] of [MATH] satisfying (3).
3.3.2 Definition Serre subcategory of an exact category is a full subcategory which is closed under exact extensions and closed under passage to admissible subobjects and admissible quotients.
Note that this property is relative to the choice of exact structure. 3.3.3 Proposition [MATH] is a Serre subcategory of [MATH] Proof.
First we show closure under exact extensions. Let [MATH] be an exact sequence in [MATH] , let [MATH] be common set of filtration data for [MATH] and [MATH] , and assume all objects be [MATH] -lean/split. We assume that [MATH] and [MATH] are objects of [MATH] , so there is a number [MATH] and a monotone function [MATH] ...
[MATH] and [MATH] for some choice of a base point [MATH] in [MATH] Therefore [EQUATION] In particular, the image [MATH] with the standard filtration [MATH] is an object of [MATH] Now
[EQUATION] Let [MATH] viewed as a subobject of [MATH] with the standard filtration. Since [MATH] for any submodule [MATH] with [MATH] , we have
[EQUATION] so [MATH] is an object of [MATH] Suppose [MATH] is an admissible monomorphism in [MATH] , which is a boundedly bicontrolled monic with [MATH] [MATH] is [MATH] -lean/split, [MATH] is [MATH] -lean/split for some [MATH] , and
[MATH] is [MATH] -insular. If [MATH] is an object of [MATH] , according to Proposition 3.3.1 [EQUATION] for some number [MATH] , an order preserving function [MATH] , and all bounded subsets [MATH] Then
[EQUATION] using the fact that [MATH] is order preserving. Since [EQUATION] we have [EQUATION] Therefore [EQUATION] so [MATH] , which is [MATH] , is also an object of [MATH]
On the other hand, let [MATH] be an admissible quotient with [MATH] and suppose [MATH] is an object of [MATH] so that there is a number [MATH] and a monotone function [MATH] such that
[EQUATION] This implies that [EQUATION] so [MATH] is also in [MATH] 3.4. Fibrewise gradings The gradings from Definition 2.4.1 can be generalized to gradings of objects from [MATH]
3.4.1 Definition Given an object [MATH] of [MATH] , a grading of [MATH] is a covariant functor [MATH] with the following properties:
(1) if [MATH] is given the standard filtration, it is an object of [MATH] (2) there is an enlargement data [MATH] such that [EQUATION]
for all subsets [MATH] of [MATH] 3.4.2 Remark If [MATH] then [MATH] is an object of [MATH] We are concerned with localizations to a specific type of subspaces of [MATH] This makes the following partial gradings sufficient and easier to work with.
3.4.3 Definition Let [MATH] be the set of all monotone functions [MATH] Let [MATH] be the subcategory of [MATH] consisting of all subsets of the form [MATH] for some choices of a subset [MATH] , a number [MATH] , and a function [MATH]
Given an object [MATH] of [MATH] , a [MATH] grading of [MATH] is a functor [MATH] with the following properties: (1) the submodule [MATH] with the standard filtration is an object of [MATH]
(2) there is an enlargement data [MATH] such that [EQUATION] for all subsets in [MATH] Since [MATH] for general subsets [MATH] , the third, largest submodule is independent of the choice of [MATH] [MATH]
We say that an object [MATH] of [MATH] is [MATH] graded if there exists a [MATH] -grading of [MATH] , but the grading itself is not specified, and define [MATH] as the full subcategory of [MATH] on [MATH] -graded filtered modules.
3.4.4 Proposition The [MATH] -graded objects in [MATH] are closed under isomorphisms. The subcategory [MATH] is closed under extensions in [MATH] Therefore, [MATH] is an exact subcategory of [MATH]
Proof. Suppose [MATH] is an isomorphism between a [MATH] -graded module [MATH] and [MATH] in [MATH] . If [MATH] and [MATH] is a [MATH] -grading for [MATH] then [MATH] . The rest of the argument closely follows the proof of Proposition 2.4.4 . We want to spell out one detail for future reference. Let
[EQUATION] be an exact sequence in [MATH] . Suppose [MATH] and [MATH] for the same set of bicontrol bound data and suppose that [MATH] and [MATH] are graded modules in [MATH] with the associated functors [MATH] and [MATH] The assignment
[EQUATION] gives a grading of [MATH] as an object of [MATH] As with the category [MATH] , the advantage of working with [MATH] as opposed to [MATH] is that we are able to localize to the grading subobjects associated to subsets from the family [MATH]
3.4.5 Lemma Let [MATH] be a submodule of a [MATH] -filtered module [MATH] in [MATH] which is lean/split with respect to the standard filtration. Then [MATH] is a [MATH] -grading of [MATH]
We will call this induced [MATH] -grading of [MATH] the standard [MATH] -grading of the submodule. Proof. The proof is reduced to checking that [MATH] is an object of [MATH] for each subset [MATH] Suppose [MATH] is the inclusion and [MATH] is the quotient of [MATH] Since [MATH] is insular by part (4) of Proposition 3.2...
is lean/split by part (5) of 3.2.7 and is insular as a submodule of insular [MATH] Locally finite generation of [MATH] follows from the same property of [MATH]
3.4.6 Proposition Suppose [MATH] is a boundedly bicontrolled epimorphism in [MATH] and suppose [MATH] is the kernel of [MATH] in [MATH] . If [MATH] is [MATH] -graded and [MATH] is lean/split with respect to the standard [MATH] -filtration then both [MATH] and [MATH] are [MATH] -graded.
Proof. The [MATH] -grading for [MATH] is given by [MATH] , where [MATH] is a chosen set of filtration data for [MATH] The argument for Lemma 3.4.5 applies directly to show that [MATH] is indeed a grading of [MATH] , and the assignment [MATH] gives a [MATH] -grading of [MATH]
This allows to characterize admissible monomorphisms in [MATH] 3.4.7 Proposition The inclusion of a subobject [MATH] in [MATH] is an admissible monomorphism if and only if [MATH] is lean/split.
Proof. Let [MATH] be a cokernel of [MATH] in [MATH] . As verified in Lemma 3.4.5 [MATH] is lean/split and insular and is, in fact, a cokernel of [MATH] in [MATH] From Proposition 3.4.6 [MATH] is graded, so it is also a cokernel of [MATH] in [MATH]
In the case [MATH] and [MATH] is a non-positive function, we will use notation [MATH] for the subset [MATH] of [MATH] . The following is a direct analogue of Corollary 2.4.8 with exactly the same proof.
3.4.8 Corollary Given an object [MATH] in [MATH] and a subset [MATH] from the family [MATH] , there is a set of enlargement data [MATH] and an admissible subobject [MATH] in [MATH] with the property that [MATH] If [MATH] is [MATH] -lean/split then the quotient [MATH] of the inclusion has the property that [MATH]
Now we can summarize the preceding results. 3.4.9 Theorem Given a graded object [MATH] in [MATH] and a subset [MATH] from the family [MATH] we assume that [MATH] is [MATH] -split and [MATH] -insular and is graded by [MATH] The submodule [MATH] has the following properties:
(1) [MATH] is graded by [MATH] (2) [MATH] for some fixed enlargement data [MATH] (3) if [MATH] is the quotient of the inclusion [MATH] and [MATH] is [MATH] -lean/split, then [MATH] is supported on [MATH]
(4) [MATH] Proof. Property (1) follows from Lemma 3.4.5 Properties (2) and (3) follow from Corollary 3.4.8 (4) follows from the fact that a [MATH] -insular filtered module is [MATH] -separated, in the sense that for any pair of subsets [MATH] and [MATH] of [MATH] such that [MATH] we have
[MATH] so [MATH] Now [EQUATION] but [MATH] , thus [MATH] 3.5. Localization fibration sequence We will use the localization theorem of Schlichting
for Serre subcategories of exact categories. These techniques require the Serre subcategory to satisfy some additional assumptions that we verify next.
3.5.1 Definition A class of morphisms [MATH] in an additive category [MATH] admits a calculus of right fractions if (1) the identity of each object is in [MATH]
(2) [MATH] is closed under composition, (3) each diagram [MATH] with [MATH] can be completed to a commutative square [EQUATION] with [MATH] , and
(4) if [MATH] is a morphism in [MATH] and [MATH] such that [MATH] then there exists [MATH] such that [MATH] In this case there is a construction of the localization
[MATH] which has the same objects as [MATH] . The morphism sets [MATH] in [MATH] consist of equivalence classes of diagrams [EQUATION]
with the equivalence relation [MATH] if there is a map [MATH] so that [MATH] and [MATH] . Let [MATH] denote the equivalence class of [MATH] . The composition of morphisms in [MATH] is defined by
[MATH] where [MATH] and [MATH] fit in the commutative square [EQUATION] from axiom 3. 3.5.2 Proposition The localization [MATH] is a category. The morphisms of the form [MATH] where
[MATH] are isomorphisms in [MATH] . The rule [MATH] gives a functor [MATH] which is universal among the functors making the morphisms [MATH]
invertible. Proof. The proofs of these facts can be found in Chapter I of . The inverse of [MATH] is [MATH] From Proposition 3.3.3 , given a subset [MATH] of [MATH] , the category [MATH] is a Serre subcategory of [MATH]
3.5.3 Proposition The restriction to [MATH] -gradings in [MATH] gives a full exact subcategory [MATH] which is a Serre subcategory of [MATH]
Proof. One only needs to observe that if the modules at the ends of the exact sequence in the proof of Proposition 3.4.4 have [MATH] -gradings in [MATH] then the displayed grading of [MATH] shows that [MATH] is an object of [MATH]
The following shorthand notation is convenient when the choice of [MATH] is clear. 3.5.4 Notation The category [MATH] is the exact subcategory of [MATH] -graded objects in [MATH] When the choice of the subset [MATH] is understood, we will use notation [MATH] for the Serre subcategory
[MATH] of [MATH] 3.5.5 Definition Define the class of weak equivalences [MATH] in [MATH] to consist of all finite compositions of admissible monomorphisms with cokernels in [MATH] and admissible epimorphisms with kernels in [MATH]
We need the class [MATH] to admits calculus of right fractions. This follows from 17 , Lemma 1.13] as soon as we prove the following fact.
A Serre subcategory [MATH] of an exact category [MATH] is right filtering if each morphism [MATH] in [MATH] , where [MATH] is an object of [MATH] , factors through an admissible epimorphism [MATH] , where [MATH] is in [MATH]
3.5.6 Lemma The subcategory [MATH] of [MATH] is right filtering. Proof. For a morphism [MATH] in [MATH] with [MATH] in [MATH] , we assume that both [MATH] and [MATH] are
[MATH] -lean/split and [MATH] -insular. Suppose [MATH] is bounded by [MATH] and let [MATH] and [MATH] be a monotone function such that
[EQUATION] Now for any characteristic set of data [MATH] for the grading [MATH] and any subset [MATH] we have [EQUATION] By part (3) of Theorem 3.4.9