1. Introduction
In classical homological algebra, homological dimensions are important invariants, and every homological dimension is defined in terms of some certain subcategory. For example, one can define projective dimension in terms of the subcategory consisting of projective objects, and define injective dimension in terms of the subcategory of consisiting injective objects in any abelian category. Resolving subcategories play important roles in approximation theory (e.g., [
1,
2]). As an important example of resolving subcategories, Auslander and Buchweitz [
3] studied the approximation theory of the subcategory consisting of maximal Cohen-Macaulay modules over an Artin algebra. Zhu [
4] studied the resolution dimension with respect to a resolving subcategory in an abelian category, and Huang [
5] introduced relative preresolving subcategories in an abelian category and defined homological dimensions relative to these subcategories. In [
6,
7], Ma, Zhao, and Huang investigated homological dimensions relative to (pre)resolving subcategories in triangulated categories with a proper class of triangles. For more references on resolution and homological dimension, see [
8,
9,
10,
11], for example.
Exact and triangulated categories are two important structures in category theory. In [
12], Nakaoka and Palu introduced the notion of extriangulated categories as a simultaneous generalization of exact categories and extension-closed subcategories of triangulated categories. After that, the study of extriangulated categories has become an active topic, and up to now, many results on exact categories and triangulated categories can be unified in the same framework, e.g., see [
8,
12,
13,
14,
15,
16]. Recently, Hu, Zhang, Zhou [
13] studied a relative homological algebra in an extriangulated category
which parallels the relative homological algebra in triangulated categories and exact categories. By specifying a class of
-triangles, which is called a proper class
of
-triangles, the authors introduced
-projective,
-injective,
-
projective and
-
injective dimensions, and discussed their properties. In abelian categories, the subcategory consisting of Gorenstein projective objects is a resolving subcategory, thus the aim of this paper is to introduce a notion of resolving subcategories in extriangulated categories, which regards the subcategory consisting of
-
projective objects as a special example. After this, we devote to further studying homological dimensions relative to a resolving subcategory in extriangulated categories which recovers lots of known results in abelian and triangulated categories, and is new in exact categories. The paper is organized as follows.
In
Section 2, we give some terminology and some preliminary results. In particular, we introduce the notion of resolving subcategories in extriangulated categories with a proper class of
-triangles.
In
Section 3, we introduce the notion of
-resolution dimension of objects relative to a resolving subcategory
, and some homological properties of resolution dimension are obtained. In particular, we obtain Auslander-Buchweitz approximation
-triangles (see Proposition 4) for objects with finite
-resolution dimension. Our main result is the following.
Theorem 1. Let be a resolving subcategory of an extriangulated category with a proper class of -triangles ξ, and a ξ-cogenerator of with . Let be the full subcategory of whose objects have finite -resolution dimension, and let (resp. ) be an nth -syzygy (resp. syzygy) of M. Assume that one of the following conditions satisfies:
- (a)
is closed under cocones of ξ-deflations.
- (b)
is closed under direct summands.
For any , if , then the following statements are equivalent:
- (1)
.
- (2)
for all .
- (3)
for all .
- (4)
for all and all .
- (5)
for all and all .
- (6)
M admits a right -approximation , where φ is a ξ-defaltion, such that there is an -triangle satisfying .
- (7)
There are two -trianglesandin ξ such that and are in and , .
As applications, in
Section 4, we will further study objects with a finite resolution dimension with respect to a resolving subcategory
. We construct adjoint pairs for two kinds of inclusion functors (see Theorems 3 and 4). Given a resolving subcategory
of
, we construct a new resolving subcategory
with a
-cogenerator
(see Theorem 5), which generalizes the Gorenstein projective subcategory
given by Hu, Zhang, and Zhou [
13] of [Definition 4.8].
Throughout this paper, all subcategories are full, additive and closed under isomorphisms.
2. Preliminaries
We first recall some notions and some needed properties of extriangulated categories from [
12].
Let
be an additive category and
a biadditive functor, where
is the category of abelian groups. Let
. An element
is called an
-extension. Two sequences of morphisms
are said to be
equivalent if there exists an isomorphism
such that
and
. We denote by
the equivalence class of
. In particular, we write
.
For an
-extension
, we briefly write
For two -extensions and , a morphism from to is a pair of morphisms with and such that .
Definition 1. ([12] of [Definition 2.9], [17]) Let be a correspondence which associates an equivalence class to each -extension . The correspondence is called a realization
of provided that it satisfies the following condition. - (R)
Let and be any pair of -extensions with Then for any morphism , there exists such that the following diagramcommutes.
Let be a realization of . If for some -extension , then we say that the sequence realizes δ; and in the condition (R), we say that the triple realizes the morphism .
For any two equivalence classes
and
, we define
Definition 2. ([12] of [Definition 2.10], [17]) A realization of is called additive
if it satisfies the following conditions. - (1)
For any , the split -extension satisfies .
- (2)
For any pair of -extensions and , we have .
Definition 3. ([12] of [Definition 2.12], [17]) The triple is called an externally triangulated
(or extriangulated
for short) category if it satisfies the following conditions. Remark 1. Please note that both exact categories and triangulated categories are extriangulated categories (see [12] of [Proposition 3.22]) and extension closed subcategories of extriangulated categories are again extriangulated (see [12] of [Remark 2.18]). Moreover, there exist extriangulated categories which are neither exact categories nor triangulated categories (see [12] of [Proposition 3.30] and [13] of [Remark 3.3]). We will use the following terminology.
Definition 4. ([12] of [Definitions 2.15 and 2.19], [17]) Let be an extriangulated category. - 1.
A sequence is called a conflation if it realizes some -extension . In this case, x is called an inflation and y is called a deflation.
- 2.
If a conflation realizes , we call the pair an -triangle, and write it in the following way. We usually do not write this “δ" if it is not used in the argument.
- 3.
Let and be any pair of -triangles. If a triplet realizes , then we write it asand call a morphism of -triangles. If above are isomorphisms, then and are said to be isomorphic.
Remark 2. We can view the collection of all -triangles together with morphisms of -triangles as an additive category. Indeed,
- (i)
Let be a morphism from to , and let be a morphism from to . The composition is defined by .
The composition is well defined. In fact, assume that and define morphisms of -extensions, then and . Thus,that is, is a morphism of -extensions. - (ii)
For an -triangle , the identity morphism is
- (iii)
The associativity of the composition is inherited by the associativity of the composition in .
- (iv)
The -triangle is an initial and terminal object.
- (v)
For objects in a category, write and for the morphisms equipping the coproduct (if it exists), and and for the morphisms equipping the product (if it exists). Now since is additive, there is an isomorphism . Now fix objects and . There are isomorphismswhere and for all and . Now let and where . Then it is easy to check: thatis the coproduct of and ; and thatis the product of and ; and that the triple is the (unique) morphisminduced by the universal property.
The following condition is analogous to the weak idempotent completeness in exact categories (see [
12] of [Condition 5.8]).
Condition (WIC) Consider the following conditions.
- (a)
Let be any composable pair of morphisms. If is an inflation, then so is f.
- (b)
Let be any composable pair of morphisms. If is a deflation, then so is g.
Example 1. (1)
If is an exact category, then Condition (WIC) is equivalent to is weakly idempotent complete (see [18] of [Proposition 7.6]).(2) If is a triangulated category, then Condition (WIC) is automatically satisfied.
Lemma 1. (c.f. [12] of [Proposition 3.15], [17]) Assume that is an extriangulated category. - (1)
Let C be an object in , and let and be any pair of -triangles. Then there is a commutative diagram in which satisfies and - (2)
Let A be an object in , and let and be any pair of -triangles. Then there is a commutative diagram in which satisfies and .
The following definitions are quoted verbatim from [
13] of [Section 3]. A class of
-triangles
is
closed under base change if for any
-triangle
and any morphism
, then any
-triangle
belongs to
.
Dually, a class of
-triangles
is
closed under cobase change if for any
-triangle
and any morphism
, then any
-triangle
belongs to
.
A class of
-triangles
is called
saturated if in the situation of Lemma 1(1), whenever
and
belong to
, then the
-triangle
belongs to
.
An -triangle is called split if . It is easy to see that it is split if and only if x is section or y is retraction.
The full subcategory consisting of the split -triangles will be denoted by .
Definition 5. ([13] of [Definition 3.1], [17]) Let
be a class of
-triangles which is closed under isomorphisms. Then
is called a
proper class of
-triangles if the following conditions hold:
- (a)
is closed under finite coproducts and .
- (b)
is closed under base change and cobase change.
- (c)
is saturated.
A proper class is a class which is not a set in general.
Definition 6. ([13] of [Definition 4.1], [17]) An object
is called
ξ-projective if for any
-triangle
in
, the induced sequence of abelian groups
is exact. Dually, we have the definition of
ξ-injective objects.
We denote by (resp., ) the full subcategory of consisting of -projective (resp., -injective) objects. It follows from the definition that and are full, additive, closed under isomorphisms and direct summands.
An extriangulated category is said to have enough ξ-projectives (resp., enough ξ-injectives) provided that for each object A there exists an -triangle (resp., ) in with (resp., ).
The ξ-projective dimension - of is defined inductively. If , then define -. For a positive integer n, one writes - provided
- (a)
there is an -triangle with and -,
- (b)
there does not exist an -triangle with and -.
Of course we set -, if - for all .
Dually we can define the ξ-injective dimension - of an object .
Definition 7. ([13] of [Definition 4.4], [17]) A ξ-exact complex is a diagramin such that for each integer n, we have for some -trianglein ξ. In particular, by saying thatis ξ-exact, it means that there are -trianglesin ξ, and for each integer , we have for some -trianglein ξ. Definition 8. ([8] of [Definition 3.1], [17]) Let M be an object in . By a ξ-projective resolution of M we mean a symbol of the form where is a ξ-exact complex, where for all and where and for all . The notion of ξ-injective coresolution of M is given dually.
Definition 9. ([8] of [Definition 3.2], [17]) Let M and N be objects in . - (1)
If we choose a ξ-projective resolution of M, by applying the functor to we have a complex of abelian groups . For any integer , the-cohomology groups
are defined as - (2)
If we choose a ξ-injective coresolution of N, by applying the functor to we have a complex of abelian groups . For any integer , the-cohomology groups
are defined as
Remark 3. (1) In fact, there is an isomorphism which is denoted by (see [8] of [Definition 3.2]). (2) Assume that has enough ξ-projective objects. Using a standard argument in homological algebra, there is a bijection Remark 4. ([8] of [Lemma 3.4]) Letbe an -triangle in ξ. If has enough ξ-projective objects and M is an object in , then there exists a long exact sequenceof abelian groups. If has enough ξ-injective objects and N is an object in , then there exists a long exact sequenceof abelian groups. For two subcategories and of , we say if (equivalently, ).
Definition 10. ([13] of [Definition 4.5], [17]) Let be a class of objects in . An -trianglein ξ is called to be -exact (resp., -exact) if for any , the induced sequence of abelian groups (resp., ) is exact in . Definition 11. ([13] of [Definition 4.6], [17]) Let be a class of objects in . A complex is called -exact (resp., -exact) if it is a ξ-exact complexin such that for each integer n we have for some -exact (resp., -exact) -trianglein ξ. A ξ-exact complex is called complete -exact (resp., complete -exact) if it is -exact (resp., -exact).
Definition 12. ([13] of [Definition 4.7], [17]) A complete ξ-projective resolution is a complete -exact complexin such that is ξ-projective for each integer n. Dually, a complete ξ-injective coresolution is a complete -exact complexin such that is ξ-injective for each integer n. Definition 13. ([13] of [Definition 4.8], [17]) Let be a complete ξ-projective resolution in . Therefore, for each integer n, there exists a -exact -trianglein ξ. The objects are called ξ-projective for each integer n. Dually if is a complete ξ-injective coresolution in , there exists a -exact -trianglein ξ for each integer n. The objects are called ξ-injective for each integer n. We denote by (resp., ) the class of -projective (resp., -injective) objects. It is obvious that ⊆ and ⊆.
Definition 14. Let and be two subcategories of with . Then is called a ξ-cogenerator of
if for any object
X in
, there exists an
-triangle
in
with
and
.
Definition 15. ([
13] of [Definition 3.4])
Letbe an -triangle in ξ. Then the morphism u (resp. v) is called a -infaltion
(resp. a -deflation
). Fix some arbitrary
-triangle
in
. We say that
is
closed under ξ-extensions if, given any such
-triangle in
as above, if
X,
Z lie in
, then
Y lie in
. We say that
is
closed under cocones of ξ-deflations (resp.
cones of ξ-inflations) if, given any such
-triangle in
as above, if
Y,
Z lie in
(resp.
X,
Y lie in
), the so too does
X (resp.
Z).
Definition 16. Let be an extriangulated category with enough -projective objects and a subcategory of . Then is called a resolving subcategory of if the following conditions are satisfied.
- (1)
.
- (2)
is closed under -extensions.
- (3)
is closed under cocones of -deflations.
Remark 5. (a) We do not require that a resolving subcategory is closed under direct summands in the above definition.
- (b)
is a resolving subcategory and closed under direct summands.
- (c)
is a resolving subcategory and closed under direct summands (see [13] of [Theorems 4.16 and 4.17]).
In the following sections, we always assume that is an extriangulated category and ξ is a proper class of -triangles in . We also assume that the extriangulated category has enough ξ-projectives and enough ξ-injectives satisfying Condition (WIC).
3. Resolution Dimension with Respect to a Resolving Subcategory
We first introduce the following definition.
Definition 17. Let be a subcategory of and . The -resolution dimension of M (with respect to ξ), written -, is defined by For a ξ-exact complexwith all , there are -triangles and with for each . The object are called an ith -syzygy of M, denoted by . In case , we have and write . In case , coincides with defined by Hu, Zhang and Zhou [13] as ξ-projective dimension, the proof is straightforward. Lemma 2. Let be a resolving subcategory of . For any object , ifandare ξ-exact complexes with all and in for , then if and only if . Proof. For
, since
has enough
-projectives, there exists a
-exact complex
for
.
First of all, by the
-exact complex (1) there are
-triangles
in
with
. Moreover, by the
-exact complex (3) there is an
-triangle
in
. Consider the following diagram
It is easy to see that
, i.e.,
is a morphism of
-extensions. Thus, by [
13] of [Lemma 4.15], there is an
-triangle
such that the following diagram
commutes. By [
13] of [Lemma 4.14], there exist morphisms
and
such that there is an
-triangle
and meanwhile, the following diagram
commutes. Repeating this process, we can obtain the following
-exact complex
Similarly, we have the following
-exact complex
Decompose the
-exact complex (4) as the
-triangle
in
and the
-exact complex
Decompose the
-exact complex (5) as the
-triangle
in
and the
-exact complex
Since
is resolving, we have that
X and
Y are objects in
by
-exact complexes (7) and (9). Moreover, by
-triangles (
6) and (
8) we have that
if and only if
if and only if
.
However, from the following
-triangles in
we have that
if and only if
, and
if and only if
. Thus,
if and only if
. □
Using the above, we can get
Proposition 1. Let be a resolving subcategory of and . Then the following statements are equivalent:
- (1)
.
- (2)
for .
- (3)
for .
Now we can compare resolution dimensions in a given -triangle in as follows.
Proposition 2. labelprop-resdim Let be a resolving subcategory of , and letbe an -triangle in ξ. Then we have the following statements for any objects A, B and C in : - (1)
-.
- (2)
-.
- (3)
-.
Proof. For any
, if
-
, by Proposition 1, we have the following
-exact complex
in
with
for
and
.
(1) Assume
-
and
, We will use induction on
m and
n. The case
is trivial. Without loss of generality, we assume
, then we can let
for
. As a similar argument to proof of Lemma 2, we can obtain the following
-exact complex
in
.
Thus, -.
(2) Assume
-
and
-
. We will use induction on
m and
n. The case
is trivial. Without loss of generality, we assume
, then we can let
for
. By [
14] of [Theorem 1], there exist a
-exact complex
and an
-triangle
in
, it follows that
by Remark 5. Thus,
-
and the desired assertion is obtained.
(3) Assume
-
and
-
. We proceed it by induction on
m and
n. The case
is trivial. Without loss of generality, we assume
, then we can let
for
. By [
14] of [Theorem 3], we have the following
-exact complex
in
, thus
-
and the desired assertion is obtained. □
We use to denote the full subcategory of whose objects have finite -resolution dimension. Following the above, we have the closure properties for the subcategory .
Remark 6. If is a resolving subcategory of , then is closed under cocones of ξ-deflations, cones of ξ-inflations and ξ-extensions.
Corollary 1. Let be a resolving subcategory of , and letbe an -triangle in ξ. - (1)
Let . Then --.
- (2)
Let . Then either or else --.
- (3)
Let and . Then --.
Proposition 3. Let and be two subcategories of with .
- (1)
.
- (2)
If is resolving, then for any , if and only if .
In particular, if , and is closed under cocones of ξ-deflations or closed under direct summands, then .
Proof. (1) Obviously.
(2) The only if part. Clearly, . Let . By assumption, we have , then , and so . Thus, .
The if part. Suppose
and
. Clearly
. Consider the following
-exact complexes
and
with
and
for all
and
. Since
, we have
by Lemma 2. Then
, and thus
and the desired equality is obtained.
Now, we assume that
and
is closed under cocones of
-deflations or closed under direct summands. Clearly,
. Conversely, let
. There exists a
-exact complex
with each
lies in
. Set
for
, where
. Since
is resolving, we have
, and hence
. Consider the following
-triangle
in
. Since
by the assumption that
, we have that the
-triangle (
10) is split by Remark 3(2). It follows that
and there exists an
-triangle
in
. Since
is closed under cocones of
-deflations or closed under direct summands by assumption, we have
. Repeating this process, we can obtain each
, hence
and
. Thus,
. □
Now we give the following definition.
Definition 18. Let be a subcategory of and M an object in . A ξ-deflation with is said to be a right -approximation of M if is exact for any .
The notion of a left -approximation of M is given dually.
We need the following easy and useful observation.
Lemma 3. Let and be two subcategories of .
- (1)
If , then . In particular, if , then .
- (2)
If , then .
Proof. let
. Then there is a
-exact complex
with each
for some nonnegative integer
n. This means that there are
-triangles
,
, and
in
with
for any
. Applying Remark 4, we can get
for any
. □
The following is an analogous theory of Auslander-Buchweitz approximations (see [
3,
6]).
Proposition 4. Let be a subcategory of closed under ξ-extensions, and let be a subcategory of such that is a ξ-cogenerator of . Then for each with , there exist two -trianglesandin ξ, where X, , and . In particular, if , then the ξ-deflation is a right -approximation of M, and the ξ-inflation is a left -approximation of M.
Proof. We will use induction on
n. The case for
is trivial. If
, there exists an
-triangle
in
with
,
. Since
is a
-cogenerator of
, there is an
-triangle
in
with
and
. By Lemma 1(2), we have the following commutative diagram
Since
is closed under cobase changes, we obtain that the
-triangle
is in
with
. Notice that
is a
-deflation, so we have that
is a
-deflation by [
13] of [Proposition 4.13], hence the
-triangle
is in
by [
12] of [Remark 3.10]. Since
is closed under
-extensions by assumption, we have
. Therefore, (
14) is the first desired
-triangle.
For
, since
is a
-cogenerator of
, there is an
-triangle
in
with
and
. By (ET4), we have the following commutative diagram
Notice that
is a
-inflation by [
13] of [Corollary 3.5], so the
-triangle
is in
. Since
is a
-deflation,
is a
-deflation by [
13] of [Proposition 4.13]. Therefore, the
-triangle
is in
with
and
, which is the second desired
-triangle.
Now suppose
. Then there is an
-triangle
in
with
and
. For
, by the induction hypothesis, we get an
-triangle
in
with
and
. By Lemma 1(2), we have the following commutative diagram
Notice that
is a
-deflation, then
is a
-deflation by [
13] of [Proposition 4.13], so the
-triangle
is in
. It follows that
from the assumption that
is closed under
-extensions. Since
is closed under cobase changes, we obtain the first desired
-triangle
in
with
and
.
For
X, since
is a
-cogenerator of
, we get the following
-triangle
in
with
and
.
By (ET4), we have the following commutative diagram
As a similar argument to that of the diagram (
15), we obtain that the
-triangles
and
are in
. Thus, (
18) is the second desired
-triangle in
with
and
.
In particular, suppose . By Lemma 3, we have . Then for any , it follows that is exact. Thus, the -deflation is a right -approximation of M. Similarly, we can prove that the -inflation is a left -approximation of M. □
Proposition 5. Keep the notion as Proposition 4. Assume with .
- (1)
If is resolving, then in the -triangles (11) and (12), we have and . In particular, if , then the ξ-deflation in the -triangle (11) is a right -approximation of M, such that . - (2)
If and is resolving, then there is an -trianglein ξ with , and . - (3)
- (a)
Let . If is a ξ-cogenerator of and is closed under ξ-extensions, then if and only if .
- (b)
If is a resolving and is a ξ-cogenerator of and , then .
- (4)
Suppose that and are resolving. If is a ξ-cogenerator of and , then M admits a right -approximation such that is an -triangle in ξ, where . In fact, we have .
Proof. (1) If
, then there is an
-triangle
. By setting
in the
-triangles (
11), we have
. If
, then
. Applying Corollary 1(2) to the
-triangle (
11) yields that
. On the other hand, since
, we have
. Thus,
.
Moreover, applying Corollary 1(1) to the
-triangle (
12) implies
. Therefore,
. Hence
.
The last assertion follows from the above argument and Proposition 4.
(2) Since , we have by Lemma 3, and so the result immediately follows from (1) and Propostion 4.
(3) (a) (⇐) Suppose . Clearly, , i.e., .
(⇒) Suppose
. Let
. By Proposition 4, there exists an
-triangle
in
with
and
. Notice that
, so
by Lemma 3, and hence
, thus
implies
. Then
, and so
. Since
, we have
by Lemma 3. Thus,
.
(b) Suppose
, by (1) and Propostion 4, there exists an
-triangle
in
with
and
. Notice that
, so
by Lemma 3, thus
and
, so
by Remark 4, and hence
. It follows that
. However,
, thus
.
(4) Suppose
, by (1), there exists an
-triangle
in
with
and
. By (2), there is an
-triangle
in
with
,
and
. Then
. By Lemma 1(2), we have the following commutative diagram
One can see that the
-triangle
is in
and
. Notice that
, so
by (3)(a). Then
for any
, and so
is exact. Thus, the
-deflation
is a right
-approximation of
M and
in the
-triangle (
20). Notice that
, so we have
by (3)(b). □
Lemma 4. Let be a subcategory of with . Assume that is closed under cocones of ξ-deflations or closed under direct summands. Then .
Proof. Clearly, .
Conversely, let
. Then there is a
-exact complex
with each
for some nonnegative integer
n. This means that there are
-triangles
,
, and
in
with
for any
. Then
yields
by Remark 4, and so the
-triangle
is split. It follows that
and there exists an
-triangle
in
. Since
is closed under cocones of
-deflations or closed under direct summands by assumption, we have
. Repeating this process, we can obtain
, hence
and
. Thus,
. □
Proposition 6. Let be a resolving subcategory of and a ξ-cogenerator of with . Assume that is closed under cocones of ξ-deflations or closed under direct summands. Then .
Proof. Clearly, and .
Now, let . Then by Lemma 3(2), we have , and hence .
On the other hand, by Proposition 4, there is an
-triangle
in
with
and
. Notice that
implies
by Remark 4, and hence
by Lemma 4. Notice that
, so the
-triangle (
21) is split, hence
. Consider the following
-triangle
in
. It follows that
from the assumption that
is resolving. Thus,
. □
Our main result is the following
Theorem 2. Let be a resolving subcategory of and a ξ-cogenerator of with . Assume that is closed under cocones of ξ-deflations or closed under direct summands. For any , if , then the following statements are equivalent:
- (1)
.
- (2)
for all .
- (3)
for all .
- (4)
for all and all .
- (5)
for all and all .
- (6)
M admits a right -approximation , where φ is a ξ-deflation, such that there is an -triangle satisfying .
- (7)
There are two -trianglesandin ξ such that , and , .
Proof. follow from Proposition 1.
follows from Proposition 5(1), and is straightforward.
follows from Proposition 5(1), and is straightforward.
Suppose
. There is a
-exact complex
with all
in
. This means that there are
-triangles
,
, and
in
with
for any
. By assumption, we have
for all
. Thus, by Remark 4,
for
.
It follows from Lemma 3.
It is clear.
Since
, by Proposition 5(1), there is an
-triangle
in
with
and
. Then
for
and
since
. Therefore,
. Please note that
, so we have the following
-exact complex
with all
. Then by Remark 4,
for
and all
, which means
. Notice that
, hence
. It follows that
from Lemma 4, so
. Thus,
. □