Definition 1.3. Let be a triangulated category. A pair of full subcategories of , , is called a -structure on if it satisfies the following properties:
and ;
;
For any object of there exists a distinguished triangle
with and .
The concept of a -structure on a triangulated category was first introduced by Beilinson, Bernstein and Deligne in [3]. The basic theory of -structures can be found in [3] and [13].
Definition 1.3. Let be a triangulated category. A pair of full subcategories of , , is called a -structure on if it satisfies the following properties:
and ;
;
For any object of there exists a distinguished triangle
with and .
The full subcategories and are often denoted by and , or simply by and , respectively; see [7] and [11].
The notion of a -structure has become widespread in the study of triangulated categories and lends itself particularly well to induction arguments in this setting, see for example [5].
We next introduce the almost dual notion of a co--structure. We note that co--structures have also recently been introduced by Bondarko in [7] where they are called weight structures.
Definition 1.4. Let be a triangulated category. A pair of full subcategories of , , will be called a co--structure on if it satisfies the following properties:
and are closed under direct summands;
and ;
;
For any object of there exists a distinguished triangle
with and .
In [7] the full subcategories and are denoted by and , respectively.
It is easy to see that is closed under direct summands if and only if is closed under direct summands; similarly for .
One can see that by interchanging the roles of and in the definition of a -structure, properties , and in Definition 1.3 become the corresponding properties in the definition of a co--structure. The inclusion of condition in Definition 1.4 is the reason why a co--structure is almost dual to a -structure rather than simply being its dual. We also note that properties and in Definitions 1.3 and 1.4 make and into examples of torsion theories in the sense of [12].
The notion of a non-degenerate co--structure can be defined in a manner analogous to that of a non-degenerate -structure.
Definition 1.5. A co--structure on a triangulated category will be called non-degenerate if we have
Recall that a full subcategory of a triangulated category is said to be closed under extensions if, whenever we have a distinguished triangle
with and objects of , then is also an object of . We next give some elementary properties of co--structures.
Proposition 1.6. Let be a triangulated category and suppose is a co--structure on . We have:
(i) For all objects of there exists an -precover .
(ii) For all objects of there exists a -preenvelope .
(iii) We have and .
(iv) is closed under extensions.
(v) is closed under extensions.
Proof: Properties (i) and (ii) are immediate consequences of the definition of a co--structure: the -precover and the -preenvelope are just the first and second morphisms in the distinguished triangle given by property of Definition 1.4. Property (iii) is a consequence of condition and the orthogonality condition of Definition 1.4, and properties (iv) and (v) are easy consequences of property (iii), see [12].
One sees in Proposition 1.6 that preenvelopes and precovers replace the truncation functors associated with -structures. In order to obtain the equalities of property (iii), and thus the fact that both halves of the co--structure are closed under extensions, we need to assume condition of Definition 1.4 which says that both halves of a co--structure are closed under direct summands.
Let be a -structure on a triangulated category . The intersection , of both halves of the -structure is called the heart of the -structure. It has the nice property that it is an abelian subcategory of . In particular, the hearts of -structures provide a means of obtaining abelian categories from triangulated categories. We define an analogous notion for co--structures.
Definition 1.7. Let be a co--structure for a triangulated category . The intersection will be called the coheart of the co--structure.
It would be hoped that the coheart of a co--structure on would be an abelian subcategory of . Unfortunately, this is not the case, see [7]. However, in this paper we present an example in which the coheart does turn out to be an abelian category.
In the next section we give some examples of co--structures.
Original source: arXiv:0705.0102v2