Almost-split sequence
Auslander-Reiten sequence
Roughly speaking, almost-split sequences are minimal non-split short exact sequences. They were introduced by M. Auslander and I. Reiten in 1974–1975 and have become a central tool in the theory of representations of finite-dimensional algebras (cf. also Representation of an associative algebra).
Let R be an Artin algebra, i.e. R is an associative ring with unity that is finitely generated as a module over its centre Z(R), which is a commutative Artinian ring.
Let C be an indecomposable non-projective finitely-generated left R-module. Then there exists a short exact sequence
\begin{equation} \tag{a1} 0 \rightarrow A \stackrel { f } { \rightarrow } B \stackrel { g } { \rightarrow } C \rightarrow 0 \end{equation}
in {}_{R}\operatorname{Mod}, the category of finitely-generated left R-modules, with the following properties:
i) A and C are indecomposable;
ii) the sequence does not split, i.e. there is no section s : C \rightarrow B of g (a homomorphism such that g s = \operatorname{id}), or, equivalently, there is no retraction of f (a homomorphism r : B \rightarrow A such that r f = \operatorname{id});
iii) given any h : Z \rightarrow C with Z indecomposable and h not an isomorphism, there is a lift of h to B (i.e. a homomorphism \tilde { h } : Z \rightarrow B in {}_{R}\operatorname{Mod} such that g \tilde { h } = h);
iv) given any j : A \rightarrow X with X indecomposable and j not an isomorphism, there is a homomorphism \tilde { j } : B \rightarrow X such that \tilde { j } f = j.
Note that if iii) (or, equivalently, iv)) were to hold for all h, not just those h that are not isomorphisms, the sequence (a1) would be split, whence "almost split" . Moreover, a sequence (a1) with these properties is uniquely determined (up to isomorphism) by C, and also by A. This is the basic Auslander–Reiten theorem on almost-split sequences, [a1], [a8], [a9], [a10], [a11].
For convenience (things also work more generally), let now R be a finite-dimensional algebra over an algebraically closed field k. The category {}_{R}\operatorname{Mod} is a Krull–Schmidt category (Krull–Remak–Schmidt category), i.e. a C \in {}_{R}\operatorname{Mod} is indecomposable if and only if {}_R\operatorname{End}(C,C), the endomorphism ring of C, is a local ring and (hence) the decomposition of a module in {}_{R}\operatorname{Mod} into indecomposables is unique up to isomorphism.
Let C be an indecomposable and consider the contravariant functor X \mapsto {}_ { R } \operatorname { Mod } ( X , C ). The morphisms g : X \rightarrow C that do not admit a section (i.e. an s : C \rightarrow X such that g s = \operatorname{id}) form a vector subspace E _ { C } ( X ) \subset {}_{R}\operatorname { Mod } ( X , C ). Let S _ { C } be the quotient functor S _ { C } = \operatorname { Mod } ( ? , C ) / E _ { C }. Then, for an indecomposable D, S _ { C } ( D ) = k if D is isomorphic to C and zero otherwise. So S _ { C } is a simple functor. (All functors \Box_R \text { Mod } ( ? , C ), E _ { C }, S _ { C } are viewed as k-functors, i.e. functors that take their values in the category of vector k-spaces.) If C is indecomposable, then (the Auslander–Reiten theorem, [a4], p.4) the simple functor S _ { C } admits a minimal projective resolution of the form
\begin{equation*} 0 \rightarrow {}_{R} \operatorname { Mod } ( ? , A ) \rightarrow {}_{R} \operatorname { Mod } ( ? , B ) \rightarrow \end{equation*}
\begin{equation*} \rightarrow {}_{R} \text { Mod } ( ? , C ) \rightarrow S _ { C } \rightarrow 0. \end{equation*}
If C is projective, A is zero, otherwise A is indecomposable.
If C is not projective, the sequence 0 \rightarrow A \rightarrow B \rightarrow C \rightarrow 0 is exact and is the almost-split sequence determined by C.
This functorial definition is used in [a5] in the somewhat more general setting of exact categories.
For a good introduction to the use of almost-split sequences, see [a6]; see also [a3], [a5] for comprehensive treatments. See also Riedtmann classification for the use of almost-split sequences and the Auslander–Reiten quiver in the classification of self-injective algebras.
The Bautista–Brunner theorem says that if R is of finite representation type and 0 \rightarrow A \rightarrow B \rightarrow C \rightarrow 0 is an almost-split sequence, then B has at most 4 terms in its decomposition into indecomposables; also, if there are indeed 4, then one of these is projective-injective. This can be generalized, [a7].
References
[a1] | M. Auslander, I. Reiten, "Stable equivalence of dualizing R-varieties I" Adv. Math. , 12 (1974) pp. 306–366 |
[a2] | M. Auslander, "The what, where, and why of almost split sequences" , Proc. ICM 1986, Berkeley , I , Amer. Math. Soc. (1987) pp. 338–345 |
[a3] | M. Auslander, I. Reiten, S.O. Smalø, "Representation theory of Artin algebras" , Cambridge Univ. Press (1995) |
[a4] | P. Gabriel, "Auslander–Reiten sequences and representation-finite algebras" V. Dlab (ed.) P. Gabriel (ed.) , Representation Theory I. Proc. Ottawa 1979 Conf. , Springer (1980) pp. 1–71 |
[a5] | P. Gabriel, A.V. Roiter, "Representations of finite-dimensional algebras" , Springer (1997) pp. Sect. 9.3 |
[a6] | I. Reiten, "The use of almost split sequences in the representation theory of Artin algebras" M. Auslander (ed.) E. Lluis (ed.) , Representation of Algebras. Proc. Puebla 1978 Workshop , Springer (1982) pp. 29–104 |
[a7] | Shiping Liu, "Almost split sequenes for non-regular modules" Fundam. Math. , 143 (1993) pp. 183–190 |
[a8] | M. Auslander, I. Reiten, "Representation theory of Artin algebras III" Commun. Algebra , 3 (1975) pp. 239–294 |
[a9] | M. Auslander, I. Reiten, "Representation theory of Artin algebras IV" Commun. Algebra , 5 (1977) pp. 443–518 |
[a10] | M. Auslander, I. Reiten, "Representation theory of Artin algebras V" Commun. Algebra , 5 (1977) pp. 519–554 |
[a11] | M. Auslander, I. Reiten, "Representation theory of Artin algebras VI" Commun. Algebra , 6 (1978) pp. 257–300 |
Almost-split sequence. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Almost-split_sequence&oldid=52543