Extriangulated length categories: torsion classes and Ο„ \tau -tilting theory Li Wang, Jiaqun Wei, Haicheng Zhang and Panyue Zhou Abstract . This paper introduces the notion of extriangulated length categories, whose prototypical examples include abelian length categories and bounded derived categories of finite dimensional algebras with finite global dimension. We prove that an extriangulated category π’œ \mathcal{A} is a length category if and only if π’œ \mathcal{A} admits a simple-minded system. Subsequently, we study the partially ordered set tor Θ ​ ( π’œ ) \mbox{\rm tor}_{\Theta}(\mathcal{A}) of torsion classes in an extriangulated length category ( π’œ , Θ ) (\mathcal{A},\Theta) from the perspective of lattice theory. It is shown that tor Θ ​ ( π’œ ) \mbox{\rm tor}_{\Theta}(\mathcal{A}) forms a complete lattice, which is further proved to be completely semidistributive and algebraic. Moreover, we describe the arrows in the Hasse quiver of tor Θ ​ ( π’œ ) \mbox{\rm tor}_{\Theta}(\mathcal{A}) using brick labeling. Finally, we introduce the concepts of support torsion classes and support Ο„ \tau -tilting subcategories in extriangulated length categories and establish a bijection between these two notions, thereby generalizing the Adachi-Iyama-Reiten bijection for functorially finite torsion classes. 2020 Mathematics Subject Classification 18G80; 18E10; 16G10; 18E05 1. Introduction The concept of abelian categories, introduced and thoroughly explored by Grothendieck, serves as a unified framework that elegantly generalizes both module categories over rings and categories of sheaves over schemes. Among these, abelian length categories [ 11 ] , which are abelian categories with the length of every object finite, play a pivotal role in representation theory. Their importance stems from their structural richness and their applications in various mathematical domains, including \tau -tilting theory [ 9 ] , lattice theory [ 3 ] , and the study of strongly unbounded types [ 16 ] . Schur’s lemma underscores their significance by showing that the set of isomorphism classes of simple objects in an abelian length category forms a semibrick, which is a combinatorial object with rich mathematical properties. In triangulated categories, the notion of a simple-minded system generalizes the role of simple objects in abelian length categories. This concept, introduced by Koenig and Liu [ 15 ] in the study of stable module categories, has been further investigated, particularly in relation to mutations [ 7 ] . Dugas [ 7 ] proved that bounded derived categories of finite dimensional algebras with finite global dimension admit simple-minded systems. This result allows us to interpret these derived categories as analogs of abelian length categories in the triangulated framework, thereby establishing a meaningful connection between the two contexts. Recently, Nakaoka and Palu [ 17 ] proposed the framework of extriangulated categories, which synthesizes and extends properties from both triangulated and exact categories. Within this broader framework, it becomes natural to ask how the notion of abelian length categories can be extended. To address this, we introduce the concept of extriangulated length categories, which formalizes the idea of abelian length categories in the setting of extriangulated categories. An extriangulated length category is defined as a pair (\mathcal{A},\Theta) , where \mathcal{A} is an extriangulated category and \Theta:\mbox{\rm Iso}\,(\mathcal{A})\rightarrow\mathbb{N} is a length function that ensures every morphism in \mathcal{A} satisfies the \Theta -admissibility condition (see Definition 3.8 ). This definition preserves the essential characteristics of abelian length categories while adapting them to the more flexible extriangulated setting. Our first main result, stated below, establishes a foundational theorem for extriangulated length categories and opens the door to further theoretical exploration and applications. Theorem A . (see Theorems 3.9 , 3.14 for details) Let \mathcal{A} be an extriangulated category. (1) If \mathcal{X} is a simple-minded system in \mathcal{A} , then (\mathcal{A},l_{\mathcal{X}}) is an extriangulated length category, where l_{\mathcal{X}} is the minimal length of \mathcal{X} -filtrations. (2) \mathcal{A} is an extriangulated length category if and only if \mathcal{A} has a simple-minded system. An important source of examples for extriangulated length categories comes from the notion of length wide subcategories (see Definition 3.15 ). These subcategories extend the concept of wide subcategories from abelian length categories to the extriangulated setting, thereby enriching the structural landscape of extriangulated categories. Analogous to their counterparts in abelian categories, length wide subcategories can be generated by semibricks, a combinatorial structure that captures key categorical properties. This connection not only deepens our understanding of the role that semibricks play in categorical frameworks but also allows us to generalize classical results into the realm of extriangulated categories. In particular, we extend the celebrated result of Ringel [ 19 ] , which describes the structure and generation of certain subcategories in abelian length categories. Our generalization shows that the underlying principles of Ringel’s theorem hold in this broader context, offering new insights and potential applications. More specifically, we establish the following theorem, which serves as a cornerstone of our study. Theorem B . (see Theorem 3.18 for details) Let \mathcal{A} be an extriangulated category. Then there exists a bijection between the following two sets. (1) The set of semibricks \mathcal{X} in \mathcal{A} . (2) The set of length wide subcategories (\mathcal{C},\Theta) in \mathcal{A} . The mutually inverse maps are given by \mathcal{X}\mapsto(\mbox{\rm{Filt}}_{\mathcal{A}}(\mathcal{X}),l_{\mathcal{X}}) and (\mathcal{C},\Theta)\mapsto\mbox{\rm sim}(\mathcal{C}) . The concept of torsion classes forms a cornerstone in the representation theory of algebras, has significant contributions and applications in areas such as \tau -tilting theory [ 1 ] , lattice theory [ 6 ] , stability conditions [ 4 ] , and Hall algebras [ 20 ] . These subcategories play a pivotal role in understanding and organizing the structural properties of various algebraic and categorical frameworks. In this paper, we extend the notion of torsion classes to the setting of extriangulated length categories (\mathcal{A},\Theta) and study their fundamental properties. Specifically, torsion classes in this context are subcategories closed under extensions and quotients (see Definition 4.1 ), preserving key homological behaviors. These subcategories exhibit elegant and useful properties, as established in Lemma 4.4 and Proposition 4.8 . Furthermore, we explore the collection \mbox{\rm torf}_{\Theta}(\mathcal{A}) , which consists of all torsion classes in \mathcal{A} , through the lens of lattice theory. This approach reveals structural insights and provides a unified framework that parallels the classical results for finite-dimensional algebras [ 6 ] . In particular, we show that the set \mbox{\rm torf}_{\Theta}(\mathcal{A}) forms a lattice with desirable combinatorial and categorical properties. These findings culminate in the following theorem, which generalizes known results and establishes a robust theoretical foundation for torsion classes in extriangulated length categories. Theorem C . (see Theorem 4.12 for details) Let (\mathcal{A},\Theta) be an extriangulated length category. (1) The set \mbox{\rm tors}_{\Theta}(\mathcal{A}) is a complete lattice. (2) The complete lattice \mbox{\rm tors}_{\Theta}(\mathcal{A}) is completely semidistributive and algebraic. For a given set \mathcal{X} , we denote its cardinality by \#\mathcal{X} . Let [\mathcal{U},\mathcal{T}] be an interval in the lattice \mbox{\rm tors}_{\Theta}(\mathcal{A}) of torsion classes in the extriangulated length category (\mathcal{A},\Theta) . We denote by {\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) the set of isomorphism classes of bricks contained in \mathcal{U}^{\perp}\cap\mathcal{T} , where \mathcal{U}^{\perp} represents the right orthogonal of \mathcal{U} in \mathcal{A} . This framework provides a natural way to study the combinatorial and structural properties of intervals within the lattice \mbox{\rm tors}_{\Theta}(\mathcal{A}) . Using this notation, we arrive at the following description of the intervals in \mbox{\rm tors}_{\Theta}(\mathcal{A}) , which connects the cardinalities of such intervals to the numbers of certain bricks. This characterization not only highlights the intrinsic link between torsion classes and bricks but also underscores the interplay between algebraic and combinatorial aspects of extriangulated length categories, offering new insights into their lattice-theoretic properties. The precise formulation of this description is presented in the following. Theorem D . (see Theorem 4.18 for details) Let (\mathcal{A},\Theta) be an extriangulated length category. Suppose that [\mathcal{U},\mathcal{T}]\subseteq\mbox{\rm tors}_{\Theta}(\mathcal{A}) . (1) We have \#[\mathcal{U},\mathcal{T}]=1 if and only if brick (\mathcal{U}^{\perp}\cap\mathcal{T})=\emptyset . (2) If \# brick (\mathcal{U}^{\perp}\cap\mathcal{T})=1 , then \#[\mathcal{U},\mathcal{T}]=2 . (3) If \#[\mathcal{U},\mathcal{T}]=2 , then there exists a brick S\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) satisfying the following conditions: (i) \Theta(S)\leq\Theta(S^{\prime}) for any S^{\prime}\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) . (ii) \mathcal{T}=\mathrm{T}_{\Theta}(\mathcal{U}\cup S) and \mathcal{U}=\mathcal{T}\cap{{}^{\perp}}S . (iii) S satisfies S\in\bigcap_{S^{\prime}\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T})}(\mbox{\rm Sub}_{\Theta}(S^{\prime})\cap\mbox{\rm Fac}_{\Theta}(S^{\prime})). Moreover, the brick S is unique up to isomorphism. Building on Theorem {\rm D} , we naturally extend the concept of brick labeling (see Definition 4.19 ) to the Hasse quiver of \mbox{\rm tors}_{\Theta}(\mathcal{A}) , offering a generalization of the framework introduced in [ 6 , Section 3.2] . This labeling provides a systematic way to assign bricks to the intervals in the lattice \mbox{\rm torf}_{\Theta}(\mathcal{A}) of torsion-free classes, enriching our understanding of its combinatorial and categorical structure. Moreover, we characterize the arrows in the Hasse quiver of \mbox{\rm tors}_{\Theta}(\mathcal{A}) using the brick labeling, as detailed in Proposition 4.22 . This characterization deepens the connection between the morphisms in the quiver and the corresponding torsion classes. Finally, we extend \tau -tilting theories to extriangulated length categories by leveraging the structure of torsion classes. To achieve this, we introduce the concepts of support torsion classes (see Definition 5.2 ) and support \tau -tilting subcategories (see Definition 5.3 ) in the context of extriangulated length categories. This notion serves as a natural analog of its counterpart in classical settings and enables us to generalize the celebrated bijection of Adachi-Iyama-Reiten (cf. [ 1 , Theorem 2.7] ) for functorially finite torsion classes. Our generalization highlights the versatility of \tau -tilting theory and its applicability in the broader framework of extriangulated categories. This development not only unifies and extends the existing results but also lays the groundwork for further exploration of the interplay between torsion classes, \tau -tilting theories, and extriangulated length categories. The precise statement of our generalized bijection is presented as follows. Theorem E . (see Theorem 5.6 for details) Let (\mathcal{A},\Theta) be an extriangulated length category with enough \Theta -projectives. Then we have a bijection between the following sets. (1) The set of support torsion classes. (2) The set of support \tau -tilting subcategories. \mathbf{Organization.} This paper is organized as follows. In Section 2, we summarize the fundamental definitions and key properties of extriangulated categories and filtration categories, and provide the foundational concepts required for subsequent discussions. Section 3 introduces the notion of extriangulated length categories and establishes several basic properties that will be utilized throughout the paper. In Section 4, we develop a lattice-theoretical framework for studying the partially ordered sets of torsion classes within extriangulated length categories, offering a structured approach to analyzing their relationships. Finally, in Section 5, we introduce the concepts of support torsion classes and support \tau -tilting subcategories. We also establish a bijection between these two notions, thereby extending \tau -tilting theories to the framework of extriangulated categories. \mathbf{Conventions~and~Notation.} Throughout this paper, we assume that all considered categories are skeletally small and Krull-Schmidt, and that the subcategories are full and closed under isomorphisms. For an additive category \mathscr{C} , we denote by \mbox{\rm Iso}\,(\mathscr{C}) the set of isomorphism class of objects in \mathscr{C} . Given an object X\in\mathscr{C} , we denote by \mbox{\rm add}(X) the full subcategory of \mathscr{C} consisting of finite direct sums of direct summands of X . For a finite dimensional algebra \Lambda over a field k , we denote by \mbox{\rm{mod}}\,\Lambda the category of finite dimensional \Lambda -modules and by D^{b}(\Lambda) the bounded derived category. 2. Preliminaries In this section, we review some notations and properties of extriangulated categories and filtration subcategories, as introduced in [ 17 ] and [ 21 ] , respectively. 2.1. Extriangulated categories In this subsection, we omit the detailed axioms and definitions on extriangulated categories, and only give terminologies and properties which we shall need later. For the precise definition and the detailed properties of extriangulated categories, we refer to [ 17 , Section 2] . An extriangulated category (\mathscr{C},\mathbb{E},\mathfrak{s}) is a triple consisting of the following data and satisfying certain conditions: (1) \mathscr{C} is an additive category, and \mathbb{E}\colon\mathscr{C}^{\rm op}\times\mathscr{C}\rightarrow{\rm Ab} is an additive bifunctor, where Ab is the category of abelian groups. (2) \mathfrak{s} is a correspondence that assigns to each \delta\in\mathbb{E}(Z,X) an equivalence class [X\to Y\to Z] of complexes in \mathscr{C} . Here two complexes X\xrightarrow{~f~}Y\xrightarrow{~g~}Z and X\xrightarrow{~f^{\prime}~}Y\xrightarrow{~g^{\prime}~}Z are equivalent if there exists an isomorphism h\colon Y\to Y^{\prime} such that the following diagram commutes: \textstyle{X\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{f} \textstyle{Y\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{g} \scriptstyle{h} \textstyle{Z\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{X\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{f^{\prime}} \textstyle{Y^{\prime}\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{g^{\prime}} \textstyle{Z.} Moreover, this triplet must satisfy the axioms (ET1)-(ET4), (ET3) op and (ET4) op as detailed in [ 17 , Definition 2.12] . In what follows, we often write \mathscr{C} for an extriangulated category (\mathscr{C},\mathbb{E},\mathfrak{s}) for simplicity. Given \delta\in\mathbb{E}(C,A) . If \mathfrak{s}(\delta)=[A\stackrel{{\scriptstyle x}}{{\longrightarrow}}B\stackrel{{\scriptstyle y}}{{\longrightarrow}}C] , then the sequence A\stackrel{{\scriptstyle x}}{{\longrightarrow}}B\stackrel{{\scriptstyle y}}{{\longrightarrow}}C is called a conflation , x is called an inflation and y is called a deflation . In this case, we call A\stackrel{{\scriptstyle x}}{{\longrightarrow}}B\stackrel{{\scriptstyle y}}{{\longrightarrow}}C\stackrel{{\scriptstyle\delta}}{{\dashrightarrow}} is an \mathbb{E} -triangle, and denote A=\mbox{\rm CoCone}(y) and C=\mbox{\rm Cone}(x) . For two subcategories \mathcal{T},\mathcal{F}\subseteq\mathscr{C} , we define \mathcal{T}^{\perp}=\{M\in\mathscr{C}~|~\mbox{\rm Hom}(X,M)=0~\text{for any}~X\in\mathcal{T}\}, {}^{\perp}\mathcal{T}=\{M\in\mathscr{C}~|~\mbox{\rm Hom}(M,X)=0~\text{for any}~X\in\mathcal{T}\}, \mathcal{T}\ast\mathcal{F}=\{M\in\mathscr{C}~|~\text{there exists an $\mathbb{E}$-triangle}~T\stackrel{{\scriptstyle}}{{\longrightarrow}}M\stackrel{{\scriptstyle}}{{\longrightarrow}}F\stackrel{{\scriptstyle}}{{\dashrightarrow}}~\text{with}~T\in\mathcal{T},F\in\mathcal{F}\}. We say that a subcategory \mathcal{T}\subseteq\mathscr{C} is extension-closed if \mathcal{T}\ast\mathcal{T}\subseteq\mathcal{T} . In this case, \mathcal{T} is an extriangulated category with extriangulated structure induced from that of \mathscr{C} (see [ 17 , Remark 2.18] ). Remark 2.1 . We know that both exact categories and triangulated categories are examples of extriangulated categories [ 17 , Example 2.13] , and that extension-closed subcategories of extriangulated categories are also extriangulated [ 17 , Remark 2.18] . Furthermore, there exist extriangulated categories that are neither exact nor triangulated, as demonstrated in [ 10 , 13 , 17 , 18 , 24 , 25 ] . 2.2. Filtration subcategories In this subsection, we always assume that \mathscr{C}=(\mathscr{C},\mathbb{E},\mathfrak{s}) is an extriangulated category. Let \mathcal{X} be a collection of objects in \mathscr{C} . The filtration subcategory \mathbf{Filt_{\mathscr{C}}(\mathcal{X})} is consisting of all objects M admitting a finite filtration of the form 0=M_{0}\stackrel{{\scriptstyle f_{0}}}{{\longrightarrow}}M_{1}\stackrel{{\scriptstyle f_{1}}}{{\longrightarrow}}M_{2}{\longrightarrow}\cdots\xrightarrow{f_{n-1}}M_{n}=M with f_{i} being an inflation and \mbox{\rm Cone}(f_{i})\in\mathcal{X} for any 0\leq i\leq n-1 . For each object M\in\mbox{\rm{Filt}}_{\mathscr{C}}(\mathcal{X}) , the minimal length of \mathcal{X} -filtrations of M is called the \mathcal{X} - length of M , which is denoted by l_{\mathcal{X}}(M) . The following lemma can be found in [ 21 ] , which is useful in the sequel. Proposition 2.2 . Let \mathcal{X} be a collection of objects in \mathscr{C} and set \mathcal{A}:=\mbox{\rm{Filt}}_{\mathscr{C}}(\mathcal{X}) . (1) For any \mathbb{E} -triangle A\stackrel{{\scriptstyle}}{{\longrightarrow}}B\stackrel{{\scriptstyle}}{{\longrightarrow}}C\stackrel{{\scriptstyle}}{{\dashrightarrow}} in \mathcal{A} , we have l_{\mathcal{X}}(B)\leq l_{\mathcal{X}}(A)+l_{\mathcal{X}}(C) . (2) \mathcal{A} is the smallest extension-closed subcategory containing \mathcal{X} in \mathscr{C} . (3) For any M\in\mathcal{A} , there exist two \mathbb{E} -triangles X_{1}\stackrel{{\scriptstyle}}{{\longrightarrow}}M\stackrel{{\scriptstyle}}{{\longrightarrow}}M^{\prime}\stackrel{{\scriptstyle}}{{\dashrightarrow}} and M^{\prime\prime}\stackrel{{\scriptstyle}}{{\longrightarrow}}M\stackrel{{\scriptstyle}}{{\longrightarrow}}X_{2}\stackrel{{\scriptstyle}}{{\dashrightarrow}} in \mathcal{A} such that X_{1},X_{2}\in\mathcal{X} and l_{\mathcal{X}}(M^{\prime})=l_{\mathcal{X}}(M^{\prime\prime})=l_{\mathcal{X}}(M)-1 . Proof.. The statements can be found as follows: (1) in [ 21 , Lemma 2.7] , (2) in [ 21 , Lemma 2.8] , and (3) in [ 21 , Lemma 2.9] . ∎ Recall that an object M\in\mathscr{C} is called a brick if its endomorphism ring is a division ring. A set \mathcal{X} of isomorphism classes of bricks in \mathscr{C} is called a semibrick if \mbox{\rm Hom}_{\mathscr{C}}(X_{1},X_{2})=0 for any two non-isomorphic objects X_{1},X_{2} in \mathcal{X} . A semibrick \mathcal{X} is said to be proper if for any X\in\mathcal{X} , there does not exist an \mathbb{E} -triangle of the form X\stackrel{{\scriptstyle}}{{\longrightarrow}}0\stackrel{{\scriptstyle}}{{\longrightarrow}}N\stackrel{{\scriptstyle}}{{\dashrightarrow}} with N\in\mbox{\rm{Filt}}_{\mathcal{A}}(\mathcal{X}) . It is straightforward to verify that the notions of semibricks and proper semibricks coincide in exact categories. However, they differ in triangulated categories (see [ 22 , Example 4.2] ). We summarize the following basic properties. Proposition 2.3 . Let \mathcal{X} be a semibrick in \mathscr{C} and set \mathcal{A}:=\mbox{\rm{Filt}}_{\mathscr{C}}(\mathcal{X}) . (1) \mathcal{A} is closed under direct summands. (2) If A=B\oplus C\in\mathcal{A} , then l_{\mathcal{X}}(A)=l_{\mathcal{X}}(B)+l_{\mathcal{X}}(C) . (3) If f:X\rightarrow M is a non-zero morphism in \mathcal{A} with X\in\mathcal{X} , then f is an inflation and l_{\mathcal{X}}(\mbox{\rm Cone}(f))=l_{\mathcal{X}}(M)-1 . Dually, if g:M\rightarrow X is a non-zero morphism in \mathcal{A} with X\in\mathcal{X} , then g is a deflation and l_{\mathcal{X}}(\mbox{\rm CoCone}(g))=l_{\mathcal{X}}(M)-1 . (4) \mathcal{X} is a proper semibrick if and only if l_{\mathcal{X}}(B)=l_{\mathcal{X}}(A)+l_{\mathcal{X}}(C) for any \mathbb{E} -triangle A\stackrel{{\scriptstyle}}{{\longrightarrow}}B\stackrel{{\scriptstyle}}{{\longrightarrow}}C\stackrel{{\scriptstyle}}{{\dashrightarrow}} in \mathcal{A} . Proof.. The first two statements can be found in [ 21 , Lemma 5.4] , (3) in [ 21 , Lemma 3.5 and Corollary 3.6] and (4) in [ 22 , Lemma 3.9] . ∎ Remark 2.4 . Let \mathcal{X} be a semibrick in \mathscr{C} and set \mathcal{A}:=\mbox{\rm{Filt}}_{\mathscr{C}}(\mathcal{X}) . By Proposition 2.2 and Proposition 2.3 , any object M\in\mathcal{A} has at least one non-zero inflation X\rightarrow M and one non-zero deflation M\rightarrow X^{\prime} with l_{\mathcal{X}}(X)=l_{\mathcal{X}}(X^{\prime})=1 . 3. Extriangulated length categories In this section, we introduce the concept of extriangulated length categories and establish their fundamental properties. We prove that extriangulated length categories correspond precisely to those categories arising from simple-minded systems. Notably, many significant categories in representation theory fall within the framework of extriangulated length categories, highlighting their broad applicability and importance. Throughout the remainder of this section, we fix an extriangulated category (\mathcal{A},\mathbb{E},\mathfrak{s}) as the basis for our discussion. 3.1. Definitions and basic properties We start with introducing the following notations. Definition 3.1 . We say that a map \Theta:\mbox{\rm Iso}\,(\mathcal{A})\rightarrow\mathbb{N} is a length function on \mathcal{A} if it satisfies the following conditions: (1) \Theta(X)=0 if and only if X\cong 0 . (2) For any \mathbb{E} -triangle X\stackrel{{\scriptstyle}}{{\longrightarrow}}L\stackrel{{\scriptstyle}}{{\longrightarrow}}M\stackrel{{\scriptstyle\delta}}{{\dashrightarrow}} in \mathcal{A} , we have \Theta(L)\leq\Theta(X)+\Theta(M) . In addition, if \delta=0 , then \Theta(L)=\Theta(X)+\Theta(M) . A length function \Theta is said to stable if it satisfies the condition \Theta(L)=\Theta(X)+\Theta(M) for any \mathbb{E} -triangle X\stackrel{{\scriptstyle}}{{\longrightarrow}}L\stackrel{{\scriptstyle}}{{\longrightarrow}}M\stackrel{{\scriptstyle}}{{\dashrightarrow}} in \mathcal{A} . We provide an elementary example of length functions. Example 3.2 . Let \mathcal{C} be a Krull-Schmidt category. One can regard \mathcal{C} as an exact category with a split exact structure, i.e., a conflation in \mathcal{C} is a kernel-cokernel pair of the form A\rightarrowtail A\oplus B\twoheadrightarrow B . Then we have a stable length function \Theta:\mbox{\rm Iso}\,(\mathcal{C})\rightarrow\mathbb{N} such that \Theta(M)=|M| for any M\in\mathcal{C} , where |M| denotes the number of non-isomorphic indecomposable direct summands of M . A semibrick \mathcal{X} in \mathcal{A} is called a simple-minded system if \mathcal{A}=\mbox{\rm{Filt}}_{\mathcal{A}}(\mathcal{X}) (see [ 21 , Definition 5.10] ). Simple-minded systems also serve as a source of length functions, as demonstrated by the following lemma. Lemma 3.3 . Let \mathcal{X} be a simple-minded system in \mathcal{A} . Then l_{\mathcal{X}} is a length function on \mathcal{A} . In addition, l_{\mathcal{X}} is stable if and only if \mathcal{X} is proper. Proof.. This immediately follows from Proposition 2.2 and Proposition 2.3 . ∎ Definition 3.4 . Let \Theta be a length function on \mathcal{A} . We say that an \mathbb{E} -triangle X\stackrel{{\scriptstyle x}}{{\longrightarrow}}L\stackrel{{\scriptstyle y}}{{\longrightarrow}}M\stackrel{{\scriptstyle\delta}}{{\dashrightarrow}} in \mathcal{A} is \Theta -stable (or stable when \Theta is clear from the context) if \Theta(L)=\Theta(X)+\Theta(M) . In this case, x is called a \Theta -inflation , y is called a \Theta -deflation and \delta is called a \Theta -extension . If there is no confusion, we depict x by the symbol X\rightarrowtail L and y by L\twoheadrightarrow M . We have the following useful characterizations of stable \mathbb{E} -triangles. Lemma 3.5 . Let \Theta be a length function on \mathcal{A} . For an \mathbb{E} -triangle \eta:~~X\stackrel{{\scriptstyle f}}{{\longrightarrow}}L\stackrel{{\scriptstyle g}}{{\longrightarrow}}M\dashrightarrow in \mathcal{A} , the following statements hold: (1) f is a \Theta -inflation if and only if g is a \Theta -deflation. (2) If \eta is stable, then f=0 if and only if X\cong 0 . (2^{\prime}) If \eta is stable, then g=0 if and only if M\cong 0 . (3) If \eta is stable, then f is an isomorphism if and only if \Theta(M)=0 . (3^{\prime}) If \eta is stable, then g is an isomorphism if and only if \Theta(X)=0 . Proof.. These follow from Definition 3.4 and [ 17 , Corollary 3.5] . ∎ Lemma 3.6 . Let \Theta be a length function on \mathcal{A} . Consider the following commutative diagrams of \mathbb{E} -triangles: (3.1) \begin{array}[]{l}\lx@xy@svg{\hbox{\raise 2.55554pt\hbox{\kern 6.75pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&&&\cr&&&\cr&&\cr&&\crcr}}}\ignorespaces{\hbox{\kern-6.75pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{A\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\hbox{\kern 1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}\hbox{\kern-1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}}}\ignorespaces{}{\hbox{\hbox{\kern 1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}\hbox{\kern-1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}}}{\hbox{\hbox{\kern 1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}\hbox{\kern-1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 13.40796pt\raise 6.1111pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.75pt\hbox{$\scriptstyle{f}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 30.75pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 30.75pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{B\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 37.7934pt\raise-14.77777pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-0.8264pt\hbox{$\scriptstyle{g}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 37.7934pt\raise-24.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces{\hbox{\kern 54.0104pt\raise 0.0pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 69.18402pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 69.18402pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{C\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 76.11526pt\raise-24.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces{\hbox{\kern 92.22012pt\raise 0.0pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 107.39374pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern 107.39374pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern-6.75pt\raise-31.27777pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces{\hbox{\kern 15.78212pt\raise-31.27777pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 30.81424pt\raise-31.27777pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 30.81424pt\raise-31.27777pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{E\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces{\hbox{\kern 34.7934pt\raise-46.91666pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 37.7934pt\raise-55.27777pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 52.53969pt\raise-25.9618pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.31596pt\hbox{$\scriptstyle{i}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 68.83679pt\raise-31.27777pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 68.83679pt\raise-31.27777pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{D\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 76.11526pt\raise-46.91666pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.63542pt\hbox{$\scriptstyle{j}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 76.11526pt\raise-55.27777pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces{\hbox{\kern 92.39374pt\raise-31.27777pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 107.39374pt\raise-31.27777pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern 107.39374pt\raise-31.27777pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern-3.0pt\raise-62.55554pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 30.88367pt\raise-62.55554pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{F\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces{\hbox{\kern 34.7934pt\raise-78.7222pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 37.7934pt\raise-89.33331pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\hbox{\kern 0.0pt\raise 1.0pt\hbox{\lx@xy@droprule}}\hbox{\kern 0.0pt\raise-1.0pt\hbox{\lx@xy@droprule}}}}\ignorespaces\ignorespaces{\hbox{\kern 53.95433pt\raise-62.55554pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces{}{\hbox{\hbox{\kern 0.0pt\raise 1.0pt\hbox{\lx@xy@droprule}}\hbox{\kern 0.0pt\raise-1.0pt\hbox{\lx@xy@droprule}}}}{\hbox{\hbox{\kern 0.0pt\raise 1.0pt\hbox{\lx@xy@droprule}}\hbox{\kern 0.0pt\raise-1.0pt\hbox{\lx@xy@droprule}}}}{\hbox{\kern 69.20554pt\raise-62.55554pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{F\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces{\hbox{\kern 73.11526pt\raise-78.7222pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 76.11526pt\raise-89.33331pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern-3.0pt\raise-91.05554pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 34.7934pt\raise-91.05554pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 73.11526pt\raise-91.05554pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\end{array} (1) If f and g are \Theta -inflations, then each \mathbb{E} -triangle in ( 3.1 ) is stable. (2) If i and j are \Theta -deflations, then each \mathbb{E} -triangle in ( 3.1 ) is stable. Proof.. We only prove the first statement and the second one can be shown similarly. First of all, by Lemma 3.5 , \Theta(A)+\Theta(C)=\Theta(B) and \Theta(B)+\Theta(F)=\Theta(E) . To see that A\stackrel{{\scriptstyle}}{{\longrightarrow}}E\stackrel{{\scriptstyle}}{{\longrightarrow}}D\dashrightarrow is stable, we calculate: \displaystyle\Theta(E) \displaystyle=\Theta(B)+\Theta(F)=\Theta(A)+\Theta(C)+\Theta(F)\geq\Theta(A)+\Theta(D)\geq\Theta(E). The preceding argument gives \Theta(A)+\Theta(D)=\Theta(E) . A similar calculation shows that \Theta(C)+\Theta(F)=\Theta(D) . ∎ Definition 3.7 . Let \Theta be a length function on \mathcal{A} . A morphism f:M\rightarrow N in \mathcal{A} is called \Theta -admissible if f admits a \Theta -decomposition (i_{f},X_{f},j_{f}) , i.e. there is a commutative diagram \textstyle{M\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{f} \scriptstyle{i_{f}} \textstyle{N} \textstyle{X_{f}\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{j_{f}} such that i_{f} is a \Theta -deflation and j_{f} is a \Theta -inflation. This motivates the following definition. Definition 3.8 . Let \Theta be a length function on \mathcal{A} . We say that ((\mathcal{A},\mathbb{E},\mathfrak{s}),\Theta) is an extriangulated length category , or simply a length category , if every morphism in \mathcal{A} is \Theta -admissible. For simplicity, we often write (\mathcal{A},\Theta) for ((\mathcal{A},\mathbb{E},\mathfrak{s}),\Theta) when \mathbb{E} and \mathfrak{s} are clear from the context. We will see later in Section 3.5 that abelian length categories and bounded derived categories of finite dimensional algebras with finite global dimension are length categories. The following theorem provides a method to obtaining length categories. Theorem 3.9 . Let \mathcal{X} be a simple-minded system in \mathcal{A} . Then (\mathcal{A},l_{\mathcal{X}}) is a length category. Proof.. By Lemma 3.3 , it suffices to show that each morphism in \mbox{\rm{Filt}}_{\mathcal{A}}(\mathcal{X}) is l_{\mathcal{X}} -admissible. To prove this, we take a morphism f:M\rightarrow N in \mbox{\rm{Filt}}_{\mathcal{A}}(\mathcal{X}) . The case of f=0 is trivial. Hence we always assume that f is a nonzero morphism. We proceed the proof by induction on l_{\mathcal{X}}(M)+l_{\mathcal{X}}(N) . The case of l_{\mathcal{X}}(M)=1 or l_{\mathcal{X}}(N)=1 follows from Proposition 2.3 . Now we consider the general case. By Proposition 2.2 , we have an l_{\mathcal{X}} -stable \mathbb{E} -triangle of the form X\stackrel{{\scriptstyle a}}{{\rightarrowtail}}M\stackrel{{\scriptstyle b}}{{\twoheadrightarrow}}\mbox{\rm Cone}(a)\stackrel{{\scriptstyle}}{{\dashrightarrow}} with X\in\mathcal{X} and l_{\mathcal{X}}(\mbox{\rm Cone}(a))=l_{\mathcal{X}}(M)-1 . We consider two cases: (\mathbf{Case~1}) fa=0 . In this case, there exists a morphism b^{\prime}:\mbox{\rm Cone}(a)\rightarrow N such that f=b^{\prime}b , i.e., we have a commutative diagram \textstyle{M\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{f} \scriptstyle{b} \textstyle{N} \textstyle{\mbox{\rm Cone}(a)\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{b^{\prime}} Note that b^{\prime}\neq 0 and l_{\mathcal{X}}(\mbox{\rm Cone}(a))=l_{\mathcal{X}}(M)-1 . By induction hypothesis, there exists an l_{\mathcal{X}} -decomposition (i_{a^{\prime}},X_{b^{\prime}},j_{b^{\prime}}) of b^{\prime} . Thus (i_{a^{\prime}}b,X_{b^{\prime}},j_{b^{\prime}}) is an l_{\mathcal{X}} -decomposition of f by Lemma 3.6 . (\mathbf{Case~2}) fa\neq 0 . In this case, by Proposition 2.3 , we have a commutative diagram (3.2) \begin{array}[]{l}\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 9.19968pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&&&\cr&&&\crcr}}}\ignorespaces{\hbox{\kern-9.19968pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{X~\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\hbox{\kern 1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}\hbox{\kern-1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}}}\ignorespaces{}{\hbox{\hbox{\kern 1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}\hbox{\kern-1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}}}{\hbox{\hbox{\kern 1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}\hbox{\kern-1.0pt\raise 0.0pt\hbox{\lx@xy@droprule}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 9.19968pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 16.03085pt\raise 4.50694pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.50694pt\hbox{$\scriptstyle{a}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 33.19968pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 33.19968pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{M\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 30.91142pt\raise-15.08334pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.75pt\hbox{$\scriptstyle{f}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 41.5955pt\raise-24.66669pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 58.72606pt\raise 5.43056pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.43056pt\hbox{$\scriptstyle{b}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 76.97746pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern-3.0pt\lower 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 76.97746pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{\mbox{\rm Cone}(a)\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 88.5996pt\raise-16.0pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.15279pt\hbox{$\scriptstyle{t}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 97.62044pt\raise-24.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces{\hbox{\kern 128.75647pt\raise 0.0pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 145.24954pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern 145.24954pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern-9.19968pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{X~\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 9.19968pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 14.10548pt\raise-25.8889pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.75pt\hbox{$\scriptstyle{fa}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 34.03302pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 34.03302pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{N\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 56.7878pt\raise-27.49306pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.50694pt\hbox{$\scriptstyle{c}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 73.99133pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern-3.0pt\lower 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 73.99133pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{\mbox{\rm Cone}(fa)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces{\hbox{\kern 130.24954pt\raise-32.0pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{$\scriptstyle{}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 145.24954pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern 145.24954pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\end{array} with X\in\mathcal{X} and l_{\mathcal{X}}(\mbox{\rm Cone}(fa))=l_{\mathcal{X}}(N)-1 . If t=0 , there exists a morphism f^{\prime}:M\rightarrow X such that f=faf^{\prime} . Note that X\in\mathcal{X} and f^{\prime}\neq 0 . Thus f^{\prime} is an l_{\mathcal{X}} -deflation and thus (f^{\prime},X,fa) is an l_{\mathcal{X}} -decomposition of f . Next, we focus on the case of t\neq 0 . Since l_{\mathcal{X}}(\mbox{\rm Cone}(a))+l_{\mathcal{X}}(\mbox{\rm Cone}(fa))1 . If X_{f}\in\Theta^{\prime}_{s} , then X_{f}\in\Theta_{s}\subseteq\Theta_{p-1} . Since P\in{{}^{\perp}}\Theta_{p-1} , we obtain i_{f}h=0 , a contradiction. Similarly, we can prove (2^{\prime}) . (3) Suppose that P\in\Theta^{\prime}_{p} or Q\in\Theta^{\prime}_{q} . By (2) and (2^{\prime}) , we have 2\leq s<\min\{m,n\} and X_{f}\notin\Theta^{\prime}_{s} . We have two cases: (\mathbf{Case~1}) . There exists a non-zero morphism a:X\rightarrow X_{f} such that X\in\Theta_{s-1} . (\mathbf{Case~2}) . There exists a non-zero morphism a:X_{f}\rightarrow X such that X\in\Theta_{s-1} . In both cases, we take a \Theta -decomposition (i_{a},X_{a},j_{a}) of a with \Theta(X_{a})=t . Clearly t\leq\Theta(X)\leq s-1 . In particular, either X\in\Theta_{1} or X\in\Theta^{\prime}_{\Theta(X)} . We divide the proof into two steps. \underline{\mathbf{Step~1}.} Suppose that P\in\Theta^{\prime}_{3} . We claim that Q\notin\Theta^{\prime}_{q} . Indeed, we have s=2 and then X\in\Theta_{1} . In (Case 1), by (1), a is a \Theta -inflation and so is gj_{f}a . If Q\in\Theta^{\prime}_{q} , then gj_{f}a=0 and thus X\cong 0 , this is a contradiction. In (Case 2), a is a \Theta -deflation and so is ai_{f}h . Since P\in\Theta^{\prime}_{3} , we get ai_{f}h=0 , this is a contradiction. Similarly, if Q\in\Theta^{\prime}_{3} , then P\notin\Theta^{\prime}_{p} . \underline{\mathbf{Step~2}.} We proceed the proof of (3) by induction on p+q . It suffices to consider the case of p+q\geq 6 with p\geq 3 and q\geq 3 . The case of p+q=6 follows from Step 2. Now consider the general case. Assume that P\in\Theta^{\prime}_{p} . In (Case 1), we obtain the following diagram: \textstyle{X\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{1} \textstyle{X\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{i_{a}} \scriptstyle{a} \textstyle{X_{f}~\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{gj_{f}} \textstyle{Q} \textstyle{X_{a}.\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{j_{a}} If t=1 , then X_{a}\in\Theta(X) . In this case, if Q\in\Theta^{\prime}_{q} , then gj_{f}j_{a}=0 , this is a contradiction. Assume that t>1 . If \Theta(X)=\Theta(X_{a}) , then X\cong X_{a} and hence gj_{f}a is a \Theta -inflation. Recall that X\in\Theta^{\prime}_{\Theta(X)} and \Theta(X)\leq s-10 . Since X_{f} is a brick and i_{f}\neq 0 , there exists a non-zero morphism s:M^{\prime}\rightarrow X_{f} such that sg=i_{f} . This implies that M^{\prime}\notin\mathcal{U} . Using Lemma 4.4 together with Lemma 3.6 , we have the following commutative diagram \textstyle{X_{f}~\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{K\vphantom{\big|}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{U\vphantom{\big|}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{X_{f}~\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{j_{f}} \textstyle{M\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{M^{\prime}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{V\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{V\ignorespaces\ignorespaces\ignorespaces\ignorespaces} where U\in\mathcal{U} and V\in\mathcal{U}^{\perp} . Since M\in\mathcal{U}^{\perp}\cap\mathcal{T} and X_{f},U\in\mathcal{T} , we have K\in\mathcal{U}^{\perp}\cap\mathcal{T} and V\in\mathcal{T} . Note that, \Theta(V)<\Theta(M) since j_{f}\neq 0 . If \Theta(K)=\Theta(M) , then K\cong M and thus M^{\prime}\cong U\in\mathcal{U} , this is a contradiction. This implies that 0<\Theta(V)<\Theta(M) . By induction hypothesis, we get K,V\in\mbox{\rm{Filt}}_{\Theta}({\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T})) and so is M . ∎ Lemma 4.15 . Let [\mathcal{U},\mathcal{T}] be an interval in \mbox{\rm tors}_{\Theta}(\mathcal{A}) . (1) We have \mathcal{T}=\mathcal{U}\ast(\mathcal{U}^{\perp}\cap\mathcal{T}) . (2) Let S\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) such that \Theta(S)\leq\Theta(S^{\prime}) for any S^{\prime}\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) . Then \mathcal{T}\cap{{}^{\perp}}S is a torsion class. Proof.. (1) It is enough to show that \mathcal{T}\subseteq\mathcal{U}\ast(\mathcal{U}^{\perp}\cap\mathcal{T}) . For any M\in\mathcal{T} , by Lemma 4.4 , there exists a stable \mathbb{E} -triangle U\stackrel{{\scriptstyle}}{{\rightarrowtail}}M\twoheadrightarrow U^{\prime}\stackrel{{\scriptstyle}}{{\dashrightarrow}} such that U\in\mathcal{U} and U^{\prime}\in\mathcal{U}^{\perp} . Since \mathcal{T} is a torsion class, we get U^{\prime}\in\mathcal{U}^{\perp}\cap\mathcal{T} . (2) It is obvious that \mathcal{T}\cap{{}^{\perp}}S is closed under extensions. We take a \Theta -deflation f:M\twoheadrightarrow N such that M\in\mathcal{T}\cap{{}^{\perp}}S . It suffices to prove N\in\mathcal{T}\cap{{}^{\perp}}S . Note that N\in\mathcal{T} since \mathcal{T} is a torsion class. If \Theta(N)=1 , then either N\in\mathcal{U} or N\in\mathcal{U}^{\perp} by Lemma 4.4 . For the former, we have N\in\mathcal{U}\subseteq{{}^{\perp}}S . For the latter, we have N\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) . Since \Theta(S)\leq\Theta(N)=1 , we have S\in\Theta_{1} . By Lemma 4.15 , {}^{\perp}S is a torsion class and thus N\in{{}^{\perp}}S . For the general case, let g:N\rightarrow S be a morphism in \mathcal{A} . We take a \Theta -decomposition (i_{g},X_{g},j_{g}) for g . If \Theta(N)=\Theta(X_{g}) , then g\cong j_{g} is a \Theta -inflation. Since S\in\mathcal{U}^{\perp} , we have N\in\mathcal{U}^{\perp}\cap\mathcal{T} . By Lemma 4.13 , there exists a brick S^{\prime}\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) such that \Theta(S^{\prime})\leq\Theta(N) . Since \Theta(S^{\prime})\leq\Theta(N)\leq\Theta(S)\leq\Theta(S^{\prime}) , we have N\cong S and hence f=0 , this is a contradiction. This implies \Theta(X_{g})<\Theta(N) . By induction hypothesis, we have X_{g}\in\mathcal{T}\cap{{}^{\perp}}S . In this case, we get j_{g}=0 and then g=0 . We conclude that N\in\mathcal{T}\cap{{}^{\perp}}S . ∎ Lemma 4.16 . If M\in\Theta_{1} , then {}^{\perp}M is a torsion class. Proof.. Clearly, it suffices to show that \mbox{\rm Fac}_{\Theta}(^{\perp}M)\subseteq{{}^{\perp}M} . To see this, we take a \Theta -deflation f:P\twoheadrightarrow Q with P\in{{}^{\perp}M} . If Q\notin{{}^{\perp}M} , then there exists a \Theta -deflation g:Q\twoheadrightarrow M by Lemma 3.11 (1). Since P\in{{}^{\perp}M} , we get fg=0 and thus M\cong 0 , this is a contradiction. This implies Q\in{{}^{\perp}M} and hence {}^{\perp}M is a torsion class. ∎ Remark 4.17 . For any S\in\mathcal{A} , {}^{\perp}S is not a torsion class in general. For instance, consider the length category (D^{b}(\Lambda),l_{\mathcal{X}}) as in Example 3.26 . We get easily P_{1}[-1]\in{{}^{\perp}P_{2}} and I_{2}[-1]\in\mbox{\rm Fac}_{l_{\mathcal{X}}}(P_{1}[-1]) . However, I_{2}[-1]\notin{{}^{\perp}P_{2}} and hence {{}^{\perp}P_{2}}\notin\mbox{\rm tors}_{l_{\mathcal{X}}}(D^{b}(\Lambda)) . For a set \mathcal{X} , we denote by \#\mathcal{X} its cardinality. Now we can state our first main result of this subsection. Theorem 4.18 . Let (\mathcal{A},\Theta) be a length category. Suppose that [\mathcal{U},\mathcal{T}]\subseteq\mbox{\rm tors}_{\Theta}(\mathcal{A}) . (1) We have \#[\mathcal{U},\mathcal{T}]=1 if and only if brick (\mathcal{U}^{\perp}\cap\mathcal{T})=\emptyset . (2) If \# brick (\mathcal{U}^{\perp}\cap\mathcal{T})=1 , then \#[\mathcal{U},\mathcal{T}]=2 . (3) If \#[\mathcal{U},\mathcal{T}]=2 , then there exists a brick S\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) satisfying the following conditions: (i) \Theta(S)\leq\Theta(S^{\prime}) for any S^{\prime}\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) . (ii) \mathcal{T}=\mathrm{T}_{\Theta}(\mathcal{U}\cup S) and \mathcal{U}=\mathcal{T}\cap{{}^{\perp}}S . (iii) S satisfies S\in\bigcap_{S^{\prime}\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T})}(\mbox{\rm Sub}_{\Theta}(S^{\prime})\cap\mbox{\rm Fac}_{\Theta}(S^{\prime})). Moreover, the brick S is unique up to isomorphism. Proof.. (1) If \#[\mathcal{U},\mathcal{T}]=1 , then brick (\mathcal{U}^{\perp}\cap\mathcal{T})=\emptyset by Lemma 4.14 . Conversely, suppose that brick (\mathcal{U}^{\perp}\cap\mathcal{T})=\emptyset . If \mathcal{U}\subsetneq\mathcal{T} , then there exists a non-zero object X\in\mathcal{U}^{\perp}\cap\mathcal{T} . Then by Lemma 4.13 there is a brick X^{\prime}\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) , this is a contradiction. (2) Suppose that \# brick (\mathcal{U}^{\perp}\cap\mathcal{T})=1 and \mathcal{U}\subseteq\mathcal{V}\subseteq\mathcal{T} . Clearly, \#{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{V})\leq 1 . If \#{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{V})=0 , then \mathcal{U}=\mathcal{V} by (1). If \#{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{V})=1 , then \mathcal{U}^{\perp}\cap\mathcal{V}=\mathcal{U}^{\perp}\cap\mathcal{T} by Lemma 4.14 . By using Lemma 4.15 (1), we have \mathcal{V}=\mathcal{U}\ast(\mathcal{U}^{\perp}\cap\mathcal{V})=\mathcal{T} . This shows that \#[\mathcal{U},\mathcal{T}]=2 . (3) For any S\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) , we set \mathcal{T}_{S}:=\mathrm{T}_{\Theta}(\mathcal{U}\cup S) and \mathcal{U}_{S}=\mathcal{T}\cap\mathrm{T}_{\Theta}(^{\perp}S) . Then one can check that \mathcal{U}\subsetneq\mathcal{T}_{S}\subseteq\mathcal{T} and \mathcal{U}\subseteq\mathcal{U}_{S}\subseteq\mathcal{T}_{S} . We can take S\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) such that \Theta(S)\leq\Theta(S^{\prime}) for any S^{\prime}\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) . Since \#[\mathcal{U},\mathcal{T}]=2 , we have \mathcal{T}=\mathcal{T}_{\mathcal{S}} . Thus, we get \mathcal{U}=\mathcal{T}\cap{{}^{\perp}}S by Lemma 4.15 . So we have shown (i) and (ii). For (iii), we take S^{\prime}\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) . Observe that \mathcal{T}=\mathcal{T}_{\mathcal{S^{\prime}}} . If S^{\prime}\in{{}^{\perp}}S , then \mathcal{T}=\mathcal{T}_{\mathcal{S}^{\prime}}=\mathrm{T}_{\Theta}(\mathcal{U}\cup S^{\prime})\subseteq\mathcal{T}\cap{{}^{\perp}}S=\mathcal{U} , this is a contradiction. Thus, there exists a non-zero morphism f:S^{\prime}\rightarrow S in \mathcal{A} . We take a \Theta -decomposition (i_{f},X_{f},j_{f}) of f . Clearly, X_{f}\in\mathcal{U}^{\perp}\cap\mathcal{T} . By Lemma 4.13 , there exists a brick S^{\prime\prime}\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) such that \Theta(S^{\prime\prime})\leq\Theta(X_{f})\leq\Theta(S) . The preceding argument implies \Theta(X_{f})=\Theta(S) and hence j_{f} is an isomorphism. Then f\cong i_{f} and thus S\in\mbox{\rm Fac}_{\Theta}(S^{\prime}) . In particular, if \Theta(S^{\prime})=\Theta(S) , then f is an isomorphism. This means that S is unique up to isomorphism. By symmetry, there exists a non-zero morphism g:S\rightarrow S^{\prime} with a \Theta -decomposition (i_{g},X_{g},j_{g}) . By using Lemma 4.13 , we have S\cong X_{g}\in\mbox{\rm Sub}_{\Theta}(S^{\prime}) . ∎ Theorem 4.18 (3) allows us to introduce brick labelling in the Hasse quiver of \mbox{\rm tors}_{\Theta}(\mathcal{A}) , which first appeared in [ 6 ] . Definition 4.19 . Let q:\mathcal{T}\rightarrow\mathcal{U} be an arrow in Hasse( \mbox{\rm tors}(\mathcal{A}) ). The brick label of q is the unique brick S\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) from Theorem 4.18 (3). In this case, we write \mathcal{T}\stackrel{{\scriptstyle S}}{{\rightarrow}}\mathcal{U} . Dually, for an arrow p:\mathcal{V}\rightarrow\mathcal{F} in Hasse( \mbox{\rm torf}_{\Theta}(\mathcal{A}) ), the brick label of p is the unique brick S\in{\rm brick}(^{\perp}\mathcal{F}\cap\mathcal{V}) (up to isomorphism) such that \Theta(S)\leq\Theta(S^{\prime}) for any S^{\prime}\in{\rm brick}(^{\perp}\mathcal{F}\cap\mathcal{V}) . In this case, we write \mathcal{V}\stackrel{{\scriptstyle S}}{{\rightarrow}}\mathcal{F} . We have the following relationship between \mbox{\rm tors}_{\Theta}(\mathcal{A}) and \mbox{\rm torf}_{\Theta}(\mathcal{A}) with respect to brick labelling. Proposition 4.20 . The isomorphism \displaystyle\mbox{\rm tors}_{\Theta}(\mathcal{A}) \displaystyle\stackrel{{\scriptstyle}}{{\longrightarrow}}\mbox{\rm torf}_{\Theta}(\mathcal{A})^{{\rm op}} \displaystyle\mathcal{T} \displaystyle\longmapsto\mathcal{T}^{\perp} of complete lattices preserves brick labelling. That is, \mathcal{T}\stackrel{{\scriptstyle S}}{{\rightarrow}}\mathcal{U} is an arrow in Hasse (\mbox{\rm tors}_{\Theta}(\mathcal{A})) if and only if \mathcal{U}^{\perp}\stackrel{{\scriptstyle S}}{{\rightarrow}}\mathcal{T}^{\perp} is an arrow in Hasse (\mbox{\rm torf}_{\Theta}(\mathcal{A})) . Proof.. The isomorphism follows immediately from Theorem 4.12 (3). In particular, p:\mathcal{T}\stackrel{{\scriptstyle}}{{\rightarrow}}\mathcal{U} is an arrow in Hasse (\mbox{\rm tors}_{\Theta}(\mathcal{A})) if and only if q:\mathcal{U}^{\perp}\stackrel{{\scriptstyle}}{{\rightarrow}}\mathcal{T}^{\perp} is an arrow in Hasse (\mbox{\rm torf}_{\Theta}(\mathcal{A})) . Note that \mathcal{U}^{\perp}\cap\mathcal{T}=\mathcal{U}^{\perp}\cap{{}^{\perp}}(\mathcal{T}^{\perp}) . Hence, the definition of brick label yields S_{p}=S_{q} . ∎ The labelled Hasse quiver of \mbox{\rm tors}_{\Theta}(\mathcal{A}) has the following global structure: Lemma 4.21 . Let X be an object in \Theta_{1} . (1) \mathcal{A}\stackrel{{\scriptstyle X}}{{\rightarrow}}{{}^{\perp}X} is an arrow in Hasse (\mbox{\rm tors}_{\Theta}(\mathcal{A})) . (1)^{\prime} \mathcal{A}\stackrel{{\scriptstyle X}}{{\rightarrow}}X{{}^{\perp}} is an arrow in Hasse (\mbox{\rm torf}_{\Theta}(\mathcal{A})) . (2) \mbox{\rm add}(X)\stackrel{{\scriptstyle X}}{{\rightarrow}}0 is an arrow in Hasse (\mbox{\rm tors}_{\Theta}(\mathcal{A})) . (2^{\prime}) \mbox{\rm add}(X)\stackrel{{\scriptstyle X}}{{\rightarrow}}0 is an arrow in Hasse (\mbox{\rm torf}_{\Theta}(\mathcal{A})) . Proof.. By Lemma 4.16 , {}^{\perp}X is a torsion class. Since {\rm brick}((^{\perp}X)^{\perp}\cap\mathcal{A})=X , \mathcal{A}\stackrel{{\scriptstyle X}}{{\rightarrow}}{{}^{\perp}X} is an arrow in Hasse (\mbox{\rm tors}_{\Theta}(\mathcal{A})) by Theorem 4.18 (2). Similarly, we can prove (1)^{\prime} . By using Proposition 4.20 , we get (2) and (2)^{\prime} . ∎ Proposition 4.22 . Let (\mathcal{A},\Theta) be a length wide category. The arrows in Hasse (\mbox{\rm tors}_{\Theta}(\mathcal{A})) starting at \mathcal{A} are precisely \mathcal{A}\stackrel{{\scriptstyle X}}{{\rightarrow}}{{}^{\perp}X} , where X runs through \Theta_{1} . Similarly, the arrows in Hasse (\mbox{\rm tors}_{\Theta}(\mathcal{A})) ending at 0 are precisely \mbox{\rm add}X\stackrel{{\scriptstyle X}}{{\rightarrow}}0 , where X runs through \Theta_{1} . Proof.. Let \mathcal{A}\stackrel{{\scriptstyle X}}{{\rightarrow}}\mathcal{U} be an arrow in Hasse (\mbox{\rm tors}_{\Theta}(\mathcal{A})) . Then Theorem 4.18 (3) implies that \mathcal{U}=\mathcal{A}\cap{{}^{\perp}}X={{}^{\perp}}X . For any 0\neq M\in{{}^{\perp}}X , there exists a \Theta -deflation M\twoheadrightarrow S for some S\in\Theta_{1} . Since {{}^{\perp}}X is a torsion class, we have S\in{{}^{\perp}}X . By Theorem 4.18 (3), this implies S\cong X since \Theta(X)\leq\Theta(S) . Thus, the first statement follows from Lemma 4.21 (1). Now, let \mathcal{T}\stackrel{{\scriptstyle X}}{{\rightarrow}}0 be an arrow in Hasse (\mbox{\rm tors}_{\Theta}(\mathcal{A})) . Then Theorem 4.18 (3) implies that \mathcal{T}=\mathrm{T}_{\Theta}(S) . By Proposition 4.8 , there exists a \Theta -deflation X\twoheadrightarrow S for some S\in\Theta_{1} . Since S\in\mathcal{T} , we have \Theta(X)\leq\Theta(S) and hence X\cong S . One can check that \mathcal{T}=\mbox{\rm add}X by Lemma 4.21 (2). By Lemma 4.21 (2) again, we obtain the second statement. ∎ Remark 4.23 . Let (\mathcal{A},\Theta) be a length category and set \Theta^{\prime}=l_{\Theta_{\infty}} . Then (\mathcal{A},\Theta^{\prime}) is a length wide category by Remark 3.16 . Hence, we can apply Proposition 4.22 to (\mathcal{A},\Theta^{\prime}) . To state our second main result, we introduce the following definition. Definition 4.24 . We say that a length category (\mathcal{A},\Theta) is standard if for any brick S in \mathcal{A} , we have \mbox{\rm Sub}_{\Theta}(S)\cap\mbox{\rm Fac}_{\Theta}(S)=\{0,S\} . Example 4.25 . (1) If \mathcal{A} is an exact length category, by [ 5 , Exercise 8.6] , we obtain that \mathcal{A} is an abelian length category. One can easily check that (\mathcal{A},l_{\rm sim(\mathcal{A})}) is standard. (2) Let \Lambda be a path algebra associated to a quiver of type A_{n} . Suppose that M,N\in\mbox{\rm ind}\,(D^{b}(\Lambda)) such that \mbox{\rm Hom}_{D^{b}(\Lambda)}(M,N)\neq 0 and \mbox{\rm Hom}_{D^{b}(\Lambda)}(N,M)\neq 0 . We claim that M\cong N . Note that D^{b}(\Lambda) is independent of the choice of orientation of A_{n} . We only need to consider the case 1\rightarrow 2\rightarrow\cdots\rightarrow n . By hypothesis, there exists an integer i such that M\cong X[i] and N\cong Y[i] for some indecomposable \Lambda -modules X,Y . For P\in\mbox{\rm ind}\,(\mbox{\rm{mod}}\,\Lambda) , we set H_{+}^{0}(P)=\{Q\in\mbox{\rm ind}\,(\mbox{\rm{mod}}\,\Lambda)~|~\mbox{\rm Hom}_{\Lambda}(P,Q)\neq 0\}. Since M\in H_{+}^{0}(N) and N\in H_{+}^{0}(M) , we obtain that M\cong N by [ 2 , Lemma 3.1] . Now, let \mathcal{X} be a semibrick in D^{b}(\Lambda) . As a consequence of Claim, (\mbox{\rm{Filt}}_{\mathcal{T}}(\mathcal{X}),l_{\mathcal{X}}) is a standard length category. In particular, (D^{b}(\Lambda),l_{\mathcal{S}}) is standard, where \mathcal{S} is a simple-minded system given in Example 3.25 . Let [\mathcal{U},\mathcal{T}] be an interval in \mbox{\rm tors}_{\Theta}(\mathcal{A}) . We define {\rm Mbrick}[\mathcal{U},\mathcal{T}]:=\{S\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T})~|~\mathcal{T}\cap{{}^{\perp}}S\in\mbox{\rm tors}_{\Theta}(\mathcal{A})\}, {\rm Jbrick}[\mathcal{U},\mathcal{T}]:=\{S\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T})~|~\mathrm{T}_{\Theta}(\mathcal{U}\cup S)\cap{{}^{\perp}}S\in\mbox{\rm tors}_{\Theta}(\mathcal{A})\}. In particular, we have {\rm Mbrick}[0,\mathcal{A}]=\{S\in{\rm brick}(\mathcal{A})~|~{{}^{\perp}}S\in\mbox{\rm tors}_{\Theta}(\mathcal{A})\} and {\rm Jbrick}[0,\mathcal{A}]=\{S\in{\rm brick}(\mathcal{A})~|~\mathrm{T}_{\Theta}(S)\cap{{}^{\perp}}S\in\mbox{\rm tors}_{\Theta}(\mathcal{A})\} . Then clearly we have {\rm Mbrick}[0,\mathcal{A}]\subseteq{\rm Jbrick}[0,\mathcal{A}] . Remark 4.26 . Suppose that \mathcal{A} is an exact category. We can check that {\rm Mbrick}[\mathcal{U},\mathcal{T}]={\rm Jbrick}[\mathcal{U},\mathcal{T}]={\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) . In particular, {\rm Mbrick}[0,\mathcal{A}]={\rm Jbrick}[0,\mathcal{A}]={\rm brick}(\mathcal{A}) . Let P=(P,\leq) be a partially ordered set. For any a\in P , we define a_{\ast}=\bigvee\{b\in P~|~bb\} . An element a\in P is completely join-irreducible if a=\bigvee L for some L\subseteq P implies a\in L . In this case, a_{\ast} is the unique element covered by a . If L is a finite set, then the converse is true. One can check that a\in P is completely join-irreducible if and only if a\neq a_{\ast} . In the same way, an element a\in P is completely meet-irreducible if a=\bigwedge L for some L\subseteq P implies a\in L , equivalently, a\neq a^{\ast} . We denote by j-irr {}^{\rm c}(P) (resp. m-irr {}^{\rm c}(P) ) the set of completely join-irreducible (resp. completely meet-irreducible) elements in P . Lemma 4.27 . Let (\mathcal{A},\Theta) be a standard length category with [\mathcal{U},\mathcal{T}]\subseteq\mbox{\rm tors}_{\Theta}(\mathcal{A}) . (1) Let S be a brick in \mathcal{U}^{\perp} and S^{\prime}\in\mathrm{T}_{\Theta}(\mathcal{U}\cup S) . Then S^{\prime}\in{{}^{\perp}}S or S\in\mbox{\rm Fac}_{\Theta}(S^{\prime}) . (2) If S\in{\rm Jbrick}[\mathcal{U},\mathcal{T}] , then \mathcal{T}_{S}:=\mathrm{T}_{\Theta}(\mathcal{U}\cup S)\in{\rm j}\text{-}{\rm irr}^{\rm c}([\mathcal{U},\mathcal{T}])~\text{and}~\mathcal{T}_{S^{\ast}}=\mathcal{T}_{S}\cap{{}^{\perp}}S. In particular, \mathcal{T}_{S}\rightarrow\mathcal{T}_{S^{\ast}} is an arrow in Hasse ([\mathcal{U},\mathcal{T}]) . (3) If S\in{\rm Mbrick}[\mathcal{U},\mathcal{T}] , then \mathcal{U}_{S}:=\mathcal{T}\cap{{}^{\perp}}S\in{\rm m}\text{-}{\rm irr}^{\rm c}([\mathcal{U},\mathcal{T}])~\text{and}~\mathcal{U}_{S^{\ast}}=\mathrm{T}_{\Theta}(\mathcal{U}_{S}\cup S). In particular, \mathcal{U}_{S^{\ast}}\rightarrow\mathcal{U}_{S} is an arrow in Hasse ([\mathcal{U},\mathcal{T}]) . Proof.. (1) If S^{\prime}\in{{}^{\perp}S} , then the statement is trivial, so we assume there exists a non-zero morphism f:S^{\prime}\rightarrow S . We take a \Theta -decomposition (i_{f},X_{f},j_{f}) for f . By Proposition 4.8 , we can check that \mathrm{T}_{\Theta}(\mathcal{U}\cup S)=\mbox{\rm{Filt}}_{\Theta}(\mathcal{U}\cup\mbox{\rm Fac}_{\Theta}(S)) . Since X_{f}\in\mathcal{U}^{\perp}\cap\mathrm{T}_{\Theta}(\mathcal{U}\cup S) , there exists a \Theta -inflation g:M\rightarrowtail X_{f} with M\in\mbox{\rm Fac}_{\Theta}(S) . Then j_{f}g is a \Theta -inflation and hence M\in\mbox{\rm Fac}_{\Theta}(S)\cap\mbox{\rm Sub}_{\Theta}(S) . Since (\mathcal{A},\Theta) is standard, we have M\cong S\cong X_{f} and then j_{f} is an isomorphism. This implies that f\cong i_{f} , thus S\in\mbox{\rm Fac}_{\Theta}(S^{\prime}) . (2) First of all, it is immediate that \mathcal{U}\subseteq\mathrm{T}_{S}\cap{{}^{\perp}}S\subsetneq\mathrm{T}_{S}\subseteq\mathcal{T} . Suppose that \mathcal{V}\in[\mathcal{U},\mathcal{T}_{S}] with \mathcal{V}\subsetneq\mathcal{T}_{S} . If \mathcal{V}\subseteq{{}^{\perp}}S , then \mathcal{V}\subseteq\mathrm{T}_{S}\cap{{}^{\perp}}S . Otherwise, there exists a non-zero morphism f:S^{\prime}\rightarrow S for some S^{\prime}\in\mathcal{V} . Thus (1) implies that S\in\mbox{\rm Fac}_{\Theta}(S^{\prime}) and then \mathcal{V}=\mathcal{T}_{S} , this is a contradiction. This shows that \mathcal{T}_{S^{\ast}}=\mathcal{T}_{S}\cap{{}^{\perp}}S and \mathcal{T}_{S}\in{\rm j}\text{-}{\rm irr}^{\rm c}([\mathcal{U},\mathcal{T}]) . Similarly, we can prove (3). ∎ The second main result of this subsection is the following. Theorem 4.28 . Let (\mathcal{A},\Theta) be a standard length category with [\mathcal{U},\mathcal{T}]\subseteq\mbox{\rm tors}_{\Theta}(\mathcal{A}) . (1) We have \#[\mathcal{U},\mathcal{T}]=2 if and only if \# brick (\mathcal{U}^{\perp}\cap\mathcal{T})=1 . In this case, there exists a unique brick S\in{\rm brick}(\mathcal{U}^{\perp}\cap\mathcal{T}) such that \mathcal{T}=\mathrm{T}_{\Theta}(\mathcal{U}\cup S),~\mathcal{U}=\mathcal{T}\cap{{}^{\perp}}S~\text{\rm and}~\mathcal{U}^{\perp}\cap\mathcal{T}=\mbox{\rm{Filt}}_{\Theta}(S). In particular, \mathcal{T}\stackrel{{\scriptstyle X}}{{\rightarrow}}\mathcal{U} is an arrow in Hasse ([\mathcal{U},\mathcal{T}]) . (2) There exists a bijection \mathcal{T}_{?}:{\rm Jbrick}[\mathcal{U},\mathcal{T}]\longrightarrow{\rm j}\text{-}{\rm irr}^{\rm c}([\mathcal{U},\mathcal{T}]);~S\longmapsto\mathrm{T}_{\Theta}(\mathcal{U}\cup S). In this case, \mathcal{T}_{S^{\ast}}=\mathcal{T}_{S}\cap{{}^{\perp}}S and \mathcal{T}_{S}\stackrel{{\scriptstyle S}}{{\rightarrow}}\mathcal{T}_{S^{\ast}} is the only arrow starting at \mathcal{T}_{S} . (3) There exists a bijection \mathcal{U}_{?}:{\rm Mbrick}[\mathcal{U},\mathcal{T}]\longrightarrow{\rm m}\text{-}{\rm irr}^{\rm c}([\mathcal{U},\mathcal{T}]);~S\longmapsto\mathcal{T}\cap{{}^{\perp}}S. In this case, \mathcal{U}_{S^{\ast}}=\mathrm{T}_{\Theta}(\mathcal{U}_{S}\cup S) and \mathcal{U}_{S^{\ast}}\stackrel{{\scriptstyle S}}{{\rightarrow}}\mathcal{U}_{S} is the only arrow ending at \mathcal{U}_{S} . (4) There exists two bijections {\rm Jbrick}[0,\mathcal{A}]\longrightarrow{\rm j}\text{-}{\rm irr}^{\rm c}(\mbox{\rm tors}_{\Theta}(\mathcal{A}));~S\longmapsto\mathrm{T}_{\Theta}(S), {\rm Mbrick}[0,\mathcal{A}]\longrightarrow{\rm m}\text{-}{\rm irr}^{\rm c}(\mbox{\rm tors}_{\Theta}(\mathcal{A}));~S\longmapsto{{}^{\perp}}S. In particular, {\rm m}\text{-}{\rm irr}^{\rm c}(\mbox{\rm tors}_{\Theta}(\mathcal{A})) is a subset of {\rm j}\text{-}{\rm irr}^{\rm c}(\mbox{\rm tors}_{\Theta}(\mathcal{A})). Proof.. (1) This follows immediately from Lemma 4.14 and Theorem 4.18 . (2) Firstly, we prove that the map \mathcal{T}_{?}:{\rm Jbrick}[\mathcal{U},\mathcal{T}]\rightarrow{\rm j}\text{-}{\rm irr}^{\rm c}([\mathcal{U},\mathcal{T}]) is well-defined. Take any S\in{\rm Jbrick}[\mathcal{U},\mathcal{T}] . By Lemma 4.27 , we have \mathcal{T}_{S}\in{\rm j}\text{-}{\rm irr}^{\rm c}([\mathcal{U},\mathcal{T}])~\text{and}~\mathcal{T}_{S^{\ast}}=\mathcal{T}_{S}\cap{{}^{\perp}}S. Suppose that there exists a brick S^{\prime}\in{\rm Jbrick}[\mathcal{U},\mathcal{T}] such that \mathcal{T}_{S^{\prime}}=\mathcal{T}_{S} . If S^{\prime}\in{{}^{\perp}S} , then \mathcal{T}_{S^{\prime}}\subseteq\mathcal{T}_{S^{\ast}}\subsetneq\mathcal{T}_{S} , this is a contradiction. Then there exists a non-zero morphism S^{\prime}\rightarrow S . Since S^{\prime}\in\mathcal{T}_{S} , we get S\in\mbox{\rm Fac}_{\Theta}(S^{\prime}) by Lemma 4.27 (1). By symmetry, we also have S^{\prime}\in\mbox{\rm Fac}_{\Theta}(S) and hence S\cong S^{\prime} . Next we show that \mathcal{T}_{?} is inverse. For any \mathcal{V}\in j-irr {}^{\rm c}([\mathcal{U},\mathcal{T}]) , we have \#[\mathcal{V}_{\ast},\mathcal{V}]=2 . By (1), there exists a unique brick S\in{\rm brick}(\mathcal{V}_{\ast}^{\perp}\cap\mathcal{V}) such that \mathcal{V}=\mathrm{T}_{\Theta}(\mathcal{V}_{\ast}\cup S),~\mathcal{V}_{\ast}=\mathcal{V}\cap{{}^{\perp}}S~\text{\rm and}~\mathcal{V}_{\ast}^{\perp}\cap\mathcal{V}=\mbox{\rm{Filt}}_{\mathcal{A}}(S). Note that \mathrm{T}_{\Theta}(\mathcal{U}\cup S)\subseteq\mathcal{V} and \mathrm{T}_{\Theta}(\mathcal{U}\cup S)\nsubseteq\mathcal{V}_{\ast} . This implies \mathcal{V}=\mathrm{T}_{\Theta}(\mathcal{U}\cup S) and hence \mathrm{T}_{\Theta}(\mathcal{U}\cup S)\cap{{}^{\perp}}S=\mathcal{V}_{\ast} is a torsion class. It follows that S\in{\rm Jbrick}[\mathcal{U},\mathcal{T}] . Then we can define a map \mathbf{L}:{\rm j}\text{-}{\rm irr}^{\rm c}([\mathcal{U},\mathcal{T}])\longrightarrow{\rm Jbrick}[\mathcal{U},\mathcal{T}];~\mathcal{V}\longmapsto S. Suppose that there exists a torsion class \mathcal{V}^{\prime}\in j-irr {}^{\rm c}([\mathcal{U},\mathcal{T}]) such that \mathbf{L}(\mathcal{V}^{\prime})=S . Then \mathcal{V}=\mathrm{T}_{\Theta}(\mathcal{U}\cup S)=\mathcal{V^{\prime}} . Hence, the map \mathbf{L} is well-defined. For any S\in{\rm Jbrick}[\mathcal{U},\mathcal{T}] , there exists an arrow \mathcal{T}_{S}\stackrel{{\scriptstyle S}}{{\rightarrow}}\mathcal{T}_{S^{\ast}} in Hasse ([\mathcal{U},\mathcal{T}]) such that {\rm brick}(\mathcal{T}_{S^{\ast}}^{\perp}\cap\mathcal{T}_{S})=S . This proves \mathbf{L}\mathcal{T}_{?}=\mbox{\rm id} . Conversely, we assume that \mathcal{V}\in j-irr {}^{\rm c}([\mathcal{U},\mathcal{T}]) . Then {\rm brick}(\mathcal{V}_{\ast}^{\perp}\cap\mathcal{V})=\{\mathbf{L}(\mathcal{V})\} and hence \mathcal{T}_{\mathbf{L}(\mathcal{V})}=\mathrm{T}_{\Theta}(\mathcal{U}\cup\mathbf{L}(V))\subseteq\mathcal{V} . Since \mathrm{T}_{\Theta}(\mathcal{U}\cup\mathbf{L}(\mathcal{V}))\nsubseteq\mathcal{V}\cap{{}^{\perp}}\mathbf{L}(\mathcal{V})=\mathcal{V}_{\ast} , we conclude that \mathcal{T}_{\mathbf{L}(\mathcal{V})}=\mathcal{V} . This proves \mathcal{T}_{?}\mathbf{L}=\mbox{\rm id} . (3) Using the analogous arguments used in (2), we can prove \mathcal{U}_{?} is a bijection. (4) It follows immediately from (2) and (3). ∎ Remark 4.29 . This is a generalization of the classical result [ 6 , Theorems 3.3, 3.4] of Demonet-Iyama-Reading-Reiten-Thomas for a finitely generated module category \mathcal{A} . We finish this subsection with a straightforward example illustrating Theorem 4.28 . Example 4.30 . Keep the notation used in Example 3.27 . By Example 4.25 (2), (\mathcal{A},l_{\mathcal{Y}}) is a standard length category. It is easy to check that {\rm Mbrick}[0,\mathcal{A}]={\rm Jbrick}[0,\mathcal{A}]={\rm brick}\mathcal{A}=\{P_{1},S_{1}[-1],S_{2}[-1],S_{3},P_{2},I_{2}[-1]\}. Let us list all 6 bricks, the corresponding completely meet-irreducible elements and completely join-irreducible elements as follows: brick completely join-irreducible element completely meet-irreducible element P_{1} \mbox{\rm add}(P_{1}) \mbox{\rm add}(\{S_{2}[-1],S_{1}[-1],I_{2}[-1]\})={{}^{\perp}P_{1}} S_{1}[-1] \mbox{\rm add}(S_{1}[-1]) \mbox{\rm add}(\{S_{2}[-1],P_{1},P_{2},S_{3}\})={{}^{\perp}S_{1}}[-1] S_{2}[-1] \mbox{\rm add}(S_{2}[-1]) \mbox{\rm add}(\{S_{1}[-1],P_{1},P_{2},S_{3},I_{2}[-1]\})={{}^{\perp}S_{2}}[-1] S_{3} \mbox{\rm add}(\{S_{3},P_{2},P_{1}\}) \mbox{\rm add}(\{P_{2},P_{1}\}) P_{2} \mbox{\rm add}(\{P_{2},P_{1}\}) \mbox{\rm add}(\{S_{2}[-1],P_{1}\}) I_{2}[-1] \mbox{\rm add}(\{I_{2}[-1],S_{1}[-1]\}) \mbox{\rm add}(\{S_{1}[-1],P_{1},P_{2},P_{3}\}) The labelled Hasse quiver of \mbox{\rm tors}_{l_{\mathcal{Y}}}(\mathcal{A}) is given by 0 {\color[rgb]{0.82,0.01,0.11}S}{{}_{2}}{\color[rgb]{0.82,0.01,0.11}[}{\color[rgb]{0.82,0.01,0.11}-1}{\color[rgb]{0.82,0.01,0.11}]} {\color[rgb]{0.82,0.01,0.11}S}{{}_{1}}{\color[rgb]{0.82,0.01,0.11}[}{\color[rgb]{0.82,0.01,0.11}-1}{\color[rgb]{0.82,0.01,0.11}]} S_{1}[-1] S_{2}[-1] P_{1} S_{2}[-1]\ P_{1} {\color[rgb]{0.82,0.01,0.11}P}{{}_{1}} S_{2}[-1] S_{2}[-1]\ S_{1}[-1]\ I_{2}[-1] S_{1}[-1] S_{2}[-1]\ P_{1}\ P_{2}\ S_{3} {\color[rgb]{0.82,0.01,0.11}P}{{}_{2}} \mathcal{A} {\color[rgb]{0.82,0.01,0.11}P}{{}_{1}} {\color[rgb]{0.82,0.01,0.11}S}{{}_{1}}{\color[rgb]{0.82,0.01,0.11}[}{\color[rgb]{0.82,0.01,0.11}-1}{\color[rgb]{0.82,0.01,0.11}]} S_{1}[-1]\ I_{2}[-1] {\color[rgb]{0.82,0.01,0.11}I}{{}_{2}}{\color[rgb]{0.82,0.01,0.11}[}{\color[rgb]{0.82,0.01,0.11}-1}{\color[rgb]{0.82,0.01,0.11}]} {\color[rgb]{0.82,0.01,0.11}S}{{}_{2}}{\color[rgb]{0.82,0.01,0.11}[}{\color[rgb]{0.82,0.01,0.11}-1}{\color[rgb]{0.82,0.01,0.11}]} S_{1}[-1]\ P_{1}\ P_{2} {\color[rgb]{0.82,0.01,0.11}P}{{}_{1}} S_{1}[-1]\ P_{1}\ P_{2}\ S_{3} {\color[rgb]{0.82,0.01,0.11}S}{{}_{3}} S_{1}[-1]\ P_{1}\ P_{2}\ S_{3}\ I_{2}[-1] {\color[rgb]{0.82,0.01,0.11}I}{{}_{2}}{\color[rgb]{0.82,0.01,0.11}[}{\color[rgb]{0.82,0.01,0.11}-1}{\color[rgb]{0.82,0.01,0.11}]} {\color[rgb]{0.82,0.01,0.11}S}{{}_{2}}{\color[rgb]{0.82,0.01,0.11}[}{\color[rgb]{0.82,0.01,0.11}-1}{\color[rgb]{0.82,0.01,0.11}]} P_{2}\ P_{1} {\color[rgb]{0.82,0.01,0.11}P}{{}_{2}} {\color[rgb]{0.82,0.01,0.11}S}{{}_{1}}{\color[rgb]{0.82,0.01,0.11}[}{\color[rgb]{0.82,0.01,0.11}-1}{\color[rgb]{0.82,0.01,0.11}]} {\color[rgb]{0.82,0.01,0.11}P}{{}_{1}} S_{3}\ P_{2}\ P_{1} {\color[rgb]{0.82,0.01,0.11}S}{{}_{3}} {\color[rgb]{0.82,0.01,0.11}S}{{}_{2}}{\color[rgb]{0.82,0.01,0.11}[}{\color[rgb]{0.82,0.01,0.11}-1}{\color[rgb]{0.82,0.01,0.11}]} {\color[rgb]{0.82,0.01,0.11}S}{{}_{1}}{\color[rgb]{0.82,0.01,0.11}[}{\color[rgb]{0.82,0.01,0.11}-1}{\color[rgb]{0.82,0.01,0.11}]} where each torsion class is represented by its non-zero indecomposable objects (unless 0 and \mathcal{A} ). 5. Support \tau -tilting subcategories in extriangulated length categories Our aim in this section is to establish a unified framework to study the \tau -tilting theory in length categories. In this section, we always assume that ((\mathcal{A},\mathbb{E},\mathfrak{s}),\Theta) is a length category. We start with the following notations. For a subcategory \mathcal{T} , we define {}^{\perp_{\Theta}}\mathcal{T}=\{M\in\mathcal{A}~|~\mathbb{E}_{\Theta}(M,T)=0~\text{for any}~T\in\mathcal{T}\}. Dually, we can define \mathcal{T}^{\perp_{\Theta}} . An object P\in\mathcal{T} is called \Theta -projective if P\in{{}^{\perp_{\Theta}}}\mathcal{T} . We denote the subcategory of \Theta -projective objects in \mathcal{T} by \mathcal{P}(\mathcal{T}) . Dually, the \Theta -injective objects are defined, and the full subcategory of \Theta -injective objects in \mathcal{T} is denoted by \mathcal{I}(\mathcal{T}) . We say \mathcal{A} has enough \Theta -projectives if for any M\in\mathcal{A} , there exists a \Theta -deflation P\twoheadrightarrow M satisfying P\in\mathcal{P}(\mathcal{A}) . Dually, we define that \mathscr{C} has enough \Theta -injectives . Remark 5.1 . Recall that ((\mathcal{A},\mathbb{E}_{\Theta},\mathfrak{s}_{\Theta}),\Theta) is a stable length category (see Proposition 3.22 ). It is easy to see that P\in\mathcal{P}(\mathcal{A}) (resp. I\in\mathcal{I}(\mathcal{A}) ) if and only if P (resp. I ) is projective (resp. injective) in (\mathcal{A},\mathbb{E}_{\Theta},\mathfrak{s}_{\Theta}) . We consider the following two conditions for a subcategory \mathcal{T} : ({\dagger}):\text{For any}~P\in\mathcal{P}(\mathcal{A}),~\text{there is a left}~\mathcal{T}\text{-approximation}~f:P\rightarrow T~\text{with}~T\in\mathcal{T}. ({\ddagger}):\text{For any}~P\in\mathcal{P}(\mathcal{A}),~\text{there is a left}~\mathcal{T}\text{-approximation}~f:P\rightarrow T~\text{with}~T\in\mathcal{P}(\mathcal{T}). Then we can introduce the following two definitions. Definition 5.2 . We say that a torsion class \mathcal{T} is support if it satisfies the condition ({\ddagger}) . Definition 5.3 . A subcategory \mathcal{T} of \mathcal{A} is a support \tau -tilting subcategory if it satisfies: (1) \mathcal{T}=\mathcal{P}({\mbox{\rm Fac}_{\Theta}(\mathcal{T})}) , (2) \mathcal{T} is \tau -rigid , i.e. \mathcal{T}\subseteq{{}^{\perp_{\Theta}}}{\mbox{\rm Fac}_{\Theta}(\mathcal{T})} , (3) \mathcal{T} satisfies the condition ({\dagger}) . We call an object T\in\mathcal{A} is a support \tau -tilting object if \mbox{\rm add}(T) is a support \tau -tilting subcategory. We denote by {\rm s}\tau - {\rm tilt(\mathcal{A}}) (resp. {\rm s} - {\rm tors_{\Theta}(\mathcal{A})} ) the set of support \tau -tilting subcategories (resp. support torsion classes) in \mathcal{A} . Remark 5.4 . When \Theta is not stable, our notion of support \tau -tilting subcategories differs from that in [ 23 , Definition 3.5] . Specifically, we only require that \mathbb{E}_{\Theta}(\mathcal{T},{\mbox{\rm Fac}_{\Theta}(\mathcal{T})})=0 , whereas [ 23 ] imposes the stronger condition \mathbb{E}(\mathcal{T},{{\rm Defl}(\mathcal{T})})=0 , where {\rm Defl}(\mathcal{T})=\{F\in\mathcal{A}\,\mid\mbox{there exists a deflation}~U\xrightarrow{~g~}F~\mbox{for some}~U\in\mathcal{T}\} . However, if \Theta is stable, then support \tau -tilting subcategories defined in [ 23 ] also qualify as support \tau -tilting subcategories in our framework (cf. [ 23 , Lemma 3.7] ). Remark 5.5 . Let \Lambda be a finite dimensional algebra. We set \Theta=l_{{\rm sim({\rm mod}\Lambda)}} . As stated in Example 3.10 , we regard (\mbox{\rm{mod}}\,\Lambda,\Theta) as a stable length category. In this case, stable \mathbb{E} -triangles are precisely short exact sequences of \mbox{\rm{mod}}\,\Lambda . (1) Let \mathcal{T} be a support torsion class. Then \mathcal{T} is functorially finite by Proposition 5.9 and [ 7 , Proposition 2.1] . Conversely, any functorially finite torsion class \mathcal{F} is support by [ 1 , Theorem 2.7] ) and [ 14 , Proposition 2.14] . This shows that support torsion classes in module category are just functorially finite torsion classes. (2) Let \mathcal{S} be a support \tau -tilting subcategory. By definition, there exists a right exact sequence (5.1) \Lambda\stackrel{{\scriptstyle f}}{{\rightarrow}}S_{1}\rightarrow S_{2}\rightarrow 0 such that f is a left \mathcal{S} -approximation with S_{1}\in\mathcal{S} . We claim that S_{2}\in\mathcal{S} . To see this, we take a \Theta -decomposition (i_{f},\mbox{\rm Im}(f),j_{f}) of f . Applying the functor \mbox{\rm Hom}(-,\mbox{\rm Fac}_{\Theta}(\mathcal{S})) to the sequence ( 5.1 ), we obtain an exact sequence \mbox{\rm Hom}(S_{1},-)|_{{\rm Fac}_{\Theta}(\mathcal{S})}\stackrel{{\scriptstyle f^{\ast}}}{{\rightarrow}}\mbox{\rm Hom}(\mbox{\rm Im}(f),-)|_{{\rm Fac}_{\Theta}(\mathcal{S})}\rightarrow{\rm Ext}^{1}(S_{2},-)|_{{\rm Fac}_{\Theta}(\mathcal{S})}\stackrel{{\scriptstyle}}{{\rightarrow}}{\rm Ext}^{1}(S_{1},-)|_{{\rm Fac}_{\Theta}(\mathcal{S})}. Note that f is a left \mbox{\rm Fac}_{\Theta}(\mathcal{S}) -approximation (see Lemma 5.7 ). One can check directly that j_{f} is also a left \mbox{\rm Fac}_{\Theta}(\mathcal{S}) -approximation. This implies that f^{\ast} is an epimorphism. On the other hand, since \mathcal{S} is \tau -rigid, we have {\rm Ext}^{1}(S_{1},-)|_{{\rm Fac}_{\Theta}(\mathcal{S})}=0 . Thus S_{2}\in\mathcal{P}({\mbox{\rm Fac}_{\Theta}(\mathcal{S})})=\mathcal{S} and the claim follows. Thus support \tau -tilting objects are precisely support \tau -tilting \Lambda -modules by [ 1 , Theorem 2.7] and [ 14 , Proposition 2.14] . We are ready to state the main theorem of this section. Theorem 5.6 . Let (\mathcal{A},\Theta) be a length category with enough \Theta -projectives. There is a bijection {\rm s}\tau\text{-}{\rm tilt(\mathcal{A})}\longrightarrow{\rm s}\text{-}\mbox{\rm tors}_{\Theta}(\mathcal{A}) given by \mathcal{S}\mapsto\mbox{\rm Fac}_{\Theta}(\mathcal{S}) with the inverse \mathcal{T}\mapsto\mathcal{P}(\mathcal{T}) . Before proving Theorem 5.6 , we need some preparations. Lemma 5.7 . Let \mathcal{T} be a \tau -rigid subcategory of \mathcal{A} . If f:P\rightarrow T is a left \mathcal{T} -approximation for some P\in\mathcal{P}(\mathcal{A}) , then f is a left \mbox{\rm Fac}_{\Theta}(\mathcal{T}) -approximation. Proof.. Clearly, T\in\mbox{\rm Fac}_{\Theta}(\mathcal{T}) . Suppose that g:P\rightarrow X is a morphism in \mathcal{A} such that X\in\mbox{\rm Fac}_{\Theta}(\mathcal{T}) . By hypothesis, there exists a \Theta -deflation h:T^{\prime}\twoheadrightarrow X for some T^{\prime}\in\mathcal{T} . Since P\in\mathcal{P}(\mathcal{A}) , there exists a morphism s:P\rightarrow T^{\prime} such that g=hs . Since f is a left \mathcal{T} -approximation, there exists a morphism t:T\rightarrow T^{\prime} such that s=tf . In particular, we have g=htf . This shows f is a left \mbox{\rm Fac}_{\Theta}(\mathcal{T}) -approximation. ∎ Proposition 5.8 . (1) Let \mathcal{T} be a \tau -rigid subcategory of \mathcal{A} . Then \mbox{\rm Fac}_{\Theta}(\mathcal{T})\in\mbox{\rm tors}_{\Theta}(\mathcal{A}) . (2) Let \mathcal{T} be a support \tau -tilting subcategory of \mathcal{A} . Then \mbox{\rm Fac}_{\Theta}(\mathcal{T})\in{\rm s}\text{-}\mbox{\rm tors}_{\Theta}(\mathcal{A}) . Proof.. (1) Let X_{1}\stackrel{{\scriptstyle}}{{\rightarrowtail}}X\twoheadrightarrow X_{2}\stackrel{{\scriptstyle}}{{\dashrightarrow}} be a stable \mathbb{E} -triangle with X_{1},X_{2}\in\mbox{\rm Fac}_{\Theta}(\mathcal{T}) . It suffices to prove X\subseteq\mbox{\rm Fac}_{\Theta}(\mathcal{T}) . Since X_{2}\in\mbox{\rm Fac}_{\Theta}(\mathcal{T}) , there exists a stable \mathbb{E} -triangle K\stackrel{{\scriptstyle}}{{\rightarrowtail}}M\twoheadrightarrow X_{2}\stackrel{{\scriptstyle}}{{\dashrightarrow}} with M\in\mathcal{T} . Consider the following commutative diagram \textstyle{K\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{a} \textstyle{K\vphantom{\big|}\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{c} \textstyle{X_{1}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{L\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{b} \textstyle{M\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{\delta} \textstyle{X_{1}\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{X\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{X_{2}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} . Since c is a \Theta -inflation, so is a by Lemma 3.20 . Thus b is a \Theta -deflation. By Lemma 3.21 , \delta\in\mathbb{E}_{\Theta}(M,X_{1})=0 and then L\cong X_{1}\oplus M\in\mbox{\rm Fac}(\mathcal{T}) . Thus X\in\mbox{\rm Fac}_{\Theta}(\mathcal{T}) and the assertion follows. (2) We note that Lemma 5.7 implies that \mbox{\rm Fac}_{\Theta}(\mathcal{T}) satisfies the condition ({\ddagger}) . Therefore, using (1), we deduce that \mbox{\rm Fac}_{\Theta}(\mathcal{T})\in{\rm s}\text{-}\mbox{\rm tors}_{\Theta}~\mathcal{A} . ∎ Proposition 5.9 . Suppose that (\mathcal{A},\Theta) has enough \Theta -projectives. If \mathcal{T}\in{\rm s}\text{-}\mbox{\rm tors}_{\Theta}(\mathcal{A}) , then \mathcal{T}=\mbox{\rm Fac}_{\Theta}(\mathcal{P}(\mathcal{T})) . In particular, we have \mathcal{P}(\mathcal{T})\in{\rm s}\tau - {\rm tilt(\mathcal{A})} . Proof.. The inclusion \mbox{\rm Fac}_{\Theta}(\mathcal{P}(\mathcal{T}))\subseteq\mathcal{T} is obvious. For each T\in\mathcal{T} , there is a \Theta -deflation f:P\twoheadrightarrow T with P\in\mathcal{P}(\mathcal{A}) . Since \mathcal{T} is support, there exists a left \mathcal{T} -approximation g:P\rightarrow T^{\prime} for some T^{\prime}\in\mathcal{P}(\mathcal{T}) . Therefore, there exists a morphism h:T^{\prime}\rightarrow T such that f=hg . By Lemma 3.20 , the morphism h is a \Theta -deflation and hence T\in\mbox{\rm Fac}_{\Theta}(\mathcal{P}(\mathcal{T})) . Then one can check that \mathcal{P}(\mathcal{T})\in{\rm s}\tau - {\rm tilt~\mathcal{A}} . ∎ Now we are ready to prove Theorem 5.6 . Proof of Theorem 5.6 . For a support \tau -tilting subcategory \mathcal{S} of \mathcal{A} , we have \mbox{\rm Fac}_{\Theta}(\mathcal{S})\in{\rm s}\text{-}\mbox{\rm tors}_{\Theta}(\mathcal{A}) by Proposition 5.8 . Note that we have \mathcal{S}=\mathcal{P}({\mbox{\rm Fac}_{\Theta}(\mathcal{S})}) . Conversely, for a given support torsion class \mathcal{T} , we have \mathcal{P}(\mathcal{T})\in{\rm s}\tau - {\rm tilt(\mathcal{A})} and \mathcal{T}=\mbox{\rm Fac}_{\Theta}(\mathcal{P}(\mathcal{T})) by Proposition 5.9 . \square Let \Lambda be a finite dimensional algebra. For a subcategory \mathcal{T} of \mbox{\rm{mod}}\,\Lambda , we denote by \mathrm{P}(\mathcal{T}) the direct sum of one copy of each of the indecomposable Ext-projective objects in \mathcal{T} up to isomorphism. We clearly have \mathcal{P}(\mathcal{T})=\mbox{\rm add}(\mathrm{P}(\mathcal{T})) . As a special case of Theorem 5.6 , we recover the following well-known fact in classical \tau -tilting theory. Corollary 5.10 . ( [ 1 , Theorem 2.7] ) Let \Lambda be a finite dimensional k -algebra over an algebraically closed field k . There is a bijection {\rm s}\tau\text{-}{\rm tilt~\Lambda}\longrightarrow{\rm f}\text{-}\mbox{\rm tors}~\Lambda given by M\mapsto\mbox{\rm Fac}(M) with the inverse \mathcal{T}\mapsto\mathrm{P}(\mathcal{T}) . Proof.. The assertion immediately follows from Remark 5.5 and Theorem 5.6 . ∎ For \mathcal{P},\mathcal{Q}\in{\rm s}\tau\text{-}{\rm tilt~\mathcal{A}} , we write \mathcal{P}\leq\mathcal{Q} if \mbox{\rm Fac}_{\Theta}(\mathcal{P})\subseteq\mbox{\rm Fac}_{\Theta}(\mathcal{Q}) . Then we have a partial order on {\rm s}\tau\text{-}{\rm tilt~(\mathcal{A})} . Then the Hasse quiver of s \tau\text{-}{\rm tilt(\mathcal{A})} and its brick labelling have the following description. Corollary 5.11 . We have that Hasse ({\rm s}\tau\text{-}{\rm tilt(\mathcal{A})})\cong{\rm Hasse}({\rm s}\text{-}\mbox{\rm tors}_{\Theta}(\mathcal{A})) is a full subquiver of {\rm Hasse}(\mbox{\rm tors}_{\Theta}(\mathcal{A})) . In this case, the brick label of \mathcal{P}\rightarrow\mathcal{Q} in {\rm s}\tau\text{-}{\rm tilt~\mathcal{A}} is the same as the brick label of \mbox{\rm Fac}(\mathcal{P})\rightarrow\mbox{\rm Fac}(\mathcal{Q}) in {\rm Hasse}({\rm s}\text{-}\mbox{\rm tors}_{\Theta}(\mathcal{A})) . Proof.. This follows immediately from Theorem 5.6 . ∎ Dually, we can define support torsion-free classes and support \tau^{-1} -tilting subcategories as follows. Definition 5.12 . We say that a torsion-free class \mathcal{F} is support if for any I\in\mathcal{I}(\mathcal{A}) , there exists a right \mathcal{F} -approximation F\rightarrow I with F\in\mbox{\rm Sub}_{\Theta}(\mathcal{F}) . Definition 5.13 . A subcategory \mathcal{F} of \mathcal{A} is a support \tau^{-1} -tilting subcategory if it satisfies: (1) \mathcal{F}=\mathcal{I}({\mbox{\rm Sub}_{\Theta}(\mathcal{F})}) , (2) \mathcal{T} is \tau^{-1} -rigid , i.e. \mathcal{T}\subseteq{\mbox{\rm Sub}(\mathcal{T})}{{}^{\perp_{\Theta}}} . (3) For each I\in\mathcal{I}(\mathcal{A}) , there exists a right \mathcal{F} -approximation F\rightarrow I with F\in\mathcal{F} . We call an object T\in\mathcal{A} is a support \tau^{-1} -tilting object if \mbox{\rm add}(T) is a support \tau^{-1} -tilting subcategory. We denote by {\rm s}\tau^{-1} - {\rm tilt(\mathcal{A})} (resp. {\rm s} - {\rm torf_{\Theta}(\mathcal{A})} ) the set of support \tau^{-1} -tilting subcategories (resp. support torsion-free classes) in \mathcal{A} . We obtain the dual version of Theorem 5.6 . Theorem 5.14 . Let (\mathcal{A},\Theta) be a length category with enough \Theta -injective objects. There is a bijection {\rm s}\tau^{-1}\text{-}{\rm tilt(\mathcal{A})}\longrightarrow{\rm s}\text{-}\mbox{\rm torf}_{\Theta}(\mathcal{A}) given by \mathcal{S}\mapsto\mbox{\rm Sub}_{\Theta}(\mathcal{S}) with the inverse \mathcal{F}\mapsto\mathcal{I}(\mathcal{F}) . Next we consider when torsion classes and torsion-free classes are support. For this purpose, we introduce the following notion. Definition 5.15 . We say that a length category (\mathcal{A},\Theta) is \tau -tilting finite if \mbox{\rm tors}_{\Theta}(\mathcal{A})={\rm s}\text{-}\mbox{\rm tors}_{\Theta}(\mathcal{A})~\text{and}~\mbox{\rm torf}_{\Theta}(\mathcal{A})={\rm s}\text{-}\mbox{\rm torf}_{\Theta}(\mathcal{A}). Remark 5.16 . (1) Suppose that \mathcal{A} is Hom-finite and there are only finitely many isomorphism classes of indecomposable objects. We can easily check that \mbox{\rm tors}_{\Theta}(\mathcal{A})={\rm s}\text{-}\mbox{\rm tors}_{\Theta}(\mathcal{A}) and \mbox{\rm torf}_{\Theta}(\mathcal{A})={\rm s}\text{-}\mbox{\rm torf}_{\Theta}(\mathcal{A}) . This means that (\mathcal{A},\Theta) is \tau -tilting finite. (2) We say a finite dimensional algebra \Lambda is \tau -tilting finite (cf. [ 6 , Definition 1.1] ) if there are only finitely many isomorphism classes of basic \tau -tilting \Lambda -modules. It was shown in [ 6 , Theorem 3.8] ) that \Lambda is \tau -tilting finite if and only if every torsion class in \mbox{\rm{mod}}\,\Lambda is functorially finite if and only if every torsion-free class in \mbox{\rm{mod}}\,\Lambda is functorially finite. Thus, our definition coincides with that in [ 6 , Definition 1.1] . When \mathcal{A} is \tau -tilting finite, we have the following bijection between support \tau -tilting subcategories and support \tau^{-1} -tilting subcategories. Theorem 5.17 . Suppose that \mathcal{A} has enough \Theta -projectives and \Theta -injectives. If \mathcal{A} is \tau -tilting finite, there is a bijection {\rm s}\tau\text{-}{\rm tilt(\mathcal{A})}\longrightarrow{\rm s}\tau^{-1}\text{-}{\rm tilt(\mathcal{A})} given by \mathcal{S}\mapsto\mathcal{I}(\mbox{\rm Fac}_{\Theta}(\mathcal{S})^{\perp}) with the inverse \mathcal{T}\mapsto\mathcal{P}(^{\perp}\mbox{\rm Sub}_{\Theta}(\mathcal{T})). Proof.. This follows from Proposition 4.6 , Theorem 5.6 and Theorem 5.14 . ∎ Remark 5.18 . In general, the converse of Theorem 5.17 is not true. For instance, if \mathcal{A}=\mbox{\rm{mod}}\,\Lambda for a finite dimensional algebra \Lambda , then we have a bijection {\rm s}\tau\text{-}{\rm tilt~\mathcal{A}}\rightarrow{\rm s}\tau^{-1}\text{-}{\rm tilt~\mathcal{A}} (cf. [ 1 , Section 2.4] ). But \mathcal{A} is \tau -tilting finite if and only if the set of bricks is finite (cf. [ 6 , Theorem 4.2] ). We end this section with an example illustrating some of our results. Example 5.19 . Keep the notation used in Example 3.27 . Recall that (\mathcal{A},l_{\mathcal{Y}}) is a stable length category. We know that its Auslander-Reiten quiver is as follows: (5.2) \begin{array}[]{l}\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 19.15274pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&&&&\cr&&&&\cr&&&&\crcr}}}\ignorespaces{\hbox{\kern-19.15274pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{{\color[rgb]{1,0,0}|}S_{2}[-1]\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\kern 46.1666pt\raise-24.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern 58.5416pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 103.93045pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{S_{1}[-1]\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\kern 161.23596pt\raise-24.87953pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern 169.88351pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 208.53107pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{P_{1}{\color[rgb]{1,0,0}|}}$}}}}}}}{\hbox{\kern-3.0pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 43.15274pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{{\color[rgb]{1,0,0}|}I_{2}[-1]\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\kern 105.8254pt\raise-8.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\kern 106.98604pt\raise-56.64372pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern 117.5832pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 161.23596pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{P_{2}{\color[rgb]{1,0,0}|}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\kern 208.53107pt\raise-7.88257pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern 217.17862pt\raise-32.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern-3.0pt\raise-64.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 58.5416pt\raise-64.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 106.98604pt\raise-64.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{{\color[rgb]{1,0,0}|}S_{3}{\color[rgb]{1,0,0}|}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\kern 161.23596pt\raise-39.13184pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern 169.88351pt\raise-64.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 217.17862pt\raise-64.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\end{array} The l_{\mathcal{Y}} -projective (resp. l_{\mathcal{Y}} -injective) objects are highlighted by a vertical line to their left (resp. to their right). By the diagram ( 5.2 ), it is easy to see that \mathcal{A} has enough l_{\mathcal{Y}} -projectives and l_{\mathcal{Y}} -injectives. Moreover, we observe that (\mathcal{A},l_{\mathcal{Y}}) is \tau -tilting finite by Remark 5.16 (1). Therefore, by Corollary 5.17 , we have bijections {\rm s}\tau\text{-}{\rm tilt(\mathcal{A})}\longleftrightarrow\mbox{\rm tor}_{l_{\mathcal{Y}}}(\mathcal{A})\longleftrightarrow\mbox{\rm torf}_{l_{\mathcal{Y}}}(\mathcal{A})\longleftrightarrow{\rm s}\tau^{-1}\text{-}{\rm tilt(\mathcal{A})}. In Table 1, we list all torsion pairs (\mathcal{T},\mathcal{F}) , the corresponding support \tau -tilting subcategories \mathcal{P}(\mathcal{T}) and support \tau^{-1} -tilting subcategories \mathcal{I}(\mathcal{F}) , where the black vertices are \mathcal{T} (resp. \mathcal{P}(\mathcal{T}) ), and the white vertices are \mathcal{F} (resp. \mathcal{I}(\mathcal{F}) ).