Torsion pairs and quasi-abelian categories (Q2058369): Difference between revisions

\textit{S. E. Dickson} gave us a torsion theory for abelian categories in 1966 (see [Trans. Am. Math. Soc. 121, 223--235 (1966; Zbl 0138.01801)]). The paper under review transfers such a theory to quasi-abelian categories and gives a number of related results in abelian categories, pre-abelian and especially in quasi-abelian categories. An additive category is \textit{pre-abelian}, if every morphism has a kernel and a cokernel. A pre-abelian category is \textit{quasi-abelian}, if cokernels are stable under pullbacks. The strategy is to define and study torsion(free) classes in quasi-abelian categories by way of torsion(free) classes in the suitably associated abelian category. A \textit{torsion pair} in a quasi-abelian category \(\mathcal{Q}\) is an ordered pair (\(\mathcal{T}, \mathcal{F}\)) of full subcategories of \(\mathcal{Q}\) with the following properties: (a) \(\Hom_\mathcal{Q}(\mathcal{T},\mathcal{F})=0\); (b) For all objects \(M\) in \(\mathcal{Q}\), there exists an exact sequence \(0\longrightarrow{}_\mathcal{T}M\longrightarrow M\longrightarrow M_\mathcal{F}\longrightarrow 0\) where \({}_\mathcal{T}M\in\mathcal{T}\) and \(M_\mathcal{F}\in\mathcal{F}\). Naturally, \(\mathcal{T}\) is called a \textit{torsion class} and \(\mathcal{F}\), a \textit{torsion-free} class. The facilitating correspondence between some torsion pairs in an abelian category \(\mathcal{L}_\mathcal{Q}\) and torsion pairs in the related quasi-abelian category \(\mathcal{Q}\) is as follows: Let (\(\mathcal{C}, \mathcal{D}\)) and (\(\mathcal{C}', \mathcal{D}'\)) be torsion pairs in an Abelian category, with an assumption \(\mathcal{C}\subseteq\mathcal{C}'\). Then \(\mathcal{C}'\cap\mathcal{D}\) is quasi-abelian and there is an inclusion preserving bijection: \begin{gather*} \{(\mathcal{X}, \mathcal{Y}) \text{ torsion pair in } \mathcal{A}\mid \mathcal{C}\subseteq\mathcal{X}\subseteq\mathcal{C}' \} \longleftrightarrow \{(\mathcal{T}, \mathcal{F}) \text{ torsion pair in } \mathcal{C}'\cap\mathcal{D}\}; \\ (\mathcal{X},\mathcal{Y})\longmapsto (\mathcal{X}\cap\mathcal{D}, \mathcal{Y}\cap\mathcal{C}'); \ (\mathcal{C}*\mathcal{T}, \mathcal{F}*\mathcal{D}')\longmapsfrom (\mathcal{T},\mathcal{F}). \end{gather*} Here, the operation \(*\) between subcategories is defined to be the category of short exact sequences where the first non-zero link is in the first category and the third link is in the second category. One of the properties valid in abelian categories that hold in quasi-abelian categories \(\mathcal{Q}\) is as follows: (\(\mathcal{T}, \mathcal{F}\)) is a torsion pair, if and only if \(\mathcal{T}^\perp=\mathcal{F}\) and \(\mathcal{T}={}^\perp\mathcal{F}\). In addition: (a) Both \(\mathcal{T}, \mathcal{F}\) are quasi-abelian categories; (b) If \(\mathcal{Q}\) is Noetherian with respect to subobjects and if it has a strong projective generator, then \(\mathcal{L}_\mathcal{Q}\cong\bmod\Lambda\), for some \(\Lambda\). Furthermore, \(\mathcal{T}\) is functorially finite if an only if \(\mathcal{F}\) is functorially finite. One of the applications exhibited is that every chain of torsion classes that satisfies some finiteness conditions in a quasi-abelian category induces a Harder-Narasimhan filtration (see [\textit{G. Harder} and \textit{M. S. Narasimhan}, Math. Ann. 212, 215--248 (1975; Zbl 0324.14006)]) in each object, unique up to an isomorphism. This paper is a part of the author's doctoral dissertation study.