Rigid modules and ICE-closed subcategories in quiver representations (Q2068142)

This paper is devoted to the classification of ICE-closed subcategories, which is a new notion of this paper, of the category \(\operatorname{\mathsf{mod}} \Lambda\) of finitely generated modules over a given hereditary artin algebra \(\Lambda\). Here, the word ICE simply stands for images, cokernels and extensions, so by definition, ICE-closed subcategories are a large generalization of torsion classes and wide subcategories. The main prior researches of this paper are [\textit{C. Ingalls} and \textit{H. Thomas}, Compos. Math. 145, No. 6, 1533--1562 (2009; Zbl 1182.16012)] for hereditary algebras and its extension [\textit{T. Adachi} et al., Compos. Math. 150, No. 3, 415--452 (2014; Zbl 1330.16004)] for arbitrary Artin algebras. Both papers studied the functorially finite torsion classes in \(\operatorname{\mathsf{mod}} \Lambda\) by constructing one-to-one correspondences between these torsion classes and the support (\(\tau\)-)tilting modules. The author of this paper used the same strategy, and was able to obtain a bijection between the set \(\operatorname{\mathsf{ice_p}} \Lambda\) of ICE-closed subcategories with enough Ext-projectives and the set \(\operatorname{\mathsf{rigid}} \Lambda\) of basic rigid modules in \(\operatorname{\mathsf{mod}} \Lambda\) when \(\Lambda\) is a hereditary artin algebra (Theorem 2.3). It is the main result of the paper. This property is remarkable, since bijections between the set \(\operatorname{\mathsf{rigid}} \Lambda\) and the set of some kind of subcategories had not often been considered nevertheless rigid modules are an important notion in the representation theory. Section 2 is devoted to giving the main result with necessary definitions, and its proof is written in Section 3. The point is the subcategory \(\operatorname{\mathsf{cok}} U\) of cokernels between all morphisms in \(\operatorname{\mathsf{add}} U\) for each rigid module \(U\), and the proof of Theorem 2.3 was done by giving another description of \(\operatorname{\mathsf{cok}} U\). In Section 4, the author explains how the main result is related to the two prior researches above, including the Ingalls-Thomas bijection between the set of functorially finite wide subcategories and the set of functorially finite torsion classes for a hereditary artin algebra. Another proof of the main result is given in Section 5 from the point of view of exceptional sequences. As a consequence, the author was able to characterize ICE-closed subcategories as torsion classes of wide subcategories when \(\Lambda\) is a representation-finite artin algebras (Corollary 5.2 (4)). By the main result, the number of ICE subcategories is the same as that of basic rigid modules for the path algebra \(kQ\) of a Dynkin quiver \(Q\). In the appendix, the author determines this number with proving that the number does not depend on the orientations of the arrows of the quiver \(Q\) in a different way from the original one [\textit{R. Marsh} et al., Trans. Am. Math. Soc. 355, No. 10, 4171--4186 (2003; Zbl 1042.52007)].