Exceptional cycles for perfect complexes over gentle algebras (Q2204838)

Let \(\mathcal T\) be a \(k\)-linear triangulated category with Serre functor \(\mathsf S\). Then an exceptional cycle \((E_1,\ldots, E_n)\) with \(n \geq 2\) consists of exceptional objects \(E_i \in \mathcal T\) such that \(E_{i+1} = \mathsf S(E_i)[m_i]\) for some integer \(m_i\) and all \(i \in \{1,\ldots,n\} = \mathbb Z/n\mathbb Z\). In the case \(n=1\), one says that a spherical objects forms a (degenerate) exceptional cycle. This notion was introduced by \textit{N. Broomhead} et al. [Math. Z. 285, No. 1--2, 39--89 (2017; Zbl 1409.18012)], but appeared already under the name spherical sequences in the article [``Braid group actions on triangulated categories'', Möbius contest (2007)] by \textit{A. I. Efimov}, see also [``Groups generated by two twists along spherical sequences'', Preprint, \url{arXiv:1901.10904}] by \textit{Y. V. Volkov}. Such sequences are interesting, because the give rise to a twist functor which turns out to be an autoequivalence of \(\mathcal T\), thus generalising spherical twists. In the article cited above by Broomhead et al., exceptional cycles were completely classfied in the case of \(\mathcal T = \mathcal D^b(\Lambda(r,n,m)\operatorname{-mod})\), where \(\Lambda(r,n,m)\) is a derived discrete algebra, as studied by \textit{D. Vossieck} [J. Algebra 243, No. 1, 168--176 (2001; Zbl 1038.16010)]. Derived discrete algebras are a subclass of gentle algebras and the article under review gives a classification of exceptional cycles in \(K^b(A\operatorname{-proj})\) where \(A\) is an indecomposable finite-dimensional gentle \(k\)-algebra. For the main result (Theorem 1.4), the authors have to exclude the case that the underlying graph of \(A\) is of type \(A_3\), but in that case a classification of exceptional cycles is given by the same authors in [Acta Math. Sin., Engl. Ser. 36, No. 3, 207--223 (2020; Zbl 1436.18011)]. The proof of the main result goes roughly along the following lines. The first observation is that any object in an exceptional cycle has to lie at the mouth. Then for \(X\) an indecomposable object in a characteristic component \(C\) of \(K^b(A\operatorname{-proj})\) with AG-invariant \((n,m)\) gives rise to the exceptional \(n\)-cycle \((X, \tau X, \ldots, \tau^{n-1} X)\). Up to shift and cyclic permutation, this is the only such cycle in \(C\). For \(n \geq 2\), these turn out to be all exceptional \(n\)-cycles. For \(n = 1\), that is, when we are speaking about spherical objects, a string complex is spherical, if and only if it is at the mouth of characteristic component with AG-invariant \((1,m)\). Whereas for a band complex \(E\), the authors show that, if \(E\) is spherical, then it has to be at the mouth of a homogeneous component. But they give an example that the converse implication is not true in general.