On the perpendicular category of a partial tilting module (Q1178875)

The main result of the notable paper is Theorem A: Let \(A\) be a connected representation infinite hereditary algebra and \(T\) a tilting module without non-zero preinjective direct summands. Then the Auslander-Reiten quiver of \(End(T)\) has precisely one preprojective component and that component is the preprojective component of a concealed algebra which has the same representation type as \(A\). This result was well known for connected tame hereditary algebras, in the wild case even the existence of a preprojective component for the tilted algebra \(End(T)\) was an open problem. The proof uses induction on the number \(n\) of simple \(A\)-modules. The statement is trivial for \(n=2\) or if \(T\) is preprojective. The induction uses perpendicular categories in the sense of \textit{W. Geigle} and \textit{H. Lenzing} [J. Algebra 144, 273-343 (1991)] and \textit{A. Schofield} [Generic representation of a quiver (Preprint)]: If \(X\) is a partial tilting module, square-free with \(k\) indecomposable direct summands, then the right perpendicular category \(X^ \bot\) is equivalent to \(\hbox{mod-}C\) where \(C\) again is hereditary with \(n-k\) simple modules. The idea of the proof now is the following: If \(T\) is a tilting module as above, not preprojective, then \(T\) has a decomposition \(T=T_ 1\oplus T_ 2\) such that (1) \(T_ 2\) is regular in \(A\)-mod, (2) if \(T_ 2^ \bot \cong\hbox{mod-}C\) and \(H:T_ 2^ \bot \to\hbox{mod-}C\) denotes the equivalence, then \(C\) is connected and wild provided \(A\) is wild. (3) \(T_ 1\bar \in T_ 2^ \bot\) and \(H(T_ 1)\) is a \(C\)-tilting module without preinjective direct summands. (4) The preprojective component of \(End_ C(H(T_ 1))\) is the preprojective component of \(End(T)\). The arguments to find this decomposition \(T=T_ 1\oplus T_ 2\) are subtle. As a Corollary the author for example gets that every tilted algebra has preprojective as well as preinjective components.