Examples and counterexamples for existence of categorical quotients (Q5925892)

If \(\mathcal A\) is a subcategory of the category \(\underline{PV}\) of complex algebraic prevarieties, and \(G\in \text{ Ob}({\mathcal A})\) is a complex algebraic group acting on \(X\in \text{ Ob}({\mathcal A})\) by \({\mathcal A}\)-morphisms, then \(p\in\text{ Mor}_{\mathcal A}(X,Y)\) is a categorical quotient if every \(G\)-invariant \({\mathcal A}\)-morphism factors uniquely through \(p\). If \(G\), \(X\) and the action are fixed, the existence of such a \(p\) remains very sensitive to the choice of \({\mathcal A}\). The paper under review illustrates this sensitivity by means of several examples, drawn from actions of subtori on complex toric varieties. In the categories \(\underline{TV}\) of toric varieties and \(\underline{TP}\) of toric prevarieties, categorical quotients always exist, as the authors have shown elsewhere. Here it is shown that the categorical quotient in \(\underline{TV}\) is also a categorical quotient in the category \(\underline{AV}\) of algebraic varieties if \(X\) is a toric variety and \(G\) is a subtorus of the big torus of codimension at most \(2\). On the other hand, innocuous-looking examples are given of \({\mathbb C}^*\)-actions on toric varieties having categorical quotients in \(\underline{AV}\) but not in \(\underline{PV}\), in \(\underline{PV}\) but not in \(\underline{AV}\), and in neither, as well as one where the categorical quotients in \(\underline{PV}\) and \(\underline{AV}\) both exists but are not the same.