A good theory of ideals in regular multi-pointed categories (Q456864): Difference between revisions

The authors introduce multi-pointed categories as a unifying generalization of the pointed and non-pointed contexts (in particular for the treatment of kernels and kernel pairs). Those categories can be defined as the ones enriched in the appropriately defined symmetric closed monoidal category Set\(_\circ\) of the multi-pointed sets, or, equivalently, as the pairs (\(\mathbb{C}, \mathcal{N}\)) where \(\mathcal{N}\) is a class of morphisms (called the null morphisms) in the category \(\mathbb{C}\) satisfying the condition [\(f\in\mathcal{N}\) or \(g\in\mathcal{N} \Rightarrow fg\in\mathcal{N}\)]. \(\mathcal{N}(X,Y)\) being the set of null morphisms from \(X\) to \(Y\), the non-pointed (``total context'') case corresponds to \(\mathcal{N}(X,Y) = \mathbb{C}(X,Y)\) and the pointed one to \(\mathcal{N}(X,Y) = \{ \ast \}\). Then a star is defined as a pair of parallel morphisms one of them being in \(\mathcal{N}\), and a kernel star of \(f\) is a star [\(\sigma_1,\sigma_2\)] such that \(f\sigma_1 = f\sigma_2\), and which is universal for this property. This (essentially) boils down to kernel pairs and kernels in the non-pointed and pointed contexts respectively. Let \(\mathbb{C}\) be a star-regular category, i.e., a regular multi-pointed category where every regular epimorphism is a coequalizer of a star, and where every morphism \(f : X \to Y\) has a \(\mathcal{N}\)-kernel (i.e., some \(k : K \to X\) such that \(fk \in \mathcal{N}\) and universal with this property). Define an ideal as an image of a kernel star (through the (Regular epi, Monic star) factorization of stars). We say that \(\mathbb{C}\) has a good theory of ideals if the mapping [\(\sigma_1,\sigma_2] \mapsto [\sigma_1k,\sigma_2k\)]*, where \(k\) is the \(\mathcal{N}\)-kernel of \(\sigma_1\), defines a bijection from kernel pairs to ideals. In the pointed context, these are essentially the ideal determined categories of [\textit{G. Janelidze} et al., Cah. Top. Géom. Différ. Catég. 51, No. 2, 115--125 (2010; Zbl 1208.18001)], and in the total context, they are the Barr exact Goursat categories (see [\textit{A. Carboni} et al., J. Pure Appl. Algebra 69, No. 3, 271--284 (1991; Zbl 0722.18005)]). Although other results are also obtained, the authors state that the main goal of the paper is to ``allow a simultaneous treatment of ideal determined and Barr exact Goursat categories. In particular, [obtaining] unified formulas for the join of two ideals [and] a unified proof of the fact that in both cases kernel pairs form modular lattices''. In addition to its intrinsic value, the paper gives useful references for and comparisons with some of the generalizations of the concepts above which were considered in the past either by the universal algebraists or the category theorists.