Interior and neighbourhood (Q2447138): Difference between revisions

Let \(P\) be a poset and denote by \(P^\uparrow\), the poset of up-closed subsets, with the reverse inclusion order. An \textit{interior operation}, \(i\), on \(P\) is a monotone map \(i:P\to P\) such that, for all \(x\in P\), \(i(x)\leq x\). A \textit{neighbourhood operation}, \(\nu\) on \(P\) is a monotone map, \(\nu:P\to P^\uparrow\) such that \(\nu(x)\subseteq \uparrow x\) for all \(x\in P\). Such an neighbourhood operation is said to be left adjoint if it admits a right adjoint, \(\nu_*\). These classes of operators are closely linked. Now let \(\mathbf{C}\) be a category with \(\mathcal{M}\) a fixed class of monomorphisms, which is assumed, in addition, to contain all isomorphisms, to be closed under composition with isomorphisms, to be pullback stable and to be closed under the formation of multiple pullbacks. This leads to notions of \(\mathcal{M}\)-subobject, interior and (left adjoint) neighbourhood operators of \(\mathcal{M}\) on \(\mathbf{C}\), these latter as certain types of lax natural transformation. After discussion of some examples, the more classical linkage mentioned above is extended to this more general setting. Some interesting consequences of this are presented and discussed.