On homotopy types of Alexandroff spaces (Q2270170)

A topological space is an Alexandroff space if the intersection of any family of open sets is open. To every Alexandroff space \(X\) is associated a preordered space \((X,\leq)\) with \(x \leq y\) iff \(y \in \overline{\{x\}}\). The restriction to \(T_0\)-topological spaces gives an equivalence between the category of \(T_0\)-Alexandroff spaces and the category of posets (partially ordered spaces) and, following the author, we will not make a difference between a \(T_0\)-Alexandroff space and its associated poset. The purpose of this article is the extension to infinite posets of certain results concerning finite posets, well known from an article of \textit{R. E. Stong} [Trans. Am. Math. Soc. 123, 325--340 (1966; Zbl 0151.29502)]. The first results concern \(C(X,Y)\), the set of continuous maps from \(X\) to \(Y\), considered as a topological space with the compact-open topology; if \(Y\) is a poset, \(C(X,Y)\) is also a poset with \(f \leq g\) if and only if \(f(x)\leq g(x)\) for all \(x\) in \(X\). When \(X\) and \(Y\) are finite posets, the compact-open topology on \(C(X,Y)\) coincides with the topology given by the order. The author shows that this is no longer true when \(X\) and \(Y\) are infinite (and so, corrects a mistake in an article of \textit{F. G. Arenas} [Acta Math. Univ. Comen., New Ser. 68, No.~1, 17--25 (1999; Zbl 0944.54018)]). More precisely, he shows that the compact-open topology on \(C(X,Y)\) is weaker than the Alexandroff topology induced by order on \(C(X,Y)\) and gives a characterization of spaces \(X\) for which the two topologies coincide (i.e. \(C(X,Y)\) is also an Alexandroff space). For finite spaces \(X\) and \(Y\), homotopies in \(C(X,Y)\) are given by chains of maps; the author extends partially this result to the general case by proving that certain families \(\{f_{\alpha}:X\to Y\}_{\alpha \leq \gamma}\) induce a homotopy from \(f_0\) to \(f_{\gamma}\) when \(\gamma\) is a countable ordinal. As an application, he proves (among other equivalences) that a poset of height 1 has a point as a strong deformation retract if and only if it is connected and contains no crowns. A second series of results concerns cores and homotopy type of posets. In the finite case, a core is a poset with no irreducible points (a point is called irreducible if it has either a unique upper cover or a unique lower cover). If a finite poset is a core, then the identity is its unique self-map homotopic to the identity. A core of a finite poset \(X\) is a subposet of \(X\) which is a core and a strong deformation retract of \(X\). Every finite poset has a core which can be obtained by dismantling irreducible points. Two finite posets have the same homotopy type if and only if they have homeomorphic cores. A finite poset is contractible if and only if it has a point as its strong deformation retract. All these results are well known from the article of R. E. Stong (cited above). In the infinite case, the author studies \(\mathcal C\)-cores, i.e. posets \(X\) such that the identity is the only comparative retraction \(r : X \to X\) (\(r\) is a retraction if \(r^2=r\) and comparative if \(r(x)\) and \(x\) are comparable for all \(x\) in \(X\)). Following methods used by \textit{B. Li} and \textit{E. C. Milner} for studying the fixed point property in infinite posets [Order 9, No.~4, 321--331 (1992; Zbl 0782.06001)] (see also chapter 4 of [\textit{B.S.W. Schröder}, Ordered sets. An introduction. Boston, MA: Birkhäuser. (2003; Zbl 1010.06001)]), for an ordinal \(\gamma\) and an infinite poset \(X\), transfinite families of comparative retractions \(\{r_{\alpha} \}_{\alpha < \gamma}\) will give an analogue to the dismantling procedure in finite posets. When such a family is infinitely composable, the inductive procedure gives a poset \(X_{\gamma}\) and one says that \(X\) is \(\mathcal C\)-dismantlable to \(X_{\gamma}\); in the opposite case, it may happen that \(X\) is \(\mathcal C\)-dismantlable to the empty set. So, the general case differs drastically from the situation in the finite setting. For example, the one-way infinite fence is contractible but has no \(C\)-core and is \(C\)-dismantlable to the empty set; on the other hand, the two-way infinite fence (also contractible) is a \(C\)-core which admits self-maps homotopic to the identity and different from the identity. However, under certain conditions on the infinite posets \(X\) and \(Y\), the author proves similar statements to those cited above in the finite setting. He gives also some results about the transfinite procedure of \(\mathcal C\)-dismantlability.