Generic inverse limits with set-valued functions (Q2673847)

Inverse limits (ILs) are a powerful tool to study metric continua. Replacing mappings by upper semi-continuous functions, generalized inverse limits (GILs) were introduced in [\textit{W. T. Ingram} and \textit{W. S. Mahavier}, Houston J. Math. 32, No. 1, 119--130 (2006; Zbl 1101.54015); \textit{W. S. Mahavier}, Topology Appl. 141, No. 1--3, 225--231 (2004; Zbl 1078.54021)]. Since then many authors have been studying the similarities and differences between ILs and GILs, and it has been shown that there are many properties of ILs that do not hold for GILs. Given an inverse limit \(X=\lim_{\leftarrow}\{X_{n},f_{n}\}\) of continua, a classical theorem says that if each \(\pi_{n}:X\rightarrow X_{n}\) is the natural projection and \(A\) is a closed subset of \(X\), then \(A\) is homeomorphic to lim\(_{\leftarrow}\{\pi_{n}(A),f_{n}\vert\pi_{n+1}(A)\}\). A simple example shows that the natural generalization of this theorem to GILs does not hold. In the paper under review the authors study GILs of the form \(X=\lim_{\leftarrow}\{I_{n},f_{n}\}\), where each \(I_{n}\) is homeomorphic to the interval \([0,1]\) and \(f_{n}\) is an upper semi-continuous function whose graph is a pseudo-arc contained in the square \(I_{n}\times I_{n+1}\), whose both projections are surjective. Given a nonempty closed subset \(A\) of \(X\), the authors define the natural GIL generated by \(A\) (taking the sets \(\pi_{n}(A)\)) and they call it the sub-generalized inverse limit, or SGIL, generated by \(A\). They show that: (a) \(X\) contains non-degenerate closed connected subsets, (b) no closed non-degenerate SGIL is connected. Then there exist many non-degenerate continua in \(X\) which are not SGIL, (c) in the space of GILs in the Hilbert cube, those GILs that are formed with pseudo-arcs bondings maps form a dense \(G_{\delta}\)-set.