Cardinal sequences of LCS spaces under GCH (Q636355)

Given a locally compact scattered (a.k.a. LCS) space \(X\), the Cantor-Bendixson hierarchy is defined as follows: \(X(0) = \{\)isolated points of \(X\}\); for \(\alpha > 0\), \(X(\alpha) = \{\)isolated points of \(X \setminus \bigcup_{\beta < \alpha}X_{\beta}\}\). The least \(\alpha\) for which \(X(\alpha) = 0\) is the height of \(X\) and is denoted \(\mathrm{ht} X\). A cardinal sequence \(f\) for a locally compact scattered space \(X\) is a function \(f: \mathrm{ht} X \rightarrow CARDS\) so that each \(f(\beta) = |X(\beta)|\). By a result of Juhasz and Weiss, cardinal sequences of length \(\leq \omega_1\) are completely characterized (basically: anything that ought to happen does happen). By results of Just and Baumgartner, there is no ZFC characterization above \(\omega_1\). But an earlier result of the authors shows that under GCH you can characterize the cardinal sequences of spaces of height \(< \omega_2\). The purpose of this paper is to explore what can happen for height \(\omega_2\) and above. Let \(\mathcal C(\alpha)\) be the collection of cardinal sequences for LCS spaces of height \(\alpha\), and let \(\mathcal C_{\lambda}(\alpha) = \{f \in \mathcal C(\alpha): f(0) = \lambda = \inf_{\beta < \alpha}|f(\beta)|\}\). The authors observe that if \(\lambda \geq \omega_2\) then \(f \in \mathcal C(\omega_2)\) iff \(f|_{\alpha} \in \mathcal C_{\lambda}(\alpha)\) for each \(\alpha < \omega_2\). What about \(\mathcal C_{\omega_1}(\omega_2)\)? This requires another definition: \(\mathcal D_{\omega}(\alpha) =\{f \in \{\omega, \omega_1\}^{\alpha}: (0) = \omega\}\). If \(\lambda > \omega\), then \(\mathcal D_{\lambda}(\alpha)= \{f \in \{\lambda, \lambda^+\}^{\alpha}: f(0) = \lambda\), and \(L = f^{-1}(\lambda)\) satisfies the following: if \(\beta \in L\) then \(\beta + 1 \in L \cup \{\alpha\}\); and if \(A \subset L\) where \(|A| = \kappa < \lambda\) then \(\sup A \in L \cup \{\alpha\}\). In a previous paper, the authors proved that, under GCH, \(\mathcal C_{\omega_1}(\delta) \subseteq \mathcal D_{\omega_1}(\delta)\) for \(\delta < \omega_2\). \textbf{Theorem.} Assume GCH. Let \(\omega_2 \leq \alpha < \omega_3\). There is a forcing \(\mathbb P\) which preserves GCH so that \(V^{\mathbb P} \Vdash \mathcal C_{\omega_1}(\delta) = \mathcal D_{\omega_1}(\delta)\) for \(\delta < \omega_3\). This is a special case of a more general theorem in the paper: \textbf{Theorem 2.} Given an uncountable regular cardinal \(\kappa\) , \(\kappa^{< \kappa} = \kappa\) and \(2^{\kappa} = \kappa^+\), then if \(\delta < \kappa^{++}\) there is a \(\kappa\)-complete \(\kappa^+\)-cc partial order \(\mathbb P\) with \(|\mathbb P| = \kappa^+\) so that \(V^{\mathbb P} \Vdash \mathcal C_{\kappa}(\delta) = \mathcal D_{\kappa}(\delta)\). The proof proceeds by using \(\mathbb P\) to force the existence of a \(\mathcal C_{\kappa}(\delta)\) universal LCS space, i.e. a space \(X\) whose cardinal sequence lies in \(\mathcal C_{\kappa}(\delta)\) and satisfying: \(\mathcal C_{\kappa}(\delta) \subset \{\)cardinal sequences of open subspaces of \(X\}\). Finally, the authors show that a consistent (with GCH) characterization of \(\mathcal C(\alpha)\) is possible for each \(\alpha\): \textbf{Theorem 3.} Suppose \(f \in V \Vdash GCH, \mathrm{dom} f = \alpha\). Then there is a cardinal-preserving GCH-preserving forcing \(\mathbb P\) in which \(V^{\mathbb P} \Vdash (f \in \mathcal C_{\alpha}\) iff there is \(n < \omega\), a decreasing sequence \(\lambda_0 >\cdots > \lambda_{n-1}\), and \(\alpha_0, \cdots \alpha_{n-1}\) with \(\alpha = \alpha_0 + \cdots \alpha_{n-1}\) so that \(f = f_0 \frown f_1\cdots \frown f_{n-1}\) where each \(f_i \in \mathcal D_{\lambda_i}(\alpha_i))\). The proof basically proceeds by finite iteration of the forcings provided by Theorem 2. A number of questions remain, in particular whether the characterization of Theorem 3 can be uniform, i.e., under GCH can you prove: \(\forall \alpha \;f \in \mathcal C_{\alpha}\) iff there is \(n < \omega\), a decreasing sequence \(\lambda_0 >D\cdots > \lambda_{n-1}\) and \(\alpha_0,\cdots \alpha_{n-1}\) with \(\alpha = \alpha_0 + \cdots \alpha_{n-1}\) so that \(f = f_0 \frown f_1\cdots \frown f_{n-1}\) where each \(f_i \in \mathcal D_{\lambda_i}(\alpha_i)\)?