Transfinite extension of dimension function \((m,n)\text{-}\dim\) (Q386880)

In [Mat. Vesn. 64, No. 4, 347--360 (2012; Zbl 1289.54104)] \textit{V. V. Fedorchuk} introduced \((m,n)\)-dim for finite values. In this paper the author extends this function to transfinite values using the ordinal function Ord. Before its definition we need some auxiliary notions. For a space \(X\) let \(u= (U_1,\dots, U_m)\) be an \(m\)-element cover of \(X\) and let \(\Phi= (F_1,\dots,F_m)\) be a family of closed subsets such that \(F_j\subset U_j\) for \(j=1,\dots,m\). Then \((\Phi, u)\) is called an \(m\)-pair in \(X\) and \(m(X)\) is the set of all \(m\)-pairs in \(X\). A closed set \(P\) is called an \(n\)-partition of an \(m\)-pair \((\Phi,u)\) iff \(P= X-\bigcup v\) for some family of open subsets \(v= (V_1,\dots,V_m)\) such that \(F_j\subset V_j\subset U_j\) for \(j=1,\dots,m\) and \(\text{ord\,}v\leq n\). A sequence \(\Sigma= \{(\Phi_i u_i)\in m(X):i=1,\dots, r\}\) is \(n\) iff there exist \(n\)-partitions \(P_i\) for \(i=1,\dots,r\) such that \(\bigcap_{i=1,\dots, r}P_i=\varnothing\), otherwise \(\Sigma\) is called \(n\)-essential. In the finite case the dimension function \((m,n)\)-dim was defined by \textit{V. V. Fedorchuk} [Valuations for dimension \((m,n)\)-dim, Mat. Vestnik. MSU (2013)] as follows: \((m,n)\)-\(\dim X=-1\) iff \(X=\varnothing\); \((m,n)\)-\(\dim X\leq k\), for some \(k= 0,1,\dots\), if every sequence \(\Sigma= \{(\Phi_i, u_i)\in m(X):i= 1,\dots, k+1\}\) is \(n\)-inessential in \(X\); \((m,n)\)-\(\dim X=\infty\) iff \((m,n)\)-\(\dim X> k\), for all integers \(k=-1,0,1,\dots\). The ordinal function Ord was defined by \textit{P. Borst} [Fundam. Math. 130, No. 1, 1--25 (1988; Zbl 0661.54035)]. Let \(T\subset\text{Fin\,}S\), where \(\text{Fin\,}S\) is the set of all finite sequences of elements of a set \(S\). For \(\sigma\in\{\varnothing\}\cup \text{Fin\,}S\) we put \(T^\sigma= \{\tau\in \text{Fin\,}S:\tau\cup\sigma\in T, \tau\cap\sigma= \varnothing\}\). Then the ordinal number \(\text{Ord\,}T\) is defined as follows: \(\text{Ord\,}T= 0\) iff \(T=\varnothing\); \(\text{Ord\,}T\leq\alpha\) iff \(\text{Ord\,}T^s<\alpha\) for every \(s\in X\); \(\text{Ord\,}T=\infty\) iff \(\text{Ord\,}T\leq\alpha\) holds for no ordinal number \(\alpha\). Putting \(M_{m(X)}= \{\Sigma\in \text{Fin\,}m(X)\): the sequence \(\Sigma\) is \(n\)-essential in \(X\}\) we define: \[ \text{tr-}(m,n)\text{-dim}= \text{Ord\,}M_{m(X)}. \] The transfinite dimension function \(\text{tr-}(m,n)\)-dim classifies al \((m,n)\)-\(C\)-spaces. By \textit{V. V. Fedorchuk} [Topology Appl. 157, No. 17, 2622--2634 (2010; Zbl 1211.54048)], a space is called a \((m,n)\)-\(C\)-spaces iff every sequence \(\{(\Phi_i,u_i)\in m(X)\): \(i=1,2,\dots\}\) contains a finite \(n\)-inessential subsequence. The dimension function tr-\((2,1)\)-\(C\)-dim coincides with the transfinite extension of dim introduced by Borst [loc. cit.]. The closed subspace theorem is proven and the following closed sum theorem: When \(X= X_1\cup X_2\) for two closed subsets \(X_1\) and \(X_2\) of \(X\). Then \[ \text{tr-}(m,n)\text{-}\dim X\leq\max(\text{tr-}(m,n)\text{-}\dim X_1,\;\text{tr-}(m,n)\text{-}\dim X_2)\oplus (\text{tr-}(m,n)\text{-}\dim X_1\cap X_2)+ 1). \] Where \(\oplus\) is the lower sum, i.e., when \(\alpha\) and \(\beta\) are ordinal numbers and \(\alpha= \alpha'+p\) and \(\beta= \beta'+ q\), where \(\alpha'\) and \(\beta'\) are limit ordinals and \(p\) and \(q\) are integers. Then if \(\alpha'>\beta': \alpha\oplus\beta= \alpha\); if \(\alpha'< \beta':\alpha\oplus \beta=\beta\); if \(\alpha'= \beta':\alpha\oplus \beta= \alpha+q= \beta+q\). The transfinite dimension function tr-\((m,n)\)-Ind is defined as follows by Fedorchuk [loc. cit. 2013]: tr-\((m,n)\)-Ind\,\(X=-1\) iff \(X=\varnothing\); tr-\((m,n)\)-Ind\,\(X\leq\alpha\) iff for every \((\Phi,u)\in m(X)\) there exists an \(n\)-partition \(P\) such that tr-\((m,n)\)-Ind\,\(P<\alpha\); tr-\((m,n)\)-Ind\,\(X=\infty\) iff tr-\((m,n)\)-Ind\,\(X\leq\alpha\) holds for no ordinal number \(\alpha\). The relation \(\dim\leq\text{Ind}\) is extended by proving tr-\((m,n)\)-\(\dim X\leq\text{tr-}(m,n)\text{-Ind\,}X\). In \textit{P. Borst} [Fundam. Math. 130, No. 2, 73--99 (1988; Zbl 0661.54036)] a compact metric space was constructed such that tr-\((m,n)\)-\(\dim X=\omega_0\) and tr-\((m,n)\)-\(\text{Ind\,} X= \omega_0+1\). The author asks the question to construct spaces with different tr-\((m,n)\)-\(\dim X\) and tr-\((m,n)\)-Ind\,\(X\) for other values of \(m\), \(n\) and \(\alpha\). The main question remains (see Fedorchuk, [loc. cit. 2010]) to prove the existence of spaces with different tr-\((m,n)\)-\(\dim X\) for different values of \(m\) and/or \(n\). In that case these functions really have their own role as transfinite dimensions.