A remark on the invariance of dimension (Q2355217)

The author, using ideas from \textit{W. Kulpa} [Am. Math. Mon. 104, No. 6, 545--550 (1997; Zbl 0891.47040)] and \textit{J. van Mill} [Infinite-dimensional topology. Prerequisites and introduction. North-Holland Mathematical Library, 43. Amsterdam etc.: North-Holland. (1989; Zbl 0663.57001)], gives a short proof of the theorem on invariance of dimension:\smallskip { Corollary 4.7.} \(I^n\simeq I^m\) if and only if \(n=m\). \smallskip Such a result can be proved from the Brouwer fixed-point theorem (given as Corollary A.2 of the current paper) which is often deduced applying algebraic topology, e.g., homotopy groups. But in this paper none of that is used. In Section 2, one finds a proof of the cubical Sperner lemma, Proposition 2.1, which is extracted from work of W. Kulpa. We will not state this result here since it requires too many definitions. It leads to:\smallskip { Theorem 3.1.} For \(i=1,\dots,n\), let \(H_i^+\) and \(H_i^-\) be closed subsets of \(I^n\) that contain the respective opposite \(i\)-faces of \(I^n\) and such that \(H_i^+\cup H_i^-=I^n\). Then \(\bigcap\{H_i^+\cap H_i^-\,|\,1\leq i\leq n\}\neq\emptyset\). \smallskip From this one gets:\smallskip { Corollary 4.2.} Whenever there are closed subsets \(C_1, \dots,C_n\) in \(I^n\) that separate the respective opposite faces of \(I^n\), then \(\bigcap\{C_i\,|\,1\leq i\leq n\}\neq \emptyset\). \smallskip This means that the set of pairs of the opposite faces of \(I^n\) is an essential family. In Proposition 4.3, it is shown that there cannot be an essential family in \(I^n\) consisting of more than \(n\) pairs of closed sets. Applying such a notion to a topological space \(X\) in general, one may define the separation-dimension \(s\)-dim\((X)\in\{-1,0,1,\dots,\infty\}\). For empty \(X\), one says that \(s\)-dim\((X)=-1\). Otherwise \(s\)-dim\((X)\) is the maximum cardinality of an essential family in \(X\) with the caveat that \(s\)-dim\((X)=\infty\) in case such a number is not finite. From the previous remarks, \(s\)-dim\((I^n)=n\), and since \(s\)-dim is a topological invariant, Corollary 4.7 is evidently true.