A Topological Dimension

From MaRDI portal





Our first requirement for a dimension function is that it should be a mapping from a class of topological spaces, including the class of all separable metric spaces, to the set \(\{-1,0,1,2,\dots, \infty\}\). We list six properties that we believe a dimension function should have if it is to capture the intuitive concept of dimensionality. Fundamental property: The set \(\emptyset\) should have dimension --1, any countable set should have dimension 0, and, for \(n \geq 1\), the space \(\mathbb{R}^ n\) should have dimension \(n\). Invariance property: The dimension should be a topological invariant. Subspace property: The dimension of a subspace should not exceed the dimension of the space of which it is a subspace. Sum property: The dimension of a countable union of closed subsets of a space should be the supremum of the dimensions of the closed subsets. Addition property: The dimension of the union of two subsets of a space should not exceed the sum of the dimensions of the subsets by more than one. Injection property: The dimension of the continuous injective image of a space should not be less than the dimension of the original space. The well-known dimension functions ind, when restricted to separable metric spaces, and Ind and dim, when restricted to totally normal spaces (the totally normal spaces have a rather technical definition, they include all normal spaces whose open sets are countable unions of closed sets), have the first five properties, but do not have the injection property. The first author's dimension function dm [the author's thesis `On dimension' (Univ. College London 1986), the author, Acta Math. Hung. 55, No. 1/2, 83-95 (1990; Zbl 0715.54024)] may enjoy all six properties; it enjoys all the properties, with the possible exception of the addition property. The objects of this paper are to introduce a new dimension function \(d\), to show that it enjoys all the six properties, and to compare it with the functions ind, Ind, dim and dm. Our investigations are not exhaustive, and we leave many questions unanswered. We certainly make no claim that the new dimension function should replace the older dimensions, but we do claim that it deserves to be studied as an equal of the others.











This page was built for publication: A Topological Dimension

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4024921)