Defining topological projective spaces and topological Grassmann spaces (Q2638524)

Various definitions of a topological projective space have been proposed. In this paper a topological projective space is a finite-dimensional projective space \({\mathbf P}\), consisting of points, lines,..., and hyperplanes, where the point set \({\mathcal T}_ 0\) and the set of lines \({\mathcal T}_ 1\) carry topologies \(\tau_ 0\) and \(\tau_ 1\) respectively such that the geometric operations of joining two distinct points by a line \(h_ 0\), of meeting two intersecting lines s, and of meeting a line with any fixed hyperplane \(g_ H\) are continuous on their domains of definition (with respect to the inherited topologies and product topologies). It was shown by \textit{J.-P. Doignon} [Math. Z. 122, 57-60 (1971; Zbl 0212.523)] that this is equivalent to the notion of a topological projective space as given by \textit{J. Misfeld} [Abh. Math. Sem. Univ. Hamburg 32, 232-263 (1968; Zbl 0164.208)] in which more continuity axioms are required. According to Misfeld's results the topology \(\tau_ 1\) is uniquely determined by the topology \(\tau_ 0\). In desarguesian spaces this leads to a topological field (and vice versa). The standard definition of a topological projective plane only requires the continuity of \(h_ 0\) and s (the continuity of \(g_ H\) is a special case of the continuity of s). The author constructs an example of a projective space of dimension at least 3 where \({\mathcal T}_ 0\) and \({\mathcal T}_ 1\) are endowed with normal topologies such that \(h_ 0\) and s are continuous but in general \(g_ H\) is not. According to earlier results by the author [Abh. Math. Sem. Univ. Hamburg 59, 125-142 (1989; Zbl 0715.51006)] such a projective space cannot be locally compact and every plane must a topological projective plane. His example is a projective space \({\mathbf P}\) over the field of rationals \({\mathbb{Q}}\), where \({\mathbb{Q}}\) is endowed with its usual topology; however the topology space \({\mathbf P}\) has been altered: there is a distinguished point Z such that all points distinct from Z keep their usual neighbourhoods but neighbourhoods of Z are blown up to cones with respect to a point \(Z_ 1\) in the real closure of \({\mathbf P}\) (there the line joining Z and \(Z_ 1\) contains no rational point except Z). Similarly neighbourhoods of lines passing through Z are replaced by cone like extensions with respect to \(Z_ 1\) whereas neighbourhoods of other lines remain unchanged. This yields topologies \(\Theta_ 0\) and \(\Theta_ 1\) on \({\mathcal T}_ 0\) and \({\mathcal T}_ 1\) respectively with respect to which \(h_ 0\) and s become continuous. Further, \(\Theta_ 0\) and \(\Theta_ 1\) are coarser than standard topologies derived from the euclidean topology on \({\mathbb{Q}}\). However, \({\mathbf P}\) is not a topological projective space in the above definition since \(\Theta_ 0\) differs from the standard topology on \({\mathcal T}_ 0\) or because \(h_ 0\) is not open. Further any hyperplane not passing through Z and not containing \(Z_ 1\) in its real closure is not closed as a point set. Further it is shown that the axioms of a topological projective space can be replaced by the following: \(h_ 0\) and s are continuous and \(h_ 0\) is an open mapping, and for each non-incident pair consisting of a point P and a hyperplance H the projection from \({\mathcal T}_ 0\setminus \{P\}\) onto H is continuous (this is a specialization of \(g_ H)\). This can be used to give a similar substitution of the axioms of topological Grassmann spaces as defined by \textit{J. Misfeld}, \textit{G. Tallini}, and the author [Rend. Math. Appl., VII. Ser. 8, 223-240 (1988; Zbl 0691.51009)].