Pseudopolar spaces of polar rank three (Q1086837)

Pseudopolar spaces are, roughly speaking, linear point-line incidence structures containing many polar spaces in the sense of \textit{F. Buekenhout} and \textit{E. Shult} [Geom. Dedicata 3, 155-170 (1974; Zbl 0286.50004)]. Call points collinear if they share a common line, and denote by \(p^{\perp}\) the set of points which are collinear with the point p. A linear incidence structure is called a polarized space of rank \(r+1\geq 3\) provided it satisfies (1) If p is a point and L a line, then p is collinear with precisely 0, 1 or all points of L; (2) If p,q are non- collinear points and \(| p^{\perp}\cap q^{\perp}| \geq 2\), then \(p^{\perp}\cap q^{\perp}\) is a polar space of rank r; (3) Certain connectedness and existence conditions, and a finite rank condition. A pseudopolar space of polar rank 3 is a polarized space of rank 3 which satisfies in addition (4) If H and K are maximal polar subspaces intersecting in at least a line, then \(H\cap K\) contains a plane if H or K is not of type \(D_ 3\), and in case H and K are both of type \(D_ 3\) there exist planes \(\alpha\) in H and \(\beta\) in K through a common line and contained in a singular subspace. The main result of this paper is a complete classification of pseudopolar spaces of rank 3 which satisfy some extra conditions. They turn out to be buildings of spherical type of one of the following types: \(A_{n,j}\) (2\(\leq j\leq n-1)\) or \(A_{2d-1,d}/<\sigma >\) with \(\sigma\) an involution, \(C_{n,n-2}\), \(D_{n,n-2}\), \(E_{7,4}\), \(E_{8,5}\), \(F_{4,1}\), or a quotient of one of the last five types, or locally a ''spread product'' of a regular spread in a classical generalized quadrangle and a spread in a projective space.