The non-existence of ovoids in the dual polar space DW(5, \(q\)) (Q1421338): Difference between revisions

Let \({\mathcal D}\) be a dual polar space of rank \(n\geq 3\), that is, the dual of a nondegenerate polar space \({\mathcal P}\) of rank \(n\). An ovoid of \({\mathcal D}\) is a set \({\mathcal O}\) of points of \({\mathcal D}\) with the property that every line of \({\mathcal D}\) meets \({\mathcal O}\) in exactly one point. Question: When does the dual space \({\mathcal D}\) contain an ovoid? In this article the authors restrict themselves to rank \(n= 3\) and finite orders \(s\), \(t\), \(t\) with \(s> 1\). With these restrictions there are three possibilities for \({\mathcal D}\): (i) the dual \(DW(5, q)\) of the polar space \(W(5, q)\) associated with a nonsingular alternating form of \(V(6, q)\) (here \(s= t= q\)); (ii) the dual \(DQ(7, q)\) of the polar space \(Q^-(7, q)\) (here \(t= q\), \(s= q^2\)); (iii) the Hamming cube \(\text{Ham}(3, s)\) (here \(t= 1\)). It is known that \(\text{Ham}(3, s)\) contains ovoids. But no ovoid is known to exist in the other cases. Indeed, it has been conjectured that no ovoid exists when \(t> 1\). See \textit{A. Pasini} and \textit{S. Shpectorov} [J. Comb. Theory, Ser. A 94, No. 2, 276--288 (2001; Zbl 0989.51008)] for results supporting this conjecture. E. E. Shult and J. Thas independently proved that if the dual polar space \(DW(5, q)\) admits an ovoid then \(q\) is odd. The authors of the present article substantially improve this result by proving no \(DW(5, q)\) admits an ovoid. To prove this important result the authors choose a point \(\infty\) of \({\mathcal D}= DW(5, q)\), define \(\Delta\) to be the collinearity graph of \({\mathcal D}\), and let \(\Gamma\) be the graph induced by \(\Delta\) on \(\Delta_3(\infty)\), the set of points in \({\mathcal D}\) at maximal distance from \(\infty\). Several properties of the graph \(\Gamma\) are derived, and they are then used to find a contradiction to the assumption that \({\mathcal D}\) admits an ovoid.