Designs and partial geometries over finite fields (Q1126439): Difference between revisions

Let \(V= V(n,q)\) denote an \(n\)-dimensional vector space over the finite field \(GF(q)\). A collection \({\mathcal B}\) of \(k\)-dimensional subspaces of \(V\) such that any \(t\)-dimensional subspace of \(V\) is contained in exactly \(\lambda\) subspaces from \({\mathcal B}\) is called a \(t- (n,k, \lambda)\)-design over \(GF (q)\). The subspaces in \({\mathcal B}\) are called the blocks of the design. It is not hard to see that if \(t=2\), one gets an ordinary 2-design (BIBD) by taking as your varieties the 2-dimensional subspaces of \(V\) (that is, the points of the projective space \(PG(n - 1,q))\) and identifying each block with the set of 1-dimensional subspaces contained in it. The first nontrivial examples of designs over finite fields, as defined above, were given by the author in Geom. Dedicta 24, 237-242 (1987; Zbl 0627.51013), and since then numerous other examples have been created. However, no such designs have been constructed for \(t=2\) and \(\lambda =1\) (hence \(k \geq 3)\). In this setting nontrivial means that \({\mathcal B}\) does not consist of all the \(k\)-dimensional subspaces of \(V\). If \({\mathcal B}\) is a nontrivial \(2-(n, k,1)\)-design over \(GF(q)\) and we associate with it the corresponding BIBD on the points of \(\Sigma = PG (n-1, q)\) as above, then each point \(P\) of \(\Sigma\) induces a \((k-1)\)-spread \(S_p = \{W/P : W \in {\mathcal B}, P \in W\}\) on \(\Sigma/P\). If this spread is a geometric spread for all points \(P\) of \(\Sigma\), then \({\mathcal B}\) is called a geometric design. Recall that a projective spread \(S\) is geometric if whenever three distinct elements \(A_1\), \(A_2\), \(A_3\) are chosen from \(S\), then either \(A_3 \cap \langle A_1, A_2 \rangle = \emptyset\) or \(A_3 \subseteq \langle A_1, A_2 \rangle\). The thrust of this paper is to show that there are no geometric designs over a finite field. The proof involves studying a purported class of partial geometries whose blocks form a collection of maximal totally isotropic subspaces of a symplectic space.