On the existence of (l,m,n)-spreads (Q2563487)

Let \(\Sigma= PG(n,K)\) denote projective \(n\)-space over the field \(K\). If \(m\) and \(l\) are integers with \(n>m> l\geq 0\), an \((l,m,n)\)-spread over \(K\) is defined to be a family of projective subspaces of \(\Sigma\) of dimension \(m\) that partition the family of all projective subspaces of dimension \(l\). Other authors have called such a family an \((l+1,m)\)-system or an \((l,m)\)-Faserung. It is easy to see that a necessary condition for the existence of such a family is the inequality \(l\geq 2m- n+1\), from which it follows that necessarily \(m<n-1\). For infinite fields \(K\), \((l,m,n)\)-spreads over \(K\) are known to exist [see \textit{A. Beutelspacher}, Geom. Dedicata 7, 499-506 (1978; Zbl 0403.51002)]. However, for finite fields \(K\) and \(l>0\), no \((l,m,n)\)-spreads over \(K\) have yet been constructed, and several nonexistence results have been demonstrated. In the paper under review more nonexistence results are obtained. Using some elementary number theory based on Zsigmondy's Theorem, it is shown that if \(l>0\) and if at least one of the integers in \(\{m-l+1, m-l+2\}\) divides neither \(n-l+1\) nor \(n-l+2\), then no \((l,m,n)\)-spreads exist over a finite field. Using some counting arguments, it is also shown that if \(n\leq 2m\), then no \((l,m,n)\)-spreads exist over a finite field. The latter result supersedes many of the previously determined nonexistence results.