On subspaces defining linear sets of maximum rank (Q6983667)

A \textit{linear set} is a set of points in a projective space that is obtained through a process called \textit{field reduction}. Let \(\mathcal{P}\) be an \((r-1)\)-dimensional projective space over the finite field \(\mathbb{F}_q^n\), hence with underlying vector space \(\mathbb{F}_{q^n}^r\). Now, let \(U \subset \mathbb{F}_{q}^{rn}\) be an \(n\)-dimensional \(\mathbb{F}_q\)-vector space. Each non-zero vector of \(U\) spans an \(\mathbb{F}_{q^n}\) vector line and hence determines a projective point in \(\mathcal{P}\). The linear set \(L_U\) is the set of projective points determined by \(U\). It is a natural question to determine all \(\mathbb{F}_q\)-linear subsets of \(V\) (e.g. of a given dimension) that determine the same points set \(L_U\).\N\NIn the nice paper under review, given two linear sets \(L_U = L_W\) with \(U\) and \(W\) of maximal dimension, the relation between \(U\) and \(W\) is investigated, and a characterization is given for \(r=2\). The following two theorems are the main results proven in the paper. We cite the theorems exactly as in the paper.\N\NTheorem 1. Let \(V=\mathbb{F}_{q^n}^r\), \(U\) and \(W\) be two \((rn-n)\)-dimensional vector spaces of \(V\) such that \(L_U=L_W\) and let \(\mathbb{F}_q\) be the maximum field of linearity for \(L_U=L_W\). Then, for every point \(P \in L=L_U=L_W\), we have that \(w_{L_U}(P)=w_{L_W}(P)\). If \(r=2\) and \(\perp\) be the polarity associated to \(\mathbf{b}:((x_1,y_1),(x_2,y_2))\in V\times V \mapsto \operatorname{Tr}(x_1y_2-x_2y_1) \in \mathbb{F}_q\), then, one of these possibilities must occur:\N\N\begin{itemize}\N\item[1.] \(W=\lambda U\) for some \(\lambda \in \mathbb{F}_{q^n}^*\);\N\item[2.] \(W=\lambda U^{\perp}\) for some \(\lambda \in \mathbb{F}_{q^n}^*\);\N\item[3.] \(n\geq 5\) and \(L_U=L_W\) is a linear set of pseudoregulus type;\N\item[4.] \(n \geq 10\) is not a prime and \(L_U=L_W\) is a linear set of generalized pseudoregulus type;\N\item[5.] \(n \geq 6\) is not a prime and \(W\) is the generalized perp of \(U\).\N\end{itemize}\N\NTheorem 2. Let \(\dim_{\mathbb{F}_{q^n}}V >2\). Let \(U\) and \(W\) be \(\mathbb{F}_q\)-vector spaces such that \(\dim_{\mathbb{F}_q} U=\dim_{\mathbb{F}_q}W=(r-1)n\) and \(L_U = L_W\) in \(\mathrm{PG}(r-1,q^n)\). Let \(U_0\) and \(W_0\) be the intersection of \(U\) and \(W\) respectively with a line \(\ell\) of \(\mathrm{PG}(r-1,q^n)\), such that \(\dim_{\mathbb{F}_q} U_0 = \dim_{\mathbb{F}_q} W_0 =n\) and \(L_{U_0} = L_{W_0}\) is a linear set of pseudoregulus type with maximum field of linearity \(\mathbb{F}_q\). Then either \(W=\lambda U\) for some \(\lambda \in \mathbb{F}_{q^n}^*\), or \(L_U=L_W\) is cone with vertex a codimension 2 subspace of \(\mathrm{PG}(r-1,q^n)\) disjoint from \(\ell\) and base \(L_{U_0}=L_{W_0}\).