On the equivalence of linear sets (Q306339)

In this paper, the authors describe and correct a mistake published by \textit{M. Lavrauw} and the reviewer [ibid. 56, No. 2--3, 89--104 (2010; Zbl 1204.51013)] concerning the equivalence of linear sets. We denote \(p_{\Gamma_i,\Lambda_i}(\Sigma_i)\) for the projection of the subgeometry \(\Sigma_i\) from \(\Gamma_i\) to \(\Lambda_i\). It is a well known result of \textit{G. Lunardon} and \textit{O. Polverino} [Forum Math. 16, No. 5, 663--669 (2004; Zbl 1072.51010)] that every linear set arises as a projection of some subgeometry. In [Zbl 1204.51013], the following was stated as a theorem: For \(i=1,2\), let \(\Sigma_i\cong \mathrm{PG}(n-1,q)\) be two canonical subgeometries of \(\Sigma^*=\mathrm{PG}(n-1,q^t)\) and let \(\Gamma_i\) be two \((n-1-r)\)-subspaces of \(\Sigma^*\), such that \(\Gamma_i\cap \Sigma_i=\emptyset\). Let \(\Lambda_i\subset \Sigma^*\setminus \Gamma_i\) be two \((r-1)\)-subspaces and let \(L_i=p_{\Gamma_i,\Lambda_i(\Sigma_i)}\). Suppose that \(L_i\) is not an \(\mathbb{F}_q\)-linear set of rank \(k<n\). Then, the following are equivalent: (i) There exists a collineation \(\alpha:\Lambda_1\rightarrow \Lambda_2\), such that \(L_1^\alpha=L_2\). (ii) There exists a collineation \(\beta\) of \(\Sigma^*\) such that \(\Sigma_1^\beta=\Sigma_2\) and \(\Gamma_1^\beta=\Gamma_2\). (iii) For all canonical subgeometries \(\Sigma\cong \mathrm{PG}(n-1,q)\) in \(\Sigma^*\), there exist collineations \(\delta\), \(\phi\), \(\psi\) of \(\Sigma^*\), such that \(\Sigma^\delta=\Sigma\), \(\Gamma_1^{\phi\delta}=\Gamma_2^\psi\) and \(\Sigma_2^{\psi}=\Sigma\). This statement is not correct, as in this paper it is shown that the implication (i)--(ii) does not necessarily hold. A minimal counterexample is given by two particular linear sets of pseudo-regulus type in \(\mathrm{PG}(1,q^5)\). The authors investigate thoroughly in which cases the implication (i) \(\Rightarrow\) (ii) does hold: they show that if condition (A) holds for a pair \((L,n)\) where \(L\) is a linear set of rank \(n\), then the implication (i) \(\Rightarrow\) (ii) certainly holds. Condition (A) is the following: (A) For any two \((n-1)\)-subspaces \(U,U'\subset \mathrm{PG}(n-1,q)\), such that \(\mathcal{B}(U)=L=\mathcal{B}(U')\), a collineation \(\gamma\in \mathrm P\Gamma L(rt,q)\) exists, such that \(U^\gamma=U'\), \(\gamma\) preserves the Desarguesian spread \(\mathcal{F}_{r,t,q}(\mathcal{P})\) and the induced map on \(\mathrm{PG}(r-1,q^t)\) is a collineation. The main result follows from a careful investigation of the properties of the map \(\Theta_s\), where \(\Theta_s: \Sigma \to \Sigma\) and \(\Theta_s(\langle x\rangle_q)=\langle x^\theta_s\rangle_q\), where \(\Sigma\) denotes \(\mathrm{PG}(n-1,q)\) and \(\theta_s\) denotes the number of points in \(\mathrm{PG}(s,q)\). Note that Condition (A) was further investigated by the reviewer in [Linear Algebra Appl. 507, 96--120 (2016; Zbl 1352.51004)].