Veronese representation of projective Hjelmslev planes over some quadratic alternative algebras (Q2302202)

This extensive and interesting paper deals with the Veronese representations of a class of projective Hjelmslev planes related to affine buildings of relative type \(\tilde{A}_2\). The authors first introduce the concept of a Hjelmslevean Veronese set of type \((d,v)\) (with \(d,v \in \mathbb{N} \cup \{-1,\infty\}\) and \(d \geq 1\)) in the following way. Consider in a projective space PG(\(d+v+2,\mathbb{K}\)) a \(v\)-space \(V\) and an ovoid \(Q_d^0\) in a \((d+1)\)-space complementary to \(V\). The union of lines joining all points of \(V\) with all points of \(Q_d^{0}\) is called a \((d,v)\)-cone with vertex \(V\) and base \(Q_d^{0}\) and this set without the vertex \(V\) is called a \((d,v)\)-tube. Next consider a spanning point set \(X\) of PG(\(N,\mathbb{K}\)), \(N > d+v+2\) together with a collection \(\Xi\) of \((d+v+2)\)-dimensional projective subspaces of PG(\(N,\mathbb{K}\)), such that, for any \(\xi \in \Xi\), the intersection \(\xi \cap X\) is a \((d,v)\)-tube \(X(\xi)\) in \(\xi\) with base \(Q_d^{0}\). The union of all vertices of those tubes is denoted by \(Y\). The pair \((X,\Xi)\) is called a \textit{Hjelmslevean Veronese set (of type \((d,v)\))} if the following properties hold: (H1) Any two distinct points \(x_1\) and \(x_2\) of \(X\) lie in at least one element of \(\Xi\) and \((\mathrm{H}_2^{\ast}\)) Any two distinct elements \(\xi_1\) and \(\xi_2\) of \(\Xi\) intersect in points of \(X \cup Y\). Moreover, \(\xi_1 \cap \xi_2 \cap X\) is non-empty. The main result of the paper is a characterization of the Veronese representations \(\mathcal{V}_2(\mathbb{K},\mathbb{A})\) of the incidence geometries \(G(2,\mathbb{A})\) over the non-split dual numbers algebra \(\mathbb{A}\). That representation is the point-subspace structure \((X, \Xi)\) defined by means of the Veronese map \(\rho: G(2,\mathbb{A}) \rightarrow\mathrm{PG}(3d+2,\mathbb{K}): (x,y,z) \mapsto \mathbb{K}(x\bar{x},y\bar{y},z\bar{z};y\bar{z},z\bar{x},x\bar{y})\) by setting \(X=\{\rho(p)|p \in \mathcal{P}\}\) and \(\Xi=\{\langle \rho(L) \rangle |L \in \mathcal{L}\}\) where \(\rho(L)\) is defined as \(\{\rho(p)\mid p \in L\}\) and incidence is given by containment made symmetric. The algebra \(\mathbb{A}\) = CD(\(\mathbb{B},0\)) is set-wise given by \(\{b+tb' \mid b,b' \in \mathbb{B}\}\) where \(\mathbb{B}\) is a unital quadratic asscociative division algebra over a field \(\mathbb{K}\), \(t^2=0\) and with multiplication rules \(a(td)=t(\bar{a}d), (tb)c=t(bc)\) and \((tb)(td)=t^2(d\bar{b})\) with \(x \mapsto \bar{x}\) the standard involution of \(\mathbb{B}\). The precise statement of the main result is as follows. Suppose \((X,\Xi)\) is a Hjelmslevean Veronese set of type \((d,v)\) such that \(X\) generates PG(\(N,\mathbb{K}\)), where \(\mathbb{K}\) is a field with \(|\mathbb{K}|>2\). Then \(d\) is a power of 2 with \(d \leq 8\) if char(\(\mathbb{K}\)) \(\not= 2\) and one of the following holds. (i) There is only one vertex \(V\) and projected from \(V\), the resulting point-subspace geometry is projectively equivalent to \(\mathcal{V}_2(\mathbb{K},\mathbb{B})\), where \(\mathbb{B}\) is a quadratic alternative division algebra over \(\mathbb{K}\) and, in particular, \(N=3d+v+1\) and \(d=\dim_{\mathbb{K}}(\mathbb{B})\); (ii) There is a quadratic associative division algebra \(\mathbb{B}\) over \(\mathbb{K}\) and two complementary subspaces \(U\) and \(W\) of \(\mathrm{PG}(n,\mathbb{K}\)), where \(U\) is possibly empty and \(\dim W=2d+2\), with \(d=v-\dim(U)=2\dim_{\mathbb{K}}(\mathbb{B})\), such that the intersection of every pair of distinct vertices is \(U\), and the structure of \((X, \Xi)\) induced in \(W\) is projectively equivalent to \(\mathcal{V}_2(\mathbb{K},\mathsf{CD}(\mathbb{B},0))\). In particular, for each \(\xi \in \Xi\), the basis of the tube \(X \cap \xi\) is always a quadric. In the special case \(v=-1\) the assumption \(|\mathbb{K}|>2\) can be dropped if a few more possibilities are added. The proof of this theorem, which is far from elementary and has been split up in several cases, takes up the bigger part of the paper. The theorem extends and completes results obtained in [\textit{J. Schillewaert} and the second author, Adv. Math. 262, 784--822 (2014; Zbl 1383.51005); Ann. Inst. Fourier 67, 2265--2305 (2017; Zbl 1410.51010); \textit{O. Krauss} et al., Mich. Math. J. 64, 819--847 (2015; Zbl 1407.51029)].