Local Schur's Lemma and commutative semifields (Q2487217)

Let \(V\) be a vector space of dimension \(n\) over the finite field \(GF(q)\) of order \(q.\) A subset \(\mathbf C\) of the set of the linear maps from \(V\) to itself is a spread set if and only if 1) \(\mathbf 0\) and \(\mathbf 1\) belong to \(\mathbf C,\) 2) if \(\mathbf X\) and \(\mathbf Y\) are two elements of \(\mathbf C,\) then \(\mathbf{X-Y}\) is non-singular; 3) \(| {\mathbf C}| =q^n.\) Let \(e\) be any non-zero element of \(V.\) If \(a\) is an element of \(V,\) then there is a unique \(L_a\) in \({\mathbf C}\) such that \(L_a(e)=a.\) Define a multiplication on \(V\) by \(a \cdot b =L_b(a);\) then \(Q=(V;+, \cdot)\) is a quasifield with unity \(e\) and any quasifield can be constructed in such a way. Moreover, \(Q\) is a semifield if and only if \(\mathbf C\) is closed under the sum between linear maps. The author studies commutative semifields, generalizing a result of Ganley [\textit{M. J. Ganley}, Geom. Dedicata 1, 103--116 (1972; Zbl 0241.50024)] which was proved using orthogonal polarities. A set of linear maps \({\mathcal R} \subset GL(V,GF(q))\) is \textit{regular} if for any pair \(x,y\) of nonzero vectors of \(V\) there is a unique element \(R\) of \(\mathcal R\) such that \(R(x)=y.\) Thus, a regular linear set containing the identity corresponds to the non-zero elements of the spread set \({\mathbf C}={\mathcal R}\cup\{\mathbf 0\}.\) If the elements of \(\mathcal R\) pairwise commute, then \({\mathcal R}\) is a field by Schur's lemma. A regular set \(\mathcal R,\) containing the identity, is said to be \textit{locally commutative} at \(x \in V\) if \(AB(x)=BA(x)\) for all elements \(A\) and \(B\) of \(\mathcal R.\) The Schur kernel of a regular set \(\mathcal R\) is the subspace \(K\) of \(V\) consisting of all points at which \(\mathcal R\) is locally commutative. If the Schur kernel is non-trivial, then \(\mathcal R\) is locally commutative at any point of \(K.\) The author proves that, if \(\mathcal R\) is a locally commutative regular set and \(K\) is its Schur kernel, then: (1) \({\mathbf C}={\mathcal R}\cup\{\mathbf 0\}\) is a spread set closed under the sum between linear maps. (2) \({\mathcal N}=\{N \in {\mathbf C}\mid N(K)\subset K\}\) is a field. (3) If \(\beta\) is a non-zero element of the Schur kernel, then \(K={\mathcal N}(\beta).\) (4) \({\mathcal N}=\{N \in {\mathbf C}\mid {\mathcal R}N \subset {\mathcal R}\}.\) If \(e\) is a non-zero element of \(K,\) and \(L_a\) is the element of \({\mathbf C}\) such that \(L_a(e)=a,\) then the quasifield \(Q,\) whose multiplication is defined by \(a \cdot b =L_b(a),\) is a commutative semifield and \(K\) is its middle nucleus. From the above mentioned results, it follows that a translation plane \(\pi\) can be coordinatized by a commutative semifield if and only if there is a spread set \(\mathbf C\) associated to \(\pi\) such that \({\mathcal R}={\mathbf C}\setminus\{\mathbf 0\}\) is a locally commutative regular set.