Splitting of differential quaternion algebras (Q6170746)

Given a ring \(\left(R,+,\cdot\right)\), an additive function \(d\) (i.e. a group morphism \(\left(R,+\right)\to \left(R,+\right)\)) satisfying the Leibniz rule \(d\left(x,y\right)=xd\left(y\right)+d(x) y\) for any \(x,y\in R\) is called a \textit{differential ring} and \(d\) is called a \textit{derivation} on \(R\). The \textit{set of constants} of \(R\) is the subring \(C_{R,d}:=\left\{x\in R:d\left(x\right)\right\}\). Every \(x\in R\) induces a derivation \(\partial_x\left(a\right):=a x-xa\) and \(C_{R,\partial_x}\) is the centralizer of \(x\) in \(R\). \(k\) will denote a field of characteristic different from \(2\) from this point onward. A \textit{central simple algebra} is a finite-dimensional \(k\)-algebra with center \(k\) and no non-trivial two sided ideals. One such algebra \(A\) \textit{splits over a field extension} \(L\mid k\) if the extension of scalars \(A\otimes_kL\) is isomorphic, as an \(L\)-algebra, to a matrix algebra \(M_m\left(L\right)\). Derivations \(\partial\) on fields \(k\) can be extended to extensions \(L\) thereof, and if \(L\mid k\) is finite then such an extension is unique and can be denoted just like its restriction \(\partial\) to \(k\). A central simple \(k\)-algebra \(A\) along with a derivation \(d\) extending \(\partial\) (i.e. restricting to \(\partial\) on \(k\)) is called a \textit{differential central simple algebra} over \(k\), and denoted \(\left(A,d\right)\). The set of all derivations \(d\) on \(A\) fulfilling this capacity are denoted \(\mathrm{Der}\left(A\mid \left(k,\partial\right)\right)\). The isomorphisms in the category \(\mathrm{CDA}_k\) of differential central algebras are \(k\)-algebra isomorphisms \(\phi:\left(A_1,d_1\right)\to \left(A_2,d_2\right)\) satisfying \(\phi\circ d_1\equiv d_2\circ \phi\) and are called \textit{differential isomorphisms}. In an earlier paper, the authors introduced the notion of extensions splitting over extensions: \(\left(A,d\right)\mid \left(k,d\right)\) \textit{splits over a differential field \(\left(A,d\right)\mid \left(k,d\right)\)} if \(\left(A\otimes_kL,d^*\right)\cong \left(M_m\left(L\right),\partial\right)\) in \(\mathrm{CDA}_L\), where \(d^*:= d\otimes \partial =d\otimes \mathrm{id}_L+\mathrm{id}_A\otimes \partial\) and \(\partial\) stands here for the coordinate-wise propagation of the derivation on \(k\) to matrices. In previous work, the authors proved the existence of differential splitting fields by way of Picard-Vessiot theory, i.e. the differential analogue of Galois theory, and characterized differential matrix algebras isomorphic to \(\left( M_m\left(k\right),d_P\right)\) (where \(d_P:=\partial+\partial_P\)) in terms of fundamental matrices of the linear system having \(P\) as a matrix. For \(\alpha,\beta\in k^*\), let \(Q=\left(\alpha,\beta\right)_k\) be a quaternion algebra with generators \(u,v\in Q^0\) such that \(u^2=\alpha,v^2=\beta\) and \(vu=-uv\), and \(d_{\left(u,v\right)}\in \mathrm{Der}\left(Q\mid \left(k,\partial\right)\right)\) a standard derivation on \(Q\), i.e. \(d(u)\in k(u)\) and \(d(v)\in k(v)\). The authors recall past results about standard derivations as well as finite and finitely generated splitting fields for extensions of \(Q\mid k\). Once this background material is introduced, the authors then proceed to construct differential splitting fields for an arbitrary differential quaternion algebra \(\left(Q,d\right)\) and to provide bounds on the transcendence degree of these splitting fields. The path towards these goals is buttressed, firstly, by a sufficient condition for nonexistence of solutions of Riccati equations which are algebraic over a transcendental extension of the coefficient field \(k\). Using this and some of the background material in the first Sections, the paper then proves a relevant Proposition linking the splitting of \(\left(Q,d\right)\) to certain Riccati and linear homogeneous equations. This Proposition is then used to construct an example of a differential splitting field of transcendence degree at most \(1\) for a particular quaternion algebra. It is also used to prove the main result in the paper -- to wit, a Theorem stating that \(\left(Q,d\right)\) is split by a differential field extension of \(\left(k,\partial\right)\) of transcendence degree at most \(3\). The next Section is devoted to examples of splitting fields with transcendence degree \(2\) and \(3\) with the aid of the tools seen in the previous Section. The sixth Section establishes a necessary and sufficient condition for \(\left(Q,d\right)\mid \left(k,\partial\right)\), where \(d=d_{\left(u,v\right)}+\partial_{au}\) (\(a\in k\)) to be split by a finite extension of \(\left(k,\partial\right)\), and proves a sufficient condition for the nonexistence of a splitting by an algebraic extension of \(k\). This Theorem draws heavily from an example amenable to study from the point of view Picard-Vessiot theory and Liouville extensions. The final Section proves necessary and conditions for a derivation on a quaternion algebra \(\left(Q,d\right)\mid \left(k,\partial\right)\) to \textit{become standard} over its differential splitting fields \(L\), i.e. to be such that \(d^*=d\otimes \partial\) is standard on the extension of scalars \(Q\otimes_k L\).