On the tensor product of composition algebras (Q5949411)

The authors investigate the tensor product \(C_1\otimes_F C_2\) of two composition algebras \(C_1\) and \(C_2\) over a field \(F\) of characteristic \(\neq 2\). First they consider the Albert forms of these algebras and obtain a necessary and sufficient condition for the similarity of the Albert forms of two algebras each of which is a tensor product of an octonion algebra and a quaternion algebra. Also an example is given where \(C_1\) and \(C_2\) are nonisomorphic octonion algebras, \(Q_1\) is a quaternion algebra which is not isomorphic to a subalgebra of either \(C_1\) or \(C_2\), and the Albert forms of \(C_1\otimes_F Q_1\) and \(C_2\otimes_F Q_1\) are similar. Some new results on the \(F\)-quadratic form defined by the Albert form are proved. A tensor product of octonion algebras is generated by its generalized alternative nucleus. Studying the generalized alternative nucleus of an algebra the authors determine the automorphism group, the derivative and the forms of a tensor product of finitely many octonion algebras. Any automorphism of such a tensor product is compatible with the canonical tensor product involution. Moreover, any such algebra does not satisfy the Skolem-Noether Theorem. Using Lie algebra-theoretic techniques the authors classify finite-dimensional simple unital algebras over an algebraically closed field of characteristic 0 which are generated by their generalized alternative nucleus. It is shown that any algebra of this kind is the tensor product of a simple associative algebra and a symmetric tensor product of octonion algebras. This result yields a variation of the Allison-Smirnov proof of the classification of finite dimensional central simple structurable algebras over a field of characteristic 0. Furthermore, an associative bilinear form on any algebra \(A\) generated by its generalized alternative nucleus can be constructed: The symmetric bilinear form \((x, y) = \text{trace} (L_xL_y)\) is associative. If A is unital, then \((x, y) = \text{trace} (L_{xy})\).