Abstract linear algebra. With revisions by Paul Place (Q1188718)

This is an interesting book on linear algebra starting from scratch and with a definite goal: That of characterizing the finite-dimensional real normed algebras as either the reals, the complex numbers, the quaternions or the octonions (also called the Cayley numbers). In this context a finite dimensional real normed algebra is a finite dimensional real vector space with an internal multiplication making it into an algebra with unit (not necessarily associative or commutative) and with a norm coming from an inner product on the space such that the norm is multiplicative on the algebra. The book is intended as a text for a first course on linear algebra. The first four chapters contain the material that a textbook on linear algebra must contain - and a lot more, - under the headlines: Algebraic preliminaries, vector spaces and linear maps, matrices and determinants, rings and polynomials and inner product spaces. The final chapter is on normed algebras. The set-up is abstract as promised by the title. Vector spaces are in principle over general fields. Dimension theory, direct sums, quotients and dual spaces are treated at an early stage while systems of linear equations are postponed till the middle of the book. Zorn's lemma is mentioned and used to establish the existence of a basis. A special feature is that when linear maps are represented by matrices, the matrix multiplication is on the right, - which is a necessity when one wants to consider linear maps over the quaternions in the final chapter and one has placed the scalars on the left. The treatment of determinants is by means of exterior algebra. The Cayley-Hamilton theorem is proved and from this spectral theorems and the Jordan form are derived. The chapter about inner product spaces contains material on the Cauchy-Schwarz inequality, Gram-Schmidt orthonormalization, the parallelogram law, orthogonal and unitary groups, stable subspaces and eigenvalues (including the spectral theorem for real symmetric matrices). The final chapter is devoted to a proof of Hurwitz's theorem, i.e. the characterization of finite-dimensional real normed algebras mentioned above. This result has been known since the beginning of the century, but a simplified treatment by \textit{R. Harvey} and \textit{H. Blaine Lawson} in an appendix in their paper [Acta Math. 148, 47-157 (1982; Zbl 0584.53021)] on ``calibrated geometries'' inspired the author to this presentation. The book contains many exercises both explicitly and implicitly. There are some misprints. The most serious ones occur in chapter IV, section B, where the following changes have to be made: ``A: \(\div ''\) has to be changed to ``a: \(+''\) on p. 120, line 11, p. 121, lines 7 and 8, p. 122, lines 12 and 13 (also change \(``=0''\) to \(``=8''\) here). The sentence on p. 121, line 14, should read: ``...substituting \(u+v\) for u in (14) and using the result from line 10 gives (13)''.