Abdenacer Makhlouf

Algebra and Applications 1


Скачать книгу

alt="image"/> of characteristic not 2 with a countable basis {un : n ∈ ℕ}. Define a multiplication and a norm on image by

image

      Then (image, ∙, n) is a composition algebra.

      In Elduque and Pérez (1997), one may find examples of infinite-dimensional composition algebras of arbitrary infinite dimension, which are even left unital.

      2.3.2. Isotropic Hurwitz algebras

      Assume now that the norm of a Hurwitz algebra (image, ∙, n) represents 0. That is, there is a non-zero element image such that n(a) = 0. This is always the case if image and image is algebraically closed.

      With a as above, take image such that image, so that n(ab, 1) = 1. Also n(ab) = n(a)n(b) = 0. By the Cayley–Hamilton equation, the non-zero element e1 := ab satisfies image, that is, e1 is an idempotent. Consider too the idempotent image, and the subalgebra image generated by e1. (1= e1 + e2).

      For any image and, as image and image, we conclude that xe1 = e2x, and in the same way, xe2 = e1x, for any image.

      But x = 1 ∙ x = e1x + e2x, and e2 ∙ (e1x)= (1 − e1) ∙ (e1x) = 0 = e1 ∙ (e2x). It follows that image splits as image with

image

      For any image, n(u) = n(e1u) = n(e1)n(u) = 0, so image, and image too, are totally isotropic subspaces of image paired by the norm. In particular, image, and this common value is either 0, 1 or 3, depending on image being 2, 4 or 8. The case of image = 0 is trivial, and the case of image = 1 is quite easy (and subsumed in the arguments below). Hence, let us assume that image is a Cayley algebra (dimension 8), so image.

      For any image and image, using [2.5] we get

image

      Hence, image is orthogonal to both image and image, so it must be contained in image. Also image.

      Besides,

image

      so that image, and also image. But for any image and image, we have

image

      and the analogues for vu. We conclude that

image

      Now, for linearly independent elements image, let image with n(u1, v) ≠ 0 = n(u2, v). Then the alternative law gives (u1u2)∙ v = −(u1v)∙ u2 +u1 ∙ (u2v + vu2) = −n(u1, v)u2 ≠ 0, so that u1u2 ≠ 0. In particular, image, and the same happens with image.

      Consider the trilinear map:

image

      This is alternating because x∙2 = 0 for any image by the Cayley–Hamilton equation, and n(xy, y) = −n(x,