Abstract definition of a root system in the symmetrisable case. (Q1421800)

Axiomatizations for the root systems of Kac-Moody algebras have been developed by various authors, such as the work of \textit{R. V. Moody} and \textit{A. Pianzola} [Trans. Am. Math. Soc. 315, 661--696 (1989; Zbl 0676.17011)] or \textit{R. Borcherds} [J. Algebra 115, 501--512 (1988; Zbl 0644.17010 )]. These axiomatizations were incorporated to a more general frame in \textit{N. Bardy} [Mém. Soc. Math. Fr. 65, 188 pp. (1996; Zbl 0880.17019)], in order to consider also imaginary roots, or double roots. However, almost all these attempts (with some exceptions which, however, require the existence of particular properties of the Tits cone) rely heavily on the notion of root basis. The aim of this article is to develop a general axiomatization for root systems of Kac-Moody algebras, analogous to that existing for finite root systems of Lie algebras, compatible with the above cited axiomatizations but which makes no reference to the root basis. The properties of the root systems according to this new axiomatic definition are largely discussed, and various types of root systems (e.g., essentially indecomposable root systems of real affine type) are analyzed in detail.