On the Mills-Seligman axioms for Lie algebras of classical type

Richard E. Block

Transactions of the American Mathematical Society · 1966 · 38 citations · 5 references

Concepts

Abstract

Introduction. Mills and Seligman in [8] presented a unified technique for the classification of those semisimple Lie algebras to which the methods of Killing and Cartan are applicable.They classified the Lie algebras L (assumed to be finite-dimensional, as will be also all algebras in the present paper), of characteristic not 2, 3, 5 or 7, which satisfy the following conditions: L has a Cartan subalgebra H such that L and H satisfy five axioms, which we may state as (i) LL = L.(ii) The center of L consists of 0 alone.(iii) H acts diagonally on L.(iv) If a is a nonzero root then LaL_a is one-dimensional.(v) If a is a nonzero root and ß is a linear functional on H, then there is a positive integer m such that ß + ma is not a root.Here of course the roots are roots with respect to H, and (iii) means that if h e H and aeLa, the root space for a nonzero root a, then ah = a(/t)a.By (ii) and (iii) H is necessarily abelian.Mills and Seligman showed that if L is over a field F of characteristic p > 7 and satisfies the above axioms with respect to some Cartan subalgebra, then L is a direct sum of simple algebras of classical type, that is, each of the simple direct summands is the analogue over £ of one of the simple Lie algebras (including the five exceptional algebras) over the complex numbers.More explicitly, the simple Lie algebras of classical type over £ are obtained as follows: start with a complex simple Lie algebra, take a basis with integral structure constants as in [4, p. 24], reduce the structure constants modulo p, then take the scalar extension to an algebra over F, and divide by the center (which is nonzero only if the initial algebra is of type Ar with p I r + 1).The result of Mills and Seligman was extended by Mills [7] to include the case of characteristics 5 and 7.Axiom (v) above, although essential for the use of root-string techniques, is rather artificial in that it does not correspond (because of the finite-dimensionality) to any step in the proof of the well-known results on complex semisimple Lie algebras.

References

5