§ 7. Ado's theorem
Recall that denotes a field of characteristic 0 and that all Lie algebras are assumed to be finite-dimensional over .
1. Coefficients of a representation
Let U be an associative algebra with unit element over K, U* the dual of the vector space U and a representation of U on a vector space E. For and , let be the corresponding coefficient of (Algebra, Chapter VIII, § 13, no. 3). Recall that and that the mapping is for fixed a homomorphism of the U-module E into the U-module U* of the coregular representation of U (loc. cit., Proposition 1); therefore the vector subspace of U* generated by the coefficients of (a subspace which we shall denote by in this paragraph) is a sub-U-module of U*. If is a family of elements generating E* over K, the mapping is an injective U-homomorphism of E into ; for for all implies for all and hence .
In particular, if U is the enveloping algebra of a Lie algebra g and is a representation of g (identified with a representation of U) on an n-dimensional vector space E, the g-module E is isomorphic to a sub-g-module of .
2. The extension theorem
Let be a Lie algebra which is the direct sum of an ideal and a subalgebra , U the enveloping algebra of g and the enveloping algebra of . There exists one and only one g-module structure on such that: (α) for and , ; (β) for and , (the latter element is certainly in since the inner derivation of U defined by x leaves and hence stable). For conditions (α) and (β) define uniquely a linear mapping of g into . It therefore suffices to verify that ; it is only necessary to consider the following cases:
(1) : then
(2) : then
(3) : then and are two derivations of whose restrictions to coincide with those of and ; hence these derivations are equal.
We shall also consider the dual representation of g on . For , is the transpose of right multiplication by x in ; the corresponding representation of is therefore the coregular representation of .
Definition 1
Let be a Lie algebra, a subalgebra of and a representation of on . A representation of on is called an extension of to if there exists an injective homomorphism of the -module into the -module . We also say that the -module is an extension of the -module .
If is finite-dimensional and is a solvable ideal of , it is necessary for the existence of a finite-dimensional extension that be contained in the largest nilpotency ideal of (§ 5, no. 3, Theorem 1).
Theorem 1 (Zassenhaus)
Let be a Lie algebra which is the direct sum of an ideal and a subalgebra and a finite-dimensional representation of whose largest nilpotency ideal contains .
(a) There exists a finite-dimensional extension of to whose largest nilpotency ideal contains that of .
(b) If for all the restriction to of is nilpotent, can be chosen such that moreover the largest nilpotency ideal of contains .
Let be the enveloping algebra of . Suppose that and have the -module structures defined at the beginning of this no.
We now show that is finite-dimensional over . Let be the space on which operates and a Jordan-Hölder series of the -module . Let be the representation of on derived from (). Let be the intersection of the kernels of the (identified with representations of ). Then
and is the largest nilpotency ideal of . By § 2, no. 6, Corollary to Proposition 6, is of finite codimension in . If , the derivation of maps into , hence into and hence into . On the other hand, clearly is a sub--module of . Hence is a sub--module of . The orthogonal of in is a finite-dimensional sub--module which contains and therefore . This shows that is finite-dimensional over . For , is obviously contained in the annihilator of the -module and hence also in the annihilator of the -module .
We saw in no. 1 that the -module is isomorphic to a sub--module of a product . Hence the -module provides a finite-dimensional extension of to . Moreover, is nilpotent for ; as is an ideal of (for it contains by hypothesis), we see that is contained in the largest nilpotency ideal of . Thus (a) is proved.
Suppose finally that for all the restriction to of is nilpotent. As the elements of operate on by derivations, there exists for all and all an integer such that ; the endomorphisms derived from on and on (which are finite-dimensional spaces) are therefore nilpotent. Thus is nilpotent for all . We have seen earlier that this is also true for . As is an ideal of containing , the sum is also an ideal of . Assertion (b) of Theorem 1 then results from the following lemma:
Lemma 1
Let be a Lie algebra which is the sum of an ideal and a subalgebra . Let be a finite-dimensional representation of . Suppose that is nilpotent for all and all . Then is nilpotent for all .
Passing to the quotient by the kernel of , may be assumed to be faithful. Then and are nilpotent and hence , which is an extension of a quotient of by , is solvable. Then and are contained in the largest nilpotency ideal of (§ 5, no. 3, Corollary 6 to Theorem 1).
For an improvement of Theorem 1, cf. Exercise 4.
3. Ado's theorem
Proposition 1
Let be a Lie algebra, its largest nilpotent ideal, a nilpotent ideal of and a finite-dimensional representation of such that every element of is nilpotent. Then admits a finite-dimensional extension to such that every element of is nilpotent.
Let be a sequence of subalgebras of such that is an ideal of of codimension 1 for (§ 4, no. 1, Proposition 1 (e)). The algebra is therefore the direct sum of and a 1-dimensional subalgebra. As is nilpotent for all , it is possible (Theorem 1) to find one by one finite-dimensional extensions of to such that every element of is nilpotent.
Let be the radical of and let be a sequence of subalgebras of such that is an ideal of of codimension 1 for (§ 5, no. 1, Proposition 2 (d)). The algebra is thus the direct sum of and a 1-dimensional subalgebra. As , it is possible (Theorem 1) to find one by one finite-dimensional extensions of to such that every element of is nilpotent.
Finally is the direct sum of and a subalgebra (§ 6, no. 8, Theorem 5). As , it is possible (Theorem 1) to find a finite-dimensional extension of to such that every element of is nilpotent.
Theorem 2
Every Lie algebra has a finite-dimensional faithful linear representation.
More precisely:
Theorem 3 (Ado)
Let be a Lie algebra and its largest nilpotent ideal. There exists a finite-dimensional faithful representation of such that every element of is nilpotent.
The 1-dimensional Lie algebra admits finite-dimensional faithful representations such that every element of is nilpotent, for example the representation
It is easily deduced that the centre of , which is a product of 1-dimensional algebras, admits a finite-dimensional faithful representation such that every element of is nilpotent. Let be a finite-dimensional extension of to such that every element of is nilpotent (Proposition 1); if denotes the kernel of , then . On the other hand, let be the adjoint representation of , whose kernel is ; every element of is nilpotent. The direct sum of and is finite-dimensional; every element of is nilpotent; and the kernel of , contained in and in , is zero, so that is faithful.
Exercises
The conventions of § 7 remain valid unless otherwise mentioned.
See the exercises for § 7.