Abstract
We make a detailed study of the Lie algebras , , of triangular polynomial derivations, their injective limit , and their completion . We classify the ideals of (all of which are characteristic ideals) and use this classification to give an explicit criterion for Lie factor algebras of and to be isomorphic. We introduce two new dimensions for (Lie) algebras and their modules: the central dimension and the uniserial dimension , and show that for all , where is the first infinite ordinal. Similar results are proved for the Lie algebras and . In particular, and .
Export citation and abstract BibTeX RIS
This work is partially supported by the Royal Society and EPSRC.
§ 1. Introduction
Throughout, module means a left module; is the set of non-negative integers; is a field of characteristic zero and is its group of units; is the polynomial algebra over , where ; are the partial derivatives (-linear derivations) of ; is the Lie algebra of -derivations of ; is the th Weyl algebra; for every integer ,
is the Lie algebra of triangular polynomial derivations (it is a Lie subalgebra of the Lie algebra ), and is its universal enveloping algebra.
Many properties of the , where , are proved in Proposition 2.1. In particular, the Lie algebras are pairwise non-isomorphic, soluble but not nilpotent, and their inner derivations are locally nilpotent. The derived series and lower central series are found for .
We introduce a new dimension for algebras and modules, the uniserial dimension (see § 4), which has turned out to be a very useful tool in the study of non-Noetherian Lie algebras, their ideals and automorphisms [1], [2].
In § 3 we give a classification of the ideals of and find an explicit basis for each. We prove that the Lie algebra is uniserial, Artinian but not Noetherian, and its uniserial dimension is equal to
where is the first infinite ordinal (Theorem 3.6). We show that the hypercentral series stabilizes precisely at step . Moreover, for every , where is explicitly given by (3.3). The central dimension of is (Theorem 3.10). It is proved that all the ideals of are characteristic (Corollary 3.11).
A Lie algebra is said to be locally nilpotent (resp. locally finite-dimensional) if every finitely generated Lie subalgebra of is nilpotent (resp. finite-dimensional). In § 4 we prove that the are locally finite-dimensional and locally nilpotent (Theorem 4.2).
In § 5, Theorem 5.1 and Corollary 5.2 give an answer to the following question.
Question. Let and be ideals of the Lie algebras and respectively. When are the Lie factor algebras and isomorphic?
The answer is given in explicit terms (via the uniserial dimensions of and ) using the classification of ideals of (Theorem 3.6). In particular, there are only countably many ideals of such that . Theorem 3.6(1) shows that every ideal of is uniquely determined by its uniserial dimension
that is, , where is the first infinite ordinal.
In § 6 we study the Lie algebra . Many of the properties of are similar to those of the but there are several differences. For example, is not soluble, not Artinian but almost Artinian, and . We obtain a classification of the ideals of (Theorem 6.2). All the ideals of are characteristic (Corollary 6.4). Corollary 6.3 is an explicit criterion for when two Lie factor algebras of are isomorphic.
In § 7 we study the topological Lie algebra , which is the completion of . Its properties diverge further from those of the and . We classify all closed ideals and all open ideals of (Theorem 7.2(1)).
§ 2. The derived series and lower central series of the Lie algebras
In this section we prove various properties of the (Proposition 2.1, Corollary 2.3) that are widely used in the rest of the paper. At the end of the section we find the image and kernel of the algebra homomorphism (see (2.11) and Theorem 2.6). In particular, it is shown that the algebra is not finitely generated and neither left nor right Noetherian.
Let be a Lie algebra over the field , and let , be ideals of . The commutator of and is the linear span in of all the elements , where and . The commutator of subspaces and of is defined in the same manner. is an ideal of , and . In particular, is called the commutator subalgebra of . Let us define recursively the following set of ideals of :
Clearly, for all . The descending chains of ideals of ,
are called the derived series and the lower central series of respectively. Notice that
Thus, the elements
form a -basis of called the canonical basis. For all , , and we have
where is the canonical free -basis of the -module and . In particular, and . The Lie algebra is the direct sum of the Abelian (infinite-dimensional when ) Lie subalgebras (that is, ) such that, for all ,
The inclusion in (2.3) is obvious but the equality follows from the fact that . By (2.3), admits a finite strictly decreasing chain of ideals
where for . By (2.3), for all ,
For every there is a canonical isomorphism of Lie algebras
In particular, . Clearly,
is an ascending chain of Lie algebras. The polynomial algebra is an -module: for all elements ,
Clearly, , . Since , the polynomial algebra is also a -module.
Let be a vector space over . A -linear map is called a locally nilpotent map if or, equivalently, for every we have for all . When is a locally nilpotent map in , we also say that acts locally nilpotently on . Every nilpotent linear map (that is, with for some ) is a locally nilpotent map but not vice versa, in general. Let be a Lie algebra. Each element determines a derivation of by the rule , . It is called the inner derivation associated with . The set of all inner derivations of is a Lie subalgebra of the Lie algebra , where . There is a short exact sequence of Lie algebras
that is, , where is the centre of and for all . An element is said to be locally nilpotent (resp. nilpotent) if the inner derivation of has that property. Let be a non-empty subset of . Then for all is called the centralizer of in . It is a Lie subalgebra of . Let be an associative algebra and a non-empty subset of . Then for all is called the centralizer of in . It is a subalgebra of .
- 1)The Lie algebra is soluble but not nilpotent.
- 2)
- 3)The lower central series of stabilizes at the first step, that is, and for all .
- 4)Each element acts locally nilpotently on the -module .
- 5)All inner derivations of are locally nilpotent.
- 6)The centre of is equal to .
- 7)The , where , are pairwise non-isomorphic.
Proof. Part 1 follows from parts 2 and 3, which in their turn follow from the decomposition and (2.3). Part 4 follows from the definition of .
Let us prove part 5. Suppose that for some . Then for some elements and . Let , and . We must show that for all . Applying (2.5) twice, we see that and . Replacing the element by , we may assume without loss of generality that and . Since , we have for some elements , , and . For all integers , by (2.5) we have
By part 4, for all . Then for all . Similarly, for all . Repeating this argument several times, we see that for all . This means that is a locally nilpotent map, as required.
Let us prove part 6. It is well known and easily provable that
It follows that
and, therefore,
The opposite inclusion is obvious. Therefore .
Part 7 follows from part 2 since the lengths of the derived series of the are distinct (and invariant under isomorphisms). □
Proposition 2.1(5) enables us to produce many automorphisms of the Lie algebra . For every element , the inner derivation is locally nilpotent and, therefore,
The automorphism group of and explicit generators for it were found in [1], and the adjoint group was shown to be a tiny part of the group .
The next lemma classifies the nilpotent inner derivations of the .
Lemma 2.2. Suppose that and . Then the following assertions are equivalent.
- 1)The map is a nilpotent derivation of .
- 2).
- 3).
Proof. The implications are obvious.
Let us prove . We claim that is not nilpotent when . Indeed, suppose that , that is, , where for all , and . Since
the derivation is not nilpotent. □
Let be a ring. A subset of is said to be multiplicative or multiplicatively closed if , and . Every associative algebra may be regarded as a Lie algebra , where is the commutator of elements . For every , the map , , is a -derivation of regarded as an associative algebra and a Lie algebra. It is called the inner derivation of associated with . Thus the associative algebra and the Lie algebra have the same set of inner derivations and the same centre .
Let be a derivation of a ring . For all elements ,
- 1)The inner derivations of the universal enveloping algebra of the Lie algebra are locally nilpotent derivations.
- 2)Every multiplicative subset of generated by an arbitrary set of elements of is a (left and right) Ore set in . Therefore .
Proof. Part 1 follows from (2.9) and Proposition 2.1(5). Part 2 is an easy corollary of (2.10) and Proposition 2.1(5). □
Example 2.4. The set is a multiplicative subset of . By Corollary 2.3(2), the ring of fractions exists.
Let be a ring, an -tuple of commuting automorphisms of ( for all ), and an -tuple of (non-zero) elements of the centre of such that for all .
The generalized Weyl algebra (briefly GWA) of degree with base ring is the ring generated by and indeterminates , subject to the following defining relations [3], [4]:
where . We say that and are the sets of defining elements and automorphisms of respectively. Given a vector , we put , where , , for and . It follows from the definition of a GWA that
is a -graded algebra ( for all ), where .
Let be the polynomial algebra in indeterminates. We define an -tuple of commuting automorphisms of by putting and for . The map
is an isomorphism of -algebras. We identify the Weyl algebra with the GWA above by means of this isomorphism. The Weyl algebra is a -graded algebra ( for all ).
The multiplicative sets and are (left and right) Ore sets of , and we have
The Weyl algebra is a -graded Lie algebra, that is, for all . By definition, and are -graded Lie subalgebras of .
The following lemma shows that contains finite-dimensional and infinite-dimensional maximal Abelian Lie subalgebras.
- 1)is a maximal Abelian Lie subalgebra of with .
- 2)The ideal of is a maximal Abelian Lie subalgebra of with .
Proof. Part 1 follows from the equalities (see (2.7)) and .
Let us prove part 2. It follows from the equality that
Then . Therefore the ideal is a maximal Abelian Lie subalgebra of . □
The inclusion induces an algebra homomorphism
The image and kernel of are found in Theorem 2.6. By Corollary 2.3(2), the homomorphism can be extended to an algebra homomorphism
where . Clearly, (since and ).
We define a relation on the set by writing for elements and of if and only if either and is arbitrary (that is, for all ), or , and . Clearly, and imply that , if and only if , and for all , implies that .
- 1)The set is a -basis for the algebra .
- 2)The kernel of is the ideal of generated by the elements , where , and .
- 3)is not surjective.
- 4)is neither finitely generated, left Noetherian, nor right Noetherian.
Proof. Let us prove part 1. The elements of are -linearly independent since they are as elements of the algebra . Put . We must show that . The algebra is generated by the elements . Using the relations
we see that is contained in the linear span, call it , of the elements
Using the commutation relations (where is the Kronecker delta), we can write each of these elements as a linear combination of elements of . Therefore, .
To prove the opposite inclusion and thus complete the proof of part 1, we must show that every element in belongs to the algebra . The element is a product
The case when is obvious. Thus, we can assume that . Every element of the Weyl algebra can be written as a sum , where . Now,
This proves part 1. Moreover, the last step implies that the set
is also a -basis of . In more detail, is the linear span of the set , whose elements are linearly independent since they are as elements of . Therefore is a -basis of . This basis will be used in the proof of part 2.
We now prove part 2. Let be the ideal of generated by the elements listed in part 2 (the putative generators for ). For every element in , that is, for and , we choose an element in the following way:
Thus we have for all . Let be the set of all elements . The elements of are linearly independent in , being the pre-images of linearly independent elements. Put . To finish the proof of part 2, it suffices to show that
(Suppose that this equality holds. Since and the set is mapped bijectively onto the basis of , these two facts necessarily imply that .) To prove the equality , we follow the line of the proof of part 1. The relations (2.2) and imply that the linear span of the elements
where
generates the algebra modulo the ideal . Using the generators for and the relations (2.2), we can write each of these elements as a linear combination of the elements defined above. Then .
Let us prove part 3. Suppose that the homomorphism is surjective. We seek a contradiction. The ideal of contains , , is an ideal of the algebra , and
The Weyl algebra is a simple infinite-dimensional algebra. So it is mapped isomorphically onto its image under the algebra homomorphism
a contradiction.
Part 4 follows from Proposition 3.13, where a stronger assertion will be proved. □
Corollary 2.7. The set (see (2.12)) is a -basis of the algebra .
Proof. This was established in the proof of part 1 of Theorem 2.6. □
§ 3. Classification of ideals of the Lie algebra
In this section we introduce the uniserial and central dimensions and give a classification of the ideals of the Lie algebra (Theorem 3.6(1)). We prove that is a uniserial Artinian non-Noetherian Lie algebra of uniserial dimension (Theorem 3.6(2)) and every ideal of is a characteristic ideal (Corollary 3.11). We also find the hypercentral series of and show that the central dimension of is given by (Theorem 3.10).
3.1. Uniserial dimension.
Let be a partially ordered set, that is, the set admits a relation satisfying the following three conditions for all :
- (i),
- (ii)and imply that ,
- (iii)and imply that .
A partially ordered set is said to be Artinian if every non-empty subset of has a least element, call it , that is, for all . A partially ordered set is totally ordered if for all elements , either or . A bijection between partially ordered sets and is an isomorphism if in implies that in . We recall that ordinal numbers are the isomorphism classes of totally ordered Artinian sets (that is, well-ordered sets). The ordinal number (isomorphism class) of a totally ordered Artinian set is denoted by . The class of all ordinal numbers is denoted by . The class is totally ordered by inclusion and is Artinian. It is endowed with an associative addition and an associative multiplication that extend the addition and multiplication of the non-negative integers. Every positive integer is identified with . We put . (More details on ordinal numbers can be found in [5].)
Definition 3.1. Let be a partially ordered set. The uniserial dimension of is the supremum of the ordinal numbers , where runs through all Artinian totally ordered subsets of .
For a Lie algebra , let (resp. ) be the set of all ideals (resp. all non-zero ideals) of . Thus . The sets and are partially ordered by inclusion. A Lie algebra is said to be Artinian (resp. Noetherian) if the partially ordered set is Artinian (resp. Noetherian). This means that every descending (resp. ascending) chain of ideals stabilizes. A Lie algebra is said to be uniserial if the partially ordered set is totally ordered. This means that for any two ideals , of the Lie algebra we have either , or .
Definition 3.2. Let be an Artinian uniserial Lie algebra. The ordinal number of the Artinian totally ordered set of non-zero ideals of is called the uniserial dimension of . For an arbitrary Lie algebra , the uniserial dimension is the supremum of , where runs through all Artinian totally ordered sets of ideals.
If is a Noetherian Lie algebra, then . Thus the uniserial dimension is a measure of the deviation from the Noetherian condition. The concept of uniserial dimension makes sense for any class of algebras (associative, Jordan, etc.).
Let be an algebra, an -module, and (resp. ) the set of all non-zero left ideals of (resp. all non-zero submodules of ). These sets are partially ordered with respect to . The left uniserial dimension of is defined as , and the uniserial dimension of is defined as .
3.2. An Artinian total ordering on the canonical basis of .
We define an Artinian total ordering on the canonical basis of by putting if and only if either , or and , for some .
Example 3.3. For we have . For we have
The following lemma is a straightforward consequence of the definition of the ordering . We write for all .
- 1)for all ,
- 2)for all such that and ,
- 3)for all such that ,
- 4)for all such that , that is, .
Let be the set of indices that parametrizes the canonical basis of . The set is an Artinian totally ordered set, where if and only if . It is isomorphic to the Artinian totally ordered set via the map . We identify the partially ordered sets and by means of this isomorphism. Clearly,
and . We put . By (2.2), if , then
By (3.2), the map
is a monomorphism of partially ordered sets ( is an order-preserving injection). We shall prove that is a bijection (Theorem 3.6(1)) and, as a result, we will have a classification of the ideals of . Each non-zero element of is a finite linear combination
where and . The elements and are called the leading term and leading coefficient of respectively, and the ordinal number denoted by and defined as the ordinal number that represents the Artinian totally ordered set , is called the ordinal degree of and is denoted by (we hope that this notation will not lead to confusion). The following assertions hold for all non-zero elements and all :
- (i)provided that ,
- (ii),
- (iii)provided that ,
- (iv)
3.3. Classification of ideals of the Lie algebra .
The following lemma plays a crucial role in the proof of Theorem 3.6.
Lemma 3.5. Let be a non-zero ideal of . Then . In particular, for all non-zero elements of , where is the ideal of generated by the element .
Proof. It suffices to prove the second assertion (that is, ) because then
To prove that , we use a double induction: first on and then, for a fixed , on . We may assume without loss of generality that the leading coefficient of is equal to 1. Notice that is not a limit ordinal.
A. Let . If , then and, therefore, , as required. Suppose that and the desired assertion holds for all non-limit ordinals such that .
Case 1: for some , that is,
for some scalars . Put . Then , that is, .
Case 2: , that is, for some element . For all we have . By Case 1, and, therefore, . Hence .
B. Suppose that and the desired assertion holds for all such that . If , then and, therefore, , as required. Suppose that and the assertion holds for all non-limit ordinals such that .
Case 1: for some , , that is, , where the dots stand for smaller terms. The following claim is a corollary of (2.2) and the definition of the total ordering on the canonical basis of .
For every non-limit ordinal such that , there are elements of , say, , of type , , such that
By induction on , it follows from this claim that .
Case 2: for some such that . For all elements with and we have
whence . By induction, . But and, therefore, .
Case 3: for some such that and . Notice that
By Case 2, . Since , we have . By (2.6), . By assumption, . Consider the element . Then the desired assertion follows by induction on . □
Given any , we have if and only if .
- 1)The map (3.3) is a bijection.
- 2)The Lie algebra is uniserial, Artinian and non-Noetherian. Its uniserial dimension is given by .
Proof. Part 1 follows at once from Lemma 3.5. Part 2 follows from part 1. □
An ideal of a Lie algebra is said to be proper (resp. cofinite) if (resp. ).
- 1)The ideal is the largest proper ideal of .
- 2)The ideal is the only proper cofinite ideal of , and .
- 3)The centre of is the least non-zero ideal of .
- 4)The ideals , , are the only finite-dimensional ideals of , and .
Proof. All these statements are easy corollaries of Theorem 3.6. □
3.4. The centralizers of ideals of .
In combination with Theorem 3.6, the following proposition describes the centralizers of all ideals of the Lie algebra . Notice that the centralizer of an ideal of a Lie algebra is also an ideal.
- 1)We have
- 2)The set of centralizers of all ideals of contains precisely elements, and the map , , is a bijection. Moreover, it is an inclusion-reversing involution, that is, for all .
- 3)We have , , , , for .
Proof. Put . If , then and by Proposition 2.1(6). If , then and, therefore, . We can assume that . To prove part 1, we use induction on . For , the only case to consider is when . In this case,
and, therefore, .
Suppose that and part 1 holds for all .
We claim that for .
Indeed, for we have and our claim is just Lemma 2.5(2). Suppose that . Then , whence . This means that the -module is also a -module and a -module since . The map
is an isomorphism of -modules (since and ), and our claim follows by induction on .
Suppose that , that is, for some . We must show that (see the claim). By the claim, the inclusions imply that . Notice that , (since ) and for all non-zero elements . It follows that , as required.
Suppose that , that is, for some . Put . Then by the claim. Therefore, . It follows from the inclusion that by the claim. Then the inclusion implies that . This means that .
Part 2 follows from part 3. Part 3 follows from part 1. □
3.5. The hypercentral series and central dimension.
For a Lie algebra over a field we define its hypercentral series recursively. We put . If is not a limit ordinal, that is, for some , then
If is a limit ordinal, then . If , then . Thus, is an ascending chain of ideals of . We put .
Definition 3.9. The minimal ordinal number (if it exists) such that , is called the central dimension of the Lie algebra and is denoted by . If there is no such , we write . The concept of central dimension makes sense for any class of algebras (associative, Jordan etc.).
A Lie algebra is central (that is, ) if and only if . Thus the central dimension measures the deviation from 'being central'. The following theorem describes the hypercentral series for the Lie algebra and gives .
Theorem 3.10. The hypercentral series stabilizes precisely at step , that is, . Moreover, for every . In particular, .
Proof. It suffices to show that for all , where . We use induction on . The case follows from Proposition 2.1(6): . Suppose that and the desired equality holds for all ordinals .
If is a limit ordinal, then
Suppose that is not a limit ordinal, that is, for some ordinal number .
Case 1: . Then we necessarily have since is not a limit ordinal but is. Clearly, by (2.2) (since by induction). Suppose that . Then necessarily (by Theorem 3.6(1)). If for some (recall that we identify and ), then and , but , a contradiction. Therefore, .
Case 2: . Notice that and by (2.6). We now complete the argument by induction on . The base of the induction is covered by Cases 1, 2 for above. □
An ideal of a Lie algebra is said to be characteristic if it is invariant under all automorphisms of , that is, for all . Clearly, an ideal is characteristic if and only if for all .
Corollary 3.11. All ideals of the Lie algebra are characteristic.
Proof. By definition, the members of the hypercentral series are characteristic ideals of . But Theorem 3.10 shows that these are all the ideals of . We can also deduce this corollary from Theorem 3.6(2). □
Corollary 3.12. For all non-zero elements and all automorphisms of we have .
The statement now follows from the fact that all the ideals are characteristic (Corollary 3.11).
3.6. The subalgebra of the Weyl algebra .
Let be an ideal of a Lie algebra . Then the ideal of the universal enveloping algebra generated by is equal to . The chain
of ideals of yields a chain of subalgebras and a chain of ideals of :
We put and , where is the algebra homomorphism (2.11). Then
is a chain of subalgebras of , and
is a chain of ideals of .
- 1)All the inclusions in (3.5) are strict. The algebra is not finitely generated.
- 2)All the inclusions in (3.6) are strict. In particular, is neither left Noetherian nor right Noetherian and does not satisfy the ascending chain condition for ideals.
Proof. 1. We use induction on . Suppose that . For every , the algebra is generated by the commuting elements since
These equalities mean that is isomorphic to the monoid algebra (via ), where is the submonoid of generated by the elements , . It follows that
This means that the inclusions are strict and, therefore, the algebra is not finitely generated. We have , is the skew polynomial algebra and for all . Therefore part 1 holds for .
Suppose that and part 1 holds for all . Let . Then is not a limit ordinal. Notice that . Hence . The elements commute. Moreover,
Therefore the algebra is commutative and is isomorphic to the monoid algebra (via ), where is the submonoid of generated by the elements . It follows that
This means that
(the inclusions are strict). Let . By Theorem 2.6(1) and (3.3),
By induction on , part 1 holds.
2. We make free use of the facts established in the proof of part 1. Arguing by induction on , we first take . For every the ideal of is generated by the commuting elements , where . By Corollary 2.7, the set is a -basis of . For each ,
It follows that
This means that
Since and , part 2 holds for .
Suppose that and part 2 holds for all . Let . Then is not a limit ordinal. Notice that . Hence . It follows from the equality and (2.2) that
Therefore,
This means that
Let . In view of (3.7) we also have
By induction on , part 2 holds. □
3.7. The Heisenberg Lie subalgebras of .
Let
be the Lie algebra of matrices, where are the elementary matrices. For every , let be the Lie algebra of upper triangular matrices, . The -linear map
is a Lie algebra monomorphism. We identify the Lie algebra with its image in .
The Heisenberg Lie algebra is the -dimensional Lie algebra with -basis , such that is a central element of and
where is the Kronecker delta. The -linear map
is a Lie algebra monomorphism. We identify with its image in . Since for , contains the , . For all positive integers and with we have .
Let be an algebra or a Lie algebra and an -module. Then is called the annihilator of . This is an ideal of . A module is said to be faithful if its annihilator is 0.
3.8. The -module .
To say that is a -module is the same as to say that is a -module. This obvious observation is important because it enables us to use the relations of the Weyl algebra in various computations with (since ). For every , the Lie algebra is a Lie subalgebra of , where is an ideal of and . In particular, is a left -module, where the action of on is given by the rule for all and . We recall that the polynomial algebra is a left -module.
- 1)The -linear map , , is an isomorphism of -modules.
- 2)The -module is indecomposable and uniserial. Moreover, and .
- 3)The set is the set of all non-zero -submodules of . We have if and only if . Moreover, for all .
- 4)The -submodules of are pairwise non-isomorphic indecomposable uniserial -modules.
Proof. To prove part 1, we notice that the map indicated is a bijection and a -homomorphism.
Let us prove part 3. By part 1, the -module can be identified with the ideal of . Under this identification, every -submodule of becomes an ideal of in , and vice versa. Part 3 now follows from part 1 and the classification of ideals of (Theorem 3.6(1)).
We now prove part 2. By part 3, the -module is uniserial and, therefore, indecomposable and . The Weyl algebra is a simple algebra, whence . Then we have because .
Let us prove part 4. By part 3 we have . Hence for all . The rest is obvious (see part 2). □
The following corollary describes the annihilators of all the -submodules of . In particular, it classifies the faithful ones.
- 1)The -submodule of is a faithful submodule if and only if .
- 2).
- 3)
Proof. We first prove part 2. The inclusion is obvious. The opposite inclusion follows since is a faithful -module (Lemma 3.14(2)), and is a -module.
To prove part 1, we notice that if is a submodule of , then . In view of this fact and part 2, to finish the proof of part 1, it suffices to show that the -module is faithful. Since , we have . Since and for all non-zero elements , we have .
Let us prove part 3. We use induction on . The base of the induction is , and there are three cases to consider: , and . The first case is part 1, the last case is obvious, and in the second,
Suppose that and part 3 holds for all . If , then the -module is faithful by part 1. If , then . If , that is, for some , then . By part 2 we have . Since , the inclusion is an inclusion of -modules. Moreover, the -submodule of can be identified with the -submodule of . The result now follows by induction on . □
Corollary 3.16. If , then the ideal is the least ideal of the Lie algebra which is a faithful -module (that is, ).
Proof. By Lemma 3.14(1) and Corollary 3.15(1), the -module is faithful, but its predecessor is not since
□
The inclusion of Lie algebras respects the total orderings on the bases and . The isomorphism , of -modules (Lemma 3.14(1)) induces a total ordering on the monomials of the polynomial algebra by the following rule: if and only if , that is, if and only if , , , and for some . This is the so-called reverse lexicographic ordering on or on ( if and only if ). By Lemma 3.4(3),(4) and Lemma 3.14(1), if (where ), then
- (i)for all such that ,
- (ii)for all and such that , that is, .
Let be the set of all non-zero submodules of the -module . This set is totally ordered with respect to . By Lemma 3.14(3), the map
is an isomorphism of totally ordered sets. Every non-zero polynomial can be written uniquely as a sum
The elements and are called the leading term and leading coefficient of respectively (with respect to the total ordering ). The ordinal number
(where ) is called the ordinal degree of . The following assertions hold for all non-zero polynomials and all :
- (i)provided that ,
- (ii),
- (iii)for all such that .
For every ordinal there is a unique representation
where and not all are equal to zero (notice the shift by 1 in the subscripts of the coefficients ). Then
where . Notice that if and only if is not a limit ordinal. The vector space possesses a largest monomial if and only if is not a limit ordinal, and in this case is the largest monomial in . The ordinal number can be written uniquely as a sum
where , and . The positive integers and are called the multiplicity and the comultiplicity of the ordinal number . The non-negative integers and are called the degree and the codegree of .
The following lemma is obvious.
Lemma 3.17. Suppose that , that is, with , , , and . Then
In particular, .
Remark 3.18. If , then the corresponding summand is absent.
Let be a vector space over . A linear map is called a Fredholm map/operator if it has finite-dimensional kernel and cokernel. Then the number
is called the index of . Let be the set of all Fredholm linear maps on . It is actually a monoid since
Let be a vector space with a countable -basis and let be a -linear map on given by the rule for all , where . For example, , and . The subalgebra of generated by the map is the polynomial algebra . Since is locally nilpotent, the algebra contains the algebra of formal power series in . The set is the group of units of the algebra . The vector space is a -module.
The following lemma is trivial.
- 1)We have
- 2)In particular, the non-zero elements of are surjective Fredholm maps.
- 3)If and , then .
Let be a vector space with -basis . We define a -linear map on by putting for all . For example, , and , . The polynomial algebra is a subalgebra of the algebra .
The following lemma is obvious.
- 1)We have (via ). In particular, the non-zero elements of are injective maps.
- 2)We have .
- 3)For all we have .
The following proposition describes the algebra of all -homomorphisms (and its group of units) of the -module . Clearly, is an embedding of -modules and (Lemma 3.19(1)). The -derivation of the polynomial algebra is also denoted by .
- 1)The map , , is an isomorphism of -algebras with inverse map , where for all and .
- 2)The map , , is an isomorphism of groups with the same inverse map as in part 1.
- 3)Every non-zero element is a surjective map with kernel , where .
- 4)For all integers we have and . In particular, is an isomorphism of -modules.
Proof. We first prove part 1. By Lemma 3.19, . Since and the maps commute, the restriction map is a well-defined homomorphism of -algebras. For all and ,
Therefore the restriction map is a monomorphism. To prove that it is surjective, we must show that for any map , its extension defined as in the statement of part 1 is a -homomorphism. The map is -linear. Hence we must only check that for all and .
Case 1: . For all and we have
Case 2: . Let and . Suppose that . Then
Suppose that . Then
(since ).
Part 2 follows from part 1. Part 3 follows from part 1 and Lemma 3.19. Part 4 follows from part 1. □
3.9. Monomial subspaces of the polynomial algebra .
Let be a subset of . The vector space is called a monomial subspace of the polynomial algebra with support . We put by definition. Clearly, .
Example 3.22. For every integer and any ordinal number , the vector space is a monomial subspace of . Hence so is their intersection
Clearly, . The following theorem describes the vector space and shows that the inclusion is always strict and .
Theorem 3.23. Suppose that , that is, , where , and . Then the following assertions hold.
- 1), where
- 2).
- 3)is a non-limit ordinal (that is, ) if and only if .
- 4)We have , and the set is a basis of the vector space .
- 5)The vector spaces are distinct. In particular, if , then .
Proof. We first prove part 1. Let be the right-hand side of the equality stated in part 1. Then (by Lemma 3.17). In particular, the set is non-empty.
Case 1: is a non-limit ordinal, that is, . In this case,
(see (3.11)), where , that is, and for . Suppose that . Then and there is an integer , , such that for all and . Then necessarily for all (since for all ) and, therefore, .
Case 2: is a limit ordinal, that is, . In this case,
(see (3.11)). Suppose that . Then if and only if for all (we recall that ) or, in other words, if and only if the following conditions hold:
- (i)for all ,
- (ii)for all .
Condition (i) is Case 1 for . Therefore,
where
Notice that and for . Condition (ii) excludes precisely the elements , that is, . The proof of part 1 is complete.
Part 2 follows from part 1. Part 3 follows from part 2.
We now prove part 4. The ordinal number is not a limit ordinal. By part 3 we have . Notice that and . By part 1,
Finally, by part 2,
and . The elements , , are the elements , , in part 1, but for the ordinal instead of . Clearly, for . Therefore the elements are -linearly independent in the vector space since the vector spaces and are monomial. These elements form a basis of the vector space since .
Part 5 follows from part 4. □
For example, for all positive integers and we have
Example 3.24. Consider the following conditions:
For every integer , any element and any ordinal number , the vector space is a monomial subspace of . Hence so is their intersection
For all ordinal numbers ,
The first inclusion follows since is a -module. If , then . The following corollary shows that these inclusions are strict and gives a -basis for every vector space .
Corollary 3.25. Suppose that , that is, , where , , , and . Then the following assertions hold.
- 1), where .
- 2)The vector spaces are distinct. In particular, if , then .
- 3).
- 4).
Proof. Part 1 follows at once from Lemma 3.17, the inclusions and Theorem 3.23.
Parts 2–4 follow from part 1. □
§ 4. The Lie algebras are locally finite-dimensional and locally nilpotent
The aim of this section is to prove Theorem 4.2. The key ideas are to use the fact that the algebra is uniserial, induction on the ordinals and Theorem 4.1, which gives sufficient conditions for a Lie algebra to be nilpotent.
Theorem 4.1. Let be an ideal of a Lie algebra such that the Lie factor algebra is a finite-dimensional nilpotent Lie algebra, is a nilpotent Lie algebra, and every element acts nilpotently on (that is, for some positive integer ). Then is a nilpotent Lie algebra.
Proof. We use induction on . The case (that is, ) is obvious.
Suppose that . This is the most important case since we will reduce the general case to this one. Then , where . Put and . The Lie algebra is nilpotent, that is,
for some integer . For all elements we have
whence . By hypothesis, the map acts nilpotently on . Thus, increasing if necessary, we can assume that . To prove that is a nilpotent Lie algebra, we must show that for some non-negative integer (the equality implies that ). It suffices to show that
for some non-negative integer since . We claim that it suffices to take . Indeed, take . Then
where is the set of finite linear combinations of the elements , where is a word of length in the alphabet containing precisely elements and elements . It suffices to show that for all .
Notice that and for all . For any with we have since and . For any with , the word is of the form , where . At least one of the numbers is greater than or equal to since otherwise we have
a contradiction. Then since and . Therefore .
Suppose that . The Lie algebra is nilpotent. We fix an ideal of such that . Let be the canonical Lie algebra epimorphism . The ideal of has codimension 1 and is an ideal of the Lie algebra . The pair satisfies the hypotheses of the theorem and . By induction on , the Lie algebra is nilpotent. Now the pair is as in the case considered above. Indeed, , and the element in the decomposition acts nilpotently on since is a finite-dimensional nilpotent Lie algebra. In more detail, put . Then for some since is a finite-dimensional nilpotent Lie algebra; for some by hypothesis. Hence . Therefore is a nilpotent Lie algebra. □
Theorem 4.2. The Lie algebras are locally finite-dimensional and locally nilpotent.
Proof. We must show that the Lie subalgebra of generated by a finite set of elements, say, , is a finite-dimensional nilpotent Lie algebra. The case is trivial. We may assume without loss of generality that , the elements are -linearly independent and
We use induction on the ordinal number . By definition, is a non-limit ordinal. The initial case is obvious since . Thus, let be a non-limit ordinal such that and assume that the result holds for all non-limit ordinals such that . Then is a non-limit ordinal and for all . Let be the Lie subalgebra of generated by the elements . By induction, is finite-dimensional and nilpotent. The inner derivation of is locally nilpotent (Proposition 2.1(5)) and . It follows that for some non-negative integer . The vector space is a finite-dimensional -invariant subspace of the ideal (see (3.3)). Let be the Lie subalgebra of generated by . By induction, is finite-dimensional and nilpotent. Then for some integer . Clearly, since and is a derivation. We see that (clearly, ; on the other hand, is a Lie subalgebra of that contains the elements , and so ). We claim that the hypotheses of Theorem 4.1 hold for the pair , whence is a nilpotent Lie algebra (by Theorem 4.1) and . Indeed, to prove the claim, it suffices to show that for all , where . We fix a number such that and . For the same reason as in the proof of Theorem 4.1, it suffices to take :
□
§ 5. The isomorphism problem for factor algebras of the Lie algebras
We already know that the Lie algebras and are not isomorphic for . The aim of this section is to answer the following question.
Question. Let and be ideals of the Lie algebras and respectively. When are the Lie algebras and isomorphic?
We first consider the case when (Theorem 5.1) and then deduce the general case from this one (Corollary 5.2). Put .
- 1)Let be an ideal of , that is, for some (Theorem 3.6), where . Then the Lie algebras and are isomorphic if and only if , where .
- 2)Let and be ideals of , that is, and for some (Theorem 3.6(1)). Then the Lie algebras and are isomorphic if and only if one of the following conditions holds:
- (a)and , where and ;
- (b)and , where , with and ;
- (c), where .
For all integers and with there is a natural isomorphism of Lie algebras
where , .
In more detail, .
Corollary 5.2. Let and be integers with , and let be ideals of , respectively. Then and are isomorphic if and only if
We recall that , where is a Lie subalgebra of and is an Abelian ideal of . We introduced the -basis for in (2.1). Define a -linear map in one of the two equivalent ways:
Lemma 5.3. For every integer , the -linear map is an epimorphism of Lie algebras and .
Proof. By the definition of we have
To finish the proof of the lemma, it suffices to show that the -linear map is a homomorphism of Lie algebras, that is,
for all elements and of the basis of (see (2.1)). Since , we have . The equality (5.3) is obvious if either (since for all ), or (since and ). In the remaining cases when and for some , the equality follows from the following three facts: is a central element of (Proposition 2.1(6)), and . Indeed, applying the inner derivation of to the equality (where ), we obtain the required equality
□
Corollary 5.4. If , then the ideal is the least ideal of such that is isomorphic to .
Proof. Clearly, and . We suppose that and seek a contradiction. Then for some . Fix an integer such that . Then . Since there is a natural epimorphism of Lie algebras , we have . By Lemma 5.3, . Hence,
a contradiction. □
Proof of Theorem 5.1. Let us prove part 1. By Lemma 5.3, for every integer the map is a Lie algebra epimorphism with kernel . Therefore . The implication follows from part 2.
To prove part 2, we use induction on . The initial step is a direct corollary of Lemma 5.3 and the classification of the ideals of . By Theorem 3.6(1), the proper ideals of are , , and . By Lemma 5.3, for all , and , and part 2 follows.
Suppose that and the result holds for all . We recall that , and . Since is uniserial and Artinian, so are all its factor algebras. Thus, if , then .
Step 1: for all .
Indeed, implies that for some . Then (Lemma 5.3) and, therefore,
whence .
Suppose that for some and .
Step 2: It suffices to consider the case when .
Indeed, if , then part 2 follows by induction on because , and . We can assume without loss of generality that . The case when is impossible by Step 1 since
a contradiction. This finishes the proof of step 2.
Step 3: In view of Lemma 5.3 we may assume that .
The idea of the proof of the theorem is to introduce, for every Lie factor algebra with , a quantity which is invariant under isomorphisms and takes distinct values for the ordinal numbers . This invariant is the uniserial dimension of certain ideals of . For each ordinal number let be the unique ideal of the uniserial Artinian Lie algebra with uniserial dimension . Every ideal of a uniserial Artinian Lie algebra is characteristic. This fact is crucial for our arguments below. Clearly,
We put and . Notice that . For every element we put .
Step 4: For all and , we have .
Indeed, the second equality follows from two facts: and, for all , .
The first equality follows since for some elements , and
Alternatively, using the equality and the fact that and are characteristic ideals of , we see that
Since and for all automorphisms , Step 4 means that the ordinal number
is an invariant (under isomorphisms) of the algebra , where is the uniserial dimension of the -module . If for some ordinals , then , whence . □
Corollary 5.5. Let be an ideal of , where , and . Then the following assertions hold.
- 1)for all .
- 2)for all and such that .
- 3)for all , where .
Proof. Part 1 is Step 1 in the proof of Theorem 5.1. Part 2 follows from part 1 and (5.1). Part 3 is trivial. □
Proof of Corollary 5.2. Notice that the uniserial dimensions , , are distinct, and if , then . The isomorphism of Lie algebras (see (5.1)) induces a bijection between the set of all non-zero ideals of the Lie algebra and the set of all ideals of the Lie algebra that properly contain the ideal :
The corollary now follows from Theorem 5.1, Corollary 5.5 and (5.4). □
§ 6. The Lie algebra
In this section we study the Lie algebra in detail. Many properties of are similar to those of , , but there are several differences. For example, is insoluble and non-Artinian (but almost Artinian) with . We obtain a classification of the ideals of (Theorem 6.2). All of them are characteristic ideals (Corollary 6.4(2)). We give an isomorphism criterion for the Lie factor algebras of (Corollary 6.3).
Let be the polynomial algebra in countably many variables, and let be the infinite Weyl algebra. The Lie algebra is a Lie subalgebra of the Lie algebra . The polynomial algebra is an -module. In particular, is a -module. The Lie algebra
is the direct sum of the Abelian (infinite-dimensional for ) Lie subalgebras . For every integer ,
is an ideal of by (2.3). Clearly, for all . contains a strictly descending chain of ideals
and for all .
- 1)is insoluble.
- 2)is locally nilpotent and locally finite-dimensional.
- 3)Each element acts locally nilpotently on the -module .
- 4)
- 5)The lower central series of stablilizes at the first step, that is, and for all .
- 6)All inner derivations of are locally nilpotent.
- 7)The centre of is 0. In particular, .
- 8)The Lie algebras and , where , are not isomorphic.
- 9)contains a copy of every nilpotent finite-dimensional Lie algebra.
- 10)The inner derivation , , is a nilpotent derivation of if and only if .
Proof. Part 1 follows from part 4. Part 2 follows from Theorem 4.2 since . Part 3 follows from Proposition 2.1(4) since and .
Parts 4, 5 follow from (2.3) and the decomposition . Part 6 follows from Proposition 2.1(5) since and .
Let us prove part 7. If , then for some , whence (Proposition 2.1(6)). Since , we must have .
Let us prove part 8. The Lie algebra is soluble (Proposition 2.1(1)), but is not (part 1). Therefore for all . By Proposition 2.1(7), the , , are pairwise non-isomorphic.
Let us prove part 9. Any finite-dimensional nilpotent Lie algebra is a subalgebra of the Lie algebra for some . The result now follows from the inclusions .
We finally prove part 10. It suffices to show that if , then the derivation is not nilpotent. We suppose that this is not the case for some and seek a contradiction. We can write the element as a sum , where for all and . In view of the Lie algebra isomorphism , the inner derivation induced by is a nilpotent derivation of . By Lemma 2.2, , a contradiction. □
The following theorem gives a list of all ideals of .
Theorem 6.2. The set of all non-zero ideals of is equal to the set , where is the ideal of as defined in (3.3), that is, . In particular, every non-zero ideal of contains an ideal for some , and if , then .
Proof. Let be a non-zero ideal of , and let be a non-zero element of . Then
for some integers , and . Let be the total degree of the polynomial , where . Fix an such that . Applying to the element , we get an element of of type . Thus, without loss of generality, we may assume from the very beginning that , that is, . Then
Hence for all . This means that . Consider the following epimorphism of Lie algebras (where ):
The image of is an ideal of such that , by the definition of the number . We may assume without loss of generality that . Then for some by Theorem 3.6(1). The following two facts are obvious:
- (i)if and only if either , or and ,
- (ii)for all .
They yield that is uniserial (that is, for any distinct ideals , of , either or ). Hence the chain
contains all the non-zero ideals of . □
In combination with Theorem 5.1 and Corollary 5.2, the following corollary gives an isomorphism criterion for the Lie factor algebras of .
- 1)An ideal of satisfies if and only if .
- 2)
In contrast to the Lie algebras , , no proper Lie factor algebra of is isomorphic to .
A Lie algebra is said to be almost Artinian if all its proper factor algebras are Artinian Lie algebras (that is, for every non-zero ideal of , the factor algebra is Artinian).
- 1)The Lie algebra is uniserial, non-Artinian, non-Noetherian, and almost Artinian. Its uniserial dimension is given by .
- 2)All the ideals of are characteristic.
Proof. We first prove part 1. We have just seen that is uniserial (see (6.3)). By Theorem 6.2, is neither Artinian, nor Noetherian, but almost Artinian (since all the are Artinian; see Theorem 3.6(2)). By Theorem 6.2, every non-zero ideal of contains the ideal for some , and the factor algebra is a uniserial Artinian Lie algebra, whence
by Theorem 3.6(2).
We now prove part 2. The ideals , , are characteristic because they form the derived series of (Proposition 6.1(4)). By Theorem 6.2 it remains to show that every ideal is characteristic. Let be an automorphism of . Since , induces an automorphism of . Then by Corollary 3.11, where is as in (6.2). Therefore, , as required. □
Let be the universal enveloping algebra of . The chain of Lie algebras gives a chain of universal enveloping algebras.
- 1)All the inner derivations of the universal enveloping algebra of are locally nilpotent.
- 2)Every multiplicative subset of which is generated by an arbitrary set of elements of is a left Ore set and a right Ore set in . Therefore, .
Proof. Both parts follow from Corollary 2.3 since . □
§ 7. The Lie algebra
In this section we study the completion of with respect to the ideal topology. Its properties diverge further from those of the , , and . For example, none of the non-zero inner derivations of is locally nilpotent, and is neither locally nilpotent nor locally finite-dimensional. The main result of this section is a classification of the closed and open ideals of the topological Lie algebra (Theorem 7.2(1)). As a result, we prove that all the open ideals and all the closed ideals of are topologically characteristic (Corollary 7.3(3)).
is a topological Lie algebra. Its topology is the ideal topology, that is, the topology with basis , where is the set of non-zero ideals of . We recall that this means that the maps , , , , and , , are continuous, where the topologies on and are the product topologies, and the topology on is the discrete topology. By Theorem 6.2, every non-zero ideal of contains the ideal for some . Hence we can take the ideals instead of all non-zero ideals in the definition of the topology on .
The completion of the topological space is a topological Lie algebra:
where is an infinite sum which is uniquely determined by its coefficients . The inclusion is an inclusion of topological Lie algebras, and the topology on is (by definition) the strongest topology on such that the map , , is continuous.
The Lie algebra contains a strictly descending chain of ideals
and for all . One can also regard as the projective limit of the projective system of Lie algebra epimorphisms
where the epimorphism is the composite of the natural epimorphism and the isomorphism . Each closed ideal of is the completion/closure of the ideal of . For every element , the set is a basis of open neighbourhoods of . These sets , , are open and closed in .
The polynomial algebra is a left -module: for all and .
- 1)The Lie algebra is insoluble.
- 2)is neither locally nilpotent nor locally finite-dimensional.
- 3)Each element acts locally nilpotently on the -module .
- 4)
- 5)The lower central series of stabilizes at the first step, that is, and for all .
- 6)None of the non-zero inner derivations of are locally nilpotent.
- 7)The centralizer is a maximal Abelian Lie subalgebra of .
- 8)The centre of is equal to 0.
- 9)The Lie algebras , where , and are pairwise non-isomorphic.
Proof. Part 1 follows from part 4.
To prove part 2, we put . Then
Hence the elements are -linearly independent. Therefore the Lie subalgebra of generated by the elements and is neither nilpotent nor finite-dimensional.
Part 3 follows from Proposition 6.1(3).
To prove part 4, we put for . The assertion follows from the following obvious facts: and for all . They immediately yield that . The reverse inclusion follows from Proposition 6.1(4) since .
Part 5 follows since and . The first equality yields that , and the second that for .
Let us prove part 7. We have for all . Since the centralizer is an Abelian Lie subalgebra of , it is automatically maximal Abelian.
To prove part 8, let be a central element of . By part 7, for all . For all we have , that is, .
Let us prove part 9. We already know that the , where , and are pairwise non-isomorphic (Proposition 6.1(8)). The are soluble (Proposition 2.1(1)), but is not (part 1). Hence for all . is locally nilpotent (Proposition 6.1(2)), but is not (part 2). Hence .
We now prove part 6. Since the centre of is equal to 0 (part 8), we must show that if is a non-zero element of , then the inner derivation is not a locally nilpotent map. Write , where , and is the smallest number in . If , then
Hence is not a nilpotent map. □
The following theorem gives a classification of the open ideals and closed ideals of the topological Lie algebra . Every non-zero closed ideal of is open and vice versa.
- 1)The set of all open ideals of is equal to the set of all non-zero closed ideals of and also to the set , where is the ideal of defined in (3.3), that is, . In particular, every non-zero closed ideal of contains the ideal for some , and if , then .
- 2)A non-zero ideal of is open if and only if it is closed.
- 3)The topological Lie algebras and are Hausdorff.
- 4)The zero ideal is a closed non-open ideal of .
Proof. We prove part 1 for open ideals only. For closed ideals, it will follow from part 2. Every open ideal of is necessarily non-zero. Clearly, the ideals , and in the statement of part 1 are open. Let be an open ideal of such that for all . We must show that for some and as in the statement of part 1. The ideal contains an open neighbourhood of the zero element, that is, for some integer . We can assume that . In view of the Lie algebra isomorphism , our claim and part 1 follow from Theorem 3.6(1) and the minimality of .
We now prove part 2. Let be an open ideal. By part 1 for open ideals, is one of the ideals listed in part 1. All the ideals listed in part 1, that is, , and , are obviously non-zero closed ideals. Hence is a non-zero closed ideal.
Let be a non-zero closed ideal in . It suffices to show that for some . This will automatically imply that is open. The ideal is the closure of the ideal of . We fix a non-zero element of , say , where , for all and . Let be the total degree of the polynomial , where . We fix an such that . Applying to the element , we get an element of of type . Thus, without loss of generality, we may assume from the outset that , that is, . For every we have
whence . For all integers , the ideal of generated by the subspace is equal to . Therefore, . It follows that the closure of the ideal belongs to , as required.
We now prove part 3. Let and be distinct elements of (resp. ), where for all . Then for some , whence (resp. ). This means that and are Hausdorff.
Part 4 is obvious. □
A topological Lie algebra is said to be closed uniserial (resp. open uniserial) if the set of all closed (resp. open) ideals of is totally ordered by inclusion. By Theorem 7.2(1), the Lie algebra is closed uniserial and open uniserial. Hence the chain
contains all the open ideals and non-zero closed ideals of .
A topological Lie algebra is said to be open Artinian (resp. closed Artinian) if the set of open (resp. closed) ideals of satisfies the descending chain condition. A topological Lie algebra is said to be open Noetherian (resp. closed Noetherian) if the set of open (resp. closed) ideals of satisfies the ascending chain condition. A topological Lie algebra is said to be open almost Artinian (resp. closed almost Artinian) if, for every non-zero open (resp. closed) ideal of , the Lie factor algebra is Artinian. For a topological Lie algebra , let
be its group of automorphisms. Every element of is an isomorphism both of the Lie algebra and of the topological space . An ideal of a topological Lie algebra is said to be topologically characteristic if for all .
- 1)The topological Lie algebra is open uniserial, closed uniserial, open almost Artinian and closed almost Artinian. It is neither open nor closed Artinian and neither open nor closed Noetherian.
- 2)The uniserial dimensions of the sets of open and closed ideals of the topological Lie algebra coincide and are equal to .
- 3)All the open/closed ideals of the Lie algebra are topologically characteristic.
Proof. Part 1 follows from Theorem 7.2(1). Part 2 follows from Theorem 7.2(1) and Corollary 6.4(1).
Let us prove part 3. The ideals are topologically characteristic since they form the derived series of (Proposition 7.1(4)). By Theorem 7.2(1), it remains to show that every ideal of is topologically characteristic. Take any . Since , induces an automorphism of the Lie factor algebra . Since and all the ideals of are characteristic (Corollary 3.11), we must have , that is, is a topologically characteristic ideal of . □