New Research In
Physical Sciences
Social Sciences
Featured Portals
Articles by Topic
Biological Sciences
Featured Portals
Articles by Topic
 Agricultural Sciences
 Anthropology
 Applied Biological Sciences
 Biochemistry
 Biophysics and Computational Biology
 Cell Biology
 Developmental Biology
 Ecology
 Environmental Sciences
 Evolution
 Genetics
 Immunology and Inflammation
 Medical Sciences
 Microbiology
 Neuroscience
 Pharmacology
 Physiology
 Plant Biology
 Population Biology
 Psychological and Cognitive Sciences
 Sustainability Science
 Systems Biology
Lie superalgebras graded by P(n) and Q(n)

Contributed by Efim I. Zelmanov, May 6, 2003
Abstract
In this article we study Lie superalgebras graded by the root systems P (n) and Q(n).
1. Introduction
In ref. 1 Berman and Moody initiated the study of Lie algebras graded by a finite root system. Let 𝒢 be a finite dimensional split simple Lie algebra over a field F of characteristic zero and have a root space decomposition relative to a split Cartan subalgebra H.
Definition 1.1: A Lie algebra L over F is graded by the root system Δ if (i) L contains , (ii) L = ∑_{α∈Δυ{0}}L_{α}, where L_{α} = {x ∈ L[h, x] = α(h)x for all h ∈ H} for α ∈ Δ υ {0}, and (iii) L_{0} = ∑_{α∈Δ}[L_{α}, L_{α}].
The motivation for the study of this class of algebras comes from the intersection matrix algebras of Slodowy (see ref. 2) and the extended affine Lie algebras (see ref. 3) that are graded by a finite root system.
Examples:

Let R be an associative unital F algebra, and let gl_{n}(R) be the Lie algebra of all n × n matrices over R with the commutator product. The subalgebra e_{n}(R) of gl_{n}(R) generated by the elements e_{ij}(a), 1 ≤ i ≠ j ≤ n, a ∈ R is graded by the root system A_{n1}. For n = 2 one has to consider a Steinberg algebra (see ref. 1) st_{2}(R) where the algebra R may be assumed to be alternative (see refs. 4 and 5).

Let R be an associative commutative unital algebra, and let 𝒢 be the split simple Lie algebra with the root system Δ. Then is a Δgraded algebra.

Generalized Tits constructions (see refs. 6 and 7) involving the split octonion algebra and the exceptional 27dimensional split Jordan algebra (see ref. 4) are G_{2} and F_{4}graded, respectively.
It is appropriate to study rootgraded Lie algebras up to central isogeny. Any perfect Lie algebra L has a universal central extension that is also perfect called a universal covering algebra of L (see ref. 8). We denote it as L̃.
Definition 1.2: Two perfect Lie algebras L_{1} and L_{2} are said to be centrally isogenous if L̃_{1} ≃ L̃_{2}.
In ref. 1 Berman and Moody proved that an arbitrary A_{n}graded algebra (n ≥ 2) is centrally isogenous with st_{n+1}(R) for some associative (for n = 2 alternative) algebra R. For other simply laced root systems Δ = D_{n}, n ≥ 4, E_{6}, E_{7}, E_{8} a Δgraded algebra is centrally isogenous with , where R is an associative commutative algebra. Rootgraded Lie algebras for nonsimply laced root systems were classified in ref. 7 (for more Jordan approach to C_{n}, B_{n} see also ref. 9). Remark that the classification of rootgraded algebras in refs. 1 and 7 led to a complete description of intersection matrix algebras. In refs. 1012 Benkart and Elduque extended the theory discussed above to Lie superalgebras and completely determined rootgraded Lie superalgebras for all finite dimensional split simple classical superalgebras except P(n), , and A(n, n) (for notation and results on Lie superalgebras see ref. 13).
Let 𝒢 be a finite dimensional split simple classical Lie superalgebra with even part . Let H be a Cartan subalgebra of , and let be the decomposition of 𝒢 into the sum of eigenspaces with respect to the action of H, .
Definition 1.3 (see ref. 10): A Lie superalgebra L over F is graded by the root system Δ if (i) L contains a subsuperalgebra ; (ii) see Definition 1.1; and (iii) see Definition 1.1.
In this article we discuss Lie superalgebras graded by the root systems of the superalgebras P(n), , n ≥ 2.
Recall that the Lie superalgebra is the superalgebra of 2(n + 1) × 2(n + 1) matrices of the type (), where a, b are (n + 1) × (n + 1) matrices, tr(b) = 0, and . The superalgebra is the universal central extension of .
The Lie superalgebra P(n  1) is the superalgebra of 2n × 2n matrices of the type (), where a, k, h are n × n matrices over F, Tr(a) = 0, k^{T} = k, h^{T} = h. The Cartan subalgebra H of consists of diagonal matrices H = {h = diag(a_{1},..., a_{n}, a_{1},..., a_{n})  a_{i} ∈ F, a_{i} = 0}. The even and odd roots are Thus P(n) is graded by the free abelian group .
Let's fix the notation for the following weight elements: For n ≠ 3 the universal covering superalgebra of P(n) is P(n). However, P(3) has a unique nontrivial universal central extension with ¢.
Keeping in mind the intersection matrix superalgebras and our focus on subspaces of nonzero weight, we will modify part i of Definition 1.3, allowing the superalgebra to contain a central extension of 𝒢 instead of 𝒢 Let be a central extension of 𝒢; that is, , and there exists an epimorphism such that Kerπ Center(). It follows from the classification (see ref. 13) that if a classical simple Lie superalgebra 𝒢 has a nontrivial central extension , then is semisimple; hence is a direct summand of . We will consider the root decomposition of L with respect to the Cartan subalgebra of the ideal of .
Definition 1.4: A Lie superalgebra L over F is graded by the root system Δ if

L contains a central extension of 𝒢,

For a Cartan subalgebra H of we have, , where for all h ∈ H} for all , and

.
Theorem 1.1.A Lie superalgebra that is graded by P(n), n ≠ 3 is isomorphic to a tensor product , where R is an associative commutative superalgebra.
In ref. 14 Cheng and Kac introduced a new superconformal algebra that they denoted as CK(6). Simultaneously and independently this superalgebra was constructed by Grozman et al. (see ref. 15). In ref. 16 for an arbitrary associative commutative superalgebra R with an even derivation d : R → R we constructed a superalgebra CK (R, d) so that CK(6) ≃ CK(F[t^{1}, t], d/dt).
Consider the associative Weyl algebra W = ∑_{i≥0} Rd^{i}, where the variable d commutes with a coefficient a ∈ R via da = d(a) + ad. We will realize the CK(R, d) as a superalgebra of 8 × 8 matrices over W. Implicitly this realization was mentioned in ref. 17.
The Lie algebra of skewsymmetric 4 × 4 matrices K_{4}(F) is a direct sum of two ideals . For an arbitrary element k ∈ K_{4}(F) we consider its decomposition k = k′ + k″ and let .
The universal central extension of P(3) can be realized as a superalgebra of 8 × 8 matrices over F[d] of the type where a, k, h are 4 × 4 matrices over F, Tra = 0, k^{T} = k, h^{T} = h, α ∈ F, and I is the identity matrix. The superalgebra CK(R, d) is a subsuperalgebra of 8 × 8 matrices over W generated by and by all matrices , where a ∈ R, 1 ≤ i ≠ j ≤ 4.
Theorem 1.2.A P(3)graded Lie superalgebra is centrally isogenous with CK(R, d), where R is an associative commutative superalgebra, and d : R → R is an even derivation. The superalgebra CK(R, d) is centrally closed; that is, .
Remark: If a Lie superalgebra L is P(3)graded in the sense of Definition 1.3, then , where R is an associative commutative superalgebra.
Classification of graded Lie superalgebras immediately follows from the classification of A_{n1}graded Lie algebras by Berman and Moody (see ref. 1) (see section 3).
Theorem 1.3.A graded Lie superalgebra is centrally isogenous to st_{n+1}(R), where is an associative or an alternative (if n = 2) unital superalgebra such that there exists.
2. Jordan Systems
Consider a system of vector spaces J = (J_{α}, α ∈ Δ) with transformations whenever α, β, α + β ∈ Δ, and for all α, β ∈ Δ. The trilinear operations ψ_{α,α,β} give rise to the mappings , where the projection of to is the mapping .
We call this system a Jordan system (see refs. 7 and 18) if the direct sum where , becomes a Lie algebra with respect to the operation:
A homomorphism between two Jordan systems, J = (J_{α}, α ∈ Δ) and J′ = (J′_{α}, α ∈ Δ), is a family of linear mappings, f = (f_{α}, α ∈ Δ), f_{α} : J_{α} → J′_{α}, which preserves the bilinear and the trilinear operations.
Every Δgraded Lie algebra gives rise to a Jordan system (L_{α}, α ∈ Δ) with the bilinear operations if α, β, α + β ∈ Δ and the trilinear operations .
In ref. 7 it was shown that two Δgraded Lie algebras and are centrally isogenous if and only if the Jordan systems (L_{α}, α ∈ Δ) and (L′_{α}, α ∈ Δ) are isomorphic.
Everything that we said about Jordan systems and their connections with Δgraded Lie algebras obviously extends to superalgebras.
Lemma 2.1.Let and be two P(n)graded Lie superalgebras, n ≥ 2. Letbe a family of linear mappings such thatfor arbitrary roots α, β ∈ Δ, α + β ∈ Δ, and arbitrary elements x_{α} ∈ L_{α}, y_{β} ∈ L_{β}. Then f is a homomorphism of Jordan supersystems.
3. A_{n1}Graded Superalgebras
Definition 3.1 (see ref. 1): Let R be a unital (super)algebra, n ≥ 3. The Steinberg Lie (super)algebra st_{n}(R) is presented by generators X_{ij}(a), 1 ≤ i, j ≤ n, i ≠ j and relators X_{ij}(αa + βb) = αX_{ij}(a) = βX_{ij}(b); [X_{ij}(a), X_{jk}(b)] = X_{ik}(ab) if i, j, k are distinct; [X_{ij}(a), X_{kt}(b)] = 0 if i ≠ t and j ≠ k.
Let L be an A_{n1}graded Lie algebra, n ≥ 3. Thus . Let e_{ij}, 1 ≤ i ≠ j ≤ n denote the matrix units from sl_{n}(F). Berman and Moody (see ref. 1) proved that there exists a unital algebra R such that the Jordan system (L_{α}, α ∈ Δ) is isomorphic to the Jordan system (X_{ij}(R), 1 ≤ i ≠ j ≤ n) of the Steinberg Lie algebra st_{n}(R). If n ≥ 4, then the algebra R is associative. For n = 3, R has to be alternative. It is proved in ref. 19 that the maps R → X_{ij}(R) are injective.
Moreover, there exists an isomorphism Ψ = (ψ_{ij}, 1 ≤ i ≠ j ≤ n) between these Jordan systems such that ψ_{ij}(e_{ij}) = X_{ij}(1), 1 ≤ i ≠ j ≤ n.
If the algebra R is associative, in particular if n ≥ 4, then st_{n}(R) is the universal central extension of the Lie algebra sl_{n}(R) generated by the matrices e_{ij}(a), 1 ≤ i ≠ j ≤ n, a ∈ R.
Following the BermanMoody proof (verbatim) we get the following.
Lemma 3.1.Letbe a Lie superalgebra, the even part of which contains sl_{n}(F) = H + Σ_{α∈An1} G_{α}. Suppose further thatL_{α}, where L_{α} = {x ∈ L  [h, x] = α(h)x}, and. Then there exists a superalgebraand an isomorphism of Jordan supersystems Ψ : (X_{ij}(R), 1 ≤ i ≠ j ≤ n) → (L_{α}, α ∈ A_{n1}) such that ψ_{ij}(X_{ij}(1)) = e_{ij}.
This lemma implies Theorem 1.3. Indeed, the root system of the superalgebra is A_{n}. Hence the graded Lie superalgebra L is A_{n}graded. Therefore L is centrally isogenous to st_{n+1}(R), where is an associative or an alternative (if n = 2) superalgebra. An A_{n}graded superalgebra graded if and only if it contains a subsuperalgebra 𝒢 isomorphic to or to such that .
Let us show that if there is with ν^{2} = 1, then L is graded.
Consider an isomorphism of Jordan systems π: (X_{ij}(R), 1 ≤ i ≠ j ≤ n + 1) → (L_{wiwj}, 1 ≤ i ≠ j ≤ n + 1). The Jordan system (X_{ij}(F1 + Fν), 1 ≤ i ≠ j ≤ n + 1) is isomorphic to the Jordan system . Hence the subsuperalgebra of L generated by the subspaces π(X_{ij}(F1 + Fν)), 1 ≤ i ≠ j ≤ n + 1, is isomorphic to .
Now suppose that . We will show that there exists with ν^{2} = 1. Let be an isomorphism of Jordan systems.
By Lemma 3.1 we can assume that
Suppose that Then
Hence which implies ν^{2} = 1. Theorem 1.3 is proved.
4. P(n  1)Graded Superalgebras: Associativity and Commutativity of R
Let be the root system be a Δgraded Lie superalgebra. Then is the Lie superalgebra graded by A_{n1}. By Lemma 3.1 there exists a superalgebra and an isomorphism of Jordan systems such that ψ_{ij}(X_{ij}(1)) = e_{wiwj}. For arbitrary 1 ≤ i ≠ j ≤ n, arbitrary a ∈ R denote ψ_{ij}(e_{ij}(a)) as e_{wiwj}(a). Clearly, .
The superalgebra L contains a central extension of the superalgebra P(n  1). If n ≠ 4, then P(n  1) does not have nontrivial central extensions, so . If n = 4, then L contains either P(3) or its universal central extension . Recall that ¢. The elements h_{wiwj}, e_{wiwj}, q_{wiwj}, q_{wi+wj} are multiplied as in P(3) except for . The element ¢ lies in the center.
For 1 ≤ i, j ≤ n, a ∈ R choose , and let .
Lemma 4.1.

For k, t ∈ {1, 2,...,n}  {i,j} we have Thus the definition of q_{wiwj}(a) does not depend on the particular choice of k.

q_{wiwj}(a) = q_{wjwi}(a).
Lemma 4.2. Let 1 ≤ i, j, k ≤ n be distinct. Then for arbitrary elements a, b ∈ R we have
Lemma 4.3.Let 1 ≤ i,j ≤ n be distinct. Then for arbitrary elements a, b ∈ R we have.
Lemma 4.4.For 1 ≤ i ≠ j ≤ n and arbitrary elements a, b ∈ R we have
Lemma 4.5..
Proof: Denote . By Lemmas 4.2 and 4.3 we have
Hence
An element has weight 2w_{i}  w_{k}  w_{t}, which is a root only when n = 3 or 4 and the integers i, k, t are distinct. In both cases and it remains to notice that 2w_{i} + w_{k}  w_{t}, 2w_{i}  2w_{k} are not roots.
Hence . Lemma 4.5 is proved.
Let 1 ≤ i ≠ j ≤ n, a ∈ R. Assuming that n ≠ 4 denotes q_{wi+wj}(a) = , where 1 ≤ k ≤ n, k ≠ i,j.
Lemma 4.6.

The definition above does not depend on the choice of k. In other words, if 1 ≤ t ≤ n, t ≠ i, j, then.

q_{wj+wi}(a) = ^{}q_{wi+wj}(a).
Remark: For n = 4 the expression may depend on the choice of k.
For n = 4 define q_{w1+w2}(a) = .
Lemma 4.7.

Let n ≠ 4. For arbitrary integers 1 ≤ i, j, k ≤ n, arbitrary a, b ∈ R we have

For n = 4, for arbitrary a, b ∈ R, we have
Lemma 4.8. The superalgebra R is associative and commutative.
Proof: Let a, b, c be arbitrary homogeneous elements from R. From and the Jacobi identity, it follows that
Let us determine the left and righthand sides separately.
The lefthand side is equal to
The righthand side is equal to
We proved that
Let c = 1. Then ab + (1)^{ab}ba = (1)^{ab}(2ba). Hence ab = (1)^{ab}ba and the superalgebra R is commutative. Hence (1)^{ab}2(ba)c = 2(ab)c = 2(1)^{ab}b(ac), and Lemma 4.8 is proved.
We will consider separately the case n = 4 in the next section. From now on in this section we assume that n ≠ 4.
Lemma 4.9.For arbitrary distinct integers 1 ≤ i, j, k ≤ n and arbitrary elements a, b ∈ R we have
Lemma 4.10.For arbitrary 1 ≤ i ≠ j ≤ n, arbitrary elements a, b ∈ R we have
Proof: Choose k ≠ i, j. We have and Lemma 4.10 is proved.
Lemma 4.11.For arbitrary 1 ≤ i, j, k, t ≤ n, i ≠ j, k ≠ t we have
Proof of Theorem 1.1: We claim that e_{wiwj}(R), 1 ≤ i ≠ j ≤ n, q_{wiwj}(R), 1 ≤ i, j ≤ n, and q_{wi+wj}(R), 1 ≤ i ≠ j ≤ n, exhaust all nonzero eigenspaces L_{α}, α ≠ 0. For α ∈ this is obvious. If α ∈ , then there exists 1 ≤ i ≠ j ≤ n such that (α  w_{i}  w_{j}) ≠ 0 and α ≠ ±(w_{i} + w_{j}). Hence which proves the assertion.
Lemmas 2.1, 4.24.5, 4.7, and 4.94.11 imply that the Jordan supersystems (L_{α}, α ∈ Δ) and (P(n)_{α}R, α ∈ Δ) are isomorphic. Hence L is centrally isogenous with P(n) R. From Proposition 6.1 (which will be proved later) it follows that L ≃ P(n) R, and Theorem 1.1 is proved.
5. ChengKac Superalgebras
Let L be a P(3)graded Lie superalgebra, , where 𝒢 = P(3) or , H is a Cartan subalgebra of , L = L_{α} is the decomposition into the sum of eigenspaces with respect to H, L_{0} = ∑_{α∈Δ} . The elements e = e_{w1w3} + e_{w2w4}, f = e_{w3w1} + e_{w4w2}, h = [e, f] = h_{w1w3} + h_{w2w4} form an sl_{2}triple with ad(h) : L → L having eigenvalues 2, 0, 2. Let L = L_{(2)} + L_{(0)} + L_{(2)} be the decomposition of L into the sum of eigenspaces with respect to ad(h). It is known (see ref. 6) that L_{(2)} with the new operation becomes a Jordan superalgebra (see refs. 13 and 16).
The Lie superalgebra L is centrally isogenous with the TitsKantorKoecher construction of this Jordan superalgebra (see refs. 6, 20, and 21).
In ref. 13 it is shown that the subsuperalgebra (, ·) of J = (L_{(2)}, ·) is isomorphic to the Jordan superalgebra The Jordan superalgebra JP(2) has a basis 1, , 1 ≤ i ≤ 3 (see ref. 16) such that = = 1, = 1,x·ν_{i} = 0, 1 ≤ i ≤ 3, [x_{i}, x] = ν_{i}, 1 ≤ i ≤ 3, x_{i}·ν_{j} = x_{i×j}, where x_{i×i} = 0, x_{1×2} = x_{2×1} = x_{3}, x_{1×3} = x_{3×1} = x_{2}, x_{2×3} = x_{3×2} = x_{1}.
The subspace ε = e_{w1w3}(R) + e_{w1w4}(R) + e_{w2w3}(R) + e_{w2w4}(R) is a subsuperalgebra of J, which is isomorphic to M_{2}(R)^{(+)} (recall that R is an associative commutative superalgebra).
Lemma 5.1.The superalgebra J is generated by JP(2) and ε.
Consider the following two commuting linear transformations on J: R(ν_{3}) the right multiplication by ν_{3} and the inner derivation The superalgebra JP(2) decomposes into a sum of eigenspaces with respect to R(ν_{3}), D(ν_{1}, ν_{2}), the weights being (0, 0), .
We say that an element a belongs to the weight (α, β) if aR(ν_{3}) = αa, aD(ν_{1}, ν_{2}) = βa. The corresponding weight space is denoted as J_{(α,β)}.
The superalgebra J also decomposes into a sum of eigenspaces with respect to R(ν_{3}), D(ν_{1}, ν_{2}) with the same weights. We have just changed a Cartan subalgebra in .
The elements are orthogonal idempotents, . If , then , hence y·e_{1} = 0, y·e_{2} = y. This implies that . Similarly, .
Denote .
Lemma 5.2.J′·J′ = (0).
Proof: We have The products belong to one of the eigenvalues with respect to D(ν_{1}, ν_{2}) and therefore are equal to zero, and Lumma 5.2 is proved.
We identify ε with M_{2}(R) and we identify R with its image in M_{2}(R).
Lemma 5.3.RD(J′, J′) = (0).
Proof: We have . Therefore and Lemma 5.3 is proved.
Lemma 5.4.If a ∈ J_{(α1,β1)}, b ∈ J_{(α2,β2)}, and c ∈ J_{(α3,β3)}, then {a, b, c} ∈ J_{(α1α2 + α3,β1 + β2 + β3)}.
Lemma 5.5.((Rν_{i})x)x_{i} = (0), 1 ≤ i ≤ 3.
Lemma 5.6.Let ν be an even element of J such that ν^{2} = 1. Then U(ν): a→{ν, a, ν} is an automorphism of order 2. The superalgebra J decomposes into the eigenspaces J = J(1) + J(1), J(i) = {a ∈ J  aU(ν) = ia}. If a ∈ J(1), then (aν)ν = a, D(a, ν) = 0.
Proof: The Macdonald identity (see ref. 4) implies that U(ν) is an automorphism. Clearly, U(ν)^{2} = U(ν^{2}) = Id. If a ∈ J(1), then aU(ν) = a(2R(ν^{2})  R(ν^{2})) = 2(aν)ν  a = a, which yields (aν)ν = a. Furthermore, D(a, ν)=D((aν)ν, ν) = ½D(aν, ν^{2}) = ½D(aν, 1) = 0, and Lemma 5.6 is proved.
Lemma 5.7.RD(J′, x) = (0).
Proof: The subspace R of M_{2}(R) is invariant under all even and odd derivations of M_{2}(R). Comparing weights it is easy to see that and therefore .. On the other hand, for a nonzero element a ∈ RD(J′, x) we always have aD(ν_{1}, ν_{2}) ≠ 0. Hence RD(J′, x) = (0), and Lemma 5.7 is proved.
Lemma 5.8.For an arbitrary element a ∈ R we have (ax_{i})x = aν_{i}, 1 ≤ i ≤ 3.
Lemma 5.9.(Rν_{i})(Rx_{i}) = (0), 1 ≤ i ≤ 3.
Proof: As stated above, it is sufficient to prove the assertion for i = 1. We have Hence we need only to check that (Rν_{1})x_{1} = (0). By Lemmas 5.7 and 5.8 for an arbitrary element a ∈ R we have aν_{1} = (ax_{1})x = (ax)x_{1}. Hence (aν_{1})x_{1} = (ax)R(). The element x_{1} lies in J′, hence by Lemma 5.3 aR(x_{1})^{2} = 0 and xR(x_{1})^{2} = 0 in JP(2). Lemma 5.9 is proved.
Lemma 5.10.For arbitrary 1 ≤ i ≤ 3; a, b ∈ R we have a(bx_{i}) = (ab)x_{i}.
Proof: By Lemma 5.9 which implies the result and proves Lemma 5.10.
Denote ′: R → R, a′ = aR(x)^{2}.
Lemma 5.11.For arbitrary 1 ≤ i ≤ 3; a, b ∈ R we have (aν_{i})(bx) = (a′b)x_{i}.
Proof: We have D(aν_{i}, b) = 0. Hence (aν_{i})(bx) = (1)^{ab}b((aν_{i})x). By Lemma 5.8 aν_{i} = (ax_{i})x. Now ((ax_{i})x)x = a′x_{i} and finally (aν_{i})(bx) = (1)^{ab}b(a′x_{i}) = (a′b)x_{i}, by Lemma 5.10, and Lemma 5.11 is proved.
Lemma 5.12.For arbitrary 1 ≤ i ≤ 3; a, b ∈ R we have a(bx) = (ab)x.
Proof: Arguing as in Lemma 5.10 and using Lemma 5.9, we have Hence, and Lemma 5.12 is proved.
Lemma 5.13. For arbitrary 1 ≤ i ≠ j ≤ 3; a, b ∈ R we have
Proof: We have D(aν_{i}, b) = 0 and D(a, x_{j}) = D(±(aν_{j})ν_{j}, x_{j}) = 0. This implies the assertion and proves Lemma 5.13.
Lemma 5.14. For arbitrary elements a, b ∈ R we have
Proof: We claim that (ax)(bx) ∈ R. Indeed, Comparing weights, we see that , and therefore . By Lemma 5.12 aR(bx)R(x) = ((ab)x)x = (ab)′ ∈ R. For an arbitrary element c ∈ R if cx_{1} = 0, then by Lemma 5.8 (cx_{1})x = cv_{1} = 0, and by Lemma 5.6 c = (cv_{1})v_{1} = 0. Hence we need to verify that ((ax)(bx))x_{1} = (a′b  ab′)x_{1}. We have Now, (ax)(bx_{1}) = a(x(bx_{1})), because D(a, bx_{1}) = D((av_{1})ν_{1}, bx_{1}) = 0. By Lemma 5.8 a(x(bx_{1})) = (1)^{b}abv_{1}. Hence, By Lemma 5.11 Clearly, (ax)ν_{1} = 0. Now, and Finally,
Lemma 5.14 is proved.
Proof of Theorem 1.2: From Lemmas 5.1 and 5.6 it follows that with multiplication defined by Lemmas 5.85.12 and 5.14. Comparing with the multiplication tables from ref. 16 we see that J is a ChengKac Jordan superalgebra. Hence L is centrally isogenous with the TitsKantorKoecher construction of the ChengKac Jordan superalgebra, that is, with a ChengKac Lie superalgebra. Central closeness will be considered in section 6. Theorem 1.2 is proved.
6. Universal Central Extensions
Let L be a Δgraded Lie superalgebra, L ⊃ a central extension of a finite dimensional simple classical Lie superalgebra corresponding to the root system Δ, H a Cartan subalgebra in . Let L̃ be a central extension of L, ψ : L̃ → L an epimorphism, .. We claim that L̃ contains a finite dimensional subsuperalgebra that is a preimage of . Indeed, choose a basis e_{1},...,e_{s} in and its preimage a_{1},..., a_{s} ∈ L̃. Then . The subsuperalgebra is a finite dimensional preimage of . We have , and the kernel of each of these homomorphisms is central. Choose a Cartan subalgebra H′ in the Levi factor of . It is easy to see that L̃ is a sum of eigenspaces with respect to the action of H′. If H = ψ(H′), then ψ is a graded epimorphism, and the mapping of the Jordan systems (L̃_{α}, α ∈ Δ) → (L_{α}, α ∈ Δ) is an isomorphism. The finite dimensional perfect superalgebra ∑_{α∈Δ} is a central extension of 𝒢. We proved that L̃ is a Δgraded Lie superalgebra. Throughout this section we will assume that Δ = P(n  1), n ≥ 3, and keep the notation of section 2. In particular, the Jordan system is isomorphic to the Jordan system (e_{ij}(R), 1 ≤ i ≠ j ≤ n). Denote .
Lemma 6.1.Let 1 ≤ i < j ≤ n. Then.
Proof: We have In particular, if i = 1, then . Now for arbitrary 1 ≤ i ≠ j ≤ n we have e_{wiwj}(R) = ,, and therefore
Lemma 6.1 is proved.
Lemma 6.2.If n ≠ 4, then L̃_{0} = S.
Proof: By the previous lemma and the results of section 4 we need only to check that . We have by Lemma 6.1, and Lemma 6.2 is proved.
Proposition 6.1. For an associative commutative superalgebra R the superalgebra L = P(n  1) R, n ≠ 4 is centrally closed.
Proof: Let x ∈ Kerψ. By Lemma 6.2 x = , a_{i} ∈ R. This implies that ψ(x) = h_{wiw1}(a_{i}) = 0 in P(n  1) R. Hence a_{2} = ··· = a_{n} = 0, and Proposition 6.1 is proved.
Now let us assume that n = 4. Recall that . Denote E = ∑_{1≤i≠j≤n} e_{wiwj}(R) + , 𝒬 = Fq_{w1+w2} + Fq_{w1+w3}. Clearly, E, and by Lemma 6.1 . For an arbitrary odd root α ∈ , choose k = 2 or 3 such that (α  w_{1}  w_{k}) ≠ 0. Then This implies that .. Hence Considering the components of weight zero we get
Proposition 6.2. For an arbitrary associative commutative superalgebra R with an even derivation d the superalgebra L = CK(R, d) is centrally closed.
Proof: Let x ∈ Kerψ. As we have seen above, a_{i}, a ∈ R. Computing the righthand side in the superalgebra CK(R, d) we get Hence a_{2} = a_{3} = a_{4} = a = 0 and Proposition 6.2 is proved.
Acknowledgments
We are grateful to G. Benkart and A. Elduque for helpful discussions and valuable remarks. This work was partially supported by Básica Física y Matemáticas Grant 20013239C0301 and Fundación para la Investigación Científica y Technológica Grant GEEXP0108 (to C.M.) and National Science Foundation Grant DMS0071834 (to E.I.Z.).
 Copyright © 2003, The National Academy of Sciences
References
 ↵
 ↵
Slodowy, P. (1986) Can. Math. Soc. Conf. Proc. 5, 361371.
 ↵
Allison, B. N., Azam, S., Berman, S., Gao, Y. & Pianzola, A. (1997) Mem. Am. Math. Soc. 126, 1122.
 ↵
Jacobson, N. (1969) Structure and Representation of Jordan Algebras (Am. Math. Soc., Providence, RI).
 ↵
Zhevlakov, K. A., Slinko, A. M., Shestakov, I. P. & Shirshov, A. I. (1982) Rings That Are Nearly Associative (Academic, New York).
 ↵
Tits, J. (1962) Indag. Math. 24, 530535.
 ↵
 ↵
Garland, H. (1984) Publ. Math. Inst. Hautes Etud. Sci. 52, 5136.
 ↵
 ↵

Benkart, G. & Elduque, A. (2003) J. Lie Theory, in press.
 ↵
Benkart, G. & Elduque, A. (2003) Selecta Math., in press.
 ↵
 ↵
 ↵
Grozman, P., Leites, D. & Shchepochkina, I. (1999) Acta Math. Vietnam. 26, 2763.
 ↵
 ↵
 ↵
Zelmanov, E. (1985) Sib. Mat. Zh. 26, 4967.
 ↵
Faulkner, J. R. (1989) Geom. Dedicata 30, 125181.
 ↵
Kantor, I. L. (1966) Tr. Semin. Vectorn. Tenzorn. Anal. 13, 310398.
 ↵