Hochschild Cohomology and Deformations of Clifford-Weyl Algebras
We give a complete study of the Clifford-Weyl algebra C(n,2k) from Bose-Fermi statistics, including Hochschild cohomology (with coefficients in itself). We show that C(n,2k) is rigid when n is even or when k ≠ 1. We find all non-trivial deformations of C(2n+1,2) and study their representations.
Gespeichert in:
| Datum: | 2009 |
|---|---|
| Hauptverfasser: | , , |
| Format: | Artikel |
| Sprache: | English |
| Veröffentlicht: |
Інститут математики НАН України
2009
|
| Schriftenreihe: | Symmetry, Integrability and Geometry: Methods and Applications |
| Online Zugang: | http://dspace.nbuv.gov.ua/handle/123456789/149177 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Назва журналу: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Zitieren: | Hochschild Cohomology and Deformations of Clifford-Weyl Algebras / I.M. Musson, G. Pinczon, R. Ushirobira // Symmetry, Integrability and Geometry: Methods and Applications. — 2009. — Т. 5. — Бібліогр.: 28 назв. — англ. |
Institution
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
irk-123456789-149177 |
|---|---|
| record_format |
dspace |
| spelling |
irk-123456789-1491772019-02-20T01:26:42Z Hochschild Cohomology and Deformations of Clifford-Weyl Algebras Musson, I.M. Pinczon, G. Ushirobira, R. We give a complete study of the Clifford-Weyl algebra C(n,2k) from Bose-Fermi statistics, including Hochschild cohomology (with coefficients in itself). We show that C(n,2k) is rigid when n is even or when k ≠ 1. We find all non-trivial deformations of C(2n+1,2) and study their representations. 2009 Article Hochschild Cohomology and Deformations of Clifford-Weyl Algebras / I.M. Musson, G. Pinczon, R. Ushirobira // Symmetry, Integrability and Geometry: Methods and Applications. — 2009. — Т. 5. — Бібліогр.: 28 назв. — англ. 1815-0659 2000 Mathematics Subject Classification: 16E40; 16G99; 16S80; 17B56; 17B10; 53D55 http://dspace.nbuv.gov.ua/handle/123456789/149177 en Symmetry, Integrability and Geometry: Methods and Applications Інститут математики НАН України |
| institution |
Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| collection |
DSpace DC |
| language |
English |
| description |
We give a complete study of the Clifford-Weyl algebra C(n,2k) from Bose-Fermi statistics, including Hochschild cohomology (with coefficients in itself). We show that C(n,2k) is rigid when n is even or when k ≠ 1. We find all non-trivial deformations of C(2n+1,2) and study their representations. |
| format |
Article |
| author |
Musson, I.M. Pinczon, G. Ushirobira, R. |
| spellingShingle |
Musson, I.M. Pinczon, G. Ushirobira, R. Hochschild Cohomology and Deformations of Clifford-Weyl Algebras Symmetry, Integrability and Geometry: Methods and Applications |
| author_facet |
Musson, I.M. Pinczon, G. Ushirobira, R. |
| author_sort |
Musson, I.M. |
| title |
Hochschild Cohomology and Deformations of Clifford-Weyl Algebras |
| title_short |
Hochschild Cohomology and Deformations of Clifford-Weyl Algebras |
| title_full |
Hochschild Cohomology and Deformations of Clifford-Weyl Algebras |
| title_fullStr |
Hochschild Cohomology and Deformations of Clifford-Weyl Algebras |
| title_full_unstemmed |
Hochschild Cohomology and Deformations of Clifford-Weyl Algebras |
| title_sort |
hochschild cohomology and deformations of clifford-weyl algebras |
| publisher |
Інститут математики НАН України |
| publishDate |
2009 |
| url |
http://dspace.nbuv.gov.ua/handle/123456789/149177 |
| citation_txt |
Hochschild Cohomology and Deformations of Clifford-Weyl Algebras / I.M. Musson, G. Pinczon, R. Ushirobira // Symmetry, Integrability and Geometry: Methods and Applications. — 2009. — Т. 5. — Бібліогр.: 28 назв. — англ. |
| series |
Symmetry, Integrability and Geometry: Methods and Applications |
| work_keys_str_mv |
AT mussonim hochschildcohomologyanddeformationsofcliffordweylalgebras AT pinczong hochschildcohomologyanddeformationsofcliffordweylalgebras AT ushirobirar hochschildcohomologyanddeformationsofcliffordweylalgebras |
| first_indexed |
2025-07-12T21:35:04Z |
| last_indexed |
2025-07-12T21:35:04Z |
| _version_ |
1837478558391336960 |
| fulltext |
Symmetry, Integrability and Geometry: Methods and Applications SIGMA 5 (2009), 028, 27 pages
Hochschild Cohomology and Deformations
of Clifford–Weyl Algebras?
Ian M. MUSSON †, Georges PINCZON ‡ and Rosane USHIROBIRA ‡
† Department of Mathematical Sciences, University of Wisconsin-Milwaukee,
Milwaukee, WI 53201-0413, USA
E-mail: musson@uwm.edu
URL: http://www.uwm.edu/∼musson/
‡ Institut de Mathématiques de Bourgogne, Université de Bourgogne,
B.P. 47870, F-21078 Dijon Cedex, France
E-mail: Georges.Pinczon@u-bourgogne.fr, Rosane.Ushirobira@u-bourgogne.fr
URL: http://www.u-bourgogne.fr/monge/phy.math/members/pinczon.htm,
http://www.u-bourgogne.fr/rosane.ushirobira
Received October 01, 2008, in final form February 25, 2009; Published online March 07, 2009
doi:10.3842/SIGMA.2009.028
Abstract. We give a complete study of the Clifford–Weyl algebra C(n, 2k) from Bose–
Fermi statistics, including Hochschild cohomology (with coefficients in itself). We show
that C(n, 2k) is rigid when n is even or when k 6= 1. We find all non-trivial deformations of
C(2n + 1, 2) and study their representations.
Key words: Hochschild cohomology; deformation theory; Clifford algebras; Weyl algebras;
Clifford–Weyl algebras; parastatistics
2000 Mathematics Subject Classification: 16E40; 16G99; 16S80; 17B56; 17B10; 53D55
Introduction
Throughout the paper, the base field is C. As usual in superalgebra theory, we denote the ring
Z/2Z by Z2.
Let C(n) be the Clifford algebra with n generators and W2k be the Weyl algebra with 2k
generators. Denote by V0 the vector space spanned by the generators of C(n). Elements of V0 will
be called Fermi-type operators. Similarly, let V1 be the vector space spanned by the generators
of W2k. Elements of V1 will be called Bose-type operators.
There exist Z2-gradations on C(n) and W2k such that Fermi and Bose-type operators all have
degree one. The Clif ford–Weyl algebra is:
C(n, 2k) := C(n)⊗Z2 W2k,
where ⊗Z2 is relative to these gradations. It unifies Fermi and Bose-type operators in a unique
algebra: as elements of C(n, 2k), they anti-commute. There is a Z2-gradation on C(n, 2k) ex-
tending the natural gradation of V = V0⊕V1, and a corresponding structure of Lie superalgebra.
Palev has shown that V generates a sub-superalgebra of C(n, 2k) isomorphic to osp(n + 1, 2k),
and introduced corresponding parastatistics relations [18]. This was an outcome of previous re-
sults by Wigner [28], Green [12] and others (see [8]). It gives an elegant algebraic interpretation
of the Green ansatz, using the Hopf structure of the enveloping algebra of osp(n + 1, 2k) [19],
and introduces a construction of representations of parastatistics relations by Verma modules
?This paper is a contribution to the Special Issue on Deformation Quantization. The full collection is available
at http://www.emis.de/journals/SIGMA/Deformation Quantization.html
mailto:musson@uwm.edu
http://www.uwm.edu/~musson/
mailto:Georges.Pinczon@u-bourgogne.fr
mailto:Rosane.Ushirobira@u-bourgogne.fr
http://www.u-bourgogne.fr/monge/phy.math/members/pinczon.htm
http://www.u-bourgogne.fr/rosane.ushirobira
http://dx.doi.org/10.3842/SIGMA.2009.028
http://www.emis.de/journals/SIGMA/Deformation_Quantization.html
2 I.M. Musson, G. Pinczon and R. Ushirobira
of osp(n + 1, 2k). It also gives an idea of what deformed (quantum) parastatistics could be:
replace C(n, 2k) by a “quantum deformation”, which still has a Hopf structure. This idea was
developed by Palev himself [19] and other authors.
The purpose of the present paper is to return to the first steps of these theories: Clifford–Weyl
algebras. At this level, there is a very natural question: does there exist non-trivial deformations
of C(n, 2k)? By a deformation, we mean a formal one, in the sense of Gerstenhaber theory [11].
It is well-known that the answer is no for C(n, 0) = C(n) and C(0, 2k) = W2k, but nothing was
done in the general case. We shall answer the question, but this is not our only goal. We also
want to introduce Clifford–Weyl algebras in a deformation quantization framework, emphasize
their periodicity behavior and how it can be used, explain where Palev’s theorem comes from,
and so on.
Before describing the content of the paper, let us answer the initial question: C(2n, 2k) is
rigid, for all n, k, C(2n + 1, 2k) is rigid if, and only if, k 6= 1, so the answer is no in these cases.
In the case of C(2n + 1, 2) there exist non-trivial deformations, that we completely describe in
the paper, including their representations.
Let us give some details of our main results. In Section 1, we recall well-known properties
of Clifford and Weyl algebras needed in the paper. In particular, we recall the deformation
quantization construction of the Weyl algebra (resp. Clifford algebra) through the Moyal product
(resp. a Moyal-type product).
In Section 2, we construct the Clifford–Weyl algebra C(n, 2k) by a similar deformation quan-
tization procedure, as a deformation of the super-exterior algebra of the Z2-graded vector space
V = V0 ⊕ V1 with an explicit Moyal-type formula for the ? -product. From this construction,
C(n, 2k) is a Z2 × Z2-graded algebra, with natural left and right Z2-gradations.
We show in Section 3, that Clifford–Weyl algebras have a periodic behavior, very similar to
the well-known, and useful, periodicity of Clifford algebras:
Periodicity Lemma 1.
C(2m + n, 2k) ' C(2m)⊗ C(n, 2k).
This simple Lemma has many consequences. For a given r ∈ N∗, denote by Mr(A) the
r × r-matrix algebra with coefficients in an algebra A. One has:
Theorem 1.
1. C(2n, 2k) 'M2n (W2k).
2. C(2n + 1, 2k) 'M2n (C(1, 2k)).
Note that C(1, 2k) is the algebra generated by W2k and the parity operator of the metaplectic
(oscillator) representation. In other words, C(1, 2k) is the smash product S2 n W2k where S2 is
the group {−1, 1} (see Remark 3.4). By Periodicity Lemma 1, it results that, though C(2n, 2k)
has been defined by anti-commuting Fermi and Bose-type operators, sitting in osp(2n + 1, 2k),
it can also be defined by a new set of commuting generators, also of Fermi and Bose-type, no
longer sitting in osp(2n+1, 2k), but generating a superalgebra of type o(2n+1)×osp(1, 2k). This
suggest that the enveloping algebra U(o(2n + 1) × osp(1, 2k)) could be used for parastatistics.
We shall not go further into this point in the present paper.
A second consequence, and the key for our purpose (to deform Clifford–Weyl algebras)
is the natural Morita equivalence (see [14]) between W2k and C(2n, 2k) (resp. C(1, 2k) and
C(2n + 1, 2k)). We shall see some consequences in Sections 5–8.
A new proof of Palev’s theorem is given in Section 4:
Theorem 2 ([18]). The sub-superalgebra of C(n, 2k) generated by Fermi and Bose-type opera-
tors is isomorphic to osp(n + 1, 2k).
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 3
The original proof uses a direct comparison of commutation rules, and does not really explain
why the result exists. This is what we want to do, and the reason for giving this new proof.
We explain it in a few words. First, we remark that C(n, 2k) is Z2 × Z2-graded, and that
the superbracket used in the Theorem 2 is the one defined by the right Z2-gradation. There
exists a twisted adjoint action ad′ of C(n, 2k) on itself, coming from the left Z2-gradation, and
a supersymmetric bilinear form of type (n + 1, 2k) on H := C ⊕ V (graded by H0 = C ⊕ V0,
H1 = V1) coming from the natural super-Poisson bracket. Then g = V ⊕ [V, V ] is a sub-
superalgebra of C(n, 2k), H is ad′(g)-stable, the bilinear form is ad′(g)-invariant, and the result
follows.
In Section 5, we go back to the initial question: to deform C(n, 2k). As well-known in defor-
mation theory, the first step is to study the Hochschild cohomology of C(n, 2k) with coefficients
in itself (see Appendix A for generalities in Hochschild cohomology and its relations with defor-
mation theory). Since Morita equivalent algebras have isomorphic Hochschild cohomology [14],
using the Periodicity Lemma 1, it results that:
Theorem 3.
1. H`(C(2n, 2k)) = {0} if ` > 0.
2. H`(C(2n + 1, 2k)) = H`(C(1, 2k)), for all `.
So we have a partial answer to our question: C(2n, 2k) cannot be non-trivially deformed,
and we are left with the case of C(1, 2k). As we mentioned before, C(1, 2k) is a smash product
S2 n W2k, but the cohomology of S2 n W2k is known, as a particular case of general results
in [1] and [21], that give the cohomology of G n W2k, when G is a finite subgroup of SP(2k).
We obtain:
Theorem 4.
1. If ` > 0 and ` 6= 2k, then
H`(C(2n + 1, 2k)) = {0}.
2. dim
(
H2k(C(2n + 1, 2k))
)
= 1. Denoting by ω1, . . . , ω2k+1 the basic Fermi-type operators
and by s the canonical symplectic form on V1, then there exists a 2k-cocycle Ω such that
H2k(C(2n + 1, 2k)) = CΩ,
Ω(X1, . . . , X2k) = sk(X1 ∧ · · · ∧X2k)ω1 · · ·ω2k+1, for Xi ∈ V1,
Ω(X1, . . . , X2k) = 0, if one Xi ∈ V0.
If a 2k-cocycle Ω′ verifies Ω′(X1, . . . , X2k) = Ω(X1, . . . , X2k) for all Xi ∈ V , then Ω′ = Ω
mod B2k.
As a consequence, C(2n+1, 2k) is rigid if k 6=1. In the case of C(2n+1, 2), since H3(C(2n+1, 2))
= {0}, there does exist non-trivial deformations, and more precisely, a universal deformation
formula (see Appendix B).
Theorem 5. Let AΛ be the C[Λ]-algebra generated by V0 = CP and V1 = span{E+, E−} with
relations:
[E+, E−]L = −1
4 + ΛP, P 2 = 1 and PE± = −E±P.
Then AΛ is a non-trivial polynomial deformation of C(1, 2) and a universal deformation
formula.
This algebra AΛ is a particular case (the simplest one) of a symplectic reflection algebra [7].
4 I.M. Musson, G. Pinczon and R. Ushirobira
Given an algebra A, any deformation AΛ of A naturally produces a deformation Mr(AΛ)
of Mr(A) and conversely, any deformation of Mr(A) is of type Mr(AΛ) up to equivalence
(see Appendix C). Therefore, Theorem 1 allows us to conclude that M2n(AΛ) is a universal
deformation formula of C(2n + 1, 2).
Another presentation of this deformation is given in Section 6. We introduce algebras AΛ(n)
and Aλ(n) (Λ formal, λ ∈ C, n ∈ N) by generators and relations, with AΛ(0) = AΛ. We show
that they can be constructed using Ore extensions, and that they have a periodic behavior:
Periodicity Lemma 2. AΛ(n) ' C(2n)⊗AΛ 'M2n(AΛ) and the same result holds for Aλ.
In Sections 7 and 8, we describe AΛ(n) and Aλ(n) using the enveloping algebra U of the Lie
superalgebra osp(1, 2) and its primitive quotients [20]. Denoting by Aλ the algebra Aλ(0), one
has:
Theorem 6.
1. AΛ(n) 'M2n(S2 n U) and Aλ(n) ' AΛ(n)/(Λ− λ)AΛ(n).
2. Let C be the Casimir element of U and Bc = U/(C − c)U , c ∈ C. Then Aλ ' Bλ2− 1
16
if
λ 6= 0 and A0 = C(1, 2).
3. If λ2 6=
(
h + 1
4
)2, 2h ∈ N, then Aλ(n) is a simple algebra.
If λ2 =
(
h + 1
4
)2, 2h ∈ N, then Aλ(n) is a primitive algebra with a unique non-zero two-
sided ideal that is the kernel of an irreducible representation πh of dimension 2n (4h + 1).
4. AΛ(n) is a FCR-algebra. Any finite-dimensional representation of Aλ(n) is completely
reducible and isotypical of type πh, if λ2 =
(
h + 1
4
)2, 2h ∈ N.
We obtain in this way all primitive quotients of AΛ(n): 22n (4h + 1)2-dimensional quotients
(2h ∈ N) and infinite-dimensional ones: Aλ(n) ' Bλ2− 1
16
if λ 6= 0 and A0(n) = C(2n + 1, 2).
The isomorphism AΛ(n) ' M2n(S2 n U) is useful to construct representations. Remark
that representations of S2 n U are merely graded representations of U . Then, from a graded
representation of osp(1, 2) on a space H, one constructs a representation of AΛ(n) on H2n
.
All irreducible finite-dimensional representations are obtained in this way: from irreducible
(4h+1)-dimensional (2h ∈ N) representations of osp(1, 2), one obtains πh. From the metaplectic
(oscillator) representation of osp(1, 2), natural infinite-dimensional representations can be also
obtained, using Dunkl-type formulas given in [13].
Finally, we extend the obtained supersymmetry by showing that AΛ is a quotient of S2 n
U(osp(2, 2)). Therefore AΛ(n) is a quotient of M2n (S2 n U(osp(2, 2)).
There are three Appendices: the first one is a short introduction to Hochschild cohomo-
logy relating it to deformation theory. In the second Appendix, we explain, with proofs, what
a “universal deformation formula” is. We show in the third Appendix, that given an algebra A,
deformations of Mr(A) are of type Mr(AΛ) up to equivalence, where AΛ is a deformation of A.
Results in Appendices B and C are known, but since we have not found a convenient reference,
short proofs are given.
1 Clifford algebras and Weyl algebras
We begin by recalling some classical properties of Clifford and Weyl algebras needed in the paper.
Throughout this section, we denote by [·, ·] the super bracket and by [·, ·]L the Lie bracket.
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 5
1.1 Weyl algebras
Let k ∈ N∗ and S2k = C[p1, q1, . . . , pk, qk] be the polynomial algebra in 2k indeterminates
equipped with the Poisson bracket:
{F,G} =
k∑
i=1
(
∂F
∂pi
∂G
∂qi
− ∂F
∂qi
∂G
∂pi
)
, ∀F,G ∈ S2k.
Let ℘ : S2k ⊗ S2k → S2k ⊗ S2k be the operator defined by:
℘ :=
k∑
i=1
(
∂
∂pi
⊗ ∂
∂qi
− ∂
∂qi
⊗ ∂
∂pi
)
.
Let m be the product of S2k and t be a formal parameter (or t ∈ C). A new associative
product m?
t
is defined by:
m?
t
:= m ◦ exp
(
t
2
℘
)
. (1)
This product m?
t
is a deformation of S2k guided by the Poisson bracket.
Definition 1.1. The Weyl algebra W2k is the vector space S2k endowed with the product
? := m?
1
, called the Moyal product.
A presentation of W2k is given by generators {p1, q1, . . . , pk, qk} and relations:
[u, v]L = {u, v} · 1, ∀u, v ∈ span{p1, q1, . . . , pk, qk}.
Structurally, W2k is central, simple, naturally Z2-graded by the parity of S2k and has a su-
pertrace [23]:
Str(F ) := F (0), ∀F ∈ W2k. (2)
The space Mk := C[x1, . . . , xk] is a faithful simple W2k-module if we realize pi as ∂
∂xi
and qi
as the multiplication by xi, for all i = 1, . . . , k. In the sequel, Mk is called the metaplectic
representation of the Weyl algebra W2k.
The algebra of operators L(Mk) appears as a completion of the Weyl algebra: W2k is the
algebra of differential operators of finite order, and any element T in L(Mk) is a differential
operator, in general of infinite order (i.e., in the formal sense, the sum is not finite), given by
the formula:
T =
∑
N
1
N !
(
m ◦ (T ⊗ S) ◦∆
(
xN
)) ∂N
∂xN
, (3)
where S is the antipode of M2k, ∆ is its co-product, xN := xn1
1 · · ·xnk
k and ∂N
∂xN := ∂n1+···+nk
∂x
n1
1 ···∂x
nk
k
if
N = (n1, . . . , nk). This formula gives the (formal) symbol of T in the normal ordering, and for
well-behaved T , its (formal) symbol in the Weyl ordering (see [23]).
From the point of view of deformation theory, W2k is rigid. More precisely, we have Hr(W2k)
= {0}, for all r > 0 [26].
We refer to [23] for more details on the Weyl algebra in the context of this section.
6 I.M. Musson, G. Pinczon and R. Ushirobira
1.2 Clifford algebras
Let n ∈ N∗ and
∧
n be the Grassmann algebra in n anti-commutative variables ω1, . . . , ωn. Recall
that
∧
n is Z-graded. Denote by ∂1, . . . , ∂n the super-derivations defined by ∂i(ωj) = δij , ∀ i, j.
The algebra
∧
n is endowed with a super Poisson bracket:
{Ω,Ω′} = 2(−1)degZ(Ω)+1
n∑
i=1
∂i(Ω) ∧ ∂i(Ω′),
for all Ω,Ω′ ∈
∧
n [22]. We define the operator ℘ of
∧
n⊗
∧
n by:
℘ :=
n∑
i=1
∂i ⊗ ∂i,
where ⊗ is the graded tensor product of operators.
Let m∧ be the product of
∧
n and t be a formal parameter (or t ∈ C). A new product m?
t
can be defined by (see [22]):
m?
t
:= m∧ ◦ exp(−t℘). (4)
Definition 1.2. The Clifford algebra C(n) is the vector space
∧
n equipped with the product
? := m?
1
.
There is a Z2-gradation on C(n) defined by degZ2
(ωi) = 1, for all i = 1, . . . , n. A presentation
of C(n) is given by basic generators ω1, . . . , ωn and relations:
[v, v′] = {v, v′} · 1, ∀ v, v′ ∈ span{ω1, , . . . , ωn}.
In particular, we have:
ω2
i = 1, ∀ i, ωi ? ωj + ωj ? ωi = 0, ∀ i 6= j and
ωi1 ∧ · · · ∧ ωip = ωi1 ? · · · ? ωip if i1 < i2 < · · · < ip, p ≤ n.
1.3 Even Clifford algebras
For i = 1, . . . , n, let Pi = ∂i and Qi = xi∧ . be respectively the operator of derivation and
multiplication of the Grassmann algebra Φn in n anti-commutative variables x1, . . . , xn. The
operators ω2j−1 = Qj +Pj and ω2j = i(Qj −Pj), j = 1, . . . , n verify the defining relations of the
Clifford algebra C(2n), so there is a homomorphism from C(2n) onto the algebra of differential
operators Diff(Φn). It is easy to see that dim(C(2n)) = dim(Diff(Φn)) = dim(L(Φn)), so we can
identify C(2n) = Diff(Φn) = M2n(C), where M2n(C) denotes the algebra of complex matrices of
order 2n. As a consequence, Φn is the unique simple C(2n)-module, called the spin representation
of C(2n).
Structurally, the even Clifford algebra C(2n) = M2n(C) is simple and its center is C. From
the point of view of deformation theory, C(2n) is rigid and we have Hr(C(2n)) = {0} if r > 0.
Since C(2n) = M2n(C), there is a natural trace on C(2n) that can be written in an analogous
way as in (2):
Tr(Ω) := 2nΩ(0), ∀Ω ∈ C(2n).
There is also a similar formula to (3) in the case of C(2n). In other words, any operator
T ∈ L(Φn) is differential and an explicit formula is given by:
T =
∑
I∈{0,1}n
(−1)θ(I,I)
(
m∧ ◦ (T ⊗ S) ◦∆
(
xI
))
∧ ∂I ,
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 7
where θ is the bilinear form on Nn associated to the matrix (θrs)n
r,s=1 with θrs = 1 if r > s and 0
otherwise, ⊗ is the non-graded tensor product of operators, S is the antipode of Φn, ∆ is its
co-product, xI := xi1
1 ∧ · · · ∧ xin
n and ∂I := ∂i1
1 ◦ · · · ◦ ∂in
n if I = (i1, . . . , in).
1.4 Periodicity of Clifford algebras
There is an algebra isomorphism between C(2n + k) and C(2n)⊗ C(k) since ω1 ⊗ 1, . . . , ω2n ⊗ 1
and inω1 ? · · · ? ω2n⊗ω′j , j = 1, . . . , k in C(2n)⊗C(k) verify the defining relations of C(2n+k)
thanks to the formula:
(ω1 ? · · · ? ω2n)2 = (−1)n.
It results that:
C(2n) ' C(2)⊗n and C(2n + 1) ' C(2n)⊗ C(1) 'M2n (C(1)) .
1.5 Odd Clifford algebras
Since C(1) ' C × C, from the isomorphism C(2n + 1) ' M2n (C(1)) it follows that C(2n + 1)
is the product C(2n) × C(2n). Therefore Hr(C(2n + 1)) = {0} if r > 0, and that implies that
C(2n + 1) is rigid.
We will make more explicit the above isomorphism: C(2n+1) ' C(2n)×C(2n). The element
z = ω1 ? · · · ? ω2n+1 is central and verifies z2 = (−1)n. Set Z = span{1, z}. Then C(2n + 1) '
Z ⊗ C(2n) as algebras. Let z+ = 1
2(1 + inz) and z− = 1
2(1 − inz). Therefore z2
± = z±,
z+ ? z− = z− ? z+ = 0 and 1 = z+ + z−. We conclude that
C(2n + 1) = z+ ? C(2n)⊕ z− ? C(2n),
that is, a reduction of C(2n + 1) in a direct sum of two ideals isomorphic to C(2n) as algebras.
It follows that C(2n + 1) has exactly two 2n-dimensional simple modules built from the spin
representation of C(2n). To give more details, we need the following lemma:
Lemma 1.3. Let P be the natural parity of Φn. Then in the spin representation of C(2n),
one has:
ω1 ? · · · ? ω2n = inP.
Proof. We set T = ω1 ? · · · ? ω2n. The operator T of Φn is diagonalizable since T 2 = (−1)n,
and it commutes with P . We denote by S0,±in and S1,±in its eigenspaces in S0 and S1 respectively,
where S = Φn. The subspaces S0,in ⊕ S1,−in and S0,−in ⊕ S1,in are C(2n)-stable since T anti-
commutes with ωi, 1 ≤ i ≤ 2n. It follows that T = ±inP . To determine the sign, we compute:
T (1) = in(Q1 + P1) ? (Q2 − P2) ? · · · ? (Qn + Pn) ? (Qn − Pn)(1) = in.
Finally, we obtain T = inP . �
The element z is central, z2 = (−1)n, hence z = ±in in any simple C(2n + 1)-module.
Since z = (ω1 ? · · · ? ω2n) ? ω2n+1, using the lemma we obtain the spin representations Φ±
n of
C(2n + 1) as follows: C(2n) ⊂ C(2n + 1) acts on Φn by the spin representation (see Section 1.3),
for Φ+
n , define ω2n+1 = P and for Φ−
n , define ω2n+1 = −P .
8 I.M. Musson, G. Pinczon and R. Ushirobira
1.6 Weyl algebras and supersymmetry
Let W = W2k = ⊕r≥0W
r be the Z-gradation of the vector space W. Recall that W is a Z2-graded
algebra. By (1), we have:
[F,G]L = {F,G}, ∀F ∈
⊕
r≤2
Wr.
Let h = h1 ⊕ h0 where h1 = W1 = span{pi, qi, i = 1, . . . , k} and h0 = W2. The super
bracket stabilizes h. Besides, h is isomorphic to the Lie superalgebra osp(1, 2k). In particular,
h0 ' sp(2k) and the adjoint action of h0 on h1 is the standard action of sp(2k) on C2k. As
a consequence, W is a semisimple h0-module for the adjoint action and W = ⊕r≥0W
r is its
reduction in isotypical components.
By (1), we have:
[v, F ] = 2vF, ∀F ∈ W2r+1 and [v, F ]L = {v, F}, ∀F ∈ W2r, v ∈ W1.
Therefore W is also semi-simple for the adjoint action of h and W = ⊕r≥0A
r is its reduction
into isotypical components, where A0 = C and Ar = W2r−1 ⊕ W2r, if r > 0. We refer to [17]
or [23] for more details.
1.7 Clifford algebras and symmetry
Let C = C(n). There is a Z-gradation on the vector space C and, as an algebra, C is Z2-graded.
By (4), we have:
[Ω,Ω′] = {Ω,Ω′}, ∀Ω ∈
⊕
r≤2
Cr.
Let g = g1 ⊕ g0 where g1 = C1 and g0 = C2. The Lie bracket stabilizes g. Moreover, g is
isomorphic to the Lie algebra o(n + 1). In particular, g0 ' o(n) and the adjoint action of g0
on g1 is the standard action of o(n) on Cn. The direct sum g = g1⊕ g0 is a Z2-gradation for the
Lie algebra g, that is [gi, gj ]L ⊂ gi+j (this is not a graded Lie algebra!). For the adjoint action,
C is a semisimple g0-module and is isomorphic to the o(n)-module
∧
Cn, whose reduction into
isotypical components is well-known (see [24] or [9]). The reduction into isotypical components
of the g-module C can be deduced, but this is simply not the subject of this paper.
2 Clifford–Weyl algebras
We recall the construction of the exterior algebra of a Z2-graded vector space V = V0 ⊕ V1: let∧
:=
∧
V0 be the exterior algebra of V0 and S := Sym(V1) be the symmetric algebra of V1. Using
their Z-gradation, define a Z× Z2 gradation on
∧
and on S by∧
(i,0) =
∧
i,
∧
(i,1) = {0} and S(i,i) = Si, S(i,j) = {0} if i 6= j.
The exterior algebra of V is the Z× Z2-graded algebra
E :=
∧
⊗
Z×Z2
S =
∧
⊗
Z
S
endowed with the product:
(Ω⊗ F ) ∧ (Ω′ ⊗ F ′) = (−1)fω′
(Ω ∧ Ω′)⊗ FF ′,
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 9
for all Ω ∈
∧
, Ω′ ∈
∧ω′
, F ∈ Sf , F ′ ∈ S. We have
A′ ∧A = (−1)aa′+bb′A ∧A′, ∀A ∈ E(a,b), A′ ∈ E(a′,b′)
and that means that E is Z× Z2-commutative.
Now, assume that the dimension of V1 is even, say dim(V1) = 2k. Set n = dim(V0).
We have defined Poisson brackets on
∧
and S in Section 1. Associated operators ℘∧ and ℘S
are respectively defined on
∧
⊗
∧
and S⊗ S.
A super Z× Z2-Poisson bracket on E is then defined by:
{Ω⊗ F,Ω′ ⊗ F ′} = (−1)fω′ ({Ω,Ω′} ⊗ FF ′ + (Ω ∧ Ω′)⊗ {F, F ′}
)
,
for all Ω ∈
∧
, Ω′ ∈
∧ω′
, F ∈ Sf , F ′ ∈ S. Now, let σ23 and ℘ be operators on E⊗ E defined by:
σ23(Ω⊗ F ⊗ Ω′ ⊗ F ′) = (−1)fω′
Ω⊗ Ω′ ⊗ F ⊗ F ′,
℘ = σ23 ◦ (−2℘∧ ⊗ Id+ Id⊗℘S) ◦ σ23,
for all Ω ∈
∧
, Ω′ ∈
∧ω′
, F ∈ Sf , F ′ ∈ S.
Let t be a formal parameter (or t ∈ C). A new product m?
t
on E is defined from these
operators and from the product mE on E by:
m?
t
:= mE ◦ exp
(
t
2
℘
)
. (5)
Since m?
t
= m∧⊗mS ◦ exp (−t℘∧)⊗ exp
(
t
2℘S
)
◦σ23, it results that m?
t
is exactly the Z2×Z2-
graded tensor algebra product:
Ct(n)⊗Z2×Z2 Wt
2k = Ct(n)⊗Z2 Wt
2k,
where ⊗Z2 means the graded tensor product with respect to left Z2-gradations, C(n)t denotes
the algebra equipped with product m?
t
and similarly for Wt
2k (see Section 1). By definition, m?
t
is a deformation of mE guided by the Poisson super bracket.
Definition 2.1. The Clifford–Weyl algebra C(n, 2k) is the vector space E endowed with the
product ? := m?
1
.
Denote by {ω1, . . . , ωn} and {p1, q1, . . . , pk, qk} respectively the basis of C(n) and W2k as in
Section 1. The algebra C(n, 2k) has a presentation given by generators {ω1, . . . , ωn, p1, q1, . . . ,
pk, qk} and relations
[ωi, ωj ]+ = 2δij , [pi, qj ]− = δij , [pi, pj ]− = [qi, qj ]− = 0 if i 6= j
and [ωi, pj ]+ = 0, [ωi, qj ]+ = 0, ∀ i, j,
where [A,B]± := A ? B ±B ? A.
When n = 2` is even, we set
Pj = 1
2(ω2j−1 + iω2j) and Qj = 1
2(ω2j−1 − iω2j) for j ≤ `.
The first two relations above become:
[Pi, Qj ]+ = δij , [Pi, Pj ]+ = [Qi, Qj ]+ = 0.
Consider now Φ` the Grassmann algebra in ` anti-commutative variables ξ1, . . . , ξ`, Mk the
polynomial algebra in k variables x1, . . . , xk and SM(`, k) the exterior algebra of the Z2-graded
10 I.M. Musson, G. Pinczon and R. Ushirobira
space W = W0⊕W1 with W0 = span{ξ1, . . . , ξ`} and W1 = span{x1, . . . , xk}. There is a C(2`, 2k)-
module structure on SM(`, k) given by Pi = ∂
∂ξi
, Qi = ξi ∧ · (1 ≤ i ≤ `), pj = ∂
∂xj
, qj = xj ·
(1 ≤ j ≤ k). Besides, SM(`, k) is a simple C(2`, 2k)-module. In the sequel, we call SM(`, k) the
spin-metaplectic representation of C(n, 2k). This provides a homomorphism from C(2`, 2k) onto
the algebra of differential operators of the Z2×Z2-graded exterior algebra SM(`, k). We will show
later that C(n, 2k) is simple, so we have actually an isomorphism, C(2`, 2k) ' Diff(SM(`, k))
and that generalizes the cases of C(2`) and W2k seen in Section 1.
When n = 2`+1 is odd, we obtain two spin-metaplectic representations SM(`, k)± of C(2`+1,
2k) by acting C(2`, 2k) on SM(`, k) as above and by setting ω2`+1 = Q for SM(`, k)+ and
ω2`+1 = −Q for SM(`, k)−, where Q is the parity:
Q(ω ⊗ f) = (−1)degZ(ω)+degZ(f)ω ⊗ f, ∀ω ∈ Φ`, f ∈ Mk.
It will be shown later that if k 6= 0, C(2` + 1, 2k) is simple and as a consequence, both
representations SM(`, k)± are faithful.
3 Periodicity of Clifford–Weyl algebras
Clifford algebras have a periodic behavior Section 1.4 and we now show that this periodicity can
be extended to Clifford–Weyl algebras. We denote by C(r), the Clifford algebra in r variables
and by W2k, the Weyl algebra constructed from 2k variables (see Section 1).
Lemma 3.1 (Periodicity Lemma 1).
C(2m + n, 2k) ' C(2m)⊗ C(n, 2k).
Proof. Let {ω1, . . . , ω2m} and {ω′1, . . . , ω′n, p1, q1, . . . , pk, qk} be respectively the set of generators
of C(2m) and C(n, 2k). Let z = imω1 ? · · · ? ω2m. So z2 = 1 and z anti-commutes with
ω1, . . . , ω2m. The following elements of C(2m)⊗C(n, 2k): ω1⊗1, . . . , ω2m⊗1, z⊗ω′1, . . . , z⊗ω′n,
z⊗ p1, . . . , z⊗ pk and z⊗ q1, . . . , z⊗ qk verify the defining relations of C(2m+n, 2k). Since they
generate C(2m)⊗ C(2n, k) as an algebra, we get an algebra homomorphism from C(2m + n, 2k)
onto C(2m)⊗ C(n, 2k).
Denote by {ω̃1, . . . , ω̃2m+n, p̃1, q̃1, . . . , p̃k, q̃k} the set of generators of C(2m + n, 2k). Let
z̃ = imω̃1 ? · · · ? ω̃2m. So z̃2 = 1, z̃ anti-commutes with ω̃1, . . . , ω̃2m+n and z̃ commutes with
p1, . . . , pk and q1, . . . , qk. The following elements of C(2m + n, 2k): ω̃1, . . . , ω̃2m, z̃ ? ω̃2m+1, . . . ,
z̃ ? ω̃2m+n, z̃ ? p̃1, z̃ ? q̃1, . . . , z̃ ? p̃k, z̃ ? q̃k verify the defining relations of C(2m)⊗ C(n, 2k), so
we get the inverse homomorphism. �
Corollary 3.2. One has:
C(2n, 2k) ' C(2n)⊗W2k 'M2n (W2k) and
C(2n + 1, 2k) ' C(2n)⊗ C(1, 2k) 'M2n (C(1, 2k)) ,
where Mr(A) denotes the r × r-matrix algebra with coefficients in an algebra A for a given
r ∈ N∗.
Corollary 3.3.
1. C(2n, 2k) is simple with center C.
2. If k 6= 0, then C(2n + 1, 2k) is simple with center C.
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 11
Proof. 1. C(2n, 2k) 'M2n(W2k) is simple since W2k is simple.
2. Since C(2n + 1, 2k) 'M2n(C(1, 2k)), it is enough to prove the result for C(1, 2k).
But C(1, 2k) ' S2 n W2k and W2k is simple, so the result is a particular case of a general
theorem in [16].
For the sake of completeness, here is a direct proof: we write C(1, 2k) = C(1)⊗
Z2
W2k where
C(1) is the Clifford algebra generated by P such that P2 = 1. Recall that using the Moyal
? -product, the Weyl algebra W = W2k can be realized as a deformation of the polynomial
algebra C[p1, q1, . . . , pk, qk]. Fix p = p1 and q = q1.
We have [p, f ]L = ∂f
∂q , ∀ f ∈ W. In addition, for all g ∈ W:
[p, P ? g]L = p ? P ? g − P ? g ? p = −P ? (p ? g + g ? p)
= −P ?
(
pg + 1
2{p, g}+ gp + 1
2{g, p}
)
= −2P ? (pg).
Let I be a non-zero two-sided ideal of C(1, 2k) and let f + P ? g ∈ I, f + P ? g 6= 0. Then
[p, f + P ? g]L ∈ I gives ∂f
∂q − 2P ? (pg) ∈ I and we can reiterate. Hence:
• if g = 0, then f ∈ I. It follows that I ∩W 6= {0};
• if g 6= 0, since there exists j such that ∂jf
∂qj = 0, one has (−1)j2jP ? (pjg) ∈ I, implying
pjg ∈ I. But pjg 6= 0, so it follows that I ∩W 6= {0} as well.
In both cases, I∩W is a non-zero ideal of the Weyl algebra W. Since W is simple, I∩W = W.
So 1 ∈ I and we conclude that I = C(1, 2k).
The center of C(1, 2k) is C since the center of W is C. �
Remark 3.4. Let us first recall what a smash product is. Let A be an algebra, G a finite group
acting on A by automorphisms and C[G] the group algebra. The smash product G n A is the
algebra with underlying space A⊗ C[G] and product defined by:
(a⊗ g)(a′ ⊗ g′) = a(g.a′)⊗ gg′, ∀ a, a′ ∈ A, g, g′ ∈ G.
Smash products used in this paper are defined from a Z2-graded algebra A and G = S2 =
{−1, 1}. Denoting by P the parity operator of A, P(a) := (−1)deg(a)a, ∀ a ∈ A, S2 acts on A by
(−1)α · a := Pα(a), ∀ a ∈ A, α = 0, 1 and there is a corresponding smash product S2 n A. It is
the algebra generated by P and A with relations Pa = P(a)P, ∀ a ∈ A and P2 = 1. It is easy to
check that S2 n A-modules and Z2-graded A-modules are exactly the same notion.
Now consider the Clifford–Weyl algebra C(1, 2k). Using the Z2-graded structure of W2k,
C(1, 2k) ' S2 n W2k. Also C(1, 2k) is isomorphic to a subalgebra of M2(W2k):
C(1, 2k) '
{(
a b
σ(b) σ(a)
)
, a, b ∈ W2k
}
,
where σ is the parity operator of W2k. In this isomorphism, ω1 ∈ C(1) is realized as the matrix(
0 1
1 0
)
and W2k as
{(
a 0
0 σ(a)
)
, a ∈ W2k
}
.
Finally, C(1, 2k) is isomorphic to the algebra generated by the parity operator P of Mk =
C[x1, . . . , xk] and W2k, realized as the algebra of differential operators of Mk (see Section 1).
4 Clifford–Weyl algebras and supersymmetry
Let us consider the Z2 × Z2-graded algebra C(n, 2k) and the subspace V = V0 ⊕ V1 where
V0 = C(n, 2k)(1,0) =
∧1
n and V1 = C(n, 2k)(1,1) = S1
2k (see Section 2 for the notation).
12 I.M. Musson, G. Pinczon and R. Ushirobira
If k = 0, then V0⊕ [V0, V0]L is a Lie algebra for the natural Lie bracket of the Clifford algebra,
isomorphic to o(n + 1) and [V0, V0]L is a Lie subalgebra isomorphic to o(n) (for details, see
Section 1).
If n = 0, then V1 ⊕ [V1, V1] is a Lie superalgebra for the natural super bracket of the Weyl
algebra, isomorphic to osp(1, 2k) and [V1, V1] is a Lie algebra isomorphic to sp(2k).
To generalize this situation, we need some notation: for an element a ∈ C(n, 2k), denote its
Z2×Z2-degree by ∆(a) := (∆1(a),∆2(a)). We consider C(n, 2k) as an algebra Z2-graded by ∆2
and we denote by [·, ·] the associated super bracket.
The proposition below shows how to realize osp(n+1, 2k) as a Lie sub-superalgebra of C(n, 2k).
This important result was first obtained by [18] for osp(2` + 1, 2k). We propose here another
method to show the same result, inspired by [23] and based on a well-chosen twisted adjoint
action.
Proposition 4.1 ([18]). Let g = V ⊕ [V, V ]. Then g is a Lie sub-superalgebra of C(n, 2k)
isomorphic to osp(n + 1, 2k). Moreover
g0 = V0 ⊕ [V0, V0]⊕ [V1, V1]
with [V0, V0] ' o(n), [V1, V1] ' sp(2k), V0 ⊕ [V0, V0] ' o(n + 1) and g0 ' o(n + 1)× sp(2k). Also,
g1 = V1 ⊕ [V0, V1]
and V1 ⊕ [V1, V1] ' osp(1, 2k). If we set h = [V0, V0]⊕ [V1, V1]⊕ [V0, V1], then h ' osp(n, 2k).
Proof. By a case by case straightforward computation, using the product formula (5), we get
the formula:
[[X, Y ], Z] = 2
(
{Y, Z}X − (−1)∆2(X)∆2(Y ){X, Z}Y
)
, ∀X, Y, Z ∈ V, (6)
where {·, ·} is the super Poisson bracket defined in Section 2.
Hence [[V, V ], V ] ⊂ V . If H ∈ [V, V ] and X, Y ∈ V , then:
[H, [X, Y ]] = [[H,X], Y ] + (−1)∆2(H)∆2(X)[X, [H,Y ]].
Using (6), we conclude that [[V, V ], [V, V ]] ⊂ [V, V ], therefore g is a Lie superalgebra and h is
a sub-superalgebra.
To prove the isomorphisms, we set V ′ = C ⊕ V . Define a non-degenerate supersymmetric
2-form (·|·) on V ′ by:
(X|Y ) := {X, Y }, ∀X, Y ∈ V and (1|1) = −2.
Then formula (6) becomes parastatistics relations:
[[X, Y ], Z] = 2
(
(Y |Z)X − (−1)∆2(X)∆2(Y )(X|Z)Y
)
, ∀X, Y, Z ∈ V. (PS)
Next, we define the ∆1-twisted adjoint representation of the Lie superalgebra C(n, 2k):
ad′(a)(b) := a ? b− (−1)∆2(a)∆2(b)+∆1(a)b ? a, ∀ a, b ∈ C(n, 2k).
It is easy to check that it is indeed a representation. If H ∈ h, ad′(H) = ad(H), writing
H = [X, Y ] and using (PS), one obtains:
(ad′(H)(Z)|T ) = −(−1)∆2(Z)∆2(H)(Z| ad′(H)(T )), ∀T ∈ V,
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 13
henceforth ad′(h)(V ′) ⊂ V ′ and h ⊂ osp(n, 2k). Since both spaces have the same dimension
n(n−1)
2 + 2nk + k(2k + 1) (see [25]), it follows h ' osp(n, 2k).
It remains to examine the action of ad′(X) on V ′ when X ∈ V . We have ad′(X)(Y ) = 0
if X ∈ Vi, Y ∈ Vj with i 6= j. Moreover, if X, Y ∈ V0, then ad′(X)(Y ) = X ? Y + Y ? X =
{X, Y } · 1 = (X|Y ). If X, Y ∈ V1, then ad′(X)(Y ) = X ? Y − Y ? X = {X, Y } · 1 = (X|Y ).
Since ad′(X)(1) = 2X, finally (ad′(X)(Y )|1) = −2(X|Y )= −(−1)∆2(X)∆2(Y )(Y | ad′(X)(1)). So
g ⊂ osp(n + 1, 2k) and both spaces have the same dimension. �
Corollary 4.2. Let V = V0 ⊕ V1 be a Z2-graded space with dim(V0) = n and dim(V1) = 2k.
Assume that V is equipped with a non-degenerate supersymmetric bilinear form (·|·). Let A be
the Z2-graded algebra generated by V = V0 ⊕ V1 and relations (PS). Then A is isomorphic to
the enveloping algebra U(osp(n + 1, 2k)).
Proof. We denote by [·, ·]A the super bracket of A. Proceeding exactly as in the proof of Propo-
sition 4.1, we show that V +[V, V ]A is a Lie superalgebra using the parastatistics relations (PS).
From the definition of A together with Proposition 4.1, there is an algebra homomorphism
from A onto C(n, 2k) that is the identity when restricted to V . This homomorphism induces a
Lie superalgebra homomorphism from V +[V, V ]A onto V⊕[V, V ] (realized in C(n, 2k) and isomor-
phic to osp(n+1, 2k) by Proposition 4.1). That implies dim(V +[V, V ]A) ≥ dim(osp(n+1, 2k)).
On the other hand, dim(V + [V, V ]A) ≤ dim(V ⊕ [V, V ]) since we can write
[V, V ]A = [V0, V0]A + [V1, V1]A + [V0, V1]A
and dim([V, V ]) = dim(V0 ∧ V0) + dim(V0 ⊗ V1) + dim(V0V1).
It results that V ⊕ [V, V ]A ' osp(n + 1, 2k). Remark that the parastatistics relations hold in
the enveloping algebra U(osp(n + 1, 2k)) since they hold in osp(n + 1, 2k). To finish, we apply
the universal property of U(osp(n + 1, 2k)). �
Remark 4.3. The result in Proposition 4.1 is helpful to obtain explicit descriptions of osp(n+1,
2k) (for instance, the root system).
Remark 4.4. As observed in [18], the fact that generators of C(n) (Fermi-type operators)
and those of W2k (Bose-type operators) anti-commute in C(n, 2k) is a main argument to prove
that the Lie sub-superalgebra that they generate is osp(n + 1, 2k). However, the periodicity of
Clifford–Weyl algebras, namely C(2n, 2k) ' C(2n) ⊗ W2k, shows that it is always possible to
obtain C(2n, 2k) from commuting Bose-type and Fermi-type operators (that will not live in the
Lie superalgebra osp(2n + 1, 2k), but rather in o(2n + 1)× osp(1, 2k)).
In the sequel, all ? products will simply be denoted by juxtaposition.
5 Cohomology of Clifford–Weyl algebras
In Appendix A, the reader can find a short introduction to Hochschild cohomology of an algebra
with coefficients in itself.
By Periodicity Lemma 1 and Corollary 3.2, we have
C(2n, 2k) 'M2n (W2k) and C(2n + 1, 2k) 'M2n (C(1, 2k)) .
But for an algebra A, Mr(A) and A have isomorphic cohomology spaces [14], so it results
that the cohomology of Clifford–Weyl algebras can be computed from the cohomology of W2k
and C(1, 2k):
Proposition 5.1.
1. H`(C(2n, 2k)) = {0} if ` > 0.
2. H`(C(2n + 1, 2k)) = H`(C(1, 2k)), for all `.
14 I.M. Musson, G. Pinczon and R. Ushirobira
Proof. It is enough to remark that H`(W2k) = {0} if ` > 0 [26]. �
We now give more details on the identifications in the above Proposition. We use the iso-
morphisms in Corollary 3.2: C(2n, 2k) ' C(2n)⊗W2k and C(2n+1, 2k) ' C(2n)⊗C(1, 2k). The
letter A denotes either W2k or C(1, 2k).
Since C(2n) is separable, we compute the cohomology of C(2n)⊗A using normalized C(2n)-
relative cochains (see [10]), that is, cochains
Ω : (C(2n)⊗A)` → C(2n)⊗A
that verify:
Ω(Ca1, a2, . . . , a`) = CΩ(a1, . . . , a`),
Ω(a1, . . . , aiC, ai+1, . . . , a`) = Ω(a1, . . . , ai, Cai+1, . . . , a`),
Ω(a1, . . . , a`C) = Ω(a1, . . . , a`)C,
Ω(a1, . . . , a`) = 0 if one ai ∈ C(2n)
for all C ∈ C(2n). Since C(2n) commutes with A, such a cochain is completely determined by
its restriction Ω̃ : A` → C(2n)⊗A verifying
CΩ̃(a1, . . . , a`) = Ω̃(a1, . . . , a`)C.
for all C ∈ C(2n). It results that Ω̃ is actually A-valued. Then the map Ω Ω̃ induces an
isomorphism [10]:
H`(C(2n)⊗A) ' H`(A).
To obtain the desired cohomology, that is, H`(C(2n, 2k)) or H`(C(2n + 1, 2k)), we use the
isomorphism φ : C(2n)⊗A → C(2n, 2k) or C(2n+1, 2k) in the Periodicity Lemma 1 (Lemma 3.1):
given a cochain Ω of C(2n) ⊗ A, we introduce a cochain φ∗(Ω) of C(2n, 2k) or C(2n + 1, 2k)
defined by
φ∗(Ω)(x1, . . . , x`) = φ
(
Ω
(
φ−1(x1), . . . , φ−1(x`)
))
,
for all x1, . . . , x` ∈ C(2n, 2k) or C(2n + 1, 2k). Then the map Ω φ∗(Ω) induces a cohomology
isomorphism.
It remains to compute the cohomology of C(1, 2k) = C(1)⊗Z2 W2k. Since C(1, 2k) = S2nW2k,
this is a particular case of a result in [1] where the cohomology of GnW2k is given for G a finite
group of symplectic linear transformations. There is an improved version of this result in [21],
that allows a better management of cocycles. Denote by P the generator of C(1) satisfying
P 2 = 1. One has:
Proposition 5.2 ([1, 21]).
1. If ` > 0 and ` 6= 2k, then
H`(C(1, 2k)) = {0}.
2. dim
(
H2k(C(1, 2k))
)
= 1. Moreover, there exists a normalized C(1)-relative cocycle θ such
that H2k(C(1, 2k)) = Cθ and
θ(X1, . . . , X2k) = sk(X1 ∧ · · · ∧X2k)P, for X1, . . . , X2k ∈ V1,
where s is the canonical symplectic form on V1. If a 2k-cocycle θ′ verifies θ′(X1, . . . , X2k) =
θ(X1, . . . , X2k) for all Xi ∈ V , then θ′ = θ mod B2k.
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 15
Proof. See [1] for the dimension of H`(C(1, 2k)). See [21] for the last claims. �
Corollary 5.3.
1. If ` > 0 and ` 6= 2k, then
H`(C(2n + 1, 2k)) = {0}.
2. Denote by ω1, . . . , ω2n, P the canonical generators of C(2n + 1) realized in C(2n + 1, 2k).
Then there exists a cocycle Ω such that such that H2k(C(2n + 1, 2k)) = CΩ,
Ω(X1, . . . , X2k) = insk(X1 ∧ · · · ∧X2k)ω1 . . . ω2k+1, for Xi ∈ V1,
Ω(X1, . . . , X2k) = 0, if one Xi ∈ V0.
If a 2k-cocycle Ω′ verifies Ω′(X1, . . . , X2k) = Ω(X1, . . . , X2k) for all Xi ∈ V , then Ω′ = Ω
mod B2k.
Proof. Proposition 5.2 provides a cocycle θ that allows us to construct a cocycle θ̂ of C(2n)⊗
C(1, 2k) such that:
θ̂(C1 ⊗ x1, . . . , C2k ⊗ x2k) = C1 · · ·C2k ⊗ θ(x1, . . . , x2k)
for x1, . . . , x2k ∈ W2k, C1, . . . , C2k ∈ C(2n). Next we compute Ω = φ∗(θ̂) using formulas in the
proof of Lemma 3.1:
Ω(X1, . . . , X2k) = φ
(
θ̂(inω1 · · ·ω2n ⊗X1, . . . , i
nω1 · · ·ω2n ⊗X2k)
)
= φ
(
(in)2k(ω1 · · ·ω2n)2ksk(X1 ∧ · · · ∧X2k)P
)
for X1, . . . , X2k ∈ V1. Since (ω1 · · ·ω2n)2 = (−1)n (see Section 1), then
Ω(X1, . . . , X2k) = φ
(
sk(X1 ∧ · · · ∧X2k)P
)
= insk(X1 ∧ · · · ∧X2k)ω1 · · ·ω2nP. �
Corollary 5.4. The Clifford–Weyl algebra C(2n + 1, 2k) is rigid if k 6= 1.
Since dim
(
H2(C(2n + 1, 2))
)
= 1 and H3(C(2n + 1, 2)) = {0}, then C(2n + 1, 2) can be non
trivially deformed by a universal deformation formula (see Appendix B). For C(1, 2), this formula
is a particular case of a symplectic reflection algebras (see [7]):
Proposition 5.5. Let AΛ be the C[Λ]-algebra generated by V0 = C P and V1 = span{E+, E−}
with relations:
[E+, E−]L = −1
4 + ΛP, P 2 = 1 and PE± = −E±P.
Then AΛ is a non-trivial polynomial deformation of C(1, 2) and a universal deformation
formula.
Proof. See [7] or [21]. �
6 Universal deformation formula of C(2n + 1, 2)
Definition 6.1. Let Aλ(n), λ ∈ C be the algebra with generators ω1, . . . , ω2n+1, E± and rela-
tions:
[E+, E−]L = −1
4 + inλω1 · · ·ω2n+1,
ωjωk + ωkωj = 2δjk (1 ≤ j, k ≤ 2n + 1),
E±ωj = −ωjE± (1 ≤ j ≤ 2n + 1).
Definition 6.2. The algebra AΛ(n), when Λ is a formal parameter, is defined in a similar way:
it is the algebra with generators ω1, . . . , ω2n+1, E±,Λ with Λ central and same relations as Aλ(n)
with λ replaced by Λ. Note that AΛ(0) = AΛ.
16 I.M. Musson, G. Pinczon and R. Ushirobira
6.1 Construction using Ore extensions
Definition 6.3. Suppose that R is an algebra, σ an automorphism of R and δ a σ-derivation
of R, that is, a linear map δ : R → R such that
δ(rs) = δ(r)s + σ(r)δ(s)
for all r, s ∈ R. Then the Ore extension R[t] is the free left R-module on the set {tn |n ≥ 0},
with multiplication defined by
tr = σ(r)t + δ(r).
Let C = C(2n + 1) be the Clifford algebra in 2n + 1 generators, ω1, . . . , ω2n+1. Consider the
polynomial ring C[Λ] where Λ commutes with all elements of C. Elements of C[Λ] are denoted
by C(Λ).
Let τ be the automorphism of C[Λ] defined by
τ(ωr) = −ωr, ∀ r and τ(Λ) = Λ.
The free C[Λ]-module C[Λ][E+] with basis {En
+ |n ∈ N} gives us a a first Ore extension with
E+C(Λ) = τ(C(Λ))E+, ∀C(Λ) ∈ C[Λ].
The following lemma is easy:
Lemma 6.4. There exists an automorphism σ of the Ore extension C[Λ][E+] satisfying:
σ(E+) = E+, σ(ωr) = −ωr, ∀ r and σ(Λ) = Λ.
Let θ be the element inω1 · · ·ω2n+1Λ in C[Λ]. So θ commutes with Λ and ωr, ∀ r and anti-
commutes with E+.
Let ∆ be the operator of C[E+] defined by
∆(f) =
f(E+)− f(−E+)
2E+
, ∀ f ∈ C[E+]
and D be the operator of C[Λ][E+] defined by
D(f(E+)C(Λ)) =
(
1
4
df
dE+
−∆(f)θ
)
C(Λ), ∀ f ∈ C[E+], C(Λ) ∈ C[Λ].
Lemma 6.5. One has D(AB) = σ(A)D(B) + D(A)B for all A,B ∈ C[Λ][E+].
Proof. This is a straightforward verification. �
From Lemmas 6.4 and 6.5, we can now construct a second Ore extension C[Λ][E+][E−] sa-
tisfying
E−A = σ(A)E− + D(A), ∀A ∈ C[Λ][E+].
It follows that:
[E+, E−]L = −1
4 + θ,
ωkωj + ωjωk = 2δjk (1 ≤ j, k ≤ 2n + 1), (7)
E±ωj = −ωjE± (1 ≤ j ≤ 2n + 1).
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 17
Proposition 6.6.
1. The Ore extension C[Λ][E+][E−] and AΛ(n) are isomorphic algebras.
2. A basis of AΛ(n) is given by:{
ωIEα
+Eβ
−Λr | I ∈ {0, 1}2n+1, α, β, r ∈ N
}
,
where ωI = ωi1
1 · · ·ω
i2n+1
2n+1 for all I = (i1, . . . , i2n+1) ∈ {0, 1}2n+1.
If Λ is replaced by a small λ (λ ∈ C) in the definition of AΛ(n), the same procedure works
to construct an Ore extension of C[E+][E−], isomorphic to Aλ(n). So
Proposition 6.7. A basis of Aλ(n) is given by:{
ωIEα
+Eβ
− | I ∈ {0, 1}2n+1, α, β ∈ N
}
.
The algebra Aλ(n) is the quotient AΛ(n)/Iλ where Iλ is the ideal AΛ(n) (Λ− λ). As a par-
ticular case, setting p = 2E− and q = 2E+, we obtain:
A0(n) ' C(2n + 1, 2) ' AΛ(n)/ΛAΛ(n).
Since AΛ(n) = A0(n)[Λ] as vector spaces, we obtain:
Proposition 6.8. The algebra AΛ(n) is a non-trivial polynomial deformation of the Clifford–
Weyl algebra A0(n) = C(2n + 1, 2).
Proof. We just have to show that the deformation is non-trivial, but that results from the fact
that the deformation cocycle is non-trivial by Corollary 5.3. �
Remark 6.9. From Corollary 5.3 and Lemma B.2, this polynomial deformation AΛ(n) is a uni-
versal deformation formula of C(2n + 1, 2).
Corollary 6.10. The center of AΛ(n) is C[Λ]. Moreover, AΛ(n) and Aλ(n) are Noetherian
algebras.
Proof. We have A0(n) ' C(2n+1, 2) with center C (Corollary 3.3). Let ã be a central element
of AΛ(n). By Proposition 6.8, we can write ã = a0 + Λb̃ with a0 ∈ A0(n) and b̃ ∈ AΛ(n).
Therefore in AΛ(n):
xa0 + Λxb̃ = a0x + Λb̃x, ∀x ∈ A0(n).
But xa0 = x × a0 + Λc̃ and a0x = a0 × x + Λd̃ where × denotes the product of A0(n). So
a0 is central in A0(n), henceforth a0 ∈ C. It follows b̃ is central in AΛ(n) and repeating the
same argument, we obtain ã ∈ C[Λ]. Finally, AΛ(n) and Aλ(n) are Noetherian since they are
constructed by Ore extensions of Noetherian algebras [15]. �
In the sequel, we denote Aλ the algebra Aλ(0). The periodicity of Clifford algebras can be
extended to the algebras AΛ(n) and Aλ(n):
Lemma 6.11 (Periodicity Lemma 2). One has
AΛ(n) ' C(2n)⊗AΛ 'M2n (AΛ) and Aλ(n) ' C(2n)⊗Aλ 'M2n (Aλ) .
18 I.M. Musson, G. Pinczon and R. Ushirobira
Proof. We denote by P and E± the generators of AΛ(0) satisfying PE± = −E±P , P 2 = 1 and
[E+, E−]L = −1
4 + ΛP . Let ω1, . . . , ω2n be the generators of C(2n).
We define ω′1, . . . , ω
′
2n+1 and E′
± elements of C(2n)⊗AΛ(0) by:
ω′i = ωi ⊗ P (1 ≤ i ≤ 2n), ω′2n+1 = inω1 · · ·ω2n ⊗ P, E′
± = 1⊗ E±.
Using (ω1 · · ·ω2n)2 = (−1)n, we check that ω′1, . . . , ω
′
2n+1 verify the defining relations of
C(2n + 1) and anti-commute with E′
±. The relation [E′
+, E′
−]L = −1
4 + inΛω′1 · · ·ω′2n+1 results
from 1⊗ P = inω′1 · · ·ω′2n+1.
Finally, this last equality and the fact that ωi = ω′i(1 ⊗ P ) imply that ω′i (1 ≤ i ≤ 2n + 1)
and E′
± generate the algebra C(2n)⊗AΛ(n).
On the other hand, if ω1, . . . , ω2n+1, E± are the generators of AΛ(n), we define ω′1, . . . , ω
′
2n,
E′
± and P ′ by:
P ′ = inω1 · · ·ω2n+1, ω′i = ωiP
′ (1 ≤ i ≤ 2n), E′
± = E±.
Since P ′ commutes with ωi, it commutes with ω′i. Since E′
± anti-commute with ωi, they anti-
commute with P ′ and commute with ω′i. The equality P ′2 = 1 follows from (ω1 · · ·ω2n+1)2 =
(−1)n and we conclude ω′2i = 1. Moreover ω′i anti-commutes with ω′j for i 6= j and [E′
+, E′
−]L =
−1
4 + inΛω1 · · ·ω2n+1 = −1
4 + ΛP ′.
All defining relations of C(2n)⊗AΛ(0) are satisfied. Moreover ωi =ω′iP
′ and in(ω′1 · · ·ω′2n+1)P
′
= i2n(ω1 · · ·ω2n)2P ′2nω2n+1 = ω2n+1. So we conclude that ω′i, E± and P ′ generate AΛ(n).
This ends the proof that AΛ(n) ' C(2n) ⊗ AΛ. Since C(2n) ' M2n(C), then AΛ(n) '
M2n (AΛ).
A similar reasoning works for Aλ(n). �
Remark 6.12. The first isomorphism in Lemma 6.11 is not a surprise: if A is an algebra, all
deformations of Mk(A) are of type Mk(AΛ) where AΛ is a deformation of A (see Appendix C).
Here, C(2n + 1, 2) 'M2n (C(1, 2)) and AΛ is a deformation of C(1, 2).
7 Algebras AΛ(n) and their representations
Let {E+, E−, Y, F,G} be the usual generators of the Lie superalgebra osp(1, 2): one has osp(1, 2)0
= span{Y, F, G}, osp(1, 2)1 = span{E+, E−} and the commutation relations
[Y, E±] = ±1
2E±, [Y, F ] = F, [Y, G] = −G, [F,G] = 2Y,
[F,E+] = [G, E−] = 0, [F,E−] = −E+, [G, E+] = −E−,
[E+, E+] = F, [E−, E−] = −G, [E+, E−] = Y,
where [·, ·] denotes the super bracket.
Let U := U(osp(1, 2)) be the enveloping algebra of osp(1, 2). Denote by θ ∈ U the ghost:
θ := 1
4 + [E+, E−]L,
where [·, ·]L denotes the Lie bracket.
Lemma 7.1 ([20, 2, 3]). The relation θE± = −E±θ holds in U .
Proof. We have θ = 1
4 + E+E− − E−E+, hence
E+θ = 1
4E+ + E2
+E− − E+E−E+, θE+ = 1
4E+ + E+E−E+ − E−E2
+.
Therefore E+θ + θE+ = 1
2E+ − [Y, E+] = 0. Similarly, we can prove that E−θ = −E−θ. �
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 19
Let us now consider the C-algebra Uϑ defined by:
Uϑ :=
〈
E+,E−, ϑ | [E+,E−]L = −1
4 + ϑ, E±ϑ = −ϑE±
〉
.
By Lemma 7.1, the enveloping algebra U is a quotient of Uϑ.
Proposition 7.2 ([13]). There exists an algebra isomorphism between U and Uϑ.
Proof. Consider the subspace V = V0⊕V1 of Uϑ, with V0 = {0} and V1 = span{E+,E−}. Define
a supersymmetric bilinear form (·, ·) on V (hence symplectic on V1) by:
(E+,E−) = −1
4 , (E+,E+) = (E−,E−) = 0.
The algebra Uϑ is Z2-graded by the Z2-gradation of V . Starting from [E+,E−] = 2E+E−+ 1
4−θ
with E±θ = −θE±, we have:
[[E+,E−],E±] = ±1
2E±.
Using the Jacobi identity, we get [[E+,E+],E−] = −2[[E+,E−],E+] = −E+ and [[E+,E+],E+] = 0.
In the same way, [[E−,E−],E+] = E− and [[E−,E−],E−] = 0. We conclude that
[[X, Y ], Z] = 2 ((Y, Z)X + (X, Z)Y ) , ∀X, Y ∈ V1.
By Corollary 4.2, we deduce a surjective algebra homomorphism from U to Uϑ and using
Lemma 7.1, we finish the proof. �
Proposition 7.3.
1. AΛ(n) ' C(2n + 1)⊗
Z2
U .
2. AΛ(n) ' C(2n)⊗ (S2 n U) 'M2n(S2 n U).
Proof. 1. Let ω1, . . . , ω2n+1 be the generators of C(2n + 1). Here C(2n + 1) is Z2-graded by
degZ2
(ωi) = 1, ∀ i. Define Λ ∈ C(2n + 1)⊗
Z2
U by
Λ = inω1 · · ·ω2n+1θ.
We see immediately that Λ is a central element and that ω1, . . . , ω2n+1, E± and Λ satisfy the
defining relations of AΛ(n). Moreover, they generate C(2n + 1)⊗
Z2
U since θ = inω1 · · ·ω2n+1Λ.
Then there exists a surjective algebra homomorphism from AΛ(n) to C(2n + 1)⊗
Z2
U .
To define the inverse map, we introduce an element θ ∈ AΛ(n) by
θ = inω1 · · ·ω2n+1Λ.
To finish the proof, we notice that elements E+, E− and θ verify the defining relations of
Uϑ ' U , hence ω1, . . . , ω2n+1, E± and θ satisfy the defining relations of C(2n + 1)⊗
Z2
U and they
generate AΛ(n).
2. The parity of U is used to define the smash product S2 n U = C(1)⊗
Z2
U . To prove (2),
apply Lemma 6.11 and (1). �
Remark 7.4. The algebra AΛ is a deformation of C(1, 2) = S2 n W2. Besides AΛ = S2 n U .
So here is a particular case where a deformation of a smash product remains a smash product.
Moreover, representations of AΛ are merely graded representations of U .
20 I.M. Musson, G. Pinczon and R. Ushirobira
Definition 7.5. An algebra A is a FCR algebra if:
1. Every finite-dimensional representation of A is completely reducible.
2. The intersection of all kernels of finite-dimensional representations is {0}.
If A is an algebra and V is an A-module, there is a corresponding structure of Mn(A)-module
on V = V n, since n×n-matrices with coefficients in A act on n-column vectors with coefficients
in V . The following well-known Lemma shows that all Mn(A)-modules are of this type:
Lemma 7.6. Given an Mn(A)-module V, there exists a corresponding A-module V such that
V ' V n, and ker(V) = Mn(ker(V )).
Proof. Let Eij be the elementary matrices in Mn. They satisfy Id =
n∑
i=1
Eii and EiiEjj =
δijEii. So V = ⊕n
i=1Wi where Wi = Eii · V. Since A and Mn commute, Wi are A-modules.
Since Eij · Wj = Eii · (Eij · Wj), Eij maps Wj into Wi and EijWk = {0} if k 6= j. Using
EjiEij = Ejj , it results that Wi and Wj are isomorphic A-modules, for all i, j. Let V be any of
the A-modules Wi, it is easy to verify that V ' V n as Mn(A)-modules. For the second claim,
use the following easy statement: if J is a two-sided ideal of Mn(A), there exists a two-sided
ideal I of A such that J = Mn(I). �
Remark 7.7. The above correspondence between A-modules and Mn(A)-modules respects
isomorphisms, irreducibility and direct sum decompositions.
Corollary 7.8. If A is an FCR algebra, then Mn(A) is also a FCR algebra.
Proof. Apply Lemma 7.6 and the remark above. �
Corollary 7.9.
1. The algebra AΛ(n) is a FCR algebra.
2. All finite-dimensional representations of Aλ(n) are completely reducible.
Proof. 1. By Periodicity Lemma 2, AΛ(n) 'M2n(AΛ) and by Proposition 7.3, AΛ'S2nU .
Representations of S2 n U are simply Z2-graded representations of U and by [6], finite-dimen-
sional ones are completely reducible. Moreover by [5], the intersection of all kernels of finite-
dimensional representations of S2 n U is {0}. So S2 n U is FCR, then apply Corollary 7.8.
2. Use Aλ(n) ' AΛ(n)/(Λ− λ)AΛ(n). �
Remark 7.10. By Lemma 7.6, the representation theory of AΛ(n) ' M2n(AΛ) is reduced to
the representation theory of AΛ ' S2 nU and therefore to the Z2-graded representation theory
of U .
Example 7.11. To study finite-dimensional representations of AΛ(n), it is enough to study
irreducible ones by Corollary 7.9. Irreducible representations of osp(1, 2) are well-known, they are
all Z2-graded: given h ∈ 1
2N, there exists an irreducible representation on a (4h+1)-dimensional
space Vh to which corresponds an irreducible representation of AΛ(n) given by 2n× 2n-matrices
with coefficients in S2 nU acting on Vh = V 2n
h , therefore of dimension 2n(4h+1). Alternatively,
sinceAΛ(n) ' C(2n)⊗AΛ, this representation is the natural action of C(2n)⊗AΛ on Vh = Φn⊗Vh
where Φn is the spin representation of C(2n) (see Section 1). This describes all finite-dimensional
representations of AΛ(n).
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 21
Example 7.12. We will now construct examples of simple Aλ-modules from the metaplectic
representation of W2. Let V = C[z] as in [13]. We define the operator ∆ of V :
∆(h) =
1
z
(h(z)− h(−z)), ∀h ∈ V.
Denote by P the parity operator of V . Define operators ρ±λ by:
ρ+
λ =
1
2
d
dz
− λ∆, ρ−λ = −1
2
z.
Then [ρ+
λ , ρ−λ ]L = −1
4 + λP , ρ±λ P = −Pρ±λ and P 2 = 1. In this way, we obtain a representa-
tion ρλ of Aλ in V such that:
ρλ(E±) = ρ±λ , ρλ(P ) = P.
We recover exactly the Z2-graded osp(1, 2)-Verma module Vλ− 1
4
of highest weight λ− 1
4 .
If λ 6= h+ 1
4 , 2h ∈ N, then ρλ(E+) does not vanish and the corresponding module is simple. If
λ = h+ 1
4 , 2h ∈ N, we have ρλ(E+)(z4h+1) = 0. Therefore Wh = span{z`, ` ≥ 4h+1} is a simple
submodule of dominant weight −
(
h + 1
2
)
, the quotient V/Wh is the simple osp(1, 2)-module of
dimension 4h + 1 and the module (V, ρh+ 1
4
) is a non-trivial extension of Wh by V/Wh (see [13]
for more details).
Denote by Vλ the Aλ-module just built. Using Aλ(n) = M2n(Aλ), define a corresponding
Aλ(n)-module by setting Vλ(n) = Φ2n⊗Vλ where Φ2n is the spin representation of C(2n). When
λ 6=
(
h + 1
4
)
, 2h ∈ N, we obtain a simple Aλ(n)-module. When λ = h+ 1
4 , 2h ∈ N, we obtain an
indecomposable Aλ(n)-module with a unique simple submodule and a unique simple quotient
of dimension 2n(4h + 1).
Since AΛ(n) ' AΛ(n)/(Λ− λ)AΛ(n), these modules are AΛ(n)-modules.
Remark 7.13. When A is a Z2-graded algebra, Mn(A) = Mn ⊗A has a natural Z2-gradation
induced by the gradation of A and deg(M) = 0 for all M ∈Mn. But algebras S2 nMn(A) and
Mn(S2 n A) have the same underlying vector space. It is easy to verify that they coincide as
algebras. Using Proposition 7.3, Remark 3.4 and Lemma 7.6, it results that all representations
of AΛ(n) are graded and obtained from graded representations of U .
8 Algebras Aλ(n)
We keep the notation of last Section. Write g = osp(1, 2) as g = g0⊕g1 where g0 = span{Y, F, G}
and g1 = span{E+, E−}, U = U(g) its enveloping algebra and θ = 1
4 + [E+, E−]L the ghost. We
have Z(g) = C[C] where Z(g) denotes the center of U , C = θ2 − 1
16 and Z(g0) = C[Q] where
Z(g0) denotes the center of U(g0), Q =
(
θ − 1
4
) (
θ + 3
4
)
[20, 2]. For c ∈ C, let Bc := U/(C− c)U .
Let us consider the C-algebra Aλ := Aλ(0). Recall that:
Aλ =
〈
E+, E−, P |P 2 = 1, [E+, E−]L = −1
4 + λP, E±P = −PE±
〉
.
If λ = 0, A0 is the Clifford–Weyl algebra C(1, 2) = S2 n W2. In general:
Proposition 8.1. One has Aλ ' Bλ2− 1
16
whenever λ 6= 0.
Proof. For u ∈ U , we denote by u its class in Bλ2− 1
16
. Therefore [E+, E−]L = −1
4 + θ and
E± θ = ±θ E±. Moreover, C = λ2 − 1
16 = θ
2 − 1
16 . Setting P = 1
λθ, one recovers exactly the
defining relations of Aλ and a map from Aλ onto Bλ2− 1
16
.
For the inverse map, one can check that elements E+ and E− in Aλ generate a superalgebra
isomorphic to g, hence a homomorphism ρ from U to Aλ. We have ρ(θ) = λP , so ρ is surjective.
Since ρ
(
C − λ2 + 1
16
)
= 0, one can define the inverse map ρ from Bλ2− 1
16
onto Aλ. �
22 I.M. Musson, G. Pinczon and R. Ushirobira
The structure of the algebra Aλ is deduced from the Proposition above and [20].
Proposition 8.2.
1. If λ2 6=
(
h + 1
4
)2, 2h ∈ N, then Aλ is a simple algebra.
2. If λ2 =
(
h + 1
4
)2, 2h ∈ N, then Aλ is a primitive algebra. Moreover, there exists a unique
non-zero two-sided ideal Iλ in Aλ of codimension (4h + 1)2, with Iλ = ker(Vh) and Vh is
the simple osp(1, 2)-module of dimension 4h + 1.
Proof. It is proved in [20] that Bc, c 6= 0 has the following properties:
• if c 6= h (2h+1)
2 , 2h ∈ N, then Bc is Z2-simple;
• if c = h (2h+1)
2 , 2h ∈ N, then Bc is primitive. Moreover, there exists a unique non-zero
Z2-graded two-sided ideal with codimension (4h + 1)2 that is the kernel of the simple
osp(1, 2)-module of dimension 4h + 1.
If λ = 0, then A0 ' C(1, 2) is simple.
If λ 6= 0, then Aλ ' Bλ2− 1
16
. It is enough to show that any two-sided ideal of Bλ2− 1
16
is
Z2-graded and then translate the results just above in term of λ. So, let I be a two-sided ideal
of Bλ2− 1
16
. We set P = 1
λθ. We have P 2 = 1 and PbP = (−1)degZ2
(b)b, ∀ b ∈ Bλ2− 1
16
. If
a = a0 + a1 ∈ I, it follows PaP = a0 − a1 ∈ I, therefore a0 and a1 ∈ I. �
Corollary 8.3.
1. If λ2 6=
(
h + 1
4
)2, 2h ∈ N, then Aλ(n) is a simple algebra.
2. If λ2 =
(
h + 1
4
)2, 2h ∈ N, then Aλ(n) is a primitive algebra. Moreover, there exists
a unique non-zero two-sided ideal in Aλ(n) of codimension 22n (4h + 1)2, that is the kernel
of the irreducible representation of dimension 2n (4h + 1).
Proof. By Lemma 6.11, Aλ(n) 'M2n (Aλ), so two-sided ideals ofAλ(n) are all of typeM2n(I),
I is a two-sided ideal of Aλ. Then apply Proposition 8.2 and Example 7.12. �
Remark 8.4. We have AΛ(n) ' AΛ(n)/(Λ− λ)AΛ(n). Moreover, the center of AΛ(n) is C[Λ],
so Corollary 8.3 lists all primitive quotients of AΛ(n).
The algebra A0 = C(1, 2) is a quotient of U(osp(2, 2)) (see Proposition 4.1). More generally:
Proposition 8.5. The algebra AΛ is a quotient of S2 n U(osp(2, 2)). Moreover, the Casimir
operator of osp(2, 2) (see [2]) vanishes in this quotient.
Proof. Using the notation in Section 6, we consider in AΛ:
K = −1
4ω1 + Λ, H0 = CK, H1 = span{E+, E−}, H = H0 ⊕H1.
Define (·|·) a supersymmetric bilinear form on H by (K|K) = 1
8 and (E+|E−) = −1
4 . It is
easy to check that relations (PS) hold in H, so by Corollary 4.2, the subalgebra of AΛ generated
by H is a quotient of U(osp(2, 2)). Now, the subalgebra of AΛ generated by H and ω1 is AΛ
itself, and it is clearly a quotient of S2 n U(osp(2, 2)). The second claim results from a direct
computation using the Casimir formula given in [2]. �
Corollary 8.6. Any graded osp(1, 2)-module can be extended to an osp(2, 2)-module (with same
underlying space).
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 23
Proof. First, remark that given a graded algebra A, graded A-modules and S2 n A-modules
are exactly the same notion.
Now, start with a graded osp(1, 2)-module with parity P . Recall that [E+, E−]L = −1
4 + θ
with θE± = −E±θ. We define Λ = θP and ω1 = P to obtain a graded AΛ-module. By
Proposition 8.5, this module is a S2 n U(osp(2, 2))-module, therefore a graded osp(2, 2)-mo-
dule. �
Remark 8.7. Let C be the Casimir element of U(osp(2, 2)). It is proved in [2] that a simple
osp(2, 2)-module is still simple as an osp(1, 2)-module if, and only if, C = 0.
A Appendix
For the convenience of the reader, we recall here some notions of Hochschild cohomology theory
relating it to Gerstenhaber deformation theory of (associative) algebras [11, 12]. See [4, 27] for
applications of deformation theory to quantization.
Let A be an (associative) algebra. By Hochschild cohomology of A, we mean Hochschild
cohomology with coefficients in A, defined as follows.
For k > 0, k-cochains are k-linear maps from Ak to A. When k = 0, 0-cochains are simply
elements of A. We denote by Mk(A) the space of k-cochains and by M(A) = ⊕k≥0M
k(A), the
space of cochains. We define the Hochschild coboundary operator d acting on M(A) by:
• if a ∈ A = M0(A), da = − ad(a) where ad(a)(b) := [a, b], for all a,b ∈ A;
• if Ω ∈ Mk(A), k > 0:
dΩ(a1, . . . , ak+1) = a1Ω(a2, . . . , ak+1)− Ω(a1a2, a3, . . . , ak+1)
+ Ω(a1, a2a3, . . . , ak+1) + · · ·+ (−1)k+1Ω(a1, . . . , ak)ak+1.
One has d2 = 0. Let B0(A) = {0}, Bk(A) = dMk−1(A), k > 0. Set Zk(A) = ker
(
d|Mk(A)
)
,
k ≥ 0 and Hk(A) = Zk(A)/Bk(A). Elements of Bk(A) (resp. Zk(A)) are k-coboundaries (resp.
k-cocycles) and Hk(A) is the kth-space of Hochschild cohomology of A. Note that H0(A) is the
center of A. Here are some examples of algebras A such that Hk(A) = {0} for all k > 0: semi-
simple algebras (e.g. algebras of finite groups, algebras of complex matrices, Clifford algebras),
Weyl algebras, etc.
A deformation of A with formal parameter Λ is a C[[Λ]]-algebra structure on A[[Λ]] defined by:
a ? b = ab +
∑
n≥1
ΛnΩn(a, b), ∀ a, b ∈ A, Ωn ∈ M2(A), ∀n.
The associativity of ? can be reinterpreted in terms of Hochschild cohomology: Ω1 ∈ Z2(A)
and when Ω1 ∈ B2(A), it can be removed by an equivalence, i.e. an isomorphism of C[[Λ]]-
algebras. When H2(A) = {0}, repeating the same argument, it results that any deformation
is equivalent to the initial product, so A is rigid. For instance, all algebras we just mentioned
above are rigid. Second, the conditions on Ωn, n ≥ 2 can be written in terms of 3-cohomology,
and it results that if H3(A) = {0}, then given any Ω1 ∈ Z2(A), there exits a deformation with
leading cocycle Ω1.
These two results are known as the rigidity and integrability theorems.
24 I.M. Musson, G. Pinczon and R. Ushirobira
B Appendix
The terminology and results presented in this Appendix are rather standard, but for the sake
of completeness we include them here with proofs.
Let A be an associative algebra with product m0. Let M(A) =
∑
k≥0
Mk(A) be the space of
multilinear maps from A to A. The space M(A) is graded, M(k) := Mk+1(A) and endowed
with the Gerstenhaber bracket, it is a graded Lie algebra [22]. Let d = − ad(m0). Since
d2 = 0, d defines a complex on M(A), the Hochschild cohomology complex of A (see [10]).
Let Z2(A) be the set of 2-cocycles, B2(A) the 2-coboundaries, and H2(A) chosen such that
Z2(A) = B2(A)⊕H2(A).
Given two vector spaces V and W , a formal map F : V → W is a power series F =
∑
k≥0
Fk
where Fk is a homogeneous polynomial function of degree k from V to W . In the sequel, we will
need essentially formal maps F : H2(A) → M(A) and we define a graded Lie algebra bracket
coming from the one defined on M(A) by:
[F, F ′] =
∑
k≥0
∑
r+s=k
[Fr, F
′
s] for F =
∑
k≥0
Fk, F ′ =
∑
k≥0
F ′
k
with [Fr, F
′
s](h) = [Fr(h), F ′
s(h)], ∀h ∈ H2(A).
Definition B.1. A universal deformation formula of A is a formal map F : Z2(A) → M2(A)
such that:
1) F = m0 + IdH2(A) +
∑
k≥2
Fk,
2) [F, F ] = 0.
If F is a universal formula of deformation, λ a formal parameter and h ∈ H2(A), then
mλ
h := F (λh) = m0 + λh +
∑
k≥2
λkFk is a deformation of m0. More generally, if we have a formal
curve in H2(A)[[λ]], h̃ =
∑
n≥1
λnhn, then
mλ
h̃
:= F (h̃(λ)) = m0 + λh1 +
∑
k≥2
λk
∑
i1+···+in=k
i1,...,in≥1,1≤n≤k
Fn(hi1 , . . . , hin)
is a deformation of m0. The Lemma below is simply a translation of the classical criterion of
integrability:
Lemma B.2. Let D2(A) be a complementary subspace of Z2(A) in M2(A). If H3(A) = {0},
then there exists a universal deformation formula
F = m0 + IdH2(A) +
∑
k≥2
Fk, with Fk ∈ D2(A), ∀ k ≥ 2.
Proof. Let σ be a section of d : M2(A) → B3(A) such that σ ◦ d is the projection onto D2(A)
along Z2(A). Step by step, we construct F verifying [F, F ] = 0, F = m0 + IdH2(A) +
∑
k≥2
Fk:
first, we find d(F2) = 1
2 [IdH2(A), IdH2(A)]. Since [IdH2(A), IdH2(A)] is valued in Z3(A) = B3(A),
define a suitable F2 = 1
2σ ◦ [IdH2(A), IdH2(A)] (remark that d ◦ σ = IdB3(A)). It is easy to see
that the remaining Fk can be constructed by the same procedure. �
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 25
Lemma B.3.
1. Let mλ be a deformation of m0. Then, up to equivalence, mλ can be written as:
mλ = m0 + h(λ) + d(λ), with h(λ) ∈ λH2[[λ]], d ∈ λ2D2[[λ]].
2. If m′λ is another deformation with
m′λ = m0 + h(λ) + d′(λ), with d′(λ) ∈ λ2D2[[λ]],
then d′(λ) = d(λ).
Proof. 1. Up to equivalence, we can assume that the leading cocycle of mλ is in H2(A),
mλ = m0 + λh1 + λ2C2 + · · · .
We have C2 = d2 + h2 + b2, d2 ∈ D2(A), h2 ∈ H2(A) and b2 ∈ B2(A). We can assume that
b2 = 0, therefore mλ = m0 + (λh1 + λ2h2) + λ2d2 + λ3C3 + · · · . Repeat the same argument to
obtain the result.
2. Let mλ = m0+λh1+λ2(h2+d2)+· · · , m′
λ = m0+λh1+λ2(h2+d′2)+· · · , then d(h2+d2) =
1
2 [h1, h1] = d(h2 + d′2), hence d(d2) = d(d′2) and that implies d2 − d′2 ∈ Z2(A) ∩ D2(A) = {0}.
Apply repeatedly the same reasoning to obtain m′
λ = mλ. �
Proposition B.4. Assume that H3(A) = {0}. Let F be a universal deformation formula
and mλ a deformation. Up to equivalence, there exists a formal curve h(λ) in H2(A)[[λ]] such
that h(0) = 0 and mλ = F (h(λ)). In other words, F characterizes all deformations of m0 up to
equivalence and up to change of formal parameter.
Proof. The existence is given by the Lemma B.2. Up to equivalence, we can assume that
mλ = m0 + h(λ) + d(λ) where h(λ) ∈ λH2[[λ]] and d(λ) ∈ λ2H2[[λ]] (Lemma B.3). But the
deformation m′λ = F (h(λ)) can be written as m′λ = m0 + h(λ) + d′(λ) with d′(λ) ∈ λ2H2[[λ]].
Henceforth m′λ = mλ by Lemma B.3. �
C Appendix
Let A be an algebra and AΛ be a deformation of A with product ? . The underlying space of AΛ
is A[[Λ]], and it is easy to check that Mn(A[[Λ]]) = Mn(A)[[Λ]]. Then Mn(AΛ) is a deformation
ofMn(A), the product is the natural one, defined by (aM) ? (a′M ′) = (a ? a′)MM ′, ∀ a, a′ ∈ A,
M,M ′ ∈Mn(A). Conversely:
Proposition C.1. Any deformation of Mn(A) is equivalent to a deformation Mn(AΛ) with AΛ
a deformation of A.
This result is known, but since we have not been able to find a reference, we give a short
proof.
Proof. We refer to [10] for relative deformation theory with respect to a separable subalgebra.
In the present case, the separable subalgebra ofMn(A) isMn, and any deformation is equivalent
to a deformation with normalized Mn-relative cochains [10], that is, cochains Ω : (Mn(A))2 →
Mn(A) that verify for all M ∈Mn, a1, a2 ∈ A:
Ω(Ma1, a2) = MΩ(a1, a2), Ω(a1M,a2) = Ω(a1,Ma2),
Ω(a1, a2M) = Ω(a1, a2)M, and Ω(x1, x2) = 0 if one xi ∈Mn.
26 I.M. Musson, G. Pinczon and R. Ushirobira
Since Mn and A commute, such a cochain is completely determined by its restriction Ω̃ : A2 →
Mn(A) that verifies M Ω̃(a1, a2) = Ω̃(a1, a2)M , ∀M ∈ Mn, a1, a2 ∈ A, and is therefore A-
valued. Summarizing, up to equivalence, we have a new product ? that satisfies
M1 ? M2 = M1M2, M1 ? a = a ? M1 = aM1, (a1M1) ? (a2M2) = (a1 ? a2)M1M2
for all M1,M2 ∈Mn, a, a1, a2 ∈ A and
a1 ? a2 = a1a2 +
∑
n≥1
ΛnCn(a1, a2), ∀ a1, a2 ∈ A
with Cn : A2 → A. So ? defines a deformation AΛ of A.
Now, we will prove that our initial deformation ? of Mn(A) is exactly the deforma-
tion Mn(AΛ): it is enough to show that (aM) ? (a′M ′) is the product of (aM) and (a′M ′)
in Mn(AΛ), for all a, a′ ∈ A, M,M ′ ∈ Mn. But this is true since aM ? a′M ′ = (a ? a′)MM ′,
that is exactly the product of Mn(AΛ). �
References
[1] Alev J., Farinati M.A., Lambre T., Solotar A.L., Homologie des invariants d’une algèbre de Weyl sous
l’action d’un groupe fini, J. Algebra 232 (2000), 564–577.
[2] Arnal D., Benamor H., Pinczon G., The structure of sl(2, 1)-supersymmetry: irreducible representations
and primitive ideals, Pacific J. Math. 165 (1994), 17–49.
[3] Arnaudon D., Bauer L., Frappat L., On Casimir’s ghost, Comm. Math. Phys. 187 (1997), 429–439,
q-alg/9605021.
[4] Bayen F., Flato M., Fronsdal C., Lichnerowicz A., Sternheimer D., Deformation theory and quantization.
I. Deformations of symplectic structures, Ann. Physics 111 (1978), 61–110.
Bayen F., Flato M., Fronsdal C., Lichnerowicz A., Sternheimer D., Deformation theory and quantization.
II. Physical applications, Ann. Physics 111 (1978), 111–151.
[5] Behr E.J., Enveloping algebras of Lie superalgebras, Pacific J. Math. 130 (1987), 9–25.
[6] Djoković D.Ž., Hochschild G., Semisimplicity of 2-graded Lie algebras. II, Illinois J. Math. 20 (1976), 134–
143.
[7] Etingof P., Ginzburg V., Symplectic reflection algebras, Calogero–Moser space, and deformed Harish-
Chandra homomorphism, Invent. Math. 147 (2002), 243–348, math.AG/0011114.
[8] Flato M., Fronsdal C., Parastatistics, highest weight osp(N,∞) modules, singleton statistics and confine-
ment, J. Geom. Phys. 6 (1989), 293–309.
[9] Fulton W., Harris J., Representation theory. A first course, Graduate Texts in Mathematics, Vol. 129,
Springer-Verlag, New York, 1991.
[10] Gerstenhaber M., Schack D.S., Algebraic cohomology and deformation theory, in Deformation Theory of
Algebras and Structures and Applications, Editors M. Hazewinkel and M. Gerstenhaber, NATO Advanced
Science Institutes Series C: Mathematical and Physical Sciences, Vol. 247, Kluwer Academic Publishers
Group, Dordrecht, 1988.
[11] Gerstenhaber M., On the deformation of rings and algebras, Ann. of Math. (2) 79 (1964), 59–103.
[12] Green H.S., A generalized method of field quantization, Phys. Rev. 90 (1953), 270–273.
[13] Lesimple M., Pinczon G., Deformations of the metaplectic representations, J. Math. Phys. 42 (2001), 1887–
1899.
[14] Loday J.-L., Cyclic homology, Grundlehren der Mathematischen Wissenschaften, Vol. 301, Springer-Verlag,
Berlin, 1992.
[15] McConnell J.C., Robson J.C., Noncommutative Noetherian rings. With the cooperation of L.W. Small,
revised ed., Graduate Studies in Mathematics, Vol. 30, American Mathematical Society, Providence, RI,
2001.
http://arxiv.org/abs/q-alg/9605021
http://arxiv.org/abs/math.AG/0011114
Hochschild Cohomology and Deformations of Clifford–Weyl Algebras 27
[16] Montgomery S., Constructing simple Lie superalgebras from associated graded algebras, J. Algebra 195
(1997), 558–579.
[17] Musson I.M., Some Lie superalgebras associated to the Weyl algebras, Proc. Amer. Math. Soc. 127 (1999),
2821–2827.
[18] Palev T., Para-Bose and para-Fermi operators as generators of orthosymplectic Lie superalgebras, J. Math.
Phys. 23 (1982), 1100–1102.
[19] Palev T., Algebraic structure of the Greens’s ansatz and its q-deformed analogue, J. Phys. A: Math. Gen.
27 (1994), 7373–7385, hep-th/9406066.
[20] Pinczon G., The enveloping algebra of the Lie superalgebra osp(1, 2), J. Algebra 132 (1990), 219–242.
[21] Pinczon G., On two theorems about symplectic reflection algebras, Lett. Math. Phys. 82 (2007), 237–253.
[22] Pinczon G., Ushirobira R., New applications of graded Lie algebras to Lie algebras, generalized Lie algebras,
and cohomology, J. Lie Theory 17 (2007), 633–667, math.RT/0507387.
[23] Pinczon G., Ushirobira R., Supertrace and superquadratic Lie structure on the Weyl algebra, and applica-
tions to formal inverse Weyl transform, Lett. Math. Phys. 74 (2005), 263–291, math.RT/0507092.
[24] Samelson H., Notes on Lie algebras, Van Nostrand Reinhold Mathematical Studies, no. 23, Van Nostrand
Reinhold Co., New York – London – Melbourne, 1969.
[25] Scheunert M., The theory of Lie superalgebras. An introduction, Lecture Notes in Mathematics, Vol. 716,
Springer-Verlag, Berlin, 1979.
[26] Sridharan R., Filtered algebras and representations of Lie algebras, Trans. Amer. Math. Soc. 100 (1961),
530–550.
[27] Sternheimer D., Quantization is deformation, Contemp. Math. 391 (2005), 331–352.
[28] Wigner E.P., Do the equations of motion determine the quantum mechanical commutation relations?, Phys.
Rev. 77 (1950), 711–712.
http://arxiv.org/abs/hep-th/9406066
http://arxiv.org/abs/math.RT/0507387
http://arxiv.org/abs/math.RT/0507092
Introduction
1 Clifford algebras and Weyl algebras
1.1 Weyl algebras
1.2 Clifford algebras
1.3 Even Clifford algebras
1.4 Periodicity of Clifford algebras
1.5 Odd Clifford algebras
1.6 Weyl algebras and supersymmetry
1.7 Clifford algebras and symmetry
2 Clifford-Weyl algebras
3 Periodicity of Clifford-Weyl algebras
4 Clifford-Weyl algebras and supersymmetry
5 Cohomology of Clifford-Weyl algebras
6 Universal deformation formula of C(2n+1,2)
6.1 Construction using Ore extensions
7 Algebras A_\Lambda(n) and their representations
8 Algebras A_\lambda(n)
A Appendix
B Appendix
C Appendix
References
|