ESI

The Erwin Schrodinger International Institute for Mathematical Physics

Boltzmanngasse 9 A-1090 Wien, Austria

The Fuzzy Supersphere

H. Grosse G. Reiter

Vienna, Preprint ESI 546 (1998)

June 1, 1999

Supported by Federal Ministry of Science and Transport, Austria Available via anonymous ftp or gopher from FTP.ESI.AC.AT or via WWW, URL: http://www.esi.ac.at

math-ph/9804013 mp-arc/98-283 ESI 546

The Fuzzy Supersphere

H.Grosse
a

a; 1

and G.Reiter

b;2 ;3

Universitat Wien, Institut fur Theoretische Physik, Boltzmanngasse 5, A-1090 Wien, Austria b Technische Universitat Graz, Institut fur Theoretische Physik, Petersgasse 16, A-8010 Graz, Austria

We introduce the fuzzy supersphere as sequence of nite-dimensional, noncommutative Z2-graded algebras tending in a suitable limit to a dense subalgebra of the 1 Z 2-graded algebra of H -functions on the (2j2)-dimensional supersphere. Noncommutative analogues of the body map (to the (fuzzy) sphere) and the super-deRham complex are introduced. In particular we reproduce the equality of the superdeRham cohomology of the supersphere and the ordinary deRham cohomology of its body on the \fuzzy level". 1991 MSC: Keywords: 17B56, 17B70, 46L87, 58A50, 58B30, 58C50, 81T60 Supermanifolds, Lie superalgebras, noncommutative di erential geometry, fuzzy sphere

Abstract

1 Introduction
Thinking about space and time noncommutative geometry 7] o ers an enormous general framework for physical model building, because one can get rid of the concept of points. The basic idea of noncommutative geometry is to formulate rst notions on di erentiable manifolds in terms of their commutative C -algebras of di erentiable, complex-valued functions in order to generalize subsequently these notions, which do not depend on the commutativity
E-mail: grosse@doppler.thp.univie.ac.at Research partly supported by the \Fonds zur Forderung der wissenschaftlichen Forschung (FWF)" through the research project P11783-PHY \Quantum Field Theory on Noncommutative Manifolds". 3 Corresponding author: Tel.: +43 316 873 8670, Fax: +43 316 873 8678, E-mail: reiter@itp.tu-graz.ac.at
1 2

to abstract, not necessarily commutative algebras. So in general one will lose the notion of points (corresponding in ordinary geometry to the spectrum of the commutative algebra of functions) and the role of general, noncommutative manifolds is played by abstract algebras. Of course there is no canonical way how to associate a noncommutative algebra with some mathematical or physical model of spacetime, phase space or some more \exotic" objects and one can contrive lots of di erent procedures. Beside that of fuzzy manifolds 39, 24, 23, 26], which is intimately related to quantization and which we will follow in the sequel, let us also mention (without being complete) a similar approach for the Minkowski space 12], quantum group motivated approaches (see for example 40]) and approaches based on posets (see for example 1]). Fuzzy manifolds are not only C -algebras but whole sequences (or more precisely directed systems) of noncommutative C -algebras, which approximate in a very speci c way the corresponding ordinary manifolds. Each of the C -algebras of such a sequence can be interpreted as description of the corresponding ordinary manifold on which localization is possible only up to a minimal length. By employing the tools of noncommutative geometry and of matrix geometry 13] in particular, it was possible to introduce on a speci c fuzzy manifold, namely the fuzzy sphere, (sections of) vector bundles, a di erential calculus, an integral, etc. and to use this for the formulation of eld theoretical models 39, 38, 21, 22, 25, 34]. When these models are quantized a very interesting feature shows up: They are nite and the fuzziness plays the role of a non-perturbative regulator, which does not break the characteristic symmetries of the corresponding continuum theories. Although supermanifolds are to some extend \baby-noncommutative geometries" they are treated and interpreted in the spirit of classical di erential and algebraic geometry. Localization is (depending on the approach to supermanifolds) more or less present and the term \super" should rather be seen as additional structure. Noncommutative generalizations should be described by Z 2-graded algebras over a ring, that depends on the class of supermanifolds under consideration. We want to mention, that there exist already several articles and books in the literature dealing with various aspects of Z 2-graded C -algebras, supersymmetry and noncommutative geometry. Without being complete let us just mention 31], where notions as cyclic cohomology and Fredholm modules are treated in the Z 2-graded setting, 14, 33], where a possibility of generalizing matrix geometry to the Z2-graded framework is presented and 32], where the concept of a spectral triple is extended to algebras which contain bosonic and fermionic degrees of freedom. Our aim in this article is to develop rst a \fuzzy variant" of the (2j2)-dimensional supersphere and subsequently an analogue to the super-deRham complex on each of the resulting Z 2-graded C -algebras. Motivated by the wish to nd an adequate language for the description of the spinor bundle on the fuzzy sphere one of the authors together with C.Klimcik and P.Presnajder already solved the rst problem 21]. Here we want to embed the description given there a little bit more in the language of supermanifolds in the sense of A.Rogers 47, 48] as well as that of graded manifolds 35, 36] in order to have a good guideline for the development of a Z 2-graded di erential calculus later on. More precisely the article is organized as follows. In the rst two chapters we describe the 2

(2j2)-dimensional supersphere as H1 -deWitt super- respectively graded manifold and establish on its Z2-graded algebra of complex-Grassmann-valued functions additional structures such as a Frechet topology, an inde nite scalar product and a \completely reducible" grade star representation of the orthosymplectic Lie superalgebra osp(1j2). The article is written as self-contained as possible, but for the basics of the theory of supermanifolds we refer the reader to the above cited original articles as well as to the excellent book 2]. In chapter 4 we endow each element of a speci c sequence of submodules of the osp(1j2)-module of complex-Grassmann-valued H1 -functions on the (2j2)-dimensional supersphere with a new Z 2-graded product and de ne by this the fuzzy supersphere. The graded-commutative limit of these products is proven. The body of the (2j2)-dimensional supersphere is the ordinary 2-sphere and chapter 5 is devoted to extend the corresponding body map to the fuzzy setting. In chapter 6 we de ne on each of the \truncated superspheres" (in the sense explained above) an analogue to the super-deRham complex by transposing the idea of derivation-based differential calculi 13, 16] and its extension used for the de nition of a di erential calculus on the fuzzy sphere 39, 37] to the Z 2-graded setting. The resulting complex is nothing else but the complex of Lie superalgebras with coe cients in a non-trivial osp(1j2)-module and it is in nite-dimensional, as it is usual for supermanifolds. The latter fact shows in particular, that the complex is completely di erent to that proposed in 14, 33]. Our construction is natural in the sense, that the noncommutative body map extends - as it is the case for graded manifolds - to a cochain map from the algebra of superforms on the truncated supersphere to the algebra of forms on its body. The nal chapter is devoted to the calculation of the cohomology corresponding with the differential complex. In particular we reproduce the equality of the super-deRham cohomology of the supersphere and the ordinary deRham cohomology of its body on the \fuzzy level". Let us nally mention some conventions which are used throughout the article. When we speak of algebras we always mean associative algebras with identity; algebra homomorphisms always preserve the identity. Left or right modules over an algebra are always assumed to be unital. There will appear lots of Z 2-graded objects. If the object is denoted by O its even part is denoted by O0 , its odd part by O1 . If e is some homogeneous element of such an object its degree will be denoted by e. Speaking of grading in the context of an ungraded object we mean, that the object is endowed with its trivial graduation. Moreover a left or right Z2graded module over a Z 2-graded, graded-commutative algebra is always assumed to be given its canonical Z 2-graded bimodule structure. There is a one-to-one correspondence between representations of a Lie (super)algebra and its universal enveloping algebra: We do not distinguish between these representations and pass freely from the language of representations to the language of modules and vice versa.

2 The (2 2)-dimensional supersphere

j

In this preliminary section we introduce the (2j2)-dimensional supersphere as H1 -deWitt supermanifold (and by this also as graded manifold) and characterize the Z 2-graded algebra 1 1 of H -functions on it as a suitable quotient of the algebra of H -functions on the (3j2)dimensional vector superspace. Moreover we endow this algebra with additional structures, such as a Frechet topology and an inde nite scalar product, which we will need later on for the formulation and proof of the \graded-commutative limit" of the fuzzy supersphere. For L 2 N let BL denote the Grassmann algebra over RL and CL the Grassmann algebra over C L . We view both of them as Z 2-graded algebras. Furthermore we introduce the set

IL := fM = (i ; ; ip)j i ; ; ip = 1; ; L; 1 p L with i <

1 1 1

< ipg f;g :

(1)

If fei gi=1; ;L is a basis of RL (or C L ) then a homogeneous basis of BL (or CL) is formed by the elements

eM := ei ei e; := 1
1

eip with M = (i1; i2;

; ip) 2 IL

(2) (3)

and

j yj

X
M 2IL

yM eM :=

X
M 2IL

jyM j

for every y 2 BL (CL), de nes a norm on BL (or CL ). The R-linear map BL ?! CL , de ned by identifying the basis elements (2) of BL with the corresponding basis elements of CL , is an injective isometric homomorphism of Z 2-graded R-Banach algebras, which we will use to interpret BL as closed and graded R-subalgebra of CL . Consequently the elements of BL are those elements of CL, which are invariant with respect to complex conjugation : CL ?! CL, de ned by 0 1

@ X yM eM A := X yM eM :
M 2In M 2In

(4)

The direct sum of BL;0 -modules

rjs := B BL L;0 r

BL;0 BL;1
r s r +s X
= +1

BL;1 ;

r; s 2 N0 ;

(5)

together with the Z2-grading

rjs := B BL; L;0 0 r j s BL;1 := BL;1 rj0 BL;0 BL 0js BL;1 BL

(6) (7)

and the norm

(xk ; ) :=

r X k=1

xk +

is called (rjs)-dimensional real vector superspace. rjs ?! Rr, The body (or augmentation) map on CL will be denoted by ; the body map BL 4

j j;

rjs ; (xk ; ) 2 BL

(xk ; ) 7! ( (xk )) by rjs . Beside the topology induced by the norm (7), the so-called ne topology, there is another important topology on the vector superspace: the coarse or deWittrjs such that the body map topology. By de nition it is the coarsest topology on BL rjs is continuous. For an arbitrary 2 R+ we de ne the (2j2)-dimensional supersphere S of \radius" as the 3j2 closed topological subspace (with respect to the ne topology) of all points of BL (L 2, xed) ful lling 3 X (8) P (xk ; ) := (xi)2 + 2 4 5 ? 2 = 0 : also 3, 5, 9, 53]). In order to do so one introduces the \north" and \south pole- - bers"

S can be endowed with the structure of a (2j2)-dimensional H1 -deWitt supermanifold (see F := (xk ; ) 2 S j 3j2(xk ; ) = (0; 0;

i=1

(9) (10) (11)

as well as their open complements

U := S nF

in S . Then the superstereographic projections

2j2 h : U ?! BL (xk ; ) 7! 1 2 4 5 x3 (x ; x ; ; )

form a H1 -deWitt atlas on S , because the transition functions

h?1 (y j ; ) = (y1)2 + (y 2 )2 + 2

3 4

(y j ; )

(12)

2j2 2j2 1 1 n ? (0; 0) on the one hand and h (U ) are coarse n ? (0; 0) ?! BL are H1 -functions BL 2j2 2j2 1 open on the other hand. The corresponding (C -)body manifold can be identi ed canonically with the 2-dimensional sphere S of radius embedded in R3 and the body projection S is then simply given by (13) S = 3j2 S :

Let us denote the sheaf of real- and complex Grassmann-valued H1 -functions on a H1 supermanifold X by H1 (?; BL ) and H1 (?; CL), respectively. We view H1 (?; BL ) as subsheaf of H1 (?; CL) in the natural way. That is, for every open subset U of the supermanifold X we de ne complex conjugation : H1 (U; CL) ?! H1 (U; CL) pointwise and characterize H1 (U; BL) as the graded R-subalgebra of -invariant elements. It should be noted, that, if X is a H1 -deWitt supermanifold with body X and body projection X , the pair (X; X H1 (?; BL )), where X H1 (?; BL) denotes the direct image of H1 (?; BL) by X , 1 is a graded manifold 2]. Consequently the map X from the Z 2-graded C -algebra H (X; CL) 1 to the (trivially Z 2-graded) C -algebra C (X; C ) of smooth complex-valued functions on X, de ned by ( X (f )) ( X (x)) := f (x) (14) for all x 2 X , is a surjective homomorphism of graded algebras. 5

The set

3j2 vanishing on the supersphere is a graded of all complex-Grassmann-valued functions on BL 3j2 3j2 1 1 ; CL)=IS according to the ideal in H (BL ; CL). We will identify H (S ; CL) and H1 (BL rst part of the following result.

IS := f 2 H1 (BLj ; CL) f S = 0
32

(15)

Lemma 1 The map : H1 (BLj ; CL)=IS ?! H1 (S ; CL), de ned by (16) (f ) := f S ; f 2 f ; is an isomorphism of Z -graded C -algebras. Moreover, for every f 2 IS there exists a H1 function g 2 H1 (BLj ; CL), such that
32 2 32

f =P g

(17)

is ful lled.

~ := Proof : Analogous to (9),(10) one introduces \north" and \south pole- - bers" F 3j2 3j2 ?1 (0; 0; R ) as well as their open complements U ~ in BL ~ := BL . Then the maps nF 0 3j2 1 ~ : U ~ ?! ? (R2 R+ ) h 3j2

qP (xi ) + 2 (xk ; ) 7! qP i x i (xi) + 2 0v u uX (x ; x ; @t (xi ) + 2

3 =1 2 4 5 3 =1 2 4 5 1 2 3 2

(18)

4 5

i=1

1v u uX x A t (xi ) + 2
3 3 2

4 5

i=1

; 4; 5 ) ;

where the square root is de ned via the Z-expansion of the ordinary square root, are \subsupermanifold charts" of the vector superspace, which one can use to conclude f jS 2 H1 (S ; CL). Obviously is an injective, even homomorphism of Z2-graded algebras and the surjectivity 3j2 of follows from the existence of H1 -partitions of the unity on coarse open coverings of BL 2]. Using once again such a coarse H1 -partitions of the unity one can conclude, that it is enough ~ . The next step is to note, that the condition f jU = 0 means to show (17) on U ~ \S ? 1 ? 1 ? 1 + 1 2 ~ ~ f h jy = = 0 for the functions f h 2 H ( 3j2(R R ); CL) and that it is enough to prove 1 f ~ h?1 = y 3 ? 2 g ; g 2 H1 ( ? (R2 R+ ); CL) : 3j2 But if one exploits the properties of the \super eld" and Z-expansion one reduces the above problem to the same problem on the \C 1 -level", which can be solved in the usual way (see 2, 6]) via integration. 2
3 2

Analogous to the case of the supersphere we can identify the C -algebra C 1 (S ; C ) with the C -algebra C 1 (R3; C ) factored by the ideal IS of all C 1 -functions on R3 vanishing on the ordinary 2-sphere S . Because of (13) and
32

?I

I ;
S

(19)

the \algebraic body map" S : H1 (S ; CL) ?! C 1 (S ; C ) is simply determined by

S (f ) = f ;

(20)

where we introduced the notation f for the equivalence class of 3j2(f ) for some f 2 f . In the spirit of the theory of graded manifolds we introduce the set Vg (X ) of complex global supervector elds on a H1 -deWitt supermanifold X as the set of graded derivations g 1 1 Derg 2-graded C -algebra H (X; CL). V (X ) forms in a natural way a C (H (X; CL)) of the Z 1 C -Lie superalgebra as well as a Z 2-graded H (X; CL)-module. In addition one should note, that there is a surjective Lie algebra homomorphism ~X from Vg (X )0 to the C -Lie algebra V(X) of complex global vector elds on the body manifold X given by ~X (D) X (f ) := X (Df ) (21) for all D 2 Vg (X )0 and all f 2 H1 (X; CL). Now let us in particular denote the partial derivatives of H1 -functions on the vector superspace 3j2 BL by @=@xk in the case of even coordinates and by @=@ in the case of odd coordinates. 3j2 is compact then If N 2 N3 BL 0 is a multiindex and K
j := max jf jK;n (xk ; )2K
32

jN j n;M 2I2

@ @x

@ @

f (xk ; )

(22)

3j2 for all f 2 H1 (BL ; CL), where jN j denotes the length of the multiindex N and we used the 3j2 ; CL). standard notation for partial derivatives of higher order, de nes a seminorm on H1 (BL 3j2 The family of all seminorms for all compact subsets K BL and all natural numbers n 2 N0 3j2 3j2 ; CL) becomes a Z2; CL) such that H1 (BL induces a locally convex topology on H1 (BL graded C -Frechet algebra 2]. For our later considerations it is important to note, that the 3j2 3j2 ; CL) of all polynomials in the coordinate projections with complex ) H1 (BL subset P (BL 3j2 ; CL) 2, 35, 54]. coe cients forms a dense graded subalgebra of H1 (BL 3j2 1 The graded ideal IS is closed in H (BL ; CL) and consequently H1 (S ; CL), endowed with the quotient topology, is also a Z 2-graded C -Frechet algebra. The topology is again induced by a family of seminorms, which are given explicitly by

jf jS jf j j ; K;n := finf 2f K;n

32

2 H1 (S ; CL) :

(23)

The set of all equivalence classes f 2 H1 (S ; CL) having a polynomial representant forms a dense graded subalgebra of H1 (S ; CL), which we denote by P (S ). C 1(R3; C ) and C 1(S ; C ) can be endowed with a C -Frechet algebra structure in an analogous way. Then the \algebraic body maps" 3j2 and S are continuous and the subalgebra P (R3) of polynomials in the coordinate projections with complex coe cients as well as its image P (S ) in C 1(S ; C ) form dense subalgebras in C 1(R3; C ) and C 1(S ; C ), respectively. 3j2 ; CL) has by de nition a unique \super eld expansion" Every f 2 H1 (BL

f = f; + f4 4 + f5 5 + f45
7

4 5

M 2I2

fM

M:

(24)

Inspired by the correspondence of delta functions and Fourier transformation and the rules of Berezin integration 6, 11] we de ne

dxd

f (xk ;

X
3

Z Z @ f; ( ; #; ') + 1 f ( ; #; ') ? f ( ; #; ') ; =1 d# sin # d' 45 ; 2 @

2 0 0 3j2 3j2 ; CL) with ; CL), where fM 2 C 1(R3; C ) are the images of fM 2 H1 (BL for all f 2 H1 (BL respect to the map (14) expressed in spherical coordinates. (25) induces a continuous, even 3j2 C -linear map H1 (BL ; CL) ?! C , which vanishes on IS according to the second part of lemma 1. Consequently the map I : H1 (S ; CL) ?! C ,

i=1

(xi )2 + 2

4 5

:=

(25)

I (f ) := dxd

f (xk ;

X
3

i=1

(xi )2 + 2

4 5

f 2f;

(26)

is well-de ned and again continuous, even and C -linear. In order to introduce an inde nite 3j2 3j2 ; CL ) ; CL) ?! H1 (BL scalar product we de ne a second \involution" (besides ) : H1 (BL via 0 1

@ X fM M A := f; + f
M 2I2

?f

+ f45

4 5

(27)

Apparently (27) is antilinear and ful lls

?f

(fg )

= (?1)fg g f = (?1)f f

(28)

3j2 ; CL). Because leaves IS invariant, induces the same for homogeneous f; g 2 H1 (BL 1 kind of \involution" on H (S ; CL), which we again denote by . Using (26) as well as we can de ne an even sesquilinear form on H1 (S ; CL) by

where the normalization has been chosen such that h1 j1 i = 1. It is continuous in both entries, non-degenerate and ful lls hf jg i = hg jf i : (30)

hf jg i := 2 I (f g) ;

(29)

3 Action of the (1 2)-dimensional orthosymplectic Lie superalgebra

j

Beside the algebra of functions on a manifold itself there is another important ingredient for the de nition of fuzzy manifolds 24, 26, 39]: The action of a Lie group respectively a Lie algebra on the algebra of functions. There are natural graded representations of the (1j2)dimensional orthosymplectic Lie superalgebra osp(1j2) on the Z2-algebras of H1 -functions on 8

the (3j2)-dimensional vector superspace as well as on the (2j2)-dimensional supersphere. They can be seen as the super-generalizations of the actions of the ordinary angular momentum on the algebras of C 1 -functions on the respective bodies 10]. These graded representations are reducible. We give their reduction into irreducible subspaces and introduce super-analogues of the spherical harmonics. It should be remarked, that the above mentioned reduction can be found mutatis mutandis also in 21] and that superspherical harmonics were studied in a di erent framework in 10]. Before analyzing the above mentioned in nite-dimensional graded representations of the orthosymplectic Lie superalgebra let us rst review some well-known facts of this Lie superalgebra and its nite-dimensional, irreducible graded representations in order to x notations and conventions. The (complex) orthosymplectic Lie superalgebra osp(1j2) is the (3j2)-dimensional Z2-graded C -vector space spanned by three even basis elements Jk ; k = 1; 2; 3; and two odd basis elements J ; = 4; 5; together with the graded Lie bracket de ned by

Ji; Jj ]g := i

X
3 5

1 Ji ; J ]g := 2

k=1

"ijk Jk
( i) J (31)

X
3

sical simple Lie superalgebra of type I 8], whose even part is isomorphic to the 2-dimensional complex special linear Lie algebra sl(2) according to (31). The triangular decomposition osp(1j2) = N? H N+ (32) of osp(1j2), where H is the Cartan subalgebra and N are the nilpotent graded Lie subsuperalgebras of osp(1j2) corresponding with the positive respectively negative roots, is chosen - as usual 4, 43] - according to H := spanC fJ3g N? := spanC fJ? ; J5g (33) N+ := spanC fJ+ ; J4g with J := J1 iJ2 : (34) Furthermore there exist two essentially di erent grade adjoint operations z ; = 0; 1; on osp(1j2) 4, 42, 43], which correspond, when restricted to the even part of osp(1j2), with the compact real form of osp(1j2)0. Explicitly they are given by Jiz := Ji z := (?1) J (35) J4 5 z +1 J5 := (?1) J4 : 9

1 (i ) J ; J ; J ]g := 2 2 i i i=1 where k are the Pauli matrices and "ijk is 3-dimensional permutation symbol. It is a basic clas-

=4

All irreducible graded representations of osp(1j2) on nite-dimensional Z 2-graded C -vector 1 spaces are highest weight modules speci ed by a highest \weight" j 2 2 N0, the so-called superspin, and a de nite degree & 2 Z2 of the unique, 1-dimensional highest weight space. Between two irreducible, nite-dimensional graded representations of osp(1j2) with the same highest weight j and the same degree & of the highest weight space there exists an isomorphism of graded osp(1j2)-representations, determined by the requirement, that the highest weight vectors are mapped onto each other 29, 30]. N and & 2 Z For xed j 2 1 2 let us denote the corresponding irreducible graded representation 2 0 j) by (j ), the nite-dimensional representation space by V (j; & ) and by e( j;j;& 2 V (j; & )& the highest weight vector, speci ed (up to a complex constant) via
j) (j ) (j ) J3 ej;j;& = je( j;j;& (j ) (j ) (j ) (j ) ej;j;& = 0 J+ ej;j;& = J4

(36)

according to (32),(33). Then the elements 42(j ?l) ?(l + m + 1) J (j ) l?m J (j ) e(j ) (37) := ?(2 5 j;j;& j + 1)?(l ? m + 1) ? and the numbers of V (j; & )& + , where the (non-negative) number l is given by l := j ? 1 2 and m run through f0; 1g respectively f?l; ?l + 1; ; l ? 1; lg, form a homogeneous basis of V (j; &). Using the graded commutation relations (31) one can deduce the action
j el;m;& +
( )

j) = m e( l;m;& + q j) = (l ? m)(l + m + 1) e( l;m+1;& + q j) = (l + m)(l ? m + 1) e( l;m?1;& + 1 pj ? m e(j ) (38) = ?2 j ? ;m+ ;& +1 r 1 (j ) (j ) J4 ej? ;m;&+1 = ? 2 j + m + 1 e(j) 2 j;m+ ;& 1 pj + m e(j ) (j ) (j ) ej;m;& = 2 J5 j ? ;m? ;& +1 r 1 (j ) (j ) (j ) J5 ej? ;m;&+1 = ? 1 2 j ? m + 2 ej;m? ;& of the homogeneous basis fJ3; J ; J g of osp(1j2) on the homogeneous basis elements (37). (37) and (38) show in particular, that the even and the odd subspace of V (j; & ) correspond to 1 irreducible sl(2)-modules with highest weights l = j and l = j ? 2 and
(j ) (j ) el;m;& + J3 (j ) (j ) J+ el;m;& + (j ) (j ) J? el;m;& + (j ) (j ) ej;m;& J4
1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2

dimC V (j; & )& = 2j + 1 dimC V (j; & )& +1 = 2j dimC V (j; & ) = 4j + 1 : Now let us introduce the graded derivations 5 3 X X 1 @ (3j2) j ( i) Ji := ?i "ijk x @xk + 2 ; =4 j;k=1 10

(39)

@ @ ;

i = 1; 2; 3 ;

J (3j2) := 1 2 k=1
3

XX
5 =4

ixk (

k)

@ ? ( ) @ ; k @ @xk

= 4; 5 ;

(40)

3j2 3j2 ) is a graded ; CL). The corresponding C -linear map (3j2) : osp(1j2) ?! Vg (BL of H1 (BL representation of osp(1j2), that leaves IS (according to lemma 1) invariant. Consequently, de ning the action via the action on representants, (3j2) induces a graded representation (S ) : osp(1j2) ?! Vg (S ) and the latter is a \grade star representation with respect to z1 ", that is, D (S) E D z (S) E f JA g = (?1)JA f JA (41) f g is ful lled for all homogeneous f ; g 2 H1 (S ; CL) and A = 1; ; 5. (3j2) as well as (S ) are reducible. In order to nd their decomposition into irreducible subspaces we note, that the even polynomials
1

?(2j + 1) s;j := Yj;j; j 0 2 ?(j + 1) j +2s for all j; s 2 N0, and the odd polynomials ?(2j + 1) s;j := Yj;j; 3 1 1 j ? 2 2 ?(j + 1 ) j + 2 +2s
2 1 2

X
3

i=1

(xk )2 + 2

!s

4 5

x1 + ix2 ;
j? 1 2

(42)

X
3

i=1

(xk )2 + 2

!s
4 5

x1 + ix2
(3 2)

x3 4 + (x1 + ix2)

;
(43)

for all j 2 N0 + ; s 2 N0, are highest weight vectors of weight vectors with di erent s but the same j are mapped under the canonical projection onto the same element of P (S ), which we will denote by Y j j;j;2j . The normalization in (42) and (43) is chosen such that E D j (44) Y j;j;2j Y j j;j;2j = 1 is ful lled. As we will see, the graded osp(1j2)-submodules
j with highest weight j . Such highest
s;j V s;j := U (osp(1j2))(3j2)Yj;j; 2j j j (S ) V := U (osp(1j2)) Y j;j;2j ;

(45)

where U (osp(1j2)) denotes the enveloping algebra of osp(1j2), constitute the irreducible direct summands of the graded representations (3j2) and (S ) .

Lemma 2 The restriction of

(3 2)

s;j , highest weight j and degree 2j of the highest weight vector. Moreover the weight vector Yj;j; 2j 3j2 3j2 ; CL) can be decomposed as Z2-graded osp(1j2)) H1 (BL dense graded subalgebra P (BL module according to M s;j 3j2 P (BL )= V : (46)
s2N 0 1N j2 2 0

j to V s;j is an irreducible graded representation with highest

Proof : Let us assume, that the restriction of (3j2) to V s;j is reducible. Then there exists a graded subrepresentation of (3j2) on a graded vector subspace W V s;j and a non-trivial

11

homogeneous w 2 W n V s;j . Because (3j2) is standard-cyclic, w is given either by

w=
or by

Nj X q=0

q J?

(3 2)

j q Y s;j ; j;j;2j

q 2 C ; Nj

6= 0 ; 6= 0 :
2j and

w=

Nj X q=0

q J?

(3 2)

j q J (3j2) Y s;j ; 5 j;j;2j

q 2 C ; Nj

Using the explicit expressions (40),(42) and (43) one nds by induction Nj
(3j2) Nj s;j ; J+ w = j Yj;j; 2j

2 C n f0g ;
j

in the rst case and Nj

2j ? 1 and
(3j2) Nj (3j2) s;j ; J+ J4 w = j Yj;j; 2j

2 C n f0g ;

s;j 2 W implying the contradiction W = V s;j . in the second case. In any case one has Yj;j; 2j 3j2 Apparently (3j2) does not only leave P (BL ) invariant but also the graded vector subspaces 3j2 3j2 n P (BL ) P (BL ) of polynomials in the coordinate projections with complex coe cients of 3j2 3 degree n 2 N0. Consequently we have V s;j P j + 2j +2s (BL 2j is of ), where the summand 2 3 1 course to be understood as 0, if j 2 N0, and as 2 , if j 2 N0 + 2 . In order to show (46) we only have to prove M 3j2 V s;j (47) P n(BL )=
3 2

for all n 2 N0. Using the fact, that highest weight vectors of V s;j and V s0 ;j 0 , with s 6= s0 ; j 6= j 0, 3 3 such that j + 2 2j + 2s = j 0 + 2 2j 0 + 2s0 = n, have to be linear independent because they are highest weight vectors to di erent highest weights, one nds, that the intersection of V s;j 3 2j + 2s = with the sum of all other vector spaces V s0 ;j 0 with s 6= s0 ; j 6= j 0 such that j + 2 0 3 s;j 0 0 j + 2 2j + 2s = n is f0g. Consequently j+ 2j+2s=n V is a well-de ned graded vector 3j2 subspace of P n(BL ). In order to show equality it is enough to show the equality of dimensions. But using (39) we nd 2n2 +2n +1 for dimC j + 2j+2s=n V s;j by induction, which is exactly 3j2 ). 2 dimC P n (BL
3 2 3 2

2j +2s=n j+ 3 2

Although the above result about the representation (3j2) is not of primary interest for itself it is of central importance for deducing the following reduction of (S ) .

Proposition 1 The restriction of

( )

est weight vector Y j j;j;2j , highest weight j and degree 2j of the highest weight vector. Moreover the dense graded subalgebra P (S ) H1 (S ; CL) can be decomposed as Z 2-graded osp(1j2)module according to M j P (S ) = V : (48)
1 j2 2 N0

S to V j is an irreducible graded representation with high-

12

Proof : The rst statement is a consequence of the facts, that Y j j;j;2j is non-trivial and that 3 j 2 the canonical projection H1 (BL ; CL) ?! H1 (S ; CL) is an surjective, even homomorphism of osp(1j2)-modules. Because of lemma 2 we only have to show, that the intersection of V j and the sum of all other vector subspaces V j 0 is trivial, in order to prove (48). But this is an easy consequence of the fact, that V j and V j 0 are irreducible Z2-graded osp(1j2)-modules with di erent highest weights. 2

According to proposition 1 and (37) the elements of V j

Yj l;m;2j +

42(j ?l) ?(l + m + 1) J (S ) l?m J (S ) Y j ; := ?(2 5 j;j;2j j + 1)?(l ? m + 1) ?

P (S ) H1 (S ; CL)
(49)

with l := j ? 1 and 2 f0; 1g, m 2 f?l; ?l + 1; ; l ? 1; lg, form a homogeneous basis of 2 j 1 V for all j 2 2 N0. We will call them superspherical harmonics. Of course, the action of osp(1j2) on the superspherical harmonics is given by (38) and using (41) and (44) we nd, that they are (pseudo)orthonormalized according to

E j0 Y Yj = (?1) 2j 1 0 0 0 0 l;m;2j+ l ;m ;2j +

jj 0

0 mm0 :

(50)

4 Truncation and the fuzzy supersphere

We truncate the direct sum (48) of graded subrepresentations of osp(1j2) on the Z2-graded algebra of complex-Grassmann-valued H1 -functions on the (2j2)-dimensional supersphere at each integer superspin and establish a new, Z 2-graded associative product on each of these truncated sums. In this way we get the fuzzy supersphere, that is a whole sequence of nitedimensional, noncommutative Z 2-graded algebras possessing a \graded-commutative limit". The procedure described above was studied in a similar way in 21]. Here we have slightly di erent conventions and we will discuss the graded-commutative limit in greater detail, which includes the introduction of noncommutative superspherical harmonics in particular. Let us rst formulate our aim, that is the basic idea of the construction of fuzzy manifolds 1 N let us introduce the truncated direct sum applied to our case. For every q 2 2

Hq :=

j2 1 N0 j 2q

Vj

(51)

of Z2-graded osp(1j2)-modules. For some in nite subset S 1 N and every q 2 S we want 2 0 to nd Z 2-graded C -algebras Aq , which are at the same time Z 2-graded osp(1j2)-modules together with even osp(1j2)-module isomorphisms q : Hq ?! Aq . Denoting by q0 q ; q; q 0 2 S ; q q 0 ; the canonical injections Hq ?! Hq0 we can introduce even, injective osp(1j2)-module homomorphisms q0 q : Aq ?! Aq0 by
q0 q := q0 q0 q

?1 q ;

(52)

13

= q00q (53) for all q; q 0; q 00 2 S ; q q 0 q 00. Consequently (Hq ; q0q ) and (Aq ; q0q ) are isomorphic directed systems of Z 2-graded osp(1j2)-modules and their direct limits can be identi ed with P (S ) (as Z2-graded osp(1j2)-modules). The corresponding homomorphisms Hq ?! P (S ) and Aq ?! P (S ) of Z2-graded osp(1j2)-modules are denoted by q and q , respectively. Now let f and f 0 be elements of P (S ). Then there will be a number p 2 Ssuch that (omitting the canonical injections p and qp as we will always do in the sequel) f ; f 0 2 Hq for all q p. We cannot form products ff 0 in the truncated sums Hq , but we can form q (f ) q (f 0 ) in the isomorphic objects Aq . A priori these products are not connected with the product ff 0 2 H1 (S ; CL) in any way, but we can map them into the direct limit according to
q00 q0 q0 q

which ful ll

?ff 0 := ? (f ) (f 0) = q q q q

?1 ? q (f ) q (f 0) q

2 P (S ) H1 (S ; CL)

(54)

and \compare" (ff 0 )q with ff 0 . More exactly f(ff 0)q gp q;q2Sis a sequence in H1 (S ; CL), whose convergence to ff 0 (with respect to the Frechet topology) can be investigated.

De nition 1 The directed system (Aq ;

is ful lled for all f ; f 0 2 P (S ).

q0q ) is said to possess a graded-commutative limit if

0 0 qlim !1 ff q = ff

(55)

Because of the algebraic structure of each Aq , we can view each Aq as \noncommutative H1 supermanifold"; the existence of the graded-commutative limit guarantees the relation with the (2j2)-dimensional supersphere. We will show in the sequel, that a directed system with graded-commutative limit really exists. We choose S= N. For all q 2 N one nds by induction dimC Hq;0 = q 2 + (q + 1)2 dimC Hq;1 = 2q (q + 1) dimC Hq = (2q + 1)2 ; (56)

from which we can conclude, that there will exist isomorphisms of Z2-graded C -vector spaces q between Hq and the Z 2-graded C -algebra End C (V ( 2 ; 1)) of all endomorphisms of the Z 2-graded q representation space of the irreducible graded osp(1j2)-representation with highest weight 2 and odd highest weight vector. q ; 1) we can introduce a scalar product by On each of the Z2-graded representation spaces V ( 2
( ) ) e( l;m;1+ el0 ;m0 ;1+
2 2

0 mm0

(57)

q ; 1) becomes a Z for all ; 0 2 f0; 1g, m 2 f?l; ; lg, m0 2 f?l0 ; ; l0g, such that V ( 2 2 graded Hilbert space. With respect to this scalar product and the grade adjoint operation z0 , q the irreducible graded representation ( ) is a grade star representation 4, 43]. Employing the
2

14

superadjoint operation z with respect to this scalar product as well as the supertrace Trs we q ; 1)) End (V ( q ; 1)) ?! C via can de ne a sesquilinear form h j i q : EndC (V ( 2 C 2
2

hf jg i q := ?Trs f zg =
2

l X X
1

The normalization has been chosen such that hIdV (q=2;1) jIdV (q=2;1) i q = 1. h j i q is even, nondegenerate and ful lls (59) hf jg i q = hg jf i q : Beside this inde nite scalar product, which has exactly the same properties as (29), we can q ; 1)) in a natural way by also establish a Z2-graded osp(1j2)-module structure on EndC (V ( 2 q q ; 1))) via q de ning ad( ) : osp(1j2) ?! Derg pl(EndC (V ( 2 C (EndC (V ( 2 ; 1)))
2 2 2 2 2

) ( ) z (?1) e( l;m;1+ f g el;m;1+ =0 m=?l

2 2

(58)

ad( ) J (f ) := J ( ) ; f g (60) q ; 1)). Here we denoted by pl(End (V ( q ; 1))) the general for all J 2 osp(1j2), f 2 EndC (V ( 2 C 2 q linear Lie superalgebra of the Z2-graded C -vector space EndC (V ( 2 ; 1)) and by ; ]g its graded q Lie bracket. Using the fact, that ( ) is a grade star representation with respect to z0 , we can q deduce, that ad( ) is a grade star representation with respect to z1 . q The graded representations ad( ) are reducible for all q 2 N. In order to nd their reduction into irreducible subspaces we proceed as we did for the in nite-dimensional case. Employing (38) and the graded Leibniz rule we can check, that the even endomorphisms
2 2 2 2 2

v u u 1 t 2j (2j )!!?(q ? j + ) J q j? J q ? 3 J q + J q J q ; (62) Yj;j; := 4 q+ ?(j + )?(q + j + ) q for all j 2 N + ; j q , are highest weight vectors of ad with highest weight j and
q ( )j
2

for all j 2 N0; j q , as well as the odd endomorphisms

+7 2 1 1 2 1 2 1 2 3 2 ( ) 2 +
1 2

j (2j ? 1)!!?(q ? j + 1) ( q ) j Yj;j;0 := 2 ?( j + 1)?(q + j + 1) J+ ;

(2)

q j

(61)

( )
2

(2) 4

( ) ( ) 2 2 + 5

normalized according to

1 2

(2 )

( )j ( )j Yj;j; Yj;j; 2j 2j
2 2

The corresponding graded osp(1j2)-submodules

= 1:
q
2

(63) (64)

V(

q)j

( )j := ad( ) (U (osp(1j2))) Yj;j; 2j

2

are the direct summands we are looking for.

Proposition 2 The restriction of adq q to V q j is an irreducible graded representation of osp(1j2) with highest weight vector Yj;j; jj , highest weight j and degree 2j of the highest weight vector. Moreover EndC (V ( q ; 1)) can be decomposed as Z -graded osp(1j2)-module according to M qj q
(2 ) 2 (2 ) (2) 2 2

EndC (V ( ; 1)) = 2 j2 15

j q

1N 2 0

V(

(65)

Proof : The proof of irreducibility can be taken over literally from lemma 2: The numbers Nj can again be determined by induction (with the same result as in lemma 2) but now by using the explicit expressions (61),(62). The fact, that we are considering only graded highest weight modules with di erent highest weights guarantees that the direct sum in (65) is well-de ned ; 1)) results from our dimensional considerations (56). 2 and the equality with EndC (V ( q 2

Consequently for every q 2 N and j q the elements 42(j ?l) ?(l + m + 1) ad( q ) J l?m ad( q ) J Y ( q ) j Yl;m;2j+ := ?(2 ? 5 j;j;2j j + 1)?(l ? m + 1)
(2)

q j

(66)

and 2 f0; 1g, m 2 f?l; ?l + 1; ; l ? 1; lg, form a homogeneous of V ( ) j , with l := j ? 1 2 q ( )j basis of V . Because of the isomorphisms (68) de ned below, we will call these elements of EndC (V ( q ; 1)) noncommutative superspherical harmonics and an analogous calculation which 2 yielded (50) shows, that they are again (pseudo)orthonormalized according to
2 2

( )j )j Yl;m; Yl(0;m 0 ;2j 0 + 2j +

2 2

q 0

0 q

= (?1)

2 1

jj 0

0 mm0 :

(67)

In consideration of proposition 2 we can introduce lots of even osp(1j2)-module isomorphisms q q : Hq ?! EndC (V ( 2 ; 1)) and we choose especially for every q 2 N the C -linear map de ned by q ( )j j (68) q (Y l;m;2j+ ) := Yl;m;2j + for all j q; 2 f0; 1g and m 2 f?l; ?l + 1; ; l ? 1; lg, because the corresponding directed ; 1)); q0q ) has the desired property. system (EndC (V ( q 2
2

a graded-commutative limit.

Proposition 3 The directed system (EndC (V ( q ; 1));

2

q0q ) corresponding with

(68) possesses

Proof : Because the topology on H1 (S ; CL) is induced by seminorms (23), superspherical harmonics - ordinary and noncommutative as well - are (pseudo)orthonormalized and (S ) and q ad( ) are grade star representations of the same type, it is enough to show
2

( )j ( )j ( )j qlim !1 Yl;m;2j + Yl0 ;m0 ;2j 0 + 0 Yl00 ;m00 ;2j 00 +

2 2 2

q 0

q 00

00 q

j j = Yj l;m;2j + Y l0 ;m0 ;2j 0 + 0 Y l00 ;m00 ;2j 00 +

00

00

(69)

N ; 0; 00 2 f0; 1g; m0 2 f?l0; ; l0g; m0 2 f?l0; ; l0g and j 2 1 N ; 2 for all j 0; j 00 2 1 2 0 2 0 0 00 0 00 0 00 0 00 f0; 1g; m 2 f?l; ; lg with jj ? j j j j + j ; m = m + m ; = 2j + 2j + 2j + 0 + 00 . q 0 0 ) j ( Now we can interpret Y j l0 ;m0 ;2j 0 + 0 and Yl0 ;m0 ;2j 0 + 0 0 as multiplication operators in P (S ) and 0 = 0; 1; m0 = ?l0 ; ; l0 g and ; 1)), respectively. Then the sets f Y j EndC (V ( q 0 ;m0 ;2j 0 + 0 j 2 l q 0 f Yl(0 ;m)0j;2j0+ 0 j 0 = 0; 1; m0 = ?l0; ; l0 g are irreducible osp(1j2)-tensor operators and we can apply the osp(1j2)-Wigner-Eckart theorem 41, 55] to conclude, that we can restrict our attention to the cases 0 = 00 = 0; m0 = j 0; m00 = j 00. We nd 0 00 0 j 00 0 j 00 Y j 0 +j 00 0 00 0 00 ; j 0; j 00 2 1 Y = c Yj j 0 00 00 00 0 0 j ;j ;2j j ;j ;2j j +j ;j +j ;2(j +j ) 2 N0 ;
2 2

16

as well as

( ) j +j j 0; j 00 2 1 Yj(0 ;j)0;j2j 0 Yj(00 ;j)00j;2j 00 = cq j 0 j 00 Yj 0 +j 00 ;j 0 +j 00 ;2(j 0 +j 00 ) ; 2 N0; where the coe cients cj 0 j 00 ; cq j 0j 00 2 R, which are also determined by the explicit expressions
2

q 0

q 00

q 0 00
2

(42),(43) and (61),(62), ful ll

1 N0. But because of (50) and (67) this proves the proposition. for all j 0; j 00 2 2

q qlim !1 cj 0 j 00

= cj 0 j 00

Consequently we have succeeded in nding a sequence of Z 2-graded C -algebras tending in 1 the limit described above to the Z2-graded C -algebra H (S ; CL) of H1 -functions on the (2j2)-dimensional supersphere S . Transposed to the language of noncommutative geometry this means, that we have approximated the (2j2)-dimensional supersphere by a sequence of noncommutative supermanifolds. Because our construction is exactly the same as the one q ; 1)); 0 ) of ordinary fuzzy manifolds 24, 26, 39] we will call the directed system (EndC (V ( 2 qq ((2j2)-dimensional) fuzzy supersphere. Each one of the Z 2-graded C -algebras and graded q osp(1j2)-modules EndC (V ( 2 ; 1)) we will call truncated ((2j2)-dimensional) supersphere and we introduce the shorter notation S ;q for it. It is worthwhile to mention the following nice property of the isomorphisms q and the truncated superspheres S ;q : For all q 2 N we nd q q 2 ( ) ( ) k Jk =: Xk ; k = 1; 2; 3; q (x ) = p q(q + 1) q q 2 = 4; 5; (70) J ( ) =: ( ) ; q( ) = p q(q + 1) and consequently
2 2 2 2

X
3

k=1

q (x
3

k )2 +
( Xk
2

q(
2

) q ( 5) ? q ( 5 ) q ( 4)
(2) 4

X
k=1

q)

(2) 5

(2) 5

(2)

= 2IdV ( q ;1) ;
2

(71)

because the left hand side is 4 2=(q (q + 1)) times the representation of the standard secondorder Casimir operator 43] of osp(1j2). That is, the de ning relation of the (2j2)-dimensional supersphere is ful lled on each truncated supersphere.

5 The noncommutative body map

The existence of a body and an \algebraic body map" is of great structural importance in the theory of supermanifolds respectively graded manifolds and one should have an fuzzy analogue. We will see, that it is very natural to interpret the fuzzy sphere as \noncommutative body" of the fuzzy supersphere. The corresponding noncommutative body map will be a surjective homomorphism of directed systems, but none of the single maps will be an algebra homomorphism (which would be impossible as map between simple algebras of di erent dimensions). The algebra homomorphism property will be recovered in the \graded-commutative limit". 17

Let us rst describe the \noncommutative body" of the fuzzy supersphere, that is the fuzzy sphere, in an adequate language (see also 39, 37, 22, 21]). The C -linear map (3) : sl(2) ?! V(R3), de ned by 3 X "ijk xj @k ; i = 1; 2; 3; (72) J(3) := ? i i is a representation of sl(2). It leaves IS invariant and induces a representation (S) : sl(2) ?! V(S ), which is a star representation with respect to the adjoint operation y corresponding with the compact real form of sl(2) and the scalar product h j i : C 1 (S ; C ) C 1 (S ; C ) ?! C , de ned by Z Z2 1 (73) hf jg i := 4 d# sin # d' f( ; #; ') g( ; #; ') ; 0 0 ?(2j + 2) x1 + ix2 j ; := 2j ?( j + 1) j
(S)

j;k=1

where f ; g are representants of f and g, expressed in spherical coordinates. The normalized (highest weight) spherical harmonics

Yjj

j 2 N0 ;
(S)

(74) to the sl(2)-

are highest weight vectors of submodules

with highest weight j , the restriction of

(S)

Vj := U (sl(2)) Yjj (75) is irreducible and the dense graded subalgebra P (S ) C 1 (S ; C ) can be decomposed as sl(2)-module according to M P (S ) = Vj : (76)
The (ordinary) spherical harmonics, given by
j := Ym j 2N0

?(j + m + 1) (S) j ?m Yjj ; ?(2j + 1)?(j ? m + 1) J?

(77)

where m 2 f?j; ?j + 1; ; j g, are orthonormal and they form a basis of Vj for every xed j 2 N0 . As in the case of the supersphere one establishes on each truncated direct sum

Cq :=

j2N0 j q

Vj ;

q 2 N0 ;

(78)

of sl(2)-modules a (noncommutative) associative product. In order to do so, let us denote q q , by by q( ) : sl(2) ?! EndC (V( q )) the irreducible sl(2)-representation with highest weight 2 2 ) corresponding with the weight space decomposife(m )gm2f? q ; ; q g the canonical basis of V( q 2 q ( ) q gm2f? q ; ; q g tion and by h j i the scalar product on V( 2 ) xed by the requirement, that fem becomes an orthonormal basis. Then
2 2 2 2 2 2 2

1 Tr f yg = 1 X e( q ) f yg e( q ) hf jg i q := q + m 1 q + 1 m=? q m
2 2 2 2

q m= 2

(79)

18

q ))) the C -Lie is a scalar product on EndC (V( q )). Moreover, denoting by DerC (EndC (V( 2 2 q )), the representation ad( q ) : sl(2) ?! algebra of derivations of the C -algebra EndC (V( 2 q ))), de ned via DerC (EndC (V( 2 h q i q (80) ad( ) J(f) := J( ) ; f )), becomes a star representation with respect to (79). for all J 2 sl(2); f 2 EndC (V( q 2 For all q 2 N0 and all j 2 N0; j q , the normalized endomorphisms
2 2 2

Yj

(2)

q j

j + 1)!!(q + 1)?(q ? j + 1) ( q ) j := 2 (2j?( J+ j + 1)?(q + j + 2)

2

(81)

are highest weight vectors of the representations ad( ) , the corresponding sl(2)-submodules
2

q q q ( )j V( ) j := ad( ) (U (sl(2)))Yj
2 2 2

(82) (83)

)) can be decomposed as sl(2)-module according to are irreducible and EndC (V( q 2

q )) = EndC (V( 2

j2N 0 j q

q V( ) j :
2

For every q 2 N we can introduce noncommutative spherical harmonics by Ym :=

(2)

q j

j ?m ( q ) j q ?(j + m + 1) ad( ) J? Yj ; ?(2j + 1)?(j ? m + 1)

2 2

(84)

where j 2 N0; j q and m 2 f?j; ; j g, and they form an orthonormal basis of EndC (V( q )). 2 q ^ Now we can de ne isometric isomorphisms q : Cq ?! EndC (V( 2 )) of sl(2)-modules by
( )j j ) := Ym ^q (Ym ;
2

j 2 N0; j q; m 2 f?j;

; jg ;
0

(85)

as well as injective sl(2)-module homomorphism ^q0q : EndC (V( q )) ?! EndC (V( q2 )); q q 0; 2 via ?1 ; (86) ^q0 q := ^q0 ^q0 q ^q q )); ^ 0 ) where ^q0 q denote the canonical injections Cq ?! Cq0 . Then (Cq ; ^q0q ) and (EndC (V( 2 qq are isomorphic directed systems of sl(2)-modules and their direct limits can be identi ed with q )); ^ 0 ) possesses a commutative limit: That P (S ). Moreover the directed system (EndC (V( 2 qq 0 0 is, for two arbitrary f ; f 2 P (S ) the sequence f( )q gp q;q2N with
?1 ^q (f ) ^q (f 0) ; ( 0 )q := ^q ^q (f ) ^q (f 0 ) = ^q

(87)

q )) ?! P (S ) and p is some number, such where ^q are the \limit homomorphisms" EndC (V( 2 that f ; f 0 2 Cp , converges to 0 2 P (S ) with respect to the Frechet topology of C 1 (S ; C ). The directed system (EndC (V( q )); ^q0q ) is called (2-dimensional) fuzzy sphere; each one of the 2 q )) is called truncated (2-dimensional) sphere and we C -algebras and sl(2)-modules EndC (V( 2 denote it by S ;q . Similar to the truncated supersphere one nds for all q 2 N q) q ( ) ^q (xk ) = p 2 J( =: Xk ; k = 1; 2; 3; (88) k q(q + 2)
2 2

19

and

k=1

X^
3

k 2 X q (x ) = k=1
3

X( k

q)
2

= 2IdV ( q ) :
2

(89)

In order to introduce a natural noncommutative analogue to the body map S : H1 (S ; CL) ?! C 1(S ; C ) we note, that S is a sl(2)-module homomorphism because of (20) and (21). Moreover the restriction of S to P (S ) is (respectively induces) a surjective homomorphism
S jP (S
)

: P (S ) ?! P (S )

(90)

of sl(2)-modules and Z 2-graded C -algebras and we have explicitly 8 (?1) > l ; <p Ym 2j + = 0; j S (Y l;m;2j+ ) = > 2l + 1 : 0; 2j + = 1; for all m 2 f?l;

(91)

; lg. Consequently
S ;q := ^q
S

?1 q : S ;q ?! S ;q

(92)

is a well-de ned surjective sl(2)-module homomorphism for all q 2 N and we will call it (noncommutative) body map of the truncated ((2j2)-dimensional) supersphere. Beside the nice feature
S ;q (Xk 2 ) = Xk2
( )

( )

S ;q (

(2)

) = 0;

k = 1; 2; 3; = 4; 5;

(93)

there are some other immediate consequences, which suggest this interpretation.

Proposition 4 (

S ;q ) is a surjective homomorphism (S ;q ; q0q ) ?! (S ;q ; ^q0q ) of directed systems and lim ?! S ;q is simply given by S jP (S ) . Moreover ? ? (f ) (f 0) = lim ? (f ) (f 0) lim ^ (94) S q S ;q q q q q?!1 q?!1 S is ful lled for all f ; f 0 2 P (S ).

Proof : For all q; q 0 2 N; q q 0 we have S ;q0 q0 q = ^q0 q S ;q as well as S jP (S ) q = ^q S ;q by construction, which proves the rst part. The \homomorphism property in the limit" follows according to

lim ^q S ;q q (f ) q (f 0 ) = q?!1 lim S (ff 0 )q = S (ff 0) = q?!1 ? = S (f ) S (f 0) = q?!1 lim S (f ) S (f 0 ) q where we used the rst part of the proposition, the (graded-)commutative limit of the fuzzy (super)sphere and the continuity of S . 2

20

6 A graded di erential calculus on the fuzzy supersphere

Let X be a H1 -deWitt supermanifold. We introduced the elements of the C -Lie superalgebra g 1 and Z 2-graded H (X; CL)-module V (X ) of complex, global supervector elds on X as graded derivations of the Z2-graded C -algebra H1 (X; CL). The latter concept can be generalized to 1 arbitrary Z2-graded C -algebras without any change, when we replace \Z 2-graded H (X; CL)module" by the formulation \Z2-graded module over the graded center Z g (H1 (X; CL)) of the Z2-graded C -algebra H1 (X; CL)", which is equivalent in the graded-commutative case. Moreover, again viewing X as graded manifold, N0-homogeneous, complex, global superdifferential forms are by de nition graded-alternating H1 (X; CL)-multilinear, or equivalently Z g (H1(S ; CL))-multilinear maps from Vg (X ) to the Z2-graded C -algebra of H1 -functions H1 (X; CL) 2, 35]. Adopting the graded center-formulation we can generalize the notion of superdi erential forms to arbitrary Z 2-graded C -algebras and take it as starting point for the development of a graded di erential calculus. This is the basic idea of the construction of derivation-based di erential calculi 13, 15, 17, 16], transferred to the Z 2-graded case. This idea can even be generalized by taking into account only a subset of all graded derivations, which is at the same time a C -Lie subsuperalgebra as well as a graded submodule over the graded center of the Z 2-graded C -algebra under consideration. We will employ, as it was done in the ungraded \fuzzy case" 39, 37, 24], a speci c variant of this generalization to develop an analogue of the super-deRham complex and the graded Cartan calculus on each of the truncated superspheres. A very natural feature of this approach will be, that the noncommutative body map extends - as in the case of graded manifolds - to a cochain map from the di erential complex on the truncated supersphere to the one on its body. According to the general principles formulated above the rst thing we need to know is the graded center Z g (S ;q ) of the truncated superspheres S ;q ; q 2 N. Analogous to the ungraded case (see for example 45]) we nd, that S ;q is graded-central,

Z g(S ;q ) = Z g(S ;q ) = C IdV

0

( 2 1)

q;

(95)

Consequently one can choose (according to the argumentation above) in principle every C g Lie subsuperalgebra of Derg C (S ;q ) as C -Lie superalgebra of supervector elds V (S ;q ) on each truncated supersphere. But there is a natural choice given by the action of the orthosymplectic Lie superalgebra osp(1j2) on S ;q ,

Vg (S ;q) := ad q (osp(1j2)) ; q 2 N:
(2 )

(96) (97)

The graded representations ad( ) are faithful by (65), such that there are natural isomorphisms
2

Vg (S ;q) = Vg (S ;q0 ) = osp(1j2) ; q; q0 2 N;

of C -Lie superalgebras and this fact will \control the growth of the graded di erential calculus with respect to q ". An additional justi cation of the choice (96) stems from a translation of the sl(2)-module homomorphism property of S ;q : We can de ne a map ~S ;q from Vg (S ;q )0 to the C -Lie 21

algebra V(S ;q ) := ad( ) (sl(2)) = sl(2) of (complex) vector elds on the truncated sphere S ;q 39, 37] via ~S ;q (D) S ;q (f ) := S ;q (Df ) ; f 2 S ;q ; (98) analogous to (21). Then ~S ;q is a surjective (in fact a bijective) Lie algebra homomorphism, q which maps ad( ) J 2 Vg (S ;q )0 to
2 2

~S ;q (ad( q )J ) = ad( q ) J ;
2 2

(99)

where J 2 sl(2) corresponds to J 2 osp(1j2)0 via the natural isomorphism. g For every natural number p 2 N let us denote by Homp 2-graded C -vector C (V (S ;q ); S ;q) the Z g g p space of all p-linear maps V (S ;q ) : : : V (S ;q ) ?! S ;q and by Sp the symmetric group on p letters. Introducing the commutation factor p : Sp Z : p: : Z2 ?! f 1g via 2
p(

; i1 ;

; ip) :=

r;s=1; ;p;r<s ?1 (r)> ?1 (s)

(?1)ir is ;

(100)

g we can de ne a representation of Sp on Homp C (V (S ;q ); S ;q) by

( ! ) (D1;

; Dp) := p( ; D1;

; Dp )!(D

(1)

;D

( )

p)

(101)

g ; Dp 2 Vg (S ;q ) 50]. for all ! 2 Homp C (V (S ;q ); S ;q ); 2 Sp and all homogeneous D1; g Now by de nition a p-linear map ! 2 Homp C (V (S ;q ); S ;q) is called graded-alternating if

! = sgn !

(102)

is ful lled for all 2 Sp and, according to the discussion at the beginning of the section, we should interpret them as p-superforms on the truncated supersphere S ;q . The set of all p-superforms on the truncated supersphere S ;q forms a graded vector subspace g g;p(S ;q ). of Homp C (V (S ;q ); S ;q ) and it will be denoted by A general superform on the truncated supersphere S ;q is an element of the direct sum
g (S ;q ) :=

p2N0

g;p (S ;q ) ;

(103)

where we set g;0(S ;q ) := S ;q . Employing the multiplicative structure of S ;q we can proceed exactly as in the case of graded manifolds 2, 35] (respectively graded Lie-Cartan pairs 27, 28, 46]) to introduce a graded wedge product on g (S ;q ). So we de ne rst for all p; p0 2 g;p0 (S ;q ) 0 ?! g;p+p0 (S ;q ) 0 by g;p (S ;q ) N0; i; i0 2 Z 2 a bilinear map ^ : i i i+i

?! ^ !0 (D ; ; D 0 ) := 1 X sgn p p0 ( ; D ; ; D p p0 ) p p p!p0! 2Sp p0 Pp (?1)i0 l D l ! (D ; ; D p )! 0(D p ; ; D

1 + + 1 +
+ =1 ( )

(104)
( +

(1)

( )

( +1)

p p0 ) )

for all homogeneous D1; ; Dp+p0 2 Vg (S ;q ) and extend these by bilinearity to g (S ;q ). With respect to it g (S ;q ) becomes a N0 Z 2-bigraded C -algebra. 22

Having built up the algebra of superforms on each truncated supersphere we can introduce the graded Cartan calculus as one does it for ordinary graded manifolds 35, 2]. As far as we are only interested in the linear structure of g (S ;q ) we are doing nothing else than Lie superalgebra cohomology of Vg (S ;q ) with values in S ;q 18, 52]. So let us follow the excellent article 52] and apply it to our case. One rst extends the action of the C -Lie superalgebra Vg (S ;q ) to g (S ;q ): For every homogeneous D 2 Vg (S ;q ) one introduces a C -linear map LD : g (S ;q ) ?! g (S ;q ) by de ning its action on bihomogeneous ! 2 g;p (S ;q ); p 2 N0; according to (LD ! ) (D1;

; Dp) := D (!(D1;
p X l=1 Vg (S ;q).

; Dp)) ?
( +

? (?1)D !

Pl?

(105)

l0 =1 Dl0 )! (D1;
1

; D; Dl]g ;

; Dp )

for all homogeneous D1; ; Dp 2 LD is a bihomogeneous endomorphism of the g bigraded C -vector space (S ;q ) of bidegree (0; D). A general graded derivation D 2 Vg (S ;q ) can be uniquely decomposed into its homogeneous components Di ; i = 0; 1; and LD := LD + LD 2 EndC ( g (S ;q )) is well-de ned. Moreover, the map L : Vg (S ;q) ?! EndC ( g (S ;q )), given by D 7! LD , is a graded representation of the C -Lie superalgebra Vg (S ;q ) on g (S ;q ). Besides this extension of Vg (S ;q ) it is useful to introduce for all D 2 Vg (S ;q ) a C -linear map {D : g (S ;q ) ?! g (S ;q ) by {D f := 0 (106) for all f 2 S ;q and by ({D ! ) (D2; ; Dp) := ! (D; D2; ; Dp) (107) for all ! 2 g;p(S ;q ); p 2 N and D2; ; Dp 2 Vg (S ;q ). For every D 2 Vg (S ;q ) it is a N0homogeneous of degree ?1 and the map { : Vg (S ;q ) ?! EndC ( g (S ;q )); D 7! {D is C -linear and Z2-even. In addition the relations
0 1

{D {D0 + (?1)DD0 {D0 {D = 0

and

(108)

(109) (LD {D0 ? {D0 LD ) ! = (?1)D! { D;D0 ]g ! are ful lled for all homogeneous D; D0 2 Vg (S ;q ) and all Z2-homogeneous ! 2 g (S ;q ). The Lie superalgebra cochain map is most elegantly introduced as C -linear map d : g (S ;q ) ?! g (S ;q ), whose action on Z 2-homogeneous 0-superforms f 2 S ;q is given by

df (D) := (?1)fD Df

(110)

for all homogeneous D 2 Vg (S ;q ) and whose action on Z 2-homogeneous p-superforms ! 2 g;p(S ;q ); p 2 N; is de ned inductively via

{D (d!) := (?1)D! LD ! ? d ({D !)

(111)

for all homogeneous D 2 Vg (S ;q ). Then d is a bihomogeneous endomorphism of bidegree (1; 0) ful lling the cochain condition d d=0 (112) 23

and commuting with LD for all D 2 Vg (S ;q ). Explicitly one nds

d!(D0;
+
0

; Dp ) =

p X

X
l<l0 p

l=0 Pl0?1 0 (?1)l +Dl0 l00 =l+1 Dl00 ! (D

(?1)l+Dl (!+

Pl?

l0 =0 Dl0 ) LDl
1

!(D0;

; Dl;

; Dp ) + ; Dl0 ;
_

(113)

; Dl?1; Dl; Dl0 ]g ;

; Dp )

for homogeneous D0; D1; ; Dp 2 Vg (S ;q ) and bihomogeneous ! 2 g (S ;q ) of bidegree (p; !) (_ denotes omission). The preceeding discussion can be summarized as follows: As far as one considers only the linear structure of g (S ;q ) the endomorphisms LD ; {D and d ful ll exactly the same relations as Lie derivative, inner product and exterior derivative in the case of graded manifolds. The latter observation stays also true if one considers the graded wedge product (104) on g (S ;q ).

bidegree (p; !) and (p0; ! 0), respectively.

Proposition 5 The relations ? LD ! ^ ! 0 = (LD !) ^ ! 0 + (?1)D! ! ^ LD ! 0 ? {D ! ^ ! 0 = (?1)D!0 ({D !) ^ ! 0 + (?1)p! ^ {D !0 ? d ! ^ ! 0 = (d!) ^ ! 0 + (?1)p! ^ d! 0 are ful lled for all homogeneous D 2 Vg (S ;q ) and all bihomogeneous !; ! 0 2

(114)
g (S ;q ) of

Proof : This can be shown exactly as in the case of graded manifolds. That is, one starts with a direct proof of the second relation and proofs the other equations inductively using the relations (109) and (111). 2

Because of the analogy to the case of graded manifolds (see 35]) we will call the endomorphisms LD ; {D and d of g (S ;q) Lie derivative and interior product (with respect to the supervector eld D 2 Vg (S ;q )) as well as exterior derivative. g;p (S ;q ); p 2 N0, of p-superforms With respect to the product ^ the Z 2-graded C -vector space g;0(S ;q ) S ;q this Z forms a Z 2-graded S ;q -bimodule. In the special case 2-graded (left as well as right) module is graded-free with homogeneous basis fIdV ( q ;1)g, of course. In g;p (S ;q ); p 2 N, a little bit closer, order to investigate the other Z 2-graded S ;q -bimodules let f Ek 2 osp(1j2)0; E 2 osp(1j2)1 j k = 1; 2; 3; = 4; 5 g be some homogeneous basis of osp(1j2) and q (115) @q;A := ad( )EA ; A = 1; ; 5; A the elements the elements of the corresponding homogeneous basis of Vg (S ;q ). Denoting by q g;1(S ;q ); A = of the dual basis to f@q;A g we can introduce homogeneous 1-superforms A q 2 1; ; 5; via A (D) := A (D)Id q (116) q q V ( ;1) for all D 2 Vg (S ;q ). By applying both sides on supervector elds we nd
2 2 2

f^

A1 ^ q

A^ B q q p ^ A q

A B A = ?(?1) q q B q ^ q P p Al = (?1)f l q A q ^
=1 1

Ap ^ f q

(117)

24

for all homogeneous f 2 S ;q and all A; B; A1;

A= q

where cA BC denote the osp(1j2)-structure constants corresponding with the basis fEA g. Introducing the graded vector subspace g;p Z g (S ;q ); p 2 N, of all p-superforms with values in the g graded center Z (S ;q ) of S ;q , we can conclude from (117) (or from the usual isomorphisms between graded-alternating maps and the graded exterior algebra 2, 50, 51, 52]), that

5 1 X A C B 2 B;C =1 cBC q ^ q ;

; Ap = 1;

; 5; p 2 N, as well as
(118)

A1 ^ q

Ap q

j (A ; ; Ap) 2 Ip j
1 1

(3 2)

(119) (120)

with

Ip j := (A ; ; Ap0 ; Ap0 ; ; Ap) j 0 p0 p; A ; ; Ap0 = 1; 2; 3;

(3 2) 1

Ap0+1 ;

; Ap = 4; 5; A1 < A2 <

+1

< Ap0 < Ap0+1

Ap?1 Ap

g;p(S ;q ), forms a homogeneous basis of g;p Z g (S ;q ). Moreover the bigraded subalgebra of g (S ) := M g;p (S ) (121) Z g ;q Z g ;q p2N0
0 g with g; Z g (S ;q ) := Z (S ;q ), is stable under the Lie derivative, the interior product as well as under exterior di erentiation according to (118). >From the preceding discussion we can conclude in particular, that the Z2-graded S ;q -modules g;p(S ;q ); p 2 N; are graded-free (as left and as right modules) and a homogeneous basis is given by (119). Consequently every ! 2 g;p(S ;q ) can be written as

!=

( 1
1

and the unique coe cients !A Ap 2 S ;q are given by !A Ap = (?1) p00(p00?1) Q5 1 !(@q;A ; ; @q;Ap ) ; (123) A=1 NA ! where p00 is the number of entries in (A1; ; Ap) greater than 3 and NA is the number of entries in (A1; ; Ap) being equal A. Similar to the case of matrix geometry ( 13, 37]) it is possible to introduce via
1 1 2 1

A ; ;Ap )2Ip

(3 2)

!A

A Ap ^ q 1 ^

Ap q

(122)

q :=

X
5

an even 1-superform on each truncated supersphere S ;q ; q 2 N, which is invariant and ful lls a super-version of the Maurer-Cartan equation.

A=1

( ) ^ EA
2

A q

(124)

Proposition 6 The de nition of q is independent of the choice of the homogeneous basis of osp(1j2). q is invariant and up to complex multiples it is the only invariant 1-superform on S ;q; q 2 N. Moreover, its exterior di erential ful lls d q= q^ q (125)
25

and the exterior di erential of each f 2 g;0 (S ;q ) can be written according to

df =

q ; f ]g

q ^f ?f ^ q:

(126)

Proof : Beside the uniqueness statement only simple calculations are involved. In order to see, 1 that q is the only invariant 1-superform on S ;q it is important to note, that g; Z g (S ;q ) is an irreducible, Z2-graded osp(1j2)-module with highest weight 1 and that the Z2-graded osp(1j2)module g;1(S ;q ) is isomorphic to the tensor product of the Z2-graded osp(1j2)-modules S ;q 1 and g; Z g (S ;q ). Then the uniqueness of q follows from proposition 2 and the \Clebsch-Gordan decomposition" of tensor products of (irreducible) Z2-graded osp(1j2)-modules 4]. 2

If X is a H1 -deWitt supermanifold with body X the body map X extends to an algebra homomorphism and cochain map from the super-deRham complex of X to the ordinary deRham complex on the body manifold X 35]. Because the noncommutative body map S ;q : S ;q ?! S ;q ; q 2 N, is no algebra homomorphism, some extension from the algebra of superforms on the truncated supersphere to the algebra of forms on the truncated sphere cannot be an algebra homomorphism. But by translating the construction of H1 -deWitt supermanifolds (respectively graded manifolds) we will introduce a cochain map, which we can interpret as noncommutative analogue to the extension of X in the graded-commutative setting. Let us give rst of all de nition and basic results of the Cartan calculus (see 13, 39, 37]) on the truncated sphere S ;q ; q 2 N, as far as they are relevant for the subsequent discussion of the body map of superforms on the truncated supersphere S ;q . For every p 2 N a p-form on the truncated sphere S ;q is a p-linear, alternating map V(S ;q ) : p: : V(S ;q ) ?! S ;q and we denote by p(S ;q ) the C -vector space of all p-forms. A general form on the truncated sphere S ;q is an element of the direct sum (S ;q ) :=

p2N0

p (S

;q )

(127)

with 0 (S ;q ) := S ;q . (S ;q ) becomes a N0-graded C -algebra, if one introduces for all p; p0 2 p0 (S ;q ) ?! p+p0 (S ;q ) by N0 a bilinear map ^ : p (S ;q ) X ?! ^ ^! ^ 0 (D1; ; Dp+p0 ) := p!1 ^ (D (1); ; D (p))^ ! 0(D (p+1); ; D (p+p0)) 0 p ! 2Sp p0 sgn ! (128) for all D1; ; Dp+p0 2 V(S ;q ), and extends these by bilinearity. Analogous to the graded case the set p Z (S ;q ); p 2 N, of p-forms with values in the center Z (S ;q ) = C IdV( q ) of S ;q forms a 0 vector subspace of p(S ;q ) and Z (S ;q ) := p2N p Z (S ;q ); Z (S ;q ) := Z (S ;q ), is a graded subalgebra of (S ;q ). Exterior di erential d, Lie derivative LD as well as interior product D (with respect to a vector eld D 2 V(S ;q )) are de ned exactly as in the case of the truncated supersphere, if one views V(S ;q ) and (S ;q ) as trivially Z2-graded. ^q;k jk = 1; 2; 3g of V(S ;q ) as For some basis fEk jk = 1; 2; 3g of sl(2) one can introduce a basis f@
+ 2 0

26

1 well as 1-forms ^ k q 2 Z (S ;q ) analogous to (115) and (116). The latter 1-forms anticommute and one nds, that the set of p-forms

n ^k
1

kp q ^ ^ ^ q j(k ; ; kp) 2 Ip
1

(3)

(129)

(130) forms for all p 2 N (trivially for p > 3, of course) a basis of the (left and right) S ;q -module p (S ;q ) as well as of the C -vector space p (S ;q ). Z (p) g;p(S ;q ) ?! In analogy to the case of graded manifolds 35] we can introduce now maps S ;q : p (S ;q ); p 2 N0; via
(3) 1 1 (p) S ;q (! )

with

Ip := f (k ; ; kp) j k ; ; kp = 1; 2; 3; k <

< kp g

~S ;q (D1);

; ~S ;q (Dp) := S ;q (! (D1 ;

; Dp))

(131)

(0) for all D1; ; Dp 2 Vg (S ;q )0 . In the case p = 0 (131) is to be understood as S ;q = S ;q , of course. For each p 2 N0 it is an even C -linear map and we can de ne its C -linear extension g (S ;q ) ?! (S ;q ), which we again denote by S . ;q

Proposition 7

S ;q is a surjective sl(2)-module homomorphism and cochain map, whose g restriction to Z g (S ;q ) is an algebra homomorphism onto Z (S ;q ).

Proof : Using the sl(2)-module homomorphism property of the body map and the fact, that ~S ;q is a Lie algebra isomorphism one nds immediately

for all D 2 Vg (S ;q )0. The action of the extension of the body map on a p-superform ! 2 g;p(S ;q ) can be described alternatively by
S ;q (! )

d S ;q = S ;q d L ~S ;q (D) S ;q = S ;q LD

(132)

0 B S ;q B @ X

X
j2) A ; ;Ap )2I(3 p
(3)

!A

A Ap ^ q 1 ^
1

1 Ap C C ^ q A=

( 1

( 1

A ; ;Ap )2Ip

S ;q

?!

A1 Ap

Ap ^^A q ^ ^^q ;

(133)

g;1(S ;q ) and ^ k 2 1(S ;q ) correspond to bases in osp(1j2) and sl(2), which are where k q2 q 0 related by the canonical isomorphism. >From (132) we can conclude in particular, that S ;q 2 is surjective and that its restriction to g Z g (S ;q ) is an algebra homomorphism.

The interpretation of the Z 2-graded C -algebras and Z 2-graded osp (1j2)-modules S ;q stems from the fact, that there are suitable osp(1j2)-module homomorphisms q0 q : S ;q ?! S ;q0 ; q; q0 2 N; q q0, constituting a directed system, whose direct limit can be identi ed with the graded subalgebra P (S ) H1 (S ; CL) and which possesses a graded-commutative 27

limit. We want to understand noncommutative p-superforms for p 2 N in a similar way as noncommutative pendants to p-superforms on the (2j2)-dimensional supersphere S . Although we reserve the precise treatment of this question to a subsequent paper we give some indications how this is done. (p) q 0, The rst thing one has to nd are maps q 0q : g;p(S ;q ) ?! g;p(S ;q0 ); q; q 0 2 N; q (p) such that ( g;p(S ;q ); q 0q ); p 2 N0; become directed systems, which are \compatible with the Cartan calculus". A natural choice for these maps (especially in consideration of a \gradedcommutative limit" of superforms) is
(p) q0 q (! )

0 p B B q0q @
( )

X
j2) A ; ;Ap )2I(3 p j

!A

Ap ^ A q1 ^ A01 ^ q

1 p C := ^ A q C A

:=

X
( 1

( 1

A ; ;Ap )2Ip

(3 2)

q0 q

?!

A1 Ap

Ap : q0

(134)

(p) Then ( g;p(S ;q ); q 0q ); p 2 N0; are directed systems of Z 2-graded osp(1j2)-modules, whose direct limits should be connected with the algebra of superforms on the (2j2)-dimensional supersphere. This problem is in fact in exactly the same way \singular" as in the case of the ordinary fuzzy sphere 39, 37]. It is more natural to interpret the elements of the direct limits as superforms on the supergroup UOSP(1j2) (see 3, 4], for example), which is a super bre bundle over the (2j2)-dimensional supersphere.

7 Cohomological considerations
According to a general theorem of 35] the super-deRham cohomology of the (2j2)-dimensional supersphere is isomorphic to the deRham cohomology of the 2-dimensional sphere. The isomorphism is induced by extension of the body map (in the sense described in the preceding chapter) to the algebra of superforms on the (2j2)-dimensional supersphere. As we will see exactly the same is true on every truncated supersphere. The cohomology M p d (135) H (S ;q) H (S ;q ) := ker im d p2N of the complex ( on the truncated
g (S ;q ); d) is our substitute for the (complexi ed) super-deRham cohomology supersphere S ;q ; q 2 N, as well as the cohomology
0

H (S ;q )

p2N0

d H p (S ;q ) := ker im d

(136)

of the complex ( (S ;q ); d) is the noncommutative pendant to the (complexi ed) deRham cohomology of the sphere. By construction H (S ;q ) is the Lie superalgebra cohomology of Vg (S ;q) with coe cients in S ;q, while H (S ;q ) is the Lie algebra cohomology of V(S ;q ) with coe cients in S ;q . 28

The cochain map S ;q induces as usual via

H ( S ;q ) : H (S ;q ) ?! H (S ;q) !] 7! S ;q (!)
a homomorphism of N0-graded C -vector spaces, which turns out to be an isomorphism.

(137)

Proposition 8 H (

S ;q ) is an isomorphism of bigraded C -vector spaces and we have explicitly

H p(S ;q) = H p(S ;q ) =

f0g ;
2

p = 0; 3 p 2 N0 n f0; 3g :

(138)

Proof : Because the (graded) representations ad( ) are faithful, H (S ;q ) is isomorphic to the Lie superalgebra cohomology H (osp(1j2); S ;q) of osp(1j2) with coe cients in S ;q as well as H (S ;q ) is isomorphic to the Lie algebra cohomology H (sl(2); S ;q ) of sl(2) with coe cients in S ;q . Moreover, using the direct sum decompositions (65) and (83), we nd 52]

H p(S ;q ) =
and

N0 j2 1 j 2q

H p(osp(1j2); V ( ) j )
2

H p (S ;q ) =

M
j2N 0 j q

H p(sl(2); V( ) j ) :
2

The standard second-order Casimir operator of osp(1j2), restricted to V ( ) j ; j 6= 0; has a q nonvanishing eigenvalue and there are no nontrivial osp(1j2)-invariant elements in V ( ) j ; j 6= 0; such that we can conclude 52]
2 2

H p(S ;q) = H p (osp(1j2); V ( ) 0 ) = H p (osp(1j2); C ) ;

2

where H (osp(1j2); C ) denotes the Lie superalgebra cohomology of osp(1j2) with trivial coe cients. The same argument leads in the case of the truncated sphere to

H p (S ;q ) = H p (sl(2); C ) :
But these cohomologies with trivial coe cients are known to be given by (138) (see 18, 19] for the case of osp(1j2) and 20] for sl(2)). 2 One should note, that (138) is not exactly the super-deRham cohomology (deRham cohomology) of the (2j2)-dimensional supersphere (2-dimensional sphere). The explicit result is rather a cohomological veri cation of the remarks we made at the end of the preceding chapter about the singular character of the identi cation of the noncommutative \limit superforms" with superforms on the (2j2)-dimensional supersphere.

29

8 Concluding Remarks
In close analogy to the construction of the fuzzy sphere and its derivation-based di erential calculus we have introduced the fuzzy supersphere together with a di erential calculus, which is based on the Lie superalgebra osp(1j2) acting on each of the truncated superspheres via graded derivations. The natural interpretation of the fuzzy sphere as noncommutative body of the fuzzy supersphere guaranteed the existence of the usual relations between H1 -deWitt supermanifolds and their bodies. In particular the noncommutative body projection induced in the same way as in the theory of graded manifolds the isomorphism between the ((graded) derivation-based) cohomologies of the fuzzy supersphere and its body. The present work can be seen as rst step towards the development of the di erential geometry and the formulation of (quantum) eld theoretical models on a non-trivial, (fuzzy) noncommutative supermanifold. From the point of view of pure supergeometry the introduction of metric and supervector bundle concepts as well as the investigation of their \limits" are the next challenging tasks. Passing from mathematics to (quantum) physical model building we will have to undertake a \categorial jump" from Z2-graded algebras of H1 -functions and their noncommutative approximations to larger classes of \functions". In our opinion the most promising program in that direction is a G-extension, graded-commutative and noncommutative as well.

Acknowledgement
The authors would like to thank W.Bulla, C.Klimcik, H.Miglbauer, F.Pauer, P.Presnajder, L.Pittner, A.Strohmaier and W.Tutschke for helpful discussions and the \Fonds zur Forderung der wissenschaftlichen Forschung (FWF)" for support funding.

References
1] A.P.Balachandran, G.Bimonte, E.Ercolessi, G.Landi, F.Lizzi, G.Sparano, P.TeotonioSobrinho. Noncommutative Lattices as Finite Approximations. Jour.Geom.Phys. 18:163, 1996. 2] C.Bartocci, U.Bruzzo, D.Hernandez-Ruiperez. The Geometry of Supermanifolds. Kluwer, 1991. 3] C.Bartocci, U.Bruzzo, G.Landi. Chern-Simons forms on principal super ber bundles. Jour.Math.Phys. 41:45, 1990. 4] F.A.Berezin, V.N.Tolstoy. The Group with Grassmann Structure UOSP(1,2). Comm. Math.Phys. 78:409, 1981. 5] P.Bryant. The Structure of De Witt Supermanifolds. Riv.Mat.Pura&Appl. 3:57, 1988. 6] F.Constantinescu, H.F.de Groote. Geometrische und algebraische Methoden der Physik: Supermannigfaltigkeiten und Virasoro-Algebren. Teubner, 1994. 30

7] A.Connes. Noncommutative Geometry. Academic Press, 1994. 8] J.F.Cornwell. Group Theory in Physics III, Supersymmetries and In nite-Dimensional Algebras. Academic Press, 1989. 9] L.Crane, J.M.Rabin. Global Properties of Supermanifolds. Comm.Math.Phys. 100: 141, 1985. 10] M.Daumens. Super angular momentum and super spherical harmonics. Jour.Math.Phys. 34:2508, 1993. 11] B.S.DeWitt. Supermanifolds. Cambridge, 1992. 12] S.Doplicher, K.Fredenhagen, J.E.Roberts. The Quantum Structure of Spacetime at the Planck Scale and Quantum Fields. Comm.Math.Phys. 172: 187, 1995. 13] M.Dubois-Violette, R.Kerner, J.Madore. Noncommutative di erential geometry of matrix algebras. J.Math.Phys. 31(2):316, 1990. 14] M.Dubois-Violette, R.Kerner, J.Madore. Super Matrix Geometry. Class.Quant.Grav. 8:1077, 1991. 15] M.Dubois-Violette. Derivations et calcul di erentiel non commutatif. C.R.Acad.Sci.Paris I 307:403, 1988. 16] M.Dubois-Violette. Some Aspects of Noncommutative Di erential Geometry. Preprint L.P.T.H.E.-Orsay 95/78, 1995. 17] M.Dubois-Violette, P.W.Michor. Connections on central bimodules in noncommutative di erential geometry. Jour.Geom.Phys. 20:218, 1996. 18] D.B.Fuks. Cohomology of In nite-Dimensional Lie Algebras. Consultants Bureau, 1986. 19] D.B.Fuks, D.A.Leites. Cohomology of Lie Superalgebras. Com.rend.Acad.bulg. 37:1595, 1984. 20] W.Greub, S.Halperin, R.Vanstone. Connections, curvature, and cohomology I-III. Academic Press, 1972/73/76. 21] H.Grosse, C.Klimcik, P.Presnajder. Field Theory on a Supersymmetric Lattice. Comm. Math.Phys. 185:155, 1997. 22] H.Grosse, C.Klimcik, P.Presnajder. Towards Finite Quantum Field Theory in Non-Commutative Geometry. Int.J.Mod.Phys.A. 35:231, 1996 23] H.Grosse, C.Klimcik, P.Presnajder. On Finite 4D Quantum Field Theory In NonCommutative Geometry. Comm.Math.Phys. 180:429, 1996. 24] H.Grosse, P.Presnajder. The Construction of Noncommutative Manifolds Using Coherent States. Lett.Math.Phys. 28:239, 1993. 31

25] H.Grosse, C.Klimcik, P.Presnajder. Topologically Nontrivial Field Con gurations in Noncommutative Geometry. Comm.Math.Phys. 178:507, 1996. 26] E.Hawkins. Quantization of Equivariant Vector Bundles. Preprint q-alg/9708030, 1997. 27] A.Jadczyk, D.Kastler. Graded Lie-Cartan pairs I. Rep.Math.Phys. 25:1, 1987. 28] A.Jadczyk, D.Kastler. Graded Lie-Cartan pairs II. The fermionic di erential calculus. Ann.Phys. 179:169, 1987. 29] V.G.Kac. Lie Superalgebras. Adv.Math. 26:8, 1977. 30] V.G.Kac. Representations of Classical Lie Superalgebras. In: Lecture Notes in Mathematics 676:597, 1978. 31] D.Kastler. Cyclic cohomology within the di erential envelope. Hermann, 1988. 32] W.Kalau, M.Walze. Supersymmetry and noncommutative geometry. Jour.Geom.Phys. 22:77, 1997. 33] R.Kerner. Graded non-commutative geometries. Jour.Geom.Phys. 11:325, 1993. 34] C.Klimcik. Gauge theories on the noncommutative sphere. Preprint hep-th/9710153, 1997. 35] B.Kostant. Graded manifolds, graded Lie theory, and prequantization. In: Lecture Notes in Mathematics 570:177, 1977. 36] D.A.Leites. Introduction to the Theory of Supermanifolds. Russ.Math.Surv. 35:1, 1980. 37] J.Madore. An Introduction to Noncommutative Di erential Geometry and its Physical Applications. Cambridge University Press, 1995. 38] J.Madore. Fuzzy Physics. Ann.Phys. 219:187, 1992. 39] J.Madore. The fuzzy sphere. Class.Quant.Grav. 9:69, 1992. 40] Y.I.Manin. Quantum Groups and Non-commutative Geometry. Les Publications du Centre de Recherche Mathematiques, Montreal, 1988. 41] L.Mezincescu. Wigner-Eckart theorem for tensor operators of graded Lie algebras. Jour.Math.Phys. 18:453, 1977. 42] W.Nahm, V.Rittenberg, M.Scheunert. Graded Lie algebras: Generalization of Hermitian representations. Jour.Math.Phys. 18:146, 1977. 43] W.Nahm, V.Rittenberg, M.Scheunert. Irreducible representations of the osp(2,1) and spl(2,1) graded Lie algebras. Jour.Math.Phys. 18:155, 1977. 44] I.Penkov, V.Serganova. Generic Irreducible Representations of Finite-Dimensional LieSuperalgebras. Int.Jour.Math. 5:389, 1994. 32

45] R.S.Pierce. Associative Algebras. Springer, 1982. 46] L.Pittner. Algebraic Foundations of Non-Commutative Di erential Geometry and Quantum Groups. Springer, 1996. 47] A.Rogers. A global theory of supermanifolds. Jour.Math.Phys. 21:1352, 1980. 48] A.Rogers. Graded Manifolds, Supermanifolds and In nite-Dimensional Grassmann Algebras. Comm.Math.Phys. 105:375, 1986. 49] M.Scheunert. The Theory of Lie Superalgebras. Lecture Notes in Mathematics 716, 1979. 50] M.Scheunert. Graded tensor calculus. Jour.Math.Phys. 24:2658, 1983. 51] M.Scheunert. Invariant supersymmetric multilinear forms and the Casimir elements of P-type Lie superalgebras. Jour.Math.Phys. 28:1180, 1987. 52] M.Scheunert, R.B.Zhang. Cohomology of Lie superalgebras and of their generalizations. J.Math.Phys. 39:5024, 1998. 53] P.Teo latto. Discrete supergroups and super Riemann surfaces. Jour.Math.Phys. 29:2389, 1988. 54] F.Treves. Topological Vector Spaces, Distributions, and Kernels. Academic Press, 1967. 55] G.-J.Zeng. On tensor operators of the Lie superalgebra OSP(1,2). Jour.Phys.A: Math. Gen. 20:5425, 1987.

33