Super-Cliord Gravity, Higher Spins, Generalized Supergeometry and much more

Carlos Castro August 2013

Center for Theoretical Studies of Physical Systems, Clark Atlanta University, Atlanta; perelmanc@hotmail.com
Abstract

An extended Orthogonal-Symplectic Cliord Algebraic formalism is developed which allows the novel construction of a graded Cliord gauge eld theory of gravity. It has a direct relationship to higher spin gauge elds, bimetric gravity, antisymmetric metrics and biconnections. In one particular case it allows a plausible mechanism to cancel the cosmological constant contribution to the action. The possibility of embedding these Orthogonal-Symplectic Cliord algebras into an innite dimensional algebra, coined Super-Cliord Algebra is described. Finally, some physical applications of the geometry of Super-Cliord spaces to Generalized Supergeometries, Double Field Theories, U -duality, 11D supergravity, M -theory, and E7 , E8 , E11 algebras are outlined. Keywords : Super-Cliord algebras; orthogonal Cliord algebras; symplectic Cliord algebras; supersymmetry; Higher spins; Bimetric gravity; Biconnections; Antisymmetric metrics; Cosmological constant; Super Cliord spaces; Generalized Super Geometry; Exceptional algebras. PACS :

Orthogonal-Symplectic Cliord Algebras

In the past years it has become more clear that extensions and/or generalizations of supersymmetry are needed. Cliord algebras are essential ingredients to attain such goals. A classication of Hermitian versus holomorphic complex and quaternionic generalized supersymmetries of M -theory was attained in [12], including the 12-dim Euclidean generalized supersymmetric F algebra and the 11-dim M theory superalgebra. 1

Polyvector super Poincare algebras like { Q , Q } =

k (k) (C 1 2 ....k ) W 1 2 .....k

(1)

were studied by [13]. The summation over k must obey certain crucial restrictions to match the degrees of freedom with the terms in the left hand side and to ensure that there is symmetry under the exchange of spinorial , indices. C is the charge conjugation (k) are the polyvector-valued momentum like generators. Polyvector matrix and W 1 2 .....k valued extensions of supersymmetry in Cliord Spaces involving spinor-tensorial super1 2 ....n charge generators Q and momentum polyvectors P1 2 ....n were analyzed in [15], [16]. Cliord-Superspace is an extension of Cliord-space and whose symmetry transformations are generalized polyvector-valued supersymmetries. The superconformal algebra su(2, 2|1) in 4D can be realized in terms of 5 5 matrices [1] and whose entries are given explicitly in terms of the gamma matrices a . The momentum, conformal boost, Lorentz and dilation generators are realized as the entries of the 4 4 matrices embedded into 5 5 matrices by setting all the entries of the 5-th column and 5-th rows to zero while identifying the entries of the 4 4 matrices by 1 a (1 5 ) , (Ka ) = 2 1 a (1 + 5 ) , a, b = 1, 2, 3, 4. (2a) 2

(Pa ) =

The Lorentz and dilation generator are 1 1 1 ab = [a , b ] , (D) = (5 ) , , = 1, 2, 3, 4. 2 4 2

(Jab ) =

(2b)

i The axial charge A generator is represented by 4 times a diagonal 5 5 matrix whose entries are (1, 1, 1, 1, 4). The 8 fermionic generators Q , S are represented by 5 5 matrices with zeros along the 4 4 block matrices and whose only nonzero entries are along the 5-th column and 5-th rows as follows

(Q )5 =

1 1 (1 5 ) , (Q ) 5 = [ (1 + 5 ) C ] , (Q )55 = 0 2 2

(3a)

(S )5 =

1 1 (1 + 5 ) , (S ) 5 = [ (1 5 ) C ] , (S )55 = 0 2 2

(3b)

where C is the charge conjugation matrix obeying C = C 1 = C T , Ca C 1 = (a )T where T denotes the transpose. In the representation chosen above C = 0 . The authors [1] have shown explicitly that the above realization of the bosonic and fermionic generators in terms of gamma matrices obey the superconformal su(2, 2|1) algebra graded commutator relations. More recently, a 6 6 matrix realization of the osp(1|4) superalgebra was 2

provide by [5]. The noncompact symplectic algebra sp(2, 2) is isomorphic to so(3, 2). The (Anti) de Sitter group is so(3, 2), so(4, 1) respectively. A (4 + N ) (4 + N ) matrix realization of the N -extended superconformal algebra su(2, 2|N ) algebra (whose even part is su(2, 2) u(N )) is also possible. In particular, a i (4 + N ) (4+ N ) matrix realization of the N spinorial supercharges Qi , S ; i = 1, 2, ...., N can be given by a generalization of eqs-(3a, 3b)
AB (Qi = )

1 1 BA 4+i (1 5 )B 4+i A , (Qi = [ (1 + 5 ) C ]B ) 2 2

(4a)

i AB ) = (S

1 1 4+i A i BA 4+i ) = , (S (1 + 5 )B [ (1 5 ) C ]B 2 2

(4b)

the other matrix components are zero, for instance

4+j (Qi ) 4+j

= 0,

i 4+j (S )

4+j

= 0, i, j = 1, 2, ...N

(4c)

when i, j = 1 one recovers the 5 5 matrix realization of eqs-(3a, 3b). We will go beyond this ordinary description of Lie superalgebras, like su(2, 2|1), in terms of the gamma matrices as displayed above, by incorporating both orthogonal and symplectic Cliord algebras into the framework of super Cliord algebras and which dif f ers from the notion of super Cliord algebras studied earlier by [3]. Orthogonal Cliord algebras are well known. What is less known is the notion of symplectic Clifford algebras [2]. A Cliord analysis approach to Superspace based on orthogonal and symplectic Cliord algebras was undertaken by [6]. A theory of quantized elds based on orthogonal and symplectic Cliord Algebras can be found in [8]. Super Cliord algebras with a Z4 grading and generalized Cliord algebras, orthogonal and symplectic were constructed by [3], [4]. We shall take a quite dierent approach and extend further the work of these authors. To our knowledge the results of this work are new. We begin by introducing the ordinary orthogonal Cliord algebra generators a , a = 1, 2, 3, ...m, and the symplectic Cliord algebra generators i , i = 1, 2, 3, ....., 2n obeying the graded commutation relations 1 { i , j } = ij = ji , [ i , j ] = ij = ji ; i, j = 1, 2, 3, ..., 2n 2 2k1 2k1 1
2k2

(5a) (5b) (5c)

= 2k1 1 = 2k2

2k2 1 2k1 1

= 0 = k1 k2

2k2

ij is a 2n 2n antisymmetric matrix consisting of diagonal blocks of 2 2 antisymmetric 1 matrices whose nonzero entries are 1. The relation 2 { i , j } = ij = ji and its implications was not considered by [6], [7]. The other commutators are 1 1 { a , b } = gab 1; [ a , b ] = ab = ba , a, b = 1, 2, 3, ...., m 2 2 3

(6)

[ a , bc ] = 2 gab c 2 gac b [ ab , cd ] = 2 gac bd + 2 gad bc 2 gbd ac + 2 gbc ad [ a , i ] = 0, [ ab , i ] = 0, [ a , ij ] = 0, [ ab , ij ] = 0 [ i , jk ] = ij k + ik j [ ij , kl ] = ik jl + il jk + jk il + jl ik

(7) (8) (9) (10) (11)

We take all the generators of the orthogonal Cliord algebra to be of even grade, while the i generator has odd grade and ij has even grade. Afterwards we shall study the case where all the odd-rank generators of the orthogonal Cliord algebra have an odd grade; and all the even-rank generators have even grade. For instance, if the grade of a is chosen to be odd, one must replace [a , i ] = 0 with {a , i } = 0 in eq-(9). The graded commutator is dened as [ A, B } = A B (1)(grade
A grade B )

BA

(12)

where the grade of the even part of the superalgebra is 0, and the grade of the odd part of the superalgebra is 1. Denoting the grades of A, B, C respectively by a, b, c, the graded Jacobi identities are given by (1)ac [ A, [ B, C } } + (1)ba [ B, [ C, A } } + (1)cb [ C, [ A, B } } = 0 (13) the above graded Jacobi identity can be also rewritten in terms of the superalgebra of derivations as [ A, [ B, C } } = [ [ A, B }, C } + (1)ab [ B, [ A, C } } (14)

In the Appendix we show explicitly that the graded Jacobi identities corresponding to the superalgebra are satised. For simplicity let us take at the moment m = 2 and 2n = 2 and dene
i ij ab A = A + Aa a + A ab + A i + A ij ; a, b = 1, 2; i, j = 1, 2

(15)

since i belongs to the odd part of the algebra, the gauge eld component Ai is taken to be anticommuting; i.e. it is Grassmannian odd, an a-number. The other eld components are 1 {i , j } = ij = ji , commuting; i.e. they are Grassmannian even, a c-number. Because 2 ab ij ji the latter belongs to the even part of the algebra. Also we have A = Aba and A = A .

1 The eld strength is dened as F = dA + 2 A A. One should note that due to the i j j i anticommuting nature of the gauge eld component Ai one has A A = A A so that wedge product

1 2

j j i Ai A i j A A j i

dx dx =

1 i j A A { i , j } dx dx = 2 (16)

j Ai A ij dx dx

involves the anticommutator {i , j } rather than the commutator [i , j ] . Hence, the eld strength component associated to the ij generator is given by
kj ij ij i F = Aij A + A k[ A ] +

1 j j i ij ji (Ai A + A A ), A = A 2

(17)

such that
ij ij ij ji F = F , F = F j i j due to the Grassmannian odd character Ai A = A A . The other components are ab ab a cb F = Aab A + 2 A c[ A ] +

(18)

1 a b ba A[ A ] , Aab = A 2

(19) (20) (21) (22)

j i i i F = Ai A + A j [ A ] a a a b F = A a A + 2 A b[ A ]

F = A A

where the [ ] denotes antisymmetrization of the indices with unit weight. A Yang-Mills like invariant Lagrangian is quadratic in the eld strengths < F F > where the bracket < > denotes extracting the unit element of the superalgebra in the product of two generators as follows < a b > = gab , < ab cd > = gac gbd gbc gad , < i j > = < a cd > = 0 (23a) (23b) (23c)

1 ij , < i jk > = 0 2 1 < ij kl > = ( ik jl + il jk ) 4 To obtain eqs-(23) one needs the initial (anti) commutators and the relations { ab , cd } = 2 abcd 2 gac gbd 2 gbc gad , { ij , k } = ij,k , { ij , kl } = ij,kl + 5 {ab , c } = 2 abc

(24) (25)

1 ( ik jl + il jk ) 2

[kl,j , i ] = ij,kl , { kl,j , i } = ki lj + li kj + ji kl obtained from the identities [A, BC ] = [A, B ] C + B [A, C ], {A, BC } = {A, B } C B [A, C ] One must introduce new generators in eqs-(25,26) obeying ij,k = ji,k = k,ij = k,ji ij,kl = kl,ij = ji,kl = ij,lk = ji,lk = lk,ji

(26)

(27)

(28a) (28b)

and as a result one may enlarge the number of terms in the decomposition of the eld A in eq-(15), using these new generators ij,k and ij,kl . The only caveat is that the super algebra will not close because the graded commutators [ ij,kl , mn,p } = [ ij,kl , mn,p ] im jn,klp + ....... [ ij,kl , mn,pq } = [ ij,kl , mn,pq ] im jn,klpq + ....... (29) (30)

furnish the new generators jn,klp , jn,klpq which were not included . The other graded commutators are ne because no new generators are introduced in [kl,j , i } = {kl,j , i } ki lj + ...., [ij,k , lm ] il jm,k + ......... [ij,kl , mn ] im jn,kl + ....., {ij,k , lm,n } il jm,kn + ..... (31)

ij,kl , one will have another ij,k and excluding Aij,kl However, by just including Aij,k problem due to the last anticommutator relation {ij,k , lm,n } il jm,kn which will preclude the exclusion of jm,kn . Therefore one will be forced to introduce an inf inite number of generators beyond i and ij . The new eld Aij,k , for example, is Grassmannian ij ij odd, it is an a-number. Its addition will also modify the expression for the terms in F due to the anticommutator contribution of the rst terms of eq(31). In the orthogonal Cliord algebra Cl(m) case the introduction of new generators stops when the number of dierent factors in the antisymmetric product of the gammas equals to m, the dimension of spacetime; i.e the rank of an antisymmetric tensor cannot exceed the spacetime dimension. Concluding, by just keeping i , ij one can explicitly verify that the super algebra closes with respect to the graded commutation relations and the graded Jacobi identities are obeyed as shown in the Appendix. In this case we have the orthogonal Cliord algebra generators 1, a , ab and the i , ij generators associated with the symplectic Cliord algebra. In the most general case when m > 2, 2n > 2 we have a super-Cliord valued gauge eld associated with the ortho-symplectic Cliord algebras of the form
a1 a2 A = A + Aa a1 a2 + Aa1 a2 a3 a1 a2 a3 + ..... + Aa1 a2 a3 ...am a1 a2 a3 ....am + a + A i1 i2 1 i2 ,j1 1 i2 ,j1 j2 Ai i1 i2 + Ai i1 i2 ,j1 + Ai i1 i2 ,j1 j2 + ...... i + A

(32)

The range of indices is a1 , a2 , a3 , .... = 1, 2, 3, ..., m and i1 , i2 , ....j1 , j2 ... = 1, 2, 3, ...., 2n. The (anti) commutators of the a , a1 a2 , ....., a1 a2 .....am generators can be found in the monograph [23]. The (anti) commutators of the innite number of s generators would require a computer algebra package. One expects to have analogous formulae to those in eqs-(5,10,11,25,26,29,30,31). The innite number of terms in (32) resembles the expansion based on the frame-like formalism, developed by [17], [18], describing the higher spin elds dynamics in terms of higher spin connection gauge elds that generalize objects such as vielbeins and spin connections in gravity. Constructing consistent gauge theories of interacting higher spin elds is a dicult and unsolved problem. Higher spin elds [18] have been conjectured to be part of the spectrum of tensionless strings. Higher conformal spin eld currents generate the W (super) algebras which are the higher spin extensions of the (super) Virasoro algebra in 2D. The analog of photons in the Extended Relativity theory in C -spaces (Cliord spaces) correspond to tensionless strings and branes [11]. The quantum Virasoro algebra generators can be constructed in terms of operators of the generalized Cliord algebras as shown by [24]. For these reasons the gauge eld theory proposed in this work deserves further scrutiny. For simplicity purposes and without loss of generality we shall retain the original decomposition of the eld A displayed in eq-(15). To sum up, from eqs-(17-22) one has for the quadratic Yang-Mills like invariant Lagrangian the following L = 1 1 i ij j kl ij F F F g g + ik jl F g g + 2 2 a ab F Fa 2 F Fab + F F
i j i ij F F g g = F Fi

(33) (34) (35)

One can rewrite and

ij kl ij ik jl F F g g = F Fij

the change in sign in (35) is due to the subtle ordering of indices in the contraction ik F kl jl = ik F kl lj = Fij (36)

the index contraction is performed among adjacent indices. One should note that i i ij F F j = 0 due to the Grassmannian odd character of F . For this reason one must contract the i, j indices with ij = ji . The contraction of the vector indices in F a F b requires the metric gab and the contraction of bivector indices in F ab F cd requires gac gbd gbc gad as displayed in eq-(23a). The Lagrangian (33) is invariant under the super gauge transformations involving the super Cliord algebra valued gauge parameter = + a a + ab ab + i i + ij ij ; ab = ba , ij = ji A = + [A , }, F = [F , } 7 (37) (38)

In component form one has

i ij j i i ij j i Ai = + 2 A j + 2 A j , F = 2 F j + 2 F j a ab b a a ab b a Aa = + 4 A b + 4 A b , F = 4 F b + 4 F b ab b bc a a b b a Aab + 4 Aac = c 4 A c + A A b a bc ac ab a b F ca + F cb 4 F = 4 F F j i j j i i jk ij + 2 Aik Aij k + 2 A k + A + A = j i jk ik ij i j + F ki + F kj + 2 F = 2 F F

(39) (40) (41) (42) (43) (44) (45)

A = , F = 0

In Appendix B we show explicitly that the Lagrangian (33) is invariant under the graded gauge transformations (39-45).

Bimetric Gravity, Biconnection, Antisymmetric Metrics and Cosmological Constant

In this section we discuss the physical implications of the eld theory constructed above. a b a If Aa e , is identied with the vielbein then the contraction e e gab = g yields the i spacetime metric. The partner Ai e is Grassmannian odd so that the contraction
j j i j i h = ei e ij = e e ij = e e ji = h

(47)

furnishes another symmetric rank two tensor h = h , that can be interpreted as another metric tensor typical of bi-metric gravity [10]. The actual contraction should be performed j as ei ij e = h so the indices are adjacent. ab ij ij The spin connection Aab has for partner A which is also Grassmannian ab ij even. The pair of connections , can be used in formulations of bi-connection extensions of ordinary gravity [9] in the same fashion that g , h are the elements in bi-metric gravity [10]. a ab ab ab When Aa e , A , one has that F dx dx coincides with the 2-form ab a b ab a (R +e[ e ] ) dx dx where R dx dx is the curvature 2-form. F dx dx becomes the Torsion 2-form. The Macdowell-Mansouri-Chamsedinne-West (MMCW) Lagrangian in 4D is given by LM M CW = (Rab + ea eb ) (Rcd + ec ed ) which in component form becomes
ab b cd c d (R + ea [ e ] ) (R + e[ e ] ) abcd

(48)

The R R terms are the topological Gauss-Bonnet curvature squared terms. R e e is the Einstein-Hilbert term and e e e e is the cosmological constant term (after one introduces a proper length scale to match units). Since the connection Aa has units of a (length)1 the proper correspondence is Aa e /L so there is an overall factor L4 in g d4 x in the action. The length scale L can front of the spacetime volume term L4 be set to the Planck scale. i j j i i j) is In 4D the following contraction involving e( e = e e + e e ; ij and trivially zero
i j ) (k l) L4 ik jl e( e e e

= 2L4

( h h + h h ) = 0

(49)

due to the antisymmetry of the epsilon tensor density. While the actual contraction involving the metric tensor in the Yang-Mills terms of eq-(35) is nonvanishing
i j ) (k l) L4 ik jl e( g = 2L4 g g ( h h + h h ) = 0 e e e g

(50)

the latter nonvanishing contraction has the form of mass-like terms for the symmetric rank 2 tensor h
2 Lmass = 2L4 ( h h + (h h ) ) > 0, h = trace (h ) = g

(51)

Since Lmass > 0 is positive denite these terms cannot be used to cancel out the cosmological constant contribution L4
abcd b c d ea e e e

= L4 |det e| = L4

|g |

(52)

There is another possibility worth exploring if one takes the grade of a to be odd a while the grade of ab still remains even. In this case we will have that Aa e is now b b a Grassmannian odd (instead of even), and a-number so that ea e = e e and the a b contraction e e gab = g[ ] yields an antisymmetric metric tensor g[ ] . In this case, a j symmetric tensor is only obtained from the contraction in eq-(47) h = ei e ij = h . If we were to identify h with a spacetime metric g( ) , in this case the emergent metric h can be seen as the condensate of two Grassmannian odd valued elds h = j ei e ij = h . A symmetric and antisymmetric metric can both be accommodated within a Hermitian complex metric g = g( ) + ig[ ] obeying
g = g g = g

(53)

When the grade of a is taken to be odd we have now that the graded commutator must be [a , i } = {a , i } = 0, instead of having [a , i ] = 0 as displayed before in eq-(9). Due to the Grassmannian odd nature of Aa we have also modif ications to the expressions for the following eld strength components
ab ab ac cb F = Aab A + 2 A[ A ]

(54)

b F = A A + Aa A gab

(55)

since [a , b } = {a , b } = 2gab 1, instead of having [a , b ] = 2ab . One should compared eqs-(54,55) with the prior expressions in eqs-(19,22). b Using the denition ea will yield mass e gab = g[ ] the quadratic terms F F [ ] terms for Kalb-Ramond like elds of the form g[ ] g = B B , after identifying B = g[ ] and raising the indices with the metric h = g ( ) . Another possibility is to raise the indices with the inverse Hermitian complex metric (g )1 = g = g ( ) + i g [ ] = g ( ) + i g [ ] (56a)

we specically used the tilde symbol and the inequality to emphasize that both g ( ) , g [ ] are nontrivial functions of both g and g[ ] . However, when a complex metric is involved one has to add the complex (Hermitian) conjugate of any given term in order to have a real-valued action. The measure associated with such Hermitian complex metric g is (|| det g ||) 2 = ( (det g ) (det g ) ) 4
1 1

(56b)

Furthermore one can also use the epsilon tensor in 4D and contract indices in the following expression which can be included in the action

L4 B B

= L4 |det g( ) | B B

|det g( ) |

(57a)

One of the most salient features of choosing the grade of a to be odd, and consequently when ea is an a-number, is that it is possible for the terms in eq-(57a) to cancel out a cosmological constant term constructed now in terms of h as L4 |det h | = L4 |det g( ) | (57b)

the contributions from h = g( ) and B = g[ ] in eqs-(57a, 57b) may cancel each other when the on-shell dynamics of the elds (associated with the action) satises

B B

|det g( ) |

+ 1 = 0

(57c)

after solving the eld equations for h = g( ) and B = g[ ] and inserting their solutions back into eq-(57c). This problem deserves further investigation.

Super-Cliord Algebras and Generalized SuperGeometry

Cliord gauge eld theories based on orthogonal Cliord Cl(m) algebras have been explored in [14] and references therein. One may assign an odd grade to the odd-valued rank 10

polyvector basis generators a , a1 a2 a3 , ...., and an even grade to the even-valued rank, like the unit element 1 and a1 a2 , a1 a2 a3 ,a4 , ..... There is an addition the odd grade i , i1 i2 ,j , .... and even grade ij , i1 i2 ,j1 j2 , ..... generators associated with the symplectic Cliord algebra. In this way one can generalize our construction to the case when m > 2; 2n > 2 and build a super gauge eld theory. Now we turn to the question of constructing a super-Cliord algebra and which is very dif f erent from the so-called super Cliord algebras with a Z4 grading proposed by [3], [4]. To proceed we turn to the the Lie superalgebra osp(m|2n) (ortho-symplectic superalgebra) whose even part is so(m) sp(2n). In particular sp(4) so(5). Since weve started with an orthogonal and a symplectic Cliord algebra, we may assign now an even grade to all the generators of the latter orthogonal-symplectic Cliord algebras. Therefore, the super-Cliord algebra in question must be a super algebra such that its even grade part must be given by the orthogonal-symplectic Cliord algebras, and its odd grade part must be an algebra whose generators Qu , u = 1, 2, 3, ..... (not to be confused with the spinorial charge generators Q , S of ordinary super conformal algebras ! ) should obey the graded commutator relations
v v [ a , Qu ] (Ma )v u Qv , [ ab , Qu ] (Mab )u Qv , [ abc , Qu ] (Mabc )u Qv , ... (58) v v [ i , Qu ] (Ni )v u Qv , [ ij , Qu ] (Nij )u Qv , [ ij,k , Qu ] (Nij,k )u Qv , ..... (59)

{ Qu , Qv } uv 1 + (La )uv a + (Lab )uv ab + (Labc )uv abc + ....... + (Li )uv i + (Lij )uv ij + (Li1 i2 ,j )uv i1 i2 ,j + (Li1 i2 ,j1 j2 )uv i1 i2 ,j1 j2 + ...... (60) where the M, N, L are suitable matrices. An analog of the above equations are the (anti) commutators of the (Anti) de Sitter superalgebra
[ Ja , Q ] (a ) Q , [ Jab , Q ] (ab ) Q , [ K, Q ] Q , ab { Q , Q } K + a Ja + Jab

(61) (62)

If one were to set the generators Qu , u = 1, 2, 3, ..... to coincide with the spinorial charge generators Q of ordinary supersymmetry by identifying the u index with the spinorial index, the Ma , Mab , Mabc , ...., La , Lab , Labc matrices in (58) could be identied with the gamma matrices a , ab , abc , .... The matrices a , ab , abc , as usual, are simply obtained by raising the indices of a , ab , abc , ... via a metric g ab and its antisymmetrized products as depicted in eq-(23a). The diculty arises then in constructing the matrices Ni , Nij , Nij,k , ....., Li , Lij , Lij,k , .... in eq-(59,60). We do not know at this stage if these matrices can be expressed in terms of the gamma matrices. This is an interesting problem that warrants further investigation. To sum up, a truly super Cliord algebra must be one whose (anti) commutators obey eqs-(58-60), in addition to the commutation relations among the s and the s generators (since now we have assigned an even grade to all of them). To considerably simplify the problem one could set to zero all the s generators except i , ij . How does one relate the Grassmannian odd Ai gauge elds to fermions, in particular to the gravitino ?. It is well known to the experts that spinors are right/left ideals 11

elements of the Cliord algebra and consequently spinors are already encoded within the Cliord algebraic structure itself. Hence, a natural correspondence (not an equality) 3 among Ai and the spin 2 -gravitino eld could be given as (omitting spinorial indices in the right hand side for simplicity)
j a ab abc + ...... (63) Ai + cabc ij A c + ca + cab i From the discussion in section 2, when Ai e , the left hand side of (63) becomes h so the mass-like squared terms in eq-(51) will have a correspondence with the quartic )2 + ..... fermionic terms ( Another correspondence is j k l cd ab b a Ai j A Al A ik c1 A A + c2 A a A b + c3 A ab A cd +... (64) c, ca , cab , ...., c1 , c2 , c3 ... are numerical coecients. A further analysis of the correspondence in eqs-(64, 64) will be left for future investigations. To nalize, following the approach of [6], [7], we introduce anti-commuting coordinates y u : y u y v = y v y u in order to initiate the construction of super-Cliord spaces. When one has commuting spacetime coordinates x , = 1, 2, ...., m and anticommuting coordinates y u , u = 1, 2, ...., 2n the exterior wedge product of super dierentials is dened as

dZ A dZ B = (1)(grade

A grade B )

dZ B dZ A

dx dx = dx dx , dy u dy v = dy v dy u , dx dy v = dy v dx , (65) and the eld strength associated with the ortho-symplectic Cliord-algebra-valued super dierential one-form A dx + Au dy u is F dx dx + Fuv dy u dy v , F = F , Fuv = Fvu (66)

Au admits an expansion similar to A in eqs-(15,32) by replacing the index with u. Now we are in a position to construct the coordinates of super Cliord spaces in terms of the polyvector-valued coordinates x[1 2 ....n ] which are fully antisymmetric under the exchange of indices, and the other set of coordinates with a mixed symmetry y u1 u2 .....ur ,v1 v2 .....vs under the index exchange. A recent study of the Extended Relativity Theory in Cliord Phase Spaces and modications of gravity at the Planck/Hubble scales, with many references therein can be found in [19]. We extended the construction of Borns Reciprocal Phase Space Relativity to the case of Cliord Spaces which involve the use of polyvector-valued coordinates and momenta x[1 2 ....n ] , p[1 2 ....n ] ; generalized vielbeins [a1 a2 ....ar ] [b1 b2 ....br ] A EM = E[1 2 .....s ] , E[1 2 .....s ] corresponding to the coordinate and momentum space directions, respectively; generalized metrics like g[1 2 ....r ] [1 2 ....s ] , ..... and the inclusion of an lower/upper (Planck/Hubble) length scale. A super Cliord valued coordinate in super Cliord spaces has a decomposition Z = x 1 + x + x1 2 1 2 + ...... + x1 2 ...m 1 2 ...m + 12

y u u + y u1 u2 u1 u2 + y u1 u2 ,v u1 u2 ,v + y u1 u2 ,v1 v2 u1 u2 ,v1 v2 + ....

(67)

We conclude with some of the salient features of the generalized super-geometry associated with super Cliord spaces : (i) A super-Cliord theory of gravity will amount to a generalized supergravity theory based on an orthogonal-symplectic Cliord algebraic structure and the geometry of superCliord spaces. [a1 a2 ....ar ] [b1 b2 ....br ] A (ii) The generalized vielbeins like EM = E[1 2 .....s ] , E[1 2 .....s ] , will cast more light to the preliminary analysis of the generalized vielbein postulate [20] that reveals tantalizing hints of new structures beyond D = 11 supergravity and ordinary space-time covariance. An E7 -valued vielbein in eleven dimensions was analyzed and they discussed the extension of these results to E8 . An E8 Gauge theory of gravity in 8D based on generalized vielbein was proposed in [14]. (iii) When the construction of super-Cliord spaces is extended to Cliord Phase Spaces [19], the coordinates X, P are now doubled, and one may connect with the U duality and double eld theory formalism of generalized geometry which has been gaining a lot attraction recently [22]. (iv) The fact that in general one has an inf inite number of generators described by the super Cliord algebras proposed in this work, one should try to nd its relation (if any) to the innite dimensional hyperbolic Kac-Moody algebras E11 which have been conjectured [21] to encode the hidden symmetries of M -theory in 11D. It was mentioned earlier that the quantum Virasoro algebra generators can be constructed in terms of operators of the generalized Cliord algebras as shown by [24]. By extending this construction using the super-Cliord algebraic approach in this work we should be able to connect with the hyperbolic Kac-Moody algebras E11 description of [21]. APPENDIX A : Graded Jacobi Identities In this Appendix we shall prove that the graded Jacobi identities are satised. We shall rstly take all the generators of the orthogonal Cliord algebra to be of even grade, while the i generator has odd grade and ij has even grade. Afterwards we shall take the odd rank generators of the orthogonal Cliord algebra to have odd grade; and the even rank generators to have even grade. The grades take the values of 0, 1 for even, odd grade respectively. Let us begin with (1)0 { i , [j , kl ] } + (1)1 { j , [kl , i ] } + (1)0 [ kl , {i , j } ] = { i , jk l + jl k } { j , ki l + li k } + 2 [ kl , ij ] = 2 jk il + 2 jl ik 2 ki jl 2 li jk + 2 ( ki lj + kj li + li kj + lj ki ) = 0 due to jk + kj = 0, jl + lj = 0 and jk = kj , ..... The graded Jacobi relation among ij , kl , pq is 13 (A.1)

(1)0 [ ij , [kl , pq ] ] + (1)0 [ kl , [pq , ij ] ] + (1)0 [ pq , [ij , kl ] ] = 0

(A.2)

it is also zero. One can verify that one ends with a collection of 48 terms, all of which can be gathered into groups (up to indices permutations) of the form ik pj ql + pj ki lq = 0 eq-(A.3) is zero because ik + ki = 0 and ql = lq , etc... (1)1 [ i , {j , k } ] + (1)1 [ j , {k , i } ] + (1)1 [ k , {i , j } ] = 2 [ i , jk ] 2 [ j , ki ] 2 [ k , ij ] = 2 ( ij k + ik j + jk i + ji k + ki j + kj i ) = 0 Similarly one obtains (1)0 [ m , [ ij , kl ] + (1)0 [ ij , [ kl , m ] + (1)0 [ kl , [ m , ij ] = 0 The graded Jacobi relation among ij , a , pq is (1)0 [ ij , [a , pq ] ] + (1)0 [ a , [ pq , ij ] ] + (1)0 [ pq , [ij , a ] ] = 0 because [a , pq ] = 0, ... and [pq , ij ] = pi qj + ..... The same result occurs with (1)0 [ ij , [ab , pq ] ] + (1)0 [ ab , [pq , ij ] ] + (1)0 [ pq , [ij , ab ] ] = 0 because [ab , pq ] = 0, .... The graded Jacobi relation among ij , a , b is (1)0 [ ij , [a , b ] ] + (1)0 [ a , [ b , ij ] ] + (1)0 [ b , [ij , a ] ] = 0 etc.... In the case that a is chosen to have on odd grade, eq-(A.8) is replaced by (1)0 [ ij , {a , b } ] +(1)0 { a , [ b , ij ] } +(1)1 { b , [ij , a ] } = [ ij , 2gab 1 ] = 0 (A.9) since [ b , ij ] = 0, ... Let us look at the graded Jacobi identities involving s , mn , r when the grade of s , r is taken to be odd, and the grade of mn is even (1)1 { s , [mn , r ] } + (1)0 [ mn , {r , s } ] + (1)0 { r [s , mn ] } = 14 (A.8) (A.7) (A.6) (A.5) (A.4) (A.3)

4 gmr gsn 4 gnr gsm + 4 gsm grn 4 gsn grm = 0 Compare (A.10) when the grade of s , r , mn are all chosen to be even [ s , [mn , r ] ] + [ mn , [r , s ] ] + [ r [s , mn ] ] = 4 gmr sn + 4 gnr sm + 4 gsm rn 4 gsn rm 4 gmr ns + 4 gnr ms + 4 gsm nr 4 gsn mr = 0

(A.10)

(A.11)

due to the antisymmetry of the bivector generators mr = rm , ..... The (anti) commutators of the a , a1 a2 , ....., a1 a2 .....am generators can be found in pages 543-545 of the monograph [23]. To check that the graded Jacobi identities are satised for the remaining of the innite number of generators i1 i2 ,j1 , i1 i2 ,j1 j2 , ...... would require a computer algebra package. They are satised for the graded commutators involving i , jk as shown in eqs-(A.1-A.5). Therefore, there is true algebraic closure in the description of the A eld, and its eld strength F components dened in eqs-(15-22) and which allowed us to proceed with the remaining analysis and results in sections 1, 2. APPENDIX B : Invariance of the Lagrangian We shall derive the invariance of the Lagrangian (33) under the graded gauge transformations (39-45). For simplicity we shall not include the spacetime indices. Let us begin with the variation 1 1 (F ij F kl ) (ik jl + il jk ) = (F ij F kl ) il jk 4 2 where F ij = F ji , .... It contains two types of terms. One of them is X1 = 2 F jm mn ni il F lk kj Due to the symmetry ni = in , eq-(B.2) is the same as X1 = 2 F jm mn in il F lk kj upon rearrangement of terms and indices, eq-(B.3) can be rewritten as X1 = 2 il in mn F jm kj F lk = 2 li in nm F mj jk F lk (B.4) (B.3) (B.2) (B.1)

the overall change of sign in (B.4) is due to 3 changes of sign due to the antisymmetry of the omega tensors. A cyclic permutation of (B.4) and after using F lk = F kl yields X1 = 2 F kl li in nm F mj jk = X1 (B.5)

therefore, after equating eq-(B.3) with eq-(B.5) one arrives at X1 = X1 X1 = 0. The reason that eq-(B.3) is the same as eq-(B.5) is because both have the same index contraction structure; i.e. the same trace. A relabeling of indices reveals that they are the same. The second type of terms in eq-(B.1) is 15

1 ik jl = 2 F i ik F kl lj j (B.6) 2 These terms (B.6) are not zero but will cancel out with those mixed terms stemming from the variation 1 (F i F j )ij . These mixed terms stemming from the latter variation are 2 given by 2 ( F i j + F j i ) F kl 1 ij ( 2 F im mn n F j + 2 F i F jm mn n ) = 2 2 F im mn n F j ij

(B.7)

due to the Grassmannian odd nature of F j and n . One must change signs as follows F j n = n F j . Hence, after reversing their ordering in (B.7) one obtains a change of sign 2 F im mn F j n ij (B.8) rearranging terms and indices in (B.8) one gets 2 F j ji F im mn n (B.9)

due to a sign change ji = ij . Therefore, upon combining (B.9) with (B.6) one has a cancellation of the mixed terms 2 F j ji F im mn n 2 F i ik F kl lj j = 0 (B.10)

since after an index relabeling one can see that the rst and second terms in (B.10) are equal but with opposite signs. These terms have the same index contraction structure. 1 (F i F j )ij which are not mixed are The terms in 2 1 2 F m mn ni ij F j (B.11) 2 due to the Grassmannian odd nature of F j and F m , after reversing their order in (B.11) one obtains a change of sign X2 = 2 X2 = 2 F j mn ni ij F m = 2 F j (ji ) in (nm ) F m = 2 F j ji in nm F m = X2 (B.12) hence from the equality of (B.11) and (B.12) (both have the same index contraction structure) one arrives at X2 = X2 X2 = 0. To sum up, after dropping the spacetime indices for convenience we have seen that the gauge variations of the quadratic terms involving F i , F jk are exactly zero. Following a similar procedure, one can show that the variation of the remaining terms F a Fa 2F ab Fab + F 2 , dropping the spacetime indices, is also zero. The variation of F F is trivially zero since F = 0. To conclude, the gauge variation of the quadratic Lagrangian (33) is zero and consequently it is invariant under the graded ortho-symplectic Cliord-valued gauge transformations. Ordinary graded gauge theories can be found in [25], for example. 16

Acknowledgments We are indebted to M. Bowers for assistance.

References
[1] M. Kaku, P. Townsend and P. van Nieuwenhuizen, Phys. Lett B 69 (1977) 304. C. Shi, G. Hanying, L. Gendao and Z. Yuanzhong, Scientia Sinica 23, no. 3 (1980) 299. [2] A. Crumeyrolle, Orthogonal and Symplectic Cliord Algebras (Springer Verlag, 1989). [3] G. Dixon, J. Math. Phys. 19 (1978) 2103. [4] G. Dixon, Letts in Math. Physics 5 (1981) 411. [5] P. Alvarez, P. Pais and J. Zanelli, Local supersymmetry without SUSY partners arXiv : 1306.1247. [6] H. De Bie and F. Sommen, A Cliord Analysis Approach to Superspace arXiv : 0707.2859 [7] F. Sommen, An extension of Cliord analysis towards supersymmetry in Clifford algebras and their applications in Mathematical Physics, vol. 2 (Ixtapa 1999), Birkhauser, Boston, 2000, pp. 199-224. [8] M. Pavsic, A Theory of Quantized Fields Based on Orthogonal and Symplectic Cliord Algebras Adv. Appl. Cliord Algebras 22 (2012) 449. [9] S. Vacaru, General Relativity and Gravitation 44 (2012) 1015. [10] I. Drummond, Phys. Rev. D 63 (2001) 043503. [11] C. Castro and M. Pavsic, Progress in Physics vol 1 (April 2005), 31. [12] F. Toppan, JHEP 0409: 016, (2004) hep-th/0406022 [13] D. Alekseevsky, V. Cortes, C. Devchand and A van Proeyen, Commun. Math. Phys. 253 (2004) 385. [14] C. Castro, Int. J. Geom. Methods in Mod. Phys. 6 (2009) 911. [15] C. Castro, Int. J. Mod. Phys A 21 (2006) 2149. [16] C. Castro, Advances in Applied Cliord Algebras 21 (2011) 661. [17] E.S. Fradkin, M.A. Vasiliev, Nucl. Phys. B 291 141 (1987) 17

[18] E.S. Fradkin, M.A. Vasiliev, Phys. Lett. B 189 (1987) 89 [19] C. Castro, The Extended Relativity Theory in Cliord Phase Spaces and Modications of Gravity at the Planck/Hubble Scales, to appear in Advances in Cliord Algebras. [20] H. Godazgar, M. Godazgar and H. Nicolai, Generalized geometry from the ground up, arXiv : 1307.8295 [21] P. West, Generalized geometry, eleven dimensions and E11 arXiv : 1111.1642. [22] D. Berman, H. Godazgar, M. Perry, and P. West, Duality Invariant Actions and Generalized Geometry arXiv : 1111. 0459. [23] K. Becker, M. Becker and J. Schwarz, String Theory and M -theory pp. 543-545 [24] E. H. Kinani, Advances in Applied Cliord Algebras 13 (2003) 127. [25] R. Kerner, Comm. Math. Phys. 91 (1983) 213.

18