International Journal of Modern Physics A Vol. 16 No.

5 (2001) 759-766 © World Scientific Publishing Company

COMPLEX GRAVITY AND NONCOMMUTATIVE GEOMETRY

ALI H. CHAMSEDDINE Center for Advanced Mathematical Sciences (CAMS) and Physics Department American University of Beirut College Hall, Room 425 P.O. Box 11-0236, Beirut, Lebanon The presence of a constant background antisymmetric tensor for open strings or D-branes forces the space-time coordinates to be noncommutative. An immediate consequence of this is that all fields get complexified. By applying this idea to gravity one discovers that the metric becomes complex. Complex gravity is constructed by gauging the symmetry U(I, D -1). The resulting action gives one specific form of nonsymmetric gravity. In contrast to other theories of nonsymmetric gravity the action is both unique and gauge invariant. It is argued that for this theory to be consistent one must prove the existence of generalized diffeomorphism invariance. The results are easily generalized to noncommutative spaces.

At planckian energies, the manifold structure of space-time will not hold, and a new geometrical setting is needed. At present there are two possible candidates to describe space-time at high energies, one is string theory and the other is noncommutative geometry.' Recently these two approaches came together when it was realized that the presence of constant background B-field for open strings or Dbranes implies that the coordinates of space-time become noncommuting.2,3,4,5,6,7,8 This result is expected to generalize to the case of a non-constant B-field. The resulting geometrical space is expected to be noncom muting and curved. The question I will address in this talk is how to describe the dynamics of the gravitational field in such spaces. One possibility is to use the tools of noncommutative geometry of Alain Connes as specified by the spectral triple (A, 1i, D) where A is an associative algebra with a * product and identity, H a Hilbert space and D a self-adjoint operator on 1i such that [D, a], a E A defines a bounded operator on H.g In this setting it is possible to develop the noncommutative analogue of Riemannian geometry. A good example of the realization of noncommutative geometry is the data encoded in superconformal field theory.' The operator D encodes the metric, differential calculus, integration and dynamics. For simple noncommutative spaces such as the noncommutative space defined by the standard model all information about the bosonic and fermionic action is encoded in the spectrum of the Dirac operator. This is known as the spectral action principle.l'' The difficulty in this approach is that in order to make progress one must know the Dirac operator. Enough information must be available about D to define geometrical quantities. In the problem at hand it is not easy to
759

Indeed the noncommutative Yang-Mills action is invariant under the gauge transformations (2) where g.lis the inverse of 9 with respect to the star product: (3) The contributions of the terms iBl"v in the star product forces the gauge fields to be complex.1) gauge fields b: The U(l. Assume that we start with the U(l.8 (4) One can imagine a general setting where the closed string theory metric arise as an effective metric coming from open strings. D . This also seems inevitable as the star product appears in the operator product expansion of the string vertex operators.8. Only conditions such as At = -AI" could be preserved under gauge transformations provided that 9 is unitary: 9 t * 9 = 9 * 9 t = 1. D . or where the D-branes become dynamical. Chamseddine guess what D should one start with. . It is not possible to restrict AI" to be real or imaginary to get the orthogonal or symplectic gauge groups as these properties are not preserved by the star product. The strategy I will adopt is to first gather information about noncommutative spaces with constant background B-fields.1) group of transformations is defined as the set of matrix transformations leaving the quadratic form (6) w. Under such circumstances one can get an effective metric of the form (5) Because of B contributions the metric must become complex. H. The operator product expansion for vertex operators is identified with the star (Moyal) product of functions on noncommutative spacesy.13 For open strings in constant background B-field the effective metric iS14. The star product is defined by f (x) * 9 (x) = e~e"v a~" a~v f (x + () 9 (x + 1]) k=1)=o (1) This definition forces the gauge fields to become complex. Open strings or D-branes in presence of constant background B-field can be realized by deforming the algebra of functions on the classical world volume. We are therefore led to investigate whether the metric can become complex.12 In this respect it was shown that noncommutative U(N) Yang-Mills theory does arise in string theory.760 A.

1) with D .1) invariant. it is not difficult to see that (16) is hermitian and U(I.. The gauge fields w: (7) b must then satisfy the condition (8) The curvature associated with this gauge field is (9) Under gauge transformations we have (10) where the matrices M are subject to the condition: (11) The curvature then transforms as R~ a = MaR I-£V d M-b1d I-£V b c C (12) Next we introduce the complex vielbein -. 1.Complex Gravity and Noncommutative Geometry 761 invariant. . The metric is defined by (17) satisfy the property gtv imaginary parts: = gVJ-L" When the metric is decomposed into its real and gl-£v = GI-£V + iBl-£v (18) the hermiticity property then implies the symmetries (19) . and its inverse e~ defined by (13) which transform as (14) It is also useful to define the complex conjugates (15) With this.1 positive entries.. D . . where za are D complex fields and 7]g = diag ( -1.

where we define gol-'v = egl-'eSv1]ab. (22) The last formula appears in the metric of the effective action in open string theory.l" which resulted in a nonsymmetric action for gravity.1v. This is equivalent to the statement that the Lagrangian should be completely expressible in terms of GI-'V and BI-'v only. with two possible contractions of the Riemann tensor.l and a and e11.17 in relation to string duality. unlike others proposed to describe non symmetric gravity+" is unique. except for the measure. H.'C a satisfy (XI-'C a)t a nonsymmetric metric = -Xl-'aC" One has to be very careful in working with (24) . after eliminating b through its equations of motion.jGv _ el-'bevwc au (va (eVel-'C _ a b _ el-'evcwb b va el-'eVC)) a (23) where Xf. In reality we have + iAfb' w: (21) In this special gauge. but the results obtained there are different from what is presented here. Similar ideas have been proposed in the past based on gauging the groups O(D. and unambiguous. and Ab = AOb From the gauge transformations of egl-' and ell-' one can easily show that the gauge parameters AOb and Afb can be chosen to make eOl-'a symmetric in J..762 A. gOf. It is worthwhile to stress that the above action. The infinitesimal gauge transformations for e~ is oe~ = Abet.l and a.vgo 1<). which can be decomposed into real and imaginary parts by writing e~ = egl-' + ie'L. One goes to the second order formalism by integrating out the spin connection and substituting for it its value in terms of the vielbein.. and use egl-' to raise and lower indices we get -2ell-'v gol-'v - IB 4" 1-'1< B).1vg!{\ = o~.W antisymmetric in J. Chamseddine The gauge invariant Hermitian action is given by (20) where e = det (e~).. The ordering of the terms in writing the action is done in a way that generalizes to the noncommutative case. 14 We can express the Lagrangian in terms of e~ only by solving the b equations of motion w: el-'eVbwC a u b + eVel-'cwb va b _1_8 . The resulting action depends only on the fields gf. The idea of a hermitian metric was first forwarded by Einstein and Strauss. D)16 and GL(D).

XIlIlP) any loss in generality. the metric gllv is symmetric and g(3llgfTll = o~ so that the inverse of M::fffT'Y is simple.Complex Gravity and Noncommutative Geometry 763 but g.. The w-equation wl<P 11+ w:1< = 1 assume.(3 -8"L"8P) (3a. because of the non symmetry of gllv this is fairly complicated and could only be solved by a perturbative expansion. (32) . It imposes a condition on the e~ w: (25) We can therefore (w:a=O). +8"L"8P "Sa -~ 4 (8"L"8P a. we point out that the trace part of b (corresponding to the U(l) part in U(D)) must decouple from the other gauge fields..X'. without gives . Before solving the W equations. that w: b is traceless 88~ (3X'. In the present case. It is thus undetermined and decouples from the Lagrangian after substituting its equation of motion.8fTs: 0") 8P 2 (3a'Y a-y(3 -~ 4 (81<8fTLP (3a'Y + 8 a-y(3LP o" K . Writing gllv = GIlV + iBllv. and defining GIlVGvp = 8~ implies that 8~ +L~ iGIlP The inverse of M::fffT'Y defined by e.81< fTLP) 8a "1(3 L1<8" 8P) a -y (3 +O(L2) +~ (L 8" 4 "I K (3 8 a P + L"8fTa 8P (3 .wgllP -I.8~.I< == Y~I< (26) We can rewrite this equation after contracting with ellce~ to get (27) By writing wp al< = wpvl<eva we get (8~8~8J + lllgfTIlO~O~) MaI<PfT (3'Y wafh = YfTPI< (28) To solve this equation we have to invert the tensor = 8a P O'Y I< 8(3 a + g(3llg a u s: 8"1 P I< (29) In the conventional case when all fields are real.. - 2GIlP BpfTGfTa Bav + O(B3) (30) (31) is evaluated to give ~ (0"'Y(3a8" + 8fT8P 81<.11 + 88~ (-X~1l 1 + 3X'ttI<) . Care also should be taken when raising and lowering indices with the metric.

LvpHJ.764 A.La a = Ov (e~eva [r BJ.w and higher order terms. and after integration by (37) This is identical to the usual expression 112 HJ. This would protect the field BJ. D-l) it is not difficult to generalize the steps that led us to the action for complex gravity to spaces where coordinates do not commute.LBvp + avBpJ.~ld *b (40) . based on the idea of gauge invariance of the group U(l.LV - M" C * R ltV d * C lVJ '1. Chamseddine This enables us to write (33) It is clear that the leading term reproduces the Einstein-Hilbert action plus contributions proportional to B. This is presently under investigation.L + opBJ.Lv. which generalizes diffeomorphism invariance to complex diffeomorphism.19 This can be traced to the fact that there is no gauge symmetry associated with the field BJ.Lvp = aJ.Lv. We then evaluate (36) When the w. For the theory to become consistent one must show that the action above has an additional gauge symmetry.<vp is substituted parts one gets back into the Lagrangian.Lvp ... It is therefore essential to identify whether there are additional symmetries present in the above proposed action. is now which transforms according to (39) R.ab J. We can check that in the flat approximation for gravity with GJ. where HJ.. First we write (34) The W/~ aequation implies the constraint XJ.LV taken to be 15 the B J. H. The later developments of nonsymmetric gravity showed that the occurrence of the trace part of the spin-connection in a linear form would result in the propagation of ghosts in the field BJ. First the gauge fields are subject to the gauge transformations -a Ma WJ.Lv - e~eJ.LV' kinetic terms.L d * M~ld *b - Ma c * UJ.LV field gets the correct J.lb is the inverse of Mg with respect to the star product. where the usual products are replaced with star products.Lv from having non-physical degrees of freedom.L b = c c * WJ.LM~lc *b >J (38) The curvature where M. or equivalently.La) =0 (35) This gives the gauge fixing condition = O.Lv. Having shown that it is possible to formulate a theory of gravity with nonsymmetric complex metric.

Douglas and A. Phys. Schupp and J. References 1. P. 18. M.-T. Douglas and C. B528. Phys. Math.030 (1999). 186. Rev.S. Nucl. Cheung and M. 203 (1989). B218. Phys. Frohlich. 731 (1997). 14. 151 (1999). J. S. H. Phys.1) gauge invariant (44) where e = det (e~). 12.H. Lett. Cannes. The connection of this action to the gravity action derived for noncommutative spaces based on spectral triples20. Chamseddine and A. J. Ann. Y. 17.H. P.032 (1999). Wess. Zachos. JHEP 9802. JHEP 9906. 13.K. B288. Fletcher and C.R. Phys. 4. C16. 161 (2000) [hepthj0001203].G. eds. Nucl. Nappi and S. Cannes.-S. D . 3.Lv. Ho. F. 6. 008 (1998). 5453 (1993). 1994). Lovelace. Schraml. 9. 8. Arfaei and M. 3722 (1995) and references therein. 525 (1987). Cannes. A. Hull. Phys. Phys. 3 (1993).21. 016 (1999). V. 003 (1998). Sheikh-Jabbari. A. Yang-Festschrift. J.LV and BJ. Yau (International Press. 1995) 10. Chamseddine and J. The U(l. Nucl. Some Elements of Noncommutative and SpaceTime Geometry. C. This action differs from the one considered in the commutative case by higher derivatives terms proportional to BJ. 15.Complex Gravity and Noncommutative Geometry 765 Next we introduce the vielbeins e~ and their inverse defined by (41) which transform to (42) The complex conjugates for the vielbeins are defined by (43) Finally we define the metric Hermitian action is gllV = (e~) t TJb' * e~. J. 731 (1946). 47. C. Seiberg and E.M. N. M. Schwarz. Math.A. Phys. B250. Liu and S. . 16. JHEP 9802. B390. 44 (1990). or whether the Lagrangian reduces to a function of GJ. Callan.Lv. D.-M. Ardalan. A. Witten.E. B528.22 remains to be made. 36. Maharana and J. C.Chu and P.Lv and their derivatives only. JHEP 9909. Krogh. Phys. Strauss.K.E. 10. Schomerus. JHEP 9902. Madore. J. Moffat. It would be very interesting to see whether these terms could be reabsorbed by redefining the field BJ. 2. 11. J. 5. Einstein and E. Math. Hoppe. W. A. C. Yost.R. Noncommutative Geometry (Academic Press. Nucl. 7. Eur. Schwarz.H. Lett. A. Comm. In order to do this one must understand the structure of Dirac operators for spaces with deformed star products. D47. C. 185 (1998). Siegel.R. Fairlie. Phys.

155. A. Phys. 36. Damour. Felder and J. D47. Frohlich. 21. Frohlich. 22. J. Phys.H. O. Phys. A. Math. Connes. 1541 (1993). Chamseddine T. 20. Rev. 36. Chamseddine. J.. 109 (1993). 6194 (1995).H. Math. H. S. A. G . 6255 (1995).766 19. Chamseddine. Comm. . Deser and J. Math. Phys. A. McCarthy. Grandjean and J.