String universality in ten dimensions

Allan Adams,1 Oliver DeWolfe,2 and Washington Taylor1 1Center for Theoretical Physics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 2Department of Physics, 390 UCB, University of Colorado, Boulder, CO 80309, USA We show that the N = 1 supergravity theories in ten dimensions with gauge groups U(1)496 248 and E8 × U(1) are not consistent quantum theories. Cancellation of anomalies cannot be made compatible with supersymmetry and abelian gauge invariance. Thus, in ten dimensions all super- symmetric theories of gravity without known inconsistencies are realized in string theory.

INTRODUCTION R F Supersymmetry and anomaly cancellation place strong R constraints on quantum theories of gravity. Such con- B straints are strongest in higher dimensions. In eleven di- R mensions there is a unique theory of gravity compatible F with supersymmetry. This theory is believed to be de- R scribed as a UV-complete quantum theory by the branch of string theory known as “M-theory”. Similarly, in ten dimensions with two supersymmetries (N = 2), there are FIG. 1: Green-Schwarz-type tree diagram arising in N = 1 su- pergravity theories in ten dimensions with abelian factors. With only two consistent supergravity theories, known as type no corresponding anomaly to cancel, these theories are not gauge IIA and IIB. Both of these are realized as limits of string invariant theory. In these highly supersymmetric situations, then, we have “string universality,” meaning that all theories without known inconsistencies are realized in string the- lation of gravitational and gauge anomalies [5] requires ory. the Green-Schwarz mechanism [6]. The only consistent As the dimension and number of supersymmetries de- choices of the gauge group G without abelian factors, creases, the range of possible theories dramatically in- G = SO(32) and G = E8 × E8, are realized as the type creases. In four space-time dimensions with one or no I and heterotic limits of string theory. It has been noted 496 248 supersymmetries, we have only a limited understanding that the gauge groups G = U(1) and G = E8×U(1) of the range of possible string compactifications, but our satisfy some of the same conditions as the SO(32) and knowledge of consistency conditions needed for UV com- E8 × E8 cases [7], and therefore might appear to be con- pletion is still weaker (see for example [1]). The term sistent, anomaly-free theories. They are thus candidates “swampland” has been used to characterize the set of for the swampland, since they have no known embedding theories which cannot be ruled out from knowledge of into string theory. In [8] Fiol noted several properties of the low-energy physics, and yet which cannot be real- these theories, related to their moduli spaces and singu- ized in string theory [2]. Given our limited knowledge of larities under compactification, suggesting they cannot both the space of string compactifications and constraints be embedded into a theory of quantum gravity. on consistent quantum gravity theories, the apparent In this note, we demonstrate that these theories with swampland is a moving target, decreasing in scope when- abelian gauge group factors indeed cannot be consistent ever new string vacuum constructions or quantum con- supersymmetric quantum theories of gravity. In brief, al- arXiv:1006.1352v2 [hep-th] 29 Oct 2014 sistency constraints are identified. Although the space of though the anomaly factorizes as needed for the Green- four dimensional theories has been so far hard to char- Schwarz mechanism, the abelian gauge fields do not par- acterize globally, it was conjectured in [3] that N = 1 ticipate in hexagon diagrams, so there are no abelian supergravity theories in six dimensions satisfy string uni- anomalies to cancel. Supersymmetry, however, requires a versality. There are, however, some theories in this class coupling of the B field to the abelian gauge bosons of the which still lie in the apparent swampland, neither prov- form BF 2, just as in the non-abelian case. In using the ably inconsistent nor as-yet realized in string theory [4], B field to cancel the gravitational and non-abelian gauge so this conjecture remains to be conclusively proven or anomalies, the standard Green-Schwarz mechanism gen- disproven. erates tree diagrams with external abelian gauge bosons In the present work, we reconsider the simplest and of the form in Figure 1 with no associated anomaly. This most symmetric class of supergravity theories where corresponds to a breakdown of gauge invariance for theo- string universality is in doubt, namely N = 1 super- ries with abelian factors. As a result, we find that ten di- gravity in ten dimensions. Supersymmetry allows the mensional supergravity theories appear to manifest string addition of Yang-Mills fields in such theories, and cancel- universality — all consistent theories have string the- 2 ory embeddings, and the 10D supergravity swampland The fermions are all Majorana-Weyl fields, and thus be- is empty. ing chiral they contribute to both gauge anomalies (the gaugino) and gravitational anomalies (gaugino, dilatino and gravitino). These contributions come from hexagon ANOMALY CANCELLATION IN NON-ABELIAN diagrams with six external gauge bosons and/or gravi- THEORIES tons coupled to the various fermi fields running in a loop. The contributions to the anomaly polynomial from the We begin by reviewing Green-Schwarz anomaly cancel- fermions for a general non-abelian gauge group were com- lation for the purely non-abelian theories. The bosonic puted by Alvarez-Gaum´eand Witten [5], and, following fields of ten-dimensional N = 1 supergravity plus super- the presentation of Polchinski [11], can be arranged to Yang-Mills consist of a metric GMN , the dilaton Φ, a the form a a a 2-form B2 and the gauge field A = A T with T the  2 3  generators of the gauge group. The action for these fields 1 6 1 4 2 (TrF ) Iˆ12 = −TrF + TrF TrF − (7) can be written as 1440 48 14400  6 4 2 2 3  1 Z  trR trR trR (trR ) Y4X8 S = e−2Φ ∗ R + 4dΦ ∧ ∗dΦ (1) +(n − 496) + + + , 2 725760 552960 1327104 768 2κ10 1 1 a a − H3 ∧ ∗H3 − F ∧ ∗F , where Tr is the trace in the adjoint representation of the 2 2 gauge group (supersymmetry requires that the gauginos √ with ∗R = d10x −GR. This theory also includes three running in the loop transform in the adjoint), tr is the fermions, a gravitino ΨM , a dilatino λ, and a gaugino trace in the fundamental of SO(1, 9), n is the total num- χ ≡ χaT a, whose contributions to the action we suppress. ber of gauge bosons and It was shown by Bergshoeff et al. [9] for the case of an 2 2 abelian gauge group, and extended to the non-abelian 2 1 2 4 (trR ) Y4 = trR − TrF ,X8 = trR + − (8) case by Chapline and Manton [10], that supersymmetry 30 4 (TrF 2)(trR2) TrF 4 (TrF 2)2 requires the field strength of the 2-form to acquire an + − . extra piece consisting of the Chern-Simons term for the 30 3 900 gauge group, The Green-Schwarz anomaly cancellation mechanism is Y possible when the first two terms in (7) vanish, so that H3 ≡ dB2 − ω3 , (2) the anomaly takes the factorized form of the final term. where For gauge groups SO(32) and E8 × E8, the first term 2 inside large parentheses in (7) vanishes due to particular ωY ≡ Aa ∧ dAa + f abcAa ∧ Ab ∧ Ac . (3) 3 3 identities of those groups, and n = 496 for both, killing the middle term. As a result, invariance under the gauge group is only For the remaining Y4X8 term to be cancelled through maintained if B transforms nontrivially under this group the Green-Schwarz mechanism, the three-form field as well. If strength must be enhanced at higher orders in the deriva- tive expansion by a Chern-Simons term in the spin con- A → A + dΛ − i[A, Λ] ,B2 → B2 + Tr(ΛF ) , (4) nection, with Λ an algebra-valued 0-form, then H3 is gauge in- Y R variant. H3 ≡ dB2 − ω3 + ω3 , (9) As is well-known, possible gauge groups for this the- ory are highly constrained by anomaly cancellation. For and the action must include at higher order the Green- theories in ten dimensions, the anomaly is conveniently Schwarz term, expressed in terms of a formal twelve-form anomaly poly- Z nomial Iˆ12(F,R), constructed out of the gauge and tan- ∆S ∼ B2 ∧ X8 . (10) gent bundle curvature 2-forms. This obeys the descent relations, To preserve local Lorentz invariance the two-form B2 must now transform as Iˆ12 = dIˆ11 , δIˆ11 = dIˆ10 , (5) where δ denotes the combined gauge and local Lorentz B2 → B2 + Tr(ΛF ) − tr(ΘR) . (11) transformation. The failure of the path integral to be Due to this modified transformation, the Green-Schwarz gauge invariant is then given by the integral of Iˆ , 10 term is not gauge invariant; instead, because Z δ log Z ∼ Iˆ10 . (6) R Y R Y Y4 = d(ω3 − ω3 ) , δ(ω3 − ω3 ) = dδB2 , (12) 3 we have from the descent relations that the gauge vari- the abelian generators drop out analogously, while E8 ation of (10) is exactly of the correct form to cancel the alone obeys the same relation used in the E8 × E8 case. anomaly in (6). Diagramatically, a tree diagram with a Correspondingly, Y4 and X8 in these cases lose all the 496 B propagator connecting BY4 and BX8 vertices is gener- traces over the U(1) generators. For G = U(1) this is 2 ated by the cross-terms in |H3| induced by (9) and the particularly simple, Green-Schwarz term, cancelling the anomaly from the (trR2)2 hexagon diagrams. Notice that the first equation in (12) (496) 2 (496) 4 Y4 = trR ,X8 = trR + , (15) is equivalent to 4 248 while for G = E8 × U(1) they take the form (8) Y = dH . (13) 4 3 but with E8 field strengths only. The only anomalies that must be cancelled, then, are gravitational and non- This coincidence between the factorized anomaly poly- abelian gauge anomalies. nomial (7) and the modified field strength required by As shown by Bergshoeff et al. [9], however, super- supersymmetry (9) is the essential relation that allows symmetry requires that each abelian factor U(1)(i) con- the Green-Schwarz mechanism to operate. tributes an abelian Chern-Simons contribution to H of Through this mechanism, the non-abelian SO(32) and 3 the form E8 × E8 supergravities can be free of quantum anoma- X (i) lies. Indeed, since these supergravities are the low-energy H3 ≡ dB2 − ω3 + ..., (16) limit of the type I and heterotic string theories, where i the higher-derivative terms in (9) and (10) arise auto- (i) (i) (i) matically, these theories seem to be consistent quantum where ω3 = A ∧ dA and ... indicates gravita- theories in the UV. tional and possible non-abelian Chern-Simons contribu- tions. As a result, the kinetic term for B2 is only gauge- invariant if B2 transforms under a general abelian gauge 10D SUPERGRAVITY WITH ABELIAN GAUGE transformation as GROUP FACTORS X (i) (i) δΛB2 = Λ F . (17) We now turn to the primary focus of this note, the i 496 theories with gauge groups G = U(1) and G = E8 × To cancel the gravitational and possible non-abelian U(1)248. These theories both contain n = 496 generators, anomalies, we need a Green-Schwarz term of the usual and the first term in large parentheses in (7) vanishes for form (10). The Green-Schwarz term is not invariant both, so they again have anomaly polynomials which take under the abelian factors in the gauge group, however, the factorized form Y4X8 [7]. Naively it then appears that transforming as the Green-Schwarz mechanism can again be brought to Z Z X (i) (i) bear. We argue that this is not the case. The crux of the δΛ B2 ∧ X8 = Λ F ∧ X8 . (18) issue is that the Green-Schwarz term needed to cancel i gravitational anomalies is required by supersymmetry to Since there is no abelian anomaly to cancel this gauge have a non-trivial abelian gauge variation, but there is variation, the abelian symmetries are explicitly vio- no abelian anomaly to cancel this variation. lated. Alternately, preserving gauge invariance under To see this, recall that the F -dependent terms in the the abelian factors forbids this Green-Schwarz term, anomaly polynomial (7) were generated by loops of gaug- whose absence would leave an uncancelled local Lorentz inos coupling to external gauge bosons. Since, however, anomaly. Supersymmetry thus allows us to preserve ei- the general coupling between a gaugino and the gauge ther abelian gauge or local Lorentz invariance, but not fields is both. The inapplicability of the Green-Schwarz mechanism Dχa = ∂χa − if abcAbχc , (14) in these cases can be viewed as the failure of the abelian a U(1) gauge field, for which all associated structure con- theories to satisfy (13), stants vanish, decouples from the gauginos and does not Y4 6= dH3 . (19) appear in hexagon diagrams at all. This means that the anomaly polynomial should be independent of all abelian The abelian fields decouple from hexagon diagrams and field strengths. hence drop out of Y4, but are kept in dH3 by the demands This is in fact already encoded in (7), since Tr is the of supersymmetry. This leads to tree-level contributions trace in the adjoint, and the adjoint of U(1) is the sin- of the form F 2R4 to the anomaly polynomial which do glet, so these traces vanish. Indeed this is why the term not correspond to any one-loop anomalies; correspond- in parentheses vanishes for these theories; for U(1)496 it ingly, the descent relations no longer imply that the vari- 248 is simply zero term by term, and for E8 × U(1) all ation of B2 can cancel the complete anomaly. Thus the 4

496 248 ten-dimensional N = 1 U(1) and E8 × U(1) theo- H. Ooguri and C. Vafa, “On the geometry of the string ries cannot be made anomaly-free by the Green-Schwarz landscape and the swampland,” Nucl. Phys. B 766, 21 mechanism. We have thus shown that there are no consis- (2007) arXiv:hep-th/0605264. tent supergravity theories in ten dimensions that cannot [3] V. Kumar and W. Taylor, “String Universality in Six Dimensions,” arXiv:0906.0987 [hep-th]. be obtained from string theory. [4] V. Kumar and W. Taylor, “A bound on 6D N = 1 su- We would like to thank Dan Freedman, Michael Green, pergravities,” JHEP 0912, 050 (2009) arXiv:0910.1586 Vijay Kumar, Joe Minahan, Daniel Park, Joe Polchinski [hep-th]. and John Schwarz for helpful discussions and correspon- D. R. Morrison, V. Kumar and W. Taylor, “Mapping dence. We would like to thank Yuji Tachikawa for point- 6D N = 1 supergravity to F-theory,” arXiv:0911.3393 ing out errors in the coefficients of the anomaly polyno- [hep-th]. mials in equations (7) and (8) in an earlier version. O.D. D. R. Morrison, V. Kumar and W. Taylor, “Global as- pects of the space of 6D N = 1 supergravities,” to appear. would like to thank the Center for Theoretical Physics at [5] L. Alvarez-Gaum´e, E. Witten, “Gravitational Anoma- MIT for hospitality. We would like to particularly thank lies,” Nucl. Phys. B234, 269 (1984). the Goosebeary’s lunch truck where this work was com- [6] M. B. Green, J. H. Schwarz, “Anomaly Cancellations in pleted. The research of A.A. and W.T. was supported by Supersymmetric D=10 Gauge Theory and Superstring the DOE under contract #DE-FC02-94ER40818. The re- Theory,” Phys. Lett. B149, 117-122 (1984). search of O.D. was supported by the DOE under contract [7] M. B. Green, J. H. Schwarz and E. Witten, “Superstring #DE-FG02-91-ER-40672. Theory. Vol. 2: Loop Amplitudes, Anomalies And Phe- nomenology,” Cambridge Univ. Pr. ( 1987) [8] B. Fiol, “Populating the swampland: The Case of 496 248 U(1) and E8 × U(1) ,” [arXiv:0809.1525 [hep-th]]. [9] E. Bergshoeff, M. de Roo, B. de Wit et al., “Ten- Dimensional Maxwell-Einstein Supergravity, Its Cur- [1] A. Adams, N. Arkani-Hamed, S. Dubovsky, A. Nicolis rents, and the Issue of Its Auxiliary Fields,” Nucl. Phys. and R. Rattazzi, “Causality, analyticity and an IR ob- B195, 97-136 (1982). struction to UV completion,” JHEP 0610, 014 (2006) [10] G. F. Chapline, N. S. Manton, “Unification of Yang-Mills arXiv:hep-th/0602178. Theory and Supergravity in Ten-Dimensions,” Phys. [2] C. Vafa, “The String landscape and the swampland,” Lett. B120, 105-109 (1983). [hep-th/0509212]. [11] J. Polchinski, “String theory. Vol. 2: Superstring theory N. Arkani-Hamed, L. Motl, A. Nicolis and C. Vafa, “The and beyond,” Cambridge Univ. Pr. (1998) string landscape, black holes and gravity as the weakest force,” JHEP 0706, 060 (2007) arXiv:hep-th/0601001.