CTP-TAMU-33/96 hep-th/9608117

M-THEORY (THE THEORY FORMERLY KNOWN AS STRINGS)1

arXiv:hep-th/9608117 v3 4 Oct 1996

M. J. Du2 Center for Theoretical Physics Texas A&M University, College Station, Texas 77843

ABSTRACT Superunication underwent a major paradigm shift in 1984 when eleven-dimensional supergravity was knocked o its pedestal by ten-dimensional superstrings. This last year has witnessed a new shift of equal proportions: perturbative ten-dimensional superstrings have in their turn been superseded by a new non-perturbative theory called M-theory, which describes supermembranes and supervebranes, which subsumes all ve consistent string theories and whose low energy limit is, ironically, eleven-dimensional supergravity. In particular, six-dimensional string/string duality follows from membrane/vebrane duality by compactiduality) or S 1 /Z2 T 4 (heterotic/Type IIA duality) or S 1 T 4 (Type IIA/Type IIA dufying M-theory on S 1 /Z2 K3 (heterotic/heterotic duality) or S 1 K3 (Type IIA/heterotic

ality).

Based on talks given at the Geometry and Physics Conference, Warwick, U. K., March 1996; the SUSY

96 conference, Maryland, U.S.A., June 1996; the CERN Duality Workshop, Geneva, Switzerland, June 1996; and the International School of Subnuclear Physics, Erice, Italy, July 1996.
2

Research supported in part by NSF Grant PHY-9411543.

Ten to eleven: it is not too late

The maximum spacetime dimension in which one can formulate a consisistent supersymmetric theory is eleven3 . For this reason in the early 1980s many physicists looked to D = 11 supergravity [2], in the hope that it might provide that superunication [3] they were all looking for. Then in 1984 superunication underwent a major paradigm shift: eleven-dimensional supergravity was knocked o its pedestal by ten-dimensional superstrings [4], and eleven dimensions fell out of favor. This last year, however, has witnessed a new shift of equal proportions: perturbative ten-dimensional superstrings have in their turn been superseded by a new non-perturbative theory called M-theory, which describes (amongst other things) supersymmetric extended objects with two spatial dimensions (supermembranes), and ve spatial dimensions (supervebranes), which subsumes all ve consistent string theories and whose low energy limit is, ironically, eleven-dimensional supergravity. The reason for this reversal of fortune of eleven dimensions is due, in large part, to the 1995 paper by Witten [5]. One of the biggest problems with D = 10 string theory [4] is that Type IIA and Type IIB. As a candidate for a unique theory of everything, this is clearly an embarrassment of riches. Witten put forward a convincing case that this distinction is just an artifact of perturbation theory and that non-perturbatively these ve theories are, in fact, just dierent corners of a deeper theory. Moreover, this deeper theory, subsequently dubbed M-theory, has D = 11 supergravity as its low energy limit! Thus the ve string theories and D = 11 supergravity represent six dierent special points4 in the moduli space of M-theory. The small parameters of perturbative string theory are provided by < e >, where is the dilaton eld, and < ei > where i are the moduli elds which arise after compactication. What makes M-theory at once intriguing and yet dicult to analyse is that in D = 11 there is neither dilaton nor moduli and hence the theory is intrinsically non-perturbative. Consequently, the ultimate meaning of M-theory is still unclear, and Witten has suggested
3

there are ve consistent string theories: Type I SO(32), heterotic SO(32), heterotic E8 E8 ,

The eld-theoretic reason is based on the prejudice that there be no massless particles with spins greater

than two [1]. However, as discussed in section (5), D = 11 emerges naturally as the maximum dimension admitting super p-branes in Minkowski signature. 4 Some authors take the phrase M -theory to refer merely to this sixth corner of the moduli space. With this denition, of course, M -theory is no more fundamental than the other ve corners. For us, M -theory means the whole kit and caboodle.

that in the meantime, M should stand for Magic, Mystery or Membrane, according to taste. The relation between the membrane and the vebrane in D = 11 is analogous to the relation between electric and magnetic charges in D = 4. In fact this is more than an analogy: electric/magnetic duality in D = 4 string theory [6, 7] follows as a consequence of string/string duality in D = 6 [8]. The main purpose of the present paper is to show how D = 6 string/string duality [9, 10, 11, 12, 13, 14, 5] follows, in its turn, as a consequence of membrane/vebrane duality in D = 11. In particular, heterotic/heterotic duality, Type IIA/heterotic duality, heterotic/Type IIA duality and Type IIA/Type IIA duality follow [16], S 1 /Z2 T 4 and S 1 T 4 , respectively. dimensional origin of string theory? from membrane/vebrane duality by compactifying M-theory on S 1 /Z2 K3 [15], S 1 K3 First, however, I want to pose the question:Should we have been surprised by the eleven-

Type II A&M theory

The importance of eleven dimensions is no doubt surprising from the point of view of perturbative string theory; from the point of view of membrane theory, however, there were already tantalizing hints in this direction: (i) K3 compactication The four-dimensional compact manifold K3 plays a ubiquitous role in much of present day M-theory. It was rst introduced as a compactifying manifold in 1983 [18] when it was realised that the number of unbroken supersymmetries surviving compactication in a Kaluza-Klein theory depends on the holonomy group of the extra dimensions. By virtue of its SU(2) holonomy, K3 preserves precisely half of the supersymmetry. This means, in particular, that an N = 2 theory on K3 has the same number of supersymmetries as an N = 1 theory on T 4 , a result which was subsequently to prove of vital importance for string/string and the D = 10 heterotic string on R10n T n [20] not only have the same supersymmetry SO(16 + n, n) SO(16 + n) SO(n) 2 duality. In 1986, it was pointed out [19] that D = 11 supergravity on R10n K3 T n3 [18]

but also the same moduli spaces of vacua, namely M=

(2.1)

It took almost a decade for this coincidence to be explained but we now know that Mtheory on R10n K3 T n3 is dual to the heterotic string on R10n T n . (ii) Superstrings in D=10 from supermembranes in D=11 Eleven dimensions received a big shot in the arm in 1987 when the D = 11 supermembrane was discovered [21]. The bosonic sector of its d = 3 worldvolume Green-Schwarz action is given by: S3 = T3 d3 1 1 ij i X M j X N GM N (X) + 2 2 1 ijk i X M j X N k X P CM N P (X) 3!

(2.2)

where T3 is the membrane tension, i (i = 1, 2, 3) are the worldvolume coordinates, ij is the worldvolume metric and X M () are the spacetime coordinates (M = 0, 1, . . . , 10). Kappa symmetry [21] then demands that the background metric GM N and background 3form potential CM N P obey the classical eld equations of D = 11 supergravity [2], whose bosonic action is I11 = 1 22 11 d11 x G RG 1 1 K4 2 2 4! 122 11 C 3 K4 K4 (2.3)

where K4 = dC3 is the 4-form eld strength. In particular, K4 obeys the eld equation 1 d K4 = K4 2 2 and the Bianchi identity dK4 = 0 (2.5) (2.4)

10) Type IIA superstring follows by wrapping the (d = 3, D = 11) supermembrane around the circle in the limit that its radius R shrinks to zero. In particular, the Green-Schwarz action of the string follows in this way from the Green-Schwarz action of the membrane. It the time) unwanted Kaluza-Klein modes which had no place in weakly coupled Type IIA theory. The D = 10 dilaton, which governs the strength of the string coupling, is just a component of the D = 11 metric. A critique of superstring orthodoxy circa 1987, and its failure to accommodate the elevendimensional supermembrane, may be found in [23]. 3 was necessary to take this R 0 limit in order to send to innity the masses of the (at

It was then pointed out [22] that in an R10 S 1 topology the weakly coupled (d = 2, D =

(iii) U-duality (when it was still non-U) Based on considerations of this D = 11 supermembrane, which on further compactication treats the dilaton and moduli elds on the same footing, it was conjectured [26] in 1990 that discrete subgroups of all the old non-compact global symmetries of compactied supergravity [24, 25] (e.g SL(2, R), O(6, 6), E7 ) should be promoted to duality symmetries of the supermembrane. Via the above wrapping around S 1 , therefore, they should be also be inherited by the Type IIA string [26]. (iv) D=11 membrane/vebrane duality In 1991, the supermembrane was recovered as an elementary solution of D = 11 supergravity which preserves half of the spacetime supersymmetry [27]. Making the three/eight split X M = (x , y m ) where = 0, 1, 2 and m = 3, ..., 10, the metric is given by ds2 = (1 + k3 /y 6)2/3 dx dx + (1 + k3 /y 6)1/3 (dy 2 + y 2d7 2 ) and the four-form eld strength by K7 K4 = 6k3 7 where the constant k3 is given by (2.7) (2.6)

211 2 T3 (2.8) 7 Here 7 is the volume form on S 7 and 7 is the volume. The mass per unit area of the k3 = membrane M3 is equal to its tension: M3 = T3 (2.9)

This elementary solution is a singular solution of the supergravity equations coupled to a supermembrane source and carries a Noether electric charge Q= 1 211
S7

(K4 + C3 K4 ) =

211 T3

(2.10)

Hence the solution saturates the Bogomolnyi bound

of the preservation of half the supersymmetries which is also intimately linked with the worldvolume kappa symmetry. The zero modes of this solution belong to a (d = 3, n = 8) supermultiplet consisting of eight scalars and eight spinors (I , I ), with I = 1, ..., 8, which correspond to the eight Goldstone bosons and their superpartners associated with breaking of the eight translations transverse to the membrane worldvolume. 4

211 M3 Q. This is a consequence

In 1992, the supervebrane was discovered as a soliton solution of D = 11 supergravity also preserving half the spacetime supersymmetry [28]. Making the six/ve split X M = (x , y m) where = 0, 1, 2, 3, 4, 5 and m = 6, ..., 10, the metric is given by ds2 = (1 + k6 /y 3)1/3 dx dx + (1 + k6 /y 3)2/3 (dy 2 + y 2d4 2 ) and the four-index eld-strength by K4 = 3k6 4 where the vebrane tension T6 is related to the constant k6 by 211 2 T6 k6 = 34 (2.13) (2.12) (2.11)

Here 4 is the volume form on S 4 and 4 is the volume. The mass per unit 5-volume of the vebrane M6 is equal to its tension: M6 = T6 (2.14)

This solitonic solution is a non-singular solution of the source-free equations and carries a topological magnetic charge P = 1 211 K4 = 211 T6 (2.15)

S4

Hence the solution saturates the Bogomolnyi bound

consequence of the preservation of half the supersymmetries. The covariant action for this D = 11 supervebrane is still unknown (see [29, 30] for recent progress) but consideration of the soliton zero modes [31, 14, 32] means that the gauged xed action must be described

211 M6 P . Once again, this is a

by the same chiral antisymmetric tensor multiplet (B , I , [IJ] ) as that of the Type IIA vebrane [33, 34]. Note that in addition to the ve scalars corresponding to the ve translational Goldstone bosons, there is also a 2-form B whose 3-form eld strength is anti-self dual and which describes three degrees of freedom. The electric and magnetic charges obey a Dirac quantization rule [35, 36] QP = 2n Or, in terms of the tensions [37, 11], 211 2 T3 T6 = 2n 5 (2.17) n = integer (2.16)

This naturally suggests a D = 11 membrane/vebrane duality. Note that this reduces the three dimensionful parameters T3 , T6 and 11 to two. Moreover, it was recently shown [16] that they are not independent. To see this, we note from (2.2) that C3 has period 2/T3 so that K4 is quantized according to K4 = 2n n = integer T3 (2.18)

Consistency of such C3 periods with the spacetime action, (2.3), gives the relation5 (2)2 2Z 3 11 2 T3 From (2.17), this may also be written as 2 T6 Z T3 2 (2.20) (2.19)

Thus the tension of the singly charged vebrane is given by 1 2 T6 = T3 2 (v) Hidden eleventh dimension We have seen how the D = 10 Type IIA string follows from D = 11. Is it possible to go the other way and discover an eleventh dimension hiding in D = 10? In 1993, it was recognized [40] that by dualizing a vector into a scalar on the gauge-xed d = 3 worldvolume of the Type IIA supermembrane, one increases the number of worldvolume scalars (i.e transverse dimensions) from 7 to 8 and hence obtains the corresponding worldvolume action of the D = 11 supermembrane. Thus the D = 10 Type IIA theory contains a hidden D = 11 Lorentz invariance! This device was subseqently used [41, 42] to demonstrate the equivalence of the actions of the D = 10 Type IIA membrane and the Dirichlet twobrane [43]. (vi) U-duality Of the conjectured Cremmer-Julia symmetries referred to in (iii) above, the case for a target space O(6, 6; Z) (T -duality) in perturbative string theory had already been made, of course [44]. Stronger evidence for an SL(2, Z) (S-duality) in string theory was subsequently provided in [6, 7] where it was pointed out that it corresponds to a non-perturbative electric/magnetic symmetry.
5

(2.21)

This corrects a factor of two error in [16] and brings us into agreement with a subsequent D-brane

derivation [38] of (2.21). I am grateful to Shanta De Alwis [39] for pointing out the source of the error.

In 1994, stronger evidence for the combination of S and T into a discrete duality of Type II strings, such as E7 (Z) in D = 4, was provided in [13], where it was dubbed Uduality. Moreover, the BPS spectrum necessary for this U-duality was given an explanation in terms of the wrapping of either the D = 11 membrane or D = 11 vebrane around the extra dimensions. This paper also conjectured a non-perturbative SL(2, Z) of the Type IIB string in D = 10. (vii) Black Holes In 1995, it was conjectured [32] that the D = 10 Type IIA superstring should be identied with the D = 11 supermembrane compactied on S 1 , even for large R. The D = 11 KaluzaKlein modes (which, as discussed in (ii) above, had no place in the perturbative Type IIA theory) were interpreted as charged extreme black holes of the Type IIA theory. (viii) D=11 membrane/vebrane duality and anomalies Membrane/vebrane duality interchanges the roles of eld equations and Bianchi identities. From (2.4), the vebrane Bianchi identity reads 1 d K7 = K 4 2 2 (2.22)

However, it was recognized in 1995 that such a Bianchi identity will in general require gravitational Chern-Simons corrections arising from a sigma-model anomaly on the vebrane worldvolume [16] 1 2 2 d K 7 = K4 + X 8 2 T6 (2.23)

where the 8-form polynomial X8 , quartic in the gravitational curvature R, describes the Lorentz d = 6 worldvolume anomaly of the D = 11 vebrane. Although the covariant vebrane action is unknown, we know that the gauge xed theory is described by the chiral
antisymmetric tensor multiplet (B , I , [IJ]), and it is a straightforward matter to read o

the anomaly polynomial from the literature. See, for example [45]. We nd X8 = 1 1 1 [ (trR2 )2 + trR4 ] 4 (2) 768 192 (2.24)

Thus membrane/vebrane duality predicts a spacetime correction to the D = 11 supermembrane action [16] I11 (Lorentz) = T3 C3 1 1 1 [ (trR2 )2 + trR4 ] 4 (2) 768 192 (2.25)

Such a correction was also derived in a somewhat dierent way in [17]. This prediction is intrinsically M-theoretic, with no counterpart in ordinary D = 11 supergravity. However, by simultaneous dimensional reduction [22] of (d = 3, D = 11) to (d = 2, D = 10) on S 1 , it translates into a corresponding prediction for the Type IIA string: I10 (Lorentz) = T2 B2 1 1 1 [ (trR2 )2 + trR4 ] 4 (2) 768 192 (2.26)

where B2 is the string 2-form and T2 = 1/2 is the string tension. As a consistency check we can compare this prediction with previous results found by explicit string one-loop calculations. These have been done in two ways: either by computing directly in D = 10 the Type IIA anomaly polynomial [46] following [47], or by compactifying to D = 2 on an 8-manifold M and computing the B2 one-point function [48]. We indeed nd agreement. Thus using D = 11 membrane/vebrane duality we have correctly reproduced the corrections to the B2 eld equations of the D = 10 Type IIA string (a mixture of treelevel and string one-loop eects) starting from the Chern-Simons corrections to the Bianchi identities of the D = 11 supervebrane (a purely tree-level eect). It would be interesting to know, on the membrane side, what calculation in D = 11 M-theory, when reduced on S 1 , corresponds to this one-loop Type IIA string amplitude calculation in D = 10. Understanding this may well throw a good deal of light on the mystery of what M-theory really is! (ix) Heterotic string from vebrane wrapped around K3 In 1995 it was shown that, when wrapped around K3 with its 19 self-dual and 3 anti-selfdual 2-forms, the d = 6 worldvolume elds of the D = 11 vebrane (or Type IIA vebrane) (B , I , [IJ] ) reduce to the d = 2 worldsheet elds of the heterotic string in D = 7 (or D = 6) [49, 50]. The 2-form yields (19, 3) left and right moving bosons, the spinors yield (0, 8) fermions and the scalars yield (5, 5) which add up to the correct worldsheet degrees of freedom of the heterotic string [49, 50]. A consistency check is provided [16] by the derivation of the Yang-Mills and Lorentz Chern-Simons corrections to the Bianchi identity of the heterotic string starting from the vebrane Bianchi identity given in (viii) . Making the seven/four split X M = (x , y m) where = 0, ..., 6 and m = 7, 8, 9, 10, the original set of D = 11 elds may be decomposed in a basis of harmonic p-forms on K3. In particular, we expand C3 as C3 (X) = C3 (x) + 1 2T3 8
I I C1 (x)2 (y)

(2.27)

I where 2 , I = 1, . . . , 22 are an integral basis of b2 harmonic two-forms on K3. Following [12], let us dene the dual string 3-form H3 by

T2 H3 = T6

K3

K7 ,

(2.28)

The dual string Lorentz anomaly polynomial, X4 , is given by X4 =

X8 =

1 1 trR2 p1 (K3) 2 192 (2) (2.29)

where p1 (K3) is the Pontryagin number of K3 p1 (K3) = 1 8 2

(2.30)

We may now integrate (2.23) over K3, using (2.21) to nd d H3 = I J [K K dIJ + trR2 ] 4 2 2 (2.31)

I I where K2 = dC1 and where dIJ is the intersection matrix on K3, given by

dIJ =

K3

I J 2 2

(2.32)

which has b+ = 3 positive and b = 19 negative eigenvalues. Thus we see that this form of 2 2 the Bianchi identity corresponds to a D = 7 toroidal compactication of a heterotic string at a generic point on the Narain lattice [20]. Thus we have reproduced the D = 7 Bianchi identity of the heterotic string, starting from the D = 11 vebrane. For use in section (3), we note that if we replace K3 by T 4 in the above derivation, the 2-form now yields (3, 3) left and right moving bosons, the spinors now yield (8, 8) fermions and the scalars again yield (5, 5) which add up to the correct worldsheet degrees of freedom of the Type IIA string. In this case, the Bianchi identity becomes dH3 = 0 as it should be. (x) N=1 in D=4 Also in 1995 it was noted [51, 52, 53, 55, 54, 56, 64] that N = 1 heterotic strings can be dual to D = 11 supergravity compactied on seven-dimensional spaces of G2 holonomy which also yield N = 1 in D = 4 [57]. (xi) Non-perturbative eects Also in 1995 it was shown [58] that membranes and vebranes of the Type IIA theory, obtained by compactication on S 1 , yield e1/gs eects, where gs is the string coupling. 9

(xii) SL(2,Z) with M-theory on R9 T 2 [38, 64]. In particular, the conjectured SL(2, Z) of the Type IIB (xiii) E8 E8 heterotic string E8 E8 heterotic string
7

Also in 1995, strong evidence was provided for identifying the Type IIB string on R9 S 1

theory discussed in (vi) above is just the modular group of the M-theory torus6

Also in 1995 (that annus mirabilis!), strong evidence was provided for identifying the on R10 with M-theory on R10 S 1 /Z2 [67].

This completes our summary of M-theory before M-theory was cool. The phrase M-

theory (though, as I hope to have shown, not the physics of M-theory) rst made its appearance in October 1995 [38, 67]. This was also the month that it was proposed [43] that the Type II p-branes carrying Ramond-Ramond charges can be given an exact conformal eld theory description via open strings with Dirichlet boundary conditions, thus heraldbeen appearing daily on the internet, including applications to black holes [62], length scales the review by Schwarz [63] for these more recent developments in M-theory, to the review by Polchinski [69] for developments in D-branes and to the paper by Aharony, Sonnenschein and Yankielowicz [70] for the connection between the two (since D-branes are intrinsically ten-dimensional and M-theory is eleven-dimensional, this is not at all obvious). Here, we wish to focus on a specic application of M-theory, namely the derivation of string/string dualities. ing the era of D branes. Since then, evidence in favor of M-theory and D-branes has

shorter than the string scale [59] and even phenomenology [60, 61]. We refer the reader to

String/string duality from M-theory

Let us consider M-theory, with its fundamental membrane and solitonic vebrane, on R6 M1 M4 where M1 is a one-dimensional compact space of radius R and M4 is a four6

Two alternative explanations of this SL(2, Z) had previously been given: (a) identifying it with the

S-duality [16] of the d = 4 Born-Infeld worldvolume theory of the self-dual Type IIB superthreebrane [65], and (b) using the four-dimensional heterotic/Type IIA/Type IIB triality [66] by noting that this SL(2, Z), while non-perturbative for the Type IIB string, is perturbative for the heterotic string. 7 It is ironic that, having hammered the nal nail in the con of D = 11 supergravity by telling us that it can never yield a chiral theory when compactied on a manifold [68], Witten pulls it out again by telling us that it does yield a chiral theory when compactied on something that is not a manifold!

10

dimensional compact space of volume V . We may obtain a fundamental string on R6 by wrapping the membrane around M1 and reducing on M4 . Let us denote fundamental string sigma-model metrics in D = 10 and D = 6 by G10 and G6 . Then from the corresponding Einstein Lagrangians G11 R11 = R3 G10 R10 = V G6 R6 R (3.1)

we may read o the strength of the string couplings in D = 10 [15] 10 2 = R3 and D = 6 6 2 = R V (3.3) (3.2)

Similarly we may obtain a solitonic string on R6 by wrapping the vebrane around M4 and reducing on M1 . Let us denote the solitonic string sigma-model metrics in D = 7 and D = 6 by G7 and G6 . Then from the corresponding Einstein Lagrangians R G11 R11 = V 3/2 G7 R7 = G6 R6 V we may read o the strength of the string couplings in D = 7 [15] 2 = V 3/2 7 and D = 6 V 2 = 6 R (3.6) (3.5) (3.4)

Thus we see that the fundamental and solitonic strings are related by a strong/weak coupling: 2 = 1/6 2 6 (3.7)

We shall be interested in M1 = S 1 (in which case from (ii) of section (2) the fundamental string will be Type IIA) or M1 = S 1 /Z2 (in which case from (xiii) of section (2) the fundamental string will be heterotic E8 E8 ). Similarly, we will be interested in M4 = T 4

(in which case from (ix) of section (2) the solitonic string will be Type IIA) or M4 = K3 (in which case from (ix) of section (2) the solitonic string will be heterotic). Thus there are four possible scenarios which are summarized in Table 1. (N+ , N ) denotes the D = 6 spacetime supersymmetries. In each case, the fundamental string will be weakly coupled as we shrink 11

(N+ , N ) (1, 0) (1, 1) (1, 1) (2, 2)

M1

M4 fundamental string dual string heterotic T ype IIA heterotic T ype IIA heterotic heterotic T ype IIA T ype IIA

S 1 /Z2 K3 S1 S 1 /Z2 S1 K3 T4 T4

Table 1: String/string dualities the size of the wrapping space M1 and the dual string will be weakly coupled as we shrink the size of the wrapping space M4 . In fact, there is in general a topological obstruction to wrapping the vebrane around M4 provided by (2.18) because the vebrane cannot wrap around a 4-manifold that has n = 08 . This is because the anti-self-dual 3-form eld strength T on the worldvolume of the vebrane obeys [41, 17] dT = K4 (3.8)

and the existence of a solution for T therefore requires that K4 must be cohomologically K3, with instanton number k in one E8 and 24 k in the other, however, the ux of K4 n = 12 k trivial. For M-theory on R6 S 1 /Z2 T 4 this is no problem. For M theory on R6 S 1 /Z2

over K3 is [15]

(3.9)

Consequently, the M-theoretic explanation of heterotic/heterotic duality requires E8 E8

with the symmetric embedding k = 12. This has some far-reaching implications. For example, the duality exchanges gauge elds that can be seen in perturbation theory with gauge elds of a non-perturbative origin [15]. The dilaton , the string -model metric GM N and 3-form eld strength H of the dual

string are related to those of the fundamental string, , GM N and H by the replacements [11, 12] = GM N GM N = e GM N
8

Actually, as recently shown in [71], the object which must have integral periods is not T3 K4 /2 but

rather T3 K4 /2 p1 /4 where p1 is the rst Pontryagin class. This will not aect our conclusions, however.

12

 H H = e H

(3.10)

In the case of heterotic/Type IIA duality and Type IIA/heterotic duality, this operation takes us from one string to the other, but in the case of heterotic/heterotic duality and Type IIA/Type IIA duality this operation is a discrete symmetry of the theory. This Type IIA/Type IIA duality is discussed in [78] and we recognize this symmetry as subgroup of the SO(5, 5; Z) U-duality [26, 13, 79] of the D = 6 Type IIA string. Vacua with (N+ , N ) = (1, 0) in D = 6 have been the subject of much interest lately. In are also the GP vacua [72, 73, 74] obtained from the SO(32) theory on K3 and the MV vacua [85, 86] obtained from F -theory [75] on Calabi-Yau. Indeed, all three categories are related by duality [80, 85, 83, 97, 84, 76, 86, 81]. In particular, the DMW heterotic strong/weak coupling duality gets mapped to a T -duality of the Type I version of the SO(32) theory, and the non-perturbative gauge symmetries of the DMW model arise from small Spin(32)/Z2 instantons in the heterotic version of the SO(32) theory [76]. Because heterotic/heterotic duality interchanges worldsheet and spacetime loop expansions or because it acts by duality on H the duality exchanges the tree level Chern-Simons contributions to the Bianchi identity dH = (2)2 X4 X4 = 1 [trR2 v trF 2 ] 4(2)2 (3.11) addition to DMW vacua [16] discussed above, obtained from M-theory on S 1 /Z2 K3, there

with the one-loop Green-Schwarz corrections to the eld equations dH = (2)2 X4 X4 = 1 [trR2 v trF 2 ] 4(2)2 (3.12)

Here F is the eld strength of the th component of the gauge group, tr denotes the trace in the fundamental representation, and v , v are constants. In fact, the Green-Schwarz anomaly cancellation mechanism in six dimensions requires that the anomaly eight-form I8 factorize as a product of four-forms, I8 = X4 X4 , (3.13)

and a six-dimensional string-string duality with the general features summarized above would exchange the two factors [12]. Moreover, supersymmetry relates the coecients v , v to the 13

gauge eld kinetic energy. In the Einstein metric Gc M N = e/2 GM N , the exact dilaton dependence of the kinetic energy of the gauge eld FM N , is [96] Lgauge = (2)3 c G v e/2 + v e/2 trFM N F M N . 8 (3.14)

So whenever one of the v is negative, there is a value of the dilaton for which the coupling constant of the corresponding gauge group diverges. This is believed to signal a phase transition associated with the appearance of tensionless strings [88, 89, 90]. This does not happen for the symmetric embedding discussed above since the perturbative gauge elds have v > 0 and v = 0 and the non-perturbative gauge elds have v = 0 and v > 0. Another kind of heterotic/heterotic duality may arise, however, in vacua where one may Higgs away that subset of gauge elds with negative v , and be left with gauge elds with v = v > 0. This happens for the non-symmetric embedding k = 14 and the appearance of non-perturbative gauge elds is not required [80, 81, 82, 85, 86]. Despite appearances, it known from F -theory that the k = 12 and k = 14 models are actually equivalent [85, 86]. Vacua with (N+ , N ) = (2, 0) arising from Type IIB on K3 also have an M-theoretic description, in terms of compactication on T 5 /Z2 [87, 17].

Four dimensions

It is interesting to consider further toroidal compactication to four dimensions, replacing R6 Higgsed, the toroidal compactication to four dimensions gives an N = 2 theory with the usual three vector multiplets S, T and U related to the four-dimensional heterotic string coupling constant and the area and shape of the T 2 . When reduced to four dimensions, the six-dimensional string-string duality (3.10) becomes [8] an operation that exchanges S symmetry. This self-duality of heterotic string vacua does not rule out the the possibility that in D = 4 they are also dual to Type IIA strings compactied on Calabi-Yau manifolds. In fact, as discussed in [93], when the gauge group is completely Higgsed, obvious candidates are provided by Calabi-Yau manifolds with hodge numbers h11 = 3 and h21 = 243, since these have the same massless eld content. Moreover, these manifolds do indeed exhibit the S T interchange symmetry [92, 91, 94]. Since the heterotic string on T 2 K3 also 14 and T , so in the case of heterotic/heterotic duality we have a discrete S T interchange by R4 T 2 . Starting with a K3 vacuum in which the E8 E8 gauge symmetry is completely

symmetry is spoiled by non-perturbative eects [95, 98].

triality symmetry, as discussed in [66]. In all known models, however, the T U interchange An interesting aspect of the Calabi-Yau manifolds X appearing in the duality between

has R to 1/R symmetries that exchange T and U, one might expect a complete S T U

heterotic strings on K3T 2 and Type IIA strings on X, is that they can always be written in the form of a K3 bration [92]. Once again, this ubiquity of K3 is presumably a consequence of the interpretation of the heterotic string as the K3 wrapping of a vebrane. Consequently, if X admits two dierent K3 brations, this would provide an alternative explanation for heterotic dual pairs in four dimensions [83, 85, 86] and this is indeed the case for the CalabiYau manifolds discussed above.

Eleven to twelve: is it still too early?

The M-theoretic origin of the Type IIB string given in (xii) of section 2 seems to require going down to nine dimensions and then back up to ten. An obvious question, therefore, is whether Type IIB admits a more direct higher-dimensional explanation, like Type IIA. Already in 1987 it was suggested [99] that the (1, 1)-signature worldsheet of the Type IIB string, moving in a (9, 1)-signature spacetime, may be descended from a supersymmetric extended object with a (2, 2)-signature worldvolume, moving in a (10, 2)-signature spacetime. This idea becomes even more appealing if one imagines that the SL(2, Z) of the Type IIB theory [13] might correspond to the modular group of a T 2 compactication from D = 12 to D = 10 just as the SL(2, Z) of S-duality corresponds to the modular group of a T 2 compactication from D = 6 to D = 4 [8]. In view of our claims that D = 11 is the maximum spacetime dimension admitting a consistent supersymmetric theory, however, this twelve dimensional idea requires some explanation. So let us begin by recalling the D = 11 argument. As a p-brane moves through spacetime, its trajectory is described by the functions X M () where X M are the spacetime coordinates (M = 0, 1, . . . , D 1) and i are the worldvolume choice by making the D = d + (D d) split coordinates (i = 0, 1, . . . , d 1). It is often convenient to make the so-called static gauge

X M () = (X (), Y m ()),

(5.1)

15

where = 0, 1, . . . , d 1 and m = d, . . . , D 1, and then setting X () = . (5.2)

Thus the only physical worldvolume degrees of freedom are given by the (D d) Y m (). So the number of on-shell bosonic degrees of freedom is NB = D d.

(5.3)

To describe the super p-brane we augment the D bosonic coordinates X M () with anticommuting fermionic coordinates (). Depending on D, this spinor could be Dirac, Weyl, Majorana or Majorana-Weyl. The fermionic -symmetry means that half of the spinor degrees of freedom are redundant and may be eliminated by a physical gauge choice. The net result is that the theory exhibits a d-dimensional worldvolume supersymmetry [100] where the number of fermionic generators is exactly half of the generators in the original spacetime supersymmetry. This partial breaking of supersymmetry is a key idea. Let M be the number of real components of the minimal spinor and N the number of supersymmetries in D spacetime dimensions and let m and n be the corresponding quantities in d worldvolume dimensions. Since -symmetry always halves the number of fermionic degrees of freedom and (for d > 2) going on-shell halves it again, the number of on-shell fermionic degrees of freedom is NF = 1 1 mn = MN. 2 4 1 1 mn = MN. 2 4 (5.4)

Worldvolume supersymmetry demands NB = NF and hence Dd= (5.5)

We note in particular that Dmax = 11 since M = 32 for D = 11 and we nd the supermembrane with d = 3. For D 12, M 64 and hence (5.5) cannot be satised. Actually, the described by scalar supermultiplets. There are also p-branes with vector and/or antisymmetric tensor supermultiplets on the worldvolume [33, 34, 40], but repeating the argument still yields Dmax = 11 where we nd a supervebrane with d = 6 [14]. The upper bound of D = 11 is thus a consequence of the jump from M = 32 to M = 64 in going from D = 11 to D = 12. However, this jump can be avoided if one is willing to pay the price of changing the signature to (10, 2) where it is possible to dene a spinor 16 above argument is strictly valid only for p-branes whose worldvolume degrees of freedom are

which is both Majorana and Weyl. A naive application of the above bose-fermi matching argument then yields Dmax = 12 where we nd an extended object with d = 4 but with (2, 2) signature [99]. The chiral nature of this object then naturally suggests a connection with the Type IIB string in D = 10, although the T 2 compactication would have to be of an unusual kind in order to preserve the chirality. Moreover, the chiral (N+ , N ) = (1, 0) supersymmetry algebra in (10, 2) involves the anti-commutator [101] {Q , Q } = M N PM N + M N P QRS Z + M N P QRS (5.6)

The absence of translations casts doubt on the naive application of the bose-fermi matching argument, and the appearance of the self-dual 6-form charge Z is suggestive of a sixbrane, rather than a threebrane. Despite all the objections one might raise to a world with two time dimensions, and despite the above problems of interpretation, the idea of a (2, 2) object moving in a (10, 2) spacetime has recently been revived in the context of F -theory [75], which involves Type IIB compactication where the axion and dilaton from the RR sector are allowed to vary on the internal manifold. Given a manifold M that has the structure of a ber bundle whose ber is T 2 and whose base is some manifold B, then F on M T ype IIB on B (5.7)

The utility of F -theory is beyond dispute and it has certainly enhanced our understanding of string dualities, but should the twelve-dimensions of F -theory be taken seriously? And if so, should F -theory be regarded as more fundamental than M-theory? Given that there seems to be no supersymmetric eld theory with SO(10, 2) Lorentz invariance [102], and given that the on-shell states carry only ten-dimensional momenta [75], the more conservative interpretation is that the twelfth dimension is merely a mathematical artifact and that F theory should simply be interpreted as a clever way of compactifying the IIB string [103]. Time will tell.

Conclusion

The overriding problem in superunication in the coming years will be to take the Mystery out of M-theory, while keeping the Magic and the Membranes. 17

References
[1] W. Nahm, Supersymmetries and their representations, Nucl. Phys. B135 (1978) 409. [2] E. Cremmer, B. Julia and J. Scherk, Supergravity in theory in 11 dimensions, Phys. Lett. 76B (1978) 409. [3] M. J. Du, B. E. W. Nilsson and C. N. Pope, Kaluza-Klein supergravity, Phys. Rep. 130 (1986) 1. [4] M. B. Green, J. H. Schwarz and E. Witten, Superstring Theory (Cambridge University Press, 1987). [5] E. Witten, String theory dynamics in various dimensions, Nucl. Phys. B443 (1995) 85. [6] A. Font, L. Ibanez, D. Lust and F. Quevedo, Strong-weak coupling duality and nonperturbative eects in string theory, Phys. Lett. B249 (1990) 35. [7] S.-J. Rey, The conning phase of superstrings and axionic strings, Phys. Rev. D43 (1991) 526. [8] M. J. Du, Strong/weak coupling duality from the dual string, Nucl. Phys. B442 (1995) 47. [9] M. J. Du and J. X. Lu, Loop expansions and string/ve-brane duality, Nucl. Phys. B357 (1991) 534. [10] M. J. Du and R. R. Khuri, Four-dimensional string/string duality, Nucl. Phys. B411 (1994) 473. [11] M. J. Du and J. X. Lu, Black and super p-branes in diverse dimensions, Nucl. Phys. B416 (1994) 301. [12] M. J. Du and R. Minasian, Putting string/string duality to the test, Nucl. Phys. B436 (1995) 507. [13] C. M. Hull and P. K. Townsend, Unity of superstring dualities, Nucl. Phys. B438 (1995) 109.

18

[14] M. J. Du, R. R. Khuri and J. X. Lu, String solitons, Phys. Rep. 259 (1995) 213. [15] M. J. Du, R. Minasian and E. Witten, Evidence for heterotic/heterotic duality, Nucl. Phys. B465 (1996) 413. [16] M. J. Du, J.T. Liu, R. Minasian, Eleven dimensional origin of string-string duality: a one-loop test, Nucl. Phys. B452 (1995) 261. [17] E. Witten, Fivebranes and M-theory on an orbifold, Nucl. Phys. B463 (1996) 383. [18] M. J. Du, B. E. W. Nilsson and C. N. Pope, Compactication of D = 11 supergravity on K3 T 3 , Phys. Lett. B129 (1983) 39. [19] M. J. Du and B. E. W. Nilsson, Four-dimensional string theory from the K3 lattice, Phys. Lett. B175 (1986) 417. [20] K. S. Narain, New heterotic string theories in uncompactied dimensions < 10, Phys. Lett. B169 (1986) 41. [21] E. Bergshoe, E. Sezgin and P. K. Townsend, Supermembranes and eleven-dimensional supergravity, Phys. Lett. B189 (1987) 75. [22] M. J. Du, P. S. Howe, T. Inami and K. S. Stelle, Superstrings in D = 10 from supermembranes in D = 11, Phys. Lett. B191 (1987) 70. [23] M. J. Du, Not the standard superstring review in Proceedings of the International School of Subnuclear Physics, The Superworld II, Erice, August 1987, (Ed. Zichichi, Plenum, 1990), CERN-TH.4749/87. [24] E. Cremmer, S. Ferrara and J. Scherk, SU(4) invariant supergravity theory, Phys. Lett. B74 (1978) 61. [25] E. Cremmer and B. Julia, The SO(8) supergravity, Nucl. Phys. B159 (1979) 141. [26] M. J. Du and J. X. Lu, Duality rotations in membrane theory, Nucl. Phys. B347 (1990) 394. [27] M. J. Du and K. S. Stelle, Multimembrane solutions of D = 11 supergravity, Phys. Lett. B253 (1991) 113. 19

[28] R. Gueven, Black p-brane solutions of d = 11 supergravity theory, Phys. Lett. B276 (1992) 49. [29] E. Bergshoe, M. de Roo and T. Ortin, hep-th/9606118. [30] P. S. Howe and E. Sezgin, Superbranes, hep-th/9607227. [31] G. W. Gibbons and P. K. Townsend, Vacuum interpolation in supergravity via super p-branes, Phys. Rev. Lett. 71 (1993) 3754. [32] P. K. Townsend, The eleven-dimensional supermembrane revisited, Phys. Lett. B350 (1995) 184. [33] C. G. Callan, J. A. Harvey and A. Strominger, World sheet approach to heterotic instantons and solitons, Nucl. Phys. B359 (1991) 611. [34] C. G. Callan, J. A. Harvey and A. Strominger, Worldbrane actions for string solitons, Nucl. Phys. B367 (1991) 60. [35] R. I. Nepomechie, Magnetic monopoles from antisymmetric tensor gauge elds, Phys. Rev. D31 (1985) 1921. [36] C. Teitelboim, Monopoles of higher rank, Phys. Lett. B167 (1986) 63. [37] M. J. Du and J. X. Lu, Elementary vebrane solutions of D = 10 supergravity, Nucl. Phys. B354 (1991) 141. [38] J. H. Schwarz, The power of M-theory, Phys. Lett. B360 (1995) 13. [39] S. P. De Alwis, A note on brane tension and M-theory, hep-th/9607011. [40] M. J. Du and J. X. Lu, Type II p-branes: the brane scan revisited, Nucl. Phys. B390 (1993) 276. [41] P. Townsend, D-Branes From M-Branes, Phys. Lett. B373 (1996) 68. [42] C. Schmidhuber, D-brane actions, Nucl. Phys. B467 (1996) 146. The eleven-dimensional ve-brane,

20

[43] J. Polchinski, Dirichlet branes and Ramond-Ramond charges, Phys. Rev. Lett. 75 (1995) 4724. [44] A. Giveon, M. Porrati and E. Rabinovici, Target space duality in string theory, Phys. Rep. 244 (1994) 77. [45] L. Alvarez-Gaum and E. Witten, Gravitational Anomalies, Nucl. Phys. B234 (1983) e 269. [46] James T. Liu, , Private Communication. [47] A. N. Schellekens and N. P. Warner, , Nucl. Phys. B287 (1987) 317. [48] C. Vafa and E. Witten, A one-loop test of string duality, Nucl. Phys. B447 (1995) 261. [49] P. K. Townsend, String/membrane duality in seven dimensions, Phys. Lett. B354 (1995) 247. [50] J. A. Harvey and A. Strominger, The heterotic string is a soliton, Nucl. Phys. B449 (1995) 535. [51] A. C. Cadavid, A. Ceresole, R. DAuria and S. Ferrara, 11-dimensional supergravity compactied on Calabi-Yau threefolds, hep-th/9506144. [52] G. Papadopoulos and P. K. Townsend, Compactication of D=11 Supergravity on spaces with exceptional holonomy, Phys. Lett. B357 (1995) 300. [53] J. H. Schwarz and A. Sen, Type IIA dual of six-dimensional CHL compactications, Nucl. Phys. B454 (1995) 427. [54] S. Chaudhuri and D. A. Lowe, Type IIA-Heterotic duals with maximal supersymmetry, Nucl. Phys. B459 (1996) 113. [55] J. A. Harvey, D. A. Lowe and A. Strominger, N = 1 string duality, Phys. Lett. B362 (1995) 65. [56] B. S. Acharya, hep-th/9508046 N = 1 Heterotic-supergravity duality and Joyce manifolds,

21

[57] M. A. Awada, M. J. Du and C. N. Pope, N = 8 supergravity breaks down to N = 1, Phys. Rev. Lett. 50 (1983) 294. [58] K. Becker, M. Becker and A. Strominger, Fivebranes, membranes and non-perturbative string theory, Nucl. Phys. B456 (1995) 130. [59] M. R. Douglas, D. Kabat, P. Pouliot and S. H. Shenker, D-branes and short distances in string theory, hep-th/9608024. [60] T. Banks and M. Dine, Couplings and scales in strongly coupled heterotic string theory, hep-th/9605136. [61] P. Horava, Gluino condensation in strongly coupled heterotic string theory,

hep-th/9608019. [62] C. Vafa and A. Strominger, Microscopic origin of the Bekenstein-Hawking entropy, hep-th/9601029. [63] J. H. Schwarz, Lectures on Superstring and M-theory dualities, hep-th/9607201. [64] P. S. Aspinwall, Some relationships between dualities in string theory, hep-th/9508154. [65] M. J. Du and J. X. Lu, The self-dual Type IIB superthreebrane, Phys. Lett. B273 (1991) 409. [66] M. J. Du, J.T. Liu and J. Rahmfeld, Four-dimensional string/string/string triality, Nucl. Phys. B459 (1996) 125. [67] P. Horava and E. Witten, Heterotic and Type I string dynamics from eleven dimensions, Nucl. Phys. B460 (1996) 506. [68] E. Witten, Fermion quantum numbers in Kaluza-Klein theory in Proceedings of the 1983 Shelter Island Conference on Quantum Field Theory and the Fundamental Problems in Physics, June 1983, (Eds. Jackiw, Khuri and Weinberg, MIT Press, 1985). [69] J. Polchinski, String Duality, hep-th/9607050. [70] O. Aharony, J. Sonnenschein and S. Yankielowicz, Interactions of strings and D-branes from M-theory, hep-th/9603009. 22

[71] E. Witten, On ux quantization in M-theory and the eective action, hep-th/9609122. [72] G. Pradisi and A. Sagnotti, Open string Orbifolds Phys. Lett. B216 (1989) 59. [73] M. Bianchi and A. Sagnotti, Twist symmetry and open string Wilson lines Nucl. Phys. B361 (1991) 519. [74] E. G. Gimon and J. Polchinski, Consistency conditions for orientifolds and D-manifolds Phys. Rev. D54 (1996) 1667. [75] C. Vafa, Evidence for F -theory, Nucl. Phys. B469 (1996) 403. [76] M. Berkooz, R. G. Leigh, J. Polchinski, J. H. Schwarz, N. Seiberg and E. Witten, Anomalies, dualities and topology of D = 6 N = 1 superstring vacua, hep-th/9605184. [77] E. Witten, Small instantons in string theory, Nucl. Phys. B460 (1996) 541. [78] A. Sen and C. Vafa, Dual pairs of Type IIA string compactications , Nucl. Phys. B455 (1995) 165. [79] R. Dijkgraaf, E. Verlinde and H. Verlinde, hep-th/9604055. [80] G. Aldazabal, A. Font, L. E. Ibanez and F. Quevado, Heterotic/heterotic duality in D = 6, 4, Phys. Lett. B380 (1996) 42. [81] G. Aldazabal, A. Font, L. E. Ibanez and A. M. Uranga, New branches of string compactications and their F -theory duals, hep-th/9607121. [82] E. Gimon and C. V. Johnson, Multiple realisations of N = 1 vacua in six dimensions, NSF-ITP-96-55 hep-th/9606176. [83] P. S. Aspinwall and M. Gross, Heterotic-heterotic string duality and multiple K3 brations, hep-th/9602118. [84] S. Ferrara, R. Minasian and A. Sagnotti, Low-energy analysis of M and F theories on Calabi-Yau manifolds, hep-th/9604097. [85] D. R. Morrison and C. Vafa, Compactications of F -theory on Calabi-Yau threefolds-I, hep-th/9602114. 23 BPS quantization of the ve-brane,

[86] D. R. Morrison and C. Vafa, Compactications of F -theory on Calabi-Yau threefolds-II, hep-th/9603161. [87] K. Dasgupta and S. Mukhi, Orbifolds of M-theory, Nucl. Phys. B465 (1996) 399. [88] O. J. Ganor and A. Hanany, Small E8 instantons and tensionless non-critical strings, hep-th/9603003. [89] N. Seiberg and E. Witten, Comments on string dynamics in six dimensions, hepth/9603003. [90] M. J. Du, H. Lu and C. N. Pope, Heterotic phase transitions and singularities of the gauge dyonic string, hep-th/9603037. [91] M. J. Du, Electric/magnetic duality and its stringy origins, hep-th/9509106. [92] A. Klemm, W. Lerche and P. Mayr, hep-th/9506112. [93] S. Kachru and C. Vafa, Exact results for N=2 compactications of heterotic strings, hep-th/9505105. [94] G. Lopez Cardoso, D. Lust and T. Mohaupt, Non-perturbative monodromies in N=2 heterotic string vacua, hep-th/9504090. [95] J. Louis, J. Sonnenschein, S. Theisen and S. Yankielowicz, Non-perturbative properties of heterotic string vacua compactied on K3 T 2 , hep-th/9606049. [96] A. Sagnotti, A note on the Green-Schwarz mechanism in open-string theories, Phys. Lett. B294 (1992) 196. [97] E. Witten, Phase transitions in M-theory and F -theory, hep-th/9603150. [98] G. Lopez Cardoso, G. Curio, D. Lust and T. Mohaupt, Instanton numbers and exchange symmetries in N = 2 dual string pairs, hep-th/9603108. [99] M. J. Du and M. Blencowe, Supermembranes and the signature of spacetime, Nucl. Phys. B310 387 (1988). K3-brations and heterotic-Type-II duality,

24

[100] A. Achucarro, J. M. Evans, P. K. Townsend and D. L. Wiltshire , Super p-branes, Phys. Lett. B198 (1997) 441. [101] I. Bars, Supersymmetry, P -brane duality and hidden dimensions, hep-th/9604139. [102] H. Nishino and E. Sezgin, Supersymmetric Yang-Mills equations in 10 + 2 dimensions, hep-th/9607185. [103] A. Sen, F -theory and orientifolds, hep-th/9605150.

25