David Berenstein - Supersymmetry: A String Theory Point of View
David Berenstein - Supersymmetry: A String Theory Point of View
David Berenstein - Supersymmetry: A String Theory Point of View
David Berenstein
Department of Physics, University of California at Santa Barbara, Santa Barbara, CA 93106, U.S.A. Abstract. In this article I describe recent advances that show the relationship between string theory and supersymmetry, and how string theory ideas have revolutionized our understanding of supersymmetric eld theories
1. INTRODUCTION
These lecture notes were presented at the XXXV Latin American School of Physics, whose topic was on Supersymmetries and its physical applications. The course lasted for ve one hour sessions. The original title that I was given to work around was "Supersymmetry in particle physics, string theory and cosmology" . I was clearly overwhelmed by the wealth on information that is available on each one of the three topics. So I decided to go with just one of the three, the one I am most familiar with, string theory, and try to keep in mind the other two as the lectures progressed. I decided that my purpose for the lectures was to explore some modern aspects of four dimensional eld theories and how these results are inspired and related to geometrical ideas in string theory. I also wanted to give a avor of what ingredients are being used to understand the problem of the cosmological constant and supersymmetry breaking, as well as the string theory ideas that are inspiring new trends in the study of phenomenological models for particle physics beyond the standard model. The choice of topics is by no means complete and reects some personal choice on what I believe are very interesting phenomena that have been found in string theory, mostly from familiarity with subjects themselves, together with constraints given by time limits. I also tried perhaps to no avail to refrain from being too technical on most parts, but not at the expense of ruining the understanding of ideas. Because of this, on some sections these lectures are very technical. Hopefully the ideas that are meant to be illustrated with the technical sections are put in a better light with these sections. The choice of references is also for the purpose of illustrating ideas and not meant to be exhaustive. I hope I have not missed giving credit where appropriate. Many facts can be learned on review papers, of which I have cited a few. The lectures are organized as follows. On lecture one I give an introduction to supersymmetry, and supersymmetric string theory in at space. I discuss how worldsheet supersymmetry on the string gives fermionic states in spacetime, and I discuss briey the superstrings with N 2 and N 1 SUSY in ten dimensions. I also introduce the notion of D-branes, whose physics occupies much of the rest of the talks. The second and third lecture will deal with max-
imally supersymmetric Yang Mills theory in four dimensions and the AdS/CFT correspondence. I will also introduce the notion of plane wave limits and I will discuss the integrable aspects of the planar diagram expansion. In lecture four I will present recent results in theories with N 1 SUSY in four dimensions which can be realized by placing D-branes on various geometries. Then I will describe how the vacuum structure of these theories is captured by studying random matrix models. Finally I will have some comments on recent attempts to understand the cosmological constant, supersymmetry breaking and the problem of nding metastable vacua in string theory.
The non-triviality of the extension of the Lorentz group is captured by the commutator
and
From these commutators it follows that the generator of time translations H, the Hamiltonian, is given by a positive denite operator
so that if there is a supersymmetric ground state, the energy of the ground state has to Q0 vanish. (A ground state is supersymmetric if Q 0 0). This fact has been argued as a natural mechanism to cancel the cosmological constant.
Q Q
Q Q
2. LECTURE 1
(1)
(2)
(3) (4)
(5)
One can have more than one supersymmetry charge, in which case we add an index to the supersymmetry charge to keep the distinction between the charges. The spacetime algebra is then J I Q Q IJ P (6) and we can also deform the Q Q commutator
I J Q Q
Z I J
where Z is called a central charge. This means Z commutes with the elements of the Lorentz group. Supersymmetries can be counted either by the total number of generators of SUSY, or by the total number of spinor charges. The rst counting is independent of dimension, while the second one depends on dimension. The convention is that N 1 SUSY in four dimensions corresponds to having four supersymmetries. Supersymmetry predicts that for every boson, there is a fermion with the same quantum numbers (except spin) and vice versa. In particular these are degenerate in mass. This is not seen in experiments. Indeed, a lot of familiar physics would be different if this were the case: electron orbits in the atom would not be stable. The electrons would decay to a lower orbit by emitting superpartners of the photon and becoming bosons. How can we reconcile this fact with low energy physics? The assumption underlying the equality of bosons and fermions is that supersymmetry is a symmetry of the ground state. However, it can be spontaneously broken. Other than being a non-trivial extension of the Lorentz group, we like supersymmetry for various technical properties of supersymmetric eld theories. If we consider loop diagrams, the fermions in the loop contribute with opposite sign to the bosons, providing natural cancelations of certain divergences and making the theory more nite. For example, a generic boson in a four dimensional renormalizable QFT has quadratic divergent corrections to its mass. However, a fermion only has logarithmic corrections to its mass. By the Bose-Fermi degeneracy, the boson mass can not receive larger corrections than the fermion. This means that these quadratic divergences cancel, and the mass gets small corrections with respect to its tree level value. This is the reason why Supersymmetry is a technically natural solution to the hierarchy problem. If at tree level the Higgs mass is small compared to some other scale (quantum gravity for example), then perturbative corrections can not make it large. Another reason for supersymmetry is that it ts the data for grand unication in the simplest models better than the corresponding theories without supersymmetry. Finally, a reason to be interested in SUSY is that it is predicted by many string theory models as part of the UV structure of the theory which makes the string theory consistent.
(7)
The minimal assumption about the dynamics of the string is that the action of the string is given by the induced volume of this surface in spacetime. We suggest reading [1, 2, 3, 4] as an introduction to strings and superstrings. By doing dimensional analysis we nd that the volume of this spacetime has engineering dimension equal to 2 ( we are counting powers of momenta), while a natural action has dimension zero. To compensate for units, we need to introduce a scale in the system. This is the string tension T
A second equivalent way to obtain the same classical physics is by introducing an auxiliary worldsheet metric , so that the action becomes
Notice that in the above action does not appear with derivatives, so that when we take its equation of motion we get constraints that let us solve for almost uniquely, and if we substitute the values of we get back the Nambu-Goto action. The action has a symmetry under local rescalings of exp f . This classical symmetry is the conformal invariance of the worldsheet action. This second version of the string action is due to Polyakov and it is the version we usually consider for quantizing the string. Strings where we have gravity on the worldsheet are called critical strings. If we want to quantize this action, we want to have the conformal symmetry to survive
$
"
d2
g x x
ls d2
(8)
det gind
(9)
(10)
the quantization. This leads to constraints from asking for the quantum corrections to the conformal symmetry to vanish. For at g this amounts to the dimension of spacetime being xed at 26. Fortunately, this is higher than four dimensions, so we can consider string theory on geometries where four of the 26 dimensions are non-compact and the other 22 are compact. For a more general metric, we get a quantum eld theory in two dimensions. The conformal invariance contains invariance under rescalings. This implies that the -model action must be invariant under the renormalization group equations. To rst order in the loop expansion on the worldsheet about a free eld theory the beta functions of g give rise to the vanishing of the Ricci tensor, so that consistency of the string theory predicts Einsteins equations. This is not assumed a priori, and it is one of the reasons why string theory is so exciting to study. We will now give a quick sketch of the quantization of the string. Notice that any Riemann surface has a conformally at metric, where is constant, and we pick a complex set of coordinates z (lightcone in Lorentz signature) so that z 1 and other components vanish. Substituting this metric in the action and varying with respect to the target space coordinates X we get very simple equations on the worldsheet: namely, the target space coordinates X satisfy free wave equations. These are given by X being holomorphic
The periodicity in 2 means we get an oscillator expansion for the solutions of the equations of motion.
and since we have gauge xed the metric, we still need to impose its equations of motion. This gives us a set of local constraints in the space of solutions (the Virasoro constraint). It is quadratic in the elds X , and takes the form
The variables p are interpreted as momenta in spacetime. Each positive integer n produces one left moving quantum and one right moving quantum of energy n, for each . These quantize the zero modes for translations. It follows that p2 gives essentially a set of integers as the allowed masses for particles. We also need to know the ground state energy of the system. This is a system where all oscillators are at zero occupation number. The ground state has a mass given by m2
p p
Oscillators
&
n exp in
& ' %
) (
where
exp in
& ' %
& '
X z
&
% %
X z
(11)
(12) (13)
(14)
(15)
(16)
which is negative, and signals a tachyonic instability in the system, so at best the bosonic string describes a consistent theory expanded about an unstable saddle point. We also get states with m2 0 and spin 2: these are identied as gravitons. These are one left oscillator and one right oscillator with n 1. This means critical string dynamics will generically include gravity. Another thing to notice is that the bosonic string we have described above does not describe fermions.
2.3. Superstrings
As we have seen, when we considered the bosonic string theory, we obtained worldsheet quantum elds for the coordinates of spacetime, and a very large symmetry algebra (the conformal group). We can speculate that increasing the worldsheet symmetries might lead to a more interesting theory. This is how one can conceive of the type II string theories. The main idea is to have supersymmetry on the worldsheet of the string, so that instead of just having the conformal group, we get its supersymmetric extension. Now, on top of gravity on the worldsheet, we have its superpartner (the gravitino), which also appears as an auxiliary eld. Also, for every X we get a pair of fermions, one left moving, one right moving , . Fixing the superconformal invariance we get 10 dimensions for at spacetime, as opposed to 26. This number is still bigger than four, and much closer to four. From the point of view of topology it is a great improvement to have only 6 extra dimensions instead of 22. Again, we consider the quantization in superconformal gauge. So the will have an oscillator expansion as well and they can have zero modes. The quantization of spin zero modes leads to representations of Clifford algebras. Here the zero modes of the become like ten-dimensional gamma matrices , so the degeneracy of the ground state implies that the ground state has spin degrees of freedom. The energy of this state is zero (the boundary conditions are supersymmetric) and leads to massless particles. We get one set of gamma matrices from the left movers and one from the right movers. This ground state is bosonic. Fermion elds are not usually observables, but their currents are (e.g. the charge of a fermion state, but not the sign of its wave function). This means we can also take antiperiodic boundary conditions (the boundary conditions are not supersymmetric). If we can do this independently on the left and right movers, then these new ground states with mixed boundary conditions will be spinors in 10D, because we only get one set of zero modes. One can check that these ground states are still massless, because half of the supersymmetries on the worldsheet are unbroken. One of these states is a spin 3 2 particle which is massless in ten dimensions! The only possibility of such a thing happening is if the ten dimensional theory in target space is supersymmetric itself. From here, worldsheet SUSY can imply spacetime SUSY, so the massless degrees of freedom of the theory lead to supergravity in ten dimensions. Consistency of the theory requires that the tachyon state is not physical, so this theory
is perturbatively stable. The projection onto physical states is called the GSO projection. The details are too technical for the scope of this paper. We will just notice that there are two ways to do this consistently: type IIA and type IIB strings, which differ on the chirality of the spacetime supersymmetries. In general one wants to compactify these theories to obtain a four dimensional world. If we require that the compactication include only a non-trivial metric and that supersymmetry survives in four at dimensions then the compactication manifold should be a Calabi-Yau geometry [5]. A problem with these theories and phenomenology is that they have too many supersymmetries: on compactifying to 4D the spectrum is non-chiral, and we would have a very hard time reconciling this theory with the standard model.
2.4. D-branes
So far I have argued that classical solutions of the heterotic string are the best candidates for 4D-particle physics, and that type II classical solutions are not. However, we can consider solitons on the type II theory which preserve some amount of SUSY and which do not involve only the metric. We can explicitly write black hole/brane solutions of type II supergravity which preserve some amount of supersymmetry, so it is important to understand if these have a nice description in terms of the type II strings. The answer to this question is yes, and the objects that are described in this simple form are called D-branes [6], see also [7]. The idea is to have a topological defect in spacetime which is the geometric locus where strings can end.
From the string theory point of view we are quantizing strings with open ends. The open ends are required to lie on the D-brane itself. These objects are physical branes: they have tension and they deform the geometry. The geometry they produce coincides with the black brane geometry in many cases. A very similar analysis for the closed type II strings can be done for the open strings ending on D-branes. It can be shown that a at dimension k D-brane preserves half of the supersymmetries in type IIA if k 0 2 4 , and for the type IIb if k 1135 . In particular, we can use them to break the total number of supersymmetries in 4D from 32 to 16. The quantization of modes on these D-branes produces massless particles with spin 1: gauge bosons!. This gives us another route towards getting gauge eld theories in 4D. If we place M branes on top of each other, we get M 2 open strings stretching between them. The interactions of the massless particles are governed by the dimensional reduction of U M , N 1, 10D gauge theory to k 1 dimensions. For k 3, this gives us N 4 SUSY in four dimensions. We can also consider congurations of D-branes at angles that reduce the supersymmetry further [8]. In this way, one is able to obtain a chiral spectrum of particles. These can in principle realize the standard model of particle physics, and the construction does not necessarily need a GUT embedding of the model into higher dimensions. Most of the modern developments in string theory in the last few years have been tied to understanding D-branes. We will explore some of the physics associated to D-branes and its relation to supersymmetry in the rest of the lectures.
$$
&
$$
However, the ground state of the string stretching between two of these D-branes will have a mass of order m l s 1 l ls ls 1 (18) where l is the typical separation of the branes. This can be made arbitrarily small compared to the string scale by moving the D-branes sufciently close to each other.
2 1
ls
(17)
These strings are essentially massless. The D3 brane is a supersymmetric D-brane (BPS) for type IIB string theory, so the degrees of freedom on the D-branes are supersymmetric. The D3 brane preserves 16 supersymmetries. Also the ground states for an open string state have spin less than or equal to one. We get that the low energy dynamics we want to capture is given exactly by N 4 SYM theory in four dimensions (the worldvolume coordinates of the D-brane). The theory is not free: the coupling constant in four dimensions does not have a scale. So it can have a nite value at very low energies. Now we will describe some of the physics associated to this theory. The stack of N branes leads to a U N gauge symmetry. The eld content is summarized by a vector particle, four weyl spinors and six scalars, all in the adjoint of U N , A , 1 4 , 1 6 . The theory has an SO 6 SU 4 R-symmetry that rotates the four chiral spinors into each other as a 4, and rotates the six scalar elds as a 6 of SU 4 . Non-renormalization theorems for sypersymmetry imply that this theory is nite. The eld theory has a superconformal symmetry and leads to an enhancement in the infrared to 32 supersymmetries when all D-branes are exactly on top of each other. The superconformal symmetry is an extension of the Lorentz group, which has additional generators: dilatations and special conformal transformations (this is like a doubling of the translations). There is a similar doubling of supersymmetries, by adding what are called special supersymmetry transformations. The R-symmetry is part of the
5
434343
43433
The Ri j are global internal symmetries of the quantum eld theory, is the generator of dilatations. L is the set of Lorentz transformations, P and Q are the translations and supersymmetries, while K S are the special conformal transformations and the special supersymmetries respectively. The bosonic part of this superconformal group is given by SO 4 2
where the rst factor is the extension of the Lorentz group to the conformal group, while the second factor is due to the R-symmetry. We will come back to explore the theory further. For the time being, this is all we need to keep in mind.
where this is the generalized potential associated to the extended charge that the Dbranes carry. The solution for the stack of D3 branes is given explicitly by the following background [9] f
F5 where
d5
SD
4 g s N 2 (23) r4 or more generally, f is a solution of the Laplace equation in the six directions transverse to the D-brane, with sources given by the positions of the parallel D-branes. The high f 1
ds2
P L K Q Q Si Si Ri j i i
(19)
0
dx2
f1
SO 6
(20)
(21)
dr2
r2 d2 5
(22)
amount of supersymmetry implies that the background does not get corrected. We can consider this geometry as a black hole/black brane solution of the supergravity equations. A standard property of black holes is that there is a redshift factor between being close to the horizon and asymptotic innity. This is seen by the ratio g00 x g00
Therefore if we add a particle which has proper energy E at a radius r, it has a global energy Eg which is reduced compared to E by the fact that clocks run at different speeds at r and at innity. States with very low global energy Eg ls 1 , do not need to have small proper energy. However, massive stringy states with this property become localized near r 0. This suggests that we can look for the infrared physics in the region of the geometry where there is a very large redshift [10]. In the metric we discussed above, this is the region where we can ignore the rst term in f and take r very small. We do this as a limit procedure. The end result is called the near-horizon geometry of the D3-brane system. The geometry we obtain in this limit procedure is given by a very simple metric. The spacetime geometry is given by a round AdS5 S5 geometry with N units of ux through the S5 . In may ways this is like a ve dimensional theory (because S 5 is compact). The cosmological constant on this effective theory is negative. The isometry group of this geometry is exactly given by SO 4 2 SO 6 , and the supergravity solution of the near horizon geometry of the stack of D3-branes also has 32 supersymmetries. Notice that this is the same result we obtained when we discussed the symmetries of the N 4 SYM theory. This result, together with some other calculations suggest that the two descriptions of the low energy dynamics (type IIB superstrings and N 4 SYM ) are the same [10]. The idea that these two systems might be describing the same physics is known as the AdS/CFT correspondence. The correspondence goes beyond the supergravity approximation. As discussed previously, we can have stringy states whose proper energy is of order ls , but with very small global energy. The question we need to answer is the following: What is the nature of this correspondence? This is a very non-trivial question and other than direct verication of the correspondence, one would want to have some set of tools which give some intuition with regards to this issue. The ideas that suggest that this is possible are related to the entropy of gravitational systems, namely that one can not put more energy in a region of spacetime than the energy required to build a black hole which lls the region under consideration. Thus the entropy of the energy inside a region of spacetime will grow at most like the area of the region under consideration [11, 12]. This suggests that one can map all of the degrees of freedom inside a region of spacetime to the boundary. This map is called a "hologram". The AdS/CFT correspondence is a holographic correspondence: the CFT is the theory living on the boundary of AdS space that encodes all of the gravitational degrees of freedom of the bulk.
1
0
@9# 6
f
1 4
(24)
The notion of boundary here is due to Penrose: one has to consider the conformal compactication of AdS5 . This is a manifold with topology S3 R in Lorentzian signature, and whose conformal structure is equivalent to a product of a round S 3 and a time direction. The S5 shrinks to zero size on the boundary. Even though we do not get a full metric on the boundary, we can still write the N 4 SYM on the corresponding conformal geometry. This is because the N 4 SYM theory is a conformal eld theory, so it only needs to be coupled to a conformal structure to be well dened.
SU 2 2 4 U N
g2 YM
g2 N YM
In the table above we have described the relations between the gauge group, coupling constant and t Hooft coupling [13] to the ux, string coupling and radius of curvature of AdS5 S5 in string units. The correspondence between gYM gopen and gs is standard. The radius can be read directly from the metric in string units, which we wrote as a limit of equation Eq. (22). Now, we want to write tests of the above correspondence. For this we want to insist on calculable properties of the eld theory and the gravitational background. Lacking a complete solution for either of the two problems, we want to see what is calculable as a perturbation expansion around a simple conguration. For type IIB string theory, we need large radius in string units and small coupling constant, so that we can perturb around a classical metric R 1 gs 1
This implies that we need gs N very large and gs very small. Both of these together imply that we need to take N very large. For SYM, we need to be in a perturbative regime. The effective perturbation expansion is in the t Hooft coupling g2 N (26) YM
G F E B EDCB EDCB
SU 2 2 4
S5
gs
R4
SYM
AdS5
S5
(25)
from summing over all colors in Feynman diagrams. This needs to be small to have calculable physics. Indeed, the ideas of t Hooft suggest that we get a double expansion in g2 N and 1 N. The rst one is characterized by planar diagrams, while the second YM one gives non-planar diagrams and which topologically suggests Riemann surfaces of higher genus. This should become the and gs expansion in the type IIB string theory on AdS5 S5 . From this (perturbative) point of view, the AdS/CFT correspondence is a strong/weak coupling duality for the t Hooft coupling constant. We now come to the main technical problem of the AdS/CFT correspondence: how to set up reliable calculations at weak coupling that are valid at strong coupling. We have already seen that both systems have the same number of free parameters and symmetries. Can we test more? This is the point where supersymmetry comes to the rescue. The supersymmetry algebra should be realized unitarily, which places some unitarity constraints on the representations. Certain of these representations are short (they preserve some supersymmetry), and they saturate a BPS bound (this is a consequence of unitarity). These can not get corrected because the bound is an inequality between the and R charge quantum numbers. However, the R charge is related to a non-abelian symmetry, which means that the charge is quantized and can not receive corrections, while in principle can. It turns out that all supergravity single particle states preserve some amount of SUSY, essentially because these are massless in ten dimensions and are naturally short (they lack the longitudinal components to make full massive multiplets). So it is natural to calculate their spectrum. The most convenient choice is to write the AdS5 S5 metric in global coordinates
Here t, , 3 are the coordinates of AdS5 , where t is associated to a timelike-killing vector and , 3 are spherical coordinates around an initial point. We have also chosen to write the S5 in similar coordinates, where is a coordinate related to a particular killing vector, and , are spherical coordinates with respect to the circle represented by the angle . Notice that the gravitational potential term as read from the metric gtt implies that the particles are conned near 0. If we want to solve for wave propagation in this geometry, this fact means that the "Schrdinger operator" will have a discrete spectrum. This time slicing gives a killing vector which is identied with the generator of dilatations in the superconformal group. The problem of studying the full supergravity modes of this geometry was studied a long time ago [14, 15], and a complete classication of states was obtained as representations of the superconformal group SU 2 2 4 . The list of representations are all in short multiplets of the symmetry algebra (half BPS). The lowest lying state of each multiplet is given by a totally symmetric traceless tensor of SO 6 , which is a scalar with respect to the rotations of the superconformal group which commute with the dilatations. This state is annihilated by all of the K and the S generators, and half of the Qs. We will now reproduce these results using the N 4 SYM theory.
&
&
&
&
&
dt 2 cosh2
d 2
sinh2 d2 3
d 2 cos2
d 2
sin2 d 3
(27)
This metric of at 4 , removing the origin is conformally equivalent to the metric of S3 , they differ by the (local) scale factor r 2 . So if we insert an operator at the origin, the scaling dimension of the operator is the eigenvalue under dilatations. This is the same as translations in the Euclidean time t log r . So local operators inserted at the origin in 4 are equivalent to states of the theory on the round sphere. I will go back and forth between these two descriptions. This property of conformal eld theories in arbitrary dimensions is called the operator-state correspondence.
These are chiral if it also holds that Q O 0 for some of the supersymmetries. States obtained from a chiral primary by application of Q P are called descendants. The conformal dimension of the operator O is the eigenvalue of the energy operator of the state O on S3 . Unitarity implies that for any primary
and equality is saturated for chiral primaries (BPS states). J is a generator of an SO 2 subgroup of SO 6 , such that there is an SO 4 which commutes with J
J 0 0 0
(31)
S
R Q R Q R Q
K O
S O
&
ds2 4
r2 d2 3
dr2 r2
I
(28)
0 P
(29)
(30)
Highest weight states of the representation algebra are the ones that saturate this bound. In the free eld theory, there is only one complex scalar eld such that its canonical dimension and J coincide. Let us call the eld Z. The half BPS states with respect to J are descendants of objects that can only be built out of Z. Remember that Z carries gauge symmetry indices Z ij in the adjoint (they are N N matrices). The operators that correspond to the BPS chiral primaries need to be gauge invariant. This means that all gauge indices need to be contracted. This can be accomplished by using a matrix notation. The list of operators are given by (multi)-traces of Z.
How does this relate to the description of gravitons? Each trace is considered as a single graviton state [16, 17]. Multi-trace objects are considered as multi-graviton states. One can prove that in the large N limit (for nite n i ), different traces are approximately orthogonal. This correspondence receives corrections in 1 N 2 , which can be interpreted as the result of interactions. The effective gravitational constant in ve dimensional AdS5 is exactly 1 N 2 in units where the radius of AdS5 is one. Do we match states with SUGRA? Yes. The state above is considered as the highest weight state of an SO 6 representation. This representation is totally symmetric and traceless, matching exactly the supergravity result.
$V $$
n1 n2 nk tr Z n1 tr Z nk
(32)
%
p
AdS giant
EF
Closed strings S giant
example, we understand their low lying uctuations both from the DBI action and from the SYM theory [23, 24]). There is still work to be done verifying these conjectures. The dynamics describing these BPS states is very simple from the gauge theory point of view. It corresponds to a gauged matrix quantum mechanics of a single N N matrix. From this point of view one nds a remarkable analogy with a Quantum hall droplet that helps describe all of these states: the strings are interpreted as edge excitations of the quantum hall droplet. The giant along AdS5 is interpreted as an electron and the giant growing along S5 is interpreted as a hole state [25]. From here, we see that already the BPS content shows that there are objects beyond supergravity which are captured by the SYM theory. This is shown pictorially in Fig. 5. In the next lecture we will go beyond supergravity states and produce a full string spectrum out of SYM theory, with many non-BPS particles.
very small region of space compared to the scale at which the geometry is varying: this is a limit where the curvature does not matter. This is a high energy limit in the sense that to localize the particle one needs a relatively high amount of kinetic energy (this follows from the uncertainty principle). From here it follows that in principle the at spacetime spectrum of strings is encoded in the AdS/CFT, but we need to think of it as a high energy limit of the spectrum of states. One can then ask the following question: is the ultra-relativistic limit of particles in a weakly curved background interesting? This question is really addressing whether in the high energy limit we get something other than a at spacetime. The answer depends on the background. There are various ways to look at this problem. The backgrounds we are dealing with have geometric curvature and uxes. These objects are tensorial. We can ask how important they are by looking at the numerical value of their components in a frame were the relativistic particle is at rest in its own frame. The background elds in this frame are related to the original coordinate system where we describe the background elds by a boost transformation. The boost transformation is an element of a non-compact SO n 1 group, and this being non-compact, a large boost results in large components for the transformation matrix . In particular, an otherwise small component of a tensor eld, can get to be very large in a different coordinate system related by a boost. This is a feature of geometries where we have time. On Euclidean geometries the group of local rotations is compact and does not lead to large components of tensors. To dene the ultra-relativistic limit, we can then take a double scaling limit where we take the background to zero in some particular coordinate system, and we make a compensating boost at the same time so that some components in the boosted system are kept nite. In a more invariant language, we have a particle with momentum p , and possibly some polarization tensors (like spin ). The background eld is characterized by some tensor T . We take the limit T 0 , so that some contraction between T with p , and perhaps stays nite. This denes the scaling of p in the limit. Notice that since the mass of the particle should roughly be xed p2 m2 , the fact that the energy goes to innity while the proper mass stays nite means that the particle trajectory is well approximated by a null curve. If we consider a point particle coupled to gravity, the natural trajectories for this limit are null geodesics. If we adapt the coordinate system to the null geodesic so that p 0 is nite, the new geometry is a non-trivial limit of the original geometrical system. The typical scalar one can keep nite is R p p . This limit is called the Penrose limit . The geometry one obtains this way is of the plane wave type, and it is a considerable simplication of the geometrical system.
We will now explicitly take the plane wave limit of the AdS5 S5 geometry [26]. We will keep a lot of the technical details in the following derivation, but the main ideas
#
S5
were already presented above. The main idea is to look at a null geodesic where we stay at a xed position in AdS5 (varying global time), and consider the motion at the speed of light along one great circle in S5 . First, we write the metric along geodesics of AdS5 S5 , parametrized by one time coordinate on AdS and one circle of S
and we want to consider a particle at the origin 0, 0. We will focus on the geometry near this trajectory. We can do this systematically by introducing coordinates x t 2 and x t. We also perform the rescaling
4dx dx
r2
y 2 dx
dy 2
dr 2
which is the plane wave background. After an extra rescaling of x we can introduce a scale . ds2 F 4dx dx 2 z 2 dx F 5678 const
2
dz 2
1234
0 limit of this geometry is at space. The What makes this limit interesting is that the free string theory can be exactly quantized in this background.
a f a e& a
z 2
z z
z 2
dx dx d d
% %
R2 x
r R
y R
&
&
&
% % % % % #
&
&
dt 2 cosh2
d 2
sinh2 d2 3
d 2 cos2
d 2
sin2 d 3
(33)
(34)
(35)
(36)
% %
(37)
(38)
Some of the supersymmetries of spacetime are realized linearly in this background, and they commute with the Hamiltonian. So the fermions are also massive. The mass comes from the ux. The exact details can be found in the paper of Metsaev [27]. Quantizing the string is straightforward: it is a free eld theory on the worldsheet, and we get an oscillator expansion. The spectrum of particles is given by p Hlc Nn
In addition we have the condition that the total momentum on the string vanishes P nNn
Here p is the lightcone Hamiltonian, and p is a conserved quantum number associated to the quantization of the zero modes of translations on x (the lightcone momentum).
where J is the rotation in the direction, one of the R-charges of the theory, exactly the one characterized by the metric Eq. (27). We therefore need operators with J nite, J R2 N, and remember also that J is the BPS condition. From these considerations it follows that we need to consider states which are almost BPS and of very large conformal dimension. Rewriting the spectrum of the string in the eld theory parameters we nd that the contribution to the energy from each oscillator is roughly given by
n
and we nd that it looks like an expansion in the tHooft coupling gN, that we can reproduce by doing perturbation theory! [28]. Notice that we are making a non-trivial statement. The large radius limit is taken rst making gN large, and the quantum numbers are taken large to compensate for the large radius limit (this is the ultra-relativistic limit we were considering). We are trying to compare these results to a perturbation expansion around gN 0. What saves the day is
i % & h h %
p p
ix ix i t p 1 J i 2 2 x R R R2
wn
4 gNn2 J2
"
& q p
% g
% &
2
g %
4n2 p
(39)
(40)
(41)
(42)
that there is a new effective expansion parameter which reduces the value of the t Hooft coupling parametrically with the quantum numbers of the particle, so that the limits are not incompatible.
where there is a nite number of r . One can interpret these as a small number of excitations scattered in the operator which is mostly Z. Planarity of the diagrams makes the position of the defects important. States with defects in different places are orthogonal in the planar theory in the free limit. All of these operators have the same conformal dimension in the free eld limit. However, when we introduce interactions we need to nd the proper basis of operators, and we end up with a problem in degenerate perturbation theory. The interaction term g2 tr Z j Z j (44) YM will make the defects hop to the left or right in the operator. Indeed, we can not get jumping by more than one site at rst order perturbation theory because each change of position would count as one extra non-planarity. To cancel it, we need to add an interaction each time we hop positions. The idea now is to interpret the position of the defects in between the Z as a lattice. The number of sites is of order N, so in the large N limit we obtain a continuum theory, as we are also taking the J interaction terms gs N . The theory on the lattice is translation invariant, mainly from the cyclic property of the trace. Therefore it is convenient to rewrite the states in a momentum basis, so we apply position dependent phases for inserting the same defect at different positions. For example, a state with one excitation of momentum n will roughly be given by
JN J
2 1 2
Obviously for n 0 the above vanishes by cyclicity of the trace. In the string theory this state has angular momentum along the string and it is not a physical state. However, we
R
6 %6
a4 0 p n
tr Z l 4 Z J
" g " %
1 J Jl 1
$$
Z
bR Q R Q
$$ r
Z r Z
(43)
"
e2 inl
(45)
FIGURE 6. Hopping interactions from planar diagrams: The dashed line indicates the defect. We see one crossing in the diagram
lc
where we have used cyclicity to keep one of the operators at zero. It is the relative phases that count. The eight bosonic oscillators of the string are supplied by the four neutral scalars under J and by the four directions in which we can take covariant derivatives. The fermions of the string are made by putting fermionic operators on the trace.
The ones with n 0 acquire an anomalous dimension: this is the quantum contribution to , J is an angular momentum so it is always quantized and xed by the free eld theory value. It is also clear that the planar perturbation theory will not notice defects that are far away from each other, so the contribution to the dimension of the operator is a sum over individual defects (dilute gas approximation). The dimension is usually split into the free eld theory part, plus the anomalous dimension . What we need to
ndefects
R
g " # d3 3 u% v t
3 a 4 a n 0 p n
tr 3 Z l 4 Z J
6
J
6 Q %
1 J 1 NJ 2 Jl 1
e2 inl
(46)
(47)
calculate is the anomalous dimension of these operators. This is a standard eld theory problem. Concentrating on one defect we have
and it has to vanish for n 0 because that operator is protected. This helps to x a 1, and one nds agreement with the rst term in the expansion of the light cone Hamiltonian of the string; we reproduce the full set of oscillator energies to rst order in the t Hooft coupling. All of this follows because we have already identied (roughly) the states that diagonalize the perturbation [28]. A fairly extensive review is found in [29].
In the free eld theory level any two words made out of the same letters in different order are degenerate, and correspond to essentially orthogonal states (up to the cyclic property of the trace). This is a property of the large N limit of free eld theory. The states 132 and 123 are different. The spectrum of anomalous dimensions is a problem in degenerate perturbation theory. At rst order this leaves the length of the trace invariant because states with different length have different scaling dimension. The fact that we restrict to planar diagrams means that to rst order the only thing that can happen is that we get a nearest neighbor interaction that can change the values of the spin labels at each point. Again, this is because going further away from the initial conguration costs us in non-planarity, which needs to be compensated by interactions. The spectrum of anomalous dimensions gives rise to a nearest neighbor Hamiltonian for the spin chain. This spin chain for scalars has SO 6 symmetry, which simplies the Hamiltonian considerably. The space of two nearest neighbor pairs decomposes into
$$
q $V
1 3 1 2
osc
gN cos 2 n J
gN n2 J 2
1312
x w
(48)
(49)
(50)
SU 4
The spin chain hamiltonian assigns a different value to each irrep. The last one can be seen to be zero, as the BPS states belong to it. The computation can be done explicitly and it was found that the spin chain Hamiltonian in question is integrable [30]. One can also have spin chains with SO 6 symmetry of the above form which are non-integrable, so this is a non-trivial fact. The integrability of the spin chain means the eigenvalues can be found analytically in implicit form by using the Bethe ansatz technique. Cases with just two impurities in the scalar sector can be solved exactly for any length of the chain. The result has been generalized to all possible single trace operators and it was found that there is an integrable spin chain with SU 2 2 4 symmetry describing the full one loop calculation [31]. From the strings on AdS5 S5 one also nds that the sigma model action is classically integrable BPR. It has been argued that both of these integrability structures are associated to the same type of symmetry (a Yangian symmetry). It is conjectured that one can connect weak coupling perturbation theory with the strong coupling string theory by a family of integrable systems, so that string theory on AdS5 S5 should be solvable!: one should be able to resum the planar perturbation expansion of SYM to obtain the string sigma model [33]. This is very hard.
(51)
y 0
5. LECTURE 4
So far we have dealt with at D3-branes in at ten dimensional space and their low energy dynamics. We have seen that this system is very rich and shows us how to obtain string dynamics from gauge theory and gives new insight into what a theory of quantum gravity could look like. However, the amount of supersymmetry is too large, so it is also important to understand less supersymmetric systems that can arise from branes. Part of the motivation is to make contact with phenomenology of particle physics. In order to do that we need to consider higher dimensional branes (let us say D5 branes) which are wrapping some compact geometrical cycle C or set of cycles, so that their low energy effective eld theory is a four dimensional supersymmetric eld theory. We need to wrap a compact cycle because the effective four dimensional coupling constant is given by a dimensional reduction formula on the world volume of the brane g2 4YM
For what we will be studying, supersymmetric cycles can be characterized by holomorphic curves or surfaces on a Calabi-Yau (CY) geometry. We will consider type IIB strings compactied on a CY manifold, and branes wrapping some of its holomorphic cycles and spacetime in the transverse directions to the brane. Since this system has less supersymmetry, there are fewer constraints on the form of the Lagrangian from the four dimensional point of view. Moreover, the system is not as protected from receiving quantum corrections to the effective action. In practical terms this means that it is a lot harder to make exact statements about the system. However, there are some quantities whose origin is essentially topological which can still be calculated exactly because there is still some amount of non-renormalization due to supersymmetry.
1 gs
dVol
(52)
To study these effects, we do not need to know the specic metric and geometry to understand some details of the low energy supersymmetric dynamics. For example, the number of massless degrees of freedom are controlled by topological aspects of the geometry, a set of N branes wrapping a cycle will still lead to a U N supersymmetric gauge theory. If the cycles are rigid (one can not deform them and keep the cycle as a holomorphic cycle), then there are no transverse motions of the brane in the higher dimensions which cost no energy. This means that in the low energy effective eld theory all we would get is pure SYM theory. Considering more cycles and branes wrapping them one can get more general theories. Also, if there are some transverse motions that are allowed, these lead to massless matter elds in the adjoint representation of the group, one for every holomorphic deformation. Our interest will be in studying the low energy dynamics of these classes of theories. The point we will take in these lectures is that the geometry of the embedding can give us a lot of information about the low energy dynamics, and in the end a lot of the vacuum structure can be obtained from understanding the geometry better. To separate the problem of gravity from the eld theory, we can consider geometries where the corresponding cycles we wrap are small, and the volume of the CY is very large. Under these circumstances most of the CY geometry does not matter, and only the geometry very close to the brane is important. We can thus take a large volume CY, even innite. This second option is done by writing a simple non-compact CY geometry and doing the analysis there. A second advantage is that the four dimensional Planck scale is sent to innity in the process, so that gravitational interactions are decoupled and can be ignored. Thus the low energy degrees of freedom are those of an ordinary supersymmetric eld theory. Instead of treating the most general case, the best thing to do is to start with an example, where we know what the low energy eld theory on the D-branes is. A particularly simple case is to begin with an orbifold space 2 2, so that the other directions form a six-dimensional at geometry. This space is the identication of at 2 under x y x y The characterization of this geometry can be done as an equation in three complex coordinates which are the holomorphic invariants of the 2 action (53) uv w2 0 Here u x2 , v y2 , w xy. This produces a singular space, where the singularity is at u v w 0 (the xed point of the 2 action), which is resolved by blowing-up the singularity: replace the singular points by a P1 S2 . If we wrap a D5-brane on the sphere, we obtain a theory which in six dimensions would have N 1 SUSY, so that in four dimensions it has N 2 supersymmetries. From here we know its Lagrangian at weak coupling. For M branes it consists of a U M gauge theory (vector multiplet) with additional scalars in the adjoint (one chiral supereld), we call this eld . represents the position of the D5-branes in the transverse directions to 2 2 and the worldvolume of the D5branes themselves, and there is also no superpotential.
#
The vacuum structure of this theory is exactly solvable due to the fact that we have N 2 SUSY. The explicit solution was found in the work of Seiberg and Witten [36]. The theory is solved by understanding the geometry of a Riemann surface associated to the gauge theory. Now take the N 2 theory and deform it slightly to N 1 SUSY, by adding a superpotential term to . This can be done as a perturbation of the N 2 system. The superpotential will be polynomial and given explicitly by W tr
i
i 1
Classically, for vacuum congurations to survive, all the eigenvalues of have to sit on the zeroes of the polynomial W 0 P (55) so now the D5-branes have some potential on the transverse directions to the orbifold. This deformation is associated to a change of the complex structure of the transverse directions to the D5-brane. The new geometry is given by [37] uv
The D5-branes can only be placed at the singularities of the above geometry: the zeroes of P z , otherwise they are not supersymmetric. The reason for this is that we have to remember that the true geometry results from blowing up the original singular locus. It is only at the singularities of the geometry above that the blow-up gives rise to a holomorphic S2 . Generically the eld is massive, unless we ne-tune the superpotential. The structure of the theory in the deep IR reduces to a product of pure SYM theory, with a gauge group U Ni . In this setup there are Ni eigenvalues of at each root of P, each eigenvalue is interpreted as a D5-brane. We have good reason to believe that pure N 1 SYM theory with group U Ni has Ni vacua [38]. The order parameter characterizing these vacua is the vacuum expectation value (v.e.v.) of the gaugino condensate
which is another element of the chiral ring. This information is holomorphic, so it is natural that is should be measurable in the geometry. In practical terms this means the equation should be deformed [37] uv
Because we can vary the coefcients of P by varying the superpotential deformation, the f associated to this process should be of lesser degree than P. One can see that P has degree k, so f has degree k 1, and it is characterized by k coefcients (one for each root of P). These should be measuring the partial gaugino condensates for each singularity. In practice, there is one complex parameter per gauge group. Roughly, this takes into account the possible phases of the gaugino condensate associated to each singularity.
w2
Pz
f z
5
tr W W
w2
Pz
k 1a
g
(54)
(56)
(57)
(58)
In general the presence of f gives rise to a smooth complex geometry. If there are no branes at a singularity, there should be no gaugino condensate associated to it, and the geometry should still be singular. This means the polynomial P2 f has some double roots left over.
where V is the superpotential of the gauge theory, and the right hand side is understood as a large N planar diagram saddle point: the right hand side generates a planar diagram expansion about a classical solution of V 0. From here we obtain the function which depends on the data of the saddle point (how many eigenvalues are at each saddle point, as a fraction of the total number of eigenvalues) and the coefcients of the potential V . 0 is a prepotential, which depends on some gaugino condensate variables in the gauge theory. The idea is that in the weak coupling approximation in the IR the gauge group is broken to U Ni , for each of which there is a gaugino condensate Si . The identication with the matrix model variables is Si gs Ni , where the right hand side is the partial t Hooft coupling in the matrix model associated to each saddle point.
U
N2 g2 s
d exp
N V gs
&
& T
U R Q T
(60)
and this is a function of the partial gaugino condensates. One then takes
W 0 (62) S to get to the vacuum structure, where one also includes a term Ni Si log Si for each partial gaugino condensate.
where m is the mass of the particle, and p are superspace momenta, while W is the covariant chiral super-eld strength for the gauge eld and gaugino elds. This propagator requires m 0, and does not apply for a chiral theory where there are no bare mass terms allowed in the Lagrangian. After using all the delta functions of the vertices on an L loop integral, and counting powers of , one gets a total measure d 2L 2 . A local term in the action will have a measure which is d 2 , so we need 2L powers of to get a non zero result. This translates into 2L powers of W . Also there can not be more powers of W than two per color loop (numbers of faces) for the result to be chiral. This is a result of some classical supereld manipulations. If we use the t Hooft double line notation (fat graphs), the number of faces of the double line graph is bounded from below F L, which singles out only planar diagrams (for orientable surfaces). Some work with Schwinger parameters shows that the momentum dependence disappears from the diagram, and one is left with the planar amplitude of the matrix model: the difference is that instead of N in the loops, one has S tr W 2 from the gaugino insertions. There is also one loop which has no W . This loop contributes the factor of Ni from tr 1 . After some combinatorics, one can see that the end result is given by summing derivatives of the prepotential. There is a second proof of the result which is more technical, and was done in the work [42]. They study the chiral ring of the supersymmetric eld theory in a lot of detail. They found that the chiral ring is generated algebraically by
tr k
tr kW
tr kW 2
&
&
p2
m2
d 4 pd 2
Weff
Ni S 0
(61)
(63)
(64)
The correlation functions of chiral elds in supersymmetric backgrounds are independent of position. Together with the cluster decomposition principle this implies that the correlation functions are just the product of v.e.v.s of the individual chiral elds on the vacuum. This is why these form a ring: multiplication of two chiral elds produces a chiral eld. The idea now is that if there is some relation to matrix models, then the observables of the matrix model should be correlated to the chiral superelds, and one can try to imitate the proof of the loop equations for the matrix model in the supersymmetric eld theory. The loop equations of the matrix model can then be interpreted as relations in the chiral ring of the eld theory. The proof goes by studying a generalized Konishi Anomaly for the (Virasoro constraint) variations
The idea is that a variation of the elds is generated by a supercurrent, so one has the conservation law for the associated current
There are two sources for chiral violation of the supercurrent conservation. The rst is the fact that the superpotential depends on the elds , so the current is not a symmetry of the theory, and there is explicit violation from the action. The second term is due to a mixed anomaly with the gauge superelds. On taking v.e.v.s on a supersymmetric background the left hand side vanishes, and the right hand side produces a set of polynomial relations in the chiral ring. Some of these coincide exactly with the planar loop equations of the matrix model associated to the superpotential W . Integrating these equations one obtains the effective superpotential again. The advantage of this second proof is that one also obtains a full dictionary between matrix model observables and the chiral ring. This permits one to also write down formal operators which correspond to the partial gaugino condensates, so these elds which arise from low energy considerations after a Higgs mechanism has been taken into account can be given a gauge invariant realization in the UV theory.
0 P
&
D2
gauge anomaly
n nW W
(65)
P0
(66)
We have seen that the deformation process follows just from eld theory alone, and does not require us to think in terms of strings in a CY geometry. This is why I spent some time describing how the proofs of the Dijkgraaf-Vafa were argued. However, the geometrical picture can help us to explain some interesting features of the eld theory. The geometrical picture of gaugino condensation process is that we have to make a change of topology. In topological gravity this was stated earlier in [43], and was shown with a supersymmetric eld theory calculation in [44]. In the new topology there is no compact holomorphic cycle in which to wrap the D-branes, so one does not see the sources of the generalized uxes (these are the D-branes). However, the change in topology allows one to have the ux without a source. From the point of view of looking at the system from asymptotic innity both of these pictures lead to the same boundary conditions, but the topological structure inside has changed. This operation can be called a brane-ux transition. It replaces D-branes by uxes. The ux through the corresponding three cycle S 3 is equal to the number of original branes we placed in the geometry. This is very similar to our discussion of the AdS5 S5 geometry, where the remnant of the fact that we have D-branes is that the ux through the S5 is quantized. Another important thing to notice is that since we have lost the D-branes, we do not have a place for open strings to end anymore. This is interpreted as connement in the gauge theory, since we do not see the charges at the end of the strings anywhere. One also sees that in the deformed geometry all of the degrees of freedom have become closed strings. These include the gauge invariant "composites" of the D-brane themselves, these states are like glueballs in QCD. This is also similar to the description of AdS 5 S5 . However, there is a lot less data on the spectrum of these strings.
connement and one usually generates an effective superpotential for these branes, see for example [45]. The upshot is that gaugino condensation produces and effective superpotential for some at directions that lifts the degeneracy of vacua, giving only a discrete set of solutions. Geometrically, the effect of the gaugino condensation is to deform the geometry, and this usually lifts the curve of singularities to something less singular. This is, one nds that instead of the full curve of singularities one only has a nite number of conifold points left over. In the simplest cases, one obtains runaway behavior: the saddle point of the effective superpotential happens at innity [46]. Usually one needs more than one extra singularity on the curve. Thus one makes two different gaugino condensation effects compete against each other. This is called a racetrack mechanism. This xes some of the moduli problems. For branes in the bulk, this does not seem to happen [47], and that one gets instead is just the deformed geometry. From the ux-geometry point of view, the uxes produce effective potentials for geometrical moduli that describe the shape of the local Calabi-Yau geometry. Thus the shape of the CY can be xed by the uxes.
6. LECTURE 5
This is the last part of the lecture notes, where I try to discuss advances in N 0 SUSY (supersymmetry breaking or no supersymmetry at all). Most of this discussion is on a much less rigorous footing than then previous lectures, but we will use extensively the intuition we have gotten from the more supersymmetric cases.
6.1. Warping
We want to begin again with AdS5 S5 as a geometric template for four dimensional eld theories. The rst description I gave in terms of at D-branes shows that we have Lorentz invariance along the four longitudinal directions to the brane, with a metric of the form ds2 dx2 f 1 2 r dr2 (67) The fact that de coefcient of dx2 varies with r is called a warp factor. The topology is still a product, but the metric is not a sum of terms for each factor independent of the other one. The observation that this theory can be interpreted as ordinary eld theory without gravity when f is 1 r 4 motivated the construction of the Randall-Sundrum scenario for solving the hierarchy problem [48]. The reason in the above geometry there is no four dimensional Newton constant is because such a constant mode deformation of the metric is non-normalizable. If we add a little geometric (UV) cutoff to the story, we can recover the constant mode for gravity, but it is weakly coupled. This would produce a hierarchy between the four and ve dimensional gravitational constant from warping. We also have this redshift factor which can reduce the mass of objects from the gravitational point of view. How strongly you feel 4D gravity depends on where you are.
u$$ &
477 & 6 77
This ingredient shows up very often in string compactications with uxes [49], although in general we do not know how to write the metrics with very much detail at all.
Start with a SUSY model: branes on a CY geometry. Do brane-ux transitions (these capture non-perturbative effects) and one can construct models which solve the moduli problem. There can still be some branes in the bulk and that lead to the SSM or something similar.
Assume that the regime one is studying is geometrical, so that the solution can tolerate geometric reasoning. This produces one value of the cosmological constant of the wrong sign.
the age of the universe. As a matter of numbers, we can ask how likely is it that we are going to get the details right. The ne tuning required by the gravitational constant is of order 10 120 . With roughly 10400 vacua one can do pretty good at getting the cosmological constant right, even if it is in a small corner of parameter space. This can still leave 10 100 vacua from which we can ask other physical requirements. The picture one should have in mind is that there is a vast landscape of valleys and hills where dynamics might take us. The parameters characterizing the landscape are the moduli elds of the string compactication. The valleys and hills indicate an effective superpotential on these elds that determine the dynamics. There are assumptions built into the picture, and although they are reasonable, there is always the possibility that they will not be generally true. The number of relevant vacua to consider is larger than the observable entropy of the universe, so it is very impractical to study them one by one until we nd the right one. We need to do a different kind of approach to study this system. So how do we explore the landscape? We could still look for the right model by choosing some region which might look attractive. However, this will only cover a very small amount of vacua of all the ones we need to consider. We can also try to study them all at once by asking statistical questions for generic vacua. This is the approach of Douglas [50, 51]. He suggests that we change perspective and ask questions of an ensemble of possible theories (presumably the most general background where we have some control). The types of questions we should be asking are of the following form Of a given family of vacua how many fall within the cosmological constant range we want? How many are perturbatively stable? How many contain three families? What is the SUSY breaking scale?
What fraction realizes the SM matter content on a brane? On these, how are the coupling constants distributed? How are these distributions correlated to each other? Do the interesting ones bunch along a particular direction? Does this describe new mechanisms to solve long standing problems?
One can keep on rening the problem further as one nds success at different stages of the program. Each of these becomes more and more involved, as we are requiring more numerical information at each step. Each of these calculations is very difcult because we do not have supersymmetry along the way to help us.
signals in accelerators (displaced vertices, or for even longer lifetimes one can expect to see very heavy particles in detectors ). This also has the advantage of reducing the number of effective parameters that are needed to describe the low energy physics. It is also a largely unexplored corner of the SSM parameter space.
ACKNOWLEDGMENTS
I rst would like to thank the organization of the XXXV Latin American School of Physics. Especially Roelof Bijker, Hugo Morales and Luis Urrutia. Also I would like to thank V. Balasubramanian, F. Cachazo, S. Cherkis, R. Corrado, R. Dijkgraaf, M. Douglas, A. Hashimoto, I. Klebanov, R. Leigh, J. Maldacena, H. Nastase, R. Roiban, N. Seiberg, A. Tseytlin, E. Witten for many discussions on physics which have shaped my current understanding of string theory and supersymmetry.
REFERENCES
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. M.B. Green, J.H. Schwarz and E. Witten, Superstring Theory. Vol. 1: Introduction, Cambridge University Press, Cambridge, 1988. M.B. Green, J.H. Schwarz and E. Witten, Superstring Theory. Vol. 2: Loop Amplitudes, Anomalies and Phenomenology, Cambridge University Press, Cambridge, 1988. J. Polchinski, String theory. Vol. 1: An introduction to the bosonic string, Cambridge University Press, Cambridge, 1998. J. Polchinski, String theory. Vol. 2: Superstring theory and beyond, Cambridge University Press, Cambridge, 1998. P. Candelas, G.T. Horowitz, A. Strominger and E. Witten, Nucl. Phys. B 258, 46 (1985). J. Polchinski, Phys. Rev. Lett. 75, 4724 (1995) [arXiv:hep-th/9510017]. C.V. Johnson, D-branes, M. Berkooz, M.R. Douglas and R.G. Leigh, Nucl. Phys. B 480, 265 (1996) [arXiv:hep-th/9606139]. G.T. Horowitz and A. Strominger, Nucl. Phys. B 360, 197 (1991). J.M. Maldacena, Adv. Theor. Math. Phys. 2, 231 (1998) [Int. J. Theor. Phys. 38, 1113 (1999)] [arXiv:hep-th/9711200]. G. t Hooft, arXiv:gr-qc/9310026. L. Susskind, J. Math. Phys. 36, 6377 (1995) [arXiv:hep-th/9409089]. G. t Hooft, Nucl. Phys. B 72, 461 (1974). H.J. Kim, L.J. Romans and P. van Nieuwenhuizen, Phys. Rev. D 32, 389 (1985). M. Gunaydin, L.J. Romans and N.P. Warner, Phys. Lett. B 154, 268 (1985). S.S. Gubser, I.R. Klebanov and A.M. Polyakov, Phys. Lett. B 428, 105 (1998) [arXiv:hepth/9802109]. E. Witten, Adv. Theor. Math. Phys. 2, 253 (1998) [arXiv:hep-th/9802150]. J. McGreevy, L. Susskind and N. Toumbas, JHEP 0006, 008 (2000) [arXiv:hep-th/0003075]. M.T. Grisaru, R.C. Myers and O. Tafjord, JHEP 0008, 040 (2000) [arXiv:hep-th/0008015]. A. Hashimoto, S. Hirano and N. Itzhaki, JHEP 0008, 051 (2000) [arXiv:hep-th/0008016]. V. Balasubramanian, M. Berkooz, A. Naqvi and M.J. Strassler, JHEP 0204, 034 (2002) [arXiv:hepth/0107119]. S. Corley, A. Jevicki and S. Ramgoolam, Adv. Theor. Math. Phys. 5, 809 (2002) [arXiv:hepth/0111222]. S.R. Das, A. Jevicki and S.D. Mathur, Phys. Rev. D 63, 024013 (2001) [arXiv:hep-th/0009019]. D. Berenstein, Nucl. Phys. B 675, 179 (2003) [arXiv:hep-th/0306090]. D. Berenstein, JHEP 0407, 018 (2004) [arXiv:hep-th/0403110].
26. M. Blau, J. Figueroa-OFarrill and G. Papadopoulos, Class. Quant. Grav. 19, 4753 (2002) [arXiv:hepth/0202111]. 27. R. Metsaev, Nucl. Phys. B 625, 70 (2002) [arXiv:hep-th/0112044]. 28. D. Berenstein, J.M. Maldacena and H. Nastase, JHEP 0204, 013 (2002) [arXiv:hep-th/0202021]. 29. J.C. Plefka, Fortsch. Phys. 52, 264 (2004) [arXiv:hep-th/0307101]. 30. J.A. Minahan and K. Zarembo, JHEP 0303, 013 (2003) [arXiv:hep-th/0212208]. 31. N. Beisert and M. Staudacher, Nucl. Phys. B 670, 439 (2003) [arXiv:hep-th/0307042]. 32. I. Bena, J. Polchinski and R. Roiban, Phys. Rev. D 69, 046002 (2004) [arXiv:hep-th/0305116]. 33. L. Dolan, C.R. Nappi and E. Witten, JHEP 0310, 017 (2003) [arXiv:hep-th/0308089]. 34. A.A. Tseytlin, arXiv:hep-th/0407218. 35. G. Arutyunov, S. Frolov and M. Staudacher, arXiv:hep-th/0406256. 36. N. Seiberg and E. Witten, Nucl. Phys. B 426, 19 (1994) [Erratum-ibid. B 430, 485 (1994)] [arXiv:hep-th/9407087]. 37. F. Cachazo, K.A. Intriligator and C. Vafa, Nucl. Phys. B 603, 3 (2001) [arXiv:hep-th/0103067]. 38. E. Witten, Adv. Theor. Math. Phys. 5, 841 (2002) [arXiv:hep-th/0006010]. 39. R. Dijkgraaf and C. Vafa, Nucl. Phys. B 644, 3 (2002) [arXiv:hep-th/0206255]. 40. R. Dijkgraaf and C. Vafa, arXiv:hep-th/0208048. 41. R. Dijkgraaf, M.T. Grisaru, C.S. Lam, C. Vafa and D. Zanon, Phys. Lett. B 573, 138 (2003) [arXiv:hep-th/0211017]. 42. F. Cachazo, M.R. Douglas, N. Seiberg and E. Witten, JHEP 0212, 071 (2002) [arXiv:hepth/0211170]. 43. R. Gopakumar and C. Vafa, Adv. Theor. Math. Phys. 3, 1415 (1999) [arXiv:hep-th/9811131]. 44. I.R. Klebanov and M.J. Strassler, JHEP 0008, 052 (2000) [arXiv:hep-th/0007191]. 45. D. Berenstein, arXiv:hep-th/0303230. 46. I. Afeck, M. Dine and N. Seiberg, Nucl. Phys. B 241, 493 (1984). 47. D. Berenstein, Phys. Lett. B 552, 255 (2003) [arXiv:hep-th/0210183]. 48. L. Randall and R. Sundrum, Phys. Rev. Lett. 83, 3370 (1999) [arXiv:hep-ph/9905221]. 49. S.B. Giddings, S. Kachru and J. Polchinski, Phys. Rev. D 66, 106006 (2002) [arXiv:hep-th/0105097]. 50. M.R. Douglas, JHEP 0305, 046 (2003) [arXiv:hep-th/0303194]. 51. M.R. Douglas, arXiv:hep-ph/0401004. 52. S. Kachru, R. Kallosh, A. Linde and S.P. Trivedi, Phys. Rev. D 68, 046005 (2003) [arXiv:hepth/0301240]. 53. R. Bousso and J. Polchinski, JHEP 0006, 006 (2000) [arXiv:hep-th/0004134]. 54. F. Denef, M.R. Douglas and B. Florea, JHEP 0406, 034 (2004) [arXiv:hep-th/0404257]. 55. N. Arkani-Hamed and S. Dimopoulos, arXiv:hep-th/0405159. 56. L. Susskind, arXiv:hep-th/0405189. 57. M.R. Douglas, arXiv:hep-th/0405279. 58. M. Dine, E. Gorbatov and S. Thomas, arXiv:hep-th/0407043.