hep-th/9710009

IASSNS-HEP-97/108

Why is the Matrix Model Correct?

arXiv:hep-th/9710009v1 1 Oct 1997

Nathan Seiberg1

School of Natural Sciences

Institute for Advanced Study
Princeton, NJ 08540, USA

We consider the compactification of M theory on a light-like circle as a limit of a com-

pactification on a small spatial circle boosted by a large amount. Assuming that the
compactification on a small spatial circle is weakly coupled type IIA theory, we derive
Susskind’s conjecture that M theory compactified on a light-like circle is given by the fi-
nite N version of the Matrix model of Banks, Fischler, Shenker and Susskind. This point
of view provides a uniform derivation of the Matrix model for M theory compactified on a
transverse torus T p for p = 0, ..., 5 and clarifies the difficulties for larger values of p.

10/97

1
[email protected]
 About a year ago Banks, Fischler, Shenker and Susskind (BFSS) [1] proposed an
amazingly simple conjecture relating M theory in the infinite momentum frame to a certain
quantum mechanical system. The extension to compactifications on tori T p for p = 1, ..., 5
was worked out in [1-5]. This proposal was based on the compactification of M theory on
N
a spatial circle of radius Rs in a sector with momentum P = Rs around that circle. In the
limit of small Rs M theory becomes the type IIA string theory and the lowest excitations
in the sector with momentum P are N D0-branes [6]. When the D0-brane velocities are
small and the string interactions are weak the D0-branes are described [7] by the minimal
supersymmetric Yang-Mills theory with sixteen supercharges (SYM). When the velocities
or the string coupling are not small, this minimal supersymmetric theory is corrected by
higher dimension operators. The suggestion of BFSS was that M theory in the infinite
momentum frame in uncompactified space is obtained by considering the minimal SYM
quantum mechanical system in the limit N, Rs , P → ∞. For compactification on T p the
proposal of [1,2] is to consider Dp-branes, which are described by SYM in p+1 dimensions,
and again to truncate to the minimal theory. This proposal raised a few questions:
1. Why is this proposal correct?
2. Why is the theory with small Rs related to the theory with large Rs ?
3. More specifically, the minimal supersymmetric theory is corrected by higher dimension
operators, which are important when Rs and the velocities are not small. Why is the
extrapolation from the minimal theory, which is valid at small Rs , correct for large
Rs ?
4. Furthermore, for p ≥ 4 the minimal theory is not renormalizable and hence it is ill
defined. Then, higher dimension operators must be included in the description. They
reflect the fact that the theory must be embedded in a larger theory with more degrees
of freedom. This theory for p = 4, 5 was found in [3-5]. The procedure to find these
extensions of the minimal theories did not appear systematic. What is then the rule
to construct the theory in different backgrounds?
Susskind noted that the finite N Matrix model enjoys some of the properties expected
to hold only in the large N limit and suggested that it is also physically meaningful [8].
He suggested that the matrix model describes M theory compactified on a light-like circle
N
of radius R with momentum P + = R. Such a compactification on a light-like circle with
finite momentum is known as the discrete light-cone and the quantization of this theory is
known as the discrete light-cone quantization (DLCQ). With a light-like circle the value
of R can be changed by a boost. Therefore, the uncompactified theory cannot be obtained

1
by simply taking R to infinity. Instead, it is obtained by taking R, N → ∞ holding P +
fixed.
In this note we will relate these two approaches to the Matrix theory. In the process
of doing so, we will derive the Matrix model and will answer the questions above. We
will also present a uniform derivation of the Matrix model for M theory on a compactified
transverse space.
We start by reviewing some trivial facts about relativistic kinematics. A compactifi-
cation on a light-like circle corresponds to the identification
     √R 
x x 2
∼ + , (1)
t t −√ R
2

where x is a spatial coordinate, e.g. x10 . We consider it as the limit of a compactification

on a space-like circle which is almost light-like
q
     R2 2    √R R2 
x x 2 + R s x 2
+ √s
2R
∼ + ≈ + (2)
t t − R
√ t − R
√
2 2

with Rs ≪ R. The light-like circle (1) is obtained from (2) as Rs → 0. This compactifica-
tion is related by a large boost with

R R2
β=p ≈ 1 − s2 (3)
R2 + 2Rs2 R

to a spatial compactification on
     
x x Rs
∼ + . (4)
t t 0

A longitudinal boost of the light-like circle (1) rescales the value of R. It also rescales
the value of the light-cone energy P − . Therefore P − is proportional to R. For small Rs
the value of P − in the system with the almost light-like circle (2) is also proportional to
R (an exception to that occurs when P − = 0 for the light-like circle; then P − can be
non-zero for the almost light-like circle). The boost (3) rescales P − to be independent of
R and of order Rs (if originally P − = 0, the resulting P − after the boost can be smaller
than order Rs ).
Following [9] (as referred to in [10]) we now consider M theory compactified on a light-
like circle (1) as the Rs → 0 limit of the compactification on an almost light-like circle (2)

2
or as the limit of the boosted circle (4). This way the DLCQ of M theory discussed in [8] is
related to the compactification on a small spatial circle as in [1]. For small Rs the theory
3
compactified on (4) is weakly coupled string theory with string coupling gs = (Rs MP ) 2 ,
and string scale Ms2 = Rs MP3 (MP is the Planck mass). For fixed energies and fixed MP ,
the limit Rs → 0 yields a complicated theory with vanishing string scale.
However, as we said above, starting with P − of order one, the effect of the boost is to
reduce P − to be of order Rs MP2 (MP2 is inserted on dimensional grounds). This is exactly
the range of energies in the discussion of [11], which was one of the motivations for the
Matrix model of BFSS [1]. In order to focus on the modes with such values of P − we
rescale the parameters of the theory. We do that by replacing the original M theory, which
is compactified on a light-like circle of radius R by another M theory, referred to as Mf
fP compactified on a spatial circle of radius Rs . The transverse
theory with Planck scale M
geometry of the original M theory is replaced by that of the Mf theory. For example, for
ei .
a compactification on a transverse torus with radii Ri the other theory has radii R
The relations between the parameters of these two theories are obtained by combining
the limit Rs → 0 with MfP → ∞ holding P − ∼ Rs M f2 fixed. Therefore we identify
P

f2 = RM 2 ,
Rs M (5)
P P

which is finite in the limit. Since the boost does not affect the transverse directions, we
identify
fP R
MP Ri = M ei (6)

and keep it fixed. In this limit the energies are finite and we find string theory with string
coupling and string scale

3 3 3
fP ) 2 = Rs4 (RM 2 ) 4
gs = (Rs M
e P
1
M f3 = Rs− 2 (RM 2 ) 23
f2 = Rs M (7)
s P P

For Rs → 0 with finite MP and R we recover weakly coupled string theory with large
string tension. This theory is very simple and is at the root of the simplification of the
Matrix model.
A sector with P + = N
R in the original M theory is mapped to a sector of momentum
N
P = Rs in the new M f theory. In terms of this latter theory it includes N D0-branes.

3
Therefore, the original M theory is mapped to the theory of D0-branes. These D0-branes
1
ei ∼ Rs2 → 0 (it is small even relative to the
move in a small transverse space of size R
1
ei M
string length R fs ∼ Rs4 → 0).

We conclude that M theory with Planck scale MP compactified on a light-like circle

of radius R and momentum P + = N f f
R is the same as M theory with Planck scale MP
compactified on a spatial circle (4) of radius Rs with N D0-branes in the limit

Rs → 0
fP → ∞
M
(8)
f2 = RM 2 = fixed
Rs M P P
fP R
M ei = MP Ri = fixed.

Here Ri should be understood as generic parameters in the transverse metric – not only
radii in a toroidal compactification. Clearly, the general discussion applies to curved space
with any number of unbroken supercharges.
For a compactification on T p we can use T duality to map the system of N D0-branes
to N Dp-branes on a torus with larger radii

1 1
Σi = = . (9)
ei M
R fs2 Ri RMP3

Note that Σi are finite when Rs → 0. The string coupling after this T duality transforma-
tion is
Y Y
fp
ges′ = ges M fp−3 R3 M 6
Σi = M Σi . (10)
s s P

The low energy dynamics of these N Dp-branes are controlled by p + 1 dimensional SYM
[7] with gauge coupling
ges′ Y
gY2 M = = R 3
M 6
P Σi , (11)
fsp−3
M
which also has a finite Rs → 0 limit. Even though the low energy dynamics of these
Dp-branes is finite in this limit, we should explore the behavior at higher energies. We will
do that shortly.
To summarize, we have mapped the original M theory problem with a light-like circle
N
of radius R and parameters MP and Ri in the sector with P + = R to a problem of N

4
Dp-branes wrapping a torus in string theory. The radii of the torus, the string coupling
and string scale are
1
Σi =
Ri RMP3
Y
fp−3 R3 M 6
ges′ = M Σi (12)
s P
1
f2 = Rs− 2 (RM 2 ) 23
M s P

and Rs → 0. Exactly this limit was analyzed recently in [12].

Let us analyze this limit for various values of p. For p = 0 the T duality which
we performed is not necessary. The theory is that of D0-branes with vanishing string
coupling and infinite string tension. Since the gauge coupling gY M is finite, the theory is
not trivial. The relevant degrees of freedom are strings stretched between the D0-branes.
The infinite string scale decouples all the oscillators on the strings. Therefore the full
theory is the minimal SYM theory. Note that closed strings or gravitons in the bulk of
space time decouple both because the string scale becomes large and because the string
coupling vanishes. This is exactly the finite N version of the Matrix model of BFSS [1].
For p = 1, 2, 3 we recover the SYM prescription of [1,2]. Again, the infinite string
scale decouples the oscillators on the strings which are stretched between the N Dp-branes.
Therefore the Lagrangian is that of the minimal SYM theory without higher order cor-
rections. For p = 3 the string coupling ges′ does not vanish, but there are still no higher
dimension operators in the 3 + 1 dimensional SYM, since they are all suppressed by inverse
fs .
powers of the string scale M
For p = 4 several new complications arise. First, the low energy SYM theory is not
renormalizable and therefore cannot give a complete description of the theory. It breaks
1
down at energies of order 2
gY
, where new degrees of freedom must be added. Second, the
M
string coupling also diverges in our limit. Therefore, in order to analyze the system we
f theory, which is an eleven dimensional
need to study the strong coupling limit of the M
theory. In this limit the D4-branes become 5-branes wrapping the eleventh dimension.
Using (12) we find that this eleven dimensional theory is compactified on a circle of finite
radius
ges′ Y
Σ5 = = R3 MP6 Σi , (13)
fs
M
but its eleven dimensional Planck scale diverges

fs
M − 16
1 ∼ Rs → ∞. (14)
(e
gs′ ) 3

5
Since the eleven dimensional Planck scale is infinite, the modes in the bulk of space-time
decouple, and the theory on the brane is a 5+1 dimensional theory. This non-trivial theory,
known as the (2, 0) field theory, was first found in [13,14]. The new degrees of freedom,
1
which we had to add at energy of order 2
gY
, can now be interpreted as associated with
M
momentum modes around the circle (13). These are related to instantons in the SYM
f theory. We have thus derived the proposal of [3,4]
theory, which are D0-branes in the M
to use this theory as a Matrix theory for M theory on T 4 .
A similar analysis applies to p = 5. Here we study the strong coupling limit of D5-
branes in IIB string theory. Using S duality of this theory we map it to NS5-branes in
weakly coupled IIB theory. We thus recover the proposal of [4,5] for the description of M
theory on T 5 in terms of a new theory obtained by studying NS5-branes in type II theory.
This theory, which can be called a non-critical string theory, is not a local quantum field
theory. In addition to the five sides of the torus Σi (9) it is characterized by the string
Q
slope α′ = gY2 M = R3 MP6 Σi . The SYM description breaks down at energies of order
1
gY M
, where new degrees of freedom are added. These degrees of freedom are the strings in
the theory. In terms of the Mf theory and its type IIB string theory, these are D1-branes.
For p = 6 the situation is more complicated [15]. Here we are led to consider the
strong coupling limit of D6-branes1 . As for p = 4, this limit is described by an eleven
dimensional theory. However, here the eleven dimensional Planck scale

fs
M 1
1 = Q 1 (15)
(e
gs′ ) 3 RMP2 ( Σi ) 3

remains finite, but the radius of the eleventh dimension diverges

ges′ −1
∼ Rs 2 → ∞. (16)
fs
M

Since the radius diverges, the D6-branes, which are Kaluza-Klein monopoles associated
with the eleventh dimension, expand and become an AN−1 singularity. The gauge coupling
of the associated SYM theory is given by the eleven dimensional Planck scale, which
remains finite. Since the eleven dimensional Planck scale is finite, there is no reason
to assume that the gravitons in the bulk of the ALE space with an AN−1 singularity
f theory point of view (before
decouple from it (see the discussion in [16-18]). From the M
the T duality transformation) these graviton modes can be identified as Kaluza-Klein

1
We thank E. Witten for very helpful discussions on this limit.

6
monopoles2 , which wrap the small T 6 . Their energy is of order Rs2 . After the boost (3)
their energy is of order Rs and it vanishes as Rs → 0. Therefore, the DLCQ theory has
an infinite number of new massless modes.
N
We conclude that M theory on T 6 and a light-like circle with momentum P + = R is
the same as M theory with Planck scale (15) compactified on T 6 with an AN−1 singularity
in the non-compact spatial directions. The eleven dimensional gravitons propagate in the
entire space. There is also an SU (N ) SYM in 6+1 dimensions describing the interactions
of gluons at the singularity. Comparing with the situation for lower values of p, we seem to
miss a decoupled 6+1 dimensional U (1) multiplet. However, considering the limit which
leads to this configuration carefully, we see that the U (1) multiplet exists. It is “smeared”
over the four dimensional non-compact space.
Unfortunately, this result is not satisfying. The Matrix theory offered a simple de-
scription of M theory, which can lead to useful computations. Here we see that for p = 6,
it goes over to a situation which is apparently as complicated as the underlying M theory.
After the completion of this work we received a paper [19] which partially overlaps
with ours. We also learned that J. Polchinski had independently reached some of these
conclusions.

Acknowledgements
We have benefited from discussions with O. Aharony, T. Banks, J. Schwarz, S. Shenker,
L. Susskind and E. Witten. This work is supported in part by #DE-FG02-90ER40542.

2
This interpretation is due to L. Susskind.

7
 References

[1] T. Banks, W. Fischler, S. H. Shenker, and L. Susskind, “M Theory As A Matrix

Model: A Conjecture,” hep-th/9610043, Phys. Rev. D55 (1997) 5112.
[2] W. Taylor IV, “D-Brane Field Theory on Compact Space,” hep-th/9611042, Phys.
Lett. B394 (1997) 283.
[3] M. Rozali, “Matrix Theory and U-Duality in Seven Dimensions,” hep-th/9702136
Phys. Lett. B400 (1997) 260.
[4] M. Berkooz, M. Rozali and N. Seiberg, “Matrix Description of M theory on T 4 and
T 5 ”, hep-th/9704089.
[5] N. Seiberg, “New Theories in Six Dimensions and Matrix Description of M-theory on
T 5 and T 5 /ZZ2 ,” hep-th/9705221.
[6] J. Polchinski, “TASI Lectures on D-Branes,” hep-th/9611050;
J. Polchinski, S. Chaudhuri and C. Johnson, “Notes on D-Branes,” hep-th/9602052.
[7] E. Witten, “Bound States of Strings and P-branes”, Nucl. Phys. B460 (1996) 335,
hep-th/9510135.
[8] L. Susskind, “Another Conjecture about M(atrix) Theory,” hep-th/9704080.
[9] S. Hellerman and J. Polchinski, work in progress.
[10] K. Becker, M. Becker, J. Polchinski and A. Tseytlin, “Higher Order Graviton Scat-
tering in M(atrix) Theory,” Phys.Rev. D56 (1997), 3174, hep-th/9706072.
[11] M. Douglas, D. Kabat, P. Pouliot, and S. Shenker, “D-branes and Short Distances in
String Theory”, Nucl. Phys. B485 (1997) 85, hep-th/9608024.
[12] J.M. Maldacena, “Branes Probing Black Holes,” hep-th/9709099.
[13] E. Witten, “Some Comments on String Dynamics,” hep-th/9507121.
[14] A. Strominger, “Open P-Branes,” hep-th/9512059, Phys. Lett. B383 (1996) 44.
[15] O. Ganor, “On the M(atrix) Model for M Theory on T 6 ,” hep-th/9709139.
[16] A. Losev, G. Moore, and S. Shatashvili, “M & m’s ,” hep-th/9707250.
[17] I. Brunner and A. Karch, “Matrix Description of M-theory on T 6 ,” hep-th/9707259.
[18] A. Hanany and G. Lifschytz, “M(atrix) Theory on T 6 and a m(atrix) Theory Descrip-
tion of KK Monopoles,” hep-th/9708037.
[19] A. Sen, “D0-Branes on T n and Matrix Theory,” hep-th/9709220.