hep-th/9704041

CALT-68-2107

Heterotic Matrix String Theory

arXiv:hep-th/9704041v2 14 Apr 1997

David A. Lowe

California Institute of Technology

Pasadena, CA 91125, USA
lowe@[Link]

M-theory suggests the large N limit of the matrix description of a collection of N Type
IA D-particles should provide a nonperturbative formulation of heterotic string theory. In
this paper states in the matrix theory corresponding to fundamental heterotic strings
are identified, and their interactions are studied. Comments are made about analogous
states in Type IIA string theory, which correspond to bound states of D-particles and
D-eightbranes.

April, 1997
1. Introduction

D-brane techniques [1] provide us with new ways to study the nonperturbative dynam-
ics of string theory. One may use these methods to test the web of strong/weak coupling
dualities that have been proposed in recent years [2]. Typically, these dualities have their
simplest interpretation when the theories are viewed as compactifications of a hypotheti-
cal eleven-dimensional theory known as M-theory [3]. D-brane techniques therefore yield
much information about the structure of M-theory. In fact, it has been conjectured the full
dynamics of M-theory with the 11th dimension decompactified may be described by the
large N matrix quantum mechanics of a system of N D-particles of Type IIA string theory
[4]. One can carry the same ideas over to the compactification of M-theory on S 1 /ZZ2 which
has been conjectured to describe the strongly-coupled dynamics of the E8 × E8 heterotic
string [5], and attempt to describe the dynamics of this system using the large N limit of
the matrix quantum mechanics of Type IA D-particles1 [7–10].
In this paper, we will continue our study of this system [9], elaborating on the spectrum
of states and their interactions. Applying a T-duality transformation along the S 1 /ZZ2
direction, the supersymmetric quantum mechanics is recast into the form of a gauge theory
in two dimensions with (0, 8) supersymmetry. The coupling constant of this gauge theory
g scales with the length, thus in the infrared this gauge theory is expected to flow to some
nontrivial conformal field theory, which we identify as a SN orbifold heterotic sigma-model.
The spectrum of the heterotic string is recovered in this formulation, and the interactions
of these states are studied. One may also consider the limit in which g is held fixed and one
treats the gauge potential as a slowly varying degree of freedom in a Born-Oppenheimer
approximation. The states required by equivalence with M-theory on S 1 /ZZ2 duly appear
in this limit. We conclude with some comments about analogous bound states of D-branes
in Type II string theory.

2. Matrix Field Theory

Our starting point will be the two-dimensional gauge theory describing the light ex-
citations of N coincident D-strings in Type IB. This system is T-dual to the system of

1
In this paper we will use the notation Type IB to refer to the usual Type I string theory, and
Type IA to refer to the theory obtained by T-dualizing Type I on S 1 [6], which has previously
been referred to as Type I′ string theory.

1
D-particles in Type IA studied in [7–9]. The relation with E8 × E8 heterotic string theory
compactified on a circle (the 9 direction) may be seen by the following chain of dualities.
T−duality
IA D − particle # ←→ IB D − string winding #
x x
(2.1)
9−11 flip S−duality
y y
T−duality
E8 × E8 Heterotic p+ ←→ SO(32) Heterotic winding #

The two-dimensional gauge theory is obtained from the massless sector of open strings
ending on D-strings and D-ninebranes, using conventional D-brane techniques [11]. The
interactions of these fields are fixed by the (0, 8) supersymmetry. The answer one obtains
is the O(N ) gauge theory with action

1
Z
S= Tr (Dµ X)2 − iθ+
T
D− θ+ + gs2 F 2 − igs2 λ− D+ λ− +
2π
 (2.2)
i 1 i j 2
2iθ+ λ− γi X − iχD+ χ + 2 [X , X ] ,
gs

in units where α′ = 1, gs is the string coupling for E8 × E8 heterotic strings, and the
γi are defined as in [12]. Here the eight scalars X i and their superpartners, the right-
moving fermionic fields θ+ lie in the symmetric rep of O(N ). This theory has a Spin(8)
R-symmetry group corresponding to the group of rotations in spacetime. The θ+ fields
transform as the 8c while the X i transform as the 8v . The gauge multiplet lies in the
adjoint of O(N ) and consists of a gauge boson Aµ and a set of eight left-moving fermionic
fields λ− which transform in the 8s . The χ fields are purely left-moving and transform as
the fundamental rep of O(N ) and in the fundamental of the SO(32) group of spacetime
gauge symmetry. The necessity of including the χ fields may be seen either from the
point of view of spacetime anomaly cancellation in Type I string theory [13,6,14], where a
background of 32 ninebranes is required, giving rise to the χ fields as 1-9 open strings, or
from the point of view of anomaly cancellation in the two-dimensional gauge theory [15].
It is natural to conjecture that this theory provides a nonperturbative definition of
uncompactified heterotic string theory. However, we know from the work of Polchinski
and Witten [16] that Type IB perturbation theory, on which this description is based, will
typically break down as the radius of the S 1 is varied due to dilaton and graviton tadpoles
not cancelling winding number by winding number. The only way to avoid this breakdown
of perturbation theory is to choose a Wilson line for the Spin(32)/ZZ2 degrees of freedom
corresponding to
1
W = (( )8 , 08 ) , (2.3)
2
2
in standard notation. In Type IA language this corresponds to putting eight D-eightbranes
on each of the orientifold planes. The SO(32) gauge symmetry is broken to SO(16) ×
SO(16). The Wilson line splits the χ degrees of freedom into two sets of sixteen, which
will have boundary conditions differing by a sign as one goes around the S 1 direction.
We have expressed the action (2.2) in terms of the string coupling of the heterotic
string. This is related to the gauge theory coupling by gs = 1/g and gs scales inversely with
the worldsheet length scale. The perturbative heterotic string theory should therefore be
recovered in the infrared limit. In this limit the gauge theory should flow to some nontrivial
superconformal fixed point. We conjecture this superconformal field theory is the heterotic
sigma model with (0, 8) supersymmetry based on the orbifold theory

(IR8 )N /(SN × (ZZ2 )N ) , (2.4)

where the SN acts by permuting the IR8 , their fermionic partners and gauge fermion degrees
of freedom. The ZZ2 ’s act by reflecting the different components of the O(N ) vectors. In
the free-string limit the space of states will be shown to correspond to the Fock space of
second-quantized heterotic strings.

3. Spectrum of States

3.1. Heterotic Strings

The techniques of [17–20] carry over straightforwardly to the present case. The Hilbert
space of the SN orbifold is decomposed into twisted sectors for each of the conjugacy classes
[g] of SN
H = ⊕[g] H[g] . (3.1)

These conjugacy classes may be represented as cycle decompositions and then the Hilbert
spaces H[g] may be further decomposed as symmetric products of Hilbert spaces H(n)
where H(n) is the ZZn invariant subspace of a single heterotic string on IR8 × S 1 with
winding number n and Wilson line W . Modding out by the additional ZZ2 ’s is achieved by
extra GSO projections acting on the gauge fermions.
This space of states is represented by n copies of the fields xiI (σ) and χrI (σ) with
I = 1, · · · , n and σ ∈ [0, 2π], with the cyclic boundary conditions

xiI (σ + 2π) = xiI+1 (σ)

(3.2)
χrI (σ + 2π) = ǫI OW χr (σ) ,

3
where ǫI = ±1 corresponds to reflection elements of O(N ) and OW is the action of the
Wilson line W on the fermions χ which is simply multiplication by −1 for r = 1 · · · 16 and
+1 for r = 17, · · · , 32. These fields may be combined into a single set of fields xi (σ) and
χr (σ) living on σ ∈ [0, 2πn] where

x(σ + 2πn) = x(σ)

(3.3)
χr (σ + 2πn) = ǫ(OW )n χr (σ) ,

and ǫ = ±1. We label the sectors with different boundary conditions on the χ’s by P for
periodic and A for antiperiodic. Since the first group of 16 χ’s can have different boundary
conditions to the second group, we need one label for each group. For n even, we have the
P P and AA sectors of the usual heterotic string, while n odd gives the AP and P A sectors.
The vacuum energy vanishes in the AP and P A twisted sectors, while in the twisted sector
with P P boundary conditions the left-movers make a nonzero contribution equal to 1/n,
and in the AA sector −1/n. When we rescale L0 + L̄0 by a factor of n so it is canonically
normalized with respect to xi (σ) the vacuum energy is 1 in the P P sector and −1 in the
AA sector as expected for a single heterotic string. The usual E8 × E8 heterotic string
has two GSO projections consisting of keeping states invariant under (−1)F1 and (−1)F2 ,
where (−1)F1 anticommutes with the first group of χr and commutes with the second,
and vice-versa for (−1)F2 . The Spin(32)/ZZ2 heterotic theory on the other hand has just
a single GSO projection consisting of (−1)F1 +F2 . This projection is to be identified with
a ZZ2 element of O(N ) which acts as −1 on vectors χrI .
Fundamental heterotic strings are obtained in a large N limit, by considering the
twisted sector for some cycle of length n with n/N finite. This corresponds to considering
a string carrying a finite longitudinal momentum p+ = n/N . Invariance under ZZn implies
that L0 − L̄0 is a multiple of n. The mass of such a state diverges as N → ∞ unless
L0 − L̄0 acting on the state vanishes. In this way, the usual level-matching condition of
the heterotic string is recovered.
What happened to the second GSO projection of the E8 × E8 heterotic string? The
theory we have constructed describes E8 × E8 heterotic strings compactified on some large
S 1 with n quanta of Kaluza-Klein momentum present. This is equivalent via T-duality to
an Spin(32)/ZZ2 heterotic string with winding number n compactified on a small S 1 with
the Wilson line W present. The worldsheet fields that emerge from the sigma model are
naturally identified with these fields, hence only one GSO projection appears. To take the
large N limit it is convenient to T -dualize the sigma model fields, which introduces the
second GSO projection, as in the usual E8 × E8 theory. We see therefore that the Hilbert
space H corresponds to the second-quantized Fock space of free E8 × E8 heterotic strings.

4
3.2. Type IA BPS Bound States

When the right-movers of the heterotic strings are in their ground states, the states
found above will be BPS saturated and using the 9-11 flip duality (2.1) may be reinter-
preted as BPS bound states of Type IA D-particles (and tensor products of such states).
The states we will be interested in here break half of the spacetime supersymmetry and
correspond to bound states at threshold of Type IA D-particles. The infrared limit of
the gauge theory amounts to considering the Born-Oppenheimer approximation in the su-
persymmetric quantum mechanics of D-particles on the branch of the moduli space when
they are stuck to one of the orientifold planes. The above results for the BPS heterotic
string states found above carry over immediately to this case. The BPS bound states of
N D-particles are to be identified with the massless states in the ZZN twisted sector. For
N even one finds the adjoint of SO(16) × SO(16) in the 8v + 8s , and a gauge singlet in
the 8v × (8v + 8s ) coming from the AA sector, while for N odd the (1, 128) + (128, 1) in
the 8v + 8s , is found in the AP and P A sectors.
One may also consider the branch when the D-particles move off of the orientifold
plane. On this branch of the moduli space the scalar A coming from the gauge field up
in two dimensions is treated as a slow variable in the Born-Oppenheimer approximation.
For the moment, let us focus on the case of two D-particles considered in [9]. There are
two distinct limits to be considered. When l11 ≪ A ≪ ls , where ls is the string length,
the relevant quantum mechanics is obtained by dimensionally reducing the N = 2 system
in two dimensions. Different boundary conditions for the χ fields in two dimensions give
rise to different sectors of the SUSY quantum mechanics with different Hamiltonia. In the
sector where all the χ fields satisfy antiperiodic boundary conditions the states found [9]
included the adjoint of SO(16) × SO(16) in the

8v + 8s (3.4)

and a gauge singlet in the

8v × (8v + 8s ) . (3.5)

The quantum numbers of these states match the massless states coming from the AA sector
of the CFT discussed above. This is a necessary condition for the exact wavefunctions to
smoothly match as one moves from one branch of the moduli space to the other. The one-
loop corrections to the effective kinetic term for A [7] are of the form gIA Ȧ2 /A3 , indicating
that corrections to this approximation set in at the eleven-dimensional Planck scale l11 . For
l11 ≪ A ≪ ls the wavefunction of these states should vary as c1 A + c2 where c1 and c2 are
constants, which can in principle be determined by matching onto the behavior as A → l11

5
and as A → ls . We know these states should spread onto the branch of moduli space at
A = 0, so we expect c2 to be nonzero. In the limit A ≫ ls , the dominant interaction is
closed string exchange, and we expect the Hamiltonian to flow to that describing a Type
IIA D-particle, interacting with its mirror image [7]. The χ modes will become massive,
and completely decouple from this Hamiltonian. The zero energy states will correspond to
a singlet of the gauge group in the (8v + 8s ) × (8v + 8c ) of Spin(8) which are the quantum
numbers of the ground state of a single Type IIA D-particle. This state does not match
onto the charged states found for A ≪ ls , so the wavefunction of (3.4) should vanish when
A ≫ ls . In this way, we see the states (3.4) are localized near the orientifold plane.
In the sector when all the χ fields satisfy periodic boundary conditions one finds all
states become massive, and this sector decouples from the low-energy description of the
D-particles. When half of the χ’s satisfy periodic boundary conditions, and half are an-
tiperiodic, a similar calculation to the one in [9] shows the normal-ordering terms appearing
in the term in the Hamiltonian linear in A vanish. In particular, the χ fields become mas-
sive and may be integrated out leaving states that are singlets under SO(16)×SO(16). The
states found in the Born-Oppenheimer approximation arise from quantizing the fermion
zero modes arising from the trace part of θ+ and the λ− , the left-moving superpartners of
A. This gives rise to the (8v + 8s ) × (8v + 8c ) of Spin(8) which are the quantum num-
bers of the ground state of a single Type IIA D-particle - as expected from duality with
M-theory. Note in this case the Hamiltonian describing these degrees of freedom is the
same for all A ≫ l11 . These states are not localized near the orientifold planes.
Let us now generalize these results to arbitrary numbers of D-particles. For N even,
we can choose A to break O(N ) → U (1) × SU (N/2). The U (1) component of A may be
treated in a similar way to the N = 2 case already discussed. In the sector where the
χ’s satisfy AP or P A boundary conditions, the other light degrees of freedom correspond
to the SU (N/2) quantum mechanics of Type IIA D-particles with fixed center of mass.
The Hamiltonian for these slow modes computed in the Born-Oppenheimer approximation
matches that of N/2 Type IIA D-particles. This is to be expected from the M-theory point
of view – far from the ends of interval, the states should match that of uncompactified
M-theory. In section 5 we will comment on finding BPS bound states for this system. In
the sector where the χ’s obey AA boundary conditions the U (1) component of A behaves
as for the N = 2 case, and again the other degrees of freedom again take the form of
the SU (N/2) quantum mechanics of N/2 Type IIA D-particles. Assuming this SU (N/2)
quantum mechanics yields one L2 normalizable bound state, one finds BPS bound states
which match onto the charged states appearing on the other branch of the moduli space
(i.e. the adjoint and gauge singlet states) and will be peaked near the orientifold plane

6
by the argument given above. Finally when χ’s obey P P boundary conditions all states
become massive and hence non-BPS.
For N = 1 the D-particle is always stuck to the orientifold, and the BPS states are
just the massless states of the CFT found in the previous section. Only AP and P A
boundary conditions are possible for this case and as mentioned before, one finds the
(1, 128) + (128, 1) of SO(16) × SO(16). For general N > 1 odd, A can be chosen to break
O(N ) → ZZ2 × U (1) × SU ((N − 1)/2). This branch of the moduli space describes one D-
particle stuck on one of the orientifold planes, and (N −1)/2 off of the orientifold. The slow
modes consist of the χ modes arising from open strings running between D-eightbranes
and the D-particle on the orientifold plane, and modes identical to (N − 1)/2 Type IIA
D-particles. The χ zero modes imply such states will always carry the spinor charges of
SO(16) × SO(16). The quantum numbers of these match the N odd states on the other
branch discussed above. The U (1) component of the gauge multiplet will behave in the
same way as the N = 2 case above, and the wavefunctions of these states will be localized
near the orientifold planes.

4. Heterotic String Interactions

The two-dimensional gauge theory (2.2) in the infrared limit gives rise to the Fock
space of free heterotic strings. Interaction terms should correspond to irrelevant perturba-
tions of the orbifold sigma model. Dijkgraaf, Verlinde, Verlinde [18] considered such terms
in the Type II case, and much of their argumentation applies here. The idea is that the
interactions correspond to transpositions of the xI eigenvalues when they coincide. In the
heterotic case one must also match the gauge fermion degrees of freedom, and transpose
these at the same time. This will give rise to splitting and joining interactions of the long
strings discussed above.
For the right-moving sector, the construction of the interaction vertex precisely mimics
that of [18]. We consider two eigenvalues x1 and x2 which are interchanged under a ZZ2
twist. In terms of the linear combinations x± = x1 ± x2 , the ZZ2 flips the sign of x− (and
likewise for the right-moving fermionic modes θ). This ZZ2 orbifold is well-known, and the
twist operators are defined by the operator product relations
1
∂xi− (z) · σ(0) ∼ z − 2 τ i (0)
1
a
θ− (z) · Σi (0) ∼ z − 2 γai ȧ Σȧ (4.1)
1
a
θ− (z) · Σȧ (0) ∼ z − 2 γai ȧ Σi .

7
One can then show [18] that the operator

VR = τ i Σi , (4.2)

is the unique, least irrelevant perturbation that preserves Spin(8) spacetime rotations and
spacetime supersymmetry. This operator has conformal weight (0, 32 ).
Now consider the left-moving sector, and bosonize the gauge fermion modes χ, to yield
sixteen bosonic coordinates xM , with M = 1, · · · , 16 living on the Cartan torus of the gauge
group. We define the ZZ2 eigenvectors xM M M
± = x1 ± x2 . The twist operator for the left-
moving spacetime coordinates x− (z̄) is constructed as above. One can also define the twist
operators for the xM − in a similar way, but now one must remember the x
M
are compact
M M
bosonic coordinates. The action x− → −x− has two fixed points, so we have two twist
operators σ̄± for each xM . We now wish to construct the least irrelevant perturbation to
the CFT preserving Spin(8) rotations, spacetime supersymmetry, and gauge invariance.
We also demand that the perturbation be invariant under z → eiθ z, which means the
conformal weight of the left-moving piece should be ( 32 , 0). The combination of left-moving
twist operators which satisfies these conditions is

X 16
Y
VL = σ̄(0) · σ̄sMM , (4.3)
M =1

where sM = ± and the sum is over all permutations of the sM .

The final interaction term that appears in the sigma model is then a sum over pairs
of eigenvalues that may be interchanged
XZ
Sint = λ d2 z(VL ⊗ VR )IJ , (4.4)
I<J

where I, J = 1, · · · , N . The coupling λ will scale inversely with the worldsheet length
scale, therefore λ will scale linearly with gs . The interaction corrections contained in the
supersymmetric gauge theory (2.2) therefore reproduce heterotic string interactions to first
order in gs .

5. Comments on Type II Bound States

5.1. Bound states of N D-particles

We saw above that the quantum mechanics describing Type IA D-particles far from
the orientifold plane reduced to that of a collection of Type IIA D-particles. The quantum

8
mechanics of Type IIA D-particles has previously been studied in [21–25]. For M(atrix)
theory to be correct for the case of eleven uncompactified dimensions, these Type IIA D-
particles must form bound states at threshold. The prediction is that one normalizable
bound state at threshold of N D-particles appears for every N , and this bound state lies
in an ultra-short multiplet of the ten-dimensional Type IIA supergravity.
If such a normalizable bound state appears in uncompactified Type IIA, then it should
show up as a normalizable bound state when we further compactify on a torus. We can
then recast the problem into a statement about the ground states of d = 2 SU (N ) Yang-
Mills theory with (8, 8) supersymmetry (here we have factored out the U (1) describing
the center of mass degrees of freedom). That is, there should be one L2 normalizable
zero-energy ground state which appears regardless of the boundary conditions [26].
We can make a heuristic argument for the existence of such a state as follows. Since
the state is normalizable (in fact it should fall off as 1/r 7 up in ten dimensions), it should
remain a bound state if we add a perturbation that only effects the large distance behavior
of the fields. In fact, if we choose the perturbation judiciously, the bound state at threshold
should become a bound state with mass gap.
The two-dimensional super Yang-Mills theory can be thought of as the dimensional
reduction of a d = 4 N = 4 theory. The vector multiplet in four dimensions can be
decomposed into a single N = 1 vector multiplet and three N = 1 chiral multiplets which
we denote by X, Y and Z, all in the adjoint rep of the gauge group. Let us assume the
perturbation of the superpotential parametrized by m
 
1 2 2 2
W = Tr X[Y, Z] − m(X + Y + Z ) , (5.1)
2

behaves in the way described. We may now use a variant of the arguments of [27]. Since
the ground state we are interested in now has a mass gap, we can actually treat m as being
large, and use semiclassical methods.
The semiclassical vacua correspond to critical points of the superpotential (5.1) which
are solutions of
[X, Y ] = mZ
[Y, Z] = mX (5.2)
[Z, X] = mY .
These equations are just the commutation relations of SU (2). There three are classes of
solutions to these equations: (i) trivial solution for which X = Y = Z = 0, (ii) non-
trivial reducible embedding of SU (2) into SU (N ), (iii) irreducible embedding of SU (2)
into SU (N ). For cases (i) and (ii) part of the gauge symmetry is unbroken and one is

9
left with a two-dimensional gauge theory with (2, 2) supersymmetry. These vacua will not
have mass gaps due to massless fermions. Case (iii) however leads to a solution which
completely breaks the gauge symmetry. In this case all fields become massive and one is
left with a single bosonic ground state. Including the fermion zero modes coming from the
center of mass degrees of freedom, one obtains a single bound state of N D-particles with
the expected spacetime quantum numbers.

5.2. Bound states of N D-particles and M D-eightbranes

Studies of solitonic solutions of Type IIA supergravity do not show any sign of super-
symmetric bound states of D-particles and D-eightbranes [28]. We can now understand
this from the point of view of the two-dimensional supersymmetric gauge theory describing
this system. As in the Type IA case, we have a theory with (0, 8) supersymmetry, this time
with U (N ) gauge multiplet (A, λ− ). The matter content consists of multiplets (X i , θ+ )
in the adjoint of U (N ), and left-moving fermions in the (N, M̄ ) + (N̄ , M ). Consider the
conformal field theory this will flow to in the infrared. One will be left with the fields
living on the Cartan torus of U (N ) corresponding to the diagonal components of X i , θ+
and λ− together with the other left-moving fermions. For all N and M this conformal
field theory will not have a massless ground state, hence no BPS bound state is possible.
Massless ground states are possible if one considers for example two sets of separated D-
eightbranes. The condition then is that there is a choice of boundary conditions on the
left-moving fundamental fermions which gives zero contribution to the vacuum energy.

Acknowledgments
I wish to thank M. Douglas, G. Papadopoulos, J. Schwarz, A. Sen and C. Vafa for
helpful discussions. This work was supported in part by DOE grant DE-FG03-92-ER40701.

Note added
While this manuscript was being completed the preprint [29] appeared in which closely
related results are reported. I understand related ideas have also been considered by R.
Dijkgraaf and S.J. Rey and by P. Horava.

10
 References

[1] J. Polchinski, “TASI Lectures on D-Branes,” hep-th/9611050; J. Polchinski, S. Chaud-

huri and C. Johnson, “Notes on D-Branes,” hep-th/9602052.
[2] E. Witten, “String Theory Dynamics in Various Dimensions,” Nucl. Phys. B443
(1995) 85, hep-th/9503124; C. Hull and P. Townsend “Unity of Superstring Duali-
ties,” Nucl. Phys. B438 (1995) 109, hep-th/9410167.
[3] J. Schwarz, “The Power of M-Theory,” Phys. Lett. B367 (1996) 97, hep-th/9510086.
[4] T. Banks, W. Fischler, S. Shenker and L. Susskind, “M-Theory as a Matrix Model: A
Conjecture,” hep-th/9610043.
[5] P. Horava and E. Witten, “Heterotic and Type I String Dynamics from Eleven Di-
mensions,” Nucl. Phys. B460 (1996) 506, hep-th/9510209.
[6] P. Horava, “Strings on Worldsheet Orbifolds,” Nucl. Phys. B327 (1989) 461; J. Dai,
R.G. Leigh and J. Polchinski, “New Connections Between String Theories,” Mod.
Phys. Lett. A4 (1989) 2073; P. Horava, “Background Duality of Open-String Models,”
Phys. Lett. B231 (1989) 251.
[7] U. Danielsson and G. Ferretti, “The Heterotic Life of the D-particle,” hep-th/9610082.
[8] S. Kachru and E. Silverstein, “On Gauge Bosons in the Matrix Model Approach to
M Theory,” hep-th/9612162.
[9] D.A. Lowe, “Bound States of Type I′ D-particles and Enhanced Gauge Symmetry,”
hep-th/9702006.
[10] N. Kim and S.J. Rey, “M(atrix) Theory on an Orbifold and Twisted Membrane,”
hep-th/9701139.
[11] M.R. Douglas, “Branes within Branes,” hep-th/9512077.
[12] M. Green, J. Schwarz and E. Witten, “Superstring Theory,” Cambridge University
Press, 1987.
[13] M. Green and J. Schwarz, “The hexagon gauge anomaly in Type I superstring theory,”
Nucl. Phys. B255 (1985) 93.
[14] Y. Arakane, H. Itoyama, H. Kunitomo and A. Tokura, “Infinity Cancellation, Type I′
Compactification and String S-Matrix Functional,” hep-th/9609151.
[15] T. Banks, S. Seiberg and E. Silverstein, “Zero and One-dimensional Probes with N=8
Supersymmetry,” hep-th/9703052.
[16] J. Polchinski and E. Witten, “Evidence for Heterotic-Type I String Duality,” Nucl.
Phys. B460 (1996) 525, hep-th/9510169.
[17] R. Dijkgraaf, G. Moore, E. Verlinde and H. Verlinde, “Elliptic Genera of Symmetric
Products and Second Quantized Strings,” hep-th/9608096.
[18] R. Dijkgraaf, E. Verlinde and H. Verlinde, “Matrix String Theory,” hep-th/9703030.
[19] T. Banks and N. Seiberg, “Strings from Matrices,” hep-th/9702187.
[20] L. Motl, “Proposals on Nonperturbative Superstring Interactions,” hep-th/9701025.

11
[21] U. Danielsson, G. Ferretti and B. Sundborg, “D-particle Dynamics and Bound States,”
Int. Jour. Mod. Phys. A11 (1996) 5463, hep-th/9603081.
[22] D. Kabat and P. Pouliot, “A Comment on Zerobrane Quantum Mechanics,” Phys.
Rev. Lett. 77 (1996) 1004, hep-th/9603127.
[23] M.R. Douglas, D. Kabat, P. Pouliot and S.H. Shenker, “D-branes and Short Distances
in String Theory,” Nucl. Phys. B485 (1997) 85, hep-th/9608024.
[24] M. Claudson and M.B. Halpern, “Supersymmetric Ground State Wave Functions,”
Nucl. Phys. B250 (1985) 689; M. Baake, P. Reinicke and V. Rittenberg, “Fierz Iden-
tities for Real Clifford Algebras and the Number of Supercharges,” J. Math. Phys. 26
(1985) 1070; R. Flume, “On Quantum Mechanics with Extended Supersymmetry and
Nonabelian Gauge Constraints,” Ann. Phys. 164 (1985) 189.
[25] B. de Wit, J. Hoppe, H. Nicolai, “On the Quantum Mechanics of Supermembranes,”
Nucl. Phys. B305 (1988) 545.
[26] A. Sen, “A Note on Marginally Stable Bound States in Type II String Theory,” Phys.
Rev. D54 (1996) 2964; hep-th/9510229.
[27] E. Witten, “Bound States of Strings and p-Branes,” Nucl. Phys. B460 (1996) 335,
hep-th/9510135.
[28] G. Papadopoulos and P.K. Townsend, “Kaluza-Klein on the Brane,” Phys. Lett. B393
(1997) 59, hep-th/9609095.
[29] T. Banks and L. Motl, “Heterotic Strings from Matrices,” hep-th/9703218.

12