SUSY Gauge Theories
4.6 Open Strings on Orbifolds
In string theory there is a complementary way to study singular spaces, and that is by using D-brane probes. In the language of D-branes, the spacetime itself arises indirectly, as the moduli space of the gauge theory living on the D-brane world volume.
To describe a Calabi-Yau threefold, it is necessary to use D p-brane probes with p :::; 3.
For concretness we will henceforth set p
=
3, the other cases being related to it via dimensional reduction. The world volume theory must haveN
= 1 supersymmetry and the singularities in the moduli spaces are resolved by turning on Flayet-Iliopoulos parameters or by deformations of the superpotential.For a general X there is no known prescription of how to determine precisely the
world volume gauge theory, with the only requirement really being that the moduli space of a single D-brane on X should in fact be X itself. With this requirement alone we can say very few things. First, N D branes at a smooth point in the Calabi-Yau space will neccesarily be described, at energies E
« :, , R'
by U(N) gauge theory on the world volume, with effectiveN
= 4 supersymmetry. When the curvatures are large and substringy there must exist an effective gauge theory description of the"compactification" manifold, as long as the probe itself is small enough to be a "good"
probe of geometry. Beyond this one must aproach the problem on a case by case basis.
Exceptions to this are orbifolds X ~
C3
/f*7, where a simplification occurs because the theory onC3 /f
is a quotient of the theory on03*
8• Studies of CC2 /Zm, andC3
/Zmshowed that in both cases. the stringy constructions provide a physical realization of such concepts as Hyper-Kahler quotients and symplectic quotient constructions respectively. It is then interesting to ask if the same phenomena we have found in the closed string theory on
C3
/Zm x Zn will persist in the open string theory as well.First, let us briefly review the general construction of D-branes probing orbifolds.
Throughout we will mostly keep the discussion at the level of low-every effective field theory on the world volume. The theory of D 3-branes on
C3 /f
is defined as a truncation of the theory onC3,
where onlyr
invariant configurations are kept in the quotient.What does
r
invariance mean? Forgetting for the moment the non-Abelian nature of the theory (or more properly, thinking about open string CFT with boundary conditions), we can associate a Chan-Paton factor i to a D-brane at z(i) EC3,
then ther
action onC3
translates intogo z(i)
=
z'(J(g)i), Vg E f.*7ln general c3 can be replaced with some other Ricci-fiat three-dimensional manifold M admit- ting a symmetry r which is useful if the probe theory on M is known. Recently, this was done for the case when M is a conifold, [49].
*s It is true in fact that the knowledge of the c3
/r
theory allows one to describe all other sin- gualrities which are toric[50]. The unsolved problem is what to do for Calabi-Yau hypersurface singularities.A
r
invariant configuration of D branes must then consist of orbits ofr
action on<C3,
a generic orbit in this case consisting of
If I
points on<C3.
Thus, to describe N D 3-branes on
<C3
/f we must start with a d=
4,N =
4 supersymmetric U(Nlfl) gauge theory, as an effective open string theory of Nlfl branes on the covering space. The action ofr
on the open string CFT induces the action on the effective bosonic degrees of freedom of the formg A ---+ 1(g)A1(g)-1,
zi ---+ i(g) (go zi) i(gt1,
( 4.4)
(4.5)
where zi are the scalar fields whose diagonal pieces parametrize the position of the D-branes on the covering space, and 1(f) is an embedding of the orbifold group
/: f---+ U(Nlfl).
At the end of the day, if
r
is a subgroup of SU(3)*9 as in the case we are interested in, the quotient theory will have N=
1 supersymmetry in d=
4.Let's now take
r
= Zm x Zn, so thatr
is generated by two elements g, and h, satisfying gm = 1, hn = 1, gh = hg.<C3
/f is then a quotient of<C3
byg : (z,z1 2 ,z)-+(emz3 2rri 1 ,z,e mz) 2 - 2rri 3
h:
(z,z,z)-+(z,enz,e nz). 1 2 3 1 2rri 2 - 2rri 3Orbits of
r
are generated by g, h, so a D-brane at a generic point in<C3
/Zm x Zn must have mn preimages on<C3.
It is convenient to label D-branes with a biindex (i, a) where i, a naturally label points on the orbits generated by g, h respectively, so that i E {O, ... ,m-1} and a E {O ... ,n-1} *10•*9f is the holonomy group of the orbifold. If r E SU(3) upon resolution of singularities in the orbifold the holonomy becomes (at most) SU(3).
*10There are also N-valued indices labeling distinct physical branes, but since we will consider here only the generic orbits of r, whatever goes through for a single brane holds for any number of them.
We will thus set N
=
1 for clarity of the text.The orbifold fixes the action of
r
on the single open string states, which is inherited fromr
action onC3,
but we must still pick an action on the Chan-Paton factors.There is a natural "geometric" choice of action which can be described as follows.
Since a generic orbit of