T-duality of open strings and D-branes

John Joseph Marchetta

March 18, 2018

1 Introduction
String theory, to say the least, gets even more interesting when dimensions become com-
pactified. In closed string theory, new physics emerge due to the fact closed strings can
wrap around circular dimensions as well as live on top of them. This displays a remarkable
0
symmetry: if you invert the radius of the compact dimension from R → αR , and exchange
the number of times the string wraps around the dimension with its quantized momentum;
both pictures become indistinguishable. This is T(toroidal)-duality and the above claims
will be laid out in Section 2. The rest of this paper will be on T-duality of open strings
and D-branes. In open string theory, the endpoints interchange Neumann and Dirichlet
boundary conditions under T-duality for each compact dimension of spacetime. This al-
ters the dimension of the brane they live on. For example, if the string has Neumann
boundary conditions on a Dp-brane wrapped in a circular dimension, in the dual world
the string would live on a D(p-1)-brane that is located at a specific point on that circle. It
turns out that this position is set by Wilson line of the Dp-brane covering the compactified
dimension. T-duality reveals some extremely remarkable (and cool) properties about elec-
tromagentic fields on D-branes. In section 6.1, we discover there is a relationship between
the strength of a constant electric field on the Dp-brane and the velocity of the dual D(p-
1)-brane traveling around the circular dimension. Since the brane must travel less than the
speed of light, there is a cutoff maximum electric field strength which happens to equal the
tension of the string. We conclude from this that D-branes must be govern by Born-Infeld
electrodynamics. In section 6.2, we will replace the electric field with a constant magnetic
field and see that, in the dual world, our brane gets rotated instead of boosted with the
titling angle determined by the magnetic field strength. The D(p-1)-brane, extended in
the orthogonal flat dimension, will continually wind around the compact dimension like
a candy cane stripe. A periodicity emerges in the flat dimension from the distance the
brane takes to make a full lap around the circle dimension. We will take this length and
wrap it into a circle to form a donut-shaped 2-torus and observe that the magnetic flux on
the torus is quantized. Lastly, we will study the T-duality of a homologically equivalent
configuration which will give us a very stringy interpretation of the quantized flux.

1
Figure 1: This is a good illustration of closed strings in a circular dimension given by
Polchinski [3]. The integers are the winding numbers and the bottom picture depicts the
string m=0 splitting into m=+1 and m=-1 strings.

2 T-duality of closed strings

2.1 Introduction to Compactification
The bosonic string requires a spacetime critical dimension of D=26. The idea of circular (or
compactified) extra dimensions has been a property of unification theories since the work
of Kaluza-Klein in 1926 [1]. We will start by compacting one of the 25 spacial dimensions
into a circle with radius R, beginning our toroidal construction.

x25 = x25 + 2πR (1)

The periodicity of the dimension will spur out some interesting physics; first being that
closed strings can now live on the compact dimension as well as wrapped around it. The
amount of times the closed string can be wrapped around is quantified by the integer m
called ”the winding number.” The string lives on the dimension when m=0.

X(τ, σ + 2π) = X(τ, σ) + m(2πR) (2)

The winding number is a topological property that describes the mapping between the
string parameter σ ∈ [0, 2π) to the coordinates in x25 [2]. The winding number must be
conserved,. For example, a m=0 string can split into a m=+1 and m=-1 as shown in Fig
[1]. The negative winding states just mean the string is mapped along the negative x25
direction.

2
 The second interesting feature is the quantized momentum, which can be understood by
examining the translation operator inducing translations in a full circle e−ipa = e−ip(a+2πR) =
e−ipa e−ip2πR , thus, |e−ip2πR | = 1 =⇒ p = Rn

It will be useful to define a new term called the winding, ω = mR α0 , since the presence
of the compacted dimension introduces a new term to our string’s coordinates,
r
α0 X e−inτ
X(τ, σ) = x0 + α0 pτ + α0 ωσ + i (αn e−inσ + α˜n einσ ); (3)
2 n
n6=0

α˜n corresponds to right moving oscillations and αn correspondence to left moving oscilla-
tions, with n > 0 annihilating excitations and 0 > n creating excitations. The additional
√
term in (Eq.3) comes from that fact that the zero modes may not be equal, α̃0 −α0 = 2α0 ω.

2.2 spectrum
q
α0
For non-compact dimensions, ω is always zero and α0 = α˜0 = 2 P. We define our
momentum for the periodic dimension as follows.
r
2 n mR
PL = α0 = + 0 (4a)
α0 R α
r
2 n mR
PR = 0
α˜0 = − 0 (4b)
α R α
The Virasoro generators are
∞
α0 PL2 X
L⊥
0 = + α−n αn (5a)
4
n=1
∞
α0 PR2
L˜⊥
X
0 = + α̃−n α̃n . (5b)
4
n=1

Before compactification, the condition L⊥ ˜⊥ ⊥ ˜⊥ = 0, where N ⊥ is

0 − L0 = 0 forced N − N
just the number operator or the second term in (Eq. 5) in disguise. If the first condition
is to hold true, that would mean N ⊥ − N˜⊥ = 0 only if the string has zero momentum,
winding number, or both. Otherwise,

N ⊥ − N˜⊥ = nm. (6)

We calculate the mass2 of the string to be

2 2
M 2 = p2 + ω 2 + (N ⊥
− N˜⊥ − 2) = n + ( mR )2 + 2 (N − Ñ − 2). (7)
α0 R2 α0 α0

3
 0
There is a beautiful symmetry of this spectrum. We can take R → αR , and since n
and m are just integers, we can just send n → m and m → n which leaves the mass
spectrum invariant. In other words, physics remains indistinguishable between compacted
0
radii R and its dual, (R̃ = αR ), upon the exchange of momentum and winding. It is
pretty remarkable a dimension with an extremely large radius is equivalent to one with
an extremely small radius. There is a few other aspects we need to touch on inorder to
complete our full transition into the dual realm. It went without saying that our string
coordinate in (Eq. 3) was constructed by adding the left and right components of the
string; X(τ, σ) = XL (τ + σ) + XR (τ − σ). Our ”dual” string coordinate is instead defined
by:
X̃(τ, σ) = XL (τ + σ) − XR (τ − σ). (8)
So that the momentum and winding number are now switched
r
0 0 α0 X e−inτ
X̃(τ, σ) = q0 + α ωτ + α pσ + i (αn e−inσ − α˜n einσ ). (9)
2 n
n6=0

The dual momentum is defined as

1 ∂ X̃
P̃τ = = T (ẊL − ẊR ), (10)
2πα0 ∂τ
1
where T = 2πα 0 is the string’s tension. Since p and w are switched, the commutation

relations are adjusted to [q0 , ω] = i and [q0 , p] = 0. T-duality is a full quantum symmetry.

3 Interlude I: Wilson line

Before I carry on with my discussion of T-duality, it’s important to discuss the new prop-
erties that arise among Maxwell potentials on compact dimensions. This will allow us to
study open string T-duality since their end points live on branes which carry gauge fields.
This usual gauge transformation leaves Maxwell equations invariant,

Ax → Ax + ∂µ χ. (11)

But not the Schrodinger equation for a charge particle as described,

→
−
∂ψ 0 1 ∇ →
−
Hψ = i = ( − q A )2 ψ + qΦψ. (12)
∂t 2m i
We must introduced a U(1) phase factor to the wave function to recover invariance,

U (x) = eiqχ(x) ; ψ 0 = U ψ, (13)

which will be the gauge parameter of our theory instead of χ(x). This can be physically seen
from the Aharonov-Bohm effect, where the wave function of a particle becomes affected

4
to a phase by the gauge potential in the absence of EM field. At first, it was thought
of as very bizarre that quantum mechanics should be effected by a potential which was
considered unphysical. Compactified dimensions are also effected by the gauge potential;
before we see this lets first take a closed curve over a constant electric field in ordinary
space.
− →
→ − →
− → −
I Z
E · d l = ( ∇ × E ) · d→−a = 0. (14)
Γ S
However, lets say Γ runs around a compact dimension. It is closed, yet there is no surface
which Γ traces out a boundary, so therefore
− →
→ −
I
E · d l 6= 0. (15)
Γ
We will study electric fields over compact dimensions in Section 6, for now lets consider
the gauge potential Ax which runs along the compact dimension. By integrating over the
potential, I
ω=q dxAx , (16)

Using the transformation in (Eq.11), (Eq.16) becomes,

I
∂χ
ω 0 → q dx(Ax + ) = ω + q(χ(x0 + 2πR) − χ(x0 ), (17)
∂x
where x0 is just some arbitrary point on the circle. The last term in (Eq. 17) does not vanish
but equals 2πm since U(x) is our gauge parameter and must be periodic U (x+2πR) = U (x).
This concludes that ω must be periodic: ω 0 = ω + 2πm. More specifically taking the values
[0, 2π), we can look at it as an angle that traces along the unit circle. This angle is used
to calculate the holonomy of the gauge field from a Wilson line defined as
H
W = eiθ = eiω = eiq dxAx
. (18)
The Wilson line is a gauge invariant quantity and becomes gauge trivial with shifts of that
are 2π periodic. We can fix the gauge χ to be linear in x,
x mx
qχ = (2πm) = . (19)
2πR R
The gauge transformation in (Eq.11) now becomes,
m
qAx (x) → qAx (x) + . (20)
R
When considering a Wilson line with constant Ax , we see that
θ
qAx = , (21)
2πR
where θ now represents the value ω in (Eq. 16) which we will see takes on a new and very
important interpretation in the world of open strings.

5
4 Open strings
4.1 T-duality
We now have the tools to study T-duality of open strings. Lets first assume we have a D25-
brane covering the entire universe. Thus our open string coordinate has only Neumann
boundary conditions (∂σ X(τ, σ)|σ=π = ∂σ X(τ, σ)|σ=0 = 0), given by
√ √ X αn
X(τ, σ) = x0 + 2α0 α0 τ + i 2α0 cosnσe−inτ . (22)
n
n6=0

We are considering x25 to be compact and this brane wraps completely around the dimen-
sion. In order to examine the effects of T-duality, we must break up our string into left
and right movers:

i √ 0 X αn −in(τ +σ)
r
1 α0
XL = (x0 + q0 ) + α0 (τ + σ) + 2α e (23a)
2 2 2 n
n6=0

i √ 0 X αn −in(τ −σ)
r
1 α0
XR = (x0 − q0 ) + α0 (τ − σ) + 2α e (23b)
2 2 2 n
n6=0

We have defined the T-dual coordinate in (Eq. 8) to be X̃ = XL − XR . For open strings

this gives us,
√ √ X αn
X̃(τ, σ) = XL − XR = q0 + 2α0 α0 σ + 2α0 sin(nσ)e−inτ . (24)
n
n6=0

Notice that this is the just coordinate of the string with Dirichlet boundary conditions
(∂τ X(τ, σ)|σ=π = ∂τ X(τ, σ)|σ=0 = 0). First, lets clearly establish the mathematical impli-
cations,
0
∂τ X̃ = XR − XL0 = ∂σ X (25a)
0
∂τ X = XR + XL0 = ∂σ X̃, (25b)

we see that the boundary conditions switch during the transformation to the dual world.
This means that the D25 brane wrapped around dimension x25 with radius R in the first
world cannot exist in its dual world. The boundary becomes Dirichlet and our D25-brane
0
reduces down to a D24-brane occupying a single point on the circle with radius R̃ = αR .
α0
(D(p); R) → (D(p − 1); ). (26)
R
If our brane is wrapped around all k circular dimensions that exist, the duality would
logically follow: D(p)-brane → D(p-k)-brane, with the new dual radii for each correspond-
ing dimension. Likewise if a string lives on a brane only in flat spacial dimension that

6
Figure 2: This is a great illustration on T-duality of open strings obtained from [4].To the
right, see that θ is the position of the brane in the dual world, given by value the line
integral of the gauge field A in the original, radius= R, world depicted to the left. The
difference between θ2 − θ1 is the length which the string is stretched given each endpoint
is on 2 different wrapped branes in the original world.

is orthogonal to k compact dimensions, then in the dual world the string will live on a
D(p+k)-brane wrapped up in the k compact dimensions.
The open string, which is a free scalar field in directions normal to the brane, can wind
around the compact dimension if the boundary conditions are Dirichlet just like the closed
string. We can take the endpoints of (Eq. 23) to see,
√ α0
X̃(τ, π) − X̃(τ, 0) = 2α0 α0 (π − 0) = 2πα0 p = 2π n = 2π R̃n. (27)
R
If the end points possess Neumann boundary conditions, the string would still be able to
wrap around the dimension but since they are able to travel anywhere along this dimension,
the string would lose its nice periodic structure.
What is of great interest-and the motivation behind section 2- is when a Maxwell gauge
field carrying brane that is wrapped in the curled dimension. From the Hamiltonian in
(Eq. 12), we see the the coupling is now tacked to the momentum term, p → p − qA. Using
our result from (Eq. 21), the momentum along the periodic direction of our string becomes

n n θ
→ − . (28)
R R 2πR
We interpret θ as the position of the dual D(p-1)brane along the compact dimension.
When the both the string’s end points are attached to the same brane they possess the
same value for θ, but have opposite charge so the effect cancels out p − qAx25 + qAx25 = p.
In order to study this new feature we will consider 2 branes, which curl around each other
in the original world. We examine Wilson line for each brane which gives us a value θ1 for

7
the first brane and θ2 for the second brane. The mass squared formula for the open string
is given by:
1
M 2 = p2 + 0 (N ⊥ − 1). (29)
α
With the momentum transformation,
[1] [2] n θ1 θ2
p − qAx25 + qAx25 = − + . (30)
R 2πR 2πR
In the dual world, both branes can only occupy a single point along the circle. If we
consider θ1 to be the dual position of the first brane and θ2 to be the position of the second
brane, then the interpretation follows that θ2 − θ1 = ∆θ is the length which the string is
stretched between the two branes, as shown in fig.[2]. The mass2 in (Eq. 29) becomes
n θ2 − θ 1 2 1
M2 = ( + ) + 0 (N ⊥ − 1). (31)
R 2πR α
For n=0, we recover the expression for a string stretched between two branes in flat di-
mensions,
X̄2 − X̄1 2 1
M2 = ( 0
) + 0 (N ⊥ − 1), (32)
2πα α
providing strong support for this interpretation. It is noteworthy to add that the sep-
arating overlapping branes, plays the same role as symmetry breaking in quantum field
theory. N overlapping branes can be expressed as a U(N) gauge field. If N=2, we get
U(2)=SU(2)XU(1), giving us the 4 massless bosons responsible for the weak interaction.
As we saw in (Eq. 31/32), separating the D-branes gives mass to string. String theory
interprets brane separation as the Higgs mechanism.

5 Interlude II: Open strings on electromagnetic background

T-duality leads to bizarre implications when background electromagnetic field permeate
the Dp-brane, which will be discussed in the next section. This section will set the stage.
The end points of open strings couple to Maxwell potentials the same way charge particles
do:
dX n dX n
Z Z Z
0
S = dτ dσL(Ẋ, X ) + dτ An (X) |σ=π − dτ An (X) |σ=π , (33)
dτ dτ
”m” and ”n” are brane world indices so they span from 0,...,p. Here we have An =
1 m
2 Fmn X , where Fmn is the electromagnetic field tensor. In order to find the equations of
motion for the string, we must vary the action and set it equal to zero, using the notation
Pµσ = ∂L/∂X µ and Pµτ = ∂L/∂ X˙µ0 , ”µ” is a spacetime index spanning from 0,...25.
0

Making the following substitutions we arrive at

Z Z
τ µ σ µ 1
δS = dτ dσ(Pµ ∂τ δX +Pµ ∂σ δX )+ dτ Fmn (δX m ∂τ X n +X m ∂τ δX n )|σ=π +.... (34)
2

8
To save room, it is only necessary to examine one of the end points since the σ = 0 term
comes in symmetrically. We will now utilize the fact that the string satisfies the Klein
Gordon wave equation ∂τ Pµτ + ∂σ Pµσ = 0. This way we can simplify things by writing out
the total derivative.
Z Z
δS = dτ dσ∂σ (Pµ δX ) + dτ δX m Fmn δτ X n |σ=π + ...
σ µ
(35)

Another thing we can handle here is that δX a = 0 where ”a” spans the directions normal
to the brane; a=p+1,...,25. Cleaning things up a bit we see that
Z
δS = dτ δX m (Pm σ
+ Fmn δτ X n )|σ=π + ... (36)

σ +F
Now that we have everything nice and tidy under one integral. We evaluate Pm n
mn ∂τ X =
0 for σ = 0, π. In static and light cone gauge our momentum takes the form Pµσ =
−T ∂σ Xµ = −1/(2πα0 )∂σ Xµ , so our boundary conditions become
∂σ Xm − 2πα0 Fmn ∂τ X n = 0. (37)
There are 3 types of boundary conditions the string can have. The first type is only
Neumann,which occurs when Fmn = 0, since the only part left is ∂σ Xm = 0. The next
type of boundary conditions are mixed with Neumann and Dirichlet. This can occur in a
situation where a string is attached to 2 branes with different dimensions. It is best to see
how they come about in (Eq. 37) with an example. Suppose we only have a magnetic field
F34 = −F43 = B. Our mix boundary conditions looks like
∂σ X3 = 2παB∂τ X 4 , (38a)
0 3
∂σ X4 = −2πα B∂τ X . (38b)
The last type is strictly Dirichlet and that can only happen if we crank our magnetic field
up to a really large value (in fact infinity) such that the first terms in (Eq 38), are negligible,
leaving ∂τ X 3or4 = 0. In this case the field is so large that the end points in the x3 and x4
directions become stuck.We will view this specific situation in more detail. Having these
boundary conditions, we are now able to continue our exploration of T-duality.

6 T-duality of D-branes
6.1 D-branes with electric fields
I propose that a D(p-1)-brane moving around a compact dimension with a velocity v is
T-dual to a Dp-brane wrapped on the circular dimension with an electric field pointed
along that direction. In fact the velocity of the dual brane determines the strength of the
electric field and vise versa. This will lead to profound implications. The proof might be
alittle hard to follow so I will give a quick run-down.

9
 • collect electric field boundary conditions

• take the dual of x25

• boost the D(p-1) brane in the dual world around the circle

• Show that the boundary conditions are equal to the Dp-brane world with an electric
field

Well lets get started. The electric field around the compact, x25 , dimension is given by
F25,0 = −F0,25 = E which yields the boundary conditions,

∂σ X0 − 2πα0 F0,25 ∂τ X 25 = 0, (39a)

∂σ X25 − 2πα0 F25,0 ∂τ X 0 = 0. (39b)

When we replace X0 = −X 0 ,E = 2πα0 E, and X 25 = X, we find

∂σ X 0 − E∂τ X 25 = 0 (40a)
0
∂σ X − E∂τ X = 0. (40b)

Now that we have both coordinates on the same footing, it would be nice to represent the
boundary conditions in matrix form. ∂τ and ∂σ will not be friendly towards this pursuit
since there is no invertible linear relation between them. The partial derivatives we will be
using are ∂+ = 12 (∂τ + ∂σ ) and ∂− = 21 (∂τ − ∂σ ). With a little algebraic manipulation, we
can solve for ∂+ X 0 and ∂+ X.
 0  " 1+E 2 2E
#  0
X 1−E 2 1−E 2 X
∂+ = 2E 1+E 2 ∂− X (41)
X 2 2
1−E 1−E

This is a very attractive form. Similarly, we can write the Neumann boundary conditions,
∂+ X 0 = ∂− X 0 , ∂+ X = ∂− X, in the form of a matrix; namely the identity.
 0    0
X 1 0 X
∂+ = ∂− (42)
X 0 1 X

Taking the dual of X, the boundary condition becomes Dirichlet, which we can write in
terms of ∂±
∂+ X = ∂+ X̃, ∂− X = −∂− X̃ (43)
In matrix form  0    0
X 1 0 X
∂+ = ∂ . (44)
X̃ 0 −1 − X̃

10
We are in the position to boost our brane along the circle to see if our proposition is true.
We’ll refer to the rest frame of the circle as S and the rest frame of the brane as S’, the
relation between them is a Lorentz transformation with boosting parameter β = vc
00
X = γ(X 0 − β X̃) (45a)
0
X̃ = γ(−βX 0 + X̃) (45b)
1
where γ = (1 − β 2 ) 2 . Putting the Lorentz transformation in matrix form we get
 00
1 −β X 0
    0
X X
0 =γ =M (46)
X̃ −β 1 X̃ X̃

We can combine our T-duality transformation with our Lorentz transformation to get our
string boundary conditions in the S frame of reference
 0    0
X −1 1 0 X
∂+ =M M ∂− (47)
X̃ 0 −1 X̃

Using (Eq. 44), we can revert back to the original dual world with ∂− X̃ → −∂− X and
∂+ X̃ → ∂+ X in matrix form,
 0      0
X −1 1 0 1 0 X
∂+ =M M ∂− , (48)
X 0 −1 0 −1 X

which becomes
 0  " 1+β 2 2β
#
X0
 
X 2 1−β 2
∂+ = 1−β
2β 1+β 2 ∂− . (49)
X 2
X
1−β 1−β 2

Comparing this with (Eq. 41), we see that

E = 2πα0 E = β, (50)

thus confirming our proposition. There is something interesting to be observed. According

to the theory of special relativity, objects cannot have a speed greater than c, so β < 1.
This leads to the inequality,
|β| 1
|E| = 0
< = T, (51)
2πα 2πα0
which means the maximum value an electric field can have equal to the tension of the
string! We can think about this intuitively by considering the two forces that act on the
end points of the string: the electric force and string’s tension. Including relativistic effects,
the effective tension of the string becomes T0 (1 − v⊥ 2 ), so for E < T we can find a value
0
for v⊥ which balances out the forces. For E = T0 , the end points stop moving.

11
 Based on this claim, we find that Maxwell’s equations aren’t suitable to describe electric
fields on the world volume of D-branes since the electron carries an infinite self energy. We
look to a new theory of non-linear electrodynamics, Born-Infeld theory, which yields a finite
electrostatic energy. The lagrangian density describing such a theory is given by
r
1
L = −b −det(ηµν + Fµν ) + b2 ,
2
(52)
b
where ηµν is the Minkowski metric. This is a good √chance to weave in the fact that the
0
string coupling also changes under T-duality, g̃ = Rα g. From section 2, the beauty of
T-duality was that it left the mass of the closed string unchanged. What about the brane?
Since the mass is equal to its tension multiplied by the volume, and the volume of the
brane clearly changes, we must accommodate this by altering the tension.

Tp−1 (g̃) = 2πRTp (g) (53)

The lagrangian for a relativistic point particle is given by

r
v2
L = −m 1 − 2 . (54)
c
We replace the mass of a particle with the mass of the brane,
r
v2
L = −Vp−1 Tp−1 (g̃) 1 − 2 . (55)
c
Taking the dual, we utilize Eq.50 and Eq. 53 to get
p
L = −Vp−1 (2πR)Tp (g) 1 − (2πα0 E)2 . (56)

We (partially) see that the lagragian density is

q
L = −Tp (g) −det(ηµν + 2πα0 Fmn ). (57)

Comparing this with Eq. 52, we find that the normalization constant b = 1/2πα0 , and the
additive contribution is no longer relevant since the rest energy of our brane is needed.
Thus Eq. 57 governs the physics of D-branes, which up until now, we have not considered
independent- dynamical objects. In the next section we will confirm that Eq. 57 is the
correct lagragian by considering magnetic fields.

6.2 Magnetic fields on D-branes

We will repeat a similar analysis with a magnetic field this time instead of an electric field.
We will see that Dp-branes carrying magnetic fields correspond to rotations rather than

12
boosts of the D(p-1) branes in the dual world. We have already computed the boundary
conditions for a magnetic field, F34 = −F43 = B, in (Eq. 38). We will stick with these
boundary conditions and just as we did for electric fields, we can re-write them in terms
of B = 2πα0 B.

∂σ X 3 = B∂τ X 4 (58a)
∂σ X 4 = −B∂τ X 3 (58b)

In terms of ∂+ and ∂− , our boundary conditions take a similar form as (Eq. 41) up to
some sign differences:
 3  " 1−B2 2B
#  3
X 1+B 2 1+B 2 X
∂+ = −2B 1−B2 ∂− . (59)
X4 2 2
X4
1+B 1+B

The x4 direction is compacted and the magnetic field covers the cylinder that it forms with
3
the x direction. We take its dual so Dirichlet boundary conditions are imposed,
 3    3
X 1 0 X
∂+ ˜4 = ∂− ˜4 . (60)
X 0 −1 X

Here, S’ is the rotated frame of reference. The transformation from S → S 0 by angel θ is

as follows: " 0 # 
cos θ sin θ X 3
 3
X3
 
X
˜04 = ˜ = R ˜4 . (61)
X − sin θ cos θ X 4 X

Recovering the boundary conditions from (Eq. 54),

 3    3
X 1 0 X
∂+ ˜4 = R−1 R∂− ˜4 , (62)
X 0 −1 X

and detting back from the dual (∂− X 4 = −∂− X̃ 4 )

 3      3
X 1 0 1 0 X
∂+ = R−1 R ∂ . (63)
X4 0 −1 0 −1 − X˜4

We see the boundary conditions we started with in (Eq.60) are expressed in terms of
rotations is given by  3    3
X cos 2θ − sin 2θ X
∂+ = ∂− . (64)
X4 sin 2θ cos 2θ X4
We can solve for B by looking at the diagonal terms first,

1 − B2
= cos 2θ. (65)
1 + B2

13
Solving this give us the relationship B = ± tan θ. By examining the off-diagonal compo-
nents we see that the correct value is

B = 2πα0 = − tan θ. (66)

It is interesting to look at this problem from the perspective of electromagnetic potentials

that we studied in section 3. One possible configuration which we can extract our B
field is to consider A4 = Bx[3] and A3 = 0. Since x4 is periodic, we know from (Eq.20)
A4 ∼ A4 + Rn4 , so the potential in that direction remains invariant under quantized leaps.
However, the defined potential increases as you move along the cylinder in the x3 direction,
n
∆A4 = B∆x3 = . (67)
R4
In the dual world this becomes even more interesting; first we let n → −n and see that

nR̃4 2πnR̃4 2πnR̃4

∆x3 = − 0
=− 0
= . (68)
αB 2πα B tanθ
When n=1, ∆x3 represents the smallest area long the cylinder the brane wraps around
x4 once. Now we have this periodic structure along the x3 , which looks like candy cane
stripes in this configuration. We now have the opportunity to enrich our toroidal structure
by making x3 periodic as well,

2π R̃4
2πR3 = ∆x3 = . (69)
tanθ
We now have an expression for the angle,

R̃4
tan θ = n , (70)
R3
which is quantized. We can argue this by showing the magnetic flux ,Φ, along our 2-torus
is quantized. The magnetic flux in the original world is,

Φ = B∆A = B(2πR3 )(2πR4 ). (71)

Writing this expression in terms of R̃4 and using our valued obtained from B we see that,

tanθ 2πα0
Φ=− 2πR3 = −2πn. (72)
2πα0 R̃4

So far we have kept the magnetic field constant. If we consider a magnetic field that is
uniform but not constant, in the dual picture the D(p-1)-brane will trace out a path that
is curved instead of a straight line.

14
Figure 3: The figure to the left shows the D(p-1) brane wrapping around x4 while traveling
along x3 . As a → a0 , b → b0 , and c → c0 we get the configuration to the right, illustrating
a possible way they can deform into each other provided by Ref.[4]

A tilted D(p-1)-brane that winds around x4 n times before it takes one lap around x3
is homologically equivalent to one D(p-1)-brane that only wraps in the x3 direction and
n D(p-1)-branes that only wrap in the x4 direction as shown in Fig.[3]. They are not
physically equivalent but only through deformations. We can study its dual, noting that
the x3 does possess T-duality; we find the transformation sends the D(p-1)-brane wrapped
in x3 to a Dp-brane covering the cylinder, and the n D(p-1)-branes circling x4 become n
D(p-2)-branes that live on the world volume of the Dp-brane.
There is a nice physical picture to help us understand. We have just realized, for
example, that a D2-brane with constant magnetic field of flux 2πn is deformation equivalent
to a D2-brane with n D0-branes on its world volume. Of course to get back from the latter
to the former we just have the D0-brane dissolve into the D2-brane. Physically speaking
the D2-brane represents a vanishing magnetic field everywhere except at the positions of
D0-branes, where the magnetic field becomes infinite with a finite flux.
Using the newly constructed 2-torus, we find the brane volume of our D(p-1)-brane
streched along the diagonal of the torus is just the volume Vp−2 multiplied by the diagonal
length. Our lagragian is just the negative rest energy of the brane.
q
L = Vp−2 Ldiag Tp−1 (g̃) = −Vp−2 (2πR3 )2 + (2π R̃4 )2 Tp−1 (g̃) (73)

˜
Using (Eq.70), we find that 2πα0 B = − R
R3 . Taking the T-dual of Eq. 73 we obtain,
4

p
L = −Vp−2 (2πR3 ) 1 + (2πα0 B)2 (2πR4 )Tp (g). (74)

Since (2πR3 )(2πR4 ) is just the volume of the torus wrapped by the Dp-brane, we divide
by the total volume to get out lagrangian density
p
L = −Tp (g) 1 + (2πα0 B)2 , (75)

confirming that Eq.72 is the correct Lagrangian density.

15
7 Conclusion and acknowledgements
We have discussed alot of astonishing consequences of string theory. As a recap, in studying
closed strings we saw a relationship emerge between compact dimensions with radius R
0
and dimensions with inverse quantity αR . The boundary conditions for an open string
interchange upon swapping its radius with its inverse. This means an open string free
to move on a compact dimension becomes fixed at one point in the dual world. This
point is determined by the gauge potential carried by the brane thats wrapped in the
compact dimension. By considering a constant electric field along the brane in the compact
dimension, the potential must vary with time, thus the position of the brane in the dual
world must vary with time, giving rise to a velocity. The velocity is determined by the
strength of the electric field and vise versa, thus a maximum velocity implies a maximum
electric field. This fact led to the insight that branes can be studied on their own as a
dynamical and gravitating objects, governed by the Born-Infeld theory of electrodyamnics.
A magnetic field along a compact dimension tilts the brane in the dual world by an angle
determined by the magnetic field strength. The length which the brane travels before
making a full loop around the circular dimension is periodic and by making it compact we
find that the magnetic flux along the area is quantized.
Aside from the impressive role T-duality in bosonic string theory, its theoretical strength
becomes more profound due to its ability to relate different types of superstring theories.
For example, Type II A superstring theory and Type II B superstring theory become
exchanged under T-duality. This relationship lead to a more fundamental theory of nature
called M-theory, which holds all 5 superstring theories by this relation [2]. There are many
√
things I left out of the discussion, one of them is the self-dual radius when R = α0 ,
which is the minimum length scale for string theory. This adds geometric structure to
the background at that distance. Ref. [5] describes that non-communitive closed string
theories are T-dual is a communitive closed string theory. .
I would like to give special thanks to my mentor, Dr. K.V. Shajesh at Southern Illinois
University-Carbondale, for encouraging and guiding my independent study of string theory
during the 2017 fall semester of undergraduate studies. I would like to thank Dr. Savdeep
Sethi, for giving me the privilege to sit in on his graduate level string theory course during
my time in ”limbo,” and his graduate student, Christen Ferko, who was willing to take
time out of his studies to answer any physics question I had with great enthusiasm.

8 References
[1] M. Grana and H. Triendl, ”String theory compactifications,”
http://www.ipht.fr/Docspht/articles/t13/042/public/Notes.pdf
[2] D. Tong, ”Lectures on String Theory,” http://www.damtp.cam.ac.uk/user/tong/string.html
[3] J. Polchinski, String Theory. Vol. 1: An Introduction to the Bosonic String. Cambridge
University Press, 1998.

16
[4] B. Zwiebach, A First Course in String Theory, Cambridge University Press, 2004.
[5] D. Lust, ”T-Duality and closed string non-commutative (doubled) geometry,” arXiv:1010.1361[hep-
th].

17