4. SUSY algebra
The SUSY algebra is
Where ππΌπ½
are called central charges and they are antisymmetric in their indices. Moreover, πππ
πΌ
π½
= ππΌπΎ
π
ππ πΎπ½
β (π β π) where ππ
=
(1, π) and ππ
= (1, βπ).
14. Electromagnetic duality
Sourcless Maxwell equations (MEs) are invariant in the transformations π·: πΈπ β π΅π , π΅π β βπΈπ . It can be generalized to
πΈπ = πΈπ cos π + π΅π sin π , π΅π = βπΈπ sin π + π΅π cos π
and MEs will still be invariant as can be seen below for the Faraday and Maxwell laws (written in term of indices and ππ‘ is time derivative);
πππππππ΅π = ππ‘πΈπ
πππππππΈπ = βππ‘π΅π
β
ππππ ππ cos π πΈπ + sin π π΅π = βππ‘ cos π π΅π β sin π πΈπ β πππππππΈπ
β²
= βππ‘π΅π
β²
ππππ ππ cos π π΅π β sin π πΈπ = ππ‘ cos π πΈπ + sin π π΅π β πππππππ΅π
β²
= ππ‘πΈπ
β²
Introduce πΉ
ππ with πΉ0π = πΈπ, π΅π =
1
2
πππππΉ
ππ. So we can visualize π· transformation as
πΈπ β π΅π β πΉ0π β
1
2
πππππΉ
ππ =β πΉ0π, π΅π β βπΈπ β
1
2
πππππΉ
ππ β βπΉ0π =
1
2
ππππ β πΉ
ππ, β πΉ
ππ =
1
2
ππππΌπ½πΉπΌπ½
So, we can see that the π· transformation as πΉ
ππ =β πΉ
ππ.
If there are no monopoles, we can take
π΅π =
1
2
ππππ πΉ
ππ = β Γ π΄ π =
1
2
ππππ πππ΄π β πππ΄π , πΈπ = πΉ0π = ββπ β
ππ¨
ππ‘ π
= π0π΄π β πππ΄0 β πΉ
ππ = πππ΄π β πππ΄π
16. Now, we choose π¨ in two patches and then them π¨π and π¨π (i.e. for Northern patch and the Southern patch). We choose π = 1 for π¨π and
π = β1 for π¨π and we have;
π¨π =
π
4ππ
1 β cos π
sin π
π, π¨π = β
π
4ππ
1 + cos π
sin π
π
The overlap region happens to be 2π = π circle and on that circle, the difference of the two potentials is
π¨π π =
π
2
β π¨π π =
π
2
=
π
2ππ
π = ββ β
π
2π
π = ββπ
So, the difference is a gauge transformation. The π(1) element ππππ
should be continuous and thus, we have the following condition;
πππ π 2π βπ 0
= 1 β π π 2π β π 0 = 2ππ, π β β€
Using the expression for π, we get
ππ = 2ππ, π β β€
Comments
ο A single monopole will ensure the quantization of electrical charge.
ο ππ = 0 and ππ = 2π represent the same π(1) element and thus, the π(1) group is compact. So, we have the following two equivalent
statements (they are the contrapositive of each other)
πππππππππ β πΆππππππ‘ π 1 ππππ’π
ππ πΆππππππ‘ π 1 ππππ’π β ππ ππππππππ π πππ’π‘ππππ
Compact π(1) group will be realised when a theory with compact gauge group is spontaneously broken to a group containing π(1) as a
product group.
17. The tβHooft, Polyakov monopole
In the Diracβs treatment of the monopole, the interior of the monopole is not studied. In order to study the interior of a monopole, we
study a monopole configuration called tβHooft, Polyakov monopole. It is nothing but a non abelian Yang Mills Higgs theory in 3 + 1
dimensions. The lagrangian is as follows;
β = β
1
4
πΉ
ππ
π
πΉ
π
ππ
+
1
2
π·πππ
π·π
ππ
β π π , π = π£ β 0
The EOM for πΉ
ππ
π
is (also called Gauss constraint) is
π·ππΉ
π
ππ
= ππππππππ·π
ππ
We define πΉππ
as
πΉ
ππ = ππ
πΉ
ππ
π
β
1
π
ππππ
ππ
π·πππ
π·πππ
where ππ
is a unit vector in the direction of ππ
and for ππ
to exist, ππ
β 0 everywhere. Note that πΉ
ππ is gauge invariant.
It can be shown that
πΉ
ππ = ππ(ππ
π΄π
π
) β ππ (ππ
π΄π
π
) β
1
π
ππππ
ππ
ππππ
ππ ππ
The proof is as follows. The definition of πΉ
ππ can be expanded and written as follows;
πΉ
ππ = ππ
πππ΄π
π
β πππ΄π
π
+ πππππ
π΄π
π
π΄π
π
β
1
π
ππππ
ππ
ππππ
+ πππππ
π΄π
π
ππ
ππππ
+ πππππ
π΄π
π
ππ
= ππ ππ
π΄π
π
β ππ ππ
π΄π
π
β
1
π
ππππ
ππ
ππππ
ππππ
+π΄π
π
ππππ
πΏππ
β ππππ
ππππ
ππππ
ππ
ππ
β π΄π
π
πΏππ
ππππ
+ ππππ
ππππ
ππππ
ππ
ππ
+ πππππ
ππ
π΄π
π
π΄π
π
β πππππ
ππ
ππππ
ππππ
π΄π
π
π΄π
π
ππ
ππ
18. The terms that we actually want are the red terms. The extra terms can be shown to be zero. For example, the first bracketed expression is;
ππππ
πΏππ
β ππππ
ππππ
ππππ
ππ
ππ
= ππππ
+ ππππ
πΏππ
πΏππ
β πΏππ
πΏππ
ππ
ππ
= ππππ
+ ππ
ππ
ππππ
β ππππ
= 0
Where I used the fact that
ππ
ππππ
=
1
2
ππ ππ
ππ
=
1
2
ππ 1 = 0
The second bracketed term vanishes by a similar proof. Now we need to prove that
ππππ
ππ
π΄π
π
π΄π
π
β ππππ
ππ
ππππ
ππππ
π΄π
π
π΄π
π
ππ
ππ
vanishes.
For this task, we can manipulate the last term as;
ππππ
ππ
ππππ
ππππ
π΄π
π
π΄π
π
ππ
ππ
= πΏππ
πΏππ
β πΏππ
πΏππ
ππ
ππππ
π΄π
π
π΄π
π
ππ
ππ
= ππππ
ππ
π΄π
π
π΄π
π
where I dropped the vanishing term ππππ
ππ
ππ
ππ
π΄π
π
π΄π
π
and used the fact that ππ
ππ
= 1.
So, we can now see that
ππππ
ππ
π΄π
π
π΄π
π
β ππππ
ππ
ππππ
ππππ
π΄π
π
π΄π
π
ππ
ππ
= ππππ
ππ
π΄π
π
π΄π
π
β ππππ
ππ
π΄π
π
π΄π
π
= 0
So, we have shown that
πΉ
ππ = ππ(ππ
π΄π
π
) β ππ (ππ
π΄π
π
) β
1
π
ππππ
ππ
ππππ
ππ ππ
21. Moduli space of BPS solution
Moduli space is the space of fixed energy and fixed topological charge solutions. The coordinates on the moduli space are called moduli or
collective coordinates.
If ππ
(π₯) is a BPS solution, so is ππ
(π₯ + π) with π fixed due to translational invariance. So, β3
is present as a product in the moduli space of
BPS monopole. Now, equation of motion π·ππΉ
π
ππ
= ππππππππ·π
ππ is expanded as;
π·π ππ
π΄π
π
β ππ
π΄π
π
+ ππππππ΄π
π
π΄π
π
= πππππππ[ππ
ππ + ππππππ΄π
π
ππ]
Set π = 0 and working in the π΄π
0
= 0 gauge, we get;
π·ππ΄π
π
+ π π, π π
= 0
The linearized form of the gauss law for deformations to the BPS solutions (πΏππ
, πΏπ΄π
π
) is obtained by letting π΄π
π
β π΄π
π
+ πΏπ΄π
π
, ππ
β ππ
+
πΏππ
and realising that only the deformations are allowed to be time dependant. We get
π·ππΏ π΄π
π
+ π π, πΏπ π
= 0
We can expand the Bogomolβnyi equation as well and then linearize it as follows;
ππππ ππ
πΏπ΄π
π
+ ππππππ΄π
π
πΏπ΄π
π
= πππΏππ
+ πππππ
π΄π
π
πΏππ
+ π πΏπ΄π, π π β πππππ·π
πΏπ΄π
π
= π·ππΏππ + π πΏπ΄π, π π
The solution to the two red equations above is πΏπ΄0
π
= 0, πΏππ
= 0, πΏπ΄π
π
= π·π π π‘ ππ
, π π‘ is arbitrary function of time.
If π = 0 then the change in π΄π
π
is a large gauge transformation (it doesnβt vanish at infinity). The corresponding π(1) element is πππ
and since
unbroken π(1) is compact, the coordinate π is on a compact π1
. So, the moduli space of the BPS monopole is;
β3
Γ π1
25. Olive Montonen and ππΏ(2, β€) duality
Set π = 0 first. Now, there can be electrically charged states (ππ, 0) and magnetic monopoles 0, ππ . The mass of an electrically charged state
is the mass of π boson i.e. ππ = π£π. The mass of a magnetic monopole is ππ = π£π β« π£π (at weak coupling). In a dual theory where the
roles of electrical and magnetic charges are replaced, we should have
π β π, ππ β ππ
Now, the dual theory should be at strong coupling as small π means large π. Moreover, this proposal of a dual theory is based on the classical
spectrum as was first given by Olive and Montonen in 1977. However, there are some problems pointed by them.
ο Potential π(π) might not remain vanishing when quantum corrections are included.
ο The states donβt match exactly as the monopoles donβt have spin 1 like the π bosons.
ο The duality is hard to test because the dual theory is at strong coupling.
We will see that the first two problems are solved by incorporating the theory in π = 4 super Yang Mills theory.
Now, let π β 0. Now, the πππ» lagrangian is;
β = β
1
4
πΉ2
β
ππ2
32π2
πΉ β πΉ β
1
2
π·πππ 2
Now, we let
ππ
β
ππ
π
, π΄π
π
β
π΄π
π
π
β πΉ
ππ
π
β
πΉ
ππ
π
π
After this rescaling of field, the lagrangian becomes (with covariant derivatives changed accordingly);
β = β
1
4π2
πΉ2
β
π
32π2
πΉ β πΉ β
1
2
π·πππ 2
26. Now consider this equivalent way of writing the first two terms of the lagrangian
=
1
32π2
πΌπ
π
2π
+
4ππ
π2
πΉ + π β πΉ 2
= β
1
8π2
πΉ2
ββ πΉ2
β
π
32π2
πΉ β πΉ =
1
4π2
πΉ2
β
π
32π2
πΉ β πΉ
where in the last step, I used the fact that πΉ2
=ββ πΉ2
as can be demonstrated
β πΉ2
=
1
4
ππππΌπ½
πΉπΌπ½ππππππΉππ
= β
1
2
πΏπ
πΌ
πΏπ
π½
β πΏπ
πΌ
πΏπ
π½
πΉπΌπ½πΉππ
= βπΉ2
So, the full lagrangian can be written in case of non zero π as;
β = β
1
32π2
πΌπ π πΉ + π β πΉ 2
β
1
2
π·πππ 2
where π =
π
2π
+
4π
π2
π
I will use a result from instanton theory. The π instantons in this theory are weighed by the factor π2ππππ
. So, physics is invariant in the
transformation π β π + 1 (corresponding to π β π + 2π). Moreover, if we set π = 0, then the electromagnetic duality (π β π) corresponds to
π β π =
4π
π
β π =
4ππ
π2
β β
π2
4ππ
= β
1
π
We can identify both of these transformations as special cases of a single general transformation
π β
ππ + π
ππ + π
, π, π, π, π β β€, ππ β ππ = 1
This transformation is known as the ππΏ(2, β€) transformation and we can identify
π β π + 1 β π = π = π = 1, π = 0
π β β
1
π
β π = π = 0, π = βπ = 1
We can see that
πΌπ π =
4π
π2
> 0
So, we are concerned with the upper half π plane only. We can also define a fundamental region such that any π in the upper half plane can be
brought into this fundamental region by an appropriate ππΏ(2, β€) transformation.
β
1
2
β€ π π π β€
1
2
, π β₯ 1
27. πΊπ³ π, β€ action on the states
The action of the ππΏ(2, β€) group on the states (ππ, ππ) states requires the electrical charge operator ππ and magnetic charge operator
ππ = πππ β
ππ
2π
ππ, ππ = πππ =
4π
π
ππ
Now, π β π + 1 β π β π + 2π and thus, we have;
ππ β π ππ β ππ β
ππ
2π
ππ β
ππ β ππ β ππ = πππ β πππ
ππ β ππ = πππ + πππ
Similarly for the π β βπβ1
transformation (for π = 0), we have
ππ = πππ β πππ β ππ β ππ = πππ β πππ
ππ = πππ β πππ β ππ β ππ = πππ + πππ
So, we have the following ππΏ(2, β€) action on states (ππ, ππ)
ππ
ππ
β
π βπ
π π
ππ
ππ
Bogomolβnyi bound revisited
The generalised Bogomolβnyi bound (proved at the end)
π2
β₯ π£2
(ππ
2
+ ππ
2
)
can entirely be written in term of π by using the definition of π and the expressions for ππ and ππ.
28. Coupling to fermions
We can add fermions into this theory with proper interactions with ππ
and π΄π
π
fields. Assume that fermions are in the β²πβ² representation of the
ππ(2) gauge group. The fermion lagrangian is;
βπ = ππππΎπ
π·ππ
π
β ππππππ
π
ππ
ππ
Where πππ
π
are the anti Hermitian generators in the β²πβ² representation. The gamma matrices are chosen as;
πΎ0
= βπ
0 π2
βπ2 0
, πΎπ
= βπ ππ
0
0 βππ
These matrices satisfy the Clifford algebra πΎπ
, πΎπ
= 2πππ
π4 as can be verified easily. The Dirac equation can be worked out from the above
lagrangian to get (using πππ = ππ
πππ
π
)
ππΎπ
π·ππ
π
β ππππππ = 0
We seek solutions of the form ππ π‘, π₯ = πβππΈπ‘
ππ(π₯) and we also set ππ π₯ = ππ
+
π₯ ππ
β
π₯ π
. Using the πΎ matrices and the Diracβs
equation, we have;
βππΈ
ππ
β
βππ
+ β
ππ
π·πππ
+
βππ
π·πππ
β
β ππππ
ππ
+
ππ
β = 0
This gives us two equations;
ππΏππππ
π·π + πππ ππ
β
= π·ππππ
β
= πΈππ
+
, ππΏππππ
π·π + πππ
β
ππ
+
= π·ππ
β
ππ
+
= πΈππ
β
,
If the above two equations have a solution with πΈ = 0 then there can be fermionic collective coordinates in the BPS moduli space. In other
words, we are looking for the kernels of π·ππ and π·ππ
β
. It is easy to see that;
ker π·ππ
β
β ker π·πππ·ππ
β
= {0}
In the last equality, I used the fact that π·πππ·ππ
β
is a positive definite operator.
31. We can see that in this multiplet, we have states which have spin 1. Moreover, if we compare the spin content of this
multiplet with the π = 4 gauge supermultiplet, there is an exact match.
There is another feature of the π = 4 theory which is worth mentioning. It is a known fact (I won't go into the details) that
the π½ function of π = 4 super Yang Mills theory vanishes and thus, the potential of the π = 4 theory won't receive quantum
corrections. Therefore, the first two problems in the Olive Montonen proposal that I mentioned are solved by incorporating 4
dimensional YMH theory in the π = 4 supersymmetric gauge theory.
39. Dyon interpretation
For dyon ππ, ππ becoming massless, π = πππ + ππ·ππ should be invariant under monodromy action. In other words
π = ππ ππ
ππ·
π
β ππ ππ π
ππ·
π
= π β ππ ππ π = ππ ππ
For ππ’ Μ, (1 0) is a left eigenvector and for πβπ’ , (1 β1) is left eigenvector (actually (πππ πππ) with integer π is a left eigenvector but
they arenβt stable as ππ and ππ have to be relatively prime).
For ππ ππ to be a left eigenvector, the monodromy matrix is
1 + 2ππππ 2ππ
2
β2ππ
2
1 β 2ππππ
This shows that (for π β β€)
πβ
βπ
ππ’πβ
π
= 1 β 4π 8π2
β2 1 + 4π
corresponds to ππ, ππ = (β2π, 1)
Moreover,
πβ
βπ
πβπ’πβ
π
= β1 β 4π 8π2
+ 2π + 8
β2 3 + 4π
corresponds to ππ, ππ = (β2π β 1,1).
40. Solution for π π’ , ππ·(π’)
Consider
π2
π
ππ§2
β π π§ π π§ = 0 (1)
Regular singular point π§ = π§0 is such that π§ β π§0
2
π(π§) is analytic at π§ = π§0.
Theorem: Non trivial constant monodromies correspond to singular points of π(π§) which are regular singular points.
Following Seiberg and Witten, π’ = 1 and thus, the singular points are π’ = Β±1, π’ β β.
The terms
1
π§ β 1
,
1
π§ + 1
in π(π§) correspond to third order poles at infinity. So, the general π(π§) is (following the convention)
π π§ = β
1
4
1 β π1
2
π§ + 1 2
+
1 β π2
2
π§ β 1 2
β
1 β π1
2
β π2
2
+ π3
2
π§ + 1 π§ β 1
, ππ > 0
The transformation
π π§ = π§ + 1
1βπ1
2 π§ β 1
1βπ2
2 πΊ
π§ + 1
2
equation (1) becomes;
π 1 β π
π2
πΊ π
ππ2
+ π β π + π + 1 π
ππΊ π
ππ
β πππΊ π = 0
where π is an independent variable and 2π = 1 β π1 β π2 + π3, 2π = 1 β π1 β π2 β π3, π = 1 β π1.
41. The solutions to this equation are hypergeometric functions
πΉ π, π, π; π =
π=0
β
π π π π
π π
ππ
π!
Where π π are Pochhammerβs symbol π 0 = 1, π π+1 = π π + 1 β¦ (π + π)
The two solutions that we pick are
πΊ1 π = βπ βπ
πΉ π, π + 1 β π, π + 1 β π;
1
π
πΊ2 π = 1 β π πβπβπ
πΉ π β π, π β π, π + 1 β π β π; 1 β π
Now, πΊ1 has trivial monodromy at infinity and πΊ2 has trivial monodromy at π = 1.
Finding π, π, π
For large π§
π π§ = β
1
4
1 β π3
2
π§2
β 4π§2
π2
π π§
ππ§2
+ 1 β π3
2
π π§ = 0
β π π§ =
π§1Β±π3
2
π3 β 0 , π π§ = π§ πΌπ π§ (π3 = 0)
The large π’ expression for ππ· resembles the π3 = 0 case. So, set π3 = 0. Around π§ = 1
π π§ = β
1
4
1 β π2
2
π§ β 1 2
β
π2
π π¦
ππ¦2
+
1 β π2
2
4π¦2
π π¦ = 0, π¦ = π§ β 1
ππ· is linear in π’ β 1 around π’ = 1 and thus, the second derivative should vanish and this, π2 = 1. β€2 symmetry implies π1 = 1.
β π = β
1
2
, π = β
1
2
, π = 0
42. β π π§ = πΊ
π§ + 1
2
β πΊ1 π’ = π
π’ + 1
2
1
2
πΉ
1
2
, β
1
2
, 1;
2
π’ + 1
, πΊ2 π’ = β
π’ β 1
2
πΉ
1
2
,
1
2
, 2;
1 β π’
2
The solutions for π and ππ· which give correct asymptotics are
ππ· π’ = βππΊ2 π’ = π
π’ β 1
2
πΉ
1
2
,
1
2
, 2;
1 β π’
2
π π’ = β2ππΊ1 π’ = 2 π’ + 1
1
2πΉ
1
2
, β
1
2
, 1;
2
π’ + 1
We can find ππ·(π) now and then, β±(π) is determined by integrating ππ· π = β±β²(π) with respect to π which gives you low energy lagrangian.
43. Further developements
Extensions
ο N. Seiberg and E. Witten added hypermultiplets in 1994.
ο A first step was taken to generalize to ππ(π) group in 1995 by A. Klemm et al.
ο Low energy action was calculated by Klemm et al. for ππ(3) in 1995.
ο A. Brandhuber et al. calculated the low energy action for ππ(2π) in 1995.
ο U. H. Danielsson et al. wrote the weak coupling limit for arbitrary gauge groups and wrote the solution for ππ(2π + 1) groups in 1995.
ο P.C. Argyres et al. worked out the description of metric on moduli space for ππ(π) theory in 1995.
ο U.H. Danielsson and B. Sundborg address the same problem with some simple groups with matter multiplets in 1996.
Tests
ο E. Witten and C. Vafa proposed a test at strong coupling in 1994.
ο A. Lossev, N. Nekrasov and S.L. Shatashvilli proposed a test of Seiberg Witten solution by instanton calculations in 1999.
ο The test proposed in 1999 was performed by N. Nekrasov (and the low instanton results matched the literature values) in 2003.
44. Proof of generalized Bogomolβnyi bound (ππ, ππ)
In the presence of electrical fields (keeping π΄0
π
= 0 still), we have the following mass for the dyon;
π =
1
2
β« π3
π₯ πΈπ
π
πΈπ
π
+ π΅π
π
π΅π
π
+ π·πππ 2
=
1
2
β« π3
π₯ πΈπ
π
β cos πΌ π·πππ
+ π΅π
π
β sin πΌ π·π ππ 2
+ β« π3
π₯ cos πΌ πΈπ
π
π·πππ
+ sin πΌ π΅π
π
π·πππ
β₯ cos πΌ ππ + sin πΌ ππ
Now, we can make the bound as tight as possible by differentiating the above expression w.r.t πΌ and setting it to zero. This gives;
tan πΌ =
ππ
ππ
β sin πΌ =
ππ
ππ
2
+ ππ
2
, cos πΌ =
ππ
ππ
2
+ ππ
2
This gives us;
π β₯ π£ ππ
2
+ ππ
2
1
2
This is the generalised Bogomolβnyi bound in terms of (ππ, ππ).
45. Acknowledgements
I would like to thank
ο My family for their support in these difficult times.
ο My supervisor and other teachers for guiding me about various things regarding the thesis.
ο My friends for their support.
ο People with whom I had useful discussions, specially Dr. K.S. Narain.
46. Proof of generalized Bogomolβnyi bound (e,g)
In the presence of electrical fields (keeping π΄π
π
= 0 still), we have the following mass for the dyon;
π =
1
2
β« π3
π₯ πΈπ
π
πΈπ
π
+ π΅π
π
π΅π
π
+ π·πππ 2
=
1
2
β« π3
π₯ πΈπ
π
β cos πΌ π·πππ
+ π΅π
π
β sin πΌ π·π ππ 2
+ β« π3
π₯ cos πΌ πΈπ
π
π·πππ
+ sin πΌ π΅π
π
π·πππ
We proved previously that π΅π
π·πππ
= ππ(π΅π
π
ππ). We can also prove that πΈπ
π
π·πππ
= ππ πΈπ
π
ππ
as follows.
We see that πΈπ
π
π·πππ
= π·π πΈπ
π
ππ
β ππ
π·ππΈπ
π
= ππ πΈπ
π
ππ
β ππ
π·ππΈπ
π
as πΈπ
π
ππ
is gauge singlet. Now, we prove that π·ππΈπ
π
= 0.
Use the zeroth component of Gaussβ constraint to get;
π·ππΉπ0
π
= πππππ
ππ
π·0ππ
= 0
The last equality follows because ππ
is time independent and π΄0
π
= 0. So, π·ππΉπ0
π
= βπ·ππΉ0π
π
= βπ·ππΈπ
π
= 0
So, we get;
π β₯ β« π3
π₯ cos πΌ ππ πΈπ
π
ππ
+ sin πΌ ππ π΅π
π
ππ
= π£[cos πΌ π + sin πΌ π]
Now, we can make the bound as tight as possible by differentiating the above expression w.r.t πΌ and setting it to zero. This gives;
tan πΌ =
π
π
β sin πΌ =
π
π2 + π2
, cos πΌ =
π
π2 + π2
This gives us;
π β₯ π£ π2
+ π2
1
2