Professional Documents
Culture Documents
Noppadol Mekareeya
INFN, sezione di Milano-Bicocca,
Piazza della Scienza 3, I-20126 Milano, Italy
E-mail: n.mekareeya@gmail.com
• The a-maximisation
Contents
1 References 1
2 Chiral multiplets 2
5 Vector multiplets 11
5.1 Some conventions 11
5.2 Gauge symmetry 12
5.3 Lagrangian for the vector multiplets 14
5.4 One-loop beta function coefficients 15
5.5 One-loop exactness of the holomorphic gauge coupling 16
8 Supersymmetric QCD 29
8.1 The classical moduli space 30
8.2 The quantum theory 32
8.2.1 Anomalous and non-anomalous global U (1) symmetries 32
8.2.2 The dynamical scale 33
8.3 The Novikov–Shifman–Vainshtein–Zakharov (NSVZ) beta function 34
8.4 Anomalous dimensions and the Banks-Zaks fixed point 38
8.5 Some important facts about the (super)conformal symmetry 39
8.6 Superconformal fixed points in SQCD 41
8.7 Seiberg duality 42
8.7.1 Some checks of the dualities 43
8.8 Quantum behaviours of SQCD 46
8.8.1 The case of Nf = Nc + 1 46
–i–
8.8.2 The case of Nf = Nc 48
8.8.3 The case of 0 < Nf < Nc 50
9 Supersymmetric index on S 3 × S 1 51
9.1 Supersymmetry on S 3 × R and S 3 × S 1 51
9.2 Indices of chiral multiplets 54
9.3 Indices of gauge theories 56
9.4 The index of SQCD 58
9.5 Some important points 59
9.6 Example: SU (2) SQCD with 3 flavours 60
9.7 A brief introduction to conformal manifolds 61
9.7.1 Example: The N = 4 SU (N ) super–Yang–Mills theory 61
9.7.2 Example: SU (3) SQCD with 6 flavours 63
10 The a-maximisation 64
10.1 The conformal anomalies 66
10.2 Statement of the problem 68
10.3 A solution by Intriligator and Wecht [1] 68
10.4 A proof of a-maximisation 69
10.5 Example: A model with the enhanced E6 flavour symmetry 71
10.6 A caution on accidental symmetries 73
10.6.1 Example: An N = 1 Lagrangian of the N = 2 Argyres–
Douglas theory of type H1 or (A1 , A3 ) 74
10.6.2 Example: The A b theory 77
A Spinors in 3 + 1 dimensions 80
1 References
The contents in this lecture note are taken, copied and/or adapted from various
references, some of them are listed below and the others are mentioned in the main
text. I claim no credits for any material in this lecture note. It is produced for the
pedagogical reason only.
There are a number of excellent lecture notes and books on the subject including
[2–9]. The note by Yuji Tachikawa [9], in particular, contains most of the topics in
this lecture note as well as more advanced material.
I do not include the material on the superspace and superfields in this note. I
refer the readers to [10].
–1–
2 Chiral multiplets
A 4d N = 1 chiral multiplet consists of a complex scalar φ, and a Weyl fermion ψα
in the same representation of the gauge group.
It is represented by a chiral superfield Φ satisfying1
Dα̇ Φ = 0 (2.4)
where
–2–
where
∂ 2 K(φ, φ† ) ∂W (φ) ∂ 2 W (φ)
g= , ∂W = , ∂ 2W = . (2.8)
∂φ† ∂φ ∂φ ∂φ ∂φ
In four spacetime dimensions, φ and ψ have mass dimensions 1 and 3/2 respec-
tively, and so the mass dimension of the superpotential W is 3.
R-symmetry
• Supersymmetry allows for a global symmetry, known as the U (1) R-symmetry.
It acts on θ and W (Φ) as
θ → eiα θ , d2 θ → e−2iα d2 θ ,
(2.9)
W → e2iα W ,
so that the R-charge assignments are R(φ) = r, R(ψ) = r−1 and R(F ) = r−2.
The R-charge of the chiral superfield Φ is the same as that of the lowest
component, i.e. R(Φ) = r.
Vacuum degeneracy
From (2.7), the classical equation of motion of F is
δS
=0 ⇒ gF † = −∂W . (2.11)
δF
Hence, the scalar potential is given by
V (φ, φ† ) = gF F † = g −1 ∂W ∂W † . (2.12)
V (φ, φ† ) ≥ 0 . (2.13)
2
|Qα |ψi|2 + |Qα̇ |ψi|2 ≥ 0
P P
The SUSY algebra, {Qα , Qα̇ } = 2Pαα̇ , implies that hψ|H|ψi ∝ α α̇
for any state |ψi.
–3–
• The supersymmetric ground states are determined by the minimum of the
scalar potential. The condition for the existence of (classical) SUSY
vacua:
∂W
=0 . (2.14)
∂φ
• In relation to the F auxiliary field, equations (2.14) are also known as the F -
terms or F-flatness condition. The space of solutions to this set of equations is
known as the flat directions or the (classical) moduli space of vacua.
• If (2.14) does not admit a solution, the theory does not have a supersymmetric
ground state and the supersymmetry is broken.
• But the sign of W can be rotated away by the U (1)R symmetry, for example
by taking α = −3π/2 in (2.9), noting that W has R-charge 2.
Suppose that we take R(Φ) = r = 2/3 (so that R(m) = 2/3 and R(λ) = 0)3 .
3π 3π
This maps φ → φe−i 2 r = −φ and ψ → ψe−i 2 (r−1) = iψ.
This map generates a Z4 discrete symmetry. (Indeed, if we raise the gener-
ator to the power of 4 we obtain the identity.)
• These two vacua are physically equivalent, since there is an unbroken symmetry
relating them.
3
One may think of the mass parameter m and the coupling λ as background (i.e. non-dynamical)
chiral multiplets, and so one can assign an R-charge to these background fields. We will further
discuss about this point later.
–4–
Example: The LH model. K = L† L + H † H and W = λ2 LH 2 (L and H stands
for light and heavy, respectively).
H = 0, L = arbitrary . (2.17)
• These classical vacua are inequivalent – there is no symmetry that relates them.
• Make use of the following U (1)Φ × U (1)R charge assignments to the fields:
U (1)Φ U (1)R
Φ 1 1
m −2 0 (3.1)
y −3 −1
W 0 2
–5–
We take the U (1)Φ charge of Φ and the U (1)R charge of Φ to be 1. The
superpotential has to be neutral under U (1)Φ , and has to carry U (1)R charge
2. The charges of m and y immediately follow.
• Note that there is no claim about quantum corrections to the Kähler potential,
which is a real function.
Example: The LH model. Wclassical = 12 mH 2 + λL2 H.
• Classically, the minimum of the scalar potential is reached at
∂H W = mH + λL2 = 0 , ∂L W = 2λLH = 0 , (3.3)
i.e. L = H = 0.
• Observe that the scalar potential contains the following mass terms (|m|2 +
4|λ|2 |L|2 )|H|2 .
• Fixing L, the first F -term in (3.3) gives H = −λL2 /m and we get Weff =
−(λ2 /2m)L4 . This procedure is called “integrating out” the heavy field H.
• The theory has a symmetry U (1)H × U (1)L × U (1)R . We assign the following
charges to the fields, with a requirement that the superpotential carries U (1)R
charge 2 and is neutral under U (1)H and U (1)L .
U (1)H U (1)L U (1)R
H 1 0 1
L 0 1 1
(3.4)
m −2 0 0
λ −1 −2 −1
W 0 0 2
–6–
• Invariance under U (1)H × U (1)L and holomorphy imply that
for some holomorphic function h. Since Weff has R-charge 2 and mass dimension
3, we conclude that
Weff ∼ λ2 L4 /m . (3.6)
Remarks:
• For concreteness, let us consider the theory with a single chiral superfield Φ
with the classical Kähler potential and superpotential
1
Kclass (Φ, Φ† ) = Φ† Φ, W (Φ) = yΦ3 , (4.1)
3
so that the classical Lagrangian is
Z Z
†
L = d θKclass (Φ, Φ ) + d2 θW (Φ) .
4
(4.2)
• Observe that the holomorphic coupling y is not invariant under the field redef-
inition, say Φ → aΦ, where a is a complex constant. However, any physical
quantity must be independent of such a field redefinition.
–7–
• To define
√ a physical coupling, look at the wavefunction renormalisation of Φ,
Φ → ZΦ:
Now let us demonstrate how to compute the beta function for y. One can
compute this perturbatively in a standard way4 , but in the following we will follow
the method of Green–Komargodski–Seiberg–Tachikawa–Wecht [13].
4
P. K. Townsend and P. van Nieuwenhuizen, Dimensional Regularization and Supersymmetry
at the Two Loop Level, Phys. Rev. D20 (1979) 1832; L. F. Abbott and M. T. Grisaru, The Three
Loop Beta Function for the Wess-Zumino Model, Nucl. Phys. B169 (1980) 415-429; A. Sen and
M. K. Sundaresan, The Four Loop Beta Function for the Wess-Zumino Model, Phys. Lett. 101B
(1981) 61-63.
–8–
• A theory of a single free chiral superfield Φ has a global U (1)Φ symmetry
and conserved current Φ† Φ. If we deform the theory by the superpotential
W = 13 yΦ3 , the U (1)Φ symmetry is broken (since the superpotential cannot
be charged under this symmetry). We will see that this gives rise to the beta
function for the coupling yP .
• Φ† Φ is not only the kinetic term but also the superfield version of the U (1)Φ
current associated with Φ → eiα Φ. The θσ µ θ components of Φ† Φ contains
j µ = φ† ∂µ φ − (∂µ φ† )φ.
• Let us write
O = Φ3 , J = Φ† Φ . (4.10)
Hence, the non-conservation equation can be written as
2
D J = qO , q=3. (4.11)
–9–
we obtain the wavefunction renormalisation, to the leading order of |y|2 ,
qNO 2π 2
µ
∆Z(µ) − ∆Z(µ0 ) = − 2 2
log |y|2 + . . . . (4.16)
NJ (4π ) µ0
Using (4.5), we obtain the anomalous dimension, to the leading order in |y|2 ,
∂∆Z 1 qNO 2 9
γ∼− = 2
|y| = 2 |y|2 > 0 . (4.17)
∂ log µ 8π NJ 4π
Note that, by definition (4.4) of yP , we have
|yP |2 = |y|2 Z −3 ∼ (1 − 3∆Z)|y|2 ∼ |y|2 . (4.18)
to the leading order in |y|2 .
• Substituting this into the exact relation (4.8) and taking yP to be real, we
obtain
3 27
βyP = yP γ ∼ 2 yP3 > 0 . (4.19)
2 8π
• Conclusion:
– 10 –
two-point-function of the currents such that Φ has charge +1.
2
• The non-conservation equation D J = qO implies the OPE (see (3.7) of [15]):
q
J(x)O(0) = O(0) + . . . (4.23)
4π 2 |x|2
5 Vector multiplets
5.1 Some conventions
• The Lie algebra of a group G is generated by the generators T A , A = 1, . . . , dim G.
6
From (4.31), we obtain the following expansion
O(x)O† (y) = COO† 1 (|x − y|, ∂y )1(y) + COO† J (|x − y|, ∂y )J(y) + . . . , (4.25)
with
qNO 1 1
COO† J (|x − y|, ∂y ) = + ... , (4.26)
NJ (4π 2 )2 |x − y|4
where . . . includes the terms with derivatives with respect to y. From which we can obtain the
three-point function
hO(x)O† (y)J(z)i = COO† 1 (|x − y|, ∂y ) h1(y)J(z)i +COO† J (|x − y|, ∂y )hJ(y)J(z)i + . . . (4.27)
| {z }
=0
where ∆O = ∆O† = 3 and ∆J = 2, and it follows from (4.26) and (4.21) that
qNO 1 NJ qNO
fOO† J = · = , (4.30)
NJ (4π 2 )2 (4π 2 )2 (4π 2 )4
– 11 –
• Example. In this note, we focus mostly on G = SU (N ). In which case,
dim G = N 2 − 1. The generators may themselves appear as N 2 − 1 matrices
in any representation of the group.
If we take T A in the fundamental representation, then each T A is an N ×
A
N matrix (Tfund )ba , where a = 1, . . . , N and b = 1, . . . , N are indices in the
fundamental and antifundamental representions of SU (N ).
A A A
Note that Tfund are traceless matrices, i.e. tr(Tfund ) = (Tfund )aa = 0, for all
A = 1, . . . , N 2 − 1.
• The gauge kinetic term of Aµ arises from the gauge covariant combination
constructed from Aµ , which is Fµν = [Dµ , Dν ]. For the abelian group, Fµν is
gauge invariant.
– 12 –
Supersymmetric theories
In a supersymmetric theory, we consider an N -component chiral multiplet φ(y µ , θ),
which transform under the gauge transformation as
Φ† eV Φ . (5.10)
• The kinetic term for the vector superfield V arises from a gauge covariant
combination constructed from V , which becomes gauge invariant for an abelian
gauge group. A minimal version is
– 13 –
The gauge kinetic term can be written as
Z
∝ d4 θ tr(Σαα̇ Σβ β̇ )αα̇ β β̇ . (5.13)
is also gauge invariant. This can be seen form the transformation rule (5.9) for
the abelian group: V → V + i(Λ − Λ† ). The integrations of Λ and Λ† over d4 θ
vanish, because they are chiral and antichiral. The term (5.16) is known as the
Fayet–Iliopoulos term.
where
θ 4πi
τ= + 2 . (5.19)
2π g
– 14 –
• In terms of the components, this can be rewritten as
Z
4
h 1 A Aµν θ A e Aµν
Sgauge = d x − 2 Fµν F + 2
Fµν F
4g 32π
(5.20)
i A 1 i
− 2 λ σ µ Dµ λA + 2 DA DA ,
g 2g
where
A A
Fµν = Fµν T = ∂µ Aν − ∂ν Aµ + i[Aµ , Aν ]
= ∂µ A A A ABC B C
Aµ Aν T A ,
ν − ∂ν Aµ + f
Dµ λ = ∂µ λ + i[Aµ , λ] , (5.21)
1
Feµν = F ρσ µνρσ .
2
– 15 –
5.5 One-loop exactness of the holomorphic gauge coupling
• τ appears in the action as a background chiral superfield. Hence, it must run
and stay holomorphic:
∂ θ 4π
βτ = + i 2 must be holomorphic in τ . (5.26)
∂ ln µ 2π g
– 16 –
βτ is perturbatively one-loop exact.
4π 2 i
i = − b0 (ln |Λ| − ln µ)
g 2 (µ) 2π
|Λ|eiθ/b0
θ 4π ib0
⇒ τ= +i 2 =− ln (5.33)
2π g 2π µ
Λ := |Λ|eiθ/b0 , (5.34)
and obtain
b0
8π 2 Λ
2πiτ (µ) − +iθ
e =e g 2 (µ) = . (5.35)
µ
Comments:
– Relation (5.35) is the dimensional transmutation; Λ is intrinsic to the
theory – it is not a cut-off.
– Since Λ has a mass dimension 1, the quantum theory is not scale invariant
even if the classical theory is.
– Although Weff is still constrained by the perturbative non-renormalization
theorem, the presence of holomorphic Λ permits, in many cases, a non-
perturbative renormalization of the superpotential. We will mention an
example due to Affleck–Dine–Seiberg [16] shortly.
2 2
– Due to the factor e−8π /g , (5.35) is regarded as the one-instanton effect.
This effect brings about the factor Λb0 .
– 17 –
Vector multiplet. Gauge fields are collected in a vector multiplet; it contains
Aµ
λ ψ (6.1)
φ
to exhibit the SU (2)R symmetry which acts on the rows; Aµ and φ are singlets
and λ, ψ are a doublet.
Let U (1)J be the Cartan generator of SU (2)R . The fields carry the following
charges under U (1)J :
• It consists of two Weyl fermions ψq and ψqe† and complex bosons q and qe†
ψq
q qe† (6.4)
ψqe†
– 18 –
• In terms of the 4d N = 1 language, these fields make up two chiral multiplets
Q containing (q, ψq ) and Q
e containing (e
q , ψqe).
The U (1)R charges for Q and Q e are R(Q) = R(Q) e = 0, and so
q†) = 0 ,
R(q) = R(e R(ψq ) = −1 , R(ψqe† ) = 1 . (6.7)
and so
R(q) = R(e
q) = 0 , R(ψq ) = R(ψqe) = −1 . (6.8)
– 19 –
• In the N = 2 notation, the N = 4 SU (Nc ) gauge theory has an N = 2
SU (Nc ) vector multiplet and one hypermultiplet in the adjoint representation
of SU (Nc ).
From (5.30), rotating the phase λ → eiα λ is equivalent to shifting the θ angle,
Thus, e2πiτ → e+2αNc i e2πiτ . Since e2πiτ ∼ Λb0 , we see that the R charge of Λ is
– 20 –
7.1 Anomaly of the R-symmetry
• As shown earlier in (5.27) and (5.28), in this theory the current associated
to this U (1)R symmetry is not conserved at the quantum level. The U (1)R
symmetry is hence anomalous and is dynamically broken.
the integer k is known as the instanton number, and the measure (7.6) is
said to be evaluated in the k-instanton background.
As discussed earlier, this transformation in the path-integral measure is equiv-
alent to the shift in the θ-angle by
θ → θ + 2Nc α . (7.8)
(i.e., α = 2πn/2Nc ) leaves the path integral invariant. Thus, at this point, we
see that, due to instanton effects,
• We’ll see soon that gaugino condensation spontaneously breaks the Z2Nc sym-
metry to Z2 in each vacua.
– 21 –
• Since Λ3Nc has an R charge 2Nc and a mass dimension 3Nc , the only object
that is compatible this is the condensate
• The quantity h(tr λλ)Nc i can be computed using the cluster decomposition,
where the last equality holds only when the separation between the xi is large
compared to the length scale Λ−1 .
However, it can be shown that htr λλ(x1 ) tr λλ(x2 ) . . . tr λλ(xNc )i is inde-
pendent of x1 , . . . xNc . In other words, the product of tr λλ has no short-
distance singularities. (In fact, tr λλ is the lowest component of the chiral su-
perfield S = tr W 2 , also known as the glueball superfield. Since the product
of the chiral superfields has no short-distance singularities, the former state-
ment follows. We will demonstrate this shortly.)
This statement means that we can always take the separation between the xi
to be large compared to Λ−1 and obtain, using (7.10),
Taking the Nc -root on both side, we see that there are Nc vacua, labelled by
m = 0, 1, . . . , Nc − 1, such that in the m-th vacuum,
cΛ3 e2πim/Nc , e
htr λλim = e c 6= 0 . (7.12)
• These Nc vacua are permuted by the Z2Nc symmetry (or equivalently by the
shift θ → θ + 2π). The generator of the group Z2Nc sends the m-th vacuum to
the (m + 1)-th and multiplies htr λλi by e2πi/Nc .
– 22 –
• Due to the condensate hλλi 6= 0, the standard expectations about the N = 1
pure SU (Nc ) gauge theory are as follows: (1) the theory has a mass gap, (2) the
theory is believed to have a confinement and (3) chiral symmetry breaking: Z2Nc
is broken to Z2 , and there are precisely Nc vacua required by this symmetry
breaking.
Statement (1) implies that fermion masses are dynamically generated. This
suggests that Z2Nc is broken down to a subgroup that allows bare masses for
elementary fermions. The largest such subgroup is Z2 .
• For the super–Yang–Mills theory with a general gauge group G, the number
of vacua is equal to the dual Coxeter number h∨ (G) of G, assuming that G is
connected, simple and simply-connected8 .
Dα̇ Wβ = 0 . (7.13)
S ≡ tr W 2 = WαA W Aα , (7.14)
we thus have
Dα̇ S = 0 . (7.15)
– 23 –
• In fact, the first term hDα̇ (. . .)i in the first line of the second equality of (7.17)
also vanishes. The argument is as follows.
Since the supercharge Qα̇ can be viewed as a differential operator associated
with the superspace derivative Dα̇ , we may write
This of course holds for any derivative with respect to xjαα̇ with j = 1, . . . , n.
• Recalling that the lowest component of S is tr λλ, we thus arrive at the con-
clusion that htr λλ(x1 ) tr λλ(x2 ) . . . tr λλ(xNc )i is independent of x1 , . . . xNc .
• Given a dynamical scale Λ, we can study the theory in the limit L Λ−1 and
L Λ−1 . In a supersymmetric theory, it can be argued that the number of
vacua in both limits should agree.
Note that L Λ−1 is weakly-coupled, so we can perform an honest computa-
tion of the counting of the vacua.
H = P 0 = {Q, Q† } . (7.20)
{(−1)F , Q} = 0 . (7.21)
– 24 –
become those with E 6= 0, but these are necessarily composed of pairs of a
fermionic and a bosonic states. In other words, this does not change tr (−1)F .
E
• ×
• ×
• (7.23)
• The limit L Λ−1 . (In the following we follow the argument in section 8 of
Witten’s 1982 paper [21].)
The system is weakly coupled. Since non-zero electric and magnetic fields
contribute to the energy of the system, we set Fµν = 0 for µ, ν = 1, 2, 3 to have
almost zero energy.
The low energy degrees of freedom are the commuting holonomies
U1 , U2 , U3 ∈ SU (Nc ) , (7.25)
corresponding to the directions x, y and z. Since these commute with each
other, they can be simultaneously diagonalised into the following form
– 25 –
– In the following, we treat θµa and aα , with a, b = 1, . . . , Nc − 1, as the
r = Nc − 1 (7.28)
ZW = tr(−1)F = r + 1 = Nc . (7.31)
– 26 –
These r + 1 states have spin 21 r. The Witten index is therefore
– In both cases, we see that the absolute value of the Witten index is
the rank of gauge group SU (Nc ) plus one:
|ZW | = r + 1 = Nc . (7.34)
|Ei
Q|Ei Q† |Ei (7.35)
† †
(Q Q − QQ )|Ei
This involves 4 states. If either Q|Ei or Q† |Ei is zero, the multiplet has 2 states. If
Q|Ei = Q† |Ei = 0, the multiplet has 1 state, and, in addition, E is zero due to
– 27 –
• Using a similar argument as above with r being the rank of Spin(Nc ) would
lead to a conclusion that |ZW | = bNc /2c + 1, but this is incorrect.
The reason is that for Spin(Nc ) with Nc ≥ 7, there are two classes of holonomies.
The first class consists of the commuting holonomies that are continuously con-
nected to the identity (similarly to the above discussion), and the other class
consists of those that are not connected to the identity. In Witten’s 1982 pa-
per [21] the second class was not taken into account, but later in 1997 Witten
himself [22] corrected this mistake and obtained the correct result.
In the following, we review the argument presented in Appendix I of [22].
• The Witten index receives two contributions, from each class of the holonomies.
– The first class, consisting of the holonomies in the Cartan torus of Spin(Nc ),
contributes rG + 1 = bNc /2c + 1 to the Witten index, by the similar ar-
gument as before.
– The second class requires more discussion.
Let us start from Nc = 7. The commuting holonomies U1 , U2 , U3 are
diagonal matrices with eigenvalues (U1 , U2 , U3 ) = (±1, ±1, ±1), with each
of the 23 − 1 = 7 combinations of signs other than (1, 1, 1) appearing with
multiplicity one. Explicitly, they are
Obviously, these matrices are not connected to the identity and does not
admit any deformation. Hence the unbroken subgroup of Spin(7) has
rank 0 in this case. Thus, the contribution of the second class of the
holonomies to the Witten index is 0 + 1 = 1. Summing the first and
second contribution together, we obtain
– 28 –
8 Supersymmetric QCD
We consider a 4d N = 1 SU (Nc ) gauge theory with Nf chiral multiplets Q in the an-
tifundamental representation of SU (Nc ) and Nf chiral multiplets in the fundamental
representation. We also take the superpotential to be zero.
where a, b, c = 1, 2, . . . , Nc and i, j, k = 1, 2, . . . , Nf .
• In components,
Nf Z h
X i
Skin = d x Dµ Q†i Dµ Qi + iψ σDψi + F †i Fi + Q† i ψi λ
4
i=1 (8.3)
i
i
λ ψ Qi + Q†i DQi + Qi → Q
ei , ψi → ψei , Fi → Fei
where the contraction of gauge indices is in each case unique, e.g. in the term
Q†i DQi the indices are contracted as
Q Q
(8.6)
e
Nf Nc Nf
At this stage, the R-symmetry can be taken to be U (1)X that rotates the phase
of the fermions as
It can be checked that the Lagrangian is invariant under these actions. The
R-charges of Q, Q
e and λ are therefore 0, 0, 1 respectively.
– 29 –
It is common to take the following linear combinations of U (1)Q and U (1)Qe :
We will shortly see that both U (1)A and U (1)X are anomalous in the quantum
theory.
The equations on the right are sometimes called the D-terms. They lead to
flat directions which we refer to as the classical moduli space Mcl
Nf ,Nc .
Case I: Nf < Nc .
e† in the form:
There exist gauge and flavour transformations that put Q and Q
a1
a2
Q∼
∼Q e† , (8.13)
.
aNf N ×N
f c
– 30 –
2. The gauge symmetry is broken SU (Nc ) → SU (Nc − Nf ), so
dimC Mcl 2
Nf <Nc = Nf . (8.14)
Mij = Qai Q
ej .
a (8.15)
Observe that Mij has Nf2 complex DOFs, precisely that of dimC Mcl
Nf <Nc .
Case II: Nf ≥ Nc .
Up to gauge and flavour rotations
a1 ea1
a
a2
e
2
. .
†
Q= , Q = , for some ak , a0k ≥ 0
e
aN c aNc
e
Nf ×Nc Nf ×Nc
2 2
|ai | − |e
ai | = r (independent of i). (8.16)
Mij = Qai Q
eja ,
a
Bi1 ...iNc = a1 ...aNc Qai11 · · · QiNNcc ,
e i1 ...iNc = a1 ...aNc Q
B e i Nc ,
e i1 · · · Q (8.17)
a1 aNc
where (∗B)iNc +1 ...iNf = N1c ! i1 ...iNf B i1 ...iNc and a ‘·’ denotes a contraction of an
upper with a lower flavour index.
– 31 –
• For Nf = Nc , the classical relations reduce to a single one:
det M − (∗B)(∗B)
e =0, (8.19)
• The moduli spaces, for all Nf ≥ Nc , are singular at the origin, where M = B =
B
e = 0.
The classical interpretation is that the SU (Nc ) gauge fields, which are massless
at the origin, need to be included for the low-energy effective theory to be
non-singular.
• We will, however, shortly see that the relation (8.19) gets modified in the
quantum theory, and there is no singularity for the case of Nf = Nc .
Moreover, for Nf < Nc , we will shortly see that quantum effects lift the classical
vacua completely, and there is no supersymmetric vacuum.
– 32 –
• Quantum mechanically, these U (1) symmetries transform the path-integral
measure and the θ-angle as in the above table; this can be computed simi-
larly to (7.6). For example, for the U (1)X symmetry,
DλDψDψe
Z
iα 4 A eB
A B A B A B
→ DλDψDψ exp
e d x F F tr(Tadj Tadj ) − Nf tr(Tfund Tfund ) − Nf tr(Tantifund Tantifund )
16π 2
Z
iα
4 A eA
= DλDψDψ exp d x F F T (adj) −N T (fund) −N T (antifund)
f f
e
16π 2
| {z } | {z } | {z }
Nc 1/2 1/2
Z
iα
= DλDψDψe exp (2Nc − Nf − Nf ) d4 x F A FeA
32π 2
= DλDψDψe exp [iα(2Nc − Nf − Nf )k] , k∈Z.
(8.23)
Hence, U (1)A and U (1)X are anomalous, whereas U (1)B is anomaly-free.
– 33 –
• The scale Λb0 = Λ3Nc −Nf transforms under the anomalous U (1)A symmetries
by virtue of the transformation of the θ-angle; see (8.22).
• On the other hand, we can work with the canonically normalised kinetic term:
Z
1 2 1 θ
dθ 2
− i 2 WαA (gVc )W A α (gVc ) + c.c. (8.29)
4 g 8π
where g is real and physical, and Vc is a real vector superfield (the subscript c
stands for “canonical”). Note that this Lagrangian is no longer holomorphic
in the combination g12 − i 8πθ 2 .
Goal: Determine how the coupling g runs, i.e. compute the beta function of
g in the canonically normalised kinetic term (8.29).
Note that such a beta function was first computed in the N = 1 gauge theory
with matter by Novikov–Shifman–Vainshtein–Zakharov (NSVZ) [23]. In the
following, we will follow the argument by Arkani-Hamed and Murayama [24].
– 34 –
Note that the wavefunction renormalisation factor ZW of W (gVc ) can be ob-
tained by simply looking at the kinetic term of the gauginos – it is simply
ZW = 1/g 2 . (8.31)
8π 2
2
8π
− Re (2πiτ ) = 2 − Nc log + Nc log(8π 2 ) = −3Nc log (|Λ|/µ) . (8.34)
g g2
We thus arrive at the “exact” NSVZ beta function for the SU (Nc ) super–
Yang–Mills theory9 :
3Nc 16π 2
β 8π2 = = − βg . (8.36)
g2 1 − g 2 Nc /8π 2 g3
i = 1, 2, 3). Below the scale M , the theory looks like the N = 1 pure SYM. This gives rise to an
explicit regularisation for pure N = 1 SYM with a cutoff M , which also manifestly preserves the
holomorphy.
– 35 –
• Let us recall the anomalous U (1)A symmetry that rotates Qi and Q ej as Qi →
1/2 ej → Ze1/2 Q
ej , where Z 1/2 , Ze1/2 = eiα is a pure phase.
Zi Qi and Q j i j
where
∂ log Zi ∂ log Zei
γi = − , ei = −
γ . (8.42)
∂ log µ ∂ log µ
• We thus arrive at the NSVZ beta function [23]:
PNf PNf
b0 + 21 i=1 γi + 12 j=1 γ
ej 16π 2
β 8π2 = = − βg . (8.43)
g2 1 − g 2 Nc /8π 2 g3
This is an “exact” relation10 , up to the yet-to-be-determined anomalous di-
mensions γi and γ
ej .
10
Note that the denominator is scheme dependent, whereas the numerator is not. We refer the
reader to footnote 9 regarding the issue of the regularisation scheme.
– 36 –
• In supersymmetric QCD, where W = 0, all charged fields are related by sym-
metry, and therefore have the same anomalous dimension γ0 . We may write
3Nc − Nf (1 − γ0 )
β 8π2 = . (8.44)
g2 1 − g 2 Nc /8π 2
Some comments.
• Higher-loop corrections to the β-function is equivalent to a Taylor expansion
in λ ≡ g 2 Nc (known as the ’t Hooft coupling).
f N
λ2 3 − Nc (1 − γ0 )
βλ = − 2 · . (8.45)
8π 1 − 8πλ2
• For a theory with gauge group G (coupling α), with charged fields φi in the
representations ri with anomalous dimensions γi and tr TrAi TrBi = T (ri )δ AB , we
have
α2 3T (adj) − i T (ri )(1 − γi ) α2 b0 + i T (ri )γi
P P
βα = − · =− · . (8.46)
2π 1 − αT (adj)/2π 2π 1 − αT (adj)/2π
– In a theory with many gauge groups and many gauge couplings, this
formula holds separately for each gauge coupling.
– The condition for the existence of the conformal fixed point, i.e. there
is a value of the coupling α = α∗ such that βα∗ = 0, is therefore
X
3T (adj) − T (ri )(1 − γi ) = 0 (8.47)
i
– 37 –
8.4 Anomalous dimensions and the Banks-Zaks fixed point
In this section, we examine the theory in the limit Nc , Nf → ∞, with the ’t Hooft
coupling λ ≡ g 2 Nc → 0 and b0 = 3Nc − Nf ∼ O(1).
λ
γ0 ∼ −c + O(λ2 ) , (8.50)
8π 2
where an explicit two-loop computation shows that c > 0 [25].
• For Nf = 3Nc − b0 , where b0 ∼ O(1), the beta function takes the form
g3 3cNc2 2
βg ∼ − b0 − g . (8.52)
16π 2 8π 2
• For b0 ≤ 0 (i.e. Nf ≥ 3Nc ), βg > 0 for small g, so the gauge coupling g flows
back to zero in the infrared (i.e., the theory is IR free).
g=0
• (8.53)
8π 2 b0
g∗2 = . (8.54)
3cNc2
– 38 –
• Since perturbation theory is an expansion in λ, all three-loop and higher cor-
rections are suppressed by higher powers of 1/Nc . Therefore, this Banks-Zaks
fixed point g∗ survives to all orders.
• Assume that O is a scalar operator, Mµν |Oi = 0. Using the last commutation
relation of (8.58), we obtain the square of the norm
– 39 –
• Now let us consider
Note that K µ has scaling dimension −1. We also have the relation
α̇ µν α̇ α̇ 3
{S , Qβ̇ } = (σ )β̇ Mµν + 2δβ̇ D − R (8.64)
2
– 40 –
Argument. We look for the highest weight representation such that Pµ† = K µ and
† α̇
Qα̇ = S . Let us assume that O is a scalar operator, so Mµν |Oi = 0. Suppose also
that the operator O has the lowest scaling dimension in the superconformal multiplet
– known as the superconformal primary. Then, |Oi is the highest weight state such
α̇
that Kµ |Oi = 0 and S α |Oi = S |Oi = 0. The square of the norm of Qα̇ |Oi is
2 α̇ α̇ 3
|Qα̇ |Oi| = hO|S Qα̇ |Oi = hO|{S , Qα̇ }|Oi = 4 ∆ − r . (8.66)
2
Hence
3
∆≥ r. (8.67)
2
The equality holds if and only if Qα̇ |Oi = 0, i.e. O is a chiral operator.
An important remark. The inequalities (8.57) and (8.67) apply strictly only to
gauge invariant operators. This is because the Hilbert space of gauge theory is
only positive-definite after imposing the gauge (=BRST) invariance.
e = Nf − Nc .
R(Q) = R(Q) (8.70)
Nf
This implies the non-anomalous R charges of Q and Q
e derived earlier in
(8.26).
• A necessary condition to have such a fixed point is that
3 3
∆(M ) = · 2R(Q) ≥ 1 ⇔ Nf ≥ Nc . (8.71)
2 2
If the bound is saturated (Nf = 23 Nc ), the meson M is free.
When Nf < 23 Nc , the theory has no fixed point, and since b0 = 3Nc − Nf > 0
the theory is asymptotically free at one-loop.
asymp. free (1-loop) BZ IR free
| •| Nf (8.72)
no non-trivial fixed point 3
2 Nc 3Nc
– 41 –
• When Nf is slightly larger than 32 Nc , the anomalous dimension γ0 is of order
1, and we lose perturbative control of the theory. We can, nevertheless, study
the theory in this regime using the Seiberg duality [28].
• Indeed, Seiberg [28] conjectured that there is always a non-trivial fixed point
for the whole interval of
3
Nc < Nf < 3Nc . (8.73)
2
This interval is also known as the conformal window.
• The electric theory. The SU (Nc ) gauge theory with Nf chiral multiplets
Q in the antifundamental representation and Nf chiral multiplets Q
e in the
fundamental representation, and zero superpotential.
Q Q
(8.74)
e
Nf Nc Nf
M0
q qe
Nf Nc0 Nf (8.76)
Comments.
• Note that the duality in the above form holds for Nc ≥ 2 and Nc0 = Nf −Nc ≥ 2.
When Nc or Nc0 is 0 or 1, a small modification needs to be made (we will discuss
about this in detail shortly).
• The magnetic theory also has a global symmetry SU (Nf ) × SU (Nf ) × U (1)B ×
U (1)R , like the electric theory. (There is also an anomalous U (1)A symmetry.)
– 42 –
The matter content of the magnetic theory and the transformation rules are
summarised below.
SU (Nc0 ) SU (Nf )1 SU (Nf )2 U (1)B U (1)R U (1)A
q N0c Nf 1 Nc /Nc0 (Nf − Nc0 )/Nf 1
qe N0c 1 Nf −Nc /Nc0 (Nf − Nc0 )/Nf 1
M0 1 Nf Nf 0 2Nc0 /Nf −2
0
(Λ0 )3Nc −Nf 1 1 1 0 0 2Nf
(8.77)
Observe that the superpotential carries R charge 2, as required.
– When 32 Nc < Nf < 3Nc , the electric and magnetic theories flows to the
same superconformal fixed point in the infrared (IR).
– When Nc +2 ≤ Nf ≤ 32 Nc , the electric theory is strongly coupled in the IR.
Nevertheless, this has a weakly coupled description, which turns out to be
the magnetic theory. (This also means that the IR free magnetic theory
has an ultraviolet completion as the asymptotic free electric theory.)
• Note also that the composite mesons (Mmag )ij := qai qeja of the magnetic
theory vanish because of the F -terms:
– 43 –
Electric theory. R(Q) = R(Q) e = 1 − Nc /Nf
⇒ R(M ) = 2(1 − Nc /Nf ) and R(B) = R(B)
e = Nc (1 − Nc /Nf ).
3. Decoupling a flavour. Let us give mass to a flavour, integrate it out and see
if the duality still holds. We add the following mass term to the superpotential
of the electric theory:
δW = mQai=Nf Q eai=Nf . (8.81)
In the energy scale far below m, we effectively have Nfnew = Nf − 1 flavours,
whereas Nc remains unchanged.
The corresponding superpotential in the magnetic theory can be obtained using
the operator map (8.78)
i=N
W = qai Mij qeia + mMi=Nff . (8.82)
i=N
From the F -term ∂W/∂Mj=Nff = 0, we obtain
a i=Nf
qi=Nf
qea +m=0 . (8.83)
a i=N
This gives a vacuum expectation value to qi=N f
and qea f . As a result, the
original SU (Nc0 ) = SU (Nf − Nc ) gauge symmetry is broken to SU (Nc0 − 1) =
SU ((Nf − 1) − Nc ) = SU (Nfnew − Nc ). The Seiberg duality is thus compatible
with decoupling a flavour.
The anomaly computed in the IR must agree with the anomaly ob-
tained from the UV calculation using the elementary fields. This is
the ’t Hooft anomaly matching condition.
Justification:
– 44 –
• This can be cured by add some massless fermions in another rep R0 that
couple only to the gauge fields of H. The rep R0 is chosen so that the
combined theory is anomaly-free, i.e. A(R) + A(R0 ) = 0.
• Suppose our original theory with rep R has a low energy description in
which the fermions are in representation R0 .
Then, the IR description of the combined theory is still anomaly-free be-
cause it comes from an anomaly-free UV theory.
Since the additional massless fermions stay the same, A(R0 ) + A(R0 ) = 0.
• A(R) = A(R0 ) is the ’t Hooft anomaly matching condition.
The quantum numbers of the elementary fermions in the electric theory are
SU (Nf )1 SU (Nf )2 U (1)B U (1)R
ψQ Nf 1 1 −Nc /Nf
(8.84)
ψQe 1 Nf −1 −Nc /Nf
λ 1 1 0 +1
Those for the magnetic theory are
SU (Nf )1 SU (Nf )2 U (1)B U (1)R
0
ψq Nf 1 Nc /Nc −1 + Nc /Nf
ψqe 1 Nf −Nc /Nc0 1 + Nc /Nf (8.85)
ψM 0 Nf Nf 0 1 − 2Nc /Nf
λ0 1 1 0 +1
From this fermion content, it is straightforward to check the matching of all of
the possible anomaly coefficients:
electric magnetic
3
(SU (Nf )1 ) Nc A(Nf ) (−Nc0 + Nf )A(Nf ) = Nc A(Nf )
(SU (Nf )1 )2 U (1)B Nc T (Nf ) · (1) = 21 Nc Nc0 T (Nf )(Nc /Nc0 ) = 12 Nc
(SU (Nf )1 )2 U (1)R Nc
Nc T (Nf )(− N f
) Nc0 T (Nf )(−1 + Nc
Nf
) Nc
+ Nf T (Nf )(1 − 2 Nf
)
= − 12 Nc2 /Nf = − 12 Nc2 /Nf
(U (1)B )2 U (1)R Nc
2Nc Nf (1)2 (− Nf
) = −2Nc2 2Nc0 Nf ( Nc 2
N0
) (−1 + Nc
Nf
) = −2Nc2
c
Nc
U (1)R grav. grav. 2Nc Nf (− Nf
) + (Nc2 − 1) 2Nc0 Nf (−1 + Nc
Nf
) Nc
+ Nf2 (1 − 2 N f
) + (Nc02 − 1)
= tr [U (1)R ] = −(Nc2 + 1) = −(Nc2 + 1)
(U (1)R )3 2Nc Nf (− Nf
) + (Nc2 − 1) 2Nc0 Nf (−1 +
Nc 3 Nc 3
Nf
) Nc 3
+ Nf2 (1 − 2 N f
) + (Nc02 − 1)
= Nc2 − 1 − 2Nc4 /Nf2 = Nc2 − 1 − 2Nc4 /Nf2
(8.86)
where in the first line we take tr[Tra {Trb , Trc }]
= A(r)d , with d a totally abc abc
symmetric tensor and independent of the representation r.13 A(r) is often re-
13
For SU (n), with n ≥ 3, the unique symmetric invariant dabc appears in the anticommutator of
representation matrices of the fundamental representation {TnA , TnB } = n1 δ AB + dABC TnC .
– 45 –
ferred to as the anomaly coefficient14 .
Comment. In the last two of these relations, the dual gauginos λ0 give a
contribution which is necessary for the success of the consistency check. The
matching requires us to take seriously the realization of the full SU (Nc0 ) vector
multiplet as a set of physical asymptotic states.
This gives vacuum expectation values (VEVs) to qi and qei for the last Nc0 − k
flavours. This breaks the SU (Nc0 = Nf − Nc ) gauge group to the SU (k) gauge
group.
Hence, the low energy effective (magnetic) theory is the SU (k) gauge theory
with Nc + k flavour. This theory is IR free.
• Since the theory is weakly coupled in the IR, we understand the IR behaviour
using the instanton computation.
0
The one-instanton effects bring about the factor (Λ0 )b0 = (Λ0 )3Nc −Nf , with Λ0
the dynamical scale of the magnetic theory.
14
If r and r are complex conjugate representations, then A(r) = −A(r). Hence, A(r) = 0 if r is
real or pseudoreal. For SU (n), A(n) = 1, A(∧2 ) = n − 4 and A(S 2 ) = n + 4, where ∧2 and S 2 are
the rank-two antisymmetric and symmetric representations, respectively.
– 46 –
To determine the effective superpotential, we can use the anomalous U (1)A
symmetry under which
0
A[(Λ0 )3Nc −Nf ] = 2Nf , A[M ] = −2 , A[m] = A[W 0 ]−A[M ] = 0−(−2) = 2 .
(8.89)
0
A combination involving Λ , M and m that is invariant under U (1)A , SU (Nc +k)
and SU (Nc0 − k) is
0 det M(Nc +k)×(Nc +k)
(Λ0 )3Nc −Nf . (8.90)
det m(Nc0 −k)×(Nc0 −k)
(Checking the U (1)A charge: 2Nf + (Nc + k)(−2) − (Nc0 − k)(2) = 0.) The
R-charge of this combination is
0
2Nc0
2Nc 0
0 + (Nc + k) − (Nc − k) 2 − = 2k (8.91)
Nf Nf
Since the superpotential must have R-charge 2, the combination (8.90) can
enter the superpotential if and only if k = 1.
• Let us focus on k = 1.
The low energy magnetic theory is an SU (1) theory with Nc +1 flavours, and the
0
factor (Λ0 )3Nc −Nf /(det m) is just a numerical factor. Hence, the superpotential
of the magnetic theory is modified to be
Note that in this special case, we can use the operator maps
where B and B e are the baryons and antibaryons in the electric theory. The
low energy superpotential of the magnetic theory is therefore
0
Weff ∝ (∗B)i Mij (∗B)
e j − det M , (8.94)
where this is written in terms of low energy variables of the electric theory.
Note the B, M , Be and the superpotential have mass dimensions Nc , 2, Nc and
3 respectively. We can restore the dynamical scale Λ for the electric theory as
0 1 h i j
i
Weff = 2Nc −1 (∗B) Mi (∗B)j − det M .
e (8.95)
Λ
– 47 –
The infrared limit of the SU (Nc ) SQCD with Nc + 1 flavours is de-
scribed by the Wess–Zumino model with (Nc + 1) × (Nc + 1) meson
matrix Mji , Nc + 1 baryons ∗B i , Nc + 1 antibaryons ∗Bei and the
superpotential
1 h i
W = 3Nc −(Nc +1) (∗B)i Mij (∗B)
e j − det M . (8.96)
Λ
– One can match the mass dimensions, the U (1)A charges and the U (1)R
charges of each side as follows:
• A comment on the SU (2) gauge group. In general, for SU (2) SQCD with
Nf ≥ 3 flavours, the quarks and antiquarks transform in the doublet (which
is a pseudoreal representation) of the SU (2) gauge group, and so the flavour
symmetry is in fact SU (2Nf ). Note that SU (2Nf ) contains SU (Nf )×SU (Nf )×
U (1)B as a subgroup, with the branching rule:
We can therefore combine the quarks Qai and antiquarks Q eia into Qa (with
I
I, J = 1, 2, . . . , 2Nf and a, b = 1, 2).
Since the baryons B = QQ and the antibaryon B e = Q eQe are quadratic, we
can combine them with the mesons M = QQ, e which is also quadratic into the
SU (2Nf ) × SU (2Nf ) antisymmetric matrix
– 48 –
3N −N 3N −(N −1)
• Let ΛNf c f and ΛNf c−1 f be the instanton factors of the theory before and
after decoupling one flavour, respectively.
Recall the following anomalous U (1)A charges:
3N −Nf 3N −(Nf −1)
A[ΛNf c ] = 2Nf , A[ΛNf c−1 ] = 2(Nf − 1) , A[m] = −2 . (8.101)
By matching the mass dimensions and the U (1)A charges, we obtain the relation
3N −(Nf −1) 3N −Nf
ΛNf c−1 = mΛNf c . (8.102)
• In conclusion, the quantum moduli space of vacua for the SU (Nc ) SQCD with
Nf = Nc flavours is generated by an Nc × Nc meson matrix M , a baryon ∗B,
an antibaryon ∗B,e subject to the relation
e = Λ2Nc .
det M − (∗B)(∗B) (8.108)
– One can match the mass dimensions, the U (1)A charges and the U (1)R
charges of each side as follows:
– 49 –
• This should be contrast with the classical moduli space, which is discussed
earlier in (8.19):
det M − (∗B)(∗B) e =0. (8.110)
The classical moduli space has a singularity at the origin, where M = B =
B
e = 0. However, this singularity is resolved in the quantum theory.
where X is the Lagrange multiplier that imposes the condition (8.108) and we
write ΛNc for the dynamical scale for the SU (Nc ) SQCD with Nc flavours.
In the SU (Nc ) gauge theory with Nf < Nc theory, the baryon and antibaryon
vanish identically, because the antisymmetrisation over Nc indices where each
index runs from 1, . . . , Nf gives zero. Plugging this back to W , we obtain
2Nc Λ2N
Nc
c
W = XΛ =m . (8.113)
det M(Nc −1)×(Nc −1)
Using a similar argument as before, the dynamical scale ΛNc −1 for the theory
with Nc − 1 flavours is related to ΛNc by
3N −(Nc −1)
c 3Nc −Nc
ΛNc −1 = mΛNc
. (8.114)
Thus, the low energy effective superpFotential for the SU (Nc ) SQCD with
Nc − 1 flavours is
3Nc −(Nc −1)
ΛNc −1
W = . (8.115)
det M(Nc −1)×(Nc −1)
• This can be easily generalised to the SU (Nc ) with Nf < Nc flavours. The
superpotential is
1
Λ3Nc −Nf
N
c −Nf
N <N
Wefff c = (Nc − Nf ) . (8.116)
det M
– 50 –
– One can match the mass dimensions and the U (1)A charges of each side
as follows:
Λb0 =3Nc −Nf det M Weff
mass dim. 3Nc − Nf 2Nf 3 (8.117)
U (1)A 2Nf 2Nf 0
• This is a notable example of non-perturbative corrections to the effective su-
perpotential.
Nf
N <Nc −N
• Notice that Wefff ∼M c −Nf
. Then, the scalar potential is
− N 2N c
V = |∂M W |2 ∼ |M | c−N f . (8.118)
9 Supersymmetric index on S 3 × S 1
9.1 Supersymmetry on S 3 × R and S 3 × S 1
The detailed discussion on rigid supersymmetry on S 3 × R and S 3 × S 1 can be found
in [30]. Here we simply summarise the argument.
The case of S 3 × R.
• The supersymmetry algebra is SU (2|1)l × SU (2)r , which is a subalgebra of the
flat space superconformal algebra SU (2, 2|1). The bosonic subalgebra of the
former is SU (2)l × SU (2)r × U (1). This is the isometry of S 3 × R.
– 51 –
• If the theory has an R-symmetry
1
H = P0 + R . (9.4)
ρ
• There are additional parameters we can turn on. These can be thought of as
background fields.
For every global non-R-symmetry U (1)s , there is a conserved current jµs and
charge Qs . We can add background gauge fields asµ by coupling them to the
currents. We turn on background gauge fields which preserve the SU (2)l ×
SU (2)r × U (1) isometry as0 = vs /ρ, where vs are dimensionless real constants.
Denoting by qis the U (1)s charge of φi , this background gauge field has the
effect of changing the parameters qi in (9.3) as
X
qi → qi + qis vs . (9.6)
s
The case of S 3 × S 1 .
– 52 –
• We compactify the Euclidean time direction to S 1 . The partition function of
the theory can be written as
ZS 3 ×S 1 = e−βESUSY IS 3 ×S 1 (9.7)
and the objects in the exponent of the index commute with all the elements of
this supergroup.
– 53 –
where R is the R-charge and Qs are the U (1)s charges of the states.
For our purpose, it will also be convenient to explicitly refine the index with
respect to the quantum number jr of SU (2)r , so the above contribution to the
index can be written as
" !#
β X
± exp − 2jl + R + µr (2jr ) + vs Qs
ρ s
Y s
2jl +R 2jr Q
= ±t y us (9.13)
s
jl +jr + R jl −jr + R s
Y
= ±p q
2 2 uQ
s ,
s
where µr is the chemical potential for the SU (2)r and we have defined
β β β
t = e− ρ , y = e− ρ µr , us = e− ρ vs
(9.14)
p = ty , q = ty −1 .
The quantities t, y, us are respectively the fugacities for 2jl + R, 2jr , and the
non-R global (flavour) symmetries.
– 54 –
• The zero modes of φ, ψ, φ† , ψ † , their charges, and their contributions to the
index I are listed below. The higher modes can be obtained by applying
spacetime derivatives on the zero modes.
Among the four spacetime derivatives ∂αα̇ (with α = ± and α̇ = ±), only two
of them, namely ∂+± , contribute non-trivially from the index.
• We then need to build up the Fock space from these one-particle states.
The bosonic Fock space is built up from the vacuum and positive powers
m n
of ∂++ ∂+− φ (with m, n ≥ 0) acting on the vacuum. The contribution to the
index is
" #
Y Y X R R s
k Y Y 1
pm+ 2 q n+ 2 uQ
s = Qs
. (9.18)
m+ 2 n+ R
R
s m,n≥0 k≥0 s m,n≥0 1 − p q 2 us
The fermionic Fock space is built up from the vacuum and ∂++ m n
∂+− ψ † (with
m, n ≥ 0) acting on the vacuum. This contributes to the index I as
Y Y
m+ 2−R n+ 2−R −Qs
1−p 2 q 2 us (9.19)
s m,n≥0
• Taking into account of the two contributions, the index for the chiral multiplet
with R-charge R is
Y Y 1 − pm+ 2−R
2 q
n+ 2−R
2 u
−Qs Y 1 s
s
Ichiral = s = Γ((pq) 2 R uQ
s ; p, q) . (9.20)
m+ R n+ R
s m,n≥0 1−p 2q 2 uQ
s s
– 55 –
where Γ(z; p, q) is known as the elliptic gamma function, defined as an
infinite product of the form
Y 1 − z −1 pm+1 q n+1
Γ(z; p, q) := mqn
. (9.21)
m,n≥0
1 − zp
• Let us consider the index of a free chiral multiplet, whose R-charge is R = 2/3:
1 the vacuum
ut2/3 the constant mode of φ
u2 t4/3 the square of the constant mode of φ (9.23)
−u−1 t4/3 the lowest mode of ψ †
u(y + y −1 )t5/3 the second mode (derivatives) of φ
• The constant mode of the gauge field A needs to be treated with care. This
mode corresponds to the holonomy of A around S 1 of S 3 × S 1 .
We call the coordinate around S 1 as the Euclidean time and denote by t.
We can take a gauge in which At = diag(At1 , At2 , . . . , AtN ) and keep it time-
independent.
In the end we need to integrate over At , projecting the spectrum to the gauge-
invariant states, but for the moment we keep them to a fixed value.
• Then the contribution from the U (N ) vector multiplet to the index can be
computed to be
" !#
β X
± exp − 2jl + R + µR (2jr ) + Ats Js
ρ s (9.24)
jl +jr + R R
Y
j −j + J
= ±p 2 q l r 2 zs s
where Js is the gauge charge at the s-th diagonal entry and we define
β t
zs = e− ρ As . (9.25)
– 56 –
• The R-charge of the gaugino is fixed at the canonical value +1 all along the
flow.
Modes ∆UV jr jl RUV RIR P0UV δU V I
(9.26)
λ± 3/2 ±1/2 0 1 1 0 0 −p, −q
Note also that the vector multiplet transform in the adjoint representation of
the gauge group.
where the terms in blue denote the contribution from the holonomy of the
gauge field around S 1 (i.e. the constant mode) and
Y
(x; p) = (1 − xpj ) . (9.28)
j≥0
For the SU (N ) vector multiplet, we need remove one contribution from the
diagonal component and zN needs to be eliminated via z1 z2 · · · zN = 1.
SU (N )
Y 1
Ivector = (p; p)N −1 (q; q)N −1 . (9.29)
Γ(za /zb ; p, q) z1 z2 ···zN =1
1≤a6=b≤N
• To compute the index of a gauge theory, we need to extract the gauge invariant
part of the combined contributions from the vector multiplet and the chiral
multiplets. We use the relation
I
dz n
z = δn0 . (9.30)
2πiz
where the factor 1/N ! in front takes into account the remaining Weyl symmetry
of the SU (N ) gauge group.
– 57 –
9.4 The index of SQCD
Let us consider the index of the electric theory: the SU (Nc ) gauge theory with Nf
flavours Q, Q
e with W = 0.
• We introduce the following fugacities for the SU (Nf ) × SU (Nf ) flavour sym-
metry:
(u1 , . . . , uNf ), (v1 , . . . , vNf ), (9.32)
with
Nf Nf
Y Y
ui = vi = 1 . (9.33)
i=1 i=1
ej are respectively ui and v −1 .
The corresponding fugacities for Q and Q i
j
• We also have U (1)B baryonic symmetry under which Q has charge 1 and Q
e
has charge −1. We introduce the fugacity b for it.
• Finally, we use (z1 , . . . , zNc ) with z1 · · · zNc = 1 for the fugacity of SU (Nc )
gauge group.
Nf Nc 0
1 N0 Nc 1 N0
− 2Nc − 2Nc −N c
Y Y
× Γ((pq) 2 f za u−1
i b
Nc0 )Γ((pq) 2 f za−1 vi b N0c ) (9.35)
i=1 a=1
Nf
Y Nc0
× Γ((pq) Nf ui vj−1 ) ,
i,j=1
– 58 –
where the terms in the last line are the contribution from the singlets M .
• The indices of the electric and magnetic theories are equal due to a highly non-
trivial identity, discovered independently by the mathematician Eric Rains (see
Theorem 4.1 of [31]):
INc ,Nf = IN0 c0 ,Nf . (9.36)
This is a very highly non-trivial test for the Seiberg duality.
– 59 –
9.6 Example: SU (2) SQCD with 3 flavours
The indices of the electric and magnetic theories are
I 3 3 Y
1 dz (p; p)(q; q) Y Y 1
±1
Y 1
±1 −1 −1
I2,3 = Q Γ((pq) 6z ui b) Γ((pq) 6z vi b )
2 2πiz ± Γ(z ±2 ) i=1 ± i=1 ±
| {z } | {z }| {z }
SU (2) v-plet Q Q
e
3 Nf Nf
1 1 1
Y Y Y
= 0
I1,3 = Γ((pq) 3 ui vj−1 ) Γ((pq) 3 u−1 2
i b ) Γ((pq) 3 vi b−2 )
i,j=1 i=1 i=1
| {z }| {z }| {z }
M ∗B ∗B
e
(9.38)
Let us write p = ty and q = ty −1 and expand the index as a power series in t. Let
us look at the first non-trivial term, which can be written in terms of the characters
of representations of SU (3) × SU (3) × U (1)B as
h i
0 SU (3) SU (3) 2 SU (3) −2 SU (3)
I2,3 = I1,3 = 1 + χ3 (u)χ3 (v) + b χ3 (u) + b χ3 (v) t2/3 + . . . .
(9.39)
• We observe that it is possible to rewrite the coefficient of t2/3 in terms of SU (6)
representations, due to the branching rule:
SU (6) −→ SU (3) × SU (3) × U (1)
(9.40)
15 −→ (3, 3)(0) ⊕ (3, 1)(2) ⊕ (1, 3)(−2) .
The 15 objects are identified with the elements of the 6×6 antisymmetric matrix
MIJ = ab QaI QbJ (with I, J = 1, . . . , 6). This matrix can be decomposed into
M in (3, 3)(0), ∗B in (3, 1)(2) and ∗B e in (1, 3)(−2).
• Since the components of M (and hence M , ∗B and ∗B) e hit the unitarity bound,
i.e. they carry R-charge 2/3 and hence scaling dimension 1, we expect that they
are free fields.
• In fact, each coefficient in the power series in t can be written in terms of the
characters of SU (6) representations17 :
0 SU (6) SU (6)
I2,3 = I1,3 = 1 + χ15 (a)t2/3 + χ105 (a)t4/3 +
SU (6) SU (6) SU (6)
(9.42)
+ (y + y −1 )χ15 (a)t5/3 + (χ490 − χ35 (a))t2 + . . . .
17
A fugacity map between SU (2N )a and SU (N )u × SU (N )v × U (1)b is as follows. Let us write
u1 = x1 , u2 = x2 x−1 −1 −1 −1 Q
1 , u3 = x3 x2 , . . . , , uN −1 = xN −1 xN −2 , uN = xN −1 (so that ui = 1) and
similarly for vi , namely v1 = y1 , v2 = y2 y1−1 , . . . , vN = yN
−1 Q
−1 (so that v i = 1). Let us denote by
SU (2N )
a1 , . . . , a2N −1 the fugacities for SU (2N )a such that χ2N (a) = a1 + a2 a−1 −1
1 + . . . + a2N −1 and by
b the fugacity for U (1)b . A fugacity map between {ai } and {ui , vi , b} is as follows:
aj = xj bj , aN = bN , aN +j = yN −j bN −j for j = 1, . . . , N − 1. (9.41)
– 60 –
where each term corresponds to the following:
SU (6)
χ15 (a) the constant modes of M
SU (6)
χ105 (a) MI1 I2 MI3 I4 subject to the F -terms I1 ...I6 MI1 I2 MI3 I4 = 0
SU (6)
(y + y −1 )χ15 (a) higher modes of M
SU (6)
χ490 (a) MI1 I2 MI3 I4 MI5 I6 subject to the F -terms (marginal operators)
SU (6)
−χ35 (a) the flavour symmetry current of SU (6)
(9.43)
– 61 –
Here A, B, C = 1, . . . , N 2 − 1 are SU (N ) adjoint indices, and dABC the sym-
metric invariant19 of SU (N ) for N ≥ 3.
• Since dABC ΦA B C
i Φj Φk are exactly marginal operators, the deformation
∆W = h(ijk) dABC ΦA B C
i Φj Φk (9.48)
dimC MC = 1 + 10 − (32 − 1) = 3 .
(9.51)
– 62 –
• For N = 2, the index of the N = 4 SU (2) SYM reads
SU (3) 4 SU (3) 5
I = 1 + χ6 (x)t 3 − χ3 (x)(y + y −1 )t 3
h
SU (3)
i (9.53)
+ 1 − χ8 (x) t2 + · · · ,
SU (3)
where the only difference from the case of N = 3 is that there is no +χ10 (x)
term at order t2 . This is because for N = 2, there is no invariant dABC . Hence
there is only one marginal coupling τ , and the conformal manifold is of
complex dimension 1.
• The magnetic theory has the same gauge group and the same number of
flavours, but there is also 36 singlets M and a non-trivial superpotential, which
fixes the R-charge of M to be 1. The index of the magnetic theory is
" 6 #
1
Y
0
I3,6 = Γ((pq) 2 ui vj−1 ) I(u−1 −1
i b, vi b ) = I3,6 . (9.55)
i,j=1
– 63 –
Each coefficient corresponds to the following objects:
SU (6) SU (6)
χ6 (u)χ6 (v) mesons Mji
SU (6) SU (6)
χ20 (u)b3 + χ20 (v)b−3 baryons and antibaryons B [i1 i2 i3 ] , B
e[i i i ]
1 2 3
SU (6) SU (6) (ij) (i e |a| j) e b
χ21 (u)χ21 (v) marginal operators: O(kl) = Qa Q(k Q|b| Ql)
SU (6) SU (6) [ij] [i e |a| j] e b
χ15 (u)χ15 (v) marginal operators: O[kl] = Qa Q [k Q|b| Ql]
SU (6) SU (6)
(y + y −1 )χ6 (u)χ6 (v) higher modes of the mesons
SU (6) SU (6)
−χ35 (u) − χ35 (v) − 1 conserved current of SU (6) × SU (6) × U (1)B
(9.57)
(ij) [ij]
Note that O(kl) and O[kl] are exactly marginal operators of the theory.
with the pairs (i, j) and (k, l) being both symmetric or anti-symmetric. Here
λkl
ij are exactly marginal couplings. Such deformations preserve conformal
invariance.
• Under such deformations, the U (1)B symmetry is preserved, but the SU (6) ×
SU (6) symmetry is broken.
• In the SU (3) SQCD with 6 flavours, there are (21 × 21) + (15 × 15) = 666
marginal operators. The conformal manifold is therefore dimC MC = 666 −
2(62 − 1) = 596 complex dimensional.
• we have in total [Nf (Nf + 1)/2]2 + [Nf (Nf − 1)/2]2 = Nf2 (Nf2 + 1)/2 marginal
operators; and
10 The a-maximisation
Important points.
– 64 –
• Supersymmetry relates the stress tensor Tµν to a U (1)R current jµR : they are
both in following supermultiplet:
β̇ β̇
Tαα̇ (x, θ, θ) = jαRα̇ (x) + θβ Sαα̇β (x) + θ S αα̇β̇ (x) + θβ θ Tαα̇β β̇ (x) + . . . , (10.1)
where the first component is the U (1)R current such that jαRα̇ = jµR σαµα̇ ; S and
S are supercurrents; and the θθ component is the stress energy tensor such
that Tαα̇β β̇ = Tµν σαµα̇ σβν β̇ .
We call the charge associated with this U (1)R current RT .
• Away from the conformal fixed point, RT is not really conserved. Its non-
conservation is related by supersymmetry to non-zero Tµµ and the lack of scale
invariance.
1. ∆(O) ≥ 23 |R∗ (O)| for all gauge invariant spin zero operators O, with ∆(O)
the exact operator dimension and R∗ (O) the operator’s U (1)R∗ charge.
2. Chiral primary operators have ∆(O) = 23 R∗ (O), and additivity of the
U (1)R charge for composite operators implies that they form a closed
OPE ring, with additive operator dimensions.
3. A unitarity bound, ∆(O) ≥ 1 for spin zero operators, implies that
2
R∗ (O) ≥ (10.2)
3
for spin zero chiral primary operators, with
2
R∗ (O) = (10.3)
3
if and only if O is a decoupled free field.
4. R∗ is anomaly free precisely if the exact NSVZ beta function vanishes,
βNSVZ (g∗ ) = 0, as is appropriate for a CFT.
– 65 –
10.1 The conformal anomalies
• The trace of the stress tensor in 4d conformal field theories is nonzero in
curved spaces:
Tµµ = aE4 − cWµνρσ
2
(10.4)
where E4 = 41 µ1 ν1 µ2 ν2 ρ1 σ1 ρ2 σ2 Rµ1 ν1 ρ1 σ1 Rµ2 ν2 ρ2 σ2 is the Euler density and Wµνρσ
2
3 1
a= (3TrR∗3 − TrR∗ ) , c= (9TrR∗3 − 5TrR∗ ) . (10.6)
32 32
This is extremely powerful, because ’t Hooft anomaly matching implies that
such ’t Hooft anomalies are constant along the RG flows20 and can
thus be evaluated in the weakly coupled UV limit.
Example: The free chiral multiplet Φ. In this case, R(Φ) = 2/3 and the
fermionic component has R-charge R(ψΦ ) = 2/3 − 1 = −1/3.
" #
3
3 1 1 1
a= 3 − − − = ,
32 3 3 48
"
3 # (10.7)
1 1 1 1
c= 9 − −5 − = .
32 3 3 24
Example: The free vector multiplet. The gaugino has R-charge R(λ) = 1.
3 3
3 (1)3 − (1) =
a= ,
32 16 (10.8)
1 1
c= 9 (1)3 − 5 (1) = .
32 8
20
Here we are supposing the the R∗ symmetry is conserved along the RG flow, rather than being
an accidental symmetry of the IR fixed point. Away from the IR fixed point, R∗ differs from the
R-current in the Tαα̇ supermultiplet.
– 66 –
Sketch of the derivation of (10.6). There are two ways to derive these relations.
1. The original way by [38]. This method uses the superspace version of
anomaly of the supercurrent:
α̇ 1 2
∇ Tαα̇ = ∇α (cW − aΞ) , (10.9)
24π 2
where W and Ξ are the superfields that contain the Weyl tensor and Euler
density. Since the superfield Tαα̇ contains also the R-current, by expanding
this superfield equation in components, we can relate a and c with the ’t Hooft
anomalies of the superconformal U (1)R symmetry.
in flat space. The form of this is completely fixed by the symmetries and Ward
identities, up to two overall normalisation coefficients.
In components, this relates the stress tensor three-point functions, and hence a
and c, and to the R-current three-point functions, and hence the tr U (1)R and
tr U (1)3R , to these two coefficients.
The relations (10.6) can then be determined by considering the special cases
of free chiral and vector superfields.
Example: The SU (Nc ) SQCD with Nf flavours, with 23 Nc < Nf < 3Nc .
• As we discussed earlier, this theory has a unique anomaly free U (1)R symmetry
under which R(Qi ) = R(Q ei ) = 1 − (Nc /Nf ).
• The a and c at the strongly coupled interacting RG fixed point can be computed
using the spectrum of the UV free theory.
32
a = 6Nf Nc [R(Q) − 1]3 − 2Nf Nc [R(Q) − 1]
3
+ 3(Nc2 − 1)(1)3 − (Nc2 − 1)(1)
!
1 Nc
= 2Nf2 −3x4 + 2x2 − 2 , with x = ,
Nf Nf
(10.11)
3
32c = 18Nf Nc [R(Q) − 1] − 10Nf Nc [R(Q) − 1]
+ 9(Nc2 − 1)(1)3 − 5(Nc2 − 1)(1)
!
2
= 2Nf2 −9x4 + 7x2 − 2 .
Nf
• The same result can be also from the magnetic theory, as expected from the
Seiberg duality.
– 67 –
The Hofman–Maldacena bounds on a/c [40, 41].
The upper and lower bounds of a/c for unitary 4d CFTs and 4d SCFTs are21
where the upper bound corresponds to free vector fields or vector multiplets, and the
lower bound corresponds to a free scalar; a free chiral multiplet; a free hypermultiplet,
respectively.
11
For 4d N = 2 SCFTs, it can also be shown that c ≥ 30 [44].
• If the theory has a large group F of non-R global flavor symmetries, with
charges FI , we can make a general R-symmetry (also known as the trial R-
symmetry) by combining any initial R-symmetry, R0 , with any linear combi-
nation of the flavour symmetries:
X
Rt = R0 + sI F I , (10.13)
I
• Question: How can we determine the particular values of the sI which produce
the special U (1)R∗ ⊂ SU (2, 2|1)?
• The value of atrial at this local maximum is then the central charge a of the
SCFT.
21
See [42] for various properties of 4d N = 3 SCFTs, including a/c. A class of 4d N = 3 SCFTs
was first constructed by the authors of [43].
– 68 –
• This procedure for finding the R-charge is called the a-maximisation.
• we have
9 tr R∗2 FI = tr FI , (10.16)
i.e. the superconformal U (1)R extremises atrial , and
• we also have
tr R∗ FI FJ < 0 , (10.17)
i.e. the extremum is a local maximum.
• However, hTµν Tσρ JFI i is proportional to the triangle anomaly (JFI · gravity2 ),
which is tr U (1)FI (see [39]).
• Another reason why the anomaly tr R∗2 FI is proportional to the anomaly (JFI ·
gravity2 ) = tr U (1)FI is because the superconformal R-symmetry and the back-
ground metric are in a single supermultiplet W containing the Weyl tensor.
• To compute the ratio between tr R∗2 FI and tr U (1)FI , it suffices to look at the
theory of a free chiral field Φ, where R(Φ) = 2/3 and the R-charge of the
fermion is −1/3. Hence, we arrive at
1
tr R∗2 FI = tr FI , (10.18)
9
as required.
– 69 –
• hT JFI JFJ i is then related to the current-current two point function hJFI JFJ i
(see [39]).
– 70 –
• The a-maximisation “intuitively implies” the a-theorem, at least for some su-
persymmetric theories.
• Consider a general RG flow, from some CFT in the UV to another CFT in the
IR. Often (but not always) the flavour symmetry group FIR of the IR theory is
a subgroup of that of the UV theory FUV , i.e. FIR ⊂ FUV , because the relevant
deformations of the UV CFT usually broke some of the flavour symmetries.
• It then follows from a-maximisation that aIR < aUV , simply because max-
imising over a subset leads to a smaller value.
• Warning. There are examples where FIR 6⊂ FUV , because of accidental sym-
metries. In all such examples, aIR < aUV is still satisfied, but this is a loophole
in the above argument.
10.5 Example: A model with the enhanced E6 flavour symmetry
This example is taken from the work of Razamat, Sela and Zafrir [45].
• Consider a theory with an SU (2) gauge group, 8 fundamental chiral multiplets
and a collection of singlets.
We split the 8 chiral fields into 6 chiral fields QA and 2 chiral fields QB , and
couple them respectively to the singlets MA and MB with the superpotential
W = αβ (MA )ij (QA )iα (QA )jβ + αβ (MB )uv (QB )αu (QB )βv , (10.24)
where α, β = 1, 2 are gauge indices, i, j, k = 1, . . . , 6 and u, v = 1, 2. Here MA
is a 6 × 6 antisymmetric matrix (15 independent components) and MB is a
2 × 2 antisymmetic matrix (1 independent component). In the literature, MA
and MB are referred to as the flipping fields.
QA QB
6 × 2 × 2 (10.25)
MA MB
This splitting breaks the original SU (8) symmetry to SU (6)A ×SU (2)B ×U (1)h ,
with the following branching rule:
8 −→ (6, 1)(1) + (1, 2)(−3) (10.26)
We will shortly see that this flavour symmetry gets enhanced to E6 × U (1)h in
the infrared.
• According to the above branching rule, the chiral fields carry the following
charges under the U (1)h symmetry:
h(QA ) = 1/2 , h(QB ) = −3/2 , h(MA ) = −1 , h(MB ) = 3 . (10.27)
and the superpotential is neutral under U (1)h as it should be.
– 71 –
• The U (1)h can mix with the R-symmetry. To determine the correct R-symmetry
we perform the a-maximisation.
Let the trial R-charges of the chiral fields to be
R(QA ) = 32 + x , R(QB ) = 23 + y ,
(10.28)
R(MA ) = 32 − 2x , R(MB ) = 23 − 2y .
where the superpotential have R-charge 2. From the NSVZ beta function, the
condition for a non-trivial fixed point (i.e. non-anomalous R-symmetry) reads
X
0 = T (adj) + T (ri )(Ri − 1)
i
1 1 (10.29)
= 2 + 6 × (R(QA ) − 1) + 2 × (R(QB ) − 1)
2 2
2
⇔ y = −3x − ,
3
and so
R(QA ) = x + 23 , R(QB ) = −3x ,
2 (10.30)
R(MA ) = −2x + 3
, R(MB ) = 6x + 2 .
Observe that (half of ) the coefficients of x in the above equations are
the non-anomalous U (1)h charges of each chiral field.
• We have
tr R3 = 12(R(QA ) − 1)3 + 4(R(QB ) − 1)3
+ 15(R(MA ) − 1)3 + 1(R(MB ) − 1)3 + 3(1)3
(10.31)
tr R = 12(R(QA ) − 1) + 4(R(QB ) − 1)
+ 15(R(MA ) − 1) + 1(R(MB ) − 1) + 3(1) .
– 72 –
• Let us compute the supersymmetric index on S 3 × S 1 :
I 6 2 Y
1 dz (p; p)(q; q) Y Y 5
Y 1
I= Q ±2
±1
Γ((pq) z xi h )
18
1/2
Γ((pq) 6 z ±1 vj−1 h−3/2 )
2 2πiz ± Γ(z ) i=1 ± j=1 ±
4 4
Y
× Γ((pq) 6 v1 v2 h3 ) Γ((pq) 9 x−1 −1 −1
i xj h ) ,
1≤i<j≤6
(10.35)
Q
where x1 , x2 , . . . , x6 (with i xi = 1) are the fugacities for SU (6)A , and v1 , v2
(with v1 v2 = 1) are the fugacities for SU (2)B , and h is the fugacity for U (1)h .
Let us write p = ty and q = t/y and expand the index as a power series in t:
h i
SU (6) SU (6)A SU (2)B
I = 1 + h−1 χ15 A (x) + χ6 (x)χ2 (v) t8/9 + h
|
−1 4/3
{zt } + . . . . (10.36)
| {z } | {z }
QA QB M B
MA
I = 1 + h−1 χ27
E6
(a)t8/9 + h−1 t4/3 + h−2 χ351
E6
(a)t16/9
(10.38)
+ h−1 χE (a)(y + y −1 )t17/9 + −χ78
E6
(a) − 1 t2 + . . . .
6
27
• One situation where accidental symmetries are readily apparent, and required,
is when a gauge invariant chiral primary operator, e.g. M = QQ, e hits or
appears to violate the unitarity bound R(M ) ≥ 2/3.
– 73 –
– The trial Rt must then include mixing with JM :
• Let us take the superpotential to be zero for now. Suppose that this theory flows
to a fixed point. The condition for the non-anomalous R-symmetry (vanishing
NSVZ beta function) is
1
0 = 2 + 2 × (R(Q) − 1) + 2(R(φ) − 1) . (10.40)
2
Its convenient to write
2 2
R(Q) = +x , R(φ) = +y , (10.41)
3 3
then (10.40) implies that
1
y = − (1 + x) . (10.42)
2
The trial a-function is then
3 1
atrial = (3 tr R3 − tr R) = 783x3 − 1269x2 − 567x + 301 .
(10.43)
32 768
where
tr R3 = 3(R(φ) − 1)3 + 4(R(Q) − 1)3 + 3
(10.44)
tr R = 3(R(φ) − 1) + 4(R(Q) − 1) + 3 .
1
√
The function atrial reaches its local maximum at x = 87 47 − 2 1009 . This
implies that
1 √
R(φ) = 1009 − 9 ≈ 0.26 . (10.45)
87
22
An N = 1 Lagrangian that flows to an N = 2 Argyres–Douglas theory was in fact first
discovered by Maruyoshi–Song in [47]. In this work the corresponding Argyres–Douglas theory is
of the type H0 or (A1 , A2 ).
– 74 –
Observe that the R-charge of the gauge invariant operator U = tr(φ2 ) is
2 √ 2
R(U ) = 1009 − 9 ≈ 0.52 < . (10.46)
87 3
This violates the unitarity bound.
and so
3 1
(3 tr R3 − tr R) = 999a3 − 837a2 − 351a + 317 . (10.50)
atrial =
32 768
1
√ 1
√
Maximising atrial gives x = 111 31 − 2 601 . This leads to R(φ) = 111 601 + 3 ≈
2
√
− 601 + 108 ≈ 1.50 > 23 .
0.25, and so R(FU ) = 111
In fact, it could be checked, for example by computing the index, that there is
no unitarity violating operator.
W = U FU + M FM . (10.51)
– Again, the vanishing NSVZ beta function condition (10.40) remains un-
changed, since FU and FM are gauge singlets. Due to the superpotential
2
R(FM ) = 2 − 2R(Q) = − 2x . (10.52)
3
– 75 –
We summarise the trial R-charges of the chiral fields as follows:
2 1 2 5
(R(Q), R(φ), R(FM ), R(FU )) = + 2x, − x, − 4x, + 2x .
3 6 3 3
(10.53)
The coefficients of x indicate the non-anomalous U (1) flavour
(non-R) charges that can be assigned to each chiral field. We will
refer to this as U (1)h .
– We have
tr R3 = 3(R(φ) − 1)3 + 4(R(Q) − 1)3 + (R(FU ) − 1)3 + (R(FM ) − 1)3 + 3
tr R = 3(R(φ) − 1) + 4(R(Q) − 1) + (R(FU ) − 1) + (R(FM ) − 1) + 3 .
(10.54)
and so
3 3
(3 tr R3 − tr R) = − 648x3 + 756x2 + 78x − 37 . (10.55)
atrial =
32 256
1
This reaches its maximum at x = − 18 , and at this values the R-charge of
the chiral fields are
5 2 8 14
(R(Q), R(φ), R(FM ), R(FU )) = , , , . (10.56)
9 9 9 9
– 76 –
• Let h be the fugacity for the non-anomalous U (1) flavour symmetry, and let
w1 , w2 (with w1 w2 = 1) be the fugacities for the SU (2) flavour symmetry. The
index for the theory (10.51) is given by
I 2
1 dz (p; p)(q; q) Y Y 5
±1
2
I= Q Γ((pq) z wi h )
9
2 2πiz ± Γ(z ±2 ) i=1 ±
w2 =w1−1
2
Q2
±1 −1 −1
Q
± Γ((pq) z wi wj h )
9 8 14
i,j=1
× 2 × Γ((pq) 9 h−4 )Γ((pq) 9 h2 )
−1
Γ((pq) 9 h ) −1
w2 =w1
SU (2)
= 1 + h−4 t8/9 − (y + y −1 )h−1 t11/9 + h3 χ3 (w)t4/3
SU (2)
+ . . . + (−χ3 (w) − 1)t2 + . . . ,
(10.59)
• From the NSVZ beta function (8.49), a condition for non-trivial fixed points of
the RG flow reads X
0 = T (adj) + T (ri )(Ri − 1) . (10.60)
i
0 = Nc + Nf (R(Q) − 1) + Nc (R(X) − 1)
(10.61)
⇔ Nf = Nf R(Q) + Nc R(X) .
R(Q) ≡ y ,
1−y Nc (10.62)
R(X) = , with x ≡ .
x Nf
– 77 –
• To simplify the calculation, we focus on the limit in which Nc 1, Nf 1
with x = Nc /Nf fixed. From (10.14), we obtain
atrial
= 6x(y − 1)3 − 2x(y − 1)
b
Nf2 | {z }
contr. from Q and Q
e
3
2 1−y 2 1−y
+ 3x −1 −x −1 (10.64)
x x
| {z }
contr. from X
2 2
+ 3x − x .
| {z }
contr. from gauginos
Plugging (10.65) into (10.64), one finds the following expression for the central
charge:
(0)
2Nf2 x2 2 2 3/2
a ≡b
atrial |y=y(0) = 18 − 90x + (20x − 1) . (10.66)
9(1 − 2x2 )2
b
• Actually, (10.65), (10.66) are not the full story, since one needs to take into
account unitarity constraints.
– 78 –
– In general, we have gauge invariant quantities Mj = QXe j−1 Q, with j =
1, 2, . . .. At some larger value of x, the operator M2 = QXQ
e reaches
R-charge 2/3, and the same procedure has to be repeated for it. More
generally, every time the R-charge of a gauge invariant chiral operator
in the theory drops below 2/3, the a-maximisation procedure has to be
modified accordingly.
– The required a-maximizations can be solved numerically (see more details
in [52]). However, this problem can be solved analytically for large x.
where
1−y
R(Mj ) = 2y + (j − 1) . (10.71)
x
Let us estimate p. We do this by requiring that R(Mp ) = 23 . The error that
arises from this is subleading in 1/x. Hence,
1−y 2 2 1 − 3y
2y + (p − 1) = ⇒ p∼ x. (10.72)
x 3 3 1−y
The second term (10.70) can then be estimated by the following integral:
Z 2(1−3y) 2
1 2 3(1−y) 1−y 1−y
N ds 2 − 3 2y + s 5 − 3 2y + s
9 f s=0 x x (10.73)
4 2
∼ Nf (1 − 3y)3 x .
9
Thus, the trial central charge is
2 3 4 3
atrial = Nf 6x(y − 1) − 10x(y − 1) + (1 − 3y) x .
b (10.74)
9
– 79 –
The local maximum of b
atrial with respect to y occurs at
√
3−1
y= , (10.75)
3
and the value of the central charge at this point is
4 √
a = (2 + 3)xNf2 .
b (10.76)
3
Appendix
A Spinors in 3 + 1 dimensions
• We will mainly use the two component spinor formalism in d = 3 + 1.
• Each Dirac spinor requires one left- and one right-handed Weyl spinor:
ψα 0
Ψ = Ψ L + ΨR ; ΨL = , ΨR = . (A.2)
0 η α̇
where
(σ µ )α̇α = {1, −σ}α̇,α , (σ µ )αβ̇ = {1, σ}α,β̇ . (A.4)
This is equivalent to writing
– 80 –
References
[1] K. A. Intriligator and B. Wecht, “The Exact superconformal R symmetry maximizes
a,” Nucl. Phys. B667 (2003) 183–200, arXiv:hep-th/0304128 [hep-th].
[2] K. A. Intriligator and N. Seiberg, “Lectures on supersymmetric gauge theories and
electric-magnetic duality,” Nucl. Phys. Proc. Suppl. 45BC (1996) 1–28,
arXiv:hep-th/9509066 [hep-th]. [,157(1995)].
[3] P. Argyres, “An Introduction to Global Supersymmetry,” 1996.
http://homepages.uc.edu/~argyrepc/cu661-gr-SUSY/susy1996.pdf.
[4] M. E. Peskin, “Duality in supersymmetric Yang-Mills theory,” in Fields, strings and
duality. Proceedings, Summer School, Theoretical Advanced Study Institute in
Elementary Particle Physics, TASI’96, Boulder, USA, June 2-28, 1996,
pp. 729–809. 1997. arXiv:hep-th/9702094 [hep-th]. http://www-public.slac.
stanford.edu/sciDoc/docMeta.aspx?slacPubNumber=SLAC-PUB-7393.
[5] P. Argyres, “An Introduction to Global Supersymmetry,” 2001.
http://homepages.uc.edu/~argyrepc/cu661-gr-SUSY/susy2001.pdf.
[6] M. J. Strassler, “An Unorthodox introduction to supersymmetric gauge theory,” in
Strings, Branes and Extra Dimensions: TASI 2001: Proceedings, pp. 561–638. 2003.
arXiv:hep-th/0309149 [hep-th].
[7] J. Terning, Modern supersymmetry: Dynamics and duality. 2006.
[8] M. Shifman, Advanced topics in quantum field theory. Cambridge Univ. Press,
Cambridge, UK, 2012. http://www.cambridge.org/mw/academic/subjects/
physics/theoretical-physics-and-mathematical-physics/
advanced-topics-quantum-field-theory-lecture-course?format=AR.
[9] Y. Tachikawa, “Lectures on 4d N =1 dynamics and related topics,” 2018.
arXiv:1812.08946 [hep-th].
[10] J. Wess and J. Bagger, Supersymmetry and supergravity. Princeton University Press,
Princeton, NJ, USA, 1992.
[11] M. T. Grisaru, W. Siegel, and M. Rocek, “Improved Methods for Supergraphs,”
Nucl. Phys. B159 (1979) 429.
[12] N. Seiberg, “Naturalness versus supersymmetric nonrenormalization theorems,”
Phys. Lett. B318 (1993) 469–475, arXiv:hep-ph/9309335 [hep-ph].
[13] D. Green, Z. Komargodski, N. Seiberg, Y. Tachikawa, and B. Wecht, “Exactly
Marginal Deformations and Global Symmetries,” JHEP 06 (2010) 106,
arXiv:1005.3546 [hep-th].
[14] E. Barnes, E. Gorbatov, K. A. Intriligator, M. Sudano, and J. Wright, “The Exact
superconformal R-symmetry minimizes tau(RR),” Nucl. Phys. B730 (2005)
210–222, arXiv:hep-th/0507137 [hep-th].
– 81 –
[15] J.-F. Fortin, K. Intriligator, and A. Stergiou, “Current OPEs in Superconformal
Theories,” JHEP 09 (2011) 071, arXiv:1107.1721 [hep-th].
[16] I. Affleck, M. Dine, and N. Seiberg, “Dynamical Supersymmetry Breaking in
Supersymmetric QCD,” Nucl. Phys. B241 (1984) 493–534.
[17] N. Dorey, T. J. Hollowood, V. V. Khoze, and M. P. Mattis, “The Calculus of many
instantons,” Phys. Rept. 371 (2002) 231–459, arXiv:hep-th/0206063 [hep-th].
[18] M. Bianchi, S. Kovacs, and G. Rossi, “Instantons and Supersymmetry,” Lect. Notes
Phys. 737 (2008) 303–470, arXiv:hep-th/0703142 [HEP-TH].
[19] O. Aharony, N. Seiberg, and Y. Tachikawa, “Reading between the lines of
four-dimensional gauge theories,” JHEP 08 (2013) 115, arXiv:1305.0318
[hep-th].
[20] F. Cachazo, M. R. Douglas, N. Seiberg, and E. Witten, “Chiral rings and anomalies
in supersymmetric gauge theory,” JHEP 0212 (2002) 071, arXiv:hep-th/0211170
[hep-th].
[21] E. Witten, “Constraints on Supersymmetry Breaking,” Nucl. Phys. B202 (1982)
253.
[22] E. Witten, “Toroidal Compactification Without Vector Structure,” JHEP 02 (1998)
006, arXiv:hep-th/9712028.
[23] V. A. Novikov, M. A. Shifman, A. I. Vainshtein, and V. I. Zakharov, “Exact
Gell-Mann-Low Function of Supersymmetric Yang-Mills Theories from Instanton
Calculus,” Nucl. Phys. B229 (1983) 381–393.
[24] N. Arkani-Hamed and H. Murayama, “Holomorphy, rescaling anomalies and exact
beta functions in supersymmetric gauge theories,” JHEP 06 (2000) 030,
arXiv:hep-th/9707133 [hep-th].
[25] M. A. Shifman and A. I. Vainshtein, “Solution of the Anomaly Puzzle in SUSY
Gauge Theories and the Wilson Operator Expansion,” Nucl. Phys. B277 (1986)
456. [Zh. Eksp. Teor. Fiz.91,723(1986)].
[26] W. E. Caswell, “Asymptotic Behavior of Nonabelian Gauge Theories to Two Loop
Order,” Phys. Rev. Lett. 33 (1974) 244.
[27] T. Banks and A. Zaks, “On the Phase Structure of Vector-Like Gauge Theories with
Massless Fermions,” Nucl. Phys. B196 (1982) 189–204.
[28] N. Seiberg, “Electric - magnetic duality in supersymmetric nonAbelian gauge
theories,” Nucl. Phys. B435 (1995) 129–146, arXiv:hep-th/9411149 [hep-th].
[29] I. Affleck, M. Dine, and N. Seiberg, “Supersymmetry Breaking by Instantons,” Phys.
Rev. Lett. 51 (1983) 1026.
[30] G. Festuccia and N. Seiberg, “Rigid Supersymmetric Theories in Curved
Superspace,” JHEP 06 (2011) 114, arXiv:1105.0689 [hep-th].
[31] E. M. Rains, “Transformations of elliptic hypergometric integrals,” 2003.
– 82 –
[32] C. Beem and A. Gadde, “The N = 1 superconformal index for class S fixed points,”
JHEP 04 (2014) 036, arXiv:1212.1467 [hep-th].
[33] B. Kol, “On conformal deformations,” JHEP 09 (2002) 046, arXiv:hep-th/0205141
[hep-th].
[34] S. Benvenuti and A. Hanany, “Conformal manifolds for the conifold and other toric
field theories,” JHEP 08 (2005) 024, arXiv:hep-th/0502043 [hep-th].
[35] B. Kol, “On Conformal Deformations II,” arXiv:1005.4408 [hep-th].
[36] R. G. Leigh and M. J. Strassler, “Exactly marginal operators and duality in
four-dimensional N=1 supersymmetric gauge theory,” Nucl. Phys. B447 (1995)
95–136, arXiv:hep-th/9503121 [hep-th].
[37] Z. Komargodski and A. Schwimmer, “On Renormalization Group Flows in Four
Dimensions,” JHEP 12 (2011) 099, arXiv:1107.3987 [hep-th].
[38] D. Anselmi, D. Z. Freedman, M. T. Grisaru, and A. A. Johansen, “Nonperturbative
formulas for central functions of supersymmetric gauge theories,” Nucl. Phys. B526
(1998) 543–571, arXiv:hep-th/9708042 [hep-th].
[39] H. Osborn, “N=1 superconformal symmetry in four-dimensional quantum field
theory,” Annals Phys. 272 (1999) 243–294, arXiv:hep-th/9808041 [hep-th].
[40] D. M. Hofman and J. Maldacena, “Conformal collider physics: Energy and charge
correlations,” JHEP 05 (2008) 012, arXiv:0803.1467 [hep-th].
[41] D. M. Hofman, D. Li, D. Meltzer, D. Poland, and F. Rejon-Barrera, “A Proof of the
Conformal Collider Bounds,” JHEP 06 (2016) 111, arXiv:1603.03771 [hep-th].
[42] O. Aharony and M. Evtikhiev, “On four dimensional N = 3 superconformal
theories,” JHEP 04 (2016) 040, arXiv:1512.03524 [hep-th].
[43] I. Garca-Etxebarria and D. Regalado, “N = 3 four dimensional field theories,”
JHEP 03 (2016) 083, arXiv:1512.06434 [hep-th].
[44] P. Liendo, I. Ramirez, and J. Seo, “Stress-tensor OPE in N = 2 superconformal
theories,” JHEP 02 (2016) 019, arXiv:1509.00033 [hep-th].
[45] S. S. Razamat, O. Sela, and G. Zafrir, “Between Symmetry and Duality in
Supersymmetric Quantum Field Theories,” Phys. Rev. Lett. 120 no. 7, (2018)
071604, arXiv:1711.02789 [hep-th].
[46] K. Maruyoshi and J. Song, “N = 1 deformations and RG flows of N = 2 SCFTs,”
JHEP 02 (2017) 075, arXiv:1607.04281 [hep-th].
[47] K. Maruyoshi and J. Song, “Enhancement of Supersymmetry via Renormalization
Group Flow and the Superconformal Index,” Phys. Rev. Lett. 118 no. 15, (2017)
151602, arXiv:1606.05632 [hep-th].
[48] S. Benvenuti and S. Giacomelli, “Supersymmetric gauge theories with decoupled
operators and chiral ring stability,” Phys. Rev. Lett. 119 no. 25, (2017) 251601,
arXiv:1706.02225 [hep-th].
– 83 –
[49] O. Aharony and Y. Tachikawa, “A Holographic computation of the central charges
of d=4, N=2 SCFTs,” JHEP 01 (2008) 037, arXiv:0711.4532 [hep-th].
[50] A. D. Shapere and Y. Tachikawa, “Central charges of N=2 superconformal field
theories in four dimensions,” JHEP 09 (2008) 109, arXiv:0804.1957 [hep-th].
[51] P. C. Argyres, M. R. Plesser, N. Seiberg, and E. Witten, “New N=2 superconformal
field theories in four-dimensions,” Nucl. Phys. B461 (1996) 71–84,
arXiv:hep-th/9511154 [hep-th].
[52] D. Kutasov, A. Parnachev, and D. A. Sahakyan, “Central charges and U(1)(R)
symmetries in N=1 superYang-Mills,” JHEP 11 (2003) 013, arXiv:hep-th/0308071
[hep-th].
– 84 –