KIAS-P22052

The shape of non-graviton operators for SU (2)

arXiv:2209.12696v4 [hep-th] 1 Jul 2024

Sunjin Choi1, Seok Kim2, Eunwoo Lee2 and Jaemo Park3

1
School of Physics, Korea Institute for Advanced Study,
85 Hoegi-ro, Dongdaemun-gu, Seoul 02455, Republic of Korea
2
Department of Physics and Astronomy & Center for Theoretical Physics,
Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul 08826, Republic of Korea
3
Department of Physics, Pohang University of Science and Technology (POSTECH),
77 Cheongam-ro, Nam-gu, Pohang 37673, Gyeongsangbuk-do, Republic of Korea

E-mails: [email protected], [email protected]

[email protected], [email protected]

Abstract

The BPS spectrum of AdS/CFT exhibits multi-gravitons at low energies, while having
black hole states at higher energies. This can be studied concretely in AdS5 /CFT4 in terms
of classical cohomologies, even in the quantum regimes at finite 1/N . Recently, Chang
and Lin found a threshold for non-graviton states in the SU (2) maximal super-Yang-Mills
theory. We explicitly construct and present this threshold cohomology.
1 Introduction and the setup

Regular BPS black holes in AdS5 × S 5 preserve 2 real Killing spinors [1], and should be dual
1
to the 16 -BPS states of the 4d U(N) maximal super-Yang-Mills theory on S 3 × R. By the
1
operator-state map of a CFT, these correspond to the 16 -BPS local operators on R4 . These
operators are defined and constructed as follows (we follow the conventions of [2, 3]). The U(N)
adjoint fields of the super-Yang-Mills theory are given by

6 real scalars : Φij = 21 ǫijkl Φkl where Φkl ≡ (Φkl )∗

i
fermions : Ψiα , Ψα̇
gauge fields : Aµ ∼ Aαβ̇ (1.1)

where i, j · · · = 1, · · · , 4 are the SU(4) R-symmetry indices, and α = ±, α̇ = ±̇ are the SU(2)l
and SU(2)r indices in SO(4) ∼ = SU(2)l × SU(2)r rotation symmetry, respectively. The system
has 16 Poincare supersymmetries Qiα and Qiα̇ . It also has 16 conformal supersymmetries which,
iα̇
in the radially quantized setup, are Hermitian conjugate to Q’s: Siα = (Qiα )† , S = (Qiα̇ )† .
1
The 16 -BPS operators are annihilated by Q ≡ Q4− and S ≡ S4− ≡ Q† . Q and S satisfy

2{Q, S} = E − (R1 + R2 + R3 + J1 + J2 ) ≥ 0 , (1.2)

where R1,2,3 are the Cartans of SO(6) ∼ = SU(4), and J1,2 are the Cartans of SO(4). The BPS
operators are made of the following set of fields and derivatives, which individually saturate
the last inequality of (1.2) (in the free theory limit):
4
φ̄m ≡ Φ4m , ψm ≡ Ψm+ , λ̄α̇ ≡ Ψα̇ , f ≡ F++ , D+α̇ . (1.3)

F++ is a component of the field strength Fαβ ∼ Fµν (σ µν )αβ , and D+α̇ are two components of
the covariant derivatives Dαβ̇ ∼ (σ µ )αβ̇ Dµ . The charges carried by these fields and derivatives
are given as follows. (RI charges of other fields φ̄m , ψm are obtained by obvious permutations.)

fields (R1 , R2 , R3 ) (J1 , J2 ) E

φ̄1 (1,0,0) (0,0) 1
ψ1 (− 12 , 12 , 21 ) 1 1
(2, 2) 3
2
λ̄±̇ ( 21 , 21 , 12 ) (± 12 , ∓ 12 ) 3
2
f (0,0,0) (1, 1) 2
D++̇ , D+−̇ (0, 0, 0) (1,0), (0,1) 1

All gauge invariant operators made of (1.3) saturate the inequality of (1.2) at strictly zero
coupling, thus providing the BPS operators of the free Yang-Mills theory. At nonzero cou-
pling, the composite operators typically acquire nonzero anomalous dimensions and many BPS
operators in the free theory become non-BPS. Questions were raised in [4, 5, 6] concerning
which BPS operators in the free theory remain BPS in the ‘weakly interacting’ regime, and
whether they can account for the strongly coupled BPS spectrum captured by the dual black

1
hole entropies. At 1-loop level, the first part of the question can be rephrased as a classical
cohomology problem in Q, which satisfies Q2 = 0. Q-actions on the BPS letters are given by

[Q, φ̄m ] = 0 , {Q, ψm } = −igYM ǫmnp [φ̄n , φ̄p ] ,

{Q, λ̄α̇ } = 0 , [Q, f ] = −igYM [ψm , φ̄m ] , [Q, D+α̇ ] = −igYM [λ̄α̇ , } . (1.4)

The second part of the question, whether the BPS operators at 1-loop level remain all the way
to the strong coupling, remains a conjecture without known evidence against it. See [7] for a
non-renormalization theorem on this issue.

With the last conjecture assumed, we expect the following structures of the large N spectrum
of cohomologies, from the AdS dual. At energies much lower than N, the spectrum is expected
to be saturated by supergravitons. This has been tested in [8, 9]. At energies of order N,
we expect less operators than gravitons because of the ‘giant graviton’ effects [10]. This is
implemented in QFT by the trace relations of finite matrices. These giant graviton states,
at least those explicitly constructed so far, are still of ‘graviton type’ (whose meaning will be
stated below). Finally, beyond certain energy level (whose threshold value is not clearly known),
we expect to have new cohomologies not of the graviton type. We expect new cohomologies
because some of them should account for the microstates of AdS black holes at energies of order
N 2 . In particular, the index [4] of this QFT exhibits a large number of operators to account
for the Bekenstein-Hawking entropy of the dual black holes. A selection of studies which gave
increasing confidence about this fact is (among many): Euclidean quantum gravity analysis
[11]; the saddle point analysis [12, 13]; the Bethe ansatz analysis [14]; explicit enumeration of
the index at finite but reasonably large N and charges [15, 16].

The graviton type BPS cohomologies are characterized as follows [3]. We first consider
the single trace operators made of scalars φ̄m ≡ (X, Y, Z). Since [φ̄m , φ̄n ] are Q-exact from
(1.4), the scalars are regarded as commuting operators inside the trace, e.g. tr(X 2 Y Z) ∼
tr(XY XZ) ∼ tr(X 2 ZY ). Then, one acts the 9 fermionic elements of SU(1, 2|3) commuting
with Q, S to obtain certain descendants [4]. Then one further acts ordinary derivatives ∂+α̇
on these single trace operators to obtain conformal descendants. All these procedures generate
new Q-cohomologies. Finally, multiplying these single-trace cohomologies generates multi-trace
cohomologies. At large N where one can ignore the trace relations, these account for all the
supergravitons. At finite N, we shall still call them ‘graviton-like’ cohomologies as they do not
give qualitatively new operators. At finite N, their growths are slower than those of large N
gravitons due to the trace relations of multi-trace operators.

Understanding where the non-graviton operators appear and how they look like have been
quite long standing problems. In particular, since the graviton-type operators can be defined
precisely at finite N, one can study this problem from the simplest toy model of SU(2) theory.
Recently, [7] made a very promising discovery on this problem. They reported the threshold
level of the SU(2) theory, from which such new operators exist. (They reported 7 non-graviton

2
states, but 6 of them are descendants of the threshold state.) The goal of this short note is
to explicitly construct representatives of such a cohomology and take a detailed look at their
structures.1 As we shall see below, one can find a simple representative of the cohomology.
It is our belief that the detailed shape of this operator will inspire us towards better analytic
studies, e.g. ansatz for new cohomologies at higher N, etc.

2 Results and discussions

For convenience, we consider the SU(2) super-Yang-Mills theory rather than U(2). The two
theories are trivially related by multiplying the free U(1) theory. In the SU(2) theory, [7]
reported that the first non-graviton type cohomology appears at n ≡ 2 3I=1 RI + 3 2i=1 Ji =
P P

24, at charges E = 19 2
, R1 = R2 = R3 = 32 , J1 = J2 = 25 . There is a unique operator in this
sector, which is not of graviton type. Furthermore, it was found that this operator is made of
7 fields. Using these data, one can easily see from the charge table that only φ̄m , ψm , f can
participate in the cohomology without any derivatives. Also, note from (1.4) that these three
operators form a closed subsector under the Q-action. Even when checking the Q-exactness of
an operator in this sector, Q should act on another operator made only of φ̄m , ψm , f to be in
this sector. Thus, we will not consider λ̄α̇ and D+α̇ from now on. By inspecting the charges,
the operator may contain terms at f 0 ψ 5 φ2 , f 1 ψ 3 φ3 and f 2 ψ 1 φ4 orders. One can further specify
which ψ’s and φ̄’s can appear in which combinations: this is helpful when listing all possible
operators, but we skip explaining this.

For the SU(2) theory,nthe adjointo fields can be represented by three-dimensional vectors
σa=1,2,3
using the Pauli matrices 2
as the basis. Our vector notation of the BPS letters (1.3)
are given as follows:
√ √
gYM 2φ̄m → φ ~ m , gYM ψm → ψ ~m , −gYM 2f → f~ . (2.1)

The Q-actions on the BPS letters (1.4) in the vector notation are written as

Qφ ~ m = 0 , Qψ ~m = 1 ǫmnp φ~n × φ
~ p , Qf~ = φ
~m × ψ
~m . (2.2)
2
For simplicity, we will drop ~ symbol from all vectors from now on. Also, we shall sometimes
use the notation φm = (φ1 , φ2, φ3 ) ≡ (X, Y, Z).

Writing down a cohomology is ambiguous by adding Q-exact terms, and here we write down
a simple representative. The unique cohomology in this sector may be represented as

O ≡ (ψ1 · X − ψ2 · Y )(ψ3 · X)ψ2 · (ψ1 × ψ1 ) + cyclic (2.3)

≡ (ψ1 · X − ψ2 · Y )(ψ3 · X)ψ2 · (ψ1 × ψ1 ) + (ψ2 · Y − ψ3 · Z)(ψ1 · Y )ψ3 · (ψ2 × ψ2 )
+(ψ3 · Z − ψ1 · X)(ψ2 · Z)ψ1 · (ψ3 × ψ3 ) ,
1
Although not reported in [7], we believe that their raw data should also contain all the information.

3
where ‘cyclic’ means adding two more terms obtained by making cyclic permutations of (X, ψ1 ),
(Y, ψ2 ) and (Z, ψ3 ). The Q-closedness of this operator is easy to check. First note that ψ3 · X
and ψ1 · X − ψ2 · Y are separately Q-closed. In fact, they are graviton-type operators, for
instance obtained by

ψ3 · X ∼ [Q2+ , X · X] ∼ [Q1+ , X · Y ] , ψ1 · X − ψ2 · Y ∼ [Q3+ , X · Y ] (2.4)

using the SU(1, 2|3) generators. The Q-action on ψ2 · (ψ1 × ψ1 ) is given by

{Q, ψ2 · (ψ1 × ψ1 )} = 2(Z · ψ1 )(X · ψ1 ) + 2(ψ1 · Y )(ψ2 · Z) − 2(ψ1 · Z)(ψ2 · Y ) . (2.5)

So the Q-action on the whole operator becomes

{Q, O} = −2(2Y )(3X)(1Z)(1X) + 2(1X)(3X)(1Y )(2Z) − 2(2Y )(3X)(1Y )(2Z)

−2(1X)(3X)(1Z)(2Z) + cyclic
= 2(1X)(1Z)(2Y )(3X) + 2(1X)(1Y )(2Z)(3X) − 2(1X)(1Y )(2Z)(3X)
−2(1X)(1Z)(2Z)(3X) + cyclic = 0 , (2.6)

where (1X) ≡ (ψ1 · X), etc., and the Fermi statistics forbids various terms like (ψ1 · X)2 or
(ψ2 · Y )2 . On the third line, we rearranged terms using the cyclic permutation.

The fact that this is not Q-exact is harder to check. We comprehensively studied all inde-
pendent Q-exact operators in this sector, and have shown that O cannot be a linear combination
of such operators. To explain this, one should first write down all operators whose Q-actions
yield operators at the desired charges. Since we want Q-exact operators made of 7 letters, we
should consider Q-action on gauge-invariant operators made of 6 fields among f, ψm , φm . One
can always make 6 vectors gauge-invariant by forming 3 inner products. (For instance, one may
use an operator of the form A · (B × C) D · (E × F ), but this can be rewritten as three inner
products.) Since we expect to find a unique cohomology [7], we can stay in the sector invariant
under the cyclic permutation symmetry, which is a symmetry of the Q-action. So the Q-exact
operators can also be restricted to cyclic invariant ones. There exist 17 independent Q-exact
and cyclic-invariant operators in this sector. They can be represented by the Q-action on the
′
following 17 independent operators O1,2,3,··· ,17 :

O1′ = (f · f )(f · X)(Y · Z) + cyclic ,

O2′ = (f · f )(X · Y )(ψ1 · ψ2 ) + cyclic ,
O3′ = (f · f )(X · ψ1 )(Y · ψ2 ) + cyclic ,
O4′ = (f · X)(f · Y )(ψ1 · ψ2 ) + cyclic ,
O5′ = (f · ψ1 )(f · ψ2 )(X · Y ) + cyclic ,
O6′ = (f · X)(f · ψ1 )(Y · ψ2 ) + cyclic ,
O7′ = (f · Y )(f · ψ1 )(X · ψ2 ) + cyclic ,

4
 O8′ = (f · X)(f · ψ2 )(Y · ψ1 ) + cyclic ,
O9′ = (f · Y )(f · ψ2 )(X · ψ1 ) + cyclic ,
′
O10 = (f · f )(X · ψ2 )(Y · ψ1 ) + cyclic ,
′
O11 = (f · X)(ψ1 · ψ2 )(ψ1 · ψ3 ) + cyclic ,
′
O12 = (f · ψ2 )(ψ1 · X)(ψ1 · ψ3 ) + cyclic ,
′
O13 = (f · ψ3 )(ψ1 · ψ2 )(ψ1 · X) + cyclic ,
′
O14 = (f · ψ1 )(X · ψ2 )(ψ1 · ψ3 ) + cyclic ,
′
O15 = (f · X)(f · ψ1 )(X · ψ1 ) + cyclic ,
′
O16 = (f · ψ1 )(ψ2 · ψ3 )(X · ψ1 ) + cyclic ,
′
O17 = (ψ1 · ψ2 )(ψ2 · ψ3 )(ψ3 · ψ1 ) . (2.7)

One can also consider cyclic-invariant Q-closed operators in this sector. Using 7 fields, one can
always take gauge-invariant operators to take the form of

(A · B)(C · D)(E · (F × G)) . (2.8)

There could be further relations between operators of this form, mostly due to the identity

A ⊗ (B × C) + B ⊗ (C × A) + C ⊗ (A × B) = [A · (B × C)]13×3 , (2.9)

which holds when A, B, C are either: all bosonic, all fermionic, two bosonic and one fermionic.
(Similar identity holds with signs changed when one field is bosonic and two are fermionic.)
In any case, writing down all possible operators of the form (2.8), we checked on a computer
all the relations between them and listed truly independent ones only. Among them, we find
18 independent Q-closed cyclic-invariant operators. 17 of them can be taken as Q-actions of
(2.7), and the remaining one can be taken as (2.3). This proves that (2.3) can be taken as the
representative of the unique cohomology in this sector found in [7].

Any linear combination between our cohomology operator (2.3) and the exact operators
given by the Q-actions on (2.7) will give alternative representation of the cohomology operator.
It would be nice if one can find other simple expressions.

We make several remarks and discussions on the future directions.

Beyond the threshold n = 24, [7] also found 6 non-graviton cohomologies at n = 25 order,
with energies E = 10, etc. These operators are essentially not new. They are the Qmα̇
superconformal descendants of the primary at E = 192
that we have discussed so far.

We also note that our analysis was substantially simplified due to the fact that the Q-
action closes within φ̄m , ψm , f . This is the subsector cannot be reached just by specifying the
charges E, RI , Ji . For instance, the charges of f are completely the same as that carried by
two derivatives D++̇ D+−̇ . What made us possible to restrict ourselves to this subsector is

5
the ‘bonus symmetry’ Y [7] of the cohomology problem, which is the number of fields in the
operator. Restricting studies within this subsector, we think there is a good chance to make
many parts of the calculations done in [7] analytically.

We find that our representative (2.3) of the cohomology may show various interesting as-
pects. For instance, in (2.3), many terms after the Q-action vanish due to Fermi statistics,
such as (ψ1 · X)2 = 0. This aspect may be used to construct generalized ansatze for new
cohomologies. It should be a subtle variant of the ‘Fermi surface’ operators discussed in [5].
On the other hand, (2.3) might also admit a ‘giant graviton’ interpretation. That is to estab-
lish non-graviton cohomologies as determinant-like operators having open spin chains ending
on them, generalizing the open giant magnons [17]. Note that, inspired by the recent giant
graviton rewriting of the index [18], it was shown that the intersecting giant gravitons with
open strings attached account for the small black hole microstates [19].

Acknowledgements

We thank Chi-Ming Chang for pointing out an error in the previous version. This work is
supported in part by a KIAS Individual Grant PG081601, PG081602 at Korea Institute for Ad-
vanced Study (SC), the National Research Foundation of Korea (NRF) Grants 2021R1A2C2012350
(SK, EL) and the National Research Foundation of Korea (NRF) Grants 2021R1A6A1A10042944,
2021R1A2C1012440 (JP).

References
[1] J. B. Gutowski and H. S. Reall, JHEP 0402, 006 (2004) doi:10.1088/1126-
6708/2004/02/006 [hep-th/0401042]; J. B. Gutowski and H. S. Reall, JHEP 0404,
048 (2004) doi:10.1088/1126-6708/2004/04/048 [hep-th/0401129]; Z. W. Chong,
M. Cvetic, H. Lu and C. N. Pope, Phys. Rev. D 72, 041901 (2005)
doi:10.1103/PhysRevD.72.041901 [hep-th/0505112]; H. K. Kunduri, J. Lucietti
and H. S. Reall, JHEP 0604, 036 (2006) doi:10.1088/1126-6708/2006/04/036
[hep-th/0601156].

[2] I. Biswas, D. Gaiotto, S. Lahiri and S. Minwalla, JHEP 12, 006 (2007)
doi:10.1088/1126-6708/2007/12/006 [arXiv:hep-th/0606087 [hep-th]].

[3] L. Grant, P. A. Grassi, S. Kim and S. Minwalla, JHEP 05, 049 (2008) doi:10.1088/1126-
6708/2008/05/049 [arXiv:0803.4183 [hep-th]].

[4] J. Kinney, J. M. Maldacena, S. Minwalla and S. Raju, Commun. Math. Phys. 275, 209
(2007) doi:10.1007/s00220-007-0258-7 [hep-th/0510251].

6
 [5] M. Berkooz, D. Reichmann and J. Simon, JHEP 01, 048 (2007) doi:10.1088/1126-
6708/2007/01/048 [arXiv:hep-th/0604023 [hep-th]].

[6] S. Minwalla, Supersymmetric States in N = 4 Yang-Mills, talk given at Strings 2006,

Beijing.

[7] C. M. Chang and Y. H. Lin, [arXiv:2209.06728 [hep-th]].

[8] R. A. Janik and M. Trzetrzelewski, Phys. Rev. D 77, 085024 (2008)

doi:10.1103/PhysRevD.77.085024 [arXiv:0712.2714 [hep-th]].

[9] C. M. Chang and X. Yin, Phys. Rev. D 88, no.10, 106005 (2013)
doi:10.1103/PhysRevD.88.106005 [arXiv:1305.6314 [hep-th]].

[10] J. McGreevy, L. Susskind and N. Toumbas, JHEP 06, 008 (2000) doi:10.1088/1126-
6708/2000/06/008 [arXiv:hep-th/0003075 [hep-th]]; M. T. Grisaru, R. C. My-
ers and O. Tafjord, JHEP 08, 040 (2000) doi:10.1088/1126-6708/2000/08/040
[arXiv:hep-th/0008015 [hep-th]]; A. Hashimoto, S. Hirano and N. Itzhaki, JHEP 08,
051 (2000) doi:10.1088/1126-6708/2000/08/051 [arXiv:hep-th/0008016 [hep-th]].

[11] A. Cabo-Bizet, D. Cassani, D. Martelli and S. Murthy, arXiv:1810.11442 [hep-th].

[12] S. Choi, J. Kim, S. Kim and J. Nahmgoong, [arXiv:1810.12067 [hep-th]].

[13] S. Choi, S. Jeong, S. Kim and E. Lee, JHEP 09, 138 (2023)
doi:10.1007/JHEP09(2023)138 [arXiv:2111.10720 [hep-th]].

[14] F. Benini and P. Milan, arXiv:1812.09613 [hep-th].

[15] S. Murthy, [arXiv:2005.10843 [hep-th]].

[16] P. Agarwal, S. Choi, J. Kim, S. Kim and J. Nahmgoong, Phys. Rev. D 103, no.12,
126006 (2021) doi:10.1103/PhysRevD.103.126006 [arXiv:2005.11240 [hep-th]].

[17] D. M. Hofman and J. M. Maldacena, J. Phys. A 39, 13095-13118 (2006)

doi:10.1088/0305-4470/39/41/S17 [arXiv:hep-th/0604135 [hep-th]].

[18] Y. Imamura, PTEP 2021, no.12, 123B05 (2021) doi:10.1093/ptep/ptab141

[arXiv:2108.12090 [hep-th]]; D. Gaiotto and J. H. Lee, [arXiv:2109.02545 [hep-th]];
S. Murthy, [arXiv:2202.06897 [hep-th]]; J. H. Lee, [arXiv:2204.09286 [hep-th]].

[19] S. Choi, S. Kim, E. Lee and J. Lee, JHEP 11, 086 (2023) doi:10.1007/JHEP11(2023)086
[arXiv:2207.05172 [hep-th]].