Version of Record: https://www.sciencedirect.

com/science/article/pii/S2405601423000226
Manuscript_41ca7ecaefacff77833dfebf325522d5

Nuclear and
Particle Physics
Proceedings
Nuclear and Particle Physics Proceedings 00 (2022) 1–13

QCD (spectral) sum rules 2022 ∗

Stephan Narison a,
a Laboratoire Univers et Particules , CNRS-IN2P3, Case 070, Place Eugène Bataillon, 34095 - Montpellier Cedex 05, France
and
Institute of High-Energy Physics of Madagascar (iHEPMAD), University of Ankatso, Antananarivo 101, Madagascar

Abstract
We present a compact review of the status of QCD spectral sum rules until 2022. We emphasize the recent progresses
for determining the QCD input parameters (α s , running quark masses, quark and gluon condensates) where their
correlations have been taken into account. Some selected phenomenological uses of the sum rules in different channels
(light and heavy quarks, gluonia/glueballs, hybrids and four-quark states) are briefly reviewed and commented. The
estimate of the 1−+ light hybrid mass is revised which confirms the hybrid nature of the π1 (1600) but not the π1 (2050).
Keywords: QCD spectral sum rules, QCD coupling α s , Hadron and Quark masses, QCD condensates.

1. Pre-QCD current algebra sum rules shall discuss in the next section, the phenomenological
 Since the famous pre-QCD Weinberg [1] (resp. success of QCD (spectral) sum rules or SVZ sum rules
Das, Mathur, Okubo (DMO) [2]) sum rules of the√1967 comes from the improvemnt of the usual dispersion re-
for predicting the axial vector A1 mass : MA1 ' 2Mρ lation which is the bridge between the high-energy QCD
assuming the asymptotic realization of chiral S U(n)L ⊗ expression and the measurable spectral function at low
S U(n)R (resp. flavour S U(n)L+R ) symmetry, spectral energy from e.g. the relation between the electomag-
sum rules have been applied successfully to study the netic spectral function and the e+ e− → hadrons cross-
masses and decay constants of hadrons. section via the optical theorem.
 The breaking of the previous sum rules by the
quark masses and QCD coupling α s have been studied
in Ref. [3, 4] where it has been shown that the 1st Wein-
berg sum rule is only broken by the running light quark
mass to order α s while the 2nd Weinberg and DMO sum
rules are broken to leading order (LO).
2. The SVZ sum rules
 Besides the Shifman-Vainshtein-Zakharov (SVZ)
(see Fig. 1) theoretical improvement of the perturbative
QCD expression introduced in 1979 [5, 6] 1 which we

∗ Compact Review talk presented at QCD22, 25th International

Conference in QCD (4-7 july 2022, Montpellier - FR) and at HEP-

MAD 22 , 12th HEP International Conference (10-16 october 2022,
Antananarivo - MG) Figure 1: Left to right : Arakady Vainshtein, SN, Valya Zakharov,
Email address: [email protected] (Stephan Narison ) Mikhael Shifman at the Munich conference (2006).
1 For reviews, see e.g. the books [7, 8] and references therein and

the recent ones in Refs.[9, 10].  This improvement has been been achieved by work-

© 2023 published by Elsevier. This manuscript is made available under the Elsevier user license
https://www.elsevier.com/open-access/userlicense/1.0/
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 2

ing with large number n of derivatives and large value – Gluon local currents G2 , G3 · · · for gluonia/glueball
of the Q2 momentum transfer Q2 but taking their ratio states,
τ ≡ n/Q2 finite leading to the so-called Borel/Laplace – Quark-gluon local currents ψ̄Gψ, ψ̄G2 ψ · · · for hy-
or Exponential sum rules (LSR) and their ratios,2 : brid states.
 The previous sum rule improvements enhance the
(−Q2 )n ∂n Π
Lc0 (τ, µ) ≡ lim 2 n
low energy contribution to the spectral integral which
Q2 , n → ∞ (n − 1)! (∂Q ) is accessible experimentally. In the often case where
n/Q2 ≡ τ the data on the spectral function are not available, one
Z tc
1 usually parametrizes it via the minimal duality ansatz :
= dt e−tτ ImΠH (t, µ) , 1
π ImΠH (t) ' fH2 MH2d δ(t − MH2 )
t>
R tc π
Lc1 dt e−tτ t ImΠH (t, µ) + “QCD continuum”θ(t − tc ), (4)
c
R10 (τ) ≡ = t>
R tc , (1)
Lc0 dt e−tτ ImΠ (t, µ)
H in order to predict the masses and couplings of the low-
t>
est ground state and in some case the ones of its ra-
where τ is the LSR variable, t > is the hadronic thresh-
dial excitations. d depends on the dimension of the
old. Here tc is the threshold of the “QCD continuum”
current, fH is the hadron decay constant normalized as
which parametrizes, from the discontinuity of the Feyn-
fπ = 132 MeV. Its accuracy has been tested in various
man diagrams, the spectral function Im ΠH (t, m2Q , µ2 ).
light and heavy quark channels e+ e− → ρ, J/ψ, Υ, . . .
 For heavy quark systems, another successful sum
where complete data are available [7, 8] and in the π-
rules are the Q20 -moments sum rules (MSR) and their
pseudoscalar channel where an improved parametriza-
ratios :
!n tion of the 3π channel within chiral perturbation theory
1 ∂ has been used [15]. Within a such parametrization, the
Mn (Q0 , µ) =
c 2
ΠH (q2 , m2Q )|q2 =−Q20
n! ∂q2 ratio of sum rules is used to extract the mass of the low-
Z tc
dt 1 est ground state as it is equal to its square. However,
= 2 n π
Im ΠH (t, µ) , this analysis cannot be done blindly without studying /
t> (t + Q0 )
checking the moments which can violate positivity for
Mcn some values of the sum rule variables (τ, n, Q20 ) though
rnc n+p = : p = 1, 2, . . . , (2)
Mcn+p their ratio can lead to a positive number identified with
the hadron mass squared.
where the mQ heavy quark mass has been exploited for
 Within the duality ansatz paametrization of the
doing the OPE in terms of 1/mQ expansion. Q20 =
spectral function, the ratios Rnc (τ) and rn,n+1
c
is approxi-
0, m2Q , .. is a free chosen scale and n is the degree of
mately equal to the hadron mass squared while the dou-
moments.
ble ratio of sum rule (DRSR) [16] :
 ΠH (t, m2Q , µ2 ) is the generic two-point correlator de-
fined as : c
R10 (τ0 )|H 0 M2 0
Z rH 0 /H ≡ c ' H2 (5)
R10 (τ)|H MH
ΠH (q2 ) = i d4 x e−iqx h0|T OH (x) (OH (0))† |0i . (3)
can be used to get the meson H and H 0 mass-splitting,
 OH (x) can be the interpolating local currents: like e.g. the one due to S U(3) breakings, provided that
– Quark bilinear local current ψ̄1 Γ12 ψ2 for mesons. the optimal value: τ00 ' τ0 . A similar quantity can be
Γ12 is any Dirac matrices which specify the quantum used for the heavy quark moments.
numbers of the corresponding meson state (and its radial
excitations), 3. Some other type of sum rules
– Quark trilinear local current ψ1 ψ2 ψ3 for baryons,  Alternative to these SVZ sum rules is the local Fi-
– Four-quark (ψ̄1 Γ12 ψ2 )(ψ̄3 Γ34 ψ4 ) or diquark anti- nite Energy Sum Rule (FESR) :
diquark (ψ̄1 Γ12 ψ̄2 )(ψ3 Γ34 ψ4 ) local current for molecules Z tc
1
or tetraquark states. Fnc (µ) = dt tn Im ΠH (t, µ) , (6)
t> π

2 Non-relativistic vesion of this sum rule has been discussed by [12,

where n ≥ 0 is the degree of the sum rule. Contrary to
13], while the inclusionof the PT α s correction to the QCD expression
LSR, FESR emphasizes the role of higher masses radial
has shown that it has the property of an inverse Laplace transform [14] excitations to the integral. This sum rule can also be de-
though the name LSR. rived from LSR by using a τ → 0 expansion. FESR can
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 3

be used to fix the value of the QCD continuum threshold  The D = 4 condensates mhψ̄ψi and the part of
tc from its dual constraint with the mass and decay con- the trace of the energy-momentum transfer : θµµ |g ≡
stant of the ground state [17, 18]. However, the accuracy mγhψ̄ψi + (1/4)βhGaµνGµνa i are known to be subtraction
of the result is destroyed when n increases due to the µ-independent where γ, β are the quark mass anoma-
needs of more information for the spectral function near lous dimension and Callan-Symanzik β-function.
the cut on the real axis. Therefore, the FESR should be – The hψ̄ψi condensate can be deduced from the well-
used with a great care for high n values.  Another sum known Gell-Mann, Oakes, Renner relation [21]:
rule is the Gaussian sum rule [17, 18]: (mu + md )hψ̄u ψu + ψ̄d ψd i = −m2π fπ2 , (10)
Z tc
1 (t+s)2 1 once the running light quark mass is known (mπ , fπ =
Gn (s, σ, µ) = √
c
dt e− 4π ImΠH (t, µ) ,(7)
4πσ t> π 132 MeV are the pion mass and decay constant) or di-
rectly from light baryon sum rules [22]. With the value
for a Gaussian
√ centered at s with a finite width reso-
of the running u, d quark masses given in Table 2, one
lution 4πσ. It has been aslo shown in Ref. [17, 18]
can deduce the value of hψ̄ψi in this table.
that the LSR can be derived from the Gaussian sum rule
– The original value of the D = 4 gluon conden-
using the ζ-prescription.  Another sum rule is the τ-
sate hα sG2 i = 0.04 GeV4 of SVZ [5, 11] has been
like-decay sum rule [19]:
claimed [12, 13] from charmonium sum rules and Fi-
Z Mτ2
ds s
!2 nite Energy Sum Rule (FESR) for e+ e− → I = 1
Rτ (Mτ ) = 1− 2 × hadrons [17, 18] to be underestimated. Recent analy-
0 Mτ2 Mτ
( ! ) sis from e+ e− → hadrons, τ-decay and charmonium
2s
1 + 2 ImΠH (s) + ImΠH (s) ,
(1) (0)
(8) confirm these claims (see the determinations in Table 1
Mτ of [23]). The present updated average is (see Table 2):
for a spin one and zero hadronic final states or its mo- hα sG2 i = (6.49 ± 0.36) × 10−2 GeV4 . (11)
ment [20]. Contrary to the usual FESR, it has the advan- The renormalization of higher dimension condensates
tage to have the threshold suppression factor near the have been studied in [24] where:
real cut. This property has enabled to extract with high  The D = 5 quark-gluon mixed condensate usu-
level of accuracy the QCD coupling at the Mτ scale. ally parametrized as ghψ̄Gψi = M02 hψ̄ψi mixes under
renormalization and runs as (α s )1/(6β1 ) in the chiral limit
4. The SVZ - OPE
m = 0. The scale :
 According to SVZ, the RHS of the two-point func-
tion can be evaluated in QCD within the Operator M02 = 0.8(2) GeV2 , (12)
Product Expansion (OPE) provided that Λ2  Q2 ≡ has been phenomenologically estimated from light
−q2 , m2Q . In this way, it reads : baryons [25–27] and heavy-light mesons [28] sum rules.
X  The D = 6 four-quark condensate mixes un-
ΠH (q2 , m2Q , µ) = C D (q2 , m2Q , µ)hOD (µ)i , (9) der renormalization with some other ones which is
D=0,2,.. not compatible with the vacuum saturation assumption
where, in addition to the usual perturbative QCD contri- used by SVZ. Its phenomenological estimate from τ-
bution (unit operator), one can add the ones due to non- decays [42], e+ e− → hadrons data [29], τ-decay [30],
perturbative gauge invariant quark and gluon conden- FESR [17, 18] and baryon [26] sum rules, leads to:
sates hOD (µ)i having a dimension D which have been ρα s hψ̄ψi2 ' 5.8(9)10−4 GeV6 : ρ ' 2 ∼ 4 , (13)
assumed to parametrize approximately the not yet un- where ρ, indicates the deviation from factorization.
der good control QCD confinement. C D are separated  Fixing the ratio hα sG2 i/ρα s hψ̄ψi2 = 106(12)
calculable Wilson coefficients in PT-QCD: GeV−2 as quoted in Ref. [40] which reduces the anal-
 The usual perturbative (PT) contribution corre- ysis to a one-parameter fit, one deduces from LSR [42]:
sponds to D = 0 while the quadratic quark mass cor-
rections enter via D = 2. hα sG2 i = (6.1 ± 0.7)10−2 GeV4 , (14)
 The quark and gluon condensates entering into the which shows the self-consistency of the previous num-
OPE up to D = 6 are successively the : bers. Some other consistency tests can be found in [42].
– D = 4 quark and gluon mhψ̄ψi and hα sG2 i,  The D = 6 g3 fabc hGaGbGc i condensate does not
– D = 5 mixed quark-gluon: hψ̄σµν λ2a Gaµν ψi mix under renormalization and behaves as (α s )23/(6 β1 ) ,
– D = 6 four-quark and three-gluon: hψ̄Γ1 ψψ̄Γ2 ψi, where β1 = −(1/2)(11 − 2n f /3) is the first coefficient of
hg3 fabcGaµρGb,ρ c,ν
ν G ρ i. the β-function and n f is number of quark flavours. The
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 4

first improvement of the estimate of the g3 fabc hGaGbGc i analysis of the lattice data of the pseudoscalar ⊕ scalar
condensate was the recent direct determination of the two-point correlators [37] lead to the average [41]:
ratio of the dimension-six gluon condensate hg3 fabcG3 i (α s /π)λ2 ' −(7 ± 3) × 10−2 GeV2 . (16)
over the dimension-four one hα sG2 i using heavy quark
sum rules with the value [31]: The existence of this D = 2 term not present in the stan-
dard OPE (absence of gauge invariant D = 2 term) has
ρG ≡ hg3 fabcG3 i/hα sG2 i = (8.2 ± 1.0) GeV2 , (15) raised some vigourous (unjustifed and emotional) re-
which differs significantly from the instanton liquid actions though its contribution is tiny in the sum rule
model estimate [32–34]. This result may question the and τ-decays [42] analyses but it has solved some para-
validity of a such result. Earlier lattice results in doxical sum rule scale puzzles [37]. This D = 2 term
pureYang-Mills found: ρG ≈ 1.2 GeV2 [35] such that it also manifests as a linear term of the heavy quark po-
is important to have new lattice results for this quantity. tentials [43] and in the SVZ-expansion [44] in some
Note however, that the value given in Eq. 15 might also AdS/QCD models. However, this term is not of In-
be an effective value of all unknown high-dimension fraRed origin like some other non-perturbative conden-
condensates not taken into account in the analysis of sates but it is dual to the sum of higher order Ultra-
[31] when requiring the fit of the data by the truncated Violet terms of the PT series as shown in [41] : better
OPE if, at that order, the OPE does not converge. We the series is known , lesser is the strength of this term
shall see here and in some examples that the effect of which can vanish after some high order terms of the PT
hg3 fabcG3 i is a small correction at the stability region series. A such term is dual to a geometric sum of the co-
where the optimal results are extracted. efficients of the PT series and its size is consistent with
 Usually, the truncation of the OPE up to D = 6 pro- the values of the known coefficients.
vides enough information for extracting the masses and  Small size instantons
couplings of the ground state hadrons with a good accu- Direct instantons are expected to be present in QCD
racy. In many papers, some classes of higher dimension for explaining the η0 − π mass shift (the so-called U(1)A
terms up to D=12 ! are added in the OPE. However, it is axial problem [45]). At large Q2 , it will be highly sup-
not clear if such term gives the dominant contributions pressed as it can be parametrized by an operator of high
compared to the omitted ones having the same dimen- dimension D = 9 [7, 8]. Its quantitative effect has been
sion. The size of these high-dimension condensates is discussed in previous QSSR literature and has lead to
not also under a good control due to the eventual viola- some controversy [7, 8]. Applied to the e+ e− → I =
tion of the vacuum saturation used for their estimate. 1 hadrons data, this effect is found to be negligible [40],
 Applied to the previous Weinberg sum rules [1], it while in the pseudoscalar channel, it gives a prediction
has been found [3] that, in the chiral limit mq = 0, the for mud (2) ≡ (1/2)(mu +md )(2) = (2.42±0.16) MeV and
1st Weinberg sum rule is spontaneously broken by the m s (2) = (63.1 ± 3.4) MeV lower than the estimate from
four-quark condensate to order α s hq̄qi2 while for non- the standard SVZ OPE of (3.95 ± 0.28) and (98.5 ± 5.5)
zero quark masses, there are contributions from mi m j MeV [46]. This instanton effect is not favoured by the
and mi hψ̄i ψi i quark condensate to these Weinberg and lattice average [61]: mud (2) = (3.6 ± 0.2) MeV and
DMO sum rules [3, 4]. m s (2) = (93.8 ± 2.4) MeV. However, one also oberves
in Ref. [46] that the ratios of masses are (almost) unaf-
5. Beyond the SVZ-OPE
fected by the presence of instanton.
 The D = 2 tachyonic gluon mass
 Duality violation
The asymptotic behaviour of the PT series is often A model of a duality violation for the spectral
expected to have an exponential behaviour (Borel sum) function has been proposed in Ref. [47] in order to
according to the large β-approximation and then alter- parametrize the oscillations observed in the spectral
nate signs are expected to be seen at large orders of function controlling τ-decay and e+ e− → I = 1 hadrons
PT. However, the known calculated terms up to or- from the data. Typically, it behaves as :
der α5s of the vector correlator D-function do not yet
show such properties. In Refs. [36–39], a phenomeno- ∆ImΠDV
H ∼ κe
−γt
sin(α + βt)θ(t − tc ), (17)
logical parametrization of these higher order terms where κ, γ, α, β are fitted parameters not based from first
due to UV-renormalons have been proposed which is principles which are channel and tc dependent. Within
parametrized by a tachyonic gluon mass squared con- this model, where the contribution is double exponential
tribution. Its phenomenological value from e+ e− → suppressed in the Laplace sum rule analysis, we expect
hadrons data [40], π-Laplace sum rule [36] and from an that in the stability region where the QCD continuum
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 5

contribution to the sum rule is minimal and where the

optimal results in this paper will be extracted, such du-
ality violations is irrelevant in the sum rule analysis. A
similar conclusion has been recently reached in the de-
termination of α s from τ-decay using a slightly different
parametrization of the spectral function [48].
6. Optimization procedure for the SVZ sum rule
 The SVZ sum rule window
The second important step in the sum rule analysis is
the extraction of the optimal result from the sum rule
as, in principle, the Laplace sum rule (LSR) variable
Figure 3: Schematic behaviour of the J/ψ mass versus the sum rule
τ, the degree of moments n and the QCD contiuum variable τN from Ref. [12, 13]. The horizontal line is the experimental
threshold tc are free external variables. In their origi- mass.
nal work, SVZ have ajusted the values of M 2 ≡ 1/τ
and tc using some guessed % contributions, for find- LO D4

ing the sum rule window where the QCD continuum 10.5 NLO D4
contribution is less than some input number while the
!D6
10
ground state one is bigger than some input number and
!D6!D8
where, in this sum rule window, the OPE is expected 9.5
to converge. The arbitrariness values of these numbers Exp

have created some doubts for non-experts on the results 9

Exp!Cont
from the sum rules, in addition to the ones on the even- 0 0.1 0.2 0.3 0.4
tual non-correctness of the non-trivial QCD-OPE ex-
pressions. Unfortunately, many sum rules practioners q
Figure 4: Behaviour of the ratio of moments R0b (m2b ) in GeV versus
continue to use this SVZ criterion.
τ in GeV−2 and for mb (mb ) = 4212 MeV from [49]. The black contin-
 τ-stability from quantum mechanics and J/Ψ uous (rep. short dashed) curves are the experimental contribution in-
cluding (resp. without) the QCD continuum (it is about the MΥ ). The
green (thick continuous) is the PT contribution including the D = 4
condensate to LO. The long dashed (red) curve is the contribution in-
cluding the α s correction to the D = 4 contribution. The short dashed
(blue) curve is the QCD expression including the D = 6 condensate
and the dot-dashed (red-wine) is the QCD expression including the
D = 8 contribution.

 tc and µ-stabilities

Later on, we have extended this τ-stability criterion

to the continuum threshold variable tc 3 and to the arbi-
trary Perturbative (PT) subtraction point µ 4 . The lowest
value of tc corresponds to the beginning of τ-stability
Figure 2: Behaviour of the ground state mass versus the time variable while the tc -stability corresponds to a complete lowest
τ for different truncation of the series from Ref. [12, 13]. The horizon- ground state dominance in the spectral integral. In our
tal line is the exact solution. ω is the harmonic oscillator frequency.
analysis, we always consider the conservative optimal
Hopefully, in the examples of harmonic oscillator in result inside this tc -region 5 .
quantum mechanics and of the non-relativistic charmo-
nium sum rules, Refs. [12, 13] have shown that the opti- 3 In many papers, the optimal value is exracted at the lowest value
mal information from the analysis from truncated series of tc ! but the result still increaes with the tc changes.
is obtained at the minimum or inflexion point in τ of the 4 One can also use the RGE resummed solution which is equivalent

approximate theoretical curves (see Figs 2 and 3 (prin- to take µ2 = 1/τ but in some case the value of τ is relatively large such
that the OPE is not well behaved. A such choice of µ value is often
ciple of minimum sensitivity of the physical parameters
outside the µ-stability region.
on the external sum rule variable τ). We illustrate in 5 √t is often identified to the mass of the 1st radial excitation
c
Figs. 4 and 5 the analysis for the Υ systems and for the which is a crude approximation as the QCD continuum smears all
B meson decay constant using relativistic sum rules. higher state contributions to the spectral function.
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 6

Figure 6: Correlation between hα s G2 i and mc (mc ) for given values of

α s and µ from J/ψ and χc .
Figure 5: Behaviour of fB from Ref. [50] versus the sum rule variable
τ for different values of the QCD continuum threshold tc and for a
given µ = 3 GeV and mb (mb ) = 4177 MeV.
pseudoscalar χc1 channel. A similar relation between
 Ground state versus the QCD continuum m̄c (m̄c ) and hα sG2 i from vector channel has been ob-
For some low values of (tc , τ), one can have some tained in [55] where using, as input, an inappropriate
flat curves ior some (apparent) minimum in τ. Then, to value of m̄c (m̄c ), has lead the authors to a too small value
check or/and to restrict the optimal region in this case, of hα sG2 i.
one also requests that the contribution of the ground
state to e.g. the spectral integral (e.g. moment sum rule)
is larger than the QCD continuum one, which one can
formulate in a rigorous way as (see e.g. [51] for some
examples of applications):
R tc
dt e−tτ Im Π(t)
RP/C ≡ R t>∞ ≥ 1. (18)
t
dt e−tτ Im Π(t)
c

7. Light quark sum rules

The light quarks sum rules have been used (for the Figure 7: Correlation between α s and hα s G2 i by requiring that the
first time) in Refs. [14, 52] to determine the light quark sum rules reproduce the (pseudo)scalar mass-splittings for given val-
ues of m̄c,b (m̄c,b ). The central value of hα s G2 i in the legend is the one
running masses. Later on, it has been applied to study of the previous average obtained in [23].
the light meson properties [11, 53, 54], to extract the
gluon hα sG2 i, the quark hψ̄ψi, the mixed quark-gluon
hψ̄Gψi and four-quark condensates. The different deter- 0.40
minations are reviewed in [7, 8, 54] while the determi- 0.35
Á
nations of the light quark masses have been updated in
0.30
asHmL

the recent work [46]. The results from light quark and
0.25
Ú

τ-decay like sum rules are quoted in Table 2.

0.20
8. Heavy quark sum rules for ordinary hadrons ‡

0.15
 Since the work of SVZ, charmonium and bot- 2 4 6 8 10
tomium sum rules have been used to determine the m@GeVD
heavy quark mass and the gluon condensates. The
Figure 8: Comparison with the running of the world average
analysis has been updated in Ref. [23] where (for the α s (MZ ) = 0.1181(11) [56, 57] of our predictions at three different
first time) the correlations between different parame- scales: Mτ for the original low moment τ-decay width [42] (open
ters (mQ , α s , hα sG2 i) have been emphasized using other circle), 2.85 GeV for Mχc0 − Mηc (full triangle) and 9.5 GeV for
channels than the vector one. Not taking into account Mχb0 − Mηb (full square) [23].
such correlations from some other channels have lead
to some apparent discrepancies among some previous  Some other interesting correlations can be found in
determinations. Ref. [23]. In particular, the value of α s at the subtrac-
 This is, e.g., the case of hα sG2 i from the J/ψ chan- tion scale µc = 2.85 GeV and µb = 9.5 GeV has been
nel where, as one can see in Fig. 6, one cannot constrain determined from the Mχ0c(0b) − Mηc(b) mass splittings tak-
accurately its value without adding the analysis of the ing into account its correlation with hα sG2 i as can be
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 7

seen in Fig. 7:
Table 2: QCD parameters from light and heavy quarks QSSR (Mo-
α s (2.85) = 0.262(9) α s (Mτ ) = 0.318(15) ments, LSR and ratios of sum rules) within stability criteria. The
running light quark masses and condensates have been evaluated at 2
α s (MZ ) = 0.1183(19)(3) ,
GeV within the SVZ expansion without instantons contributions dis-
α s (9.50) = 0.180(8) α s (Mτ ) = 0.312(27) favoured by lattice results [61].
α s (MZ ) = 0.1175(32)(3) . (19)
Parameters Values Sources Ref.
The last error is due to the running procedure. We have Heavy
requested that the method reproduces within the errors α s (MZ ) 0.1181(16)(3) Mχ0c,b −Mηc,b [23]
the experimental mass-splittings by about (2 ∼ 3) MeV. mc (mc ) [MeV] 1266(6) D, Bc ⊕ (see Table 1)
The geometric mean of the two previous values of α s is : J/ψ, χc1 , ηc
mb (mb ) [MeV] 4196(8) B, Bc ⊕ Υ (see Table 1)
α s (Mτ ) = 0.317(15) α s (MZ ) = 0.1181(19)(3), (20) hα s G2 i [GeV4 ] 6.49(35)10−2 Light, Heavy [9, 23]
which is (surprisingly) in a very good agreement with hg3 G3 i/hα s G2 i 8.2(1.0)[GeV2 ] J/ψ [31]
the world average [56, 57] : Light
µ̂ψ [MeV] 253(6) Light [7, 8, 46, 62]
α s (MZ ) = 0.1181(11) . (21)
hψ̄ψi(2) [MeV]3 −(276 ± 7)3 – –
and where one observes a nice 1/Log-behaviour (see
Fig. 7) as expected from asymptotic freedom. κ ≡ h s̄si/hd̄di 0.74(6) Light, Heavy[7, 8, 46, 62–64]
m̂u [MeV] 3.05 ± 0.32 Light [7, 8, 46, 62]
Table 1: Values of mc (mc ) and mb (mb ) in units of MeV coming from m̂d [MeV] 6.10 ± 0.57 – –
most recent QSSR analysis based on stability criteria. Some other m̂ s [MeV] 114(6) – –
determinations can be found in PDG [56]. mu (2) [MeV] 2.64 ± 0.28 – –
md (2) [MeV] 5.27 ± 0.49 – –
Masses Values Sources Ref. m s (2) [MeV] 98.5 ± 5.5 – –
mc (mc ) 1256(30) J/ψ family Ratios of LSR [23] M02 [GeV2 ] 0.8(2) Light, Heavy [7, 8, 25–28]
1266(16) Mχ0c −Mηc Ratios of LSR [23] ρα s hψ̄ψi2 × 104 5.8(9)[GeV6 ] Light, τ [18, 26, 29, 42]
1264(6) J/ψ family MOM & Ratios of MOM [58]
1286(66) MD Ratios of LSR [50]
1286(16) MBc Ratios of LSR [59]  Bc -like mesons
1266(6) Average [59] Similar analysis has been done for the Bc -like
mb (mb ) 4192(17) Υ family Ratios of LSR [23] mesons.
4188(8) Υ family MOM & Ratios of MOM [58]
The Bc mass prediction taking into account the cor-
4236(69) MB Ratios of MOM & of LSR [50]
4213(59) MB Ratio of HQET-LSR [60]
relation between mc (mc ) and mb (mb ) is given in Fig. 9.
4202(7) MBc Ratios of LSR [59] The resulting values of mc (mc ) and mb (mb ) are quoted
4196(8) Average [59] in Table 1.
 As mentioned earlier in Section 4, the ratio
hg3G3 i/hα sG2 i condensate has been also extracted (for
the first time) from charmonium sum rules where a 6350
large deviation from the dilute gas instanton estimate 6300
MBc @MeVD

m c Hm c L@MeVD
has been observed.
6250
 These determinations of the QCD parameters from 1282
heavy quark sum rules have been reviewed recently in 6200 1276
1252
M B c @DATAD
Ref. [9] and summarized in the Tables 1 and 2. 6150
9. Heavy-Light quark sum rules 6100
mb Hmb L@MeVD
4180 4200 4220 4240 4260
 D, B-like mesons
In this channel, the sum rules have been used to
extract the decay constants of the D, D∗ and of their
Figure 9: MBc as function of mb (mb ) for different values of mc (mc ),
strange partners D s , D∗s mesons, their chiral partners and at the stability point µ=7.5 GeV and for the range of τ-stability values
their B-like analogue. The masses of the observed D, D∗ τ = (0.30 − 0.32) GeV−2 .
mesons and their B-like analogue have been used to de-
termine the running mc (mc ) and mb (mb ) quark masses
(see Table 1). The average values of the decay constants The masses and couplings of the Bc -like mesons from
and the SU(3) breaking ratios come from [10] and refer- LSR [65] are given in Table 1 and 5 which are compared
ences therein. The results are compiled in Table 4. with the Heavy Quark Symmetry [65], Potential models
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 8

Table 3: Heavy-light decay constants fH within µ and tc -stability at N2LO. fH associated to the divergence of U(1)A axial current
is normalized as fπ = 132 MeV. which reads for n f flavours:
Channel Values [MeV] Bounds [MeV] fH s / fH
∂µ J5µ (x) =
X
D 204(6) ≤ 218(2) 1.170(23)
2mi ψ̄i ψi + 2n f Q(x) , (24)
i=u,d,s
Ds 243(5) ≤ 254(2) –
where :  α 
B 204(5) ≤ 235(4) 1.154(21)
µνρσGµν ρσ
s
Bs 235(4) ≤ 251(6) – Q(x) ≡ a (x)G a (x), (25)
16π
D∗ 250(8) ≤ 266(8) 1.215(30) a = 1, ..., 8 is the colour index. In earlier works [72, 73],
D∗s 290(11) ≤ 287(18) – this sum rule has been used to extract the gluon compo-
B∗ 210(6) ≤ 295(15) 1.020(11) nent of the η0 -mass and decay constant, the topological
B∗s 221(7) ≤ 317(17) – charge χ(0) ≡ ψP (0)/(8π) and its slope χ0 (0). A recent
D∗0 220(11) 0.922(15) update including N2LO PT corrections lead, in the chi-
D∗s0 202(15) – ral limit, to (in units of MeV) [74] :
B∗0 278(12) 1.064(10)
fη1 = 905(72), χ0 (0)|chiral = 24.3(3.4) , (26)
p
B∗s0 255(15) –
compared to the one at NLO [75] and the one from pure
Yang-Mills [73]: χ0 (0)|Y M = −(7 ± 3) MeV. p This re-
Table 4: Values of the masses from LSR and HQS compared with sult√is also smaller than the OZI value: χ0 (0)|OZI =
lattice and potential models (PM) results.
fπ / 6 = 38 MeV 6 . Used in the relation with the
Channel LSR HQS Lattice [66] PM [67] proton singlet form factor G A (0)(Q2 ) appearing in the
Masses first moment of the polarised proton structure function
B∗c (1−− ) 6451(86) 6315(1) 6331(7) 6330(20) [68] g1P [76, 77] (Ellis-Jaffe sum rule [78]), one obtains at
B∗0c (0++ ) 6689(198) 6723(29) 6712(19) 6693 N2LO [74]:
B1 (1++ ) 6794(128) χ0 (0)|LSR
p
6730(8) 6736(18) 6731
Bc2 (2++ ) – 6741(8) – 7007 G A |LSR (Q = 10 GeV ) = GA |OZI p
(0) 2 2 (0)
χ0 (0)|OZI
B∗0 (0++ ) 5701(196) [69] 5733
= (0.340 ± 0.050), (27)
and Lattice calculations. An upper bound for the the after running χ0 (0)|LSR from 2 to 10 GeV2 where
Bc (2S ) coupling is also derived [70]: A |OZI = 0.579 ± 0.021. As a result, the first moment
G(0)
of the polarized proton structure function reads:
fBc (2S ) ≤ (139 ± 6) MeV. (22) Z 1
Γ1p (Q2 = 10 GeV2 ) ≡ dx g1P (x, Q2 )
10. Gluonia / Glueballs sum rules 0

 NSVZ [71] have also used the sum rules to pre- = (0.144 ± 0.005), (28)
dict qualitatively the gluonia scale which has been re- in excellent agreement with the world average (0.145 ±
cently revisited in [36] including the tachyonic gluon 0.014) [79] and the recent COMPASS [80] and HER-
mass contribution. This effect has restored the large MES data [81].
value of some scales found by NSVZ. Since then, some
progress has been done for more quantitative predic-  Pseudoscalar digluonia spectrum
tions of the spectra.
Using a one resonance parametrization of the spectral
 Pseudoscalar digluonium and the proton spin function, one obtains from L0 and R10 [82]:
This channel is described by the two-point function: fP ' (8 ∼ 12) MeV, MP ' (2.05 ± 0.10) GeV, (29)
Z
ψP (q ) = (8π) i d4 x eiqx h0|Q(x)Q† (0)|0i (23)
2 2 and the upper bound from the positivity of the spectral
function:
MP ' (2.34 ± 0.42) GeV. (30)
Table 5: Values of the decay constants fH in units of MeV using as
The mass prediction is comparable with the ones from
input the values of the masses from LSR and HQS quoted in Table 1. other sum rules determinations [83–85].
Masses Bc (0−− ) B∗c (1−− ) B∗0c (0++ ) B1c (1++ ) B∗0 (0++ )
LSR 371(17) [70] 442(44) 155(17) 274(23) – 6 f is normalized as f = 93 MeV like other gluonia decay con-
η1 π
HQS – 387(15) 158(9) 266(14) 271(26) [10]
stants in this section of gluonia.
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 9

Parametrizing the spectral function beyond the one obtains, by saturating the spectral function by the σB
resonance model, a recent analysis shows that there can and its radial excitation σ0B , the low-energy sum rules:
be more states above the η0 (see Table 2 of [74]): √
√
X X 2 fS
P1 (1397 ± 81), P01 (1541 ± 118) , P2 (2751 ± 140),(31) gS π+ π− 2 fS ' 0, gS π+ π− ' 1, (39)
S ≡σ ,σ0 S ≡σ ,σ0
4 MG2
B B
where the highest mass is comparable with the lattice B B

+ −
value [86]. The corresponding couplings are: where gS π+ π− is the S π π coupling normalized as:
fP1 = 594(144), fP01 = 205(282), fP2 = 500(42). (32) 3 |gσB π+ π− |2 4m2
!1/2
Γ[σB → π+ π− +2π0 ] = 1 − 2π .(40)
P01is expected to the the radial excitation of P1 which is 2 16πMσB MσB
(a posteriori) justified by its weak coupling to the cur-
Fixing for definiteness Mσ0 = 1.37 GeV, one deduces
rent via its decay constant. These unmixed states may
for MσB ' 1.07 GeV:
explain the nature of the observed η(1405), η(1495) and
η(1760) while the P2 (2751) remains to be discovered. Γ(σB → π+ π− + π0 π0 ) ' 873 MeV. (41)
 Scalar digluonia spectra
This result has motivated the interpretation that the
Similar analysis has been done in the scalar channel
broad σ/ f0 (500) can be a good candidate for a low
by working with the two-point function:
Z mass scalar gluonium [82, 87]. Different analysis of
 †
ψG (q2 ) = 16i d4 x eiqx h0| θµµ (x) θµµ (0)|0i (33) the γγ and ππ scatterings data [88–90] have confirmed
 
G G
this result where, within a Breit-Wigner or On-shell
built from the gluon component of the trace of the parametrization of the data, the σ(500), in the com-
energy-momentum tensor: plex plane, becomes a 920 MeV resonance with a width
1 of about 700 MeV in the real axis. The presence of a
θµµ = β(α s )Gµν
X
a G µν + (1 + γm ) mi ψ̄i ψi ,
a
(34)
4 u,d,s
low mass glueball state around (0.6-1.25) GeV are ex-
pected from some other sum rules analysis [84, 91], in
with : γm = 2α s /π + · · · is the quark mass anomalous the Dragon model of [89, 92] and from a strong cou-
dimension and β(α s ) is the β-function. Working with the pling analysis of the gluon propagator [93]. A double
substracted L0 and unsubtracted sum rule L−1 leads to resonance around 0.8 and 1.6 GeV is also found from a
inconsistencies for a “one resonance” parametrization Gaussian sum rule where the lighter has a large width
of the spectral function as the two sum rules stabilize [91] and from dispersive approach [94].
at very different values of τ due to the strong effect of However, though we have found two gluonia candi-
the subtraction constant ψG (0) to L−1 [71] which pushes dates : σB (1.0) and G(1.5), these two states are not suffi-
its stabilty to lower values of τ making it less sensitive cient to explain the numerous I = 0 scalar states below
to the low mass hadron contrary to usual expectations. 2 GeV. In Ref. [90], we have extended the analysis by
Starting the analysis with the unsubtracted high degree parametrizing the spectral function beyond “two reso-
c
of moments R21 , and Lc1,2 , one obtains [82, 87]: nances” model and using various higher moments. In
MG = (1.50±0.19) GeV , fG = (390±145) MeV,(35) this way, we obtain in units of MeV:
and : MG ≤ (2.16 ± 0.16) GeV, after using positivity at MσB = 1070(126), MG1 = (1548 ± 121),
the minimum of R21 c
. Using this result into L−1,0 within fσB = 456(157), fG1 = 365(110), (42)
a “two resonances” parametrization of the spectral func-
tion, one obtains: and the radial excitations :
f σB ≈ 1.0 GeV, MσB ≈ 1.0 GeV, (36) Mσ0B = 1110(117), MG01 = 1563(141),
where we have used [71]: MG2 = 2992(221). (43)
16
ψG (0)|LET ' − β1 hα sG2 i = (1.46 ± 0.08) GeV4 ,(37) Their effective coupling which might be the sum of
π
from low-energy theorem (LET) to be checked later on. some other higher mass radial excitations are:
We use the previous result into the vertex function : fσe0f f = 648(216), fGe 0f f = 1000(230)
Z ∞ B 1
µ dt 1
V(q ) ≡ hπ|θµ |πi =
2
ImV(t), (38) fGe 2f f = 797(74). (44)
0 t − q − i π
2

with q ≡ p1 − p2 where V(0) = 2m2π . Using the fact One can compare fσe0f f with the one fσ0 (1370) = 329(30)
B
that V(0) = 0 in the chiral limit and V 0 (0) = 1, one MeV obtained from low-energy vertex sum rule.
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 10

One can observe the one to one correspondence of 11. Hybrid mesons sum rules
these states with their chiral pseudoscalar partners ob-  Light hybrids
tained in the previous section. One expects that the ob-
served σ(500), f0 (980), f0 (1370), f0 (1500), f0 (1710) They are described by the mixed light quark- gluon
emerge from a mixing among these gluonia or/and with operator:
quarkonia states, while the large width of σB → ππ is OV(A)
µ = gψ̄γµ (γµ γ5 )λaGa ψ, (49)
due to the OZI violation in this channel. where the hadron coupling is normalized as:
 Conformal charge and its slope h0|OV(A)
µ |Hi = µ fH MH2 . In the pioneer works [101],
some controversies on the QCD expression of the
The moments Lc−1 and Lc0 have been also used to ex- two-point functions have been resolved in Ref. [102].
tract the conformal charge ψG (0) in order to check the From this corrected expression, one has obained the
LET result in Eq. 37 while its slope ψG0 (0) has been ex- predictions [8, 102] :
tracted from L2 c ⊕ ψG (0) One obtains:
Mρ̃ (1−+ ) = (1.4 ∼ 1.6) GeV, Mη̃ (0−− ) ' 3.8 GeV.(50)
ψG (0) = (2.09 ± 0.29) GeV ,4

ψG0 (0) = −(0.2 ± 0.3) GeV2 , (45) The mass of the exotic 0+− hybrid related to the current:

compared to the LET estimate in Eq. 37. Oµ (V) = gψ̄γµ γ5 λaG̃a ψ, (51)

 2++ Tensor digluonium has been also found for Gaussian sum rule within a two-
resonance model to be [103]:
This channel is described by the two-point function
associated to the gluon component of the energy mo- M1 (0+− ) ' 2.6 GeV, M2 (0+− ) ' 3.57 GeV. (52)
mentum tensor:
80
1
= −GρµGρν + gµνGρσGρσ
Á
Ú

g
θµν
Ï

(46) 70
4
f1-+ @MeVD

60
‡

Contrary to the case of (pseudo)scalar channels, it can- m2 = 1êt

t c @GeV 2 D
not be affected by the eventual contribution due to in- 50
‡
1
2
4

stantons. However, in Ref. [36], it has been shown that 40

Ï

Ú
5
7
8
Á 9

the eventual contribution of the tachyonic gluon mass in

t @GeV-2D
0.2 0.4 0.6 0.8 1.0 1.2 1.4

the OPE restores the universality of the gluonium scales

for diffrent channels. Its QCD expression has been ob- Figure 10: Behaviour of the 1−+ light hybrid coupling from the mo-
tained by Ref. [71] to LO. Within a “one resonance” ment L1 versus τ for different values of tc . .
parametrization, one obtains to LO [82]:
5
MT = (2.0 ± 0.1) GeV, fT ' (80 ± 14) MeV. (47) ‡ m=t
t c @GeV 2 D
4
Ï
M1-+@GeVD

1
2
 Trigluonium sum rules
‡

3 4
Ï 5

Since the pioneer work of Ref. [95] from the analy- 7

2 8
Ú

sis of the two-point function of the scalar g3 fabcGaGbGc Ú

Ï
‡

trigluonium current which predicts. [8, 95] : 1

t @GeV-2D
0.2 0.4 0.6 0.8 1.0 1.2 1.4
MG3 ' 3.1 GeV, fG3 ' 62 MeV, (48)
and where the digluonium-trigluonium scalar mixing is Figure 11: Behaviour of the 1−+ light hybrid mass from the moment
tiny (θ ' 40 ), there are new sum rules applications in R21 versus τ for different values of tc . .
different channels [96–100].
Though there are some technical points to be solved The 1−+ meson parameters have been updated in
among different results (check of the QCD expressions Ref. [104, 105] by including the tachyonic gluon mass
and of the regions for extracting the optimal results), contribution where a mass about (1.6-1.8) GeV has been
the main feature is that these trigluonium bound states found. However, in Ref. [106], it is claimed that the use
are higher than the digluonia analogue by about (1–2) of the new value of the hg3G3 i gluon condensate from
GeV which is intuitively expected for high dimension charmonium sum rules (see Table 2) shifts the mass to
operator currents. (1.72-2.60) GeV. However, as previously mentioned in
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 11

2.0
mass [104, 105] dereases the mass by 20 MeV and in-
1.9
creases the coupling by 10 MeV. Using a two reso-
1.8
M1-+ @GeVD

Ú
nance model and subtracting the lowest ground state
1.7 Ú
Ú Ú Ú
contribution, we obtain from R21 a nice inflexion point
1.6
for τ ' (0.3 ∼ 0.6) GeV−2 (see Fig. 13) and a min-
1.5
imum for the coupling from L2 similar to Fig. 10 at
1.4
4 5 6 7 8 9 10 τ ' (0.24 ∼ 0.28) GeV−2 . This leads to the result for
tc @GeV2 D the 1st radial excitation:
Figure 12: 1−+ light hybrid mass from the moment R21 at the optimal
values of τ versus tc .
Mρ̃0 ' (2.8 ± 0.5) GeV, fρ̃0 ' (37 ± 11) MeV. (55)

5.0 The previous results confirm that the π1 (1600) is a good

4.5 m=t 1−+ hybrid state candidate [105] but not the π1 (2050).
t c @GeV 2 D
‡

4.0
M1-+@GeVD

 Heavy-Light hybrids
3.5 ‡
7
8

3.0 9 The spectra of heavy-light hybrids have been stud-

2.5 ied in Refs [107] and revised in [8, 108]. The unmixed
2.0 states are expected to be in the range 3.40 to 5.07 (resp.
1.5 7.01 to 8.6) GeV for charm (resp. bottom) channels
t @GeV-2D
0.2 0.4 0.6 0.8 1.0
where there are discrepancies to be clarified for some
results of the two groups. Mixing with ordinary mesons
Figure 13: Mass of the 1−+ light hybrid 1st radial excitation from the
is expected to shift upward the mass by about 0.02 to
moment R21 versus τfor different values of tc .
0.05 (resp. 0.19 to 0.74) GeV for the charm (resp. bot-
tom) channels [108].
Refs. [8, 104], the authors use a sum rule which is sen-
sitive to the value of the two-point function subtrac-  Heavy hybrids
tion [101]: The spectra of heavy hybrids have been studied in
16π Refs [109] and revised in Ref. [8, 110]. The lightest
q2 Π(1)
V (0)|q2 =0 ≈ α s hψ̄ψi2 , (53)
9 (resp. heaviest) charmonium state is the 1−− (3.36) (resp.
which cancels the contribution of the four-quark con- 0−− (5.51)). The analogue for the bottomium states are
densate in the OPE. This effect is not taken into ac- the 1−− (9.70) (resp. 0−− (11.48)). In this analysis, the
count in the analysis. To avoid such ambiguities , we negative parity states J PC = (0, 1, 2)−+ and 1−− are
work with higher moments where the contribution of lighter than states with positive C and P parities J PC =
the D = 6 condensates disappear to LO in the sum (0, 1)−+ , (0, 1, 2)++ [110].
rule. This can be achieved by working with q2n Π(q2 ) for
n ≥ 1. The analysis is shown in Fig 10 for the coupling 12. Heavy Four-quark / Molecule sum rules
from the moment L1 and in Fig 11 for the mass from  Spectra of compact four-quarks and molecules have
the ratio R21 . The coupling exhibits nice τ-stability (the been also extracted from QSSR two-point functions
same from L2 ) but not R21 . We extract the optimal value while there are some attempts to determine their widths
of the mass at the minimum of L1,2 . The mean from the using vertex or light cone sum rules [99, 111].
two determinations of the mass using τ from L1,2 are  Though phenomenologically successful for predict-
shown in Fig. 12 for tc ≥ M1−+ . A similar analysis is ing the X, T, Y, Z, DK, fully charmed,... spectra com-
done for the coupling. We deduce the most reliable de- pared with the data, the common weakness of the exist-
terminations independent on the d = 6 condensates: ing works, besides the (handwaving) choice of the sum
Mρ̃ = (1.69 ± 0.13) GeV, fρ̃ ' (52 ± 22) MeV, (54) rule window, is the use of perturbative LO QCD expres-
sion where the definition of the heavy quark mass which
for tc = (4 ∼ 9) GeV2 , where the stability is reached. plays a crucial role is ill-defined. The common prefered
It confirms and improves previous predictions [8, 102, (but unjustified) choice of different authors is the MS
104–106]. The errors have been added quadratically. running mass instead of the pole/on-shell quark mass
For ∆Mρ̃ (resp. ∆ fρ̃ ), 0.11 (resp. 0.2 MeV) is due which would (a priori) be more natural here as the spec-
to the sum rule analysis while 0.07 GeV (resp. 22 tral function is evaluated within an on-shell scheme.
MeV) comes from QCD. We have used Λ = (340 ± 28)  An exception is the series of works in Refs. [112,
MeV. We note that the inclusion of the tachyonic gluon 113] where the factorised NLO PT contributions have
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 12

been considered and where well-founded stability crite-  To my personal opinion, QCD spectral sum rules
ria have been used for extracting the optimal results. In can still have a long lifetime for studying successfully
this analysis, it has been found that the NLO corrections the properties of hadrons and for extracting the QCD
are (in most cases) small for the mass due to some par- parameters, provided, that we continuously improve the
tial cancellations in the ratio of moments. This explains method by doing a more careful job !
(a posteriori) the success of the LO analysis for the mass  QSSR predictions are based on QCD first principles
predictions. However, the NLO corrections can be large and complement the Lattice calculations. They are often
for the decay constants and affect the width predictions. successful and have been obtained many years before
 One can notice tha the method cannot distin- the lattice ones !
guish accurately the compact four-quark states from the
References
molecule ones leading Refs. [112, 113] to identify the
[1] S. Weinberg, Phys. Rev. Lett. 18 (1967) 507.
experimental states with the Tetramole states having [2] T. Das, V.S. Mathur, S. Okubo, Phys. Rev. Lett. 19 (1967) 470.
their mean masses. [3] E.G. Floratos, S. Narison and E. de Rafael, Nucl. Phys. B155
(1979) 155.
13. Concluding comments [4] S. Narison and E. de Rafael, Nucl. Phys B169 (1980) 253.
[5] M.A. Shifman, A.I. Vainshtein, V.I. Zakharov, Nucl. Phys.
B147 (1979) 385; ibid, Nucl. Phys. B147 (1979) 448.
 Within the last ten years, there has been inten- [6] V.I. Zakharov, talk given at the Sakurai’s Price, Int. J. Mod
sive horizontal applications of QSSR for extracting the .Phys. A14, (1999) 4865.
hadron masses and couplings from two-point functions [7] S. Narison, QCD spectral sum rules, World Sci. Lect. Notes
built with local currents and their widths using vertex or Phys. 26 (1989) 1.
[8] S. Narison, QCD as a theory of hadrons, Cambridge Monogr.
light cone sum rules. Unfortunately, only some few new Part. Phys. Nucl. Phys. Cosmol. 17. (2004) 1-778 [hep-
vertical applications for a much deeper improvements of ph/0205006].
the sum rules have been done. [9] S. Narison, Nucl. Part. Phys. Proc.312-317 (2021) 87; ibid,
258-259 (2015) 189.
 One can also notice the inflation of including higher
[10] S. Narison, Nucl. Part. Phys. Proc. 270-272 (2016) 143.
dimension condensates (sometimes until D=12) which [11] L.J. Reinders, H.R. Rubinstein, S. Yazaki, Phys. Reports 127
may be a good point if all contributions of a given di- (1985) 1.
mension are included and if a care has been properly [12] J.S. Bell, R.A. Bertlmann, Nucl. Phys. B177, (1981) 218; ibid,
Nucl. Phys. B187, (1981) 285.
done on the mixing of these given dimension conden- [13] R.A. Bertlmann, Acta Phys. Austriaca 53, (1981) 305.
sates under renormalization as illustrated by the case of [14] S. Narison, E. de Rafael, Phys. Lett. B522, (2001) 266.
D = 6 condensates [24]. One may also expect a large vi- [15] J. Bijnens, J. Prades, E. de Rafael, Phys. Lett. B348 (1995) 226.
olation of the vacuum saturation estimate of these con- [16] S. Narison, Phys. Lett. B210 (1988) 238; ibid, Phys. Lett. B337
(1994) 166.
densates like in the case of four-quark ones. Instead, [17] R.A. Bertlmann, G. Launer and E. de Rafael, Nucl. Phys. B250,
it is more useful to include NLO corrections to the PT (1985) 61.
and D ≤ 6 condensates contributions and to check care- [18] R.A. Bertlmann, C.A. Dominguez, M. Loewe, M. Perrottet and
fully some Wilson coefficients not automatically gener- E. de Rafael, Z. Phys. C39 (1988) 231.
[19] E. Braaten, S. Narison, A. Pich, Nucl. Phys. B373 (1992) 581.
ated by the quark propagator in external gluon fields. [20] F. Le Diberder, A. Pich, Phys. Lett. B286 (1992) 147; B289
 Another weak point is the optimization procedure (1992) 165.
based on the sum rule window of SVZ. Manyt results [21] M. Gell-Mann, R.J. Oakes, B. Renner, Phys. Rev. 175 (1968)
2195.
are alos extracted at the lowest value of tc where the re- [22] H.G. Dosch, S. Narison, Phys. Lett. B417 (1998) 173.
sult still increases with tc . Besides this point, the error [23] S. Narison, Int. J. Mod. Phys. A33 (2018) no. 10, 185004; Ad-
analysis is often done in a sloppy way where more de- dendum: Int. J. Mod. Phys. A33 (2018) no.10, 1850045.
tails sources are not provided. [24] S. Narison, R. Tarrach, Phys. Lett. B125 (1983) 217.
[25] B.L. Ioffe, Prog. Part. Nucl. Phys. 56 (2006) 232; ibid, B188
 Some authors continue also to use old obsolete es- (1981) 317; ibid, B191 (1981) 591.
timates of the QCD parameters obtained by SVZ which [26] H.G. Dosch, Non-Perturbative Methods (Montpellier 1985); Y.
(to my opinion) SVZ themselves will not consider se- Chung et al.Z. Phys. C25 (1984) 151; H.G. Dosch, M. Jamin,
riously at present . One should be aware that, since S. Narison, Phys. Lett. B220 (1989) 251.
[27] A.A. Ovchinnikov, A.A. Pivovarov, Yad. Fiz. 48 (1988) 1135.
1979, a lot of efforts have been devoted to improve [28] S. Narison, Phys. Lett. B605 (2005) 319.
their values, while during the 1980-90 period, several [29] G. Launer, S. Narison, R. Tarrach, Z. Phys. C26 (1984) 433.
works have improved our understanding of the sum [30] C.A. Dominguez, J. Sola, Z. Phys. C40 (1988) 63.
[31] S. Narison, Phys. Lett. B693 (2010) 559, erratum ibid, B705
rules. However, reading recent papers, one has the im-
(2011) 544; ibid, B706 (2012) 412; ibid, B707 (2012) 259.
pression that no theoretical progress has been done since [32] S.N. Nikolaev, A.V. Radyushkin, Phys. Lett. B124 (1983) 243.
the SVZ discovery and the clock has stopped in 1979 ! [33] T. Schafer, E.V. Shuryak, Rev. Mod. Phys. 70 (1998) 323.
 Stephan Narison / Nuclear and Particle Physics Proceedings 00 (2022) 1–13 13

[34] B.L. Ioffe, A.V. Samsonov, Phys. At. Nucl. 63 (2000) 1448. [77] G. M. Shore, Lect. Notes Phys. 737 (2008) 235.
[35] A. Di Giacomo, Non-pertubative Methods, ed. Narison, World [78] J. R. Ellis, R. L. Jaffe, Phys. Rev. D9 (1974) 1444, [Erratum-
Scientific (1985); A. Di Giacomo, G.C. Rossi, Phys. Lett. B100 ibid. D10 (1974) 1669].
(1981) 481; M. D’Elia, A. Di Giacomo, E. Meggiolaro, Phys. [79] D. Adams et al., [SMC Collaboration], Phys. Lett. B329 (1994)
Lett. B408 (1997) 315. 399, Phys. Lett. B339 (1994) 332 (Erratum).
[36] K.G. Chetyrkin, S. Narison, V.I. Zakharov, Nucl. Phys. B550 [80] V. Y. Alexakhin et al. [COMPASS Collaboration], Phys. Lett.
(1999) 353. B647 (2007) 8.
[37] S. Narison, V.I. Zakharov, Phys. Lett. B522 (2001) 266. [81] A. Airapetian et al. [HERMES Collaboration], Phys. Rev. D75
[38] V.I. Zakharov, Nucl. Phys. Proc. Suppl. 164 (2007) 240. (2007) 012007.
[39] S. Narison, Nucl. Phys. Proc. Suppl. 164 (2007) 225. [82] S. Narison, Nucl. Phys. B509 (1998) 312; ; ibid, Nucl. Phys.
[40] S. Narison, Phys. Lett. B300 (1993) 293; ibid B361 (1995) 121. Proc. Suppl. 64 (1998) 210; Phys. Rev. D73 (2006) 114024.
[41] S. Narison, V.I. Zakharov, Phys. Lett. B679 (2009) 355. [83] D. Asner et al., Phys. Lett. B296 (1992) 171.
[42] S. Narison, Phys. Lett. B673 (2009) 30. [84] H. Forkel, Phys.Rev. D 71 (2005) 054008; Phys. Rev. D64
[43] O. Andreev, Phys. Rev. D73 (2006) 107901; O. Andreev, (2001) 034015.
V.I. Zakharov, Phys. Rev. D74 (2006) 025023; ibid, D76 [85] H.-X. Chen, W. Chen, S.-L. Zhu, Phys. Rev. D104 (2021)
(2007)047705. 094050.
[44] F. Jugeau, S. Narison, H. Ratsimbarison, Phys. Lett. B722 [86] E. Gregory et al, JHEP 10 (2012) 170.
(2013) 111. [87] S. Narison, G. Veneziano, Int. J. Mod. Phys. A 4 (1989) 2751.
[45] G ’t Hooft, Nucl. Phys. B72 (1974) 461 ; G. Veneziano, —it [88] G. Mennessier, S. Narison, W. Ochs, Phys. Lett. B665 (2008)
Nucl. Phys. B117 (1976) 519; E. Witten, Nucl. Phys. B156 205; Nucl. Phys. Proc. Suppl. 238 (2008) 181; G. Mennessier,
(1979) 461; ibid Nucl. Phys. B160 (1979) 57. S. Narison, X.G Wang, Phys. Lett. B688 (2010) 59.
[46] S. Narison, Phys. Lett. B738 (2014) 346. [89] For a review, see e.g. W. Ochs, J. Phys. G40 (2013) 043001.
[47] O. Cata, M. Golterman, S. Peris, Phys. Rev. D77 (2008) [90] S. Narison, Nucl.Phys. A1017 (2022) 122337.
093006; ibid, D79 (2009) 053002; D. Boito et al., Phys.Rev. [91] D. Harnett, T.G. Steele, Nucl.Phys. A695 (2001) 279.
D84 (2011) 113006. [92] P. Minkowski, W. Ochs, Eur. Phys. J. C9 (1999) 283.
[48] A. Pich, A. Rodriguez-Sanchez, JHEP 07 (2022) 145. [93] M. Frasca, Nucl. Phys. Proc. Suppl. 207-208 (2010) 196.
[49] S. Narison, Phys. Lett. B707 (2012) 259. [94] Hsiang-na Li, Phys. Rev. D104 (2021) 11, 114017.
[50] S. Narison, Phys. Lett. B718 (2013) 1321; ibid, Nucl. Phys. [95] J.Latorre,S.Narison, S.Paban,Phys. Lett. B191(1987)437.
(Proc.Suppl.) 234 (2013) 187. [96] A. Pimikov et al., Phys. Rev. D 96 (2017) 11, 114024; Phys.
[51] R.M. Albuquerque, S. Narison and D. Rabetiarivony (to ap- Rev. Lett. 119 (2017) 7, 079101.
pear); R.M. Albuquerque et al., (to appear). [97] L. Tang, C-F. Qiao, Nucl. Phys. B904 (2016) 282; C.-F. Qiao,
[52] C. Becchi, S. Narison, E. de Rafael and F.J. Ynduràin, Z. Phys. Nucl.Part.Phys.Proc. 318-323(2022) 127.
C8 (1981) 335. [98] H.-X. Chen, W. Chen, S.-L. Zhu, Nucl.Part.Phys.Proc. 318-
[53] S. Narison, Nucl.Phys. B182 (1981) 59. 323 (2022) 122; ibid. Phys.Rev. D104 (2021) 9, 094050.
[54] S. Narison, Revista Nuovo Cimento, 10, n.2 (1987)1. [99] H.-X. Chen et al., arXiv:2204.02649v2 (2022).
[55] B.L. Ioffe, K.N. Zyablyuk, Eur. Phys. J. C27 (2003) 229; B.L. [100] S. Narison, oral contribution at QCD 21-Montpellier-FR.
Ioffe, Prog. Part. Nucl. Phys. 56 (2006) 232. [101] I.I Balitsky, D.I. D’Yakonov and A.V. Yung, Phys. Lett. B112
[56] R.L Workman et al. (Particle Data Group), Prog. Theor. Exp. (1982) 71; J. Govaerts, F. de Viron, D. Gusbin, J.Weyers, Nucl.
Phys. 2022 (2022) 083C01. Phys. B248 (1984) 1; J.I Latorre, S. Narison, P. Pascual, R.
[57] S. Bethke, Nucl. Part. Phys. Proc. 282-284 (2017)149. Tarrach, Phys. Lett. B147 (1984) 169.
[58] S. Narison, Phys. Lett. B784 (2018) 261. [102] J.I Latorre, S. Narison, P. Pascual, Z. Phys. C34 (1987) 347.
[59] S. Narison, Phys. Lett. B802 (2020) 135221. [103] J. Ho, R. Berg, W. Chen, D. Harnett, T.G. Steele Phys. Rev.
[60] S. Narison, Phys. Lett. B721 (2013) 269. D98 (2018) 9, 096020.
[61] S. Aoki et al., arXiv:1310.8555 [hep-lat] (2013). [104] K.G. Chetyrkin, S. Narison, Phys. Lett. B485 (2000) 145.
[62] S. Narison, Int. J. Mod. Phys. A30 (2015) no.20, 1550116. [105] S. Narison, Phys. Lett. B675 (2009) 319.
[63] R.M. Albuquerque, S. Narison, Phys. Lett. B694 (2010) 217. [106] Z.-R. Huang, H.-Y. Jin, T. G. Steele, Z.-F. Zhang, JHEP 1504
[64] R.M. Albuquerque, S. Narison, M. Nielsen, Phys. Lett. B684 (2015) 004; Nucl.Part.Phys.Proc. 294-296(2018) 113.
(2010) 236. [107] J. Govaerts, L.J. Reinders, J.Weyers, Nucl. Phys. B262 (1985)
[65] S. Narison, Phys.Lett. B 807 (2020) 135522. 575.
[66] N. Mathur, M. Padmanath and S. Mondal, Phys. Rev. Lett. 121 [108] J. Ho, D. Harnett, T.G. Steele JHEP 1309 (2013) 019; ibid,
(2018) 20, 202002. Nucl.Part.Phys.Proc. 294-296(2018) 75.
[67] E. Eichten, C. Quigg, Phys.Rev. D99 (2019) 5, 054025. [109] J. Govaerts, L.J. Reinders, H. Rubinstein, J.Weyers, Nucl.
[68] E. Bagan, H.G Dosch, P. Gosdzinsky, S. Narison, J.-M. Phys. B258 (1985) 215.
Richard, Z. Phys. C64 (1994) 57. [110] W. Chen, R. Kleiv, T. G. Steele, B. Bulthuis, D. Harnett, et al.,
[69] S. Narison, Phys. Lett. B605 (2005) 319. JHEP 09 (2013) 019.
[70] S. Narison, Phys. Lett. B802 (2020) 135221. [111] For recent reviews, see e.g.: R.M. Albuquerque et al., J. Phys.
[71] V.A. Novikov et al., Nucl. Phys. B165 (1980) 67; ibid, B191 G 46 (2019) 9, 093002; M. Nielsen, talk at this conferemce; R.
(1981) 301. Albuquerque and and D. Rabetiarivony talks.
[72] S. Narison, Phys. Lett. B125 (1983) 501; ibid Z. Phys. C26 [112] R.M. Albuquerque et al., Int. J. Mod. Phys. A31 (2016) 36,
(1984) 209. 1650196; Int. J. Mod. Phys. A31 (2016) 17, 1650093; Int. J.
[73] S. Narison, Phys.Lett. B255 (1991) 101. Mod. Phys. A33 (2018) 16, 1850082; Phys. Rev. D102 (2020)
[74] S. Narison, Nucl.Phys. A1020 (2022) 122393. 9, 094001; Nucl.Phys. A1007 (2021) 122113.
[75] S. Narison, G. M. Shore, G. Veneziano, Nucl.Phys. B433 [113] R.M. Albuquerque, S. Narison and D. Rabetiarivony, Phys.
(1995) 209. Rev. D103 (2021) 7, 074015; Nucl.Phys. A1023 (2022)
[76] G.M. Shore, G. Veneziano, Nucl.Phys. B381 (1992) 23. 122451.