betaFeedback

WorksPhysics and Astronomy

Multiplicity distributions in QCD jets and jet topics

Abstract

We evaluate the Koba-Nielsen-Olesen (KNO) scaling functions for quark- and gluon-initiated jets by incorporating energy conservation into the Double Logarithmic Approximation (DLA). The resulting modified DLA (MDLA) expressions differ substantially from the DLA predictions and qualitatively align with the recently proposed QCD-inspired expressions, albeit with some quantitative differences. By fixing the two parameters in the MDLA expressions, we show that the inclusive charged-particle multiplicity distributions of the two leading jets in pppp collisions at s=13\sqrt{s} = 13 TeV, measured by ATLAS over a wide jet pTp_T range of 0.10.1-2.52.5 TeV, are well described within experimental uncertainties and consistent with PYTHIA simulations. This conclusion is further supported by direct comparisons with quark- and gluon-initiated jet distributions extracted via jet topics, though the propagated uncertainties from experimental data remain sizable.

Prepared for submission to JHEP

Introduction

QCD jets are collimated sprays of hadrons with fluctuating particle numbers, and their multiplicity distributions, that is, the probability distribution P(n)P(n) for producing nn particles, constitute an essential feature. Studying these distributions provides valuable insights into the dynamics of Quantum Chromodynamics (QCD), particularly the mechanisms of parton branching and hadronization in the high-virtuality regime. An expected simplification of multiplicity distributions in high-energy QCD jets arises from the concept of Koba–Nielsen–Olesen (KNO) scaling [1]. This scaling hypothesis states that at asymptotically high energy, the probability distribution of final-state multiplicities P(n)P(n) becomes self-similar when rescaled by the mean multiplicity nˉ\bar{n}, such that nˉP(n)\bar{n}P(n) depends only on the scaling variable x=n/nˉx = n/\bar{n} and approaches a universal function Ψ(x)\Psi(x), that is,

nˉP(n)=Ψ(n/nˉ)(1)\bar{n}P(n) = \Psi\left(n/\bar{n}\right) \tag*{(1)}

with ∫dx Ψ(x)=1\int dx\,\Psi(x) = 1. It was originally proposed on general grounds of scale invariance in quantum field theory [2, 3], and found a natural interpretation within the framework of QCD as a consequence of the cascade structure of parton showers in QCD jets [4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15].

Multiplicity distributions are expected to exhibit universality across different high-energy collision systems, including e+e−e^{+}e^{-}, Deep Inelastic Scattering (DIS), and pppp collisions, when dominated by high-energy jets. In particular, measurements of hadronic final states in e+e−e^{+}e^{-} annihilation over a wide energy range, from PETRA to LEP [16, 17, 18, 19], showed that charged-particle multiplicity distributions approximately collapse onto a universal KNO curve. Nonetheless, deviations from exact KNO scaling are expected, arising from the superposition of jets with different flavors and topologies [20, 21], as well as from the running of the strong coupling [14, 15]. Similar approximate KNO behavior is observed in DIS at HERA [22], where both theoretical predictions [23, 24] and experimental data support scaling in the current hemisphere at large momentum transfer. Early low-energy pppp collisions, from bubble chamber experiments to ISR data, also reported approximate scaling of multiplicity distributions [25, 26]. In high-energy hadronic collisions, however, LHC data [27, 28, 29, 30, 31] show that multiplicity distributions broaden with increasing energy and no longer follow KNO scaling. This breakdown is not surprising, since contributions from mixed event topologies of hard processes, such as quark- and gluon-initiated jets [13, 14], as well as soft processes, can lead to scaling violation (for recent discussions, see ref. [32] and references therein). KNO scaling is suggested to hold within QCD jets in pppp collisions [11, 12], while such universality is anticipated to reemerge when focusing specifically on high-energy QCD jets with proper quark–gluon discrimination [13, 14].

The KNO scaling functions are different for quark- and gluon-initiated jets, referred to as quark and gluon jets below, and have been calculated in perturbative QCD [6, 8, 9, 10, 13, 14, 15]. In the double logarithmic approximation (DLA), where the dominant soft and collinear emissions are resummed with strong ordering in both energy and angle, the forms of the scaling functions Ψ(x)\Psi(x) for gluon and quark jets can be evaluated explicitly [6, 8, 9]. These scaling functions are exact in the asymptotic high-energy (high-virtuality) limit, and KNO scaling within DLA has recently been confirmed to hold to a good approximation by numerically solving the multiplicity distributions in QCD jets for a broad jet pTp_T range relevant to LHC energies [13].

However, the DLA KNO scaling functions do not provide a reliable description of the measured multiplicity distributions [7, 10, 13, 14, 15]. To remedy this, the DLA for gluon jets was first improved by enforcing energy conservation in the parton cascade [10]. This modified double logarithmic approximation (MDLA) has recently been generalized to jets initiated by any source, including quarks and antiquarks, achieving notable phenomenological success [14, 15]. Meanwhile, the KNO scaling functions have also been extracted using PYTHIA [11, 13], showing good agreement with LHC measurements within experimental uncertainties [13, 33]. One of the main goals of this work is to compare quark and gluon KNO scaling functions obtained within MDLA and PYTHIA, and to explore their phenomenological implications.

Recent advances in jet physics and jet substructure at the LHC [34, 35] provide an unprecedented opportunity to investigate multiplicity fluctuations in QCD jets at very high transverse momenta, both theoretically and experimentally. A variety of methods for quark–gluon discrimination have been proposed [36, 37, 38, 39, 40, 41, 42, 43, 44] and applied in LHC analyses [33, 45, 46, 47]. If proven effective for multiplicity distributions, these techniques would enable direct experimental tests of the universality of the KNO scaling functions for quark and gluon jets, as illustrated using energy–correlation functions [38] in ref. [13]. Another central motivation of this paper is to investigate KNO scaling in QCD jets using jet topics [42, 43], which is well suited to extracting multiplicity distributions and can be implemented without additional theoretical input.

The rest of this paper is structured as follows. In § 2, we start with a review of the generating function method for multiplicity distributions within QCD jets, as well as a summary of the DLA results for parton multiplicity distributions and mean multiplicities. We then calculate the KNO scaling function for both quark and gluon jets within the MDLA, i.e., by incorporating energy conservation into the DLA, following refs. [6, 8, 10], as an alternative to the recent calculations in refs. [14, 15]. In § 3, we present a phenomenological study of inclusive charged-particle multiplicity distributions of two leading jets in pppp collisions at s=13\sqrt{s} = 13 TeV, combining the MDLA KNO scaling functions with leading-order cross sections and the calculations of mean multiplicities, and compare our theoretical results with ATLAS data from ref. [33]. As a consistency check, we also directly compare the MDLA KNO scaling functions for quark and gluon jets with those extracted from ATLAS measurements using topic modeling [42, 43], which is presented in § 4. We conclude in § 5 with a summary of results and outlook for future directions. Finally, for completeness, we collect the next-to-next-to-next-to-leading-order (N3^{3}LO) results for mean parton multiplicities in QCD jets from refs. [48, 49] in Appendix § 7, which are used in our phenomenological studies in § 3.

Multiplicity distributions and KNO scaling in QCD jets

In this section, we begin with a brief review, for the sole purpose of completeness in our subsequent discussion, of the method of generating functions (GFs) [4, 5, 6, 9, 50, 51] and the DLA results for parton multiplicity distributions within QCD jets [6, 13, 52, 53]. We then evaluate the KNO scaling functions within MDLA for both quark and gluon jets by building on the treatment of gluon jets presented in ref. [10], providing an alternative to the recent calculations in refs. [14, 15].

Review of the generating function method

The GF method serves as a powerful and convenient tool for describing the evolution of QCD jets produced in high-energy hard scatterings. To support the forthcoming discussion, we outline the basics of the method and summarize the DLA results of parton multiplicities within QCD jets here. For a more comprehensive discussion, interested readers are referred to refs. [9, 50, 51].

The definition of generating functions

For a QCD jet initiated by a parton aa at initial scale QQ, the GF is defined as

Za(u,Q)≡∑n=0∞unPa(n,Q)withZa(1,Q)=1,(2)Z_a(u,Q) \equiv\sum_{n=0}^{\infty} u^n P_a(n,Q) \quad\text{with} \quad Z_a(1,Q)=1, \tag*{(2)}

where the multiplicity probability distribution Pa(n,Q)P_a(n,Q) denotes the probability of finding nn particles. By successive differentiation of the GF, one can recover the probability distributions:

Pa(n,Q)=1n!∂n∂unZa(u,Q)∣u=0(3)P_a(n,Q)=\left.\frac{1}{n!}\frac{\partial^n}{\partial u^n}Z_a(u,Q)\right|_{u=0} \tag*{(3)}

Analogous to the integral form of the DGLAP equation, the GFs at next-to-leading logarithmic (NLL) order obey [50]

Za(u,Q)=Za(u,Q0)Δa(Q,Q0)+∫Q0QdQˉQˉΔa(Q,Q0)Δa(Qˉ,Q0)×∫dzαsπP^a→bc(z)Zb(u,zQˉ)Zc(u,(1−z)Qˉ),(4)\begin{aligned} Z_a(u,Q)={}&Z_a(u,Q_0)\Delta_a(Q,Q_0)+\int_{Q_0}^{Q}\frac{\mathrm{d}\bar{Q}}{\bar{Q}}\frac{\Delta_a(Q,Q_0)}{\Delta_a(\bar{Q},Q_0)}\\ &\times\int\mathrm{d}z\frac{\alpha_s}{\pi}\widehat{P}_{a\to bc}(z)Z_b(u,z\bar{Q})Z_c(u,(1-z)\bar{Q}), \tag*{(4)} \end{aligned}

where αs\alpha_s is evaluated at the scale z(1−z)Qˉz(1-z)\bar{Q} if the running of the strong coupling is taken into account, the repeated indices bb and cc run over all parton species, Q0<QQ_0<Q denotes another momentum scale, the splitting functions are given by

P^g→gg(z)=2CA[1−zz+z1−z+z(1−z)],\widehat{P}_{g\to gg}(z)=2C_A\left[\frac{1-z}{z}+\frac{z}{1-z}+z(1-z)\right],
P^g→qqˉ(z)=TF[z2+(1−z)2],\widehat{P}_{g\to q\bar{q}}(z)=T_F\left[z^2+(1-z)^2\right],
P^q→gq(z)=CF[1+(1−z)2z].(5)\widehat{P}_{q\to gq}(z)=C_F\left[\frac{1+(1-z)^2}{z}\right]. \tag*{(5)}

and the Sudakov form factors are defined as

Δa(Q,Q0)=exp⁡[−∑bc∫Q0QdQˉQˉ∫dzαsπP^a→bc(z)].(6)\Delta_a(Q,Q_0)=\exp\left[-\sum_{bc}\int_{Q_0}^{Q}\frac{\mathrm{d}\bar{Q}}{\bar{Q}}\int\mathrm{d}z\frac{\alpha_s}{\pi}\widehat{P}_{a\to bc}(z)\right]. \tag*{(6)}

Equivalently, by differentiating eq. (4) with respect to QQ, one obtains, at O(αs)O(\alpha_s) [51],

∂∂ln⁡QZa(u,Q)=∫dzαsπP^a→bc(z)[Zb(u,zQ)Zc(u,(1−z)Q)−Za(u,Q)].(7)\frac{\partial}{\partial\ln Q}Z_a(u,Q)=\int\mathrm{d}z\frac{\alpha_s}{\pi}\widehat{P}_{a\to bc}(z)\left[Z_b(u,zQ)Z_c(u,(1-z)Q)-Z_a(u,Q)\right]. \tag*{(7)}

The first term on the right-hand side of eq. (4), involving the form factor Δa\Delta_a, corresponds to the process in which the initial parton aa remains unchanged, while the second term accounts for the splitting process a→bca\to bc. When coherent branching (i.e., angular ordering) [9, 54] is incorporated into the second term, the formalism properly accounts for both soft and collinear enhancements. Denote the momenta of the daughter parton bb and its parent aa by kk and pp, respectively. In this case, the evolution variable is most conveniently defined as [50]

Q2=p2z(1−z)=k⊥2z2(1−z)2,(8)Q^2=\frac{p^2}{z(1-z)}=\frac{k_\perp^2}{z^2(1-z)^2}, \tag*{(8)}

where the daughter parton bb carries a longitudinal momentum fraction zz, and k⊥k_\perp represents the transverse momentum of the daughter parton.

Parton multiplicities in QCD jets within DLA

Focusing solely on soft and collinear logarithmic contributions, the splitting function simplifies to P^a→ga(z)≈2Ca/z\hat{P}_{a\to ga}(z) \approx2C_a/z. And the condition for coherent branching can be expressed as [9]

zQ=k0θ>k⊥>Q0,(9)zQ = k^{0}\theta> k_{\perp} > Q_{0}, \tag*{(9)}

where kk is the momentum of the emitted soft gluon, θ\theta is the angle relative to the preceding branching, and the momentum scale of the parent parton is given by Q=pθQ = p\theta, with pp denoting the parent parton’s momentum.

Within DLA, eq. (7) becomes

∂∂ln⁡QZa(u,Q)=Za(u,Q)ca∫dzzγ02[Zg(u,zQ)−1]withγ0=2Ncαsπ,(10)\frac{\partial}{\partial\ln Q}Z_a(u,Q)=Z_a(u,Q)c_a\int\frac{\mathrm{d}z}{z}\gamma_0^2\left[Z_g(u,zQ)-1\right]\quad\text{with}\quad\gamma_0=\sqrt{\frac{2N_c\alpha_s}{\pi}}, \tag*{(10)}

where cg=1c_g=1 for gluon jets and cq≡CF/CAc_q\equiv C_F/C_A for quark jets with the Casimir factors CA=NcC_A=N_c and CF=(Nc2−1)/(2Nc)C_F=(N_c^2-1)/(2N_c). It has the formal solution

Za(u,Q)=Za(u,Q0)exp⁡{∫Q0QdQˉQˉ∫dzz2αs(zQˉ)Caπ[Zg(u,zQˉ)−1]}.(11)Z_a(u,Q)=Z_a(u,Q_0)\exp\left\{\int_{Q_0}^{Q}\frac{\mathrm{d}\bar{Q}}{\bar{Q}}\int\frac{\mathrm{d}z}{z}\frac{2\alpha_s(z\bar{Q})C_a}{\pi}\left[Z_g(u,z\bar{Q})-1\right]\right\}. \tag*{(11)}

For hadron multiplicities, the GF necessarily includes non-perturbative contributions. In this section, we instead focus on parton multiplicities at the reference virtuality Q0Q_0, where the system contains a single parton. This sets the initial condition: Za(u,Q0)=uZ_a(u,Q_0)=u.1 Restricting to the DLA phase space in eq. (9), in terms of k⊥=zQˉk_{\perp}=z\bar{Q} and zz, one obtains [9]

Za(u,y)=uexp⁡{ca∫0ydyˉ (y−yˉ)γ02[Zg(u,yˉ)−1]},(12)Z_a(u,y)=u\exp\left\{c_a\int_0^y\mathrm{d}\bar{y}\,(y-\bar{y})\gamma_0^2\left[Z_g(u,\bar{y})-1\right]\right\}, \tag*{(12)}

where y≡ln⁡(Q/Q0)y\equiv\ln(Q/Q_0) and yˉ≡ln⁡(k⊥/Q0)\bar{y}\equiv\ln(k_{\perp}/Q_0). From the GFs given in eq. (12), we obtain the following recursive relation between Pa(n)P_a(n)’s [13]:

Pa(1,Q)=exp⁡{−ca∫0ydyˉ (y−yˉ)γ02},P_a(1,Q)=\exp\left\{-c_a\int_0^y\mathrm{d}\bar{y}\,(y-\bar{y})\gamma_0^2\right\},
Pa(n+1,Q)=ca∑k=1nknPa(n+1−k,Q)∫0ydyˉ (y−yˉ)γ02Pg(k,yˉ).(13)P_a(n+1,Q)=c_a\sum_{k=1}^{n}\frac{k}{n}P_a(n+1-k,Q)\int_0^y\mathrm{d}\bar{y}\,(y-\bar{y})\gamma_0^2P_g(k,\bar{y}). \tag*{(13)}

The mean multiplicity, denoted by nˉa\bar{n}_a, can also be derived by taking the first derivative of the GFs with respect to uu at u=1u=1:

nˉa(Q)=1+ca∫0ydyˉ (y−yˉ)γ02nˉg(yˉ),(14)\bar{n}_a(Q)=1+c_a\int_0^y\mathrm{d}\bar{y}\,(y-\bar{y})\gamma_0^2\bar{n}_g(\bar{y}), \tag*{(14)}

where the normalization condition of the GFs has been applied in the final step. Accordingly, one has

nˉq(Q)−1=ca[nˉg(Q)−1].(15)\bar{n}_q(Q)-1=c_a[\bar{n}_g(Q)-1]. \tag*{(15)}

Differentiating eq. (14) twice with respect to the evolution variable yy for gluon jets gives

∂2∂y2nˉg=γ02nˉg(16)\frac{\partial^2}{\partial y^2}\bar{n}_g=\gamma_0^2\bar{n}_g \tag*{(16)}

with the boundary conditions nˉg′(0)=0\bar{n}'_g(0)=0 and nˉg(0)=1\bar{n}_g(0)=1. This second-order ordinary differential equation admits the analytic solution: [9, 52, 53]

nˉg={cosh⁡(γ0y)for fixed couplingz1[I1(z1)K0(z2)+K1(z1)I0(z2)]for running coupling(17)\bar{n}_g = \begin{cases} \cosh(\gamma_0 y) & \text{for fixed coupling}\\ z_1\left[I_1(z_1)K_0(z_2)+K_1(z_1)I_0(z_2)\right] & \text{for running coupling} \end{cases} \tag*{(17)}

where z1≡Ay+λz_1 \equiv A\sqrt{y+\lambda} and z2≡Aλz_2 \equiv A\sqrt{\lambda}, with λ≡ln⁡(Q0/Λ)\lambda\equiv\ln(Q_0/\Lambda), A≡16Nc/β0A \equiv\sqrt{16N_c/\beta_0}, and β0=(11Nc−2nf)/3\beta_0=(11N_c-2n_f)/3. Alternatively, given Pa(n,Q)P_a(n,Q) numerically evaluated according to eq. (13), the mean parton multiplicities in quark and gluon jets can be obtained via:

nˉa(Q)=∑n=1∞nPa(n,Q).(18)\bar{n}_a(Q)=\sum_{n=1}^{\infty}nP_a(n,Q). \tag*{(18)}

The agreement of these two approaches has been verified in ref. [13].

KNO scaling within QCD jets

In the asymptotic limit Q→∞Q\to\infty, KNO scaling in multiplicity distributions within QCD jets naturally follows the definition of the GFs [4, 5, 6, 9, 51]. As previously verified in ref. [13], nˉa(Q)Pa(n,Q)\bar{n}_a(Q)P_a(n,Q) for partons within DLA, given in eq. (13), indeed converge to universal scaling functions for quark and gluon jets respectively for Q≫Q0Q\gg Q_0. However, they are quite different from those for hadrons in PYTHIA simulations and the LHC data. In this subsection, we evaluate the KNO scaling functions within MDLA for both quark and gluon jets, improving the DLA results [6, 8, 9] by imposing energy conservation [10, 14, 15].

Definition of KNO scaling functions for quark and gluon jets

Following refs. [6, 9], we first outline the definition of, and the method for evaluating, the KNO scaling functions for QCD jets, which holds exactly in the asymptotic limit Q→∞Q\to\infty. Setting u=exp⁡{−β/nˉa}u=\exp\{-\beta/\bar{n}_a\} in the definition of GFs in eq. (2) and taking the limit Q→∞Q\to\infty in ZaZ_a yields

Φa(β)≡lim⁡Q→∞Za(e−βnˉa,Q)=lim⁡Q→∞∑n=0∞e−βnnˉanˉa[nˉaPa(n,Q)]=∫0∞dx Ψa(x)e−βx=∑k=0∞(−β)kk!fa(k)(19)\Phi_a(\beta)\equiv\lim_{Q\to\infty}Z_a\left(e^{-\frac{\beta}{\bar{n}_a}},Q\right)=\lim_{Q\to\infty}\sum_{n=0}^{\infty}\frac{e^{-\beta\frac{n}{\bar{n}_a}}}{\bar{n}_a}\left[\bar{n}_aP_a(n,Q)\right] =\int_0^\infty dx\,\Psi_a(x)e^{-\beta x} =\sum_{k=0}^{\infty}\frac{(-\beta)^k}{k!}f_a^{(k)} \tag*{(19)}

where x≡n/nˉax\equiv n/\bar{n}_a, and

fa(k)≡∫0∞dx xkΨa(x)(20)f_a^{(k)}\equiv\int_0^\infty dx\,x^k\Psi_a(x) \tag*{(20)}

with

fa(0)=1=fa(1).(21)f_a^{(0)} = 1 = f_a^{(1)}. \tag*{(21)}

Accordingly, the asymptotic KNO scaling functions Ψa(x)\Psi_a(x) are given by the inverse Laplace transform of Φa(β)\Phi_a(\beta)

Ψa(n/nˉa)≡lim⁡Q→∞[nˉaPa(n,Q)]=∫dβ2πiΦa(β)eβnnˉa,(22)\Psi_a(n/\bar n_a) \equiv\lim_{Q\to\infty}[\bar n_a P_a(n,Q)] = \int\frac{d\beta}{2\pi i}\Phi_a(\beta)e^{\beta\frac{n}{\bar n_a}}, \tag*{(22)}

where the integral over β\beta runs parallel to the imaginary axis, with all singularities of Φa\Phi_a lying to the left in the complex plane. This definition thus extends the original KNO scaling in eq. (1) to jets initiated by specific parton species.

One way to evaluate the above KNO scaling functions is to boil it down to evaluating fa(k)f_a^{(k)}. In terms of multiplicity correlators:

na(k)(Q)≡∂k∂ukZa(u,Q)∣u=1=∑n=k∞n!(n−k)!Pa(n,Q),(23)n_a^{(k)}(Q) \equiv\left.\frac{\partial^k}{\partial u^k}Z_a(u,Q)\right|_{u=1} = \sum_{n=k}^{\infty}\frac{n!}{(n-k)!}P_a(n,Q), \tag*{(23)}

one has

Za(u,Q)=∑k=0∞(u−1)kk!na(k)(Q).(24)Z_a(u,Q)=\sum_{k=0}^{\infty}\frac{(u-1)^k}{k!}n_a^{(k)}(Q). \tag*{(24)}

From this expression of ZaZ_a, one obtains

Φa(β)=lim⁡Q→∞∑k=0∞(e−βnˉa(Q)−1)kk!na(k)(Q)=∑k=0∞(−β)kk!lim⁡Q→∞na(k)(Q)[nˉa(Q)]k.(25)\Phi_a(\beta)=\lim_{Q\to\infty}\sum_{k=0}^{\infty}\frac{\left(e^{-\frac{\beta}{\bar n_a(Q)}}-1\right)^k}{k!}n_a^{(k)}(Q)=\sum_{k=0}^{\infty}\frac{(-\beta)^k}{k!}\lim_{Q\to\infty}\frac{n_a^{(k)}(Q)}{[\bar n_a(Q)]^k}. \tag*{(25)}

Comparing with the power series expansion in eq. (19) which yields

fa(k)=lim⁡Q→∞na(k)(Q)[nˉa(Q)]k.(26)f_a^{(k)}=\lim_{Q\to\infty}\frac{n_a^{(k)}(Q)}{[\bar n_a(Q)]^k}. \tag*{(26)}

KNO scaling functions within MDLA

Similar to the evolution equation for the GFs within DLA in eq. (10), incorporating the approximation P^a→ga(z)≈2Ca/z\hat P_{a\to ga}(z)\approx2C_a/z along with energy conservation in the NLL evolution equation in eq. (7) yields

∂∂ln⁡QZa(u,Q)=ca∫dzzγ02[Zg(u,zQ)Za(u,(1−z)Q)−Za(u,Q)].(27)\frac{\partial}{\partial\ln Q}Z_a(u,Q)=c_a\int\frac{dz}{z}\gamma_0^2\left[Z_g(u,zQ)Z_a(u,(1-z)Q)-Z_a(u,Q)\right]. \tag*{(27)}

By differentiating both sides of this equation at u=1u=1, one has

∂na(m)(Q)∂ln⁡Q=ca∫dzzγ02[∑k=0mCmkng(k)(zQ)na(m−k)((1−z)Q)−na(m)(Q)],(28)\frac{\partial n_a^{(m)}(Q)}{\partial\ln Q}=c_a\int\frac{dz}{z}\gamma_0^2\left[\sum_{k=0}^{m}C_m^k n_g^{(k)}(zQ)n_a^{(m-k)}((1-z)Q)-n_a^{(m)}(Q)\right], \tag*{(28)}

where the binomial coefficient is denoted as

Cmk≡m!k!(m−k)!.(29)C_m^k\equiv\frac{m!}{k!(m-k)!}. \tag*{(29)}

As we are interested in the large QQ limit, na(m)(Q)n_a^{(m)}(Q) can be expressed in terms of fa(m)f_a^{(m)} and nˉa(Q)\bar n_a(Q) according to eq. (26) in the above equation. Following ref. [10], we consider only the corrections arising from energy conservation, while neglecting the effects of the running coupling and higher-order corrections to the anomalous dimension γ0\gamma_0. In this case, one can approximate the mean multiplicities as

nˉa(Q)≈ca(Q/Q0)γ0/2,(30)\bar n_a(Q) \approx c_a(Q/Q_0)^{\gamma_0/2}, \tag*{(30)}

following their fixed-coupling expressions in eq. (17).

Under the above approximation, for m≥2m \ge2, after integrating out zz on the right-hand side of eq. (28), one obtains

cam−1{m+caγ0[ψ(γ0m+1)−ψ(1)]}fa(m)−fg(m)m=γ0∑k=1m−1CmkΓ(γ0k)Γ(γ0(m−k)+1)Γ(γ0m+1)cam−kfg(k)fa(m−k),(31)c_a^{m-1}\left\{m+c_a\gamma_0\left[\psi(\gamma_0m+1)-\psi(1)\right]\right\}f_a^{(m)}-\frac{f_g^{(m)}}{m} =\gamma_0\sum_{k=1}^{m-1}C_m^k\frac{\Gamma(\gamma_0k)\Gamma(\gamma_0(m-k)+1)}{\Gamma(\gamma_0m+1)}c_a^{m-k}f_g^{(k)}f_a^{(m-k)}, \tag*{(31)}

where the digamma function ψ(z)≡Γ′(z)/Γ(z)\psi(z)\equiv\Gamma'(z)/\Gamma(z). Following ref. [10], we neglect the ψ\psi terms. Then, starting with fa(0)=fa(1)=1f_a^{(0)}=f_a^{(1)}=1, one can iteratively obtain, for m≥2m\ge2,

fg(m)=γ0mm2−1∑k=1m−1CmkΓ(γ0k)Γ(γ0(m−k)+1)Γ(γ0m+1)fg(k)fg(m−k),f_g^{(m)}=\frac{\gamma_0m}{m^2-1}\sum_{k=1}^{m-1}C_m^k\frac{\Gamma(\gamma_0k)\Gamma(\gamma_0(m-k)+1)}{\Gamma(\gamma_0m+1)}f_g^{(k)}f_g^{(m-k)},
fq(m)=cq1−mm2−1fg(m)+γ0∑k=1m−11mCmkΓ(γ0k)Γ(γ0(m−k)+1)Γ(γ0m+1)cq1−kfg(k)fq(m−k).(32)f_q^{(m)}=\frac{c_q^{1-m}}{m^2-1}f_g^{(m)}+\gamma_0\sum_{k=1}^{m-1}\frac{1}{m}C_m^k\frac{\Gamma(\gamma_0k)\Gamma(\gamma_0(m-k)+1)}{\Gamma(\gamma_0m+1)}c_q^{1-k}f_g^{(k)}f_q^{(m-k)}. \tag*{(32)}

The DLA results [6, 9] are recovered in the limit γ0→0\gamma_0\to0.

Given the expressions fa(m)f_a^{(m)} above, the KNO scaling functions Ψa(x)\Psi_a(x) can be presented using an expansion in Laguerre polynomials [8], which obey the relations

Lm(x)=∑k=0mCmk(−1)kk!xkand∫0∞e−xLn(x)Lm(x)=δmn.(33)L_m(x)=\sum_{k=0}^{m}C_m^k\frac{(-1)^k}{k!}x^k \qquad\text{and}\qquad \int_0^\infty e^{-x}L_n(x)L_m(x)=\delta_{mn}. \tag*{(33)}

That is, Ψa(x)\Psi_a(x) can be approximated as

Ψa(x)≈e−x∑m=0Nf^a(m)Lm(x),(34)\Psi_a(x)\approx e^{-x}\sum_{m=0}^{N}\hat f_a^{(m)}L_m(x), \tag*{(34)}

where NN denotes the truncation order of the expansion and according to the relations in eq. (33), the expansion coefficients are given by

f^a(m)=∫0∞dxLm(x)Ψa(x)=∑k=0mCmk(−1)kk!fa(m).(35)\hat f_a^{(m)}=\int_0^\infty dxL_m(x)\Psi_a(x)=\sum_{k=0}^{m}C_m^k\frac{(-1)^k}{k!}f_a^{(m)}. \tag*{(35)}

We observe that for N≥30N\ge30, the series converges reliably for both quark and gluon jets. All results presented below are evaluated with N=40N=40.

The left panel of Figure 1 verifies the validity of KNO scaling in DLA with a fixed αs\alpha_s, which is assumed in the preceding MDLA derivation. This is verified by directly evaluating Pa(n)P_a(n), as given in eq. (13) with Q0=0.5Q_0 = 0.5 GeV, over a broad pTp_T range from 0.1 to 2.5 TeV. With running coupling, the resulting curves of nˉaPa(n)\bar{n}_a P_a(n) all converge to some universal functions for quark and gluon jets, respectively, though with slight deviations from the asymptotic DLA scaling functions [13]. With fixed coupling, we find that nˉaPa(n)\bar{n}_a P_a(n) shows some dependence on the value of αs\alpha_s and the results with αs=0.2\alpha_s = 0.2 converge to the DLA scaling functions, as shown in this plot. Here, the asymptotic KNO functions within DLA are given by eq. (34) with fa(n)f_a^{(n)} evaluated in the limit γ0→0\gamma_0 \to0, which agree with the results of refs. [8, 10].

Figure 1. KNO scaling functions for quark and gluon jets. Left: Verification of KNO scaling within DLA by directly evaluating Pa(n)P_a(n) according to eq. (13) with Q0=0.5Q_0 = 0.5 GeV, using a fixed coupling αs=0.2\alpha_s = 0.2, over a pTp_T range from 0.1 to 2.5 TeV (solid), and comparing them to the DLA KNO scaling functions derived in the asymptotic limit Q→∞Q \to\infty (dashed). Within this pTp_T range, we observe that the curves of nˉaPa(n)\bar{n}_a P_a(n) for different pTp_T still show some dependence on the value of αs\alpha_s. With the chosen value, they all converge to the DLA KNO scaling functions. Right: Comparison of the MDLA KNO scaling functions evaluated from eq. (28) (solid) with the QCD-inspired forms in ref. [15] (dashed) for γ0=0.4=γ\gamma_0 = 0.4 = \gamma and cq=2/3c_q = 2/3.

The right panel of Figure 1 compares the MDLA scaling functions evaluated above with the QCD-inspired expressions recently proposed in ref. [15]. The comparison is shown for γ0=0.4\gamma_0 = 0.4 [10] (with γ=0.4\gamma= 0.4 in the results of ref. [15]) and cq=2/3c_q = 2/3 [15]. For both quark and gluon jets, including energy conservation significantly shifts the peak of the scaling functions toward larger values of xx. These significant discrepancies highlight the need to go beyond DLA to obtain reliable KNO scaling functions. Our results, obtained using Laguerre polynomial expansions, agree with the expression for gluon jets given in ref. [10], except in the x≪1x \ll1 region, where our method may yield negative values, like the DLA results in ref. [8]. Hence, for the remainder of the paper, we adopt the analytic form from ref. [10] for gluon jets. In comparison with recently suggested forms in ref. [15], our results for gluon jets are slightly different from that at x≃1x \simeq1, starting around the peak region, while the differences for quark jets are more pronounced. We find that this difference can be traced back to the fact that the coefficients fq(m)f_q^{(m)}, calculated iteratively above, no longer satisfy the relation within DLA: Φq(β)=[Φg(β/cq)]cq\Phi_q(\beta) = [\Phi_g(\beta/c_q)]^{c_q}, which is used to obtain the quark-jet KNO functions in refs. [14, 15].

Comparison with KNO scaling functions using PYTHIA

As shown by the results above, while the mean parton multiplicity and the parton multiplicity distribution functions exhibit a logarithmic dependence on Q0Q_0, the asymptotic KNO scaling functions remain infrared safe. Consequently, one may not expect logarithmically enhanced corrections when transitioning from the parton level to the hadron level. In this subsection, we compare the above results at the parton level with those for charged particles extracted in PYTHIA simulations [55], which exhibit KNO scaling to a very reasonable level in both quark and gluon jets [13].

In our PYTHIA simulations of pppp collisions [13], the event selection procedure follows the methodology in ref. [33]. Particles satisfying pT>0.5p_T > 0.5 GeV and ∣η∣<2.5|\eta| < 2.5 are used to reconstruct the two leading jets via the anti-ktk_t algorithm [56] with a radius parameter R=0.4R = 0.4, implemented in the FastJet package [57]. The jets are required to satisfy ∣η∣<2.1|\eta| < 2.1, and the leading and subleading jets must have a transverse momentum ratio pTlead/pTsublead<1.5p_T^{\mathrm{lead}}/p_T^{\mathrm{sublead}} < 1.5. The PYTHIA samples are generated using the A14 tune, which has been shown to provide a good description for charged-particle multiplicity distribution observed by ATLAS [33]. In our analysis, the charged particles from jet constituents are used to define the multiplicity distributions, while jet pTp_T is reconstructed from all particles inside the jet. In addition, quark and gluon jets are identified based on their spatial proximity to the initiating parton from the underlying 2→22 \to2 hard scattering event.

The left panel of Figure 2 compares our MDLA results with PYTHIA simulations that include both MPI and hadronization, facilitating comparison with ATLAS data in the next sections. We find that incorporating energy conservation indeed improves significantly the agreement between the KNO scaling functions and the PYTHIA results for both quark and gluon jets. Inspired by refs. [14, 15], we take cqc_q as a fitting parameter rather than fixing it to the DLA value CF/CAC_F/C_A. With γ0=0.43\gamma_0 = 0.43 and cq=0.8c_q = 0.8, the MDLA predictions achieve good agreement not only with PYTHIA simulations but also, as will be shown later, with the ATLAS measurements reported in ref. [33]. In details, we first determine γ0\gamma_0 by fitting the PYTHIA results over the entire pTp_T range for gluon jets. Keeping this value of γ0\gamma_0 fixed, we then extract cqc_q by fitting the PYTHIA results over the entire pTp_T range for quark jets. The fitted cqc_q turns out larger than the chosen value of 2/32/3 in ref. [15], reflecting quantitative differences between the two results.

Figure 2. KNO scaling functions for quark and gluon jets in PYTHIA and MDLA. Left: Charged-particle multiplicity distributions for quark and gluon jets in PYTHIA [13] (solid) versus MDLA results (dashed) with γ0=0.43\gamma_0 = 0.43 and cq=0.8c_q = 0.8. Right: Negative binomial distributions (NBD, solid), with k=5.4k = 5.4 for quark jets and k=8.5k = 8.5 for gluon jets (see eq. (36)) compared to the same MDLA results (dashed).

Note that setting the momentum scale in αs\alpha_s to Q=RpTQ = Rp_T, using LHAPDF [58], we obtain γ0\gamma_0 values in the range 0.41–0.51 across the plotted pTp_T interval.2 Our choice of γ0=0.43\gamma_0 = 0.43 indeed lies in this range although allowing for this running produces a broader spread than is seen in the PYTHIA results. Since the MDLA scaling functions with γ0=0.43\gamma_0 = 0.43 and cq=0.8c_q = 0.8 can very well describe the experimental data in ref. [33], we proceed with this phenomenological approach and defer a more refined analysis beyond DLA to future work.

Similar to the DLA results (see, e.g., ref. [8]), we find that the MDLA KNO functions are well approximated by negative binomial distributions (NBDs) of the form

ΨNB(x)=xk−1e−kxkkΓ(k),(36)\Psi^{NB}(x) = x^{k-1}e^{-kx}\frac{k^k}{\Gamma(k)}, \tag*{(36)}

where kk serves as a fit parameter (see also ref. [10]). The right panel of Figure 2 demonstrates this agreement, where the MDLA results are reproduced using k=5.4k = 5.4 for quark jets and k=8.5k = 8.5 for gluon jets, exhibiting only minor deviations.

Inclusive multiplicity distributions in pppp collisions at the LHC

In pppp collisions, distinguishing quark and gluon jets to study their multiplicity distributions is a nontrivial challenge [34, 35], which we defer to the next section. As an alternative, the validity of KNO scaling can be inferred through inclusive multiplicity measurements of QCD jets [12, 13, 14]. In this section, we compare our theoretical calculations, based on the MDLA scaling functions evaluated above, with ATLAS measurements of the inclusive multiplicity distributions of the two leading jets at s=13\sqrt{s} = 13 TeV [33].

Theoretical calculations

Inclusive multiplicity distributions inside jets in pppp collisions have been measured at the LHC [33, 59]. These inclusive distributions can be represented as a combination of contributions from quark and gluon jets [13, 14]:

P(n)=rqPq(n)+rgPg(n),nˉ=rqnˉq+rgnˉg,(37)P(n) = r_qP_q(n) + r_gP_g(n), \\ \bar{n} = r_q\bar{n}_q + r_g\bar{n}_g, \tag*{(37)}

where the coefficient

ra=dσa/dpTdσq/dpT+dσg/dpT(38)r_a = \frac{d\sigma_a/dp_T}{d\sigma_q/dp_T + d\sigma_g/dp_T} \tag*{(38)}

denotes the fraction of jets initiated by a parton of type aa at a given jet transverse momentum pTp_T, and dσa/dpTd\sigma_a/dp_T represents the corresponding differential cross section.

To establish a theoretical baseline, we begin with the leading-order (LO) differential cross section in pppp collisions, which is factorized as a convolution of parton distribution functions (PDFs) with the partonic hard-scattering cross sections, summed over all contributing subprocesses ab→cdab \to cd:

dσdpT=2pT∑a,b,c,d∫dycdyddxafa/p(xa,μ2)xbfb/p(xb,μ2)dσ^ab→cddt^,(39)\frac{d\sigma}{dp_T} = 2p_T \sum_{a,b,c,d} \int dy_c dy_d dx_a f_{a/p}(x_a,\mu^2) x_b f_{b/p}(x_b,\mu^2) \frac{d\hat{\sigma}_{ab\to cd}}{d\hat{t}}, \tag*{(39)}

where ycy_c and ydy_d are the rapidities of the final-state partons cc and dd that initiate the jets, and the longitudinal momentum fractions xax_a and xbx_b of the incoming partons are determined by

xa=12xT(eyc+eyd),xb=12xT(e−yc+e−yd)(40)x_a = \frac{1}{2}x_T(e^{y_c}+e^{y_d}), \qquad x_b = \frac{1}{2}x_T(e^{-y_c}+e^{-y_d}) \tag*{(40)}

with xT≡2pT/sx_T \equiv2p_T/\sqrt{s}. The PDFs fa/p(xa,μ2)f_{a/p}(x_a,\mu^2) and fb/p(xb,μ2)f_{b/p}(x_b,\mu^2) are evaluated at the factorization scale μ2=pT2\mu^2 = p_T^2. The partonic cross sections, denoted as dσ^ab→cd/dt^d\hat{\sigma}_{ab\to cd}/d\hat{t} and found in the standard textbooks such as ref. [60], are expressed in terms of the Mandelstam variables:

s^=xaxbs,t^=−xapTse−yc,u^=−xbpTseyc.(41)\hat{s} = x_a x_b s, \qquad\hat{t} = -x_a p_T \sqrt{s}e^{-y_c}, \qquad\hat{u} = -x_b p_T \sqrt{s}e^{y_c}. \tag*{(41)}

In our analysis, we explicitly separate the contributions from quark and gluon jet channels by examining each partonic subprocess ab→cdab \to cd and evaluating their corresponding partonic cross sections. This enables us to compute the differential cross sections dσq/dpTd\sigma_q/dp_T for quark jets and dσg/dpTd\sigma_g/dp_T for gluon jets, which in turn allows us to extract the quark and gluon fractions rqr_q and rgr_g, as defined in eq. (38). For multiplicity distributions, we utilize the MDLA results as presented in the previous section. Based on the theoretical framework outlined above, we investigate the mean multiplicities and multiplicity probability distributions within jets across a broad jet transverse momentum range of pT=0.1p_T = 0.1–2.52.5 TeV in pppp collisions at s=13\sqrt{s} = 13 TeV. Our theoretical predictions are to be compared with results from PYTHIA simulations and experimental measurements from ATLAS [33].

Mean charged-particle multiplicity

In this subsection, we briefly summarize the results of mean charged-particle multiplicities used in our studies of inclusive multiplicity distributions, as given in eq. (37). Specifically, we employ the DLA results with running coupling, given in eqs. (15) and (17), as well as the next-to-next-to-next-to-leading-order (N3^3LO) results from refs. [48, 49], which are summarized in appendix A (see also ref. [51] for a review). To facilitate a meaningful comparison between parton-level theoretical predictions and experimental charged-particle data in ref. [33], we make use of the Local Parton–Hadron Duality (LPHD) hypothesis [61, 62].

Within this approach, we introduce a normalization factor KLPHDK_{\mathrm{LPHD}} for the DLA results, defined as the ratio of the mean charged-particle multiplicity to the mean parton multiplicity. The corresponding normalization parameter in the N3^3LO results is denoted by KK.

To confront the theoretical calculations for the inclusive mean charged-particle multiplicity with the measurements in ref. [33], we first calculate the fractions of quark and gluon jets at LO using eqs. (38) and (39). Here, we consider two leading jets with ∣η∣<2.1|\eta| < 2.1 over the pTp_T range 0.1–2.5 TeV in pppp collisions at s=13\sqrt{s} = 13 TeV. The PDFs are taken from the CT18NLO via LHAPDF [58]. Using these extracted jet fractions, we then fix the parameters in the theoretical calculations by fitting to the experimental data for the inclusive mean charged-particle multiplicity in jets with radius R=0.4R = 0.4.

The left panel of Figure 3 compares the theoretical results of the inclusive mean multiplicity with ATLAS data and PYTHIA simulations. The DLA mean multiplicities depend on two parameters, Q0Q_0 and KLPHDK_{\mathrm{LPHD}}. We fix Q0=0.5Q_0 = 0.5 GeV, to be consistent with the discussions in § 2.2.2 and ref. [13], and determine KLPHDK_{\mathrm{LPHD}} through a χ2\chi^2 fit to the ATLAS data [33], obtaining KLPHD=0.80K_{\mathrm{LPHD}} = 0.80 with χ2/d.o.f.=23.1/13=1.78\chi^2/\mathrm{d.o.f.} = 23.1/13 = 1.78.3 For the N3^3LO results, the two parameters Q0Q_0 and KK are both determined via fits to the ATLAS data, yielding K=0.0353K = 0.0353 and Q0=0.053Q_0 = 0.053 GeV with χ2/d.o.f.=0.904/12=0.075\chi^2/\mathrm{d.o.f.} = 0.904/12 = 0.075. As shown in the plot, the N3^3LO predictions provide an impressively accurate description of the data, while the DLA results remain reasonable but tend to underestimate nˉch\bar{n}_{\mathrm{ch}} at low pTp_T and overestimate it at high pTp_T. The PYTHIA simulations, with MPI and hadronization included, also reproduce the measurements well, aside from a mild overshoot at high pTp_T.

Figure 3. Mean charged-particle multiplicity within QCD jets at the LHC. Left: Inclusive mean charged-particle multiplicity as a function of jet transverse momentum pTp_T, comparing ATLAS data [33], PYTHIA simulations, DLA, and N3^3LO results. Right: Mean charged-particle multiplicity in quark and gluon jets from PYTHIA, DLA, and N3^3LO predictions. For the DLA results, KLPHD=0.80K_{\mathrm{LPHD}} = 0.80 is obtained by fitting the inclusive mean multiplicity data shown in the left panel with Q0=0.5Q_0 = 0.5 GeV held fixed, while for the N3^3LO results K=0.0353K = 0.0353 and Q0=0.053Q_0 = 0.053 GeV are both determined from fits to the same data.

With the parameters fixed as above, the right panel of Figure 3 compares the mean charged-particle multiplicities in quark and gluon jets from DLA, N3^3LO, and PYTHIA. Overall, in comparison with the DLA curves, the N3^3LO results get closer to the PYTHIA results (the corresponding experimental results are discussed in the next section). Figure 4 shows the corresponding ratio of mean multiplicities in gluon and quark jets. Consistent with the behavior observed in the right panel of Figure 3, the N3^3LO ratio lies noticeably below the DLA prediction. In the next subsection, we employ both the DLA and N3^3LO mean multiplicities to investigate the sensitivity of inclusive multiplicity distributions to these mean multiplicities.

Figure 4. Ratio of mean multiplicities between gluon and quark jets. In the case of DLA, this ratio is not constant at the expected value of CA/CF=9/4C_A/C_F = 9/4, indicated by the horizontal dashed line, because the leading partons are included in the mean multiplicity given in eq. (15).

Inclusive multiplicity distributions

In this subsection, we present a detailed investigation of inclusive multiplicity distributions in two leading jets produced in pppp collisions at s=13\sqrt{s} = 13 TeV. Based on eqs. (3.1) and (3.2), and on the assumption that KNO scaling holds separately for quark and gluon jets, the inclusive multiplicity distributions in KNO form are expressed as

nˉP(n)=nˉ[rqPq(n)+rgPg(n)]=rqnˉnˉqΨq(xnˉ/nˉq)+rgnˉnˉgΨg(xnˉ/nˉg),(42)\bar{n}P(n) = \bar{n}[r_qP_q(n) + r_gP_g(n)] = r_q\frac{\bar{n}}{\bar{n}_q}\Psi_q(x\bar{n}/\bar{n}_q) + r_g\frac{\bar{n}}{\bar{n}_g}\Psi_g(x\bar{n}/\bar{n}_g), \tag*{(42)}

where we have used the relation n=xnˉ=xqnˉq=xgnˉgn = x\bar{n} = x_q\bar{n}_q = x_g\bar{n}_g. Here, the fractions rqr_q and rgr_g are determined from the LO cross section, as in the mean multiplicity calculations. For the KNO scaling functions Ψa\Psi_a, we adopt the MDLA results with γ0=0.43\gamma_0 = 0.43 and cq=0.8c_q = 0.8, as described in § 2.2.2, since the DLA predictions clearly fail to describe the measurements reported in [33]. The remaining relevant quantity is the ratio nˉg/nˉq\bar{n}_g/\bar{n}_q appearing in nˉ/nˉa\bar{n}/\bar{n}_a.

Figure 5 presents our theoretical predictions for charged-particle multiplicity distributions in KNO form: nˉchP(nch)\bar{n}_{ch}P(n_{ch}). For the ratio of mean multiplicities in gluon and quark jets, we employ both the DLA and N3N^{3}LO results, shown in figure 4. The differences between these predictions can be understood from eq. (42) and from the fact that the N3N^{3}LO multiplicity ratio is smaller than that in DLA across the entire pTp_T range. At low pTp_T, gluon jets dominate. When the contribution from quark jets can be neglected, the DLA curve is expected to be broader, lower, and shifted to the right relative to the N3N^{3}LO curve. This is because the coefficient in front of Ψg\Psi_g in the last line of eq. (42), proportional to nˉ/nˉg=rg+rqnˉq/nˉg\bar{n}/\bar{n}_g = r_g + r_q\bar{n}_q/\bar{n}_g, which also enters the argument of Ψg\Psi_g, decreases as nˉg/nˉq\bar{n}_g/\bar{n}_q increases. Conversely, at high pTp_T, quark jets dominate. When gluon-jet contributions are negligible, the DLA curve is expected to become narrower, higher, and shifted to the left relative to the N3N^{3}LO curve, since the coefficient of the quark-jet KNO scaling function, proportional to nˉ/nˉq=rq+rgnˉg/nˉq\bar{n}/\bar{n}_q = r_q + r_g\bar{n}_g/\bar{n}_q, increases with nˉg/nˉq\bar{n}_g/\bar{n}_q, which also affects the argument of Ψq\Psi_q. The eight panels of the figure, arranged from top left with 0.1 TeV<pT<0.2 TeV0.1\ \mathrm{TeV} < p_T < 0.2\ \mathrm{TeV} to bottom right with 2.0 TeV<pT<2.5 TeV2.0\ \mathrm{TeV} < p_T < 2.5\ \mathrm{TeV}, quantitatively illustrate the evolution between these two extremes.

Figure 5. Investigation of KNO scaling via inclusive multiplicity distributions in pp collisions. Shown here are our theoretical results using the MDLA KNO scaling functions (ΨMDLA\Psi_{\mathrm{MDLA}}), the LO cross section (σLO\sigma_{\mathrm{LO}}), and the ratios of mean multiplicities between gluon and quark jets in DLA (nˉDLA\bar{n}_{\mathrm{DLA}}) and N3^{3}LO (nˉN3LO\bar{n}_{\mathrm{N^{3}LO}}), compared with the ATLAS measurements of nˉchP(nch)\bar{n}_{\mathrm{ch}}P(n_{\mathrm{ch}}) reported in ref. [33] across eight pTp_T bins.

In Figure 5, our theoretical predictions are also compared with the measurements reported in ref. [33]. The results obtained with the N3N^{3}LO mean multiplicities show very good agreement with the experimental data across all reported pTp_T bins. In contrast, predictions using DLA mean multiplicities provide a good description only in the range 0.1 TeV<pT<0.2 TeV0.1\ \mathrm{TeV} < p_T < 0.2\ \mathrm{TeV}, as shown in the top left panel. For the remaining pTp_T bins, clear discrepancies appear, even when accounting for the relatively large uncertainties and coarse binning (Δnch=4\Delta n_{ch} = 4).

Figure 6 presents the same theoretical results as in Figure 5, now divided by the corresponding theoretical inclusive mean multiplicities and compared directly with the P(nch)P(n_{ch}) measurements from ref. [33]. The predictions with N3N^{3}LO mean multiplicities continue to agree very well with the experimental data across the entire pTp_T range, as the N3N^{3}LO inclusive mean multiplicity provides a very good description of the ATLAS measurements, as shown in the left panel of Figure 3. In contrast, results obtained with DLA mean multiplicities again show less accurate agreement, though with a noticeable improvement for 2.0 TeV<pT<2.5 TeV2.0\ \mathrm{TeV} < p_T < 2.5\ \mathrm{TeV} (and a deterioration for 0.1 TeV<pT<0.2 TeV0.1\ \mathrm{TeV} < p_T < 0.2\ \mathrm{TeV}). This behavior reflects the deviations of the DLA inclusive mean multiplicity from the data in the low- and high-pTp_T bins.

Figure 6. Inclusive multiplicity distributions in pp collisions. Our theoretical predictions, shown in comparison to the P(nch)P(n_{\mathrm{ch}}) measurements reported in ref. [33], are the same as those in Figure 5, divided by the corresponding theoretical results of the inclusive mean multiplicity. The figure also displays the PYTHIA results, with a bin width of Δnch=4\Delta n_{\mathrm{ch}} = 4 in each panel.

The same figure also shows comparisons with PYTHIA simulations. The PYTHIA P(nch)P(n_{ch}) distributions are presented using the same bin width of Δnch=4\Delta n_{ch} = 4 as in the ATLAS measurements. As shown in this figure, our theoretical predictions are consistent with the PYTHIA results across all pTp_T bins. The agreement with PYTHIA is expected, as the values of γ0\gamma_0 and cqc_q were determined by fitting to PYTHIA results in ref. [13], as shown in Figure 2. In turn, the consistency between our theoretical predictions and experimental measurements validates our approach of fixing γ0\gamma_0 and cqc_q based on PYTHIA simulations, although a more sophisticated determination of these parameters could be achieved by fitting directly to experimental data.

Thus, the agreement of our theoretical results with the ATLAS measurements provides

further evidence that the measurements of ref. [33] can be consistently described in terms of two distinct KNO scaling functions, as presented within MDLA in § 2.2.2, in line with our earlier observation using PYTHIA in ref. [13]. Nonetheless, our theoretical calculations can be further refined, as it relies on the LO cross section, neglects KNO scaling violations induced by the running coupling (see ref. [14] for a discussion at large nch/nˉchn_{\mathrm{ch}}/\bar{n}_{\mathrm{ch}}) and treats cqc_q as a free fitting parameter. A more systematic analysis, which would require improved theoretical inputs for each quantity involved, as discussed at the beginning of this subsection, is deferred to future work.

Multiplicity distributions in quark and gluon jets versus jet topics

As discussed above, the analysis of inclusive multiplicity distributions requires additional theoretical inputs, such as rar_a and nˉa\bar{n}_a, besides the KNO scaling functions, which makes establishing KNO scaling less straightforward. A more direct investigation of KNO scaling in QCD jets can be carried out alongside quark–gluon discrimination using jet substructure techniques (see refs. [34, 35] for recent reviews). The feasibility of this approach has been demonstrated with PYTHIA simulations in ref. [13], where energy–correlation functions (ECFs) [38] are employed to probe KNO scaling. However, because ECF-based studies select jets within a constrained phase space, the MDLA results presented in § 2 are not directly applicable. In this section, we instead compare the MDLA KNO scaling functions for quark and gluon jets with results obtained via the topic modeling method [42, 43] applied to the two jet samples measured by ATLAS in ref. [33].

In ref. [33], the two selected leading jets are separated into two samples based on the pseudorapidities of the two jets: the more forward jet sample and the more central jet sample. The charged-particle multiplicity distributions with a bin width of Δnch=4\Delta n_{\mathrm{ch}} = 4 are measured for both jet samples, denoted as hifh_i^f and hich_i^c, where ii labels the nchn_{\mathrm{ch}} bins. According to topic modeling [42, 43], the multiplicity distributions in quark and gluon jets can be approximated by the multiplicity distributions for topics 1 (T1T_1) and 2 (T2T_2):

hiT1=hif−min⁡j{hjf/hjc}hic1−min⁡j{hjf/hjc},hiT2=hic−min⁡j{hjc/hjf}hif1−min⁡j{hjc/hjf}.(43)h_i^{T_1} = \frac{h_i^f - \min_j\{h_j^f/h_j^c\}h_i^c}{1-\min_j\{h_j^f/h_j^c\}}, \qquad h_i^{T_2} = \frac{h_i^c - \min_j\{h_j^c/h_j^f\}h_i^f}{1-\min_j\{h_j^c/h_j^f\}}. \tag*{(43)}

As a test of this method for discriminating the quark and gluon jets, we also generate two jet samples in the same way using PYTHIA simulations.

In our data analysis, we do not retain all nchn_{\mathrm{ch}} bins. For the PYTHIA samples, we impose the condition hjf,c>N/Nh_j^{f,c} > \sqrt{N}/N (with NN the sample size) to exclude bins dominated by statistical fluctuations. For the experimental data, we discard bins with large uncertainties. Defining ra/b≡hja/hjbr_{a/b} \equiv h_j^a/h_j^b, where aa and bb correspond to two different jet samples, the propagated uncertainty of ra/br_{a/b} from the statistical and systematic errors in hjah_j^a and hjbh_j^b is given by

σra/b2=1(hjb)2σhja2+(hja)2(hjb)4σhjb2,(44)\sigma_{r_{a/b}}^2 = \frac{1}{(h_j^b)^2}\sigma_{h_j^a}^2 + \frac{(h_j^a)^2}{(h_j^b)^4}\sigma_{h_j^b}^2, \tag*{(44)}

where σoj2\sigma_{o_j}^{2} represents the variance (squared uncertainty) associated with a data point ojo_j. Throughout our discussion, uncertainties are treated as uncorrelated. When evaluating κa/b≡min⁡j{hja/hjb}\kappa_{a/b} \equiv\min_j\{h_j^a/h_j^b\}, only those nchn_{ch} bins are retained for which the total relative uncertainty, calculated as the square root of the sum of the squares of the propagated statistical and systematic errors, sufficiently small. We find that one obtains stable results when requiring:

σra/bra/b<25%.(45)\frac{\sigma_{r_{a/b}}}{r_{a/b}} < 25\%. \tag*{(45)}

This threshold can be lowered, for example, to 15%, but using an even smaller value may lead to deviations from the results presented below in certain pTp_T bins, highlighting the need for improved experimental precision. Conversely, using a larger threshold can result in κa/b\kappa_{a/b} being determined randomly by nchn_{ch} bins with very small probabilities (<0.005< 0.005), producing unreliable results.

Figure 7 compares the MDLA results, divided by nˉa\bar{n}_a, with the multiplicity distributions for T1T_1 and T2T_2 extracted from ATLAS data, as well as the corresponding PYTHIA results. Here, nˉq\bar{n}_q and nˉg\bar{n}_g are taken as the mean multiplicities for T1T_1 and T2T_2, respectively. To validate the topic modeling method, we first compare the charged-particle distributions of quark and gluon jets identified through the hard processes in PYTHIA, as implemented in [13], with those obtained via topic modeling applied to PYTHIA-generated jet samples. Overall, the two results are consistent across all pTp_T bins, though the degree of agreement varies among different pTp_T ranges. For example, in the range 0.3 TeV<pT<0.4 TeV0.3\ \mathrm{TeV} < p_T < 0.4\ \mathrm{TeV}, as shown in the top left panel, when restricting to nchn_{ch} bins with probability above 0.01 to eliminate those dominated by statistical uncertainties, the gluon jet results agree at the percent level, and for quark jets only a single nchn_{ch} bin shows a relative deviation exceeding 10%.

Figure 7. Charge-particle multiplicity distributions in quark (topic 1) and gluon (topic 2) jets. The plots compare the MDLA results (dashed) with the multiplicity distributions for topics 1 and 2 extracted from ATLAS data [33] (dots). In the experimental results, the error bars indicate statistical uncertainties, while the shaded bands correspond to systematic uncertainties. For comparison, predictions from PYTHIA are also shown, where quark and gluon jets are identified according to the underlying hard processes [13] (dotted) and via the topic modeling approach [42, 43] (solid).

Applying the same procedure to the two jet samples from ATLAS by imposing the selection criteria in eq. (45), we obtain the multiplicity distributions for the jet topics, shown as data points with both statistical errors (error bars) and systematical errors (shaded bands) in Figure 7. The zeros of the extracted distributions for the topics indicate which nchn_{ch} bin determines the minimum value of κa/b\kappa_{a/b}. For T2T_2, these zeros do not occur in the outermost nchn_{ch} bins, reflecting the impact of experimental uncertainties and coarse binning (Δnch=4\Delta n_{ch} = 4), which can be expected to account for the rise of the central values of P(nch)P(n_{ch}) in the smaller nchn_{ch} bins. Here, error propagation is performed using the standard formula; for example, for the T1T_1 distribution,

σhi2 T1=1(1−κf/c)2σhif2+κf/c2(1−κf/c)2σhic2+(hif−hic)2(1−κf/c)4σκf/c2.(46)\sigma_{h_i}^{2\,T_1} = \frac{1}{(1-\kappa_{f/c})^2}\sigma_{h_i^f}^{2} + \frac{\kappa_{f/c}^{2}}{(1-\kappa_{f/c})^2}\sigma_{h_i^c}^{2} + \frac{(h_i^f-h_i^c)^2}{(1-\kappa_{f/c})^4}\sigma_{\kappa_{f/c}}^{2}. \tag*{(46)}

Large uncertainties appear in the very low- and high-pTp_T regions, where the two jet samples, dominated by gluon and quark jets, respectively, become increasingly similar (i.e., κf/c→1\kappa_{f/c} \to1), which amplifies the propagated errors according to the above expression.

As shown in Figure 7, the extracted jet-topic distributions from the ATLAS samples are consistent with the PYTHIA results once uncertainties are taken into account. Notably, for 0.9 TeV<pT<1.0 TeV0.9\ \mathrm{TeV} < p_T < 1.0\ \mathrm{TeV}, the extracted distributions from experimental data and PYTHIA,

whether topic modeling is used or not, are consistent, differing from the results exemplified in ref. [33], which we attribute to our exclusion of nchn_{ch} bins with large uncertainties. At higher pTp_T (see the bottom right panel), the agreement between our experimental extractions and the PYTHIA results deteriorates, as the central values of the two ATLAS jet samples become more alike than those in PYTHIA.

As shown in Figure 8, the mean multiplicities extracted from ATLAS jet samples via topic modeling differ from those reported in ref. [33] using the same method, due to the exclusion of nchn_{ch} bins with large uncertainties. In contrast, they agree with the results obtained via the alternative method using the nominal quark and gluon fractions from the PYTHIA simulation provided by ATLAS in ref. [33] and are consistent with the PYTHIA predictions within propagated uncertainties across the full pTp_T range, although the central values lie below the PYTHIA results at higher pTp_T for gluon jets. Here, the large uncertainties for topics 1 and 2 at both low and high pTp_T arise from the near-similarity of the two jet samples, as mentioned above.

Figure 8. Mean charged-particle multiplicities in quark (topic 1) and gluon (topic 2) jets. The figure shows the average multiplicities for topics 1 and 2 extracted by applying topic modeling to the ATLAS more-forward and more-central jet samples (dots; statistical errors shown as error bars, systematic uncertainties as shaded bands), compared with the ATLAS results using the nominal quark and gluon fractions from the PYTHIA simulation (triangles) in ref. [33], with experimental uncertainties omitted, as well as with PYTHIA results from ref. [13] (dotted) and PYTHIA results obtained via topic modeling (solid).

Overall, the MDLA scaling functions (with γ0=0.43\gamma_0 = 0.43 and cq=0.8c_q = 0.8) presented in § 2.2.2 show good agreement with the multiplicity distributions extracted from ATLAS data using jet topics, within the propagated experimental uncertainties. A notable discrepancy appears for gluon jets at low nchn_{ch} in the 1.6 TeV <pT<< p_T < 2.0 TeV bin (see the bottom right panel of Figure 7). This is consistent with the mean multiplicities in Figure 8, where the central values of nˉg\bar{n}_g at high pTp_T are smaller than those predicted by PYTHIA. Since Pg(n)=Ψg(n/nˉg)/nˉgP_g(n) = \Psi_g(n/\bar{n}_g)/\bar{n}_g, this shift explains why the MDLA curve lies higher and more to the left relative to PYTHIA and the experimental results. We have also verified that, when nˉg\bar{n}_g from PYTHIA is used, the MDLA results agree very well with both the ATLAS extractions at large nchn_{\mathrm{ch}} and the PYTHIA predictions.

As summarized in Figure 9, the extracted multiplicity distributions for the two jet topics exhibit reasonably distinct scaling behavior. For the PYTHIA results, they are consistent with the distributions extracted in ref. [13], and the MDLA predictions provide a good description when the slight spread in PYTHIA results is neglected. For the extracted ATLAS results, the central values also display similar scaling, with quark jets (Topic 1) agreeing better with the MDLA results than gluon jets (Topic 2). While caution is required when drawing final conclusions due to experimental uncertainties and coarse binning as explicitly shown in Figure 7 and Figure 8, these observations demonstrate that jet topic modeling offers a promising approach for extracting multiplicity distributions and supporting studies of KNO scaling.

Figure 9. KNO scaling in jet topics. Left: Multiplicity distributions extracted from the two ATLAS jet samples [33] using topic modeling, with propagated experimental uncertainties (see Figure 7) omitted for clarity. Right: Corresponding distributions from PYTHIA jet samples. In both panels, the bin width is nch=4n_{\mathrm{ch}}=4, and the results are compared with the MDLA scaling functions (dashed).

Summary and outlook

Using an expansion in Laguerre polynomials [8] and following the calculations for gluon jets in ref. [10], we obtain the KNO scaling functions for quark and gluon jets by incorporating energy conservation into the Double Logarithmic Approximation (DLA) of the generating function evolution equation. The resulting modified DLA (MDLA) expressions deviate markedly from the DLA results, supporting earlier conclusions that the latter fail to provide a reliable description of experimental data [7, 10, 13, 14, 15], although recent DLA-based studies have successfully explained the observed KNO scaling in DIS [23, 24]. For gluon jets, our results are consistent with ref. [10], while for both quark and gluon jets they exhibit quantitative differences compared with the recently suggested QCD-inspired expressions in ref. [15].

The MDLA KNO scaling function for gluon jets depends solely on the anomalous dimension γ0\gamma_0, while the quark KNO scaling function additionally depends on the ratio of mean multiplicities of quark and gluon jets in the asymptotic high-virtuality limit, cqc_q.

Inspired by recent studies in refs. [14, 15], we treat cqc_q as a fitting parameter in our analysis of jets in pp collisions, rather than using the DLA value of CF/CAC_F/C_A. We find that by fixing γ0\gamma_0 and cqc_q, the MDLA results can very well reproduce the approximate KNO scaling functions for both quark and gluon jets in the two leading jets of pp collisions at s=13\sqrt{s}=13 TeV, as extracted using PYTHIA for a broad jet pTp_T range of 0.1–2.5 TeV in ref. [13]. Furthermore, we observe that combining the MDLA scaling functions with the leading-order cross section and the N3^3LO mean multiplicities in refs. [48, 49] provides a very good description of the inclusive charged-particle multiplicity distributions measured by ATLAS [33].

Describing the inclusive multiplicity distributions requires additional theoretical input, such as the fraction of jets initiated by each parton type and mean multiplicities. This extra dependence may introduce uncertainties when evaluating the validity of KNO scaling. To address this and as a consistency check, we utilize jet substructure techniques to distinguish quark and gluon jets. Specifically, we apply the topic modeling method [42, 43] to two jet samples provided by ATLAS [33] to extract multiplicity distributions for jet topics, serving as an approximation for quark and gluon jets. We find that these results are consistent with our MDLA results, although the experimental uncertainties remain significant.

Several aspects of our study could be further improved. First, a complete calculation of the KNO scaling functions for quark and gluon jets beyond the DLA approximation is still lacking. This is important because, although MDLA only incorporates energy conservation compared to DLA, it already introduces significant modifications. Moreover, describing the high-multiplicity tail observed by ATLAS [33], where hadronic multiplicities exceed the average by a factor of 3 or more, would require detailed knowledge of hard jet constituents [14], which lies beyond both DLA and MDLA. A key issue to address beyond DLA is how to perform a direct calculation using the same jet definition (e.g., the anti-ktk_t algorithm) employed in experiments. For inclusive multiplicity distributions, higher-order dijet cross sections are available [63]. In addition, the discrepancies between theoretical predictions and the measured mean multiplicities need to be resolved to achieve better agreement. One promising approach for high-precision calculations involves infrared-safe multiplicity definitions [64, 65], although multiplicity distributions within this approach are yet to be studied. Notably, KNO scaling functions are infrared safe within both DLA and MDLA, and parton-level results have been used for comparisons with experimental data in this work (see also refs. [14, 15]). At the same time, non-perturbative effects are observed in PYTHIA simulations [11, 13], highlighting the importance of clarifying the role of hadronization effects [61, 66]. Another promising avenue is to explore how studies of multiplicity distributions in quark and gluon jets can benefit from complementary methods of quark–gluon discrimination [36, 37, 39, 40, 41, 44]. On the experimental side, more precise measurements are needed, and the nchn_{\rm ch} binning should be refined.

Acknowledgments

X.D. thanks Dr. Wen-Ya Wu for valuable discussions regarding the treatment of experimental data. This work is supported by the European Research Council under project ERC-2018-ADG-835105 YoctoLHC; by Maria de Maez tu excellence unit grant CEX2023001318-M and projects PID2020-119632GB-I00 and PID2023-152762NB-I00 funded by MI-CIU/AEI/10.13039/501100011033; and by ERDF/EU. It has received funding from Xunta de Galicia (CIGUS Network of Research Centres). X.D. is supported by the China Scholarship Council under Grant No. 202306100165. L.C. is supported by the Marie Skłodowska-Curie Actions Postdoctoral Fellowships under Grant No. 101210595. X.D. and G.M. are supported by the National Natural Science Foundation of China under Grants No. 12147101, No. 12547102, No. 12325507, the National Key Research and Development Program of China under Grant No. 2022YFA1604900, and the Guangdong Major Project of Basic and Applied Basic Research under Grant No. 2020B0301030008. B.W. acknowledges the support of the Ramón y Cajal program with the Grant No. RYC2021-032271-I and the support of Xunta de Galicia under the ED431F 2023/10 project.

Mean multiplicities at N3LO

To clarify the role of the parameters appearing in the N3^{3}LO results [48, 49], we provide a summary in this appendix. The mean multiplicity for gluon jets takes the form

nˉg=Ky−a1C2exp⁡[2Cy+δg(y)](47)\bar{n}_{g}=K y^{-a_{1}C^{2}}\exp[2C\sqrt{y}+\delta_{g}(y)] \tag*{(47)}

where KK is the normalization constant, y≡ln⁡(Q/Q0)y\equiv\ln(Q/Q_{0}) (as defined in eq. (12)), C=4Nc/β0C=\sqrt{4N_{c}/\beta_{0}}, and

δg(y)=Cy{2a2C2+β1β02[ln⁡(2y)+2]}+C2y{a3C2−a1β1β02[ln⁡(2y)+1]}.(48)\delta_{g}(y)=\frac{C}{\sqrt{y}}\left\{2a_{2}C^{2}+\frac{\beta_{1}}{\beta_{0}^{2}}[\ln(2y)+2]\right\}+\frac{C^{2}}{y}\left\{a_{3}C^{2}-\frac{a_{1}\beta_{1}}{\beta_{0}^{2}}[\ln(2y)+1]\right\}. \tag*{(48)}

Here, the QCD β\beta-function coefficients are given by β0=(11Nc−2nf)/3\beta_{0}=(11N_{c}-2n_{f})/3, β1=[17Nc2−nf(5Nc+3CF)]/3\beta_{1}=[17N_{c}^{2}-n_{f}(5N_{c}+3C_{F})]/3. Analogously, the mean multiplicity for quark jets takes the form

nˉq=Kr0y−a1C2exp⁡[2Cy+δq(y)](49)\bar{n}_{q}=\frac{K}{r_{0}}y^{-a_{1}C^{2}}\exp[2C\sqrt{y}+\delta_{q}(y)] \tag*{(49)}

with

r0=NcCF,δq(y)=δg(y)+Cyr1+C2y(r2+r122).(50)r_{0}=\frac{N_{c}}{C_{F}},\qquad\delta_{q}(y)=\delta_{g}(y)+\frac{C}{\sqrt{y}}r_{1}+\frac{C^{2}}{y}\left(r_{2}+\frac{r_{1}^{2}}{2}\right). \tag*{(50)}

The coefficients ai,ria_{i},r_{i} depend on the number of flavors nfn_{f} and are tabulated in refs. [49]. In our analysis, these coefficients are evaluated using nf=5n_{f}=5, with a1=0.314a_{1}=0.314, a2=−0.301a_{2}=-0.301, a3=0.112a_{3}=0.112, r1=0.198r_{1}=0.198, r2=0.510r_{2}=0.510, and r3=−0.041r_{3}=-0.041.

References

  1. [1]Z. Koba, H.B. Nielsen and P. Olesen, Scaling of multiplicity distributions in high-energy hadron collisions, Nucl. Phys. B 40 (1972) 317.DOI
  2. [2]A.M. Polyakov, A Similarity hypothesis in the strong interactions. 1. Multiple hadron production in e+ e- annihilation, Zh. Eksp. Teor. Fiz. 59 (1970) 542.
  3. [3]R.P. Feynman, Very high-energy collisions of hadrons, Phys. Rev. Lett. 23 (1969) 1415.DOI
  4. [4]K. Konishi, A. Ukawa and G. Veneziano, Jet Calculus: A Simple Algorithm for Resolving QCD Jets, Nucl. Phys. B 157 (1979) 45.DOI
  5. [5]A. Bassetto, M. Ciafaloni and G. Marchesini, Inelastic Distributions and Color Structure in Perturbative QCD, Nucl. Phys. B 163 (1980) 477.DOI
  6. [6]Y.L. Dokshitzer, V.S. Fadin and V.A. Khoze, On the Sensitivity of the Inclusive Distributions in Parton Jets to the Coherence Effects in QCD Gluon Cascades, Z. Phys. C 18 (1983) 37.DOI
  7. [7]E.D. Malaza and B.R. Webber, QCD CORRECTIONS TO JET MULTIPLICITY MOMENTS, Phys. Lett. B 149 (1984) 501.DOI
  8. [8]A. Bassetto, KNO SCALING IN QCD JETS AND THE NEGATIVE BINOMIAL DISTRIBUTION, Nucl. Phys. B 303 (1988) 703.DOI
  9. [9]Y.L. Dokshitzer, V.A. Khoze, A.H. Mueller and S.I. Troian, Basics of perturbative QCD, Editions Frontieres (1991).
  10. [10]Y.L. Dokshitzer, Improved QCD treatment of the KNO phenomenon, Phys. Lett. B 305 (1993) 295.DOI
  11. [11]R. Vertesi, A. Gemes and G.G. Barnafoldi, Koba-Nielsen-Olesen-like scaling within a jet in proton-proton collisions at LHC energies, Phys. Rev. D 103 (2021) L051503 [2012.01132].DOI
  12. [12]G.R. Germano, F.S. Navarra, G. Wilk and Z. Wlodarczyk, Emergence of Koba-Nielsen-Olsen scaling in multiplicity distributions in jets produced at the LHC, Phys. Rev. D 110 (2024) 034026 [2406.04856].DOI
  13. [13]X.-P. Duan, L. Chen, G.-L. Ma, C.A. Salgado and B. Wu, KNO scaling in quark and gluon jets at the LHC, 2503.24200.
  14. [14]Y.L. Dokshitzer and B.R. Webber, Hadron multiplicity fluctuations in perturbative QCD, JHEP 08 (2025) 168 [2505.00652].
  15. [15]Y.L. Dokshitzer and B.R. Webber, QCD-inspired description of multiplicity distributions in jets, JHEP 10 (2025) 114 [2507.07691].arxiv.org/abs/2507.07691
  16. [16]TASSO collaboration, Jet Production and Fragmentation in e+ e- Annihilation at 12-GeV to 43-GeV, Z. Phys. C 22 (1984) 307.DOI
  17. [17]TASSO collaboration, Charged Multiplicity Distributions and Correlations in e+ e- Annihilation at PETRA Energies, Z. Phys. C 45 (1989) 193.DOI
  18. [18]DELPHI collaboration, Charged particle multiplicity distributions in Z0 hadronic decays, Z. Phys. C 50 (1991) 185.DOI
  19. [19]DELPHI collaboration, Charged particle multiplicity distributions in restricted rapidity intervals in Z0 hadronic decays., Z. Phys. C 52 (1991) 271.
  20. [20]ALEPH collaboration, Measurements of the charged particle multiplicity distribution in restricted rapidity intervals, Z. Phys. C 69 (1995) 15.
  21. [21]A. Giovannini and R. Ugoccioni, Soft and semihard components structure in multiparticle production in high-energy collisions, Nucl. Phys. B Proc. Suppl. 71 (1999) 201 [hep-ph/9710361].DOI
  22. [22]H1 collaboration, Measurement of charged particle multiplicity distributions in DIS at HERA and its implication to entanglement entropy of partons, Eur. Phys. J. C 81 (2021) 212 [2011.01812].arxiv.org/abs/2011.01812
  23. [23]Y. Liu, M.A. Nowak and I. Zahed, Mueller’s dipole wave function in QCD: Emergent Koba-Nielsen-Olesen scaling in the double logarithm limit, Phys. Rev. D 108 (2023) 034017 [2211.05169].DOI
  24. [24]Y. Liu, M.A. Nowak and I. Zahed, Universality of Koba-Nielsen-Olesen scaling in QCD at high energy and entanglement, 2302.01380.arxiv.org/abs/2302.01380
  25. [25]P. Slattery, Evidence for the Onset of Semiinclusive Scaling in Proton Proton Collisions in the 50-GeV/c - 300-GeV/c Momentum Range, Phys. Rev. Lett. 29 (1972) 1624.DOI
  26. [26]AMES-BOLOGNA-CERN-DORTMUND-HEIDELBERG-WARSAW collaboration, Charged Multiplicity Distribution in p p Interactions at ISR Energies, Phys. Rev. D 30 (1984) 528.DOI
  27. [27]UA5 collaboration, A New Empirical Regularity for Multiplicity Distributions in Place of KNO Scaling, Phys. Lett. B 160 (1985) 199.
  28. [28]UA5 collaboration, Charged Particle Multiplicity Distributions at 200-GeV and 900-GeV Center-Of-Mass Energy, Z. Phys. C 43 (1989) 357.DOI
  29. [29]CMS collaboration, Charged Particle Multiplicities in pp Interactions at √s = 0.9, 2.36, and 7 TeV, JHEP 01 (2011) 079 [1011.5531].DOI
  30. [30]J.F. Grosse-Oetringhaus and K. Reygers, Charged-Particle Multiplicity in Proton-Proton Collisions, J. Phys. G 37 (2010) 083001 [0912.0023].
  31. [31]ALICE collaboration, Charged-particle multiplicity distributions over a wide pseudorapidity range in proton-proton collisions at √s = 0.9, 7, and 8 TeV, Eur. Phys. J. C 77 (2017) 852 [1708.01435].arxiv.org/abs/1708.01435
  32. [32]H.R. Martins-Fontes and F.S. Navarra, Soft and semihard components of multiplicity distributions in the κT factorization approach, 2506.17127.
  33. [33]ATLAS collaboration, Properties of jet fragmentation using charged particles measured with the ATLAS detector in pp collisions at √s = 13 TeV, Phys. Rev. D 100 (2019) 052011 [1906.09254].DOI
  34. [34]S. Marzani, G. Soyez and M. Spannowsky, Looking inside jets: an introduction to jet substructure and boosted-object phenomenology, vol. 958, Springer (2019), 10.1007/978-3-030-15709-8, [1901.10342].DOI
  35. [35]A.J. Larkoski, QCD masterclass lectures on jet physics and machine learning, Eur. Phys. J. C 84 (2024) 1117 [2407.04897].arxiv.org/abs/2407.04897
  36. [36]J. Gallicchio and M.D. Schwartz, Quark and Gluon Tagging at the LHC, Phys. Rev. Lett. 107 (2011) 172001 [1106.3076].DOI
  37. [37]J. Gallicchio and M.D. Schwartz, Quark and Gluon Jet Substructure, JHEP 04 (2013) 090 [1211.7038].DOI
  38. [38]A.J. Larkoski, G.P. Salam and J. Thaler, Energy Correlation Functions for Jet Substructure, JHEP 06 (2013) 108 [1305.0007].DOI
  39. [39]A.J. Larkoski, J. Thaler and W.J. Waalewijn, Gaining (Mutual) Information about Quark/Gluon Discrimination, JHEP 11 (2014) 129 [1408.3122].
  40. [40]P. Gras, S. Höche, D. Kar, A. Larkoski, L. Lönnblad, S. Plätzer et al., Systematics of quark/gluon tagging, JHEP 07 (2017) 091 [1704.03878].DOI
  41. [41]E.M. Metodiev, B. Nachman and J. Thaler, Classification without labels: Learning from mixed samples in high energy physics, JHEP 10 (2017) 174 [1708.02949].DOI
  42. [42]E.M. Metodiev and J. Thaler, Jet Topics: Disentangling Quarks and Gluons at Colliders, Phys. Rev. Lett. 120 (2018) 241602 [1802.00008].DOI
  43. [43]P.T. Komiske, E.M. Metodiev and J. Thaler, An operational definition of quark and gluon jets, JHEP 11 (2018) 059 [1809.01140].DOI
  44. [44]F.A. Dreyer, G. Soyez and A. Takacs, Quarks and gluons in the Lund plane, JHEP 08 (2022) 177 [2112.09140].DOI
  45. [45]CMS collaboration, Performance of quark/gluon discrimination in 8 TeV pp data, .
  46. [46]ATLAS collaboration, Light-quark and gluon jet discrimination in pp collisions at √s = 7 TeV with the ATLAS detector, Eur. Phys. J. C 74 (2014) 3023 [1405.6583].DOI
  47. [47]ATLAS collaboration, Discrimination of Light Quark and Gluon Jets in pp collisions at √s = 8 TeV with the ATLAS Detector, .
  48. [48]I.M. Dremin and J.W. Gary, Energy dependence of mean multiplicities in gluon and quark jets at the next-to-next-to-next-to leading order, Phys. Lett. B 459 (1999) 341 [hep-ph/9905477].DOI
  49. [49]A. Capella, I.M. Dremin, J.W. Gary, V.A. Nechitailo and J. Tran Thanh Van, Evolution of average multiplicities of quark and gluon jets, Phys. Rev. D 61 (2000) 074009 [hep-ph/9910226].DOI
  50. [50]R.K. Ellis, W.J. Stirling and B.R. Webber, QCD and collider physics, vol. 8, Cambridge University Press (2, 2011), 10.1017/CBO9780511628788.
  51. [51]I.M. Dremin and J.W. Gary, Hadron multiplicities, Phys. Rept. 349 (2001) 301 [hep-ph/0004215].DOI
  52. [52]Y.L. Dokshitzer, V.S. Fadin and V.A. Khoze, Double Logs of Perturbative QCD for Parton Jets and Soft Hadron Spectra, Z. Phys. C 15 (1982) 325.DOI
  53. [53]Y.L. Dokshitzer, V.S. Fadin and V.A. Khoze, Coherent Effects in the Perturbative QCD Parton Jets, Phys. Lett. B 115 (1982) 242.DOI
  54. [54]Y.L. Dokshitzer, V.A. Khoze, S.I. Troian and A.H. Mueller, QCD Coherence in High-Energy Reactions, Rev. Mod. Phys. 60 (1988) 373.DOI
  55. [55]C. Bierlich et al., A comprehensive guide to the physics and usage of PYTHIA 8.3, SciPost Phys. Codeb. 2022 (2022) 8 [2203.11601].arxiv.org/abs/2203.11601
  56. [56]M. Cacciari, G.P. Salam and G. Soyez, The anti-kt jet clustering algorithm, JHEP 04 (2008) 063 [0802.1189].
  57. [57]M. Cacciari, G.P. Salam and G. Soyez, FastJet User Manual, Eur. Phys. J. C 72 (2012) 1896 [1111.6097].DOI
  58. [58]A. Buckley, J. Ferrando, S. Lloyd, K. Nordström, B. Page, M. Rüfenacht et al., LHAPDF6: parton density access in the LHC precision era, Eur. Phys. J. C 75 (2015) 132 [1412.7420].DOI
  59. [59]ATLAS collaboration, Properties of jets measured from tracks in proton-proton collisions at center-of-mass energy √s = 7 TeV with the ATLAS detector, Phys. Rev. D 84 (2011) 064001 [1107.3311].
  60. [60]M.E. Peskin and D.V. Schroeder, An Introduction to quantum field theory, Addison-Wesley, Reading, USA (1995), 10.1201/9780429503559.
  61. [61]Y.I. Azimov, Y.L. Dokshitzer, V.A. Khoze and S.I. Troyan, Similarity of Parton and Hadron Spectra in QCD Jets, Z. Phys. C 27 (1985) 65.DOI
  62. [62]Y.L. Dokshitzer, V.A. Khoze and S.I. Troyan, Specific features of heavy quark production. LPHD approach to heavy particle spectra, Phys. Rev. D 53 (1996) 89 [hep-ph/9506425].DOI
  63. [63]J. Currie, A. Gehrmann-De Ridder, T. Gehrmann, E.W.N. Glover, A. Huss and J. Pires, Precise predictions for dijet production at the LHC, Phys. Rev. Lett. 119 (2017) 152001 [1705.10271].DOI
  64. [64]R. Medves, A. Soto-Ontoso and G. Soyez, Lund and Cambridge multiplicities for precision physics, JHEP 10 (2022) 156 [2205.02861].DOI
  65. [65]R. Medves, A. Soto-Ontoso and G. Soyez, Lund multiplicity in QCD jets, JHEP 04 (2023) 104 [2212.05076].DOI
  66. [66]Z.-B. Kang, R. Kao and A.J. Larkoski, Multiplicity scaling of fragmentation function, Phys. Rev. D 109 (2024) 054039 [2305.13359].DOI

Paper details

vlnc:W_9vkcgsgg
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents