betaFeedback

WorksPhysics and Astronomy

Energy Correlators in V+XV + X as a Benchmark Observable for Precision QCD

Abstract

We propose projected energy correlators measured on the recoiling QCD radiation of a Z/γZ/γ as a benchmark observable for precision QCD at the LHC. Using the Z/γZ/γ as a hard scale prevents the need for a jet algorithm, simplifying both the perturbative and non-perturbative corrections. We develop a framework to combine state-of-the-art fixed-order amplitudes, high order resummation, and universal non-perturbative matrix elements. Our approach is based on numerically computed inclusive hard functions, allowing flexibility in the process and the inclusion of realistic experimental cuts. We perform detailed numerical studies of the projected energy correlators at next-to-leading order + next-to-next-to-leading logarithm (NLO+NNLL) to verify the stability of our setup. We present numerical results at NLO+NNLL, which are the first complete matched predictions for energy correlators at the LHC at this order. We discuss the prospects for extensions to higher orders, outlining a path to NNLO calculations of energy correlators at the LHC.

Prepared for submission to JHEP

Cavendish-HEP-26/04

MIT-CTP 6030

IFJPAN-IV-2026-14

Introduction

The Large Hadron Collider (LHC), together with its high-luminosity upgrade (HL-LHC), will deliver an enormous dataset with the potential to sharpen our understanding of the dynamics of Quantum Chromodynamics (QCD) at the TeV scale, probing increasingly extreme regions of phase space with ever-greater precision. Turning this data into precise tests of the Standard Model is theoretically demanding. Precision calculations of QCD observables at hadron colliders rest on three ingredients: high-multiplicity, high-loop scattering amplitudes; efficient subtraction schemes that turn these amplitudes into predictions for physical, infrared-safe observables; and factorization theorems that organize the dynamics across the disparate scales involved. Over the past decade each of these directions has advanced dramatically, and they are now beginning to converge. In this paper we argue that they come together in a particularly clean way for a single observable: the projected energy correlator measured on the QCD radiation that recoils against an electroweak boson. We develop the theoretical framework needed to predict it, and present the first complete matched results for energy correlators at the LHC with unparalleled accuracy.

This convergence is most visible in the recent conversion of two-loop amplitudes for 2→32 \to3 scattering into NNLO phenomenology. Fully differential predictions are now available for a wide range of final states, including pp→γγγpp \to\gamma\gamma\gamma at leading [1] and full [2] color, pp→γjjpp \to\gamma jj [3], and pp→jjjpp \to jjj [4, 5], and they have been used to compute hadron-collider event shapes in both multijet [5] and diphoton [6] final states. The reach has extended to processes with a massive leg, such as the associated production of an electroweak boson with a bottom-quark pair, pp→Wbbˉpp \to Wb\bar{b} [7, 8], with related processes matched to parton showers [9, 10], and even to two massive legs in pp→ttˉjpp \to t\bar{t}j [11]. This mirrors the history of e+e−e^+e^- colliders, where the two-loop three-parton amplitudes had long been known and it was the construction of suitable infrared subtraction schemes that finally enabled the calculation of event shapes at NNLO [12, 13, 14, 15], ushering in an era of precision QCD. It is natural to ask whether these same ingredients can now be brought to bear on collider processes at the LHC.

In parallel, jet substructure at the LHC has grown into one of the most active areas of collider physics [16, 17]. Building on the extensive study of event shapes in the controlled environment of e+e−e^+e^- colliders, it has greatly broadened the set of available observables and adapted them to the hadron collider environment, probing QCD in regimes that were previously inaccessible. Combining this progress with modern fixed-order calculations has proven difficult: many jet substructure observables are either too intricate to compute at high orders, or are contaminated by large contributions from the underlying event and other soft hadronic activity. The soft-drop groomed jet mass [18, 19] was among the first jet substructure observables to be matched to fixed order at next-to-leading order. However, the grooming procedure complicates the perturbative description and the treatment of non-perturbative corrections [20, 21], and prospect of going to even higher order and possibility of precision comparison with data remain uncertain.

Energy correlators sidestep several of these difficulties. Building on the energy-energy correlator of Refs. [22, 23, 24, 25] applied to e+e−e^+e^- colliders, it was proposed in Refs. [26, 27] to reformulate jet substructure in terms of correlation functions of energy flow operators; see Ref. [28] for a recent review. They admit a clean field-theoretic definition and can be computed on charged-particle (track) information with correspondingly improved angular resolution [26, 29, 30]. Of particular interest are the projected NN-point correlators [26], which retain only the largest pairwise angle among the NN particles; varying NN yields a family of observables whose ratios furnish additional, theoretically robust probes [27], in analogy with the classic 3/23/2 jet ratios. Because they weight final-state particles by their energy, energy correlators are dominated by the energetic core of the radiation and are correspondingly insensitive to soft contamination from pileup and the underlying event. Furthermore, for precision QCD, in the collinear limit these projected correlators exhibit a simple, approximately power-law scaling in the opening angle, with an exponent set by the timelike anomalous dimensions that is proportional to αs\alpha_s at leading order [26, 31]. This makes the slope of the distribution a theoretically well-controlled handle on the strong coupling constant αs\alpha_s at hadron colliders.

The promise of this approach has already been borne out in data. Projected energy correlators and their ratios were measured inside high-energy jets at the LHC by CMS [32], enabling one of the most precise extractions of αs\alpha_s from jet substructure to date, and they have also been measured in ZZ-boson-tagged jets in heavy-ion collisions [33]. Complementary information has come from the transverse energy-energy correlator, which is defined on the full hadronic final state and has been used to determine αs\alpha_s at the LHC [34], although accessing the collinear scaling regime that makes energy correlators such precise probes of αs\alpha_s is difficult for an observable defined on the entire event. Together, these measurements make the experimental case for a precision jet substructure program at the LHC. Realizing it requires a theoretical description with control over both the perturbative and the non-perturbative components in the hadron collider environment.

The observable we study in this paper is especially well suited to closing this gap. We propose measuring projected energy correlators on the QCD radiation recoiling against an electroweak boson, pp→V+Xpp \to V+X with V=γ,ZV=\gamma,Z. The key point is that the VV supplies the hard scale of the problem on its own. As a result, no jet algorithm is needed to reconstruct the hard scattering, and the correlator can be measured directly on the hadronic recoil without reference to a jet axis. This greatly simplifies the observable: there is no jet algorithm to complicate the perturbative description, and none of the power corrections in the jet radius that accompany substructure measured on a reconstructed jet. The result is an especially clean setting in which to combine the three ingredients of precision QCD identified above. The observable is illustrated in Figure 1: the boson recoils against the QCD radiation produced in the collision, and the correlation of energy flow is measured on that recoiling radiation. The process also places jet substructure in direct contact with the frontier of multi-leg scattering amplitudes. Reaching NNLO accuracy, for instance, requires the two-loop amplitude for the boson together with four partons, a five-point amplitude with a single off-shell color-singlet leg, of the kind that has recently become available [35].

Figure 1. Left: Illustration of the projected energy correlator in V+XV+X. An electroweak boson V=γ,ZV=\gamma,Z recoils against the QCD radiation produced in the collision, and the correlation of energy flow in the recoiling radiation is measured by the energy flow operators E(n⃗i)\mathcal{E}(\vec{n}_i). Right: State of the art calculation of the two-point energy correlator, at NLO+NNLL, and including leading non-perturbative corrections.

At the heart of this program is a factorization theorem for the energy correlators in the collinear limit, derived for the two-point case in Ref. [31] and generalized to the projected NN-point energy correlators in the point correlators in Ref. [26], with the extension to hadron colliders following Ref. [36, 37]. It expresses the observable as a convolution of a jet function, which captures the collinear dynamics and is universal across processes, with an inclusive hard function, the same partonic coefficient function that governs single-inclusive identified-hadron production. The entire process dependence resides in the hard function. In e+e−e^{+}e^{-} collisions both ingredients are known analytically to high order, which has enabled high-precision calculations of the collinear energy correlator [38, 39]. At a hadron collider, and in the presence of realistic fiducial cuts, the hard function can no longer be obtained analytically; what is needed instead is a numerical framework able to compute it for a generic initial state and arbitrary selection cuts, and to perform the matching to fixed order. Such a framework is already at hand. Within the Stripper subtraction scheme [40, 41, 42], extended to fragmentation processes in Refs. [43, 44, 45, 46], one can compute both the inclusive hard functions and the full fixed-order cross sections required for the matching. In Ref. [47] we initiated a framework that interfaces these numerically computed ingredients with the analytic factorization theorems for jet observables, illustrating it with single-inclusive small-radius jet production at NNLO+NNLL accuracy.

In the present work we build on this framework and apply it to projected energy correlators in V+XV+X. We assemble a complete calculational framework that combines state-of-the-art fixed-order amplitudes and subtractions, collinear resummation, and non-perturbative power corrections. The perturbative and non-perturbative components can be fed in together and evolved consistently: the leading non-perturbative correction is controlled by the same universal matrix element that governs e+e−e^{+}e^{-} event shapes [48], and we include its renormalization-group running alongside higher-order hard functions. We then present detailed predictions for the projected energy correlators at NLO+NNLL. These are the first complete matched predictions for energy correlators at the LHC built on exact fixed-order hard functions, and they reach a level of perturbative control comparable to the best-understood jet substructure observables. Achieving genuine NNLL accuracy requires the two-loop constants of the energy-correlator jet functions, which we incorporate [39, 49]. Our numerical studies are carried out for the photon channel, pp→γ+Xpp \to\gamma+ X, where the hard scale is set by a directly reconstructed, color-neutral boson. Furthermore, we study in detail the structure of the perturbative power corrections to the energy correlator, that is, the terms suppressed by powers of the opening angle relative to the leading-power singular behavior. The resummation does not capture these terms, which are instead supplied by the matching to fixed order; their size controls how deep into the singular region the fixed-order computation must be carried, and they can modify the slope of the distribution sensitive to αs\alpha_s. We find that they are under good control and take a simple form, with coefficients that depend linearly on the number of points NN. This is encouraging for an extension to NNLO, since the fixed-order computation then need not be pushed deep into the singular region where subtractions are most delicate; the remaining ingredient, the two-loop five-point one-mass (V+V+jets) amplitudes, is now available [35], which would also provide a striking illustration of their phenomenological relevance.

Together, these results chart a path towards genuine precision jet substructure at the LHC, uniting high-precision data with modern scattering amplitudes, subtraction schemes, and factorization. With perturbative and non-perturbative control already on par with the most thoroughly studied jet substructure observables, and a clear route to NNLO, the energy correlator in V+XV+X is poised to become a benchmark observable for precision QCD at the LHC. The remainder of this paper is organized as follows. In Sec. 2 we define the projected energy correlators and the fiducial selection cuts that we impose for the V+XV+X process. In Sec. 3 we develop the theoretical framework, discussing the fixed-order calculation, the collinear factorization theorem for energy correlators in V+XV+X, the numerical evaluation of the inclusive hard functions, the matching of the resummed and fixed-order results, and the inclusion of non-perturbative power corrections. In Sec. 4 we present our results for the photon channel, studying in detail the structure of perturbative power corrections and presenting NLO+NNLL predictions for the energy correlator in γ+X\gamma+X. We conclude and discuss future directions in Sec. 5.

Projected Energy Correlators in V+XV+X Processes

In this section, we define the projected energy correlators for the pp→V+Xpp\to V+X process, together with the fiducial selection cuts imposed in our analysis. We first briefly review the projected correlators in e+e−e^+e^- collisions, where they were introduced, and their generalization to hadron colliders using jets.

Energy correlators characterize the distribution of energy in the final state of a collision through correlation functions of energy flow operators [50, 51, 52],

E(n⃗)=∫0∞dtlim⁡r→∞r2niT0i(t,rn⃗),(1)\mathcal{E}(\vec{n})=\int_0^\infty dt\lim_{r\to\infty}r^2 n_i T_{0i}(t,r\vec{n}), \tag*{(1)}

each of which acts as an idealized calorimeter, recording the energy deposited over time in the lightlike direction n⃗\vec{n}. The simplest such observable is the two-point energy-energy correlator (EEC), introduced long ago in the context of e+e−e^+e^- annihilation [22, 23, 24, 25]. This was generalized to the projected NN-point correlators (ENC) also in e+e−e^+e^- [26], obtained by correlating NN energy flow operators and integrating over all pairwise angles while keeping the largest one, xLx_L, fixed:

dσe+e−[N]dxL=∫∏i=1NdΩn⃗i δ(xL−max⁡1≤i<j≤Nxij)∫d4xQNeiq⋅x⟨0∣O†(x)E(n⃗1)⋯E(n⃗N)O(0)∣0⟩.(2)\frac{d\sigma_{e^+e^-}^{[N]}}{dx_L}=\int\prod_{i=1}^{N}d\Omega_{\vec{n}_i}\,\delta\left(x_L-\max_{1\leq i<j\leq N}x_{ij}\right)\int\frac{d^4x}{Q^N}e^{iq\cdot x}\langle0|\mathcal{O}^{\dagger}(x)\mathcal{E}(\vec{n}_1)\cdots\mathcal{E}(\vec{n}_N)\mathcal{O}(0)|0\rangle. \tag*{(2)}

Here xij=(1−n⃗i⋅n⃗j)/2x_{ij}=(1-\vec{n}_i\cdot\vec{n}_j)/2 is the angular distance between detectors ii and jj, dΩn⃗=14πsin⁡θ dθ dϕd\Omega_{\vec{n}}=\frac{1}{4\pi}\sin\theta\,d\theta\,d\phi is the area element on the celestial sphere, and O\mathcal{O} is the source operator for the process under consideration. In e+e−e^+e^- annihilation the source is the electroweak current, whose virtuality q2=Q2q^2=Q^2 fixes the hard scale of the measurement. In perturbation theory, Eq. (2) can be rewritten as a sum of energy-weighted cross sections over final states,

dσe+e−[N]dxL=12Q2∑m∫dΦm∑Xm∣MXm∣2×∑1≤i1,…,iN≤m∏a=1NEiaQN δ(xL−max⁡{xi1i2,xi1i3,…,xiN−1iN}),(3)\begin{aligned} \frac{d\sigma_{e^+e^-}^{[N]}}{dx_L}=\frac{1}{2Q^2}\sum_m\int d\Phi_m\sum_{X_m}|\mathcal{M}_{X_m}|^2 \\ &\times\sum_{1\leq i_1,\ldots,i_N\leq m}\frac{\prod_{a=1}^{N}E_{i_a}}{Q^N}\,\delta\left(x_L-\max\{x_{i_1i_2},x_{i_1i_3},\ldots,x_{i_{N-1}i_N}\}\right), \tag*{(3)} \end{aligned}

where dΦmd\Phi_m is the massless Lorentz-invariant phase space of the mm-particle final state XmX_m and ∣MXm∣2|\mathcal{M}_{X_m}|^2 is the corresponding squared amplitude. The indices iai_a are allowed to coincide, producing contact terms. In e+e−e^{+}e^{-}, the projected correlator is defined over the entire angular region, from the collinear limit xL→0x_L \to0 to the back-to-back limit xL→1x_L \to1. For later convenience, we also define the cumulant of ENC by integrating the distribution,

Σe+e−[N](xL)≡∫0xLdxL′ dσe+e−[N]dxL′.(4)\Sigma_{e^{+}e^{-}}^{[N]}(x_L) \equiv\int_{0}^{x_L} dx'_L \, \frac{d\sigma_{e^{+}e^{-}}^{[N]}}{dx'_L}. \tag*{(4)}

Two ingredients implicit in Eq. (3) are especially worth preserving at a hadron collider: the energy weights are normalized to the hard scale QQ, which the colliding beams fix once and for all, and the observable is a function of the final state alone, requiring no auxiliary algorithmic construction.

At a hadron collider there is no direct analogue of the global scale QQ, since the scale of the underlying hard partonic scattering varies event by event. Jets provide the standard means of tagging the hard scattering: a high transverse momentum jet, reconstructed with a jet algorithm such as anti-kTk_T [53] with radius parameter RR, serves as a proxy for a hard parton, and its substructure has grown into a rich field in its own right [16, 17]. Energy correlators were brought to the LHC within this setting, as one such substructure observable, by measuring the correlations of energy flow among the particles inside the jet [27],

dσjet[N]dRL=∫dσpp→jet+X∑i1,…,iN∈jet∏a=1Npia,T(pjet,T)Nδ(RL−max⁡{Ri1i2,Ri1i3,…,RiN−1iN}),(5)\frac{d\sigma_{\mathrm{jet}}^{[N]}}{dR_L} = \int d\sigma_{pp\to\mathrm{jet}+X} \sum_{i_1,\ldots,i_N\in\mathrm{jet}} \frac{\prod_{a=1}^{N} p_{i_a,T}}{(p_{\mathrm{jet},T})^N} \delta\left(R_L-\max\{R_{i_1i_2},R_{i_1i_3},\ldots,R_{i_{N-1}i_N}\}\right), \tag*{(5)}

where the angles are now measured with the boost-invariant rapidity-azimuth distance Rij=Δηij2+Δϕij2R_{ij}=\sqrt{\Delta\eta_{ij}^{2}+\Delta\phi_{ij}^{2}}, with Δηij=ηi−ηj\Delta\eta_{ij}=\eta_i-\eta_j and Δϕij=ϕi−ϕj\Delta\phi_{ij}=\phi_i-\phi_j the rapidity and azimuthal differences between particles ii and jj, and RLR_L denotes the largest such distance. Here dσpp→jet+Xd\sigma_{pp\to\mathrm{jet}+X} is shorthand for the fully differential single-inclusive jet production cross section, the analogue of the explicit phase-space integrals and squared amplitudes in Eq. (3), now including the parton distribution functions, flux factors and a summation over all jets passing the kinematic selection of the measurement. The jet transverse momentum pTjetp_T^{\mathrm{jet}} plays the role of the hard scale, both selecting the hard scattering and normalizing the energy weights. This strategy has proven remarkably powerful and underlies the success of the jet substructure program. For energy correlators in particular, the enormous rates of high-pTp_T jets give access to the collinear scaling regime at TeV energies, across a wide range of angular scales. It has driven a broad and successful measurement program by several collaborations [32, 54, 55] in pppp and even to growing measurements in heavy-ion collisions as well [33, 56, 57].

At the same time, measuring correlators inside jets introduces structure that is not intrinsic to the scattering itself. The jet algorithm and the radius RR are parameters of the analysis rather than of the process: nothing in the underlying scattering singles out a particular value of RR, which enters instead as an external choice that defines the object whose constituents are correlated. For a precision program this has several consequences. First, the jet introduces the additional scale pTRp_TR, and the theoretical description acquires algorithm-dependent ingredients: in the factorized description of in-jet correlators, the hard production is connected to the collinear dynamics by matching coefficients that depend explicitly on the jet algorithm, the radius RR, and the scale pTRp_TR [37, 58]. Second, the normalization of the energy weights in Eq. (5) is the jet pTp_T itself, an algorithm-dependent quantity with its own perturbative and non-perturbative corrections, rather than a scale set by the hard scattering. Third, the jet boundary limits the angular range: as RLR_L approaches RR, the collinear power expansion breaks down and the correlator is distorted by clustering effects at the jet edge, introducing power corrections in RL/RR_L/R, potentially together with clustering and non-global logarithms, so that only the region RL≪RR_L \ll R is cleanly accessible, in contrast to the full angular coverage available in e+e−e^+e^-.

The V+XV+X process removes this auxiliary structure. In pp→V+Xpp \to V+X with V=γ,ZV=\gamma,Z, the electroweak boson is reconstructed independently of the hadronic activity and its transverse momentum pV,Tp_{V,T} is a genuine scale of the hard scattering, determined by the interaction itself rather than imposed as a parameter of the analysis. In this sense, V+XV+X is the closest hadron-collider analogue of e+e−e^+e^- annihilation, which likewise “tags” the hard interaction with an electroweak boson, and can be viewed as a crossed version of it. In e+e−e^+e^-, the hadronic final state is sourced by a timelike virtual γ∗/Z\gamma^*/Z of virtuality q2=Q2q^2=Q^2, which sets the hard scale. In V+XV+X, the same electroweak current appears in the final state as an on-shell boson, a real photon (q2=0q^2=0) or a ZZ near MZM_Z, that recoils against the QCD radiation whose correlations we measure, with the hard scale now set by the transverse momentum pV,Tp_{V,T}, as illustrated in Figure 1. We therefore define the projected energy correlators in V+XV+X on the recoiling radiation, with the energy weights normalized to the boson transverse momentum:

dσ[N]dRL=∑m∑Xm∫dσpp→V+Xm∑1≤i1,…,iN≤m∏a=1Npia,TpV,TNδ(RL−max⁡{Ri1i2,Ri1i3,…,RiN−1iN}).(6)\frac{d\sigma^{[N]}}{dR_L} = \sum_m \sum_{X_m} \int d\sigma_{pp\to V+X_m} \sum_{1\leq i_1,\ldots,i_N\leq m} \frac{\prod_{a=1}^{N}p_{i_a,T}}{p_{V,T}^{N}} \delta\left(R_L-\max\{R_{i_1i_2},R_{i_1i_3},\ldots,R_{i_{N-1}i_N}\}\right). \tag*{(6)}

Here dσpp→V+Xmd\sigma_{pp\to V+X_m} is the fully differential cross section for producing the boson together with an mm-particle hadronic final state XmX_m, including the parton distribution functions and flux factors, and the correlator sums run over all final-state particles except the boson (or its decay products). We give the explicit perturbative expansion of Eq. (6) in § 3.

No jet algorithm enters this definition: there is no clustering, no radius, and no scale pTRp_TR. As in e+e−e^+e^-, the correlator is a property of the final state alone, tagged by a hard scale of the scattering, and it is not partitioned by any jet boundary. The collinear regime RL→0R_L\to0 connects continuously out to wide angles. Finally, pV,Tp_{V,T} is fixed purely by the boson kinematics and is therefore independent of the reconstruction of hadronic energy. Relative to the in-jet case, this removes the jet-radius power corrections and the jet-energy-scale non-perturbative corrections and simplifies the perturbative description, while the leading non-perturbative correction to the correlator shape itself remains the universal collinear matrix element shared with e+e−e^+e^- event shapes.

In this work, we focus our numerical studies on the photon channel, pp→γ+Xpp\to\gamma+X, which can be resolved with precise resolution even at the highest transverse momenta. The identical framework applies to Z+XZ+X with Z→ℓ+ℓ−Z\to\ell^+\ell^-, which avoids the photon isolation requirements discussed below and provides an exceptionally clean tag; we foresee it as a central target for future measurements. For the correlator measurement, an important consideration is that the energy weights in Eq. (6) enhance configurations in which the hadronic system is much harder than the photon. In particular, dijet-like configurations (though can be computed within our framework and are included), in which the photon is relatively soft and the remaining final-state particles recoil against each other in the partonic center-of-mass frame, receive weights ∏i=1Npia,T/pγ,TN>1\prod_{i=1}^{N} p_{i a,T}/p_{\gamma,T}^{N} > 1, and their contribution grows with NN. To prevent these dijet configurations completely dominating, we impose a stringent lower cut on the photon transverse momentum, pγ,T>500p_{\gamma,T} > 500 GeV, together with ∣ηγ∣<2.4|\eta_{\gamma}| < 2.4. All other final-state particles are also required to satisfy ∣η∣<2.4|\eta| < 2.4. Furthermore, the perturbative definition of a prompt photon requires an isolation prescription, which suppresses contributions in which the photon is produced collinear to QCD radiation and which would otherwise require non-perturbative photon fragmentation functions. We require both hard cone and smooth cone isolation, in the so-called hybrid cone approach [59, 60]. Only particles with ∣η∣<2.4|\eta| < 2.4 contribute to the isolation requirements. In the hard cone, the scalar sum of the transverse momenta of all particles with ΔRγ,particle<0.2\Delta R_{\gamma,\mathrm{particle}} < 0.2 must be less than 0.09 pγ,T0.09\,p_{\gamma,T}. In the smooth cone, the scalar sum of the transverse momenta of all particles with ΔRγ,particle<r\Delta R_{\gamma,\mathrm{particle}} < r must be smaller than 0.15(1−cos⁡r)/(1−cos⁡0.1) pγ,T0.15(1-\cos r)/(1-\cos0.1)\,p_{\gamma,T} for all r<0.1r < 0.1. All of these isolation criteria can be implemented within the fixed-order computation using Monte Carlo integration, which is described in detail in § 3.1.

It is also interesting to study the ratios of the projected NN-point correlators to the 2-point correlator,

dσ[N]/dRLdσ[2]/dRL.(7)\frac{\mathrm{d}\sigma^{[N]}/\mathrm{d}R_L}{\mathrm{d}\sigma^{[2]}/\mathrm{d}R_L}. \tag*{(7)}

There are several advantages to using ratios of correlators as observables. First, uncertainties that affect only the overall normalization of a measurement, such as the luminosity uncertainty, cancel between numerator and denominator. Second, the leading non-perturbative corrections, which are controlled by the same universal matrix element in the numerator and the denominator, partially cancel in the ratio. Finally, the ratios increase the relative sensitivity to the value of the strong coupling [39]. These features make the ratio correlators prime candidates for extracting the strong coupling from LHC data.

Theoretical Framework

In this section, we will introduce the theoretical framework for the calculation of energy correlators in hadronic processes. While we will focus on the process, pp→γ+Xpp \to\gamma+ X at the LHC, our framework is designed to enable the calculation of the energy correlator in a wide variety of processes at the LHC and other colliders. Our aim is a prediction that is accurate across the range of the observable, from the small-angle region RL→0R_L \to0 to moderate angles RL∼O(1)R_L \sim\mathcal{O}(1), which we obtain by combining a fixed-order calculation of the correlator with the resummation of the logarithms that dominate at small angles.

At fixed order in αs\alpha_s, the projected correlator is computed directly from the QCD matrix elements for a photon and partons, with the energy weighting of Eq. (6) imposed on the final state. This provides the complete result at moderate angles, RL∼O(1)R_L \sim\mathcal{O}(1), where the distribution is not dominated by the collinear singularity and the power corrections in RLR_L are substantial rather than negligible. In § 3.1, we describe how the STRIPPER subtraction scheme [40, 41, 42] was used to carry out these computations.

In the small-angle collinear region, RL→0R_L \to0, the observable obeys a factorization theorem into a perturbatively calculable jet function, which captures the collinear dynamics and is universal across processes, and a hard function that can also be computed in Stripper. We present the factorization and its resummation in § 3.2 and the numerical evaluation of the process-dependent hard functions in § 3.3. Non-perturbative power corrections are incorporated in § 3.4.

Fixed Order Calculations Using the Stripper Framework

The fixed-order expansion in αs\alpha_s of the projected energy-energy correlator defined in Eq. (6) for the process pp→γ+Xpp \to\gamma+ X reads:

dσ[N]dRL=dσ[N](0)dRL+dσ[N](1)dRL+dσ[N](2)dRL+O(αs4)(8)\frac{d\sigma^{[N]}}{dR_L} = \frac{d\sigma^{[N](0)}}{dR_L} + \frac{d\sigma^{[N](1)}}{dR_L} + \frac{d\sigma^{[N](2)}}{dR_L} + \mathcal{O}(\alpha_s^4) \tag*{(8)}

where the superscript (i)(i) labels successive orders in the strong coupling, beginning with the lowest, O(αs)\mathcal{O}(\alpha_s), term dσ[N](0)/dRLd\sigma^{[N](0)}/dR_L. An important feature of the projected correlator is that, away from the endpoint at RL=0R_L = 0, its differential distribution starts one order in αs\alpha_s higher than the cross section. At the lowest order, a single parton recoils against the photon and carries the entire hadronic transverse momentum, so all of the energy weight is deposited at zero separation and contributes only to a contact piece ∝δ(RL)\propto\delta(R_L). The first genuine contribution at RL≠0R_L \ne0, which we take to define the leading order (LO) of the correlator, therefore appears at O(αs2)\mathcal{O}(\alpha_s^2), and its next-to-leading order (NLO) at O(αs3)\mathcal{O}(\alpha_s^3). Equivalently, the LO and NLO of the differential correlator dσ[N]/dRLd\sigma^{[N]}/dR_L correspond to the NLO and NNLO of the pp→γ+Xpp \to\gamma+ X cross section, or of its cumulant Σ[N]\Sigma^{[N]}.

In this subsection, we follow closely the notation in Ref. [41]. The perturbative coefficients of the hadronic correlator dσab[N](i)dRL\frac{d\sigma_{ab}^{[N](i)}}{dR_L} expressed in terms of partonic coefficients dσ^ab[N](i)dRL\frac{d\hat{\sigma}_{ab}^{[N](i)}}{dR_L} are given by

dσ[N](i)dRL=∑ab∫ ⁣ ⁣∫01dx1dx2 fa(x1,μF)fb(x2,μF)dσ^ab[N](i)dRL(x1,x2,μF,μR,αs(μR)).(9)\frac{d\sigma^{[N](i)}}{dR_L}=\sum_{ab}\int\!\!\int_0^1 dx_1dx_2\,f_a(x_1,\mu_F)f_b(x_2,\mu_F)\frac{d\hat{\sigma}_{ab}^{[N](i)}}{dR_L}(x_1,x_2,\mu_F,\mu_R,\alpha_s(\mu_R)) . \tag*{(9)}

Here we made explicit the dependence on the renormalization and factorization scales (μR\mu_R and μF\mu_F, respectively) and on the initial-state momentum fractions x1x_1 and x2x_2. For brevity, these are omitted in the following expressions. At lowest (B)orn order, i.e. O(αs)\mathcal{O}(\alpha_s), we have

dσ^ab[N](0)dRL=dσ^ab[N]BdRL=12s^1Nab∫dΦ2 ⟨Mab→γk(0)∣Mab→γk(0)⟩F1[N](RL)(10)\frac{d\hat{\sigma}_{ab}^{[N](0)}}{dR_L}=\frac{d\hat{\sigma}_{ab}^{[N]\mathrm{B}}}{dR_L}=\frac{1}{2\hat{s}}\frac{1}{N_{ab}}\int d\Phi_2\,\langle\mathcal{M}_{ab\to\gamma k}^{(0)}|\mathcal{M}_{ab\to\gamma k}^{(0)}\rangle\mathcal{F}_1^{[N]}(R_L) \tag*{(10)}

with implied sum over all partons kk allowed by the initial state partons abab.1 We used the abbreviation

Fm[N](RL)≡∑1≤i1,…,iN≤m∏a=1Npia,Tpγ,TNδ(RL−max⁡{Ri1i2,Ri1i3,…,RiN−1iN})(11)\mathcal{F}_m^{[N]}(R_L)\equiv\sum_{1\le i_1,\ldots,i_N\le m}\frac{\prod_{a=1}^{N}p_{i_a,T}}{p_{\gamma,T}^{N}}\delta\left(R_L-\max\{R_{i_1i_2},R_{i_1i_3},\ldots,R_{i_{N-1}i_N}\}\right) \tag*{(11)}

for the definition of NN-point projected energy correlators. We also use ⟨M(j)∣M(i)⟩\langle\mathcal{M}^{(j)}|\mathcal{M}^{(i)}\rangle to denote the interference between the ii- and jj-loop matrix elements. Finally, NabN_{ab} denotes the number of incoming partonic spin and color degrees of freedom, whereas s^=(pa+pb)2\hat{s}=(p_a+p_b)^2. Examples for contributing Feynman diagrams at this order are shown in the first row in Figure 2. With only one parton in the final state, this only contributes to a δ(RL)\delta(R_L) in projected correlators.

Figure 2. Example Feynman diagrams contributing to the fixed-order predictions at O(αs)\mathcal{O}(\alpha_s) (top row), O(αs2)\mathcal{O}(\alpha_s^2) (middle row), and O(αs3)\mathcal{O}(\alpha_s^3) (bottom row). With a single parton in the final state, the diagrams in the top row contribute only to the δ(RL)\delta(R_L) endpoint, so the middle and bottom rows define the LO and NLO of the differential projected correlators at RL≠0R_L \ne0.

The leading contribution of the projected correlator for RL≠0R_L\ne0 starts at O(αs2)\mathcal{O}(\alpha_s^2) and receives contributions from tree-level ij→γ+klij\to\gamma+kl matrix elements. One-loop corrections of ij→γ+kij\to\gamma+k again contribute to the coefficient of δ(RL)\delta(R_L) only. The partonic coefficient reads

dσ^ab[N](1)dRL=dσ^ab[N]RdRL+dσ^ab[N]VdRL+dσ^ab[N]CdRL,(12)\frac{d\hat{\sigma}_{ab}^{[N](1)}}{dR_L}=\frac{d\hat{\sigma}_{ab}^{[N]\mathrm{R}}}{dR_L}+\frac{d\hat{\sigma}_{ab}^{[N]\mathrm{V}}}{dR_L}+\frac{d\hat{\sigma}_{ab}^{[N]\mathrm{C}}}{dR_L}, \tag*{(12)}

where σ^abR\hat{\sigma}_{ab}^{\mathrm{R}} and σ^abV\hat{\sigma}_{ab}^{\mathrm{V}} refer to real emission and virtual corrections, respectively, and σ^abC\hat{\sigma}_{ab}^{\mathrm{C}} to collinear beam factorization terms to cancel initial state collinear singularities. Explicitly, these contributions are given by

dσ^ab[N]RdRL=12s^1Nab∫dΦ3 ⟨Mab→γkl(0)∣Mab→γkl(0)⟩F2[N](RL),(13)\frac{d\hat{\sigma}_{ab}^{[N]\mathrm{R}}}{dR_L}=\frac{1}{2\hat{s}}\frac{1}{N_{ab}}\int d\Phi_3\,\langle\mathcal{M}_{ab\to\gamma kl}^{(0)}|\mathcal{M}_{ab\to\gamma kl}^{(0)}\rangle\mathcal{F}_2^{[N]}(R_L), \tag*{(13)}
dσ^ab[N]VdRL=12s^1Nab∫dΦ2 2ℜ⟨Mab→γk(0)∣Mab→γk(1)⟩F1[N](RL),\frac{d\hat{\sigma}_{ab}^{[N]\mathrm{V}}}{dR_L}=\frac{1}{2\hat{s}}\frac{1}{N_{ab}}\int d\Phi_2\,2\Re\langle\mathcal{M}_{ab\to\gamma k}^{(0)}|\mathcal{M}_{ab\to\gamma k}^{(1)}\rangle\mathcal{F}_1^{[N]}(R_L),
dσ^ab[N]CdRL=αs2π1ϵ(μR2μF2)ϵ∑c∫01dz[Pca(0)(z)σ^cb[N]B(zx1,x2)+Pcb(0)(z)σ^ac[N]B(x1,zx2)].\frac{d\hat{\sigma}_{ab}^{[N]\mathrm{C}}}{dR_L}=\frac{\alpha_s}{2\pi}\frac{1}{\epsilon}\left(\frac{\mu_R^2}{\mu_F^2}\right)^\epsilon\sum_c\int_0^1 dz\left[P_{ca}^{(0)}(z)\hat{\sigma}_{cb}^{[N]\mathrm{B}}(zx_1,x_2)+P_{cb}^{(0)}(z)\hat{\sigma}_{ac}^{[N]\mathrm{B}}(x_1,zx_2)\right].

Again, we imply the sum over all partonic states kk and klkl allowed by the initial state abab. The virtual and collinear factorization counterterms contribute only a δ(RL)\delta(R_L) because they have only one parton in the final state. Example Feynman diagrams for the real and virtual corrections are shown in the middle row of Figure 2.

At NLO for the projected correlators (which corresponds to NNLO for the pp→γ+Xpp \to\gamma+ X process), corrections from double real emissions, virtual corrections to single real emissions, and two-loop contributions have to be taken into account. The NNLO coefficient can decompose in

dσ^ab[N](2)dRL=dσ^ab[N]RRdRL+dσ^ab[N]RVdRL+dσ^ab[N]VVdRL+dσ^ab[N]C1dRL+dσ^ab[N]C2dRL.(14)\frac{d\hat{\sigma}_{ab}^{[N](2)}}{dR_L} = \frac{d\hat{\sigma}_{ab}^{[N]\mathrm{RR}}}{dR_L} + \frac{d\hat{\sigma}_{ab}^{[N]\mathrm{RV}}}{dR_L} + \frac{d\hat{\sigma}_{ab}^{[N]\mathrm{VV}}}{dR_L} + \frac{d\hat{\sigma}_{ab}^{[N]\mathrm{C1}}}{dR_L} + \frac{d\hat{\sigma}_{ab}^{[N]\mathrm{C2}}}{dR_L}. \tag*{(14)}

The expressions for the double real (RR), real virtual (RV), and double virtual (VV) contributions are

dσ^ab[N]RRdRL=12s^1Nab∫dΦ4⟨Mab→γklm(0)∣Mab→γklm(0)⟩F3[N](RL),(15)\frac{d\hat{\sigma}_{ab}^{[N]\mathrm{RR}}}{dR_L} = \frac{1}{2\hat{s}}\frac{1}{N_{ab}}\int d\Phi_4 \langle\mathcal{M}_{ab\to\gamma klm}^{(0)}|\mathcal{M}_{ab\to\gamma klm}^{(0)}\rangle\mathcal{F}_3^{[N]}(R_L), \tag*{(15)}
dσ^ab[N]RVdRL=12s^1Nab∫dΦ3 2ℜ⟨Mab→γkl(0)∣Mab→γkl(1)⟩F2[N](RL),\frac{d\hat{\sigma}_{ab}^{[N]\mathrm{RV}}}{dR_L} = \frac{1}{2\hat{s}}\frac{1}{N_{ab}}\int d\Phi_3\,2\Re\langle\mathcal{M}_{ab\to\gamma kl}^{(0)}|\mathcal{M}_{ab\to\gamma kl}^{(1)}\rangle\mathcal{F}_2^{[N]}(R_L),
dσ^ab[N]VVdRL=12s^1Nab∫dΦ2(2ℜ⟨Mab→γk(0)∣Mab→γk(2)⟩+⟨Mab→γk(1)∣Mab→γk(1)⟩)F1[N](RL).\frac{d\hat{\sigma}_{ab}^{[N]\mathrm{VV}}}{dR_L} = \frac{1}{2\hat{s}}\frac{1}{N_{ab}}\int d\Phi_2\left(2\Re\langle\mathcal{M}_{ab\to\gamma k}^{(0)}|\mathcal{M}_{ab\to\gamma k}^{(2)}\rangle+ \langle\mathcal{M}_{ab\to\gamma k}^{(1)}|\mathcal{M}_{ab\to\gamma k}^{(1)}\rangle\right)\mathcal{F}_1^{[N]}(R_L).

In the interest of brevity, we do not spell out the expression for the collinear factorization contributions (C1 and C2), which can easily be extracted from Ref. [41]. Here, the double virtual corrections and the C2 contribution contribute only to the endpoint. Example diagrams for the real emission and virtual correction at this order can be found in the lowest row in Figure 2. Following this, we can define all NkN^{k}LO orders for the projected correlators at fixed order. While the differential ENC does not receive any contributions from terms proportional to δ(RL)\delta(R_L), the cumulant, defined in Eq. (4), does. In practice, we evaluate the Nk+1N^{k+1}LO cumulant distribution Σ[N]\Sigma^{[N]} by keeping all contributions that are proportional to δ(RL)\delta(R_L), i.e., subtraction terms and virtual contributions with tree-level kinematics. Using a NNLO subtraction scheme thus allows us to evaluate the NLO differential and NNLO cumulative distributions at fixed order in perturbation theory. Ingredients to push the differential ENC one order higher are available [3] but their incorporation is computationally very challenging and beyond the scope of the current work. The incorporation of the progress made on three-loop amplitudes for pp→γ+jetpp \to\gamma+ \text{jet} [61], needed for the cumulative distribution, is lacking a method of computing N3^{3}LO cross sections at hadron colliders, i.e. an N3^{3}LO subtraction scheme. While such a scheme is available for jet production in lepton collisions [62], so far, no such scheme has been constructed for hadronic collisions.

We employ the sector-improved residue subtraction scheme, as implemented in the STRIPPER framework [40, 41, 42], to handle infrared divergences up to NNLO QCD for the pp→γ+Xpp \to\gamma+ X process, or equivalently, next-to-leading order for the projected energy-energy correlator. All tree-level matrix elements are provided by the AVHLIB library [63, 64], while all required one-loop matrix elements are publicly available in the OpenLoops2 library [65, 66]. The two-loop matrix elements have been taken from Ref. [67] and newly implemented in a numerical code. For the numerical results presented in the following sections, we use the PDF4LHC21 PDF set [68] to provide αs\alpha_s and the PDF values. We work in nf=5n_f = 5 massless QCD, i.e., we don’t include contributions from top-quark loops. Further, we are using the dynamical scale choice μR=μF=pγ,T\mu_R = \mu_F = p_{\gamma,T}. The fiducial phase space of the photon was described in § 2. To estimate uncertainties arising from missing higher-order terms in the fixed-order predictions, we exploit a 7-point scale variation around the central scale choice by a factor of 2. The quoted uncertainties are derived from the envelope of these variations. For ratios, we use correlated variations in the numerator and denominator.

While this fixed-order expansion is well behaved at moderate RLR_L, in the small-angle limit RL→0R_L \to0 each order develops large singular logarithms of RLR_L. These logarithms must be resummed to all orders to obtain a reliable prediction, which we achieve through the collinear factorization theorem described next.

Collinear Factorization Theorem

In this subsection, we review the collinear factorization theorem for projected energy correlators. In fixed-order predictions, the ENC distribution receives large logarithmic enhancements at each order:

dσ^[N]dRL=∑L=1∞∑j=−1L−1(αs4π)LeL,jLj(RL),(16)\frac{d\hat{\sigma}^{[N]}}{dR_L}=\sum_{L=1}^{\infty}\sum_{j=-1}^{L-1}\left(\frac{\alpha_s}{4\pi}\right)^L e_{L,j}\mathcal{L}^j(R_L), \tag*{(16)}

where L−1(RL)=δ(RL)\mathcal{L}^{-1}(R_L)=\delta(R_L), Lj(RL)=[ln⁡j(RL)/RL]+\mathcal{L}^j(R_L)=\left[\ln^j(R_L)/R_L\right]_+.

Here eL,je_{L,j} are the constants that are associated with the logarithmic terms.

For ENC cumulants, the plus distributions Lj(RL)\mathcal{L}^j(R_L) in the partonic coefficient dσ^/dRLd\hat{\sigma}/dR_L will be mapped to ln⁡j+1RL\ln^{j+1}R_L in the partonic coefficient cumulant Σ^(RL)\hat{\Sigma}(R_L). The existence of these large logarithms can spoil the convergence of perturbation theory when αsln⁡RL∼1\alpha_s\ln R_L\sim1, and thus the resummation to all orders is necessary.

The collinear factorization for the 2-point correlator, EEC, is first derived in Ref. [31] and generalized to the NN-point projected correlator in Ref. [26] for e+e−e^+e^- collisions and hadronic Higgs decays. Using the largest angle xLx_L at e+e−e^+e^- annihilation, the ENC cumulant can be written as (see Eq. (4))

Σ^e+e−[N](xL)=∫01dx xNJ⃗[N](ln⁡xLxQ2μ2,μ)⋅H⃗e+e−(x,ln⁡Q2μ2,μ).(17)\hat{\Sigma}^{[N]}_{e^+e^-}(x_L)=\int_0^1 dx\,x^N\vec{J}^{[N]}\left(\ln\frac{x_LxQ^2}{\mu^2},\mu\right)\cdot\vec{H}_{e^+e^-}\left(x,\ln\frac{Q^2}{\mu^2},\mu\right). \tag*{(17)}

Here the hard function H⃗e+e−={He+e−,q,He+e−,g}\vec{H}_{e^+e^-}=\{H_{e^+e^-,q},H_{e^+e^-,g}\} is the single-inclusive fragmentation function fixing the energy fraction of one final-state parton. J⃗[N]\vec{J}^{[N]} is the ENC jet function, which can be defined in terms of operators in Soft-Collinear Effective Theory (SCET) [69, 70, 71, 72, 73]:

Jq[N](xL,Q,μ2)=∫dl+2π12NCTr⁡∫d4x eil⋅x⟨0∤nˉ2χn(x)M^ENC[N]δ(Q+nˉ⋅P)δ2(P⊥)χˉn(0)∣0⟩,Jg[N](xL,Q,μ2)=∫dl+2π12(NC2−1)Tr⁡∫d4x eil⋅x⟨0∣Bn,⊥a,μ(x)M^ENC[N]δ(Q+nˉ⋅P)δ2(P⊥)Bn,⊥a,μ(0)∣0⟩.(18)\begin{aligned} J_q^{[N]}(x_L,Q,\mu^2)&=\int\frac{dl^+}{2\pi}\frac{1}{2N_C}\operatorname{Tr}\int d^4x\,e^{il\cdot x}\langle0|\frac{\not{\bar n}}{2}\chi_n(x)\widehat{\mathcal{M}}_{\mathrm{ENC}}^{[N]}\delta(Q+\bar n\cdot\mathcal{P})\delta^2(\mathcal{P}_\perp)\bar{\chi}_n(0)|0\rangle,\\ J_g^{[N]}(x_L,Q,\mu^2)&=\int\frac{dl^+}{2\pi}\frac{1}{2(N_C^2-1)}\operatorname{Tr}\int d^4x\,e^{il\cdot x}\langle0|\mathcal{B}_{n,\perp}^{a,\mu}(x)\widehat{\mathcal{M}}_{\mathrm{ENC}}^{[N]}\delta(Q+\bar n\cdot\mathcal{P})\delta^2(\mathcal{P}_\perp)\mathcal{B}_{n,\perp}^{a,\mu}(0)|0\rangle. \tag*{(18)} \end{aligned}

where χn≡Wn†ξn\chi_n\equiv W_n^\dagger\xi_n is the collinear quark and Bn,⊥μ≡1g1nˉ⋅PWn†[inˉ⋅Dn,iD⊥μWn]\mathcal{B}_{n,\perp}^{\mu}\equiv\frac{1}{g}\frac{1}{\bar n\cdot\mathcal{P}}W_n^\dagger[i\bar n\cdot D_n,iD_{\perp}^{\mu}W_n] is the collinear gluon, and Pn⊥μ\mathcal{P}_{n\perp}^{\mu} form a complete set of collinear gauge invariant building blocks [74] in SCET.

Here, M^ENC[N]\widehat{\mathcal{M}}_{\mathrm{ENC}}^{[N]} is the measurement operator for the projected correlators, defined by its action on an arbitrary mm-particle state ∣Xm⟩\lvert X_m\rangle,

M^ENC[N]∣Xm⟩=∑1≤i1,…,iN≤m∏a=1NEiaQNΘ(xL−max⁡{xi1i2,xi1i3,…,xiN−1iN})∣Xm⟩(19)\widehat{\mathcal{M}}_{\mathrm{ENC}}^{[N]}\lvert X_m\rangle= \sum_{1\le i_1,\ldots,i_N\le m}\frac{\prod_{a=1}^{N}E_{i_a}}{Q^N}\Theta\left(x_L-\max\{x_{i_1i_2},x_{i_1i_3},\ldots,x_{i_{N-1}i_N}\}\right)\lvert X_m\rangle \tag*{(19)}

where xi1i2=(1−cos⁡θi1i2)/2≈θi1i22/4x_{i_1i_2}=(1-\cos\theta_{i_1i_2})/2\approx\theta_{i_1i_2}^2/4 is the angular distance between particle i1i_1 and i2i_2. In perturbation theory, such jet functions can be either computed via SCET Feynman rules or extracted from fixed-order ENC distributions.

At hadron colliders, the factorization theorem is similar to Eq. (17), except that the angle xLx_L is replaced by the rapidity-azimuthal distance RLR_L, i.e. xL→RL2/4x_L\to R_L^2/4, and one trades energies for transverse momenta as Ei→pi,TE_i\to p_{i,T}. We also need to convolve with the initial PDFs. This allows us to write the factorization formula as

dσ[N]dRL=ddRL[∑ab∫01dx1dx2 fa(x1,μF)fb(x2,μF)×∑c=q,g∫dz zN×dσ^ab→cγ+X(x1,x2,z,μF,μR,μJ,αs(μR))×J⃗c[N](ln⁡RL2z2pγ,T2μJ2,ln⁡μRμJ,αs(μR))].(20)\frac{d\sigma^{[N]}}{dR_L}=\frac{d}{dR_L}\left[\sum_{a b}\int_0^1 dx_1dx_2\,f_a(x_1,\mu_F)f_b(x_2,\mu_F)\times\sum_{c=q,g}\int dz\,z^N\times d\widehat{\sigma}_{ab\to c\gamma+X}(x_1,x_2,z,\mu_F,\mu_R,\mu_J,\alpha_s(\mu_R))\times\vec{J}_c^{[N]}\left(\ln\frac{R_L^2z^2p_{\gamma,T}^2}{\mu_J^2},\ln\frac{\mu_R}{\mu_J},\alpha_s(\mu_R)\right)\right]. \tag*{(20)}

Here dσ^ab→cγ+Xd\widehat{\sigma}_{ab\to c\gamma+X} is the hard function for producing a parton cc in association with a photon and it is fully differential in the Stripper framework. Note that the fiducial phase space introduced in § 2 is implicitly included in the hard functions and z=pi,T/pγ,Tz=p_{i,T}/p_{\gamma,T} is the energy fraction. Since the factorization is performed in the cumulant level, we take the derivative with respect to RLR_L to obtain the distribution. In addition to the renormalization scale μR\mu_R and PDF factorization scale μF\mu_F, we also introduce the μJ\mu_J scale for ENC factorization.

The ENC jet functions satisfy the modified DGLAP evolution,

ddln⁡μ2J⃗[N](ln⁡RL2z2pγ,T2μ2,αs(μ))=∫01dy yNJ⃗[N](ln⁡RL2y2z2pγ,T2μ2,αs(μ))⋅P^(y,αs(μ)),(21)\frac{d}{d\ln\mu^2}\vec{J}^{[N]}\left(\ln\frac{R_L^2z^2p_{\gamma,T}^2}{\mu^2},\alpha_s(\mu)\right)=\int_0^1dy\,y^N\vec{J}^{[N]}\left(\ln\frac{R_L^2y^2z^2p_{\gamma,T}^2}{\mu^2},\alpha_s(\mu)\right)\cdot\widehat{P}(y,\alpha_s(\mu)), \tag*{(21)}

where P^(y,αs(μ))\widehat{P}(y,\alpha_s(\mu)) is the singlet timelike splitting matrix, now known in full through three loops [75, 76] and partially at four-loop [77, 78, 79, 80, 81, 82]. Since obtaining a closed-form solution for DGLAP evolution is challenging, we use the expanded expression. In practice, we use the following vector ansatz in the flavor space at some scale μ\mu

J⃗[N](ln⁡RL2z2pγ,T2μ2,αs(μ))=∑n=0∞∑m=0n(αs(μ)4π)nj⃗n,m[N]m!ln⁡mRL2z2pγ,T2μ2,(22)\vec{J}^{[N]}\left(\ln\frac{R_L^2z^2p_{\gamma,T}^2}{\mu^2},\alpha_s(\mu)\right)=\sum_{n=0}^{\infty}\sum_{m=0}^{n}\left(\frac{\alpha_s(\mu)}{4\pi}\right)^n\frac{\vec{j}_{n,m}^{[N]}}{m!}\ln^m\frac{R_L^2z^2p_{\gamma,T}^2}{\mu^2}, \tag*{(22)}

plug into the RG equation in Eq. (21) and solve the coefficients j⃗n,m[N]\vec{j}_{n,m}^{[N]} by collecting the expression at each (n,m)(n,m). Note that j⃗n,0[N]\vec{j}_{n,0}^{[N]} are the jet boundary constants obtained from fixed-order calculations. The NLO result j⃗1,0[N]\vec{j}_{1,0}^{[N]} for arbitrary NN is calculated in Ref. [26]. At NNLO, j⃗2,0[N]\vec{j}_{2,0}^{[N]}, the N=3N=3 constant is derived in Ref. [49], and the 4≤N≤64\le N\le6 is computed recently in Ref. [39]. With these ingredients at hand, we now obtain the complete NNLL jet functions. In App. A, we provide the expressions for the ENC jet functions.

For resummation, we set μ=μJ\mu=\mu_J in the solution in Eq. (22) and re-express αs(μJ)\alpha_s(\mu_J) in terms of αs(μR)\alpha_s(\mu_R) via β\beta function. Plugging into Eq. (20), the remaining step for resummation is to perform the convolution with PDFs and hard functions, which we will discuss in the next section.

For resummation order, we define NkLL\mathrm{N}^k\mathrm{LL} through (k+1)(k+1)-loop timelike splitting kernels, kk-loop hard and jet boundary functions and (k+1)(k+1)-loop β\beta function, and match it to Nk−1LO\mathrm{N}^{k-1}\mathrm{LO} in fixed-order.

Note that for the log counting, we rewrite ln⁡(RL2z2pγ,T2/μ2)\ln(R_L^2 z^2 p_{\gamma,T}^2/\mu^2) as ln⁡RL2\ln R_L^2 and ln⁡(z2pγ,T2/μ2)\ln(z^2 p_{\gamma,T}^2/\mu^2), and count ln⁡RL2\ln R_L^2 as a large log, while keeping ln⁡(z2pγ,T2/μ2)\ln(z^2 p_{\gamma,T}^2/\mu^2) as O(1)\mathcal{O}(1) contribution. To the logarithmic order we claim, this will not change the predicted logarithms, but only change the higher-order terms, which is captured by the theory uncertainties.

A diagrammatic summary of the contributions needed to achieve a certain fixed-order or resummation precision within our perturbative counting is shown in Figure 3.

Figure 3. Illustration of the perturbative orders for the projected correlators. The first column shows example fixed-order contributions at O(αs1,2,3)\mathcal{O}(\alpha_s^{1,2,3}), and the remaining columns show their factorized counterparts, organized by the loop orders of the hard and jet functions, H(i)⊗J(j)\mathcal{H}^{(i)} \otimes\mathcal{J}^{(j)}. Contributions of the same color combine into the fixed-order and resummation accuracies listed in the legend, for both the cumulant Σ\Sigma and the differential distribution dσ/dRLd\sigma/dR_L.

Numerical Evaluation of Hard Functions

Now let us discuss the definition of the hard function and its numerical evaluation for pp→V+Xpp\to V+X process. In e+e−e^+e^- annihilation, the hard function for projected energy correlators is the singleinclusive hadron fragmentation function [83], which is also a vector in the flavor space and depends on an outgoing parton energy fraction x=2p⋅qQ2x = \frac{2p\cdot q}{Q^2}, where qq is the total momentum and pp is the parton momentum. At leading order, the e+e−e^+e^- hard function is H⃗ee(0)(x)={2δ(1−x),0}\vec{H}_{ee}^{(0)}(x) = \{2\delta(1-x),0\}, obtained from the Born process e+e−→qqˉe^+e^- \to q\bar{q}. At one-loop, it starts to have non-trivial energy dependence

12Hee,q(1)(x)=αs(Q)4πCF[(4π23−9)δ(1−x)+4[ln⁡(1−x)1−x]++(4ln⁡(x)−32)(2[11−x]+−x−1)−9x2−2(x+1)ln⁡(1−x)+72],\frac{1}{2}H_{ee,q}^{(1)}(x) = \frac{\alpha_s(Q)}{4\pi}C_F\left[\left(\frac{4\pi^2}{3}-9\right)\delta(1-x)+4\left[\frac{\ln(1-x)}{1-x}\right]_+ + \left(4\ln(x)-\frac{3}{2}\right)\left(2\left[\frac{1}{1-x}\right]_+ -x-1\right)-\frac{9x}{2}-2(x+1)\ln(1-x)+\frac{7}{2}\right],
Hee,g(1)(x)=αs(Q)4πCF[4(x2−2x+2)ln⁡(1−x)x+8(x2−2x+2)ln⁡(x)x].(23)H_{ee,g}^{(1)}(x)=\frac{\alpha_s(Q)}{4\pi}C_F\left[\frac{4(x^2-2x+2)\ln(1-x)}{x}+\frac{8(x^2-2x+2)\ln(x)}{x}\right]. \tag*{(23)}

The factor 1/21/2 in front of the quark channel indicates for identical contribution from anti-quark, since projected ENC jet function does not distinguish quark and anti-quark flavor. The two-loop correction is computed in Ref. [83], and very recently, the three-loop result is also available [84].

To compute the hard functions for pp→V+Xpp \to V+X process, one can use the implementation of fragmentation in Stripper [43, 44, 45, 47]. This implementation can be used to calculate convolutions of hard functions with arbitrary distribution-valued fragmentation functions. By taking the fragmentation functions to be δ\delta-distributions, one can numerically obtain all the partonic hard functions separately. Implementing various fiducial cuts as we wish using the Stripper implementation, we can compute the hard function dσVi+Xd\sigma_{V i+X}, the fiducial single-inclusive cross section for producing fragmenting parton ii in association with a vector boson VV, fully differential in the phase space of the final state. Here, we take V=γV=\gamma with the fiducial cuts described in § 2.

In practice, what we really need is this hard function convolved with the jet functions discussed in § 3.2. After the generalization performed in Ref. [47], the convolution can involve arbitrary functions of the renormalization and final-state factorization scales and a characteristic scale which can be an arbitrary function of the kinematics. As discussed in § 3.2, the jet RGEs are solved iteratively and truncated at a higher order. We have properly restored the dependence on the renormalization scale μR\mu_R and the ENC factorization scale μJ\mu_J.

So in the end, the leading-power computation reduces to computing weighted fiducial cross sections of the type

∫dσγi+X(pi,Tpγ,T)Nαsl(μR)ln⁡m(pi,T2μJ2)ln⁡n(μR2μJ2),(24)\int d\sigma_{\gamma i+X}\left(\frac{p_{i,T}}{p_{\gamma,T}}\right)^N \alpha_s^l(\mu_R)\ln^m\left(\frac{p_{i,T}^2}{\mu_J^2}\right)\ln^n\left(\frac{\mu_R^2}{\mu_J^2}\right), \tag*{(24)}

where dσγi+Xd\sigma_{\gamma i+X} is the fiducial single-inclusive cross section for producing parton ii in association with a photon, fully differential in the phase space of the final state, and we used z=pi,T/pγ,Tz=p_{i,T}/p_{\gamma,T}. We will be interested in NN-point correlators up to N=6N=6. For NNLL accuracy with our log counting, only m,n=0,1,2m,n=0,1,2 are needed. In fact, since we will only be interested in values of μR\mu_R and μJ\mu_J which are identical up to a constant factor, the weighting by arbitrary powers of ln⁡(μR2/μJ2)\ln(\mu_R^2/\mu_J^2) can be easily performed a posteriori, and only n=0n=0 needs to be explicitly computed during the Monte Carlo integration. ll starts at 0 and we truncate it at l=14l=14. We checked that this leads to a truncation error well below 1 permille for the values of RLR_L considered here.

Once all of these weighted cross sections are computed, they can be combined with the appropriate powers of ln⁡RL\ln R_L and the corresponding values of the perturbative coefficients of the jet functions to obtain the final results for the projected energy correlators. Note that although we compute the generalized moments numerically through the Stripper framework, the final resummation results are nevertheless analytic functions of RLR_L, allowing us to take derivatives and change the values of RLR_L considered in post. We further note that, due to the very inclusive nature of the moments, their Monte Carlo integration converges very quickly: only about 40k core-h were needed for the unmatched results presented in this work.

Non-perturbative Power Corrections

To obtain final predictions for the projected energy correlators, we must also include non-perturbative power corrections. The structure of the leading non-perturbative corrections to event-shape observables was understood in the seminal work of Korchemsky and Sterman [85], and this analysis has recently been generalized to the projected NN-point energy correlators [48, 86]. The leading non-perturbative power correction, suppressed by ΛQCD/pT\Lambda_{\mathrm{QCD}}/p_T, is characterized by the soft matrix element

Ω1κ≡1Nκ⟨0∣tr⁡[Y‾n†Ynˉ†ET(0)YnˉYn]∣0⟩,κ=q,g.(25)\Omega_{1\kappa} \equiv\frac{1}{N_\kappa}\langle0| \operatorname{tr}\left[\overline{Y}_n^\dagger Y_{\bar n}^\dagger\mathcal{E}_T(0)Y_{\bar n}Y_n\right] |0\rangle,\qquad\kappa=q,g . \tag*{(25)}

Here YnY_n and Y‾nˉ\overline{Y}_{\bar n} are soft Wilson lines along the two back-to-back lightlike directions nμn^\mu and nˉμ\bar n^\mu, and ET(0)\mathcal{E}_T(0) is the transverse energy flow operator evaluated at the origin. The color normalization factors are Nq=NcN_q=N_c and Ng=Nc2−1N_g=N_c^2-1 for the quark and gluon channels, respectively. Note that the Ω1κ\Omega_{1\kappa} are the same non-perturbative parameters that enter e+e−e^+e^- event shapes such as thrust, heavy jet mass, and the CC-parameter [87, 88, 89]. They are scheme dependent; throughout this work we adopt the MS‾\overline{\mathrm{MS}} scheme and denote them by Ω‾1κ\overline{\Omega}_{1\kappa}.

In the collinear factorization of § 3.2, the leading power correction can be incorporated into the jet functions as

Jκ[N](ln⁡RL2x2pT2μ2,μ)=J^κ[N](ln⁡RL2x2pT2μ2,μ)−NxpTRLJ^NPκ[N−1](ln⁡RL2x2pT2μ2,μ),(26)J^{\kappa[N]}\left(\ln\frac{R_L^2x^2p_T^2}{\mu^2},\mu\right)=\hat{J}^{\kappa[N]}\left(\ln\frac{R_L^2x^2p_T^2}{\mu^2},\mu\right)-\frac{N}{xp_TR_L}\hat{J}_{\mathrm{NP}}^{\kappa[N-1]}\left(\ln\frac{R_L^2x^2p_T^2}{\mu^2},\mu\right), \tag*{(26)}

where J^\hat{J} and J^NP\hat{J}_{\mathrm{NP}} are the perturbative and non-perturbative jet functions, respectively. Due to the prefactor 1/x1/x in the second term, J^NPκ[N−1]\hat{J}_{\mathrm{NP}}^{\kappa[N-1]} obeys the same evolution equation as the (N−1)(N-1)-point perturbative jet function J^κ[N−1]\hat{J}^{\kappa[N-1]}. At leading order in αs\alpha_s, the non-perturbative jet function at its canonical scale μJ0=pTRL\mu_J^0=p_TR_L are given by

J^NPκ[N−1](ln⁡RL2pT2(μJ0)2,μJ0)=Ω‾1κ+O(αs).(27)\hat{J}_{\mathrm{NP}}^{\kappa[N-1]}\left(\ln\frac{R_L^2p_T^2}{(\mu_J^0)^2},\mu_J^0\right)=\overline{\Omega}_{1\kappa}+\mathcal{O}(\alpha_s). \tag*{(27)}

In this work, we use Ω‾1q=0.305\overline{\Omega}_{1q}=0.305 GeV, obtained from the global thrust fit of Ref. [88] and converted to the MS‾\overline{\mathrm{MS}} scheme, and apply Casimir scaling for the gluon channel: Ω‾1g=CA/CFΩ‾1q\overline{\Omega}_{1g}=C_A/C_F\overline{\Omega}_{1q}. Since Ω‾1g\overline{\Omega}_{1g} has not been determined from data, we additionally vary it by a factor of two around this Casimir-scaling value and display the resulting uncertainty as a separate band in our final predictions below. The constants of the non-perturbative jet functions beyond leading order are unknown; we therefore model them by the corresponding constants of the (N−1)(N-1)-point perturbative jet function, rescaled by an overall factor of Ωˉ1κ\bar{\Omega}_{1\kappa}, and assign each coefficient an independent 100% uncertainty. Including these O(αs)\mathcal{O}(\alpha_s) corrections to the non-perturbative constants is particularly important and consistent for our observable. As discussed in § 2, the energy weights enhance configurations in which the boson is relatively soft compared to the hadronic system, and such configurations first appear in the O(αs)\mathcal{O}(\alpha_s) corrections. In the fixed-order singular expansion of the non-perturbative power correction, they therefore enter through the NLO hard function convolved with the LO non-perturbative jet function, and consistency of the order counting then requires the NLO non-perturbative jet function to be included as well. Evolving the non-perturbative jet functions with Eq. (21) and convolving them with the hard functions as in Eq. (20), we obtain the final prediction for the non-perturbative power correction to the ENC in the collinear limit.

In practice, these non-perturbative power corrections can be obtained using the same strategy of computing generalized moments used to compute the singular perturbative contribution described around Eq. (24). There are only two differences. First, the non-perturbative corrections to the NN-point correlator should be computed using the weight zN−1z^{N-1}, instead of zNz^N. Second, they should be computed using an additional factor 1/pγ,T1/p_{\gamma,T}. While the first difference is easily taken into account by just including one moment lower than needed for the purely perturbative parts (i.e. by also computing N=1N=1 in our case), the second difference can only be properly taken into account by computing a second set of moments, since pγ,Tp_{\gamma,T} changes from event to event, and this additional factor can only be included on the fly. We therefore compute all moments for N=1,…,6N=1,\ldots,6, both with and without an additional factor 1/pγ,T1/p_{\gamma,T}. The inclusion of the non-perturbative corrections is then fully analogous to the computation of the resummed singular contribution.

So far we have only considered the non-perturbative corrections to the singular distribution. In principle, one should also include the corrections to the non-singular contribution in the fixed-order region. However, the RLR_L range studied in this work lies mostly in the collinear regime, where the non-singular piece is very small (see § 4.1). We therefore simply extend the non-perturbative corrections obtained from the jet functions above across the fixed-order region. The final matched predictions for the projected energy correlators can then be written as

dσ[N]dRL=dσ^match[N]dRL+dσNP[N]dRL,withdσ^match[N]dRL=dσ^resum[N]dRL+dσ^ns[N]dRL.(28)\frac{d\sigma^{[N]}}{dR_L}=\frac{d\hat{\sigma}^{[N]}_{\mathrm{match}}}{dR_L}+\frac{d\sigma^{[N]}_{\mathrm{NP}}}{dR_L},\qquad\text{with}\qquad\frac{d\hat{\sigma}^{[N]}_{\mathrm{match}}}{dR_L}=\frac{d\hat{\sigma}^{[N]}_{\mathrm{resum}}}{dR_L}+\frac{d\hat{\sigma}^{[N]}_{\mathrm{ns}}}{dR_L}. \tag*{(28)}

Numerical Results

In this section, we apply our approach to derive numerical results for the projected energy correlators in γ+X\gamma+X events at the LHC. Our main results are shown in Figure 4, presenting both the differential distributions for the energy correlators and their ratios. In the following sections, we will present the detailed contributions from resummation, fixed-order matching, and non-perturbative corrections, as well as study the perturbative convergence.

Figure 4. Left panel: NLO+NNLL predictions for projected energy correlators with all values of NN, where the leading non-perturbative corrections are included. Right panel: The ratio of the NN-point correlators with respect to the 2-point correlator.

As a caveat throughout this section, all uncertainties are estimated through scale variations. While this is consistent with what is done in the literature, care must be taken in interpreting these estimates, particularly for the case of ratio observables. A better approach is provided by Theory Nuisance Parameters (TNPs) [90, 91], whose application we plan to investigate in the future.

Singular and Non-Singular Distributions

We begin by exploring the singular and non-singular distributions for the projected energy correlator observables in γ+X\gamma+X. This is important both for validating our factorization theorem and for understanding the size of the perturbative power corrections to it. In Figure 5, we show the complete numerical calculation of the two-point energy correlator, compared with the singular prediction derived from our factorization theorem and the numerically computed hard function, at both LO and NLO. We also show the non-singular contribution, which is defined as the difference between the complete fixed-order calculation and the singular contribution. It is also referred to as the perturbative power correction. Analogous distributions up to the projected six-point correlator can be found in App. § 8.§ 1.

Figure 5. The full, singular and non-singular two-point energy correlators at LO (left) and the NLO corrections (right). Analogous distributions for higher-point correlators can be found in App. § 8.§ 1.

First of all, we find good agreement between the complete numerical result and the singular result as predicted by our factorization theorem, in the limit RL→0R_L \to0. This provides a highly non-trivial check on our factorization theorem. It also illustrates the numerical stability of the subtraction scheme. Note that these plots show only the perturbative predictions. In reality, the distribution becomes non-perturbative at sufficiently small values of RL∼ΛQCD/pTR_L \sim\Lambda_{\mathrm{QCD}}/p_T and thus the result in Figure 5 should not be interpreted as an actual prediction for the distribution below this value, but rather as a test of our factorization theorem.

We can fit the numerical fixed-order data with the following form,

ENCfull(n)=ENCsing(n)×f(n)(RL)≡ENCsing(n)×∑i=0∞ai(n)RL2i,(29)\mathrm{ENC}_{\mathrm{full}}^{(n)} = \mathrm{ENC}_{\mathrm{sing}}^{(n)} \times f^{(n)}(R_L) \equiv\mathrm{ENC}_{\mathrm{sing}}^{(n)} \times\sum_{i=0}^{\infty} a_i^{(n)} R_L^{2i}, \tag*{(29)}

where nn represents the order O(αsn)\mathcal{O}(\alpha_s^n), ENCsing(n)\mathrm{ENC}_{\mathrm{sing}}^{(n)} is the singular expansion from the factorization theorem at order nn and ai(n)a_i^{(n)} are unknown coefficients. The non-singular contribution, namely the perturbative power corrections, is captured by the series expansion in RL2R_L^2.

The agreement in the RL→0R_L \to0 limit shown in Figure 5 implies that a0(n)=1a_0^{(n)} = 1 for any nn. In Figure 6, we perform the fit for a1(n)a_1^{(n)} and a2(n)a_2^{(n)} at both LO and NLO, for N=2N = 2–6. In [31], it was found that setting a1(n)=1/4a_1^{(n)} = 1/4 and a2(n)=1/24a_2^{(n)} = 1/24 captures the non-singular contribution very well for the EEC in e+e−e^+e^- collisions. Interestingly, for pp→γ+Xpp \to\gamma+ X, we find a1(1)≈1a_1^{(1)} \approx1 and a2(1)≈0a_2^{(1)} \approx0. This means that at LO, the coefficient of the next-to-leading power (NLP) correction is almost the same as that of the LP (singular) expansion, and the higher powers almost vanish. At NLO, we find a1(2)≈0.7a_1^{(2)} \approx0.7 and a2(2)≈0.1a_2^{(2)} \approx0.1, and thus the NLP coefficient is still very close to the LP one. As NN increases, the higher powers contribute more. Lastly, we find that at both orders the coefficients are, to a good approximation, linear in NN: the gray dashed lines in Figure 6 show linear fits, which describe the coefficients very well over the full range N=2N = 2–66. This linearity could potentially be used to recover the full RLR_L-dependence for any NN from the factorization theorem together with calculations at just two values of NN.

Fitted coefficients at LO and NLO as a function of N

Figure 6. Coefficients a1(n)a_1^{(n)} (red) and a2(n)a_2^{(n)} (blue) at LO (left) and NLO (right) as a function of NN. The values are obtained by fitting to numerical data from Stripper. The gray dashed lines show linear fits in NN.

Matched resummation results

In this subsection, we present results for matched projected energy correlators in γ+X\gamma+ X events at the LHC. We study in detail the effect of the matching, both on the full energy correlator distributions and on the ratio observables. This is of particular interest in light of recent strong coupling extractions from ratios of projected correlators [32, 49]: the scaling region from which αs\alpha_s is extracted can receive important contributions from non-singular terms, which must be accounted for in a precision extraction. Our results are the first complete matched calculations of projected energy correlators at this order at hadron colliders.

The matching formula for combining resummation and fixed-order is given in Eq. (28). In principle, we need to design a profile function (scale) to turn off the resummation properly when the singular contribution is no longer dominant [87, 88, 89, 92] (for an application of profile scales to the track-based EEC, see Ref. [93]). Given that we are focusing primarily on the collinear region in this work, we do not turn off the resummation but instead add the non-singular corrections additively as in Eq. (28). Note that the non-singular contribution, dσ^ns[N]=dσ^FO[N]−dσ^sing[N]d\hat{\sigma}_{\mathrm{ns}}^{[N]} = d\hat{\sigma}_{\mathrm{FO}}^{[N]} - d\hat{\sigma}_{\mathrm{sing}}^{[N]}, is precisely the perturbative power correction fit in Eq. (29). We leave the design of profile scales to future work.

In Figure 7 we show results for the matched energy correlators in γ+X\gamma+ X for two- and three-point energy correlators, and analogous results up to six-point projected correlators can be found in App. § 8.§ 2. We present three resummation orders, LL, LO+NLL and NLO+NNLL. For comparison, we also include two fixed-order distributions. Aside from the lowest-order predictions, LO and LL, we observe good perturbative convergence: the scale uncertainty bands of the NLO, LO+NLL and NLO+NNLL predictions all overlap and the uncertainties are reduced by a factor of 2-3, down to about 10%, when going from NLO or LO+NLL to NLO+NNLL. There are two exceptions to the good perturbative behavior. First, the NLO prediction starts to deviate from the resummed predictions in the small-angle limit. Of course, this is expected, since fixed-order calculations break down in that limit. The second exception is the large-angle limit, where we observe large perturbative corrections going from LO+NLL to NLO+NNLL. In this region, however, the distribution is dominated by the fixed-order contribution, for which large NLO corrections are not unusual. In the small angle regime, these distributions exhibit the characteristic enhancement associated with a scaling behavior in the collinear limit. This behavior is identical to their behavior in e+e−e^{+}e^{-} collisions, or within jets.

Figure 7. Differential matched distributions for the two-point (left) and three-point (right) correlators through NLO+NNLL. Analogous distributions for higher-point correlators can be found in App. § 8.§ 2.

In Figure 8 we show ratios of the three- through six-point projected correlators with respect to the two-point correlator at different orders. These ratios were originally proposed in Ref. [26] to isolate the scaling behavior in the collinear limit of the energy correlators, analogous to three/two jet ratios. Again, we include fixed-order curves together with the resummed distributions. Note that the fixed-order results do not even qualitatively describe the shape of the distribution, indicating that resummation is essential to obtain the correct behavior of the energy correlators in the collinear limit.

Figure 8. The ratios of the differential NN-point correlators to the differential two-point correlator through NLO(+NNLL).

To assess the stability of the slope, we compare these ratios (dσENC/dRL)/(dσE2C/dRL)(d\sigma^{\mathrm{ENC}}/dR_L)/(d\sigma^{\mathrm{E2C}}/dR_L) at each order with the highest-order NLO+NNLL prediction. In Figure 9, we show the cases of the three- and four-point correlators. For 10−2<RL<10−110^{-2} < R_L < 10^{-1}, avoiding the non-perturbative and fixed-order regimes, the slope changes by about 10% between LO+NLL and NLO+NNLL for the three-point case, indicating an uncertainty on the slope smaller than 10% at NLO+NNLL, consistent with the scale uncertainties at that order, which correspond to an uncertainty on the slope of about 7%. For the four-point case, the slope only changes negligibly between LO+NLL and NLO+NNLL, while the scale uncertainties indicate a roughly 5% uncertainty on the slope at NLO+NNLL. For both the three-point and the four-point case, the change in slope between LL and LO+NLL is much larger, on the order of 30%, than the change between LO+NLL and NLO+NNLL, suggesting good perturbative convergence.

Figure 9. The ratios of the differential three- (left) and four-point (right) correlators to the differential 2-point correlator through NLO+NNLL, shown as differences relative to the NLO+NNLL curves. Analogous distributions for higher-point correlators can be found in App. § 8.

In Figure 10, we investigate the impact of matching corrections, namely the non-singular contributions. For both the EEC and E3C, we plot the pure NNLL resummation and the matched NLO+NNLL result, normalized to the former. For RL<0.3R_L < 0.3, we find that the matched resummed distributions are close to the pure resummation, while at larger RLR_L, non-singular contributions from the fixed-order calculation become important. In particular, when RLR_L is O(1)\mathcal{O}(1), the matching correction reaches around 50% for N=2N = 2 and N=3N = 3-point correlators. As NN increases, the non-singular correction also increases, approaching up to 80% for the six-point correlator. The distributions beyond three-point are provided in App. § 8.§ 4. In Figure 11, we provide the identical figures for the ratio correlators. In both cases, we find that the non-singular corrections are significantly smaller: when RLR_L is O(1)\mathcal{O}(1), the E3C/EEC ratio receives ≲4%\lesssim4\% and the E4C/EEC ratio ≲7%\lesssim7\% corrections from matching.

Figure 10. Impact of matching corrections in ENC resummation. The left panel shows EEC and the right panel is E3C. In the top panel, we compare the NLO fixed-order with its singular expansion, normalized to the latter. In the lower panel, we show the matched NLO+NNLL and pure NNLL resummations, both normalized to the latter. We find important non-singular corrections at larger RLR_L.

Figure 11. The impact of matching corrections to resummation similar to Figure 10, but for the ratio correlators.

Complete Predictions with Non-Perturbative Corrections

Finally, in Figure 12 we provide complete predictions for the ratios of the projected energy correlators in γ+X\gamma+ X events, including fixed-order matching, resummation, and non-perturbative corrections. These provide state-of-the-art predictions for energy correlators at hadron colliders. The non-perturbative corrections suppress the ratios at small RLR_L, where the leading power correction to the distribution, scaling as ∼NΩˉ1κ/(pγ,TRL2)\sim N\bar{\Omega}_{1\kappa}/(p_{\gamma,T}R_L^2), is largest, while leaving the large-angle region essentially unchanged; this effectively increases the slope of the ratio distributions. The effect grows with NN and is sizable, reaching several tens of percent at RL∼10−2R_L \sim10^{-2} and remaining at the ∼10\sim10–20%20\% level in the scaling region around RL∼10−1R_L \sim10^{-1}. An accurate treatment of the non-perturbative corrections is thus essential for a reliable extraction of αs\alpha_s from the slope of these ratios as neglecting them would lead to an erroneously larger extracted value of αs\alpha_s. Besides the scale uncertainty, our predictions include the two non-perturbative uncertainties discussed in § 3.4, from the modeled higher-order constants of the non-perturbative jet functions and from the variation of Ωˉ1g\bar{\Omega}_{1g} around its Casimir-scaling value. We therefore conclude that for complete predictions of the projected energy correlators one needs matching, resummation, and non-perturbative corrections. The framework developed in this paper achieves all three, opening the door to precision phenomenology with energy correlators.

Figure 12. The ratios of the differential NN-point correlators to the differential two-point correlator at NLO+NNLL with and without non-perturbative corrections. The red band indicates the scale uncertainty, while the green and blue bands indicate the uncertainties from the unknown higher-order constants of the non-perturbative jet functions and from the unknown value of Ω1g\Omega_{1g}, respectively, both estimated as described in § 3.4.

Conclusions and Outlook

In this paper we have proposed energy correlators measured on the radiation recoiling against a vector boson as a precision QCD observable at the LHC. Achieving precision predictions for this observable requires combining state-of-the-art ingredients from scattering amplitudes, numerical subtraction schemes, and factorization/resummation. To enable the calculation of the energy correlator for a broad set of final states at hadron colliders, we have developed a modular framework, combining analytic resummation with a numerical subtraction scheme based on Stripper [40, 41, 42]. This makes matched, resummed calculations of energy correlators possible in a wide class of processes for which the NLO/NNLO matrix elements are known.

We illustrated our framework by presenting results for the projected energy correlators at NLO+NNLL, incorporating the leading non-perturbative corrections. This is the first complete calculation of the energy correlator at a hadron collider that incorporates matching at this order. Since energy correlators are being used [32] to measure the strong coupling constant at hadron colliders, it is essential to understand the importance of matching corrections. We performed detailed numerical studies of the impact of fixed-order matching and resummation on our results in an experimentally realistic phase space. We considered both the energy correlator distributions themselves and their ratios. For the distributions, the matching corrections grow to the 50–80% level at wide angles, while for the ratios they are suppressed by roughly an order of magnitude; the scale uncertainties of the distributions at NLO+NNLL shrink to about 10%. The non-perturbative corrections are also significant, suppressing the ratios at small RLR_L and thereby increase their slope, and neglecting them would bias an extraction of αs\alpha_s toward larger values.

There are several clear ways in which our calculation can be extended. Most importantly, the primary goal of this paper, namely developing an interface between analytic resummation for energy correlators and numerical hard functions, was motivated by the desire to extend the calculations presented in this paper to include NNLO matching. The availability of NNLO 2→32 \to3 scattering amplitudes (as discussed in detail in the introduction) makes this feasible, and we hope it will provide a beautiful illustration of their phenomenological importance. Numerically, this will be challenging, as it will require the evaluation of these amplitudes in singular regions of phase space. To assess this feasibility, we performed a detailed analysis of the structure of perturbative power corrections for the EEC observable. The size of perturbative power corrections, not captured by our leading power factorization theorem, dictates how far into the singular region the fixed-order calculation must be performed. Our study supports the feasibility of an NNLO calculation, since it will not need to be run too deep into singular regions.

Another clear way in which our analysis will need to be improved is to better understand the behavior of theoretical uncertainties for the ratios of projected energy correlators. In this paper, these were estimated using scale variations; however, this is clearly not ideal due to the strong correlations between the observables. A better approach is provided by Theory Nuisance Parameters (TNPs) [90, 91], whose application we plan to investigate further in the future.

Although we have highlighted the clean nature of the V+XV+X process, our approach is also applicable to the study of energy correlators in identified jets. Since these have been directly used for extractions of the strong coupling constant, it will be important to study the effect of matching and perturbative power corrections for jets identified with the anti-kTk_T algorithm [53]. This will be presented in future work.

We believe that the framework developed in this paper will enable the calculation of energy correlators for a wide variety of processes at the LHC and beyond.

Acknowledgments

We thank Jennifer Roloff, Matt Leblanc, Simon Rothman, Gavin Salam for useful discussions. This work was performed in part using the Cambridge Service for Data Driven Discovery (CSD3), part of which is operated by the University of Cambridge Research Computing on behalf of the STFC DiRAC HPC Facility (www.dirac.ac.uk). The DiRAC component of CSD3 was supported by STFC grants ST/P002307/1, ST/R002452/1 and ST/R00689X/1. T.G. is supported by STFC consolidated HEP theory grants ST/T000694/1 and ST/X000664/1. K.L. is supported by the U.S. Department of Energy under contract DE-AC02-06CH11357. I.M. is supported by the DOE Early Career Award DE-SC0025581, and the Sloan Foundation. R.P is funded by the European Union (ERC, STAPLE, 101222253). Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council. Neither the European Union nor the granting authority can be held responsible for them. X.Z. is supported by the MIT Pappalardo Fellowship. This work was also supported in part by the Aspen Center for Physics under NSF grant PHY-2210452 and by the Kavli Institute for Theoretical Physics under NSF grant PHY-2309135. We also thank the Erwin Schrödinger International Institute for Mathematics and Physics for support.

Perturbative Ingredients

In this appendix, we provide the expressions for ENC jet functions in Eq. (22). First of all, the anomalous dimensions in collinear resummation are defined via the timelike splitting functions,

γ˙ij(L)(N)=−∫01dx xNPij(L)(x),γ¨ij(L)(N)=−∫01dx ln⁡x xNPij(L)(x),γ...ij(L)(N)=−∫01dx ln⁡2x xNPij(L)(x).(30)\dot{\gamma}_{ij}^{(L)}(N)=-\int_{0}^{1}\mathrm{d}x\,x^{N}P_{ij}^{(L)}(x), \\ \ddot{\gamma}_{ij}^{(L)}(N)=-\int_{0}^{1}\mathrm{d}x\,\ln x\,x^{N}P_{ij}^{(L)}(x), \\ \dddot{\gamma}_{ij}^{(L)}(N)=-\int_{0}^{1}\mathrm{d}x\,\ln^{2}x\,x^{N}P_{ij}^{(L)}(x). \tag*{(30)}

where we expand the splitting functions in αs/(4π)\alpha_s/(4\pi),

Pij(x)=∑L=0∞(αs4π)L+1Pij(L)(x).(31)P_{ij}(x)=\sum_{L=0}^{\infty}\left(\frac{\alpha_s}{4\pi}\right)^{L+1}P_{ij}^{(L)}(x). \tag*{(31)}

The analytic expressions of γij(L)\gamma_{ij}^{(L)} up to three loops and up to N=6N=6 (for 6-point correlators) can be found in Ref. [39].

As in Eq. (22), the resummed jet functions can be expressed in terms of αs(μ)\alpha_s(\mu) series,

j⃗[N](ln⁡RLQ2μ2,μ)=∑n=0∞∑m=0n(αs(μ)4π)nj⃗n,m[N]m!ln⁡mRLQ2μ2.(32)\vec{j}^{[N]}\left(\ln\frac{R_LQ^2}{\mu^2},\mu\right)=\sum_{n=0}^{\infty}\sum_{m=0}^{n}\left(\frac{\alpha_s(\mu)}{4\pi}\right)^n\frac{\vec{j}_{n,m}^{[N]}}{m!}\ln^m\frac{R_LQ^2}{\mu^2}. \tag*{(32)}

Here we present the coefficients j⃗n,m[N]\vec{j}_{n,m}^{[N]} for the first three orders. For the quark channel, we find,

jq,0,0[N]=jq,0[N],jq,1,1[N]=γgq[0] jq,0[N]+γqq[0] jq,0[N],jq,1,0[N]=jq,1[N],jq,2,2[N]=γgg[0]γgq[0]jg,0[N]−β0γgq[0]jg,0[N]+γgq[0]γqq[0]jg,0[N]+γgq[0]γqg[0]jq,0[N]−β0γqq[0]jq,0[N]+(γqq[0])2jq,0[N],jq,2,1[N]=γgg[0]γ˙gq[0]jg,0[N]+γgq[0]γ˙qq[0]jg,0[N]+γqg[0]γ˙gq[0]jq,0[N]+γqq[0]γ˙qq[0]jq,0[N]+γgq[0]jg,1[N]+γqq[0]jq,1[N]+γgq[1]jg,0[N]+γqq[1]jq,0[N]−β0jq,1[N],jq,2,0[N]=jq,2[N],jq,3,3[N]=−3β0γgg[0]γgq[0]jg,0[N]+(γgq[0])2γqq[0]jg,0[N]+γgg[0]γgq[0]γqq[0]jg,0[N]+2β02γgq[0]jg,0[N].j_{q,0,0}^{[N]}=j_{q,0}^{[N]}, \\ j_{q,1,1}^{[N]}=\gamma_{gq}^{[0]}\,j_{q,0}^{[N]}+\gamma_{qq}^{[0]}\,j_{q,0}^{[N]}, \\ j_{q,1,0}^{[N]}=j_{q,1}^{[N]}, \\ j_{q,2,2}^{[N]}=\gamma_{gg}^{[0]}\gamma_{gq}^{[0]}j_{g,0}^{[N]}-\beta_0\gamma_{gq}^{[0]}j_{g,0}^{[N]}+\gamma_{gq}^{[0]}\gamma_{qq}^{[0]}j_{g,0}^{[N]}+\gamma_{gq}^{[0]}\gamma_{qg}^{[0]}j_{q,0}^{[N]}-\beta_0\gamma_{qq}^{[0]}j_{q,0}^{[N]}+\left(\gamma_{qq}^{[0]}\right)^2j_{q,0}^{[N]}, \\ j_{q,2,1}^{[N]}=\gamma_{gg}^{[0]}\dot{\gamma}_{gq}^{[0]}j_{g,0}^{[N]}+\gamma_{gq}^{[0]}\dot{\gamma}_{qq}^{[0]}j_{g,0}^{[N]}+\gamma_{qg}^{[0]}\dot{\gamma}_{gq}^{[0]}j_{q,0}^{[N]}+\gamma_{qq}^{[0]}\dot{\gamma}_{qq}^{[0]}j_{q,0}^{[N]}+\gamma_{gq}^{[0]}j_{g,1}^{[N]}+\gamma_{qq}^{[0]}j_{q,1}^{[N]}+\gamma_{gq}^{[1]}j_{g,0}^{[N]}+\gamma_{qq}^{[1]}j_{q,0}^{[N]}-\beta_0j_{q,1}^{[N]}, \\ j_{q,2,0}^{[N]}=j_{q,2}^{[N]}, \\ j_{q,3,3}^{[N]}=-3\beta_0\gamma_{gg}^{[0]}\gamma_{gq}^{[0]}j_{g,0}^{[N]}+\left(\gamma_{gq}^{[0]}\right)^2\gamma_{qq}^{[0]}j_{g,0}^{[N]}+\gamma_{gg}^{[0]}\gamma_{gq}^{[0]}\gamma_{qq}^{[0]}j_{g,0}^{[N]}+2\beta_0^2\gamma_{gq}^{[0]}j_{g,0}^{[N]}.
+(γqq[0])2γqq[0]j˙q,0[N]−3β0γqq[0]γqq[0]jq,0[N]+γqq[0](γqq[0])2jq,0[N]+γqq[0]γqq[0]γqq[0]jq,0[N]+2γqq[0]γqq[0]γqq[0]jq,0[N]+2β02γqq[0]jq,0[N]−3β0(γqq[0])2jq,0[N]+(γqq[0])3jq,0[N],jq,3,2[N]=−3β0γ˙gg[0]γqq[0]jq,0[N]+γgg[0]γqq[0]γ˙qq[0]jq,0[N]+(γgg[0])2γqq[0]jq,0[N]+γgg[0]γqq[0]γqq[0]jq,0[N]+γgg[0]γqq[0]γ˙qq[0]jq,0[N]+γgg[0]γqq[0]γ˙qq[0]jq,0[N]+(γgg[0])2γ˙qq[0]jq,0[N]−3β0γgg[0]γqq[0]jq,0[N]+2γqq[0]γqq[0]γqq[0]jq,0[N]+γgg[0]γqq[0]γ˙qq[0]jq,0[N]+γqq[0]γqq[0]γ˙qq[0]jq,0[N]−3β0γqq[0]γqq[0]jq,0[N]+2(γqq[0])2γqq[0]jq,0[N]+γgg[0]γqg[1]jg,0[N]+γqg[0]γgg[1]jg,0[N]+γqq[0]γqg[1]jg,0[N]+γqq[0]γqq[1]jq,0[N]+γqg[0]γqq[1]jq,0[N]+γqq[0]γqg[1]jq,0[N]+2γqq[0]γqg[1]jg,0[N]−β1γqq[0]jq,0[N]−β1γqq[0]jq,0[N]+γgg[0]γqq[0]jq,1[N]−3β0γqq[0]jq,1[N]+(γqq[0])2jq,1[N]−2β0γqg[1]jg,0[N]−2β0γqq[1]jq,0[N]+2β02jq,1[N],(33)\begin{aligned} &+ \left(\gamma_{qq}^{[0]}\right)^2 \gamma_{qq}^{[0]} \dot{j}_{q,0}^{[N]} - 3\beta_0 \gamma_{qq}^{[0]} \gamma_{qq}^{[0]} j_{q,0}^{[N]} + \gamma_{qq}^{[0]} \left(\gamma_{qq}^{[0]}\right)^2 j_{q,0}^{[N]} + \gamma_{qq}^{[0]} \gamma_{qq}^{[0]} \gamma_{qq}^{[0]} j_{q,0}^{[N]} \\ &+ 2\gamma_{qq}^{[0]} \gamma_{qq}^{[0]} \gamma_{qq}^{[0]} j_{q,0}^{[N]} + 2\beta_0^2 \gamma_{qq}^{[0]} j_{q,0}^{[N]} - 3\beta_0 \left(\gamma_{qq}^{[0]}\right)^2 j_{q,0}^{[N]} + \left(\gamma_{qq}^{[0]}\right)^3 j_{q,0}^{[N]}, \\ j_{q,3,2}^{[N]}={}&-3\beta_0 \dot{\gamma}_{gg}^{[0]} \gamma_{qq}^{[0]} j_{q,0}^{[N]}+\gamma_{gg}^{[0]}\gamma_{qq}^{[0]}\dot{\gamma}_{qq}^{[0]}j_{q,0}^{[N]}+\left(\gamma_{gg}^{[0]}\right)^2\gamma_{qq}^{[0]}j_{q,0}^{[N]}+\gamma_{gg}^{[0]}\gamma_{qq}^{[0]}\gamma_{qq}^{[0]}j_{q,0}^{[N]}\\ &+\gamma_{gg}^{[0]}\gamma_{qq}^{[0]}\dot{\gamma}_{qq}^{[0]}j_{q,0}^{[N]}+\gamma_{gg}^{[0]}\gamma_{qq}^{[0]}\dot{\gamma}_{qq}^{[0]}j_{q,0}^{[N]}+\left(\gamma_{gg}^{[0]}\right)^2\dot{\gamma}_{qq}^{[0]}j_{q,0}^{[N]}-3\beta_0\gamma_{gg}^{[0]}\gamma_{qq}^{[0]}j_{q,0}^{[N]}\\ &+2\gamma_{qq}^{[0]}\gamma_{qq}^{[0]}\gamma_{qq}^{[0]}j_{q,0}^{[N]}+\gamma_{gg}^{[0]}\gamma_{qq}^{[0]}\dot{\gamma}_{qq}^{[0]}j_{q,0}^{[N]}+\gamma_{qq}^{[0]}\gamma_{qq}^{[0]}\dot{\gamma}_{qq}^{[0]}j_{q,0}^{[N]}-3\beta_0\gamma_{qq}^{[0]}\gamma_{qq}^{[0]}j_{q,0}^{[N]}\\ &+2\left(\gamma_{qq}^{[0]}\right)^2\gamma_{qq}^{[0]}j_{q,0}^{[N]}+\gamma_{gg}^{[0]}\gamma_{qg}^{[1]}j_{g,0}^{[N]}+\gamma_{qg}^{[0]}\gamma_{gg}^{[1]}j_{g,0}^{[N]}+\gamma_{qq}^{[0]}\gamma_{qg}^{[1]}j_{g,0}^{[N]}+\gamma_{qq}^{[0]}\gamma_{qq}^{[1]}j_{q,0}^{[N]}\\ &+\gamma_{qg}^{[0]}\gamma_{qq}^{[1]}j_{q,0}^{[N]}+\gamma_{qq}^{[0]}\gamma_{qg}^{[1]}j_{q,0}^{[N]}+2\gamma_{qq}^{[0]}\gamma_{qg}^{[1]}j_{g,0}^{[N]}-\beta_1\gamma_{qq}^{[0]}j_{q,0}^{[N]}-\beta_1\gamma_{qq}^{[0]}j_{q,0}^{[N]}\\ &+\gamma_{gg}^{[0]}\gamma_{qq}^{[0]}j_{q,1}^{[N]}-3\beta_0\gamma_{qq}^{[0]}j_{q,1}^{[N]}+\left(\gamma_{qq}^{[0]}\right)^2j_{q,1}^{[N]}-2\beta_0\gamma_{qg}^{[1]}j_{g,0}^{[N]}-2\beta_0\gamma_{qq}^{[1]}j_{q,0}^{[N]}+2\beta_0^2j_{q,1}^{[N]}, \tag*{(33)} \end{aligned}

where we have omitted the NN dependence in the anomalous dimensions. For the gluon channel, we have the same expressions with the following replacements:

jq,i[N]↔jg,i[N],i=0,1,2j_{q,i}^{[N]}\leftrightarrow j_{g,i}^{[N]},\qquad i=0,1,2
γqq[i]↔γgg[i],γqg[i]↔γgq[i],γ˙qq[i]↔γ˙gg[i],γ˙qg[i]↔γ˙gq[i],γ¨qq[i]↔γ¨gg[i],γ¨qg[i]↔γ¨gq[i].(34)\gamma_{qq}^{[i]}\leftrightarrow\gamma_{gg}^{[i]},\qquad\gamma_{qg}^{[i]}\leftrightarrow\gamma_{gq}^{[i]},\qquad\dot{\gamma}_{qq}^{[i]}\leftrightarrow\dot{\gamma}_{gg}^{[i]},\qquad\dot{\gamma}_{qg}^{[i]}\leftrightarrow\dot{\gamma}_{gq}^{[i]},\qquad\ddot{\gamma}_{qq}^{[i]}\leftrightarrow\ddot{\gamma}_{gg}^{[i]},\qquad\ddot{\gamma}_{qg}^{[i]}\leftrightarrow\ddot{\gamma}_{gq}^{[i]} . \tag*{(34)}

jq,i[N]j_{q,i}^{[N]} and jg,i[N]j_{g,i}^{[N]} with i=0,1,2i=0,1,2 are the perturbative jet constants. At tree-level, we have jq,0[N]=1j_{q,0}^{[N]}=1 and jg,0[N]=1j_{g,0}^{[N]}=1. At one-loop, the result with full NN dependence is [26]

jq,1[N]=CF((N−1)(N(13N+37)+12)N(N+1)2−4(ψ(0)(N)+γ)2−12(ψ(0)(N)+γ)N+1+12ψ(1)(N)−2π2)j_{q,1}^{[N]}=C_F\left(\frac{(N-1)(N(13N+37)+12)}{N(N+1)^2}-4\left(\psi^{(0)}(N)+\gamma\right)^2-\frac{12\left(\psi^{(0)}(N)+\gamma\right)}{N+1}+12\psi^{(1)}(N)-2\pi^2\right)
jg,1[N]=CA{19N(N+1)2(N+2)2(N+3)2[2N(67N6+804N5+3634N4+7380N3+4723N2−5520N−9π2(N+1)2(N+2)2(N+3)2−8712)−4752]−4(ψ(0)(N)+γ)2−8(N(2N+9)+11)(ψ(0)(N)+γ)(N+1)(N+2)(N+3)+12ψ(1)(N)}+nf9(N+1)2(N+2)2(N+3)2(−23N6−276N5−1190N4−2376N3−1703N2+1644N+864N+36(N+1)(N+2)(N+3)(N(N+3)+4)(ψ(0)(N)+γ)+3060).(35)\begin{aligned} j_{g,1}^{[N]}=C_A\{\frac{1}{9N(N+1)^2(N+2)^2(N+3)^2}[2N(67N^6+804N^5+3634N^4+7380N^3 \\ &+4723N^{2}-5520N-9\pi^{2}(N+1)^{2}(N+2)^{2}(N+3)^{2}-8712)-4752] -4(\psi^{(0)}(N)+\gamma)^{2} -\frac{8(N(2N+9)+11)(\psi^{(0)}(N)+\gamma)}{(N+1)(N+2)(N+3)}+12\psi^{(1)}(N)\} +\frac{n_f}{9(N+1)^{2}(N+2)^{2}(N+3)^{2}}(-23N^{6}-276N^{5}-1190N^{4}-2376N^{3}-1703N^{2} +1644N+\frac{864}{N}+36(N+1)(N+2)(N+3)(N(N+3)+4)(\psi^{(0)}(N)+\gamma)+3060). \tag*{(35)} \end{aligned}

At two-loop, we don’t have the closed form in NN; instead, we compute them semi-analytically up to N=6N=6 [39]

jq,2[3]=CFTFnf(98.416094±0.001584)+CF2(171.153969±0.018553)+CFCA(−210.211343±0.008717),jq,2[4]=CFTFnf(124.967567±0.002166)+CF2(254.438849±0.018860)+CFCA(−269.979451±0.009042),jq,2[5]=CFTFnf(147.891331±0.002602)+CF2(339.713439±0.021279)+CFCA(−323.170146±0.009702),jq,2[6]=CFTFnf(168.459145±0.003037)+CF2(426.020756±0.024441)+CFCA(−371.824995±0.011163),(36)\begin{aligned} j_{q,2}^{[3]}&=C_FT_Fn_f(98.416094\pm0.001584)+C_F^{2}(171.153969\pm0.018553)\\ &\quad+C_FC_A(-210.211343\pm0.008717),\\ j_{q,2}^{[4]}&=C_FT_Fn_f(124.967567\pm0.002166)+C_F^{2}(254.438849\pm0.018860)\\ &\quad+C_FC_A(-269.979451\pm0.009042),\\ j_{q,2}^{[5]}&=C_FT_Fn_f(147.891331\pm0.002602)+C_F^{2}(339.713439\pm0.021279)\\ &\quad+C_FC_A(-323.170146\pm0.009702),\\ j_{q,2}^{[6]}&=C_FT_Fn_f(168.459145\pm0.003037)+C_F^{2}(426.020756\pm0.024441)\\ &\quad+C_FC_A(-371.824995\pm0.011163), \tag*{(36)} \end{aligned}
jg,2[3]=CFTFnf(−16.427368±0.001614)+CATFnf(140.389919±0.003094)+CA2(−47.854829±0.021790)+nf2TF2(4688375−8π215),jg,2[4]=CFTFnf(−21.592489±0.001800)+CATFnf(182.634493±0.003985)+CA2(−27.054523±0.022288)+nf2TF2(526896563472875−608π2945),jg,2[5]=CFTFnf(−25.583483±0.001906)+CATFnf(219.793744±0.007128)+CA2(2.207608±0.036991)+nf2TF2(11872972694575−136π2189),jg,2[6]=CFTFnf(−28.753959±0.002008)+CATFnf(253.496395±0.005500)+CA2(37.086330±0.026721)+nf2TF2(34827077218753525−440π2567).(37)\begin{aligned} j_{g,2}^{[3]}&=C_FT_Fn_f(-16.427368\pm0.001614)+C_AT_Fn_f(140.389919\pm0.003094)\\ &\quad+C_A^{2}(-47.854829\pm0.021790)+n_f^{2}T_F^{2}\left(\frac{4688}{375}-\frac{8\pi^{2}}{15}\right),\\ j_{g,2}^{[4]}&=C_FT_Fn_f(-21.592489\pm0.001800)+C_AT_Fn_f(182.634493\pm0.003985)\\ &\quad+C_A^{2}(-27.054523\pm0.022288)+n_f^{2}T_F^{2}\left(\frac{52689656}{3472875}-\frac{608\pi^{2}}{945}\right),\\ j_{g,2}^{[5]}&=C_FT_Fn_f(-25.583483\pm0.001906)+C_AT_Fn_f(219.793744\pm0.007128)\\ &\quad+C_A^{2}(2.207608\pm0.036991)+n_f^{2}T_F^{2}\left(\frac{11872972}{694575}-\frac{136\pi^{2}}{189}\right),\\ j_{g,2}^{[6]}&=C_FT_Fn_f(-28.753959\pm0.002008)+C_AT_Fn_f(253.496395\pm0.005500)\\ &\quad+C_A^{2}(37.086330\pm0.026721)+n_f^{2}T_F^{2}\left(\frac{348270772}{18753525}-\frac{440\pi^{2}}{567}\right). \tag*{(37)} \end{aligned}

Additional Plots

In this appendix, we provide additional plots supplementing those in the main text: the comparisons shown there are extended to the complete set of projected correlators up to six points and, where noted, to additional perturbative orders and to cumulative distributions. The appendix is organized into subsections mirroring the structure of Sec. 4.

Singular and Non-Singular Distributions

In Figure 13, we extend the comparison of the full, singular, and non-singular distributions in Figure 5 to all projected correlators up to six points.

Figure 13. The comparisons of the full, singular and non-singular projected NN-point correlators, for N=2N = 2–66 (top to bottom), at LO (left) and NLO (right).

Matched Distributions

In Figure 14, we show the cumulative and differential matched distributions for all projected correlators up to six points, extending Figure 7.

Figure 14. The cumulative (left) and differential (right) NN-point correlators through (N)NLO(+NNLL), for N=2N = 2–6 (top to bottom).

Slope Stability of the Ratio Distributions

In Figure 15, the ratio distributions at each order are compared with the highest-order NLO+NNLL predictions for all ratio correlators (N=3N = 3–66), extending Figure 9.

Figure 15. The ratios of the differential NN-point correlators to the differential two-point correlator through NLO+NNLL, shown as differences relative to the NLO+NNLL curves.

Impact of Matching Corrections

In Figure 16 and Figure 17, we show the impact of the matching corrections on the differential distributions (N=2N = 2–66) and on the ratio correlators (N=3N = 3–66), extending Figure 10 and Figure 11.

Figure 16. The effect of the matching (non-singular) corrections on the differential ENCs at LO(+NLL) (left) and NLO(+NNLL) (right), for N=2N = 2–66 (top to bottom).

Figure 17. The effect of the matching (non-singular) corrections on the differential ENC/EEC ratios at LO(+NLL) (left) and NLO(+NNLL) (right), for N=3N = 3–66 (top to bottom).

References

  1. [1]H. A. Chawdhry, M. Czakon, A. Mitov, and R. Poncelet, NNLO QCD corrections to three-photon production at the LHC, JHEP 02 (2020) 057, [arXiv:1911.00479].DOI
  2. [2]M. Czакон and R. Poncelet, How much color do we really need? Two-loop subleading-color effects in photon and jet physics, arXiv:2512.17591.arxiv.org/abs/2512.17591
  3. [3]S. Badger, M. Czакон, H. B. Hartanto, R. Moodie, T. Peraro, R. Poncelet, and S. Zoia, Isolated photon production in association with a jet pair through next-to-next-to-leading order in QCD, JHEP 10 (2023) 071, [arXiv:2304.06682].DOI
  4. [4]M. Czакон, A. Mitov, and R. Poncelet, Next-to-Next-to-Leading Order Study of Three-Jet Production at the LHC, Phys. Rev. Lett. 127 (2021), no. 15 152001, [arXiv:2106.05331].DOI
  5. [5]M. Alvarez, J. Cantero, M. Czакон, J. Llorente, A. Mitov, and R. Poncelet, NNLO QCD corrections to event shapes at the LHC, JHEP 03 (2023) 129, arXiv:2301.01086.DOI
  6. [6]F. Buccioni, X. Chen, W.-J. Feng, T. Gehrmann, A. Huss, and M. Marcoli, Precise Predictions for Event Shapes in Diphoton Production at the LHC, Phys. Rev. Lett. 134 (2025), no. 17 171901, [arXiv:2501.14021].arxiv.org/abs/2501.14021
  7. [7]H. B. Hartanto, R. Poncelet, A. Popescu, and S. Zoia, Next-to-next-to-leading order QCD corrections to Wb̄ production at the LHC, Phys. Rev. D 106 (2022), no. 7 074016, [arXiv:2205.01687].
  8. [8]L. Buonocore, S. Devoto, S. Kallweit, J. Mazzitelli, L. Rottoli, and C. Savoini, Associated production of a W boson and massive bottom quarks at next-to-next-to-leading order in QCD, Phys. Rev. D 107 (2023), no. 7 074032, [arXiv:2212.04954].DOI
  9. [9]J. Mazzitelli, V. Sotnikov, and M. Wiesemann, Next-to-next-to-leading order event generation for Z-boson production in association with a bottom-quark pair, Phys. Rev. D 112 (2025), no. 5 056031, [arXiv:2404.08598].arxiv.org/abs/2404.08598
  10. [10]C. Biello, J. Mazzitelli, A. Sankar, M. Wiesemann, and G. Zanderighi, Higgs boson production in association with massive bottom quarks at NNLO+PS, JHEP 04 (2025) 088, [arXiv:2412.09510].arxiv.org/abs/2412.09510
  11. [11]S. Badger, C. Brancaccio, M. Becchetti, M. Czакон, H. B. Hartanto, R. Poncelet, and S. Zoia, Higher-order QCD corrections to top-quark pair production in association with a jet, arXiv:2511.11431.arxiv.org/abs/2511.11431
  12. [12]A. Gehrmann-De Ridder, T. Gehrmann, E. W. N. Glover, and G. Heinrich, Infrared structure of e⁺e⁻ → 3 jets at NNLO, JHEP 11 (2007) 058, [arXiv:0710.0346].
  13. [13]A. Gehrmann-De Ridder, T. Gehrmann, E. W. N. Glover, and G. Heinrich, NNLO corrections to event shapes in e⁺e⁻ annihilation, JHEP 12 (2007) 094, [arXiv:0711.4711].
  14. [14]A. Gehrmann-De Ridder, T. Gehrmann, E. W. N. Glover, and G. Heinrich, Second-order QCD corrections to the thrust distribution, Phys. Rev. Lett. 99 (2007) 132002, [arXiv:0707.1285].
  15. [15]S. Weinzierl, NNLO corrections to 3-jet observables in electron-positron annihilation, Phys. Rev. Lett. 101 (2008) 162001, [arXiv:0807.3241].DOI
  16. [16]A. J. Larkoski, I. Moult, and B. Nachman, Jet Substructure at the Large Hadron Collider: A Review of Recent Advances in Theory and Machine Learning, arXiv:1709.04464.DOI
  17. [17]R. Kogler et al., Jet Substructure at the Large Hadron Collider: Experimental Review, Rev. Mod. Phys. 91 (2019), no. 4 045003, [arXiv:1803.06991].DOI
  18. [18]C. Frye, A. J. Larkoski, M. D. Schwartz, and K. Yan, Factorization for groomed jet substructure beyond the next-to-leading logarithm, JHEP 07 (2016) 064, [arXiv:1603.09338].
  19. [19]C. Frye, A. J. Larkoski, M. D. Schwartz, and K. Yan, Precision physics with pile-up insensitive observables, arXiv:1603.06375.
  20. [20]A. H. Hoang, S. Mantry, A. Pathak, and I. W. Stewart, Nonperturbative Corrections to Soft Drop Jet Mass, arXiv:1906.11843.DOI
  21. [21]A. Ferdinand, K. Lee, and A. Pathak, Field-theoretic analysis of hadronization using soft drop jet mass, Phys. Rev. D 108 (2023), no. 11 L111501, [arXiv:2301.03605].arxiv.org/abs/2301.03605
  22. [22]C. Basham, L. Brown, S. Ellis, and S. Love, Energy Correlations in electron-Positron Annihilation in Quantum Chromodynamics: Asymptotically Free Perturbation Theory, Phys. Rev. D 19 (1979) 2018.DOI
  23. [23]C. L. Basham, L. S. Brown, S. D. Ellis, and S. T. Love, Energy Correlations in Perturbative Quantum Chromodynamics: A Conjecture for All Orders, Phys. Lett. B 85 (1979) 297–299.DOI
  24. [24]C. L. Basham, L. S. Brown, S. D. Ellis, and S. T. Love, Electron - Positron Annihilation Energy Pattern in Quantum Chromodynamics: Asymptotically Free Perturbation Theory, Phys. Rev. D 17 (1978) 2298.DOI
  25. [25]C. L. Basham, L. S. Brown, S. D. Ellis, and S. T. Love, Energy Correlations in electron - Positron Annihilation: Testing QCD, Phys. Rev. Lett. 41 (1978) 1585.DOI
  26. [26]H. Chen, I. Moult, X. Zhang, and H. X. Zhu, Rethinking jets with energy correlators: Tracks, resummation, and analytic continuation, Phys. Rev. D 102 (2020), no. 5 054012, [arXiv:2004.11381].DOI
  27. [27]K. Lee, B. Meçaj, and I. Moult, Conformal Colliders Meet the LHC, arXiv:2205.03414.DOI
  28. [28]I. Moult and H. X. Zhu, Energy Correlators: A Journey From Theory to Experiment, arXiv:2506.09119.arxiv.org/abs/2506.09119
  29. [29]P. T. Komiske, I. Moult, J. Thaler, and H. X. Zhu, Analyzing N-Point Energy Correlators inside Jets with CMS Open Data, Phys. Rev. Lett. 130 (2023), no. 5 051901, [arXiv:2201.07800].DOI
  30. [30]K. Lee, I. Moult, and W. J. Waalewijn, Putting Jet Substructure on Track(s), arXiv:2607.00087.arxiv.org/abs/2607.00087
  31. [31]L. J. Dixon, I. Moult, and H. X. Zhu, Collinear limit of the energy-energy correlator, Phys. Rev. D 100 (2019), no. 1 014009, [arXiv:1905.01310].DOI
  32. [32]CMS Collaboration, A. Havrapetvan et al., Measurement of Energy Correlators inside Jets and Determination of the Strong Coupling αS(mZ), Phys. Rev. Lett. 133 (2024), no. 7 071903, [arXiv:2402.13864].DOI
  33. [33]CMS Collaboration, Energy-energy correlation in Z boson tagged PbPb and pp collisions at √sNN = 5.02 TeV, tech. rep., 2025.
  34. [34]ATLAS Collaboration, G. Aad et al., Determination of the strong coupling constant from transverse energy–energy correlations in multijet events at √s = 13 TeV with the ATLAS detector, JHEP 07 (2023) 085, [arXiv:2301.09351].DOI
  35. [35]G. De Laurentis, H. Ita, B. Page, and V. Sotnikov, Compact two-loop QCD corrections for Vjj production in proton collisions, JHEP 06 (2025) 093, [arXiv:2503.10595].arxiv.org/abs/2503.10595
  36. [36]K. Lee, B. Meçaj, and I. Moult, Conformal collider physics meets LHC data, Phys. Rev. D 111 (2025), no. 1 L011502, [arXiv:2205.03414].
  37. [37]K. Lee, I. Moult, and X. Zhang, Revisiting single inclusive jet production: timelike factorization and reciprocity, JHEP 05 (2025) 129, [arXiv:2409.19045].arxiv.org/abs/2409.19045
  38. [38]Electron-Positron Alliance Collaboration, H. Bossi et al., Energy Correlators from Partons to Hadrons: Unveiling the Dynamics of the Strong Interactions with Archival ALEPH Data, arXiv:2511.00149.
  39. [39]K. Lee, Y. Li, Z. Xu, and X. Zhang, Projected Energy Correlators: Two-Loop Jet Functions and NNLL Resummation, arXiv:2606.02714.arxiv.org/abs/2606.02714
  40. [40]M. Czakon, A novel subtraction scheme for double-real radiation at NNLO, Phys. Lett. B 693 (2010) 259–268, [arXiv:1005.0274].DOI
  41. [41]M. Czakon and D. Heymes, Four-dimensional formulation of the sector-improved residue subtraction scheme, Nucl. Phys. B 890 (2014) 152–227, [arXiv:1408.2500].DOI
  42. [42]M. Czakon, A. van Hameren, A. Mitov, and R. Poncet, Single-jet inclusive rates with exact color at O (αₛ⁴), JHEP 10 (2019) 262, [arXiv:1907.12911].DOI
  43. [43]M. Czakon, T. Generet, A. Mitov, and R. Poncet, B-hadron production in NNLO QCD: application to LHC t̄t events with leptonic decays, JHEP 10 (2021) 216, [arXiv:2102.08267].DOI
  44. [44]M. Czakon, T. Generet, A. Mitov, and R. Poncet, NNLO B-fragmentation fits and their application to t̄t production and decay at the LHC, JHEP 03 (2023) 251, [arXiv:2210.06078].DOI
  45. [45]M. Czakon, T. Generet, A. Mitov, and R. Poncet, Open B-Hadron Production at Hadron Colliders in QCD at Next-to-Next-to-Leading-Order and Next-to-Next-to-Leading-Logarithmic Accuracy, Phys. Rev. Lett. 135 (2025), no. 16 161903, [arXiv:2411.09684].DOI
  46. [46]M. Czakon, T. Generet, A. Mitov, and R. Poncet, Identified Hadron Production at Hadron Colliders in Next-to-Next-to-Leading-Order QCD, Phys. Rev. Lett. 135 (2025), no. 17 17, [arXiv:2503.11489].DOI
  47. [47]T. Generet, K. Lee, I. Moult, R. Ponclet, and X. Zhang, Small radius inclusive jet production at the LHC through NNLO+NNLL, JHEP 08 (2025) 015, [arXiv:2503.21866].arxiv.org/abs/2503.21866
  48. [48]K. Lee, A. Pathak, I. W. Stewart, and Z. Sun, Nonperturbative Effects in Energy Correlators: From Characterizing Confinement Transition to Improving αₛ Extraction, Phys. Rev. Lett. 133 (2024), no. 23 231902, [arXiv:2405.19396].DOI
  49. [49]W. Chen, J. Gao, Y. Li, Z. Xu, X. Zhang, and H. X. Zhu, NNLL resummation for projected three-point energy correlator, JHEP 05 (2024) 043, [arXiv:2307.07510].arxiv.org/abs/2307.07510
  50. [50]N. Sveshnikov and F. Tkachov, Jets and quantum field theory, Phys. Lett. B 382 (1996) 403–408, [hep-ph/9512370].
  51. [51]F. V. Tkachov, Measuring multi - jet structure of hadronic energy flow or What is a jet?, Int. J. Mod. Phys. A 12 (1997) 5411–5529, [hep-ph/9601308].DOI
  52. [52]D. M. Hofman and J. Maldacena, Conformal collider physics: Energy and charge correlations, JHEP 05 (2008) 012, [arXiv:0803.1467].DOI
  53. [53]M. Cacciari, G. P. Salam, and G. Soyez, The anti-kₜ jet clustering algorithm, JHEP 04 (2008) 063, [arXiv:0802.1189].DOI
  54. [54]ALICE Collaboration, S. Acharya et al., Exposing the parton-hadron transition within jets with energy-energy correlators in pp collisions at √s = 5.02 TeV, arXiv:2409.12687.DOI
  55. [55]STAR Collaboration, B. E. Aboona et al., Measurement of Two-Point Energy Correlators within Jets in p+p Collisions at √s = 200 GeV, Phys. Rev. Lett. 135 (2025), no. 11 111901, [arXiv:2502.15925].arxiv.org/abs/2502.15925
  56. [56]CMS Collaboration, V. Chekhovsky et al., Observation of nuclear modification of energy-energy correlators inside jets in heavy ion collisions, Phys. Lett. B 866 (2025) 139556, [arXiv:2503.19993].DOI
  57. [57]ALICE Collaboration, D. Ali Hassan Abdallah et al., Energy-energy correlators in p-Pb collisions at √sNN = 5.02 TeV, arXiv:2606.18143.arxiv.org/abs/2606.18143
  58. [58]Z.-B. Kang, F. Ringer, and I. Vitev, The semi-inclusive jet function in SCET and small radius resummation for inclusive jet production, JHEP 10 (2016) 125, [arXiv:1606.06732].DOI
  59. [59]F. Siegert, A practical guide to event generation for prompt photon production with Sherpa, J. Phys. G 44 (2017), no. 4 044007, [arXiv:1611.07226].DOI
  60. [60]X. Chen, T. Gehrmann, N. Glover, M. Höfer, and A. Huss, Isolated photon and photon+jet production at NNLO QCD accuracy, JHEP 04 (2020) 166, [arXiv:1904.01044].DOI
  61. [61]T. Gehrmann, P. Jakubčík, C. C. Mella, N. Syrrakos, and L. Tancredi, Planar three-loop QCD helicity amplitudes for V+jet production at hadron colliders, Phys. Lett. B 848 (2024) 138369, [arXiv:2307.15405].arxiv.org/abs/2307.15405
  62. [62]X. Chen, P. Jakubčík, M. Marcoli, and G. Stagnitto, Jet production at electron-positron colliders at next-to-next-to-next-to-leading order in QCD, Phys. Lett. B 869 (2025) 139804, [arXiv:2505.10618].DOI
  63. [63]A. van Hameren, “avhlib.” https://bitbucket.org/hameren/avhlib.
  64. [64]M. Bury and A. van Hameren, Numerical evaluation of multi-gluon amplitudes for High Energy Factorization, Comput. Phys. Commun. 196 (2015) 592–598, [arXiv:1503.08612].arxiv.org/abs/1503.08612
  65. [65]F. Cascioli, P. Maierhofer, and S. Pozzorini, Scattering Amplitudes with Open Loops, Phys. Rev. Lett. 108 (2012) 111601, [arXiv:1111.5206].DOI
  66. [66]F. Buccioni, J.-N. Lang, J. M. Lindert, P. Maierhöfer, S. Pozzorini, H. Zhang, and M. F. Zoller, OpenLoops 2, Eur. Phys. J. C 79 (2019), no. 10 866, [arXiv:1907.13071].DOI
  67. [67]E. W. N. Glover and M. E. Tejeda-Yeomans, Two loop QCD helicity amplitudes for massless quark massless gauge boson scattering, JHEP 06 (2003) 033, [hep-ph/0304169].DOI
  68. [68]PDF4LHC Working Group Collaboration, R. D. Ball et al., The PDF4LHC21 combination of global PDF fits for the LHC Run III, J. Phys. G 49 (2022), no. 8 080501, [arXiv:2203.05506].
  69. [69]C. W. Bauer, S. Fleming, D. Pirjol, and I. W. Stewart, An Effective field theory for collinear and soft gluons: Heavy to light decays, Phys. Rev. D 63 (2001) 114020, [hep-ph/0011336].DOI
  70. [70]C. W. Bauer, S. Fleming, and M. E. Luke, Summing Sudakov logarithms in B → Xsγ in effective field theory, Phys. Rev. D 63 (2000) 014006, [hep-ph/0005275].DOI
  71. [71]C. W. Bauer, D. Pirjol, and I. W. Stewart, Soft collinear factorization in effective field theory, Phys. Rev. D 65 (2002) 054022, [hep-ph/0109045].DOI
  72. [72]C. W. Bauer and I. W. Stewart, Invariant operators in collinear effective theory, Phys. Lett. B516 (2001) 134–142, [hep-ph/0107001].DOI
  73. [73]M. Beneke, A. P. Chapovsky, M. Diehl, and T. Feldmann, Soft collinear effective theory and heavy to light currents beyond leading power, Nucl. Phys. B 643 (2002) 431–476, [hep-ph/0206152].DOI
  74. [74]C. Marcantonini and I. W. Stewart, Reparametrization Invariant Collinear Operators, Phys. Rev. D 79 (2009) 065028, [arXiv:0809.1093].DOI
  75. [75]A. A. Almasy, S. Moch, and A. Vogt, On the Next-to-Next-to-Leading Order Evolution of Flavour-Singlet Fragmentation Functions, Nucl. Phys. B 854 (2012) 133–152, [arXiv:1107.2263].DOI
  76. [76]H. Chen, T.-Z. Yang, H. X. Zhu, and Y. J. Zhu, Analytic Continuation and Reciprocity Relation for Collinear Splitting in QCD, Chin. Phys. C 45 (2021), no. 4 043101, [arXiv:2006.10534].DOI
  77. [77]T. Gehrmann, A. von Manteuffel, V. Sotnikov, and T.-Z. Yang, Complete N² contributions to four-loop pure-singlet splitting functions, JHEP 01 (2024) 029, [arXiv:2308.07958].DOI
  78. [78]G. Falcioni, F. Herzog, S. Moch, and A. Vogt, Four-loop splitting functions in QCD – The quark-quark case, Phys. Lett. B 842 (2023) 137944, [arXiv:2302.07593].DOI
  79. [79]G. Falcioni, F. Herzog, S. Moch, and A. Vogt, Four-loop splitting functions in QCD – The gluon-to-quark case, Phys. Lett. B 846 (2023) 138215, [arXiv:2307.04158].DOI
  80. [80]G. Falcioni, F. Herzog, S. Moch, A. Pelloni, and A. Vogt, Four-loop splitting functions in QCD – The quark-to-gluon case, Phys. Lett. B 856 (2024) 138906, [arXiv:2404.09701].arxiv.org/abs/2404.09701
  81. [81]G. Falcioni, F. Herzog, S. Moch, A. Pelloni, and A. Vogt, Four-loop splitting functions in QCD – the gluon-gluon case –, Phys. Lett. B 860 (2025) 139194, [arXiv:2410.08089].arxiv.org/abs/2410.08089
  82. [82]G. Falcioni, F. Herzog, S. Moch, A. Pelloni, and A. Vogt, Additional results on the four-loop flavour-singlet splitting functions in QCD, Phys. Lett. B 875 (2026) 140278, [arXiv:2512.10783].arxiv.org/abs/2512.10783
  83. [83]A. Mitov, S. Moch, and A. Vogt, Next-to-Next-to-Leading Order Evolution of Non-Singlet Fragmentation Functions, Phys. Lett. B638 (2006) 61–67, [hep-ph/0604053].DOI
  84. [84]C.-Q. He, H. Xing, T.-Z. Yang, and H. X. Zhu, Single-Inclusive Hadron Production in Electron-Positron Annihilation at Next-to-Next-to-Next-to-Leading Order in QCD, Phys. Rev. Lett. 135 (2025), no. 10 101901, [arXiv:2503.20441].DOI
  85. [85]G. P. Korchemsky and G. F. Sterman, Power corrections to event shapes and factorization, Nucl. Phys. B 555 (1999) 335–351, [hep-ph/9902341].
  86. [86]H. Chen, P. F. Monni, Z. Xu, and H. X. Zhu, Scaling Violation in Power Corrections to Energy Correlators from the Light-Ray Operator Product Expansion, Phys. Rev. Lett. 133 (2024), no. 23 231901, [arXiv:2406.06668].DOI
  87. [87]A. H. Hoang, D. W. Kolodrubetz, V. Mateu, and I. W. Stewart, C-parameter distribution at N³LL′ including power corrections, Phys. Rev. D 91 (2015), no. 9 094017, [arXiv:1411.6633].
  88. [88]M. A. Benitez, A. H. Hoang, V. Mateu, I. W. Stewart, and G. Vita, On Determining αₛ(mZ) from Dijets in e⁺e⁻ Thrust, JHEP 07 (2025) 249, [arXiv:2412.15164].
  89. [89]M. A. Benitez, A. Bhattacharya, A. H. Hoang, V. Mateu, M. D. Schwartz, I. W. Stewart, and X. Zhang, A Precise Determination of αₛ from the Heavy Jet Mass Distribution, arXiv:2502.12253.DOI
  90. [90]F. J. Tackmann, Beyond scale variations: perturbative theory uncertainties from nuisance parameters, JHEP 08 (2025) 098, [arXiv:2411.18606].arxiv.org/abs/2411.18606
  91. [91]T. Cridge, G. Marinelli, and F. J. Tackmann, Theory Uncertainties in the Extraction of αₛ from Drell-Yan at Small Transverse Momentum, arXiv:2506.13874.arxiv.org/abs/2506.13874
  92. [92]R. Abbate, M. Ficking, A. H. Hoang, V. Mateu, and I. W. Stewart, Thrust at N³LL with Power Corrections and a Precision Global Fit for α_s(m_Z), Phys. Rev. D83 (2011) 074021, [arXiv:1006.3080].
  93. [93]M. Jaarsma, Y. Li, I. Moult, W. J. Waalewijn, and H. X. Zhu, From DGLAP to Sudakov: Precision Predictions for Energy-Energy Correlators, arXiv:2512.11950.arxiv.org/abs/2512.11950

Paper details

vlnc:W_xnnq7ygg
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 8

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