betaFeedback

WorksPhysics and Astronomy

T-odd Parton Distribution Functions and Azimuthal Anisotropy at High Transverse Momentum in pp-pp and pp-AA Collisions

Abstract

Various azimuthal anisotropies (v1,v2,v3,v4v_1, v_2, v_3, v_4), at high transverse momentum (high-pTp_T), are shown to arise from the asymmetric scattering of transverse polarized quarks and gluons, arising from unpolarized nucleons (the Boer-Mulders' effect) and resulting in unpolarized hadrons (the Collins effect). Combined with the asymmetric scattering of partons from polarization independent but transverse momentum dependent (TMD) distributions, we obtain a possible mechanism to understand the azimuthal anisotropy of hadrons at large transverse momentum observed in pp-pp collisions. Constraining the ratio of polarization dependent TMD distributions to polarization independent distributions by comparing with the data from pp-pp collisions, we find that scaling the acquired transverse momentum of the initial state partons from the proton, due to prescattering before the hard interaction, with the length of the nucleus (k⊥2∝A1/3k_\perp^2 \propto A^{1/3}), straightforwardly yields the azimuthal anisotropy at high pTp_T in pp-AA collisions.

Introduction

One of the foremost signals of the formation of a quark-gluon plasma (QGP) during a heavy ion collision is the elliptic flow of the bulk medium [1, 2, 3, 4, 5], in semi-central collisions.1 Due to the geometry of the collision, the overlap region of the two nuclei is not circular, and contains azimuthal anisotropies in density. Pressure gradients created by these spatial inhomogeneities lead to azimuthal anisotropies in the momentum distribution [6], quantified by the Fourier coefficients of the differential hadronic yield,

dNdϕ=dNdϕ(1+∑n2vncos⁡[n(ϕ−Ψn)]).(1)\frac{dN}{d\phi} = \frac{dN}{d\phi}\left(1 + \sum_n 2v_n \cos[n(\phi-\Psi_n)]\right). \tag*{(1)}

In the above equation, ϕ\phi is the azimuthal angle and Ψn\Psi_n is the nthn^{\mathrm{th}} order event plane angle, and dNdN is the differential yield of hadrons, in some range of transverse momentum and rapidity.

Due to the azimuthally anisotropic spatial density distribution in semi-central heavy-ion collisions, hard partons produced in rare collisions, tend to experience azimuthally asymmetric energy loss [7, 8, 9]. This leads to an azimuthal anisotropy in both high transverse momentum (high pTp_T) hadron [10] and jet distributions [11, 12]. This high pTp_T anisotropy, goes hand-in-hand with the angle-integrated, centrality-dependent suppression of the high pTp_T jet and hadron spectrum in heavy-ion collisions (with

greater suppression appearing in the more central collisions) [13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23].

Simulations of jet energy loss in a fluid medium [24, 25, 26] compute and obtain both the angle integrated and angle dependent suppression by changing the underlying space-time fluid profile used to compute the local jet transport coefficients [27, 28, 29, 30]. The underlying fluid profile is obtained by rigorous comparison of bulk evolution simulations with experimental data [31, 32, 33, 34].

Recently, measurements of pp-pp and pp-PbPb collision events (referred to as small systems), with high multiplicity (N≳90N \gtrsim90), yield a variety of azimuthal anisotropies [35, 36, 37] in the soft bulk sector (pT≲3 GeVp_T \lesssim3\ \mathrm{GeV}). This indicates the possible formation of a QGP in these small systems. There have been several theoretical efforts to understand the origin of the low-pTp_T azimuthal anisotropies in small systems [38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48]. The corroborating evidence from high pTp_T probes is, however, inconclusive. Experimental data on hadrons at large pTp_T show a significant azimuthal anisotropy [49, 50], in both pp-pp (measured up to 10 GeV), and pp-PbPb (measured up to 50 GeV) collisions. However, there is no evidence of suppression in the angle integrated spectrum in pp-PbPb collisions (at high-pT≳5 GeVp_T \gtrsim5\ \mathrm{GeV}) [51, 52, 53].

The origin of an azimuthal anisotropy at high-pTp_T, in the absence of any suppression of the angle integrated spectrum in pp-AA, can be found in the presence of the same anisotropy in pp-pp collisions themselves. In the current effort, this anisotropy will arise due to the existence of a small transverse momentum in the initial state. Partons in the proton carry a small transverse momentum k⊥k_\perp, in addition to the large longitudinal momentum, which is a fraction (xx) of the proton’s momentum, i.e. p+=xP+p^+ = xP^+.

This small transverse momentum will lead to azimuthal anisotropies in both the polarization dependent and polarization averaged cross sections. In this Paper, we study how transverse momentum dependent parton distribution functions (TMDPDFs) and fragmentation functions (TMDFFs) [54], can generate azimuthal anisotropy at high pTpT in pp-pp and pp-Pb collisions. In particular, we find that the combination of the Boer-Mulders’ process of polarized partons from unpolarized hadrons [55] and the Collins’ process [56] of polarized parton fragmentation into unpolarized hadrons lead to a large azimuthal anisotropy. The small transverse momentum k⊥k_{\perp} in pp-pp will be enhanced by a factor of A1/3A^{1/3} (with AA the mass number of the nucleus) in pp-A collisions, to account for the possibility of soft pre-scattering of the hard parton from the proton within the nucleus, prior to the hard scattering that leads to jet formation.

Due to the small system size, momentum conservation in the initial state becomes more apparent in the particle distributions from the final state in pp-pp and pp-A: The transverse momentum k⊥k_{\perp} of the hard parton, that will participate in hard scattering is balanced by the remainder of the proton. As a result, the net transverse momentum of the produced dijet system is partially balanced by the net transverse momentum of the bulk system. Simulations using the PYTHIA generator [57] will be used to show that an intrinsic transverse momentum will yield both greater particle production over a range of rapidities, as well as the correlation between the dijet pair momentum (qT=pjet 1+pjet 2q_T = p_{\mathrm{jet}\ 1} + p_{\mathrm{jet}\ 2}) and the event plane angle(s). Momentum conservation will correlate the cos⁡(nϕ)\cos(n\phi) asymmetries of the hard sector with the event plane of the bulk system, leading to corresponding vnv_n’s in the hard sector. PYTHIA simulations, currently do not include polarization dependent matrix elements. Thus, our final results will be semi-analytic, assuming that the dijet pair angle is completely correlated with the event plane angle(s).

The paper is organized as follows: In Sec. II, we present the theoretical framework of polarization dependent and polarization averaged scattering with TMDPDFs and TMDFFs. We also present the matrix elements for the processes that will be used. In Sec. III, we present a phenomenological description of TMDPDF and TMDFFs that will be used to numerically evaluate these quantities. In Sec. IV, we present the overarching phenomenological picture of how an elliptic anisotropy can originate in a small system, in the presence of hard scattering. We also carry out a modified PYTHIA simulation to demonstrate this picture and correlate the event plane angle(s) with the direction of the net momentum of the dijets (qT=pjet 1+pjet 2q_T = p_{\mathrm{jet}\ 1} + p_{\mathrm{jet}\ 2}). In § 0.5, we present our results for the cross section and the azimuthal anisotropy, and compare with data from the ALICE and ATLAS collaborations. A summary and outlook is presented in § 0.6.

Theoretical Framework

In the standard collinear factorization of hard processes [58], the outgoing hadron distributions are azimuthally symmetric. The two incoming partons that participate in the hard scattering, approach head-to-head, perfectly collinearly, and leave the hard scattering back-to-back. There is no preferred direction in the azimuthal plane of the initial state. Instead, if one were to include the more realistic transverse momentum dependent (TMD-) PDFs [59, 60, 61, 62], and FFs [63, 64, 65, 66], the two incoming partons (a,ba,b) are no longer perfectly collinear. Each has a transverse momentum (k⊥a,k⊥bk_{\perp}^{a},k_{\perp}^{b}). As a result, the direction k⊥a+k⊥bk_{\perp}^{a}+k_{\perp}^{b} breaks the azimuthal symmetry of the process, resulting in momentum correlations in the transverse plane of the hadronic scattering.

Following [64, 65], we assume that at high energies the momentum hierarchy between the longitudinal momenta and the transverse momenta allows the pion production cross section to factorize into a soft parton correlator and a hard scattering matrix element

dσdyd2PT=∑a,b,c,d∫dxadxbdz2π2z3d2k⊥ad2k⊥bd3k⊥c δ(k⊥c⋅p^c)J(k⊥c)Γσμ(xa,k⊥a)Γαν(xb,k⊥b)M^μνρ(M^σαβ)∗Δβρ(z,k⊥c)δ(s^+t^+u^).(2)\frac{d\sigma}{dy d^{2}P_{T}} = \sum_{a,b,c,d}\int\frac{dx_{a}dx_{b}dz}{2\pi^{2}z^{3}}d^{2}k_{\perp a}d^{2}k_{\perp b}d^{3}k_{\perp c}\,\delta(k_{\perp c}\cdot\hat{p}_{c})J(k_{\perp c})\Gamma^{\sigma\mu}(x_{a},k_{\perp a})\Gamma^{\alpha\nu}(x_{b},k_{\perp b})\hat{M}_{\mu\nu\rho}\left(\hat{M}_{\sigma\alpha\beta}\right)^{*}\Delta^{\beta\rho}(z,k_{\perp c})\delta(\hat{s}+\hat{t}+\hat{u}) . \tag*{(2)}

where J(k⊥c)=(Ec2+pc2−k⊥c2)24(pc2−k⊥c2)J(k_{\perp c})=\frac{\left(E_{c}^{2}+\sqrt{p_{c}^{2}-k_{\perp c}^{2}}\right)^{2}}{4\left(p_{c}^{2}-k_{\perp c}^{2}\right)} and the Mandelstam variables of the partonic system are given by (s^,t^,u^)(\hat{s},\hat{t},\hat{u}). In the equation above, M^μνρ(M^σαβ)∗\hat{M}_{\mu\nu\rho}\left(\hat{M}_{\sigma\alpha\beta}\right)^{*} is the hard partonic squared amplitude, Γ\Gamma are the initial state correlators and Δ\Delta are the final state fragmentation correlator.

The leading twist gluon generalized PDF reads [67]

ΓPμν(x,k⊥)=∫d(ξ⋅P)d2ξT(2π)3ei(xP⋅ξ)⟨P∣Tr⁡[Fμν(0)Fστ(ξ)]∣P⟩nρnσ(P⋅n)2=−gTμνfg(x,k⊥)+(k⊥μk⊥νMp2+gTμνk⊥22Mp2)hg⊥(x,k⊥2)2x,(3)\Gamma_{P}^{\mu\nu}(x,k_{\perp}) = \int\frac{d(\xi\cdot P)d^{2}\xi_{T}}{(2\pi)^{3}}e^{i(xP\cdot\xi)}\frac{\langle P|\operatorname{Tr}\left[F^{\mu\nu}(0)F^{\sigma\tau}(\xi)\right]|P\rangle n_{\rho}n_{\sigma}}{(P\cdot n)^{2}} = \frac{-g_{T}^{\mu\nu}f_{g}(x,k_{\perp})+\left(\frac{k_{\perp}^{\mu}k_{\perp}^{\nu}}{M_{p}^{2}}+g_{T}^{\mu\nu}\frac{k_{\perp}^{2}}{2M_{p}^{2}}\right)h_{g}^{\perp}(x,k_{\perp}^{2})}{2x}, \tag*{(3)}

where f(x,k⊥)f(x,k_{\perp}) is the unpolarized gluon distribution. The gluon Boer-Mulders’ function h⊥(x,k⊥2)h^{\perp}(x,k_{\perp}^{2}) can be interpreted as the distribution of linearly polarized gluons in the nucleon. Using the following gluon polarization vectors,

ϵλ1μ=(0,−λ1e−iϕ,−iλ1eiϕ,0),(4)\epsilon_{\lambda_{1}}^{\mu}=(0,-\lambda_{1}e^{-i\phi},-i\lambda_{1}e^{i\phi},0), \tag*{(4)}
ϵλ2μ∗=(0,−λ2e−iϕ,−iλ2eiϕ,0),(5)\epsilon_{\lambda_{2}}^{\mu*}=(0,-\lambda_{2}e^{-i\phi},-i\lambda_{2}e^{i\phi},0), \tag*{(5)}

we can project the gluon generalized PDF onto gluon helicity states

ΓPλ1,λ2(x,k⊥)=−δλ1,λ2fg(x,k⊥)+k⊥22Mp2δλ1,−λ2hg⊥(x,k⊥2)2x.(6)\Gamma_{P}^{\lambda_{1},\lambda_{2}}(x,k_{\perp}) = \frac{-\delta^{\lambda_{1},\lambda_{2}}f_{g}(x,k_{\perp})+\frac{k_{\perp}^{2}}{2M_{p}^{2}}\delta^{\lambda_{1},-\lambda_{2}}h_{g}^{\perp}(x,k_{\perp}^{2})}{2x}. \tag*{(6)}

The quark generalized PDF is given by

ΓPμν(x,k⊥)=∫d(ξ⋅P)d2ξT(2π)3ei(p⋅ξ)⟨P∣ψˉ(0)ψ(ξ)∣P⟩nμnν(p⋅n)2=12[p̸fq(x,k⊥)+hq⊥(x,k⊥)σμνk⊥μPν2Mp],(7)\Gamma_P^{\mu\nu}(x,k_\perp)=\int\frac{d(\xi\cdot P)d^2\xi_T}{(2\pi)^3}e^{i(p\cdot\xi)}\frac{\langle P|\bar{\psi}(0)\psi(\xi)|P\rangle n^\mu n^\nu}{(p\cdot n)^2}=\frac{1}{2}\left[\not{p}f_q(x,k_\perp)+h_q^\perp(x,k_\perp)\frac{\sigma^{\mu\nu}k_\perp^\mu P^\nu}{2M_p}\right], \tag*{(7)}

where σμν=[γμ,γν]\sigma_{\mu\nu}=[\gamma_\mu,\gamma_\nu]. Similar to the gluon, the distribution of unpolarized quarks is given by fq(x,k⊥)f_q(x,k_\perp). The time-reversal odd function hq⊥(x,k⊥)h_q^\perp(x,k_\perp) is the quark Boer-Mulders’ function, which can be interpreted as the distribution of transversely polarized quarks in an unpolarized proton.

Analogously, the fragmentation correlators are written

Δgλ1,λ2(z,k⊥)=−δλ1,λ2Dg(z,k⊥2)+δλ1,−λ2k⊥22Mπ2Hg⊥(z,k⊥2)2/z,(8)\Delta_g^{\lambda_1,\lambda_2}(z,k_\perp)=\frac{-\delta^{\lambda_1,\lambda_2}D_g(z,k_\perp^2)+\delta^{\lambda_1,-\lambda_2}\frac{k_\perp^2}{2M_\pi^2}H_g^\perp(z,k_\perp^2)}{2/z}, \tag*{(8)}
Δq(z,k⊥)=p̸Dq(z,k⊥2)+σμνk⊥μPν2MpHq⊥(z,k⊥2)2/z.(9)\Delta_q(z,k_\perp)=\frac{\not{p}D_q(z,k_\perp^2)+\frac{\sigma^{\mu\nu}k_\perp^\mu P^\nu}{2M_p}H_q^\perp(z,k_\perp^2)}{2/z}. \tag*{(9)}

The TMD fragmentation of unpolarized gluons or quarks to unpolarized pions with momentum zpc+k⊥Czp_c+k_{\perp C} is given by Dg/q(z,k⊥2)D_{g/q}(z,k_\perp^2). The equivalent to the BM function is the Collins’ function Hg/q⊥(z,k⊥2)H_{g/q}^\perp(z,k_\perp^2), which describes the fragmentation of a transversely polarized gluon or transversely polarized quarks to an unpolarized pion [56].

Unpolarized partons cross section

For the case of unpolarized partons, the cross section from Eq. (2) can be simplified further by setting the same helicities between the matrix element and its complex conjugate, leading to

dσdyd2PT=∑a,b,c,d∫dxadxbdz16π2xaxbz2sd2k⊥ad2k⊥bd2k⊥Cδ(k⊥C⋅p^c)J(k⊥C)fa/A(xa,k⊥a)fb/B(xb,k⊥b)∣M^abcd∣2DC/c(z,k⊥C)δ(s^+t^+u^),(10)\frac{d\sigma}{dy d^2P_T}=\sum_{a,b,c,d}\int\frac{dx_a dx_b dz}{16\pi^2x_ax_bz^2s}d^2k_{\perp a}d^2k_{\perp b}d^2k_{\perp C}\delta(k_{\perp C}\cdot\hat{p}_c)J(k_{\perp C})f_{a/A}(x_a,k_{\perp a})f_{b/B}(x_b,k_{\perp b})|\hat{M}_{ab}^{cd}|^2D_{C/c}(z,k_{\perp C})\delta(\hat{s}+\hat{t}+\hat{u}), \tag*{(10)}

where ∣M^abcd∣2|\hat{M}_{ab}^{cd}|^2 is now the usual unpolarized partonic squared matrix element that can be found in the literature [68].

Due to the transverse momenta of the partons, the partonic center of mass frame (q=ka+kbq=k_a+k_b) is off-center in the transverse plane of the hadronic scattering. This leads to transverse momentum correlations between the final state partons. Taking the limit of small transverse momentum k⊥a≪pak_{\perp a}\ll p_a, the incoming partonic momenta, with a small component in the xx-direction, can be written pa=pa(1,sin⁡θa,0,cos⁡θa)p_a=p_a(1,\sin\theta_a,0,\cos\theta_a), where θa\theta_a is the polar angle in the hadronic center of mass frame. Consider a typical partonic matrix element squared, where the outgoing parton cc is produced at mid-rapidity (θc=π/2\theta_c=\pi/2), along an azimuthal angle ϕc\phi_c,

M∝s^t^≃−s^2papc(1−θacos⁡(ϕc)).(11)\mathcal{M}\propto\frac{\hat{s}}{\hat{t}}\simeq\frac{-\hat{s}}{2p_ap_c(1-\theta_a\cos(\phi_c))}. \tag*{(11)}

By expanding the denominator in the small angle limit θa≪1\theta_a\ll1, we obtain

M≃−s^4papc[2+θa2+2θacos⁡(ϕc)+θa2cos⁡(2ϕc)].(12)\mathcal{M}\simeq\frac{-\hat{s}}{4p_ap_c}\left[2+\theta_a^2+2\theta_a\cos(\phi_c)+\theta_a^2\cos(2\phi_c)\right]. \tag*{(12)}

These azimuthal correlations generated by the intrinsic transverse momentum are suppressed by powers of the large transverse momentum, as θa=k⊥a/pa\theta_a=k_{\perp a}/p_a. Thus, while azimuthal anisotropies do arise in unpolarized cross sections, they are suppressed by powers of the hard scale, and have a diminishing contribution at high-pTp_T.

Polarized partonic cross section

Since polarized distributions change the helicity between the matrix elements and its complex conjugate (c.f. Eqs. (6)-(8)), inclusive unpolarized pion production requires two polarized distributions of the same parton species. Either both initial partons are polarized or one initial parton and one final parton are polarized, which we will refer to as BM ⊗\otimes BM or BM ⊗\otimes C effects, respectively. In the following four sections we will present the relevant contributions for gluon initiated processes and quark initiated processes.

Gluon initiated: Boer-Mulders' Collins' effect

We start with the BM ⊗\otimes C effect, where a polarized gluon, stemming from a Boer-Mulders’ function, scatters with an unpolarized gluon, producing a polarized gluon that fragments to an unpolarized pion, with a Collins’ function. Here, the matrix element times the PDFs and FF is written as,

ΣBM⊗C=Re⁡∑λ=±1HC/g⊥(1)(z,k⊥C)[hg/A⊥(1)(xa,k⊥a2)fg/B(xb,k⊥b2)M^λ,λλ,λ(M^−λ,−λλ,λ)∗+fg/A(xa,k⊥a2)hg/B⊥(1)(xb,k⊥b2)M^λ,λλ,−λ(M^−λ,−λλ,−λ)∗].(13)\Sigma^{\mathrm{BM}\otimes\mathrm{C}}=\operatorname{Re}\sum_{\lambda=\pm1}H_{C/g}^{\perp(1)}(z,k_{\perp C})\left[h_{g/A}^{\perp(1)}(x_a,k_{\perp a}^2)f_{g/B}(x_b,k_{\perp b}^2)\hat{M}_{\lambda,\lambda}^{\lambda,\lambda}(\hat{M}_{-\lambda,-\lambda}^{\lambda,\lambda})^*+f_{g/A}(x_a,k_{\perp a}^2)h_{g/B}^{\perp(1)}(x_b,k_{\perp b}^2)\hat{M}_{\lambda,\lambda}^{\lambda,-\lambda}(\hat{M}_{-\lambda,-\lambda}^{\lambda,-\lambda})^*\right]. \tag*{(13)}

We list the Matrix element times the PDFs and FF for gluon-gluon

Tgg→ggBM⊗C=gs4N2N2−1t^2+t^u^+u^2t^2e−4iλ(ϕ12−ϕ23),(14)\mathcal{T}_{gg\to gg}^{\mathrm{BM}\otimes\mathrm{C}}=g_s^4\frac{N^2}{N^2-1}\frac{\hat{t}^2+\hat{t}\hat{u}+\hat{u}^2}{\hat{t}^2}e^{-4i\lambda(\phi_{12}-\phi_{23})}, \tag*{(14)}

where using the momentum angle p=(p,θ,ϕ)\boldsymbol{p}=(p,\theta,\phi), we write

tan⁡ϕij=tan⁡ϕj−ϕi2sin⁡θj−θi2sin⁡θj+θi2.(15)\tan\phi_{ij}=\tan\frac{\phi_j-\phi_i}{2}\frac{\sin\frac{\theta_j-\theta_i}{2}}{\sin\frac{\theta_j+\theta_i}{2}}. \tag*{(15)}

After performing the helicity sums, the contribution to the cross section becomes

Σgg→ggBM⊗C=2D^C/g(z,k⊥C)[h^g/A⊥(1)(xa,k⊥a2)fg/B(xb,k⊥b2)M^10M^20cos⁡(4(ϕ12−ϕ23))+fg/A(xa,k⊥a2)h^g/B⊥(1)(xb,k⊥b2)M^10M^30cos⁡(4(ϕ12−ϕ13))].(16)\Sigma^{\mathrm{BM}\otimes\mathrm{C}}_{gg\to gg}=2\widehat{D}^{\mathrm{C}/g}(z,k_{\perp C})\left[\widehat{h}^{\perp(1)}_{g/A}(x_a,k_{\perp a}^{2})f_{g/B}(x_b,k_{\perp b}^{2})\widehat{M}^{0}_{1}\widehat{M}^{0}_{2}\cos\left(4(\phi_{12}-\phi_{23})\right)+f_{g/A}(x_a,k_{\perp a}^{2})\widehat{h}^{\perp(1)}_{g/B}(x_b,k_{\perp b}^{2})\widehat{M}^{0}_{1}\widehat{M}^{0}_{3}\cos\left(4(\phi_{12}-\phi_{13})\right)\right]. \tag*{(16)}

We define matrix elements stripped of their angular correlations,

M^10M^20=92gs4[1+ut+u2t2].(17)\widehat{M}^{0}_{1}\widehat{M}^{0}_{2}=\frac{9}{2}g_s^{4}\left[1+\frac{u}{t}+\frac{u^{2}}{t^{2}}\right]. \tag*{(17)}
M^10M^30=92gs4[1+tu+t2u2].(18)\widehat{M}^{0}_{1}\widehat{M}^{0}_{3}=\frac{9}{2}g_s^{4}\left[1+\frac{t}{u}+\frac{t^{2}}{u^{2}}\right]. \tag*{(18)}
Gluon initiated: Double Boer-Mulders' effect

Another contribution to the cross section comes from the scattering of two polarized gluons leading to unpolarized gluons in the final state. The matrix element times the PDFs and FF is now written

Σgg→ggBM⊗BM=∑λ=±1h^g/A⊥(1)(xa,k⊥a2)h^g/B⊥(1)(xb,k⊥b2)M^λ,λλ,λ(M^−λ,−λλ,λ)∗DC/g(z,k⊥C),(19)\Sigma^{\mathrm{BM}\otimes\mathrm{BM}}_{gg\to gg}=\sum_{\lambda=\pm1}\widehat{h}^{\perp(1)}_{g/A}(x_a,k_{\perp a}^{2})\widehat{h}^{\perp(1)}_{g/B}(x_b,k_{\perp b}^{2})\widehat{M}^{\lambda,\lambda}_{\lambda,\lambda}\left(\widehat{M}^{\lambda,\lambda}_{-\lambda,-\lambda}\right)^{*}D_{\mathrm{C}/g}(z,k_{\perp C}), \tag*{(19)}

with the following matrix element

Tgg→ggBM⊗BM=gs4N2N2−1t2+tu+u2s2e−4λi(ϕ23−ϕ13).(20)\mathcal{T}^{\mathrm{BM}\otimes\mathrm{BM}}_{gg\to gg}=g_s^{4}\frac{N^{2}}{N^{2}-1}\frac{t^{2}+tu+u^{2}}{s^{2}}e^{-4\lambda i(\phi_{23}-\phi_{13})}. \tag*{(20)}

Performing the helicity sums, yields

Σgg→ggBM⊗BM=4h^g/A⊥(1)(xa,k⊥a2)h^g/B⊥(1)(xb,k⊥b2)DC/g(z,k⊥C)M^20M^30cos⁡(4(ϕ13−ϕ23)),(21)\Sigma^{\mathrm{BM}\otimes\mathrm{BM}}_{gg\to gg}=4\widehat{h}^{\perp(1)}_{g/A}(x_a,k_{\perp a}^{2})\widehat{h}^{\perp(1)}_{g/B}(x_b,k_{\perp b}^{2})D_{\mathrm{C}/g}(z,k_{\perp C})\widehat{M}^{0}_{2}\widehat{M}^{0}_{3}\cos\left(4(\phi_{13}-\phi_{23})\right), \tag*{(21)}

with

M^20M^30=92gs4[t2+tu+u2s2].(22)\widehat{M}^{0}_{2}\widehat{M}^{0}_{3}=\frac{9}{2}g_s^{4}\left[\frac{t^{2}+tu+u^{2}}{s^{2}}\right]. \tag*{(22)}
Quark initiated: Boer-Mulders' Collins' effect

For the quark-quark scattering, the square matrix element of the BM⊗CBM\otimes C effect lead to odd number of Dirac matrices and only the interference of the tt- and uu-channel diagrams contribute to the cross section. The matrix element times the PDFs and FF is written

Σqq→qqBM⊗C=2HC/q⊥(1)(z,k⊥C)[M^10M^20hq/A⊥(1)(xa,k⊥a)fq/B(xb,k⊥b)+M^10M^30fq/A(xa,k⊥a)hq/B⊥(1)(xb,k⊥b)].(23)\Sigma^{\mathrm{BM}\otimes\mathrm{C}}_{qq\to qq}=2H^{\perp(1)}_{\mathrm{C}/q}(z,k_{\perp C})\left[\widehat{M}^{0}_{1}\widehat{M}^{0}_{2}h^{\perp(1)}_{q/A}(x_a,k_{\perp a})f_{q/B}(x_b,k_{\perp b})+\widehat{M}^{0}_{1}\widehat{M}^{0}_{3}f_{q/A}(x_a,k_{\perp a})h^{\perp(1)}_{q/B}(x_b,k_{\perp b})\right]. \tag*{(23)}

The matrix elements are given by the trace of ten Dirac matrices which evaluates to

M^10M^20=−8271tu[t(k⊥a⋅p2)(k⊥C⋅p4)+t(k⊥a⋅p2)(k⊥C⋅p2)−su(k⊥a⋅k⊥C)].(24)\widehat{M}^{0}_{1}\widehat{M}^{0}_{2}=-\frac{8}{27}\frac{1}{tu}\left[t(k_{\perp a}\cdot p_{2})(k_{\perp C}\cdot p_{4})+t(k_{\perp a}\cdot p_{2})(k_{\perp C}\cdot p_{2})-su(k_{\perp a}\cdot k_{\perp C})\right]. \tag*{(24)}
M^10M^30=−8271tu[u(k⊥b⋅p1)(k⊥C⋅p4)+u(k⊥b⋅p1)(k⊥C⋅p1)−st(k⊥b⋅k⊥C)].(25)\widehat{M}^{0}_{1}\widehat{M}^{0}_{3}=-\frac{8}{27}\frac{1}{tu}\left[u(k_{\perp b}\cdot p_{1})(k_{\perp C}\cdot p_{4})+u(k_{\perp b}\cdot p_{1})(k_{\perp C}\cdot p_{1})-st(k_{\perp b}\cdot k_{\perp C})\right]. \tag*{(25)}

The scalar products above lead to an azimuthal correlation of the form cos⁡(ϕk⊥a)\cos(\phi_{k_{\perp a}}) and cos⁡(ϕk⊥b)\cos(\phi_{k_{\perp b}}), which produces a contribution to the v1v_1 in the azimuthal distribution (c.f. Appendix A).

Quark initiated: Double Boer-Mulders' effect

Similarly, the quark BM⊗BMBM\otimes BM effect is given by the interference of the tt- and uu-channel diagrams which leads to the following Dirac trace

Σqq→qqBM⊗C=2DC/q(z,k⊥C)hq/A⊥(1)(xa,k⊥a)hq/B⊥(1)(xb,k⊥b)M^20M^30,(26)\Sigma^{\mathrm{BM}\otimes\mathrm{C}}_{qq\to qq}=2D_{\mathrm{C}/q}(z,k_{\perp C})h^{\perp(1)}_{q/A}(x_a,k_{\perp a})h^{\perp(1)}_{q/B}(x_b,k_{\perp b})\widehat{M}^{0}_{2}\widehat{M}^{0}_{3}, \tag*{(26)}

with the matrix element

M^20M^30=−8271tu[s(k⊥a⋅p3)(k⊥b⋅p4)+s(k⊥a⋅p4)(k⊥b⋅p3)+tu(k⊥a⋅k⊥b)].(27)\widehat{M}^{0}_{2}\widehat{M}^{0}_{3}=-\frac{8}{27}\frac{1}{tu}\left[s(k_{\perp a}\cdot p_{3})(k_{\perp b}\cdot p_{4})+s(k_{\perp a}\cdot p_{4})(k_{\perp b}\cdot p_{3})+tu(k_{\perp a}\cdot k_{\perp b})\right]. \tag*{(27)}

Here the scalar products lead to azimuthal correlation of the form cos⁡(ϕk⊥a)×cos⁡(ϕk⊥b)\cos(\phi_{k_{\perp a}})\times\cos(\phi_{k_{\perp b}}) which produces a contribution to the v2v_2 in the azimuthal distribution (c.f. A). All the matrix elements listed above for quark and gluon scattering will have to be convoluted with the appropriate TMDPDF and TMDFF, according to Eqs. (2). These non-perturbative functions, typically obtained from fits to data from other experiments, are not well constrained in many cases. In the subsequent section, we will describe how these functions are parameterized and how we modify these for use in pp-AA collisions. Final numerical results are presented in § 0.5.

Phenomenology of TMDPDF and TMDFF

In the preceding section, we outlined the matrix elements of all the processes that could contribute to high-pTp_T hadron production in pp-pp (and pp-AA) collisions. These include both polarization averaged and polarization dependent contributions. To complete the process, these matrix elements have to be convoluted with the appropriate TMDPDF and TMDFF.

In this section, we will discuss our chosen parameterizations for both the polarization averaged and polarization dependent PDF and FF. It will come as no surprise that some of these functions are not well constrained by experimental data. Our choice of parameterization is further constrained by the fact that our parameterization also has to be easily modified to the case of pp-AA. As implemented below, this will force us to use factorized (xx and k⊥k_{\perp}) distributions.

Mean transverse momentum

There have been several experimental and Lattice QCD results that have enabled multi-dimensional parameterizations of TMDPDFs [f(x,k⊥2,Q2)f(x,k_{\perp}^{2},Q^{2}), where the xx and k⊥k_{\perp} dependence is not factorized] [69, 70]. However, there are no parameterizations of un-factorized TMDPDFs including nuclear effects, that can be used for pp-AA collisions. There are two specific effects that concern us, the modification of the xx distribution of partons in the nucleus AA, referred to as shadowing [71], and the modification of the transverse momentum of a parton in the pp undergoing pre-scattering in the nucleus prior to the hard exchange that leads to dijets, referred to as the Cronin effect [72].

In the absence of a pre-scattering formalism that can simultaneously describe both these effects, we are left with a factorized parameterization of both these effects.

In the literature, for the case of factorized distributions, the transverse momentum dependence is typically approximated by a Gaussian Ansatz [66, 73]

f(x,k⊥,Q2)=4π⟨k⊥2⟩e−k⊥2⟨k⊥2⟩f(x,Q2).(28)f(x,k_{\perp},Q^{2}) = \frac{4\pi}{\langle k_{\perp}^{2}\rangle}e^{-\frac{k_{\perp}^{2}}{\langle k_{\perp}^{2}\rangle}}f(x,Q^{2}) . \tag*{(28)}

In the presence of a large nucleus, partons may undergo scattering off multiple nucleons before the hard exchange that leads to the energetic dijets. For the case of polarization averaged processes, it has been demonstrated that the extra acquired mean squared transverse momentum in pp-AA collisions (Q2⊥,A2Q_{2\perp,A}^{2}) is proportional to the length traversed, i.e., Q2⊥,A2∼A1/3Q_{2\perp,A}^{2} \sim A^{1/3} [74, 75, 76, 77, 78, 79]. Thus the total mean transverse momentum from pre-scattering, prior to the hard interaction may be expressed as

⟨k⊥2⟩pA≃Q2⊥,A2+⟨k⊥2⟩pp.(29)\langle k_{\perp}^{2}\rangle_{pA} \simeq Q_{2\perp,A}^{2} + \langle k_{\perp}^{2}\rangle_{pp} . \tag*{(29)}

In the equation above, ⟨k⊥2⟩pp\langle k_{\perp}^{2}\rangle_{pp} is the mean transverse momentum from the TMDPDF of the proton prior to interaction with the nucleus. In this first attempt to calculate the multiple moments of the elliptic anisotropy, we will invoke the Ansatz:

Q2⊥,A2∼A1/3⟨k⊥2⟩pp≫⟨k⊥2⟩pp.(30)Q_{2\perp,A}^{2} \sim A^{1/3}\langle k_{\perp}^{2}\rangle_{pp} \gg\langle k_{\perp}^{2}\rangle_{pp}. \tag*{(30)}

There are two aspects to this: The stronger statement is that the transverse momentum accrued from scattering off each nucleon is nearly equal to the intrinsic transverse momentum of typical partons in a nucleon, and that the nucleus in the pp-AA case is large, i.e., A1/3≫1A^{1/3} \gg1.

Given the above assumptions, for calculational convenience, we can incorporate the extra transverse momentum from pre-scattering within an effective transverse momentum distribution of the parton (from the proton) before hard scattering in pp-AA. Thus for the case of pp-AA, the parton distribution function in the proton is modified as:

fpA(x,k⊥,Q2)=4πA1/3⟨k⊥2⟩ppe−k⊥2A1/3⟨k⊥2⟩ppfpA(x,Q2).(31)f_{pA}(x,k_{\perp},Q^{2})=\frac{4\pi}{A^{1/3}\langle k_{\perp}^{2}\rangle_{pp}}e^{-\frac{k_{\perp}^{2}}{A^{1/3}\langle k_{\perp}^{2}\rangle_{pp}}}f_{pA}(x,Q^{2}) . \tag*{(31)}

In the above equation, we will employ an nPDF fpA(x,Q2)f_{pA}(x,Q^{2}) for the AA ion side.

In order to make sure the partons are generally collinear with the proton, we impose a kinematic constraint on the transverse momentum of the partons [65]

k⊥≤s×min⁡(x,x(1−x)).(32)k_{\perp}\leq\sqrt{s}\times\min\left(x,\sqrt{x(1-x)}\right) . \tag*{(32)}

While the factorization of the transverse momentum simplifies the calculation significantly, due to kinematic constraints the low-xx partons will not contribute to the cross section.

In an attempt to capture the complex relation between the transverse momentum and the momentum fraction, we use an xx-dependent ⟨k⊥2⟩\langle k_{\perp}^{2}\rangle ansatz. A parameterization of the mean transverse momentum in pp-pp collisions was presented in Ref. [66] as follows

⟨k⊥2⟩1/2(x)=⟨k⊥2⟩01/2x0.8(1−x)0.15.(33)\langle k_{\perp}^{2}\rangle^{1/2}(x)=\langle k_{\perp}^{2}\rangle_{0}^{1/2}x^{0.8}(1-x)^{0.15} . \tag*{(33)}

However, this ansatz can become large at intermediate xx as shown by the doted lines in Figure 1.

Different models of the mean transverse momentum for $p$-$p$ and $p$-$A$ collisions

Figure 1. Different models of the mean transverse momentum ⟨k⊥2⟩\langle k_{\perp}^{2}\rangle for pp-pp and pp-AA collisions where one side is enhanced by A1/3A^{1/3}. The dot-dashed lines indicate the case of xx-independent ⟨k⊥2⟩=A1/3\langle k_{\perp}^{2}\rangle= A^{1/3} GeV. Solid lines indicate the xx-dependent model of Eq. (34). The dotted lines indicate what would happen to this model if the model of x<x0x < x_{0} is allowed to apply for the entire range of xx. The solid blue line is the hard kinematic bound in Eq. (32).

Multiple scattering in deconfined matter leads to transverse momentum broadening governed by the transport coefficient q^\hat{q} [80]. The mean transverse momentum is proportional to q^L\sqrt{\hat{q}L}, where LL is the length of the medium. Perturbatively, q^\hat{q} is governed by a logarithmic dependence (log⁡E/E0)(\log E/E_{0}) on the energy of the hard parton EE and a soft screening scale E0E_{0} [80]. Guided by this result, we modify the xx-dependent ansatz of Ref. [66] as follows2,

⟨k⊥2⟩1/2(x)=⟨k⊥2⟩01/2((1−x)ss0)0.15A1/6×{(xx0ss0)0.8,if x≤x0[1+b0log⁡(xx0ss0)]1/2,if x>x0(34)\begin{aligned} \langle k_{\perp}^{2}\rangle^{1/2}(x)&=\langle k_{\perp}^{2}\rangle_{0}^{1/2}\left((1-x)\sqrt{\frac{s}{s_{0}}}\right)^{0.15}A^{1/6}\\ &\quad\times\begin{cases} \left(\frac{x}{x_{0}}\sqrt{\frac{s}{s_{0}}}\right)^{0.8}, & \text{if }x\leq x_{0}\\ \left[1+b_{0}\log\left(\frac{x}{x_{0}}\sqrt{\frac{s}{s_{0}}}\right)\right]^{1/2}, & \text{if }x>x_{0} \end{cases} \tag*{(34)} \end{aligned}

where we take ⟨k⊥2⟩0=0.9 GeV2\langle k_{\perp}^{2}\rangle_{0}=0.9\ \mathrm{GeV}^{2}, x0=10−3x_{0}=10^{-3}, s0=8.16 TeV\sqrt{s_{0}}=8.16\ \mathrm{TeV} and b0=0.1b_{0}=0.1. This is indicated by the solid red line in Figure 1. Using these constants one can reproduce the distribution of Ref. [66] at x≤x0x\leq x_{0}. Throughout this manuscript, unless stated otherwise, we will use this xx-dependent ansatz, we refer to [81] for results using a constant ⟨k⊥2⟩\langle k_{\perp}^{2}\rangle.

Since (34) involves non-trivial xx dependence, the true mean transverse momentum of the PDF is not directly the value of ⟨k⊥2⟩1/2(x)\langle k_{\perp}^{2}\rangle^{1/2}(x). To obtain what will refer to as the effective mean transverse momentum, one needs to integrate the full PDF as follows

⟨kT2⟩(Q)=∫dxa⟨k⊥2⟩(xa)f(xa,Q)∫dxaf(xa,Q),(35)\langle k_{T}^{2}\rangle(Q)=\frac{\int dx_{a}\langle k_{\perp}^{2}\rangle(x_{a})f(x_{a},Q)}{\int dx_{a}f(x_{a},Q)} , \tag*{(35)}

where QQ is the hard scale of the PDF. In contrast to Figure 1, the curve we obtain in Figure 2 leads to a mean transverse momentum lower than 1 GeV for all scales. This is because the support of the PDF becomes more and more concentrated at low xx as the scale increases. We point out these TMDPDF’s, even those for the case of pp-pp collisions, are phenomenological, with coefficients dialed to fit data. These do not obey the TMD evolution equations [82].

Mean transverse momentum as a function of the hard scale $Q$

Figure 2. Mean transverse momentum as a function of the hard scale QQ, obtained by carrying out the averaging of ⟨k⊥2⟩(x,Q2)\langle k_{\perp}^{2}\rangle(x,Q^{2}), over the allowed range of xx as described in Eq. (35). Due to the shift of parton distribution functions to smaller xx at larger QQ, the mean transverse momentum gradually reduces with increasing QQ.

Fragmentation Functions

The theoretical description of the production of pions from the hadronic scatterings, requires fragmentation functions as described in Eq. (8). Collinear and TMD unpolarized fragmentation functions have been studied extensively and are well constrained by data [83, 84, 85]. In order to simplify the analysis, we will employ the same factorized Gaussian Ansatz for the TMD fragmentation functions, as we have done for PDFs, i.e.

DC/c(z,k⊥C2)=4π⟨k⊥C2⟩e−k⊥C2⟨k⊥C2⟩DC/c(z).(36)D_{C/c}(z,k_{\perp C}^{2})=\frac{4\pi}{\langle k_{\perp C}^{2}\rangle}e^{-\frac{k_{\perp C}^{2}}{\langle k_{\perp C}^{2}\rangle}}D_{C/c}(z) . \tag*{(36)}

One can also consider a zz-dependent ⟨k⊥C2⟩\langle k_{\perp C}^{2}\rangle ansatz for the fragmentation functions [66]. However, this plays a very minor role for the processes that are being studied. Since the pion transverse momentum k⊥Ck_{\perp C} is transverse to the momentum of the parton cc, the fragmentation functions will not contribute to the azimuthal anisotropies in the plane transverse to the beam axis. For the zz-dependence of the fragmentation functions, we will use the KKP parameterization [83]. Similar to the case of the TMDPDF’s, these TMDFF’s are phenomenological and do not obey TMD evolution [82].

Polarized distributions

While the unpolarized collinear and TMD distributions can be extracted from global fits to data, the polarized Boer-Mulders’ and Collins’ functions arise in sub-leading effects making them harder to extract. However, constraints on the positivity of the density matrix lead to a bound on the polarized distributions known as the Soffer bound [64, 86].

hq⊥(1)(x,k⊥2)≡k⊥22Mp2hq⊥(x,k⊥2)≤fq(x,k⊥2)(37)h_q^{\perp(1)}(x,k_\perp^2) \equiv\frac{k_\perp^2}{2M_p^2}h_q^\perp(x,k_\perp^2) \leq f_q(x,k_\perp^2) \tag*{(37)}
hg⊥(1)(x,k⊥2)≡k⊥22Mp2hg⊥(x,k⊥2)≤fg(x,k⊥2)(38)h_g^{\perp(1)}(x,k_\perp^2) \equiv\frac{k_\perp^2}{2M_p^2}h_g^\perp(x,k_\perp^2) \leq f_g(x,k_\perp^2) \tag*{(38)}
Hq⊥(1)(x,k⊥2)≡k⊥22Mp2Hq⊥(x,k⊥2)≤Dq(x,k⊥2)(39)H_q^{\perp(1)}(x,k_\perp^2) \equiv\frac{k_\perp^2}{2M_p^2}H_q^\perp(x,k_\perp^2) \leq D_q(x,k_\perp^2) \tag*{(39)}
Hg⊥(1)(x,k⊥2)≡k⊥22Mp2Hg⊥(x,k⊥2)≤Dg(x,k⊥2).(40)H_g^{\perp(1)}(x,k_\perp^2) \equiv\frac{k_\perp^2}{2M_p^2}H_g^\perp(x,k_\perp^2) \leq D_g(x,k_\perp^2). \tag*{(40)}

For the present work, we will consider the polarized distributions to be proportional to the unpolarized distributions, i.e. hq⊥(1)(x,k⊥2)=bfq(x,k⊥2)h_q^{\perp(1)}(x,k_\perp^2)=b f_q(x,k_\perp^2) and Hq⊥(1)(x,k⊥2)=BDq(x,k⊥2)H_q^{\perp(1)}(x,k_\perp^2)=B D_q(x,k_\perp^2). Since all the processes that we consider are proportional to the product of two polarized distributions, we will only modulate the overall factor of3^{3} B⋅bB \cdot b. Once may assume that B≃bB \simeq b. We present results for different values (0.2≤B⋅b≤0.40.2 \leq B \cdot b \leq0.4) in § 0.5.

Bulk simulations and elliptic flow

In the preceding section, we have described the 2-to-2 scattering of polarized and unpolarized partons (a+b→c+da+b \to c+d), with transverse momentum in the initial state. If the two out-going hard partons have transverse momenta pT,cp_{T,c} and pT,dp_{T,d}, then there will be an asymmetry in the azimuthal angle ϕ\phi between qT=pT,c+pT,d\boldsymbol{q}_T=\boldsymbol{p}_{T,c}+\boldsymbol{p}_{T,d} and QT=pT,c−pT,d≈2pT,c\boldsymbol{Q}_T=\boldsymbol{p}_{T,c}-\boldsymbol{p}_{T,d}\approx2\boldsymbol{p}_{T,c} [48, 59, 60].

However, the angle ϕqT\phi_{q_T}, which indicates the direction of the vector qT\boldsymbol{q}_T, varies from event to event. The azimuthal asymmetry in the angle between qT\boldsymbol{q}_T and QT\boldsymbol{Q}_T will yield anisotropy coefficients (vnv_n) if there is a correlation between ϕqT\phi_{q_T} and the event plane angles Ψn\Psi_n. That is, if the system can be reoriented in each event, based on the detected particles in that event. In this section, we will demonstrate that ϕqT\phi_{q_T} is strongly correlated with the event plane angles of the colliding system.

Elliptic flow in small systems

Azimuthal angular distributions of particles are expressed in terms of the Fourier series

dNdϕdpTdy=dNdpTdy(1+2∑nvncos⁡[n(ϕ−Ψn)])(41)\frac{dN}{d\phi dp_Tdy}=\frac{dN}{dp_Tdy}\left(1+2\sum_n v_n\cos[n(\phi-\Psi_n)]\right) \tag*{(41)}

where the Ψn\Psi_n are event plane angles, with dNdN the differential yield of particles with momentum given as pz=pT2+m2sinh⁡(y)p_z=\sqrt{p_T^2+m^2}\sinh(y) and px=pTcos⁡(ϕ),py=pTsin⁡(ϕ)p_x=p_T\cos(\phi),p_y=p_T\sin(\phi).

In the case of a heavy-ion collision, the origin of azimuthal anisotropies in the momentum distribution of particles is now well understood: An initial geometric asymmetry, i.e. a spatially asymmetric density distribution leads to pressure gradients, which in turn lead to an asymmetric momentum distribution. In the case of small systems, there is no geometrical asymmetry in the initial state. In events without hard scattering, any asymmetries are generated entirely from fluctuations in the initial state [87]. However, the presence of a hard scattering leads to a much larger multiplicity, that overwhelms the yield of particles produced in entirely soft interactions [88]. In the case of collisions with hard partonic scattering, multiplicity or transverse energy production at almost all rapidities is dominated by hadrons that originate in the hard scattering, or are connected, in some way, to the hard scattering, e.g., in the breaking of strings that stretch from the hard high pTp_T parton to the remnants that go down the beam line [89].

Particle production, and the eventual momentum asymmetries, that are effected by the hard sub-process, may be described as follows: Let us consider a hard parton carrying momentum fraction xax_a of the proton AA. If the hard parton possesses, in addition, a transverse momentum k⊥,a\boldsymbol{k}_{\perp,a}, orthogonal to the proton’s momentum, the remaining partons (and any other constituents) inside the proton must have the same magnitude of transverse momentum, but in the opposite direction, i.e., −k⊥,a-\boldsymbol{k}_{\perp,a}.

On the left side of Figure 3 we present the initial state of a proton-proton scattering where the blue proton with a solid boundary is coming out of the page, the red proton with the dotted boundary goes into the page. The hard parton is indicated by the solid blue/red dot and the remainder of the proton is indicated by the blue/red blob. The balanced transverse momenta between the hard parton and the remainder, in the blue proton, are indicated by the opposing blue arrows. The opposing red arrows indicate a similar situation in the red proton. In the illustration, we show the hard parton as physically separated from the remainder of the photon. This is just for illustration, the hard parton may indeed be within the vicinity of the remainder.

Figure 3. Left: A cartoon of a pp-pp collision with relative transverse momentum in the initial state. The blue proton indicated by the solid blue circle comes out of the page, and contains a hard parton (blue dot) with transverse momentum, recoiling off the rest of the proton represented by the blue blob moving in the opposite direction. The net transverse momentum of the proton is zero, and so the hard parton and the remainder of the proton are balanced in transverse momentum. The red proton going into the page has a similar situation with its hard parton and remainder also balanced in transverse momentum. Right: the final state after the collision can be separated in a part containing the dijet pair resulting from the scattering of the two hard partons recoiling off the remainder of the colliding system.

After the scattering, hard partons produce 2 back-to-back jets in the right panel of Figure 3. The center of mass of the final state partons is shifted by the transverse momentum qT=k⊥,a+k⊥,b=p⊥,c+p⊥,d\boldsymbol{q}_T = \boldsymbol{k}_{\perp,a}+\boldsymbol{k}_{\perp,b} = \boldsymbol{p}_{\perp,c}+\boldsymbol{p}_{\perp,d}. While the entire remainder does not participate in the collision, one expects that the total transverse momentum of the bulk is generally opposite to the hard scattering ∼−qT\sim-\boldsymbol{q}_T.

In the illustration described above (and in Figure 3), we have considered the case of a single 2-to-2 hard scattering, with the remainder of the colliding system being soft. This does not have to be the case, the remainder of the participating system may engender multiple hard scatterings as well. In this case, the dots in our illustration indicate the two partons that undergo the hardest exchange. The blob will represent the remainder of hard and soft scatterings. None of our considerations above are affected by this. We will investigate these assumptions further by considering a simple PYTHIA [57] simulation in the subsequent subsec. IV B.

PYTHIA simulations

In this paper, we have studied the azimuthal distribution of high-pTp_T hadrons produced in the scattering of hard partons that possess transverse momentum in the initial state. To experimentally determine this azimuthal distribution, one needs a reference direction in the azimuthal plane. Since the incoming state consists of unpolarized protons, no reference angle is provided by the initial state; it has to be determined from the distribution of hadrons in the final state.

In the preceding subsection, we argued that the vector qT=k⊥,a+k⊥,b=p⊥,c+p⊥,d\boldsymbol{q}_T = \boldsymbol{k}_{\perp,a}+\boldsymbol{k}_{\perp,b} = \boldsymbol{p}_{\perp,c}+\boldsymbol{p}_{\perp,d} would be correlated with the event plane angle(s). In this subsection, we demonstrate this with a PYTHIA simulation [57]. As will be demonstrated in the next section, the largest component of our calculated azimuthal asymmetry is for the case of polarization dependent scattering. Such simulations are not possible using PYTHIA, or any other event generator that correlates hard and soft particle production. Our simulations will thus demonstrate the angular correlation only for the case of polarization averaged scattering, which will constitute a small component of the calculated azimuthal anisotropy. Also, in PYTHIA simulations with multiple interactions turned on, the initial transverse momentum of each parton that eventually encounters a hard scattering, is balanced of its sole remnant and not distributed among other partons and remnants in the same proton.

To overcome the above limitations of typical PYTHIA simulations, we set up events where initial state radiation, final state radiation and multiple parton interactions are turned off. Each proton now consists of a hard parton that will participate in the hard scattering (small dots in Figure 3), other partons that do not participate in the hard scatterings, and diquarks, that along with the non-participating partons, constitute the remnant or reminder of the proton (blobs in Figure 3).

The entire events thus consist of two hard partons which undergo a hard scattering and remnants that proceed down the beam line. To simulate TMD effects, we generate a transverse momentum kTik_{T_i} for each remnant, from a Gaussian distribution with a width of ⟨k⊥2⟩1/2\langle k_{\perp}^{2}\rangle^{1/2}. The total transverse momentum of all remnants qT=∑ikTi\boldsymbol{q}_T = \sum_i \boldsymbol{k}_{T_i} is compensated by applying a Lorentz boost to the center of mass frame of the hard scattering with the velocity

v=qT(Ea+Eb)2+qT2 .(42)\boldsymbol{v} = \frac{\boldsymbol{q}_T}{\sqrt{(E_a+E_b)^2+\boldsymbol{q}_T^2}} \, . \tag*{(42)}

This leads to initial state partons with a transverse momentum following a Gaussian distribution with a width comparable to the remnants’ width. By using a boost to generate the transverse momentum of the initial state partons, we ensure that the kinematics of the scattering is unmodified, such that we ensure energy-momentum conservation in the hard-scattering while balancing the transverse momentum between the hard scattering and the remnants partons.

The above equation also does not work in the case there is initial state radiation, as the hard partons are then not directly balanced off the remnants. The remnants and final state partons are hadronized in PYTHIA using the Lund string model [90], where strings are connected between the hard outgoing partons and the beam remnants.

Event plane angle

As described in the previous subsection, we have generated events where the hard scattering system of two partons has a net transverse momentum, qT=p⊥,c+p⊥,dq_T=\mathbf{p}_{\perp,c}+\mathbf{p}_{\perp,d}, where p⊥,c\mathbf{p}_{\perp,c} and p⊥,d\mathbf{p}_{\perp,d} are the large momenta of the outgoing partons, which are now no longer back-to-back. The remainder of the event, which consists of beam remnants that are connected via strings with the hard partons and remnants which are not connected to a hard scattering, will recoil with a net transverse momentum −qT-q_T.

Once the entire event is hadronized, the event plane angle Ψn\Psi_n in Eq. (41) can be obtained using the flow vector QnQ_n [91] defined as follows,

Qn=1M∑jMeinϕj,(43)Q_n=\frac{1}{\sqrt{M}}\sum_j^M e^{in\phi_j}, \tag*{(43)}

where the sum runs over all hadrons in the event (typically over some rapidity range) and ϕj\phi_j is the azimuthal angle of the hadron jj. The event plane angle is then defined as the Euler angle of the flow vector QnQ_n as follows

Ψn=1narg⁡[Qn].(44)\Psi_n=\frac{1}{n}\arg[Q_n]. \tag*{(44)}

In Figure 4, we compute correlation histograms of the values of the event plane angle Ψn\Psi_n and the azimuthal angle of the total transverse momentum qTq_T. For both Ψ1\Psi_1 and Ψ2\Psi_2, we observe a clear correlation between the two angles which becomes more pronounced as the mean transverse momentum is increased ⟨kT2⟩1/2=5\langle k_T^2\rangle^{1/2}=5 GeV. This implies that in the presence of hard scattering, leading to two high pTp_T jets, the net transverse momentum of the jets qT=p⊥,c+p⊥,dq_T=\mathbf{p}_{\perp,c}+\mathbf{p}_{\perp,d}, is tightly correlated with the 1st and

Figure 4. Histogram of the correlation between the event plane angle and the azimuthal angle of the total transverse momentum for ⟨kT2⟩1/2=1\langle k_T^2\rangle^{1/2}=1 GeV (left) and ⟨kT2⟩1/2=5\langle k_T^2\rangle^{1/2}=5 GeV (right) using the event plane angle Ψ1\Psi_1 (top) and Ψ2\Psi_2 (bottom).

2nd order event plane (it is also correlated with higher order event planes, though more weakly). Thus, in events with a hard scattering, the direction of the qTq_T vector can be determined, by analyzing all the hadrons in an event, and thus the azimuthal distribution of p⊥,c/d\mathbf{p}_{\perp,c/d} with respect of qTq_T can indeed be determined.

Centrality selection

While not a central issue in our calculations, our PYTHIA simulations reveal another interesting feature of these small system collisions: The correlation between the transverse momentum of the incoming partons and the transverse energy produced in the rapidity of forward calorimeters. There are several methods used by experiments to select the centrality of the collision. One such method is based on the transverse energy in the forward calorimeter (FoCal) which measures hadrons in the pseudo-rapidity range 3.9≤∣η∣≤5.03.9\leq|\eta|\leq5.0. Transverse momentum of the initial hard partons can lead to additional transverse energy in the FoCal.

Using the results of the simulations in the previous subsection, we compute the transverse energy in each event as,

ET=∑iEisin⁡θi,(45)E_T=\sum_i E_i\sin\theta_i, \tag*{(45)}

where EiE_i is the energy of the final state hadron and θi\theta_i is the angle between the hadron momentum and the zz-axis. The sum runs over all hadrons in the event within a prescribed range in (pseudo-)rapidity. The average transverse energy ⟨ET⟩\langle E_T\rangle is shown in Figure 5 as a function of the average transverse momentum of the initial hard parton ⟨k⊥2⟩1/2\langle k_{\perp}^{2}\rangle^{1/2}. We find a clear linear relation between the initial transverse momentum of the hard parton and the transverse energy in the forward direction.

Average transverse energy in the forward direction

Figure 5. Average transverse energy ⟨ET⟩=Esin⁡θ\langle E_T\rangle=E\sin\theta in the forward direction 3≤∣η∣≤103\leq|\eta|\leq10 as a function of the average transverse momentum squared of the initial hard parton ⟨kT2⟩1/2\langle k_T^2\rangle^{1/2}.

Experimental data on the elliptic anisotropy, from the ATLAS detector [50], that will be compared against, are extracted from the most central (0-5%) events in pp-PbPb collisions. In these experiments, centrality is determined from the amount of transverse energy deposited in the FoCal. The results of our simulations above indicate that larger transverse energy deposited in the FoCal isolates events where the mean intrinsic transverse momentum of the initial state partons is larger. These are events where the correlation between the direction of qT\boldsymbol{q}_T and the event plane angles is more enhanced (See Figure 4). This further enhances our argument that the angle of qT\boldsymbol{q}_T can be determined in experiment.

Results

In the previous section, we argued that events with hard scattering that contain a dijet pair recoiling off the remainder of the colliding protons will have a well defined event plane, with several event plane angles coinciding with the angle of the dijet momentum qT=p⊥,c+p⊥,d\boldsymbol{q}_T=\boldsymbol{p}_{\perp,c}+\boldsymbol{p}_{\perp,d} (where p⊥,c\boldsymbol{p}_{\perp,c} and p⊥,d\boldsymbol{p}_{\perp,d} are the large transverse momenta of the out-going jets, see Figure 3). We further argued that more central events isolated via larger transverse energy deposition in the forward calorimeter, enhances events with larger intrinsic transverse momentum in the initial state, leading to events with a larger qT\boldsymbol{q}_T.

In the calculations that follow, we will assume that qT\boldsymbol{q}_T is coincident with the event plane angle(s). In reality, the event plane angle will be strongly correlated with the direction of qT\boldsymbol{q}_T. To compensate for this assumed coincidence, we will keep the mean ⟨k⊥⟩\langle k_{\perp}\rangle, for the case of the xx-dependent k⊥k_{\perp} distribution, somewhat below 1 GeV in the case of pp-pp collisions, as shown in Figure 2. This xx-dependent k⊥k_{\perp} distribution is used for the preponderance of spectra and azimuthal anisotropy calculations that follow, and is enhanced by a factor of A1/3A^{1/3} for the case of pp-PbPb. For comparison purposes only, we also present the pion spectrum in pp-pp and pp-PbPb collisions (Figure 7) using the fixed ⟨k⊥⟩=1\langle k_{\perp}\rangle=1 GeV2^{2} prescription. Thus, we are using values of ⟨k⊥2⟩\langle k_{\perp}^{2}\rangle that are closer to the lower limit considered in our PYTHIA simulations.

Pion production cross section in p-p and p-Pb collisions

Figure 7. Pion production cross section using gluon-gluon scattering only in pp-pp at 5.02 TeV (top) and pp-PbPb at 8.16 TeV (bottom). The unpolarized contribution (red dashed line squares) is dominant over the BM⊗CBM \otimes C (purple lines) and the BM⊗BMBM \otimes BM contributions (green lines). The cross section using an xx-dependent ⟨k⊥2⟩\langle k_{\perp}^{2}\rangle reproduces the collinear results (blue lines) at high pT≲5p_{T} \lesssim5 GeV. Experiment results from ALICE [92] and ATLAS [93].

Beyond this, there will be no further reference to the PYTHIA simulations. No further simulations will be carried out. Our results will be from analytical calculations. The anisotropy coefficients will be calculated assuming that qT\boldsymbol{q}_T coincides with the direction of the event plane. We remind the reader that a majority of the calculated anisotropy will result from polarization dependent scattering; and currently, there is no event generator that can simulate polarization dependent processes. Thus, there is no alternative other than semi-analytical calculations.

For our calculations, we employ the nCTEQ parameterization [95] for integrated PDFs [f(x,Q2)][f(x,Q^2)] and leading order KKP [83] for FFs. While we will use either a constant or xx-dependent mean transverse momentum for the TMD-PDFs, we will only employ a constant ⟨k⊥2⟩1/2=0.8\langle k_{\perp}^{2}\rangle^{1/2}=0.8 GeV for the pion’s TMD-FFs, following the work of Ref. [65, 66].

When using a constant transverse momentum ⟨k⊥2⟩\langle k_{\perp}^{2}\rangle the Gaussian k⊥k_{\perp} dependence integrates out, recovering the results from collinear factorization. We display the angular integrated cross section, using full lines in Figure 6, for different partonic scatterings ab→cdab\to cd, with the fragmenting parton cc. As expected, the gluon PDFs are dominant at lower values of the Q2Q^2 scales; this is partially responsible for the gg→gggg\to gg partonic scattering contribution remaining dominant up to pT≲80p_T\lesssim80 GeV, beyond which quark initiated processes start to contribute. We also include the experimental data from the ALICE collaboration [92], which, given the limited range in pTp_T, is almost completely described by the gg→gggg\to gg process.

Pion production cross section in p-p at 5.02 TeV

Figure 6. Pion production cross section in pp-pp at 5.02 TeV. The top panel displays unpolarized contributions from gg→gggg \to gg (blue), qg→qgqg \to qg (red), qq→qqqq \to qq (green) and gq→gqgq \to gq (orange) partonic scatterings. The bottom panel displays the ratio to the gg→gggg \to gg scattering which dominates the cross section for pT≲90p_T\lesssim90 GeV. Compared with ALICE data [92].

In Figure 7, we further decompose the gg→gggg\to gg cross section with xx-dependent ⟨k⊥2⟩\langle k_{\perp}^{2}\rangle into the polarization averaged (dashed red line squares), and polarization dependent contributions which include a Boer-Mulders distribution with a Collins fragmentation function (BM⊗CBM\otimes C in green solid) and a double Boer-Mulders distribution (BM⊗BMBM\otimes BM in maroon solid). The upper panel represents pp-pp at sNN=5.02\sqrt{s_{NN}}=5.02 TeV, while the lower panel is for pp-PbPb collisions, at sNN=5.02\sqrt{s_{NN}}=5.02 TeV. In blue lines we present the cross section for fixed ⟨k⊥2⟩\langle k_{\perp}^{2}\rangle which recovered the collinear result as the Gaussian profile integrates to unity.

Since polarization phases do not cancel for the polarized cross sections, the angular integrated spectrum is suppressed compared to the unpolarized cross section. The xx-dependence leads to a deviation from the collinear cross section at low pTp_{T}, visibly for the case of pp-PbPb (and

barely for the case of pp-pp). However, at pT≳5p_{T} \gtrsim5 GeV, the effect of the transverse momentum on the cross section becomes negligible. Above this pTp_{T} (≳5\gtrsim5 GeV), the cross section is indistinguishable from the collinear result as shown in Figure 7. As mentioned in Sec. III, for the case of pp-PbPb, shown in the bottom panel of Figure 7, the ⟨k⊥2⟩\langle k_{\perp}^{2}\rangle for the proton is enhanced by A1/3A^{1/3} to account for multiple soft interaction in the PbPb prior to the hard scattering.

We present the nuclear modification factor RpPbR_{pPb} in Figure 8 for pion production. No significant modification is observed between the pp-pp and pp-PbPb spectra consistent with the minimum bias ALICE experimental result [94].

Nuclear modification factor for pion production in p-Pb

Figure 8. Nuclear modification factor RpPbR_{pPb} for pion production in pp-PbPb at 5.02 TeV. We compare with ALICE results [94].

The cross section can be written as an expansion in the azimuthal angle ϕπ\phi_{\pi} as follows

dσdpTdydΔϕ=dσdpTdy(1+2∑n=1∞vncos⁡(nΔϕ))(46)\frac{d\sigma}{dp_{T}dy d\Delta\phi}=\frac{d\sigma}{dp_{T}dy}\left(1+2\sum_{n=1}^{\infty}v_{n}\cos(n\Delta\phi)\right) \tag*{(46)}

where Δϕ≡ϕπ−ϕqT\Delta\phi\equiv\phi_{\pi}-\phi_{q_{T}} is the azimuthal angle of the pion ϕπ\phi_{\pi} with respect to the total transverse momentum of the hard scattering qT\mathbf{q}_{T}. The flow coefficients vnv_{n}, can be obtained by integrating over the azimuthal angle Δϕ\Delta\phi as follows

vn=⟨cos⁡(nΔϕ)⟩=∫dΔϕcos⁡(nΔϕ)dσdpTdydΔϕ∫dΔϕdσdpTdydΔϕ(47)v_{n}=\langle\cos(n\Delta\phi)\rangle=\frac{\int d\Delta\phi\cos(n\Delta\phi)\frac{d\sigma}{dp_{T}dy d\Delta\phi}}{\int d\Delta\phi\frac{d\sigma}{dp_{T}dy d\Delta\phi}} \tag*{(47)}

The elliptic flow coefficient v2v_{2} obtained is displayed in Figure 9 (pp-pp in the top panel and pp-PbPb on the bottom panel). The uncertainty over the functional form of the mean transverse momentum is reflected in the red (pp-pp) and green (pp-PbPb) filled areas which is the bound between the mean transverse momentum $0.1\ \mathrm{GeV}\leq\langle k_{\perp}^{2}\rangle_{0}\leq 0.90.9 GeV2^{2} and using B⋅b=0.2−0.5B\cdot b=0.2-0.5 for the polarized contribution. Within uncertainties, the elliptic flow coefficient v2v_2 is consistent with the experimental data from ATLAS [49, 50] for pT≳6p_T\gtrsim6 GeV.

Elliptic flow coefficient $v_2$ for pion production in $p$-$p$ at 5.02 TeV and $p$-Pb at 8.16 TeV

Figure 9. Elliptic flow coefficient v2v_2 for pion production in pp-pp at 5.02 TeV (top) and pp-Pb at 8.16 TeV (bottom). The green and red filled areas are the uncertainty over the functional form of the mean transverse momentum between ⟨k⊥2⟩0=0.1−0.9\langle k_{\perp}^{2}\rangle_{0}=0.1-0.9 GeV2^{2} using B⋅b=0.2−0.5B\cdot b=0.2-0.5 for the polarized contributions.

To investigate the importance of the polarization dependent terms versus polarization averaged terms, we present the different contributions to the elliptic flow the xx-dependent mean transverse momentum in Figure 10 with ⟨k⊥2⟩0=0.9\langle k_{\perp}^{2}\rangle_{0}=0.9 GeV2^{2}. The unpolarized contribution

The different contributions to the elliptic flow coefficient $v_2$

Figure 10. The different contributions to the elliptic flow coefficient v2v_2 in pp-Pb at 8.16 TeV using the xx-dependent ⟨k⊥2⟩1/2(x)\langle k_{\perp}^{2}\rangle^{1/2}(x) GeV and B⋅b=0.4B\cdot b=0.4 for the unpolarized distribution in pp-pp (top) and pp-Pb (bottom). The hatched blue area is the unpolarized contribution, the green area is the BM⊗CBM\otimes C terms.

is shown in blue hatched area, and the BM⊗CBM\otimes C terms are shown in green area. Although we use B⋅b=0.5B\cdot b=0.5 for the polarized contribution the BM⊗CBM\otimes C terms are dominant compared to the unpolarized contribution at high-pTp_T. We observe that while the unpolarized contribution becomes negligible when pT≫k⊥p_T\gg k_{\perp}, the BM⊗CBM\otimes C terms continue to contribute, and lead to a large azimuthal anisotropy coefficient, even at large transverse momentum. This because the unpolarized contribution is power suppressed.

To further explore the azimuthal anisotropies arising from TMDPDFs, we compute higher order harmonics v3v_3 and v4v_4 which are displayed in Figure 15. The experimental results of v3v_3 and v4v_4 display a different behavior than v2v_2. While v2v_2 starts decreasing with pTp_T already at pT≳5p_T \gtrsim5 GeV, v3v_3 and v4v_4 are increasing up to the highest data points (pT≃10p_T \simeq10 and pT≃4p_T \simeq4, respectively). However, the data displays large uncertainties at high-pTp_T and is limited to pT≤10p_T \leq10 GeV. While experimental data is only available for pT≤10p_T \leq10 GeV, the polarized contributions lead to non-zero coefficients at high-pTp_T. Markedly, v4v_4 is entirely driven by the BM⊗BMBM \otimes BM terms at high-pTp_T, which does not contribute to any vnv_n coefficient with n<4n < 4.

Azimuthal asymmetry plots for $v_3$ and $v_4$ in p-Pb collisions

Figure 15. Same as Figure 10 but for v3v_3 (top) and v4v_4 (bottom). The red filled area is for BM⊗BMBM \otimes BM contribution.

To obtain further insight into the origin of different vnv_n’s from different terms, we compute the azimuthal angular distribution of different contributions to the cross section, at a fixed pTp_T. In Figure 11, we present the azimuthal angle distribution dσdϕ\frac{d\sigma}{d\phi} for gg→gggg \to gg scattering in pp-pp and pp-PbPb collisions. We find that for the pp-pp scattering the unpolarized contribution has a large constant term and a small v1v_1 contribution. While the BM⊗CBM \otimes C and BM⊗BMBM \otimes BM terms contribute predominantly to v2v_2 and v4v_4, respectively. Conversely, for pp-PbPb collisions, while different terms display similar symmetries, they gain an additional elliptic shape due to the asymmetry of the transverse momentum distribution, leading to a larger v2v_2. We remind the reader that all terms in Figure 11 are for the process gg→gggg \to gg (quark scattering contributions are included in the appendix).

Azimuthal angle distributions for unpolarized, $BM \otimes C$, and $BM \otimes BM$ contributions in $p$-$p$ and $p$-$Pb$ collisions

Figure 11. Azimuthal angle distribution dσdϕ\frac{d\sigma}{d\phi} for gg→gggg \to gg scattering pion production in pp-pp at 5.02 TeV (top) and pp-PbPb at 8.16 TeV (bottom) at a fixed pT=6p_T = 6 GeV. The panels from left to right represents the unpolarized, BM⊗CBM \otimes C and BM⊗BMBM \otimes BM contributions.

Predictions at RHIC

We investigate azimuthal anisotropies at RHIC energies for pp-pp and pp-AuAu collisions at sNN=200\sqrt{s_{NN}} = 200 GeV. The nuclear modification factor RpAuR_{pAu} was measured by the PHENIX collaboration [96] and is shown in Figure 12. While our results is restricted to LO matrix elements, they are compatible with NLO calculations obtained using the same nCTEQ PDFs in [96]. We observe minimal modification between pp-pp and pp-AuAu collisions, consistent with the experimental data.

Nuclear modification factor $R_{pAu}$ for $p$-$p$ and $p$-$Au$ collisions at $\sqrt{s}=200$ GeV

Figure 12. Same as Fig. 8 for pp-pp and pp-AuAu at s=200\sqrt{s}=200 GeV. Calculations include shadowing as included in the nCTEQ PDF [95]. The experimental points are from PHENIX [96].

At RHIC energies with s=200\sqrt{s}=200 GeV, the quark FFs become dominant at higher pTp_T such that the quark initiated processes are more important as shown in Figure 13. As such, we compute the azimuthal anisotropy v2v_2 for pp-pp and pp-AuAu collisions at s=200\sqrt{s}=200 GeV in Figure 14 using both gg→gggg \to gg and qq→qqqq \to qq. While the cross section for compton scattering is also important, the spin-dependent processes will require a BM⊗\otimesC contribution on the quark line which should contribute a factor cos⁡ϕ\cos\phi which is less important for v2v_2. We neglect these terms for our calculations where their contribution should be within our theoretical uncertainties.

Cross sections and ratios for $p$-$p$ collisions at 200 GeV

Figure 13. Same as Fig. 6 for pp-pp at s=200\sqrt{s}=200 GeV.

Azimuthal asymmetry as a function of transverse momentum for p-p and p-Au collisions at 200 GeV

Figure 14. Same as Figure 9 for pp-pp and pp-AuAu at 200 GeV.

The results are similar to what we observe at LHC energies, with non-zero v2v_2 at high-pTp_T. However, limited by the collision energy both the cross section and the azimuthal anisotropy dies off at very high-pTp_T. These results show that azimuthal anisotropies present a crucial avenue to explore the transverse momentum distribution of partons in the initial state at RHIC energies.

Summary and Outlook

In hard scattering processes calculated within the ambit of collinear factorization, the incoming partons possess no transverse momentum with respect to the direction of the incoming nucleons (in pp-pp or pp-PbPb collisions). In such processes, the outgoing partons and hadrons demonstrate no azimuthal anisotropy. However, in the event the incoming partons have a small amount of transverse momentum, anisotropies in the azimuthal distribution arise. The presence of transverse momentum in the initial state parton distribution, both with and without the presence of transverse polarization (linear polarization for gluons), leads to a scattering cross section that is not isotropic in azimuthal angle.

In this paper, we have presented detailed results on the first attempt to calculate these anisotropies and compare them to experimentally measured Fourier coefficients (of these anisotropies) in pp-pp and pp-PbPb collisions. This attempt presents complete details of our calculations, which were briefly presented in Ref. [81]. In the range of pTp_T explored by experimental data, the dominant channel is gg→gggg \to gg (Figure 6). Within this channel, the dominant contribution to the angle integrated cross section is the unpolarized process (Figure 7). However, gg→gggg \to gg, with one gluon linearly polarized (Boer-Mulders’ distribution), followed by the fragmentation of the linearly polarized gluon to an unpolarized hadron (Collins’ function), provides the largest contribution to the elliptic coefficient v2v_2 (Figure 9), followed by the polarization averaged process (Figure 10).

We calculate using a modified version of the xx-dependent mean ⟨k⊥2⟩\langle k_\perp^2\rangle model of TMDPDF from Ref. [66]. The typical mean ⟨k⊥2⟩\langle k_\perp^2\rangle is set to be ∼0.9 GeV2\sim0.9\ \mathrm{GeV}^2 in pp-pp for the xx-dependent ansatz (Figure 2). These values are enhanced by A1/3A^{1/3} of the pp in pp-AA to account for prescattering off multiple nucleons prior to the hard exchange that produces the dijet.

The A1/3A^{1/3} enhancement required a factorized form for the TMDPDF (separate functions for xx and k⊥k_\perp dependence), as multiple scattering based calculations for polarization dependent TMDs do not exist at this time. We further vary the ratio between the polarization dependent and independent distributions (0.2≤b⋅B≤0.50.2 \leq b \cdot B \leq0.5) to obtain a range of results for v2v_2. This is done as the exact magnitudes of the Boer-Mulders and Collins functions are not well known, at this time. In the absence of a bulk simulation which determines the event plane, we have argued in Sec. III that there is a strong correlation between the event plane angle and that of the dijet momentum qTq_T. Calculations of vnv_n were then done using the direction of qTq_T as the event plane angle.

In spite of these overlapping aspects of ignorance, our modest range of values for the Boer-Mulders’ and Collins’ functions seems to straightforwardly encapsulate the v2v_2 in pp-pp. Even more encouraging is the A1/3A^{1/3} enhancement for pp-AA collisions reproducing the larger v2v_2 in pp-PbPb. Our calculations show a mild tension with the measured v3v_3, and the comparisons with v4v_4 are inconclusive, as these have not been measured to a sufficiently high pTp_T.

We remind the reader that this is the first calculation of the azimuthal anisotropy at high pTp_T, arising from transverse momentum of the incoming partons in the initial state, in pp-pp and pp-AA collisions. As such, no solace should be drawn from the agreement with the measured v2v_2, and no alarm should be perceived in the tension with the measured v3v_3. The results of this paper, and the accompanying letter [81], should be taken as an argument for further exploration of TMDPDFs as a source of azimuthal anisotropy in small system collisions. Without having done a full study, our naïve expectation is that these effects will be diminished in systems with an extended final state, such as in heavy-ion collisions, where energy loss may completely wash out these initial state effects. In fact, it may even be interesting to study the interference of this effect with the typical anisotropy from energy loss in semi-central collisions of intermediate size ions, e.g., O+OO+O collisions [97].

Restricting our attention to pp-pp and pp-AA collisions, we point out that all parton distribution functions used were leading twist and there could be sizable contributions at lower pTp_T from higher twist contributions [98]. These contributions could also be important for v3v_3 (and higher harmonics). As one moves towards a description of the remainder of the proton participating in the collision [99], yet another contribution could originate in generalized transverse momentum dependent distributions [100].

The A1/3A^{1/3} enhancement in the case of pp-AA is only known for the case of polarization averaged scattering, and has to be carried out for polarization dependent scattering. Also, one should carefully check for the possibility of energy loss in the case of multiple scattering of partons in a large nucleus prior to the hard scattering that leads to high pTp_T jets [74, 101]. These calculations will rigorously test the hypothesised broadening without energy loss, required for the existence of an elliptic anisotropy with no modification to the angle integrated spectrum.

Most prior calculations of multiple scattering of partons in nuclear matter [74, 75, 76, 101, 102] have not considered gluons as the hard parton. For the currently measured pTp_T range, gluon scattering is the most dominant effect, and thus, many of these calculations will have to be repeated for gluons. While gluon-gluon scattering is by far the largest contribution in the considered pTp_T range, quark initiated processes will begin to become comparable at higher pTp_T (see the Appendix). Quark mass effects (not necessarily for heavy-flavors) may allow for additional enhancement [60].

With further improvement in semi-analytical calculations, these effects will eventually have to be studied via extensive simulations, where both energy-momentum and angular momentum correlations between the hard and soft sectors can be thoroughly investigated. Any simulation, similar to the semi-analytic calculations in this paper, will have to assume factorization between hard and soft sectors in transverse momentum dependent high-pTp_T hadron production. Such studies will of course suffer from the lack of a formal proof of factorization in traverse momentum dependent processes in pp-pp (and by extension pp-AA) collisions [103].

Acknowledgments:

A. M. is indebted to S. Pratt and J. Putschke for discussions and for suggesting hard scattering as a source of the elliptic anisotropy. Both authors are members of the JETSCAPE collaboration and have benefited from internal discussions. This work is supported in part by the U.S. Department of Energy under Grant No. DE-SC0013460. I. S. was funded as a part of the European Research Council project ERC-2018-ADG-835105 YoctoLHC, and as a part of the Center of Excellence in Quark Matter of the Academy of Finland (project 346325).

Quark Scattering

Since the gluon PDFs and FFs are dominant at small scales, we have found that gluon-gluon scattering is the only relevant partonic scattering at pTp_T range of the experimental results. For completeness, we present the results obtained for quark scattering at high-pTp_T. In Figure 16, we present the azimuthal angle distribution dσdϕ\frac{d\sigma}{d\phi} for qq→qqq q \to q q scattering in pp-pp and pp-PbPb collisions. While the unpolarized contribution displays a similar shape as the gluon scattering, the polarized contributions are dominated by lower order coefficients than the gluon scattering. For the quark the BM⊗CBM \otimes C terms are dominated by v1v_1, while the BM⊗BMBM \otimes BM terms are dominated by v2v_2.

Azimuthal angle distributions for unpolarized, $BM \otimes C$, and $BM \otimes BM$ contributions in p-p and p-Pb collisions

Figure 16. Azimuthal angle distribution dσdϕ\frac{d\sigma}{d\phi} for qq→qqqq \to qq scattering pion production in pp-pp at 5.02 TeV (top) and pp-PbPb at 8.16 TeV (bottom) at a fixed pT=100p_T=100 GeV. The panels from left to right represents the unpolarized, BM⊗CBM \otimes C and BM⊗BMBM \otimes BM contributions.

Computing the azimuthal anisotropy v2v_2, we find that the qq→qqq q \to q q scattering is subdominant compared to the gluon scattering as shown in Figure 17. However, the BM⊗BMBM \otimes BM term is increasing with pTp_T only surpassing the gluon contribution at pT≫200p_T \gg200 GeV.

Azimuthal anisotropy $v_2$ for pion production in p-Pb at 8.16 TeV

Figure 17. Azimuthal anisotropy v2v_2 for pion production in pp-PbPb at 8.16 TeV using a fixed ⟨k2⟩1/2=1\langle k^2\rangle^{1/2}=1 GeV and B⋅b=0.4B\cdot b=0.4 for the unpolarized distribution. The blue lines are the gg→gggg \to gg contribution, the green lines are the qq→qqqq \to qq contribution. Full lines represent the unpolarized contribution, dashed lines represent BM⊗CBM \otimes C terms and dashed lines circles represent BM⊗BMBM \otimes BM terms.

Kinematics of the cross section integration

While we use a different approach to obtain the matrix element, the kinematics we use are based on [65, 66, 104].

We reproduce the kinematics here.

We use the hadronic center of mass reference frame with Mandelstam variables (s,t,u)(s,t,u), where the hadronic momenta are given by

pA+=−pB−=s2,pA−=pB+=0,(48)p_A^+ = -p_B^- = \sqrt{\frac{s}{2}}, \qquad p_A^- = p_B^+ = 0 , \tag*{(48)}
pC=(pT2+pL2,pT,0,pL).(49)p_C = \left(\sqrt{p_T^2+p_L^2},p_T,0,p_L\right) . \tag*{(49)}

The partonic scattering a+b→c+da+b\to c+d is

pa=pA+(k⊥a22xas,k⊥acos⁡ϕa,k⊥asin⁡ϕa,−k⊥a22xas)(50)p_a=p_A+\left(\frac{k_{\perp a}^{2}}{2x_a\sqrt{s}},k_{\perp a}\cos\phi_a,k_{\perp a}\sin\phi_a,-\frac{k_{\perp a}^{2}}{2x_a\sqrt{s}}\right) \tag*{(50)}
pb=pB+(k⊥b22xbs,k⊥bcos⁡ϕb,k⊥bsin⁡ϕb,k⊥b22xbs)(51)p_b=p_B+\left(\frac{k_{\perp b}^{2}}{2x_b\sqrt{s}},k_{\perp b}\cos\phi_b,k_{\perp b}\sin\phi_b,\frac{k_{\perp b}^{2}}{2x_b\sqrt{s}}\right) \tag*{(51)}

The detected hadron CC has light-cone momentum fraction z=pC+pc+z=\frac{p_C^+}{p_c^+} and transverse momentum k⊥Ck_{\perp C} with respect to the parton cc. Using the delta function δ(k⊥C⋅p^c)\delta(k_{\perp C}\cdot\hat{p}_c) from Eq. (2), we find that the partonic momentum is given by

Ec=EC+EC2−k⊥C22z,(52)E_c=\frac{E_C+\sqrt{E_C^2-k_{\perp C}^2}}{2z}, \tag*{(52)}
pc=pC−k⊥CEC2−k⊥C2(pC−k⊥C).(53)\mathbf{p}_c=\frac{p_C-k_{\perp C}}{\sqrt{E_C^2-k_{\perp C}^2}}(p_C-k_{\perp C}). \tag*{(53)}

The integration phase space can be written as

∫dΩ=∫dxadxbdz2π2z3sd2k⊥ad2k⊥bd3k⊥Cδ(k⊥C⋅p^c)J(k⊥C)δ(s^+t^+u^).(54)\int d\Omega=\int\frac{dx_a dx_b dz}{2\pi^2 z^3s}d^2k_{\perp a}d^2k_{\perp b}d^3k_{\perp C}\delta(k_{\perp C}\cdot\hat{p}_c)J(k_{\perp C})\delta(\hat{s}+\hat{t}+\hat{u}). \tag*{(54)}

We define arbitrary transverse momentum k⊥Ck_{\perp C} in the hadronic center of mass frame as follows

k⊥C=k⊥C(sin⁡θk⊥Ccos⁡ϕk⊥C,sin⁡θk⊥Csin⁡ϕk⊥C,cos⁡θk⊥C).(55)k_{\perp C}=k_{\perp C}(\sin\theta_{k_{\perp C}}\cos\phi_{k_{\perp C}},\sin\theta_{k_{\perp C}}\sin\phi_{k_{\perp C}},\cos\theta_{k_{\perp C}}). \tag*{(55)}

The momentum of the outgoing parton cc is related to the pion’s momentum pπ=pT(1,0,sinh⁡y)p_\pi=p_T(1,0,\sinh y) as follows

pc=pπ−k⊥C.(56)p_c=p_\pi-k_{\perp C}. \tag*{(56)}

The transverse momentum delta function can be performed to define the azimuthal angle of k⊥Ck_{\perp C}

cos⁡ϕk⊥C=k⊥C−pTsinh⁡ycos⁡θk⊥CpTsin⁡θk⊥C,(57)\cos\phi_{k_{\perp C}}=\frac{k_{\perp C}-p_T\sinh y\cos\theta_{k_{\perp C}}}{p_T\sin\theta_{k_{\perp C}}}, \tag*{(57)}
sin⁡ϕk⊥C±=±1−cos⁡2ϕk⊥C,(58)\sin\phi_{k_{\perp C}}^{\pm}=\pm\sqrt{1-\cos^2\phi_{k_{\perp C}}}, \tag*{(58)}

leading to the Jacobian

∫d3k⊥Cδ(k⊥C⋅p^c)f(k⊥C)=∫k⊥Cdk⊥Cdθk⊥CpC2−k⊥C2pT∣sin⁡ϕk⊥C∣[f(k⊥C+)+f(k⊥C−)],(59)\int d^3k_{\perp C}\delta(k_{\perp C}\cdot\hat{p}_c)f(k_{\perp C})=\int k_{\perp C}dk_{\perp C}d\theta_{k_{\perp C}}\frac{\sqrt{p_C^2-k_{\perp C}^2}}{p_T|\sin\phi_{k_{\perp C}}|}\left[f(k_{\perp C}^{+})+f(k_{\perp C}^{-})\right], \tag*{(59)}

where one needs to sum over the two solutions sin⁡ϕk⊥C±\sin\phi_{k_{\perp C}}^{\pm}.

Using the definition of the hadronic momentum, we obtain the Jacobian J(k⊥C)J(k_{\perp C}) for the transformation from the parton cc to the hadron CC as follows

d3pcEc=1z2J(k⊥C)d3pCEC,(60)\frac{d^3p_c}{E_c}=\frac{1}{z^2}J(k_{\perp C})\frac{d^3p_C}{E_C}, \tag*{(60)}

obtaining

J(k⊥C)=(EC2+pC2−k⊥C2)24(pC2−k⊥C2).(61)J(k_{\perp C})=\frac{\left(E_C^2+\sqrt{p_C^2-k_{\perp C}^2}\right)^2}{4(p_C^2-k_{\perp C}^2)}. \tag*{(61)}

References

References

  1. [1]J. Adams et al., STAR, Phys. Rev. C 72, 014904 (2005), arXiv:nucl-ex/0409033.
  2. [2]S. S. Adler et al., PHENIX, Phys. Rev. Lett. 91, 182301 (2003), arXiv:nucl-ex/0305013.DOI
  3. [3]K. Aamodt et al., ALICE, Phys. Rev. Lett. 105, 252302 (2010), arXiv:1011.3914.
  4. [4]S. Chatrchyan et al., CMS, Phys. Rev. C 87, 014902 (2013), arXiv:1204.1409.DOI
  5. [5]G. Aad et al., ATLAS, Phys. Rev. C 86, 014907 (2012), arXiv:1203.3087.
  6. [6]P. F. Kolb, J. Sollfrank, and U. W. Heinz, Phys. Rev. C 62, 054909 (2000), hep-ph/0006129.
  7. [7]X.-N. Wang, Phys. Lett. B 595, 165 (2004), arXiv:nucl-th/0305010.DOI
  8. [8]A. Majumder, Phys. Rev. C75, 021901 (2007), arXiv:nucl-th/0608043.DOI
  9. [9]S. A. Bass et al., Phys.Rev. C79, 024901 (2009), arXiv:0808.0908.DOI
  10. [10]M. Aaboud et al., ATLAS, Eur. Phys. J. C 78, 997 (2018), arXiv:1808.03951.
  11. [11]G. Aad et al., ATLAS, Phys. Rev. Lett. 111, 152301 (2013), arXiv:1306.6469.
  12. [12]J. Adam et al., ALICE, Phys. Lett. B 753, 511 (2016), arXiv:1509.07334.
  13. [13]G. Aad et al., ATLAS, Phys. Lett. B 719, 220 (2013), arXiv:1208.1967.
  14. [14]S. Chatrchyan et al., CMS, Phys. Rev. C 84, 024906 (2011), arXiv:1102.1957.
  15. [15]B. Abelev et al., ALICE, JHEP 03, 013 (2014), arXiv:1311.0633.
  16. [16]G. Aad et al., ATLAS, JHEP 09, 050 (2015), arXiv:1504.04337.
  17. [17]S. Chatrchyan et al., CMS, Eur. Phys. J. C 72, 1945 (2012), arXiv:1202.2554.
  18. [18]K. Aamodt et al., ALICE, Phys. Lett. B 696, 30 (2011), arXiv:1012.1004.
  19. [19]J. Adams et al., STAR, Phys. Rev. Lett. 91, 172302 (2003), arXiv:nucl-ex/0305015.DOI
  20. [20]K. Adcox et al., PHENIX, Phys. Rev. Lett. 88, 022301 (2002), arXiv:nucl-ex/0109003.DOI
  21. [21]K. M. Burke et al., JET Collaboration, Phys. Rev. C 90, 014909 (2014), arXiv:1312.5003.DOI
  22. [22]A. Majumder and M. Van Leeuwen, Prog. Part. Nucl. Phys. 66, 41 (2011), arXiv:1002.2206.DOI
  23. [23]S. Cao, A. Majumder, R. Modarresi-Yazdi, I. Soudi, and Y. Tachibana, Int. J. Mod. Phys. E 33, 2430002 (2024), arXiv:2401.10026.arxiv.org/abs/2401.10026
  24. [24]J. Noronha-Hostler, B. Betz, J. Noronha, and M. Gyulassy, Phys. Rev. Lett. 116, 252301 (2016), arXiv:1602.03788.
  25. [25]A. Kumar, A. Majumder, and C. Shen, Phys. Rev. C 101, 034908 (2020), arXiv:1909.03178.DOI
  26. [26]Y. He et al., Phys. Rev. C 106, 044904 (2022), arXiv:221.08.08408.
  27. [27]R. Baier, Nucl. Phys. A715, 209 (2003), arXiv:hep-ph/0209038.DOI
  28. [28]A. Majumder, Phys. Rev. C87, 034905 (2013).
  29. [29]A. Kumar, A. Majumder, and J. H. Weber, Phys. Rev. D 106, 034505 (2022), arXiv:2010.14463.arxiv.org/abs/2010.14463
  30. [30]A. Majumder, Phys. Rev. C80, 031902 (2009), arXiv:0810.4967.DOI
  31. [31]H. Song, S. A. Bass, U. Heinz, T. Hirano, and C. Shen, Phys. Rev. Lett. 106, 192301 (2011), arXiv:1011.2783, [Erratum: Phys. Rev. Lett.109,139904(2012)].DOI
  32. [32]C. Shen et al., Comput. Phys. Commun. 199, 61 (2016), arXiv:1409.8164.arxiv.org/abs/1409.8164
  33. [33]J. E. Bernhard, J. S. Moreland, and S. A. Bass, Nature Phys. 15, 1113 (2019).
  34. [34]J. S. Moreland, J. E. Bernhard, and S. A. Bass, Phys. Rev. C 101, 024911 (2020), arXiv:1808.02106.DOI
  35. [35]M. Aaboud et al., ATLAS, Eur. Phys. J. C 77, 428 (2017), arXiv:1705.04176.
  36. [36]B. B. Abelev et al., ALICE, Phys. Rev. C 90, 054901 (2014), arXiv:1406.2474.
  37. [37]V. Khachatryan et al., CMS, Phys. Rev. Lett. 115, 012301 (2015), arXiv:1502.05382.
  38. [38]A. Dumitru et al., Phys. Lett. B 697, 21 (2011), arXiv:1009.5295.DOI
  39. [39]A. Kovner and M. Lublinsky, Int. J. Mod. Phys. E 22, 1330001 (2013), arXiv:1211.1928.DOI
  40. [40]M. Gyulassy, P. Levai, I. Vitev, and T. S. Biro, Phys. Rev. D 90, 054025 (2014), arXiv:1405.7825.DOI
  41. [41]T. Lappi, B. Schenke, S. Schlichting, and R. Venugopalan, JHEP 01, 061 (2016), arXiv:1509.03499.DOI
  42. [42]T. Altinoluk, N. Armesto, G. Beuf, A. Kovner, and M. Lublinsky, Phys. Lett. B 751, 448 (2015), arXiv:1503.07126.arxiv.org/abs/1503.07126
  43. [43]A. Dumitru, T. Lappi, and V. Skokov, Phys. Rev. Lett. 115, 252301 (2015), arXiv:1508.04438.DOI
  44. [44]T. Altinoluk, N. Armesto, G. Beuf, A. Kovner, and M. Lublinsky, Phys. Rev. D 95, 034025 (2017), arXiv:1610.03020.DOI
  45. [45]Y. Hagiwara, Y. Hatta, B.-W. Xiao, and F. Yuan, Phys. Lett. B 771, 374 (2017), arXiv:1701.04254.DOI
  46. [46]T. Altinoluk, N. Armesto, A. Kovner, and M. Lublinsky, Eur. Phys. J. C 78, 702 (2018), arXiv:1805.07739.DOI
  47. [47]T. Altinoluk, N. Armesto, A. Kovner, M. Lublinsky, and V. V. Skokov, Eur. Phys. J. C 81, 583 (2021), arXiv:2012.01810.DOI
  48. [48]Y. Hatta, B.-W. Xiao, F. Yuan, and J. Zhou, Phys. Rev. Lett. 126, 142001 (2021), arXiv:2010.10774.DOI
  49. [49]M. Aaboud et al., ATLAS, Phys. Rev. C 96, 024908 (2017), arXiv:1609.06213.
  50. [50]G. Aad et al., ATLAS, Eur. Phys. J. C 80, 73 (2020),
  51. [51]V. Khachatryan et al., CMS, JHEP 04, 039 (2017), arXiv:1611.01664.
  52. [52]J. Adam et al., ALICE, Phys. Lett. B 749, 68 (2015), arXiv:1503.00681.
  53. [53]ATLAS, (2022), arXiv:2211.15257.arxiv.org/abs/2211.15257
  54. [54]J. C. Collins and D. E. Soper, Nucl. Phys. B194, 445 (1982).
  55. [55]D. Boer and P. J. Mulders, Phys. Rev. D 57, 5780 (1998), arXiv:hep-ph/9711485.DOI
  56. [56]J. C. Collins, Nucl. Phys. B 396, 161 (1993), arXiv:hep-ph/9208213.DOI
  57. [57]C. Bierlich et al., (2022), arXiv:2203.11601.arxiv.org/abs/2203.11601
  58. [58]J. C. Collins, D. E. Soper, and G. Sterman, Nucl. Phys. B261, 104 (1985).
  59. [59]D. Boer, S. J. Brodsky, P. J. Mulders, and C. Pisano, Phys. Rev. Lett. 106, 132001 (2011), arXiv:1011.4225.DOI
  60. [60]C. Pisano, D. Boer, S. J. Brodsky, M. G. A. Buffing, and P. J. Mulders, JHEP 10, 024 (2013), arXiv:1307.3417.DOI
  61. [61]W. J. den Dunnen, J. P. Lansberg, C. Pisano, and M. Schlegel, Phys. Rev. Lett. 112, 212001 (2014), arXiv:1401.7611.DOI
  62. [62]D. Boer and W. J. den Dunnen, Nucl. Phys. B 886, 421 (2014), arXiv:1404.6753.DOI
  63. [63]A. Bacchetta, M. Boglione, A. Henneman, and P. J. Mulders, Phys. Rev. Lett. 85, 712 (2000), arXiv:hep-ph/9912490.DOI
  64. [64]M. Anselmino, M. Boglione, U. D’Alessio, E. Leader, and F. Murgia, Phys. Rev. D 71, 014002 (2005), arXiv:hep-ph/0408356.DOI
  65. [65]M. Anselmino et al., Phys. Rev. D 73, 014020 (2006), arXiv:hep-ph/0509035.DOI
  66. [66]U. D’Alessio and F. Murgia, Phys. Rev. D 70, 074009 (2004), arXiv:hep-ph/0408092.
  67. [67]P. J. Mulders and J. Rodrigues, Phys. Rev. D 63, 094021 (2001), arXiv:hep-ph/0009343.DOI
  68. [68]R. D. Field, Redwood City, USA: Addison-Wesley (1989) 366 p. (Frontiers in physics, 77).
  69. [69]A. Bermudez Martinez et al., Phys. Rev. D 99, 074008 (2019), arXiv:1804.11152.DOI
  70. [70]A. Bermudez Martinez, F. Hautmann, and M. L. Mangano, Phys. Lett. B 822, 136700 (2021), arXiv:2107.01224.DOI
  71. [71]R. G. Arnold et al., Phys. Rev. Lett. 52, 727 (1984).
  72. [72]J. W. Cronin et al., E100, Phys. Rev. D 11, 3105 (1975).
  73. [73]J. P. Gamber and G. R. Goldstein, Phys. Lett. B 650, 362 (2007), arXiv:hep-ph/0506127.
  74. [74]X. Guo, Phys. Rev. D 58, 036001 (1998), arXiv:hep-ph/9711453.DOI
  75. [75]R. J. Fries, Phys. Rev. D 68, 074013 (2003), arXiv:hep-ph/0209275.DOI
  76. [76]A. Majumder and B. Muller, Phys. Rev. C77, 054903 (2008), arXiv:0705.1147.DOI
  77. [77]Y. V. Kovchegov, Phys. Rev. D 60, 034008 (1999), arXiv:hep-ph/9901281.DOI
  78. [78]A. H. Mueller, Nucl. Phys. B 335, 115 (1990).
  79. [79]L. V. Gribov, E. M. Levin, and M. G. Ryskin, Phys. Rept. 100, 1 (1983).
  80. [80]P. Arnold and W. Xiao, Phys. Rev. D78, 125008 (2008), arXiv:0810.1026.DOI
  81. [81]I. Soudi and A. Majumder, Phys. Lett. B 859, 139105 (2024), arXiv:2308.14702.arxiv.org/abs/2308.14702
  82. [82]J. Collins, Foundations of Perturbative QCD Cambridge Monographs on Particle Physics, Nuclear Physics and Cosmology (Cambridge University Press, 2011).
  83. [83]B. A. Kniehl, G. Kramer, and B. Potter, Nucl. Phys. B 582, 514 (2000), hep-ph/0010289.
  84. [84]S. M. Abyat and T. C. Rogers, Phys. Rev. D 83, 114042 (2011), arXiv:1101.5057.
  85. [85]P. C. Barry et al., Jefferson Lab Angular Momentum (JAM), Phys. Rev. D 108, L091504 (2023), arXiv:2302.01192.DOI
  86. [86]J. Soffer, Phys. Rev. Lett. 74, 1292 (1995), arXiv:hep-ph/9409254.DOI
  87. [87]H. Mäntysaari, B. Schenke, C. Shen, and P. Tribedy, Phys. Lett. B 772, 681 (2017), arXiv:1705.03177.DOI
  88. [88]C. M. Butter et al., The Underlying Event, in HERA and the LHC: A Workshop on the Implications of HERA for LHC Physics: CERN - DESY Workshop 2004/2005 (Midterm Meeting, CERN, 11-13 October 2004; Final Meeting, DESY, 17-21 January 2005), pp. 192–217, Geneva, 2005, CERN.
  89. [89]T. Sjostrand and M. van Zijl, Phys. Rev. D 36, 2019 (1987).
  90. [90]B. Andersson, G. Gustafson, G. Ingelman, and T. Sjostrand, Phys. Rept. 97, 31 (1983).
  91. [91]N. Borghini, P. M. Dinh, and J.-Y. Ollitrault, Phys. Rev. C 64, 054901 (2001), arXiv:nucl-th/0105040.DOI
  92. [92]S. Acharya et al., ALICE, Phys. Rev. C 101, 044907 (2020), arXiv:1910.07678.
  93. [93]G. Aad et al., ATLAS, Phys. Lett. B 763, 313 (2016), arXiv:1605.06436.
  94. [94]J. Adam et al., ALICE, Phys. Lett. B 760, 720 (2016), arXiv:1601.03658.
  95. [95]K. Kovarik et al., Phys. Rev. D 93, 085037 (2016), arXiv:1509.00792.DOI
  96. [96]U. A. Acharya et al., PHENIX, Phys. Rev. C 105, 064902 (2022), arXiv:2111.05756.DOI
  97. [97]J. Brewer, A. Mazeliauskas, and W. van der Schee, Opportunities of OO and pO collisions at the LHC, in Opportunities of OO and pO collisions at the LHC, 2021, arXiv:2103.01939.arxiv.org/abs/2103.01939
  98. [98]D. Boer, P. J. Mulders, and F. Pijlman, Nucl. Phys. B 667, 201 (2003), arXiv:hep-ph/0303034.DOI
  99. [99]Jetscape et al., JETSCAPE, PoS HardProbes2023, 128 (2024), arXiv:2308.02560.
  100. [100]D. Boer, T. Van Daal, P. J. Mulders, and E. Petreska, JHEP 07, 140 (2018), arXiv:1805.05219.DOI
  101. [101]R. J. Fries, A. Schäfer, E. Stein, and B. Muller, Nucl. Phys. B 582, 537 (2000), arXiv:hep-ph/0002074.DOI
  102. [102]X.-N. Wang and X.-F. Guo, Nucl. Phys. A696, 788 (2001), arXiv:hep-ph/0102230.DOI
  103. [103]T. C. Rogers and P. J. Mulders, Phys. Rev. D 81, 094006 (2010), arXiv:1001.2977.DOI
  104. [104]A. P. Contogouris, R. Gaskell, and S. Papadopoulos, Phys. Rev. D 17, 2314 (1978).

Paper details

vlnc:W_ddtx5dr8
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents