Extraction of partonic transverse momentum distributions from semi-inclusive deep-inelastic scattering, Drell-Yan and Z-boson production

Journal of High Energy Physics, Jun 2017

We present an extraction of unpolarized partonic transverse momentum distributions (TMDs) from a simultaneous fit of available data measured in semi-inclusive deep-inelastic scattering, Drell-Yan and Z boson production. To connect data at different scales, we use TMD evolution at next-to-leading logarithmic accuracy. The analysis is restricted to the low-transverse-momentum region, with no matching to fixed-order calculations at high transverse momentum. We introduce specific choices to deal with TMD evolution at low scales, of the order of 1 GeV2. This could be considered as a first attempt at a global fit of TMDs.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:


Extraction of partonic transverse momentum distributions from semi-inclusive deep-inelastic scattering, Drell-Yan and Z-boson production

Revised: May Extraction of partonic transverse momentum distributions from semi-inclusive deep-inelastic scattering, Drell-Yan and Z-boson production Alessandro Bacchetta 0 1 2 Filippo Delcarro 0 1 2 Cristian Pisano 0 1 2 Marco Radici 0 1 2 Andrea Signori 0 1 3 0 12000 Jefferson Avenue , Newport News, VA 23606 , U.S.A 1 via Bassi 6, I-27100 Pavia , Italy 2 INFN Sezione di Pavia 3 Theory Center, Thomas Jefferson National Accelerator Facility We present an extraction of unpolarized partonic transverse momentum distributions (TMDs) from a simultaneous fit of available data measured in semi-inclusive deep-inelastic scattering, Drell-Yan and Z boson production. To connect data at different scales, we use TMD evolution at next-to-leading logarithmic accuracy. The analysis is restricted to the low-transverse-momentum region, with no matching to fixed-order calculations at high transverse momentum. We introduce specific choices to deal with TMD evolution at low scales, of the order of 1 GeV2. This could be considered as a first attempt at a global fit of TMDs. aDipartimento di Fisica; Università di Pavia - 3 Data analysis 3.1 Semi-inclusive DIS data Semi-inclusive DIS Drell-Yan and Z production TMDs and their evolution 3.1.1 3.1.2 Hermes data Compass data Low-energy Drell-Yan data Z-boson production data The replica method 2.1 2.2 2.3 3.2 3.3 3.4 4.1 4.2 4.3 4 Results 5 1 Conclusions Introduction 1 Introduction 2 Formalism Agreement between data and theory Transverse momentum dependence at 1 GeV Stability of our results Parton distribution functions describe the internal structure of the nucleon in terms of its elementary constituents (quarks and gluons). They cannot be easily computed from first principles, because they require the ability to carry out Quantum Chromodynamics (QCD) calculations in its nonperturbative regime. Many experimental observables in hard scattering experiments involving hadrons are related to parton distribution functions (PDFs) and fragmentation functions (FFs), in a way that is specified by factorization theorems (see, e.g., refs. [ 1, 2 ]). These theorems also elucidate the universality properties of PDFs and FFs (i.e., the fact that they are the same in different processes) and their evolution equations (i.e., how they get modified by the change in the hard scale of the process). Availability of measurements of different processes in different experiments makes it possible to test factorization theorems and extract PDFs and FFs through so-called global fits. On the other side, the knowledge of PDFs and FFs allows us to make predictions for other hard hadronic processes. These general statements apply equally well to standard collinear PDFs and FFs and to transverse-momentum-dependent parton distribution functions (TMD PDFs) and fragmentation functions (TMD FFs). Collinear PDFs describe the distribution of partons integrated over all components of partonic momentum except the one collinear to the parent hadron; hence, collinear PDFs are functions of the parton longitudinal momentum fraction x. TMD PDFs (or TMDs for short) include also the dependence on the transverse – 1 – ? momentum k . They can be interpreted as three-dimensional generalizations of collinear PDFs. Similar arguments apply to collinear FFs and TMD FFs [3]. There are several differences between collinear and TMD distributions. From the formal point of view, factorization theorems for the two types of functions are different, implying also different universality properties and evolution equations [4]. From the experimental point of view, observables related to TMDs require the measurement of some transverse momentum component much smaller than the hard scale of the process [5, 6]. For instance, Deep-Inelastic Scattering (DIS) is characterized by a hard scale represented by the 4-momentum squared of the virtual photon ( Q2). In inclusive DIS this is the only scale If Ph2T of the process, and only collinear PDFs and FFs can be accessed. In semi-inclusive DIS (SIDIS) also the transverse momentum of the outgoing hadron (PhT ) can be measured [7, 8]. Q2, TMD factorization can be applied and the process is sensitive to TMDs [ 2 ]. If polarization is taken into account, several TMDs can be introduced [7, 9–12]. At? tempts to extract some of them have already been presented in the past [13–21]. In this work, we focus on the simplest ones, i.e., the unpolarized TMD PDF f1q(x; k?2) and the unpolarized TMD FF D1q!h(z; P 2 ), where z is the fractional energy carried by the detected hadron h, k? is the transverse momentum of the parton with respect to the parent hadron, and P ? is the transverse momentum of the produced hadron with respect to the parent parton. Despite their simplicity, the phenomenology of these unpolarized TMDs presents several challenges [22]: the choice of a functional form for the nonperturbative components of TMDs, the inclusion of a possible dependence on partonic flavor [23], the implementation of TMD evolution [4, 24], the matching to fixed-order calculations in collinear factorization [25]. We take into consideration three kinds of processes: SIDIS, Drell-Yan processes (DY) and the production of Z bosons. To date, they represent all possible processes where experimental information is available for unpolarized TMD extractions. The only important process currently missing is electron-positron annihilation, which is particularly important for the determination of TMD FFs [24]. This work can therefore be considered as the first attempt at a global fit of TMDs. The paper is organized as follows. In section 2, the general formalism for TMDs in SIDIS, DY processes, and Z production is briefly outlined, including a description of the assumptions and approximations in the phenomenological implementation of TMD evolution equations. In section 3, the criteria for selecting the data analyzed in the fit are summarized and commented. In section 4, the results of our global fit are presented and discussed. In section 5, we summarize the results and present an outlook for future improvements. 2 2.1 Formalism Semi-inclusive DIS In one-particle SIDIS, a lepton ` with momentum l scatters off a hadron target N with mass M and momentum P . In the final state, the scattered lepton momentum l0 is measured together with one hadron h with mass Mh and momentum Ph. The corresponding reaction – 2 – formula is invariants where `(l) + N (P ) ! `(l0) + h(Ph) + X : The space-like momentum transfer is q = l l0, with Q2 = q2. We introduce the usual mhN (x; z; jPhT j; Q2) = d Nh =(dxdzdjPhT jdQ2) d DIS=(dxdQ2) ; where d Nh is the differential cross section for the SIDIS process and d DIS is the corresponding inclusive one, and where PhT is the component of Ph transverse to q (we follow here the notation suggested in ref. [26]). In the single-photon-exchange approximation, the multiplicities can be written as ratios of structure functions (see ref. [8] for details): mhN (x; z; jPhT j; Q2) = 2 jPhT jFUU;T (x; z; Ph2T ; Q2) + 2 "jPhT jFUU;L(x; z; Ph2T ; Q2) FT (x; Q2) + "FL(x; Q2) x = Q2 2 P q ; y = P q P l ; z = P Ph P q ; = 2M x Q : The available data refer to SIDIS hadron multiplicities, namely to the differential number of hadrons produced per corresponding inclusive DIS event. In terms of cross sections, we define the multiplicities as HJEP06(217)8 (2.1) (2.2) (2.3) ; (2.4) (2.5) (2.6) " = 1 y 4 Q2. In these limits, the structure function FUU;L of eq. (2.4) can be neglected [27]. The structure function FL in the denominator contains contributions involving powers of the strong coupling constant S at an order that goes beyond the level reached in this analysis; hence, it will be consistently neglected (for measurements and estimates of the FL structure function see, e.g., refs. [28, 29] and references therein). To express the structure functions in terms of TMD PDFs and FFs, we rely on the factorized formula for SIDIS [ 2, 30–37 ] (see figure 1 for a graphical representation of the involved transverse momenta): FUU;T (x; z; Ph2T ; Q2) = HUU;T (Q2) a X a x Z ? d2k d2P f ? 1a x; k?2; Q2 D1a~ h z; P?2 ; Q2 (2) zk? PhT +P ? + YUU;T Q2; Ph2T ? (not measured). The struck parton fragments into a hadron, ? (not measured). The total measured transversemomentum of the final hadron is PhT . When Q2 is very large, the longitudinal components are all much larger than the transverse components. In this regime, PhT zk? + P?. Here, HUU;T is the hard scattering part; f1a(x; k2 ; Q2) is the TMD PDF of unpolarized ? partons with flavor a in an unpolarized proton, carrying longitudinal momentum fraction x and transverse momentum k?. The D1a~ h(z; P 2 ; Q2) is the TMD FF describing the fragmentation of an unpolarized parton with flavor a into an unpolarized hadron h carrying ? longitudinal momentum fraction z and transverse momentum P ? (see figure 1). TMDs generally depend on two energy scales [ 2 ], which enter via the renormalization of ultraviolet and rapidity divergencies. In this work we choose them to be equal and set them to Q2. The term YUU;T is introduced to ensure a matching to the perturbative fixed-order calculations at higher transverse momenta. In our analysis, we neglect any correction of the order of M 2=Q2 or higher to eq. (2.6). At large Q2 this is well justified. However, fixed-target DIS experiments typically collect a large amount of data at relatively low Q2 values, where these assumptions should be all tested in future studies. The reliability of the theoretical description of SIDIS at low Q2 has been recently discussed in refs. [39, 40]. Eq. (2.6) can be expanded in powers of S. In the present analysis, we will consider only the terms at order S0. In this case HUU;T (Q2) a corrections include large logarithms L e2a and YUU;T log z2Q2=Ph2T , so that SL 0. However, perturbative 1. In the present analysis, we will take into account all leading and Next-to-Leading Logarithms (NLL).1 In these approximations ( S0 and NLL), only the first term in eq. (2.6) is relevant (often in the literature this has been called W term). We expect this term to provide a a detailed treatment of the matching to the high Ph2T good description of the structure function only in the region where Ph2T Q2 region to future investigations Q2. We leave (see, e.g., ref. [25]). 1We remark that formulas at NNLL are available in the literature [41]. – 4 – transform of the part of eq. (2.6) involving TMDs. The structure function thus reduces to FUU;T (x; z; Ph2T ; Q2) 2 d T T J0 T jPhT j=z f~1a x; T2 ; Q2 D~ 1a~ h z; T2 ; Q2 ; where we introduced the Fourier transforms of the TMD PDF and FF according to f~1a x; T2 ; Q2 = D~ 1a~ h z; T2 ; Q2 = Z 1 0 djk?jjk?jJ0 T jk?j f1a x; k?2; Q2 ; Z 1 djzP2?j jP?jJ0 T jP?j=z D1a~ h z; P?2 ; Q2 : 0 2.2 Drell-Yan and Z production In a Drell-Yan process, two hadrons A and B with momenta PA and PB collide at a centerof-mass energy squared s = (PA + PB)2 and produce a virtual photon or a Z boson plus hadrons. The boson decays into a lepton-antilepton pair. The reaction formula is A(PA) + B(PB) ! [ =Z + X !]`+(l) + ` (l0) + X: The invariant mass of the virtual photon is Q2 = q2 with q = l + l0. We introduce the rapidity of the virtual photon/Z boson (2.7) (2.8) (2.9) (2.10) (2.11) (2.12) (2.13) (2.14) where the z direction is defined along the momentum of hadron A (see figure 2). The cross section can be written in terms of structure functions [42, 43]. For our purposes, we need the unpolarized cross section integrated over d and over the azimuthal angle of the virtual photon, d dQ2 dqT2 d = 0 ;Z FU1U + 21 FU2U : The elementary cross sections are 0 = 4 2 e2m ; 3Q2s 0Z = s sin2 2 em W cos2 W BR(Z ! `+` ) (Q2 MZ2 ); where W is Weinberg’s angle, MZ is the mass of the Z boson, and BR(Z ! `+` ) is the branching ratio for the Z boson decay in two leptons. We adopted the narrow-width approximation, i.e., we neglect contributions for Q2 6= MZ2 . We used the values sin2 MZ = 91:18 GeV, and BR(Z ! `+` ) = 3:366 [44]. Similarly to the SIDIS case, in the W = 0:2313, kinematic limit qT2 Q2 the structure function FU2U can be neglected (for measurement and estimates of this structure function see, e.g., ref. [45] and references therein). The longitudinal momentum fractions of the annihilating quarks can be written in terms of rapidity in the following way 1 2 = log q0 + qz q0 qz ; xA = p e ; Q s xB = p e : Q s – 5 – from two hadrons collide. They have transverse momenta k?A and k?B (not measured). They produce a virtual photon with (measured) transverse momentum qT = k?A + k?B with respect to the hadron collision axis. Some experiments use the variable xF , which is connected to the other variables by the following relations p s xF Q 2 = sinh 1 ; xA = r Q2 s x 2 4 + F + xF 2 ; xB = xA xF : (2.15) The structure function FU1U can be written as (see figure 2 for a graphical representation of the involved transverse momenta) FU1U (xA; xB; qT2 ; Q2) = HU1aU (Q2) X a Z + YU1U Q2; qT2 + O M 2=Q2 : d2k?Ad2k?Bf1a xA; k?2A; Q2 f1a xB; k?2B; Q2 (2) k?A qT +k?B PB q quark kA qT nucleon 2We remind the reader that the value of weak isospin I3 is equal to +1=2 for u, c, t and 1=2 for d, s, b. – 6 – As in the SIDIS case, in our analysis we neglect the YUU term and we consider the hard coefficients only up to leading order in the couplings, i.e., where2 HUU; (Q2) 1a e 2 a ; Nc HUU;Z (Q2) 1a Va2 + A2a ; Nc Va = I3a 2ea sin2 W ; Aa = I3a : The structure function can be conveniently expressed as a Fourier transform of the right-hand side of eq. (2.16) as FU1U (xA; xB; qT2 ; Q2) 2 X a HUU 1a Z 1 0 d T T J0 T jqT j f ~1a xA; T2 ; Q2 f~1a xB; T2 ; Q2 : (2.16) (2.17) (2.18) (2.19) fe1a(x; T2 ; Q2) = Ca=i f1i (x; ; b2 ) De1a!h(z; T2 ; Q2) = C^a=i D1i!h (z; ; b2 ) We choose the scale b to be where E is the Euler constant and b = 2e E X i=q;q;g X i=q;q;g eS( b2;Q2) eS( b2;Q2) Q2 2 b Q2 2 b ( T ; min; max) = max K( ; b) K( ; b) Q2 gK( T ) Q2 0 Q2 gK( T ) Q2 0 fe1NP(x; T2 ) ; a De1aN!Ph(z; T2 ) : ; 1 1 e T4 = m4ax !1=4 e T4 = m4in : (2.20) (2.21) (2.22) (2.23) Evolution equations quantitatively describe the connection between different values for the energy scales. In the following we will set their initial values to as Q2, so that only Q2 and b2 need to be specified in a TMD distribution. Following the formalism of refs. [ 2, 34 ], the unpolarized TMD distribution and fragmentation functions in configuration space for a parton with flavor a at a certain scale Q2 can be written as b2 and their final values This variable replaces the simple dependence upon T in the perturbative parts of the TMD definitions of eqs. (2.20), (2.21). In fact, at large T these parts are no longer reliable. Therefore, the is chosen to saturate on the maximum value max, as suggested by the CSS formalism [ 2, 34 ]. On the other hand, at small T the TMD formalism is not valid and should be matched to the fixed-order collinear calculations. The way the matching is implemented is not unique. In any case, the TMD contribution can be arbitrarily modified at small T . In our approach, we choose to saturate at the minimum value min / 1=Q. With the appropriate choices, for T = 0 the Sudakov exponent vanishes, as it should [46, 47]. Our choice partially corresponds to modifying the resummed logarithms as in ref. [48] and to other similar modifications proposed in the literature [25, 49]. One advantage of these kind of prescriptions is that by integrating over the impact parameter T , the collinear expression for the cross section in terms of collinear PDFs is recovered, at least at leading order [25]. The choice of the functional form in eq. (2.23) is arbitrary. In the original CSS article [31], a functional form based on a square-root saturation was adopted. With our choice, rapidly saturates to its limiting values min and max, leaving a reasonably wide region where it is equal to T . This is particularly critical for low values of Q, where min and max are very close. We remind the reader that there are different schemes available to deal with the high- T region, such as the so-called “complex- prescription” [50] or an extrapolation of the perturbative small- T calculation to the large T region based on dynamical power corrections [51]. – 7 – The values of max and min could be regarded as boundaries between the regions of “small” and “large” distances. They are not completely arbitrary, but there is also no way to determine them precisely. We choose to fix them to the values max = 2e E GeV 1 1:123 GeV 1 ; min = 2e E =Q : (2.24) The motivations are the following: with the above choices, the scale b is constrained between 1 GeV and Q, so that the collinear PDFs are never computed at a scale lower than 1 GeV and the lower limit of the integrals contained in the definition of the perturbative Sudakov factor (see eq. (2.30)) can never become larger than the upper limit; at Q = Q0 = 1 GeV, max = min and there are no evolution effects; the TMD is simply given by the corresponding collinear function multiplied by a nonperturbative contribution depending on k? (plus possible corrections of order S from the Wilson coefficients). At NLL accuracy, for our choice of scales K( ; b) = 0. Similarly, the C and C^ are perturbatively calculable Wilson coefficients for the TMD distribution and fragmentation functions, respectively. They are convoluted with the corresponding collinear functions according to Ca=i f1i (x; ; b2) = C^a=i D1i!h (z; ; b2) = Z 1 du x u Ca=i u x Z 1 du C^a=i u z z u ; ; S ; ; S 2 b 2 b f1i(u; b2) ; D1i!h(u; b2) : In the present analysis, we consider only the leading-order term in the S expansion for C and C^, i.e., Ca=i u x ; ; S 2 b reduces to ai (1 x=u); C^a=i u z ; ; S 2 b ai (1 z=u): (2.27) As a consequence of the choices we made, the expression for the evolved TMD functions fe1a(x; T2 ; Q2) = f1a(x; b2) eS( b2;Q2) egK( T ) ln(Q2=Q20) fe1NP(x; T2 ) ; a De1a!h(z; T2 ; Q2) = D1a!h(z; b2) eS( b2;Q2) egK( T ) ln(Q2=Q20) De1aN!Ph(z; T2 ) : The Sudakov exponent S can be written as S( b2; Q2) = A S( 2) ln + B S( 2) ; Q2 2 where the functions A and B have a perturbative expansions of the form A S( 2) = 1 X Ak k=1 B S( 2) = 1 X Bk k=1 S k : Z Q2 d 2 2 b S 2 k ; – 8 – (2.25) (2.26) (2.28) (2.29) (2.30) (2.31) To NLL accuracy, we need the following terms [ 31, 52 ] A1 = CF ; A2 = CF CA 1 2 67 18 2 6 5 9 Nf ; 3 2 B1 = CF : (2.32) We use the approximate analytic expression for S at NLO with the QCD = 340 MeV, 296 MeV, 214 MeV for three, four, five flavors, respectively, corresponding to a value of S(MZ ) = 0:117. We fix the flavor thresholds at mc = 1:5 GeV and mb = 4:7 GeV. The integration of the Sudakov exponent in eq. (2.30) can be done analytically (for the complete expressions see, e.g., refs. [36, 53, 54]). Following refs. [55–57], for the nonperturbative Sudakov factor we make the traditional HJEP06(217)8 choice at low Q2. with g2 a free parameter. Recently, several alternative forms have been proposed [58, 59]. Also, recent theoretical studies aimed at calculating this term using nonperturbative methods [60]. All these choices should be tested in future studies. In ref. [61], a good agreement with data was achieved even without this term, but this is not possible when including data In this analysis, for the collinear PDFs f1a we adopt the GJR08FFnloE set [62] through the LHAPDF library [63], and for the collinear fragmentation functions the DSS14 NLO set for pions [64] and the DSS07 NLO set for kaons [65].3 We will comment on the use of other PDF sets in section 4.3. We parametrize the intrinsic nonperturbative parts of the TMDs in the following ways gK ( T ) = After performing the anti-Fourier transform, the f1NP and D1NP in momentum space correspond to (2.34) : (2.35) (2.36) (2.37) f1NP(x; k?2) = a D1aN!Ph(z; P 2 ) = ? 1 1 1 + k g1a + g3a!h + 2 ? g1a 1 k 2 ? 2 e g1a ; F =z2 g42a!h P 2 P 2 e g3a?!h + F z?2 e g4a?!h : P 2 The TMD PDF at the starting scale is therefore a normalized sum of a Gaussian with variance g1 and the same Gaussian weighted by a factor k2 . The TMD FF at the starting scale is a normalized sum of a Gaussian with variance g3 and a second Gaussian with variance g4 weighted by a factor F P?2 =z2. The choice of this particular functional forms is motivated by model calculations: the weighted Gaussian in the TMD PDF could arise ? 3After the completion of our analysis, a new set of kaon fragmentation function was presented in ref. [66]. – 9 – from the presence of components of the quark wave function with angular momentum L = 1 [67–71]. Similar features occur in models of fragmentation functions [38, 67, 72]. The Gaussian width of the TMD distributions may depend on the parton flavor a [23, 38, 73]. In the present analysis, however, we assume they are flavor independent. The justification for this choice is that most of the data we are considering are not sufficiently sensitive to flavor differences, leading to unclear results. We will devote attention to this issue in further studies. Finally, we assume that the Gaussian width of the TMD depends on the fractional longitudinal momentum x according to HJEP06(217)8 (2.38) (2.39) (2.40) g1(x) = N1 (1 (1 x) x x^) x^ ; g3;4(z) = N3;4 (z^ + ) (1 (z + ) (1 z) z^) ; where ; ; and N1 functions we have g1(x^) with x^ = 0:1, are free parameters. Similarly, for fragmentation where ; ; ; and N3;4 g3;4(z^) with z^ = 0:5 are free parameters. can be computed analytically: The average transverse momentum squared for the distributions in eq. (2.36) and (2.37) k ?2 (x) = g1(x) + 2 g12(x) 1 + g1(x) ; P?2 (z) = g32(z) + 2 F g43(z) g3(z) + F g42(z) : 3 Data analysis The main goals of our work are to extract information about intrinsic transverse momenta, to study the evolution of TMD parton distributions and fragmentation functions over a large enough range of energy, and to test their universality among different processes. To achieve this we included measurements taken from SIDIS, Drell-Yan and Z boson production from different experimental collaborations at different energy scales. In this section we describe the data sets considered for each process and the applied kinematic cuts. Table 1 refers to the data sets for SIDIS off proton target (Hermes experiment) and presents their kinematic ranges. The same holds for table 2, table 3, table 4 for SIDIS off deuteron (Hermes and Compass experiments), Drell-Yan events at low energy and Z boson production respectively. If not specified otherwise, the theoretical formulas are computed at the average values of the kinematic variables in each bin. 3.1 Semi-inclusive DIS data The SIDIS data are taken from Hermes [74] and Compass [75] experiments. Both data sets have already been analyzed in previous works, e.g., refs. [23, 76], however they have never been fitted together, including also the contributions deriving from TMD evolution. The application of the TMD formalism to SIDIS depends on the capability of identifying the current fragmentation region. This task has been recently discussed in ref. [39], where the authors point out a possible overlap among different fragmentation regions when the Cuts Points Max. Q2 x range Hermes Hermes Hermes Hermes p ! + p ! Cuts Cuts Points p s Q range Kin. var. [79] 45 19.4 GeV 4-9 GeV =0.40 [79] 45 23.8 GeV 4-9 GeV =0.21 qT < 0:2 Q + 0:5 GeV [79] 78 27.4 GeV =0.03 E605 [80] 35 38.8 GeV xF = 0:1 5-9, 11-14 GeV 7-9, 10.5-11.5 GeV with different center-of-mass energies. hard scale Q is sufficiently low. In this paper we do not tackle this problem and we leave it to future studies. As described in tables 1 and 2, we identify the current fragmentation region operating a cut on z only, namely 0:2 < z < 0:74. HJEP06(217)8 Reference Cuts Points p s Normalization [81] 31 Another requirement for the applicability of TMD factorization is the presence of two separate scales in the process. In SIDIS, those are the Q2 and Ph2T , which should satisfy the condition Ph2T Q2, or more precisely Ph2T =z2 Q2. We implement this condition by imposing PhT < min[0:2 Q; 0:7 Qz] + 0:5 GeV. With this choice, Ph2T is always smaller than Q2=3, but in a few bins (at low Q2 and z) Ph2T =z2 may become larger than Q2. The applicability of TMD factorization in this case could be questioned. However, as we will explain further in section 4.3, we can obtain a fit that can describe a wide region of PhT and can also perform very well in a restricted region, where TMD factorization certainly holds. All these choices are summarized in tables 1 and 2. 3.1.1 Hermes data Hermes hadron multiplicities are measured in a fixed target experiment, colliding a 27:6 GeV lepton beam on a hydrogen (p) or deuterium (D) gas target, for a total of 2688 points. These are grouped in bins of (x; z; Q2; PhT ) with the average values of (x; Q2) ranging from about (0:04; 1:25 GeV2) to (0:4; 9:2 GeV2). The collinear energy fraction z in eq. (2.2) ranges in 0:1 z 0:9. The transverse momentum of the detected hadron satisfies 0:1 GeV jPhT j 1:3 GeV. The peculiarity of Hermes SIDIS experiment lies in the ability of its detector to distinguish between pions and kaons in the final state, in addition to determining their momenta and charges. We consider eight different combinations of target (p; D) and detected charged hadron ( ; K ). The Hermes collaboration published two distinct sets, characterized by the inclusion or subtraction of the vector meson contribution. In our work we considered only the data set where this contribution has been subtracted. 3.1.2 Compass data The Compass collaboration extracted multiplicities for charge-separated but unidentified hadrons produced in SIDIS off a deuteron (6LiD) target [75]. The number of data points is an order of magnitude higher compared to the Hermes experiment. The data are organized in multidimensional bins of (x; z; Q2; PhT ), they cover a range in (x; Q2) from about (0:005; 1:11 GeV2) to (0:09; 7:57 GeV2) and the interval 0:2 z 0:8. The multiplicities published by Compass are affected by normalization errors (see the erratum to ref. [75]). In order to avoid this issue, we divide the data in each bin in (x; z; Q2) by the data point with the lowest Ph2T in the bin. As a result, we define the normalized multiplicity as mnorm(x; z; Ph2T ; Q2) = mhN (x; z; Ph2T ; Q2) mhN (x; z; min[Ph2T ]; Q2) ; (3.1) where the multiplicity mhN is defined in eq. (2.3). When fitting normalized multiplicities, the first data point of each bin is considered as a fixed constraint and excluded from the degrees of freedom. We analyze Drell-Yan events collected by fixed-target experiments at low-energy. These data sets have been considered also in previous works, e.g., in refs. [56, 57, 77, 78]. We used data sets from the E288 experiment [79], which measured the invariant dimuon cross section Ed3 =dq3 for the production of + pairs from the collision of a proton beam with a fixed target, either composed of Cu or Pt. The measurements were performed using proton incident energies of 200, 300 and 400 GeV, producing three different data sets. Their respective center of mass energies are p s = 19:4; 23:8; 27:4 GeV. We also included the set p of measurements Ed3 =dq3 from E605 [80], extracted from the collision of a proton beam with an energy of 800 GeV ( s = 38:8 GeV) on a copper fixed target. The explored Q values are higher compared to the SIDIS case, as can be seen in table 3. E288 provides data at fixed rapidity, whereas E605 provides data at fixed xF = 0:1. We can apply TMD factorization if qT2 Q2, where qT is the transverse momentum of the intermediate electroweak boson, reconstructed from the kinematics of the final state leptons. We choose qT < 0:2 Q + 0:5 GeV. As suggested in ref. [79], we consider the target nuclei as an incoherent ensemble composed 40% by protons and 60% by neutrons. As we already observed, results from E288 and E605 experiments are reported as Edd33q ; this variable is related to the differential cross section of eq. (2.12) in the following way: HJEP06(217)8 Ed3 d3q = d 3 d d qT dqT ) d 2 d d(qT2 ) ; Ed3 d3q = where is the polar angle of qT and the third term is the average over . Therefore, the invariant dimuon cross section can be obtained from eq. (2.12) integrating over Q2 and adding a factor 1= to the result Numerically we checked that integrating in Q2 only the prefactor q (see eq. (2.13)) introduces only a negligible error in the theoretical estimates. We also assume that em does not change within the experimental bin. Therefore, for Drell-Yan we obtain where Qi;f are the lower and upper values in the experimental bin. (3.2) (3.3) (3.4) In order to reach higher Q and qT values, we also consider Z boson production in collider experiments at Tevatron. We analyze data from CDF and D0, collected during Tevatron collaborations studied the differential cross section for the production of an e+e pair from pp collision through an intermediate Z vector boson, namely pp ! Z ! e+e + X. The invariant mass distribution peaks at the Z-pole, Q MZ , while the transverse momentum of the exchanged Z ranges in 0 < qT < 20 GeV. We use the same kinematic cut applied to Drell-Yan events: qT < 0:2 Q + 0:5 GeV = 18:7 GeV, since Q is fixed to MZ . The observable measured in CDF and D0 is apart from the case of D0 Run II, for which the published data refer to 1= d =dqT . In order to work with the same observable, we multiply the D0 Run II data by the total cross section of the process exp = 255:8 16 pb [85]. In this case, we add in quadrature the uncertainties of the total cross section and of the published data. We normalize our functional form with the factors listed in table 4. These are the same normalization factors used in ref. [78], computed by comparing the experimental total cross section with the theoretical results based on the code of ref. [86]. These factors are not precisely consistent with our formulas. In fact, as we will discuss in section 4.3 a 5% increase in these factors would improve the agreement with data, without affecting the TMD parameters. 3.4 The replica method Our fit is based on the replica method. In this section we describe it and we give a definition of the 2 function minimized by the fit procedure. The fit and the error analysis are carried out using a similar Monte Carlo approach as in refs. [23, 87, 88] and taking inspiration from the work of the Neural-Network PDF (NNPDF) collaboration (see, e.g., refs. [89–91]). The approach consists in creating M replicas of the data points. In each replica (denoted by the index r), each data point i is shifted by a Gaussian noise with the same variance as the measurement. Each replica, therefore, represents a possible outcome of an independent experimental measurement, which we denote by mhN;r(x; z; Ph2T ; Q2). The number of replicas is chosen so that the mean and standard deviation of the set of replicas accurately reproduces the original data points. In this case 200 replicas are sufficient for the purpose. The error for each replica is taken to be equal to the error on the original data points. This is consistent with the fact that the variance of the M replicas should reproduce the variance of the A minimization procedure is applied to each replica separately, by minimizing the original data points. following error function: Er2(fpg) = X i mhN;r(xi; zi; Ph2T i; Qi2) mhN;theo(xi; zi; Ph2T i; fpg) 2 mhN;2stat + mhN;2sys (xi; zi; Ph2T i; Qi2)+ mhN;theo(xi; zi; Ph2T i) 2 : (3.6) The sum runs over the i experimental points, including all species of targets N and finalstate hadrons h. In each z bin for each replica the values of the collinear fragmentation functions D1a~ h are independently modified with a Gaussian noise with standard deviation equal to the theoretical error D1a~ h. In this work we rely on different parametrizations for D1a~ h: for pions we use the DSEHS analysis [64] at NLO in parametrization [65] at LO in S. The uncertainties in ref. [92]; they represents the only source of uncertainty in S; for kaons we use the DSS D1a~ h are estimated from the plots mhN;theo. Statistical and systematic experimental uncertainties mhN;stat and mhN;sys are taken from the experimental collaborations. We do not take into account the covariance among different kinematic bins. We minimize the error function in eq. (3.6) with Minuit [93]. In each replica we randomize the starting point of the minimization, to better sample the space of fit parameters. The final outcome is a set of M different vectors of best-fit parameters, fp0rg; r = 1; : : : M, with which we can calculate any observable, its mean, and its standard deviation. The distribution of these values needs not to be necessarily Gaussian. In fact, in this case the 1 values of 2/d.o.f. should be peaked around one. confidence interval is different from the 68% interval. The latter can simply be computed for each experimental point by rejecting the largest and the lowest 16% of the M values. Although the minimization is performed on the function defined in eq. (3.6), the agreement of the M replicas with the original data is expressed in terms of a 2 function defined as in eq. (3.6) but with the replacement mhN;r ! mhN , i.e., with respect to the original data set. If the model is able to give a good description of the data, the distribution of the M 4 Our work aims at simultaneously fitting for the first time data sets related to different experiments. In the past, only fits related either to SIDIS or hadronic collisions have been presented. Here we mention a selection of recent existing analyses. In ref. [23], the authors fitted Hermes multiplicities only (taking into account a total of 1538 points) without taking into account QCD evolution. In that work, a flavor decomposition in transverse momentum of the unpolarized TMDs and an analysis of the kinematic dependence of the intrinsic average square transverse momenta were presented. In ref. [76] the authors fitted Hermes and Compass multiplicities separately (576 and 6284 points respectively), without TMD evolution and introducing an ad-hoc normalization for Compass data. A fit of SIDIS data including TMD evolution was performed on measurements by the H1 collaboration of the so-called transverse energy flow [55, 94]. Looking at data from hadronic collisions, Konychev and Nadolsky [57] fitted data of low-energy Drell-Yan events and Z-boson production at Tevatron, taking into account TMD evolution at NLL accuracy (this is the most recent of a series of important papers on the subject [56, 77, 95]). They fitted in total 98 points. Contrary to our approach, Konychev and Nadolsky studied the quality of the fit as a function of max. They found that the best value for max is 1:5 GeV 1 (to be compared to our choice max 1:123 GeV 1, see section 2.3). Comparisons of best-fit values in the nonperturbative Sudakov form factors are delicate, since the functional form is different from ours. In 2014 D’Alesio, Echevarria, Melis, Scimemi performed a fit [78] of Drell-Yan data and Z-boson production data at Tevatron, Parameters 11 12629 363 1:55 0:05 focusing in particular on the role of the nonperturbative contribution to the kernel of TMD evolution. This is the fit with the highest accuracy in TMD evolution performed up to date (NNLL in the Sudakov exponent and O( S) in the Wilson coefficients). In the same year Echevarria, Idilbi, Kang and Vitev [15] presented a parametrization of the unpolarized TMD that described qualitatively well some bins of Hermes and Compass data, together with Drell-Yan and Z-production data. A similar result was presented by Sun, Isaacson, Yuan and Yuan [96]. In the following, we detail the results of a fit to the data sets described in section 3 with a flavor-independent configuration for the transverse momentum dependence of unpolarized TMDs. In table 5 we present the total 2. The number of degrees of freedom (d.o.f.) is given by the number of data points analyzed reduced by the number of free parameters in the error function. The overall quality of the fit is good, with a global 2/d.o.f. = 1:55 0:05. Uncertainties are computed as the 68% confidence level (C.L.) from the replica methodology. 4.1 Agreement between data and theory The partition of the global 2 among SIDIS off a proton, SIDIS off a deuteron, Drell-Yan and Z production events is given in table 6, 7, 8, 9 respectively. Semi-inclusive DIS. For SIDIS at Hermes off a proton, most of the contribution to the 2 comes from events with a + in the final state. In ref. [23] the high 2 was attributed to the poor agreement between experiment and theory at the level of the collinear multiplicities. In this work we use a newer parametrization of the collinear FFs (DSEHS [64]), based on a fit which includes Hermes collinear pion multiplicities. In spite of this improvement, the contribution to 2 from Hermes data is higher then in ref. [23], because the present fit includes data from other experiments (Hermes represents less than 20% of the whole data set). The bins with the worst agreement are at low Q2. As we will discuss in section 4.3, we think that the main reason for the large 2 at Hermes is a normalization difference. This may also be due to the fact that we are computing our theoretical estimates at the average values of the kinematic variables, instead of integrating the multiplicities in each bin. Kaon multiplicities have in general a lower 2, due to the bigger statistical errors and the large uncertainties for the kaon FFs. For pion production off a deuteron at Hermes the 2 is lower with respect to the production off a proton, but still compatible within uncertainties. For kaon production off a deuteron the 2 is higher with respect to the scattering off a proton. The difference is especially large for K . h SIDIS at Compass involves scattering off deuteron only, D ! h , and we identify . The quality of the agreement between theory and Compass data is better than in the case of pion production at Hermes. This depends on at least two factors: first, our fit is essentially driven by the Compass data, which represent about 75% of the whole data Points 2=points Hermes D ! + 190 set; second, the observable that we fit in this case is the normalized multiplicity, defined in eq. (3.1). This automatically eliminates most of the discrepancy between theory and data due to normalization. Figure 3 presents the agreement between the theoretical formula in (2.3) and the Hermes multiplicities for production of pions off a proton and a deuteron. Different hxi, hzi and hQ2i bins are displayed as a function of the transverse momentum of the detected hadron PhT . The grey bands are an envelope of the 200 replica of best-fit curves. For every point in PhT we apply a 68% C.L. selection criterion. Points marked with different symbols and colors correspond to different hzi values. There is a strong correlation between hxi and hQ2i that does not allow us to explore the x and Q2 dependence of the TMDs separately. Studying the contributions to the 2/points as a function of the kinematics, we notice that the 2(Q2) tends to improve as we move to higher Q2 values, where the kinematic approximations of factorization are more reliable. Moreover, usually the 2(z) increases at lower z values. Figure 4 has same contents and notation as in figure 3 but for kaons in the final state. In this case, the trend of the agreement as a function of Q2 is not as clear as for the case of pions: good agreement is found also at low Q2. In figure 5 we present Compass normalized multiplicities (see eq. (3.1)) for production of off a deuteron for different hxi, hzi, and hQ2i bins as a function of the transverse momentum of the detected hadron PhT . The open marker around the first PhT point in each panel indicates that the first value is fixed and not fitted. The correlation between x and Q2 is less strong than at Hermes and this allows us to study different hxi bins at fixed hQ2i. For the highest Q2 bins, the agreement is good for all hxi, hzi and Ph2T . In bins at lower Q2, the descriptions gets worse, especially at low and high z. For fixed hQ2i and high hzi, a good agreement is recovered moving to higher hxi bins (see, e.g., the third line from the top in figure 5). Figure 6 has same content and notation as in figure 5, but for h+ + . The same comments on the agreement between theory and the data apply. 〈 〉= 〈 〉= ( ( ( ( = ) = ) 〈 〉= 〈 〉= = ) = ) 〈 〉= 〈 〉= ( ( = ) = ) hxi, hzi, and hQ2i bins as a function of the transverse momentum of the detected hadron PhT . For clarity, each hzi bin has been shifted by an offset indicated in the legend. Drell-Yan and Z production. The low energy Drell-Yan data collected by the E288 and E605 experiments at Fermilab have large error bands (see figure 7). This is why the 2 values in table 8 are rather low compared to the other data sets. The agreement is also good for Z boson production, see table 9. The statistics from Run-II is higher, which generates smaller experimental uncertainties and higher 2, especially for the CDF experiment. Figure 7 displays the cross section for DY events differential with respect to the transverse momentum qT of the virtual photon, its invariant mass Q2 and rapidity y. As for the 〈 〉= 〈 〉= ( ( ( ( = ) = ) 〈 〉= 〈 〉= = ) = ) 〈 〉= 〈 〉= ( ( = ) = ) hxi, hzi, and hQ2i bins as a function of the transverse momentum of the detected hadron PhT . For clarity, each hzi bin has been shifted by an offset indicated in the legend. case of SIDIS, the grey bands are the 68% C.L. envelope of the 200 replicas of the fit function. The four panels represents different values for the rapidity y or xF (see eq. (2.15)). In each panel, we have plots for different Q2 values. The lower is Q, the less points in qT we fit (see also section 3.2). The hard scale lies in the region 4:5 < hQi < 13:5 GeV. This region is of particular importance, since these “moderate” Q values should be high enough to safely apply factorization and, at the same time, low enough in order for the nonperturbative effects to not be shaded by transverse momentum resummation. 〈 〉= 〈 〉= 〈 〉= 〈 〉= 〈 〉= 〈 〉= 〈 〉= ( ( ( ( ( ( ( = ) = ) = ) = ) = ) = ) = ) 〈 〉=〈 〉= 〈 〉= 〈 〉= different hxi, hzi, and hQ2i bins as a function of the transverse momentum of the detected hadron PhT . Multiplicities are normalized to the first bin in PhT for each hzi value (see (3.1)). For clarity, each hzi bin has been shifted by an offset indicated in the legend. In figure 8 we compare the cross section differential with respect to the transverse momentum qT of the virtual Z (namely eq. (2.12) integrated over ) with data from CDF and D0 at Tevatron Run I and II. Due to the higher Q = MZ , the range explored in qT is much larger compared to all the other observables considered. The tails of the distributions deviate from a Gaussian behavior, as it is also evident in the bins at higher Q2 in figure 7. The band from the replica methodology in this case is much narrower, due to the reduced 〈 〉= 〈 〉= 〈 〉= 〈 〉= 〈 〉= 〈 〉= 〈 〉= ( ( ( ( ( ( ( = ) = ) = ) = ) = ) = ) = ) 〈 〉=〈 〉= 〈 〉= 〈 〉= different hxi, hzi, and hQ2i bins as a function of the transverse momentum of the detected hadron PhT . Multiplicities are normalized to the first bin in PhT for each hzi value (see (3.1)). For clarity, each hzi bin has been shifted by an offset indicated in the legend. sensitivity to the intrinsic transverse momenta at Q = MZ and to the limited range of best-fit values for the parameter g2, which controls soft-gluon emission. As an effect of TMD evolution, the peak shifts from 1 GeV for Drell-Yan events in figure 7 to 5 GeV in figure 8. The position of the peak is affected both by the perturbative and the nonperturbative part of the Sudakov exponent (see section 2.3 and [22]). Most of the contributions to the 2 comes from normalization effects and not from the shape in qT (see section 4.3). 〈〈〈 〉〉〉=== ( ( ( = ) = ) = ) 〈〈〈 〉〉〉=== ( ( ( = ) = ) =- ) 〈〈〈 〉〉〉=== ( ( ( =- ) =- ) =- ) Points 2=points 0:99 45 E605 35 8 and for different hQi bins. For clarity, each hQi bin has been normalized (the first data point has been set always equal to 1) and then shifted by an offset indicated in the legend. E288 [200] E288 [300] E288 [400] energy. The labels in square brackets were introduced in section 3.2. Transverse momentum dependence at 1 GeV The variables min and max delimit the range in T where transverse momentum resummation is computed perturbatively. The g2 parameter enters the nonperturbative Sudakov exponent and quantifies the amount of transverse momentum due to soft gluon radiation that is not included in the perturbative part of the Sudakov form factor. As already explained in section 2.3, in this work we fix the value for min and max in such a way that at Q = 1 GeV the unpolarized TMDs coincide with their nonperturbative input. We leave g2 Table 10 summarizes the chosen values of min, max and the best-fit value for g2. The latter is given as an average with 68% C.L. uncertainty computed over the set of 200 replicas. We also quote the results obtained from replica 105, since its parameters are very D0) with two different values for the center-of-mass energy (ps = 1:8 TeV and p produced from pp collisions at Tevatron. The four panels refer to different experiments (CDF and s = 1:96 TeV). In this case the band is narrow due to the narrow range for the best-fit values of g2. All replicas Replica 105 max [GeV 1 ] min [GeV 1 ] g2 [GeV2] (fixed) 2e E 2e E (fixed) 2e E =Q 2e E =Q 0:13 0:01 0:128 close to the mean values of all replicas. We obtain a value g2 = 0:13 0:018) obtained in ref. [57], where however no SIDIS data was taken into consideration, and smaller than the value (g2 = 0:16) chosen in ref. [15]. We stress however that our prescriptions involving both min and max are different from previous works. Table 11 collects the best-fit values of parameters in the nonperturbative part of the TMDs at Q = 1 GeV (see eqs. (2.34) and (2.35)); as for g2, we give the average value over the full set of replicas and the standard deviation based on a 68% C.L. (see section 3.4), and we also quote the value of replica 105. In figure 9 we compare different extractions of partonic transverse momenta. The hor? izontal axis shows the value of the average transverse momentum squared for the incoming parton, k 2 (x = 0:1) (see eq. (2.40)). The vertical axis shows the value of P 2 (z = 0:5), the average transverse momentum squared acquired during the fragmentation process (see eq. (2.40)). The white square (label 1) indicates the average values of the two quantities obtained in the present analysis at Q2 = 1 GeV2. Each black dot around the white square is an outcome of one replica. The red region around the white square contains the 68% of the replicas that are closest to the average value. The same applies to the white circle and the ? TMD PDFs 0:285 N3 [GeV2] 0:212 0:86 0:78 0:39 F PDFs, hk?2i(x = 0:1), in different phenomenological extractions. (1): average values (white square) obtained in the present analysis, values obtained from each replica (black dots) and 68% C.L. area (red); (2) results from ref. [23], (3) results from ref. [97], (4) results from ref. [76] for Hermes data, (5) results from ref. [76] for Hermes data at high z, (6) results from ref. [76] for normalized Compass data, (7) results from ref. [76] for normalized Compass data at high z, (8) results from ref. [15]. orange region around it (label 2), related to the flavor-independent version of the analysis in ref. [23], obtained by fitting only Hermes SIDIS data at an average hQ2i = 2:4 GeV2 and neglecting QCD evolution. A strong anticorrelation between the transverse momenta is evident in this older analysis. In our new analysis, the inclusion of Drell-Yan and Z production data adds physical information about TMD PDFs, free from the influence of TMD ? FFs. This reduces significantly the correlation between k ? 2 (x = 0:1) and P 2 (z = 0:5). The 68% confidence region is smaller than in the older analysis. The average values of k 2 (x = 0:1) are similar and compatible within error bands. The values of P 2 (z = 0:5) in the present analysis turn out to be larger than in the older analysis, an effect that is due mainly to Compass data. It must be kept in mind that the two analyses lead also to differences in the x and z dependence of the transverse momentum squared. This dependence is ? ? 2.31 0.80 C.L. envelope of the full sets of best-fit curves. The data used in the fit approximately cover the range 5 10 3 . x . 0:5 and 0:2 . z . 0:7. HJEP06(217)8 p ! 5.18 1.94 p ! D ! 2 (x) and figure 10 (b) for P 2 (z). The bands are computed as the 68% C.L. envelope of the full sets of curves from the 200 replicas. Comparison with ? other extractions are presented and the legend is detailed in the caption of figure 9. Stability of our results In this subsection we discuss the effect of modifying some of the choices we made in our default fit. Instead of repeating the fitting procedure with different choices, we limit ourselves to checking how the 2 of a single replica is affected by the modifications. As starting point we choose replica 105, which, as discussed above, is one of the most representative among the whole replica set. The global 2=d.o.f. of replica 105 is 1.51. We keep all parameters fixed, without performing any new minimization, and we compute the 2=d.o.f. after the modifications described in the following. First of all, we analyze Hermes data with the same strategy as Compass, i.e., we normalize Hermes data to the value of the first bin in PhT . In this case, the global 2=d.o.f. reduces sharply to 1.27. The partial 2 for the different SIDIS processes measured at Hermes are shown in table 12. This confirms that normalization effects are the main contribution to the 2 of SIDIS data and have minor effects on TMD-related parameters. In fact, even if we perform a new fit with this modification, the 2 does not improve significantly and parameters do not change much. We consider the effect of changing the normalization of the Z-boson data: if we increase the normalization factors quoted in the last row of table 4 by 5%, the 2 quoted in the last row of table 9 drops to 0.66, 0.52, 0.65, 0.68. This effect is also already visible by eye in figure 8: the theoretical curves are systematically below the experimental data points, but the shape is reproduced very well. We consider the sensitivity of our results to the parameterizations adopted for the collinear quark PDFs. The 2=d.o.f. varies from its original value 1.51, obtained with the NLO GJR 2008 parametrization [62], to 1.84 using NLO MSTW 2008 [98], and 1.85 using NLO CJ12 [99]. In both cases, the agreement with Hermes and Z boson data is not affected significanlty, the agreement with Compass data becomes slightly worse, and the agreement with DY data becomes clearly worse. An extremely important point is the choice of kinematic cuts. Our default choices are listed in tables 1–4. We consider also more stringent kinematic cuts on SIDIS data: Q2 > 1:5 GeV2 and 0:25 < z < 0:6 instead of Q2 > 1:4 GeV2 and 0:2 < z < 0:7, leaving the other ones unchanged. The number of bins with these cuts reduces from 8059 to 5679 and the 2=d.o.f. decreases to the value 1.23. In addition, if we replace the constraint PhT < Min[0:2 Q; 0:7 Qz] + 0:5 GeV with PhT < Min[0:2 Q; 0:5 Qz] + 0:3 GeV, the number of bins reduces to 3380 and the 2=d.o.f. decreases further to 0.96. By adopting the even stricter cut PhT < 0:2 Qz, the number of bins drops to only 477, with a 2/d.o.f. =1.02. We can conclude that our fit, obtained by fitting data in an extended kinematic region, where TMD factorization may be questioned, works extremely well also in a narrower region, where TMD factorization is expected to be under control. 5 In this work we demonstrated for the first time that it is possible to perform a simultaneous fit of unpolarized TMD PDFs and FFs to data of SIDIS, Drell-Yan and Z boson production at small transverse momentum collected by different experiments. This constitutes the factorization and with the implementation of TMD evolution at NLL accuracy. first attempt towards a global fit of f1a(x; k?2) and D1a!h(z; P 2 ) in the context of TMD ? We extracted unpolarized TMDs using 8059 data points with 11 free parameters using a replica methodology. We selected data with Q2 > 1:4 GeV2 and 0:2 < z < 0:7. We restricted our fit to the small transverse momentum region, selecting the maximum value of transverse momentum on the basis of phenomenological considerations (see section 3). With these choices, we included regions where TMD factorization could be questioned, but we checked that our results describe very well the regions where TMD factorization is supposed to hold. The average 2/d.o.f. is 1:55 0:05 and can be improved up to 1.02 restricting the kinematic cuts, without changing the parameters (see section 4.3). Most of the discrepancies between experimental data and theory comes from the normalization and not from the transverse momentum shape. Our fit is performed assuming that the intrinsic transverse momentum dependence of TMD PDFs and FFs can be parametrized by a normalized linear combination of a Gaussian and a weighted Gaussian. We considered that the widths of the Gaussians depend on the longitudinal momenta. We neglected a possible flavor dependence. For the nonperturbative component of TMD evolution, we adopted the choice most often used in the literature (see section 2.3). We plan to release grids of the parametrizations studied in this work via TMDlib [100] to facilitate phenomenological studies for present and future experiments. HJEP06(217)8 In future studies, different functional forms for all the nonperturbative ingredients should be explored, including also a possible flavor dependence of the intrinsic transverse momenta. A more precise analysis from the perturbative point of view is also needed, which should in principle make it possible to relax the tension in the normalization and to describe data at higher transverse momenta. Moreover, the description at low transverse momentum should be properly matched to the collinear fixed-order calculations at high transverse momentum. Together with an improved theoretical framework, in order to better understand the formalism more experimental data is needed. It would be particularly useful to extend the coverage in x, z, rapidity, and Q2. The 12 GeV physics program at Jefferson Lab [101] will be very important to constrain TMD distributions at large x. Additional data from SIDIS (at Compass, at a future Electron-Ion Collider), Drell-Yan (at Compass, at Fermilab), Z=W production (at LHC, RHIC, and at A Fixed-Target Experiment at the LHC [ 102 ]) will be very important. Measurements related to unpolarized TMD FFs at e+e colliders (at Belle-II, BES-III, at a future International Linear Collider) will be invaluable, since they are presently missing. Our work focused on quark TMDs. We remark that at present almost nothing is known experimentally about gluon TMDs [11, 103], because they typically require higher-energy scattering processes and they are harder to isolate as compared to quark distributions. Several promising measurements have been proposed in order to extract both the unpolarized and linearly polarized gluon TMDs inside an unpolarized proton. The cleanest possibility would be to look at dijet and heavy quark pair production in electron-proton collisions at a future EIC [104, 105]. Other proposals include isolated photon-pair production at RHIC [106] and quarkonium production at the LHC [107–110]. Testing the formalism of TMD factorization and understanding the structure of unpolarized TMDs is only the first crucial step in the exploration of the 3D proton structure in momentum space and this work opens the way to global determinations of TMDs. Building on this, we can proceed to deepen our understanding of hadron structure via polarized structure function and asymmetries (see, e.g., refs. [111, 112] and references therein) and, at the same time, to test the impact of hadron structure in precision measurements at high-energies, such as at the LHC. A detailed mapping of hadron structure is essential to interpret data from hadronic collisions, which are among the most powerful tools to look for footprints of new physics. Acknowledgments Discussions with Giuseppe Bozzi and Alexey Vladimirov are gratefully acknowledged. This work is supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant agreement No. 647981, 3DSPIN). AS acknowledges support from U.S. Department of Energy contract DE-AC05-06OR23177, under which Jefferson Science Associates, LLC, manages and operates Jefferson Lab. The work of AS has been funded partly also by the program of the Stichting voor Fundamenteel Onderzoek der Materie (FOM), which is financially supported by the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO). HJEP06(217)8 Open Access. This article is distributed under the terms of the Creative Commons Attribution License (CC-BY 4.0), which permits any use, distribution and reproduction in any medium, provided the original author(s) and source are credited. [http://www.cambridge.org/de/knowledge/isbn/item5756723] [INSPIRE]. [3] R. Angeles-Martinez et al., Transverse Momentum Dependent (TMD) parton distribution functions: status and prospects, Acta Phys. Polon. B 46 (2015) 2501 [arXiv:1507.05267] HJEP06(217)8 (2016) 163 [INSPIRE]. [4] T.C. Rogers, An Overview of Transverse Momentum Dependent Factorization and Evolution, Eur. Phys. J. A 52 (2016) 153 [arXiv:1509.04766] [INSPIRE]. [5] A. Bacchetta, Where do we stand with a 3-D picture of the proton?, Eur. Phys. J. A 52 [6] M. Radici, Imaging the proton, AIP Conf. Proc. 1735 (2016) 020003 [INSPIRE]. [7] P.J. Mulders and R.D. Tangerman, The Complete tree level result up to order 1/Q for polarized deep inelastic leptoproduction, Nucl. Phys. B 461 (1996) 197 [Erratum ibid. B 484 (1997) 538] [hep-ph/9510301] [INSPIRE]. [8] A. Bacchetta, M. Diehl, K. Goeke, A. Metz, P.J. Mulders and M. Schlegel, Semi-inclusive deep inelastic scattering at small transverse momentum, JHEP 02 (2007) 093 [hep-ph/0611265] [INSPIRE]. [9] D. Boer and P.J. Mulders, Time reversal odd distribution functions in leptoproduction, Phys. Rev. D 57 (1998) 5780 [hep-ph/9711485] [INSPIRE]. [10] A. Bacchetta and P.J. Mulders, Deep inelastic leptoproduction of spin-one hadrons, Phys. Rev. D 62 (2000) 114004 [hep-ph/0007120] [INSPIRE]. [11] P.J. Mulders and J. Rodrigues, Transverse momentum dependence in gluon distribution and fragmentation functions, Phys. Rev. D 63 (2001) 094021 [hep-ph/0009343] [INSPIRE]. [12] D. Boer, S. Cotogno, T. van Daal, P.J. Mulders, A. Signori and Y.-J. Zhou, Gluon and Wilson loop TMDs for hadrons of spin 1, JHEP 10 (2016) 013 [arXiv:1607.01654] [13] A. Bacchetta and M. Radici, Constraining quark angular momentum through semi-inclusive measurements, Phys. Rev. Lett. 107 (2011) 212001 [arXiv:1107.5755] [INSPIRE]. [14] M. Anselmino, M. Boglione and S. Melis, A Strategy towards the extraction of the Sivers function with TMD evolution, Phys. Rev. D 86 (2012) 014028 [arXiv:1204.1239] [15] M.G. Echevarria, A. Idilbi, Z.-B. Kang and I. Vitev, QCD Evolution of the Sivers Asymmetry, Phys. Rev. D 89 (2014) 074013 [arXiv:1401.5078] [INSPIRE]. [16] M. Anselmino, M. Boglione, U. D’Alesio, F. Murgia and A. Prokudin, Study of the sign change of the Sivers function from STAR Collaboration W/Z production data, JHEP 04 (2017) 046 [arXiv:1612.06413] [INSPIRE]. [17] Z. Lu and I. Schmidt, Updating Boer-Mulders functions from unpolarized pd and pp of azimuthal asymmetries in unpolarized semi-inclusive deep inelastic scattering, Phys. Rev. D 91 (2015) 074019 [arXiv:1502.04214] [INSPIRE]. [19] C. Lefky and A. Prokudin, Extraction of the distribution function h1?T from experimental data, Phys. Rev. D 91 (2015) 034010 [arXiv:1411.0580] [INSPIRE]. Simultaneous extraction of transversity and Collins functions from new SIDIS and e+e data, Phys. Rev. D 87 (2013) 094019 [arXiv:1303.3822] [INSPIRE]. [21] Z.-B. Kang, A. Prokudin, P. Sun and F. Yuan, Extraction of Quark Transversity Distribution and Collins Fragmentation Functions with QCD Evolution, Phys. Rev. D 93 (2016) 014009 [arXiv:1505.05589] [INSPIRE]. [22] A. Signori, Flavor and Evolution Effects in TMD Phenomenology: Manifestation of Hadron Structure in High-Energy Scattering Processes, Ph.D. Thesis, Vrije U., Amsterdam (2016), [23] A. Signori, A. Bacchetta, M. Radici and G. Schnell, Investigations into the flavor dependence of partonic transverse momentum, JHEP 11 (2013) 194 [arXiv:1309.3507] [24] A. Bacchetta, M.G. Echevarria, P.J.G. Mulders, M. Radici and A. Signori, Effects of TMD evolution and partonic flavor on e+e annihilation into hadrons, JHEP 11 (2015) 076 [arXiv:1508.00402] [INSPIRE]. [25] J. Collins, L. Gamberg, A. Prokudin, T.C. Rogers, N. Sato and B. Wang, Relating Transverse Momentum Dependent and Collinear Factorization Theorems in a Generalized Formalism, Phys. Rev. D 94 (2016) 034014 [arXiv:1605.00671] [INSPIRE]. [26] D. Boer et al., Gluons and the quark sea at high energies: Distributions, polarization, tomography, arXiv:1108.1713 [INSPIRE]. [27] A. Bacchetta, D. Boer, M. Diehl and P.J. Mulders, Matches and mismatches in the descriptions of semi-inclusive processes at low and high transverse momentum, JHEP 08 (2008) 023 [arXiv:0803.0227] [INSPIRE]. [28] ZEUS collaboration, S. Chekanov et al., Measurement of the Longitudinal Proton Structure Function at HERA, Phys. Lett. B 682 (2009) 8 [arXiv:0904.1092] [INSPIRE]. [29] H1 collaboration, V. Andreev et al., Measurement of inclusive ep cross sections at high Q2 at p Eur. Phys. J. C 74 (2014) 2814 [arXiv:1312.4821] [INSPIRE]. s = 225 and 252 GeV and of the longitudinal proton structure function FL at HERA, [30] J.C. Collins and D.E. Soper, Back-To-Back Jets in QCD, Nucl. Phys. B 193 (1981) 381 [Erratum ibid. B 213 (1983) 545] [INSPIRE]. [31] J.C. Collins, D.E. Soper and G.F. Sterman, Transverse Momentum Distribution in Drell-Yan Pair and W and Z Boson Production, Nucl. Phys. B 250 (1985) 199 [INSPIRE]. [32] X.-d. Ji and F. Yuan, Parton distributions in light cone gauge: Where are the final state interactions?, Phys. Lett. B 543 (2002) 66 [hep-ph/0206057] [INSPIRE]. [33] X.-d. Ji, J.-p. Ma and F. Yuan, QCD factorization for semi-inclusive deep-inelastic scattering at low transverse momentum, Phys. Rev. D 71 (2005) 034005 [hep-ph/0404183] [34] S.M. Aybat and T.C. Rogers, TMD Parton Distribution and Fragmentation Functions with QCD Evolution, Phys. Rev. D 83 (2011) 114042 [arXiv:1101.5057] [INSPIRE]. [35] M.G. Echevarria, A. Idilbi and I. Scimemi, Factorization Theorem For Drell-Yan At Low qT And Transverse Momentum Distributions On-The-Light-Cone, JHEP 07 (2012) 002 [arXiv:1111.4996] [INSPIRE]. Transverse Momentum Dependent Distribution Functions (TMDs) at NNLL, Eur. Phys. J. C 73 (2013) 2636 [arXiv:1208.1281] [INSPIRE]. [37] J.C. Collins and T.C. Rogers, Equality of Two Definitions for Transverse Momentum Dependent Parton Distribution Functions, Phys. Rev. D 87 (2013) 034018 [arXiv:1210.2100] [INSPIRE]. [38] H.H. Matevosyan, W. Bentz, I.C. Cloet and A.W. Thomas, Transverse Momentum Dependent Fragmentation and Quark Distribution Functions from the NJL-jet Model, Phys. Rev. D 85 (2012) 014021 [arXiv:1111.1740] [INSPIRE]. Kinematics of Current Region Fragmentation in Semi-Inclusive Deeply Inelastic Scattering, Phys. Lett. B 766 (2017) 245 [arXiv:1611.10329] [INSPIRE]. [40] E. Moffat, W. Melnitchouk, T.C. Rogers and N. Sato, What are the low-Q and large-x boundaries of collinear QCD factorization theorems?, Phys. Rev. D 95 (2017) 096008 [arXiv:1702.03955] [INSPIRE]. [41] M.G. Echevarria, I. Scimemi and A. Vladimirov, Unpolarized Transverse Momentum Dependent Parton Distribution and Fragmentation Functions at next-to-next-to-leading order, JHEP 09 (2016) 004 [arXiv:1604.07869] [INSPIRE]. [42] D. Boer and W. Vogelsang, Drell-Yan lepton angular distribution at small transverse momentum, Phys. Rev. D 74 (2006) 014004 [hep-ph/0604177] [INSPIRE]. [43] S. Arnold, A. Metz and M. Schlegel, Dilepton production from polarized hadron hadron collisions, Phys. Rev. D 79 (2009) 034005 [arXiv:0809.2262] [INSPIRE]. [44] Particle Data Group collaboration, C. Patrignani et al., Review of Particle Physics, Chin. Phys. C 40 (2016) 100001 [INSPIRE]. [45] M. Lambertsen and W. Vogelsang, Drell-Yan lepton angular distributions in perturbative QCD, Phys. Rev. D 93 (2016) 114013 [arXiv:1605.02625] [INSPIRE]. [46] G. Parisi and R. Petronzio, Small Transverse Momentum Distributions in Hard Processes, Nucl. Phys. B 154 (1979) 427 [INSPIRE]. [47] G. Altarelli, R.K. Ellis, M. Greco and G. Martinelli, Vector Boson Production at Colliders: A Theoretical Reappraisal, Nucl. Phys. B 246 (1984) 12 [INSPIRE]. [48] G. Bozzi, S. Catani, G. Ferrera, D. de Florian and M. Grazzini, Production of Drell-Yan lepton pairs in hadron collisions: Transverse-momentum resummation at next-to-next-to-leading logarithmic accuracy, Phys. Lett. B 696 (2011) 207 [arXiv:1007.2351] [INSPIRE]. [49] D. Boer and W.J. den Dunnen, TMD evolution and the Higgs transverse momentum distribution, Nucl. Phys. B 886 (2014) 421 [arXiv:1404.6753] [INSPIRE]. [50] E. Laenen, G.F. Sterman and W. Vogelsang, Higher order QCD corrections in prompt photon production, Phys. Rev. Lett. 84 (2000) 4296 [hep-ph/0002078] [INSPIRE]. HJEP06(217)8 Small Transverse Momentum, Nucl. Phys. B 244 (1984) 337 [INSPIRE]. the transverse momentum distributions of massive vector bosons in hadronic collisions, Nucl. Phys. B 542 (1999) 311 [hep-ph/9809367] [INSPIRE]. and the spectrum of the Higgs boson at the LHC, Nucl. Phys. B 737 (2006) 73 [hep-ph/0508068] [INSPIRE]. [55] P.M. Nadolsky, D.R. Stump and C.P. Yuan, Semiinclusive hadron production at HERA: The effect of QCD gluon resummation, Phys. Rev. D 61 (2000) 014003 [Erratum ibid. D 64 (2001) 059903] [hep-ph/9906280] [INSPIRE]. [56] F. Landry, R. Brock, P.M. Nadolsky and C.P. Yuan, Tevatron Run-1 Z boson data and Collins-Soper-Sterman resummation formalism, Phys. Rev. D 67 (2003) 073016 [hep-ph/0212159] [INSPIRE]. [57] A.V. Konychev and P.M. Nadolsky, Universality of the Collins-Soper-Sterman nonperturbative function in gauge boson production, Phys. Lett. B 633 (2006) 710 [hep-ph/0506225] [INSPIRE]. [58] C.A. Aidala, B. Field, L.P. Gamberg and T.C. Rogers, Limits on transverse momentum dependent evolution from semi-inclusive deep inelastic scattering at moderate Q, Phys. Rev. D 89 (2014) 094002 [arXiv:1401.2654] [INSPIRE]. [59] J. Collins and T. Rogers, Understanding the large-distance behavior of transverse-momentum-dependent parton densities and the Collins-Soper evolution kernel, Phys. Rev. D 91 (2015) 074020 [arXiv:1412.3820] [INSPIRE]. [60] I. Scimemi and A. Vladimirov, Power corrections and renormalons in Transverse Momentum Distributions, JHEP 03 (2017) 002 [arXiv:1609.06047] [INSPIRE]. [61] U. D’Alesio, M.G. Echevarria, S. Melis and I. Scimemi, Non-perturbative QCD effects in qT spectra of Drell-Yan and Z-boson production, JHEP 11 (2014) 098 [arXiv:1407.3311] [INSPIRE]. [62] M. Gluck, P. Jimenez-Delgado and E. Reya, Dynamical parton distributions of the nucleon and very small-x physics, Eur. Phys. J. C 53 (2008) 355 [arXiv:0709.0614] [INSPIRE]. [63] A. Buckley et al., LHAPDF6: parton density access in the LHC precision era, Eur. Phys. J. C 75 (2015) 132 [arXiv:1412.7420] [INSPIRE]. [64] D. de Florian, R. Sassot, M. Epele, R.J. Hernández-Pinto and M. Stratmann, Parton-to-Pion Fragmentation Reloaded, Phys. Rev. D 91 (2015) 014035 [arXiv:1702.06353] [INSPIRE]. Parton-to-Kaon Fragmentation Revisited, Phys. Rev. D 95 (2017) 094019 function for pions and kaons in a spectator model, Phys. Lett. B 659 (2008) 234 [68] B. Pasquini, S. Cazzaniga and S. Boffi, Transverse momentum dependent parton distributions in a light-cone quark model, Phys. Rev. D 78 (2008) 034025 [69] H. Avakian, A.V. Efremov, P. Schweitzer and F. Yuan, The transverse momentum dependent distribution functions in the bag model, Phys. Rev. D 81 (2010) 074035 [70] A. Bacchetta, M. Radici, F. Conti and M. Guagnelli, Weighted azimuthal asymmetries in a diquark spectator model, Eur. Phys. J. A 45 (2010) 373 [arXiv:1003.1328] [INSPIRE]. 161 [arXiv:1510.02567] [INSPIRE]. [72] A. Bacchetta, R. Kundu, A. Metz and P.J. Mulders, Estimate of the Collins fragmentation function in a chiral invariant approach, Phys. Rev. D 65 (2002) 094021 [hep-ph/0201091] [73] P. Schweitzer, M. Strikman and C. Weiss, Intrinsic transverse momentum and parton correlations from dynamical chiral symmetry breaking, JHEP 01 (2013) 163 [arXiv:1210.1267] [INSPIRE]. [74] HERMES collaboration, A. Airapetian et al., Multiplicities of charged pions and kaons from semi-inclusive deep-inelastic scattering by the proton and the deuteron, Phys. Rev. D 87 (2013) 074029 [arXiv:1212.5407] [INSPIRE]. [75] COMPASS collaboration, C. Adolph et al., Hadron Transverse Momentum Distributions in Muon Deep Inelastic Scattering at 160 GeV/c, Eur. Phys. J. C 73 (2013) 2531 [Erratum ibid. C 75 (2015) 94] [arXiv:1305.7317] [INSPIRE]. [76] M. Anselmino, M. Boglione, J.O. Gonzalez Hernandez, S. Melis and A. Prokudin, Unpolarised Transverse Momentum Dependent Distribution and Fragmentation Functions from SIDIS Multiplicities, JHEP 04 (2014) 005 [arXiv:1312.6261] [INSPIRE]. [77] F. Landry, R. Brock, G. Ladinsky and C.P. Yuan, New fits for the nonperturbative parameters in the CSS resummation formalism, Phys. Rev. D 63 (2001) 013004 [hep-ph/9905391] [INSPIRE]. [78] U. D’Alesio, M.G. Echevarria, S. Melis and I. Scimemi, Non-perturbative QCD effects in qT spectra of Drell-Yan and Z-boson production, JHEP 11 (2014) 098 [arXiv:1407.3311] [79] A.S. Ito et al., Measurement of the Continuum of Dimuons Produced in High-Energy Proton-Nucleus Collisions, Phys. Rev. D 23 (1981) 604 [INSPIRE]. [80] G. Moreno et al., Dimuon production in proton-copper collisions at ps = 38:8-GeV, Phys. Rev. D 43 (1991) 2815 [INSPIRE]. (2000) 845 [hep-ex/0001021] [INSPIRE]. [81] CDF collaboration, T. Affolder et al., The transverse momentum and total cross section of pairs in the Z boson region from pp collisions at ps = 1:8 TeV, Phys. Rev. Lett. 84 [82] D0 collaboration, B. Abbott et al., Measurement of the inclusive differential cross section for Z bosons as a function of transverse momentum in pp collisions at p s = 1:8 TeV, Phys. Rev. D 61 (2000) 032004 [hep-ex/9907009] [INSPIRE]. in the Z-boson region from pp collisions at p [83] CDF collaboration, T. Aaltonen et al., Transverse momentum cross section of e+e pairs s = 1:96 TeV, Phys. Rev. D 86 (2012) 052010 [84] D0 collaboration, V.M. Abazov et al., Measurement of the shape of the boson transverse momentum distribution in pp ! Z= ! e+e Phys. Rev. Lett. 100 (2008) 102002 [arXiv:0712.0803] [INSPIRE]. + X events produced at p s = 1:96-TeV, [85] CDF collaboration, A. Abulencia et al., Measurements of inclusive W and Z cross sections in pp collisions at p s = 1:96-TeV, J. Phys. G 34 (2007) 2457 [hep-ex/0508029] [INSPIRE]. hadron colliders: a fully exclusive QCD calculation at NNLO, Phys. Rev. Lett. 103 (2009) HJEP06(217)8 082001 [arXiv:0903.2120] [INSPIRE]. [87] A. Bacchetta, A. Courtoy and M. Radici, First extraction of valence transversities in a collinear framework, JHEP 03 (2013) 119 [arXiv:1212.3568] [INSPIRE]. [88] M. Radici, A. Courtoy, A. Bacchetta and M. Guagnelli, Improved extraction of valence transversity distributions from inclusive dihadron production, JHEP 05 (2015) 123 [arXiv:1503.03495] [INSPIRE]. [89] S. Forte, L. Garrido, J.I. Latorre and A. Piccione, Neural network parametrization of deep inelastic structure functions, JHEP 05 (2002) 062 [hep-ph/0204232] [INSPIRE]. [90] NNPDF collaboration, R.D. Ball et al., A determination of parton distributions with faithful uncertainty estimation, Nucl. Phys. B 809 (2009) 1 [Erratum ibid. B 816 (2009) 293] [arXiv:0808.1231] [INSPIRE]. [91] R.D. Ball et al., A first unbiased global NLO determination of parton distributions and their uncertainties, Nucl. Phys. B 838 (2010) 136 [arXiv:1002.4407] [INSPIRE]. [92] M. Epele, R. Llubaroff, R. Sassot and M. Stratmann, Uncertainties in pion and kaon fragmentation functions, Phys. Rev. D 86 (2012) 074028 [arXiv:1209.3240] [INSPIRE]. [93] F. James and M. Roos, Minuit: A System for Function Minimization and Analysis of the Parameter Errors and Correlations, Comput. Phys. Commun. 10 (1975) 343 [INSPIRE]. [94] H1 collaboration, S. Aid et al., Transverse energy and forward jet production in the low x regime at HERA, Phys. Lett. B 356 (1995) 118 [hep-ex/9506012] [INSPIRE]. [95] G.A. Ladinsky and C.P. Yuan, The Nonperturbative regime in QCD resummation for gauge boson production at hadron colliders, Phys. Rev. D 50 (1994) R4239 [hep-ph/9311341] [96] P. Sun, J. Isaacson, C.P. Yuan and F. Yuan, Universal Non-perturbative Functions for SIDIS and Drell-Yan Processes, arXiv:1406.3073 [INSPIRE]. [97] P. Schweitzer, T. Teckentrup and A. Metz, Intrinsic transverse parton momenta in deeply inelastic reactions, Phys. Rev. D 81 (2010) 094019 [arXiv:1003.2190] [INSPIRE]. [98] A.D. Martin, W.J. Stirling, R.S. Thorne and G. Watt, Parton distributions for the LHC, Eur. Phys. J. C 63 (2009) 189 [arXiv:0901.0002] [INSPIRE]. [99] J.F. Owens, A. Accardi and W. Melnitchouk, Global parton distributions with nuclear and finite-Q2 corrections, Phys. Rev. D 87 (2013) 094012 [arXiv:1212.1702] [INSPIRE]. [100] F. Hautmann et al., TMDlib and TMDplotter: library and plotting tools for transverse-momentum-dependent parton distributions, Eur. Phys. J. C 74 (2014) 3220 [arXiv:1408.3015] [INSPIRE]. HJEP06(217)8 [arXiv:1202.6585] [INSPIRE]. gluon TMDPDFs and the Higgs qT -distribution, JHEP 07 (2015) 158 [arXiv:1502.05354] gluons and photons in unpolarized collider experiments, JHEP 10 (2013) 024 [arXiv:1307.3417] [INSPIRE]. Photon Pair Production, Phys. Rev. Lett. 107 (2011) 062001 [arXiv:1103.3861] [INSPIRE]. LHCb and AFTER, Phys. Rev. D 86 (2012) 094007 [arXiv:1208.3642] [INSPIRE]. Dynamics and Polarization of Gluons inside the Proton at the LHC, Phys. Rev. Lett. 112 (2014) 212001 [arXiv:1401.7611] [INSPIRE]. (J= ) at the LHC as a probe of gluon transverse momentum dependent distributions, Nucl. Phys. B 920 (2017) 192 [arXiv:1702.00305] [INSPIRE]. Eur. Phys. J. A 52 (2016) 156 [arXiv:1512.05379] [INSPIRE]. Eur. Phys. J. A 52 (2016) 154 [arXiv:1511.06924] [INSPIRE]. [1] J.C. Collins , D.E. Soper and G.F. Sterman , Factorization of Hard Processes in QCD , Adv. Ser. Direct. High Energy Phys. 5 ( 1989 ) 1 [hep- ph/0409313] [INSPIRE]. [2] J. Collins , Foundations of perturbative QCD , Cambridge University Press ( 2011 ) Drell-Yan data , Phys. Rev. D 81 ( 2010 ) 034023 [arXiv: 0912 . 2031 ] [INSPIRE]. [18] V. Barone , M. Boglione , J.O. Gonzalez Hernandez and S. Melis , Phenomenological analysis [20] M. Anselmino , M. Boglione , U. D'Alesio , S. Melis , F. Murgia and A. Prokudin , [36] M.G. Echevarria , A. Idilbi , A. Schäfer and I. Scimemi , Model-Independent Evolution of [39] M. Boglione , J. Collins, L. Gamberg , J.O. Gonzalez-Hernandez , T.C. Rogers and N. Sato , QT distributions, Phys. Rev. D 63 ( 2001 ) 114011 [ hep -ph/0012348] [INSPIRE]. [52] C.T.H. Davies and W.J. Stirling , Nonleading Corrections to the Drell-Yan Cross-Section at [53] S. Frixione , P. Nason and G. Ridolfi, Problems in the resummation of soft gluon effects in [54] G. Bozzi , S. Catani , D. de Florian and M. Grazzini , Transverse-momentum resummation [65] D. de Florian , R. Sassot and M. Stratmann , Global analysis of fragmentation functions for pions and kaons and their uncertainties , Phys. Rev. D 75 ( 2007 ) 114010 [ hep -ph/0703242] [66] D. de Florian , M. Epele , R.J. Hernandez-Pinto , R. Sassot and M. Stratmann , [71] M. Burkardt and B. Pasquini , Modelling the nucleon structure , Eur. Phys. J. A 52 ( 2016 ) [86] S. Catani , L. Cieri , G. Ferrera , D. de Florian and M. Grazzini , Vector boson production at [104] D. Boer , S.J. Brodsky , P.J. Mulders and C. Pisano , Direct Probes of Linearly Polarized Gluons inside Unpolarized Hadrons , Phys. Rev. Lett . 106 ( 2011 ) 132001 [arXiv: 1011 .4225] [101] J. Dudek et al., Physics Opportunities with the 12 GeV Upgrade at Jefferson Lab, Eur. Phys. J. A 48 ( 2012 ) 187 [arXiv: 1208 .1244] [INSPIRE]. [102] S.J. Brodsky , F. Fleuret , C. Hadjidakis and J.P. Lansberg , Physics Opportunities of a Fixed-Target Experiment using the LHC Beams , Phys. Rept . 522 ( 2013 ) 239 [103] M.G. Echevarria , T. Kasemets , P.J. Mulders and C. Pisano , QCD evolution of (un )polarized [105] C. Pisano , D. Boer , S.J. Brodsky , M.G.A. Buffing and P.J. Mulders , Linear polarization of [106] J.-W. Qiu , M. Schlegel and W. Vogelsang , Probing Gluonic Spin-Orbit Correlations in [109] A. Signori , Gluon TMDs in quarkonium production, Few Body Syst . 57 ( 2016 ) 651 [110] J.-P. Lansberg , C. Pisano and M. Schlegel , Associated production of a dilepton and a

This is a preview of a remote PDF: https://link.springer.com/content/pdf/10.1007%2FJHEP06%282017%29081.pdf

Alessandro Bacchetta, Filippo Delcarro, Cristian Pisano, Marco Radici, Andrea Signori. Extraction of partonic transverse momentum distributions from semi-inclusive deep-inelastic scattering, Drell-Yan and Z-boson production, Journal of High Energy Physics, 2017, 81, DOI: 10.1007/JHEP06(2017)081