IR-Improved DGLAP-CS Theory

Abstract

We show that it is possible to improve the infrared aspects of the standard treatment of the DGLAP-CS evolution theory to take into account a large class of higher-order corrections that significantly improve the precision of the theory for any given level of fixed-order calculation of its respective kernels. We illustrate the size of the effects we resum using the moments of the parton distributions.

---

In the preparation of the physics for the precision QCD × EW (electroweak) [1–10] LHC physics studies, all aspects of the calculation of the cross sections and distributions for the would-be physical observables must be re-examined if precision tags such as that envisioned for the luminosity theoretical precision are to be realized, that is, 1% cross section predictions for single heavy gauge boson production in 14 TeV pp collisions when that heavy gauge boson decays into a light lepton pair. The QCD [11–21] evolution of the structure functions from the typical reference scale of data input, 𝜇₀∼1–2 GeV, to the respective hard scale is one step that warrants further study, as it is well-known to many. Many authors [22–25] have provided excellent realizations of this evolution in the recent literature. Here, we will re-examine the infrared aspects of the basic evolution theory itself as it is represented via the approach of [17–21] to that theory to try to improve the treatment to a level consistent with the new era of precision QCD × EW physics needed for the LHC physics objectives.

Throughout the discussion, then, we work in the parton model; and we focus on the kernels of what in the literature are commonly referred to as the DGLAP [17–21] evolution equations for the respective parton distributions. These equations, under Mellin transformation, are entirely implied by those of the Callan-Symanzik-type [11–13] analyzed in [15, 16] in their classic analysis of the deep inelastic scattering processes. Thus, henceforward, we will refer to these equations as the DGLAP-CS equations.

Specifically, the motivation for the improvement which we develop can be seen already in the basic results in [17–21] for the kernels that determine the evolution of the structure functions by the attendant DGLAP-CS evolution of the corresponding parton densities by the standard methodology. Consider the evolution of the non-singlet (NS) parton density function 𝑞^NS(𝑥), where 𝑥 can be identified as Bjorken's variable as usual. The basic starting point of our analysis is the infrared divergence in the kernel that determines this evolution:𝑑𝑞^NS(𝑥,𝑡)=𝛼𝑑𝑡^𝑠(𝑡)2𝜋¹_𝑥𝑑𝑦𝑦𝑞^NS(𝑦,𝑡)𝑃_𝑞𝑞(𝑥/𝑦), (1) where the well-known result for the kernel 𝑃_𝑞𝑞(𝑧) is, for 𝑧<1,𝑃_𝑞𝑞(𝑧)=𝐶_𝐹1+𝑧²1−𝑧 (2)when we set 𝑡=ln𝜇²/𝜇²₀ for some reference scale 𝜇₀ with which we study evolution to the scale of interest 𝜇. (We will generally follow [26] and set 𝜇₀=Λ_QCD without loss of content since 𝑑𝑡=𝑑𝑡^′ when 𝑡=ln𝜇²/Λ²_QCD, 𝑡^′=ln𝜇²/𝜇²₀) for fixed values of Λ_QCD, 𝜇₀.) Here, 𝐶_𝐹=(𝑁²_𝑐−1)/(2𝑁_𝑐) is the quark color representation's quadratic Casimir invariant, where 𝑁_𝑐 is the number of colors and so that it is just 3. This kernel has an unintegrable IR singularity at 𝑧=1, which is the point of zero energy gluon emission; and this is as it should be. The standard treatment of this very physical effect is to regularize it by the replacement1→1(1−𝑧)(1−𝑧)₊ (3)with the distribution 1/(1−𝑧)₊ defined so that for any suitable test function 𝑓(𝑧) we have¹₀𝑑𝑧𝑓(𝑧)(1−𝑧)₊=¹₀𝑑𝑧𝑓(𝑧)−𝑓(1)(1−𝑧). (4)A possible representation of 1/(1−𝑧)₊ is seen to be1(1−𝑧)₊=1(1−𝑧)𝜃(1−𝜖−𝑧)+ln𝜖𝛿(1−𝑧) (5)with the understanding that 𝜖↓0. We use the notation 𝜃(𝑥) for the step function from 0 for 𝑥<0 to 1 for 𝑥≥0 and 𝛿(𝑥) is Dirac's delta function. The final result for 𝑃_𝑞𝑞(𝑧) is then obtained by imposing the physical requirement [17–21] that¹₀𝑑𝑧𝑃_𝑞𝑞(𝑧)=0, (6)which is satisfied by adding the effects of virtual corrections at 𝑧=1 so that finally𝑃_𝑞𝑞(𝑧)=𝐶_𝐹(1+𝑧²(1−𝑧)₊+32𝛿(1−𝑧)). (7)Note that we can write the last result as𝑃_𝑞𝑞(𝑧)=𝐶_𝐹(1+𝑧²)1−𝑧₊ (8)from which it follows that (6) holds identically.

The smooth behavior in the original real emission result from the Feynman rules, with a divergent 1/(1−𝑧) behavior as 𝑧→1, has been replaced with a mathematical artifact: the regime 1−𝜖<𝑧<1 now has no probability at all; and at 𝑧=1 we have a large negative integrable contribution so that we end-up finally with a finite (zero) value for the total integral of 𝑃_𝑞𝑞(𝑧). This mathematical artifact is what we wish to improve here; for, in the precision studies of 𝑍 physics [27–32] at LEP1, it has been found that such mathematical artifacts can indeed impair the precision tag which one can achieve with a given fixed order of perturbation theory. An analogous case is now well-known in the theory of QCD higher-order corrections, where the FNAL data on 𝑝_𝑇 spectra clearly show the need for improvement of fixed-order results by resumming large logs associated with soft gluons [33, 34]. For reference, note that at the LHC, 2 TeV partons are realistic so that 𝑧≅0.001 means ∼2–3 GeV soft gluons, which are clearly above the LHC detector thresholds (here we intend the combined effect of such gluons), in complete analogy with the situation at LEP where 𝑧≅0.001 meant ∼100 MeV photons which were also above the LEP detector thresholds—just as resummation was necessary to describe this view of the LEP data, so too we may argue it will be necessary to describe the LHC data on the corresponding view; more importantly, why should we have to set 𝑃_𝑞𝑞(𝑧) to 0 for 1−𝜖<𝑧<1 when it actually has its largest values in this very regime?

By mathematical artifact we do not mean that there is an error in the computations that lead to it; indeed, it is well-known that this +-function behavior is exactly what one gets at 𝒪(𝛼_𝑠) for the bremsstrahlung process. The artifact is that the behavior of the differential spectrum of the process for 𝑧→1 in 𝒪(𝛼_𝑠) is unintegrable and has to be cut-off; and thus this spectrum is only poorly represented by the 𝒪(𝛼_𝑠) calculation; for, the resummation of the large soft higher-order effects as we present below changes the 𝑧→1 behavior nontrivially, as from our resummation we will find that the 1/(1−𝑧)- behavior is modified to (1−𝑧)^𝛾−1, 𝛾>0. This is a testable effect, as we have seen in its QED analogs in 𝑍 physics at LEP1 [27–32]: if the experimentalist measures the cross section for bremsstrahlung for gluons (photons) down to energy fraction 𝜖₀, 𝜖₀>0, in our new resummed theory presented below, the result will approach a finite value from below as −𝜖^𝛾₀ whereas the 𝒪(𝛼_𝑠) +-function prediction would increase without limit as −ln𝜖₀. The exponentiated result has been verified by the data at LEP1.

To illustrate the issue, consider the QED example of the Bonneau-Martin cross section formula for the process 𝑒⁺(𝑝₁)+𝑒⁻(𝑝₂)→𝜇⁺(𝑞₁)+𝜇⁻(𝑞₂)(+𝛾(𝑘)):𝜎=¹₀𝑑𝑣𝜌_BM(𝑣)𝜎^Born(𝑠(1−𝑣)), (9)where𝜌_BM(𝑣)=𝛿(𝑣)(1+𝛾ln𝜖+Δ⁽¹⁾_𝑉𝑆)+𝛾𝜃(𝑣−𝜖)1+(1−𝑣)²,2𝑣𝛾=2𝛼𝜋𝑠(𝐿−1),𝐿=ln𝑚²_𝑒,Δ⁽¹⁾_𝑉𝑆𝛼=(𝜋3)(21𝐿−2+3𝜋²) (10)and for the single photon emission 𝑣=2𝑘⁰/√𝑠 in the center of momentum system (cms), with 𝑠=(𝑝₁+𝑝₂)² as usual. The parameter 𝜖↓0 then defines the +-distribution in the single photon emission𝛾1+(1−𝑣)²2𝑣₊ (11)just as we have indicated above for the single gluon emission in 𝑃_𝑞𝑞. The result (9) is inadequate for precision LEP physics and has to be replaced with an exponentiated formula such as that obtained from substituting [28–32] 𝜌_BM(𝑣)→𝜌_YFS(𝑣) with𝜌_YFS(𝑣)=𝐹_YFS(𝛾)𝑒^𝛿YFS𝛾𝑣^𝛾−1(1+Δ_𝑆⁽¹⁾+Δ_𝐻⁽¹⁾Δ(𝑣)),_𝑆(1)=𝛼𝜋Δ(𝐿−1),_𝐻(1)1(𝑣)=𝑣(−1+2𝛿𝑣),_YFS𝛼=(𝜋1)(21𝐿−1+3𝜋²𝐹),_YFS𝑒(𝛾)=^−𝐶𝛾,Γ(1+𝛾)𝐶=0.57721566…, (12)where 𝐶 is Euler's constant and Γ(𝑤) is Euler's gamma function. (See [27] for a complete discussion of all available variants of this substitution.) We see as advertised that the +-function has been replaced with an integrable function in 𝑣 for 𝑣↓0. See [27] for more discussion of this phenomenology.

The important point is that the traditional resummations in 𝑁-moment space for the DGLAP-CS kernels address only the short-distance contributions to their higher-order corrections. The deep question we deal with in this paper concerns, then, how much of the complete soft limit of the DGLAP-CS kernels is contained in the anomalous dimensions of the leading twist operators in Wilson's expansion, an expansion which resides on the very tip of the light-cone? Are all of the effects of the very soft gluon emission, involving, as they most certainly do, arbitrarily long wavelength quanta, representable by the physics at the tip of the light-cone? The Heisenberg uncertainty principle surely tells us that answer cannot be affirmative. In this paper, we calculate these long-wavelength gluon effects on the DGLAP-CS kernels that are not included (see the discussion below) in the standard treatment of Wilson's expansion. We therefore do not contradict the results of the large 𝑁-moment space resummations such as that presented in [35] nor do we contradict the renormalon chain-type resummation as done in [36].

We employ the exact rearrangement of the Feynman series for QCD as it has been shown in [37–48]. For completeness, as this QCD exponentiation theory is not generally familiar, we reproduce its essential aspects in our appendix. The idea is to sum up the leading IR terms in the corrections to 𝑃_𝑞𝑞 with the goal that they will render integrable the IR singularity that we have in its lowest-order form. This will remove the need for mathematical artifacts and exhibit more accurately the true predictions of the full QCD theory in terms of fully physical results.

As we explain in detail in the appendix for the specific example of 𝑄^′𝑄→𝑄^′′′𝑄^′′+𝑛𝐺, if ℳ^(𝑛) is the amplitude for any process 𝐴→𝐵+𝑛𝐺, the application of amplitude-based resummation as derived in [37–48] leads to the exact resultℳ^(𝑛)=𝑒^{𝛼𝑠𝐵QCD∞}_𝑗=0m_𝑗^(𝑛), (13)where we have defined𝛼_𝑠(𝑄)𝐵_QCD=𝑑⁴𝑘(𝑘²−𝜆²𝑆+𝑖𝜖)^QCD(𝑘), (14)where the amplitudes {m_𝑗^(𝑛)} are free of the IR singularities that are contained in the virtual IR function 𝐵_QCD. Here, 𝑗 is the loop index and the virtual IR emission function 𝑆_QCD(𝑘) is defined in the appendix. Upon squaring the amplitude in (13) and using the standard methods, we get the cross section representation, specializing to 𝐴=𝑄^′(𝑝₁)𝑄(𝑞₁), 𝐵=𝑄^′′′(𝑝₂)𝑄^′′(𝑞₂) for definiteness:𝑑𝜎^𝑛=𝑒^{2𝛼𝑠Re𝐵QCD}𝑛!^𝑛_𝑚=1𝑑³𝑘_𝑚(𝑘²_𝑚+𝜆²)^1/2𝛿(𝑝₁+𝑞₁−𝑝₂−𝑞₂−^𝑛_𝑖=1𝑘_𝑖)×𝜌^(𝑛)(𝑝₁,𝑞₁,𝑝₂,𝑞₂,𝑘₁,…,𝑘_𝑛)𝑑³𝑝₂𝑑³𝑞₂𝑝⁰₂𝑞⁰₂, (15)where we have defined𝜌^(𝑛)(𝑝₁,𝑞₁,𝑝₂,𝑞₂,𝑘₁,…,𝑘_𝑛)=_color,spin‖^∞_𝑗=0m_𝑗^(𝑛)‖² (16)in the incoming 𝑄(𝑞₁)𝑄^′(𝑝₁) cms system and 𝜆 is an IR regulator mass only (it is not a parameter in the Lagrangian)—see the appendix for more details (Some kinematical factors are absorbed into the normalization of the amplitudes.) We show in the appendix that, upon summing over 𝑛, we can extract the dominant real emission contributions from the 𝜌^(𝑛) to arrive at the master formula𝑑𝜎_exp=_𝑛𝑑𝜎^𝑛=𝑒^ΣIR(QCD)∞_𝑛=01𝑛!^𝑛_𝑗=1𝑑³𝑘_𝑗𝑘_𝑗𝑑⁴𝑦(2𝜋)⁴𝑒^{𝑖𝑦⋅(𝑝1+𝑞1−𝑝2−𝑞2−∑𝑘𝑗)+𝐷QCD}∗𝛽^{0𝑥002𝑑𝑐}_𝑛(𝑘₁,…,𝑘_𝑛)𝑑³𝑝₂𝑝⁰₂𝑑³𝑞₂𝑞⁰₂, (17) where now the hard gluon residuals ^0𝑥0𝛽^02𝑑𝑐_𝑛(𝑘₁,…,𝑘_𝑛) are defined in the appendix and are free of IR singularities to all orders in 𝛼_𝑠(𝑄), 𝑄 is the relevant hard scale andΣ_IR(QCD)=2𝛼_𝑠Re𝐵_QCD+2𝛼𝐵_𝑠^{0𝑥002𝑑𝑐}_QCD(𝐾_max),2𝛼𝐵_𝑠^{0𝑥002𝑑𝑐}_QCD(𝐾_max𝑑)=∫³𝑘𝑘^{00𝑥002𝑑𝑐}𝑆_QCD(𝑘)𝜃(𝐾_max𝐷−𝑘),_QCD𝑑=∫³𝑘𝑘^{0𝑥002𝑑𝑐}𝑆_QCD(𝑘)[𝑒^{−𝑖𝑦⋅𝑘}−𝜃(𝐾_max−𝑘)], (18)where the real IR function ^0𝑥0𝑆^02𝑑𝑐_QCD(𝑘) is defined in the appendix. Note that (17) is independent of 𝐾_max.

Here, we apply the QCD exponentiation master formula in (17) (see also [37–45]), following the analogous discussion then for QED in [28–32], to the gluon emission transition that corresponds to 𝑃_𝑞𝑞(𝑧), that is, to the squared amplitude for 𝑞→𝑞(𝑧)+𝐺(1−𝑧) so that in the appendix one replaces everywhere the squared amplitudes for the 𝑄^′𝑄→𝑄^′′′𝑄^′′ processes with those for the former one plus its 𝑛𝐺 analoga with the attendant changes in the phase space and kinematics dictated by the standard methods; this implies that in [17, equation (53)] we have from (17) the replacement (see Figure 1)𝑃_𝐵𝐴=𝑃⁰_𝐵𝐴≡12𝑧(1−𝑧)_spins|𝑉_{𝐴→𝐵+𝐶}|²𝑝²_⟂⇒𝑃_𝐵𝐴=12𝑧(1−𝑧)_spins|𝑉_{𝐴→𝐵+𝐶}|²𝑝²_⟂𝑧^𝛾𝑞𝐹_YFS(𝛾_𝑞)𝑒^{(1/2)𝛿𝑞}, (19)where 𝐴=𝑞, 𝐵=𝐺, 𝐶=𝑞, and 𝑉_{𝐴→𝐵+𝐶} is the lowest-order amplitude for 𝑞→𝐺(𝑧)+𝑞(1−𝑧), so that we get the unnormalized exponentiated result𝑃_𝑞𝑞(𝑧)=𝐶_𝐹𝐹_YFS(𝛾_𝑞)𝑒^{(1/2)𝛿𝑞}1+𝑧²1−𝑧(1−𝑧)^𝛾𝑞, (20)where [28–32, 37–48] (note, 𝑡=ln(𝜇²/Λ²_QCD))𝛾_𝑞=𝐶_𝐹𝛼_𝑠𝜋𝑡=4𝐶_𝐹𝛽₀,𝛿_𝑞=𝛾_𝑞2+𝛼_𝑠𝐶_𝐹𝜋(𝜋²3−12), (21)and 𝐹_YFS(𝑧) was already defined. Here,𝛽₀2=11−3𝑛_𝑓, (22)where 𝑛_𝑓 is the number of active quark flavors. The function 𝐹_YFS(𝑧) was already introduced by Yennie et al. [49, 50] in their analysis of the IR behavior of QED. We see immediately, as illustrated above for QED, that the exponentiation has removed the unintegrable IR divergence at 𝑧=1. For reference, we note that we have in (20) resummed the terms 𝒪(ln^𝑘(1−𝑧)𝑡^ℓ𝛼^𝑛_𝑠), 𝑛≥ℓ≥𝑘, which originate in the IR regime and which exponentiate. (Following the standard LEP Yellow Book [27] convention, we do not include the order of the first nonzero term in counting the order of its higher-order corrections.) The important point is that we have not dropped outright the terms that do not exponentiate but have organized them into the residuals ^0𝑥0𝛽^02𝑑𝑐_𝑚 in the analog of (17). The application of (17) to obtain (20) proceeds as follows. First, the exponent in the exponential factor in front of the expression on the RHS of (17) is readily seen to be from (A.16), using the well-known results for the respective real and virtual infrared functions from [37–48]:Σ_IR(QCD)=2𝛼_𝑠Re𝐵_QCD+2𝛼𝐵_𝑠^{0𝑥002𝑑𝑐}_QCD(𝐾_max)=12(2𝐶_𝐹𝛼_𝑠𝜋𝐾𝑡ln^max𝐸+𝐶_𝐹𝛼_𝑠𝛼2𝜋𝑡+^𝑠𝐶_𝐹𝜋(𝜋²3−12)), (23)where on the RHS of the last result we have already applied the DGLAP-CS synthesization procedure in [39] to remove the collinear singularities, lnΛ²_QCD/𝑚²_𝑞−1, in accordance with the standard QCD factorization theorems [51–55]. This means that, identifying the LHS of (17) as the sum over final states and average over initial states of the respective process divided by the incident flux and replacing that incident flux by the respective initial state density according to the standard methods for the process 𝑞→𝑞(1−𝑧)+𝐺(𝑧), occurring in the context of a hard scattering at scale 𝑄 as it is for [17, equation (53)], the soft gluon effects for energy fraction <𝑧≡𝐾_max/𝐸 give the result, from (17), that, working through to the ^0𝑥0𝛽^02𝑑𝑐₁-level and using 𝑞₂ to represent the momentum conservation via other degrees of freedom for the attendant hard process𝛼_𝑠(𝑡)𝑃2𝜋^𝐵𝐴𝑑𝑡𝑑𝑧=𝑒^{ΣIR(QCD)(𝑧)}{𝛽^{0𝑥002𝑑𝑐}₀𝑑⁴𝑦(2𝜋)⁴𝑒^{{𝑖𝑦⋅(𝑝1−𝑝2)+∫𝑘<𝐾max(𝑑3𝑘/𝑘)0𝑥002𝑑𝑐𝑆QCD(𝑘)[𝑒−𝑖𝑦⋅𝑘−1]}}+𝑑³𝑘₁𝑘₁^{0𝑥002𝑑𝑐}𝛽₁(𝑘₁)𝑑⁴𝑦(2𝜋)⁴𝑒^{{𝑖𝑦⋅(𝑝1−𝑝2−𝑘1)+∫𝑘<𝐾max(𝑑3𝑘/𝑘)0𝑥002𝑑𝑐𝑆QCD(𝑘)[𝑒−𝑖𝑦⋅𝑘−1]}+⋯}}𝑑³𝑝₂𝑝⁰₂𝑑³𝑞₂𝑞⁰₂=𝑒^{ΣIR(QCD)(𝑧)}{𝛽^{0𝑥002𝑑𝑐}₀^∞_−∞𝑑𝑦𝑒(2𝜋)^{{𝑖𝑦⋅(𝐸1−𝐸2)+∫𝑘<𝐾max(𝑑3𝑘/𝑘)0𝑥002𝑑𝑐𝑆QCD(𝑘)[𝑒−𝑖𝑦𝑘−1]}}+𝑑³𝑘₁𝑘₁^{0𝑥002𝑑𝑐}𝛽₁(𝑘₁)^∞_−∞𝑑𝑦𝑒(2𝜋)^{{𝑖𝑦⋅(𝐸1−𝐸2−𝑘01)+∫𝑘<𝐾max(𝑑3𝑘/𝑘)0𝑥002𝑑𝑐𝑆QCD(𝑘)[𝑒−𝑖𝑦⋅𝑘−1]}}𝑑+⋯}³𝑝₂𝑝⁰₂𝑞⁰₂, (24)where we set 𝐸_𝑖=𝑝⁰_𝑖, 𝑖=1,2, and the real infrared function ^0𝑥0𝑆^02𝑑𝑐_QCD(𝑘) is well-known as well:^0𝑥0𝑆^02𝑑𝑐_QCD𝛼(𝑘)=−^𝑠𝐶_𝐹8𝜋²(𝑝₁𝑘𝑝₁−𝑝₂𝑘𝑝₂)²|_{DGLAP-CSsynthesized} (25)and we indicate as above that the DGLAP-CS synthesization procedure in [39] is to be applied to its evaluation to remove its collinear singularities; we are using the kinematics of [17] in their computation of 𝑃_𝐵𝐴(𝑧) in their (53), so that the relevant value of 𝑘²_⟂ is indeed 𝑄². It means that the computation can also be seen to correspond to computing the IR function for the standard 𝑡-channel kinematics and taking 1/2 of the result to match the single line emission in 𝑃_𝐺𝑞. The two important integrals needed in (24) were already studied in [49, 50]:𝐼_YFS(𝑧𝐸,0)=^∞_−∞𝑑𝑦𝑒2𝜋^{[𝑖𝑦(𝑧𝐸)+∫𝑘<𝑧𝐸(𝑑3𝑘/𝑘)0𝑥002𝑑𝑐𝑆QCD(𝑘)(𝑒−𝑖𝑦𝑘−1)]}=𝐹_YFS(𝛾_𝑞)𝛾_𝑞,𝐼𝑧𝐸_YFS(𝑧𝐸,𝑘₁)=^∞_−∞𝑑𝑦𝑒2𝜋^{[𝑖𝑦(𝑧𝐸−𝑘1)+∫𝑘<𝑧𝐸(𝑑3𝑘/𝑘)0𝑥002𝑑𝑐𝑆QCD(𝑘)(𝑒−𝑖𝑦𝑘−1)]}=(𝑧𝐸𝑧𝐸−𝑘₁)^1−𝛾𝑞𝐼_YFS(𝑧𝐸,0). (26)

(a)

(b)

Figure 1 

In (a), we show the usual process $q\hspace{0.17em}\to \hspace{0.17em}q(1-z)+G(z)$; in (b), we show its multiple gluon improvement $q\hspace{0.17em}\to \hspace{0.17em}q(1-z)+G_1({\xi }_1)+\cdots +G_n({\xi }_n)$, $z={\sum }_j{\xi }_j$.

When we introduce the results in (26) into (24), we can identify the factor(𝛽^{0𝑥002𝑑𝑐}₀𝛾_𝑞+𝑧𝐸𝑑𝑘₁𝑘₁𝑑Ω𝛽₁^{0𝑥002𝑑𝑐}₁(𝑘₁)(𝑧𝐸𝑧𝐸−𝑘₁)^1−𝛾𝑞𝛾_𝑞)𝑑𝑧𝐸³𝑝₂𝐸₂𝑞⁰₂𝛼=𝑑𝑡^𝑠(𝑡)𝑃2𝜋^0𝐵𝐴𝑑𝑧+𝒪(𝛼²_𝑠), (27)where 𝑃⁰_𝐵𝐴 is the unexponentiated result in the first line of (19). This leads us finally to the exponentiated result in the second line of (19) by elementary differentiation:𝑃_𝐵𝐴=𝑃⁰_𝐵𝐴𝑧^𝛾𝑞𝐹_YFS(𝛾_𝑞)𝑒^{(1/2)𝛿𝑞}. (28)

The following observations are in order. First, unlike the light-cone gauge or light-like Wilson line singularity artifacts discussed in [56] for unintegrated definitions of parton density functions, the analyses just presented, both for the QED case and for the QCD case, show that the 1/(1−𝑧) real emission singularity in 𝑃_𝑞𝑞 (it would be in 𝑃_𝑒𝑒 in the analog QED case) is a genuine property of soft radiation, it is gauge invariant. Second, from the explicit results for the exponent Σ_IR(QCD) in (23) and the results in (18), we see that the gluon mass regulator 𝜆 has completely canceled from our cross section, which is also then gauge invariant because we never introduced 𝜆 into the QCD Lagrangian—we only used 𝜆 to define IR singularities so that the Slavnov-Taylor, Ward-Takahashi identities were all the time maintained. Use of the 𝑛-dimensional regulator methods of [57, 58] gives the same results as our use of 𝜆.

Here, we also may note how one can see that the terms we exponentiate are not included in the standard treatment of Wilson's expansion: from the standard methods [59, 60], the 𝑁th moment of the invariants 𝑇_𝑖,ℓ, 𝑖=𝐿,2,3, ℓ=𝑞,𝐺, of the forward Compton amplitude in DIS, where we recall the structure functions 𝐹_𝑖, 𝑖=𝐿,2,3, satisfy (1/2𝜋)ℑ𝑇_𝑖=(1/2𝑥_𝐵𝑗)𝐹_𝑖, 𝑖=2, 𝐿, (1/2𝜋)ℑ𝑇₃=𝐹₃, is projected by𝒫_𝑁𝑞≡[^{𝜇1⋯𝑞^𝜇𝑁}𝜕𝑁!𝜕𝑝^𝜇1⋯^𝑁𝜕𝑝^𝜇𝑁]|_𝑝=0 (29)with 𝑥_𝐵𝑗=𝑄²/(2𝑞𝑝) in the standard DIS notation; this projects the coefficient of 1/(2𝑥_𝐵𝑗)^𝑁. For the dominant terms which we resum here, the characteristic behavior would correspond formally to 𝛾_𝑞-dependent anomalous dimensions associated with the respective coefficient whereas by definition Wilson's expansion does not contain such. In more phenomenologically familiar language, it is well-known that the parton model used in this paper to calculate the large distance effects that improve the kernels contains such effects whereas Wilson's expansion does not; for example, the parton model can be used for Drell-Yan processes, whereas Wilson's expansion cannot. Similarly, any Wilson-expansion guided procedure used to infer the kernels via inverse Mellin transformation, by calculating the coefficient of (1/𝑧)^𝑛 in Wilson's expansion, will necessarily omit the dominant IR terms which we resum. Here, we stress that we refer to the properties of the expansion of the invariant functions 𝑇_𝑖, not to the expansion of the kernels themselves, as the latter are related to the respective anomalous dimension matrix elements by inverse Mellin transformations.

The normalization condition in (6) then gives us the final expression𝑃_𝑞𝑞(𝑧)=𝐶_𝐹𝐹_YFS(𝛾_𝑞)𝑒^{(1/2)𝛿𝑞}[1+𝑧²1−𝑧(1−𝑧)^𝛾𝑞−𝑓_𝑞(𝛾_𝑞)𝛿(1−𝑧)], (30)where𝑓_𝑞(𝛾_𝑞2)=𝛾_𝑞−2𝛾_𝑞+1+1𝛾_𝑞+2. (31)The latter result is then our IR-improved kernel for NS DGLAP-CS evolution in QCD. We note that the appearance of the integrable function (1−𝑧)^{−1+𝛾𝑞} in the place of 1/(1−𝑧)₊ was already anticipated by Gribov and Lipatov in [18–21]. Here, we have calculated the value of 𝛾_𝑞 in a systematic rearrangement of the QCD perturbation theory that allows one to work to any exact order in the theory without dropping any part of the theory's perturbation series.

The standard DGLAP-CS theory tells us that the kernel 𝑃_𝐺𝑞(𝑧) is related to 𝑃_𝑞𝑞(1−𝑧) directly: for 𝑧<1, we have𝑃_𝐺𝑞(𝑧)=𝑃_𝑞𝑞(1−𝑧)=𝐶_𝐹𝐹_YFS(𝛾_𝑞)𝑒^{(1/2)𝛿𝑞}1+(1−𝑧)²𝑧𝑧^𝛾𝑞. (32)This then brings us to our first nontrivial check of the new IR-improved theory; for, the conservation of momentum tells us that¹₀𝑑𝑧𝑧(𝑃_𝐺𝑞(𝑧)+𝑃_𝑞𝑞(𝑧))=0. (33)In view of new results in (30), (32), we note that, for any 𝑃_𝑞𝑞(𝑧) which satisfies the normalization condition (6) and which is related to 𝑃_𝐺𝑞(𝑧) via the relation𝑃_𝐺𝑞(𝑧)=𝑃_𝑞𝑞(1−𝑧), (34)𝑧<1, we have the following result:𝐼=¹₀𝑑𝑧𝑧(𝑃_𝑞𝑞(𝑧)+𝑃_𝐺𝑞=(𝑧))₁₀𝑑𝑧(((𝑧−1)+1)𝑃_𝑞𝑞(𝑧)+𝑧𝑃_𝑞𝑞=(1−𝑧))₁₀𝑑𝑧[−(1−𝑧)𝑃_𝑞𝑞(𝑧)+𝑧𝑃_𝑞𝑞(1−𝑧)+𝑃_𝑞𝑞(𝑧)]. (35)The integral of the first term in square brackets on the RHS of this last equation is transformed to the negative of the integral of the second one by the change of variable 𝑧→1−𝑧 so that they exactly cancel while the third term integrates to zero by the normalization condition (6). Thus𝐼=0 (36)and the quark momentum sum rule is satisfied. Since our new results for 𝑃_𝑞𝑞, 𝑃_𝐺𝑞 satisfy the conditions for 𝐼=0, we conclude that the quark momentum sum rule holds for them as well.

Having improved the IR divergence properties of 𝑃_𝑞𝑞(𝑧) and 𝑃_𝐺𝑞(𝑧), we now turn to 𝑃_𝐺𝐺(𝑧) and 𝑃_𝑞𝐺(𝑧). We first note that the standard formula for 𝑃_𝑞𝐺(𝑧),𝑃_𝑞𝐺1(𝑧)=2(𝑧²+(1−𝑧)²), (37)is already well-behaved (integrable) in the IR regime. Thus, we do not need to improve it here to make it integrable; and we note that the singular contributions in the other kernels are expected to dominate the evolution effects in any case. We do not exclude improving it for the best precision [61] and we return to this point presently.

This brings us then to 𝑃_𝐺𝐺(𝑧). Its lowest-order form is𝑃_𝐺𝐺(𝑧)=2𝐶_𝐺(1−𝑧𝑧+𝑧1−𝑧+𝑧(1−𝑧)) (38)which again exhibits unintegrable IR singularities at both 𝑧=1 and 𝑧=0. (Here, 𝐶_𝐺 is the gluon quadratic Casimir invariant, so that it is just 𝑁_𝑐=3.) If we repeat the QCD exponentiation calculation carried-out above by using the color representation for the gluon rather than that for the quarks, that is, if we apply the exponentiation analysis in the appendix to the squared amplitude for the process 𝐺→𝐺(𝑧)+𝐺(1−𝑧), we get the exponentiated unnormalized result 𝑃_𝐺𝐺(𝑧)=2𝐶_𝐺𝐹_YFS(𝛾_𝐺)𝑒^{(1/2)𝛿𝐺}×(1−𝑧𝑧𝑧^𝛾𝐺+𝑧1−𝑧(1−𝑧)^𝛾𝐺+12(𝑧^1+𝛾𝐺(1−𝑧)+𝑧(1−𝑧)^1+𝛾𝐺)) (39)wherein we obtain the 𝛾_𝐺 and 𝛿_𝐺 from the expressions for 𝛾_𝑞 and 𝛿_𝑞 by the substitution 𝐶_𝐹→𝐶_𝐺:𝛾_𝐺=𝐶_𝐺𝛼_𝑠𝜋𝑡=4𝐶_𝐺𝛽₀,𝛿_𝐺=𝛾_𝐺2+𝛼_𝑠𝐶_𝐺𝜋(𝜋²3−12). (40)We see again that exponentiation has again made the singularities at 𝑧=1 and 𝑧=0 integrable.

To normalize 𝑃_𝐺𝐺, we take into account the virtual corrections such that the gluon momentum sum rule¹₀𝑑𝑧𝑧(2𝑛_𝑓𝑃_𝑞𝐺(𝑧)+𝑃_𝐺𝐺(𝑧))=0 (41)is satisfied. This gives us finally the IR-improved result𝑃_𝐺𝐺(𝑧)=2𝐶_𝐺𝐹_YFS(𝛾_𝐺)𝑒^{(1/2)𝛿𝐺}×{1−𝑧𝑧𝑧^𝛾𝐺+𝑧1−𝑧(1−𝑧)^𝛾𝐺+12(𝑧^1+𝛾𝐺(1−𝑧)+𝑧(1−𝑧)^1+𝛾𝐺)−𝑓_𝐺(𝛾_𝐺)𝛿(1−𝑧)}, (42)where for 𝑓_𝐺(𝛾_𝐺) we get𝑓_𝐺(𝛾_𝐺𝑛)=^𝑓6𝐶_𝐺𝐹_YFS(𝛾_𝐺)𝑒^{−(1/2)𝛿𝐺}+2𝛾_𝐺(1+𝛾_𝐺)(2+𝛾_𝐺)+1(1+𝛾_𝐺)(2+𝛾_𝐺)+12(3+𝛾_𝐺)(4+𝛾_𝐺)+1(2+𝛾_𝐺)(3+𝛾_𝐺)(4+𝛾_𝐺). (43)It is these improved results in (30), (32), (42) for 𝑃_𝑞𝑞(𝑧), 𝑃_𝐺𝑞(𝑧), and 𝑃_𝐺𝐺(𝑧) that we use together with the standard result in (38) for 𝑃_𝑞𝐺(𝑧) as the IR-improved DGLAP-CS theory.

For clarity we summarize at this point the new IR-improved kernel set as follows:𝑃^exp_𝑞𝑞(𝑧)=𝐶_𝐹𝐹_YFS(𝛾_𝑞)𝑒^{(1/2)𝛿𝑞}[1+𝑧²1−𝑧(1−𝑧)^𝛾𝑞−𝑓_𝑞(𝛾_𝑞𝑃)𝛿(1−𝑧)],_exp𝐺𝑞(𝑧)=𝐶_𝐹𝐹_YFS(𝛾_𝑞)𝑒^{(1/2)𝛿𝑞}1+(1−𝑧)²𝑧𝑧^𝛾𝑞,𝑃^exp_𝐺𝐺(𝑧)=2𝐶_𝐺𝐹_YFS(𝛾_𝐺)𝑒^{(1/2)𝛿𝐺}×{1−𝑧𝑧𝑧^𝛾𝐺+𝑧1−𝑧(1−𝑧)^𝛾𝐺+12(𝑧^1+𝛾𝐺(1−𝑧)+𝑧(1−𝑧)^1+𝛾𝐺)−𝑓_𝐺(𝛾_𝐺𝑃)𝛿(1−𝑧)},_𝑞𝐺1(𝑧)=2(𝑧²+(1−𝑧)²), (44)(45)(46)(47)where we have introduced the superscript exp to denote the exponentiated results henceforth.

Returning now to the improvement of 𝑃_𝑞𝐺(𝑧), let us record it as well for the sake of completeness and of providing better precision. Applying (17) to the process 𝐺→𝑞+𝑞, we get the exponentiated result𝑃^exp_𝑞𝐺(𝑧)=𝐹_YFS(𝛾_𝐺)𝑒^{(1/2)𝛿𝐺}12{𝑧²(1−𝑧)^𝛾𝐺+(1−𝑧)²𝑧^𝛾𝐺}. (48)The gluon momentum sum rule then gives the new normalization constant for the 𝑃^exp_𝐺𝐺 via the result𝑓_𝐺(𝛾_𝐺𝑛)=^𝑓𝐶_𝐺1(1+𝛾_𝐺)(2+𝛾_𝐺)(3+𝛾_𝐺)+2𝛾_𝐺(1+𝛾_𝐺)(2+𝛾_𝐺)+1(1+𝛾_𝐺)(2+𝛾_𝐺)+12(3+𝛾_𝐺)(4+𝛾_𝐺)+1(2+𝛾_𝐺)(3+𝛾_𝐺)(4+𝛾_𝐺). (49)The constant 𝑓_𝐺 should be substituted for 𝑓_𝐺 in 𝑃^exp_𝐺𝐺 whenever the exponentiated result in (48) is used. These results, (47); (48); and (49), are our new improved DGLAP-CS kernel set, with the option exponentiating 𝑃_𝑞𝐺 as well. Let us now look into their effects on the moments of the structure functions by discussing the corresponding effects on the moments of the parton distributions.

We know that moments of the kernels determine the exponents in the logarithmic variation [15–21] of the moments of the quark distributions and, thereby, of the moments of the structure functions themselves. To wit, in the nonsinglet case, we have𝑑𝑀^NS_𝑛(𝑡)=𝛼𝑑𝑡^𝑠(𝑡)𝐴2𝜋^NS𝑛𝑀^NS_𝑛(𝑡), (50)where𝑀^NS_𝑛(𝑡)=¹₀𝑑𝑧𝑧^𝑛−1𝑞^NS(𝑧,𝑡) (51)and the quantity 𝐴^NS_𝑛 is given by𝐴^NS_𝑛=¹₀𝑑𝑧𝑧^𝑛−1𝑃^exp_𝑞𝑞(𝑧)=𝐶_𝐹𝐹_YFS(𝛾_𝑞)𝑒^{(1/2)𝛿𝑞}[𝐵(𝑛,𝛾_𝑞)+𝐵(𝑛+2,𝛾_𝑞)−𝑓_𝑞(𝛾_𝑞)] (52)where 𝐵(𝑥,𝑦) is the beta function given by𝐵(𝑥,𝑦)=Γ(𝑥)Γ(𝑦)/Γ(𝑥+𝑦). (53)This should be compared to the un-IR-improved result [15–21]:𝐴^NS𝑜_𝑛≡𝐶_𝐹1[−2+1𝑛(𝑛+1)−2^𝑛_𝑗=21𝑗]. (54)The asymptotic behavior for large 𝑛 is now very different, as the IR-improved exponent approaches a constant, a multiple of −𝑓_𝑞, as we would expect as 𝑛→∞ because lim_𝑛→∞𝑧^𝑛−1=0 for 0≤𝑧<1 whereas, as it is well-known, the un-IR-improved result in (54) diverges as −2𝐶_𝐹ln𝑛 as 𝑛→∞. The two results are also different at finite 𝑛: for 𝑛=2 we get, for example, for 𝛼_𝑠≅.118 [62],𝐴^NS₂=𝐶_𝐹𝐶(−1.33),un-IR-improved,_𝐹(−0.966),IR-improved, (55)so that the effects we have calculated are important for all 𝑛 in general. For completeness, we note that the solution to (50) is given by the standard methods as𝑀^NS_𝑛(𝑡)=𝑀^NS_𝑛(𝑡₀)𝑒^{∫𝑡𝑡0𝑑𝑡′(𝛼𝑠(𝑡′)/2𝜋)𝐴NS𝑛(𝑡′)}=𝑀^NS_𝑛(𝑡₀)𝑒^{𝑎𝑛[𝐸𝑖((1/2)𝛿1𝛼𝑠(𝑡0))−𝐸𝑖((1/2)𝛿1𝛼𝑠(𝑡))⇒0𝑡,𝑡largewith0𝑡≫𝑡𝑀NS𝑛(𝑡0)(}𝛼_𝑠(𝑡₀)𝛼_𝑠(𝑡)^{)𝑎′𝑛,]} (56)where 𝐸𝑖(𝑥)=∫^𝑥_−∞𝑑𝑟𝑒^𝑟/𝑟 is the exponential integral function,𝑎_𝑛=2𝐶_𝐹𝛽₀𝐹_YFS(𝛾_𝑞)𝑒^𝛾𝑞/4[𝐵(𝑛,𝛾_𝑞)+𝐵(𝑛+2,𝛾_𝑞)−𝑓_𝑞(𝛾_𝑞)],𝑎^′_𝑛=𝑎_𝑛𝛿(1+¹2(𝛼_𝑠(𝑡₀)−𝛼_𝑠(𝑡))ln(𝛼_𝑠(𝑡₀)/𝛼_𝑠)(𝑡)) (57)with𝛿₁=𝐶_𝐹𝜋(𝜋²3−12). (58)We can compare with the un-IR-improved result in which the last line in (56) holds exactly with 𝑎^′_𝑛=2𝐴^NS𝑜_𝑛/𝛽₀. Phenomenologically, for 𝑛=2, taking 𝑄₀=2 GeV and evolving to 𝑄=100 GeV, if we set Λ_QCD≅.2GeV and use 𝑛_𝑓=5 for definiteness of illustration, we see from (56), (57) that we get a shift of the respective evolved NS moment by ∼5%, which is of some interest in view of the expected HERA precision [63]. (Although HERA is shutdown, HERA data analysis continues as the H1 and ZEUS combine their data to improve their results accordingly.)

We give now the remaining elements of the anomalous dimension matrix in its “best” IR-improved form for completeness:𝐴_𝑛^𝐺𝑞=¹₀𝑑𝑧𝑧^𝑛−1𝑃^exp_𝐺𝑞(𝑧)=𝐶_𝐹𝐹_YFS(𝛾_𝑞)𝑒^{(1/2)𝛿𝑞}[1𝑛+𝛾_𝑞−1+𝐵(3,𝑛+𝛾_𝑞𝐴−1)],_𝑛𝐺𝐺=¹₀𝑑𝑧𝑧^𝑛−1𝑃^exp_𝐺𝐺(𝑧)=2𝐶_𝐺𝐹_YFS(𝛾_𝐺)𝑒^{(1/2)𝛿𝐺}{𝐵(𝑛+1,𝛾_𝐺)+𝐵(𝑛+𝛾_𝐺+1−1,2)2(𝐵(𝑛+1,𝛾_𝐺+2)+𝐵(𝑛+𝛾_𝐺+1,2))−𝑓(𝛾_𝐺)},2𝑛_𝑓𝐴_𝑛^𝑞𝐺=2𝑛_𝑓¹₀𝑑𝑧𝑧^𝑛−1𝑃^exp_𝑞𝐺(𝑧)=2𝑇(𝐹)𝐹_YFS(𝛾_𝐺)𝑒^{(1/2)𝛿𝐺}(𝐵(𝑛+2,1+𝛾_𝐺)+𝐵(𝑛+𝛾_𝐺,3)), (59)(60)(61)where 𝑇(𝐹)=(1/2)𝑛_𝑓. We note that the unexponentiated value of the last result in (61) is a well-known one [15–21], 2𝑇(𝐹)(2+𝑛+𝑛²)/𝑛(𝑛+1)(𝑛+2), and it would be used whenever we do not choose to exponentiate 𝑃_𝑞𝐺. We will investigate the further implications of these IR-improved results for LHC physics elsewhere [61].

In the discussion so far, we have used the lowest-order DGLAP-CS kernel set to illustrate how important the resummation which we present here can be. In the literature [64–74], there are now exact results up to 𝒪(𝛼³_𝑠) for the DGLAP-CS kernels. The question naturally arises as to the relationship of our work to these fixed-order exact results. We stress first that we are presenting an improvement of the fixed-order results such that the singular pieces of the any exact fixed-order result, that is, the 1/(1−𝑧)₊ parts, are exponentiated so that they are replaced with integrable functions proportional to (1−𝑧)^𝛾−1 with 𝛾 positive as we have illustrated above. Since the series of logs which we resum to accomplish this has the structure 𝛼^ℓ_𝑠𝑡^ℓln^𝑛(1−𝑧), ℓ≥𝑛, these terms are not already present in the results in [64–74]. As we use the formula in (17), there will be no double counting if we implement our IR-improvement of the exact fixed-order results in [64–74]. The detailed discussion of the application of our theory to the results in [64–74] will appear elsewhere [61]. For reference, we note that the higher-order kernel corrections in [64–74] are perturbatively related to the leading-order kernels, so one can expect that the size of the exponentiation effects illustrated above will only be perturbatively modified by the higher-order kernel corrections, leaving the same qualitative behavior in general.

In the interest of specificness, let us illustrate the IR-improvement of 𝑃_𝑞𝑞 when calculated to three loops using the results in [64–74]. Considering the nonsinglet case for definiteness (a similar analysis holds for the singlet case) we write in the notation of the latter references:𝑃⁺_ns=𝑃^𝑣_𝑞𝑞+𝑃^𝑣_𝑞𝑞≡^∞_𝑛=0(𝛼_𝑠)4𝜋^𝑛+1𝑃^(𝑛)+_ns, (62)where at order 𝒪(𝛼_𝑠), we have𝑃⁽⁰⁾⁺_ns(𝑧)=2𝐶_𝐹{1+𝑧²(1−𝑧)₊+32𝛿(1−𝑧)} (63)which shows that 𝑃⁽⁰⁾⁺_ns(𝑧) agrees with the unexponentiated result in (7) for 𝑃_𝑞𝑞 except for an overall factor of 2. We use this latter identification to connect our work with that in [64–74] in the standard methodology. In [64–74], exact results are given for 𝑃⁽¹⁾⁺_ns(𝑧), and in [73, 74] exact results are given for 𝑃⁽²⁾⁺_ns(𝑧). When we apply the result in (17) to the squared amplitudes for the processes 𝑞→𝑞+𝑋, 𝑞→𝑞+𝑋^′, we get the exponentiated result𝑃^+,exp_ns𝛼(𝑧)=(^𝑠4𝜋)2𝑃^exp_𝑞𝑞(𝑧)+𝐹_YFS(𝛾_𝑞)𝑒^{(1/2)𝛿𝑞}𝛼×[(^𝑠)4𝜋²{(1−𝑧)^𝛾𝑞𝑃⁽¹⁾⁺_ns(𝑧)+𝐵₂𝛼𝛿(1−𝑧)}+(^𝑠)4𝜋³{(1−𝑧)^𝛾𝑞𝑃⁽²⁾⁺_ns(𝑧)+𝐵₃𝛿(1−𝑧)}], (64)where 𝑃^exp_𝑞𝑞(𝑧) is given in (47) and the resummed residuals 𝑃^(𝑖)+_ns, 𝑖=1,2, are related to the exact results for 𝑃^(𝑖)+_ns, 𝑖=1,2, as follows:𝑃^(𝑖)+_ns(𝑧)=𝑃^(𝑖)+_ns(𝑧)−𝐵_1+𝑖𝛿(1−𝑧)+Δ^(𝑖)+_ns(𝑧), (65)whereΔ⁽¹⁾⁺_ns(𝑧)=−4𝐶_𝐹𝜋𝛿₁{1+𝑧²1−𝑧−𝑓_𝑞Δ𝛿(1−𝑧)},_(2)+ns(𝑧)=−4𝐶_𝐹(𝜋𝛿₁)²{1+𝑧²1−𝑧−𝑓_𝑞𝛿(1−𝑧)}−2𝜋𝛿₁𝑃⁽¹⁾⁺_ns(𝑧),𝐵₂=𝐵₂+4𝐶_𝐹𝜋𝛿₁𝑓_𝑞,𝐵₃=𝐵₃+4𝐶_𝐹(𝜋𝛿₁)²𝑓_𝑞−2𝜋𝛿₁𝐵₂. (66)(67)Here, the constants 𝐵_𝑖, 𝑖=2,3, are given by the results in [64–74] as𝐵₂=4𝐶_𝐺𝐶_𝐹(17+24113𝜁₂−3𝜁₃)−4𝐶_𝐹𝑛_𝑓(1+2123𝜁₂)+4𝐶²_𝐹(38−3𝜁₂+6𝜁₃𝐵),₃=16𝐶_𝐺𝐶_𝐹𝑛_𝑓(54−167𝜁54²+1𝜁20²²+25𝜁18³)+16𝐶_𝐺𝐶²_𝐹(15164+𝜁₂𝜁₃−205𝜁24²−247𝜁60²²+211𝜁12³+152𝜁₅)+16𝐶²_𝐺𝐶_𝐹(−1657+576281𝜁27²−18𝜁²₂−979𝜁₃+52𝜁₅)+16𝐶_𝐹𝑛²_𝐹(−17+5144𝜁27²−19𝜁₃)+16𝐶²_𝐹𝑛_𝐹(−23+516𝜁12²+29𝜁30²²−176𝜁₃)+16𝐶³_𝐹(2932−2𝜁₂𝜁₃+98𝜁₂+185𝜁²₂+174𝜁₃−15𝜁₅), (68)where 𝜁_𝑛 is the Riemann zeta function evaluated at argument 𝑛. In arriving at the result in (64), we use the fact that the 𝑀𝑆 results for the higher-order kernels do not contain any of the powers of 𝛾_𝑞 that we have resummed, so that the only issue for their improvement is the factor 𝑒^𝛿𝑞/2, which then has to have the coefficients in the results for the higher-order kernels adjusted so that there is no double counting. It is in this way that we have derived the results in (65)–(67). The detailed phenomenological consequences of the fully exponentiated 2- and 3-loop DGLAP-CS kernel set will appear elsewhere [61].

In summary, we have used exact rearrangement of the QCD Feynman series to isolate and resum the leading IR contributions to the physical processes that generate the evolution kernels in DGLAP-CS theory. In this way, we have obviated the need to employ artificial mathematical regularization of the attendant IR singularities as the theory's higher-order corrections naturally tame these singularities. The resulting IR-improved anomalous dimension matrix behaves more physically for large 𝑛 and receives significant effects at finite 𝑛 from the exponentiation.

We in principle can make contact with the moment-space resummation results in [75] but we stress that these results have necessarily been obtained after making a Mellin transform of the mathematical artifact which we address in this paper. Thus, the results in [75] do not in any way contradict the analysis in this paper.

We note that the program of improvement of the hadron cross section calculations for LHC physics advanced herein should be distinguished from the results in [76–78]. Indeed, recalling the standard hadron cross section formula𝜎=_𝑖,𝑗𝑑𝑥₁𝑑𝑥₂𝐹_𝑖(𝑥₁)𝐹_𝑗(𝑥₂)𝜎(𝑥₁𝑥₂𝑠), (69)where {𝐹_ℓ(𝑥)} are the respective parton densities and 𝜎(𝑥₁𝑥₂𝑠) is the respective reduced hard parton cross section, the resummation results in [76–78] address, by summing the large logs in Mellin transform space, the 𝑥₁𝑥₂→1 limit of 𝜎(𝑥₁𝑥₂𝑠) whereas the results above address the improvement, by resummation in 𝑥-space, of the calculation of the parton densities {𝐹_𝑖(𝑥)} for all values of x. Thus, the program of improvement presented above is entirely complementary to that in [76–78] and both programs of improvement are needed for precision LHC physics. The situation can be illustrated by comparing the results in [79] with our results herein. The key observation can already be made from (2.1) in the latter paper, wherein it is made manifest that the resummation carried out therein, as an application of the methods in [76–78], is a resummation for the large 𝑁-Mellin space limit of the Mellin transform of the hard scattering coefficient function so that all of the IR effects in the parton densities are not included in this resummation. What we deal with here is however resummation of the IR effects in the kernels which generate exactly these IR effects in these parton densities directly in configuration space so that we work on a complementary aspect the formula (69) and this we do directly in 𝑥-space rather than in 𝑁-Mellin space. There is then no contradiction or repetition between our results and those in [79].

The usual factorization theorems for mass singularities in QCD are fully consistent with our results: we act on the Feynman series for the hadron-hadron scattering in (69) after the mass singularities have been factorized into the parton densities, as our resummation is multiplicative in character. What one has to note is that, since the methods of [76–78], which are also consistent with the QCD factorization theorems, apply to the hard scattering coefficients, there is always the possibility to use them to improve any hard scattering effect where soft gluons are important. In particular, it is possible to use these methods to resum the soft gluon effects on the hard scattering contribution which one assigns in one's scheme to the kernels for example, as one can see in [79]. The resummation of the effects which we address, involving as they do terms of the form 𝛼^𝑛_𝑠𝑡^𝑛ln^𝑛(1−𝑧), is genuinely associated with the external line initial-state parton density evolution aspects of the kernels, and is not addressed by the methods in [76–78]. Both resummations obtain because of the exclusive limit 1−𝑧→0. One [76–78] is focused on the effects which remain after those associated with initial-state collinear singularities are removed so that they can be addressed by analyzing the respective hard coefficient function; and the other (that presented herein) is inclusive and allows one to focus on the effects associated with the initial-state collinear singularities as well as effects associated with the hard scattering coefficient, as we show now in the appendix by analyzing the result of [80] in our framework. From the discussion in the appendix, we see manifestly that there is no double counting of effects between the two approaches when they are used properly.

Finally, we address the issue of the relationship between the rearrangement that we have made of the exact leading-logs in the QCD perturbation theory and the usual treatment in the nonexponentiated DGLAP-CS theory. If one expands out the exponentiated kernels, using the distribution identity(1−𝑧)^𝑎−1=1𝑎1𝛿(1−𝑧)+(1−𝑧)₊+^∞_𝑗=1𝑎^𝑗[𝑗!ln^𝑗(1−𝑧)]1−𝑧₊, (70)one can see that for example 𝑃_𝑞𝑞 and 𝑃^exp_𝑞𝑞 agree to leading order, so that the leading log series which they generate for the respective NS parton distributions also agree through leading order in (𝛼_𝑠/𝜋)𝐿, where 𝐿 is the respective big log in momentum-space. At higher orders then, we have a different result for the {𝐹_𝑖}, let us denote them by {𝐹^′_𝑖}, and a different result for the reduced cross section, let us denote it by 𝜎^′, such that we get the same perturbative QCD cross section𝜎=∑_𝑖,𝑗∫𝑑𝑥₁𝑑𝑥₂𝐹_𝑖(𝑥₁)𝐹_𝑗(𝑥₂)𝜎(𝑥₁𝑥₂𝑠)=∑_𝑖,𝑗∫𝑑𝑥₁𝑑𝑥₂𝐹^′_𝑖(𝑥₁)𝐹^′_𝑗(𝑥₂)𝜎^′(𝑥₁𝑥₂𝑠) (71)order by order in perturbation theory. The exponentiated kernels are used to factorize the mass singularities from the unfactorized reduced cross section and this generates 𝜎^′ instead of the usual 𝜎 whose factorized form is generated using the usual DGLAP-CS kernels. We thus have the same leading log series for 𝜎 as does the usual calculation with unexponentiated DGLAP-CS kernels. We have an important advantage: the lack of +-functions in the generation of the configuration space functions {𝐹^′_𝑖,𝜎^′} means that these functions lend themselves more readily to Monte Carlo realization to arbitrarily soft radiative effects, both for the generation of the parton shower associated to the {𝐹^′_𝑖} and for the attendant remaining radiative effects in 𝜎^′. Further consequences of our results for LHC physics will be presented elsewhere[61].

Note-added
The application of exact, amplitude-based YFS-style resummation to non-Abelian gauge theories is done for the first time in [37–48]. In [81, 82], cancellation of IR singularities for QCD is approached from the KLN theorem perspective. As far as QED itself is concerned, the treatment in [81] is just the case of a singlet form-factor in which the exponentiated virtual IR function that is finally exhibited is not gauge invariant. The exponentiation of the soft real emission processes which cancel these virtual IR singularities is then done as an approximate treatment of the real emission processes in which momentum conservation for the soft real emission is ignored. In [82], the exponentiation and cancellation of IR singularities are demonstrated for any number of external electron lines as an approximate representation of the respective amplitudes in which the IR divergent terms are retained—finite terms are dropped. Thus, in neither case is the exact YFS theory for QED presented for the entire theory. Finally, we note that the discussion in [83] is a complete version of that in [82] but it still treats soft real photon emission in same soft photon approximation, so that it is not an exact rearrangement of the theory such as we have in the YFS formulation.

Acknowledgments

The authors thank Professor S. Jadach for useful discussions and Professor W. Hollik for the kind hospitality of the Max-Planck-Institut, Munich, wherein a part of this work was completed. Work partly supported by US DOE grant DE-FG02-05ER41399 and by NATO grant PST.CLG.980342.