PaperlayerView exact revision on arXiv

Source-linked AI summary

Calculation of prompt diphoton production cross sections at Tevatron and LHC energies

C. Balázs, E. L. Berger, P. M. Nadolsky, C. -P. Yuan

arXiv:0704.0001v1hep-ph

TL;DR

Diphoton production requires precise theoretical control because a narrow low-mass Higgs signal appears above substantial background. This paper develops a fully differential QCD calculation including higher-order subprocesses and gluon-radiation resummation, finding excellent agreement with data and improved Higgs-signal sensitivity through event selection.

  • Problem

    A precise theoretical understanding of diphoton production is needed because a narrow low-mass Higgs signal appears above substantial background.

  • Method

    The paper presents a fully differential QCD calculation incorporating leading- and next-to-leading-order direct-production subprocesses and higher-order contributions.

  • Results

    The calculation shows excellent agreement with data and provides reliable diphoton transverse-momentum dependence at small and intermediate values where the cross section is greatest.

  • Takeaways & Limitations

    Contrasting Higgs-decay and QCD diphoton distributions indicates that judicious event selection can enhance sensitivity to the Higgs signal.

Abstract

from arXiv · show

A fully differential calculation in perturbative quantum chromodynamics is presented for the production of massive photon pairs at hadron colliders. All next-to-leading order perturbative contributions from quark-antiquark, gluon-(anti)quark, and gluon-gluon subprocesses are included, as well as all-orders resummation of initial-state gluon radiation valid at next-to-next-to-leading logarithmic accuracy. The region of phase space is specified in which the calculation is most reliable. Good agreement is demonstrated with data from the Fermilab Tevatron, and predictions are made for more detailed tests with CDF and DO data. Predictions are shown for distributions of diphoton pairs produced at the energy of the Large Hadron Collider (LHC). Distributions of the diphoton pairs from the decay of a Higgs boson are contrasted with those produced from QCD processes at the LHC, showing that enhanced sensitivity to the signal can be obtained with judicious selection of events.

I. INTR ODUCTION

The paper develops a quantitative QCD calculation of isolated diphoton production at hadron colliders, motivated by the need to distinguish a narrow low-mass Higgs signal from substantial Standard Model backgrounds. It includes NLO subprocess contributions, all-orders resummation of initial-state gluon radiation, fragmentation effects, and validation against Tevatron data [1].

  • Scope: The calculation predicts invariant-mass, transverse-momentum, rapidity, and angular distributions for continuum diphoton production in proton-antiproton and proton-proton collisions.It targets hadron-collider energies relevant to the Tevatron and LHC.
  • Method: All parton-parton subprocess contributions are computed through next-to-leading order in perturbative QCD, whose large LHC corrections are necessary for quantitatively trustworthy predictions.The calculation also resums initial-state soft and collinear logarithms from gluon radiation to all orders in αs.
  • Method: Resumming initial-state gluon-radiation logarithms is essential for physically meaningful diphoton transverse-momentum predictions at small and intermediate QT, where the cross section is large.
  • Photon definition: The study analyzes final-state collinearly enhanced fragmentation contributions, while focusing on isolated high-energy photons separated from appreciable hadronic remnants.Such isolation targets photons originating directly in hard QCD scattering rather than nonperturbative sources of non-isolated photons.
  • Validation: Good agreement with isolated diphoton production data from the Fermilab Tevatron [1] supports the calculation’s predictions at LHC energies.The paper presents this comparison as increasing confidence in its LHC predictions.

Single−photon fragmentation +... … T lnp (Q2/Q2

The calculation combines direct diphoton subprocesses through NLO with NNLL resummation of initial-state radiation, while treating final-state fragmentation through isolation and auxiliary prescriptions. It provides fully differential predictions in the diphoton variables, is reliable for QT < Q, and improves the description of Tevatron QT distributions while informing LHC signal selection.

  • Single−photon fragmentation +...: The study includes leading- and next-to-leading-order direct subprocesses, including q¯q, qg, gg, and gluon-(anti)quark channels, plus subleading quark-scattering contributions.The gg channel includes one-loop gg → γγg diagrams from Refs. [7, 8] and two-loop four-leg diagrams from Refs. [9, 10].
  • Single−photon fragmentation +...: Final-state collinear singularities arise when a photon is collinear with a final-state parton, but isolation largely suppresses fragmentation contributions for the isolated-photon cross section.A complete treatment beyond lowest order would require joint resummation of initial- and final-state logarithmic singularities.
  • Single−photon fragmentation +...: For QT > Eiso, quasi-experimental isolation avoids the qg final-state collinear singularity; for QT < Eiso, an auxiliary regulator approximates the full NLO direct-plus-fragmentation rate.Subtraction and smooth-cone isolation prescriptions give similar predictions at the Tevatron and the LHC.
  • I I. THEOR Y O VER VIEW: Resummation is important for successfully describing physical QT distributions and stabilizing estimates of experimental acceptance effects on diphoton invariant-mass distributions.The requirement QT < Q further suppresses final-state fragmentation beyond the reduction from isolation, while selected LHC events can enhance Higgs-signal sensitivity over QCD continuum production.
  • A. Notation: The framework computes the fully differential diphoton cross section in Q, y, QT, and Collins–Soper photon angles, with the pair decay described in the Collins–Soper frame [15].The stated differential observable is dσ/(dQ^2dydQ_T^2dΩ*).
  • T lnp (Q2/Q2: The finite-order treatment uses asymptotic small-QT approximations and phase-space slicing, while the CSS result performs an all-orders two-dimensional Fourier transform in impact-parameter space.The resummed form separates a perturbative factor evaluated at b* from a nonperturbative exponent exp(−FNP(Q, b)).

the region QT ∼Q,

The calculation matches resummed and fixed-order predictions across QT, using W+Y below the crossing point and NLO above it, while resummation remains essential near kinematic cuts. Its photon-fragmentation treatment preserves isolated-cross-section behavior but leaves residual sensitivity to isolation modeling.

  • the region QT ∼Q,: Below the crossing point, the calculation uses W+Y, while above it the final prediction is the NLO cross section.The W+Y result includes resummed contributions at lower QT; the crossing point is where it falls below the finite-order prediction.
  • the region QT ∼Q,: The CSS and CFG resummation schemes differ numerically only slightly for diphoton production at the Tevatron and LHC [3].Scheme variation probes unaccounted-for NNNLL effects.
  • the region QT ∼Q,: The resummed cross section integrates approximately to the NLO cross section, with typically a few-percent correction from higher-order logarithmic terms.This agreement applies to observables permitting QT integration, such as the large-Q region of the diphoton invariant-mass distribution.
  • the region QT ∼Q,: Resummation is essential near kinematic cutoffs, where the NLO cross section becomes unstable but the resummed result remains smooth.This enables credible estimates of experimental-acceptance effects in diphoton invariant-mass and other distributions.
  • C. Final-state photon fragmentation: The adopted photon-fragmentation prescription preserves a continuous isolated cross section and reproduces the integrated qg rate from DIPHOX at small QT.Fragmentation contributions can remain moderately important in parts of phase space, while their magnitude depends on approximately modeled isolation parameters.

I I I. COMP ARISONS WITH D A T A AND PREDICTIONS

The calculation is implemented by matching resummed and NLO cross sections on grids and evaluating fully differential predictions with ResBos. Tevatron comparisons use CDF-like photon selections, reveal correlations between diphoton kinematics, and identify the region where the framework is most reliable.

  • Implementation: Matched resummed and NLO cross sections are computed on Q, Q_T, and y grids, then integrated with the ResBos program for fully differential predictions.The calculation uses five active quark flavors and the CTEQ6M NLO parton distribution set.
  • Results for Run 2 at the Tevatron: For CDF comparisons at 1.96 TeV, the calculation applies the experimental photon thresholds and isolation requirements, and also predicts approximately DØ-like selection constraints.The CDF thresholds are 14 GeV and 13 GeV for the harder and softer photons, respectively.
  • Results for Run 2 at the Tevatron: The event simulation shows that Δϕ tracks Q_T−Q: Q_T<Q events tend toward Δϕ>π/2, whereas Q_T>Q events tend toward Δϕ<π/2.The Born-limit configuration Q_T=0 corresponds to Δϕ=π.
  • Results for Run 2 at the Tevatron: The framework is most applicable for Q_T≲Q, where large Δϕ events dominate, while Q_T→0 requires resummation of logarithms log(Q/Q_T) and Q_T>Q involves complementary physics.The fixed-order calculation develops singularities as Q_T approaches zero and involves two hard scales, Q_T and Q.
  • Tevatron cross sections: The study compares resummed and finite-order predictions for the Tevatron diphoton invariant-mass distribution, with finite-order accuracy specified separately across the q̄q+qg and gg+gq channels.The finite-order calculation uses phase-space slicing and is integrated over all Q_T for the dσ/dQ distribution.

Q (GeV)

For the Q range shown, resummed and fixed-order invariant-mass distributions are close because the NNLO correction is relatively small, and both agree with CDF data within experimental uncertainties. At small Q, photon transverse-momentum cuts shape dσ/dQ and make finite-order predictions somewhat unstable through its correlation with the Q_T spectrum.

  • Q (GeV): The resummed and fixed-order invariant-mass distributions are close in normalization and shape, and both agree with CDF data within experimental uncertainties.The relatively small NNLO correction produces the similarity between the predictions in the displayed Q range.
  • Q (GeV): At small Q, transverse-momentum cuts on the photons affect the shape of dσ/dQ and correlate the invariant-mass distribution with the Q_T spectrum.These cuts also produce the characteristic shoulder near Q ≈ 27 GeV described in the preceding subsection.
  • Q (GeV): This correlation makes finite-order predictions for dσ/dQ somewhat unstable.The instability follows from the dependence of the invariant-mass distribution on the Q_T spectrum induced by the photon cuts.

QT (GeV) … Q (GeV)

Resummation resolves the low-Q_T singular behavior of the fixed-order diphoton prediction and improves agreement with Tevatron data, while the calculation is most reliable for Q_T < Q. The Q_T > Q and small-Δϕ regions remain sensitive to unimplemented diphoton fragmentation and other higher-order effects.

  • QT (GeV): Resummation improves agreement with CDF data at the lowest Q_T and for Q_T = 12−32 GeV, respectively lowering and increasing the predicted rate.The fixed-order prediction is disfavored at low Q_T, whereas resummed logarithmic terms regulate the small-Q_T region and alter the rate across the measured bins.
  • Resummed (NNLL): The resummed W + Y prediction is finite as Q_T approaches zero and is matched to the fixed-order result where the latter is reliable at large Q_T.W resums the initial-state logarithmic singularities, while Y supplies the remaining fixed-order terms; the prediction switches to P at their crossing point.
  • Resummed (qq: About 75% of the Tevatron rate with CDF cuts and 84% with DØ cuts comes from q̄q + qg + q̄g initial states, with the remainder from gg + gq channels.The gg + gq contribution falls steeply beyond Q_T > 22 GeV because the gluon parton distribution decreases rapidly at the relevant momentum fraction.
  • QT (GeV): The fixed-order distribution is ill-defined at Δϕ = π, while resummation increases the cross section near Δϕ = 2.5 rad and better agrees with data.At small Δϕ ≲ π/2, the fixed-order and resummed predictions coincide because of the matching procedure.
  • Q (GeV): The Q_T > Q region is populated mainly by small-Δϕ events and may contain fragmentation and other radiative contributions absent from the implemented diphoton calculation.Both fixed-order and resummed predictions may become deficient for Q_T ≫ Q because small-Q diphoton fragmentation is not yet modeled.
  • Q (GeV): Applying Q_T < Q preserves most of the cross section and restricts comparison to the phase-space region where the predictions are most valid.The authors expect substantially better agreement with data after imposing this selection.
  • DIPHOX (qq: The fixed-order q̄q + qg prediction agrees well with DIPHOX’s direct contribution, especially at large Q_T where both use the same fixed-order treatment.The difference in gg-channel accuracy has little impact because gg + gq is not dominant, particularly at high Q_T.
  • DIPHOX (qq: DIPHOX fragmentation contributions are small for nominal isolation but become comparable to direct contributions when Q_T is at or below the isolation energy.The stated isolation setup is E_iso/p_Tγ = 0.07, ΔR_cone = 0.4, ΔR_γγ > 0.3, and photon transverse-momentum thresholds of 21 and 20 GeV.

QT (GeV)

The resummed calculation agrees with DIPHOX over most Q_T values, while the Q_T < Q requirement suppresses fragmentation-sensitive regions and restricts comparisons to theoretically well-understood kinematics. In the CDF shoulder region, predictions with E_T^iso = 1 GeV underestimate data, whereas increasing isolation energy enhances fragmentation contributions and can improve agreement.

  • QT (GeV): The resummed description reproduces the integrated DIPHOX rate for 0 ≤ Q_T ≤ E_T^iso and agrees with DIPHOX inclusive rates for most Q values.Figure 7 compares resummed and DIPHOX invariant-mass and transverse-momentum distributions, including all channels and the q̄q + qg channel.
  • QT (GeV): For Q ≲ 27 GeV, Q_T > 25 GeV, and Δϕ < 1 rad, the resummed and DIPHOX cross sections with E_T^iso = 1 GeV underestimate CDF data within two standard deviations.The DIPHOX rate can be raised in this shoulder region by using E_T^iso = 4 GeV and μ_F = μ_R = Q/2.
  • QT (GeV): 400%: the one-fragmentation contribution increases on average when E_T^iso is raised from 1 to 4 GeV, further enhancing the shoulder-region rate.The direct contribution is weakly sensitive to E_T^iso, while the one-fragmentation contribution is approximately proportional to it.
  • QT (GeV): The Q_T < Q cut efficiently suppresses the fragmentation shoulder and small-Δϕ region while discarding only a small fraction of events.It provides a comparison region where the theory is well understood and has small uncertainty.

Q (GeV)

Resummed diphoton transverse-momentum distributions broaden as invariant mass Q increases. The average ⟨Q_T⟩ rises linearly for 30 < Q < 80 GeV, then saturates at higher Q, while proposed cuts remove fragmentation effects and reduce scale dependence.

  • Theoretical stability: Requiring Q_T < Q greatly reduces dependence of differential cross sections on the choice of factorization scale μ_F, to the typical size of higher-order corrections.This identifies the cut region as more stable for the calculation.
  • Tevatron cuts: Applying Q_T < Q or Q > 27 GeV cuts is predicted to eliminate the low-Δϕ and intermediate-Q_T enhancement from fragmentation contributions.The study urges CDF and DØ to apply these cuts in future Tevatron analyses.
  • Q dependence: The predicted resummed Q_T distributions broaden with increasing diphoton invariant mass Q.Figure 9(a) shows normalized distributions for different Q bins, reflecting the expected mass dependence.
  • Average transverse momentum: ⟨Q_T⟩ increases linearly over 30 < Q < 80 GeV, approaches ⟨Q_T/Q⟩ = 1 below about 30 GeV, and saturates near Q ∼ 80 GeV and above.The low-Q behavior results from cuts suppressing production at small Q_T, whereas high-Q saturation reflects PDF x dependence and other factors.

B. Results for the LHC · T > 40 (25) Ge V for the harder (softer) photon, (26) · QT (GeV)

For 14 TeV LHC predictions, the calculation applies ATLAS-inspired photon selections optimized for Higgs searches and studies resummed diphoton transverse-momentum distributions across invariant-mass bins. The average diphoton transverse momentum increases with invariant mass, but its growth rate decreases monotonically, while the cuts retain many Higgs events above 115 GeV.

  • B. Results for the LHC: The LHC study uses photon cuts based on ATLAS simulations of h →γγ at S = 14 TeV.
  • T > 40 (25) GeV for the harder (softer) photon, (26): The Higgs-search selection requires less than Eiso transverse energy within ΔR = 0.4 around each photon and photon separation above 0.4.These requirements impose a looser isolation restriction than in the Tevatron study.
  • QT (GeV): Figure 10 shows resummed diphoton transverse-momentum distributions for several invariant-mass bins at the LHC with the stated cuts imposed.
  • QT (GeV): The Higgs-optimized cuts may require adjustment for perturbative-QCD tests across the full accessible γγ invariant-mass range.
  • QT (GeV): The photon cuts optimized for Higgs searches preserve a large fraction of Higgs events with Q > 115 GeV.
  • QT (GeV): The photon cuts may be too restrictive for γγ production studies at smaller Q because the final-state photons most likely originate from about Q/2.
  • QT (GeV): Harder- and softer-photon cuts are necessary in fixed-order calculations but not required in the resummed calculation, where the associated instabilities are eliminated.Symmetric cuts could increase the γγ event sample in experimental analyses.
  • QT (GeV): The average γγ transverse momentum grows with Q, while the growth rate decreases monotonically with Q.The trend is demonstrated by the resummed distributions and Fig. 11.

Q (GeV) · QT (GeV)

At the LHC, diphoton production is dominated by the qg channel across invariant-mass ranges, while resummation and nonperturbative-model dependences are generally small under stated conditions. For QT > 80 GeV, the resummed q̄q+qg cross section approaches the direct fixed-order result, alongside comparisons with DIPHOX predictions.

  • Q (GeV): The Q and Δϕ diphoton distributions depend on scattering-subchannel combinations and theoretical-parameter choices, as discussed in Refs. [2, 3].The supplied passage does not provide numerical comparisons for these distributions.
  • Q (GeV): The qg contribution accounts for about 50% of the fixed-order (NLO) diphoton production rate across all Q ranges.This fraction depends on the factorization scheme and scale, and separating q̄q and qg contributions is not meaningful in the resummation calculation [3].
  • Q (GeV): The gg + gq contribution is about 25% at Q ∼80 GeV and decreases at larger Q.The Q ∼80 GeV feature corresponds to a cutoff-induced location in dσ/dQ from the photon transverse-momentum cuts.
  • Q (GeV): The cross sections have small resummation-scheme dependence, while nonperturbative-model dependence can be neglected if the nonperturbative function does not vary strongly with x [3].This behavior is also reported at the Tevatron [3].
  • Q (GeV): For QT > Eiso, quasi-experimental isolation removes direct NLO events containing collinear final-state photons and partons when QT > Eiso.The comparison with DIPHOX uses transverse-momentum and invariant-mass distributions in the q̄q + qg channel.
  • Q (GeV): At QT > 80 GeV, the resummed q̄q+qg cross section reduces to the direct fixed-order cross section.This comparison is made against DIPHOX predictions for the direct-plus-fragmentation treatment.
  • QT (GeV): The QT subsection presents a comparison between resummed q̄q + qg and DIPHOX direct-plus-fragmentation predictions.The supplied caption identifies these curves but provides no numerical values.

Q (GeV)

The resummed and DIPHOX q̄q+qg invariant-mass distributions agree within 10–20% over most Q values, with differences reaching a factor of 2 at the lowest Q. Including resummed gg+gq contributions improves agreement in the full γγ prediction, while reliability is greatest for Q_T < Q.

  • Q (GeV): 25%: Explicit single-photon fragmentation contributes about 25% of the full DIPHOX rate for 60 < Q_T < 120 GeV and grows approximately linearly with E_iso.Fragmentation and direct contributions are treated differently between the resummation and DIPHOX calculations, affecting their differential comparison.
  • Q (GeV): 10–20%: Resummed and DIPHOX q̄q+qg cross sections agree within 10–20% at most Q values, with the resummed rate consistently lower and differing by a factor of 2 at the lowest Q.The largest low-Q discrepancy occurs where photon-fragmentation contributions are large relative to the direct rate.
  • Q (GeV): 9% and 20%: The LO gg and resummed gg+gqS contributions constitute about 9% and 20% of the total rate, respectively.Including gg+gqS brings the resummed and DIPHOX invariant-mass distributions closer together.
  • Q (GeV): The resummed and DIPHOX rates reasonably agree for 1.5 ≲ Δϕ ≲ 2.5, while resummation remains finite as Δϕ approaches π and fixed-order DIPHOX rates diverge.For Δϕ < 1.5, DIPHOX is enhanced by photon-fragmentation contributions and theoretical uncertainties increase.
  • Q (GeV): The calculation captures the dominant γγ-production contributions at the LHC, with predictions most reliable when Q_T < Q because photon-fragmentation contributions are suppressed.The integrated DIPHOX rate is more stable under E_iso variations than its differential distributions, particularly because of differing E_iso dependence across Q_T regions.

QT (GeV)

At the LHC, direct qg scattering is the leading channel in the region relevant to Higgs searches, with its enhancement driven mainly by nonsingular phase-space contributions. The Higgs signal rate is substantially smaller than the QCD continuum background.

  • Direct qg scattering is the leading scattering channel in the region relevant for the Higgs boson search at the LHC.
  • The enhanced qg rate is driven predominantly by nonsingular phase-space contributions rather than final-state collinear radiation.
  • The q¯q + qg direct rate is only weakly sensitive to adjustments in the isolation parameter Eiso.
  • Contributions to qg scattering from photon fragmentation may be non-negligible and should be computed when LHC data become available.
  • C. Comparison with Higgs boson signal distributions: The Higgs signal cross section times branching ratio is substantially smaller than the QCD continuum background.The comparison considers diphoton production near Q = 130 GeV, with the signal evaluated at mH = 130 GeV and the selection QT < Q.

5 Ge V, resp e tiv ely . The a v erage v alues of QT

Over 0–75 GeV, the QT spectra differ in shape, with average values of 26 and 23 GeV; these differences reflect distinct leading initial-state Sudakov terms and final-state effects.

  • The QT spectra differ in shape over the 0–75 GeV range, with average values of 26 and 23 GeV.
  • The spectral differences can be attributed to the distinct structure of leading terms in the initial-state Sudakov exponents and to final-state effects.

QT (GeV)

At NNLL accuracy, Higgs and continuum diphoton distributions differ through their initial-state, final-state, and spin-correlation structures. These differences provide several kinematic discriminators, including QT, azimuthal, and rapidity-related distributions.

  • QT distribution: About 80% of the diphoton rate comes from q¯q + qg, whose smaller Sudakov coefficient produces narrower QT distributions than gg + gq radiation.Final-state collinear radiation in the qg channel hardens the continuum QT distribution under nominal ATLAS cuts, reducing its difference from the Higgs signal; effective isolation may lessen this impact.
  • Azimuthal discrimination: Without isolation, the spin-0 Higgs signal is central in ϕ*, whereas the QCD background peaks near ϕ* = 0 and π from the final-state qg singularity.Isolation produces a broad signal peak near ϕ* = π/2 and suppresses both signal and background for sin ϕ* < sin ∆R; selecting sufficiently large ϕ* reduces qg background.
  • Rapidity and angular correlations: The background peaks at zero rapidity difference, while the Higgs signal is nearly flat over a wide range, reflecting different spin correlations.The distinction is even more pronounced in the related scattering-angle variable, and the signal tends toward larger |ϕ3T − ϕ4T| than the background.
  • QT selection and resummation: The qualitative differences among rapidity, Collins–Soper scattering-angle, and azimuthal distributions persist after imposing QT > 10 GeV.The resummed calculation does not show the finite-order kinematic singularity near ∆y ≈ 2, because the discontinuity is resummed.
  • Overall discrimination: Resummed distributions provide good discriminators between the Higgs boson signal and continuum background in a simultaneous kinematic analysis.The signal-to-background ratio may be enhanced by restricting QT > 10 GeV, while combining several distributions is more efficient than relying on one alone.

IV. CONCLUSIONS … 3CATRNf

The paper presents a fully differential diphoton calculation combining NLO hard-scattering subprocesses with NNLL initial-state gluon resummation, obtaining reliable low- and intermediate-Q_T predictions and good agreement with collider data [2, 3]. It also identifies Higgs-sensitive distributions, limitations from photon isolation and fragmentation, and directions for future measurements and resummation improvements.

  • IV. CONCLUSIONS: The NLO calculation with NNLL initial-state gluon resummation gives reliable Q_T predictions where Q_T is small or intermediate and the cross section is largest.Resummation is essential for realistic Q_T dependence and stable estimates of experimental acceptance effects.
  • IV. CONCLUSIONS: The predictions show excellent agreement with published Tevatron data and are insensitive to the resummation scheme and nonperturbative functions in the region Q_T ≲ Q.The study recommends more differential measurements of the Q_T distribution versus Q and of mean transverse momentum versus Q.
  • IV. CONCLUSIONS: Distinctive Higgs-signal and continuum-background shapes suggest increased discovery significance through simultaneous likelihood analysis of several kinematic distributions, particularly resummed Q_T.The paper presents continuum diphoton mass, transverse-momentum, and angular predictions at LHC energy.
  • IV. CONCLUSIONS: At the LHC, the calculation includes NLO gg + gqS contributions, which are generally non-negligible and make predictions more accurate than fixed-order alternatives for low-Q_T-sensitive distributions.The qg + q̄g channels also become more important at the LHC than at the Tevatron.
  • IV. CONCLUSIONS: Applying Q_T < Q largely suppresses fragmentation contributions and is expected to improve agreement with current and future data, including removal of the Tevatron shoulder enhancement.Agreement deteriorates in the small-Q, Δϕ < π/2 region, where higher-order direct and fragmentation contributions can strongly modify rates.
  • IV. CONCLUSIONS: The gg channel contains a cos 2ϕ* spin-flip contribution whose Q_T dependence may become measurable with larger diphoton samples, while Q_T ≫ Q could probe the two-photon fragmentation function Dγγ(z1, z2).The high-Q_T region contains additional log(Q/Q_T) singularities associated with fragmentation of a parton into a light photon pair.
  • 3CATRNf: The appendices summarize perturbative coefficients and asymptotic small-Q_T cross sections, including gg + gqS spin-flip terms from interference of opposite gluon polarizations in the helicity-amplitude formalism [3].The coefficients and splitting functions are organized for the q̄q + qg and gg + gqS initial states.
Loading 0704.0001v1…