sign in sign up
<<
Home , Photon

arXiv:1607.03838v2 [hep-ph] 20 Dec 2016 Keywords: etosfravreyo P nlsae n locets vi creates, also and states final e UPC We detector of role. variety major a a for sections plays p geometry can collision interactions the two-photon interactions, while states, final tonuclear al;te e they cally; rpitsbitdt optrPyisCommunications Computer to submitted Preprint aueo rbe:Tecosscinfrultra- for cross-section pho- the The integrating by obtained is collisions problem: peripheral of are states. 3.0 final DPMJET some for and needed 8.2 PYTHIA Routines: External 2.5 17.8, 11.2, Classification: param- files. the input sample for chal- in MB not eters 5 Around should but computers. modern size, lenge problem on Depends RAM: Linux System: Operating workstations and test PCs Computer: including C program, Language: distributed Programming in test 2,965,010 bytes including data,etc.: of program, distributed No. in lines 53,188 data,etc.: of No. rga banbefo http: from obtainable Program (v2.2) STARlight Title: Program summary Program vi electromagnetically, interact nuclei colliding significant two a the been have (UPCs) collisions Ultra-peripheral Abstract TRih:AMneCrosmlto rga o ultra-peri for program simulation Carlo Monte A STARlight: ffi ciency. lr-eihrlcliin,Pooula interactions Photonuclear collisions, Ultra-peripheral ff cieyms ahohr htnceritrcin produ interactions Photonuclear other. each miss ectively ++ // starlight.hepforge.org arneBree ainlLbrtr,Bree A 94720 CA, Berkeley Laboratory, National Berkeley Lawrence riho nvriy mh,N 87 USA 68178 NE Omaha, University, Creighton ieUiest,HutnT,721USA 77251 TX, Houston University, Rice ae ee n uiGorbunov Yuri and Seger Janet nvriyo egn Norway Bergen, of University eaiitcions relativistic oyButterworth Joey pne .Klein R. Spencer okmNystrand Joakim / w-htno htnceritrcin,btnthadroni not but interactions, photonuclear or two-photon a oreo td tRI n h H.I hs collisions, these In LHC. the and RHIC at study of source ouelpo rmsnpis rsnl eos nthese In . single or pairs, or roduce rsn rga,SAlgt htcluae h cross- the calculates that STARlight, program, a present ot al iuain vnsfruei determining in use for events simulation, Carlo Monte a ucino nlsaerpdt n as h dimen- The mass. and rapidity state as final cross-section of production function the a of dimen- table two look-up a generates sional program for- The ASCII method: in Solution events these outputs mat. It . of and in consideration spins produced appropriate with unstable reaction, any the of decay the cludes e fms ), mass of ter as cal- of interest, it of function First, reaction a the process. for step cross-sections two the a culates is program - The interact not ically. de- do parameter nuclei colliding impact the also that pendent) is (which requirement subject the space, to parameter impact transverse in fluxes ton o tde fdtco e detector or of constraints studies for kinematic specific determine to within used be cross-sections can which events Carlo Monte ates gener- STARlight Second, ). transverse state w-htninteractions Two-photon , evco eo ttso oegnrlpho- general more or states meson vector ce W poo-oeo rtopoo cen- two-photon or (photon- USA Y fia tt aiiy and rapidity) state (final ffi inis h eodse in- step second The ciencies. hrlcliin of collisions pheral eebr2,2016 21, December p T (final - sions of the table are selectable, allowing the user to ally referred to as two-photon interactions, or photonu- choose the desired accuracy. For certain final states, a clear. second two-dimensional look-up table, giving the trans- In a two-photon interaction, the two electromagnetic verse momentum distribution, as a function of rapidity, fields, carried by the two nuclei, collide. These fields is also used. With these look-up tables, the program couple to any particles that carry , so a wide va- generates final states. Particle decays and the final an- riety of final states are possible. For relativistic - gular distributions are calculated for each event. clei, the electric and magnetic fields are perpendic- Restrictions: The program is focused on ultra- ular, and the configuration may be represented as a relativistic collisions at Brookhaven’s RHIC (Relativis- flux of almost-real photons, following the Weizs¨acker- tic Heavy Ion Collider) and CERN’s LHC (Large Williams method. At an impact parameter b from a rel- Hadron Collider), with final states that are visible in a ativistic nucleus with charge Z and Lorentz boost γ, the central detector. At lower (i.e. at the CERN photon is [1] SPS), caution should be exercised because STARlight does not account for the longitudinal momentum trans- Z2α k 1 1 N(k, b) = K2() + K2(x) (1) fer to the nucleus; this is larger at low beam energies. π2 (~c)2 γ2 1 γ2 0 Running time: The running time depends on the binning   where k is the , x = kb/γ~c, α 1/137 is used for the look-up tables and the number of events ≈ requested. Table generation is typically less than 15 the electromagnetic fine structure constant and K1 and minutes on a single core, and event generation is quick K0 are modified Bessel functions. The output N is the – generating 100,000 events takes only a few minutes. number of photons per unit area, per unit energy. The Running the providedsample input file to produce 1,000 first term gives the flux of transversely polarized pho- rho mesons at RHIC energies takes 7 minutes; run- tons; it dominates for relativistic nuclei, and it is the ning the provided sample input file∼ to produce 1,000 only one considered further here. As long as the ions - pairs at RHIC energies takes 2 min- are moderately relativistic, the second term is unimpor- utes. ∼ tant. References: http://starlight.hepforge.org and references Through these two-photon interactions, pairs of lep- in this article. tons, or electroweak may be produced. If a pair is produced, they hadronize into a sin- gle meson (with the appropriate /parity), a meson 1. Introduction pair or a more complex final state. The cross-section depends on the overlap of the two photon fields and on Ultra-peripheral collisions (UPCs) [1] are an active the two-photon coupling to the final state. STARlight field of study at the Relativistic Heavy-Ion Collider can generate a number of single-meson and lepton pair (RHIC) at Brookhaven National Laboratory [2, 3, 4, final states. 5, 6, 7], at the Fermilab Tevatron [8] and at the Large In photonuclear interactions, the most common ex- Hadron Collider (LHC) [9, 10, 11, 12, 13, 14, 15, 16]. clusive final state is a . In the vector meson In UPCs, the two nuclei interact electromagnetically dominance picture, photons fluctuate to qq pairs, which rather than hadronically by physically missing each scatter elastically from the target nucleus. For heavy other. Two-photon and photonuclear interactions can nuclear targets, the for from indi- occur. STARlight is capable of simulating a range of vidual add coherently when the 4 momentum − two-photon and photonuclear interactions: two-photon transfer q is small ( q < ~/RA), leading to a large cross- | | production of single mesons and photoproduction of a section. The treatment of coherence is handled via a variety of vector mesons. It can also interface with the Glauber formalism which accounts for multiple interac- DPMJET III to generate general pho- tions within the nucleus [17]. tonuclear interactions. Short-lived mesons (including STARlight also simulates photon scattering at higher vector mesons) are usually decayed within STARlight pT , where the amplitudes from the multiple target nu- and take into account the spin state. For more com- cleons add incoherently. plex cases, such as mesons with multiple final states, In photonuclear UPCs, either nucleus can emit the PYTHIA is used for the decays, but the spin informa- photon, with the other nucleus serving as a target. tion is lost. These possibilities are indistinguishable. Exchanging In ultra-peripheral collisions, two types of interac- the photon emitter and target introduces a tions can be distinguished: purely electromagnetic, usu- exp( i~k ~b) to account for the position difference. Here − · 2 ~k and ~b are the photon and impact parameter vectors skin depth of 0.53 fm is used for all nuclei. For nu- [18]. For symmetric collisions (e.g. -proton or clei, with Z 6, a Gaussian mass distribution is used. ≤ gold-gold), swapping the nucleus is equivalent to a par- Each projectile beam energy are defined by the ity transformation. Vector mesons are negative parity, Lorentz factor γ set by the user. The input beam en- so the amplitudes subtract and, for p 0, vector me- ergies can be asymmetric for the two projectiles, but in- T → son production disappears. For proton- col- ternally all calculations are performed in the center of lisions, swapping the nuclei is a charge-parity transfor- mass frame. Before the output is written, the final state mation; since vector mesons are CP positive, the two particles are boosted to the laboratory frame specified amplitudes add [19]. by the user. STARlight is optimized for RHIC and LHC One important aspect of ultra-peripheral collisions is energies, although other energy ranges should be usable. that, at least for heavy-ions, multiple interactions are STARlight 2.2 should closely reproduce the calcula- possible between a single ion pair. This has important tions in the original publications [17, 18, 23], with the consequences; it allows for studies involving correla- following exceptions. First, for some nuclei, the nuclear tions among the multiple final state particles. For ex- radii are set to the measured values, while previously a ample, one can study the azimuthal angular correlations parameterization assuming R A1/3 was always used; ∝ between the emitted in mutual Coulomb disso- this new radius is used exclusively. This reduces the ciation [20] or rapidity correlations for states like ρ0ρ0 cross-section for gold beams by about 7%, but has al- [21]. most no effect for lead beams. Second, the parameter- Experimentally, UPCs accompanied by single or izations of the γ + p J/ψ + p and γ + p Υ+ p → → multiple Coulomb excitation have proven to be very im- cross-sections have been improved, as described below. portant. The nuclear decays produce neutrons which are easily visible in zero degree calorimeters, greatly sim- plifying triggering. Many experimental results to-date 2. Photon spectra and form factors have relied on these neutrons for triggering. Because of the kinematics of the photon emission and absorption The photon spectrum is calculated in impact parame- [20], to a large degree, each photon acts independently, ter space by integrating Eq. 1 over all impact parameters and the cross-section for multiple reactions is given by under the assumption that the beam projectiles do not interact. For photonuclear interactions, the flux is

2 σ123 = d bP1(b)P2(b)...Pn(b) (2) dNγ(k) 2 Z = d bPNOHAD(~b)N(k,~b) (3) dk Z where the P (b) are the probabilities for the different i and for two-photon interactions, subreactions, as a function of impact parameter. These probabilities may be substantial; a unitarization pro- 2 d Nγγ(k1, k2) cedure corrects for the possibilities of multiple sub- = + dk1dk2 reactions, such as the production of two e e− pairs. STARlight can calculate cross-sections and generate 2 2 2 ~ ~ ~ ~ d b1d b2PNOHAD( b1 b2 )N(k1, b1)N(k2, b2) events for two-photon and photonuclear interactions ac- Z Z | − | companied by mutual Coulomb excitation. These are (4) threeor fourphotoninteractions: oneor two forthe pho- ~ tonuclear or two-photon interaction, plus one to excite Here, N(k1, b1) is the photon density for photons with ~ ~ each nucleus. It can also calculate the cross-sections for energy k1 at position b1 from nucleus 1 and N(k2, b2) is reactions accompanied by mutual excitation to a Giant the corresponding density from nucleus 2. The vectors Dipole Resonance (GDR) interactions; the GDR excita- ~b1 and ~b2 have their origins at the center of each nu- tions usually decay by single emission. cleus. So, b~ b~ is the impact parameter between the | 1 − 2| STARlight can simulate the collision of two dissim- two nuclei. The requirement that there be no hadronic ilar nuclei; both Z and A are selectable by the user. interactions reduces the usable luminosity. The actual Heavy nuclei are modeled either as hard spheres, or cross-section is sensitive to how this non-interaction is following the Woods-Saxon mass distribution. For se- defined [23, 24, 25]. The probability of not having a lected nuclei (gold, lead, copper), STARlight uses mea- hadronic interaction at impact parameter b, PNOHAD(~b), sured parameters for the nuclear radius [22]. Otherwise, is implemented in different ways for nucleus-nucleus, the Woods-Saxon radius is set to R = 1.2 fm A1/3. A proton-nucleus and proton-proton collisions. A · 3 STARlight uses a trick to lower the dimensionality of 0.7 fm. This is very close to the the Woods-Saxon 2 2 of the integral in Eq. 4. The integral d b1 d b2 is distribution, and the form factor may be determined an- replaced with db1 db2 dθ PNOHADR(b) whereR b = alytically [17]. For a 4-momentum transfer q, 2 2 R R R b1 + b2 2b1b2cos(θ) [24, 26]. 4πρ0 1 q − F(q) = sin(qRA) qRAcos(qRA) . In nucleus-nucleus collisions, the probability to have Aq3 − 1 + a2q2 no hadronic interaction at impact parameter b is   (10) Here, ρ0 is the nuclear density. σNNTAA(~b) PNOHAD(~b) = e− (5) For light nuclei, with Z 6, a Gaussian form factor is used. Proton-proton and proton-nucleuscollisions≤ are ~ where TAA(b) is the nuclear overlap function, which is treated separately, as discussed below, with the dipole calculated from the nuclear density profiles. The nu- form factor for the proton [19, 28] clear densities are assumed to follow a Woods-Saxon 1 distribution, and the -nucleon interaction cross- F(q) = . (11) sections follow the parameterization (1 + q2/(0.71GeV2))2 for pp collisions for proton-proton center of mass ener- gies √s above 7 GeV [27]: 3. Two-photon interactions σ = (33.73+0.2838 ln2(r)+13.67r 0.412 7.77r 0.5626)mb − − The two photon energies, k and k , are related to the − (6) 1 2 2 final state W and rapidity Y through where r = s/1GeV . The use of PNOHAD(b) reduces the γγ luminosities by up to 20% compared to the simpler W = 4k k (12) requirement that b > 2R [23], because there is still a 1 2 A p substantial probability of two nuclei interacting hadron- and ically at impact parameters larger than 2R . 1 k A Y = ln( 1 ). (13) In proton-nucleus collisions, the hadronic interaction 2 k2 probability is also calculated from the Glauber model. 2 The probability of having no hadronic interactions is The two-photon luminosity d N/dk1dk2 in Eq. 4 can thus be transformed to d2N/dWdY. For two-photon in-

σNNTA(~b) teractions, the cross-section to produce a final state X PNOHAD(~b) = e− (7) factorizes into a two-photon cross-section and a two- where TA(~b) is the nuclear thickness function. For large photon luminosity [1]: nuclei, applying this probability function is roughly 2 d Nγγ equivalent to requiring b > RA. This probability is used σ = σ(γ1γ2 X)dYdW . (14) whether the photon is emitted by the nucleus or the pro- Z Z dWdY → ton. So, the effective photon flux from the proton in a The limits on Y (a symmetric region is assumed) and proton-nucleus collision is considerably smaller than if W in the integrals| | of Eq. 14 are set by the user. the flux is calculated from just the proton form factor In differential form, Eq. 14 becomes [28]. In proton-proton collisions the probability of having 2 2 d σ d Nγγ no hadronic interactions is calculated from = σ(γ γ X) . (15) dWdY dWdY 1 2 → P (~b) = 1 Γ(s,~b) 2 (8) NOHAD | − | The look-up table produced during the first phase of the simulation is based on this equation, which thus pro- where Γ(s,~b) is the Fourier transform of the pp elastic vides the basis for the invariant mass and rapidity distri- scattering [29]. This is modeled by an expo- butions. nential b2/2b The final state transverse momentum pT is the vector Γ(s,~b) = e− 0 (9) sum of the pT of the two photons. The pT distribution 2 with b0 = 19.8 GeV− . This interaction probability cor- of the photons depends on the nuclear form factor [18, responds roughly to a cut in minimum impact parameter 26, 30, 31] as defined by of b > 1.4 fm. The nuclear form factor is the convolution of a hard- dN(k, p ) 2F2(Q2 = p2 )p3 T = T T . (16) sphere form factor with a with a range 2 2 2 2 dpT (2π) ((k/γ) + pT ) 4 Because the pT distribution depends on the photon en- is that it loses the effects of the correlations between the ergy, it cannot be pre-calculatedfor every condition. For lepton spins. simplicity, STARlight determines the photon pT using Examples of the kinematic distributions are shown rejection sampling. The azimuthal angle in the trans- in Figs. 1 and 2. Figure 1 shows the rapidity of verse plane is assumed to be uniformly distributed. each daughter from electron-positron pairs produced in Three types of two-photon interactions are imple- Xn Xn Au+Au collisions at √sNN = 200 GeV. Figure − + mented currently: two-photon production of lepton 2 shows the transverse momentum from µ−µ produced pairs, single meson production and decay, and the re- in Pb+Pb collisions at LHC energies. action γγ ρ0ρ0, followed by ρ decays. → 3.1. Lepton pairs 6 STARlight creates lepton pairs using the equivalent Au+Au → Au+Au+e+e- s = 200 GeV 40 photon approximation (EPA) approach. This is a low- NN est order calculation, where the photons are treated as 4 35 massless. The kinematic distributions are in generally 30 good agreement with the data, except that the EPA ap- 2 proach finds a lower average pT than in the data. A full 25 electrodynamic calculation, including the pho- 0 ton virtuality, found better agreement with the data [4]. 20

Later calculations showed that the result includes Rapidity of Daughter 2 room for an additional contribution to the cross-section, -2 15 due to higher order corrections [32, 33]. These correc- 10 ff tions are expected, but their e ects on the kinematic dis- -4 tributions are not well known. Still, STARlight lepton 5 pairs are useful for most purposes. -6 0 The cross-section to produce a pair of with -6 -4 -2 0 2 4 6 lepton mass m and pair invariant mass W is given by the Rapidity of Daughter 1 Breit-Wheeler formula [34]

+ σ(γγ l l−) = Figure 1: Rapidity of one track vs. rapidity of the other → for low-mass electron-positron pairs (pair mass between 4πα2 8m2 16m4 W + √W2 4m2 2 + ln ( − ) (17) 0.1 and 0.3 GeV/c2) produced in Xn Xn Au+Au col- W2 W2 − W4 2m −   lisions at RHIC energies. Pair rapidity is restricted to 4m2 4m2 1 1 + . ypair < 0.1. − r − W2 W2 | |  In the region W 2m, the cross-section drops very rapidly with increasing≥ W, so it is necessary to use a 3.2. Mesonic final states 2 very large number of bins ( 100 bins/GeVc− ) in W to The cross-section to produce a single meson from two accurately evaluate the total≥ cross-section. photons depends on its two-photon width Γγγ, mass MR, The angular distribution of these lepton pairs is given total width Γ and spin J [35] by [34] ΓγγΓ 2 2 2 4m 2 2 4m σγγ(W) = 8π(2J + 1) (19) 4m (1 2 ) sin (θ)cos (θ) + 2 2 2 2 = + − W W (W MR) G(θ) 2 4 1 2 . − − W2 4m 2 θ 2 (1 (1 W2 )cos ( )) 2 Γγγ  − − 8π (2J + 1) δ(W MR), (20) (18) ≈ 2M2 − where θ is the angle between the beam direction and R one of the leptons, in the lepton-lepton center of mass where Eq. 20 applies only for narrow resonances. frame. Here, we neglect the effect of the photon pT on STARlight can simulate either narrow or wide reso- the angular distribution. nances; in the latter case, the user can select the max- All three leptons are treated as stable; τ (and, if de- imum and minimum W. sired, µ) decays must be handled externally. One weak- STARlight simulates the production of a variety of ness of this approach (to be corrected in later versions) spin 0 and spin 2 mesons, using standard values of Γγγ. 5 4. Overview of photonuclear interactions 4500

4000 STARlight simulates the photoproduction and decay Pb+Pb → Pb+Pb+µ+µ- 3500 of a variety of vector meson final states: ρ, ω, φ, J/ψ, s = 5.02 TeV, M > 1.0 GeV [mb/(GeV/c)] 3000 nn inv ψ′, Υ(1S ), Υ(2S ) and the Υ(3S ). It can also simu- T late ρ+direct π+π photoproduction, including interfer- /dp 2500 − σ d 2000 ence between the two channels. The cross-sections are γ 1500 mostly based on parameterized HERA data for p Vp. It can also simulate general photonuclear interac-→ 1000 tions using DPMJET III. 500 The ρ0 and ω cross-sections, σ(γ + p V + p), are 0 0 0.05 0.1 0.15 0.2 0.25 parameterized as a function of the γp center→ of mass, p [GeV/c] T Wγp:

ǫ η Figure 2: Transverse momentum distribution for σ(γ + p V + p) = σP Wγp + σM Wγp . (22) pairs with pair mass greater than 1 GeV/c2 produced in → · · Pb+Pb collisions at LHC energies. The first term is for Pomeron exchange and the second for meson exchange. The other mesons are produced solely via Pomeron exchange, so σM = 0. Short-lived mesons, like the f0(980) and f2(1270), are For the J/ψ, ψ′ and Υ states, the power law is supple- decayed according to their spin and parity. The mesons mented with a factor that accountsfor the near-threshold are formed by the fusion of two spin 1 photons, with decrease in the cross-section: the photon spins either aligned (for spin 2 final states) 2 2 or antialigned (for spin 0) along the beam axis. For spin (mp + mV ) σ γ+ + = σ ǫ . ( p V p) P 1 2 Wγp (23) 0, the decays are isotropic, while for spin 2, the meson → · − Wγp  · spin leads to the angular distribution     This is shown in Fig. 3a. The cross-sections for ψ(2S ) dN and Υ(2S, 3S ) states are assumed to have the same sin5(θ). (21) dθ ∝ shape, with mV replaced by their respective masses. The values of the parameters were obtained from fitting ex- This distribution is sampled using rejection sampling. perimental data and are given in Table 1. The rapidity Thef2′(1525)is a special case, in that 50%ofthe time distributions for Υ(1S ), Υ(2S ) and Υ(3S ) produced in it decays to KsKs. These Ks’s are written to the output pp collisions at √s = 13 TeV are shown in Fig. 3b. file, and must be decayed by an external program, such For most of the particles, the masses and widths are as GEANT. the standard Particle Data Group values. The excep- For some two-photon final states, STARlight uses tion is the more complex ρ′, with considerable evidence PYTHIA (8.2 or higher) to handle particle decays. for multiple overlapping resonances [37]. Because key These are the η, η′, a2(1320) and the ηc. The advan- details about individual states and the interference are tage of using PYTHIA is that it can include multiple poorly known, STARlight uses a single resonance with branching ratios and more complicated final states; the a mass of 1540 MeV and a width of 570 MeV, chosen disadvantage is that the spin directional information is to match STAR observations [2]. The ρ′-photon cou- lost. pling and nuclear interaction cross-section are arbitrar- ily taken to be the same as for the ρ. The vector mesons are decayed assuming that the 0 0 3.3. ρ ρ photon is parallel with the beam axis. The same nuclear breakup options as for the two- STARlight can simulate the reaction γγ ρ0ρ0 at photon interactions are available. threshold (i.e., both ρ0 are produced with zero→ momen- tum in their center of mass frame) with σ 0 0 = 100 ρ ρ 4.1. Coherent vector meson production nb. This mode is included to simulate the excess which has been observed in two-photon production, which has The cross-sections for coherent production on nuclear been hypothesized to stem from a 4-quark state [36]. targets are determined using a classical Glauber calcu- 6 Table 1: Vector meson parameters for Wγp, in GeV. The of PROD MODE, the event generation is always done ρ, ω and φ parameters are from Ref. [38], while the J/ψ with a Breit-Wigner invariant mass distribution. values are obtained from a fit of data from [39, 40] to STARlight calculates a two-dimensional look-up ta- Eq. 23. The ρ′ characteristics are not well known, so ble, covering W and Y, and then generates W, Y pairs it is arbitrarily given the same parameters as the ρ. The for each event by sampling from the table. The W and Y Υ(1S ) cross-section uses the parameterization in Eq. 23 range and number of bins are settable, or STARlight can and Fig. 3a, while the Υ(2S ) and Υ(3S ) cross-sections pick a W range to cover the vector meson in question. are scaled from this, based on their couplings to e+e − There are multiple options for generating the vec- (i.e. σ 1/ f 2), per Eq. 10 of Ref. [17]. The ψ(2S ) V tor meson p spectra. STARlight can generate the p cross-section∝ is taken to be 0.166 that of the J/ψ, fol- T T spectra assuming that photoproduction on the two nu- lowing Ref. [41]. Previous versions of STARlight used clei is independent as shown in Fig. 4 for J/ψ pro- older, slightly different values for the J/ψ [17]. duced in Pb+Pb collisions at LHC energies, or in- Vector Meson σ ǫ σ η cluding the interference between the two production P M sites [18]; this is controlled by the user via the IN- ρ0 & ρ 5.0 µb 0.22 26.0 µb 1.23 ′ TERFERENCE PARAMETER. When interference is ω 0.55 µb 0.22 18.0 µb 1.92 turned on, it can also generate spectra including in- φ 0.34 µb 0.22 – – terference, but with an altered strength, set by the in- J/ψ 4.06nb 0.65 – – put parameter IF STRENGTH. To simulate proton anti- ψ(2S) 0.674nb 0.65 – – proton collisions, where the interference has the op- Υ(1S) 6.4pb 0.74 – – posite sign to proton-proton collisions [19], one sets Υ(2S) 2.9pb 0.74 – – IF STRENGTH= 1. When interference is used, the Υ(2S) 2.1pb 0.74 – – − pT spectrum is calculated in another two-dimensional look-up table, in Y and pT ; the range and number of lation [17]: pT bins in this table are settable via the input parameter INT PT N BINS; the number of Y bins is the same as

dNγ(k) for the cross-section table. σ(AA AAV) = 2 dk σ(γA VA) → Z dk → ∞ dNγ(k) ∞ dσ(γA VA) = 2 γ+p → ϒ(1S)+p 2 dk dt → F(t) , 25 p+p → p+p+ϒ s = 13 TeV [pb] Z0 dk Zt dt t=0| | σ ϒ 3 (1S) min /dy (pb) 10 ZEUS σ 20 d ϒ H1 (2S) (24) LHCb ϒ 15 (3S) 102

10 where F(t) is the nuclear form factor. The latter formula 10 5 is useful because it shows how the p distribution of the T 0 −8 −6 −4 −2 0 2 4 6 8 1 3 1 10 102 10 y vector mesons is determined. W [GeV] The photon flux dNγ(k)/dk is given by integrating Eq.

9 b) 25 b) (1) over impact parameter space, subject to the condi- µ Pb+Pb → Pb+Pb+ϒ (coherent) µ Pb+Pb → Pb+Pb+ϒ (incoherent) 8 ϒ ϒ s = 5.1 TeV /dy ( /dy ( (1S) (1S) sNN = 5.1 TeV NN σ 20 σ 7 tion that the two nuclei do not interact hadronically. For d ϒ d ϒ(2S) (2S) 6 ϒ 15 ϒ(3S) (3S) each impact parameter, the photon flux striking the nu- 5 4 10 cleus is determined and spread evenly over the nucleus. 3 2 5 The calculation of the cross-section can be done 1 0 0 −4 −2 0 2 4 −6 −4 −2 0 2 4 6 assuming narrow resonances, in which case tmin = y y 2 2 (MV /4kγ) , or it can be done by convoluting the pho- ton spectrum with the Breit-Wigner shape [17]. This is Figure 3: A fit to the available data for the cross-section decided by the user-selected parameter PROD MODE. for Upsilon production as a function of energy in γ- Setting PROD MODE=2 gives the narrow resonance p interactions is shown in the upper left [42, 43, 16]. approximation and setting PROD MODE=3 uses a STARlight uses this fit as the basis for calculating the wide resonance. The difference is significant only for photonuclear cross-section. Rapidity distributions for the ρ0, where the cross-section is reduced by about 5% Upsilon mesons produced at LHC energies are shown in heavy-ion collisions when a wide resonance is used. for pp collisions (upper right), coherent Pb+Pb col- The setting of PROD MODE only affects the cal- lisions (lower left) and incoherent Pb+Pb collisions culation of the cross section. Regardless of the value (lower right). 7 The transverse momentum is calculated in the same way as for coherent production, but with the nuclear form factor replaced with a nucleon form factor. For 500 Pb+Pb → Pb+Pb+J/ψ the light vector mesons, ρ0 and ω, a dipole form factor 400 snn = 5.02 TeV

[mb/(GeV/c)] is used: T 2 1 /dp 300 F(Q ) = . (27) σ + 2 2 2 d (1 Q /Q0) 200 This corresponds to a Fourier transform of an exponen- tial distribution. The value of Q0 depends on the 100 width of this distribution. The width of the matter dis- tribution is chosen such that the RMS radius is equal to 0 0 0.05 0.1 0.15 0.2 0.25 the squared sum of the proton and RMS radii, lead- p [GeV/c] T 2 2 ing to Q0 = 0.45 GeV . For the heavier vector mesons, a narrower transverse momentum distribution has been Figure 4: Transverse momentum distribution for J/ψ observed in γ + p V + p data and for these an expo- mesons produced at LHC energies. nential is used for the→ form factor [38]: 2 bQ2 F(Q ) = e− (28) 2 2 4.2. Incoherent vector meson production with b = 7.0 GeV− for φ, b = 4.0 GeV− for J/ψ and 2 The incoherent photonuclear cross-section is calcu- Υ, and b = 4.3 GeV− for ψ(2S). The distribution used lated under the assumption that it scales the same way for the light vector mesons is slightly wider than what as the total inelastic vector meson nucleus cross-section. one gets from an exponential distribution in Q2 with 2 This can be written b 10 GeV− , as has been found previously [47], but it gives≈ a good description of data on ρ0 production from σ (γ + A V + A ) σ (V + A) inc → ∗ = inel . (25) ALICE [12]. σ(γ + p V + p) σ (V + p) → inel π+π π+π Assuming and using the clas- 4.3. − − final states sical Glauber expression for the total inelastic cross- The structure of the 4π final state is somewhat com- section, this can also be expressed as plex in that the bulk of the cross-section likely comes from the decay of two or more excited ρ states. In the 4πα σ T(b) 2 absence of a detailed model, the 4π final state is decayed σ (γ + A V + A) = 1 e− VN db . inc f 2 following a distribution. Because the fi- → v Z  −  (26) nal state depends on the ρ′ nucleon cross-section σρ N, The term inside the bracket corresponds to the proba- if there are two different vector− mesons (with different− bility of having at least one vector meson-nucleon in- σρ N ), then the actual 4π mass spectrum and substruc- teraction at impact parameter b. It is thus assumed in ture− could dependon the ion species being collided. The the model that one or more vector meson-nucleon in- invariant mass distribution of 4π produced in Au+Au teractions lead to the emergence of a real vector me- collisions at √sNN = 200 GeV is shown in Fig. 5. son and that it is not destroyed by multiple interac- tions. Other authors have assumed that incoherent pro- 4.4. Interference and pT spectra duction occurs only if the vector meson reacts with The vector meson pT is the vector sum of the photon exactly one nucleon [44] or calculated the incoherent and Pomeron pT . The photon pT is as discussed in the cross-section based on event-by-event variations in the section on two-photon interactions. The Pomeron pT is positions of the nucleons (i.e the variations in the sec- determined by the form factor of the target nucleus, fol- ond moment of the amplitudes) [45]. Other authors lowing the distribution in Eq. 24 with the form factor in have made a direct connection between incoherent in- Eq. 10. The production amplitude is evenly distributed teractions and nuclear breakup, leading to neutron emis- throughout the nucleus (i.e. absorption is neglected). sion [46]. These approaches may lead to a reduction Interference between vector meson production on the in the incoherent cross-section at small pT ; this is not two nuclear targets is implemented via the cross-section present in STARlight. Figure 3d shows the Υ(1S ), formula: Υ(2S ) and Υ(3S ) production in incoherent Pb+Pb col- 2 σ(y, pT ) = d b[σ(y, pT ) cσ( y, pT )] (29) lisions at √sNN = 5.1 GeV. Z − − 8 but particle production at very forward rapidities, char-

+ acterized by low photon energies, cannot be completely 600 → π π-π+π- Au+Au Au+Au+ sNN = 200 GeV

/GeV) described. 2 500 ) (mb c - 400 π 5. Photonuclear breakup + π - π

+ 300 π Experimentally, it is usually easier to study ultra-

/dM( 200 peripheral collisions when they are accompanied by σ d photonuclear breakup. The breakup reaction produces 100 neutrons near the beam rapidity; these neutrons are eas- 0 ily detected in forward calorimeters [2, 3, 4]. 0 0.5 1 1.5 2 2.5 3 3.5 M(π+π-π+π-) (GeV/c2) Multiple interactions proceed via multiple indepen- dent photons [20] and are incorporated following Eq. Figure 5: Parent mass distribution for 4-pion final state 2. Since nuclear breakup depends only on the nucleus- produced in Au+Au collisions at RHIC energies. The 3 nucleus separation, it can be included without adding 2 any new integrals. Instead, the two-photon luminosity GeV/c cutoff is due to the choice of WMAX. and the photonuclear cross-section are both multiplied by a probability factor, PX(b), giving the probability of where σ(y, pT ) gives the pT spectrum without inter- the desired nuclear breakup at that impact parameter. ference. Here, c depends on the sign of the com- STARlight considers two types of nuclear breakup: bination: c=1 for ion-ion collisions, and c = 1 Xn, which is a breakup leading to any number of neu- for proton-antiproton collisions (e. g. at the Teva-− trons, and Giant Dipole Resonance (GDR) excitation, tron). In STARlight, c is user settable (input parame- which usually leads to a single neutron emission, 1n ter IF STRENGTH); it can be any . In this [50]. One may require that one or both nuclei break approach, Eq. 29 assumes that the production phase up and the selection is handled by the input parameter is independent of energy. This should be satisfied for BREAKUP MODE. Pomeron exchange, but may fail at low energies where The cross-section for a nucleus to be excited is [51, production proceeds via meson exchange. 52] For incoherent photoproduction, this interference is d3n(b, k) P1(b) = dk σγA A (k). (30) not important, and is therefore not implemented. Z dkd2b → ∗ The subscript ’1’ shows that this is a non-unitarized 4.5. General photonuclear interactions probability, and for high enough energy collisions, P1(b) can exceed one. The actual probability is given STARlight is interfaced with the DPMJET III [48] by a unitarization procedure. The excitation cross- Monte Carlo to simulate general photonuclear inter- section σγA A (k) was determined using experimental actions γ + A X. This includes photonuclear → ∗ → data [53]. Because of this, the excitation code works jet production and production of soft . Pho- only for gold or lead beams. For proton-proton or ton emission from one or both nuclei can be han- proton-nucleus interactions, only the modes that place dled, and the photon spectrum is calculated for the no restriction on Coulomb excitation are implemented. two cases as described in [49]. Single photon ex- An interaction with photon energy less than 30 change events are characterized by a rapidity gap on MeV/c (in the target rest frame) is assumed to lead to the side of the photon-emitting nucleus. The min- a Giant-Dipole resonance and, consequently, to a single imum and maximum photon energies (in the rest emitted neutron. Higher energy photons are assumed to frame of the target nucleus) must be set by the user produce higher excitations, with multiple neutrons. through the input parameters MIN ENERGY and MAX GAMMA ENERGY (in GeV). DPMJET re- quires that the photon energy is greater than 6 GeV 6. Program overview so MIN GAMMA ENERGY should be set higher than this value. A minimum photon energy of 6 GeV is STARlight runs in two phases. The first phase cal- sufficient to simulate particle production around mid- culates cross-sections and other kinematic distributions, rapidity ( η 4.5 in Pb+Pb interaction at the LHC), and stores them in a series of look-uptables. The second | | ≤ 9 phase uses these look-up tables to rapidly simulate nu- Utils contains root macros that can be used to convert clear interactions. Because of this two-stage approach, the output to a root tree, and to produce a simple analy- it can generate large numbers of events very rapidly. sis There is also a collection of top-level files outside the 6.1. Look-up table structure folder structure including The first look-up table to be generated is a CMakeLists, a build file two-dimensional table for event mass and rapidity, • d2N/dYdW. For photonuclear vector meson produc- Makefile created by CMake tion, the shape of the mass distribution is a relativis- • tic Breit-Wigner distribution [54] with mass and width Readme contains instructions for installing • taken from the PDG. For two-photon production of lep- STARlight, a brief description of the physics ton pairs, the shape of the mass distribution is calcu- incorporated in the code, and a description of input lated from the Breit-Wheeler formula (Eq. 18) convo- parameters and supported channels luted with the photon fluxes from each nucleus. For starlightDoxyfile.conf gives the settings used by two-photon production of single mesons, the mass dis- • tribution is a delta function at the PDG mass. the doxygen documentation system. doxygen doc- Additionally, for photonuclear interactions, a sec- umentation may be created (in the doxygen direc- tory) via ”doxygen starlightDoxyfile.conf” ond table stores the pT distribution, in the form of 2 d N/dpT dY. In actuality, the pT depends slightly on W, The driver program Main.cpp instantiates, initial- but neglecting this small correlation does not adversely izes and runs starlightStandalone.cpp. starlightStan- affect the accuracy of STARlight. For γγ interactions, dalone.cpp in turns calls methods from STARlight to the pT distribution is generated semi-analytically, and initialize and produce events of the selected channel and no table is needed. decay them as needed. The final daughter particles are The range of these tables in Y and W, and the num- boosted to the lab frame and written out in ASCII for- ber of bins for each dimension, are all user settable. mat to the file slight.out. STARlight stores integral tables that are normalized to The init method in the starlight class reads the input 1.0. This simplifies the look-up - for each variable, parameters from the file slight.in and uses these to de- STARlight selects a random number, finds the appro- termine the beam-beam system and the interaction type priate locations in the table that straddle that number, (e.g., photon-nucleusvs. two-photon, coherent vs. inco- and then linearly interpolates between the two nearest herent). It checks whether appropriate differential lumi- values. For two-dimensional tables, an additional one- nosity tables (dN/dWdY, with granularity set by input dimensional array holds a collapsed version of the table, parameters specifying the number of bins in W and Y) just containing dσ/dY. Using this, a value of Y is de- exist, and calls the appropriate method to create them if termined. A second random number is chosen, and the theydonot. Thedifferential luminosity tables are stored value of W or pT is found by looking at the two rows in the file slight.txt. It instantiates the appropriate event in the two-dimensional tables that straddle the chosen channel and calculates the cross-section. value of Y. Then, linear interpolation is performed on Important classes used in the init method of the two rows, to determine two values of W correspond- STARlight: ing to the Y values for those rows. Finally, another linear interpolation is performed, in Y, to determine the final Nucleus: includes methods to determine nuclear W. Or course, the same value of Y is used to determine • radius, thickness and form factor. W and pT . Beam: derives from nucleus; includes methods to • 6.2. Overview of the program structure calculate photon flux, based on impact parameter and photon energy. STARlight is coded in C++. The package contains files in several folders: Beambeamsystem: includes methods to boost to Src contains the C++ code • CMS, calculate probability of breakup. Include contains the header files that are needed by the C++ code photonNucleusCrossSection: base class with Config contains example input files • methods to calculate cross-section for coherent Cmake modules contains cmake macros photon-nucleus interactions. 10 photonNucleusLuminosity: derives from photon- starlightDpmJet: derives from eventChannel; in- • NucleusCrossSection, calculates differential lumi- • cludes methods to generate photonuclear events nosity tables for coherent photon-nucleus interac- using DPMJET. tions. nBodyPhaseSpaceGen: used for 4-particle final incoherentphotonNucleusLuminosity: derives • state generation. • from photonNucleusCrossSection, calculates differential luminosity tables for incoherent The utils directory includes a few examples which photon-nucleus interactions. may be useful for some users. ConvertStarlightAsci- iToTree.C converts STARlight output to a ROOT TTree, narrowResonanceCrossSection: derives from pho- while Analyze.cxx reads STARlight output and pro- • tonNucleusCrossSection, includes a method to find duces some useful histograms. AnalyzeTree.cxx reads the total cross-section for a narrow resonance. the output from ConvertStarlightAsciiToTree and pro- duces similar histograms. wideResonanceCrossSection: derives from pho- • tonNucleusCrossSection, includes a method to find the total cross-section for a wide resonance. 7. Description of input data

incoherentVMCrossSection: derives from pho- STARlight runs are controlled through the slight.in • tonNucleusCrossSection, includes a method to file. The input parameters are listed below, with typical find the total cross-section for incoherent photon- values for LHC Pb+Pb running given in parentheses. nucleus interactions. 1. baseFileName: The name of the output files. twophotonluminosity.cpp: calculates the two pho- STARlight will copy the input slight.in to base- • ton luminosity table, as a function of W and Y. FileName.in, and produce output files baseFile- Name.txt and baseFileName.out. (slight) The run method in starlightStandalone calls the pro- 2. BEAM 1 Z: Charge of beam one projectile. (82) duceEvent method in STARlight. The produceEvent method contains an iterative loop to generate the re- 3. BEAM 1 A Atomic number of beam one projec- quested number of events using the method appropri- tile. (208) ate to the event channel. All methods derive from the 4. BEAM 2 Z: Charge of beam two projectile. (82) eventChannel class. Each method randomly picks an 5. BEAM 2 A: Atomic number of beam two projec- energy and rapidity within the selected range, produces tile. (208) the parent system, and decays it where appropriate. 6. BEAM 1 GAMMA: Lorentz boost for beam one starlightStandalone then boosts the final state particles projectile (pz>0). (1470) into the lab frame and writes them into the ASCII file 7. BEAM 2 GAMMA: Lorentz boost for beam two slight.out projectile (pz<0). (1470) Important classes used in the produceEvent method 8. W MAX: Maximum value for the gamma-gamma of STARlight: center of mass energy, W = 4E1E2, in GeV. Set- ting W MAX = -1 tells STARlight to use the de- eventChannel: derives from readLuminosity; base fault value specified in inputParameters.cpp (rec- • class with methods to produce events and keep ommended for single meson production). For sin- track of the number attempted and accepted. gle mesons, the default W MAX is the particle Gammaavectormeson: derives from eventChannel, mass plus five times the width. For lepton pairs, • γ γ includes methods to produce vector mesons from ~ 1 2 the default W MAX is given by 2 c R R . (-1) q 1 2 photon-nucleus interactions. 9. W MIN: Minimum value for the gamma-gamma Gammagammaleptonpair: derives from center of mass energy, W = 4E1E2, in GeV. Set- • eventChannel, includes methods to create lepton ting W MIN = -1 tells STARlight to use the de- pairs from photon-photon interactions. fault value specified in inputParameters.cpp (rec- ommended for single meson production). The de- Gammagammasingle: derives from eventChannel; fault W MIN is the larger of the kinematic limit (e. • includes methods to create single-meson events in g. 2mπ for ρ decays) or the particle mass minus five photon-photon interactions. times the width. (-1) 11 10. W N BINS: Specifies the number of W bins to BREAKUP MODE =2: requires Coulomb store in the look-up tables. (40) • break-up of both nuclei, with no restriction 11. RAP MAX: Maximum rapidity of produced parti- on the number of neutrons emitted by either cle. (8) nucleus (XnXn). 12. RAP N BINS: Number of rapidity bins used in the BREAKUP MODE = 3: requires Coulomb cross-section calculation. (80) • break-up of both nuclei, but requires that a 13. PROD MODE: Allows the user to select from a single neutron is emitted from each nucleus variety of production modes. (2) (1n1n). PROD MODE=1: Two-photon interaction. BREAKUP MODE = 4: requires Coulomb • • break-up of neither nucleus. (0n0n) PROD MODE=2: Coherent photonuclear • BREAKUP MODE = 5: requires that there vector meson production assuming narrow • resonances. In pA collisions, this option be no hadronic break up, no restriction on means that the proton emits the photon and Coulomb break-up (This is similar to option that the gamma-A interaction is coherent. 1, but with the actual hadronic interaction probability). PROD MODE=3: Coherent photonuclear • BREAKUP MODE = 6: requires Coulomb vector meson production assuming wide res- • onances. This option should be used for ex- break up of one or both nuclei, with no re- clusive ρ0 production. striction on the number of neutrons emitted (XnXn + 0nXn + Xn0n). PROD MODE=4: Incoherent photonuclear • BREAKUP MODE = 7: requires Coulomb vector meson production. In pA collisions, • this option means that the nucleus emits the break up of only one nucleus, with no re- photon. striction on the number of neutrons emitted (0nXn+ Xn0n). PROD MODE=5: Photonuclear one photon • exchange uses DPMJET single. 18. INTERFERENCE: Specifies whether interference based on the ambiguity of which nucleus emits the PROD MODE=6: Photonuclear two photon ff • photon is included. The e ect of this interference exchange (both nuclei excited) uses DPMJET is only visible at very small transverse momentum. double. 0 = interference off,1 = interference on. (0) PROD MODE=7: Photonuclear single pho- 19. IF STRENGTH: If interference is turned on, spec- • ton exchange uses DPMJET single proton ifies the percentage of interference. The range is mode. -1.0 to 1.0.; 1 is the standard value for ion-ioncolli- sions, while -1.0 is expected for proton-antiproton 14. N EVENTS: Number of events to be produced. collisions. (1) (1000) 20. INT PT MAX: Used only when the interference 15. PROD PID: For PROD MODE 1 through 4, this option above is turned on. This specifies the maxi- selects the channel to be produced, in PDG nota- mum transverse momentum considered, in GeV/c. tion. Currently supported options are given in Ta- (0.24) bles 2 and 3. (443013) 21. INT PT N BINS: Used only when the interference 16. RND SEED: Seed for random number generator. option above is turned on. This specifies the num- (34533) ber of bins in transverse momentum to use. (120) 17. BREAKUP MODE: Specifies the way nuclear 22. INT PT WIDTH: Used only when the interference break-up is handled. This option only works for option above is turned on. This specifies the width lead or gold. In proton-proton or proton-nucleus of bins in transverse momentum to use. (0) collisions, options 2 - 7 automatically default to option 5. (5) The following parameters are optional. The default BREAKUP MODE = 1: hard sphere nuclei value is shown after the parameter name. • (no hadronic break-up if impact parameter is 23. CUT PT: Specifies whether the user chooses to greater than the sum of nuclear radii, no re- place restrictions on the transverse momentum of striction on Coulomb break-up). the decay products. 0= no, 1 = yes. (0) 12 24. PT MIN: If a transverse momentum cut is ap- plied, this specifies the minimum value produced, in GeV/c. (1.0) 25. PT MAX: If a transverse momentum cut is ap- plied, this specifies the maximum value produced, Table 2: List of two-photon channels in STARlight. in GeV/c. (3.0) 26. CUT ETA: Specifies whether the user chooses to Channel Produced PROD PID + place restrictions on the pseudorapidity of the de- e e− pair 11 + cay products. 0= no, 1 = yes. (0) µ µ− pair 13 + 27. ETA MIN: If a pseudorapidity cut is applied, this τ τ− pair 15 0 specifies the minimum value produced. (-10) ρ pair 33 28. ETA MAX: If a pseudorapidity cut is applied, this a2(1320) decayed by PYTHIA 115 specifies the maximum value produced. (10) η decayed by PYTHIA 221 f (1270) π+π 225 29. XSEC METHOD: Determines which method is 2 → − used to calculate the cross-section for γγ cross- η’ decayed by PYTHIA 331 0 sections. XSEC METHOD=0 is faster, but works f ’(1525) K+K (50%), K0K (50%) 335 2 → − only for symmetric collisions (i. e. with identical ηc decayed by PYTHIA 441 nuclei). XSEC METHOD=1 always works, but is f0(980) 9010221 slower. (0)

The following parameters are used only when in- terfacing with the PYTHIA and/or DPMJET inter- faces: 30. MIN GAMMA ENERGY: Allows the user to set the minimum photon energy (in GeV) in the rest frame of the target nucleus. The default is 6.0 GeV and it should never be set below this value since DPMJET was not designed to handle low energy Table 3: List of vector meson channels in STARlight. interactions. 31. MAX GAMMA ENERGY: Allows the user to set Channel Produced PROD PID 0 + the maximum photon energy (in GeV) in the rest ρ π π− 113 0 → + frame of the target nucleus. The default is 60000.0 ρ + direct π π− production, with 913 + GeV. interference. The direct π π− frac- 32. PYTHIA PARAMS: Used to supply input param- tion is from ZEUS data. + eters to the PYTHIA interface. This takes a ω π π− 223 → + to pass on semi-colon separated parameters to Φ K K− 333 → + PYTHIA 6. eg: ”mstj(1)=0;paru(13)=0.1”(the de- J/ψ e e− 443011 → + fault is a blank string ” ”) J/ψ µ µ− 443013 → + 33. PYTHIA FULL EVENT RECORD: Determines ψ(2S) e e− 444011 → + whether the full event record from PYTHIA is ψ(2S) µ µ− 444013 → + written to slight.out. true = yes, false = no (false). Υ(1S) e e− 553011 → + The additional information added is as follows: Υ(1S) µ µ− 553013 → + daughter production vertex (x [mm], y [mm], z Υ(2S) e e− 554011 → + [mm], t [mm/c]), mother1, mother2, daughter1, Υ(2S) µ µ− 554013 → + daughter2, PYTHIA particle status code. The Υ(3S) e e− 555011 → + PYTHIA 8 Particle Properties page describes in Υ(3S) µ µ− 555013 →+ + more detail the properties of mother, daughter, and ρ′ π π−π π− 999 status code designations. → Currently supported production channels in STARlight are listed in Tables 2 (two-photon channels) and 3 (vector meson channels). 13 8. Description of output BREAKUP MODE = 2 INTERFERENCE = 0 STARlight outputs an ASCII file named slight.out. IF STRENGTH = 0. For each event, a summary line is printed, with the for- INT PT MAX = 0.24 mat INT PT N BINS =120 EVENT: n ntracks nvertices , where n is the event number (starting with 1), ntracks Additional input files used to generate the figures in is the number of tracks in the event, and nvertices is this paper are available in the code distribution, in the the number of vertices in the event (STARlight does not config folder. currently produce events with more than one vertex). This is followed by a line describing the vertex, of the format 10. Summary VERTEX: x y z t nv nproc nparent ndaughters , where x, y, z and t are the 4-vector components of the STARlight is a versatile C++ code which can sim- vertex location, nv is the vertex number, nproc is a num- ulate a variety of important ultra-peripheral collisions ber intended to represent the physical process (always between relativistic heavy nuclei. The output in terms set to 0), nparent is the track number of the parent track of cross-sections, rapidity and pT distributions etc. can (0 for primary vertex) and ndaughters is the number of and has been tested against experimental results. Events daughter tracks from this vertex. generated by STARlight have also been used by sev- This is followed by a series of lines describing each eral experiments to calculate corrections for acceptance of the daughter tracks emanating from this vertex. Each and efficiency; in these calculations, it is important to track line has the format have an event generator which provide reasonable phase TRACK: GPID px py py nev ntr stopv PDGPID , space distributions of the produced particles. STARlight where GPID is the Geant particle id code, px, py and pz has been found to describe most important features of are the three vector components of the track’s momen- the data, from both RHIC and the LHC, very well. tum, nev is the event number, ntr is the number of this track within the vertex (starting with 0), stopv is the ver- 11. Acknowledgments tex number where track ends (0 if track does not termi- nate within the event), and PDGPID is the Monte Carlo This C++ code was developed from a FORTRAN particle ID code endorsed by the Particle Data Group. version. Some of the γγ FORTRAN code was written by Evan Scannapieco, and incorporated an earlier basis 9. Description of test data written by Kai Hencken. We thank them both profusely. We thank Boris Grube for useful discussions on pho- The following parameters were used to produce Fig. toproduction of the 4π final state. Kyrre Skjerdal and 1. Øystein Djuvsland added some pieces of code and we thank them for their contribution. baseFileName = ee RHIC This work was funded by the U.S. Department of En- BEAM 1 Z = 79 ergy under contract numbers DE-AC-76SF00098, DE- BEAM 1 A = 197 FG02-96ER40991 and DE-FG02-10ER41666. BEAM 2 Z = 79 BEAM 2 A = 197 BEAM 1 GAMMA = 108.4 References BEAM 2 GAMMA = 108.4 [1] C. A. Bertulani, S. R. Klein and J. Nystrand, Ann. Rev. Nucl. W MAX = 0.3 Part. Sci. 55, 271 (2005); G. Baur et al., Phys. Rept. 364, 359 W MIN = 0.1 (2002); F. Krauss, M. Greiner and G. Soff, Prog. Part. Nucl. W N BINS = 100 Phys. 39, 503 (1997); A. J. Baltz et al., Phys. Rept. 458, 1 (2008). RAP MAX = 1.3 [2] B. I. Abelev et al.[STAR Collaboration], Phys. Rev. C81, RAP N BINS = 100 044901 (2010). PROD MODE = 1 [3] B. I. Abelev et al.[STAR Collaboration], Phys. Rev. C77, N EVENTS = 1000000 034910 (2008); C. Adler et al., Phys. Rev. Lett. 89, 272302 (2002). PROD PID = 11 [4] J. Adams et al.[STAR Collaboration], Phys. Rev. C70, 031902 RND SEED = 5574533 (2004). 14 [5] B. I. Abelev et al.[STAR Collaboration], Phys. Rev. Lett. 102, (1998); S. Chekanov et al. [ZEUS Collaboration], Phys. Lett. 112301 (2009). B680, 4 (2009). [6] G. Agakishiev et al.[STAR Collaboration], Phys. Rev. C85, [43] C. Adloff et al. [H1 Collaboration], Phys. Lett. 483, 23 (2000). 014910 (2012). [44] T. Lappi and H. Mantysaari, Phys. Rev. C87, 032201 (2013). [7] A. Afanasiev et al.[PHENIX Collaboration], Phys. Lett. B679, [45] T. Toll and T. Ullrich, Comput. Phys. Commun. 185, 1835 321 (2009). (2014); T. Toll and T. Ullrich, Phys. Rev. C87, 024913 (2013). [8] T. Aaltonen et al.[CDF Collaboration], Phys. Rev. Lett. 102, [46] M. Strikman, M. Tverskoy and M. Zhalov, Phys. Lett. B 626, 72 242001 (2009). (2005). [9] B. Abelev et al. [ALICE Collaboration], Phys. Lett. B718 [47] S. Aid et al. [H1 Collaboration], Phys. Lett. B 463, 3 (1996); (2013) 1273. J. Breitweg et al. [ZEUS Collaboration], Eur. Phys. J 2, 247 [10] E. Abbas et al. [ALICE Collaboration], Eur. Phys. J. C73 (2013) (1998). 2617. [48] R. Engel, J. Ranft and S. Roesler, Phys. Rev. D55, 6957 (1997); [11] B. B. Abelev et al. [ALICE Collaboration], Phys. Rev. Lett. 113 Phys. Rev. D57, 2889 (1998). (2014) 232504. [49] O. Djuvsland, J. Nystrand, Phys. Rev. C83, 041901 (2011). [12] J. Adam et al. [ALICE Collaboration], JHEP 1509 (2015) 095. [50] B. L. Berman and S. C. Fultz, Rev. Mod. Phys. 47, 713 (1975). [13] J. Adam et al. [ALICE Collaboration], Phys. Lett. B751, 358 [51] A. J. Baltz, S. R. Klein and J. Nystrand, Phys. Rev. Lett. 89, (2015). 012302 (2002). [14] R. Aaij et al. [LHCb Collaboration], J. Phys. G40 (2013) [52] A. J. Baltz, S. R. Klein, J. Nystrand and Y. Gorbunov, Phys. Rev. 045001. C80, 044902 (2009). [15] R. Aaij et al. [LHCb Collaboration], J. Phys. G41 (2014) [53] A. J. Baltz, M. J. Rhoades-Brown and J. Weneser, Phys. Rev. 055002. E54, 4233 (1996). [16] R. Aaij et al. [LHCb Collaboration], JHEP 1509 (2015) 084. [54] J.D. Jackson, Nuovo Cimento 34, 6692 (1964). [17] S. Klein and J. Nystrand, Phys. Rev. C60, 014903 (1999). [18] S. Klein and J. Nystrand, Phys. Rev. Lett. 84, 2330 (2000); S. Klein and J. Nystrand, Phys. Lett. A308, 323 (2003); K. Hencken and G. Baur, Phys. Rev. Lett. 96, 012303 (2006). [19] S. Klein and J. Nystrand, Phys. Rev. Lett. 92, 142003 (2004). [20] G. Baur et al., Nucl. Phys. A729, 787 (2003). [21] M. Klusek-Gawenda and A. Szczurek, Phys. Rev. C 89, 024912 (2014). [22] C. W. deJager, H. deVries and C. deVries, At. Data Nucl. Data Tables 14, 495 (1974). [23] J. Nystrand and S. Klein, preprint arXiv:9811007. [24] R. N. Cahn and J. D. Jackson, Phys. Rev. D42, 3690 (1990). [25] K. Hencken, D. Trautmann and G. Baur, Zeit. f. Physik 68, 473 (1995). [26] G. Baur and L. G. Ferreira-Filho, Nucl. Phys. A518, 768 (1990). [27] K. A. Olive et al. [Particle Data Group], Chin. Phys. C38, 090001 (2014). [28] M. Drees and D. Zeppenfeld, Phys. Rev. D39, 2536 (1989). [29] L. Frankfurt, C. E. Hyde, M. Strikman and C. Weiss, Phys. Rev. D75 (2007) 054009. [30] K. Hencken, note NUCL-TH-9503004. [31] M. Vidovic, M. Greiner, C. Best and G. Soff, Phys. Rev. C47, 2308 (1993). [32] A. J. Baltz, Phys. Rev. C80, 034901 (2009). [33] S. R. Klein, Radiat. Phys. Chem. 75, 696 (2006). [34] S. Brodsky, T. Kinoshita and H. Terazawa, Phys. Rev. D4, 1532 (1971). [35] W. Greiner and J. Reinhardt, , Springer-Verlag, 1994. [36] N. N. Achasov and G. N. Shestakov, Sov. Phys. Usp. 34, 471 (1991) [Usp. Fiz. Nauk 161, 53 (1991)]. [37] See the mini-review “The ρ(1450) and the ρ(1700)” by S. Eidel- man and J. J. Hernandez-Rey on pg. 620 of W. -M. Yao et al., [Particle Data Group], J. Phys. G33, 1 (2006). [38] J. A. Crittenden, Exclusive Production of Neutral Vector Mesons at the Electron-Proton Collider HERA, Springer-Verlag, Berlin, 1997. [39] S. Chekanov et al. [ZEUS Collaboration], Eur. Phys. J. C24 (2002) 345. [40] A. Aktas et al. [H1 Collaboration], Eur. Phys. J. C46 (2006) 585. [41] C. Adloff et al. [H1 Collaboration], Phys. Lett. B541, 251 (2002). [42] J. Breitweg et al. [ZEUS Collaboration], Phys. Lett. B437, 432

15