PreprinttypesetinJHEPstyle-PAPERVERSION 1 About the helix structure of the Lund string 1 0 2 n a J 2 1 Sˇ´arka Todorova-Nov´a ] h Tufts University p E-mail: [email protected] - p e h Abstract: The helix structure of the Lund string, first derived from studies devoted to [ the emission of soft gluons at the end of the parton cascade, may beat the origin of certain 1 v characteristic discrepancies observed in the low transverse momentum region at LEP and 7 LHC.Astudyoftherelationbetweendifferentheliximplementationsandobservableeffects 0 4 ispresented. Themodelisextendedtocoveramultipartonstringtopology(resultofparton 2 shower), and compared with the experimental data. It is found that a helix-ordered string . 1 with aregular winding(proportionalto theenergy densitystored in thestring), is favoured 0 1 by the inclusive single-particle spectra measured in the hadronic decay of Z0. 1 : v Keywords: hadronization,string, screwiness. i X r a Contents 1. Introduction 1 2. Fragmentation of the Lund string with helix structure 2 3. Parametrization of the helix string: theory 3 3.1 The Lund helix model 3 3.2 The modified helix model 5 4. Parametrization of the helix string: phenomenology 6 5. Extension of helix model on multiparton string topology 9 6. Model tuning and comparison with data 10 7. Other observables 11 8. Conclusions 13 A. Pythia implementation of the modified helix scenario 14 1. Introduction TheLundfragmentation model[1,4]usestheconceptofstringwithuniformenergydensity to modelthe confiningcolour field between partons carryingcomplementary colour charge. The string is viewed as being composed of straight pieces stretched between individual partons according to the colour flow. The fragmentation of the string proceeds via the tunneling effect (creation of a pair of a quark and an antiquark from vacuum) with a probability given by the fragmentation function. The sequence of string break-ups defines the final set of hadrons, each built from a qq¯ pair (ev.diquark in case of baryons) and a piece of string between the two neighbouring string break-ups. The longitudinal hadron momenta stem directly from the space-time difference between the break-ups. The model gives a fair description of the available high-energy hadronic data and is therefore widely used in experimental particle physics. It reproduces particularly well the particlemultiplicity andlongitudinalprofile(jetformation)buttherearecertaincharacter- istic discrepancies between data and simulation which suggest the treatment of transverse momenta may not be entirely adequate ( more in Section 6). It is therefore interesting to develop and study alternative models. A very interesting work devoted to the study of the properties of the emission of soft gluonswaspublishedbyAnderssonetal. sometimeago[2]. Undertheassumptionthatthe – 1 – generating current has a tendency to emit as many soft gluons as possible, and due to the constraint imposed on the emission angle by helicity conservation, it was shown that the optimal packing of emitted gluons in the phase space corresponds to a helix-like ordered gluon chain. Such a structure of the colour field cannot be expressed through gluonic excitations of the string and it needs to be implemented as an internal string property. 2. Fragmentation of the Lund string with helix structure The implementation of a string with a helix structure radically changes the way hadrons acquire their transverse momentum. In the conventional Lund model [1], the transverse momentumofthehadronisthe(vectorial) sumofthetransversemomentaofthe(di)quarks whichwerecreatedviaatunnelingprocessduringthebreakupofthestring. Thetransverse momentaofnewlycreatedpartonsarerandomlysampledfromagaussiandistribution(with adjustable width) and their azimuthal direction is random. In the helix ordered string, hadrons obtain their transverse momentum from the shape of the colour field itself, so that there is in principle no need to assign a momentum to new quarksinthestringbreakup. Ifwepicturethecolourfieldasastreamofsoftgluonsordered at emission, we get the hadron transverse momentum by integration over the transverse momenta of soft gluons emitted in between the string break-up points which define the hadron, see Fig.1: Φj ~p (hadron) = ~p (gluon)dΦ t Z t Φi where Φ is the ’phase’ of the helix (azimuthal angle) in the break-up point i(j). i(j) a) b) Figure 1: a) The helix structure of the string carriedby colour connected chain of soft gluons. b) After fragmentation, the transverse momentum of a direct hadron is the integral of the transverse momenta of the soft gluons, integrated over the corresponding string piece. Thetransverse momentum thehadron carries is thus entirely definedby the properties of the helix ordered field. This additional constraint translates into a loss of azimuthal degree of freedom in the string break-up, arguably the most significant consequence of the implementation of the helix string model in the fragmentation process. The underlying – 2 – helix structure is reflected in correlations between transverse and longitudinal components of hadrons which may lead to experimentally observable effects, depending on the actual form of the helix string. So far, only one type of helix string parametrization was put under scrutiny [2, 3], and no convincing experimental evidence in favour of the helix string was obtained. The purpose of the present paper is to introduce an alternative helix string parametrization, to study and compare the observable effects, and to show that the helix string model (after modification) describes the hadronic data better than the standard Lund fragmentation model. 3. Parametrization of the helix string: theory As a reminder, and for the sake of clarity, we reiterate the properties of the helix string introducedin[2]. Certainaspectsoftheoriginalimplementationwhichwerenotnecessarily addressed in the original paper, but which are relevant for the discussion, will be pointed out. 3.1 The Lund helix model In [2], the phase difference of the helix windingwas related to the rapidity difference of the emitting current by the formulae: ∆y ∆Φ = , (3.1) τ where ∆Φ is the difference in helix phase between two points along the string, ∆y is their rapidity difference, and τ is a parameter. In the Lund model, the rapidity at a given point along the string is defined as k+p+ y = 0.5 ln( ), (3.2) k−p− where p+,− are the initial light-cone momenta of the endpoint quarks and k+,− their fractions defining a position along the string, see Fig.2. The rapidity difference between two points along the string is then k+k− i j ∆y = 0.5 ln( ) (3.3) k−k+ i j and it is related to the angular difference of points in the string diagram (Fig.2). The evolution of the phase of the helix string defined according to (3.1) is illustrated for a simple qq¯stringin the stringdiagram Fig.3. Thephase is fixed by a random choice at one point of the diagram (in our example Φ=0 at [k+,k−]=[1,1]) and is calculated for the rest of the diagram from Eq.3.1 with the help of Eq.3.2. The parameter τ is set to 0.3 for definiteness, its value is irrelevant for discussion of the qualitative features of the model. Please note that the density of helix winding increases with the distance from the string center, and becomes infinite near the turning points ([k+,k−]=[0,1]/[1,0]) where Eq. 3.3 – 3 – Figure 2: Evolution of the string in the rest frame of the qq¯pair, including the first string break- up. The partons lose their momentum as they separate and the string - the confining field - is created. The space-time coordinates can be obtained from the relation [t,x]=(k+p++k−p−)/κ (κ∼ 1 GeV/fm). The x direction is parallel to the string axis. Figure 3: On the left: the phase of the helix winding along the string according to the Lund prescription (Eq.3.1) for parameter τ=0.3. The phase was fixed by a random choice at one point. Onthe right: forbetterillustrationofthephaseevolutioninthe space-timecoordinates,thepoints with equalphases areconnectedby lines in equidistantphase intervales. The singularitypresentin the model at the endpoint of the string is not graphically emphasized. contains a singularity. This singularity does not affect the modelling of simple qq¯strings as long as the endpoint quarks do not acquire transverse momentum (a default solution in Pythia [4]) but the model needs some sort of regularization in case of multiparton string – 4 – configurations. (All studies in [2] were done using a simple qq¯string.) 3.2 The modified helix model The presence of a singularity in the Lund helix model is one of the reasons why we wish to take a second look at the definition of the helix model. Also the requirement of the homogenity of the string field which lies at the heart of the Lund fragmentation model seems to be poorly satisfied, given the difference of the helix winding at the middle of the stringandnearthestringendpoints. Itshouldbeemphasizedthatwestrictlyadheretothe central idea of [2], namely the emergence of a helix-ordered gluon chain at the end of the parton cascade, and that we are merely looking into details of the helix parametrization. We derive the alternative helix model studying the properties of an elementary dipole in the gluon chain on the basis of equation 3.3 from [2]. The squared mass of the dipole formed by colour connected gluons can be written as s = k2 2 [cosh(∆y)−cos(∆Φ)] (3.4) j,j+1 T where the transverse momenta of both gluons are set to k (for simplicity) ,∆y is the T difference in rapidity, and ∆Φ difference in azimuthal angle between gluons. The original proposal for the helix string neglected the azimuthal difference in the search for gluon packing which would minimize the gluon distance yet satisfy the helicity conservation laws, andthedistancebetween gluons wasparametrized withthehelpof their rapiditydifference. Hereweintendtoreversetheapproachanddevelopahelixmodelwhich minimizes the rapidity difference between soft gluons and where the gluons are separated mainly in the transverse plane. Under the assumption ∆Φ>> ∆y ≈ 0 (3.5) equation ( 3.4) reads ∆Φ s = k2 2 [1−cos(∆Φ)] = 4 k2 sin2( ) (3.6) j,j+1 T T 2 and the distance d between gluons, as introduced in [2], becomes ∆Φ d = s /k2 = 2 | sin( ) | (3.7) j,j+1 q j,j+1 T 2 The condition d ≥ 11, derived from helicity conservation restrictions, is satisfied for 6 ∆Φ > 2.3 rad. Since there is no constraint on the length of the gluon chain ordered in azimuthal angle, the number of soft gluons in the chain will depend on k and the energy T available for string build-up. It has to be stressed however that we assume the emergence of the ordered helix field occurs in parallel with the ’homogenization’ of the string field in which the interactions between field quanta redistribute the longitudinal momenta of field creating partons, and that we can describe the string with the help of uniform energy density and string tension, much as the standard Lund fragmentation model does. . We set the difference in the helix phase to be proportional to the energy stored in between two points along the string – 5 – ∆Φ = S (∆k+ +∆k−) M /2, (3.8) 0 where M stands for the invariant mass of the string, S[rad/GeV] is a parameter, and 0 fractions ∆k+ = |k+−k+ |, ∆k− = |k− −k− | define the size of the string piece. j j+1 j j+1 Figure 4: Onthe left: the phase ofthe modified helix (Eq.3.8)forparameterS=0.5rad/GeVand invariantmass of the string M0=91.22GeV. The phase was fixed by a randomchoice at one point. Onthe right: forbetterillustrationofthephaseevolutioninthe space-timecoordinates,thepoints with equal phases are connected by lines in equidistant phase intervales. As shown in the string diagram of Fig.4, the definition Eq.3.8 corresponds to a helix with a constant pitch along the string (proportional to the energy density of the string). The phase in the modified helix scenario is constant in time for a given point along the stringaxis,formingastationarywave(similartotheinterferencepatternduetoanemission from two sources). 4. Parametrization of the helix string: phenomenology Inthissectionweturnourattention toobservableeffects relatedtothehelixorderedstring. We shall first study a simple quark-antiquark system to get a better understanding of the differences between models. In the Lund helix model [2], the phase difference is directly related to the rapidity difference. The fragmentation of the Lund string produces roughly one hadron per unit of rapidity. The hadrons, therefore, obtain – on average – a transverse momentum of about the same size, i.e. roughly independent of y , and the helix-like structure should be visible in their azimuthal angle difference as a function of the rapidity ordering. – 6 – The observable which should reveal such a behaviour was introduced in [2] Screwiness(ω) = P | exp(i(ωy −Φ ))|2, (4.1) e j j X X e j where y ,Φ stand for the rapidity and the azimuthal angle of final hadrons, P is a nor- j j e malization factor and the parameter ω is the characteristic frequency of the helix rotation. The first sum goes over hadronic events, the second one over hadrons in a single event. The expected signal for charged final particles is shown in Fig. 5. The presence of a Lund helix field is visible as a peak at ω ∼ 1/τ, but the significance of the peak, with respect to standard Lundfragmentation, decreases with τ. There is some screwiness signal in themodifiedhelixscenario, too, butitcomes inaformofamulti-peak structuredifficult to interpret. (It is worth remembering that the experimental study of screwiness [3] found a few percent difference between data and the standard Lund model but the signal did not exhibit a single peak shape.) The observable effects stemming from a modified helix parametrization are however not restricted to the hadron ordering in the azimuthal angle (though we will come back to the question in section 7). The modified helix introduces a strong correlation between the size of the fraction of string forming a hadron (i.e., the energy of the hadron in the string c.m.s.) and the size of Figure 5: Screwiness signal obtained at the generator level in hadronic Z0 decay (without parton shower), for various values of parameter τ in Lund helix model (3.1), and compared to standard Lund string fragmentation(histogram). A peak is expected atω ∼1/τ but its significance is small for low τ values (< 0.3). Modified helix model (3.8) produces a multi-peak pattern (dashed line). Screwiness is calculated from final charged hadrons with p > 0.15 GeV/c and |y|<2. – 7 – its transverse momentum. In the rest frame of the string, ∆Φ SE p2(hadron) = 4 r2 sin2 = 4 r2 sin2 had. (4.2) T 2 2 wherer[GeV] (the’radius’of thehelix) is aparameter and ∆Φis thehelix phasedifference between the two break-ups which created the hadron. The correlations are visible in Fig. 6, where a clear structure appears in the dis- tributionofdirecthadrons. Experimentally, weneverobservesuchaclearpictureofstring fragmentation, because it is smeared by the parton shower and decays. Still, these cor- relationsleavetraceintheinclusivep spec- T tra,as showninFig.7. Duetotheexponen- tially falling distribution of hadron energy in fragmentation which governs the size of Figure 6: Correlations between the transverse transverse momentum through Eq. 4.2, the momentum of a direct hadron and its energy in modified helix model creates more hadrons the string c.m.s. in the modified helix model, with very low p but less in the region p ≈ t t r=(0.4±0.1)GeV/c,S=0.5 rad/GeV. 0.4GeV/c where the peak of the uncorre- lated, gaussian distribution lies. Figure 7: Inclusive transverse momentum distribution of direct charged pions , for the modi- fied helix scenario (3.8) with different helix pitch, compared to the standard fragmentation. For comparison,thedistributionobtainedwiththeLundhelixmodel(3.1)isshown,too(dashedline). – 8 – For comparison, Fig. 7 also shows the spectrum obtained from the Lund helix model with τ = 0.7, which exhibits qualitatively similar behaviour, albeit attenuated, as the modified helix model. (Generally speaking, we may expect observables designed for one helix scenario to show some effect in the other scenario, too, but weaker and somewhat distorted, as we saw in the case of the screwiness measure.) The effect of the modified helix scenario on the inclusive p is strong enough to be T readily visibleinexperimentaldata. Asamatter of fact, acharacteristic discrepancyinthe low p distribution is visible both in LEP [5] and LHC [6] data, but before performing a T direct comparison of dataandthemodel, we needto makesurethemodelhandlesproperly the multi-parton string we use for description of the real data. 5. Extension of helix model on multiparton string topology The comparison of the helix string model with data requires the model to be extended to cover not only the simple case of qq¯ system but also an arbitrary multiparton configuration corresponding to the emission of hard gluons from the quark-antiquark dipole. This is actually the most complicated part of the model implementation which requires some ad- ditional assumptions to be made. The solution adopted for the modi- fied helix scenario consists of two steps: Figure 8: An illustration of helix phase evolution of the modified helix model in case of presence of a first, the multiparton system is followed hard gluon kink on the string. in space-time (every parton looses about 1 GeV of its energy per fermi in favour of the developing string field, the energy loss of gluon is twice as much because the gluon participates in thecreation of two adjacent stringpieces) inordertofindtheway thestring breaks into pieces, and to evaluate their respective masses (Fig.8). Every string piece is formed by a combination of the fractions of momenta of 2 partons, combinations and frac- tions depend on the distribution of partons in the phase space. The second step consists in calculation of the combined helix phase difference between endpoint quarks ∆Φ = S M , (5.1) i X i where the sum runs over all (ordered) string pieces and M is the mass of the i-th string i piece. Since the phase of the modified helix string is constant at a given point along the string, it can be easily calculated from the relative distance from string endpoints. A convenient way of doing this is to use the energy fraction. For a given string break-up, for example, one can calculate the total energy of hadrons on the left and right side. The sum of hadron energies being identical to the sum of energies of ordered string pieces, it – 9 –