HERWIG

HERWIG is a general purpose MC generator and its philosophy is to describe in a universal way the parton showers (where perturbative QCD applies) and the hadronization phase (that is modelled with cluster fragmentation). Therefore the differences between DIS and photoproduction lies only in the generation of the hard subprocess, which needs specific phase space cuts to define the base kine- matics and the selection of definite set(s) of parton density distributions. For tuning purposes most of the relevant parameters have then to be thought as common to both the processes. The most recent version agreed upon the authors is 5.8d, which is found in http://surya11.cern.ch/users/seymour/herwig/. It has several important updates for DIS and photoproduction and it is warmly suggested.

Here is a list of the main parameters and switches of HERWIG which are relevant for a tuning of this last version 5.8d.

General Parameters

* QCDLAM (def.: 0.18 GeV)	Lambda QCD (tuned at LEP, a more recent fit
                                gives 0.163)

* RMASS(1-6) (def.: 0.32,0.32,0.5,1.8,5.2,170. GeV)                    
                                Quark masses, used in all stages of the 
                                processing. For light flavours the values define
                                the quark shower cutoff when added to VQCUT.

* RMASS(13) (def.: 0.75 GeV)    Gluon effective mass (tuned at LEP, 
                                a more recent fit gives 0.65), it defines
                                the shower cutoff for gluons when added to 
                                VGCUT. It has to be at least twice the mass 
                                of the lightest quarks.

* VQCUT (def.: 0.48 GeV)        Quark virtuality cutoff (tuned at LEP).
* VGCUT (def.: 0.10 GeV)        Gluon virtuality cutoff (tuned at LEP).

                                With these settings the cutoff for shower 
                                termination is Q_g=VGCUT+RMASS(13)=0.85 GeV for 
                                gluons and Q_q=VQCUT+RMASS(i)=0.8 GeV for 
                                quarks. The cutoffs have to satisfy the 
                                condition: 1/Q_i + 1/Q_j < 1/QCDL3 for either i
                                or j or both gluons and QCDL3 the equivalent 
                                3-flavour Lambda, to avoid divergences in the
                                parton showers.
                                 
* VPCUT (def.: 0.4 GeV)         Photon virtuality cutoff (range 0.1-1 GeV from 
                                LEP).

* CLMAX (def.: 3.35 GeV)        Maximum cluster mass parameter 
                               (tuned at LEP, a more recent fit gives 3.48 GeV).
* CLPOW (def.: 2.0)             Power in maximun cluster mass 
                                (recent LEP fit: 1.49).
* CLSMR (def.: 0.0)             Width of gaussian angle smearing in cluster 
                                decays (recent LEP fit: 0.36).
* PSPLT (def.: 1.0)             Power in mass spectrum of splitted clusters.
                                Not much relevant at LEP, it governs the tail of
                                high mass clusters (which are supposed to be a
                                few).

* PWT(3) (def.: 1.0)		Weight for s quarks (recent LEP fits: 0.83)
* PWT(7) (def.: 1.0)            Weight for diquarks (recent LEP fits: 0.74)
* DECWT  (def.: 1.0)            Weight for decuplet baryons 
                                (recent LEP fits: 0.77)


Parameters which could be tuned at HERA:


* QSPAC (def.: 2.5 GeV)         Lower Q limit in spacelike evolution. 
                                It can be lowered down to the initial scale of
                                the chosen p.d.f. set when this is below
                                2.5 GeV, with the limit of a minimum scale of 
                                0.8-1.0 GeV like the infrared cutoff in showers.

* PTRMS (def.: 0.0 GeV)         R.M.S. Intrinsic Pt of partons in incoming 
                                hadrons (suggested range 0.-1. GeV).

* BTCLM (def.: 1.0 GeV)         Mass parameter in the fragmentation of remnants
                                (with the default a nice agreement of version
                                 5.8d with HERA Et flows is obtained).

* ISPAC (integer switch):       Meaning of QSPAC              
                 0 (def.):      same as previous versions of the program: 
                                at QSPAC the spacelike evolution stops and 
                                branching of non-valence to valence partons 
                                is forced.
                 2 (suggested): spacelike evolution continues to the infrared 
                                cutoff but the structure functions are freezed 
                                at QSPAC.

Soft Underlying Event option

HERWIG can generate a minimum bias soft collision superimposed to the hard scattering. Presently this model is suggested for resolved photoproduction, where a lack of activity is observed with most of the generators. On the contrary it is recommended to switch it off for DIS, by running process IPROC=19000 rather than 9000 (this is particularly true from version 5.8d onward), since it largely overestimates the energy flow in DIS at low x. The Soft Underlying Event generation is controlled by two parameters: 
* ENSOF (def.: 1.)            Charged multiplicity is that given from a P-Pbar
                              parametrization at energy ENSOF*Sqrt(s)

* PRSOF (def.: 1.)            Probability of the Soft Underlying Event

For photoproduction only

* PHOMAS (def.: 0.0 GeV)        Damping parameter for off-shell photon structure
                                function (suggested range: QCDLAM - 1.0 GeV)

* BREIT (logical switch, def.: .TRUE.) 
                                It defines the frame where kinematic 
                                reconstruction takes place, defaulting to the 
                                Breit frame in DIS. For photoproduction it is
                                suggested to use BREIT=.FALSE., in that case the
                                hadronic c.m. is used. This avoids possible 
                                thresholds effects due to the implemented prote-
                                ction against numerical problems.

Leif Lönnblad