HERWIG Version 6.5

Here is the source code.  Two include files are needed: HERWIG65.INC and herwig6521.inc.
This version runs successfully with Jimmy versions 4.0 and later, which can be obtained from the Jimmy web page.

The previous version was HERWIG 6.520, of August 16th 2010. Here is the source code.  The changes from version 6.520 to 6.521 are listed below.

The previous version was HERWIG 6.510, of October 31st 2005. Here is the source code.  The changes from version 6.510 to 6.520 are listed below.

The previous version was HERWIG 6.507, of March 8th 2005. Here is the source code.  The changes from version 6.507 to 6.510 are listed below.

The previous version was HERWIG 6.506, of November 3rd 2004. Here is the source code.  The changes from version 6.506 to 6.507 are listed below.

The previous version was HERWIG 6.505, of March 31st 2004. Here is the source code.  The changes from version 6.505 to 6.506 are listed below.

The version before that was HERWIG 6.504, of August 12th 2003. Here is the source code.  The changes from version 6.504 to 6.505 are listed below.

The version before that was HERWIG 6.503, of July 11th 2003. Here is the source code.  The changes from version 6.503 to 6.504 are listed below.

The version before that was HERWIG 6.500, of October 16th 2002. Here is the source code.  The changes from version 6.500 to 6.503 are listed below.

Note that since the include files are backwards compatible, there is no problem with using the latest include files with any of the 6.5xx versions.

Here is the documentation and release note.

Here is a simple test job that generates QCD 2->2 events at the Tevatron.

There is also a test job that shows how to read in SUSY input files and generate SUSY events at the LHC. To run it you will need the file sugra_pt2.in from the ISAWIG web page. Note that it produces a few warning messages when the decay tables are processed. This is due to the fact that both the built-in decay table and the one read in from file contain entries for top decays, and is not a problem: the duplicates are simply deleted.

• Alignment of initial state radiation cones (affects HWBJCO)
• Some things needed by MC@NLO (affects HWBJCO)
• Bug fix in initial-state spin correlations (affects HWBSPA)
• Bug fix in finding gauge boson pairs (affects HWDBOS)
• Bug fix in heavy object decay correlations (affects HWDHO2)
• Les Houches underlying event (affects HWHGUP)
• Bug fix in Les Houches interface for 2->1 processes (affects HWHGUP)
• Bug fix: top lifetime (affects HWUDAT)
• Bug fix for tops with Les Houches (affects HWHGUP)
• Bug fix in spinor routines to avoid divide by zero (affects HWH2F1, HWH2F2, HWH2F3)
• Bug fix in error severity for negative energy underlying events (affects HWMULT)
• Bug fix for Les Houches Higgs decay (affects HWDHIG)

• The use of the running quark mass for q qbar -> Higgs, this is to avoid the unphysically large contribution to cross section u ubar -> Higgs was giving using the constituent quark masses. (New routine HWURQM and affects HWHIGS). Note that this is not done for any other Higgs production or decay processes.
• Changes to allow the forcing of gauge boson decays in processes using the Les Houches interface (affects HWDBOS)
• A new flag ITOPRD to allow the use of PHOTOS to generate QED radiation in top quark pair production and decay. The default ITOPRD=0 is not to use PHOTOS whereas ITOPRD=1 will use PHOTOS. It should be noted that there may be a problem with double counting if the default HERWIG photon radiation in the parton shower is also switched on, however the new option will give radition for the leptons in the decay of the W which the HERWIG treatment does not. (New routines HWPHTP and HWPHTT and affects HWBCON, HWBTOP, HWDHO4, HWDTAU and HWUINC).

• New feature:
  • A major update of the Jimmy generator for multiple parton scattering has been made. Although this is still a separate package (available from the Jimmy web page) its incorporation into HERWIG is now much smoother - no HERWIG routines need to be modified or replaced. Moreover, it has been modified to run correctly in 'underlying event' mode, i.e. it can attach additional scatters to a high-pt scattering event or to events of other types, rather than always running in 'minimum bias' mode as before
• Improvements:
  • Value of MODMAX (size of MODBOS) increased to 50. Note that since MODBOS is the last member of its common block, this does not move the position of any other variables in the include file
  • Calculation of minimum invariant mass needed for a given hard process separated off from HWEGAM to a new routine HWEGAS
  • Several new features added to HWHSCT and new routine HWHSCU added to implement the underlying event mode of Jimmy
• Bug fixes:
  • Several modifications (in HWCFOR, HWCHAD and HWURES) to improve the calculation of the threshold for partonic decays of b baryons and B_c mesons
  • Hard scale variable EMSCA was not set correctly in gauge boson pair production (HWHGBS and HWHGBF)
  • In many places in the code, in several recently added processes, tests on IPROC were not performed correctly, leading to different behaviour in the same hard process, depending on whether the underlying event was switched on (IPROC<10000) or off (IPROC>10000)
  • Probability of backwards evolution to an 'anomalous' photon was not correctly calculated. New probability is somewhat larger at large x-gamma, but similar at small x-gamma (HWSFBR)
  • Protection against corruption of photon production cross sections due to infinitesimal parton distribution functions (HWHPHO and HWHPH2)
  • Corrected angular distributions in gg->tbH (HWHIGQ)
  • Corrected cross section in Higgs+jet production (HWHIGA)
  • Added protection against clusters of unknown flavour combinations (HWCHAD)

• New feature:
  • Two new parameters, PDFX0 and PDFPOW, have been provided to control the probing of parton distribution functions at extremely small x values, potentially below where they are valid. For values of x below PDFX0, x f(x,Q^2) is replaced by PDFX0 f(PDFX0,Q^2) (x/PDFX0)^PDFPOW. The default value of PDFX0 (D=0) means that the pdfs are unmodified and that the value of PDFPOW (D=0) is irrelevant. For 'valence-like' distributions at small x, set PDFX0=1d-5 for example, and PDFPOW=0. This feature is mainly needed in conjunction with Jimmy since, for PTJIM=2GeV at the LHC for example, x values down to 2x10^-8 are probed, whereas most pdf sets are not considered reliable below x values of about 10^-5
• Improvements:
  • Spin correlations in the decay of the W/Z in W/Z+Higgs events are now included
  • In several gauge boson production processes, hard coded limits on the mass distribution have been replaced by GAMMAX [D=10] widths
  • Generation of Breit-Wigner distribution is changed from m to m^2
  • Infrared cutoffs on large x values implemented in Drell-Yan matrix element correction routine HWBDYP. Makes a negligible difference
  • Running electromagnetic coupling used in W+jet routine
  • Branching fraction replaced by running partial width divided by m times nominal total width in W production routine HWHWPR. This is only different in the rare events in which the chosen W mass is above the top threshold
  • More accurate calculation of hadron remnant mass used in HWSBRN
  • PDFSET is only called if the pdf set has changed since the last call, saving cpu time
• Bug fixes:
  • Kinematics were not calculated correctly in Drell-Yan matrix element correction routine HWBDYP. Has a tiny effect on W/Z kinematics, but more significant on the distributions of their decay products, giving too hard a lepton pt distribution at very high pt for example
  • In several Higgs and/or gauge boson production processes, including W+jet, W/Z+H and H+jet, Breit-Wigner was generated twice, leading to a too wide mass distribution by a factor of SQRT(2)
  • Mass distribution of W/Z in W/Z+H and W/Z+jet production were incorrect by a factor m^2/m_0^2, where m_0 is the nominal mass
  • Subprocess bg->Wt was previously included in W+jet routine in the massless quark approximation. This subprocess has been switched off
  • Off-shell hadron remnants sometimes led to momentum non-conservation in secondary scatters generated by Jimmy. Fixed by shuffling small amount of momentum between the two remnants in HWHREM

• Bug fixes:
  • Space-time production position of hadrons from the underlying event was previously non-sensical. Now the cluster position is chosen according to a Gaussian distribution in its rest-frame and the hadron positions correctly take account of the position of the primary interaction point. Implemented in HWMEVT
  • Space-time production position of leptons and photons from gauge and Higgs boson decays was previously set to the origin. Now calculated correctly taking account of the position of the primary interaction point. Implemented in HWBJCO, HWCFOR, HWDBOS and HWDHIG
  • A bug has been found in the interface to Jimmy that led to errors (stable quarks in the final state and/or divide-by-zero crashes) when adding multiple scatters to Drell-Yan-type processes. Fixed in HWHSCT
  • MRST pdfs did not previously freeze at QSPAC in the ISPAC.GT.0 options, as they should. Fixed in HWSFUN
  • Printed version number was incorrect in several places in the LaTeX and html output formats. Fixed in HWUDPR and HWUEPR
  • A minor improvement to the formatting of the printed event record for minimum bias events, implemented in HWUEPR

• Bug fixes:
  • A bug has been found in the kinematic reconstruction of parton showers, that had particularly severe consequences for top quark decay. This step involves reshuffling a small amount of momentum between jets in order to restore overall momentum conservation. It is not Lorentz invariant, so the jet momenta must be boosted to and from the frame in which it is performed. In versions 6.504 to 6.507, these two boosts were performed via different intermediate frames, and therefore a (Thomas) rotation was induced. This lead to a significant shift in the direction of the b jet, even in events in which there was no radiation from it.
    Note that this fix has already been circulated to several experimental collaborations, who were asked to call the fixed version 6.508.
  • A bug in the logic of the azimuthal correlations between two back-to-back jets was present in versions 6.503 to 6.507, leading to a slight (a few parts per mille) asymmetry in the azimuthal distribution of produced hadrons in hadron collisions.
  • A minor bug was fixed in tau decays using TAUOLA, which had lead to Lorentz-noninvariance of some spin effects.
  • A variable, TMPRN, was used, but not declared, as an array in HWHSCT, leading to crashes or unpredictable results when using Jimmy on some machines.
  • Our particle numbering distinguishes Standard Model and BSM Higgs bosons, while the pdg use the same number for the SM and lightest CP even BSM Higgs. Our numbering is now only used internally and is converted to the pdg convention at the end of event processing.
• Changes made to reduce machine/compiler dependence:
  • Multiple-ENTRY routines have been replaced by independent routines.
  • Alternate return points have been removed. Note that in particular, this affects HWWARN, so users who use HWWARN should change their call accordingly:
           CALL HWWARN('HWANAL',ICODE,*999)
    should become
           CALL HWWARN('HWANAL',ICODE)
           IF (ICODE.LT.0.OR.(ICODE.GE.50.AND.ICODE.LT.200)) GOTO 999
  • An arithmetic IF statement has been removed.
  • All variables in DATA statements have also been SAVEd.
  • Line numbers have been removed from END statements.
  Several of these are necessary to compile under gcc4.0. Note that the first point necessitated moving the random number seeds from a SAVEd local variable to a COMMON block. We had some concerns that this may slow the program, but in fact on several compilers and machines we find that the new version is actually faster. Nevertheless, if you encounter significantly slower performance with the new version, please inform us, with details of the machine, compiler and compiler options.
  We are grateful to Mikhail Kirsanov and the GENSER group of the LCG project at CERN for their help with some of these changes.

• New options:
  • Added the option to indicate that the input Les Houches event already includes a truncated shower (as needed for the POWHEG method of Nason et al). If TRUNSH=.TRUE., the gluon's maximum opening angle is the angle to its nearer colour partner. If TRUNSH=.FALSE. (the default), the behaviour is as in previous versions: the maximum opening angle is the angle to one of its two colour partners, chosen with equal probability.
  • Added the option to indicate that resonances in an input Les Houches event should be parton showered with an upper scale of SCALUP (if PTVETO=.TRUE.) rather than with an upper scale of the mass of the resonance (if PTVETO=.FALSE., the default).
  • Modified Les Houches Accord routine HWHGUP to work with photoproduction in MC@NLO.
  • Introduced parameter DQXTRA (default 0.2 GeV), added to spectator mass in partonic decays of b- and c-flavoured baryons only, to facilitate cluster decay.
• Bug fixes:
• Included complete spin correlations in H->WW (previously used average over azimuth around WW axis).
• Added vertex information for tau decays performed by TAUOLA. (Thanks to ATLAS for providing this fix.)
• Swapped decay modes of BL_S10 meson and its antiparticle (which previously violated beauty and strangeness conservation).
• Fixed bug in Higgs lineshape option: affects gluon fusion cross section by ~9% for 700 GeV Higgs for example, much less for lower masses, also less for other Higgs production processes.
• Phase space and matrix elements were incorrect in VBF to Higgs processes, which previously led to forward-backward asymmetric jet distributions.
• Factorization scale for VBF to Higgs changed to Higgs mass (was W mass).
• Fixed bug in H->WW spin correlations - Lorentz boost should be done in two steps to avoid 'Thomas rotation' effect.
• Improved treatment of mass effects in Drell-Yan matrix element correction routine HWBDYP, which led to strange rapidity distribution in low-mass Drell-Yan at the LHC (i.e. when one of the incoming partons is at very low x).
• Fixed book-keeping bug in Les Houches Accord routine HWHGUP and potential infinite loop there if given an inconsistent event. (Thanks to ATLAS for providing fixes for both.)
• Fixed bug in W production routine HWHWPR if GAMMAX>m_W/Gamma_W.
• Fixed an inconsistency in use of IPRINT when it is greater than 2.
• Added missing SAVE statement for IQ1 in HWHHVY.
• Added RETURN if HWBGUP (showering of Les Houches events) fails.
• In HWETWO, redefined x as fraction of p3, not p+, to prevent small-x partons from going in opposite direction to their parent hadrons.
• Fixed bug in HWBJCO which sometimes led to non-conservation of rapidity of hard process when this was demanded (PRESPL=.FALSE.).
• Fixed potentially unsafe code in HWCGSP (gluon splitting).
• Fixed sign error in polarization correlations in e+e- --> ZH.
• Inserted missing identical-particle factor of 1/2 in diphoton production.
• Changes made to reduce machine/compiler dependence:
• Fixed minor compilation warnings on gfortran -Wall. Meaningless use of some dummy arguments added, others removed, to reduce warnings.
• Removed tab characters from .inc file.
• Moved DATA statements in HWDSM2.

Version 6.521 has two bug fixes and a new feature:

• A fix to handle leptons with off-shell (zero) masses input from a Les Houches file.
• A fix for top decays to non-b quarks.
• A parameter FSHWR (default 1.0) multiplying the scale at which PDFs are evaluated in initial-state parton showering ONLY, for use in studies of scale dependence.