Version 1.1.0 includes all bugfixes for previous versions as well as new features.

Sherpa 1.1.0

Version 1.1.0 includes all bugfixes for previous versions as well as new features (see Changelog below).

Download of Sherpa 1.1.0

The Sherpa package and a short primer on how to get started, can be downloaded here:

A Changelog for important differences between the previous version and this version is provided below.


User Interface changes

  • Particle properties are no longer specified in Particle.dat and Hadron.dat. Instead they are initialized in the code of the corresponding model and in the fragmentation handler. Overwriting particle/hadron properties is possible in Model.dat/Fragmentation.dat respectively using the tags MASS[<kfcode>]=<mass>, WIDTH[<kfcode>]=<width>, ACTIVE[<kfcode>]=<active>, STABLE[<kfcode>]=<stable> and MASSIVE[<kfcode>]=<massive>. These can also be specified on the command line. Particles and their properties will be displayed during the initialization in output level 2, while hadron properties only get displayed in output level 4.
  • QED radiation is turned on by default from the hard process as well as from hadron decays (see new physics modules below). Both of them can be turned off by the switches FS_QED_SCHEME=0 in Shower.dat and YFS_MODE=0 in Fragmentation.dat respectively.
  • The b-quark is set massless by default, and is thus contained in the jet container 93.
  • The tau-lepton is also set massless by default. Both can be overridden by the above tags.
  • Specifying parameters with blanks instead of underscores is not allowed anymore :!: For users owning old setups, we provide a migration script in TOOLS/update_datfiles (see also note about "single file" setup below)
  • The parameter Num. Accuracy changes to NUM_ACCURACY. This change is handled by the migration script as well.
  • The kf codes have been changed to comply with those given by the PDG. Most importantly, this means W<sup>+</sup> -23 -> 24, W<sup>-</sup> 23 -> -24, Z 24 -> 23, but also many non-SM particles have changed. The migration script also handles their replacement in Processes.dat (but not in Selector.dat or any other files:!:).
  • The meaning of the option to specify an electroweak or strong order for a process in Processes.dat has been changed to the exact number which is given. Before, the given number specified the maximum order with which Feynman diagrams were supposed to be generated.
  • The parameter to set the SUSY Les Houches accord input file has been changed to SLHA_INPUT (also handled by the migration script).
  • The default of the interaction model parameter SIGNAL_MODEL in ME.dat has been changed to the value of the parameter MODEL defined in Model.dat. Accordingly, the MODEL switch now defines both the model parameters and interactions to be initialized. The syntax for the available model implementations can be infered by setting SHOW_MODEL_SYNTAX=1 on the commandline when running Sherpa.

New physics modules

  • AHADIC -- a new fragmentation handler.
  • HADRONS -- a new hadron decay handler (in addition to the previously provided tau decays).
  • PHOTONS -- a new soft photon radiation handler for QED radiation off hadron decays.

New features

  • A new universal selector is provided, based on the Variable class. The variable syntax can be viewed running Sherpa SHOW_VARIABLE_SYNTAX=1. The selector syntax is

    "<variable>" <kf1>,..,<kfn> <min1>,<max1>;..;<minn>,<maxn> {<fix#1>,..,<fix#m>}|[<order1>,..,<orderm>]
  • Tag replacing in input files is enabled. Tags are specified on the command line using the syntax <Tag>:=<Value>.

  • Readers employ fallback settings for input file lookup. The search sequence is

      - PATH
      - <prefix>/share/SHERPA-MC/PATH
      - <prefix>/share/SHERPA-MC
  • A new option is provided, to set a limit for memory consumption. The corresponding tag on the command line or in Run.dat is RLIMIT_AS. It accepts a pure number, meaning the corresponding fraction of the total memory, a specification in per cent or settings in MB and GB. Examples would e.g. be RLIMIT_AS = 50 % and RLIMIT_AS = 4 GB. Note that the space between the number and the unit is mandatory, even on the command line. By default, the limit is set to the total memory minus 100 MB.
  • The grid_makelibs script is provided to allow process code compilation via the Grid see Sherpa Grid page.
  • Multithreading in Sherpa can be enabled by

configure --enable-multithread

In the current implementation different threads run in parallel for evaluating hard matrix elements and the corresponding phase space weight. This feature should only be enabled on multi-processor and / or multi-core processor machines. However, so far no significant loss in performance has been observed for single-core single processor computers. The overall computational cost in multithreaded mode is always higher than for single threads, so if only very simple processes are to be evaluated, Sherpa should better be configured for single-threaded mode (the default).

  • Dynamic add-on library loading is enabled using the tag SHERPA_LDADD = <libname 1> ... <libname n> in Run.dat or on the command line. Any object provided via getter functions within Sherpa (e. g. models) can thus be dumped into an external library and loaded at runtime. An example can be found [documentation:frameworks:external_rng here]
  • PDF information is added to the GenEvent? created by our interface to HepMC2.
  • Our interface to HepMC2 also allows for event output according to the HepMC::IO_Ascii format now.
  • Algebra interpretation is enabled in Selector.dat and the VarObs in Analysis.dat.
  • Interfacing external RNGs is enabled through the External_RNG class. See [documentation:frameworks:external_rng documentation] for details.
  • The full setup for Sherpa can be put into a single file using sectioning. The default section, which will be interpreted equivalent to Run.dat starts with (run){ and ends with }(run). The corresponding file can be generated from an existing setup using the TOOLS/update_datfiles script. This script handles the syntax migration as well (see above).
  • More reasonable default values in the code for all parameters specified by dat-files.


For process declarations a specific ordering is needed:

  • For electron-photon collisions make sure the photon is on the first beam, bunch and parton!
  • For electron-proton collisions make sure the electron is on the first beam, bunch and parton!

A few non-standard setups are not working. Do not even try:

  • Parton shower in DIS--like situations, i.e. with only one incoming leg that showers.
  • Substructure of photons.
  • Laser backscattering beyond the region where CompAZ is valid (up to 500 GeV).