Chapter 1 General remarks and History
The increase of complexity in the scattering processes that are of interest at the next generation of colliders calls for new tools for automatic computation and event generation. Many different problems have to be tackled simultaneously: The program has to be versatile and necessarily involves a large degree of automatization since the list of multi-particle processes to be considered is much longer than can be included in a hard-coded process library like PYTHIA . The precision required for the predictions makes the traditional distinction between signal and background processes obsolete, since interference effects often cannot be neglected. Thus, massive multi-particle phase space has to be handled in the presence of many resonances and nearby singularities. Since detector effects need to be studied, the program has to have a convenient user interface, and it must be able to generate unweighted events with reasonable efficiency.
During the workshops of the ECFA/DESY study for a future Linear Collider it became obvious that no single existing package was able to meet all these needs. Therefore, the initial idea of WHIZARD was to combine known packages for generating matrix elements with a program which is able to treat generic phase space, integrate and generate events. The included matrix element generator is O’Mega  (for historic reasons and for internal compatibility and cross checks) slightly modified late 1990s’ versions of CompHEP  and MadGraph  are also available). O’Mega covers the whole set of processes that currently can be handled automatically (at tree level). The latter problem could be solved with the help of the VAMP  integration program, which extends the VEGAS algorithm to multi-channel parameterizations and thus makes it possible to handle the complex singularity patterns of multi-particle phase space in a uniform way.
The task for WHIZARD was to provide the actual phase space parameterizations, Jacobians and transformations, provide a consistent environment and to make the programs communicate with each other by common interfaces. The user had to be given a simple setup with common configuration and parameter definition files, commands to run all programs consistently without the need for manual intervention (a simple make should suffice), and an analysis system which allows for rapid inspection of the results as well as for interfacing hadronization and detector simulation programs. The program had to keep full track of beam polarization and include beamstrahlung and initial-state radiation. For hadronization, an interface to external programs should be included. Finally, it should allow for flavor summation in the final state: usually, many different processes contribute to a single final state that cannot be distinguished experimentally, and thus should be covered in a single run.
These goals have to a large extent been achieved by the current release. However, there is still room for improvement. Most prominent, for a longer time period, parts of the support for QCD amplitudes was given by the O’Mega matrix element generator only in the leading Nc approximation. This problem has been successfully solved with the 1.41 release in July 2005. However, the treatment of QCD effects still is (and probably will ever be) incomplete, which always is a problem of the matrix element generators, the shower generators and the hadronization models which e.g. do not treat interfering color structures correctly in all cases. Furthermore, these programs obviously limit WHIZARD to the physics models they can support.
The original acronym WHIZARD stands for
W, HIggs, Z, And Respective Decays
which is the class of processes (electroweak processes at e+e− colliders) the program was originally designed for. However, the present version has a much larger range of applicability, especially focusing on beyond the Standard Model (BSM) physics. Not that WHIZARD had originally been working with MadGraph and CompHEP as matrix element generators, while O’Mega joined in summer 2001. From version 1.90 on O’Mega, which had before been installed externally and its environment variables properly set within WHIZARD has been attached to WHIZARD. From that version on, MadGraph and CompHEP are switched off per default and will only be used for compatibility and consistency tests.
From January 2007 on, the new development branch of WHIZARD 2.0.0 has been opened and the new version 2.0.0 has finally been released in April 2010. Many design deficiencies of version 1 like event-dependent scales, vast support of hadron collider physics, factorized decays, new event formats like HepMC etc. have been included in the new release. With version 1.94 released in February 2010 we decided to freeze WHIZARD version 1, meaning that there will be no new feature developments in the version 1 branch. Further releases will only contain updated documentation, regression and bug fixes. For a description of the features, installation and running of WHIZARD 2 confer the abovementioned webpages and the manual and distribution files of WHIZARD version 2 directly.