mcplots.cern.ch

March 2018 - A. Karneyeu, P. Skands
Reference: Eur Phys J C74 (2014) 1 (arXiv:1306.3436)

Note! donate your unused CPU cycles to mcplots, via the Test4Theory project (based on LHC@home).

← Select beam, process, and observable

Navigate these pages by using the menu to the left. The default for each topic is a comparison of a small number of models to available data, but look for links at the top of each page for comparisons with more tunes/generators. Scroll down each page to see plots at other collider energies. To choose specific generator version(s), use the Generator Versions link towards the top of the menu (the default is to just display the most recent ones). More plots will be added, as new tunes become available, and as the available data increases.

Note: For a description in layman's terms, and/or to find out how to donate your unused CPU cycles to help generate more statistics for mcplots see the Test4Theory project, the first volunteer cloud computing project to be based on Virtual Machines via the LHC@home platform.


MCPLOTS

MCPLOTS is intended as a simple browsable repository of MC (Monte Carlo) plots comparing High Energy Physics event generators to a wide variety of available experimental data, for tuning and reference purposes. Apart from individual plots contained in papers and presentations, there has not previously been any central database where people can quickly see how tune X of version Y of generator Z looks on distribution D. The idea with mcplots is to provide such a repository, mainly based on the Rivet analysis tool.

On each page, selected by clicking in the menu on the left, you will see a number of plots at different collider energies, comparing a set of default generators/tunes to each other and to the data (if there is any). If data for several different fiducial cuts are available at the same energy, these will be arranged horizontally. At the top of the page, you can select other generator/tune groups. The default ("Main") is a comparison of the main generators. Each additional tab then gives access to illustrations of more specialized tunes/variations for each generator.

The millions of different possible combinations of beam energies, cuts, generators, versions, and tunes, means that a lot of attention has been paid to how to simplify the site, in particular how to organize the site visually, and how to minimize the number of user clicks/inputs required to get what you want. The current site is based on a prototype developed by P. Skands in the context of workshops in Les Houches and Perugia [LH08, Ska10], but obviously it also draws heavily on the lessons and tools developed as part of the JetWeb project, in particular the Rivet analysis tool. The result is intended to look and feel much like any old-style browsable/clickable set of pages, with the more advanced technologies it is based on having been hidden 'under the hood', for the most part.

Ultimately, we aim to include a possibility for more interactive use of the site, with a special "USER" generator/tune group in which users can compose their own comparisons and even upload their own files for overplotting.


Missing an Interesting Plot / Tune / Generator?

Write to P. Skands (peter.skands.AT.monash.edu) if you are interested in adding more analyses, distributions, generators, tunes, etc to this site. We have endeavoured to make this as easy as possible. The steps are roughly as follows:

  • Start by checking out a copy of the mcplots SVN repository:
    svn co https://svn.cern.ch/reps/mcplots/trunk
        mcplots-trunk
    This will allow you to experiment with modifying files locally, do test runs (most easily on CERN lxplus which has the required architecture in place) and communicate changes easily to the mcplots authors, using the output of svn diff.
  • If your plot is based on Rivet, the procedure is almost automatic, with the user just having to furnish a few additional pieces of information; see section 4 in arXiv:1306.3436 for explicit instructions. Note: to keep the mcplots site tidy and well organised, it is important to make judicious use of the distinction between 'observables' and 'cuts'. The former define what labels will appear in the left-hand menus, while the latter will appear together as different plots on the same 'plot plage', for the chosen observable label. As an example, for a set of differential jet shape measurements using different pT cuts and different jet algorithms / R values, we would normally use 'Differential Jet Shape' as the 'observable' for all of them and let the different detailed definitions be referred to as different 'cuts'; i.e., we use the word 'cut' in a general sense to refer to different specific definitions of the same generic 'observable'. This avoids cluttering the left-hand menus up with observable definitions that only differ by small details, and also allows to see plots that only differ by such small details together side by side on the same plot page.
  • If you want to add a new generator, this requires a little bit more work. The automated computation back end and volunteer computing setup make use of the LCG GENSER releases, so having the generator included in such a release is normally the first step. See section 6 in arXiv:1306.3436 for further explicit instructions.
  • To test if an updated set of MCPLOTS configuration files can be used to successfully generate events, you can use the runRivet.sh script in local or lxplus mode; see section 5 in arXiv:1306.3436
  • for further instructions.
  • Finally, to test how updates to the database would look on the web site, you would need access to a php-enabled web server. Detailed instructions on how to do this are beyond the scope of this help text. Get in touch with the authors, and we can test your modifications eg on our mcplots-dev test server.

If you are having trouble, or if documentation is lacking for what you want to do, just write to us for information on how to get your favourite plot / tune / generator up on this site.

Problems with the Site?

Write to A. Karneyeu (anton.karneyeu.AT.cern.ch) if you are experiencing technical problems with this site, or to report problems with interpreting the contents of the site, such as menu names, insufficiently labeled cuts or plots, ranges, etc.


Support

The MCPlots site is being developed with participation by

Funds may be available to support development and integration of new features / new analyses / new tools on the site.

Write to P. Skands (peter.skands.AT.monash.edu) if you have a project to propose and would be interested in a short-term stay at CERN.