Installing the Professor system

Note

If you have access to the CERN AFS file system you can try to use a version of Professor installed under /afs/cern.ch/sw/lcg/external/MCGenerators/professor.

Required packages

In order to use the Professor tools you will need the following external packages:

Numpy

The numpy datatypes are used for most numerical calculations.

SciPy

The SciPy implementation of the matrix pseudo-inversion is used.

PyMinuit or PyMinuit2

PyMinuit is used to interface with the well known MINUIT package.

Matplotlib

Most of the plotting tools require matplotlib to produce output.

prof-I requires the wxWindows backend for matplotlib.

Except for PyMinuit, all current Linux distributions should come with recent enough versions of these packages. Make sure that you get the correct version of PyMinuit: PyMinuit2 is probably the version you want, which connects to a recent version of ROOT. You may have to persuade its build system to find your ROOT system’s header files if they are not in a standard location.

Many Professor features also require the Python packages and plotting scripts which come with the Rivet MC validation toolkit. As Rivet is anyway the normal source of input data for Professor tuning, we highly recommend that you install Rivet. Detailed instructions are available on the Rivet website, and there is a handy bootstrapping script available to get everyting up and running as quickly as possible. Sorry, Professor doesn’t have a clever script like that!

Installing Professor

There is not currently a Professor release tarball: you should get the code from SVN by e.g.:

svn checkout http://svn.hepforge.org/professor/tags/professor-1.0.1

or, if you want to get the development version:

svn checkout http://svn.hepforge.org/professor/trunk/ professor-trunk

This will create a directory like professor-XXX, containing a bin and professor directory. The bin directory contains all scripts that come with Professor (and that are prefixed by prof-). The professor directory contains the library modules that are shared by the Professor scripts.

To make the Professor scripts and modules available you have several options:

1. Quick and dirty: Add the bin directory to your $PATH and professor to your $PYTHONPATH by doing:

   export PATH="/path/to/trunk/bin:$PATH"
   export PYTHONPATH="/path/to/professor-XXX:$PYTHONPATH"

   in your shell (for C-shells the above commands must be modified). This should allow you to run the Professor scripts from the command line, e.g.:

   prof-tune -h
   

   To make this persistent between shell sessions you can add the above lines to your shell’s runtime configuration, e.g. $HOME/.bashrc.

   The down-side of this is that your environment variables might get a bit overloaded and on a multi-user system this has to be done by every user.

2. More systematic: Alternatively, you can use the setup.py script that is distributed with Professor and offers the usual options, try:

   ./setup.py --help

   in the trunk directory.

   --prefix allows you to give a directory under which Professor is installed similar to the usual .configure && make && make install triple jump, e.g.:

   ./setup.py --prefix $HOME/local install

   will install the scripts in $HOME/local/bin and the modules in $HOME/local/lib/pythonXY/site-packages/ where XY is the Python version available at this instance.

Programmable completion in bash

Another useful installation detail is the prof-completion file in the Professor source directory. This provides more intelligent command-line tab-completion for the Professor scripts in bash shell terminals, with awareness of script option flags and arguments. prof-completion can be sourced explicitly with:

source prof-completion

It will be installed into $prefix/etc/bash_completion.d by installation method number 2 above, from where it may be automatically picked up by the bash shell programmable command-line completion system. As with all shell configurations, unless it is automatically read at shell initialisation or added to your $HOME/.bashrc file, it will need to be explicitly sourced in each new terminal where you wish to get its benefits.