Skip to content

FastJet: a code for fast kt clustering, and more

Matteo Cacciari

hep-pharXiv:hep-ph/0607071

Abstract

Two main classes of jet clustering algorithms, cone and kt, are briefly discussed. It is argued that the former can be often cumbersome to define and implement, and difficult to analyze in terms of its behaviour with respect to soft and collinear emissions. The latter, on the other hand, enjoys a very simple definition, and can be easily shown to be infrared and collinear safe. Its single potential shortcoming, a computational complexity believed to scale like the number of particles to the cube (N3), is overcome by introducing a new geometrical algorithm that reduces it to N ln N. A practical implementation of this approach to kt-clustering, FastJet, is shown to be orders of magnitude faster than all other present codes, opening the way to the use of kt-clustering even in highly populated heavy ion events.

Create a lesson