This page is only relevant for FastJet 2 users
Updated 2011-05-21 with links to a couple of new tools
For the next version of FastJet we plan to include a set of
"Tools" to help deal with jets subsequent to finding them.
On this page we're starting to collect some of the ingredients
that may go into the tools part of the next release. We're
not quite sure what the final interfaces will look like, so
anything that's below should be taken as subject to future
evolution (and the documentation isn't quite there yet either).
This page also includes links to other people's code that
provides related functionality.
To get started with these tools, have a look also at
the Boost 2010 tutorial.
If you have requests, or code that you would like to contribute, let
Boosted object finding
(Butterworth, Cox and Forshaw).
You just need to call
the ClusterSequence::exclusive_subdmerge(...) method on
- Higgs tagging as performed
(Butterworth, Davison, Salam and Rubin) is
available in any recent distribution of FastJet as
example/fastjet_boosted_higgs.cc, which also illustrates the
- The Johns Hopkins Top Tagger
Kaplan, Rehermann, Schwartz and Tweedie), JHTopTagger.hh. The
method is by the Johns Hopkins group, the code is by the
- The Centre of Mass Top Tagger, by Gavin Salam
(unpublished), which performs comparably to the Johns Hopkins
one, but involves fewer parameters and sculpts the W mass
distribution less in background
- The Seattle Pruning method of
(Ellis, Vermillion and Walsh) is
available as a FastJet plugin from its authors'
- A simple generic Cambridge/Aachen subjet finder (as
Butterworth, Ellis, Raklev and Salam), CASubJet.hh,
designed to have fewer tunable parameters than the boosted Higgs
tagger and to give relatively unsculpted backgrounds (dm/m).
- The HEPTopTagger, based
and 1006.2833 is
available on request from Tilman Plehn,
- The multivariate W-jet tagger from Cui, Han and Schwartz
has a page of its own.
Some of the code extracts above make use of
helper class and
all the ones from the FastJet group assume that the
ClusterSequence is based on the Cambridge/Aachen algorithm (though
they might not always check that assumption...).
- Filtering, introduced in
for boosted particle reconstruction, and applicable also for
limiting UE/pileup in normal jets:
breaks a jet (or several jets) into
subjets on an angular scale Rfilt and keeps only
the nfilt hardest.
The code, Filter.hh
and Filter.cc, by the
FastJet authors, is to be used for post-processing jets, and
also includes two further options:
subtracting noise before choosing which subjets to filter
2) keeping subjets above some pt cut, which is closely related to the
"Trimming" proposal by Krohn, Thaler & Wang (next entry) and
which they found improves over just tuning nfilt
and Rfilt in dijet mass reconstructions.
The version of 2010-11-05 augments FilteredJet so that it
has a constituents() member function (the exact interface will
still evolve in the future and is previewed in the 3.0alpha2
release of FastJet).
- Trimming, arXiv:0912.1342
by Krohn, Thaler & Wang, available as a plugin from their