source: git/external/fastjet/plugins/CMSIterativeCone/README@ ef8f739

Note on the implementation of the CMS Iterative Plugin

The code is initially the CMSSW implementation in
CMSIterativeConeAlgorithm.cc in /CMSSW/RecoJets/JetAlgorithms
(more precisely, Revision 14 of the cc file)

To avoid importing too many files from CMSSW, we did a couple of
modifications starting from the CMSIterativeConeAlgorithm file:

 - directly used the FastJet PseudoJet instead of the InputItem
   present initially. Note that access to the info then has to be done
   using . instead of -> which implies a few small modifications here
   and there.

 - added a small piece of code to compute Et. That peice of code is
   the same as tracked down in the CMS code. It comes from the Root
   Lorentzvector, more precisely, the PxPyPzE4D implementation.

 - The ordering in Et, NumericSafeGreaterByEt, is imported from
   EtComparator.h in PhysicsTools/Utilities/

 - The DeltaR2 computation is imported from DeltaR.cc and DeltaPhi.cc
   in CMSSW/DataFormats/Math

 - finally, instead of the ProtoJet construction, we use the
   ClusterSequence-compatible clustering. That one follows the line of
   what is done in the SISCone plugin. Note that ve've checked in the
   CMSSZ code that the protojet is indeed reconstructed using E-scheme
   recombination (while the stability test is done using Et-scheme)

Finally, we want to thank Andreas Oehler for pointing us to the CMS
code.