How to do jet flavor matching in sherpa samples?

[Si.Xie@cern.ch]

Hi,

I would like to know if there is a standard way to associate jets with partons in the decay chain in order to figure out if a particular jet was a b-jet, charm, or otherwise.

In CMS we are using an algorithm given by the example code below, however I believe that this may assume pythia decay chains which is maybe different from sherpa (??).

​http://cmssw.cvs.cern.ch/cgi-bin/cmssw.cgi/CMSSW/PhysicsTools/JetMCAlgos/plugins/JetPartonMatcher.cc?revision=1.12&view=markup

Using this jet flavor matching scheme, I get a btagging efficiency of only 25% with the sherpa sample, while in a pythia qcd sample it is about 65%, so I think the matching scheme is not working correctly.

I have also tried a very simple scheme of just looking for a status 3 b quark inside the jet cone (deltaR < 0.5), but this yields similar btagging efficiency of 30% instead of 60-70%. Looking at a few example events, it wasn't very clear to me how the shower is exactly evolving as sometimes you have also status 11 bquarks showing up later on.

So, I'm wondering what kind of algorithm is typically used to trace whether a particular (gen)jet for example is a heavy flavor jet in sherpa.

Thanks. Si

[marek]

Hi Si,

to give you an overwiew, the Sherpa status codes are (1-4 are HepMC (quasi-)standard): 1 - stable particles (leaving the generator without decaying) 2 - unstable particles (physical particle decaying inside Sherpa) 3 - particle associated with the hard interaction, ie. matrix element 4 - beam particle 11 - documentation particle (not necessarily physical), e.g. particles from the parton shower, intermediate states, helper particles, etc.

This means, when you do a jet-parton matching (in deltaR) using hadron level b-tagged jets as input you will find b-quarks with status 3 in cases where the b-quark was part of the hard interaction and b-quarks with status 11 in cases where the b-quark originated in the parton shower (as it contains g->bb splittings). Of course, if the b-quark is part of the hard matrix element (with status 3) it also enters and leaves the shower, giving rise to another entry with status 11 in the event record. Please note that Sherpa does not/cannot resolve the individual shower steps in the event record, but only gives its final list of produced partons.

Also not, that once you use a multijet merged sample (CKKW) the technical distinction of a parton being produced in the matrix element or the parton shower becomes meaningless, as it is highly dependent of your merging scale Qcut.

In the end, the most sensible way to establish the quality of your tagging algorithm is to check whether in the ancestry of the particles clustered into a jet is a B-meson (for b-jets) or a charm meson (for c-jets), as only mesons are well defined physical states and always come with the well-defined generator independent status 2. Especially since the ancestry from the hadronic to the partonic regime does not follow a simple tree like structure and neither do the intra-partonic interactions.

Hope this helps, Marek