[Rivet] undocumented behaviour in wfinder/zfinder (bug ?)

[Frank Siegert]

Hi Nicolas,

On 02/08/11 18:34, Nicolas Morange wrote:
> I've read the new code, and it seems correct to me. Thanks !

Thanks for looking at the code again and for your very thorough 
analysis, it is much appreciated.

> Maybe one little thing (I don't understand fully how projections work): is it
> normal that line 124 of ZFinder.cc is commented ?

As you noticed, I removed the comment, because the projection system 
will take care of that automatically (in fact, it wouldn't even work if 
those lines were active).

> I was able to discover this bug because I'm studying Z+b, where the other
> leptons in the event play a much more important role than in Z+jets, because
> of the semi-leptonic decays. The overall effect was ~3% on the inclusive
> cross-section.

Hopefully it'll work properly now, please let us know if you spot any 
more strangeness while using Rivet.

Cheers,
Frank


>
> Cheers,
> Nicolas
>
>
> Le Mardi 2 Août 2011 12:12:06 Frank Siegert a écrit :
>> Hi Nicolas,
>>
>> I have revamped the implementation of [WZ]Finder to correctly exclude
>> particles from the RemainingFinalState:
>>
>>     http://projects.hepforge.org/rivet/trac/changeset/3268
>>
>> IMO it's much easier to read now, so maybe you could take another look
>> and check whether you see any remaining inconsistency. You can browse
>> that revision at
>>
>>     http://projects.hepforge.org/rivet/trac/browser/trunk?rev=3268
>>
>> The idea now is that the WFinder and ZFinder are FinalState projections
>> which have the original constituents as _theParticles and the W or Z as
>> _bosons. This makes it easier to exclude the original particles from the
>> remaining final state while still allowing for re-using the existing
>> projections (one of which, LeptonClusters, is also much simpler now).
>>
>> These changes don't seem to have a large effect, but they do have an
>> effect (very small) of increasing jet multiplicity, as would be expected
>> since we include more particles in the jets now:
>>
>> http://fsiegert.web.cern.ch/fsiegert/tmp/wzfinder
>>
>> This will be released very soon as Rivet 1.6.0.
>>
>> Best,
>> Frank
>>
>> On 29/07/11 19:54, Nicolas Morange wrote:
>>> Hi Frank,
>>>
>>> Thanks for your quick answer.
>>> For what I see all the other leptons are removed, for
>>> LeptonClustersConstituents (which contains the particles that will
>>> eventually be removed) doesn't know about pT and eta cuts. That seems
>>> normal, as only the clustered leptons have to pass the cuts.
>>>
>>> As a quick&dirty workaround I'm trying to merge :
>>> zfinder.remainingFinalState()
>>> + zfinder.getProjection<FinalState>("LeptonClusters")
>>> - zfinder.constituentsFinalState()
>>> which adds the composite clustered leptons to the final state.
>>> that's not completely correct but that should do a decent job.
>>>
>>> Cheers,
>>> Nicolas
>>>
>>> Le Vendredi 29 Juillet 2011 19:30:41 Frank Siegert a écrit :
>>>> Hi Nicolas,
>>>>
>>>> I'm afraid you are right, there is an inconsistency. It is not
>>>> removing
>>>> all leptons though, but "only" the ones which pass the initial lepton
>>>> cuts. But that's still more than the ones used as Z, and thus wrong.
>>>>
>>>> It's not obvious to me how to solve this with the current approach of
>>>> chaining existing projections (LeptonClusters and InvMassFinalState in
>>>> this case). I would need to be able to access the original particles
>>>> which form the two LeptonClusters accepted by IMFS. So there doesn't
>>>> seem to be a way around duplicating LC and IMFS into a combined
>>>> projection, which might as well be the ZFinder then.
>>>>
>>>> I'll think about this for an hour and if I (or anybody else on this
>>>> list) haven't come up with a better solution by then, I'll start
>>>> working
>>>> on it.
>>>>
>>>> Cheers,
>>>> Frank
>>>>
>>>> On 29/07/11 16:46, Andy Buckley wrote:
>>>>> Hi Nicholas,
>>>>>
>>>>> I'm a bit uncertain about exactly how the remainingFS works for
>>>>> WFinder
>>>>> and ZFinder... hopefully someone can clarify and check that there is
>>>>> no
>>>>> bug: Frank S, can you help?
>>>>>
>>>>> We're just about to make a new Rivet release, and I just uploaded
>>>>> the
>>>>> release tarballs to the HepForge downloads area... but I'll wait for
>>>>> confirmation that there's not a bug before making the release
>>>>> official.
>>>>>
>>>>> Cheers,
>>>>> Andy
>>>>>
>>>>> On 29/07/11 15:29, Nicolas Morange wrote:
>>>>>> Hello,
>>>>>>
>>>>>> I'm writing a Z+jets analysis.
>>>>>> Thus I use a ZFinder whose remainingFinalState() is chained to a
>>>>>> FastJet. However, the code in ZFinder
>>>>>>
>>>>>> 00044 LeptonClusters leptons(fs, bareleptons, dRmax,
>>>>>> 00045 clusterPhotons, excludePhotonsFromRFS,
>>>>>> 00046 etaRanges, pTmin);
>>>>>> 00047 addProjection(leptons, "LeptonClusters");
>>>>>> 00048 InvMassFinalState imfs(leptons, std::make_pair(pid, -pid),
>>>>>> m2_min,
>>>>>> m2_max);
>>>>>> 00049 addProjection(imfs, "IMFS");
>>>>>> 00050
>>>>>> 00051 VetoedFinalState remainingFS;
>>>>>> 00052
>>>>>> remainingFS.addVetoOnThisFinalState(leptons.constituentsFinalState
>>>>>> ());
>>>>>> 00053 addProjection(remainingFS, "RFS");
>>>>>>
>>>>>> gives me the impression that not only the Z leptons are removed
>>>>>> from
>>>>>> the
>>>>>> remainingFS, but also the other leptons in the event.
>>>>>> Is this correct ?
>>>>>> And how should I do to remove only the Z leptons (and associated
>>>>>> photons) from
>>>>>> the final state ?
>>>>>>
>>>>>> Cheers,
>>>>>> Nicolas
>>>>>> _______________________________________________
>>>>>> Rivet mailing list
>>>>>> Rivet at projects.hepforge.org
>>>>>> http://www.hepforge.org/lists/listinfo/rivet