[Rivet] ATLAS ttbar+jets analysis

[Andy Buckley]

On 22/07/14 15:49, Alexander Grohsjean wrote:
> Dear Andy, dear all,
> 
> I checked out the dev version and modified my stuff to get it working.
> (mainly ClusteredLepton was changed to DressedLepton).
> Attached you can find my modified/added files that are running in this
> version.
> 
> There are 3 points which affect rivet in general (except the new
> projection), so I added this to the README but would like to discuss it
> now.
> I added a p T sorting to dressedleptons, something that I couldn't find.
> If it is not my mistake and I missed it, I think
> that is something usefull to add as other projections can be sorted.

There are already sorting routines, including sortByPt, for all
containers of classes that behave like FourMomentum. I'll change the
code to do that.

> I changed the containsb function in Jet.cc to include ghost tagging. Not
> sure how you like to get this into rivet.
> There are various way of doing it and I am sure you have a prefered
> option. You can easily follow my modifications,
> there are detailed in the file. Same for adding the ghost b hadrons in
> FastJets.cc. Maybe you also want to have the same
> for c jets?

Yes, this was started a long time ago by James Monk but was never
finished. I rewrote it last week along with other Rivet::Jet /
fastjet::PseudoJet interoperability improvements, and it also does c and
tau tagging, so I should just be able to use that functionality directly
and skip these patches.

> Not sure what I can check with Roman apart from the validation I already
> did (object level for 5000 events looking at jets, leptons, cuts  and
> the final plots I provided)?
> Maybe it is useful to run, once everything is in, on a small sample and
> check it, but apart from that,
> I am not sure I can do more. Let me know.

Sounds like it's already sorted. Thanks.

> Regarding the jet gap fraction analysis. Officially (rivet page) it is
> clearly written that one needs dilepton events.
> The problem with the projection was when running on at least one lepton
> events, like we have them usually in ttbar @ 7 TeV.
> I assume Kiran et al. were using a home-made filter. In that case there
> is no problem.
> Now if you are running on ttbar events without filter, the projection
> would select you ll events and you can compare it with the data we have.
> But from a technical point everything is ok, the page clearly says
> dilepton.

Thanks again. I also discussed this in an MC physics / tuning meeting
with Stefano Camarda, to see if there would be a way to run this
analysis before the new Rivet is available. Seems not -- which is ok, I
just wanted to know if there was a pragmatic shortcut to get it into
tuning asap.

I'll merge in a version of FromElectroweakDecay now, and let you know if
I've got any more questions. Thanks.

Andy


> Am 22.07.2014 13:33, schrieb Andy Buckley:
>> On 22/07/14 11:56, Alexander Grohsjean wrote:
>>> Dear all,
>>>
>>> I was prodividing the tools that we changed in a tar bal with just the
>>> modified/added files.
>>> I summarized quickly the changes in a README in the main path.
>>> So I must admit that I am not sure what is missing here. Diff should be
>>> very easy to run and to
>>> see the changes providing this?
>> The issue is that we need a minimal diff against the latest version --
>> ideally against the 2.1.x branch head since other things have changed
>> and we don't want to just copy your files in place and overwrite those
>> other developments.
>>
>>> Changing names "FromElecroweakDecay" is perfectly fine with us, these
>>> were
>>> just historically.
>>> I started developing in 2.1.0, then updated to 2.1.1 at some point but
>>> didn't switch to 2.1.2 as this happened after my validation. Should I
>>> now run it
>>> in 2.1.2?
>> Since it's not just a new analysis, working from the *development*
>> version (i.e. the target for 2.1.3, which has evolved since 2.1.2) would
>> help us a lot with integrating these changes.
>>
>> You can get the branch head like this:
>> hg clone https://rivet.hepforge.org/hg/rivet -b release-2-0
>> then make changes and commit them if you need, and point us at your
>> cloned repo when ready. Ask if you have any questions!
>>
>>> For validation, I attached the same distributions that we have in the
>>> paper (blue and red with ct10).
>>> Should I provide the log-files from object by object comparisons?
>>> These are the internal notes:
>>> Jet multiplicity supporting note
>>> https://cds.cern.ch/record/1532076
>>> Jet pT supporting note
>>> https://cds.cern.ch/record/1545583
>> I think that's for ATLAS internal validation purposes... I'm wearing my
>> Rivet hat here, which means that I assume you and Roman have checked
>> everything and we just need to deal with the technicalities. Although
>> since there are new projections we will be pickier than with just
>> accepting a new analysis ;-)
>>
>> By the way, I saw a report from Stefano Camarda that at least the
>> important ttbar jet veto analysis (and maybe also the ttbar jet shapes)
>> do not properly require "prompt" leptons and hence the results differ
>> due to the allowed W decay channels. Could you also fix these to use the
>> FromElectroweakDecay projection?
>>
>> Thanks,
>> Andy
>>
>>
>>> Am 21.07.2014 20:59, schrieb roman lysak:
>>>>    Hi Andy,
>>>>
>>>>
>>>>
>>>> On 21/07/14 16:14, Andy Buckley wrote:
>>>>> Hi Roman,
>>>>>
>>>>> I've seen this analysis already and realised the issue. This is a case
>>>>> where it would have been nice if we could have worked with the authors
>>>>> to discuss the new projections and get them directly into the Rivet
>>>>> trunk rather than need to do it retrospectively.
>>>>>
>>>>> It would help us if you/they could provide diffs with respect to the
>>>>> latest Rivet version -- have these modifications been made on top of
>>>>> version 2.1.2?
>>>> they have been made w.r.t. version 2.1.1, as far as I know.
>>>>
>>>>>    We need to make sure that we don't undo our own
>>>>> developments when merging this. Having looked at the source of the
>>>>> FromElectroweakDecay projection, it doesn't actually do what that name
>>>>> suggests, so I would like to change that to match the sort of scheme
>>>>> that we've used for Particle.fromDecay(), or perhaps define IsPrompt /
>>>>> IsNonPrompt particle classifiers.
>>>>>
>>>>> Getting a new Rivet out with these features and some others in time
>>>>> for
>>>>> the BOOST conference in mid-August is high on my priority list, so
>>>>> I'll
>>>>> be back in touch. But if you can talk with Will and Alexander (right?)
>>>> right, cc-ing to them, so that the communication is hopefully quicker
>>>>
>>>>> to make minimal patches (or ideally an hg branch that we can clone,
>>>>> modify and merge) that we can apply, that would help a lot.
>>>> Alex, Will, could you try to do as suggested by Andy, i.e. at least
>>>> try to compare to Rivet 2.1.2?
>>>>
>>>> Thanks a lot,
>>>>    Roman
>>>>
>>>>> Thanks,
>>>>> Andy
>>>>>
>>>>>
>>>>> On 21/07/14 15:03, Roman Lysak wrote:
>>>>>>     Dear Rivet authors,
>>>>>>
>>>>>> in ATLAS, we've got another analysis we would like to eventually get
>>>>>> included into Rivet (right now, it's being validated): ttbar+jets
>>>>>> analysis.
>>>>>> However, while implementing this analysis, the authors made
>>>>>> changes to
>>>>>> some core Rivet routines (FastJet, Jet, and DressedLepton
>>>>>> projections)
>>>>>> and also added one new Projection (FromElectroweakDecay). I'm
>>>>>> attaching
>>>>>> the changes they made.
>>>>>>
>>>>>> We would like to ask you, what would be the best way to proceed:
>>>>>> whether
>>>>>> you would be willing to accept any of the updates to the core
>>>>>> routines
>>>>>> or you would prefer to have everything implemented inside the
>>>>>> analysis
>>>>>> routine (in the second case, the validation/re-validation will
>>>>>> probably
>>>>>> take longer, obviously :)).
>>>>>>
>>>>>> Thanks,
>>>>>>     Roman
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Rivet mailing list
>>>>>> Rivet at projects.hepforge.org
>>>>>> https://www.hepforge.org/lists/listinfo/rivet
>>>>>>
>>
> 


-- 
Dr Andy Buckley, Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow / PH Dept, CERN