Search for Fourth Generation Quarks at LHC in light b' sector


With previous our studies at the _B_-factories, hints of physics beyond the Standard Model (SM) are emerging. The 4th generation quarks could be one of the candidates. This study covers the low mass region of the possible b' scenario.

The 4th generation quarks are so called b' and t' which are theoretically viewed as heavy b -like and heavy t -like particles. The searching scenario can be divided into light b' search and heavy b' search depending on the b' mass is below or above the tW mass threshold. In the light b' case, b' -> cW likely dominants for low b' mass. The possibility of Flavor Changing Neutral Current (FCNC) b' -> bZ at % level or higher makes the program very interesting. This is our focused topic.

MC Samples

  • 1_3_x and 1_4_x study is done with home made MC (obsoleted)

  • 1_6_x study uses the CSA07 MC production (obsoleted)

  • 2_2_x study now use Summer08 and Fall08 MC production
    • /Exotica_BPrimeCW_M200/Summer08_IDEAL_V9_v2/GEN-SIM-RECO
    • /Exotica_BPrimeCW_M225/Summer08_IDEAL_V9_v2/GEN-SIM-RECO
    • /Exotica_BPrimeCW_M250/Summer08_IDEAL_V9_v2/GEN-SIM-RECO
    • /TTJets-madgraph/Fall08_IDEAL_V9_v1/GEN-SIM-RECO
    • /ZJets-madgraph/Fall08_IDEAL_V9_v1/GEN-SIM-RECO
    • /WJets-madgraph/Fall08_IDEAL_V9_v1/GEN-SIM-RECO
    • /VQQ-madgraph/Fall08_IDEAL_V9_v1/GEN-SIM-RECO
    • /WW_2l/Summer08_IDEAL_V9_AODSIM_v1/AODSIM
    • /WZ_incl/Summer08_IDEAL_V9_AODSIM_v1/AODSIM
    • /ZZ_2l2n/Summer08_IDEAL_V9_AODSIM_v1/AODSIM
    • /ZZ_4l/Summer08_IDEAL_V9_AODSIM_v1/AODSIM

Physics Process Generator Cross-section Processed evt. Status Comments
b'b'->bZcW, m(b')=200 GeV pythia6 2.84 pb 10.7K done BR(bZ/cW)=50%
b'b'->bZcW, m(b')=225 GeV pythia6 1.59 pb 10.3K done BR(bZ/cW)=50%
b'b'->bZcW, m(b')=250 GeV pythia6 0.89 pb 10.1K done BR(bZ/cW)=50%
tt + Jets MadGraph 317 pb 973K done RECO
Z + Jets MadGraph 3700 pb 1074K done RECO
W + Jets MadGraph 40000 pb 9683K done RECO
V + QQ MadGraph 289 pb 1021K done RECO
WW -> 2l Pythia6 3.65 pb 106K done AOD
WZ -> incl. Pythia6 30.6 pb 249K done AOD
ZZ -> 4l Pythia6 0.05 pb 267K done AOD
ZZ -> 2l2n Pythia6 0.09 pb 113K done AOD

* Cross-section including sub-branching fractions.

Analysis Procedures

Analysis code

The analysis is first starting with CMSSW_1_2_3 using all basic physics objects. Later we migrate to CMSSW_1_3_6 and CMSSW_1_6_12 as new software available and the old ones obsoleted.

The latest version of light b' analysis code is using the CMSSW_2_2_3 Physics Analysis Toolkit (PAT) objects. The source code including Grid job scripts (CRAB) can be found here: MyG4Kit.tgz: MyG4Ana in PAT

People in Interests

| Institution | Related People |RECO
National Taiwan University Yuan (John) CHAO (light b' contact)*; Rong-Syang Lu and Kai-Feng Chen (heavy b' sector); Paolo Bartalini (generator contact); Wei-Shu (George) Hou (theoretical direction)
National Central University Chai-Ming Kuo, Wan-Ting Chen (b'b'->cWcW working group)

Issues and Tasks

  • Analysis note draft with CMSSW_1_6_12

  • New updated draft with CMSSW_2_2_3

  • Analysis pre-approved in Exotica, ARC review in progress

Reviewing Questions and Answers

Answers to questions risen by Analysis Review Committee:

> Concerning the PAS, there has been significant progress from the version
> available at the time of the pre-approval to the most recent version
> (version 3 of the 8/03/2009) ... but, as it stands, the PAS still 
> contains a
> number of internal CMS jargon [e.g. things like "HLT_Ele15_SW_L1R" on
> page 3 or "PAT::Electron" on page 4 etc.] which should be cleared away
> We would like to remind that the PAS is meant to be a letter-style paper
> at a quality close to refereed papers. Some sentences that have been
> apparently copy/pasted from the Analysis Note version should be 
> re-written
> for the PAS public release.
The updated PAS has those jargons removed.

> The analysis topic is important and could probably be better
> motivated in the introduction of the PAS.
The introduction part has been enriched and more supporting references added.

> The hint from B-factories that you mention as a motivation is fine but
> you must then give a reference [a paper or conference talk with refereed
> proceedings].
Have put back the Kpi Acp publication in Nature.

> You then should motivate, as you have done already in parts,
> the emphasis of this analysis on the mass range  m_b' < m_t +m_W
> and the fact that you will consider the process pp -> b' b' + X with
> b'decaying in  bZ or  cW.
More description text added.

> You say (PAS line 26) that "the study ... is focused on the low-mass
> region where" the "b' goes to cW and bZ" ... and in the next line
> that there "can be up to three high-PT charged leptons in the
> event".  This is wrong at this stage as you can get four leptons
> and two b quarks in the final state with  b'b' -> bZbZ -> bb 4l,
> with a small cross-section x BR but for a very clean signal.
> In section 2.1 you then talk only about bz + cW and monoleptonic,
> dileptonic and trileptonic final states ..  You later on say (line 51)
> that for the generation you assume for convenience a
> BR(b'-> bZ) = 50% and that for the analysis (line 124) you
> take BR(b'-> bZ) = 10%.
Text has re-written.

> It is hard to justify neglecting the four lepton final state in
> this context since the existence of FCNC processses and
> the b' -> bZ decay is a necessity for this analysis ... and the
> branching ratio is arbitrary. You circumvent this somehow
> with the assumption BR(b'-> bZ) = 10% so that BR^2 ~ 0.
A new check has done with fast sim sample on b'->bZ assuming
BR(b'-> bZ) = 10%. The expected yield is about 0.5 event while ~30 for
our study target. Z decays to all leptons and neutrios are included.

> You should start with you basic assumption, namely that
> BR for b' is 100% in bZ or cW. Discuss the possible final
> states containing high PT leptons (up to 4), explain then
> that that this analysis focuses on tri-lepton final states,
> which implies that you assume both sizeable BR(b' -> bZ)
> AND cW(b' -> cW) and Z to ll plus W to el'.
Text has re-written.

> The Mb' [from 200 to 250 GeV/c2] and BR range [5 to 20%]
> you consider for your final PAS plot (Fig.4) from is probably fine.
A quick check for events Mb'>200 for assumed b'=250 gives me ScP
significance ~1.8 for 1/fb data. (originally ~1.1)

> QCD:
> No particular filtering or enhancement (e.g. "e.m. enhanced
> by looking for at least a hard pi0, or explicit request for
> at least one high PT muon) has been used. As a consequence
> only a ridiculously small equivalent cross-section could be
> simulated for QCD ... turning it around: if a single QCD event
> had survived from your very small equivalent integrated
> luminosity,  a huge background would emerge from QCD.
The EM enriched MC has been checked. Details listed in table.

> Could you profit from summer 08 inclusive PT_hat sample ?
> Could you cross check the background estimate e.g. with one of the CSA08
> JetXX samples,
The statistics of them are fairly low. So we didn't check.

The factorizing method currently used gives us general enough and low enough
contribution estimation on QCD events. That's why we didn't check further.
Further that at the time this updating study is done, there's no much related
background samples available.

> ZZ / Zbb:
> are the Z boson in the background simulation allowed to decay
> in all three generation leptons ? Have you considered

> an "enhanced" Zbb sample with a third lepton in on the b-jet ?
> Zbb remains a major background in other 3-lepton and 4 lepton
> analyses ... surprisingly small compared to Z+jets in this analysis.
> It is not clear (in latter section) how ZZ-> lltautau with l =e,mu is
> suppressed.
I can confirm that three gen. are in the Z decays in the official sample.
ZZ -> 2l2tau is checked as all three species of leptons are included.
The W_mt and only one W cut will reject most of the events.
Zbb is included in VQQ which V=W, Z and Q=b, c.

> The way ambiguity are resolved (or not) should be clearly stated
> in this section. The are also small "misprints", like in the caption
> of Fig. 3 where you say that you require N^jets > 2 (!!!) ... while
> you probably require instead N^jets >= 2.
Oh. You are right. Thanks for pointing out.

> You start by collecting isolated leptons (electrons and muons)
> and you calculated missing ET corrected for muons. Fine.
> Now, how to you resolve ambiguities if you are left with three
> or more lepton of the same flavour (say e+ e- e-) with both
> combinations within the allowed Z mass windows ?
If there are two Z candidate or more with mass 60-120 GeV window,
the event is dropped.

> Do you keep track of non-matching signs and flavours ?
> Events where only one "Z mass" can be found in the
> combinatorial but with wrong charge or flavour combinations
> will probably be precious to control (if only for sanity) the
> fake multi-lepton background from QCD. You should explain
> how you can make use of such events.
The wrong charge and flavours of Z decays will be indeed a very
good monitor of the background, in particular if the background rates
would be unexpectedly considerably higher.
I keep enough information for this and had a quick check. 
Just that I do not have any statistics with the present background predictions.

> Why don't you exploit this information from a sliding
> window ??? Is it to be more "model independent"
This is for robustness for a first search, and these sliding window strategy is
deferred to a future study.

> somehow ??? It is not clear to us how you can justify
> this given your assumptions with b' forced to
> decay (100%) in b' -> bZ or cW.
We are having separated studies on b'b'->bZ / cW (100%).

> What are the goals of the "Data Driven Analysis" (Section 6) in the
> context of this search analysis? You present a shopping list but
> do not seem to make use of the gain in knowledge.
The the results of data-driven analysis is actually used to improve systematical
uncertainties. We have re-written the table for this.

> In other terms, what analysis strategy would you adopt if you had
> real data? You would compare data-MC for different selections/control
> samples: which ones? How would you determine normalizations and /or
> shapes from data in particular for your main background
> components, WZ and Z+X ?
The matrix method used in this data-driven analysis is actually adopted from
the WZ production study. By estimating the rate of true/fake leptons passing
the isolation cut using separated data, we can extract the normalization of
WZ and Z+X using just the number of events in the two defined boxes.

> Why the "if" ? To be "conservative" ? Do you believe (or not)
> that you can reduce / better control the systematics with the data
> driven methods in this channel ? If yes (we assume) then we
> should show what you can gain (in addition to sanity checks) in
> terms of significance or rejection limits.
We "are" using the result from data-driven analysis and giving the results.
The table in PAS is updated to avoid confusion.

> Those quoted in the paper (sect.7) are determined from MC.
> Why? Many measurements, of eg cross-sections of SM process, will be
> available while analysis-specific uncertainties on selection
> efficiencies and
> background contributions can be determined from data.
> For cross-section uncertainties (in sect.7.2) why not using the
> (expected to be)
> measured cross-section uncertainties (with 1/fb and 200/pb for which
> you quote
> the final results)?
> For uncertainties on lepton isolation and selections (sect. 7.3) and
> jetMET scale
> and resolution (sect.7.4), how do you expect to evaluate them
> from real data (ie with data-driven methods, see previous comment) ?
Surely we can have better understanding on the SM background cross-sections.
Just that as discussed in ARC meeting, one may have only ~200/pb results
usable when giving 1/fb results. Also to my convenience, the numbers estimated
here can apply on 200/pb.

> How do you deal with flavour dependent systelatics
> (JES for light quarks, gluons and b--jets) ...
> Could'nt you make use of Zbb to control b--jet resolution ?
We are expecting JES working group to give us the estimation. Though their
conclusion in note is very vague.

> for 1 fb-1 and the 1fb-1 results for an exclusion. You might
> prefer to call this section "Results" and discuss first the
> significance for an observation and then the exclusion
> (as is done in current version).
Ya. Ok.

> Now Table 13 is made assuming BR(b' -> bZ) = 10% !!!
> ... which is abitrary ! .... Why not also at BR(b' -> bZ) = 50%
> which brings 2.8 times more signal in the tri-lepton final state ??
Though out the note, BR(b' -> bZ) = 10% is used. As mentioned above, giving
the numbers used in PYTHIA generation is misleading and will be removed. In
fig. 4, BR(b' -> bZ) = 5-20% are given.

We need to see some hint first before we can apply more sophisticated studies.

> Please find below some requests, comments and suggestions concerning
> the paper draft CMS PAS EXO-08-013 (CMS AN 2008/079) following the
> discussions between the ARCs and the author(s) of Friday 27/03.

> number of internal CMS jargon [e.g. things like "HLT_Ele15_SW_L1R" on
> page 3 or "PAT::Electron" on page 4 etc.] which should be cleared away
> We would like to remind that the PAS is meant to be a letter-style
I'm improving this part. Those "PAT::XXX" are actually explained in the text.
To avoid confusion, I can just remove the names.

* Questions from the 2nd ARC meeting:
> The QCD EM sample study is only put in AN, not in PAS.
We think the results are mainly supporting our results with PYTHIA and the
limited number of pages for PAS. We will put a statement in PAS saying the
estimated contribution is negligible.

> The 20% on MET resolution seems too big and double-counting with JES.
The MET is correlated with JES but it's not 100%. MET also depends on the
Muon measurements. Therefore we smeared the MET by 20% and quote the difference
as systematics. Surely this is a bit over-conservative and we can use a less
conservative number.

> We'd like to see the break down numbers of the cuts. What's the most effective
cut to reduce Zbb. Please show the isolation value vs. pt for the W candidate
The isolation requirement does reduce large portion of the Zbb background.
We can see the breakdown numbers in the attached PDF. Here is a short summary:
                       ttbar          VQQ
  any e                2.0%           24.2%
  any mu               2.3%           22.9%

  One Z                0.2%           9.4%
  One W                4.7E-6         2.0E-5
  2+ jets              3.1E-6         2.9E-6

  (the last 3 cuts are accumulated)

These are based on the events after Z, MET and W_mt selection, but #
of jets cut is not applied to gain some statistics. There's no clear
clustering and no clear pt dependence though.

Questions from conveners:

> - what do you do if you have 3 leptons and two of these combine to 2 Z's do
you drop the event or make a choice. It's dropped.

> I buy the argument for W+Z - and it's OK that the results are smaller than the
> expectation. For the Z+jets, I think you shoudl remove the loose leptons from
> the list of jets and then you should be fine estimating the Z & 2 jets background.
Ok. Here you are.
I remove EID and trk/HCAL/ECAL cut on Z candidates daughters and Jet/lepton isolation:

p_T hat        20        30        80
one W        145(2.8E-5)    1405(1.2E-4)    1460(2.7E-4)
one Z        11(2.2E-6)    890(7.2E-5)    7311(1.4E-3)
Jet #>=2    46303(9.1E-3)    482082(4.3E-2)    700278(1.3E-1)

Deriving from above:
one W one Z    <<0.1        0.1        2.0
Z+Jets        0.1        34.7        980.4
W+Jets        1.3        57.8        189.0

MC results:
one W one Z    0        0        0
Z+Jets        2        53        1160
W+Jets        3        224        734

Now the Z+jets estimation servers better. But W+jets still have a ~3x factor.
The factorization still holds as the cross check with EM enriched sample gives me
negligible contribution. Details on the cross check are added into the latest
version of AN.

>> table 1 & 4.   I realized only now that you have 10K events in your signal
>> generation and the cross sections are 35-113 pb
10K events correspond to 0.89-2.84 pb. The sub-BF of 50% for both b'->bZ
and cW are embedded by PYTHIA. The 1/fb equiv. yields are re-calculated with
BF(b'->bZ)=10%, BF(b'-cW)=90%.

113 pb is b' inclusive cross-section.
113 * 0.5 (b'->bZ) * 0.5 (b'->cW) * (0.033549+0.033549+0.033473) (Z->ll)
= 2.84 pb (given by PYTHIA) 

>> Selection: this was asked in the ARC meeting: do we take events with exactly
>> 3 and only 3 leptons? We should say that
There can be more than 3 leptons but only one W and one Z candidates allowed.
> in case you have 4 leptons and only two combiner to a Z, what do you do?
> How do you know the two others could not be from a W?
> in other words, under which condtion do you keep events with 4 leptons?
If only one lepton form a W within M_t(W) window, the event is kept.

>> lines 158-162:  how often do you get the combination right by taking these
>> jet/lepton pairs?
76% if bZ or cW is correct.

>> line 173:  'small statistics'... you mean in the data, so that a data driven
>> technique can not be performed?
small statistics on side band of Z. 

>> line 264-266:  Not sure what is meant by this sentence. Explain better please
sqrt(N) used as uncertainty.

>> It's arbitrarily given based on the TDR/PAS estimation on the x-s uncertainties.
>> The x-s measurements for 1/fb should be better than this 10%.
> based on what? Be careful what you use here, some of these cross section
> uncertainties include also the scale
> uncertainty, not just the PDF. It looks we cannot defend this number too strongly
> Better be careful with the words here...
OK. Taking TOP-08-002 PAS for example, di-leptonic ttbar production measurement
is expected to have ~8% (stat.) error for 100/pb data. (JES is about 3% from
AN-2008/016) Please note that the central mass energy is 14 TeV in this study.
As the cross-section is about 1/2 for 10 TeV case and we probably will quote 200/pb
number when doing 1/fb study, I think 10% here is a safe number. 


  • Latest PAS on CADI system. (need iCMS login)

  • CMS Analysis Note AN 2008/079 AN 2008/079 (CMS internal only)

  • Final version of EPS09 poster on b' searches.

Reports in the Past Meetings

-- YuanChao - 16 Apr 2009

  • An MC event of b'b'->cWbZ->meejj:
Topic attachments
I Attachment History Action Size Date Who Comment
Compressed Zip archivetgz MyG4Kit.tgz r1 manage 14.4 K 2009-01-09 - 02:04 YuanChao MyG4Ana in PAT
PNGpng bprime_bZcW_200GeV_10TeV_3D_3s.png r1 manage 81.9 K 2009-07-31 - 18:04 YuanChao An MC event of b'b'->cWbZ->meejj
PDFpdf poster_eps09_final.pdf r1 manage 1736.1 K 2009-07-18 - 17:08 YuanChao Final version of EPS09 poster on b' searches.
Edit | Attach | Watch | Print version | History: r12 < r11 < r10 < r9 < r8 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r12 - 2009-07-31 - YuanChao
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Main All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2021 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback