NLO s-channel studies

1. Basic Info

Used generator: ONETOP CTQ 6.6
Histograms can be found at

2. NLO corrections

NPART directory name name description
20 tree Born level LO
21 svci soft and virtual corrections to initial state soft + virtual + crossing contribution
22 svcf soft and virtual corrections to final state soft + virtual correction
23 svcd soft and virtual corrections to top quark decay soft + virtual correction
24 inita real initial state corrections gluon radiated from initial quarks
25 final real final state corrections gluon radiated from top or bbar
26 initb real initial state corrections u quark comes from left and dbar from gluon from right
27 initc real initial state corrections u quark comes from right and dbar from gluon from left
28 initd real initial state corrections u quark comes from gluon from right and dbar from left
29 inite real initial state corrections u quark comes from gluon from left and dbar from right
30 decay real decay corrections gluon radiated from top decay process

3. How to...

...produce events

  1. in
    - choose how many events you want to produce (recommendation: not more than 500.000 a time because of crash risk)
    - change random number seed
    - choose top/ antitop
    - choose s-/t-channel (recommendation: s-channel, as it is much cooler)
  2. to run, type: ./
  3. when it is done and you want root files instead of ntuples
    type: h2root inia.ntuple
    (if doesn't work: either wrong root version, or paths incorrect (look, where root, its libraries and cernlib is on the system))

...make histograms

  1. in
    - choose your output file
    - several options for running: can be found in driver and nloanalysis
    - -lofile $NTUPLEDIR/stre.root makes it possible to run over one file only
  2. to run, type:

...make pretty graphs

  1. create output directory
  2. in macros/s-channel/makeObjectPlots.C:
    - change name of input file, output directory
    - choose gif/eps option
    - compare numbers of folders in your root file with those in macro, change, if necessary
  3. to run, type: root, .L makeObjectPlots.C, makeAllPlots() or whichever plots you want to make

...get cross section measurements etc.

  1. create output directory
  2. to run macros/s-channel/measureEfficiencies.C type analogous to above

4. How to...on the tier

...produce events

  1. my directory on the tier: /home/sheim/100Mio/stnlo_ctq6.6_top_tier
  2. choose options in (see before)
  3. in
    • random number seed range
    • directory for logfiles
  4. in
    • directory to store root-tuples (OUTPUTDIR)
  5. type: (wants new root library -opt because of h2root)

...make histograms

  1. my directory on the tier: /home/sheim/100Mio/NLOAnalysis
  2. in
    • choose random number seed
    • directory for logfiles
    • directory in which you store your root-tuples
    • directory for output root files
  3. in
    • still NTUPLEDIR
    • runme options (see above)
  4. type: (wants old root library, I don't understand that)

...add histograms

  1. in /msu/data/dzero there is
  2. change FILEDIR and the filenames
  3. type "" (wants new library)

5. a) Changes 2008

Major changes

  • new cut class: Cut_Basic_LHC.cpp with the proposed cuts in the ATLAS CSC note ("Prospect for single top measurement in ATLAS")
  • loose cuts (preselection cuts):
    • deltaR > 0.4 between jets/leptopn (isolation cut)
    • abs(eta) (lepton) < 2.5
    • exactly one lepton pT>30 GeV
    • 2 or 3 jets, >30 GeV
    • eta for jets <5
    • at least one b tagged jet
    • MET > 20 GeV
  • tight cuts:
    • 2 jets b-tagged
    • 50 GeV pt cut for all jets
    • eta < 2.5 for jets
    • lepton cuts unchanged

  • ATLAS cuts different
  • 1. loose
    • 2 jets only
    • energy in cone around lepton deltaR=0.2 be less than 6 GeV
    • veto cut on another lepton with pT >10 GeV
    • opening angle between jets: 0.5 <deltaR(b,bbar)<4.0
    • total jet transverse momentum HT: 80 <= HT(jets) <= 220 geV
    • lepton/neutrino transverse momentum: 60 <= MET + pT (lep) <130 GeV
  • 2. tight
    • jet veto on any third jet > 15 GeV

  • CMS cuts
  • 2. tight
    • 1 lepton pT>19 GeV
    • eta (el) < 2.4
    • eta (muon) < 2.1
    • eta(jet) < 2.5
    • exactly two jets pT> 50 GeV (!)
    • MET> 30 GeV
    • veto cut on further lepton with pt > 10 GeV
    • veto cut on further jet with pT> 20 GeV
    • further cuts after reconstruction; total hardness HT, MT, M(t), sigmaT

  • option for event mirroring: GetNextMirrorEvent() and readMirrorEvent() in Particles.cpp and included in nloanalysis.cpp in the loop.
    It takes every event a second time with -px and -pz, to enforce LHC symmetry.
    Careful: You will have to divide the cross section by two!
    for using it, just add -mirror in

  • ONETOP update: two more initial corrections for the LHC
    now: inita, initb, initc, initd, inite
    furthermore change of numbering of corrections (see table above)
    => change in Particles.hpp, myTH1F, myTH2F

  • -applybtag in seriously screws up weights

  • added new Histogramming class: Histogramming_Boost to look at eta and polarization distributions
    for two cases:
    pz(W intermediate) > 0 and pz(W intermediate) < 0

  • added new function to makeObjectPlots.C: makeBoostPlots()

  • added new function to measureEfficiencies.C: makePolMeasurements()

Minor stuff

  • plots for eta now only from 0 to 5 range as symmetric for LHC
    (change only in root macros, the original histograms are still from -5 to 5)

5. b) Changes 2009

  • I changed my cuts to 30 GeV for every jet as some distributions have residue between 15 and 30 GeV otherwise, this comes from the misinterpretation of a gluon jet as quark jet.

6. Interesting physics found in the plots

1. Polarization

For the top quark:
For pz(W intermediate = Wint)>2500 GeV and using pz(incoming u quark >0) we get a very high polarization in the parton level, as a high pz Wint follows the direction of the incoming u quark most of the time, that is by taking pz(u quark)>0, we are guessing the right direction of the u quark, which is necessary for the Optimal basis.
The eta cut on the lepton takes away events for those high Wint energies and deforms the plot.
For the antitop:
If we still use pz(incoming u(bar) quark)>0, taking pz(Wint) < -2500 gives a high polarization, though not as high as for the top.
The eta cut cuts away events here, too and deforms the plot, though in a different way than for the top.

2. Init corrections to top quark rapidity

The different initial corrections for the top quark rapidity peak at different values, as they describe different physics

3. Width/shape of beta and bbareta for top/antitop

The more a particle is boosted in z direction the wider is its eta distribution.
Below are estimates of the widths of the tree level eta distributions

  top antitop
b/bbar from top decay 1.6 0
bbar/b 0.1 1.2

We actually get two peaks for the eta distributions: One for u coming from left and one for u coming from right. The top is boosted stronger than the antitop as p(u) is approx. 2*p(d) due to the parton distribution function. That is why the widths are generally wider for the top than the antitop. The effect is opposite for top and antitop which can be explained by looking at the spin correlation plots. Here the bbar that is produced with the top is "antiboosted" while the b produced with the antitop is boosted.

4. Cuts

My first set of plots is produced with the ATLAS Cuts from the ATLAS CSC note ("Prospect for single top cross -section measurement in ATLAS"), can also be found on the WhiteBoard . The Pt cuts on the jets are the following: 30 GeV on two jets and if there is an additional jet it has to have at least 15 GeV. One b-tagged jet with 30 GeV required, if there is a second b-tagged jet it is not required that it has 30 GeV, but it has to have at least 15 GeV (minimum required energies for all jets). If you look at the b-jet pt-distributions after cuts you see that they are zero up to 15 GeV, then there is a small contribution from the cases where the b-tagged jet is the third jet and the gluon jet is actually over the 30 GeV limit (that doesn't happen too often, so this contribution is very small). At pt>30 GeV the distribution should look the same as before cuts.

The second set of plots will have a 30 GeV cut on every jet. It will be interesting to see how the number of jets for 2-jet and 3-jets changes with the higher cut on the third jet. If we require all jets to have minimum pt of 30 GeV, the small contribution between 15 and 30 GeV in the jet-pt plots vanishes.

But even with the stricter jet cuts, the parton level distributions after cuts (which are not really experimentally relevant, but interesting to understand the event generation), do have a contribution between 0 and 30 GeV, as the jet clustering doesn't always capture the emitted gluon and combines it with the b-jet, so that the real parton in fact has a smaller pt than the one that is detected as a jet. The contribution between 0 and 30 GeV consists only of real NLO corrections, as the virtual corrections are always fully captured by the jet clustering and the tree level doesn't have a third jet. If we look at the b pt distribution of the alpha-s corrections after cuts, the sdecay contribution looks very bumpy. This can be explained by looking at the real and virtual corrections separately: the virtual corrections are negative, and cut at 30 GeV, the real corrections are always positive and only reduced under 30 GeV, but now, with the negative correction missing, getting pretty big and then falling once the virtual correction kicks in.

It is also interesting to look at the top mass reconstruction. Here, the decay correction looks very bumpy. Both real and virtual corrections shift the top mass up a bit, but again, we do not capture all the emitted gluons and thus reconstruct some of the b-quarks with too little energy, so the real correction is not shifted up as much as the virtual correction, so that it dominates at lower masses, whereas the virtual correction kicks in at higher masses, making the distribution going first up and then down.

The top pt after cuts is another "illegal" plot to look at as it is basically parton level aftercut, which we will never measure in real life. I was struck by the weird shape and the dominance of the initial corrections though. As we have a cut of 30 GeV on the bbar jet now, the top can only have a pt smaller than 30 GeV if p.ex. the bbar emits a gluon, which is not included in the jet, and they are back to back, so that the bbar appears to have a greater momentum than it actually has (this happens very seldomly, so that this contribution is very small. The other way of getting a top with a pt less than 30 GeV is if one of the quark is produced by a gluon (and an (anti) quark is emitted as third jet), so that the produced quark in fact has a pt that is nonzero (which is allowed as it has the third jet as counter part). This process boosts the top and the bbar, so it can happen that the bbar is detected with a momentum of 30 GeV, because of the boost, but the top (goes into the other direction and therefore) has a smaller pt.

7. Cross section for top mass 175 GeV and 5 mio produced events:

- used makeXSTable() in measureEfficiencies.C
- input: "schanCtq6.6/lhc_top_loose_04_5mio.root"
and "schanCtq6.6/lhc_antitop_loose_04_5mio.root"
=> mirrored => I divided all given numbers by 2
=> converted to pb => divided all numbers by 1000
=> 10 input files (with 500000 events each) => divided all numbers by 10
=> we are only looking at the electron branch => multiply by 9 to get full cross section

The table below is actually the data from the log file summed over the 10 input files,
multiplied by 9 (branching ratio), because I wanted more digits after the decimal point than
makeXS gave me.

  Top cross section [pb] Antitop cross section [pb]
tree 4.2984 2.6172
NLO 5.9166 3.6261

8. Dependence of Cross Section on Top Mass & Scale

1. Top Mass Dependence

Now I will make a table for my 500.000 event - top mass = 175 measurement (which I used to determine
the dependence of the cross section on the top mass) to calculate the difference between the two measurements.
I will shift the mass - dependence graphs by this difference as the 5 mio measurement is more precise.
This table is the output of the log file (schan_51.check), multiplied by 9 (because of the branching ratio)

  Top cross section [pb] Antitop cross section [pb]
tree 4.2948 2.6163
NLO 5.9256 3.6306

So we have the following offsets:

  offset Top cross section [pb] offset Antitop cross section [pb]
tree 0.0036 0.0009
NLO -0.009 -0.0045

  1. I produced 500000 events with an input top mass from 168 - 182 GeV (my random number seed was 123):
    script: heim/work/hx7/sarah/onetop/stnlo_ctq6.6_top_masses/
    => loops over top masses, substitutes itop1=14 and itop2=14 with numbers from 0 (=168) to 28 (=182)
    => stores the summary part (last part, cross sections for NLO and tree) of every produced schan_51.check into a textfile
  2. I used heim/work/hx7/sarah/onetop/stnlo_ctq6.6_top_masses/ to plot top mass dependence on cross section
  3. also calculates the ratio of top/antitop cross section for NLO and tree level and plots it vs the top mass
  4. If you use it multiplies the cross sections by 9, according to the branching ratio and shifts the curves according to the offset (see above)
    It will also make a fit and plot only the fitted lines.

The ratio for the cross section top/antitop is around 1.6 and seems to be rising for higher top masses.
A possible explanation could be that the phase space for the top production is bigger as
the incoming u quark (top production) has a higher energy than the incoming down quark (antitop production).
With higher top mass this effect becomes more important.
It also has to be understood why the ratio is not 2:1=2 (2 u quarks and 1 d quark) or bigger, but significantly smaller.

With sheim/work/hx7/sarah/onetop/onetop/lhc/analysis/macros/s-channel/makeTopAntitopPlots.C
I plotted the ratio of the top/antitop cross section vs the pz of the intermediate W.
As for higher energies our calculations don't seem to work for the NLO corrections,
I plotted only until ca 2300 GeV and stuffed the rest (up to ca 7000, the produced histogram does only have this range)
into the last bin.

2. Scale dependence

I produced 500000 events with a top mass of 175 GeV and a scaling factor between 0.5 and 2 (in steps of 0.1). For this I used my script: heim/work/hx7/sarah/onetop/stnlo_ctq6.6_top_masses/

Interestingly my cross sections seem to rise with a rising scaling factor, whereas for Tevatron setting the cross sections fall with rising scaling factor.

100. Outline of the paper

Let's fetz...Don't forget:
  • how to cite atlas note?
  • page numbers
  • citations (to compare cuts, pictures, cross sections and top mass)
  • Fig.
  • W boson
  • s-channel single top quark production
  • take titles out of used graphs => in CombineHistos, SetTitle("");
  • some legends have to be shifted: (just go to makeObjectPlots... and change legend corner, can be found in CombineHistos.C)
  • legend corners: 10 upper left, 20 upper right, 11 lower left, 21 lower right, 31 center, 32 upper right-hand
    • HT sum
    • (optimal basis boosted 4000)
    • histo_topptsums_parton
  • weird plots
    • histo_bEff_compare_aftercuts => doubly printed?
    • (histo_bJetbbarJetEta_aftercuts => doubly printed?)
    • histo_bgTopPtsums_aftercuts => looks illegal, but is "from bg"
    • topmass_compare => why still in fb?

  • top - antitop
  • only parton level plots or reco after cuts (never parton after cuts!)
  • intermediate W?
  • "modified" narrow width for top quark
  • fixed value for top quark mass used
  • polarization helicity basis tbbar(j) => t-channel appendix paper...???
  • discuss different init contributions? (at least virtual and real?)



  • tevatron results so far (discovery)
  • single top production modes at LHC
  • properties single top => spin, Vtb
    • new physics (top mass closest to ew symmetry breaking scale, affects all three channels differently => independent tests, discrimination between models):
    • sensitivity to additional bosons (see Tait, CP's paper)
    • charged Higgs boson (Lucotte, Chevalier 2008 & yue zhou)
    • modified Wtb and Wqq' coupling (new yuan, aguilar saavedra)
    • four fermion interactions (new yuan)
    • FCNC (flavour changing neutral current) (production or decay
  • new physics background => higgs
    • Wt background for H-> WW and for charged Higgs
    • s-channel?
    • t-channel?
  • signature?
  • motivation nlo
  • generator description => cite old paper
  • previous calculations, Sullivan...with jets instead of partons???
  • top - antitop asymmetry
  • outline paper


  • we need exact theoretical prediction for cross section for: Vtb, new physics
  • we present numerical results...
  • cteq6.6
  • generator properties: values of constants (schan_51.check)
  • electron only, applies to muon, too
  • factor 9 / 9.2 ?
  • comment on top quark mass = 175?
  • we adopt phase space slicing method => theoretical cutoff dependence (cite old paper), choose 5 GeV^2
  • absolute cross section

A Inclusive cross section
  • total cross section: add up contributions from born level, init,,,,
  • explicit diagrams in theo-paper (cao)
  • finite widths of t, W
  • cross section table with INIT, FINAL...contributions (measure efficiencies)
  • discuss effects of corrections, sdec small -> cite old paper
  • discuss top-antitop difference

B Top quark mass and Renormalizing/Factorization scale
  • uncertanties: top quark mass, renormalization/factorization scales * top mass => want to know how top mass accuracy affects cross section prediction
  • explain renormalization, factorization scale
  • we work at fixed order => scale dependence
  • top mass dependence => cross section changes roughly 10% for +-5 GeV
  • top - antitop (ratios) =>
  • scale dependence => NLO reduces scale dependence (makes sense as scale dependence comes from higher orders)
  • top - antitop (ratios)
  • uncertainty due to NLO predictions (charlie)


  • final state objects
  • only leptonic decay channel
  • signature: one charged lepton, missing transverse energy and two or three jets
  • parton level => only approx of kinematic acceptance (no detector effects as b-tagging efficiency, jet energy resolution)
  • jet algorithm (jet must be infra-red safe) => cite old paper
  • list used cuts, cite prospects single top lhc note (contains both atlas and cms) => do it like in old paper
  • table from measureEfficiencies
  • interpret table
  • low acceptance => due to pt cuts (reference to pt...) reduce to 20 Gev would help a lot???
  • fig 4 => slice of lego plot jet pT threshold => measureEfficiencies...
  • interpret
  • for the following discussion: loose cut set


    • kinematic properties, repeat signature
    • jet identification
  1. Final State Object Distributions
    • final state distributions after event reco and after applying loose set
  • 1.
    • lepton, missing ET: fig 5 electron, neutrino ? or too much repetition? I need pt distribution anyways...for cut explanation
    • top, anti, interpret
  • 2.
    • b, bbar jets (from partons): fig 7 b, bbar
    • compare b, bbar jets eta distribution, explain differences, look at top, antitop, spin correlation picture? also INIT...shifts
  • 3.
    • Ht => INIT, FNAL, DEC contribution shifts

B. Event Reconstruction

    • check: ANTI important/different?
    • correlations between particles => reconstruct object completely, including intermediate particles
    • W boson reco from lepton and MET (no pz of neutrino, but require W mass) => two solutions, take smaller (see old paper, cite)
    • top quark reco => b jets problem => best jet => with third jet: bg jet including gluon (cite old paper, repeat some of explanation)
    • best jet + b-tagging or Meb or eta separation?
    • percentage best jet real b jet => Integral over both whole area and bg area => divide = 80 %
    • fig 9: best jet is b jet...
    • fig 10: top peak wth best jet... top quark width comment ...two competing effects
    • fig 11: top pt, eta at parton level and after cuts

C. Kinematic and Spin Correlations / Top quark polarization
  • 1. bbar correlations
    • fig 12: transverse momentum, diff between best, non -best
  • 2. Top quark polarization measurements
    • explain correlations for top polarization
    • explain bases, show histos
    • helicity
    • explain tbbarj, tbbar frames for helicity basis (see t-channel paper appendix) (=> in makeObjectPlotsblblabla, copy numbers into table)
    • show histos, before after cuts, anti important? 3 jet events
    • optimal
    • problematic optimal basis
    • => (basically beamline basis)
    • => polarization for >2500 GeV
    • degree of polarization
    • table of polarizations (add >2500 to it)
    • => pol depending on Wz => prescription?

D. Distributions for three-jet event
  • large number (see Fig...)
  • 1. kinematic distributions of the extra jet
  • production stage vs. decay stage gluon radiation
  • fig 16: jet 3 distributions (pt, eta)
  • How do I figure out the percentage for the third jet to be from the gluon?
  • fig 17: delta R, angular
  • prescription for dealing with extra jet?

V. SINGLE TOP PRODUCTION AS BACKGROUND TO SM HIGGS SEARCHES ??? => maybe rather single top and new physics

  • fig 19, 20: invariant mass of bbar, angular distance bbar


  • Summarize important results of chapters
  • Cross Section, Acceptance, Kinematic distributions (bbar, reconstruction and top, top polarization, three jet events)
  • a lot of three-jet events
  • largest contribution to cross section: INIT corrections
  • 6% range dependence on top mass
  • signature
  • acceptance 30 %
  • reconstruction best-jet algorithm
  • different eta distributions top antitop
  • correlations

-- SarahHeim - 30 Oct 2008

Topic revision: r53 - 16 Oct 2009, TomRockwell
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback