Searches for double partons

Lee Pondrom

University of Wisconsin

July 23, 2012

Rick Field’s definition of the ‘underlying event’

Single vertex dijet or Z+jet event

Charged tracks in the transverse region

• R Field studied the underlying event and tuned Pythia to match the charged track activity observed in the transverse region.

• (PRD 82, 034001,(2010).• Track pT > .5 GeV, track |η|<1• Parameters track multiplicity, scalar sum

track pT.• He looked at Z D-Y data to define the

transverse region.

RField’s Z transverse data

X

X X X

X =jet20,50,70, and 100 data

Use jets to define the axis and check Field’s Z results

Transverse track activity depends slowly on pTZ or jet1 ET

• About 90% of the ∑pT and track number plots are energy independent.

• Underlying event activity is the same for pTZ ≈ jet1ET

• If double parton hard scattering exists, a good place to look for it would be in the transverse region.

• Try scalar ∑transverse track pT>15 GeV as a ‘trigger’

To see if it works, try jet events with two vertices

• Use jet100 jet50, and jet20 data• Require jets one and two to be on the first

vertex. Exactly two vertices per event.• Extra jets three and four can be anywhere• Separate the two vertices by at least 10

cm.• Require vtx2 to have at least 3 charged

tracks, with pT>.5 GeV and |η|<1.• Second vtx σ ≈ 6 mb.

Two vertex event with 2 jets on primary vtx and 2 jets on 2nd vtx

Second vertex is ‘minbias’

• CDF minbias defined by the CLC σ ≈ 36 mb

• Track requirements here are different, in particular |η|<1, so σ is smaller, although 6 mb is a soft number – it depends on instantaneous luminosity taken to be 2E32.

• For the jet stntuples analyzed, the avge probability of two vertices is 30%.

Transverse tracks on primary and secondary vertices

Transverse track activity

• The first vertex transverse tracks are defined with respect to the azimuth φ of the highest ET

jet: /3<Δφ(jet-track)<2/3 • second vertex transverse tracks are defined in

the same way with respect to the same jet – highest ET jet on vtx1, track on vtx2.

• Jet activity ‘triggered’ by ∑transtrackpT>15 GeV is similar in jet20 and on 2nd vtx.

• ~90% of all ‘triggers’ have a third jet ET>5 GeV

Fraction of ∑transtrackpT>15GeV vs ET jet1

∑transtrackpT>15 GeV

• Note that the fraction increases from .001 to .015 going from minbias (plotted as

ET=5GeV) to jet20.

Using σ ≈ 6 mb for the 2nd vertex, the cross section for the ∑pT>15 GeV ‘trigger’ is

σ ≈ 6 b. about 5 times larger than the effective cross section for jet20.

Jet activity ‘triggered’ by ∑pT>15 GeV – 2nd vtx and jet20

Jet activity ‘triggered’ by ∑pT>15 GeV 2nd vtx compared to jet20

• Jet4 is similar 2nd vtx compared to jet20

• Δφ12 is the azimuthal angular correlation for the two leading jets. For the 2nd vtx the ‘trigger’ has no effect on Δφ12, which is on the other vertex. But with transverse tracks on the same vertex, Δφ12 is totally wiped out for jet20. Effect for jet100 is less traumatic.

Δφ dijet angular correlations-main jet on vtx 1, jets 3&4 on vtx 2

Δφ jet-jet angular correlations after transtrack ∑pT>15 GeV

• Δφ23 is shifted into the transverse region. The third jet appears near 90 degrees to the second one. Jet 2 and jet 3 are on different vertices.

• Δφ14 shows a similar shift. Relative to the primary axis, jets 3 and 4, on the second vertex, are in the ‘transverse’ region.

Δφ for jets 3&4 on 2nd vtx, selected by ∑transtrackpT>15GeV

Δφ34 with transtrack∑pT>15 GeV

• For the 2nd vtx Δφ34 has a distinct peak near . (Sum 2.4<φ<)/all events=3.5E-4

• Jets 3&4 are not required to be on vtx2. Events with Δφ<1 are probably on vtx1

• Δφ12 for jet20 data without the ∑pT>15 GeV requirement is shown for comparison of the jet-jet angular correlation. Jet20 is sharper – the average jet ET’s are higher.

•

Jet ET on 2nd vertex with ∑trackpT>15 GeV ‘trigger’

Jet20 Data Stntuple gjt1bk & gjt1bj 3E6 events

• Require only one vertex• Require at least two jets with |η|<1. ET1>20 GeV.

Other jet ET>5 GeV• Apply level 5 jet energy corrections• Events• Jet1&2 Jet3 Jet4 Lum(live)• 110203 61769 21174 151694/nb• 56% 19%• Prescaled σ ≈ 0.7 nb; unprescaled σ≈1.2b

Jet20 data Jet ET

Jet20 data Δφ distributions

Following Rick Field, define the transverse region relative to jet1φ

• Look at charged tracks with |η|<1, pT>.5 GeV, and with /3<Δφ<2/3, where Δφ is the azimuthal angle between the track and the highest ET (trigger) jet.

• These tracks are sensitive to the underlying event, and hence at least in part depend on multiparton interactions.

Jet20 data properties of the transverse tracks

Jet20 transverse track pT cut

• Based on the idea that the transverse region has some sensitivity to what is going on in the event other than the two primary jets, we make a cut on the scalar sum of track pT>15 GeV.

• This cut leaves 1611events – 1.5% of all dijets. The fraction increases with jet energy.

• The cut moves jet3 into the Δφ region of the tracks.

Effect of the ∑transtrackpT>15GeV on the jet φ distributions

Effect of the ∑transtrack pT>15 GeV cut on jet φ

• Of the 1611events, 1495, or 93%, have jet3ET >5 GeV, and these jets are clustered around /2 relative to the trigger jet.

• 15 GeV is too high relative to the main jet activity, so the correlation Δφ12 is strongly perturbed.

Δφ34 and the high pT transverse tracks

• The idea is that jets 3 and 4 could be result of independent scattering of two other partons.

• If that is true, a good place to look is in the Δφ region of the underlying event.

• The ∑transtrack pT>15 GeV serves as a ‘trigger’.

• Δφ34 should peak near .

Δφ34 and jet3ET before and after track∑pT>15 GeV cut

Enhancement near Δφ=?

• Normalizing the two distributions to Δφ<1.5 gives a difference 2.4<Δφ<3.2 of

• -8±17 events.

• Transverse jet energies for jets3 and 4 are increased by the track pT cut

Look at jet100 data 1E6 events gjt4bk & Pythia bt0stb

• Same requirements: only one good vertex, trigger jet ET>100 GeV, level5 jetECorr

• Yields for 1E6 events• Jet1&2 jet3 jet4 pT>15GeV Lum• 170710 101231 35034 10247 126342/nb• Jet3 and jet4 fractions same as jet20• pT>15 GeV fraction 4x larger than jet20• σ ≈ 1.3 nb no prescale

Jet100 data and Pythia ET

Jet100 & Pythia transtrack pT and Δφ12

Jet100 data effect of the transverse tracks on Δφ12 and Δφ13

Jet100 effect of transverse tracks on jetET3 and Δφ34

Jet100 & Pythia effect of ∑pT>15 GeV cut on transverse tracks

Jet100 effect of transverse tracks

Jet100 similar to jet20. Pythia & data agree.

Perturbation of Δφ12 considerably less than for jet20.

Δφ13 shifts so that jet3 is /2 away from jet1

Jet3 ET shifts to larger values

Compare jet100 and Pythia Δφ34 before and after ∑pT>15GeV cut

Jet100 data and Pythia Δφ34

• The data and Pythia agree qualitatively in the shapes of the Δφ34 angular distributions before and after the ∑pT>15 GeV cut on the scalar sum of transverse tracks.

• Near Δφ34≈ Pythia has a smaller excess than the data.

Now compare Δφ34 before and after ∑pT>15 GeV cut

Excess near Δφ34≈

• Normalize the plots to .5<Δφ34<1.5

• Subtract (after cut)-(before cut) 2.4<Δφ34<3.2.

• Jet100 data difference = 295±50 events

• Pythia difference = 54±30 events

Does this excess mean anything?

• There are 170710 jet100 good dijet events• So the excess 2.4<Δφ34< is 0.0017±0.0003.• Pythia excess is smaller: 0.0007±0.0004. • If the number of MPI’s per hard scatter is 5,

which comes from Field’s analysis of Drell Yan (PRD 82,034001(2010)), the probability of a second hard scatter is 0.00034±0.00005, or about 3.5E-4 for the jet100 data.

Compare to the 2nd vertex Δφ34

• The 2.4<Δφ< fraction for 2nd vtx =3.5E-4, which agrees with the peak fraction observed in the jet100 data (but not in Pythia).

• The dip in Δφ34 near 1.7 radians for the ‘trigger’ events in jet100 is of unknown origin, but is reproduced by Pythia.

• Subtract (after ∑pT –before ∑pT) and compare to 2nd vertex peak.

Compare jet100 excess with 2nd vertex

Look at Jet50 data and Pythia

• Jet20 data are too low ET relative to the ∑transtrackpT>15 GeV cut.

• Jet50 data are higher, and the Pythia file bt0srb has 4.5E6 events, while bt0stb (jet100) has only 1E6 events, so we have better statistics for the monte carlo.

• Same procedure as jet20 and jet100

Jet transverse energies after level5 jet energy corrections

Jet-jet Δφ correlations jet50 and Pythia

Jet50 data and Pythia

Jet50 data and Pythia after ∑transtrackpT>15 GeV cut

Jet50 data and Pythia after ∑transtrackpT>15 GeV cut

Δφ34 for jet50 data and Pythia w/wo ∑transtrackpT>15 GeV

Jet50 data and Pythia agree well

• The excess near Δφ34 = when the transverse track ‘trigger’ is applied now appears in both the data and Pythia.

• The excess normalized to the number of events is:

• jet50 jet100

• Data 0.0016±0.0002 0.0017±0.0003

• Pythia 0.0009±0.0002 0.0007±0.0004

So what?

• ∑pT>15 GeV trigger gives back to back low ET jets on the second vertex

• ∑pT>15 GeV trigger also gives an enhancement in Δφ34≈ in jet100 and jet50 data. Pythia jet50 shows about ½ the effect. Pythia jet100 lacks statistics.

• The enhancement is consistent with the 2nd vertex, if there are 5 2nd vertices inside a jet100 or jet50 event.

Effective cross section

• Some folks like to quote an effective cross section, defined by the ratio σDP=σAσB/σeff.

• Divide by σA: ε = σB/σeff, or σeff=σB/ε, where σB = 6 b = cross section for the second vertex, and ε = 0.0016±0.0002.

• ε is energy independent-same for jet50 and jet100, which is consistent with DPI.

• These numbers give σeff=3.8±0.5 mb,Δφ=2/3 • Scaling in φ, σeff=11.4±1.5 mb, in agreement

with D0 (PRD 81 052012 (2010)).

Extend the study to Drell-Yan pairs

• Use high pT muon trigger stntuples

• 5E8 events total available.

• 1E6 events takes about 4 hours to analyze

• So 5E8 would take 3 months of steady computing, unless I can speed things up.

• 5E7, or 10%, analyzed so far

± pair yields from high pT muon trigger Stntuples

• Require two muons opposite charge |η|<1.

• Eliminate events with cosmic rays

• Require at least one CMU*CMP muon

• Require at least one jet ET>5 GeV

• 48894 events 30GeV<m<130GeV

• 40567 events 80GeV<m<100 GeV

• 28811 events Z pair pT>10 GeV

D-Y mass and pT plots

D-Y mass and pT

• Gauss fit to peak at 90.8 GeV, width 3.8 GeV

• Pair pT compared to recoil jet ET with level5 jet energy corrections is much harder than jet20 ET spectrum

Compare to Pythia D-Y

Δφ and ΔET for track pair and recoil jets

Data compared to Pythia D-Y

Δφ and ΔET for track pair and recoil jets

• Pair pT>10GeV. Central Δφ peak consistent with jet-jet Δφ12 for jet20 data.

• Δφ=|(recoil jetsφ) – tracksφ| - is asymmetric, with a tail towards negative values, ie jets and tracks in the same quadrant.

• ΔET is nearly symmetric, with a shift such that jet ET is about 4 GeV low relative to the tracks, even with level5 jetEcorr.

• Pythia agrees with both plots.

Scalar sum pT for transverse tracks Z data, jet20data, and Pythia

Transverse track scalar ∑pT relative to the pair axis

• The trans track ∑pT distributions for D-Y pairs and for QCD jet20 dijets look very similar. pTZ > 10 GeV required to define ‘transverse track’.

• Again Pythia agrees well with the data

Effect of pTZ on recoil jet ET

Effect of pTZ on recoil dijet invariant mass

Effect of pTZ on jet ET

• There are plenty of dijets with ET>5 GeV in events with pTZ<10 GeV

• Requiring pTZ>10 GeV suppresses the low ET jets, particularly for jet 1.

• Dijet invariant mass for recoil jets 1&2 depends on pTZ ; jets 2&3 not so much.

• pTZ<10 <m12>=15.6;<m23>=12.7 (GeV)

• pTZ>10 <m12>=22.3;<m23>=14.7 (GeV)

Effect of pTZ on recoil jets Δφ distributions

Closer look at recoil jet Δφ12

Closer look at Δφ12 near

• Normalizing the blue curve to the black from 0<Δφ12<2 radians gives an excess from 2.5<Δφ12< of 884 events-3% of all Z’s with pTZ>10 GeV. Too large to be DP.

• The difference in the two histograms is compared to jet20 data in the second plot.

• It is possible for two jets to be created in D-Y events where the pT of the pair is less than 10 GeV. For very low pair pT the two jets have to balance each other.

scalar∑transtrack pT>15 GeV

• For Z-> data with pTZ>10 GeV define the transverse tracks after all muons have been removed as those tracks with pT>.5 GeV and |η|<1 at /3<Δφ<2/3 relative to pTZ..

• Then require the scalar ∑pT of these tracks to be >15GeV, as a ‘trigger’ analogous to the jet20 and jet100 analyses.

Effect of ∑transtrackpT>15 GeV on recoil jets ET

Effect of ∑transtrackpT>15 GeV

Effect of ∑transtrackpT>15 GeV on recoil jet Δφ distributions

Compare D-Y Z-> data and Jet20 data, pTZ>10 GeV

Compare D-Y Z-> and Jet20 data, pTZ>10 GeV

Discussion

• Remember that jet1 in the jet20 and jet100 data becomes the Z in the D-Y data, so jet20 jet2->D-Y jet1, jet20 jet3->D-Y jet2, etc.

Jet20 Δφ34 is similar to D-Y Δφ23 Jet20 Δφ23 is dominated by gluon radiation

off jet 2 to form jet3, while for D-Y Δφ12 is dominated at low jetET by ISR from initial quarks, so they don’t look similar.

D-Y and jet20 for scalar∑transtrackpT>15 GeV

More discussion

• However, there is no sign of an enhancement in Δφ(jets2-3) near . In fact, Δφ(jets2-3) looks more isotropic with the ∑pT>15 GeV cut than without it.

• This represents about 10% of the total available high pT muon data, but 30% of the integrated luminosity (3/fb)

Keep looking -79927 Z->

More Δφ plots 79927 Z’s

The last set including Δφ34

High statistics PYTHIA runs-new histograms

High statistics Pythia runs

Pythia Δφ jets1&2 and Δφ jets 2&3

Subtract (with-without) the ∑transtrackpT requirement

Jets 1 and 2 align in the transverse region to balance the pTZ

No sign of double parton scattering

• The behavior of the ∑transtrackpT>15 GeV trigger on the Z sample affects jets 1 and 2.

• jet 3 moves near jet2, to help conserve pT.• Jets 1 and 2 move into the transverse

region of φ, and Δφ12~2*/3, resulting in a cancellation of pTZ.

• Hence there is no evidence for two independent hard scatters.

Pythia recoil jet ET for jets1&2

Pythia recoil jet3 ET and Δφ

Pythia ET jets 1,2&3

• Notice how much broader the ET distributions are with the ∑transtrackpT>15 GeV ‘trigger’.

• This is counter to the notion of a second independent hard scatter, which would result in steeper falling ET distributions, like jet20.

Still no effect, even with jets 3 and 4

• Try cutting harder pTZ>20 GeV, to see if that helps the Δφ(pair-jets_total) plot.

• Try using the entire D-Y mass range 30-130 GeV.

• Look at Δφ for the jets – muon pair, as well as jet-jet.