Analisi di dati del Run I

dal top al bottom

con un pizzico di charm

P. Giromini

Congregazione per la Dottrina della Fede

5 analisi completate fra il 1999 ed il 2002 Usano campioni di dati diversi, ma hanno un filo logico

comune 1: Heavy flavor properties of jets produced in p p

interactions at s1/2=1.8 TeV (blessata, draft in preparazione) 2: Studio di decadimenti semileptonici sequenziali

(b l c X, c l s X) (blessing frozen dopo 3 mesi di presentazioni, non si trovano sbagli ma il risultato non piace, analogo dei superjet)

3: Misura di time-integrated mixing parameter di adroni con beauty (presentata, blessing parte inizi ottobre)

4: Studio dello spettro di massa invariante di dimuoni fra la J/ e la (1S) (presentata, in coda)

5: Studio di eventi con J/ +leptone (analogo del BC) (presentata, in coda)

1Heavy flavor properties of jets produced

in p p interactions at s1/2=1.8 TeV

A small step forward in resolving the long-standing

discrepancy between the measured and predicted b-

cross section at the Tevatron

Jets with heavy flavor Over several years CDF has been comparing the

fraction of jets with heavy flavor (b and c quarks) to a simulation based upon the Herwig and CLEO (QQ) Monte Carlo generators

Heavy flavor-identification : EfficiencySECondary VerTeX (SECVTX) 50%Jet-ProBability (JPB) 50%

Data sets : W+ jet events, generic-jet data (JET20, JET50, and JET 100), di-jet events with one jet containing a lepton (low-pT lepton sample)

Jets with heavy flavor We have used the low-pT lepton

sample to calibrate the data-to-simulation scale factors for the SECVTX and JPB tagging algorithms

We have used generic-jet data to tune the parton-level cross sections evaluated in Herwig within experimental and theoretical uncertainties

W+jet events P~50%

PRD 64, 032002 (2001)

Jets with heavy flavor We also identify heavy flavors

by searching jets for semileptonic decays (SoftLeptonTagging ) efficiency 20%

PRD 65, 052007 (2002)

P~0.4%

P~50%

Anomalous W+ 2,3 jet events with a supertag

The kinematics of these events has a 10-6 probability of being consistent with the SM simulation [PRD 64, 032004 (2002)]

Some of us took these events very seriously; hep-ph/0109020 shows that the superjets can be modeled by postulating the existence of a low mass, strong interacting object which decays with a semileptonic branching ratio of the order of 1 and a lifetime of the order of 1 ps

Since there are no limit to the existence of a charge –1/3 scalar quark with mass smaller than 7 GeV/c2

[PRL 86, 1963 (2001)], the supersymmetric partner of the bottom quark is a potential candidate

Light sbottom (bs) Lot of very recent buzz hep-ph/0007318 uses it to resolve the long-standing

discrepancy between the measured and predicted value of R for 5 < s1/2 < 10 GeV at e+ e- colliders

PRL 86, 4231 (2001) uses it in conjunction with a light gluino which decays to b bs to explain the difference of a factor of 2 between the measured b-quark production cross section and the NLO prediction

If light bs existed, Run 1 has produced 109 pairs; why we did’t see them ?

Strategy the NLO calculation of p p bsbs predicts =

19.2 b for a squark mass of 3.6 GeV/c2 (Prospino MC generator program) .

The bb production cross section at the Tevatron is = 48.1 b (MNR)

The cc production cross section at the Tevatron is = 2748.5 b (MNR)

The NLO calculation has a >50% uncertainty because of the renormalization scale

Strategy We have adjusted the heavy flavor production cross sections

calculated by Herwig within the theoretical and experimental uncertainties to reproduce the rate of SECVTX and JPB tags observed in generic-jet data.

In that study we have used jets with with uncorrected ET>15 GeV and ||<1.5; they correspond to partons with transverse energy approximately larger than 18 GeV

For partons with transverse energy larger than 18 GeV, = 84 nb , = 298 nb , and = 487 nb (10% contamination)

we could have easily tuned the Herwig generator to explain in terms of SM processes an additional 10% pair production of scalar quarks: f= 382 nb , and f= 487 nb

Strategy What if there is a bs quark with a 100% semileptonic

branching ratio In b-quark decays, a lepton is produced in 37% of the cases In c-quark decays, a lepton is produced in 21% of the cases By requiring that at least one jet in the previous QCD data

contains a lepton (inclusive low-pT lepton sample or generic-jet data with SLT tags), the production cross sections are = 84 nb , = 110 nb , and = 102 nb (28% contamination)

By fitting a simulated pseudo-experiment which includes sbottom production with a conventional QCD simulation using rates of SECVTX and JPB tags, one might find f= 194 nb , and f= 102 nb

Strategy As a next step, we select events in which an additional jet

contain a soft lepton (57% contamination) If the data are due to bottom and charmed quark production

only, we will find = 93.4 nb (71.8 nb due to b-quarks and 21.6 nb due to c-quarks) in the data and the simulation

If the data contain bs production, the cross section in the data will be = 146.3 nb (40.7 nb due to b-quarks, 21.6 nb due to c-quarks, and 84.0 nb due to scalar quarks)

In generic-jet data with SLT tags, one would find = 244 nb without a scalar quark, or , with a scalar quark, = 296 nb in the data and = 244 nb in the simulation

Use generic-jet data to cancel out experimental uncertainties (efficiencies, fake removal,…)

Data sample Events with 2 or more jets with ET > 15

GeV and at least two SVX tracks (taggable,|

one electron with ET> 8 GeV or one muon with pT > 8 GeV/c contained in one of the jets

Require I > 0.1 Reject conversions Apply all lepton quality cuts used in

the high-pT lepton sample

68544 events with an electron jet and 14966 events with a muon jet

away jet

lepton jet

l

Perform a detailed comparison between data and simulation using SECVTX, and JPB tags on both the lepton- and away-jets

Differently from previous CDF analyses, this study checks at the same time the cross section for producing at least 1 b with |imperfect NLO calculation), 1 b +1 b with |apparently robust NLO calculation)

Then we check the semileptonic branching ratio of heavy flavor hadrons by counting the number of a-jets with SLT tags in the data and in the simulation

Strategy

Mistags and tagging efficiencies PRD 64, 032002 (2001) Mistags (tags in a jet without heavy flavor) are evaluated with

parametrized probability functions derived in generic-jet data. We estimate a 10% uncertainty.

Since we use a parametrized simulation of the detector, we have measured the data-to simulation scale factor for the tagging efficiency of the SECVTX and JPB algorithms.These factors were determined with a 6% accuracy and implemented into the simulation.

The SLT simulation uses efficiencies for each selection cut measured using data; we estimate a 10% uncertainty, which includes the uncertainty on the semileptonic branching ratio

The SLT efficiency in supertags is corrected for the data-to-simulation scale factor measured in generic-jet data: (85±5)%

Simulation Use the Herwig generator program (option 1500, generic 2 2 hard

scattering with pT > 13 GeV/c) bb and cc production are generated through processes of order

such as qq bb Processes of order are implemented through flavor excitation

diagrams, such as g b g b, or gluon splitting, in which the process g g g g is followed by g bb

Use MRS (G) PDF’s The bottom and charmed hadrons are decayed with QQ (version

9_1) We select simulated events which contain hadrons with heavy flavor

and at least one lepton with pT > 8 GeV/c These events are passed through QFL, a parametrized simulation of

the CDF detector and treated as real data We have simulated 27156 electron events (98.9 pb-1) and 7267

muon events (55.1 pb-1) with heavy flavor

Evaluation of the heavy flavor content of the data

Before tagging, approximately 50% of the lepton jets do not contain heavy flavor; they are mostly due to fake leptons

Mistags in the lepton-jets and away jets are evaluated with a parametrized probability and removed

The fraction (1-hf) of events in which the l-jet does not contain heavy flavor is not simulated. In these events, away-jets can have tags due to heavy flavor. Their rates are estimated using a parametrized probability of finding a tag due to heavy flavor in generic-jet data. Using a sample of l-jets containing electrons due to identified conversions, we estimate a 10% accuracy. It is a slight overestimate.

away jet

lepton jet

l

Fit of the simulation to the data Use 6 fit parameters corresponding to the direct, flavor excitation and

gluon splitting production cross sections evaluated by Herwig for b and c-quarks

Ke and Kaccount for the luminosity and b-direct production The parameters bf, bg, c, cf, cg account for the remaining production

cross sections, relative to the b-direct production The ratio of b to c direct production constrained to the default value

(about 1) within 14% the ratio of b to c flavor excitation constrained to the default value

(about 0.5) with a 28% uncertainty bg constrained to (1.4±0.19) cg constrained to (1.35±0.36) The tagging efficiencies are also fit parameters, and are constrained

to their measured values within their uncertainties (6% for b-quarks, 28% for c-quarks)

Fit result

2/DOF=4.6/9

Fit result

Fhf = (45.3±1.9)% for electrons

Fhf = (59.7±3.6)% for muons

Kinematics

SECVTX tagged

KinematicsSECVTX tagged

KinematicsL -jet SECVTX tagged

A-jet

A-jet with

SECVTX tags

KinematicsL -jet SECVTX tagged

KinematicsL -jet SECVTX tagged

A-jet with

SECVTX tags

Comparison of a-jets with SLT tags in the data and the normalized simulation

SEEN 1137±140.0 (±51.0

STAT.)

EXPECTED 746.9±75.0 (SYST)

SEEN 453±29.4 (±25 STAT.)

EXPECTED 316.5±25.4 (SYST)

(±15.8 SLT efficiency, ±20 fit)

Systematics (away-jets with SLT tags)

In events due to heavy flavor, there is an excess of 391 a-jets with a SLT tag with respect to the simulation (1137.8 observed and 746.9 expected), having removed 619.3 fake tags [the events in which the l-jet does not have heavy flavor contain 901.9±91 a-jet with SLT tags (74% fake+ 26% heavy flavor): slight overestimate].

If one could increase the fake rate in events with heavy flavor by 60%, the excess would disappear. However, in generic-jet data, the fake rate is already 74% of the SLT tagging rate.

Since fakes are approximately 74% of the SLT rate, the 10% uncertainty of the fake removal was evaluated by comparing observed rates of SLT tags to the parametrized prediction in all QCD samples. Most of the 10% comes from the fact that different QCD sample have slightly different heavy flavor purity

Systematics (fake SLT tags)

The heavy flavor content of generic-jet data has been evaluated using SECVTX and JPB tags

In generic-jet data the number of SLT tags due to heavy flavor is therefore known with a 13% error, mostly due to the 10% uncertainty of the SLT tagging efficiency

Therefore the real uncertainty on the fake rate is no larger than 2.6%

Data – simulated H.F. = 15783±423 fakes

Parametrized SLT fakes 15570

Systematics (fake SLT tags) Away-jets in the inclusive lepton have a higher

heavy flavor content (26%) than generic-jet data (13%) .

Could the fake rate in jets with heavy flavor be anomalously large ? Could the SLT efficiency or the semileptonic branching ratio in the simulation be grossly wrong ?

Jets with SECVTX or JPB tags in generic-jet data have a heavy flavor content ranging from 86% (JET 20) to 71% (JET 100) . The rate of SLT tags in these jets is not higher than in the simulation

This does not support simulation deficiencies as an explanation for the discrepancy

h.f . fakes

low pT (SLT) 1138 619

SECVTX+SLT 944 507

JPB+SLT 1167 856

Conclusions We have measured the heavy flavor content of the low pT

inclusive lepton sample by comparing rates of SECVTX and JPB tags in the data and the simulation

We find good agreement between the data and the simulation tuned within the experimental and theoretical uncertainties

We find a 50% excess of a-jets with SLT tags due to heavy flavor with respect to the simulation; the discrepancy is a 3.5 systematic effect due to the uncertainty of the SLT efficiency and background subtraction. However, comparisons of analogous tagging rates in generic-jet data and their simulation do not support any increase of the efficiency or background subtraction beyond the quoted systematic uncertainties

Conclusions A discrepancy of this kind and size is expected, and was the

motivation for this study, if pairs of light scalar quarks with a 100% semileptonic branching ratio were produced at the Tevatron

The data cannot exclude alternate explanations for this discrepancy

Previously published measurements support the possibility, born out of the present work, that approximately 50% of the presumed semileptonic decays of heavy flavor hadrons produced at the Tevatron are due to unconventional sources

correlated +b-jet cross section

PRD 53, 1051 (1996) 90% of the cross section at

Data are 1.5 times larger

than the NLO calculation The NLO cross section is

not very sensitive to the scale

The NLO value is approximately equal to the Born value

bb•BR

bb correlations (dimuons)

PRD 55, 2547 (1997) Data are 2.2 times larger

than the NLO calculation D0 has a similar result The dependence of the

NLO prediction on the scale is less than 20%

Born and NLO values are within a few percents

bb•BR2

2Study of sequential semileptonic decays of b-hadrons

produced at the Tevatron

Given the previous anomaly, it is of interest to study sequential semileptonic decays of hadrons with heavy flavor

We use the same data set and simulation of the previous analysis

Comparison of data and simulation

Data: 1447±65 dileptons (±44 stat. and ±48 syst.)

Sim.: 1181±129 dileptons (±51 fit and sim. stat., ±118 SLT eff.)

2 effect

Dilepton kinematics

Dilepton kinematics

Systematics (fake dileptons) “Misidentified leptons are expected to be present in

equal amount in OS and SS dileptons” [PRD 49].The technique of subtracting SS from OS dileptons to remove this background has been used many times in CDF [PRD 49 (1994), PRD 59 (1999)].

In contrast with previous analyses which used this technique, we now find a discrepancy between the data and the simulation prediction. We want to investigate further. Use the electron sample only (larger statistics)

Fake dileptons The heavy flavor simulation of the inclusive electron sample, after

normalization, contain 955 ± 108 OS and 63 ± 9 SS dileptons In the data, there are 1450 OS and 399 SS dileptons To get rid of this discrepancy, one would like to explain in terms of

background 495 ± 114 OS and 276 ± 20 SS dileptons (R=1.79±0.43) On average, a jet contain the same number of positive and negative

lepton-candidate tracks. When searching for a second lepton in a jet in which one track has

been already identified with an electron, one expects the number of OS candidates to be larger than the number of SS candidates [Nc(OS)=54938 and Nc(SS)=34744] (R=1.58)

If this is correct, there is no dilepton excess, but a number of previous analyses are wrong

Dilepton background (conclusions)

Applying the parametrized fake probability to the Nc(OS) candidates one gets OS=302 ± 30 [used in PRD 60, (1999)]

However, the standard fake parametrization derived from generic-jet data might not be adequate to describe the fake rate in jets which already contain a lepton with pT> 8 GeV/c

Use generic-jet data to derive a fake probability for jets which already contain a lepton

Dileptons (generic-jet data)

Dilepton background (conclusions)

When applying P(OS) and P(SS) of generic-jet data to the candidate tracks, one predicts 236 ± 44 OS and 229 ± 14 SS fake dileptons

Note that b-hadron mixing is not simulated; when using =0.118, the simulation predicts 915 ± 105 OS and 103 ± 15 SS dileptons due to heavy flavor; the difference between the data and the heavy flavor expectation is 535 ± 111 OS and 236 ± 24 SS dileptons

P(SS) of generic-jet data provides the correct prediction of SS dileptons

A background of 276 events with a 10% accuracy seems to be a fair estimate of the fake OS dileptons

Sequential b-decays

According to the simulation, most of the dileptons in the same jet are contributed by semileptonic cascade decays of b-hadrons. They are simulated with the QQ Monte Carlo program. How good is it ?

Help from DELPHI (2000-060 PHYS 861, 2000) DELPHI compares 573474 hadronic Z-decays to a JETSET 7.3

simulation consisting of 992988 events normalized to the same luminosity of the data

Use events with |cos T|<0.95, where T is the thrust axis Select events with two leptons with pT > 3 GeV/c and | |<1 in the same

emisphere as defined by the thrust axis. Divide events in OS and SS First show agreement between DELPHI data and simulation. Then,

show agreement between the DELPHI and our simulations at the Z-pole

DELPHI

Sequential b-decays

DELPHI simulation: comparison

between OS-SS dileptons and

sequential b-decays identified at

generator level

Comparison between our and DELPHI

simulation at the Z-pole

3Ratio of like-sign to opposite-sign dileptons

Recent publications by Ed Berger et al. explore an explanation within the context of MSSM to the discrepancy between the measured cross section and the NLO prediction

Hep-ph/0103145 postulates the existence of a light gluino which decay to b and

The pair production of gluinos provides an inclusive b-cross section comparable to conventional QCD production

However, pair production of gluinos and subsequent decays to b quarks also leads to an increase of like-sign dileptons

bb

b~

The like-sign dilepton increase could be confused with an enhanced B0-B0 mixing and in a value of the time-integrated mixing parameter, larger than the world average 0.118±0.005

Hep-ph/0103145 predicts and, after comparing to the CDF Run 1A result of 0.131±0.026, concludes that a better measurement is needed in Run II

Other results: UA1 measured and CDF (Run 0)

We repeat the Run 1A analysis using also Run 1B dimuons and e-

Method of analysis Use the same method of RUN 1A: fit the impact

parameter distributions with the expected distributions for various sources

The main sources of dileptons are semileptonic decays of charmed and bottom mesons, and prompt decays of onia and Drell-Yan production

Impact parameter distributions for b and c quarks are derived from a simulation which uses Herwig (option 1500), QQ version 9_1, and QFL’

Prompt shape The prompt shape is derived

using muons from 1S) decays (9.28< M <9.6 GeV)

The background is removed using dimuons with 9.04< M <9.2 and 9.54< M <9.7 GeV

Shapes

Ratio of numbers of leptons with d<0.008 to that with d>0.008 cm

b quarks 0.85 c quarks 2.03 prompt 32.3

Shapes Since there are two leptons the fit is performed in the

two-dimensional space of the impact parameters The impact parameters of each lepton are

uncorrelated variables Use a binned maximum log likelihood method with

the same binning used for the Run 1A analysis Fit BB, CC, PP, BP, and CP combinations

Dimuons with invariant mass > 5 GeV/c2

18420 OS 9279 LS

Removed 6264 OS and 1302 LS with 9.2 < M <10.5 GeV

Fit result R=BB(LS)/BB(OS)

=0.600±0.022 CC/BB(LS+OS)=0.33 ±0.03

e-events

4331 LS7802 OS

e-events

R=BB(LS)/BB(OS) =0.597±0.028

CC/BB(LS+OS)=0.23 ±0.03

Time integrated mixing parameter

R=

With f=2 f f) f= 0.123 ± 0.015 (1A syst.) and f=0.2157±0.0226 (syst.)

f= f e+ ffe ff e = 0.06, f= 0.142 and f=0.185±0.0161 (syst.)

= 0.168±0.011 (dimuons) = 0.184±0.013 (e )

f [21- ] +f) (1-f) [22f

(stat.) ±0.01 (syst.)

bb cross section (dimuons) the number BB(OS+LS) corresponds to a cross section

which is a factor of 2.2 larger than the NLO prediction D0 has a similar result For dimuons b-quarks with pT> 6.5 GeV/c and |

the ratio NLO/Born(LO) is 1.05 (not 2 as for the inclusive b cross section)

The dependence on thescale (is ±20% and not ±50% as for the inclusive b cross section

the inclusive NLO cross section can be easily adjusted by a factor of two, but not the NLO cross section for 2 b-quarks with pT> 6.5 GeV/c and |

Ratio of cc to bb production (dimuons)

CC/BB(LS+OS)=0.33 ±0.013 (present fit) D0 has also a measurement of the bb cross section using dimuons

with pT> 4 GeV/c and |The various contributions are disentangled with fits to the pT

rel distributions. D0 finds CC/BB(LS+OS)=0.31 ±0.09 In the simulation CC/BB(LS+OS)=0.18 ±0.02 for us, and 0.13 ±0.03

for D0 Is the Herwig simulation all right ? UA1 also measured the bb cross section using dimuons with pT> 3

GeV/c. The various contributions are disentangled with fits to the pTrel

distributions, but the ratio CC/BB(LS+OS) is forced to be 0.13±0.06 as expected from the ISAJET simulation

Ratio of cc to bb production CC/BB(LS+OS)=0.23 ±0.03 in edata In the esimulation (15000 events generated with option 1500 of

Herwig) CC/BB(LS+OS)=0.13 How reliable is the simulation ? In the NLO calculation (MNR) the ratio of the bb to cc cross section

for producing heavy quarks with |and transverse momentum large enough to produce an electron with ET >5 GeV and a muon with pT >3 GeV/c is 0.72

The CASCADE generator (hep-ph/0109146) models correctly the inclusive b-cross section measured by CDF and DO .With this generator we find BB/CC=0.74

In Herwig, this ratio is 0.73 We start seeing another problem: the cc cross section is a factor of 4

larger than the NLO calculation

Deviation from the conventional QCD expectation

The measured bb cross section (2 x NLO) and the value of agree with the Berger’s prediction

However, if Berger was right we also should have found CC/BB(LS+OS)=0.06 and not 0.23 ±0.03 as in edata

4Search for 1-- bound states of scalar quarks

SPEAR set a ll= 100 eVlimit to such states The expectation is ll= 10-30 eV(Nappi) In the reasonable assumption that production and

decay proceed via three-gluon exchange B= Y(1s) B x R, with R= (mY/m)3 x Y

Y =1.32 KeV and R=2-6 % We have 104 Y(1S) and one expects >200

eV, also in a simulation

eV in a simulation

A 150 MeV bin (±2 contains 9000 events

Background reduction As seen in the previous

analysis, 78% of the dimuons with invariant mass larger than 5 GeV are due to heavy flavor

Cut on isolation and impact parameter significance

Tune the cuts on the Y(1S) to have >90% efficiency

The background is cut by a factor of 3

Fit result Resolution fixed (38

MeV) Mass 7.2 GeV 250±61 events 3.5 effect

5: search for Bc sb sb is approximately 15% of b b

~

b (bs) c

c c

ls

J/

In the assumption that the bs semileptonic branching ratio is 100% for a bs quark and 20% for a b quark the channel J/+l+X is equally contributed by the Bc and Bc

~

Bc discovery Uses electrons with pT >2

GeV/c and |1.0 , but muons with pT > 3 GeV/c and |0.6

Signal of about 20 events; given the lifetime the cross section x BR is a factor of 2 larger than expected in units of B J/ K

By using muons with |1.0 and pT > 2 GeV/c we increase the acceptance by more than a factor of 4 without degrading the signal to noise ratio

J/+l+X

Slides di supporto

Data

Heavy flavors in the simulation are identified at generator level

Fit result-parameter corr. coeff.

For

NLO and Herwig calculations Herwig ignores interference terms between the Born

approximation and the NLO diagrams, and evaluates a gluon splitting+flavor excitation contribution which is a factor of 3 larger than the Born approximation.

In the NLO calculation the contribution of the Born cross section and of the gluon splitting+flavor excitation are approximately equal using the renormalization scale when using the scale the scale the NLO calculation gets closer to Herwig.

The fact that the ratio between NLO and Born is about two and is not stable as a function of the renormalization scale is taken by the experts as an indication that NNLO corrections are important

The relevance of the Herwig result, which models the data, is the indication that the effect of NNLO correction should be that of canceling the interference terms

NLO and Herwig calculations However, in this specific analysis we are interested in comparing

rates of a-jet with heavy flavor (signaled by SLT or SECVTX tags) in events in which the l-jet has also heavy flavor

These jet have | |<1 and corresponds to partons with ET > 18 GeV In this case Herwig evaluates that the gluon splitting+flavor

excitation contribution are 40% of the Born contribution and not a factor of 3 higher

For this type of kinematics, the ratio of the NLO to Born calculations is also of the order of 1.1-1.3. In addition, for this topology, the NLO calculation depends little on the choice of and it appears to meet general criteria of robustness.

KinematicsL -jet SECVTX tagged

A-jet

A-jet with

SECVTX tags

KinematicsL -jet SECVTX tagged

A-jet with

SECVTX tags

KinematicsL -jet SECVTX tagged

A-jet with

SECVTX tags

fragmentation in generic-jet data

550,000 generic-jet events in the data and in the Herwig simulation (JET20, JET50, and JET100). 1324 supertags in the data1342 simulated supertags

away-jets with SLT tags

No dependence of the result from the fit normalization

Nb and Nc are the numbers of predicted a-jets with bottom and charmed flavor

bJPB=0.40, c

JPB=0.20, bSLT=0.058, c

SLT=0.029 HFTSLT(a-jet)= b

SLT (Nb + 0.5 Nc) bJPB /b

JPB = b

SLT /bJPB HFTJPB(data) =

bSLT /b

JPB (5126.6±146.7) =743±78

Independent of the heavy flavor composition of the fitted simulation

No dependence of the result from the fit normalization

In other words Remove, e.g., the 14% constraint on c/b=1 The fit returns c=2.8±1.6 However, HFTSLT(a-jet) =597.6±69.3 603±66

(electrons) 149±21 156±21 (muons)

Systematics (away-jets with SLT tags) In the fitted simulation, 79% of the a-jets with an SLT tag due to

heavy flavor are attributed to b-production The fraction of SLT tagged jets is higher for b-production (2%) than

for c-production (1%) If one could increase the b-purity from 79% to 100%, the predicted

number of SLT tagged a-jets would increase from 747 to 839, closer to the observed 1138 ± 120

Check the b-purity by comparing the numbers of l+ D0, l +D±, and l +non-prompt J/combinations in the data and the fitted simulation

b-purity

D0 : 126.0 ± 15.5 in the data and 139.9 ± 15.0 in the simulation

D± : 73.7 ± 17.8 and 68.5 ± 14.1

J/: 90.8 ± 10.1 and 101.9 ± 11.4

Ratio of the b-purity in the simulation to that in the data is 1.09 ± 0.11

D0

D±

Same flavor OS-SS dileptons 2.6 < mee< 3.6 GeV/c2

2.9 < m< 3.3 GeV/c2

259 ± 17.2 and 209.2 ± 23.7 (before tagging)

89.7 ± 10.5 and 100.5 ± 12.4 (SECVTX)

90.8 ± 10.1 and 101.9 ± 11.4 (JPB)

Cross check with J/ mesons from B-decays

In generic-jet data we do not observe any excess of jets with SLT tags or supertags

We do observe an excess after enriching the b-purity of the QCD data by requiring a lepton-jet

We study a sample of jets recoiling J/ mesons from B-decays. We use the same J/ data set and selection used for the measurement of the J/ lifetime and fraction from B-decays

1163 J/ over a background of 1179 events estimated from the side-bands (SB)

J/ lifetime

The number of J/ mesons from B-decays is N= ()-SB+–SB-) =561, which is 48% of the initial sample

In the 572 away-jets we find 48.0 ± 15.1SECVTX, 61.7 ± 17.3 JPB tags, and –9.4 ± 14.4 SLT tags

In the simulation we expect 8.1 ± 1.1 SLT tags

Dilepton kinematics