Yu Bai (IHEP, Beijing) On behalf of the SUSY Weak Production team ACCM 23, Mar 3, 2013 Search of SUSY Weak Production with hadronic taus Search of SUSY

Yu Bai (IHEP, Beijing)On behalf of the SUSY Weak Production team

ACCM 23, Mar 3, 2013

Search of SUSY Weak Production with hadronic taus

SUSY Approval Meeting 2

Support note: https://cds.cern.ch/record/1500884

Editors:Anyes TaffardChristophe ClementFederica LeggerXuai Zhuang

Editorial Board:Dan Tovey (chair)Shirkma BresslerMasahiro KuzeSadrine Laplace

Present study on 20.69 fb-1 , √s=8 TeV Data

Conference note: https://cds.cern.ch/record/1519195


Outline

• Introduction

• Signal grids

• Object and event selection

• Signal region definition

• Background estimation

• Results

• Interpretation

• Summary


Introduction

Either for natural SUSY, or heavy scalars, it is fairly generic for the gauginos to be light.

Higgs into di-photon rate can be enhanced via staus without changing the Higgs to WW/ZZ rates, so light stau with large mixing may help

First study of direct gaugino through the final states with at least two taus, which is complementary of emu final states .


Signal GridsDirect Gaugino Decay

pMSSM Signal Grid: • Heavy Squarks and Gluinos• tan β = 50• M1 = 50 GeV• M2 and μ ranging from 100 to

500 GeV• Mass of stau fixed at 95 GeV,

other sleptons are heavy

Mode A (chargino-neutralino decay):

Mode C (chargino-chargino decay):


Objects Pre-Selection

Electrons and muons• Base line Selection• Overlap removal

Taus• Baseline• Overlap removal• Loose tauid• Tight tauid(at least 1 in each event)

MET : Egamma10NoTau

Trigger : Di-tau trigger || soft-met trigger• Di-tau Trigger: EF_tau29Ti_medium1_tau20Ti_Medium_1• Soft-met Trigger: EF_xe80_tclcw

Jet :• Baseline• Overlap removal • L25: light jet• B20: bjet• F30: forward jet


SR definition Two SRs have been defined

require at least 1 OS tau pair

one SR is after jet veto

another SR is only applying

bjet veto

Suppress Z+jets BG through Z-veto

MET>40 GeV to suppress fake tau background

Apply large mT2 cut to enhance SUSY sensitivity

mT2 cut has been optimized

SUSY Approval Meeting

SR Optimization - mode A

mT2 > 80 GeV mT2 > 90 GeV mT2 > 100 GeV

m

T2 je

t vet

o

mT2

b-ve

to

• mT2 jet-veto: pick mT2 >90 GeV (similar performance)• mT2 b-veto: mT2 >100 GeV (best performance)• Other models are checked, consistent with mode A

8

5fb-1 data usedmT2 jet vetoMt2>80GeV: 3Mt2>90GeV: 2Mt2>100GeV: 1mT2 b-vetoMt2>80GeV: 20Mt2>90GeV: 4Mt2>100GeV: 2


Background components in SRs

Two classes of backgrounds: fake tau (qcd+W, ~75-80%) and real tau background (top, Z+jets, diboson)

Fake tau background: estimated with data-driven approach (ABCD method） Real tau background: use MC simulation


Background Estimation : fake tau background

• Fake Background

• QCD di-jet (2 fake tau)

• W+jets (1 fake tau)

• Can be estimated together due to same fake

tau character

• Not modeled well in MC due to fake rate

• Huge cross section, limited MC statistics


The ABCD method

Use transfer factor (TF) from QCD events (low mT2 region from data) and take the TF difference between QCD and W+jets as systematics

• Variables:

• mT2 , tauid• Base line control region

• A high mT2, loose tauid• B low mT2,, loose tauid• C low mT2, tight+medium tauid

• extrapolation from A to D through a TF (C/B)

• Alternative ABCD method checked (W CR-AB)


Correlation between tauid and mT2

mT2 jet veto mT2 b-veto

No strong correlation between tau id and mT2

The correlation has been taken into account to syst.


Fake tau background purity


2 lo

ose

taui

d ti

ght t

au v

eto

1 tig

ht ta

u +

1med

ium

tau

• In QCD+W CR-C,B (low mT2 region) , QCD + W purity is high (>96%)

• In QCD+W CR-A (high mT2 region), QCD+W purity is ~90%

• non fake tau bkg in CR-A has been subtracted from MC and taken into account to syst.


Signal contamination in region AmT2 jet-veto mT2 b-veto

Mod

e A

Mod

e C

The SUSY contamination in the QCD+W control region A is around 10% for SR OS-mT2 and OS-mT2-nobjet


Fake tau background estimation: Result


Good Data/MC Agreement in non-signal region (mT2 < 90 GeV)

QCD+W Results:In mT2 -jet veto region : • 8.38+/-2.98

In mT2 b-veto region : • 12.23+/-4.48


Kinematic distributions

Good data/MC agreement for leading and next leading tau pt and eta distributions

The ABCD method works well


Validation of QCD+W estimation I

We validated QCD+W estimation in different SM bkg enriched region.

The QCD+W control region definition is close to SM BG validation regions but use loose tau id


Validation of QCD+W estimation I

Good data/MC agreement in all BG enriched VRs


Validation of QCD+W estimation II: Alternative ABCD Method

We also checked the QCD+W estimation using CRs with different bkg components (1L+1M w/o tight).

Good data/MC agreement in alternative ABCD method

The ABCD method proves to be reliable even change the components in the control regions


Real Tau Background Estimation and Validation

top and Z+jets:

No event left at one of the SRs due to low statistics use “ABCD”-like MC driven estimation as default CR definition is similar as QCD+W CR definition but from

MC events (loose tau id, remove MET cut)

Diboson has reasonable statistics, use the MC simulation directly and validated with above “ABCD”-like method

we use two different sets of SFs for real and fake taus. This is why we don't check that all taus are real when we use MC


Real Tau Background

The numbers in SRs are consistent between MC prediction and MC-driven estimation within statistical uncertainty

SUSY Approval Meeting

Results Results from SM

Bkg estimation are compared with the total number of data events in table 19

mT2 distribution for data and SM bkg in eash SR is at fig 20.

The observed and expected number of events in the table (together with uncertainty) are used to calculate the exclusion limit

22


Model independent upper limits on the visible cross-section

Table 22 shows the model-independent upper limits based on the observed and expected number of events in each SR


Best Exclusion Limits – simplified model

Masses of degenerate and are excluded up to 300-330 GeV in the chargino-neutralino simplified model for light (below 50-100 GeV); In the chargino-chargino simplified model, we exclude masses of 200< <330GeV for very light ( below 30-50GeV)


Best Exclusion Limits - pMSSM

Holes formed due to relatively lower signal yield with higher statistical err.


Summary Two signal regions with high mT2 cut have been defined

and optimized No significant excess observed in the signal regions.

Given exclusion limits with pMSSM and simplified model (mode A,C)

Masses of degenerate and are excluded up to 300-330 GeV in the chargino-neutralino simplified model for light (below 50-100 GeV)

In the chargino-chargino simplified model, we exclude masses of 200< <330GeV for very light (below 30-50GeV)

Aiming at Moriond CONF note


Backup


Acceptance*Efficiency of at least two taus for signal grid


Triggers

Using following Oring trigger:

• EF tau29Ti medium1 tau20Ti medium1 => 2taus, pT

leading > 40 , pTnext-leading > 25 GeV

• EF xe80 tclcw, =>MET> 150 GeV

Good Turn-on Curve

Efficiency of EF_xe80_tclcw

Trigger Matching• Leading tau matches to EF tau29Ti medium1• Next Leading tau matches to tau20Ti medium1

Tau Trigger Reweighting

• Use TauSF from tau performance group recommendation


Datasets Data : 20.69 fb-1 , √s=8 TeV (JetTauEtmiss Stream)

Standard Model MC：


Objects and Events Selection


Overlap removal


mT2 without MET cut

mT2 b-vetomT2 jet-veto


Systematic Uncertainty of QCD+W estimation

Correlation between tauid and mT2 : difference between high mT2 region and low mT2 region

• high mT2 : 40 GeV< mT2 < 90 GeV• low mT2 : mT2 < 40 GeV

transfer factor difference: •

• T_W : from MC• f_W: from fitting on W MC mT2 distribution• W fraction in region A: 17%

Non QCD/W backgrounds in region A• Systematic uncertainty taken from background study

Statistic uncertainty in region A


W fraction estimation

OS-mT2

OS-mT2 -nobjet


The result of fake tau background estimation


mt2 distribution of alternative ABCD method


Correlation between tauid and mt2 in alternative ABCD method


Results from alternative ABCD method


Result with alternative JVF Cut(>0.25)

D: SR (medium + tight tau, mT2>90GeV) A: Top/Z/Diboson Control Region: close to SR except

using loose tauID and remove met cut CB: Normalization Region, close to D,A except

40<mT2<80GeV for OS mT2 region

We can extrapolate from Top/Z/Diboson Control region A to signal region D trough a transfer factor from C/B: D = A* C/B

A D

B C

Loose Medium Tau ID

MT2Signal Region

Top/Z/Diboson Control Region

2012/11/20 41

ABCD-like method

Normalization Region

SUSY BG Meeting


mt2 distribution of Z+jets


Z+jets mt2 fit


Z+jets systematic uncertainty: mt2 jet-veto


Z+jets systematic uncertainty: mt2 b-veto


mt2 distribution of top


mt2 fit of top


Top estimation systematic uncertainty : mt2 jet-veto


Top estimation systematic uncertainty : mt2 b-veto


Diboson samples and events number(raw number) in SR


Diboson systematic uncertainty : mt2 jet-veto


Diboson systematic uncertainty : mt2 b-veto


ABCD-like diboson estimation


Acceptance of at least two taus for pMSSM


Acceptance of at least two taus for modeA/C


Signal Optimisation

• Signal Region– SR@OSmt2 mt2>90GeV– SR@OSmt2-nobjet mt2>100GeV

• Signal Mode– Pmssm– Simplified modeA– Simplified modeC


Signal Optimization• Background estimation in Signal region:

– The number of events with 5 fb-1 are scaled to 20.69 fb-1.

– Systematical uncertainty is quoted as 15%.


SR opt with 5fb-1 data@OSMT2

modeA

modeC

pMSSM

Data (5fb-1) Mt2>80GeV: 3 Mt2>90GeV: 2 Mt2>100GeV: 1

Mt2>80GeV Mt2>90GeV Mt2>100GeV


SR opt with 5fb-1 data@OSMT2-nobjet

modeA

modeC

pMSSM

Data (5fb-1) Mt2>80GeV: 20 Mt2>90GeV: 4 Mt2>100GeV: 2

Mt2>80GeV Mt2>90GeV Mt2>100GeV


Zn check for two modeA grids• OSmt2 modeA• mt2>80@ C1 = 150 N1 = 50 s = 23.3351 b = 11.9228 significance = 4.51045• mt2>90@ C1 = 150 N1 = 50 s = 18.8384 b = 7.94852 significance = 4.51581• mt2>100@ C1 = 150 N1 = 50 s = 16.2078 b = 3.97426 significance = 5.20883

• mt2>80@ C1 = 250 N1 = 100 s = 9.86776 b = 11.9228 significance = 2.10468• mt2>90@ C1 = 250 N1 = 100 s = 6.83389 b = 7.94852 significance = 1.83262• mt2>100@ C1 = 250 N1 = 100 s = 4.65025 b = 3.97426 significance = 1.76165

• OSmt2-nobjet modeA• mt2>80@ C1 = 150 N1 = 50 s = 41.1117 b = 79.4852 significance = 2.35261• mt2>90@ C1 = 150 N1 = 50 s = 32.6729 b = 15.897 significance = 5.19816• mt2>100@ C1 = 150 N1 = 50 s = 24.8696 b = 7.94852 significance = 5.65192

• mt2>80@ C1 = 250 N1 = 100 s = 18.947 b = 79.4852 significance = 1.10004• mt2>90@ C1 = 250 N1 = 100 s = 12.4417 b = 15.897 significance = 2.23432• mt2>100@ C1 = 250 N1 = 100 s = 9.16587 b = 7.94852 significance = 2.41341


Expected signal yields(21fb-1)SR@OSmt2: pMSSM


Expected signal yields(21fb-1) SR@OSmt2-nobjet: pMSSM


Signal yields at x/y axis (OSmt2)

• RunNb=164475 mu=100;M2=100 signal yield=0+0RunNb=164476 mu=110;M2=100 signal yield=39.9135+26.3983RunNb=164477 mu=120;M2=100 signal yield=40.9171+25.6674RunNb=164478 mu=140;M2=100 signal yield=16.2225+15.4187RunNb=164479 mu=160;M2=100 signal yield=0RunNb=164480 mu=180;M2=100 signal yield=0RunNb=164481 mu=210;M2=100 signal yield=0RunNb=164482 mu=250;M2=100 signal yield=0RunNb=164483 mu=300;M2=100 signal yield=0RunNb=164484 mu=350;M2=100 signal yield=0RunNb=164485 mu=400;M2=100 signal yield=0RunNb=164486 mu=450;M2=100 signal yield=0RunNb=164487 mu=500;M2=100 signal yield=0RunNb=164488 mu=100;M2=110 signal yield=0RunNb=164501 mu=100;M2=120 signal yield=50.8324+27.6802RunNb=164527 mu=100;M2=160 signal yield=1.46859+3.08171RunNb=164540 mu=100;M2=180 signal yield=6.33867+6.13021RunNb=164553 mu=100;M2=210 signal yield=13.1495+8.33708RunNb=164566 mu=100;M2=250 signal yield=4.85258+4.5601RunNb=164579 mu=100;M2=300 signal yield=27.8572+11.0293RunNb=164592 mu=100;M2=350 signal yield=9.70794+7.27852RunNb=164605 mu=100;M2=400 signal yield=0RunNb=164618 mu=100;M2=450 signal yield=0RunNb=164631 mu=100;M2=500 signal yield=0


Expected signal yields(21fb-1) OSmt2-jetveto(left) / OSmt2-bjetveto(right)

modeA

modeC

modeA

modeC


Cross-section upper limits for Dstau samples

• A rough idea of the value. • More will be done soon.• Better exclusion cross-section upper limits is

provided by OSmt2 jet veto SR.


164604:mu=500,M2=350• Cut 0 : 10000 no cut• Cut 1 : 10000.000977 GRL cut• Cut 2 : 10000.000977 LAr hole veto + TTC incomplete event veto• Cut 3 : 9953.8867188 jet cleaning + LarError cut • Cut 4 : 9868.9423828 >= 1 primary vertex with >4 tracks• Cut 5 : 9735.7275391 cosmic muon veto+ Bad muon veto• Cut 6 : 3053.902832 at least 2leptons• Cut 7 : 822.21057129 tau-tau (pt>40,25GeV or MET>150GeV) ***• Cut 8 : 430.58502197 tau-tau with trigger match• Cut 22 : 78.932563782 //SR4: OS && jet veto &&Z veto && MET>40• Cut 24 : 30.94043541 && mT2_new > 90 ***• Cut 25 : 224.74725342 SR5:OS && bjet veto && zveto• Cut 26 : 185.11817932 && MET>40GeV• Cut 27 : 49.082584381 && mT2 > 100 GeV


164602:mu=400,M2=350• Cut 0 : 10000 no cut• Cut 1 : 10000.000977 GRL cut• Cut 2 : 10000.000977 LAr hole veto + TTC incomplete event veto• Cut 3 : 9963.6582031 jet cleaning + LarError cut • Cut 4 : 9873.9707031 >= 1 primary vertex with >4 tracks• Cut 5 : 9744.5654297 cosmic muon veto+ Bad muon veto• Cut 6 : 3345.3520508 at least 2leptons• Cut 7 : 819.33990479 tau-tau (pt>40,25GeV or MET>150GeV) ***• Cut 8 : 480.61099243 tau-tau with trigger match• Cut 22 : 77.047554016 //SR4: OS && jet veto &&Z veto && MET>40• Cut 24 : 18.065355301 && mT2_new > 90 ***• Cut 25 : 220.07788086 SR5:OS && bjet veto && zveto• Cut 26 : 177.69471741 && MET>40GeV• Cut 27 : 34.182556152 && mT2 > 100 GeV


Exclusion Limits – mode A

Masses of degenerate and are excluded up to 300-330 GeV in the chargino-neutralino simplified model for light (below 50-100 GeV)


Exclusion Limits – mode C

In the chargino-chargino simplified model, we exclude masses of 200< <330GeV for very light ( below 30-50GeV)


Exclusion Limits - pMSSM

Holes formed due to relatively lower signal yield with higher statistical err.

Documents

Yu Bai (IHEP, Beijing) On behalf of the SUSY Weak Production team ACCM 23, Mar 3, 2013 Search of SUSY Weak Production with hadronic taus Search of SUSY