Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

35
19/11/2007 SOCLE@Clermont 1 Manqi Ruan Discussing & Support: Francois, Roman, Vincent, Advisor: Z. ZHANG (LAL) & Y. GAO (Tsinghua)) Higgs mass measurement through μ channel of Higgs strahlungs process (e + e - HZμμH)

description

Higgs mass measurement through μ channel of Higgs strahlungs process (e + e - HZ μμ H ). Manqi Ruan Discussing & Support: Francois, Roman, Vincent, Advisor: Z. ZHANG (LAL) & Y. GAO (Tsinghua)). Outline. Motivation & Software introduction Preselection cuts at generator level - PowerPoint PPT Presentation

Transcript of Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

Page 1: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 1

Manqi RuanDiscussing & Support: Francois, Roman, Vincent, Advisor: Z. ZHANG (LAL) & Y. GAO (Tsinghua))

Higgs mass measurement through μ channel of Higgs strahlungs process

(e+e-HZμμH)

Page 2: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 2

Outline• Motivation & Software introduction

• Preselection cuts at generator level

• Higgs Mass & Xsection determination– Result for SM Higgs– Result if Higgs can decay invisibly

• Summary

Page 3: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 3

Motivation: • Higgs strahlungs process:

• Higgs Recoil Mass

• X section measurement: coupling strength

sEmsm ZZh 222

LNg /2

Higgs invisible decay

Higgs SM decay

μ +

μ -

Page 4: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 4

Sqrt(s) = 230GeV:

Recall the analytic form of error on Higgs mass:

δmh2 = sqrt{((4Ep1-mZ

2)p1k(p1))2 + ((4Ep2-mZ2)p2k(p2))2 } ~ p2 ;

Small sqrt(s) means better Higgs mass resolution!

Motivation: Why Higgs strahlung:

Only muon momentum information is needed inHiggs strahlung channel analysis

A model independent (without any assumption on Higgs decay) analysis can be applied

Any potentially model dependentcut will not be used here

Beam polarization will increase the signal cross section by 58%. (electron 80%, positron, 40%)ISR effect will reduce the cross section with sqrt(s)<300GeV while increase it a little at higher energy

Xse

c/fb

Hmass = 120 GeV

Page 5: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 5

Software chain• Generator: whizard-1.50, pythia 6.4.13 ( with Guinea-Pig

to simulate BS effect: Thanks to cecile! );

• Full Simulation: Mokka-v06-04. with LDC01_sc detector conception (184 TPC layer), the accuracy of tracking system to 5E-5 at δ(1/P) on average

• Reconstruction & Analysis: MarlinReco/Marlin, ROOT;

• Fit: RooFit package

Page 6: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 6

(Invisible) ZH(fb) 6.62 (3310 evt) 7.78 (3890) 4.87 (2435)

eeZZμμvv (fb) 15.42 (7710) 14.61 (7305) 10.18 (5090)

eeWWμμvv (fb) 207.96 (104k) 198.42 (99.2k) 146.44 (73.2k)

Sqrt(s) 230GeV 250GeV 350GeV

(SM) ZH(fb) 6.62 (3310 evt) 7.78 (3890) 4.87 (2435)

eeZZ (fb) 1.34k (672k) 1.27k (635k) 0.856k (428k)

eeWW (fb) 15.86k (7.93M) 15.61k (7.81M) 1.155k (5.77M)

eeqq (fb) 57.6k (28.8M) 52.2k (26.1M) 22.63k (11.3M)

X section of main BG

Non-Polarized beam at 500 fb-1; ISR, BS actived

Page 7: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 7

Pre cuts introduction

• If Higgs invisible decay, no pre cut

• If N_track > 3, (SM Higgs decay), apply following pre cut (pions are included here for the PID have a chance ~1% to misidentify the a pion as a muon)– Energetic pion/muon (E1>15GeV )

– Exist another pion/muon (with energy E2), together with the most energetic pion/muon to form a invariant mass > 70 GeV

– Dynamic cut: 2E1+E2<180&&2E1+3E2>200

– cos(θmumu)>-0.95

Page 8: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 8

Higgs SM decay

Huge SM background:(qq, WW, ZZ)Pre cut is needed beforeProcess to Full Simulation!

Current muon id efficiency ~ 93.6% purity ~ 99%(O.Martin, RDR)

Page 9: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 9

• 3 Pre cuts to reduce the qq back ground– Energetic muons > 15GeV– Invariant mass of Muons >

70 GeV– cos(θmumu)> -0.95

• A few hundreds qq Events survive, far from signal region (115 -140GeV): qq back ground vanishes after pre cut selection

qq background

Page 10: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 10

ev μv τv qq

ev

μv

τv

qq

ee μμ ττ vv qq

evμμττ

νν

qq

ZZ & WW background

Blue: background for Higgs invisible decayGray: background for Higgs invisible decay through tau leptonic decayLight green: background for Higgs SM decayRed, pink and light blue: possible background for Higgs SM decay (pion be misidentified as muon & muon from bb, cc)Yellow and Dark Green: background for Higgs SM decay: H ττ

ZZ WW

Z decay ratio: ~3% to lepton pairs (each), ~20% to neutrino pairs, ~70% to qq

W decay ratio: ~10% to lepton pairs (each), ~70% to qq

Page 11: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 11

Pre cuts effect on WW & ZZ background

• Effect: 56k (of 7.9M) WW events & 11.3k (of 672k) ZZ events survive– Of the 56k WW events;

• 220 events with 2 muon• 46.7k events with muon + pion • 10.55k events with 2 pions• With current PID ~ 400 WW events remains after require both

muons identified (reconstruction efficiency ~ 64%)

– Of the 11.3k ZZ events;• 10.26k μμqq events (majority)• 410 ττqq events • 162 qqqq events• 1.06k 4-lepton events• 8.6k ZZ events remains after reconstruction with require both

muons identified (reconstruction efficiency ~ 79%)

Page 12: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 12

S+B for Higgs SM decay. Br(SM)=100%

Page 13: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 13

ZH ZZ WWTotal EventNum at 500 fb-1 3310 11k 630

Both muon identified 2714 (82.0%) 8638 (78.5%) 400 (63.5%)

|cos(θ)|<0.99 2710 8621 400

Emu>20 2693 8531 318

2E1+E2<182&&2E1+3E2>202 2672 7218 289

cos(θmumu)>-0.95 2462 7022 259

|mZ-mlepton|<10 2363 (71.4%) 5739 (5.74%) 80 (12.7%)

Cuts Chain in Higgs SM decay

Require both muon be identified will reduce our efficiency by ~17%, but also reduce greatly the back ground a muon is more easily to be misidentified in forward region Cut applied on the reconstructed data is more strict than the pre cut at MC truth level

Page 14: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 14

S+B after cuts & Gaussian like background

Page 15: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 15

Fit 2 parameters with likelihood method : mH & Fraction

mH = 119.999 GeV ± 18.0 MeV

Frac = 54.93%±1.4% (δσ ~ 0.19 fb)

Page 16: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 16

Higgs Invisible decay

Exotic Model beyond SM:

SUSY ?Extra dimension ?Heavy neutrino ?

…Main background

e+e- WW, ZZ μμvv

Higgs Decay Br with a forth neutrino

mHwith mass = 50 GeV

K.Belotsky hep-ph/0210153 (2002)

Page 17: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 17

S+B for Higgs invisible decay. Br(inv)=100%

Page 18: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 18

Cuts Chain in Higgs invisible decayZH ZZ WW

X Section (fb) 6.62 15.42 207.96

Total EventNum at 500 fb-1 3310 7710 103980

Reconstructed event num & efficiency (with mH resolved)

3260 (98.5%) 7614 (98.8%) 96661 (93.0%)

|cos(θ)|<0.99 3230 (97.6%) 7566 (98.1%) 96157 (92.5%)

|mZ-mlepton|<10 3091 (93.4%) 7134 (92.5%) 11570 (11.1%)

Emu>20 3091 (93.4%) 7129 (92.5%) 11570 (11.1%)

cos(θmumu)<-0.4 3091 (93.4%) 4765 (61.8%) 6868 (6.61%)

Total energy<110 3086 (93.2%) 1762 (22.9%) 4827 (4.64%)

Cut on W mass resolve:

(2<Ratio<4)

2874 (86.8%) 1165 (15.1%) 3278 (3.15%)

Page 19: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 19

S+B after cuts & Gaussian like background

Page 20: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 20

Fit 2 parameters with likelihood method : mH & Fraction

δmH = 16.5 MeV

δ(Frac) = 1.2% (δσ ~ 0.18 fb)

Page 21: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 21

For arbitrary Br(inv)Combination the result fromHiggs invisible and visible decay( Br(inv) + Br(visible) =100% )Red, invisible part contributionBlue, visible part contributionGreen, overall result

δσ < 0.23 fb

σ ~ 6.6 fb

δmH < 18 MeV

Page 22: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 22

Summary• Accuracy of Higgs mass and cross section measurement through

eeHZHμμ with Higgs SM decay and Higgs invisible decay assumption have been studied.

• An accuracy of about 18MeV on Higgs mass measurement can be achieved at 500 fb-1 with non polarized beam. ( Details: δ(mH) = 13.6 MeV for Pure signal, δ(mH) = 16.5 MeV with Br(inv) = 100%, δ(mH) = 18 MeV with Br(SM) = 100% ). Overall reconstruction efficiency is 71.4% for SM Higgs decay case, and 86.8% for Higgs invisible decay

• Cross section can be measured to an accuracy of 3% level

• It is foreseen to improve a lot with beam polarization for it will not only reduce the WW background but also increase ~58% the cross section of Higgs strahlung channel (electron, 80%, positron, 40%).

• To do: Generate samples with enough statistic to optimize the fit. Optimize the cuts to save signal efficiency. (Especially for Higgs SM decay, with using the jet information. )

Page 23: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 23

Back Up Slides

Page 24: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 24

Fit Algorithm• Generate 2 samples with mH=119GeV and mH=121 GeV

assumption (currently 10k statistic each).

• Getting PDF f(119,m), f(121,m): smooth the Higgs recoil mass spectrum to a PDF function.

• Assume the higgs mass is ma ( 119< ma <121); then the expected PDF f(ma,m) will be a linear combination of f(119,m-(ma-119)) and f(121,m+(121-ma)): (Shift the PDF(119) & PDF(121) to expected position and make a linear combination)

f(ma) = 0.5*((ma -119)* f(121,m+(121-ma))+ (121-ma)* f(119,m-(ma-119)) )

• Use likelihood method to fit ma

Page 25: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 25

How the likelihood method works:

Smoothed mH Spectrum with mH=121/119 assumption:Then shift to measured mH andtogether form an expected PDF at mH=120 (Green)Match the sample data nicely

sample with 10k statistic each

Page 26: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 26

Red/Green/Blue: Smoothed mH Spectrum with mH=121/120/119 assumption and original Histogram (show in error bar);Yellow: expected PDF at mH=120, calculated from f(119) and f(121) match to the sample (Green) nicely. 10k statistic for mH=121 & mH=119 cases; 6k statistic for mH=120 case.

Page 27: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 27

mH measurementwith no background

IO test at 500fb-1

Use toy MC provide by RooFit:

Num of events per run is determinedBy a gaussian distributionGaus(N, sqrt(N)). Totally 500 run

Error on mH nicely agreed withError on sample means distribution:

IO test fine

δmH = 13.7 MeV

Page 28: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 28

mH measured: 119.966GeV±16.6MeVTotal Event Num with 115<mH<140:2871 events + 3202 background;Signal Fraction = 2871/(2871+3202) = 47.275%Measured Fraction = 47.0% ± 1.2%

Fit 2 parameters with likelihood method : mH & Fraction

There is a displacement of about 2 sigma (34 MeV,) which could be eliminate using much meticulous samples (large statistic + detailed Higgs mass assumption)

δmH = 16.5 MeV

δ(Frac) = 1.2% (δσ ~ 0.18 fb)

Page 29: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 29

IO test with toy MC at 500fb-1. Br(invisible) = 100%. 1000 Samples

δmH = 16.5 MeV

δ(Frac) = 1.2%

Truth: 47.275%.Measured: 47.28%±1.2%

Page 30: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 30

X Section measurement Br(inv)=100%

• TotalEventNr*fraction=BranchingRatio*efficiency*luminosity*Xsection• Efficiency = 86.83%; luminosity = 500 fb-1

• Result: 6.613±0.181 fb. Accuracy at 2.7% level

MCTruth:σ: 6.62 fbBr: 100% (invisible decay)

Page 31: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 31

Dynamical Pre CutFor SM Higgs decay

2*E1+E2<180 && 2*E1+3*E2>200

Effect: Signal Channel: 98.8%Back Ground WW: 25.4% ZZ: 31.9%

Signal

WW

ZZ

Page 32: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 32

Effect of individual Cuts for Higgs invisible decay

ZH ZZ WWX Section (fb) 6.62 15.42 207.96

Total EventNum at 500 fb-1 3310 7710 103980

Reconstructed event num & efficiency (with mH resolved)

3260 (98.5%) 7614 (98.8%) 96661 (93.0%)

|cos(θ)|<0.99 3229 (97.6%) 7566 (98.1%) 96157 (92.5%)

|mZ-mlepton|<10 3120 (94.3%) 7175 (93.1%) 11626 (11.2%)

Emu>20 3248 (98.1%) 7584 (98.4%) 94128 (90.5%)

cos(θmumu)<-0.4 3211 (97.0%) 5123 (66.4%) 74202 (71.4%)

Total energy<110 3253 (98.3%) 2056 (26.7%) 25107 (24.1%)

Cut on W mass resolve:

(2<Ratio<4)

2956 (89.3%) 1407 (18.2%) 32444 (31.2%)

Page 33: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 33

Cut on W mass resolve: (2<Ratio<4)

2<Ratio<4

Page 34: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 34

Page 35: Manqi Ruan Discussing & Support: Francois, Roman, Vincent,

19/11/2007 SOCLE@Clermont 35

Momentum resolution vary with energy: 10o, 15o, 20o, 30o ,40o, 60o, 80o polar angle: yellow curve is the Fast simulation result from M.Berggren

• Thanks to hengne’s work on the SIT radiation length, we have gain a factor of 5% improvement in the mu momentum resolution

5*10-5