lcws2013 higgs recoil mass study at 350 gev apr 27, 2014 jacqueline yan komamiya lab, univ. of tokyo...

LCWS2013

Higgs Recoil Mass Study at 350 GeV

Apr 27 , 2014

Jacqueline Yan Komamiya Lab, Univ. of Tokyo

ILC@ 富山 12013/07/20

What’s New This Week

improve (correct) fitting method re-evaluate data selection performance and cross section error

Compare alternative polarization scenarios

Comparison with sqrt(s)=250 GeV Summary & Plans

Goal: precise measurement of • Higgs mass

• cross section σH

recoil mass study using e+e- recoil mass study using e+e- Zh Zh μ+μ-h μ+μ-h

Ec.m.s. = 250 GeV, L = 250 fb-1Ec.m.s. = 350 GeV, L = 333 fb-1Ec.m.s. = 350 GeV, L = 333 fb-1

polarization: (e-, e+) = (- 0.8, + 0.3)

Signal sample: Pe2e2h_.eL.pR & Pe2e2h._eR.pL

relevant BG process for Zmumu• 4f_ZZ_leptonic• 4f_ZZ_semileptonic• 2f_Z_leptonic• 4f_WW_leptonic• 4f_WW_semileptonic• 4fSingleZee_leptonic• 4fSingleZnunu_leptonic• 4f_ZZWWMix_leptonic• 6f backgrounds (sqrt(s)=350 GeV)

after all cuts, dominant BG are: sqrt(s) = 250 GeV : #1) 2f_Z_l #2) 4f_ZZ_sl #3) 4f_ZZWWMix_l

sqrt(s) = 350 GeV : #1) 4f_ZZ_sl #2) 2f_Z_l #3) 4f_WW_sl no ttbar BG left after data selection

Δσ/σ ～ 4.1 %

fitted recoil mass : Mh =125.7 GeV +/- 144 MeV

recoil mass

calculated recoil mass with correction for 14 mrad beam crossing angle

BG: 3rd order polynomial signal : GPET: 5 parameters :

Gaus (left-side) , Gaus + expo (right side)

Final fitting: float only height and mean, fix others from BG shape and 1st time fitting

1st time: fit only signal float all 5 GPET pars

Final Selection for 350 GeV • 84 GeV < M_mumu < 98 GeV • 10 GeV < pT_mumu < 140 GeV• coplanarity < 3• |cos(θ_Zpro)| < 0.91 (Z production angle)• 120 GeV < Mrecoil < 140 GeV

Muon Selection •reject neutrals • Ptot > 5 GeV• small E_cluster / P_total < 0.5• opposite charge Best track selection : cos(track angle) < 0.98 & |D0/δD0| < 5

Best Z Candidate Selection 2 mu candidates with opposite chargechoose pair with invariant mass closest to Z mass

Evaluate performance in recoil mass range of 120– 140 GeV

calculate recoil mass with correction for 14 mrad beam crossing angle

Sig + BG

fitting using weighted binsFitting in wide range 115-150 GeV

Sig + BG

Integrated fitted func in (120 – 140 GeV)

error : 3.15/76.67 = 4.1 % Nsig = 1080+/-44 <ε> = 48.9 +/- 0.5 %

Nbg= 3467S/BG = 0.32 significance = 16.2

Very close to result from just counting under histogram

•<Nsig = 1089+/-45 •<ε> = 48.9 +/- 0.5 %•S/BG = 0.31• significance = 16.1

Compare Alternative Polarization Scenarios

See supplementary EXCEL files (attached to meeting page)

Comparisons (0,0) seems best for both Δσ/σ(+0.8, -0.3) yields best S/N , Δσ/σ not bad eitherWW BGs significantly suppressed (< 1/10 of (-0.8, *0.3)) other major BGs less also (esp for eLpR) (< ½ of (-0.8, +0.3) even though statistics is lower, some BG process is suppressed ? no big difference between (-0.8, + 0.3) and (-0.8, 0) no big difference in cut efficiency for each individual BG process is e+ polarization really necessary (practical)? Re-consider for 250 GeV (accelerator issues)

　 　 ε 　 Δσ/σ 　 Nsig 　 S/N significance350 GeV 　 　 　 　 　 　 　 　 　

(-0.8,+0.3) 48.9+/-0.5% 4.10%　 1089+/-45 0.31 16.1　

(-0.8,0) 　 47.6+/-0.5% 4.00%　 865+/-34 　 0.32 14.4　

(0,0) 　 47.7+/-0.5% 3.10%　 737+/-23 　 0.37 14.1　

(+0.8,-0.3) 47.8+/-0.5% 3.70%　 738+/-27 　 0.48 15.4　

250 GeV 　 　 　 　 　 　 　 　 　

(-0.8,+0.3) 69.9+/-0.5% 4.10%　 1752+/-72 0.26 19　

(-0.8,0) 　 67.2+/-0.5% 4.00%　 1390+/-56 0.26 17　

(0,0) 　 66.9+/-0.5% 3.20%　 1183+/-38 0.3 16.6　

350 GeV (-0.8, + 0.3)(-0.8, 0)

(0 , 0)(+0.8, -0.3)

Compare with Results for 250 GeV

250 GeV (-0.8, + 0.3)

(0, 0)

(-0.8, 0)

Δσ/σ ～ 4.1 %

fitted recoil mass : Mh =125.6 GeV +/- 144 MeV

recoil mass

calculated recoil mass with correction for 14 mrad beam crossing angle

BG: 3rd order polynomial signal : GPET: 5 parameters :

Gaus (left-side) , Gaus + expo (right side)

Final fitting: float only height and mean, fix others from BG shape and 1st time fitting

1st time: fit only signal float all 5 GPET pars

Final Selection for 250 GeV analysis after filling root files

• 84 GeV < M_mumu < 98 GeV • 10 GeV < pT_mumu < 70 GeV• |cos(θ_Zpro)| < 0.91 (Z production angle)• 120 GeV < Mrecoil < 140 GeV• also tried 123 GeV – 135 GeV

Muon Selection •reject neutrals • Ptot > 5 GeV• small E_cluster / P_total < 0.5• opposite charge Best track selection cos(track angle) < 0.98 |D0/δD0| < 5

Best Z Candidate Selection 2 mu candidates with opposite charge if several possibilities : choose pair with invariant mass closest to Z mass

Evaluate data selection efficiency in within range of 120– 140 GeV

calculate recoil mass with correction for 14 mrad beam crossing angle

optimization of data selection method for Higgs recoil mass study using e+e- Zh μ+μ-h @ Ec.m.s =350 GeV, L = 333 fb-1

• compared with @ Ec.m.s. = 250 GeV, L = 250 fb-1

•350 GeV: • ε_sig = 48,9 +/-0.5 %, S/B ～ 0.31, significance ～ 16.1 Δσmeas> / <σmeas> = 4.1 % • fitted recoil mass : 125.7 GeV +/- 144 MeV

250 GeV: • ε_sig = 69.9 +/-0.5 %, S/B ～ 0.26, significance ～ 19 Δσmeas> / <σmeas> = 4.1 % fitted recoil mass : 125.4 GeV +/- 44 MeV

• After improvement of fitting, efficiency and Δσ/σ seem much improved

Compared different polarization scenarios : (-0.8, 0.3) vs (+ 0.8, -0.3) vs (0 ,0)• (0,0) is best for Δσ/σ , not realistic (?)• (+0.8, -0.3) seems to have better S/N (?), but not practical (?)• not much difference between (-0.8,0) amd (-0.8, +0.3) •Need to balance physics merits with accelerator (e+ production) issues

Summary

Use Toy MC to evaluate cross section error may resolve some statistical fluctuations

Deeper study of which process is suppressed by alternative polarization scenarios

improve data selection Implement Pt_bal and likelihood cut ?

Plans

BACKUP

• preliminary comparison of cut efficiency between MC truth and reconstructed for 350 GeV signal and a few dominant BGs : integrated under histograms, counted in region 123-135 GeV

Rec 　cut 　 signal eff 4f_ZZ_sl eff 2f_Z_l effraw 　 2288 100% 188087 100.00% 2226361 100.00%only best mu pair 2214 97% 25217 13.41% 329581 14.80%cos(trackAng)<0.98 2202 96% 19906 10.58% 305146 13.71%84 <M_inv <98 1824 80% 5314 2.83% 94671 4.25%10 <P_Tdl<140 1817 79% 5198 2.76% 26063 1.17%copl < 3 　 1790 78% 4853 2.58% 22766 1.02%cos(θZ)<0.91 1707 75% 3672 1.95% 10765 0.48%120 GeV <M_rec <140 GeV 1089 48% 1133 0.60% 1050 0.05%MC 　cut 　 signal eff 4f_ZZ_sl eff 2f_Z_l effraw 　 2288 100% 188087 100.00% 2226361 100.00%only best mu pair 2288 100% 26219 13.94% 417982 18.77%cos(trackAng)<0.98 2208 97% 17385 9.24% 306297 13.76%84 <M_inv <98 1981 87% 5115 2.72% 102691 4.61%10 <P_Tdl<140 1945 85% 5006 2.66% 24539 1.10%copl < 3 　 1945 85% 4691 2.49% 24539 1.10%cos(θZ)<0.91 1852 81% 3599 1.91% 11813 0.53%120 GeV <M_rec <140 GeV 1256 55% 1056 0.56% 986 0.04%

MC

reconstructed

Black : muon-Green: muon +

Muon track angle sqrt(s) = 350 GeV

more forward/ asymmetrical than 350 GeV

MC

reconstructed

Black : muon-Green: muon +

Muon track angle sqrt(s) = 250 GeV

(0, 0)(+0.8, -0.3)

(-0.8, + 0.3)(-0.8, 0)

recoil mass cut 120-140 GeVfit in 114-140 GeV<n> = 1115 +/- 46<ε> = 42.8 +/- 1.7 %

<n>/sqrt(<n>+<B>) = 14.7<n>/<B>=0.24

Δσmeas> / <σmeas> = 4.1 %

250 GeV

Compare 250 GeV & 350 GeV: integrate fitted functions

Calculate Error of σmeas from relative error of height fitting

recoil mass cut 120-140 GeVfit in 115 – 150 GeV<n> = 1145 +/- 54<ε> = 51.4 +/- 2.4 %

<n>/sqrt(<n>+<B>) = 17.3<n>/<B>=0.35

Δσmeas> / <σmeas> = 4.7 %

350 GeV

recoil mass cut : 123 – 135 GeVfit in 114-140 GeV<n> = 1169 +/- 48<ε> = 44.9+/- 1.8 %

<n>/sqrt(<n>+<B>) = 18.5<n>/<B>=0.42

Δσmeas> / <σmeas> = 4.1 %

recoil mass cut 120-140 GeVfit in 114-140 GeV<n> = 1115 +/- 50<ε> = 50.1 +/- 2.2 %

<n>/sqrt(<n>+<B>) = 17.2<n>/<B>=0.36

Δσmeas> / <σmeas> = 4.5 %

Compare different polarization scenarios

sqrt(s) = 350 GeV L = 333 fb-1

For now, keep same cut parameters as (-0.8, +0.3) (they could be optimized)

(0, 0)Pol weight(eLpR) = 0.5*0.5Pol weight(eRpL) = 0.5*0.5

<n> = 737 +/- 23<ε> = 33.1+/- 1.0 %

<n>/sqrt(<n>+<B>) = 14.2<n>/<B>=0.37

Δσmeas> / <σmeas> = 3.1 %

(+0.8, -0.3) Pol weight(eLpR) = 0.1*0.35Pol weight(eRpL) = 0.9*0.65

n> = 737 +/- 27<ε> = 33.1+/- 1.2 %

<n>/sqrt(<n>+<B>) = 15.8<n>/<B>=0.51

Δσmeas> / <σmeas> = 3.7 %

(-0.8, + 0.3)Pol weight(eLpR) = 0.9*0.65Pol weight(eRpL) = 0.1*0.35

<n> = 1080 +/- 44<ε> = 48.5 +/- 2.0 %

<n>/sqrt(<n>+<B>) = 16.2<n>/<B>=0.32

Δσmeas> / <σmeas> = 4.1 %

(-0.8, 0)Pol weight(eLpR) = 0.9*0.5Pol weight(eRpL) = 0.1*0.5

<n> = 857 +/- 34<ε> = 38.5+/- 1.5 %

<n>/sqrt(<n>+<B>) = 14.5<n>/<B>=0.32

Δσmeas> / <σmeas> = 4.0 %

Signal efficiency 61 % S/N 0.39 Significance 〜 21.2

Evaluation in 123 – 135 GeV region sqrt(s) = 250 GeV count under histograms

DBD Samplesevent weight = pol_weight * ( process_cross_section * assumed_integrated_luminosity )/ ( number_of_reconstructed_events )

sqrt(s) = 250 GeV/grid/ilc/prod/ilc/mc-dbd/ild/dst-merged/250-TDR_ws/

20 x Higher statistics data : /hsm/ilc/grid/storm/user/a/amiyamot/myprod/ild/dst-merged /250-TDR_ws/reference: meta files and diagrams in http://ilcsoft.desy.de/dbd/generated/other.html

sqrt(s) = 350 GeV/grid/ilc/prod/ilc/mc-dbd/ild/dst-merged/350-TDR_ws/List of samples: http://www-jlc.kek.jp/~miyamoto/CDS/prod_status/REC_ILD_o1_v05_350GeV.html

LumosityTDR baselinesqrt(s)=250 GeV, Lumi=0.75 x 10^34 cm^-2 s^-1 assume L = 250 fb-1sqrt(s)=350 GeV, Lumi=1.0 x 10^34 cm^-2 s^-1 assume L = 333 fb-1

Polarization: (-0.8, + 0.3) compare with (+0.8, -0.3)

stacked, eLpR

stacked, eRpL

250 GeV

recoil mass

after implementing all cuts

Black : MC

Red: reconstructed

pT

Invariant mass

Black : MC

Red: reconstructed

Z production angle

coplanarity

dilepton PT, 350 GeV

do cut : 10 GeV< pT_dl < 140 GeV

Signal , 250 GeV: 0 – 80 GeV peak at about 60 GeV

Before cuts)

Signal , 350 GeV: 0 – 140 GeV peak at about 135 GeV

Before cuts)