Limit update 1fb-1 extrapolation

Thanks for inputs fromCamila, Jose, Liwen, Bertrand, Li!

Yuji Enari

2011.1.6 LPNHE Higgs meeting

Input: Signal yield (Camila, Jose)

• Camila got numbers, and put them into datacard file

110 GeV 115 GeV 120 GeV 130 GeV 140 GeV

ε[%] Nev ε[%] Nev ε[%] Nev ε[%] Nev ε[%] Nev

ggH 33.8 13.21 34.7 13.40 36.2 14.86 38.5 12.28 39.4 9.27

VBF 41.8 1.15 42.8 1.21 43.5 1.36 44.7 1.17 45.4 0.93

WH 0.72 0.69 0.70 0.50 0.34

ZH 0.39 0.37 0.39 0.28 0.19

ttH 0.10 0.10 0.10 0.07 0.05

Total 15.57 15.77 17.41 14.30 10.78

BG slope (comfirmed by Li)• What we use is Marumi’s number. Here is

output from Li xi = -0.02+/- 0.01(stats) ± 0.01(syst)

Signal shape

• Same as before, from Marc.

95% CL Limit with 1fb-1.Mass point 95%CL +2 s 95%CL +1 s 95%CL 95%CL -1 s 95%CL -2 s

MH=110 GeV 0.95 1.35 2.22 3.65 4.94MH=115 GeV 0.94 1.20 1.97 3.28 4.55MH=120 GeV 0.92 0.98 1.83 2.85 3.90MH=130 GeV 0.93 1.16 1.97 3.25 4.56MH=140 GeV 0.97 1.58 2.65 3.15 5.79

• Script to make plot!

Systematic uncertainty• Reduced set of systematics.

Source

Nominal

Gaussian only

aCB +50%

aCB -50%

Resol . 0.5%

Resol. 1.0%

Lumi (-11%)

Photon (-1%)

Source

Nominal

Gaussian only

Smeared parameter

Lumi (-11%)

Got all mass points!

From Liwen!

Including systematics

Mass point Nominal Gaussian only Smeared param Lumi (-11%)

MH=110 GeV 2.22 1.99 3.51 2.51

MH=115 GeV 1.97 1.90 2.82 2.25

MH=120 GeV 1.83 ------ 2.32 1.88

MH=130 GeV 1.97 1.90 2.82 2.05

MH=140 GeV 2.65 2.59 3.64 2.66

Whooops!

• Forgot to change the range. Better to update them.• Ran 100*100 toy in each, about 24 hours to finish.• Now another set of job running.

Systematic inclusion to the result

• Now trying to make all numbers which have done for sensitivity study.– Evaluate effect of systematic one by one.

• Need to combine. – Easiest solution is same as sensitivity study.– Otherwise, we need to generate S+B hypothesis in toy

MC to evaluate all systematic effect at simultaneously. • No sys in BG, but “xi” is nuisance parameter in the

TOY.

Backup

Limit setting in Hfitter

Slides from Francesco, Lydia, Nansi, 18/1/2010

A technical information• We are using CCALI, If you would like to play

– /sps/atlas/e/enari/Fitters/README_HowToRunHfitterAtLyon– Nicola’s Twiki (https://twiki.cern.ch/twiki/bin/view/AtlasProtected/Hfitter) is very useful!

• You can submit job by running run_sub.csh in HGGfitter/run directory.– You can control # of job and # of toy by changing

run_sub.csh and template/test_v1.C• Effect of statistics:

# of Toy 100 200 500 10000

95% CL limit ratio 13.3 13.0 13.3 13.4

This number is low BG case (scaled down to current BG level) with 36 pb-1.

95%CL limit ratio to SM with L=36pb-1

Mass point 95%CL +1 s 95%CL 95%CL -1 s

MH=110 GeV 9.06 16.4 27.3

MH=115 GeV 8.21 14.7 24.5

MH=120 GeV 7.39 13.4 22.8

MH=130 GeV 7.42 13.5 22.6

MH=140 GeV 8.98 16.4 28.3

Just scaling to L=1.0 fb-1, 95% CL upper limit is ~ 3.

Systematic effect for MH = 120 GeVSource 95%CL +1 s 95%CL 95%CL -1 s

Nominal 7.39 13.4 22.8

Gaussian only 6.97 12.7 21.4

aCB +50% 7.06 12.8 21.8

aCB -50% 9.06 16.4 27.7

Resol . 0.5% 7.40 13.3 23.0

Resol. 1.0% 7.96 14.5 24.4

Lumi (-11%) 8.84 16.1 27.2

Photon (-1%) 8.00 14.5 24.6