Full Simulation Study ofSearch for Invisible Higgs Decayswith the ILD Detector at the ILC
Akimasa Ishikawa(Tohoku University)
Just started on Monday.
Invisible Higgs Decay• In the SM, an invisible Higgs decay is H
ZZ* 4n process and its BF is small ~0.1%• If we found sizable invisible Higgs decays, it is
clear new physics signal.– The decay products are dark matter candidates.
• At the LHC, one can search for invisible Higgs decays by using recoil mass from Z or summing up BFs of observed decay modes with some assumptions.– The upper limit is O(10%).
• At the ILC, we can search for invisible Higgs decays using a recoil mass technique with model independent way!– e+e- ZH
ZeeH PPP
known measured
Signal and Backgrounds• Signal
– Pseudo signal : ZH, Zqq, HZZ*4n• Also we have Zee, mm, tt, H4n samples
• Backgrounds– I found qqll, qqln and qqnn final states are the dominant backgrounds. (other
backgrounds also studies but the results mostly omitted from the slides)
– Wen, Wqq– Znn, Zqq– WW semileptonic, Wqq, Wln– ZZ semileptonic, Zqq, Zll– nnH, generic H decays– qqH, generic H decays
MC setup and Samples• Generator : Wizard
– for both signal and backgrounds– ECM = 250GeV– Higgs mass 125GeV– Polarization of P(e+,e-)=(+30%,-80%), (-30%, +80%)
• Throughout the slides, “Left” and “Right” polarization • Samples
– Official DBD samples• Interference is considered, ex WWqqen and enW enqq
– Half of the samples are used for cut determination. The other to be used for efficiency calculation.
mode ZZ nnZ WW enW eeZ"Right" Cross Section [fb] 982 5 2783 684 3992
[fb] Wen sl Znn sl WW sl ZZ sl nnH qqH qqH H4n
eeHH4n
mmHH4n
ttHH4n
“Left” 161.2 271.8 12436 856.9 489.5 210.2 0.224 0.0116 0.0111 0.0111
“Right” 102.2 92.5 869 467.2 275.6 142.0 0.151 0.0079 0.0075 0.0075
Overview of the Selections• Durham jet algorithm
– Two-jet reconstruction• No isolated leptons• No forward going electrons
– It is found ineffective with full simulator• Z mass reconstructed from di-jet
– Also used for Likelihood ratio cut
• cos(qZ)– Production angle of Z– Just apply < 0.99 cut to eliminate peaky eeZ background before making likelihood ratio
• Likelihood ratio of Z mass, cos(qZ), cos(qhel)– cos(qhel) : Helicity angle of Z
• Recoil mass– The final plot– Fitted to set upper limit (not yet).
• The figures to be shown are only eLpR pol. config., not scaled to 250fb -1
Optimization is needed !
No Isolated Lepton• Selection
– Common• P > 10 GeV• Cone cosq >0.98• IP3D < 0.01
– Electron• 1>EEcal / ( EECal + EHCal) >0.9• 1.4 > E/p > 0.7
– Muon• EEcal / ( EECal + EHCal) < 0.4• E/p < 0.3
• Identification efficiency is not high.• We need to optimize the selection
especially leptons going closely to jets.
signal
ZZ slOne Z mm or ttThe other qq
• To eliminate low multiplicity events like Znn, Ztt– NPFO > 16– Ntrack > 6
Number of PFOs and Tracks
signal Znn Zll
Ntrack
NPFO
Z mass• 80GeV < MZ < 100GeV
– Sigma Mean = 90.7GeV, sigma = 10.6GeV
signal
enW nnZ
WW sl
ZZ slqqH
nnH
Fast sim signal
Z mass for enW• The distributions for fast and full simulations are quite
different, I guess due to the lepton ID selections.
WW sl
Fast sim Full sim
Polar angle of Z• We know eeZ is small but can be eliminated by polar angle of Z.• And some backgrounds can be reduced.
– To be included to Likelihood ratio cut
signaleeZ nnZ sl
Recoil Mass• 120 < Mrecoil < 140GeV
– We found • qqH is new peaking background.• nnH is new background peaking close to signal region
signal
enWnnZ
WW slZZ sl
qqH
nnH
Cut Summary• Number of events scaled to 250fb-1
“Right” Wen sl Znn sl WW sl ZZ sl nnH qqH qqH, H4nPseudo signal
No cut 25546 23124 189596 116797 10646 35488 37.87 (1.000)
No lepton 9475 21792 94327 83528 9491 30184 35.84 (0.946)
Trk and PFO 9409 21188 93993 81666 8815 30181 35.58 (0.940)
MZ 1414 15036 11615 35010 639 48 28.89 (0.763)
cosqZ 1349 14850 11388 31476 653 47 28.75 (0.759)
MRecoil 453 1504 869 3309 276 36 23.06 (0.609)
“Left” Wen sl Znn sl WW sl ZZ sl nnH qqH qqH H4nPseudo signal
No cut 40297 67951 2748230 214232 19383 52546 56.07 (1.000)
No lepton 13512 64247 1376700 155472 17297 44609 53.14 (0.948)
Trk and PFO 13431 62401 1371880 152834 16092 44605 52.75 (0.941)
MZ 1567 45186 162967 70813 1201 64 42.50 (0.758)
cosqZ 1548 44656 159303 65418 1195 64 42.31 (0.755)
MRecoil 526 7423 12436 6578 490 43 34.37 (0.613)
Comparison with fast simulator• Upper is s in ab from old study, lower is #events with 250fb-1 with new one.
– Sorry. Multiply 4 for new one to compare with old one.
• Total background s=22fb for old one, s=18fb for new one.
“Right”Events/250fb-1
qqH, H4nPseudo signal
ZZ sl Znn sl WW sl Wen sl nnH qqH
No cut 37.87 (1.000) 116797 23124 189596 25546 10646 35488
No lepton 35.84 (0.946) 83528 21792 94327 9475 9491 30184
Trk and PFO 35.58 (0.940) 81666 21188 93993 9409 8815 30181
MZ 28.89 (0.763) 35010 15036 11615 1414 639 48
cosqZ 28.75 (0.759) 31476 14850 11388 1349 653 47
MRecoil 23.06 (0.609) 3309 1504 869 453 276 36
Cut / s[ab] Signal [%] ZZ nnZ WW enW eeZNo cut 100.00 982000 5000 2783000 684000 3992000Isolated lepton 95.55 835853 4767 1745868 545711 2878061Forward electron veto
95.55 586409 3971 1269207 455945 1709386
Z mass 48.92 156627 1966 34445 26718 752847Cos(qZ) 48.54 153362 1920 33041 25806 8651
LR 48.04 150800 1893 30629 24231 6542Recoil mass 42.10 20073 705 3695 7733 430
Summary and Plan• Just started.
• Found new peaking background, qqH• No significant difference from the result with fast simulation
• Optimize the lepton ID• Likelihood ratio cut• Toy MC study for an upper limit • Temporal result to be Snowmass, hopefully
• Add Z leptonic channel