Study of Higgs Selfcouplings at ILC J. Tian (Tsinghua U.) Y. Takubo (Tohuko U.) K. Fujii (KEK) Y. Gao (Tsinghua U.) status investigating the full simulated backgrounds llbbH and ννbbH which are not considered in the analyses of ZHH(llHH) and ZHH(ννHH) at 500 GeV. met a problem about the generators of llbbH and ννbbH used by the ILD mass production. previous conclusion + − e + e → ZHH → (l¯l)(bb̄)(bb̄) 4.93 signal events with 0.50 background events: ZHH excess significance: 3.6σ +3.5 5.4 54% for cross section 68% confidence level: −2.3 precision for Higgs self-coupling: 97% llbbH not considered: 14.5 ab, 19.4 llbbbb expected (comparing to current 25.6 llbbbb), if we double the current backgrounds: 3.2σ Mistake: cross section of llbbH was half estimated (previous only for eebbH), should be 30.0 ab 3 eebbH background no beam polarization ! Ecm = 500GeV, MH = 120GeV Ldt = 2ab−1 full simulation of eebbH with the generator used by ILD mass production succeeded on KEKCC normalized MC expected pre-selection econe1+econe2<30 econec1+econec2<10 80<mz<100 mvallqq>0.8 mvalvbbqq>0.8 mvallbbbb>0.5 llhh(llbbbb) 19700 21.18(9.5) 17.50(7.83) 15.67(7.30) 13.82(6.21) 10.47(5.78) 7.37(5.19) 4.93(3.55) llbbbb 10924 25.64 9.31 7.79 6.18 4.12 3.67 0.50 eebbH 4986 29.0 18.6 16.2 14.0 11.2 9.10 4.71 the efficiency of last cut should similar for llbbbb and eebbH (slight higher), but here eebbH is similar with signal ! 4 strange distributions scatter plot of M(H1) versus M(H2) for eebbH background (before cuts) mh1 mh1:mh2 200 180 don’t they look like signal? 160 140 120 100 80 60 40 40 60 80 100 120 140 160 180 200 mh2 these events look like background from eeZH 5 strange distributions scatter plot of M(H1) versus M(H2) for eebbH background (after cuts) mh1:mh2 {econe1+econe2<30&&econec1+econec2<10&&mz>80&&mz<100&&mvallqq>0.8&&mvalvbbqq>0.8&&mvallbbbb>0.5} mh1 180 160 140 120 100 80 40 60 80 100 120 140 160 mh2 6 strange distributions (cont’) forward angle for eebbH background (before cuts) angular distribution of the boson with biggest momentum 0.02 llhh(llbbbb) 0.018 eebbh 0.016 Normalized 0.014 0.012 0.01 0.008 0.006 0.004 0.002 0 −1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1 cos!1st No enhancement at |cos|=1 (don’t like eebbbb case) 7 angular distribution of the boson with highest momentum forward angle for eebbbb background (before cuts) angular distribution of the boson with biggest momentum 0.05 llhh(llbbbb) llbbbb Normalized 0.04 0.03 0.02 0.01 −1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1 cos!1st enhancement at |cos|=1 is very obvious 38 similar situation for ννbbH (BG in ννHH analysis) scatter plot of M(H1) versus M(H2) for ννbbH background (before cuts) mh1 mh1:mh2 200 180 160 140 120 100 80 60 40 40 60 80 100 120 140 160 180 200 mh2 higher statistics, more obvious signal region is 9 check the Feynman diagrams(MadGraph) except the QED process and fusion process, there are 32 diagrams leading contributions come from ZZH and ZHH! 10 check the Feynman diagrams(MadGraph) except the QED process and fusion process, there are 32 diagrams leading contributions come from ZZH and ZHH! 10 check the Feynman diagrams(MadGraph) except the QED process and fusion process, there are 32 diagrams leading contributions come from ZZH and ZHH! 10 summery this situation happens also in the ee ----> eebbbb process, where both ZHH and ZZH contribute to the Feynman diagrams. In the full simulation framework, Whizard is responsible for the generator and M(H)=∞ is set for the non-Higgs process. So it is safe for the eebbbb background. for eebbH process, M(H)=120GeV is set. So the contribution from ZHH involved. I think this is the reason that large parts of eebbH events really look like signal. from MadGraph, we know the leading contribution to eebbH should come from ZZH except ZHH. This part is the real background. But Whizard seems only consider the final state, we need Physsim to generate the ZZH events for full simulation. Suehara-san and Miyamoto-san are helping me to use Physsim for full simulation, still some problem about the stdhep file generation process. 11 backup discussion once we know the real contamination for llbbH comes from ZZH, we can estimate it from the cross section obtained by Physsim. σ(ZZH) = 0.494 fb, the expected number of llbbH events should be (2 ab-1) 0.494×2000×0.067×0.15×2 = 19.8; the expected number of llbbbb from llbbH shoud be 19.8×0.67 = 13.3; according to the efficiencies of llbbbb background and llHH signal, a rough estimation for llbbH: ~ 1 event. That means 4.9 signal events with total 1.5 background events, excess significance will change to 2.6σ (previous estimation: 4.9 S + 1.0 B = 3.2σ) anyway, we need real full simulated ZZH data to give the answer and I think the neural-net could be optimized if ZZH events available. 13