Complementary M H Measurements at ILC

Transcription

1 Complementary M H Measurements at ILC (i) Threshold cross-section, ii) Direct reconstruction Graham W. Wilson October 26, 217 Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

2 Introduction Higgs Mass Measurements LHC is measuring, and will continue to measure M H using direct mass reconstruction in the γγ and 4-lepton channels. Current PDG average has an error of 24 MeV based on RunI data. ILC can measure the Higgs mass cleanly using the recoil mass in the dilepton channels. H2 precision estimate is 14 MeV. Sensitivity per fb 1 depends on beam spectrum. Higher s of limited utility. Today: Alternative and Complementary ILC Higgs Mass Measurements M H from the Threshold Cross-Section Direct reconstruction of M H at ILC. Assume 1M produced Higgs particles over the course of the program. Focus for now on 4l and µµ. For direct reconstruction, the issue for each channel, is can one compete statistically and systematically in terms of M H precision with LHC. M = σ M / N Typically with a factor of 1 more events at LHC, one needs ILC detector mass resolution to be 1 times better than LHC. LHC systematic wall?.1.1 σ M? Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

3 Motivation for Precise Higgs Mass Measurement Why measure M H as precisely as possible? Why not? It is naively not as important as a test of the SM as the top mass and the W mass, but it is something that is directly accessible in a relatively optimized way at the initial-stage of an ILC starting at s = 25 GeV. Parametric uncertainties from M H enter coupling determinations due to the strong phase-space related dependence in the ZZ* and WW* decays. See eg. Almeida, Lee, Pokorski, Wells, PRD89, 336 (214). ( ΓZZ Γ ZZ ( ΓWW Γ WW ) ( ) MH = 15.3 M H ) ( ) MH = 13.7 So for.1% parametric uncertainty from M H on these partial widths, need error on M H of 8 MeV. M H (1) (2) Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

4 Introduction: M H from Threshold Cross-section Nominal threshold for ZH production is M H + M Z GeV 25 Measurement of the ZH cross-section can be used to determine M H. M H will likely be known to at least 1 MeV. So no need to scan if sole model-dependence is M H Modest statistics - so statistics dominated measurement. Less need and less advantage (than WW threshold) for a (polarized) scan. cross section [fb] e + +e ZH = 125 GeV m H = 4.3 MeV Γ H = 1 MeV Γ H = 1 GeV Γ H = 1 GeV Γ H s [GeV] Optimal s GeV. How accurate? How much lumi? Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

5 Threshold Cross-section Statistical Sensitivity Study Calculate e + e ν µ ν µ H (ZH-like) cross-sections with Whizard 2.5 B(Z ν µ ν µ ) =.67 Include ISR. For now have neglected beam-spread and beam-strahlung Study various beam polarization scenarios Statistical sensitivity from signal statistics depends on the statistical error on the cross-section measurement and the slope of the cross-section dependence on M H M H = σ dσ dm H 1 1 εl Similar considerations for M W at LEP161 (unpolarized beams) led to M W = 29 MeV / εl(fb 1 ) Cross-section (fb) ZH Production (ECM=216.5 GeV) 8: MH (GeV) Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

6 Threshold Cross-section Study Method Polynomial fits to e + e ν µ ν µ H cross-section vs M H 125. GeV for each s Errors are Whizard integration errors Increase polynomial degree if resulting χ 2 lowered by more than 1 Statistical sensitivity coefficient, at M H = 125. GeV, α 1 σ dσ dm H, is given by A/ A1. Cross-section (fb) ZH Production (ECM=216.5 GeV) -8, / 6 A A A A A E-1 A E-1 A6.9448E-2 A7.1264E-1 A E-2 A E-2 A1.7765E-3 Example for (-8% e, +3% e + ) at s = GeV: α = 1.8 GeV/ fb α ZH α B(Z ν µ ν µ ) = 28 MeV/ fb σ = 3.6 fb. σ ZH = 53.3 fb MH (GeV) Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

7 Optimize s Setting Check s dependence of Higgs mass sensitivity for ZH Optimal s GeV Roughly nominal +.5 GeV (same as WW) Statistical sensitivity factor, 28 MeV/ fb, is ten times less precise (cf 29 MeV/ fb). Substantial integrated luminosity needed to target 25 MeV and below. Higgs Mass Sensitivity (GeV fb -1/2 ) Optimize Center-of-Mass Energy (m H = 125. GeV) For -8%(L) e, +3%(R) e + s (GeV) Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

8 Check e + e ν e ν e H Process has contributions from ZH and WW-fusion. WW-fusion contribution dilutes M H sensitivity significantly. (1.58 GeV/ fb cf 1.8 GeV/ fb for e + e ν µ ν µ H) Cross-section (fb) nue-nue-h Production (ECM=216.5 GeV) -8%, +3% / 12 A E A E-1 A E-1 2 A3.74E E-2 A E E Higgs Mass Sensitivity (GeV fb -1/2 ) Optimize Center-of-Mass Energy (m H = 125. GeV) MH (GeV) s (GeV) [Initial indication: e + e e + e H is ZH-like with no significant dilution from ZZ fusion] Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

9 Separating ZH from WW-fusion Events per bin 12 1 ECM=216.5 GeV. ZH. H to 4-l Entries Mean RMS Underflow Overflow Events per bin ECM=216.5 GeV. nue nue H. H to 4-l Entries Mean RMS Underflow Overflow p (GeV) Higgs p (GeV) Higgs Scope for kinematic cuts to reduce WW-fusion contribution and/or categorize events as more or less ZH-like recuperating some of the lost sensitivity It may be feasible by kinematic cuts/criteria to increase the slope of σ vs M H in some channels (suppressing ISR, mass cuts on the Z etc.) NB So far no considerations at all about background contributions NOR on event selection efficiency. Study naively assumes 1% efficiency for all channels, and no background Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

10 Higgs Sensitivity Numbers from Threshold Cross-section For various beam polarization configurations. For ZH, beam polarization is of some use. Mass Sensitivities ( s = GeV, M H = 125. GeV) Setting α HZ [MeV/ fb] M H [MeV] (5 fb 1 ) M H [MeV] (.5 ab 1 ) 8, , , , , , , , , , , , NB. Signal statistical error only. 1% efficiency, no background. Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

11 To do If you agree that this is of some interest, need some estimate of backgrounds, to clarify feasibility. Clearly need more channel-by-channel event selection procedure to keep backgrounds under control. Note that running at threshold is likely not so attractive for other measurements... Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

12 Direct Reconstruction. LHC 4-lepton mass What can we do at ILC? Can we use other channels? σ M 2. GeV Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

13 Direct Reconstruction. ILC 4-lepton mass Events per bin ECM=25 GeV. nnh. H to 4-l Entries 4 Mean RMS Underflow 641 Overflow 584 Yes indeed. An ILC detector gives mass resolution about 1 times better than LHC! m Higgs (GeV) σ M.19 GeV Estimated using PYTHIA8 events including FSR. Assumes tracking-like momentum resolution for electrons and muons. ECAL-like resolution for photons. Possibly OK for µµµµ, likely underestimate for eeµµ and eeee. Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

14 Numerology LHC cross-sections are substantial. ggfusion (8 TeV) = 21 pb ggfusion (13 TeV) = 49 pb ggfusion (14 TeV) = 55 pb. acceptance*efficiency modest - more so (I think) at 13 TeV. Backgrounds often substantial (especially γγ) Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

15 Simulations Generated Higgs events both with Whizard and Pythia8. 4-vector smearing based on ILD type tracker, ECAL, HCAL and PFA resolutions. Pythia8 samples include FSR (final-state radiation). Most promising channels appear to be Higgs decays to µ + µ and Higgs to 4 leptons (assume for now the same p resolution for e and µ...). The 2-body decay has higher momenta (worse resolution). Four-leptons have lower momenta (better resolution). Here Whizard ZH samples with NO FSR. ECM=25 GeV. ZH. H to mu-mu s=25 GeV. ZH with H 4l Events/.2 GeV Entries 1 Mean Std Dev χ 2 / ndf / 91 Constant Events/.2 GeV Entries 1 Mean Std Dev χ 2 / ndf / 8 Constant Mean Mean Sigma Sigma Di-Muon Mass (GeV) σ M =.168 GeV Four-Lepton Mass (GeV) σ M =.121 GeV Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

16 s dependence? Whizard ZH samples (No FSR) with H µµµµ. At s = 25 GeV, σ M =.121 GeV. Now s = 35 GeV and s = 5 GeV. Events/.2 GeV s=35 GeV. ZH with H 4l Entries 1 Mean Std Dev χ 2 / ndf / 83 Constant Mean Sigma Events/.2 GeV s=5 GeV. ZH with H 4l Entries 1 Mean Std Dev χ 2 / ndf / 85 Constant Mean Sigma Four-Lepton Mass (GeV) Four-Lepton Mass (GeV) σ M =.134 GeV σ M =.156 GeV This is in contrast to the recoil mass. And additionally the WW-fusion production can be utilized at high s (lower p H compared to ZH at high s) Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

17 Simulations with FSR WW-fusion events with Pythia8. These include FSR (final-state radiation). ISR photons are flagged at truth level and not included in the mass estimation. Most promising channels appear to be Higgs decays to µ + µ and Higgs to 4 leptons. The 2-body decay has higher momenta (worse resolution). Four-leptons have lower momenta (better resolution). But more leptons so more FSR. Following are µ + µ and 4l based on tracker + FSR photons (4e, 2e2µ and 4µ). Better ECAL resolution would help... Estimate σ M from Voigtian fit (BW convolved with Gaussian). Events per bin 16 Entries ECM=25 GeV. nnh. H to mu mu Mean RMS Underflow 12 Overflow Entries 1 Events per bin ECM=25 GeV. nnh. H to 4-l Mean RMS Underflow 168 Overflow (GeV) m Higgs (GeV) m Higgs σ M =.21 GeV σ M =.19 GeV Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

18 Full simulations with FSR: Work in progress Use ILD full simulation and reconstruction. Detector ILD l4 v2. ilcsoft v Two samples at s = 25 GeV with Pythia8 including FSR. ZH for decay to µ + µ and WW-fusion for decay to 4µ. First look. Use mass estimate using just the muonic system ignoring any FSR photons. Mass resolutions estimated for events where the muon tracks alone satisfy a Higgs mass constraint,(p fit >.5%). Most of rejected events will have significant FSR and will need different treatment. Events/3 MeV s = 25 GeV, Z H, H µ + µ hrecomass Entries 7597 Mean 125 Std Dev χ / ndf / 81 Constant Mean Sigma Events/3 MeV s = 25 GeV, ν e ν e H, H Z Z* 4µ hrecomass Entries 656 Mean 125 Std Dev χ / ndf / 78 Constant Mean Sigma Measured Mass (GeV) Measured Mass (GeV) σ M =.23 GeV σ M =.16 GeV Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

19 BRs, mass resolution estimates, M guesstimates Assume 1M produced (and detected) Higgs particles for ILC. Neglect background. LHC mass resolution estimate based on CMS. Mass Uncertainty Prospects Decay Mode BR (SM) LHC σ M [GeV] σ M [GeV] M H [MeV] bb.5824 X 3.1 (4.1) cc.289 X 3.1 (18.2) gg.819 X 3.1 (1.8) γγ µ + µ X µ + µ µ + µ e + e µ + µ (.19) 24.7 e + e e + e (.22) 38.6 Others X Averaging the four leptonic channels one gets an overall statistical uncertainty of 1.9 MeV. Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

20 Caveats Other M H Measurement Methods at ILC Full reconstruction of 4-jet events was studied for TESLA. Potential for direct reconstruction with H bb but this is a systematically challenging approach. Event Selection 4l channel. Common wisdom - should be straightforward. I agree. µ + µ channel. To date, ILC studies have not demonstrated efficient isolation of this signal from significant residual backgrounds. It has yet to be shown that this channel can be isolated cleanly. Studies have emphasized ν e ν e H at s >= 5 GeV. My take: I believe that ZH production at s = 25 GeV should be more tractable and deserves in-depth study for this µ + µ channel Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

21 Summary and Conclusions Threshold Threshold Cross-Section Statistical Sensitivity Factor is 28 MeV/ fb 1 M H uncertainty from threshold below 5 MeV will likely need at least 5 fb 1 Significant dedicated luminosity needed to be competitive with recoil mass Firmer conclusions would need more detailed study with event selection and background estimates Would be good to understand if threshold shape is sensitive to EFT parameters. Direct Reconstruction Direct reconstruction of the Higgs mass in golden channels looks to be very competitive with LHC and other ILC methods. One single event can give a mass resolution better than the current world average. This deserves more study with full simulation - started. Electron reconstruction (tracking) needs more attention. ECAL energy resolution is an issue for FSR reconstruction. Even better momentum resolution very valuable. Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21

22 Backup Slides Graham W. Wilson Complementary M H Measurements at ILC October 26, / 21