γ Detection efficiency for DM Searches with Single-Multi γ

1. γ Detection efficiency for DM Searches with Single-Multi γ OUTLINE • Motivation and Goal • Measurement Method of Systematic error of γDetection • Efficiency. • Full Simulation Results using: • e⁺ e⁻→ µ⁺µ⁻γ • e⁺ e⁻→ e⁺ e⁻γ • Summary and Prospects J-J.Blaising, LAPP/IN2P3

2. Motivation and Goal At CLIC at 500 GeV, with ∫L=500 fb⁻¹ could perform a high precision measurement of the e⁺ e⁻→ ννγcross section. Left plot dN/dEγ: for e⁺ e⁻→ ννγ events; high energy ISR γs from Z return events and low energy γs Eγ < 200 GeV. Right plot dN/dEγ: e⁺ e⁻→ χ̃⁰₁ χ̃⁰₁γ, (model III, mχ̃⁰=100 GeV). An excess in the low E part of the measured dN/dEγ spectrum w.r.t the SM spectrum would be a hint of BSM physics; e.gSusy, large ED… J-J.Blaising, LAPP/IN2P3

3. Motivation and Goal • For 10 GeV < Eγ< 200 GeV, σ(e⁺ e⁻→ ν νγ) =2414 fb. • with ∫L=500 fb⁻¹, a sensitivity of ~ 20 fb can be reached provided that the systematic errors on: • the detection efficiency εγ (signal) • the veto efficiency εv (backgrounds) • are controlled with an accuracy ~ 10⁻³ . • To minimize the dependence from the MC the γdetection and identification efficiency should me measured using “data”. • In e⁺ e⁻→ µ⁺µ⁻, ( e⁺ e⁻) interactions (µ⁺+µ⁻)t, (e±) ~ 0. • In e⁺ e⁻→ µ⁺µ⁻γ, (e⁺ e⁻γ) interactions (µ⁺+⁻)t, (e±) = (∑γ)t. • The correlation between (∑γ)t and(L⁺+L⁻ )tis used to tag events with γs. J-J.Blaising, LAPP/IN2P3

4. Tagging Method • In e⁺ e⁻→ µ⁺µ⁻γ interactions • there are two types of γ’s: • High energy ISR γs (Z return) • Plot (∑γ)tvs (µ⁺+⁻)t • Requiring (µ⁺+⁻)t> 45 GeV • Tags events with 10⁰< θγ < 170⁰ • Low energy ISR and FSR γs • Plot (∑γ)t vs (µ⁺+⁻)t • Requiring (µ⁺+⁻)t > 10 GeV • Tags events with 10⁰< θγ < 170⁰ • Same for e⁺ e⁻→ e⁺ e⁻γ interactions J-J.Blaising, LAPP/IN2P3

5. Simulationand Reconstruction • Events generated with whizard1.95 • Simulation and reconstruction using: • CLIC_ILD_CDR geometry • Mokkaand Marlin CDR software versions J-J.Blaising, LAPP/IN2P3

6. e⁺ e⁻→ µ⁺µ⁻γ Rand Tμ±invariant mass Mμ± good R/T agreement. Right : ∆P=P(T)-P(R) rms=0.9 GeV (µ⁺+⁻)t(R) vs (µ⁺+⁻)t(T) using the selection ofslide 4. good R/T correlation; all events have γ, θγ>10⁰ J-J.Blaising, LAPP/IN2P3

7. γ Detection Efficiencye⁺ e⁻→ µ⁺µ⁻γ Left: (∑γ)t (R) vs (µ⁺+⁻)t (R) ; events with γ (blue) ; without γ (red) good correlation; 48 events without γ=> εγ=0.998 ; 27 have a N Pfo Right: R and T dN/dEγ of the most energetic γ. Eγ (T) smeared assuming ∆E/E=0.009+.25/√Eγ; reasonable agreement. J-J.Blaising, LAPP/IN2P3

8. e⁺ e⁻→ µ⁺µ⁻γdN/d∆Eγ Left : dN/d∆Eγ of most energetic γ with R/T θ match ; rms=6.1 GeV Underflow: 2 or 3 True γ ~ same θγ reconstructed as one γ; ok. Overflow: bad measurements or γs broken into γ +N (45% of evts) Right: dN/d∆Eγ for events without N, ∆Eγ improved; rms=4.5 GeV J-J.Blaising, LAPP/IN2P3

9. e⁺ e⁻→ µ⁺ µ⁻γSummary The good μ± momentum resolution allows tagging of events with 10 < θγ < 170 ⁰ using the correlation between (∑γ)tand (µ⁺+⁻)t. In ~ 45% of events Pandora breaks γ into γ + N; it degrades ∆Eγbut the efficiency measurement is still possible. Assuming ∫L=500 fb⁻¹ e⁺ e⁻→ µ⁺ µ⁻γ events allow the γ detection and identification efficiency measurement with statistical accuracy of 4.2 10⁻³ The γangular distribution is flat => in the forward region 10 < θγ < 30 the statistical accuracy is ~ 1% J-J.Blaising, LAPP/IN2P3

10. e⁺ e⁻→ e⁺ e⁻(γ) Left: (⁺+⁻)t (R) vs (⁺+⁻)t (T) using selection of slide 4. No correlation at low (⁺+⁻)t ; evtsθγ<10 Right : ∆P=P(T)-P(R) ; long low side tail, bad rms dN/dMe± ; bad R/T agreement at high M. Origin is Bremsstrahlung in material J-J.Blaising, LAPP/IN2P3

11. e⁺ e⁻→ e⁺ e⁻(γ) Left: (⁺+⁻)t (R) vs (⁺+⁻)t (T) using selection of slide 4 and rejecting the events with bremsstrahlung (simulation info) Good correlation, all events have γ with θγ>10⁰ Right : ∆P=P(T)-P(R) =2.8 GeV dN/dMe±, good R/T agreement J-J.Blaising, LAPP/IN2P3

12. e± Momentum Resolutione⁺ e⁻→ e⁺ e⁻(γ) ∆P = 2.8 GeVfor e± events and 0.9 GeV for μ± events; why? Left: dN/dθe for events with ∆P<5 GeV and ∆P > 5 GeV; ∆P > 5 GeVcorrelated with low θe values. More F tracks in e± events It affects the momentum resolution of high momentum e± Right plot: dN/dEe J-J.Blaising, LAPP/IN2P3

13. γ Detection Efficiencye⁺ e⁻→ e⁺ e⁻(γ) (∑γ)t (R) vs (⁺+⁻)t (R); events: with γ (blue) ;without γ (red) Left : selection (⁺+⁻)t> 10 GeV; no Bremsstrahlung; εγ=0.985 Right: selection (⁺+⁻)t> 15 GeV; no Bremsstrahlung; εγ=0.993 Better correlation for low (⁺+⁻)t values. 16 events without γ, but 10 have a N; 66 % of events have a N PFO J-J.Blaising, LAPP/IN2P3

14. e⁺ e⁻→ e⁺ e⁻(γ)dN/dθγ and dN/dEγ Left: R and TdN/dθγ ; strongly peaked forward Right: R and TdN/dEγ; energy range covered down to 10 GeV adequate for ISR γ efficiency measurement J-J.Blaising, LAPP/IN2P3

15. e⁺ e⁻→ e⁺ e⁻(γ)dN/d∆Eγ Left: dN/d∆Eγ; Underflow: 2 or 3 True γ ~ same θreconstructed as one γ (ok). Overflow: γ broken into γ+N or bad measurement. Right: dN/d∆Eγ; events without N; still low side tail. Debug [-40,-30]: Pe± well measured; e/γ confusion due to θγ ~ θe ? Not crucial for the γ efficiency measurement but understanding the origin would improve the γ energy resolution. J-J.Blaising, LAPP/IN2P3

16. e⁺ e⁻→ e⁺e⁻γSummary • The e± momentum measurement is strongly affected by bremsstrahlung in material. • To reach a momentum resolution allowing the tagging of radiative events requires: • Measurement of e± after γ radiation • Or rejection of events with bremsstrahlung electrons • In this study bremsstrahlung events identified using simulation info. Not applicable with real data => need a dedicated Marlinprocessor allowing their identification. J-J.Blaising, LAPP/IN2P3

17. e⁺ e⁻→ e⁺e⁻γSummary In ~ 66% of events Pandora breaks γ into γ + N; it degrades ∆Eγ; Despite this bias, assuming ∫L=500 fb⁻¹, e⁺ e⁻→ e⁺e⁻γ eventsallow the γ detection and identification efficiency measurement with a statistical accuracy <10⁻³ (provided that bremsstrahlung events are rejected or the momentum resolution of bremsstrahlung tracks is improved). To provide some input about γ into γ + N breaking I viewed some events => J-J.Blaising, LAPP/IN2P3

18. e⁺ e⁻→ e⁺e⁻γEvents with N e ± well measured; Eγ (T)=69.9 GeV, Eγ (R) =44.9 GeV ; En(R)=4.9 GeV ; Why is there a N? ; why ∆Eγ = 20 GeV?;Leakage in Hcal ? J-J.Blaising, LAPP/IN2P3

19. e⁺ e⁻→ e⁺e⁻γEvents with N e ± well measured, Eγ (T)=24.1 GeV, Eγ (R) =2.5 GeV; En(R)=2.7 GeV Why is there a N? because the γ and e are close to adetector crack ? J-J.Blaising, LAPP/IN2P3

20. Prospects • Can one improve the momentum resolution of e± with bremsstrahlung in material or identify such events? I asked F.Gaede to help to address this issue. • Can one optimize Pandora to minimize γinto γ + N breaking for isolated μ, e ? According to J.Marshall the Pandora version used for the ECAL optimization studies could reduce the γ misidentification => redo analysis. • If these issues are addressed successfully, next step: estimate forward tagging systematic error reachable using radiative Bhabha events. J-J.Blaising, LAPP/IN2P3

21. Backup J-J.Blaising, LAPP/IN2P3

22. Event Tagging J-J.Blaising, LAPP/IN2P3

23. e⁺ e⁻→ µ⁺µ⁻γEvent Tagging • At 500 GeV: √s/2 =250, Requiring: • Pµ⁻⁺ > 242 GeV => Eγ1 < 8 GeV • and Pµ⁺⁻ > 200 GeV => Eγ2 < 50 GeV • and (Pµ⁺ + Pµ⁻)t > 9 GeV => Eγ > 9 Gev • =>γradiated by µ⁺⁻ • with 9 < Eγ< 50 Gev • Sin(θγmin) = PtCut/Eγ (PtCut=9 GeV) • For Eγ=50 GeV => θγ=10⁰ • For Eγ=20 GeV => θγ=27⁰ • For all other events requiring • (Pµ⁺ + Pµ⁻)t > 45 GeV => Eγ>45 GeV • => Select high E γs • For Eγ=250 GeV => θγ=10⁰ Cut Cut J-J.Blaising, LAPP/IN2P3

24. e⁺ e⁻→ e⁺ e⁻(γ) ILC Andre suggested that more recent ILC tracking software couldimprove ∆Pe±. => Same analysis using ILC_LCD geometry, Mokka080003 and Marlin 0116. Left plot: dN/dΔPe± ; no improvement w.r.t slide 15. Right plot: (θγT-θγR) vs θγT ; θγR problem fixed J-J.Blaising, LAPP/IN2P3