Benchmarking the SiD. Tim Barklow SLAC Sep 27, 2005

Transcription

1 Benchmarking the SiD Tim Barklow SLAC Sep 27,

2 There is an effort underway at SLAC to do physics benchmarking of the SiD. Activities include: Evaluation and parameterization of the output of event reconstruction software applied to the fully simulated (Geant4) SiD detector Selection of physics benchmark reactions Physics analysis of reconstructed objects 2

3 Why Do Physics Benchmarking? Optimize individual detector designs where is optimal physics/$ as function of component radius and length, B-field, etc.? Help evalulate performance of different detector concepts we re not at this point yet, but eventually we will have to do this. Further strengthen physics case for the ILC despite 15+ years of research, not all processes have been studied an in-depth look at a previously studied physics measurement can cut both ways: more realistic simulation of detector and background may worsen resolution, but inclusion of overlooked signal reactions and decay modes can lead to an improvement. 3

4 SiD Detector outline Whole Detector ~ 12m X 12m X 12m 4

5 SiD Detector outline A high performance detector for the LC Uncompromised performance BUT Constrained & Rational cost This is simulated SiD 00 5

6 Detector Design Issues to be Addressed by Physics Benchmarking (I) Physical dimensions & B-field Tracker (aka Momenter) performance momentum resolution how much is enough? how much multiple scattering is acceptable? tracking efficiency* as function of polar angle, track density, track origin forward region behavior Calorimeter performance granularity, E jet resolution*, MIP tracking *Combined VTX+TRK+CAL performance 6

7 Fraction of the photon(s) energy per event, closer to a charged track than some distance 1.5 x the pad size Fraction Distance in cm BR 2 does not by itself set performance. Pixel size (and Moliere radius) are also very important. 7

8 EMCAL Si/W pixel size: prototypes are 16 mm 2 readout chip: designed for 12 mm 2 How small can we go?? 2-4 mm 2? Need a physics argument for smaller pixels. ρ-> π + π o 8

9 9

10 Detector Design Issues to be Addressed by Physics Benchmarking (II) Vtx detector inner radius, number of layers mechanical design, sensor technology Alignment and Calibration Is Z-pole running required? Alternatives such as? Background true and false track finding efficiency timing-based background veto ee eν W, eez, γ Z e 10

11 Illustration of bunch timing tag Yellow = muons Red = electrons Green = charged hadrons Black = Neutral Hadrons Blue = photons with E > 100 MeV 150 bunch crossings (5% of train) 98 events 920 GeV detected energy 125 detected charged tracks 1 bunch crossing 11

12 WWS (World Wide Study of Physics and Detectors for the ILC) Formed Committee to Develop Physics Benchmark List: 12

13 Physics Benchmark Processes 13

14 Physics Benchmark Processes Reduced Benchmark List : SiD plans initially to study all of these reactions plus ee ττ ρρνν at s= 1 TeV* τ τ *addresses issue of ultimate EM calorimeter granularity 14

15 SiD Benchmarking Tools MC Data sets (stdhep files) of all SM processes at Ecm=500 GeV assuming nominal ILC machine parameters About 50 fb -1 with e- pol=+/- 90% available at ftp://ftp-glast.slac.stanford.edu/glast.u32/simdet_output/simd401xx/whizdata.stdhep (-90% e- pol) ftp://ftp-glast.slac.stanford.edu/glast.u32/simdet_output/simd402xx/whizdata.stdhep (+90% e- pol) 1 ab -1 on SLAC mass storage with all initial e+,e- polarization states Many Monte Carlos (Pythia, Whizard) for producing additional stdhep files Fast MC which takes stdhep files as input and outputs the same kind of reconstructed particle LCIO objects that full event reconstruction software produces (LCIO bindings exist for C++, JAVA, FORTRAN ). 15

16 Fast MC Detector Simulation (I) In the context of SiD benchmarking the Fast Monte Carlo should be considered a Fast Physics Object Monte Carlo. It emulates the bottom line performance of the event reconstruction software in producing the electron, muon, charged hadron, photon and neutral hadron physics objects. Status of Fast MC used by SiD: Tracker simulation uses parameterized covariance matrices based on tracker geometry and material Electron and muon id given by min energy + overall efficiency Photon and neutral hadron energies & angles smeared using single particle EM & hadronic energy & angle resolutions. Photons and neutral hadrons also have min energy and overall efficiency within detector volume. 16

17 Fast MC Detector Simulation (II) Fast MC with nominal single particle calorimeter response gives 17%/sqrt(E) jet energy resolution. This can be tuned to any value by varying the single particle EM & hadronic calorimeter energy resolutions and by replacing charged particle tracker momentum with calorimeter energy a certain fraction of the time. Will improve the parameterization of calorimeter response as we learn more from the particle flow algorithm studies. 17

18 Full Reconstruction/Analysis Overview Java based reconstruction and analysis package Runs standalone or inside Java Analysis Studio (JAS) Fast MC Smeared tracks and calorimeter clusters Full Event Reconstruction detector readout digitization (CCD pixels & Si μ-strips) ab initio track finding and fitting for ~arbitrary geometries multiple calorimeter clustering algorithms Individual Particle reconstruction (cluster-track association) Analysis Tools (including WIRED event display) Physics Tools (Vertex Finding, Jet Finding, Flavor Tagging) Beam Background Overlays at detector hit level 18

19 Full Reconstruction/Analysis Java Analysis Studio (JAS) provides a framework for event visualization (with WIRED) and reconstruction. 19

20 Software CD We have developed a CD containing simulation and reconstruction software as well as documentation and tutorials. In addition, a small amount of data is available on this CD. Full Detector simulation is available through slic (GUI available for Windows). Reconstruction/analysis via org.lcsim & JAS. 20

21 Examples of Tracker (Momenter) Performance Benchmarking 21

22 ee + ZH + μ μ X a = b = Δ M = 103 MeV h a = b = Δ M = 85 MeV h δ p p s = 350 GeV L= 500 t 2 t = a fb 1 p t b sinθ Δσ and ΔBR(h X) Zh relatively independent of ab, since you must use events outside peak to maximize statistics Recoil Mass (GeV) a = b = Δ M = 153 MeV Recoil Mass (GeV) h Recoil Mass (GeV) a = b = Δ M = 273 MeV Recoil Mass (GeV) h 22

23 ee + ZH + μ μ X s = 350 GeV L= 500 fb 1 δ p p t 2 t = a p ΔM vs b h t b sinθ ΔM vs a h ΔM h (MeV) Recoil technique provides best Higgs mass measurement if there is signficant branching ratio for decays with invisible particles a 10 or b

24 Beam Energy Profiles Ebeam (incoming) = 250 GeV Before Collision After Collision Lumi Weighted 50 ppm < E < 250 ppm bias CM σ Ecm 0.1% δ B 4.3% δ B 1.6% E beam (GeV) E beam (GeV) E beam (GeV) Center of Mass Energy Error Requirements Top mass: 200 ppm (ΔM t =35 Mev) Higgs mass: 200 ppm (ΔM H =60 MeV for 120 GeV Higgs) Giga-Z program: 50 ppm 24

25 Measure Ecm with Detector using + + ee μ μ ( γ) ΔE cm (MeV) ΔE Zγ vs b angles only ΔE μμ vs b ΔE μμ vs a ΔE Zγ vs a s = 350 GeV L= 100 fb 1 δ p p t 2 t = a p t b sinθ a 10 or b

26 Examples of Calorimeter Performance Benchmarking 26

27 ee + ZH qqbb δ E E jet jet = 0.3 Δ = 42 MeV M h δ E E jet jet = 0.4 Δ = 46 MeV M h s = 350 GeV L= 500 fb 1 M bb (GeV) δ E E jet jet = 0.5 Δ = 48 MeV M h M bb (GeV) δ E E jet jet = 0.6 Δ = 50 MeV M h Δ E/ E = 60% 30% equiv to 1.4 Lumi M bb (GeV) M bb (GeV) 27

28 ghhh = 2 6 μ /v Standard Model: M = 2λv = 2μ H V ( φ) = μφ + λφ e e ZHH qqbbbb s = 500 GeV, L=1000 Δ E/ E = 60% 30% equiv to 4 Lumi C. Castanier et al. hep-ex/ fb -1 ( Δ E/ E = 60%) 28 (%)

29 Summary Physics benchmarking is an important part of the ILC detector design process. SiD Benchmarking project would be an excellent entry point into ILC physics and detector studies Please contact any of the SLAC SiD people if you are interested. My is 29