Measuring the CP state of tau pairs from Higgs decay at ILC in ILD - - PowerPoint PPT Presentation

1

Measuring the CP state of tau pairs from Higgs decay at ILC in ILD

still preliminary, preparing for Santander

Daniel Jeans May/June 2016

2

CP Violation needed to explain baryon asymmetry of universe Currently known sources of CPV not sufficient Higgs of the minimal SM is CP even eigenstate more complex scenarios may also have CP odd states ( H125 being pure CP odd is ~ruled out by LHC) In the case of CP violation in the Higgs sector, H may not be an eigenstate of CP could the Higgs sector be an additional source of CPV?

3

Tree-level coupling of Higgs to fermions L = g f ( cos ψ + i γ5 sin ψ ) f H CP conserving coupling: ψ=0 violating ψ≠0

CP violation via loops can be searched for in H → ZZ, WW

4

projection of spin on some axis: ↑, ↓ spin state of pair of spin ½ particles produced by spin-0 parent: (1/√2) ( |↑↓> + e2iψ |↓↑> ) ψ = 0 : CP even eigenstate ψ = π/2 : CP odd eigenstate

• therwise a mixture

5

distribution of spins (s) of fermions (f) from spin-zero parent spin components parallel / perpendicular to flight direction s z / s⊥

Γ( CP even→ f+ f - ) 1 − s ∼

+ z s z + s+ ⊥ s ⊥

Γ( CP odd → f+ f - ) 1 − s ∼

+ z s z - s+ ⊥ s ⊥

CP state affects transverse spin correlations

6

To measure this, we need fermions: with an appreciable coupling to Higgs whose spins we can (at least partially) reconstruct neutrinos → difficult to observe electrons, muons → difficult to observe spin lighter quarks → hadronisation washes out spin correl tau lepton and top quark → potentially OK significant coupling to H distribution of decay products → spin information

7

tau decay; polarimeter vector Γ ( τ → X ) ~ ( 1 + a h (X) · s ) Polarimeter vector h couples to spin s depends on momenta of τ decay products factor a depends on decay mode: maximal for hadronic decays, smaller for leptonic decays h can be easily calculated from visible daughters for τ± → π± ν and τ± → π± π0 ν if τ momentum is known (BR ~ 11% and 25% respectively)

8

General strategy select Higgs-strahlung events, with H → tau tau Fully reconstruct tau momenta

I know how to do it for Higgs-strahlung with hadronic taus decays

Reconstruct tau polarimeter vectors

I know how to do it in τ±→ π± ν and τ± → π± π0 ν decay modes

look at correlation between transverse components

• f polarimeter vectors

Extract CP-violating angle ψ using full detector simulation and reconstruction

9

simulation, reconstruction

whizard 2.2.8, CIRCE2 beams, ISR e+ e- → f+f-τ+τ- (τ+τ- from 125 GeV Higgs) e+ e- → f+f-τ+τ- (τ+τ- not from Higgs) f = mu, e, uds (some generator level cuts, particularly for e+e-τ+τ-) pythia v8.212 for hadronisation & FSR tauola c++ v1.1.4

• 1. signal decays: τ±→ π± ν and/or τ± → π± π0 ν “rho / ρ”
• 2. all τ± decays

include spin correlations (HSM, HCP(ψ=π/4) , non-H) Mokka simulation ILD_o1_v05 standard Marlin/ILDConfig reconstruction ilcsoft v01-17-09,

standard Pandora steering (latest photon reco) background overlay

10

electron or muon charged hadron photon

Leptonic Z decay channels

11

electron or muon charged hadron photon

look for Z → e+e-, mu+mu- same flavour leptons (PandoraPFA PID)

• pposite charge

associate FSR invariant mass not too far from mZ require exactly 1 Z candidate

reject Z → ee with very forward electrons to reduce non-H backgrounds

12

electron or muon charged hadron photon

Identify hadronic tau candidates >=2 additional charged PFOs 2 most energetic → tau seeds require oppositely charged seeds not identified as e/mu group remaining photons into pi0s if m < mtau with a tau seed

use mass constrained kinematic pi0 fit

add unpaired photons to nearest tau candidate if resulting mass < mtau

13

electron or muon charged hadron photon

reject events with too much charged / neutral-hadron energy or pT in addition to Z and tau candidates

(some from underlying event allowed)

14

electron or muon charged hadron photon

select tau → pi, tau → rho decays based on number of photons/pi0 and visible mass of tau jet reconstruct tau momenta using impact parameters, measured momenta imposing tau mass, pT balance (details in NIM A 810 p51)

require successful reconstruction

select H → tau tau require tau-tau mass consistent with mH require Z recoil mass consistent with mH measure CP properties reconstruct tau polarimeters using measured momenta and reconstructed ptau correlation of transverse components of polarimeters → CP

15

leptonic Z channel look for Z → μ+μ- , e+e- : use Pandora PID, and FSR recovery require exactly one good Z candidate look for two single-prong taus decays: require at least 2 additional charged PFOs assume two most energetic are from tau decays not muon-like or electron-like

• pposite charge

treat as tau jet seeds not too much energy/pT in additional charged PFOs not too much energy/pT in neutral hadron PFOs starting with highest energy photon candidates, try to make pi0 consistent with tau mass apply mass constraint to pi0s: keep if good probability apply cuts on invariant mass and number of photons in tau jet add remaining “orphan” photons to tau jet if they don't take it over the tau mass

16

main updates since last time: increased MC statistics include all tau decay modes more recent reconstruction no cheating of PID add Z → electron channel changes to event selection change to final likelihood fitting

17

SIGNAL PROCESS, SIGNAL TAU DECAY SIGNAL PROCESS, OTHER TAU DECAY BACKGROUND PROCESS (non-H)

Z → e e Z → mu mu

lepton-lepton mass (after selection)

Events / bin

showing LR polarisation, scaled to H20 @ 250 GeV: 1350 fb-1

Events / bin

(stacked histograms)

lepton-lepton mass [GeV] lepton-lepton mass [GeV]

18

Z → e e Z → mu mu

recoil mass (after selection)

SIGNAL PROCESS, SIGNAL TAU DECAY SIGNAL PROCESS, OTHER TAU DECAY BACKGROUND PROCESS (non-H)

Events / bin Events / bin

showing LR polarisation, scaled to H20 @ 250 GeV: 1350 fb-1

recoil mass [GeV] recoil mass [GeV]

19

Z → e e Z → mu mu

tau-tau mass (after selection)

SIGNAL PROCESS, SIGNAL TAU DECAY SIGNAL PROCESS, OTHER TAU DECAY BACKGROUND PROCESS (non-H)

Events / bin Events / bin

showing LR polarisation, scaled to H20 @ 250 GeV: 1350 fb-1

tau-tau mass [GeV] tau-tau mass [GeV]

20

Z → e e Z → mu mu

transverse polarimeter correlation (after selection)

SIGNAL PROCESS, SIGNAL TAU DECAY SIGNAL PROCESS, OTHER TAU DECAY BACKGROUND PROCESS (non-H)

Events / bin Events / bin

showing LR polarisation, scaled to H20 @ 250 GeV: 1350 fb-1 I am surprised there is more BG in muon channel: need to check transverse polarimeter correlation [rad]

21

*** lumi 0 ECOM= 250 GeV Lumi= 1350 fb-1 Pol (e-,e+)=(-0.80, 0.30) 1.35e+06 XSEC #EVTS EFF process 0 eett_nonH sel step 0 15.5409 20980.2 100 sel step 1 15.3813 20764.7 98.973 sel step 2 0.642117 866.858 4.13179 sel step 3 0.00366338 4.94556 0.0235725 process 1 eett_SMHiggs_signal sel step 0 0.0874662 118.079 100 sel step 1 0.0872358 117.768 99.7366 sel step 2 0.0484956 65.4691 55.445 sel step 3 0.0326782 44.1155 37.3609 process 2 eett_SMHiggs_othertau sel step 0 0.589066 795.239 100 sel step 1 0.582987 787.033 98.968 sel step 2 0.0584068 78.8492 9.91515 sel step 3 0.00905114 12.219 1.53652

e e tau tau channel

selection step 0 = all ; 1 ~ find Z ; 2 ~ presel, find & fit tau tau ; 3 ~ full selection

nonH BG SIGNAL

• ther tau

decays

22

*** lumi 0 ECOM= 250 GeV Lumi= 1350 fb-1 Pol (e-,e+)=(-0.80, 0.30) 1.35e+06 XSEC #EVTS EFF process 8 mmtt_nonH sel step 0 20.8433 28138.5 100 sel step 1 20.8201 28107.1 99.8884 sel step 2 3.27458 4420.69 15.7105 sel step 3 0.0164947 22.2679 0.0791366 process 9 mmtt_SMHiggs_signal sel step 0 0.0837425 113.052 100 sel step 1 0.0834467 112.653 99.6467 sel step 2 0.0492233 66.4514 58.7793 sel step 3 0.036084 48.7134 43.0892 process 10 mmtt_SMHiggs_othertau sel step 0 0.560827 757.116 100 sel step 1 0.553412 747.106 98.6778 sel step 2 0.0519277 70.1023 9.25912 sel step 3 0.00875604 11.8207 1.56127

mu mu tau tau channel

selection step 0 = all ; 1 ~ find Z ; 2 ~ presel, find & fit tau tau ; 3 = full selection

nonH BG SIGNAL

• ther tau

decays

23

transverse polarimeter correlation (delta Phi) [rad]

full selection efficiency

colours = different tau decay modes (Z → electron channel)

Does the selection bias the delta-phi distribution? → looks flat → OK

24

Standard Model Higgs CP violating

transverse polarimeter correlation [rad] Events / bin

signal only full reco & selection signal only

25

estimate sensitivity using toy MC experiments assume BG is flat → fit the level from MC assume SIG is sinusoidal → fit the “contrast” from MC make toy MC experiments using Poisson(expected # of events) unbinned maximum likelihood fit to extract CP phase (fix the phase) delta Phi [rad]

26

delta Phi [rad]

a few toy MC examples

27

results of pseudoexperiments (muon channel) scaled to H20 LR polarisation RL polarisation

extracted phase (SM=0) [rad] uncertainty from fit [rad] pull distribution Expected uncertainty (looks about 2x worse than my last report)

28

results of pseudoexperiments (electron channel) scaled to H20 LR polarisation RL polarisation

extracted phase (SM=0) [rad] uncertainty from fit [rad] pull distribution a few failed fits (I guess due to low statistics)

29

Summary

measurement in quite good shape in leptonic channels still some cross-checking needed some more background MC needed → on its way still working on hadronic Z decay channel events being simulated & reconstructed simple analysis is ~ready probably not optimal Plan to show leptonic results @ Santander meeting perhaps also a quick glimpse of the hadronic if analysis proceeds smoothly

30

• ld slides & backups

31

if no leptonic Z found, apply “hadronic” selection use TauFinder (T. Suehara) to look for tau jets require exactly two single prong tau jets of

• pposite charge

assume all other PFOs are in recoiling system

32

Then apply my tau fitting procedure NIM A810 (2016) 51-58 arXiv:1507.01700 6 unknowns per tau pair: two neutrino 3-momenta 6 constraints: tau invariant mass (2) tau impact parameter plane (2) ← requires good knowledge of IP event transverse momentum (x,y = 2) ← little effect from ISR [method minimizes pT, other constraints applied exactly] typically get several solutions: choose only those with positive tau decay lengths and pT < 1 GeV

• f those, choose one with tau-tau mass closest to 125 GeV

calculate polarimeter vector: treat as rho decay if >0 photons [probably too simple, treat as pi decay if 0 photons could be improved]

33

• nly 250 GeV for now [pessimistic]
• nly Z → mu+ mu- for now [pessimistic]

Scale data to full H20 program 1350 fb-1 @ polarisation -80, +30 450 fb-1 @ polarisation +80, -30

• nly tau decays to π± & (π± π0)

[a little over-optimistic, there may be some cross-talk from other modes]

34

mu-mu invariant mass [GeV]

non-Higgs, HSM = ππ + πρ + ρρ, HCP

events per bin

MC reco

35

recoil mass [GeV] non-Higgs, HSM = ππ + πρ + ρρ, HCP

MC reco

events per bin

36

tau-tau mass [GeV]

events per bin

non-Higgs, HSM = ππ + πρ + ρρ, HCP

MC reco

37

Use these 3 observables mu-mu mass, mass recoiling against mu-mu tau-tau mass to reject non-H background visible mass of tau[GeV] reconstructed, selected

non-Higgs, HSM = ππ + πρ + ρρ, HCP

38

tau+ (tau-tau rest frame) tau-

polarim + polarim -

θ-

Longitudinal polarimeter components

angle between polarimeter and momentum (in tau-tau rest frame)

remember:

Γ(H, A→ f+ f - ) 1 − s ∼

+ z s

z ± s+

⊥ s ⊥

sin ( θ+) * sin ( θ-)

remove these badly reconstructed events

non-H HSM (CP even) H (CP mix) HSM (pi pi) HSM (pi rho) HSM (rho rho)

θ+

reconstructed

39

transverse spin correlations non-H HSM (CP even) H (CP mix)

tau - momentum tau- polarimeter tau+ polarimeter

δφ

δφ [rad]

L = g f ( cos ψ + i γ5 sin ψ ) f H

HSM (CP even) has ψ = 0 H (CP mix) has ψ = pi/4 clear modulation, depends on ψ

expect f(δφ) ~ 1 – A (δφ - 2ψ) non-H background ~flat

MC

Events / bin

40

transverse spin correlations non-H HSM (CP even) H (CP mix)

tau - momentum tau- polarimeter tau+ polarimeter

δφ

δφ [rad]

L = g f ( cos ψ + i γ5 sin ψ ) f H

HSM (CP even) has ψ = 0 H (CP mix) has ψ = pi/4

reconstruction: loses statistics loses contrast but still looks OK

reco

Events / bin

41

process final efficiency final events (-0.80, +0.30) final events (+0.80, -0.30) HSM (pi pi) 60 % 6 1 HSM (pi rho) 48 % 23 5 HSM (rho rho) 40 % 22 5 HSM (sum) 51 % 51 11 non-H 0.2 % 10 2

efficiencies, numbers of events after H20 250 GeV only, Z → mu mu only

42

δφ [rad]

reco

Events / bin

Sensitivity Fit red curve with a * ( 1 – b * cos ( dPhi ) ) Assume BG is flat estimate expected number

• f signal, BG events

run toy MC experiments using these inputs Unbinned maximum likelihood fit to resulting dPhi values

43

toy MC results

extracted phase 2ψ [rad] error on phase [rad]

pull

Full H20 @ 250 GeV, Z → mu mu Pe-, e+ = -0.80, +0.30 +0.80, -0.30 mean error on 2ψ ~ 0.45 (0.69) radians ψ ~ 0.23 (0.35) radians ~ 13 (20) degrees Pull looks a little strange...

preliminary

44

First look at hadronic Z decays (uds only)

δφ [rad]

reco

Events / bin

more events less strong modulation non-H HSM (CP even) H (CP mix) HSM (pi pi) HSM (pi rho) HSM (rho rho)

very prelim

45

First look at hadronic Z decays (uds only)

very prelim

process final efficiency final events (-0.80, +0.30) final events (+0.80, -0.30) HSM (pi pi) 29 % 37 8 HSM (pi rho) 28 % 163 37 HSM (rho rho) 25 % 170 38 HSM (sum) 27 % 370 83 non-H 0.6 % 109 21

efficiencies, numbers of events after H20 250 GeV only, Z →uds only

46

First look at hadronic Z decays (uds only)

extracted phase 2ψ [rad] error on phase [rad]

pull

Full H20 @ 250 GeV, Z →uds Pe-, e+ = -0.80, +0.30 +0.80, -0.30 mean error on 2ψ ~ 0.32 (0.61) radians ψ ~ 0.16 (0.31) radians ~ 9 (18) degrees Pull looks not perfect...

very prelim

Toy MC results

47

CP nature of tau pairs from Higgs(-strahlung) decays fully reconstruct tau momenta and polarimeter vectors in τ±→ π± ν and τ± → π± π0 ν decay modes

(total BR ~ 36%/tau ~ 13%/Higgs)

correlation between transverse components sensitive to CP some effects ignored for now: beam backgrounds cross-talk from other tau decay modes

• ther backgrounds (not μμττ final state: expect to be small)

H20 Z→mu mu @ 250 GeV gives precision on CP violating angle ~ 13 deg ← preliminary expect Z → electron mode to have similar power hadronic Z decay less clean, but more statistics. seems to have slightly better sensitivity (~9 deg ← very preliminary)

48

Plans: refine analysis (esp. hadronic Z decays) use Z → electrons Higgs-strahlung @ 500 GeV