Measurements of spin and parity of the new boson in ATLAS Kirill Prokofiev (NYU)
Outline Present spin results Short term plans Long term plans page 2
Introduc>on ATLAS and CMS have observed a new Higgs- like resonance produced in pp collisions. In Standard Model the Higgs boson is spin- 0 CP- even par>cle. Produc>on mechanisms: gluon- gluon fusion (spin- 0,2) and/or qqbar produc>on (spin- 1,2). (some of the) Observed decays: γγ, ZZ, WW. Observa>on of the two- photon decay excludes spin- 1 hypothesis (Landau - Yang). However there are loopholes. Measurement of proper>es: deduce spin and parity from measured distribu>ons of kinema>c observables. Kinema>c observables: only one angle cos θ* for γγ or two- fermion decay. More for ZZ and WW. page 3
Present spin studies in ATLAS In 2012 ATLAS has presented two major studies of the spin and parity of the Higgs- like resonance around 126 GeV. Decays: H- >ZZ- >4l and H- >γγ. Spin and parity hypotheses considered: 0 +, 0 -, graviton- like tensor with minimal couplings 2 m+, pseudo- tensor 2 -. 2 m + and 2 - produc>on. gg- >X: g 1 =1; qq- >X: ρ 12. 2 m + decay g 1 =g 5 =1. 2 - decay: g 8 =g 9 =1. Follow in general the formalism described in the JHU papers: Y. Gao, et al., Spin determina/on of single- produced resonances at hadron colliders, Phys. Rev. D81 (2010) 075022, arxiv:1001.3396 [hep- ph] S. Bolognesi, et al., On the spin and parity of a single- produced resonance at the LHC, Phys. Rev. D86 (2012) 21. For the 2012 studies, ATLAS used the JHU generator. page 4
Present spin studies in ATLAS Which Spin- 2 models makes sense? The decay amplitude has 10 form factors. Produc>on mechanism can also vary: gg, qq, VBF. General idea: Given the number of possibili>es, we cannot exclude generic spin- 2. We should start with the model with minimal couplings and exclude it in favor of the SM hypothesis, which is rela>vely well defined. If during this study we observe something funny have a deeper look in spin- 2 models. It is possible that both ggf and qq produc>on mechanisms contribute to the spin- 2 state. The possible mixtures should thus be studied. page 5
ATL- CONF- 2012-168 Two photon decay channel Study based on the single produc>on angle: cos θ*. page 6 13 v - 1 at 8 TeV Considered models: 0 + and 2 + m Both qqbar and ggf produc>on mechanisms are considered for the 2 + m state. No categoriza>on: 123.8 GeV<m γγ <128.6 GeV
ATL- CONF- 2012-168 Two photon decay channel page 7 Expected p- value: 3.4% (1.8σ) Observed p- value for 0 + hypothesis: 29% (0.55 σ) Observed p- value for 2 + m hypothesis: 8.4% (1.4 σ) qqbar scan: no discrimina>on power for 75% qq and higher in two- photon channel.
ATL- CONF- 2012-169 Four lepton decay channel page 8 4.6 v - 1 at 7 TeV and 13 v - 1 at 8 TeV Sensi>vity to all Spin- parity combina>ons. 5 produc>on and decay angles and reconstructed masses of the intermediate Z s: m 12 and m 34. Θ* of the first Z- boson. Φ and Φ 1 between the decay planes defined in the Higgs rest frame. Θ 1 and Θ 2 of the nega>ve leptons defined in the corresponding Z rest frame.
ATL- CONF- 2012-169 Four lepton decay channel page 9 Crea>ng a single mul>variate discriminant using Spin/Parity dependent observables (angles and masses). Two methods: BDT and pseudo- MELA. Different treatment of correla>ons and acceptance, different systema>cs. Two bins in m 4l with high and low S/B ra>o: (115 GeV <m 4l <121 GeV) + (127 GeV <m 4l <130 GeV); AND (121 GeV <m 4l <127 GeV). Test sta>s>c: ra>o of profiled likelihoods.
ATL- CONF- 2012-169 Expected p 0 (Nσ) (BDT) Four lepton decay channel Expected and observed p 0 values to exclude various spin and parity hypotheses. The shaded column shows the p 0 to exclude spin and parity hypotheses in favor of the 0 + state. Both methods show comparable results. 0+ is the favorite hypothesis preferred by data. 0- is disfavored by all other hypotheses. Data slightly prefer 2 + over 2 -. 100% ggf produc>on assumed for the spin- 2. Observed p 0 (Nσ) (BDT) Observed p 0 (Nσ) (pseudo- MELA) page 10
What we try to agree on A common spin and CP studies strategy should be defined among CMS, ATLAS and theory community in the near future. Theore>cally sound set of benchmarks. Easy to compare and eventually to combine among experiments. Priority 1: spin 0 +, 1 + and 2 + (to disentangle if the observed resonance is spin 0, 1 or 2). Start with the spin- 2 models with minimal couplings and trying to exclude those in favor of the spin 0 hypothesis. If anything funny is found on the way, give a closer look to spin- 2 models. Priority 1: exclude spin 0 - to improve current result on the parity of the resonance. page 11
What we try to agree on Priority 2: spin 1 - : Excluding this unlikely scenario. Priority 3: spin 2 - : Excluding the generic 2- hypothesis a.k.a. 2 - h. S. Bolognesi, et al., On the spin and parity of a single- produced resonance at the LHC, Phys. Rev. D86 (2012) 21. Produc>on processes: Priority 1: ggf and qq produc>on for the spin- 2. Priority 2: VBF. Currently possible for the γγ, at some point will also become available for the ZZ. Priority 3: WH/ZH and }H. page 12
Further inves>ga>ons By the end of 2012 the LHC has delivered: 5.6 v - 1 at 7 TeV and 23.3 v - 1 at 8 TeV. With this amount of data each experiment can exclude most of the reasonable spin and parity models. Establish the dominant spin 0, exclude spin 1, exclude spin 2 + m and 2- h Spin 1 is in principle forbidden by Landau- Yang theorem. Decay channels we can study: ZZ- >4l (spin and parity), γγ(spin), WW (spin and parity). Separately and in combina>on. Large contribu>ons of qq spin- 2 produc>on can also be excluded (ZZ, WW). CP nature of Higgs through the VBF produc>on: T.Plehn et al., arxiv:hep- ph/0105325v2 Spin studies in VH H- >bb: Ellis et al., arxiv:1208.6002v1 page 13
ATLAS- PHYS- PUB- 2012-004 CP- viola>on in Higgs sector European Strategy for Par>cle Physics. Study of the feasibility to exclude large CP- viola>ng contribu>ons with H- >ZZ*- >4l decay at large luminosi>es. General amplitude of interac>on between Spin- 0 boson and 2 vector bosons: A(X VV ) ~ (a 1 M X 2 g µν + a 2 (q 1 + q 2 ) µ (q 1 + q 2 ) ν + a 3 ε µναβ q 1 α q 2 β )ε 1 *µ ε 2 *ν a 1 and a 2 are associated with coupling of CP- even Higgs to a pair of vector bosons; a 3 is associated with that of a CP- odd Higgs boson. CP- conserving tree- level SM: a 1 = 1, a 2 = a 3 = 0. CP- viola>on: a 3 0, given a 1 0 and/or a 2 0. In general a i can be momentum- dependent form factors that may be generated in loops with new heavy par>cles. It is always possible to select a 1 to be real. a 2 and a 3 are in general complex. page 14
ATLAS- PHYS- PUB- 2012-004 CP- viola>on in Higgs sector Studying the sensi>vity to the CP- viola>ng effects. Choose the form factor a 1 = 1 and vary a 3. The form factor a 2 is set to 0 to simplify the analysis. Generator level Monte Carlo study. Monte Carlo: JHU at 14 TeV for the signal and MadGraph for the ZZ background. Pythia showering (AU2 CTEQ6L1). Smearing func>ons to simulate detector resolu>on effects. Trigger and reconstruc>on efficiencies are accounted for by assigning event weights. page 15
ATLAS- PHYS- PUB- 2012-004 CP- viola>on in Higgs sector Event selec>on in general matches the discovery analysis (H- >ZZ- >4l sec>on). Phys. Le}. B716 (2012) 1-29 Analysis: Applying two independent BDT s to separate spin and parity states and to reject the ZZ background. First BDT is trained to separate spin- CP states using angular and mass variables: cosϑ 1, cosϑ 2, φ, cosϑ *, φ 1. Second BDT is trained to separate Higgs signal from the ZZ background using kinema>c variables. The analysis is done in two bins of the second BDT. The measurement is divide in two channels with high and low S/B. Applied per FS. Stat test: profiled likelihood on the combina>on of 8 channels. page 16
ATLAS- PHYS- PUB- 2012-004 CP- viola>on in Higgs sector Expected separa>on in number of Gaussian σ between the pure 0 + hypothesis and the mixed hypothesis as a func>on of a 3. Signal region: 100 GeV to 150 GeV. The ZZ background is scaled to the total background expecta>on. Exclusion a 3 = 6+6i wrt 0+ Exclusion a 3 = 6i wrt 0+ Exclusion a 3 = 4+4i wrt 0+ 100 A - 1 3.0 2.4 2.2 200 A - 1 4.2 3.3 3.1 300 A - 1 5.2 4.1 3.8 The LHC is approved un>l 300 v - 1. Form factors much smaller than presented can be excluded with luminosi>es higher than planned for the LHC. Given the latest result for the pure spin/cp states, this result may be improved. Also the Matrix Element Likelihood fit might be more sensi>ve than this old version of the BDT- based analysis. page 17
ATLAS- PHYS- PUB- 2012-004 H- >μμ CP- odd amplitudes are naturally expected to be suppressed in the HVV coupling. The channels containing the HVV coupling in the final state are hence not ideal for the CP- studies. Naturally remain the channels where there is no HVV couplings in the final state. Studies were done for }H, H- >μμ and }A, A- >μμ. Too li}le sta>s>cs. About 1σ separa>on between pure CP- even and CP- odd states at 3000 v - 1. page 18
Summary First spin and parity results start appearing in LHC experiments. No decisive conclusion yet, but data start looking more like 0 +. Further studies (23.3 v - 1 + 5.6 v - 1 and beyond) should help us to: Excluding all popular alterna>ve hypotheses both in combina>ons of channels and in each channel alone. Understand the gg/qq produc>on mechanism for spin- 2. Start working with VBF, VH. CP- viola>on and tensor structure of the HVV vertex Exclusion of large CP- viola>ng form factors will require a lot of data (hundreds of v - 1 ). Most probably, with 23.3 v - 1 + 5.6 v - 1 it will be possible to set upper limit on the CP- viola>ng form factor. Precise measurement will require much bigger data samples. page 19
Backup page 20
Status of the Monte Carlo generators JHU: LO generator (2.2.6). Includes lepton interference in the Final State. The effect is less that 10%. Allows to model CP- mixtures with various strength of anomalous and CP- viola>ng couplings. The matrix element can be extracted for a par>cular model, making the LO MC re- weigh>ng possible: fast modeling of mixtures with various anomalous coupling strengths. ggf and qqbar produc>on mechanisms. MG5: LO. Currently implements 0+, 0-, 1+, 1-, 2+ which is slightly different from the 2 + m model proposed by JHU. Preliminary studies show that the distribu>ons of observables are consistent with JHU (ZZ- >4l) Implementa>on of the CP- mixtures is in progress. ggf and qqbar produc>on mechansms; VH, VBF, }H in progress. amc@nlo. Final states: gamma gamma, but not yet ZZ and WW. page 21
page 22
page 23