%$ h \rightarrow A^0A0 \rightarrow b\overline{b} \tau^+\tau-$% in VBF at LHC (Les Houches Project)

Hypothesis

We assume SM-like weak vector boson fusion production of %$h$% with a xsec of order 2 to 4 pb depending on the mass. NLO corrections are available but have been shown to be small (typically of order 5 to 10% at most). We assume that %$Br(h\rightarrow A^{0 A}0)=1$%. It is a simplifying hypothesis, often justified from the theoretical point of view (%$A^{0$% is much heavier than light fermions). Anyway, if it's not true, all signal xsec should simply be scaled by this Br.}

We also assume that %$A^{0$% is a Higgs like scalar coupling to the mass, so that}

Numerically, this gives %$Br(A^{0\rightarrow \tau}+\tau^{-)=8\%$% and %$Br(A}0\rightarrow b\overline{b})=92\%$% if one takes %$m_b=3.5$% Gev which is a reasonable value at the scale %$m_{A^{0}$%. These BR could be modified if the coupling of %$A}0$% to charm is sizable, but this is not the case with SM like couplings or in large %$\tan(\beta)$% type II models like MSSM.

The total BR for the decay chain %$ h \rightarrow A^0A0 \rightarrow b\overline{b} \tau^+\tau-$% is thus 2 * 8% * 92%=15%, and the typical xsec is thus ranging from 300 to 450 fb.

Benchmark point

For the LH project, we take %$m_{A^{0}$% at 50 Gev and %$m_{h}$% at 120 Gev. The mass of the %$h$% is light enough to have a good production xsec and to avoid to consider its decays into WW or ZZ. The mass of the %$A}0$% is light enough to open the decay %$h\rightarrow A^{0 A}0$% but heavy enough to give a good pT and a good separation for b's and tau's.

Monte Carlo production

Methodology

Signal and backgrounds are generated separately using MG/ME v4.1. The PDF is CTEQ6L1 and the renormalization and factorization scales are set to 120 Gev. All the following generation cuts are applied. Jet means non b jets and leptons means e, %$\mu$% and %$\tau$%.

• Minimal pT of 20 Gev for the jets and 10 Gev for leptons and b's
• Maximum rapidity of 5 for jets and 2.5 for leptons and b's
• Minimal deltaR>0.3 for all objects
• Maximal invariant mass of 80 Gev for the tau's pair (to avoid the Z peak)
• Minimal invariant mass of 700 Gev for the two jets (VBF cut)
• Delta eta >4 for the two jets (VBF cut)

Signal

100k events with tau's decayed into leptons only: ​Events

The cross section after tau decay into leptons is 9.5fb. There is no cut on the tau decay products. Plots before tau decay are ​here.

If you need to split LHEF files, please use Software.SplitEvents

Considered backgrounds

Irreducible: QCD %$2\tau 2b 2j$% background

44k events with tau's decayed into leptons only ​Events

The cross section after tau decay into leptons is 1fb. There is no cut on the tau decay products. Plots before tau decay are ​here.

Nearly irreducible QCD %$2e 2b 2j$% and %$2mu 2b 2j$% background

Irreducible if no tau tag and no missing Et cut. Cross section is 8.7fb (for each) with cuts on leptons same as cuts on tau's for signal.

A 25k events muon sample: ​Events

Another 26k events muon sample ​Events

I propose to use one of these two muon samples as an electron sample (just need to switch the PDG code in the LHEF file since the kinematic is the same).

Some ​Plots for the whole 50k sample.

Reducible tt + 2j with t>Wb and both W going to e, mu or tau and tau going to e or mu

A 100k event sample with W's in e or mu (xsec=2300fb): ​Events

A 50k event sample with W+ in tau's (and tau+ in e or mu) and W- in e or mu (xsec=404fb): ​Events

A 50k event sample with W- in tau's (and tau- in e or mu) and W+ in e or mu (xsec=404fb): ​Events

A 50k event sample with W- and W+ in tau's (and tau's in e or mu) (xsec=70fb): ​Events, another one ​Events

Without any cut on leptonic final state, the total xsec is 3200 fb! But the kinematic is very different for both leptons pairs and b pairs, large invariant masses and large deltaR are typically preferred. Also, there is a lot (half of it if I'm right) of mismatched pairs e-mu (instead of e-e or mu-mu) which should be easy to remove. The amount of Et is also typically much higher (neutrinos are coming from W's), so a max Et cut should also be rather efficient. So there is maybe a hope...

Not considered backgrounds

QED (i.e. alpha_EW^{4) backgrounds}

Should be small (see ​this paper) and anyway impossible to simulate completely.

QCD %$2b 4j$% background for jets faking tau's

Not considered here since tau's are decaying into leptons.

2ta4j without b's (mistagging)

Probably not necessary. CMS TDR I gives a b mistagging probability around %$10^{{-2}$% for gluons and light quark jets (for a tagging efficiency of 0.5) and %$10}{-1}$% for c jets. The cross sections of 2ta4j and 2ta2c2j with the strong VBF cuts are probably to small to give sizable contribution after double mistag.

Summary of the situation…

A 10fb signal with a 1fb irreducible bkg, two 9fb nearly irreducible background which could be efficiently cut by a mass window on leptons and a delta R cut on b's. A huge 3.2 pb tt background with very different kinematic but same final state...

References

A good generic reference for VBF is the corresponding section in ​Djouadi. It gives different references for experimental analysis of %$h\rightarrow \tau^+\tau-$% in VBF which could be considered as a good starting point (but ATLAS related!).

Les Houches 02 proceeding:

​Azuleos

ATLAS notes

​phys-2002-018 ​phys-2002-004

Parton level Papers:

• PRD59 014037​

• PRD61 093005​

PLB454 297

CMS

To be checked: Physics TDR and Notes

Why is it interesting ?

Theoretical point of view: if %$m_{A^{0}>2 m_b$%, most of NMSSM signal with only tau's and mu's decays are suppressed.}

Pheno point of view: signal xsec similar to (or even better than) SM one for VBF with h in tau's which has been shown to be a quite interesting channel (4.5 sigmas in each tau's leptonic mode, 6 sigmas for the combined analysis after 30 fb-1). Jet veto cannot be applied straightforward but a "non b jet" veto is an interesting new concept (check if it is new!). The additional double b tag could maybe compensate the less effective background rejection without decreasing too much the signal (25%).

Final Version of our paper

• pdf​: Our contribution to Les Houches Proceedings