B&Q: B hadrons for spectroscopy

A set of selectors for dominant B-hadron decay modes to be used in various X -> BY searches, where X may either decay in PV or have noticeable flight distance. Thus, B-hadrons have to be unbiased by PV requirements. Idea is to save the whole event (with persistreco=True) therefore selections need to be as tight as possible to bring rate down. In the end must be useful for both spectroscopy studies and calibration for flavour tagging.

List of decay modes currently considered (click to expand)

B+ modes (BRs):
- J/psi K+ (0.10%)
- D~0 pi+ (0.47%)
- D~0 3pi (0.56%)
- D- pi+ pi- (1.1%)
- D~0 mu+ (2.3+5.6=7.9%)

B0 modes (BRs):
- J/psi K- pi+ (0.12%)
- D- pi+ (0.25%)
- D~0 pi+ pi- (0.27% = 0.27*0.67{D*-} + 0.09{NR})
- D- 3pi (0.6%)
- D- mu+ (2.2%)

Bs modes (BRs):
- J/psi K+ K- (0.08%)
- Ds- pi+ (0.30%)
- Ds- 3pi (0.61%)
- D~0 K- pi+ (0.10%)
- Ds-(*) mu+ (2.4+5.3=7.7%)

Lb modes (BRs):
- J/psi p K (0.03%)
- Lc+ pi- (0.49%)
- Lc+ 3pi (0.76%)
- Lc+ mu- (6.2%)
- ?? Lc+ pi+pi- mu- (5.6%)

Bc modes:
- J/psi pi+ (X)
- J/psi 3pi (2.4*X)
- J/psi mu+ (20*X)

Xib-:
- J/psi Xi-(->Lambda pi-)
- J/psi p K- K- (not seen yet)
- Xic0 pi- (seen)
- Lc+ K- pi- (not seen yet)
- Xic0 mu- (not seen yet)

Xib0:
- J/psi X- pi+ (not seen yet)
- J/psi p K- K- pi+ (not seen yet)
- Xic+ pi- (not seen yet)
- Lc+ K- pi+ pi- (not seen yet)
- Xic+ mu- (not seen yet)

Omb-:
- J/psi Omega (seen)
- J/psi p+ K- K- K- pi+ (not seen yet)
- Omc0 pi- (seen)
- Xic+ K- pi- (seen)
- Lc+ K- K- pi+ pi- (not seen yet)
- Omc0 mu- (not seen yet)
- Xic+ K- mu- (not seen yet)

Tried to rely on selectors from B2OC and SL, however the existing selectors didn't allow to tune specific cuts (mainly tightening of cuts or releasing requirements on B direction). Therefore had to create local copies of tight selectors for Charm and B->Dh and B->Dmu. Later a convergence with Charm/B2OC/SL is possible once they allow to incorporate our tighter (and B PV/DIRA unbiased) selections. Resulting tight_ versions of Charm selectors can be a solution for #534.

In process had to fix BcToJpsiPi/PiPiPi lines to use detached_pions instead of prompt. Plus, suggested using cleaned version of BcToJpsiMu (for persistreco=True) with tighter requirements on PIDmu and M.

Estimation of expected rates considering:

• Lumi ~ 2x10^33 1/s 1/m^2 = 2x10^9 1/s 1/bn,
• x-section(pp->bb~) ~ 144 mub, x-section(pp->Bc) ~ 3.8x10^-3 * x-section(pp->bb~),
• typical BR (b->c+X, c->hadr) ~ 1% x 5% x 2 ~ 10^-3,
• typical eff ~ 1%

gives signal rate ~2.8 Hz for all b->C+hadr and ~10mHz for Bc. For semileptonic modes can expect rates 10x higher - i.e. ~30Hz. Real rates (which include background) may be ~10-100 times higher, i.e. O(20-300)Hz.

Estimation on 100k hlt1-filtered minbias events gives:

• 100-200Hz (tight version of cuts) / 100-400Hz (loose version of cuts) for fully reconstructed XForSpectrosopy lines, where X=Bu,Bd,Bs,Bc,Lb,Xibm,Xib0,Omegab;
• 10-50Hz for semi-leptonic XForSpectrosopy lines, where X=Bs,Bc,Lb,Xib,Omegab and 220Hz for Bud;

See difference between tight/loose versions of cuts in comments below. In short loose version of cut is expected to have 15-50% higher efficiency than tight version depending on particular channel.

Things no note:

• currently most of candidates come from lines with 3 pions - Bs2DsmPiPiPi, Lb2LcPiPiPi give >20 candidates per 100k events (55-65% of all candidates). Tightening of corresponding selections may be considered. Has checked that soft pions (with PT<100MeV if forming Sigma_c->Lcpi) increase candidate rate by less than 10%.