D* Partial Reconstruction on the Recoil

Motivation:

Present B0 ->D*+lv measurement is limited by systematics. I consider the pros of a measurement based on partial reconstruction on the recoil of the B0 Semi-Exclusive reconstructed sample.

Partial reconstruction means tag only lepton and soft pion; use correlation between soft pion and D* to tag events. This works in unbiased events:




~ 800 000 signal events tagged in 80 fb-1 (neat efficiency*BR ~ 0.55 % ).

S/B ~ 1/1 . Background is 30% BB combinatorics, 15% continuum, 5% peaking B+

a) Why partial reco ?

    1. Statistics: ~ factor 10 compared to full reco. This might be crucial on the recoil (where about 100 000 B0B0 are tagged). With the efficiency quoted above, would obtain ~ 600 events , with S/N ~ 5-6. Cuts however can be relaxed (Daniele quotes 3000 evts). I assume 1000 and a statistical error of ~ +-3%

    2. Systematic: in the recoil, the main issue of partial reco (background) is significantly reduced. By not reconstructing the D0, many benefits are obtained.




Recoil + Partial reco can help in reducing the dominant sources of syst. Error.

  1. Normalisation (f00). At present 2.7% from PDG. Using last BaBar measurement,R+0 = 1.11+-0.037 +0.035 gives f00 = 0.473+-0.012, i.e. +-2.5% relative error on the normalisation, not much better than the error now quoted (+- 2.7% from PDG). The measurement in the recoil would depend on the knowledge of the B0 sample composition (ARGUS function for background, peaking B+ ). The precision should be at least comparable and however independent, therefore decreasing the overall error when averaged.

  2. D0 final states: two source of errors

    1. Knowledge of BR (mostly D0 -> Kp ) ~ 2.0 % not likely to decrease in the future. This is completely removed with partial reconstruction

    2. Tracking+PID: depending on the final state, 2 to 4 more particles in addition to the lepton and the soft pion. Can be ~ halved with partial reconstruction (only lepton and soft pion tracking, no K PID).

  3. D** . Usually determined from cos(BY) (or equivalently Mv2), i.e. from as full as possible kinematic reconstruction of the decay. Pros and cons.

Cons:

    1. D* is not fully reconstructed; must parameterise it using soft pion 3-momentum (usual trick in partial reconstruction, however significan loss of resolution)

Pros

    1. B0 direction is known in the recoil : this should entirely compensates for the problem above

    2. In untagged sample~2/3rd of D** come from B+ (Isospin). In B0 tagged sample, this background is reduced to ~ 1/3th .

D** model+fit error should therefore reduce by ~ 50 %



Conclusions

Accounting for all these effects, the total “internal” systematic error should decrease from 4.9 to 3.6 %, the “external” systematic error from 3.5 to 0.7 % , to this last the uncorrelated error due to the knowledge of the compostion of the B0 semiexclusive sample must be added. In addition a ~ 3% statistical error should be considered. All included, the use of a partially reconstructed sample on the recoil would allow to reduce the total error by ~ sqrt(2) on the Branching ratio.

Is it worth considering a partially reconstructed sample from minimum bias events ?