11/03/2003 Andrei Nomerotski  

this is an update on the previous study - we figured out that the 6 pb^-1 used in that study were biased by the muon skimming. below
we use 2 pb^-1 of unbiased data.

Can we use even looser muons than now?

We use the following cuts on muon candidates - they are referred below as "usual cuts"
1) nseg > 1
2) chi^2 local >-0.5 (i.e. converged fit exists)
3) CFT and SMT hits
4) central rank = 1 (this is obsolete in p14 and is not used in the below)
5) lepton Pt>2 GeV, Ptotal >3 GeV

after Sergey's D* loose selections here is what we have in unbiased 2 pb^-1 with usual cuts. 
The plot is the mass difference between D* and D0 candidates.

 

 same as usual cuts but nseg < 2 (those are MTC aka calorimetric muons nseg=0 and Layer A muons nseg=1).
no Pt or P cuts applied since nseg<2 is normally somewhat softer stuff. 

 

this is what we have without any muon requirement

 

this is what we have for tight muons as defined by muon ID group

 
 
the number of signal event for all above plot is summarized in the table below

selection	entries	signal
tight muons	1009	213
BBN usual	1113	223
nseg=0||1, any Pt or P	1682	84
any muon (no cuts at all)	3107	342

i.e. potential gain in the B->l D*X yield from adding looser muons is  (342-223)/223 = 53% at expense of smaller S/B and bigger skimmed sample.

another thing we checked is if there is a double counting in the sample. in the tight muon + D* (selected as mDst-mkpi < 0.15)sample there are 7 pairs with the same muon and 12 pairs with the same D*.  The pairs with the same muon is just normal combinatorial bkg - this will be accounted for by the bkg subtraction. However the pairs with the same D* are a double counting of the signal. total # of entries  was 307 so the double counting is present at the level of 12/307 = 3.9%.