ciao Daniele,
very interesting study...
a few comments:
* the resolution on deltaM is quite poor. This might be just normal, but
can you quantify the resolution (it is about 0.8 MeV usually, when
everything is reconstructed).
* do you understand why the signal has a peak (which is actually much
narrower than the background one)?
* from your numbers it does not look like a good reason to cut on it ...
smarter comments tomorrow after sleeping a bit more...
bye
ric
On Tue, 6 Aug 2002, Daniele del Re wrote:
>
>
> Hi all,
>
> these are the very preliminary results of the study on the cut on deltaM.
>
> DeltaM is defined as the difference between the X system mass and
> the mass obtained removing one pion with the right charge from the X
> system itself.
>
> DeltaM = M(X) - M(X - 1 pion)
>
> Since we can have many pions with the correct charge I select the
> combination with the minimum DeltaM.
>
> If the X system corresponds to a D* that decays in D0pi+ this DeltaM
> should should be compatible with DeltaM = M(D*) - M(D0) = 0.145.
>
> I used the "old" production (I mean the one with the overlap bug we
> started to run 10 days ago). The MC used is B0 cocktail and B+ cocktail
> for the background and Vubmix+breco for the signal.
>
>
> These are the results.
>
> B0 cocktail
> -----------
>
> In
>
> http://www.slac.stanford.edu/~daniele/vub/deltamstudy/b0cocktail.ps
>
> you find
>
> TOP LEFT: DeltaM after the lepton cut for b->clnu events (with ntrk > 1)
> TOP RIGHT: same plot with zoom
>
> BOTTOM LEFT: DeltaM after all analysis cuts for b->clnu events (with ntrk > 1)
> BOTTOM RIGHT: same plot with zoom
>
>
> In
>
> http://www.slac.stanford.edu/~daniele/vub/deltamstudy/b0cocktail_MD0.ps
>
> you see the M(X - 1 pion) distribution
>
> TOP: lepton cuts
> BOTTOM: all cuts
>
> the D0 peak is clearly visible
>
>
> Finally in
>
> http://www.slac.stanford.edu/~daniele/vub/deltamstudy/b0cocktail_mx.ps
>
> you can see the Mx distribution for b->clnu events after all cuts (black)
> and adding a cut on deltaM (.14<deltaM<.16).
>
> This cut is very good to reject Vcb events in the peak of the
> distribution but less in the left tail. This makes sense since the events
> at low Mx usually have either just one charged track or missed tracks.
>
> For the efficiencies see later.
>
>
>
> B+ cocktail
> -----------
>
> Same plots in
>
> http://www.slac.stanford.edu/~daniele/vub/deltamstudy/bpcocktail.ps
> http://www.slac.stanford.edu/~daniele/vub/deltamstudy/bpcocktail_MD0.ps
> http://www.slac.stanford.edu/~daniele/vub/deltamstudy/bpcocktail_mx.ps
>
>
> Signal
> ------
>
> Again same plots in
>
> http://www.slac.stanford.edu/~daniele/vub/deltamstudy/sigocktail.ps
> http://www.slac.stanford.edu/~daniele/vub/deltamstudy/sigocktail_MD0.ps
> http://www.slac.stanford.edu/~daniele/vub/deltamstudy/sigocktail_mx.ps
>
>
> Efficiencies (.14<deltaM<.16)
> ------------
>
> - B0 cocktail (b->clnu)
>
> Eps (no cut on MX) = 76%
> Eps (MX<1.6) ~90%
>
>
> - B+ cocktail (b->clnu)
>
> Eps (no cut on MX) = 97%
> Eps (MX<1.6) ~97%
>
>
> - Signal (b->ulnu)
>
> Eps (no cut on MX) = 98.4%
> Eps (MX<1.6) = 98.5%
>
>
> Conclusions:
>
> This cut seems to be useful for B0's even if it is less effective at low
> Mx.
> I will try to investigate the possibility to apply a looser cut on delta
> M (to include D** soft pions) and to study pions with the "wrong" charge.
> I will try to perform a similar study removing pi0's and photons (for Bch)
> even if it should be more challenging.
>
>
> Daniele
>
>
>
|