Hi Antonio,
this is already a good outline. Please find my comments below:
Page 1: I would change the title
What about 'Studies on MES fits' ?
Page 4: I don't understand what you mean with
* 'peaking BBbar bckground with the same shape as signal'
* 'B events reconstructed in different decay modes then are
actually decaying into'
Page 5: * 'crystall ball' --> 'crystal ball'
* You show a fit for signal only. You should specify what
MC here exactly means, that is, what the procedure is to
get signal-only events. (truth associator)
Page 7: * It is important to explain which parameters are free in the
MC fit
* The same holds for the fit on data
* If possible please provide the chi^2 for the fit on data
Page 8: * Besides the fit instability you should also say that the
statistical error on the signal yield is significantly
larger when having three PDF components in the M_ES fit
resulting in a rather large statistical error on Vub
Page 9: * Although we write the correction factor now as you are
doing it my personal understanding of it is rather:
Fix the S/P from MC,enriched and correct for possible
data-MC differences using a control sample where we have
sufficiently large statistics:
(S/P)_data,depleted/(S/P)_MC,depleted
But this is a matter of taste.
Page 10: * 'Enpoint' --> 'Endpoint'
* You should give a reason why we keep the endpoint fixed
to the same value for peaking and comb. BG
* Cut out 'parameters' in the line 'For high statistics ...'
to save on line
* 'survirve' --> 'survive'
* Add P+ and the last line and indicate that we have only studied
so far the other variables
Page 11: * It is not clear what the different plots are standing for.
Please provide information on MX bins.
* You say here that there is no strong q^2 dependence in these
plots which is not really the case.
If one takes the double ratio (Page 12) and S/P for data
depleted (Page 13) then one can claim that there is no
strong q^2 dependence.
* BTW: There is this funny behaviour in the second bin where
S/P on the MC depleted sample does not change at all
over 6 q^2 bins...
Page 13: * Do we have also the full correction factor for the mX-q^2
studies as a function of q^2 for the four mX bins?
Page 15: * In the table: what is meant by computed, respectively, fitted?
* It would be nice to see a fit without correction factor
in order to see the effect
Page 17: * 'orter' --> 'order'
* It is not clear which chi^2 is meant. Why is the Ndof for the
second order polynomial 7 and for the others it is 8?
What is missing is our conclusion to be discussed in today's meeting.
In this context, we need to compare the systematic errors when fixing
S/P with the statistical errors we have in BAD1383.
* Concerning studies on integrated purity cuts:
You can add a transparency saying that we have studied if the correction
factors as a function of MX change their functional dependence with
higher purity cut. From what we have seen so far this is not the case.
In particular, it does not become flat. hence, we don't gain from such
a purity cut. It would be good if we had plots to demonstrate this.
* It would also be good if you prepared a slide discussing a bit the
several re-weighting factors
* I could prepare a slide showing what Wolfgang has done so far
concerning the fallback solution and we had this to your talk.
Cheers,
Heiko
On Tue, 12 Sep 2006, Antonio Petrella wrote:
> Hi all,
>
> here is the first version. I need your feedback (I don't know if there
> is useless stuff) and also need to understand what to add on integrated
> purity studies.
>
> http://www.slac.stanford.edu/~petrella/antonio.pdf
>
> Thank you
> Antonio
>
