 |
 |
Ciao Daniele, > The blinding is applied, the fit to three shapes is on, the reweighting > for the B and D decays is not applied. The central value is going to > change since the correct trk and neutral smearing on MC (fix from Urs) has > to be applied and the files used to train the fit will be updated in order > to take into account the correct fraction B0/Bcha. > For sure ,these numbers are fine to look at the stability of the fit. > > > GOOD NEWS: this plot looks encouraging. The BRBR variation is of the order > +-10%, a factor two less than the theoretical uncertainty due to the Mx > cut. From this plot we could extract systematics to take into account > possible violation of the duality in that Mx region. Remember that with > the three shapes fit we are not cutting on MX but all the events below > MxCUT are put together in a single bin and the full Mx region is used to > extract the ratio of BR's. Since we are not actually cutting, we could > start thinking about extracting the value of BRBR from this whole set of > fits. > For instance > > central value = BRBR > spread = syst on BRBR due to duality violation. I am not completely sure I understand what you mean. The fact that you are fitting for the shapes means that you are testing that the shape on data is in agreement with the shape in the MC, which I agree is a very good news. On the contrary the uncorrelated errors among these points are very tough to estime because the events used are the same. I would say that there is no statistical component, but there is only the theoretical error. Therefore I think this plot makes us confident that the theoretical error is correct, but I would not over interpret it. > > LITTLE BAD NEWS: there is a jump (at 1.58GeV) around our present cut on > Mx. I will try to figure out if this is due to a fit instability. > once more, it is good to check that there is no failed fit, but otherwise we should only decide which value to pick up according to the ooptimization procedure (which does NOT make use of this plot) ad get whatever we get (since it is within errors the central value does not matter) ciao ric