 |
 |
Hi all, just to start the discussion, here is a proposal on how to fit simultaneously BRBR, mb and a. Please read it carefully: questions and feedback or alternative ideas are very welcome!!!! Currently we are performing a chisquare fit to our data distribution assuming the shapes for vub vcb and oth from MC. This results in something like: chi2 ~ sum_i (data_i - par[0]vub_i - par[1]vcb_i -par[2]oth_i)^2/sigma_i^2 where the constraint par[2] = 1 - par[0] - par[1] may or may not be applied. In the attempt of a simultaneous fit the above formula should be formally changed into: chi2 ~ sum_i (data_i - par[0]*vub_i(mb,a) - par[1]vcb_i -par[2]oth_i)^2/sigma_i^2 where vub_i(mb,a) contains the bin-by-bin dependence of mX on mb and a (par[3] and par[4] in the fit)... The problem is to determine an analytical form for vub(mb,a) in each bin. Unfortunately, there is no functional dependence between mX or q2 - either the true or the reconstructed ones -, and Mb / a. So we propose to extract an empirical vub(mb,a) by looking at bins of mX (or q2) in different MC samples generated with several values of mb and a. We can obtain vub(mb,a) with e.g. a polinomial fit to such distributions in each bin. Then we can implement vub(mb,a) directly in the chi2 fit and minimize the chisquare simultaneously for the background shapes normalizations, mb and a. The systematics coming from vub(mb,a) PDF determination [polin. fit] needs to be estimated as soon as possible to see if that way of extract it is feasible and actually reduces the impact of theoretical error. Comments, questions and suggestions about this proposal are very very welcomed. Concezio, Virginia and Alessio