 |
 |
Hi all, I implemented a new option in the fit that performs a fit based on a three components fit: b->ulnu, b->clnu and other on the full Mx spectrum (this was actually the first implementation I proposed many months ago). All the three shapes are fixed from the MC and they have a big bin 0-1.6 GeV. After the fit you can do the subtraction of the background as well in order to know the structure of the signal Mx distribution. This method should be actually identical to the one with the background subtraction but in this case you don't care about the assumed BR(b->ulnu) to get the right b->ulnu contribution for Mx>1.6GeV since it is automatically included according to the fitted ratio of BR's. Then this method should be more robust and you don't have to perform the fit in many iteration to get the right tail at high Mx. This method is also useful in the fit with multiplicity categories. In this case the subtraction should introduce a very large systematics since you don't know the ratio of BR in each multiplicity category and you can make a large error in the fraction of signal events you are assuming for Mx>1.6GeV. First tests show that the value obtained with this option is pretty compatible with the one using the subtraction. The error is slightly lower (probably the additional information in the first bin makes the error smaller). You can see an example of the output on 300fb-1 (cocktail) in http://www.slac.stanford.edu/~daniele/vub/MCshapefitresults_nocat.eps This implementation could replace the old one. We should discuss this in the next meeting. Daniele