[RooFit] Incorrect low-yield bins drawn when using blinding

Roofit and RooStats
#1

Hello all, I’m having a strange problem when attempting to do a blind fit of some data. I’m forcing a blind region in the middle of my mass window, and the bins adjacent to the blind region have strangely low-yields. I went to investigate and found (by doing a pull distribution of my fit) that the pull in these adjacent bins was quite normal, not even placing among the highest-pull bins in the mass window.

So the only conclusion that I can come to is that the script that I’m using isn’t drawing these bins correctly. After all, I know that there’s nothing wrong with the dataset, otherwise the pull values would be off as well. I think that this is a problem with the blinding procedure, or some strange RooFit artifact…

Any help understanding and fixing this problem would be very much appreciated! I’ve linked my macro and an example of an output graph created by it, showing the weird bins adjacent to the blind region.

Cheers!
output_graph.pdf (20.6 KB)
Blind_fit_2011.cc (10.8 KB)

#2

Hi,

It looks to me to be a binning problem. are you sure that the blind regions starts from the lower edge of a bin and ends to the upper edge ?
From the plot it seems to me that start from the bin center

Best Regards

Lorenzo

#3

Hello Lorenzo,
you are absolutely right, defining the bin number to a smaller value (92 instead of 100), the problem disappears. Thanks for the quick reply!

Cheers,
Julian