Hi Kerstin,
in the 2D fit we varied the mx binning like this (MXBIN is the cut in Mx):
0,MXBIN,2.2,2.8,5 (default)
0,MXBIN,2.2,2.5,5 (move down 4th bin)
Variable bins: define
x_1 = ((5. - MXBIN)/3.) + MXBIN;
x_2 = ((5. - MXBIN)/3.*2.) + MXBIN;
binning is: {0,MXBIN,x_1,x_2,5.};
The associated (absolute) error on BRBR is less than 0.002.
Virginia and Concezio.
Kerstin Tackmann wrote:
>Hi Concezio and Virginia,
>
>thanks. Actually it is most interesting to us how you varied the mx
>binning in the 1D fits since this is what we need to do as well. Could you
>point to to what you did there?
>
>Thanks,
>Kerstin
>
>
>
>On Tue, 4 May 2004, Concezio Bozzi wrote:
>
>
>
>>Hi Kerstin,
>>we don't know how the binning was varied in BAD540 for systematic studies.
>>In our 2D fits, we are using 4x4 bins in (mx,q2) such that the first mx
>>bin goes from 0 to the mx cut, and the last q2 bin goes from the q2 cut
>>to 26. We have investigated some different possibilities for the q2
>>binning. Our default is to divide the q2 range between 0 and the q2 cut
>>in 3 equal-size bins, but this gives unstable results when the q2 cut is
>>lower than, say, 5. We therefore used constant q2 bins evenly
>>distributed (0,2,4,6,8,10,12,14,26), with more stable results at low q2
>>cuts. For higher q2 cuts, the two sets of q2 binnings give almost the
>>same results.
>>We also varied the mx binning with very small variations of the results.
>>Cheers, Concezio and Virginia.
>>
>>Kerstin Tackmann wrote:
>>
>>
>>
>>>Hi all,
>>>
>>>we have a question concerning the uncertainty from binning effects on the
>>>mX spectrum fits. In BAD 540 you say that you varied the bin sizes in a
>>>wide range and that you increased the number of bins. Is there some more
>>>explixit information on the actual binnings you used for the estimation of
>>>this uncertainty?
>>>Virginia and Concezio, are you going to use the values as they were used
>>>for BAD 540?
>>>
>>>Thanks,
>>>Kerstin
>>>
>>>
>>>
>>>
>>
>>
>>
>>
>>
>
>
>
|