[Rivet] histogram normalization for a CMS analysis

[Albert Knutsson]

Hi Andy,

thanks for the answer.

Then we just fill the integrals for data and MC, necessary for the same 
shape comparison as in the paper, in separate histograms in the plugin. 
I guess that is preferable, instead of some print-outs in the end of the 
run...

Cheers,
Albert



On 11/1/11 5:41 PM, Andy Buckley wrote:
> On 01/11/11 15:16, Albert Knutsson wrote:
>> Dear Rivet developers,
>>
>> we are about to finish a Rivet plugin for "Measurement of BB Angular
>> Correlations"
>> http://cdsweb.cern.ch/record/1329400/files/BPH-10-010-arxiv.pdf
>>
>> We are a bit unsure about how to do the normalization of the MC in the
>> Rivet plugin.
>>
>> In the publication only a shape comparison is made: The MC is normalized
>> to the data in a certain region, and a 47% common systematic
>> normalization error is not included on the data points.
>>
>> Would you prefer to have the MC in the rivet plugin normalized as in the
>> publication (for shape comparison)?
>>
>> or
>>
>> Would you prefer to have a MC cross-section prediction from the plug-in?
>> 47% uncert extra uncert then need to be added on the points in the data
>> aida-file.
>
> Hi Albert,
>
> Our "policy" is (at least for now) to output the MC normalizations 
> from Rivet analyses -- the user is then free to do "post-processing" 
> normalisation changes with the method of their choice. I think it's 
> probably good enough to say explicitly that there should be an extra 
> 47% uncertainty in the Description field in the .info file.
>
> As with everything else, this will probably be reviewed when we 
> finally get our improved histogramming interface into action!
>
> Hope that helps!
> Andy
>