[Rivet] setSumOfWeights

[Silvan Kuttimalai]

Ok, then we will implement it in this way. The only caveat I can see is 
that the cross section error cannot be set externally. It's probably not 
huge issue but the error on the cross section shown in the output of 
"MC_XS" would be off, for example. Would you be able to add such a function?

Best,
Silvan

On 03/15/2018 01:16 PM, Andy Buckley wrote:
> That seems simpler to me! No obvious issues: this way I think it's 
> guaranteed that the sum(w) and sum(w**2) terms will both be accurate, 
> and they should combine together consistently when you merge.
>
> Cheers,
> Andy
>
> *Dr Andy Buckley, Lecturer / Royal Society University Research Fellow*
> Particle Physics Experiment Group, University of Glasgow
>
> 	
>
> On Mar 15 2018, at 6:49 pm, Silvan Kuttimalai 
> <silvan at slac.stanford.edu> wrote:
>
>
>     Hi Andy,
>
>     thanks for the quick response! If I understand the issue
>     correctly, then
>     we might just as well re-set the total cross sections for our
>     sub-contributions and leave the sum of weights untouched. To me this
>     even seems like to conceptually better way of doing it. I just ran a
>     test and it appears to work for us. Would you see any issues in this
>     approach?
>
>     Best regards,
>     Silvan
>
>     On 03/15/2018 03:55 AM, Andy Buckley wrote:
>
>         Hi Silvan,
>
>         The issue is that more than the sum of weights alone is needed
>         now.
>         Can you also supply the sum of squared weights? If so, I'll add an
>         extra argument and an implementation to version 2.6.1.
>
>         Andy
>
>         *Dr Andy Buckley, Lecturer / Royal Society University Research
>         Fellow*
>         Particle Physics Experiment Group, University of Glasgow
>
>
>
>         On Mar 14 2018, at 11:45 pm, Silvan Kuttimalai
>         <silvan at slac.stanford.edu> wrote:
>
>
>         Dear Rivet Authors,
>
>         we noted that in Rivet version 2.6.0 the method
>         AnalysisHandler::setSumOfWeights is not present anymore. We
>         used this
>         method in Sherpa in order to automatically split up our rivet
>         analysis
>         output by contributions from H and S events in MC at NLO for
>         example. To
>         get the normalization of these sub contributions right, we set
>         the sum
>         of weights explicitly at the end of the run. Is there an
>         alternative for
>         this procedure in the new version?
>
>         Many thanks,
>         Silvan
>         _______________________________________________
>         Rivet mailing list
>         Rivet at projects.hepforge.org
>         https://www.hepforge.org/lists/listinfo/rivet
>