Hello Riveters,<br><br>I'm cleaning up the soon-to-be-committed Atlas ET flow Rivet routine and am trying to determine what the official Rivet policy is for normalisation. I'm sure this point has been raised before, but it's this: if the routine normalises plots itself and a user wishes to run parallel jobs, the resulting output when merged with hadd will not be correct. The only obvious way to correctly merge would be to ensure each selected the same number of events and take an average. I've
had a look at an official Atlas routine:<br>
<br><a href="https://rivet.hepforge.org/code/dev/a00287_source.html" target="_blank">https://rivet.hepforge.org/code/dev/a00287_source.html</a><br><br>And
it seems that things are self-contained, eg. the plots are normalised within the routine. My question is this - when it comes to official Rivet routines, is it standard policy to assume a "single job" environment, and leave it to the user to merge normalised histograms correctly?<br>
<br>Thanks for the help; if this question is misdirected, please point me to the appropriate forum.<br><br>Cheers,<br>Peter