Weight normalisation for unweighted events

[Frank Siegert]

Our unweighted events have a relatively unpredictable weight normalisation which is somehow related to the sum of the cross section (or maximum?) of all processes. This is of course not wrong, since for a given sample it doesn't matter whether I scale all weights, but I would suggest to change that at least in the HepMC output for two reasons:

1. I don't want to be asked about it by users anymore ("Are these events really unweighted?").
2. ATLAS had some technical issues with large weights (>1e9) (don't ask!) which would be avoided that way.

So it would help if unweighted events are normalised such that they have typically weight=1 with the following natural exceptions/modifications of course:

• MC@NLO events have +-1
• MENLOPS has local kfactor on higher jet multis
• Enhance_Factor's will appear as 1/enh

I hope it should be simple to divide out the current normalisation or at least make it accessible in the SP blob such that I can divide it out in the HepMC or ATLAS interface?

[Stefan Hoeche]

There is now a double "Weight_Norm" attached to the signal blob.