|
[Rivet] [Rivet-svn] r2720 - trunk/src/ProjectionsAndy Buckley andy.buckley at ed.ac.ukTue Oct 19 13:42:06 BST 2010
On 19/10/10 11:59, Frank Siegert wrote: > 2010/10/19 Andy Buckley <andy.buckley at ed.ac.uk>: >> On 19/10/10 10:23, blackhole at projects.hepforge.org wrote: >>> Author: richardn Date: Tue Oct 19 10:23:23 2010 New Revision: 2720 >>> >>> Log: if more than one W candiate take the one nearest to mW rather >>> than throwing the event away. Brings Herwig++ NLO into agreement with >>> D0_2000_S4480767 rather than being a factor of two low. >> >> Interesting: I based the W finder on the Z finder, for which I thought >> that the rate of multiple candidates inside the mass window was >> extremely low. This suggests that ~half of the W events lead to multiple >> candidates: I'm surprised. And also the weight counting in that analysis >> can't have been spot-on if vetoing multi-candidate events was affecting >> the overall normalisation. Maybe worth changing ZFinder to have the same >> behaviour, i.e. choose closest-to-mZ. Frank? > > D0_2000_S4480767 is one of the analyses which doesn't have a real mass > window, but is corrected to the fully inclusive phase space. So this > problem shouldn't be as critical for Z's (which will always have to > have at least some kind of mass window), but I agree that it might be > nicer to do it correctly. The question for me is just what to define > as correctly... unfortunately the papers aren't very clear on their > procedure and up to now I always thought that the difference is not > noticeable. If it is as pronounced as the factor of 2 that Peter > found, we should probably be careful simply switching all of them to > the "nearest mass" procedure, shouldn't we? Good point ;) I was again relying on the vague memory that you had done some testing and the effect was very small. Probably because of the mas windowing. I think that the Z and W finders should do the same thing -- or possibly have a re-settable config flag to choose the strategy. And we should make sure that the final normalisation is not affected by this: without a mass window I'm not surprised that there are duplicates... probably *miles* away from the W mass! Any other W analyses without a mass window, or is this an isolated incident? Peter has already changed the WFinder default, as the commit message indicates. Andy -- Dr Andy Buckley SUPA Advanced Research Fellow Particle Physics Experiment Group, University of Edinburgh The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.
More information about the Rivet mailing list |