Opened 8 years ago
Last modified 7 years ago
#365 new defect
Incomplete variation weights for MEPS@NLO samples
| Reported by: | Frank Siegert | Owned by: | Enrico Bothmann |
|---|---|---|---|
| Priority: | major | Milestone: | rel-2.3.0 |
| Component: | Hadron Decays | Version: | 2.2.1 |
| Keywords: | Cc: | marek |
Description
The newly implemented variation weights for MEPS@NLO samples fail to take into account the shower weight, which can come from the B/Bbar in (2.14) of arXiv:1207.5030 as long as CSS_USE_BWW=1 (the default).
This has two consequences:
1.) The variation weights are incomplete, i.e. a trivial 1.,1. variation does not always reproduce the nominal weight.
Trivial workaround for users of 2.2.1: Include the 1.,1. variation in your runs and multiply all variation weights by the following ratio (e.g. in HepMC terms assuming the default PDF):
event->weights[0]/event->weights["MUR1_MUF1_PDF261000"]
2.) The variation of the B/Bbar term is missing in the variation weights and would thus not give perfect closure. This will make even less of a difference, but should be fixed for completeness in the future.
In both cases the difference is (fortunately) beyond the NLO accuracy of the method.
1.) is fixed on rel-2-2-2 and trunk with r29367, r29366, respectively.