Hui-Chi’s talk:
Jonathon comments that this confirms that the way CMS and ATLAS derive their uncertainties is very similar
Frank makes the point that having some defined source of an uncertainty, such as the migration around a certain boundary, is much more well-defined and therefore more likely to be able to attract e.g. improved calculations from theorists in future.
Frank also asks a technical question about considering >=1 jet bins for the pTH uncertainties on S11. Ans: this has been tried and is being considered, possibility as (or similar to) method 2 as described on the slide.
Subsequent discussion on the synchronisation:
ATLAS proposes a single nuisance for migrations across the pT(Hjj) boundary, whilst CMS currently has one type A nuisance in the mjj < 350 region and two type B nuisances in the mjj > 350 region, one for each of 350-700 and >700. It is pointed out by Haider that these latter two already have very similar values, and Frank says clearly the underlying source is the same for both and also the type A nuisance. Jonathon & Ed agree, and it would be simple for CMS to modify this to correlate the existing three sources. It seems to be agreed that given we expect the main source of type A uncertainty in these bins to arise from the pT(Hjj) distribution, this CMS type A nuisance can be treated as an effective pT(Hjj) boundary migration nuisance and as such the correlation both within CMS and also across experiments makes sense. It should be noted that the CMS and ATLAS methods and magnitudes of the uncertainties here seem to be compatible.
For the mjj boundary nuisances, ATLAS proposes one for each boundary so four in total (two solid, two dashed). CMS currently has type B nuisances corresponding to the two solid boundaries, but then just one type A uncertainty for the > 700 region. There was no clear consensus on what should be done here. The existing CMS nuisance could simply be correlated with whichever of the ATLAS nuisances is seen to have the greater impact. Alternatively, CMS could split its existing nuisance into two (with different nuisances affecting different analysis categories), and thereby produce two nuisances that could be appropriately correlated with ATLAS’ sources. This would probably need some follow-up discussion on the details.