Meeting with LHCHiggsXS WG representatives - 15/07!!!!
* Shoulder/second peak explanation:
The second peak was seen by using many generators!
By using MC@NLO, few different options (corresponding to different propagators) were checked -> generally, the second peak appears if the BRs are not calculated at the generated mass!
The black curve - in spite of the still-present PDF enhancement at small masses, the net effect is that of a suppression, chiefly because of the presence of the partial width computed at the generated mass! Still there is some model dependence as the BRs are the SM ones.
When producing signal only (with no BG and interference) -> the presence of a second mass peak implies a choice of the propagator hard to justify
on physical grounds. Among other things, there is no other resonance in the computation, and hence no propagator which could give rise to
such a peak!
A second low-mass peak (when not performing a complete computation) is the signal of a strong dependence on arbitrary choices made at the level of the propagator, and therefore such a peak is NOT physical -> don’t use it as is!!!
“Solution” -> when producing a 2TeV Higgs with 200GeV width, around the 2TeV BW peak both the black and the pink curve (corresponding to right/wrong BRs)
behave the same -> so the model independence is recovered and therefore we should restrict the range to this area!
This also solves the problem of the interference effects that are important in the lower part of the mass spectrum (and would also be strongly model dependent).
Far away from the resonance, all models will behave differently and one shouldn’t look there at all when giving model independent results.
* Proposed plan:
1. Produce results for NWA (we have everything for that).
2. Produce “pseudo model independent” results ->
A. Take for example BW implementation of powheg with wide width (which will produce the shoulder)
BUT restrict yourself to the peak area only (~5%/10%/20%m[GeV] -> compare).
This will be done in few mass points with few width values -> m: 600GeV,1.5TeV,2TeV width: 5,10,20%.
Also, don’t try to look for “too” wide width -> choose reasonable width values based on the resolutions AND the theory inputs.
We will generate few width for few mass points, but we need to smear it for other mass/width values ->
B. We can generate CPS/BW with large width such that we have enough population in all the bins ->
Then suppress (by using the propagator) the events out of the peak with event by event weight
to obtain the proper shape of BW and keep the correct kinematics correct.
This will be done once for specific masses to ensure the propagator reweighting is working.
3. Produce results for full models like EWS, 2HDM etc (more tools become available and will be reviewed by Nikolas and Eleni for YR4).
1. Which width values to choose?
We try 5%,10%,20%.
2. Where to cut?
On the width that was used to generate -> based on Pamela check.
3. If we ignore the interference and cut for the signal around the peak ->
we might introduce bias when we use the control regions from data (for fits, fakes, corrections etc) which might include interference.
We need to make sure we make it clear that the results are conservative ONLY if the interference is positive!
1. Check with MC@NLO the comparison of with right/wrong BRs for the same mass with few different width values to see where to cut -> Pamela will check.
That’ll answer if we can cut on the width we used for the generation AND if it will make sense to go to 3TeV with 20%.
2. Check the 3TeV with 20% using powheg -> verify there are still events to reweight -> Pamela.
3. How large can the width be? Check with BW*PDF for the mass and see where to stop?! -> phenomenologist.