Les Houches

0-lepton + MET + nJets SUSY search (3.2/fb). HEPData entry: https://www.hepdata.net/record/ins1458270

Has a rivet routine ! https://rivet.hepforge.org/analyses/ATLAS_2016_I1458270

On HEPdata, tables 4-10 contain the Meff distributions which contain the SRs, with info about the Data, SM Bkg (inc uncertainties split by stat and syst). This is perfect for CONTUR !

Here is one, but there are 7 SRs…

Unfortunately the rivet routine only quotes the integrated signal in those regions and is not linked to the HEPdata entry. This was fairly trivial to change by filling a Histo1D, and using the ref data from HEPData as ref.

The exclusion power of using this 3.2/fb search in CONTUR was tested using a Z'-mediated DM model.Under different treatements of MC, Data errors. (No correlations yet!)

CS=chi2(All previous assume background=data, background error=0), CSD=as CS but assuming background, error=data error: CSTD=chi2 using theory, where available, otherwise background and error taken from measurement.

And it gives a similar sensitivity to a Jet+MET measurement at the same luminosity:

Caveat: The actual search integrates over all bins in the SR and at the moment we are just picking the most significant bin. So the fact that it's less performant is not surprising. Two options: add some dummy histograms which integrate over SRs… or wait for correlation developments which David is working on!

Th next step is that we can update to the full 36/fb version which is on HEPdata here: https://www.hepdata.net/record/77891

but does not yet have a Rivet routine… this should be an easy copy over from the 3/fb version.

Has many more SRs: o(20)!

Andy, Louie, Tanya … can we write a rivet routine for this and use it in contur? Good model to try this on is the Third Family Hypercharge Model studied in 1904.10954 (UFO files)

The HEPData entry is here: https://www.hepdata.net/record/ins1725190?version=1

We want to reproduce the data from tables 1 and 2. Resolution functions are given in tables 23 and 24 as a function of truth Mll. These last tables basically give the values for the parameters of functions.

Then we'd need to figure out the stat correlations if we wanted to do a resonant search, Idea is to fill Bootstraps for o(1000) pseudo experiments and see how events migrate. For a given Mll, sample from the resulting resolution function.

Andy working on this.

Themore detailed interference study has its own page here.