Indico & Vidyo Page

Joey

Hi. I wanted to share with the conveners some ideas for studies to be done before, at and after Les Houches that will have an impact not only on SM physics but on BSM as well. For that reason, and for the reason that we would like to have more continuity between the 1st and 2nd sessions, I am sending this to the complete list of conveners. The list of ideas below is incomplete, so other ideas are welcome.

A key study that can be carried out at this year's Les Houches is a detailed investigation of where fixed order predictions can be considered reliable, where resummation/parton shower effects are important, and how the implementations of the resummation/parton shower effects may differ. This is a source of great confusion among both experimentalists and theorists I believe. There has not been to date (as far as I know) quantitative comparisons of fixed order predictions to ME+PS predictions (with varied levels of merging). In Powheg inclusive jet production, the addition of parton showering seems to have a noticeable impact in almost all kinematic regions. This has influenced at least the opinions of the experimental community. If this is true, then we must understand why resummation effects are important for regions which don't (naively) have large logs. I don't know to what extent that the impact of parton showers can be varied within Powheg. Sherpa gives some understanding of resummation effects by allowing a variation of the scale mu_Q; if resummation effects are not important, then the variation of an observable with mu_Q should be minimal.

At Les Houches, we should be able to test Sherpa, Madgraph5_aMC@NLO and Powheg. Each should be run without(/with) any non-perturbative effects, since we are trying to understand perturbative effects. Sherpa and Madgraph5_aMC@NLO can be run with merged samples of 1 or 2 additional jets at NLO. We can examine the impacts of varying the renormalization, factorization and (where possible) the resummation scales as a function of how much merging has been done.

One testbed for studying these effects are for the case of Higgs(+jets), where gosam has provided fixed order ntuples for 1,2 and 3 jets at NLO. In addition, we should have the complete Higgs+>=1 jet NNLO prediction. Frank Petriello will be at Les Houches during the first week. We should make sure that we have NNLO predictions that can be directly compared (where relevant) to the framework we are setting up at NLO. We can consider similar distributions as were used for the ATLAS Higgs+jets paper (and for which a Rivet routine exists, and which I will distribute). Take for example, Higgs+>=3 jets at NLO. Compare the pT distributions for jet 1, 2 and 3 to the ME+PS predictions with merging information from 0, 1 and 2 jets at NLO (0 being the inclusive Higgs cross section). Where are there deviations for ME+PS predictions from the fixed order predictions? What are the relative resummation scale dependences? Is jet 3 (or 2) affected differently than jet 1? If PS effects are important, does the size of the effects agree within the different frameworks? If not, why not? We're aiming for precision physics measurements, so that level of precision must be present in the theoretical predictions.

Can we also use the large cross section W/Z+jets for comparison to better understand these questions?

Another testbed is inclusive jet production. There, it appears that in Sherpa there are no major changes from fixed order predictions, but no quantitative comparisons have been carried out yet. The resummation scale uncertainty, though, seems relatively large. This should be understood. Are we just lucky (if fixed order = MEPS@NLO) that the central resummation scale gives good agreement? Does the resummation scale uncertainty change as more matrix element information is included? Inclusive jet production is very important to global PDF fits; so far we are assuming that fixed order (+non-pert effects) is all that is needed to describe the data.

A Sherpa study (probably there are others as well, but I don't know them off-hand) has indicated that fixed order NLO does not do a good job of describing dijet cross sections where different cuts are placed on the leading and next-to-leading jet. Of course, some effects are expected at threshold, but the difference between NLO and NLO+PS exists out to fairly high jet pT values. On the other hand, NLO closely approximates NLO+PS for inclusive jet measurements. (This agrees with the original studies performed by both CDF and QCD theorists in the 90's that indicated that NLO(+non-perturbative corrections) did a reasonable job of describing the jet shape for inclusive jets.) For boson+jet(s) (W/Z/Higgs), almost always the jets are ranked-ordered, so that we look at the pT of the lead jet, the next-to-leading jet, and so on. Studies have also indicated that PS corrections to fixed order predictions are limited in this case, perhaps because of the less steeply falling pT spectra. This has to be quantitatively understood.

A related question(s) is in regards to vetoed cross sections. If I look at the cross section for Higgs+>=1 jet, where the jet pT cut is 30 GeV/c, do I have to treat this as a vetoed cross section? What are the size of the effects? If this is an exclusive cross section, then there aren't very many inclusive cross sections left. How can we test this? If some of the pi^2 terms are resummed, a la Tackmann, does this really represent the NNLO correction?

Another related question is with regards to scale dependence. A (universally used) central scale of HT/2 seems to work reasonably well, but perhaps does not have a rigorous understanding of why it should be the central scale (and why a variation of a factor of 2 around it is also reasonable). Here we may be able to use CKKW/MINLO tools to shed some light on this issue.

We should have a good start on these comparisons before Les Houches, so that the discussion at Les Houches can be informed. We also need Rivet routines for the data analyses that we would like to test these theory variations on.

Sasha I would add couple of more related issues might be important for the Higgs production generation:

1. we know from the last Les Houches 2013 how big was a difference between

 generator predictions of gg->h+2jets passing VBF h selections. The
 proposal they made was to use in data the gluon initiated process and
 check data with the different MC. I have discussed it a bit with
 Frixione and looks like tt+jj with VBF selected jets may be a good
 benchmark for Monte-Carlos.

2. Z+b(b) production important benchmark for SUSY H+b(b). Both ATLAS and

 CMS 8 TeV analyses see the pTZ a bit harder than LO/NLO+PS predictions
 see slide 22 of this talk for example:

http://indico.cern.ch/event/331452/session/4/contribution/25/material/slides/1.pdf

 We see with MG5_aMC@NLO that Higgs pT in H+b(b) is sensitive to the
 shower scale. So tuning it with Z+b(b) data may help to descrive properly
 Higgs pT in H+b(b).
 

Josh CMS has produced/is producing large sets of events for ttbar+0,1,2j NLO with mg_aMC@NLO+pythia8 (FXFX merging) at both 8 and 13TeV. Since that is quite CPU intensive we can provide the lhe files if we have some common space somewhere where I can put them.