The discussion focused on assessing and controlling the theoretical 
uncertainties in the prediction of the tt+jets background (with both light 
flavour, LF, and heavy flavour, HF, jets), and followed the series of
questions raised in the talk given by Aurelio Juste at  the last meeting 
(see https://indico.cern.ch/conferenceDisplay.py?confId=189783).


1) How to assess the systematic error when using tools such as 
   ALPGEN to model the background?
  
   Would it be possible to learn anything by comparing LO+PS to
   existing parton level NLO calculations?

   Can this help in having a more comprehensive list of theoretical 
   and experimental errors in each control region?

  For instance, the question was raised of how can we treat the
  regime of small b-bbar angles (in ttbb production)?
  Can this be better studied having the NLO calculation available?

These questions are difficult to answer on the basis of parton level 
NLO calculations of tt+jets, which still unavoidably have a very limited
final state multiplicity and do not include the decay of the final state
top quarks.

The necessity of having a more accurate/realistic description of the
final state particle multiplicity is also dictated by the fact that
the experimental analyses are predominantly based on the events' kinematic.

The goal would be to have the NLO calculation for both tt+jj and
tt+HF interfaced with a PS Monte Carlo. 
The experiments timeframe woudl ideally have this as soon as possible,
but no later than six months from now.
Z. Trocsanyi and collaborators are working at interfacing the NLO calculation 
of tt+bb with PS  using the POWHEG method. They expect results by the end 
of Summer 2012. This would represent a crucial improvement and could
time in well with the experimental needs.
More groups may be working in similar directions, but for the moment the
interface of tt+jj with PS seems harder and not expected on the same timescale.


2) In the meanwhile, how can the existing NLO parton level calculations 
    be used in the analysis to improve the background modeling? 

The study of the NLO cross sections for tt+jets (LF and HF jets) gives 
important indications on how to reduce some systematic theoretical 
uncertanties. 
For instance, in the case of tt+bb it shows how using a dynamical scale
enormously reduces the residual uncertainty from scale dependence,
and it is therefore highly recommendable even for packages based on
LO+ME (e.g. ALPGEN).


3)  Are the experimental analyses going to consider both a standard
     and a boosted-Higgs regimes?

For 7 and 8 TeV the boosted-Higgs regime is not viable, there is not
enough statistics. So, fo the moment only the standard regime will be considered.


4) Aurelio Juste raised the point that it would be interesting to have
   a general study of the systematic error(s) implicit in all NLO PS
   Monte Carlo, in a process indepedent way.