MC samples:
/tas/vimartin/SingleLepton2011/output/V00-04-13/ttsl_smallTree.root
/tas/vimartin/SingleLepton2011/output/V00-04-13/ttdl_smallTree.root
Hadronic top reconstruction with NO CUTS ( just to plot )
For the sigma(pt) neded in the W mass constraint fit the function getJetMetSys (in
NtupleMacros /CORE/jetSelections.cc) has been copied TO THE SELECTION CODE and MODIFIED (return result*0.01 instead of 1+flag*result*0.01). Added also MC matching.
Distribution of three jet combinations:
The following cuts have been selected:
M_jj | M_jjj | tag | ttsl efficiency | ttdl efficiency | significance S/sqrt(S+B) |
50 - 120 | 130 - 200 | > 0.12 | 0.954 | 0.867 | 1.59 |
60 - 110 | 140 - 190 | > 0.14 | 0.908 | 0.755 | 1.59 |
70 - 110 | 150 - 190 | > 0.16 | 0.830 | 0.627 | 1.55 |
Significance is estimated from the numbers in here:
https://indico.cern.ch/getFile.py/access?contribId=4&resId=0&materialId=slides&confId=174826, considering ttsl_efficiency the same in signal, reducing ttdl bkg by ttdl_efficiency and everything else fixed.
For the hadronic top reconstruction I have included the MBL variable
(the mass of the lepton plus the other b jet), and added a cut on the b tag as follows:
I remove candidates where the W jets are more "b tagged" than the other b jets,
this worked a little better than requiring the b jets to be b tagged.
Then I apply the following cuts on this variables:
Pt - Minimum Pt of the jets
M_jj - Invariant mass of the W jets
M_jjj - Invariant mass of the Top candidate
drW -
DeltaR between Wjets
dRt -
DeltaR between the reconstructed top and l+jet
Pt M_jj M_jjj dRW dRt mlb eff tt->l jets (%) eff tt->ll (%)
10 50 - 120 140 - 200 2.5 1.5 150 75 52
20 50 - 120 140 - 200 2.5 1.5 150 68 42
30 50 - 120 140 - 200 2.5 1.5 150 54 27
The "efficiency" is the number of events that satisfy all cuts divided by the events that
satisfy the preliminary cuts ( 4 jets, 1 b tag , isolation and so on).
I am still optimizing the cuts, but not much will be gained. I think this is the best
we can do without including more variables (suggestions?) or making a more sophisticated
discriminator.
--
RicardoVillalba - 26 Sep 2012