Adding the Jet Finder to the W algo

Run jet finder over all muDst's in st_W sample, and generate jet trees.

       -Jet Finder parameters: Same as 2006 jet analysis (except jet Pt min = 3.5) -> details at the bottom of the page

                                             Use Pibero's 100% tower energy correciton for tracks       

Note:  All the figures on the page were made using BTOW ideal gains.

Data

Figure 1: Jet Phi vs Eta for all jets in events where a track is matched to 2x2 cluster with ET > 15 GeV

Figure 2: Jet PT for all jets in events where a track is matched to 2x2 cluster with ET > 15 GeV

Possibility for cut on new ptBalance variable

    Definition: ptBalance = |lepton 2x2 cluster ET vector + (PT vector sum of all jets outside near cone)|.

Figure 3: PT Balance vs 2x2 Cluster ET  (filled before away side pt cut)

Definition: awayside PT = sum TPC + EMC PT in deltaPhi=0.7 window opposite in phi from the electron candidate.  

Figure 4:  ptBalance vs away side pT

Attached below is a pdf file of ptBalance vs awayside pt, sorted by the cluster ET bins used in figure 6 and 8.

How can we best optimize the cut on ptBalance and awayside pt?  Based on the plots above (as well as W and W+jet MC samples below) try double cut on

     ptBalance > 15 GeV

     awayside pt < 30 GeV

The results of this choice of cuts are shown in figure 5 and 6.

Figure 5:  Input plots for background subtracted yield.

Upper Left: lepton cluster spectrum after ptBalance > 15 and awayside pt < 30

Upper Right: lepton cluster spectrum after ptBalance > 15 and awayside pt < 30 (EEMC is not include in jet finder or calculation of awayside pt)

Lower Left: lepton cluster spectrum rejected by including EEMC

Lower Right: lepton cluster spectrum after inverting cut (ie. ptBalance < 15 and awayside pt > 30) 

Figure 6: Final Background subtracted yield for ptBalance > 15 and awayside pt < 30

The plot above is made the same way as was done for the plot Joe showed at DNP.  Subtract "2nd endcap" background from input W spectrum, then subtract normalized spectrum from the inverted cut (QCD backg est) to get final background subtracted W yield.  The QCD background is normalized to W candidates in the cluster ET range [15,19].  (Same plot normalized to cluster ET range [15,21]  here)

For comparison, I generated similar plots for the exact cuts used in the DNP preliminary result. These plots are different from what Joe showed because I used the SL09g production.

Figure 7: Input plots for background subtracted yield (DNP cuts)

Figure 8: Final Background subtracted yield (DNP cuts)

Can we improve the ptBalance variable by using a "signed" version?  The idea is to Determine to the magnitude and direction of the ptBalance vector relative to electron cluster.

Definition: sPtBalance = |(lepton cluster ET vector) * (ptBalance PT vector )| / |lepton cluster ET|

Figure 9: signed PT Balance vs awayside pT

W MC sample (40K input events)

Figure 10: PT Balance vs 2x2 Cluster ET (lepton candidate)

Figure 11:  ptBalance vs away side pT

Figure 12: signed PT Balance vs awayside pT

W+jet MC sample (16K input events)

Figure 13: PT Balance vs 2x2 Cluster ET (lepton candidate)

Figure 14:  ptBalance vs away side pT

Figure 15: signed PT Balance vs awayside pT

Parameters used to generate jet trees:

    //set the analysis cuts: (see StJetMaker/StppJetAnalyzer.h -> class StppAnaPars )
    StppAnaPars* anapars = new StppAnaPars();
    anapars->setFlagMin(0); //track->flag() > 0
    anapars->setNhits(12); //track->nHitsFit()>10
    anapars->setCutPtMin(0.2); //track->pt() > 0.2
    anapars->setAbsEtaMax(2.0); //abs(track->eta())<1.6
    anapars->setJetPtMin(3.5);
    anapars->setJetEtaMax(100.0);
    anapars->setJetEtaMin(0);
    anapars->setJetNmin(0);
   
    //Setup the cone finder (See StJetFinder/StConeJetFinder.h -> class StConePars)
    StConePars* cpars = new StConePars();
    cpars->setGridSpacing(105, -3.0, 3.0, 120, -pi, pi);  //include EEMC
    cpars->setConeRadius(0.7);
    cpars->setSeedEtMin(0.5);
    cpars->setAssocEtMin(0.1);
    cpars->setSplitFraction(0.5);
    cpars->setPerformMinimization(true);
    cpars->setAddMidpoints(true);
    cpars->setRequireStableMidpoints(true);
    cpars->setDoSplitMerge(true);
    cpars->setDebug(true);