TPC Track Finding Efficiency for W cross section
 

Data Sample: Run 9 pp500 SL09g production of L2W

MC Sample:   rcf10016 (pythia QCD w/ partonic pT > 35 GeV)

Figure 1: Last hit on track for Data after simple track cuts (nFitPts, nFitFrac, Rin, Rout, and pt > 5) 

Figure 2: Last hit on track for MC after simple track cuts (nFitPts, nFitFrac, Rin, Rout, and pt > 5)

We need to compensate for the inefficiencies in the TPC tracking seen in the data that aren't modeled in the MC simulation.  This should be done in an eta and phi dependent way.

Figure 3: Ratio of tracks in MC sample to Data in each eta-phi bin

The hot spots at eta=0 is the relatively large inefficiency in the data probably due to the deconvoluted clusters not accounted for in the MC.

Note: The bins in figure 3 are 0.22 wide in eta in the range [-1.1,1.1] and 15 degrees wide (1/2 a TPC sector) in phi.

To correct the MC to more accurately represent the inefficiencies we see in the TPC for the data I used the following procedure:

1. Correct for different efficiency in phi for differend sectors (only applied for eta slices with |eta| > 0.44)

• Make slices in eta from figure 3, and normalize each sector to the 'good sectors' in that eta bin.  The good sectors have the maximal yield, and so the assumption is the track reconstruction efficiency in the data and MC match reasonably well in this eta-phi bin.

• Here are some of the eta slice plots and the red line shows the ratio in the 'good sectors' mentioned above.  Anything above the red line is an inefficiency in the data that is not accounted for in the MC.
• These corrections are then applied to the data to match the efficiency in phi for each eta bin.  The result is seen below in figure 4

Figure 4: Last hit on track for Data after correcting |eta|>0.44 bins. 

          2.    Correct for inefficiency near eta=0 in the data

• Make slices in phi from figure 4 for each pair of sectors.
• Normalize inner 4 eta bins (eta=[-.44,.44]) to the average of the 3rd (eta=[-.66,-.44]) and 8th (eta=[.44,.66])eta bins.  This assume distribution of tracks near mid rapidity is ~flat in eta.  The red line in these phi slice plots is the average of the 3rd and 8th eta bins mentioned above, and deficits from this line are inefficiencies in the data that are not accounted for in the MC.

With the corrections from 1) and 2) above we now have an eta-phi "inefficiency map" that can be applied to the MC to approximately match the data.  This was applied by statistically removing tracks from the MC based on the inefficiency in the eta-phi bin each track fell in.  The resulting ratio of QCD MC tracks to data tracks is found below.

Figure 5: Ratio of tracks in MC sample to Data in each eta-phi bin after all inefficiency corrections

Comparing fig 3) and 5) the balance of MC and data tracks in each eta-phi bin is much better than before.  I put fig 3) and fig 5) on the same z scale so it is clear how the inefficiency corrections have smoothed out this ratio of MC tracks to data tracks.  There appear to still be some inefficiencies not accounted for near mid rapidity, but it is better than it was.

Note:  No attempt was made to correct sectors 4 (eta>0, phi=-0.5) and 11 (eta>0, phi=2) since the timestamps used in generating the MC already introduced a sizable inefficiency due to dead padrows.  It appears this already made these sectors less efficient in the MC than the data.

Conclusion

The result of applying this eta and phi dependent track removal on the MC samples is a reduction of the reconstruction efficiency by ~15%.