Cut Variation Tests


To test the stability of the cross section measurement to changes in the analysis cuts and, if necessary, assign a systematic uncertainty for cut variations.


At many points in the measurements I make cuts on the data.  For example, I place a maximum Zgg cut of 0.7 on all pion candidates.  These cuts are motivated by our understanding of the detector and underlying physics, but the specific location of each cut is somewhat arbitrary.  The cuts could move by some small amount and still be well-motivated.  The measurement should be relatively stable with respect to changing analysis cuts.  To test this, I take the three most important cuts, Zgg, inv. mass window, and z vertex, and vary them by some small amount in either direction.  I then repeat the analysis from start to finish to discern the effect (if any) these cut changes have on the final cross section.  The procedure is similar to that used to test the BEMC energy scale systematic (seen here)


Instead of showing the cross section measurement for each individual cut change, I will plot the results in what I hope is a more illuminating way.  Below is nine plots, each one representing a Pt bin in my analysis.  The Pt range in GeV/c is shown in the upper right hand corner.  Plotted on each canvas are the (Delta Sigma)/(Sigma) points for each of the cut variations (see key below.)  Also note the solid lines indicating the statistical error of the nominal cross section measurement in that bin.



point x position   Cut Change

2                         Invariant Mass Window of plus/minus 4sigma (nominal is 3sigma)

3                         Zgg - 10%

4                         Z vertex cut + 10%

7                         Invariant Mass Window of plus/minus 2sigma (nominal is 3sigma)

8                         Zgg cut - 10%

9                         Z vertex - 10%  


Broadly speaking, the points on the left side of the dashed-dotted line at 5.5 are cut changes that ought to increase raw yield and cuts on the right side ought to decrease raw yield.  Of course an increase (decrease) in raw yield does not always translate into an increase (decrease) in cross section because the raw yields are corrected.  Note that for most bins the effect for all cut changes is small (on the same order as the statistical uncertainty.)  Other systematics (BEMC energy scale and yield extraction) dominate the uncertainty.