Inclusive or Leading?

This page compares a measurement of A_{LL} for the leading charged pion opposite a trigger jet with a measurement that counts all charged pions opposite that jet.

First, let’s look at the π+ and π- multiplicities in the away-side jet. I calculated the multiplicities by imposing all of the cuts that I usually employ in the inclusive A_{LL} analysis, except for the “pT < 10 GeV/c” cut.

Figure 1

So, I’m defining “leading” as the highest pT track (of either charge sign) that satisfies all my charged pion cuts (I’m open to alternative definitions). There could be a higher pT track in the event (or even in that jet) — if it doesn’t pass the DCA, nHits, PID cuts, etc. I ignore it. Here’s the fraction of away-side charged pions which meet this definition of “leading” as a function of pT:

Figure 2
leading fraction

Here are the pT spectra for the case when a π+ is leading and when a π- is leading, and the ratio of the two distributions:

Figure 3a/b
pT distributions charge ratio

I started looking into the leading pion measurement because of comments at July 24 Spin PWG meeting about “non-statistical” fluctuations in the SSAs vs. time. I realized that if I split the plot up into low pT and high pT distributions, all of the “non-statistical” stuff was in the low pT:

Figure 4a/b
low pT SSA high pT SSA

I thought that perhaps I was underestimating the size of the error bars in my measurement because I wasn’t taking the correlations between multiple pions from the same event into account, so I tried plotting the SSAs for the leading pions. Here’s a comparison of the same SSA for inclusive (left panel) and leading (right panel) away-side A_{L}:

Figure 5a/b
inclusive SSA leading SSA

You can also inspect the full SSAs versus fill for the inclusive and leading measurements, where you’ll see that the χ2 is better behaved for the leading measurements across the board. Integrated over pT the leading measurement has 2/3 of the stats of the inclusive measurement, but of course most of the loss comes in the lowest bin where we have plenty of stats to spare.

Update 2008-08-01

I tried redefining leading as follows:

  1. find the away-side jet in a dijet
  2. find the leading particle in that jet
  3. check if it passes all charged pion cuts

Here’s a plot of how often the leading particle in the away-side jet ends up identified as a charged pion:

Figure 6

It seems reasonable to me. However, if I look at the stats I see a significant drop from my previous definition of leading to this one:

Figure 7a/b
lead<em>spectra fraction</em>survive

There are at least three factors at work here:

  1. I’m rejecting events if the lead particle in the jet is neutral,
  2. or fails pion PID cuts.
  3. I’m requiring a reconstructed dijet event.

I don’t know which of these is primarily responsible for the drop in stats right now. I suspect it’s not (1), given the relatively small number of leading particles which are neutral in my “Fate” plot. Note that neutrals in out jet finder are towers, not tower clusters or reconstructed particles.