This page details an apparant problem with the 100% subtraction scheme as implemented in the 2009 200GeV jet trees.

At the most basic level, jets at STAR are created by combining the four-momenta of tracks and towers in some cone radius. When a track points at a fired tower, a certain amount of the tracks momentum is subtracted from the energy in the tower to avoid double counting energy. In previous years, a MIPs worth of energy was subtracted from a tower whenever a track pointed to it. For 2009 however, the full momentum of the track was subtracted from the energy of the tower. Studies comparing the MIP subtraction scheme to the 100% subtraction scheme can be found here.

As part of my investigations into reconstructing jets in the endcap, I developed a system to match jets from the 12 point jet tree branch to the emc only branch (brief description and plots here). Once a jet from one branch was matched to the same jet in the other branch, I compared various quantities in one branch against those in the other. One of the comparisons I looked at was the Pt difference between the tower only component of the 12 point jet and the emc only branch jet. That quantity can be seen in fig 1 ploted vs the detector eta of the 12 point jet.

Figure 1: Pt of towers in 12 point jet - Pt of emc only jet Vs detector eta of 12 point jet.

When I showed the above plot at a local meeting, the question was raised as to why there was such a sharp locus at zero. We are looking at matched jets, so the towers in the 12 point jet should be roughly the same towers as in the emc only jet. One would expect the Pt of the towers in the 12 point branch to be systematically lower than the Pt of the emc only towers because the some of the 12 point towers will have tracks pointing to them and thus have the track momentum subtracted from the tower energy whereas the towers in the emc branch will have no such subtraction because there are no tracks. When I looked at this in detail, I found that in most cases the pt difference was less than 10^-7. In fact, from looking at a couple dozen jets, it appears that the only time the Pt difference deviates from zero is when the towers included in the jets did not exactly match up between branches.

I have linked a text file showing the output for a couple of dozen events. For each event, the 12 point tower / emc only jet Pt difference is shown along with the number of tracks and towers in each jet and the number of towers not shared between the jets in different branches. Then each track, 12 point tower, and emc only tower is listed along with tower / detector id (or id of hit tower in case of track), Pt, energy, eta, ADC, and pedestal. As can be seen, the tower Pts for towers in the 12 point branch and emc branch are identical, even if the 12 point tower has a track pointing to it.

As a further check, I worked out the energy for a tower without a track pointing to it and for a tower with a track pointing to it using the ADC and pedestal information printed out from the jet finder. I got the gain by looking in the DB browsing interface page located on the software & computing page. The barrel gains can be found in the calibration table for day 12-15-2008 00:00:05, direct link here.

Event ID: 183501

Tower ID: 3507  ADC = 47  Ped = 22.66  Gain = 0.0157254  (ADC-Ped)*Gain = 0.382756

The energy I calculated by hand is exactly the same as the energy reported from the jet finder. This tower did not have a track pointing to it.

Tower ID: 3547  ADC = 70  Ped = 23.16  Gain = 0.0148844  (ADC-Ped)*Gain = 0.697185

Again, the energy I calculated by hand is the same as the energy reported from the jet finder. This time however, there is a track pointing to the tower. The track has a Pt of 2.18168 and an eta of -0.220835 which means it has a momentum (P) of 2.235.

Pibero mentioned that the ADC should be the 'raw' tower ADC, that is before any track momentum subtraction is applied. If this is the case, the above example would seem to indicate that the 100% subtraction scheme is not being applied to the 12 point branch.

Problem Solved:

Pibero has found a subtle bug in the jet maker code which was preventing the subtraction scheme from being changed from its default state, which was no subtraction. A brief explanation can be found at this thread. I have done a quick check on a run produced after the fix and it appears that the subtraction scheme is now working as advertized. Compare this text file with the one above.