In testing whether the simulated trigger is working like expected, Renee suggested the following plots:

1. tower Et for data with hardware trigger applied
2. tower Et for data with L0+L2 software trigger applied
3. tower Et for data with L0+L2 software trigger + your L2 mockup with data threshold
4. histogram of tower id's that passed 2) but did not pass 3)
5. tower Et for filtered MC
6. tower Et for filtered MC with L0+L2 software trigger
7. tower Et for filtered MC with L0+L2 software trigger + you L2 mockup with MC thresholds

Note: in writing the data EEMC trees, a minimal filter was chosen rather than cutting on any specific triggers.  Instead all the triggers are stored and the trigger cut can be applied later.  The "minimal filter" was chosen to be the L2-Mockup trigger with thresholds 2.0 and 4.0.  For MC data, the filter was not applied.  So plots (1)-(4) include this minimal L2-Mockup trigger cut.  The following plots suggest that the values are sufficently low, though (if requested by the Spin group) the threshold values for the "data filter" can be reduced further.  It takes on the order of 1 cpu-day to remake the data trees.

Note: the plots, except (4), are the Et of all towers over all events, with the only cut being excluding Et == 0 (i.e., towers in an event with adc less than 3 sigma above pedestal).

The data sets are 1 run of data and 50 runs of the lowest (3-6 GeV) partonic pT bin Pythia.

Plots (1)-(3)

The distribution includes two contributions: the part closely related to the trigger, and a low energy portion, due to lower energy towers from other particles (or leakage) in the event.  Adding the simulated trigger cut over the hardware cut (and "filter" of 2.0, 4.0) makes almost no perceptable change, while changing the mockup from (2, 4) GeV to (3.7, 5.2) GeV makes a significant impace, with a much sharper edge on the lower part of the upper distribution (the one closely related to the trigger).

Plot (4) was looked at, but there tower indices span the whole range, with no obvious patterns.

Plots (5)-(7)

With no cut, the distribution is monotonically decreasing.  The basic shapes of the other two (red and blue) lines are the same as in data.

Data vs. Pythia

The following plot shows the blue lines from above data and Pythia plots shown on the same plot, with Pythia normalized to the data in the range of 3-10 GeV.

The shapes are very similar, but there is a relatively small upward shift in the Monte Carlo distribtion.  Part of this discrepancy could be the use of only one partonic pT bin.  Another way to visualize this is by taking the ratio of the two lines in the above plot, data over Pythia.

There is some statistically significant deviations around 3-4 GeV, in the trigger turn-on region, but again this could be due to only using the lowest partonic pT bin.  Certainly, the discrepancy at ET > 8 GeV is due to this fact.

Conclusions

1. The simulated trigger is not doing anything "crazy", but generally gets the right shape.
2. There are small differences in the positioning of the turn-on region.
3. There is no need to use the L2 mockup trigger, other than as a filter in making the EEMC trees
   1. Should apply the same cut to MC while we're at it, or 1/0.97 times the cut, but shouldn't be making much difference.

Consequences of item (3) include:

• The values of the cross section and asymetries will change very little by using the lower thresholds
• The data/MC kinematical comparison will appear worse
  • Must be due to some other factor besides the trigger
  • Could be missing material in the simulation--very difficult to decifer.
  • Background subtracton (and thus asymetries) account for differences in data/MC shapes in systematic related to the residual
  • Unfolding matrix, reconstruction efficiency and trigger efficiency (i.e. the cross section) may be affected
    • Very difficult to quantify the effect further than previous studies--which indicated a negligible contribution to the systematic uncertainty
    • For preliminary, can hedge that "further systematics are being considered" and then think more about what to do