Trigger Emulator/Simulator Test for Trigger ID 137641

Abstract: We evaluate the L2 gamma trigger simulator in the StTriggerUtilities package.  To validate the trigger simulator, we compare the decisions made by the simulation software to the online trigger decision using an independently triggered event sample, i.e. events which satisfied either the EJP, BJP or dijet triggers.

1. Select events from fills 7863 - 7957
2. Select triggers 137273,137652,137551,137222, and 137622
3. No requirements on trigger ID 137641 (L2gamma in the EEMC)
4. Run the StTriggerSimuMaker
5. Run StEEmcPool/StEEmcPi0Maker
6. Produce invariant mass spectrum for all events, and events which satisfied both online and emulated L2gamma trigger (id=137641)

Figure 1 -- Invariant mass spectrum for all EJP+BJP+dijet triggers (solid line), subset of events which satisfied the online L2gamma trigger (red hatch) and subset of events which satisfied the emulated L2gamma trigger (blue hatch).  Red lines at 0.1 and 0.18 GeV indicate mass window for further event selection below.

7. The online and emulated triggers differ only in a few (<1%) events.
8. To examine which events are being dropped, we select events with a reconstructed neutral pion decay for further study.  We count the number of reconstructed pions which satisfied the emulated trigger and the number which satisfied the online trigger.  We plot these versus the transverse momentum of the neutral pion.
We also plot the ratio of the two samples versus the pT of the pion.

Figure 2 -- At left: (a) pT distribution for events which satisfied the online (red hatch) and emulated (blue hatch) L2 gamma triggers.  The red line at 5.2 indicates the L2 trigger threshold.  At right: (b) ratio of the number of emulated to the number of online L2gamma triggers, versus pT of the reconstructed pi0 candidate.  Note the supressed zero.

9.  Figure 2(b) shows that the online and emulated triggers reach the same result >98% of the time over all pT, with the largest deviation (~2%) occuring well below the nominal trigger threshold.
10. Finally, we look at the number of reconstructed pions versus run number. Single event anamolies are found in some runs, and are consistent with the prescale accept for the L2gamma trigger (which we know is not handled properly in the simulator). Again, plots in the file have red hatching for online trigger and blue for emulated trigger.

Conclusions:  We have examined the decisions made by the online L2 gamma trigger and the offline emulation of this trigger for an indepedntly triggered event sample.  The decisions reached by the emulator differ from the online decisions by <1%.  The largest discrepancies occur well below the nominal trigger threshold and are <2%.

NOTE: Code which generated above plots has been archived to HPSS at PDSF.  See ~jwebb/02-08-2008-emulator-test.tar on HPSS.

To Do

1. Add in trigger ID (id=6) from the rest of the ppLong2 run period
    - needed for A_LL(pi0) studies
    - needed by those doing prompt gamma cross sections
2. Add in trigger IDs from ppTrans
    - needed for A_N(pi0)
    - needed for those doing prompt gamma cross sections