It does not appear the software min-bias trigger is actually filtering out events. Until an explanation is found, I have sorted the BBC-coincidence states by referring to log files from running over the μDST's. Additionally, I read out the bunch-crossing ID's from the μDST's, which allows me to document those events with discrepant bunch-crossings.

Figure 1

Of the 7472 events passing both the hardware and software eemc-http trigger, only 32 do not involve a BBC-coincidence. This is significantly fewer than I had expected, however, this may serve as an indication of how efficient the trigger and cuts are at removing non-collision background. However, it also brings to light an unexpected problem. Only 40 of the ID-0 events actually turn out to be unfilled-bunch events, leaving 217 as misidentified filled-bunch events. This means 3% of events have a discrepant bunch-crossing. It, thus, becomes important to verify this does not translate to a misidentified spin-state.

Figure 2

The fact that a large majority of the ID-0 excess is due to misidentified filled bunches explains why ID-0 appeared to have a typical rate of di-photons. Figure 2 shows that rather few unfilled-bunch events return di-photons. Figure 2 also demonstrates that only 1 non-coincidence event returns a di-photon, making this study quite difficult.

Figure 3

The lone non-coincidence event occurs at lower η in the top hemisphere of the EEMC above the signal mass window.

Figure 4

It is difficult to glean qualitative information about abort-gap events given the dearth of statistics.

Figure 5

Unfortunately, it may be rather difficult to learn much from this study, given the small number of statistics.