After much consternation with job scheduling, I have finally followed up on Renee's request to look at towers above threshold and trigger rates. I have run over the full transverse dataset, specifying production P11id and trigger set-up name pp500_production_2011. A full run list (981 runs) can be found at

/star/u/drach09/Run11/Jets/fromKevin/runNumbers.list

I study, here, the number of towers above trigger threshold for BHT1, BHT2, and BHT3 triggers. For BHT1 and BHT2, I look at these in general and when the BHT1 and BHT2 triggers fired. For BHT3 I look at the N_{tow above thresh} distributions in general and when the BHT1 trigger fired. Additionally, I also look at the trigger rates. Here, I follow Pibero's procedure of considering N_trig*prescale/N_minbias*prescale. I look at the rates for BH1, BHT2, JP0, JP1, JP2*L2JetHigh, and AJP.

Figure 1: N_{tow above thresh} (All Triggers)

In Fig. 1 I plot the number of towers above trigger thresholds for all events in a run. There appear to be two rather high outliers: ID 469 and 620. These throw off the vertical scale, so I plot zoomed versions on the right-hand side. In the zoomed version there appears to be a high fill early on and perhaps another around ID 600. Additionally, there are outliers here and there as well as some runs which appear to be empty. This seems like a job for the run logs and the submission logs.

Figure 2: N_{tow above thresh} (Select Triggers)

In Fig. 2, I post the number of towers above threshold for select triggers. I look at towers above BHT1 threshold for BHT1 triggers, BHT2 threshold for BHT2 triggers, and BHT3 threshold for BHT1 triggers. This last selection is arbitrary, as I did not find a BHT3 trigger for these data. Run ID 469 seems to have come back to earth, while ID 620 remains high. The fill-dependent behavior also seems to have come back to earth, somewhat. Admittedly, I am a bit surprised that there are runs averaging below a tower per trigger. This doesn't seem right, however, perhaps the information is still useful.

Figure 3: Trigger Rates

In Fig. 3 I plot the trigger rate, defined as N_trig*scaler_trig/N_MB*scaler_MB. Oddly enough, Run ID 620 appears to be a low outlier, here, while a new run (ID 759) appears high. It looks possible that we have some high rate fills, again.

Next Step

I probably need to check the logs for these strange runs to make sure the behavior is real. Also, I should check the run logs to see if anything unusual happened during these runs. Then, it is probably time to produce jet trees for the well-behaved runs.