This page details my studies on implimenting eta dependant thresholds into Matt's dijet filtering code. I have modified Matt's code to implement eta dependant thresholds on the Pt of the two jets. This allows me to keep Matt's thresholds for events in which both jets are in the barrel while using lower thresholds for events in which one or both jets go into the endcap.

Pythia filter thresholds:

• If one jet is in barrel (eta<1.0) and one jet is in endcap (eta>=1.0)
  • High pt jet > 5.0 and low pt jet > 4.0
• If both jets are in endcap
  • High pt jet > 4.0 and low pt jet > 3.0
• If both jets are in barrel
  • High pt jet > 10.0 and low pt jet > 7.0

BFC Trigger Filter:

This filter runs in the BFC and accepts events which fire either the JP1, AJP, or BHT3 triggers. The filter also prints out all of the bits from the L2 emc DSM (EM201). I can look at which bits fired to get an idea of whether the event was barrel-barrel, endcap-endcap, or barrel-endcap. The conditions I use are:

• BB: If ((BJP1 || BAJP1) && (!EJP1 && !EAJP1))
• EE: If ((!BJP1 && !BAJP1) && (EJP1 || EAJP1))
• BE: If (((BJP1 || BAJP1) && (EJP1 || EAJP1)) || (JP1 && !BJP1 && !EJP1))
  

The following table shows the performance of the filter. The first column gives the Pt bin. The second column shows the number of events processed, out of a maximum of 25000. The third column shows how many events passed the pythia filter using the conditions given above. The next three columns show how the events in the third column break down according to the pythia conditions defined at the top of the page. Note: There is a slight ambiguity in that the pytia filter can define two sets of dijets, one having its energy increased by a certain factor, and these two sets don't necessarily have to fall in the same categories. The seventh column gives the number of events which pass the pythia filter and the trigger filter. The next three columns show how the events in column seven break down into barrel-barrel, endcap-endcap, and barrel-endcap events using the criteria defined above in the BFC Trigger Filter section. Column eleven shows the number of events which failed the pythia trigger but had a dijet found after the big full chain. Column twelve gives the number of events which fail the pythia filter, pass the trigger filter and have a dijet found after the big full chain.

Table 1: Results of filtering.

Changed Pythia Thresholds

After looking at the numbers in the above table I decided to raise the BE pythia filter thresholds from 5 and 4 GeV to 6 and 5 GeV. I include the L0 thresholds used for reference. Note: these thresholds come from run 10179006 and all have been lowered by 1 adc count. GeV = (adc-5)*0.236

• BEMC-JP-th0 = 19 adc = 3.3 GeV
• BEMC-JP-th1 = 27 adc = 5.2 GeV
• BEMC-JP-th2 = 35 adc = 7.1 GeV
• EEMC-JP-th0 = 17 adc = 2.8 GeV
• EEMC-JP-th1 = 24 adc = 4.5 GeV
• EEMC-JP-th2 = 31 adc = 6.1 GeV

The following table is the same as the one above with the addition of two columns at the end. The thirteenth column gives the number of events which fail pythia, pass the trigger filter but had no dijet after the BFC. The last column contains events which pass the trigger filter regardless of other conditions.

Table 2: Filter performance using new thresholds

Table 3: Filter CPU usage and Disk footprint