Home » Groups » STAR Protected

Dijet Filter Performance

Updated on Fri, 2010-04-16 09:23. Originally created by mattheww on 2010-03-23 10:24.

I created a dijet Filter using the StMCFilter framework. The filter classifies events into 3 categories: signal, could reconstruct as signal, background.

Signal events have a dijet with the following characteristics:

both jet |eta| < 1.3

delta phi > 2

max pt > 10.0

min pt > 7.0

dijet M > 10.0

 

Events classified as could reconstruct have a dijet with the same cuts, but the particle energies in the jets were increased by small amounts to make a fast simulation of possible detector effects.

Background events did not pass either group.

All events are passed, but the filter result is saved. The events are then passed to the TriggerFilter which accepts events that would have fired the JP1, AJP or BHT3 triggers. The gains were lowered so that events needed for energy scale studies would still pass.

 The following table has the results for each partonic pT bin. 50,000 events were thrown in starsim in each bin. Some jobs failed for whatever reason, and I did not try to recoup them. The second column lists the number of events reconstructed by BFC. Column 3 is the number of events that passed the PYTHIA filter out of 50,000 thrown. Columns 4, 5, and 6 show the number of events in certain groups out of the total in Column 2 that were reconstructed in BFC. Column 7 is for reference how many events are needed to simulate 1 pb^(-1) for that partonic pT bin.

Bin Events Reconstructed in BFC Events Passing PYTHIA Filter (out of 50,000) Events Passing PYTHIA Filter and Trigger Filter Events Failing PYTHIA Filter but had dijets after BFC Events Failing PYTHIA Filter but had dijets after BFC and passed the Trigger Events passing Trigger Filter Events Needed for 1pb^(-1)
2-3  28500 20  0 0 0 10  8.1*10^9
3-4  32997 125 0 0 0 64  1.3*10^9
4-5 32998 323 10 0 0 199  3.1*10^8
5-7  37986 1112 68 5 1 592  1.4*10^8
7-9  49916 5032 579 26 5 2712  2.3*10^7
9-11 48792 14992 2830 89 21 7039  5.5*10^6
11-15 48462 26496 8569 94 39 14647  2.2*10^6
15-25 48563 37118 23239 141 82 30263  3.9*10^5
25-1000 47714 47272 43665 63 54 45902  1.1*10^4

 

 

UPDATE

After a number of fixes were made to the Filter code in StRoot, I was able to run some longer tests with the filters on, allowing me to calculate the total accept %, the time needed to run the simulation, and the disk footprint. At a jet meeting where this was discussed, the proposal to do full filtering on 2-9 in partonic pt, only trigger filter in on 9-15, and no filtering above 15 GeV in partonic pT was made. For this proposal, the resulting disk space needed for 1 pb^(-1) is 170 GB (for MuDst and pythia.root files only) and 12.1 CPU years.

Groups: