Results from the cone jet finder


1. Algorithm


This finder is based on the cone jet finder for the Tevatron Run II, as described by the Run II jet physics group in hep-ex/0005012. The current implementation of the algorithm uses each tower as seedpoint (seedless algorithm). The jet radius for the cone finder is 0.7. The found protojet cones are iterated until stable (number of max iterations 100) and are not required to be completly in the accepted phasespace. A splitting/merging step as described in the reference is performed. Please note that this step leads to asymmetric jets (towers are not necessary in the a cone with radius 0.7, the geometrical and the ET weighted position must not be the same). A detailed description of the cone jet finder can be found in section 5.1.

2. Results

2.1 PYTHIA events with 100 GeV/c hard scattering (status 11/13/02)

PYTHIA at 5.5 TeV was used to generate 100,000 events. Standard settings + CKIN(3,100), CKIN(4,100.1), CKIN(7,-1), CKIN(8,1), MSEL(1), MSTP(111,1), MSTP(151,0). This should produce only events with a hard parton scattering with pT~100GeV/c in the rapidity range -1<y<1.

The cone jet finder was run on each event. To get some rough estimate of the acceptance, only particles with -1 < eta < 1 where used. No further cuts on the particles were applied in this stage. Relativistic particles are assumed, i.e. ET=pT.

Figure 1 shows the number of found jets per event. The average number of found jets per event is around 6. Figures 2 and 3 show the position in eta-phi and the ET of the jets. The eta-phi distribution is basically flat and the Et distribution shows a paek at low ET (ET ->0 GeV) and a second peak around ET ~100 GeV (please note the logarithmis scale in this plot). Figures 4 and 5 show the position in eta-phi and the ET of the highest E T  jet in the event. The eta-phi distribution is again  basically flat. The Et distribution shows a clear peak at ET~100 GeV with a very small width. We've found our jets :-) The small peak at low ET comes probably from events where the scattered parton fragmented mainly outside the eta-acceptance.



Fig 1: Number of jets per event
Fig 2: eta-phi distribution of found jets


Fig 3: ET distribution of found jets


Fig 4: eta-hi distribution of found jet with highest ET


Fig 5: ET distribution of found jet with highest E T

The further analysis uses only jets which are the jet with the highest E T in the event and which have an ET in the interval 85 < ET < 110 (the colored area in figure 5). Figures 6 shows the ratio between ET from charged particles and total E T , figure 7 the ratio between ET from "electromagnetic" particles (electrons, positions, gammas). Is this the right definition of EM-particles??? Please note that there is an overlap between charged particles and EM particles with this definition. The mean ET (charged)/E T ratio is ~0.65, the mean ET(EM)/E T ratio ~0.25.

Figure 8 shows the ET distribution of the ET from charged particles only. The expected peak at ~60 GeV is clearly visible. This is what ALICE can approximatly see, when a perfect experiment (or one with hadron+EMCAL ;-)) sees the blue region from figure 5.


Fig 6: Ratio between ET(charged) and total ET


Fig 7: Ratio between ET(EM) and total ET


Fig 8: ET(charged) distribution

The number of charged particles in the selected jets is plotted in figure 9. The mean charged particle multiplicity in the analysed jets is 10, the distribution looks like expected. Figure 10 shows the pT spectrum of the charged particles in the selected jets. The combined information is shown in figure 11, which shows number the pT of the charged particles in a jet vs number of charged particles in the jet. Please note that all plots are not properly normalized, I still have to fix this...


Fig 9: Number of charged particles


Fig 10: pT spectrum of charged particles


Fig 11: pT of charged particles vs number of charged particles

A first look at the trigger: The simple trigger is defined as nParticles with a transverse momentum larger than PtThreshold . For the first look, nParticles is fixed to 4, the dependence on this cut will be studied later. The PtTreshold was 1, 2, 3, 4 and 5 GeV/c.

Figure 12 and 13 show the results of such a trigger. Figure 12 shows the number of total event (black) and the number of events which passed the trigger for increasing PtThreshold as function of ET(charged). Figure 13 shows the efficiency of the trigger for the different PtThresholds . The efficiency for high PtThresholds is clearly disappointing and therefore more studies are needed to tune the trigger algorithm (see below).



Fig 12: Number of jets triggered (black all, red pT>1, green >2 , blue >3 , yellow >4, purple >5 GeV/c)


Fig 13: Jet trigger efficiency


The two tuneable parameters of the algorithm are nParticles and PtThreshold. These two parameters are highly correlated. This is shown in Figure 14. The color shows the number of events which contain a jet with ET(charged) and have nPart charged particles in the jet. The upper row shows the plots for PtThreshold = 1, 2, 3 GeV/c, the lower row for 4 and 5 GeV/c. One clearly seed the expected behaviour of less and less charged particles per jet with increasing PtThreshold. This is can be more clearly seen in figure 15, which shows the number of charged particles above PtThreshold for all jets.


Fig 14: Distribution of events with a ET(charged) jet and nPart charged Particles > PtThreshold in this jet.


Fig 15: number of charged particles above PtThreshold (color codes see text)

The efficiency of the trigger algorithm for the different PtTreshold s and nParticles can be easliy calculated from figure 14. The total number of jets is given by the sum of all bins, the number of triggered jets by the number of bins above nParticles. The efficiency is defined as number of triggered jets divided by number of total jets. Figure 16 shows the results as a function of nPart for the PtThresholds from figure 15. (Black: PtThreshold 1 GeV/c, red 2 GeV/c, green 3 GeV/c, blue 4 GeV/c, yellow 5 GeV/c, please note the change of color code compared to figure 12 and 13). I would guess that this is the plot we wanted all the time. The question where to set the threshold depends on the rate on would expect from HIJING background events (see section 7.1.4 -- no results there yet :-( ).

Some first remarks on the interpretation: In a real enviroment, not all tracks will be reconstructed (track reconstruction efficiency). Therefore one would like to be as independed as possible from the number of tracks above threshold as possible to have a small systematic error. This is even more important for jets in AA, because we search for a modification of the fragmentation function. As one can see, this is only true for the lowest threshold of 1 GeV/c, which will probably cause some problems with the background rate. A possible solution is to look for number of tracks in a cone. As we have seen from the charged jetfinder results (section 7.1.3, figure 7), the pT in a radius 0.3 is ~0.8 of the total pT, indicating that the "high"-pT particles are still contained in the cone. We have to repeat the cone radius dependence analysis for this jet algorithm. This should results in the cone-radius dependence of figures 14,15,16 and the same for the HIJING background events.


Fig 16: trigger efficiency as function of number of charged particles and PtThreshold.


Conclusions:


Thorsten Kollegger
Constantin Loizides
IKF - University of Frankfurt
Last updated: 11/13/2002 18:55 pm EST