# Run 9 200GeV Dijet Kinematics

Updated on Thu, 2014-12-11 18:11. Originally created by pagebs on 2014-12-05 03:26.

Look at the initial kinematics probed in the dijet measurements ...

I want to look at the x1 and x2 values probed by the dijet asymmetry measurement. I look at the detector level simulation, and apply all dijet selection criteria and cuts as well as requiring that the detector level dijet match to a parton level dijet. For each event which passes the above criteria, I record x1 and x2, where the association between which parton corresponds to x1 and which to x2 is defined as:

For events with both jets in the east barrel: x1 is the parton with negative z momentum and x2 is the parton with positive z momentum

For events with both jets in the west barrel: x1 is the parton with positive z momentum and x2 is the parton with negative z momentum

For events with one jet in the east and the other in the west barrel: x1 is the parton with positive z momentum and x2 is the parton with negative z momentum

There appears to be two ways to get the partonic x values from the simulation which are slightly different for reasons I do not understand. I refer to these two methods as the particle method and the tree method:

Particle Method: Here I calculate x by looking at the z momentum of particles 5 and 6 (those involved in the hard scattering) from the PYTHIA record. I then calculate x by dividing that momentum by the z momentum of its parent proton (particles 1 and 2).

Tree Method: The skim trees contain a class called StPythiaEvent where various pieces of information from the pythia record are stored including the partonic A_LL information and the TParticle array for all particles. It also contains explicit values for x1 and x2 which are closely correlated, but not identical to the x values obtained from the particle method (see figures 1 and 2).

Figure 1: Scatter plots showing the relationship between the x values obtained from the Tree Method (y-axis) and the x values obtained from the Particle Method. The 8 panels each show a different detector level dijet mass bin. Both x1 and x2 are shown on the same plot.

Figure 2: The top 8 panels show the x1 distributions for the EEWW topology for the Particle (red) and Tree (blue) methods of getting x for the 8 detector level mass bins. The bottom 8 panels show the corresponding x2 distributions. The plots for the EW topology can be found here (x1 x2).

As can be seen, the differences between the Particle and Tree methods seem to be rather minor, with the largest discrepencies showing up in the low x region of the x2 EEWW distributions. I have not been able to trace back the code to determing where the Tree x1 and x2 values are filled from, so (for now at least) I will use the Particle Method from here on out as I know exactly where those numbers are coming from.

By using the Particle Method, I can also make cuts on the parton type when plotting x. For the comparisons below, I will show x values obtained from all partons and the x values obtained from only gluons.

Figure 3: The top 8 panels show the x1 distributions for the EEWW topology for all partons (red) and gluons only (blue) for the 8 detector level mass bins. The bottom 8 panels show the corresponding x2 distributions. The plots for the EW topology can be found here (x1 x2).

Figure 4: Comparison of the x1 (red) and x2 (blue) distributions for the EEWW (left hand side) and EW (right hand side) topologies for two mass bins (16-19 GeV bin is solid curves and 58-82 GeV bin is dashed curves). The top panels are for all partons and the bottom panels are for gluons only. A pdf which compares the 16-19 GeV bin with all other mass bins can be found here.

Figure 5: The average x1 (red) and x2 (blue) values as a function of the invariant mass of the detector level dijet. Left panels are for the EEWW topology and the right panels are for the EW topology. The top panels are for all partons and the bottom panels are for gluons only.

Carl wanted to see the x_Gluon distributions plotted on a linear scale. Figure 6 shows the same information shown in the bottom panel of Figure 4 in a linear scale and with the high mass histograms scaled up by a factor of 500.

Figure 6: Comparison of the x1 (red) and x2 (blue) distributions for the EEWW (left hand side) and EW (right hand side) topologies for two mass bins (16-19 GeV bin is solid curves and 58-82 GeV bin is dashed curves). The 58-82 GeV histograms have been scaled up by a factor of 500.

I wanted to look a little deeper at the difference between the Tree method (Renee confirmed that the x1 and x2 values stored in StPythiaEvent are actually PARI 33 and 34 from Pythia) and the Particle method. To do this, I generated 4 small (10 event) standalone Pythia samples with ISR and kT turned on and off:

ISR = On, kT = On

ISR = On, kT = Off

ISR = Off, kT = On

ISR = Off, kT = Off

Some quick observations: Turning off the ISR makes lines 3 & 5 and lines 4 & 6 identical as well as making PARI(31)=PARI(33) and PARI(32)=PARI(34). The Pythia manual says this should be true (Page 64), so it is nice that we see it. Even with the ISR off, the x calculated from the particle z momenta does not equal that stored in PARI(31-34) unless the primordial k_T is turned off as well. For the events with the ISR on but the k_T turned off, we see that the x calculated from the z momentum of particles 3 and 4 (before ISR is applied) is identical to the x values stored in PARI(31&32) which are the x values of the initial state parton shower initiators. However, once the ISR is applied and gives the system some transverse kick (particles 5 and 6), we see that the x values calculated from particle z and the x values in PARI(33&34) no longer match exactly.

If I am interpreting this correctly, the x values stored in the PARI structure are those from the pure colinear 2->2 scattering before any transverse momentum kick from either the ISR or intrinsic k_T.

I want to look at the x1 and x2 values probed by the dijet asymmetry measurement. I look at the detector level simulation, and apply all dijet selection criteria and cuts as well as requiring that the detector level dijet match to a parton level dijet. For each event which passes the above criteria, I record x1 and x2, where the association between which parton corresponds to x1 and which to x2 is defined as:

For events with both jets in the east barrel: x1 is the parton with negative z momentum and x2 is the parton with positive z momentum

For events with both jets in the west barrel: x1 is the parton with positive z momentum and x2 is the parton with negative z momentum

For events with one jet in the east and the other in the west barrel: x1 is the parton with positive z momentum and x2 is the parton with negative z momentum

There appears to be two ways to get the partonic x values from the simulation which are slightly different for reasons I do not understand. I refer to these two methods as the particle method and the tree method:

Particle Method: Here I calculate x by looking at the z momentum of particles 5 and 6 (those involved in the hard scattering) from the PYTHIA record. I then calculate x by dividing that momentum by the z momentum of its parent proton (particles 1 and 2).

Tree Method: The skim trees contain a class called StPythiaEvent where various pieces of information from the pythia record are stored including the partonic A_LL information and the TParticle array for all particles. It also contains explicit values for x1 and x2 which are closely correlated, but not identical to the x values obtained from the particle method (see figures 1 and 2).

Figure 1: Scatter plots showing the relationship between the x values obtained from the Tree Method (y-axis) and the x values obtained from the Particle Method. The 8 panels each show a different detector level dijet mass bin. Both x1 and x2 are shown on the same plot.

Figure 2: The top 8 panels show the x1 distributions for the EEWW topology for the Particle (red) and Tree (blue) methods of getting x for the 8 detector level mass bins. The bottom 8 panels show the corresponding x2 distributions. The plots for the EW topology can be found here (x1 x2).

As can be seen, the differences between the Particle and Tree methods seem to be rather minor, with the largest discrepencies showing up in the low x region of the x2 EEWW distributions. I have not been able to trace back the code to determing where the Tree x1 and x2 values are filled from, so (for now at least) I will use the Particle Method from here on out as I know exactly where those numbers are coming from.

By using the Particle Method, I can also make cuts on the parton type when plotting x. For the comparisons below, I will show x values obtained from all partons and the x values obtained from only gluons.

Figure 3: The top 8 panels show the x1 distributions for the EEWW topology for all partons (red) and gluons only (blue) for the 8 detector level mass bins. The bottom 8 panels show the corresponding x2 distributions. The plots for the EW topology can be found here (x1 x2).

Figure 4: Comparison of the x1 (red) and x2 (blue) distributions for the EEWW (left hand side) and EW (right hand side) topologies for two mass bins (16-19 GeV bin is solid curves and 58-82 GeV bin is dashed curves). The top panels are for all partons and the bottom panels are for gluons only. A pdf which compares the 16-19 GeV bin with all other mass bins can be found here.

Figure 5: The average x1 (red) and x2 (blue) values as a function of the invariant mass of the detector level dijet. Left panels are for the EEWW topology and the right panels are for the EW topology. The top panels are for all partons and the bottom panels are for gluons only.

Carl wanted to see the x_Gluon distributions plotted on a linear scale. Figure 6 shows the same information shown in the bottom panel of Figure 4 in a linear scale and with the high mass histograms scaled up by a factor of 500.

Figure 6: Comparison of the x1 (red) and x2 (blue) distributions for the EEWW (left hand side) and EW (right hand side) topologies for two mass bins (16-19 GeV bin is solid curves and 58-82 GeV bin is dashed curves). The 58-82 GeV histograms have been scaled up by a factor of 500.

I wanted to look a little deeper at the difference between the Tree method (Renee confirmed that the x1 and x2 values stored in StPythiaEvent are actually PARI 33 and 34 from Pythia) and the Particle method. To do this, I generated 4 small (10 event) standalone Pythia samples with ISR and kT turned on and off:

ISR = On, kT = On

ISR = On, kT = Off

ISR = Off, kT = On

ISR = Off, kT = Off

Some quick observations: Turning off the ISR makes lines 3 & 5 and lines 4 & 6 identical as well as making PARI(31)=PARI(33) and PARI(32)=PARI(34). The Pythia manual says this should be true (Page 64), so it is nice that we see it. Even with the ISR off, the x calculated from the particle z momenta does not equal that stored in PARI(31-34) unless the primordial k_T is turned off as well. For the events with the ISR on but the k_T turned off, we see that the x calculated from the z momentum of particles 3 and 4 (before ISR is applied) is identical to the x values stored in PARI(31&32) which are the x values of the initial state parton shower initiators. However, once the ISR is applied and gives the system some transverse kick (particles 5 and 6), we see that the x values calculated from particle z and the x values in PARI(33&34) no longer match exactly.

If I am interpreting this correctly, the x values stored in the PARI structure are those from the pure colinear 2->2 scattering before any transverse momentum kick from either the ISR or intrinsic k_T.

»

- pagebs's blog
- Login or register to post comments