Pythia study for FMS simulation production

 Pythia Study for FMS Simulation Production

 

A discussed last week, I looked at two different tunes of Pythia with two different settings of primordial <kT>. They are,

1. tuneA with <kT>=1 GeV

2. tuneA with <kT>=0.5 GeV

3. Perugia0 with <kT>=1 GeV

4. Perugia0 with <kT>=0.5GeV

For all of the above, Pythia version 6.4 was used.

The comparison is against the measured pi0 cross-section at three different <eta> points, at 4.0, 3.8, and 3.3.

 

First, the comparison among the 4 settings. 

 

Fig. 1. Tune and <kT> mix and match

Observation 1: The cross-section does not depend on the choice of the tune much. 

Observation 2: The cross-section does depend on <kT>, and more so at higher <eta>.

 

Since the tune does not matter much for the current purpose, we now compare Perugia0 against the measured FPD pi0 corss-sections.

 

Fig. 2, Data vs. MC, Perugia 0

Observation 1: <kT>=1.0 GeV overshoots the data by almost an order of magnitude at <eta>=4.0.

Observation 2: <kT>=0.5 GeV still overshoots the data somewhat at <eta>=4.0, but it is below the data at <eta>=3.8. Overall, Pythia shows factor of ~2 smaller difference between 4.0 and 3.8 compared to the data.

Observation 3: <eta>=3.3 shows good agreement for both <kT>.

 

Beyond this, we want to look at what type of processes produces these observed pi0s. I've played this game a few times before, but I'll go over it again here.

We will be looking at two variables that I defined to measure how well the momentum of the pi0 in the FMS matches with initial or final state partons. 

1. pZmatch = pZ(parton) / pZ(pi0)

2. pTmatch = pT(parton) / pT(pi0) * cos (theta)

The idea is that both of these will be ~1 or greater if that parton was the one that fragmented to the observed pi0.

 

Fig. 3. pZmatch vs. pTmatch for incoming (towards the FMS) HS (Hard Scattered) final state parton. (Perugia0, <kT>=0.5 GeV)

 

As can be seen, there is a blob there indicating clear hard scattering events. But many many events do not show such pattern. 

The next thing to do is to track down the parentage and see if that works. 

 

Fig. 4. Parentage tracking through Pythia record.  (Perugia0, <kT>=0.5 GeV)

Apart from the ones that do not lead to hard scattering (which I do not understand), we see a pretty clear pattern where things either lead to a FS parton, or an incident proton. 

The obvious assumption would be that the ones leading to FS is the hard scattering, and ones leading to incident proton is the beam remnant. 

Except, it's not that simple. If we look at the pMatch plots for "Hard Scattering" and "Beam Remnant" according to the above definition,

 

Fig. 5. pTmatch vs. pZmatch for incoming HS final state parton, for pi0s whose first parent is a final state parton. (Perugia0, <kT>=0.5 GeV)

 

The HS blob is more evident, but still, we have many events that are not correlated with a FS parton. These events do not show correlation with the other FS parton either. 

 

Fig. 6. pTmatch vs. pZmatch for incoming HS final state parton, for pi0s whose first parent is an incident proton. (Perugia0, <kT>=0.5 GeV)

Some of the pi0s that lead to incident protons actually show good correlation with a final state particle.  

This is especially strange because the parentage does not propagate through the hard scattering. Oddly enough, based on the prcoess id, the events in the HS-looking blob are dominantly gg->gg events.  

I also tried a different method of parentage tracking, by recording the particle with the lowest line number (oldest) whose momentum matches the pi0 reasonably well. 

When I do this, as expected many events that lead to either FS or incident proton do not go all the way back. 

 

Fig. 7. Oldest particle with matching momentum vs. parentage by pythia record (line numbers).

 

To summarize, 

1. Parentage tracking through pythia record seems an unreliable way of telling whether a pi0 was a hard scattered or beam remnant. 

2. While momentum matching seems more robust, a significant portion of the pi0s do not match anything from HS or incident protons. 

3. For these "mystery" events, usually the pi0 momentum matches quite well to a string (id=92) that produced it, but nothing before that string has pT or pZ that is nearly enough to make that pi0. I did not search for things that have larger line numbers than the pi0, but this should not be an issue, I don't think.

 

Finally, for fun, I plotted cross-section for what I loosely considered Hard Scattering events for<kT>=0.5. The condition is that EITHER the official parentage leads to final state OR one of the FS parton matches the pi0 momentum. (Probably too loose)

 

Fig. 8.  Data vs. MC, "Probably Hard Scattering" only.

So this matches the <eta>=4.0 quite well now, and 3.3 is as good as always, but 3.8 is way below the data. 

But if I tried a much more stringent definition of HS, simulation is far below the data, and consequently also far below pQCD calculations.

 

Fig. 9.  Data vs. MC, "Definitely Hard Scattering" only.

Next, I plan to look at the "jetty" structure at Pythia level and see if that correlates with any of this confusion.